Assignment 1: exposing the gaps—how does the united states compare in

Ace your studies with our custom writing services! We've got your back for top grades and timely submissions, so you can say goodbye to the stress. Trust us to get you there!


Order a Similar Paper Order a Different Paper

This week, you explore many resources that shine a light on the health of the United States (U.S.) population as it compares to the populations of many other countries. Where is America leading the way in population health? Where do other countries outperform the U.S.? How do key health indicators differ among individual states?

Among developed nations, the U.S. is the only country that offers no support to families during the most critical periods in which health is determined (i.e., during the first months and years of a newborn’s life). Given the importance of early life, countries that use the political structure to provide support for their populations during that critical period of growth and development tend to be healthier than those who leave parenting to market forces. It is, therefore, not surprising that the U.S. ranks 33rd in life expectancy, according to the United Health Foundation (United Health Foundation, 2013).

This week you examine health indicators used to measure the health of the U.S. population and contrast them to other countries around the world. You also consider various determinants of health within different states in the U.S. as well as across continents.When the health of a population is measured by various mortality indicators such as life expectancy, infant or child mortality, or the chances of surviving to retirement, surprising trends emerge. Health, as measured by longevity, appears to be declining in substantial segments of the U.S. population, especially for women (United Health Foundation, 2013). These findings receive little attention in most public health efforts or in the mainstream media, at least in the United States.

For this Assignment, you select health indicators used to measure the health of the U.S. population and contrast them to other countries around the world. You compare various determinants of health within different states in the United States as well as across continents.

To prepare for this Assignment, complete the readings and view the media in your Learning Resources. Install the free Gapminder Desktop tool and experiment plotting different health outcomes against various determinants already loaded along the two axes. Using the various health ranking resources provided, select two key health indicators for which the United States ranks lower than other nations.

The Assignment (3–4 pages):

  • Provide a brief description of the two health indicators you selected, citing specific sources.
  • Explain how the U.S. ranks on these indicators compared to other nations.
  • Explain two factors that might influence those rankings and the relative standing of the U.S. compared to the other nations.
  • Determine which two states rank the best and which two states rank the worst for those indicators. Describe factors you believe might contribute to those relative rankings among the states.
  • Share any insights you gained or conclusions you drew as a result of making these comparisons.
  • Expand on your insights utilizing the Learning Resources.

https://www.who.int/countries/usa/

www.thelancet.com Vol 389 April 8, 2017 1453

Series

America: Equity and Equality in Health 3

Structural racism and health inequities in the USA: evidence
and interventions
Zinzi D Bailey, Nancy Krieger, Madina Agénor, Jasmine Graves, Natalia Linos, Mary T Bassett

Despite growing interest in understanding how social factors drive poor health outcomes, many academics, policy
makers, scientists, elected officials, journalists, and others responsible for defining and responding to the public
discourse remain reluctant to identify racism as a root cause of racial health inequities. In this conceptual report, the
third in a Series on equity and equality in health in the USA, we use a contemporary and historical perspective to
discuss research and interventions that grapple with the implications of what is known as structural racism on
population health and health inequities. Structural racism refers to the totality of ways in which societies foster racial
discrimination through mutually reinforcing systems of housing, education, employment, earnings, benefits, credit,
media, health care, and criminal justice. These patterns and practices in turn reinforce discriminatory beliefs, values,
and distribution of resources. We argue that a focus on structural racism offers a concrete, feasible, and promising
approach towards advancing health equity and improving population health.

Introduction
Racial and ethnic inequalities, including health
inequities, are well documented in the USA (table),1–5 and
have been a part of government statistics since the
founding of colonial America.6–8 However, controversies
abound over explanations for these inequities.6–8 In this
report, we offer a perspective not often found in the
medical literature or taught to students of health
sciences, by focusing on structural racism (panel 1)9–11 as
a key determinant of population health.9,10,12,13 To explore
this determinant of health and health equity, we examine
a range of disciplines and sectors, including but not
limited to medicine, public health, housing, and human
resources. Our focus is the USA.

Although there is growing interest in understanding
how social factors drive poor health outcomes,14 and
directed investigation in social science and social
epidemiology into the interconnected systems of
discrimination,9,10,12,13 many academics, policy makers,
scientists, elected officials, and others responsible for
defining and responding to the public discourse remain
resistant to identify racism as a root cause of racial health
inequities.9,10,13 For example, in a Web of Science search
done on Sept 7, 2016, with the term “race” in conjunction
with “health”, “disease”, “medicine”, or “public health”,
47 855 articles were retrieved. However, when “race” was
replaced by “racial discrimination”, only 2061 articles
were located, and only 1996 articles were found when it
was replaced by “racism”. Furthermore, when “race” was
replaced by “structural or systematic racism”, only
195 articles were identified (ie, 0·4% of those identified
with the search term “race”).

To date, the small body of empirical research on racial
discrimination and health has focused primarily on the
stress of perceived unfair treatment as experienced
by individuals (interpersonal racism).9,10,12,15–18 Such
inequitable suffering matters, but a broad, societal

view—one that identifies and seeks to alter how such
racism contributes to poor health—is required to
understand, prevent, and address the harms related to
structural racism. There is a rich social science literature
conceptualising structural racism,8–10,19 but this research
has not been adequately integrated into medical and
scientific literature geared towards clinicians and other
health professionals.9,10,12,13 In this report, we examine
what constitutes structural racism, explore evidence of
how it harms health, and provide examples of
interventions that can reduce its impact. Our central
argument is that a focus on structural racism is essential
to advance health equity and improve population health.

Structural racism: a brief introduction
Any account of structural racism within the USA must start
with the experiences of black people and the Indigenous
people of North America. It was on these two groups that
the initial colonisers of North America (the English, French,

Lancet 2017; 389: 1453–63

See Editorial page 1369

See Comment pages 1376
and 1378

This is the third in a Series of
five papers about equity and
equality in health in the USA

New York City Department of
Health and Mental Hygiene,
Long Island City, NY, USA
(Z D Bailey ScD, N Linos ScD,
M T Bassett MD); Department
of Social and Behavioral
Sciences, Harvard T.H. Chan
School of Public Health,
Boston, MA, USA
(Prof N Krieger PhD,
M Agénor ScD); and Bard Prison
Initiative, Annandale-on-
Hudson, NY, USA
(J Graves MPH)

Correspondence to:
Dr Mary T Bassett,
42-09 28th Street, Long Island
City, NY 11101, USA
[email protected]

See Online for infographic
www.thelancet.com/
infographics/us-health

Search strategy and selection criteria

An overarching search strategy was not used; instead, we
drew on our collective experience and specific searches for
different sections to update or amplify the completeness of
our review of the published literature. To identify review
articles on racism and health, we searched Web of Science,
PubMed, and Google Scholar using the search terms “racism
AND health” or “racial discrimination AND health” or
“structural racism AND health”. Only review articles
published in English between Jan 1, 2000, and Feb 23, 2016,
were considered. We identified additional sources by
performing selected searches in the databases listed above
and the Google and DuckDuckGo search engines. These
searches were further supplemented from our own
knowledge of this subject.

Series

1454 www.thelancet.com Vol 389 April 8, 2017

Dutch, and Spanish) first promulgated genocide and
enslavement, and created both legal and tacit systems of
racial oppression.8,20,21 Our report focuses primarily on the
experiences of black Americans, since most research on
racism and health has focused on this racialised group. We
recognise, however, that Native Americans and other people
of colour in the USA—including Latinos, Asian Americans,
and Pacific Islanders—have also been the target of health-
harming racial discrimination, combined with anti-
immigrant and religious (eg, anti-Muslim) discrimination.8
Although issues of immigration and nativism are beyond
the scope of this report, our analysis is applicable to the
structural discrimination experienced not only by these
groups but also by societally defined and racialised groups
in other countries with systems of oppression that have led
to health inequities.9,14,16,22

Racial ideology and the categorisation of racialised
social groups
As with many other race-conscious societies, the USA has a
long history as a slaveholding republic and as a colonial-
settler nation.8,19–21 The modern concept of “race” emerged at
the cusp of the country’s nationhood, as early European
settlers sought to preserve an economy largely on the basis
of the labour of enslaved African people and their
descendants while upholding the universal rights of

“man”.6,8,19,23,24 To reconcile this contradiction, the colonists
established legal categories based on the premise that black
and Native American individuals were different, less than
human, and innately, intellectually, and morally inferior—
and therefore subordinate—to white individuals.8,19–21,23
Buttressing this concept of racial classification has been a
long legacy of now discredited scientific theory and inquiry,
constructed around the primary assumption that “race”
was an innate and fixed characteristic and an inherently
hierarchical category.6,8,9,19,23 This manufactured concept of
race used ostensibly visible phenotypic characteristics and
ancestry to justify systems of oppression and privilege.6,8,19
Similar processes in other racialised societies, such as
those of South Africa and Brazil, have produced country-
specific racial hierarchies, which ascribe human value on
the basis of proximity to whiteness.22 Furthermore, since
the 18th century, scientific racism rooted in Aryan or white
supremacy became a blueprint for many other mani-
festations of society-specific scientific racism around the
world.6,22,25

The continuing role of ostensibly colour-blind laws and
policies
In the USA, since the passage of the 1960s civil rights
laws,8,20 government complicity in the promotion of racial
discrimination is typically viewed as belonging to the
past. Examples of such de jure discrimination include
the legalisation and enforcement of slavery, the Jim Crow
laws enacted in the 1870s (which legalised racial
discrimination in reaction to the civil rights and social
gains attained by the newly freed black population in the
short Reconstruction period after the US Civil War), the
forcible removal of Indigenous people from their lands,
and the forcible transfer of Indigenous children from
their families to punitive so-called boarding schools
designed to strip them of their culture.8,19–21,26,27

However, this standard view overlooks the long reach of
past practices and the impact of contemporary practices of
institutional racism in both the public and private sector;
such practices have been and continue to be realised by
purportedly colour-blind policies that do not explicitly
mention “race” but bear racist intent or consequences, or
both.28–30 Institutional racism in one sector reinforces it in
other sectors, forming a large, interconnected system of
structural racism whereby unfair discriminatory practices
and inequities in the health and criminal justice systems
and in labour and housing markets bolster unfair
discriminatory practices and inequities in the educational
system, and vice versa.10 One key example, with ongoing
intergenerational effects, is the historic Social Security Act
of 1935, which created an important system of
employment-based old-age insurance and unemployment
compensation.8,20 The Act also, however, deliberately
excluded agricultural workers and domestic servants—
occupations largely held by black men and women. This
accommodation was made to secure the votes of
Democrats in the South and thus ensure passage of the

Key messages

• Racial/ethnic health inequities in the USA are well documented, but controversies over
explanations of these inequities persist.

• To date, in the small body of empirical research on racism and health, most studies
have focused on interpersonal racial/ethnic discrimination, with comparatively less
emphasis on investigating the health effects of structural racism.

• Structural racism involves interconnected institutions, whose linkages are historically
rooted and culturally reinforced. It refers to the totality of ways in which societies
foster racial discrimination, through mutually reinforcing inequitable systems
(in housing, education, employment, earnings, benefits, credit, media, health care,
criminal justice, and so on) that in turn reinforce discriminatory beliefs, values, and
distribution of resources, which together affect the risk of adverse health outcomes.

• One example of structural racism pertains to the ongoing residential segregation of black
Americans, which is associated with adverse birth outcomes, increased exposure to air
pollutants, decreased longevity, increased risk of chronic disease, and increased rates of
homicide and other crime. Residential segregation also systematically shapes health-care
access, utilisation, and quality at the neighbourhood, health-care system, provider, and
individual levels.

• Several avenues exist for potentially efficacious solutions, including the use of a
focused external force that acts on multiple sectors at once (eg, place-based
multisector initiatives such as Purpose Built Communities, Promise Neighborhoods,
and Choice Neighborhoods), disruption of leverage points within a sector that might
have ripple effects in the system (eg, reforming drug policy and reducing excessive
incarceration), and divorcing institutions from the racial discrimination system
(eg, by training the next generation of health professionals about structural racism).

• A focus on structural racism offers a concrete, feasible, and promising approach towards
advancing health equity and improving population health. Without a vision of health
equity and the commitment to tackle structural racism, health inequities will persist.

Series

www.thelancet.com Vol 389 April 8, 2017 1455

Act. This racially motivated exclusion afforded the
primarily white recipients additional opportunities to
acquire wealth and pass it on to their children, while those
excluded were unable to do so and instead often became
dependent on their children after retirement, thereby
further curtailing the intergenerational accumulation of
assets.8,20 The net result has been an entrenchment of
racial economic inequities that persist to this day.8,10,20,29,30

Another example is the War on Drugs and tough-on-
crime policies enacted in the 1970s and 1980s (labelled
“The new Jim Crow”).28 Without ever referring to “race” by
itself, these policies stereotyped black Americans as drug
addicts—despite similar prevalence of illicit drug use
among white Americans—and disproportionately targeted
black people for incarceration.28,30 The legacy of these
policies is that the annual rate of incarceration of black
men is 3·8–10·5 times greater than that of white men,
across all age groups;31 moreover, in 2014, almost 3% of all
black men in the USA were serving sentences of at least
1 year in prison.31

Structural racism in the private sector
Institutional racism also continues unabated in the
private sector, especially in housing and employment,
underpinning the structural racism of the ostensibly
colour-blind policies in the public sector.32–34 In their
review of the evidence on discrimination in four domains
(employment, housing, credit markets, and consumer
markets), Pager and Shepherd33 argue that discrimination

in the rental and housing markets against black and
Latino communities remains pervasive, even though
intentional redlining is no longer legal (the term
redlining is derived from the legal practice initiated in
1934 by the Federal Housing Administration, which
involved marking maps with red lines to delineate
neighbourhoods where mortgages were denied to
marginalised, racialised groups to steer them away from

Total White non-Hispanic Asian* Hispanic or
Latino

Black non-
Hispanic†

Native American
or Alaska Native

Wealth: median household assets (2011) $68 828 $110 500 $89 339 $7683 $6314 NR

Poverty: proportion living below poverty level, all ages
(2014); children <18 years (2014)

14·8%; 21·0% 10·1%; 12·0% 12·0%; 12·0% 23·6%; 32·0% 26·2%; 38·0% 28·3%; 35·0%

Unemployment rate (2014) 6·2% 5·3% 5·0% 7·4% 11·3% 11·3%

Incarceration: male inmates per 100 000 (2008) 982 610 185 836 3611 1573

Proportion with no health insurance, age <65 years (2014) 13·3% 13·3% 10·8% 25·5% 13·7% 28·3%

Infant mortality per 1000 livebirths (2013) 6·0 5·1 4·1 5·0 10·8 7·6

Self-assessed health status (age-adjusted): proportion with
fair or poor health (2014)

8·9% 8·3% 7·3% 12·2% 13·6% 14·1%

Potential life lost: person-years per 100 000 before the age
of 75 years (2014)

6621·1 6659·4 2954·4 4676·8 9490·6 6954·0

Proportion reporting serious psychological distress‡ in the
past 30 days, age ≥18 years, age-adjusted (2013–14)

3·4% 3·4% 3·5% 1·9% 4·5% 5·4%

Life expectancy at birth (2014), years 78·8 79·0 NR 81·8 75·6 NR

Diabetes-related mortality: age-adjusted mortality per
100 000 (2014)

20·9 19·3 15·0 25·1 37·3 31·3

Mortality related to heart disease: age-adjusted mortality
per 100 000 (2014)

167·0 165·9 86·1 116·0 206·3 119·1

NR=not reported. *Economic data and data on self-reported health and psychological distress are for Asians only; all other health data reported combine Asians and Pacific Islanders. †Wealth, poverty, and
potential life lost before the age of 75 years are reported for the black population only; all other data are for the black non-Hispanic population. ‡Serious psychological distress in the past 30 days among adults
aged 18 years and older is measured using the Kessler 6 scale (range=0–24; serious psychological distress: ≥13). Sources: wealth data taken from the US Census;1 poverty data for adults taken from the National
Center for Health Statistics,2 and poverty data for children taken from the National Center for Education Statistics;3 unemployment data taken from the US Bureau of Labor Statistics;4 incarceration data taken
from the Kaiser Family Foundation;5 data on uninsured individuals taken from the National Center for Health Statistics;2 data on infant mortality, self-assessed health status, potential life lost, serious
psychological distress, life expectancy, diabetes-related mortality, and mortality related to heart disease taken from the National Center for Health Statistics.2

Table: Social and health inequities in the USA

Panel 1: Definitions of structural racism and institutional racism

Many academics use structural racism and institutional racism interchangeably, but we
consider these terms as two separate concepts.

Structural racism refers to “the totality of ways in which societies foster [racial]
discrimination, via mutually reinforcing [inequitable] systems…(eg, in housing,
education, employment, earnings, benefits, credit, media, health care, criminal justice,
etc) that in turn reinforce discriminatory beliefs, values, and distribution of resources”,
reflected in history, culture, and interconnected institutions.9 This definition is similar to
the “über discrimination” described by Reskin.10

Within this comprehensive definition, institutional racism refers specifically to racially
adverse “discriminatory policies and practices carried out…[within and between
individual] state or non-state institutions” on the basis of racialised group membership.9

Some of these institutional policies and practices explicitly name race (eg, de jure Jim
Crow laws, which required schools and medical facilities to be racially segregated, and
restricted certain neighbourhoods to be white-only), but many do not (eg, employer
practices of screening applications on seemingly neutral codes, such as telephone area
codes or ZIP codes, because of presumptions about which racial groups live where).11

Series

1456 www.thelancet.com Vol 389 April 8, 2017

white neighbourhoods). Additionally, strong evidence
from experimental audit studies reveals continued racial
discrimination in hiring decisions. In one study that
used identical résumés, which differed only in the name
of the applicant, hiring managers called back those with
traditionally white names (eg, Brad or Emily) 50% more
often than those with traditionally black names (eg,
Jamal or Lakisha).33 In another study that used mailed
résumés, white applicants with criminal records were
called back more often than were black applicants
without criminal records.33 Ongoing de facto racial
segregation in the workforce is partly why black
Americans, on average, have lower wages than those of
white Americans.35

As this brief summary suggests, structural racism is an
ongoing—and not just historical—concern across
multiple systems. We next consider the implications of
such systemic racism on population health.

Health consequences of structural racism:
evidence and evidence gaps
Contemporary scholarship has established multiple
pathways by which racism harms health, involving
adverse physical, social, and economic exposures, as well
as maladaptive coping behaviours and stereotype threats
(panel 2).9,12,13,15–18,21,30,32–50 Typically concurrent, these
exposures can accumulate over the life course and across
generations.

To date, research on racial discrimination and health
has focused primarily on interpersonal discrimination as
a psychosocial stressor.9,16–18 The strongest evidence in the
scientific literature is for adverse effects on psychological
wellbeing, mental health, and related health practices
(eg, sleep disturbance, eating patterns, and the
consumption of psychoactive substances, including
cigarettes, alcohol, and drugs), as summarised in
panel 3.9,12,15,16,18,35,51–58 Furthermore, growing research is
linking interpersonal racism to various biomarkers of
disease and wellbeing, including allostatic load,
inflammatory markers, and hormonal dysregulation.16,18

Here, we focus instead on adverse health effects of
structural racism through two distinct but related pathways
emphasised in the literature: residential segregation and
health-care quality and access.9,12,13,18 Both of these
pathways include actionable leverage points to reduce
exposure and promote health equity. A third relevant
pathway, discriminatory incarceration,28,30,35 is only briefly
mentioned since it is discussed elsewhere in this Series
by Wildeman and Wang.59

Residential segregation
As a reflection and reinforcement of structural and
institutional racism, most residents in the USA have
grown up in, and continue to live in, racialised and
economically segregated neighbourhoods.29,33,34,60 Analysis
of 2010 US Census data has found that “the average
white person in metropolitan America lives in a
neighborhood that is 75% white”, whereas “a typical
African American lives in a neighborhood that is only
35% white (not much different from 1940) and as much
as 45% black”.61 The literature on racial residential
segregation and poor health32,34,36,37,62–68 examines several
direct and indirect pathways through which structural
racism harms health, including the high concentration
of dilapidated housing in neighbourhoods that people of
colour reside in,62,63 the substandard quality of the social64
and built65 environment, exposure to pollutants and
toxins,36,37,65 limited opportunities for high-quality
education and decent employment,34,66 and restricted
access to quality health care.65 Health outcomes
associated with residential segregation documented

Panel 2: Pathways between racism and health9,12,13,16–18

Economic injustice and social deprivation8,9,12,32–35

Examples include residential, educational, and occupational segregation of marginalised,
racialised groups to low-quality neighbourhoods, schools, and jobs (both historical
de jure discrimination and contemporary de facto discrimination), reduced salary for the
same work, and reduced rates of promotion despite similar performance evaluations

Environmental and occupational health inequities9,36–38

Examples include strategic placement of bus garages and toxic waste sites in or close to
neighbourhoods where marginalised, racialised groups predominantly reside, selective
government failure to prevent lead leaching into drinking water (as in Flint, MI, in 2015–16),
and disproportionate exposure of workers of colour to occupational hazards

Psychosocial trauma9,15,16,18

Examples include interpersonal racial discrimination, micro-aggressions (small, often
unintentional racial slights and insults, such as a judge asking a black defence attorney
“Can you wait outside until your attorney gets here?”), and exposure to racist media
coverage, including social media

Targeted marketing of health-harming substances9,30,39

Examples include legal substances such as cigarettes and sugar-sweetened beverages, and
illegal substances such as heroin and illicit opioids

Inadequate health care9,17,40–45

Examples include inadequate access to health insurance and health-care facilities, and
substandard medical treatment due to implicit or explicit racial bias or discrimination

State-sanctioned violence and alienation from property and traditional lands9,21,30,46–48

Examples include police violence, forced so-called urban renewal (the use of eminent
domain to force the relocation of urban communities of colour), and the genocide and
forced removal of Native Americans

Political exclusion49,50

Examples include voter restrictions (eg, for former felons and through identification
requirements)

Maladaptive coping behaviours9,16,18

Examples include increased tobacco and alcohol consumption on the part of
marginalised, racialised groups

Stereotype threats15–18

Examples include stigma of inferiority, leading to physiological arousal, and an impaired
patient–provider relationship

Series

www.thelancet.com Vol 389 April 8, 2017 1457

among black Americans include adverse birth
outcomes,32 increased exposure to air pollutants,36
decreased longevity,34,66 increased risk of chronic
disease,32,34,64 and increased rates of homicide and other
crime.66,67 These adverse outcomes far outweigh any
benefits deriving from social support or political power
that accrue from the clustering of black Americans (or
other oppressed racialised groups) in adjoining
neighbourhoods.63,68 Residential segregation is thus a
foundation of structural racism and contributes to
racialised health inequities.

Moreover, analysis of residential segregation requires
addressing the intertwined occurrences of residential
segregation by both racialised group and class.60,69,70 In the

USA there has been a shift from macrosegregation to
microsegregation, whereby “blacks and whites became
more evenly distributed across states and counties during
the first two-thirds of the twentieth century, [and] … less
evenly distributed at the city and neighborhood levels”.60
Highlighting the need to think about smaller geographies,
researchers have also noted that, as income inequality has
increased, people at the top and bottom of the
socioeconomic distribution have increasingly become
spatially isolated,69,70 such that “middle-class blacks are less
able than their white counterparts to translate their higher
economic status into desirable residential conditions”.34

In recognition of the trend towards microsegregation
and increased social polarisation, public health

Panel 3: Dominant approaches to studying racial discrimination as a psychosocial stressor and associated adverse health
outcomes, with counterexamples of research on measures of structural racism

Racism and stress
To date, racism has primarily been conceptualised as a
psychosocial stressor in the health science literature, and the
strongest and most consistent evidence of its adverse health
effects concerns mental health, as detailed in several
comprehensive, systematic reviews.9,12,15,16,18 In one such review,16
published in 2015, the authors found that self-reported racism
was positively associated with increased levels of negative
mental health, including all individual mental health outcomes
except for positive affect (eg, depression, anxiety, distress,
psychological stress, negative affect, and post-traumatic stress),
and negatively associated with positive mental health (eg,
self-esteem, life satisfaction, control and mastery, and
wellbeing). After adjusting for publication bias, the association
between reported racism and mental health remained twice as
large as that for physical health, which was driven primarily by
obesity outcomes. There is growing evidence that experiences
of racism are associated with poor sleep outcomes, which could
be linked to both mental and physical health.51

Stress pathways
Much of the research on interpersonal racism and health has
posited that racism is a social stressor that operates through
diverse stress pathways, including physiological, psychological,
and behavioural pathways. Experiences that are perceived as
racist act as social stressors, which can initiate a set of
neurobiological and behavioural responses (ie, coping
behaviours) that can affect mental and physical health. These
experiences can be chronic and include everyday hassles of
receiving poor service at restaurants, being followed or not
helped in stores, and generally being treated with less respect
and consideration than others. Acute experiences of violence,
harassment, and other threatening behaviour are also included
in this category. However, although such exposures are most
likely to garner media attention, the common, chronic
experiences of discrimination are more consistently associated
with poor health outcomes than are acute experiences,9,15,16,18

probably reflecting how brain chemistry and general

metabolism change in response to chronic stressors.15 There is
burgeoning evidence linking experiences of discrimination to
biomarkers of disease and wellbeing, including allostatic load,
telomere length, cortisol dysregulation, and inflammatory
markers.9,16,18

Reliance on self-reports of exposure to racial discrimination
Most of the research on racial discrimination and health has
relied on self-reported measures, although some studies have
used vignettes or experimental situations. Evidence suggests
that because of well known cognitive biases, including social
desirability, self-reported data are likely to provide an
underestimate of actual exposure, leading to underestimates of
the magnitude of the association of racial discrimination with,
and its impact on, adverse health outcomes.9,18 Some immigrant
groups, moreover, might be less likely than others to recognise
racist interactions, or less likely to attribute discriminatory
behaviour to racism as opposed to language skills, immigration
status, or chance.9,52

Counterexamples of research on measures of structural racism
Although small in comparison with psychosocial approaches,
an emerging body of research has begun to investigate the
relationship between health and four domains of state-level
structural racism: political participation, employment and job
status, educational attainment, and judicial treatment,
including incarceration.9,12,16,35,53–58 Black people living in states
with higher levels of structural racism in these domains were
more likely than those living in states with lower levels of
structural racism to self-report a myocardial infarction in the
previous year; meanwhile, the same association for white
people was null or protective.57Another study that used the
same measures found a positive association between structural
racism at the state level and the odds of births that were small
for gestational age in both black and white women.58 Such
measures could be used to build the evidence base regarding
the connections between structural and institutional racism
and health, and highlight areas for intervention. Priority should
be given to expanding this type of research.

Series

1458 www.thelancet.com Vol 389 April 8, 2017

researchers have recently begun to use the Index of
Concentration at the Extremes (ICE).70 This measure was
introduced into the sociological literature in 200169 and
was designed to measure economic polarisation—the
extent to which a population is concentrated into the
extremes of wealth or impoverishment—by taking the
difference between the number of affluent and poor
households in an area and dividing it by the total number
of households in the area.70 Moreover, these areas can be
measured at multiple levels (eg, census tract, city
neighbourhood, and county). New innovations include
the development of an ICE for racialised economic
segregation, which uses data on the joint distribution of
income and race/ethnicity. Research done in New York
City, for example, has shown that ICE measures that
captured both income and racialised group yielded larger
risk ratios, at both the neighbourhood and census tract
levels, for infant mortality, premature mortality, and
diabetes mortality than an ICE solely for income or the
poverty level.70

Underscoring the need for explicit analysis of the
health burden of residential segregation (regardless of
how it is measured) and neighbourhood disinvestment,
there is evidence to suggest that these structurally driven,
place-based exposures harm economic opportunity and,
when coupled with inadequate gun control, contribute to
the lethal burden of gun violence and crime in
predominantly black and Latino neighbourhoods71,72 and
in impoverished Native American reservations.21 In turn,
the violence and crime in these neighbourhoods
reinforces the intergenerational legacy of racialised
punitive policing,8,20,21,28,31 perpetuating vicious cycles of
further community depletion and adverse health
outcomes.8,9,28,30,31,35,59

Discriminatory incarceration
The penal institutions that constitute the US criminal
justice system—police departments, court systems,
correctional agencies, parole and probation departments,
and sentencing boards—have established policies and
practices that are ostensibly colour-blind yet they
criminalise communities of colour (eg, through day-to-
day practices such as stop and frisk) and disproportionately
incarcerate black men, women, and children.30 As
reviewed in this Series by Wildeman and Wang,59 each
component of the criminal justice continuum—from
arrest to re-entry—carries various health consequences,
and a growing body of literature has documented severe
adverse health outcomes associated with incarceration
on the individual, their families, and neighbourhoods.
What should not be lost in the explication of these
outcomes is their roots in structural racism; the present
disproportionate representation of black people in the
penal system is reminiscent of the Black Codes and
convict leasing practices from the colonial period.8,26 New
freedoms afforded to black people following the US Civil
War were promptly undone by laws that selectively

criminalised unemployment, vagrancy, and loitering.26
The resultant prison population effectively re-established
free labour for Southern states to rebuild infrastructure.73
The effects of mass incarceration, as traced by Wildeman
and Wang59 from the 1970s, are best understood as a
continuation of racialised imprisonment8,10,20 rather than
as an emergent process.28 Moreover, as noted previously,
strong feedback mechanisms exist between inequities in
incarceration, employment, and health on a population
level.30,35,59

Health-care quality and access
Interpersonal racism, bias, and discrimination in health-
care settings can directly affect health through poor
health care. Almost 15 years ago, the Institute of Medicine
Report titled Unequal Treatment: Confronting Racial and
Ethnic Disparities in Health Care40 documented systematic
and pervasive bias in the treatment of people of colour,
resulting in substandard care. Evidence continues to
support this finding.41–44

However, it would be short sighted to view these
problems solely as a matter of institutional and
interpersonal discrimination within health-care
settings.17,40–44 Instead, it is essential to understand the
broad context within which health-care systems operate,
including the potentially disparate settings in which
health-care professionals and their patients reside.
Specifically, residential segregation systematically shapes
health-care access, utilisation, and quality at the
neighbourhood, health-care system, provider, and
individual levels.45 The socioeconomic disadvantage
resulting from systematic disinvestment in public and
private sectors renders it difficult to attract primary-care
providers and specialists to predominantly black
neighbourhoods.40,45 Likewise, health-promoting resources
are inadequately invested into these neighbourhoods.
Health-care infrastructure and services are inequitably
distributed, resulting in predominantly black neighbour
hoods having lower-quality facilities with fewer clinicians
than those in other neighbourhoods. Moreover, most of
these clinicians have lower clinical and educational
qualifications than those in other neighbourhoods. This
inequitable system is likely to disproportionately expose
black residents to racially biased services.45

Addressing structural racism to advance health
equity
Although efforts to counter institutional racism and
residential segregation in the housing market and
medical care system require initiatives focused on these
institutions, such initiatives are not sufficient. Also
needed is intersectoral work, especially that which is
guided by transdisciplinary frameworks and action.
Analytical insights derived from a systems perspective
suggest several avenues for efficacious solutions,
including the use of a focused external force that acts on
multiple subsystems (ie, sectors) at once, disruption of

Series

www.thelancet.com Vol 389 April 8, 2017 1459

leverage points (ie, key points of intervention within a
sector that could be important for maintenance of the
system, both within and outside the particular sector in
question), and divorcing institutions from the racial
discrimination system.10 We highlight some promising,
concrete, intersectoral examples of each of these types of
solutions, which have the potential to reduce, if not
remove, the burden of structural racism on population
health.

Place-based, multisector, equity-oriented initiatives
Health and health equity are substantially influenced by
the places where people live, work, play, and pray.14 Yet,
the USA has high levels of racialised economic
segregation.69,70 Within this context, multisector, place-
based partnerships focusing on equity can be an effective
means of placing pressure on the systems of structural
racism operating in a specific geographical region.
Place-based initiatives create structures for reinvesting in
neighbourhoods that have long been sidelined. Several
initiatives have combined public and private partners
from multiple sectors to achieve community-specific
changes.74 These community-specific, multisector
interventions that seek neighbourhood-wide coverage
have thus far focused primarily on predominantly black
and Latino neighbourhoods, and also on Native American
reservations, that have experienced high levels of poverty,
health-limiting built environments, and substandard
resources for schools and housing as a result of
generations of structural racism.

Established in 2009, Purpose Built Communities is
exploring the redevelopment of more than 20 high-need
neighbourhoods with the use of a model based on their
original 1995 development site: the East Lake
neighbourhood of Atlanta, GA.74 About 20 years ago, a
private philanthropist partnered with the president of the
Atlanta Housing Authority, a resident leader, and several
community business leaders to revitalise the area by
razing a violent, poorly maintained public housing
development and rebuilding a new mixed-income
development, which involved temporary displacement of
residents during construction. Unlike other attempts at
rebuilding public housing, this development’s planning
and rollout was organised and backed by a dedicated
non-profit and focused on high-quality construction and
on safe walkways and streets. The effort included a
cradle-to-college educational curriculum, and a
combination of facilities, programmes, and services
prioritised by community residents to promote healthy
behaviours, create jobs, and reduce crime in the short
term, and break the cycle of intergenerational poverty
concentrated in this community in the long term.74

With active involvement of community residents, by
2015, crime had declined by 95% (compared with a
50% overall decline in Atlanta), the employment rate
among families in public housing increased from 13%
to 70%, capital investments increased from no

investment (over the course of 30 years before the
project) to US$123 million, property values in the
surrounding area increased, and new grocery stores,
banks, and other businesses opened.74 The evidence of
changes in the social determinants related to health
inequities is striking; to date, no health impact
assessment has been done, although it is clearly
warranted. Future place-based interventions should
build in health equity impact assessments from the
start. Two federal initiatives launched in 2010 have
followed similar principles: the US Department of
Education’s Promise Neighborhood initiative and the
US Department of Housing and Urban Development’s
Choice Neighborhood initiative. Results of health impact
assessments are eagerly awaited.

Short of full-scale community redevelopment, data
suggest that improvements in housing lead to
improvements in health. In New York City, individuals
and families on a low income are able to enter lotteries
for affordable housing units. Data from the New York
City Housing and Neighborhood Study,75 which assessed
the impact of re-housing on those who won the lottery
compared with those who did not, showed reductions in
depression and asthma exacerbations. Although results
among adolescents were mixed, findings from the
Moving to Opportunity study,76,77 in which vouchers for
housing were randomly allocated, suggest that housing
mobility policies that enable voluntary movement out of
deprived neighbourhoods can result in long-term
improvements in health and social outcomes.

Building government and public support for large-
scale initiatives to counter structural racism is both
necessary and possible. In May, 2016, the Government
Alliance for Race and Equity (GARE) and the non-profit
Living Cities jointly launched Racial Equity Here, a
$3 million initiative to help five cities (Albuquerque, NM,
Austin, TX, Grand Rapids, MI, Louisville, KY, and
Philadelphia, PA) improve racial equity, building on
approaches such as Seattle’s Race and Social Justice
Initiative, which has explicitly recognised the links
between racial equity and health equity.78 As the Mayor of
Austin, Steve Adler, noted, “Government helped create a
lot of the inequities, it institutionalized them. It’s
important for the government, the city government to
address racial inequity, not just because of the conditions,
but also because we helped create it.”78

Advocating for policy reform
With the recognition that mass incarceration is a system
used to subordinate black people,10,28,30 efforts to reduce
discriminatory criminal sanctions on drug use (a leverage
point) are also beginning to gain traction. From the 1980s
to 2010, the federal government sentencing guidelines
mandated penalties for crimes related to crack cocaine
(a cheaper formulation more common in black
communities than in other communities) that were
100 times harsher than sentences for crimes involving

For more on Seattle’s Race and
Social Justice Initiative see
http://www.seattle.gov/rsji

For more on Promise
Neighhorhoods see
https://www2.ed.gov/programs/
promiseneighborhoods/index.
html

For more on Choice
Neighborhoods see
https://portal.hud.gov/
hudportal/HUD?src=/program_
offices/public_indian_housing/
programs/ph/cn

Series

1460 www.thelancet.com Vol 389 April 8, 2017

the pharmacologically identical substance in powder
form, effectively targeting black people for prolonged
prison sentences.30 In the first sentencing breakthrough
in decades—the Fair Sentencing Act of 2010—the crack-
to-powder penalty ratio was reduced to 18:1, shrinking
the disparity but not eliminating it.30 Meanwhile,
prescription opioids, which are fuelling the current
opioid epidemic among white people, have been relatively
unregulated. It was not until opioid addicts from white
communities started being incarcerated and dying in
large numbers that the national narrative shifted from
penalisation to treatment—a clear demonstration of the
racialised nature of the War on Drugs.79

The past decade has also witnessed new bipartisan
efforts, across the country, to reduce the number of
people who are imprisoned. For example, California has
sought to address its unconstitutionally overcrowded
prisons through several legislative initiatives, including
Proposition 47.80 This ballot initiative, passed in
November, 2014, commutes drug possession felonies
(and a few minor offenses) to misdemeanours. It also
allows people serving a sentence for an eligible felony
conviction to petition the court for resentencing. With
the disproportionate impact of drug arrests, prosecutions,
and convictions on black and Latino men and women,
Proposition 47 is likely to reduce racial inequities in
sentencing. Since 2014, more than 4000 people have been
released under this initiative and California has reduced
overcrowding in prisons; however, racial inequities and
health effects have not yet been assessed.81

Training the next generation of health professionals
Structural racism has developed over centuries and is
deeply embedded in the thoughts and behaviours of people
in the USA and other countries,6,8,10,22,25 with its influence
extending to how health sciences are taught and the
routine practices of health agencies and health-care
providers.6,7,13,82–85 An analysis of structural racism is
required to recognise these problems and change them.
Fortunately, a new wave of public health and medical
students, galvanised by protests over police killings and
the Black Lives Matter movement, have been advocating to
ensure that medical and public health schools incorporate
essential pedagogy about racism and health into standard
coursework, as one step towards divorcing medical and
public health institutions from their supportive roles in the
system of structural racism.13,82–84,86 Similarly, several public
health agencies have begun to reform their institutional
structure and organisational culture.

The standard practice for teaching about race and
health in medical and public health schools is one in
which race is often discussed, but conversations about
racism are sidelined, with scant hours (if any) devoted to
social epidemiologists, medical anthropologists, social
scientists, or historians who focus on racism and
health.82–84 Few scientific and medical textbooks include
discussions of how racism affects the conceptualisation

of race or an analysis of racial inequality in relation to
health and other outcomes.85 Although many medical
schools now include diversity training and provide
instruction on cultural competency, such instruction is
often brief (and sometimes delivered online). Moreover,
the programmes typically focus on individual
responsibility to counteract interpersonal discrimination;
the goal is for individuals to increase their sensitivity to,
and knowledge about, other racial/ethnic groups.87,88 The
emphasis is therefore on “others”, in a way that could
inadvertently contribute to racial stereotyping, as opposed
to critical self-reflection about the participants’ positions
in their societies’ race relations.

By contrast, approaches based on structural
competency,83 cultural humility,89 and cultural safety46,90,91—
which have been implemented in health professionals’
training in several countries such as Canada and New
Zealand—encourage a lifelong commitment to self-
reflection and mutual exchange in engaging power
imbalances along the lines of cultural differences. These
approaches emphasise the value of gaining knowledge
about structural racism, internalised scripts of racial
superiority and inferiority, and the cultural and power
contexts of health professionals and their patients or
clients. Tying interactions between patients and health-
care providers to population-level inequalities requires
skilled instruction and considerable time, far beyond that
patched together for short training courses in cultural
competency.83 These approaches also require that health
professionals be informed by scholarship from diverse
disciplines about the origins and perpetuation of—as well
as remedies to counter—structural racism. It remains the
charge of those committed to exploring and reversing
structural racism to connect how these forms of social
inequality translate into health and health-care inequities,
within and across generations.9,13,82,86

Professional education about structural racism after
graduate school also matters, especially for clinical and
public health practitioners whose decisions affect peoples’
health daily.13,92 As Hardeman and colleagues13 advocate,
health professionals already practising in the field can
still “learn, understand, and accept” the contemporary
and historical basis of structural racism in the USA,
understand how structural racism shapes our overarching
narrative around inequities, define and call out racism
when it is present, and contribute to the understanding of
equity through clinical care and health research from the
perspective of marginalised groups and with a healthy
dose of cultural humility. Several local health departments
have already incorporated anti-racism training into staff
professional development, and introduced internal
reforms to drive organisational change.92,93 For example,
in the mid-1990s the Alameda County Public Health
Department began to place neighbourhood offices in
areas with poor health outcomes. Over time, these offices
drove changes in the department, including additional
community involvement, staff trainings on anti-racism, a

Series

www.thelancet.com Vol 389 April 8, 2017 1461

new unit and a strategic plan to incorporate equity into
their work, and an increased presence of the health
department in local activism.92 The Boston Public Health
Commission has also engaged in organisational change,
launching a Racial Justice and Health Equity Initiative
that incorporates an anti-racism advisory committee, the
development of a health equity framework, anti-racism
training and professional development, and a forthcoming
evaluation of its activities.93 As institutional reform is
closely associated with other models of productive
practices—including quality improvement, collective
impact, community engagement, and community
mobilisation—application of an anti-racism lens should
not only be judged on its moral merits but also on its
contributions to organisational effectiveness. We
anticipate that forthcoming evidence will continue to
support the view that removing racism from institutions
is essential to protect and promote the health of our
increasingly diverse communities.

Conclusion
Since the American colonial period, public and private
institutions have reinforced each other, maintaining
racial hierarchies that have allowed white Americans,
across generations, to earn more and consolidate more
wealth than non-white Americans, and maintain political
dominance. This structural racism has had a substantial
role in shaping the distribution of social determinants of
health and the population health profile of the USA,
including persistent health inequities. The stark reality is
that research investigating the relationship between
structural racism and population health outcomes has
been scant, and even less work has been done to assess
the health impacts of the few interventions and policy
changes that could help dismantle structural racism.

We can, however, look to history as a guide. Notably,
the handful of studies on the impact of the abolition of
Jim Crow laws have consistently shown improvements in
mortality in the black community, and converging
mortality between black and white communities in the
15 years after the passage of the 1964 Civil Rights Act.53–56
We recognise that efforts to implement reforms to
dismantle structural racism have repeatedly encountered
serious obstacles and backlash from institutions,
communities, and individuals seeking to preserve their
racial privilege.8,20,26,30 However, as Frederick Douglass
famously said in his 1857 address on the struggle against
slavery in the USA, the West India emancipation, and the
backlash that ensued: “Power concedes nothing without
a demand.”94

Without a vision of health equity and the commitment
to tackle structural racism, health inequities will persist,
thwarting efforts to eliminate disparities and improve the
health of all groups—the overarching goals for US health
policy as enunciated by the official Healthy People 2020
objectives. The challenge is great, but rising to this
challenge lies at the heart of our mission and our

commitment, as health professionals, to prevent
avoidable suffering, care for those who are unwell, and
create conditions in which all can truly thrive.
Contributors
All authors contributed to the conceptualisation of the manuscript,
literature search, and writing of this report. ZDB, NK, and MTB took the
lead in ensuring coherence of the text, including the selection of
appropriate data, and in data interpretation.

Declaration of interests
We declare no competing interests.

Acknowledgments
NK’s work is supported in part by an American Cancer Society Clinical
Research Professor Award.

References
1 US Census Bureau. Detailed tables on wealth and ownership assets:

2011. http://www.census.gov/people/wealth/data/dtables.html
(accessed Jan 25, 2017).

2 National Center for Health Statistics. Health, United States,
2015: with special feature on racial and ethnic health disparities.
May, 2016. http://www.cdc.gov/nchs/data/hus/hus15.pdf (accessed
Jan 25, 2017).

3 National Center for Education Statistics. Family characteristics of
school-age children. May, 2016. http://nces.ed.gov/programs/coe/
pdf/coe_cce.pdf (accessed Jan 25, 2017).

4 Bureau of Labor Statistics. Labor Force Characteristics by Race and
Ethnicity, 2014. Report 1057. November, 2015. https://www.bls.gov/
opub/reports/race-and-ethnicity/archive/labor-force-characteristics-
by-race-and-ethnicity-2014.pdf (accessed Jan 25, 2017).

5 James C, Salganicoff A, Ranji U, Goodwin A, Duckett P. Putting
men’s health care disparities on the map: examining racial and
ethnic disparities at the state level. Menlo Park, CA: Kaiser Family
Foundation, 2012. https://kaiserfamilyfoundation.files.wordpress.
com/2013/01/8344.pdf (accessed Feb 20, 2017).

6 Hammonds EM, Herzig RM. The nature of difference: sciences of
race in the United States from Jefferson to genomics.
Cambridge, MA: MIT Press, 2008.

7 Krieger N, Bassett M. The health of black folk: disease, class, and
ideology in science. Mon Rev 1986; 38: 74–85.

8 Omi M, Winant H. Racial formation in the United States, 3rd edn.
New York, NY: Routledge/Taylor & Francis Group, 2015.

9 Krieger N. Discrimination and health inequities. Int J Health Serv
2014; 44: 643–710.

10 Reskin B. The race discrimination system. Annu Rev Sociol
2012; 38: 17–35.

11 US Equal Employment Opportunity Commission. Facts about race/
color discrimination. January, 1999. https://www.eeoc.gov/facts/fs-
race.pdf (accessed Jan 25, 2017).

12 Gee GC, Ford, CL. Structural racism and health inequities.
Du Bois Rev 2011; 8: 115–32.

13 Hardeman RR, Medina EM, Kozhimannil KB. Structural racism
and supporting black lives — the role of health professionals.
N Engl J Med 2016; 375: 2113–15.

14 Marmot M, Friel S, Bell R, Houweling TAJ, Taylor S, on behalf of
the Commission on Social Determinants of Health. Closing the gap
in a generation: health equity through action on the social
determinants of health. Lancet 2008; 372: 1661–69.

15 Berger M, Sarnyai Z. “More than skin deep’’: stress neurobiology
and mental health consequences of racial discrimination.
Stress 2014; 18: 1–10.

16 Paradies Y, Ben J, Denson N, et al. Racism as a determinant of health:
a systematic review and meta-analysis. PLoS One 2015; 10: e0138511.

17 Paradies Y, Truong M, Priest N. A systematic review of the extent
and measurement of healthcare provider racism. J Gen Intern Med
2014; 29: 364–87.

18 Williams DR, Mohammed SA. Racism and health I: pathways and
scientific evidence. Am Behav Sci 2013; 57: 1152–73.

19 Frederickson GM. Racism: a short history. Princeton, NJ: Princeton
University Press, 2003.

20 Zinn H. A people’s history of the United States: 1492–present.
New York, NY: Harper Perennial, 2015.

For more on Healthy People
2020 see https://www.
healthypeople.gov/2020/About-
Healthy-People

Series

1462 www.thelancet.com Vol 389 April 8, 2017

21 Dunbar-Ortiz R. An indigenous peoples’ history of the
United States. Boston, MA: Beacon Press, 2015.

22 Loveman M. Making “Race” and nation in the United States,
South Africa, and Brazil: taking “making” seriously. Theory Soc
1999; 28: 903–27.

23 Higginbotham AL Jr. Shades of freedom: racial politics and
presumptions of the American legal process, vol 2. New York, NY:
Oxford University Press, 1996.

24 Waldstreicher D. Slavery’s constitution: from revolution to
ratification. New York, NY: Hill & Wang, 2010.

25 Anderson W. Racial conceptions in the global south. Isis 2014;
105: 782–92.

26 Foner E. A short history of Reconstruction, 1863–1877.
New York, NY: Harper & Row, 1990.

27 Du Bois WEB. Black folk then and now: an essay in the history and
sociology of the Negro race. New York, NY: Henry Holt & Co, 1939.

28 Alexander M. The new Jim Crow: mass incarceration in the age of
colorblindness. New York, NY: The New Press, 2010.

29 Massey DS, Denton NA. The dimensions of residential segregation.
Soc Forces 1988; 67: 281–315.

30 Hinton EK. From the war on poverty to the war on crime:
the making of mass incarceration in America. Cambridge,
MA: Harvard University Press, 2016.

31 Carson EA. Prisoners in 2014. NCJ 248955. Sept 17, 2015.
Washington, DC: Bureau of Justice Statistics, Office of Justice
Programs, Department of Justice, 2015. https://www.bjs.gov/index.
cfm?ty=pbdetail&iid=5387 (accessed Jan 19, 2017).

32 Acevedo-Garcia D, Lochner KA, Osypuk TL, Subramanian SV.
Future directions in residential segregation and health research:
a multilevel approach. Am J Public Health 2003; 93: 215–21.

33 Pager D, Shepherd H. The sociology of discrimination: racial
discrimination in employment, housing, credit, and consumer
markets. Annu Rev Sociol 2008; 34: 181–209.

34 Williams DR, Collins C. Racial residential segregation:
a fundamental cause of racial disparities in health.
Public Health Rep 2001; 116: 404.

35 Western B, Pettit B. Black-white wage inequality, employment rates,
and incarceration. Am J Sociol 2005; 111: 553–78.

36 Bravo MA, Anthopolos R, Bell ML, Miranda ML. Racial isolation
and exposure to airborne particulate matter and ozone in
understudied US populations: environmental justice applications of
downscaled numerical model output. Environ Int 2016;
92–93: 247–55.

37 Mohai P, Pellow D, Roberts JT. Environmental justice.
Annu Rev Env Resour 2009; 34: 405–30.

38 Siqueira CE, Gaydos M, Monforton C, et al. Effects of social,
economic, and labor policies on occupational health disparities.
Am J Ind Med 2014; 57: 557–72.

39 Grier SA, Kumanyika S. Targeted marketing and public health.
Annu Rev Public Health 2010; 31: 349–69.

40 Institute of Medicine. Unequal treatment: confronting racial and
ethnic disparities in health care. Washington, DC: The National
Academies Press, 2003.

41 Shavers VL, Fagan P, Jones D, et al. The state of research on racial/
ethnic discrimination in the receipt of health care.
Am J Public Health 2012; 102: 953–66.

42 Flores G. Racial and ethnic disparities in the health and health care
of children. Pediatrics 2010; 125: e979–1020.

43 Blair IV, Havranek EP, Price DW, et al. Assessment of biases against
Latinos and African Americans among primary care providers and
community members. Am J Public Health 2013; 103: 92–98.

44 Puumala SE, Burgess KM, Kharbanda AB, et al. The role of bias by
emergency department providers in care for American Indian
children. Med Care 2016; 54: 562–69.

45 White K, Haas JS, Williams DR. Elucidating the role of place in
health care disparities: the example of racial/ethnic residential
segregation. Health Serv Res 2012; 47: 1278–99.

46 Browne AJ, Smye VL, Varcoe C. The relevance of postcolonial
theoretical perspectives to research in Aboriginal health.
Can J Nurs Res 2005; 37: 16–37.

47 Fullilove MT. Root shock: how tearing up city neighborhoods hurts
America, and what we can do about it. New York, NY: New Village
Press, 2016.

48 Chaney C, Robertson RV. Racism and police brutality in America.
J Afr Am Stud 2013; 17: 480–505.

49 Purtle J. Felon disenfranchisement in the United States:
a health equity perspective. Am J Public Health 2013; 103: 632–37.

50 Blessett B. Disenfranchisement: historical underpinnings and
contemporary manifestations. Public Administration Quarterly 2015;
39: 3.

51 Slopen N, Lewis TT, Williams DR. Discrimination and sleep:
a systematic review. Sleep Med 2016; 18: 88–95.

52 Carlisle SK. Perceived discrimination and chronic health in adults
from nine ethnic subgroups in the USA. Ethn Health 2015; 20: 309–26.

53 Almond D, Chay KY, Greenstone M. Civil rights, the war on poverty,
and black-white convergence in infant mortality in the rural South and
Mississippi. Dec 31, 2006. Cambridge, MA: Massachusetts Institute of
Technology, 2006. https://dspace.mit.edu/handle/1721.1/63330
(accessed Jan 25, 2017).

54 Krieger N, Chen JT, Coull B, Waterman PD, Beckfield J.
The unique impact of abolition of Jim Crow laws on reducing
inequities in infant death rates and implications for choice of
comparison groups in analyzing societal determinants of health.
Am J Public Health 2013; 103: 2234–44.

55 Krieger N, Chen JT, Coull BA, Beckfield J, Kiang MV,
Waterman PD. Jim Crow and premature mortality among the US
black and white population, 1960–2009: an age–period–cohort
analysis. Epidemiology 2014; 25: 494–504.

56 Chay KY, Greenstone M. The convergence in black-white infant
mortality rates during the 1960’s. Am Econ Rev 2000; 90: 326–32.

57 Lukachko A, Hatzenbuehler ML, Keyes KM. Structural racism and
myocardial infarction in the United States. Soc Sci Med 2014;
103: 42–50.

58 Wallace ME, Mendola P, Liu D, Grantz KL. Joint effects of structural
racism and income inequality on small-for-gestational-age birth.
Am J Public Health 2015; 105: 1681–88.

59 Wildeman C, Wang EA. Mass incarceration, public health, and
widening inequality in the USA. Lancet 2017; 389: 1464–74.

60 Massey DS, Rothwell J, Domina T. The changing bases of segregation
in the United States. Ann Am Acad Polit Soc Sci 2009; 626: 74–90.

61 Logan JR, Stults BJ. The persistence of segregation in the
metropolis: new findings from the 2010 census. March 24, 2011.
http://www.s4.brown.edu/us2010/Data/Report/report2.pdf
(accessed Jan 19, 2017).

62 White K, Borrell LN. Racial/ethnic residential segregation: framing
the context of health risk and health disparities. Health Place 2011;
17: 438–48.

63 Kramer MR, Hogue CR. Is segregation bad for your health?
Epidemiol Rev 2009; 31: 178–94.

64 Kershaw KN, Roux AVD, Burgard SA, Lisabeth LD, Mujahid MS,
Schulz AJ. Metropolitan-level racial residential segregation and
black-white disparities in hypertension. Am J Epidemiol 2011;
174: 537–45.

65 Landrine H, Corral I. Separate and unequal: residential segregation
and black health disparities. Ethn Dis 2009; 19: 179–84.

66 Collins CA, Williams DR. Segregation and mortality: the deadly
effects of racism? Sociol Forum 1999; 14: 495–523.

67 Krivo LJ, Byron RA, Calder CA, et al. Patterns of local segregation: do
they matter for neighborhood crime? Soc Sci Res 2015; 54: 303–18.

68 Mendez DD, Hogan VK, Culhane JF. Institutional racism,
neighborhood factors, stress, and preterm birth. Ethn Health 2014;
19: 479–99.

69 Massey DS. The prodigal paradigm returns: ecology comes back to
sociology. In: Booth A, Crouter AC, eds. Does it take a village?
Community effects on children, adolescents, and families.
Mahwah, NJ: Lawrence Erlbaum Associates, 2001: 41–48.

70 Krieger N, Waterman PD, Spasojevic J, Li W, Maduro G,
Van Wye G. Public health monitoring of privilege and deprivation
with the Index of Concentration at the Extremes (ICE).
Am J Public Health 2016; 106: 256–63.

71 Burgason KA, Thomas SA, Berthelot ER. Nature of violence:
a multilevel analysis of gun use and victim injury in violent
interpersonal encounters. J Interpers Violence 2014; 29: 371–93.

72 Ulmer JT, Harris CT, Steffensmeier D. Racial and ethnic disparities
in structural disadvantage and crime: white, black, and hispanic
comparisons. Soc Sci Q 2012; 93: 799–819.

Series

www.thelancet.com Vol 389 April 8, 2017 1463

73 Lichtenstein A. Twice the work of free labor: the political economy
of convict labor in the New South. New York, NY: Verso, 1996.

74 Purpose Built Communities. Purpose Built Communities at the
Chautauqua Institution. Sept 28, 2015. http://
purposebuiltcommunities.org/news-press/purpose-built-
communities-at-the-chautauqua-institution (accessed Jan 25, 2017).

75 Gaumer E, Jacobowitz A, Brooks-Gunn J. The impact of affordable
housing on the well-being of low-income households. Public Policy
Analysis and Management Conference; Miami, FL; Nov 14, 2015.
https://appam.confex.com/appam/2015/webprogram/Paper14806.
html (accessed Jan 25, 2017).

76 Chetty R, Hendren N, Katz LF. The effects of exposure to better
neighborhoods on children: new evidence from the Moving to
Opportunity experiment. Am Econ Rev 2016; 106: 855–902.

77 Ludwig J, Liebman JB, Kling JR, et al. What can we learn about
neighborhood effects from the Moving to Opportunity experiment.
Am J Sociol 2008; 114: 144–88.

78 Abello OP. 5 cities get support for dismantling systemic racism.
Next City (Philadelphia, PA), May 31, 2016. https://nextcity.org/
daily/entry/living-cities-racial-inequity (accessed Jan 25, 2017).

79 Netherland J, Hansen H. White opioids: pharmaceutical race and
the war on drugs that wasn’t. BioSocieties 2016; published online
Jan 11. DOI:10.1057/biosoc.2015.46.

80 Legislative Analyst’s Office. The 2015–16 budget: implementation of
Proposition 47. Sacramento, CA: Legislative Analyst’s Office, 2015.
http://www.lao.ca.gov/reports/2015/budget/prop47/
implementation-prop47-021715.pdf (accessed Jan 25, 2017).

81 Stanford Justice Advocacy Project. Proposition 47 progress report:
year one implementation. Stanford, CA: Stanford Law School, 2015.
https://www-cdn.law.stanford.edu/wp-content/uploads/2015/10/
Prop-47-report.pdf (accessed Jan 25, 2017).

82 Hart A. Anti-racism in public health education: a student-driven
model for changes in a Master’s of Public Health Program.
American Public Health Association Annual Meeting; Chicago, IL;
Nov 3, 2015. https://apha.confex.com/apha/143am/webprogram/
Paper323616.html (accessed Jan 25, 2017).

83 Metzl J, Roberts D. Structural competency meets structural racism:
race, politics, and the structure of medical knowledge.
Virtual Mentor 2014; 16: 674.

84 Anderson W. Teaching race at medical school: social scientists on
the margin. Soc Stud Sci 2008; 38: 785–800.

85 Morning AJ. The nature of race: how scientists think and teach
about human difference. Berkeley, CA: University of California
Press, 2011.

86 Bassett MT. # BlackLivesMatter—a challenge to the medical and
public health communities. N Engl J Med 2015; 372: 1085–87.

87 Sturm S. The architecture of inclusion: advancing workplace equity
in higher education. Harv J L & Gender 2006; 29: 247–334.

88 Cross TL, Bazron BJ, Dennis KW, Isaacs MR. Towards a culturally
competent system of care: a monograph on effective services for
minority children who are severely emotionally disturbed.
Washington, DC: CASSP Technical Assistance Center, Georgetown
University Child Development Center, 1989.

89 Tervalon M, Murray-Garcia J. Cultural humility versus cultural
competence: a critical distinction in defining physician training
outcomes in multicultural education. J Health Care Poor Underserved
1998; 9: 117–25.

90 Darroch F, Giles A, Sanderson P, et al. The United States does
CAIR about cultural safety examining cultural safety within
Indigenous health contexts in Canada and the United States.
J Transcult Nurs 2016; published online Feb 25.
DOI:10.1177/1043659616634170.

91 Papps E, Ramsden I. Cultural safety in nursing: the New Zealand
experience. Int J Qual Health C 1996; 8: 491–97.

92 National Association of County and City Health Officials.
Expanding the boundaries: health equity and public health practice.
Washington, DC: National Association of County and City Health
Officials, 2014. http://www.dialogue4health.org/uploads/resources/
Expanding_the_Boundaries_Final_508_091814.pdf (accessed
Jan 25, 2017).

93 Boston Public Health Commission. The Racial Justice and Health
Equity Initiative: 2015 overview. http://www.bphc.org/whatwedo/
health-equity-social-justice/racial-justice-health-equity-initiative/
Documents/RJHEI%202015%20Overview%20FINAL.pdf (accessed
Jan 25, 2017).

94 Douglass F. Two Speeches, by Frederick Douglass: one on
West India emancipation, delivered at Canandaigua, Aug. 4th, and
the other on the Dred Scott Decision, delivered in New York, on the
occasion of the anniversary of the American Abolition Society,
May, 1857. Rochester, NY: C. P. Dewey, 1857.

  • Structural racism and health inequities in the USA: evidence and interventions
    • Introduction
    • Structural racism: a brief introduction
      • Racial ideology and the categorisation of racialised social groups
      • The continuing role of ostensibly colour-blind laws and policies
      • Structural racism in the private sector
    • Health consequences of structural racism: evidence and evidence gaps
      • Residential segregation
      • Discriminatory incarceration
      • Health-care quality and access
      • Addressing structural racism to advance health equity
      • Place-based, multisector, equity-oriented initiatives
      • Advocating for policy reform
      • Training the next generation of health professionals
    • Conclusion
    • Acknowledgments
    • References

EDITORIAL Open Access

Looking forward to the next 15 years:
innovation and new pathways for research
in health equity
Ana Lorena Ruano1,2* , Efrat Shadmi3, John Furler4, Krishna Rao5, Miguel San Sebastián6, Manuela Villar Uribe5

and Leiyu Shi7

Abstract

Since our launch in 2002, the International Journal for Equity in Health (IJEqH) has furthered our collective
understanding of equity in health and health services by providing a platform on which academics and
practitioners can share their work. Today, we celebrate our fifteenth anniversary with an article collection
that presents a call for new and novel research in equity in health and we invite our authors to use new
approaches and methods, and to focus on emerging areas of research related to health equity in order to
set the stage for the next fifteen years of health equity research.
Our anniversary issue provides a platform for expanding the conceptualization, diversity of populations and
study designs, and for increasing the use of novel methodologies in the field. The IJEqH has helped to
support the wider group of researchers, policymakers and practitioners with a commitment to social justice
and equity but there is still more to do. With the help of the highly committed editorial team and editorial
board, the innovative work of researchers worldwide, and the countless of hours dedicated by hundreds of
reviewers, we are confident in the IJEqH’s ability to continue supporting the dissemination of health equity
research for years to come.

Keywords: Equity, Indigenous peoples, Right to health, Refugees, Non-Nationals, SDGs, Qualitative methods,
Social justice, Cooperation

Since our launch in 2002, the International Journal for
Equity in Health (IJEqH) has furthered our collective un-
derstanding of equity in health and health services by pro-
viding a platform on which academics and practitioners
can share their work. Our mandate continues to be the
publication of political, policy-related, economic, social
and health services research that focuses on the systematic
differences in the distribution of one or more aspects of
health in population groups defined demographically, geo-
graphically or socially. Our commitment to giving voice to
authors from high-, middle- and low-income countries re-
mains just as strong today as when Prof Barbara Starfield
founded the journal fifteen years ago.

Understanding and acting on global and local inequity
remains one of the greatest issues of our times, and our
efforts to collect and publish research on particular
topics of concern are reflected in our three long-running
thematic series. The first of these, which focuses on mul-
timorbidity (MM), studies the co-occurrence of health
conditions in individuals and populations who experi-
ence high burdens of disease and who already find them-
selves in disadvantaged circumstances. The series is
edited by Prof Efrat Shadmi, our co-Editor-in-Chief [1],
and stems from Prof Starfield’s work showing that ‘only
a person-focused (rather than a disease-focused) view of
morbidity, in which multiple illnesses interact in myriad
ways, can accurately depict the much greater impact of
illness among socially disadvantaged people and the
nature of the interventions that are required to ad-
equately manage the increased vulnerability to and inter-
actions among diseases’ [2]. The series contributes to

* Correspondence: [email protected]
1Center for the Study of Equity and Governance in Health Systems, CEGSS,
Guatemala City, Guatemala
2Center for International Health, University of Bergen, Bergen, Norway
Full list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Ruano et al. International Journal for Equity in Health (2017) 16:35
DOI 10.1186/s12939-017-0531-0

the growing body of knowledge on MM and equity and
presents an international perspective on the prevalence
of MM in diverse population groups [3], its determi-
nants [4] and outcomes [5].
Our second thematic series deals with the unprece-

dented economic crisis that has affected Europe, mainly
in the southern countries, and is edited by our Associate
Editor, Prof Miguel San Sebastián, and Guest Editor Dr
Antonio Escolar-Pujolar. While certain macroeconomic
recovery has been observed in the last couple of years,
the economic situation remains unstable. This is particu-
larly true for the Mediterranean countries, which still
face numerous social and economic challenges [6]. This
series aims to raise awareness of the impacts of the eco-
nomic crisis on population health and, especially, on
health inequalities. The articles within the series show-
case the diversity and complexity of the impact of the
crisis on health [7], but further studies are needed to
better understand the link between health and the pol-
icies emanating from a dominant socioeconomic model
that continuously generates socioeconomic inequalities
that affect the most vulnerable groups in society [8].
Finally, our thematic series on interventions in primary

health care that improve equity of outcomes in health
showcases the potential contributions of continuous,
comprehensive, and coordinated primary health care.
Prof Starfield’s seminal work demonstrated that effective
primary care is associated with improved access to
health care services, reduced hospitalizations, cost-
effectiveness and enhanced equity [9–12]. In recent
years, many countries, organizations, communities, and
world agencies have implemented primary care reforms
and interventions to improve healthcare access, quality,
health outcomes and equity in health. This series aims
to create a space where we can document and analyze
primary health care reforms and interventions that com-
munities, organizations, countries and world agencies
have undertaken [13] and is edited by our Associate
Editor, Dr John Furler; our Managing Editor, Dr Ana
Lorena Ruano; and Prof Leiyu Shi, co-Editor-in-Chief.
Today we celebrate our fifteenth anniversary with an

article collection that presents a call for new and novel
research in equity in health. The papers presented here
invite our authors to use new approaches and methods,
and to focus on emerging areas of research related to
health equity. This is the case of the article by Fridman
and Gostin [14], who use the Framework Convention on
Global Health to call for a profound questioning of
health policies at the global, national and local levels and
who ask for research to guide actions geared towards
truly reducing inequity in health through activism that
can help reshape power dynamics. This is in line with
Rasanathan and Diaz’s commentary [15], which focuses
on Sustainable Development Goals (SDGs) and the need

for strengthened development, testing, and implementa-
tion of possible solutions for improving equity levels
around the globe. Brolan et al. [16] use a rights-based
approach and state that the SDGs can be truly trans-
formative if they are made operational in all countries
and if they include nationals and non-nationals alike.
This is especially urgent given the conflicts in the
Middle East and Africa, as well as the escalating violence
in Central America, which has forced hundreds of thou-
sands of adults and unaccompanied minors to flee in
perilous conditions and with little immediate hope of
improving their lives and those of their loved ones.
Using systems thinking and engaging in complexity

studies, Hernánez et al. [17] call for a novel approach to
improving the health status of indigenous peoples, who
remain among the most excluded and marginalized
population groups all over the world [18]. The authors
invite us to move from reductionist approaches that
frame indigenous health as a set of poor health indica-
tors and instead focus on holistic, integrated approaches
that address the root causes of inequity both inside and
outside the health sector. This approach raises the need
to expand our methodological and theoretical toolbox,
particularly when it comes to social sciences and qualita-
tive approaches. Hannefeld et al., on behalf of the
SHAPES thematic working group from Health Systems
Global, also focus on the role of different methodologies
and approaches to enriching our understanding of in-
equity [19]. Without a network of researchers of diverse
backgrounds and methodological interests, our compre-
hension of issues that are central to improving the qual-
ity of health services and the tools we have at our
disposal to strengthen health systems would be greatly
diminished. Because of this, our anniversary issue en-
shrines our commitment to publish high-quality qualita-
tive and social science research.
Social justice is at the heart of improving equity in

health, and Devia et al. [20] examine the role of
community-based participatory research in addressing
the social determinants of health through working with
the underlying causes of the inequitable distribution of
resources and power structures. Flores et al. [21] show
the urgent need for health professionals to be better
equipped to address the cultural diversity of the popula-
tions seeking care in many countries today. Providing
adequate support and training to new physicians and
other health cadres can not only help ensure academic
and professional success but also improve the quality
and equity of care provided.
The last paper in our anniversary issue focuses on the

role of competition and collaboration in achieving
greater levels of health equity. Chang and Fraser [22]
argue for an agency-based approach, and warn that a
zero-sum mentality with respect to competition leads to

Ruano et al. International Journal for Equity in Health (2017) 16:35 Page 2 of 3

ethically questionable and less effective cooperation.
Market-driven approaches to healthcare contribute
greatly to the growing inequities around the world.
Setting the stage for the next fifteen years of health

equity research, our anniversary issue provides a plat-
form for expanding the conceptualization, diversity of
populations and study designs, and for increasing the
use of novel methodologies in the field. Recent global
changes that instigate major demographic [23] and eco-
nomic transformations [7] call for expanding the role of
research in illuminating injustice, developing and testing
interventions to expand the evidence base on how to
eliminate inequity in health and health care, and guiding
policy that will lead to better outcomes for diverse popu-
lation groups worldwide. Fifteen years on, much has
been achieved. The journal has helped support the wider
group of researchers, policymakers and practitioners
with a commitment to social justice and equity. There is
still more to do. With the help of the highly committed
editorial team and editorial board, the innovative work
of researchers worldwide, and the countless of hours
dedicated by hundreds of reviewers, we are confident in
the IJEqH’s ability to continue supporting the dissemin-
ation of health equity research for years to come.

Acknowledgements
Not applicable.

Funding
Not applicable.

Availability of data and materials
Not applicable.

Authors’ contributions
ALR drafted the first version of this manuscript. ALR, ES, JF, KR, MSS, MVU
and LS all contributed equally. All authors approve this paper.

Competing interests
The authors declare they have no competing interests.

Consent for publication
All authors consent to publication of this paper.

Ethics approval and consent to participate
Not applicable.

Author details
1Center for the Study of Equity and Governance in Health Systems, CEGSS,
Guatemala City, Guatemala. 2Center for International Health, University of
Bergen, Bergen, Norway. 3The Cheryl Spencer Department of Nursing,
Faculty of Socail Welfare and Health Sciences, the University of Haifa, Haifa,
Israel. 4Department of General Practice, Faculty of Medicine, Dentistry and
Health Sciences University of Melbourne, Melbourne, Australia. 5Johns
Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA.
6Department of Public Health and Clinical Medicine, Unit of Epidemiology
and Global Health Umeå University, SE-901 87 Umeå, Sweden. 7Johns
Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA.

Received: 7 February 2017 Accepted: 13 February 2017

References
1. Shadmi E. Multimorbidity and equity in health. Int J Equity Health. 2013;

12(1):59.
2. Starfield B. The hidden inequity in health care. Int J Equity Health. 2011;

10(1):15.
3. Boutayeb A, Boutayeb S, Boutayeb W. Multi-morbidity of non

communicable diseases and equity in WHO Eastern Mediterranean
countries. Int J Equity Health. 2013;12(1):60.

4. Reis-Santos B, Gomes T, Macedo LR, Horta BL, Riley LW, Maciel EL.
Prevalence and patterns of multimorbidity among tuberculosis patients in
Brazil: a cross-sectional study. Int J Equity Health. 2013;12(1):61.

5. Alaba O, Chola L. The social determinants of multimorbidity in South Africa.
Int J Equity Health. 2013;12(1):63.

6. Escolar-Pujolar A, Bacigalupe A, San SM. European economic crisis and
health inequities: research challenges in an uncertain scenario. Int J Equity
Health. 2014;13(1):59.

7. Escolar Pujolar A, Bacigalupe A, San SM. Looking beyond the veil of the
European crisis – the need to uncover the structural causes of health
inequalities. Int J Equity Health. 2016;15(1):39.

8. Bacigalupe A, Escolar-Pujolar A. The impact of economic crises on social
inequalities in health: what do we know so far? Int J Equity Health. 2014;
13(1):52.

9. Macinko JA, Starfield B. Annotated bibliography on equity in health,
1980–2001. Int J Equity Health. 2002;1(1):1.

10. Macinko J, Starfield B, Shi L. The Contribution of Primary Care Systems to
Health Outcomes within Organization for Economic Cooperation and
Development (OECD) Countries, 1970–1998. Health Serv Res. 2003;38(3):
831–65.

11. Starfield B. Is primary care essential? The Lancet. 344(8930):1129–33
12. Starfield B. Primary care: balancing health needs, services, and technology.

USA: Oxford University Press; 1998.
13. Ruano AL, Furler J, Shi L. Interventions in Primary Care and their

contributions to improving equity in health. Int J Equity Health. 2015;14(1):
153.

14. Friedman E, Gostin LO. From local adaptation to activism and global
solidarity: framing research and innovation agenda towards true health
equity. Int J Equity Health. 2017;16(18)

15. Rasanathan K, Diaz T. Research on health equity in the SDG era: the urgent
need for greater focus on implementation. Int J Equity Health. 2016;15(202)

16. Brolan C, Forman L, Dagron S, Hammonds R, Waris A, Latif L, et al. The right
to health of non-nationals and displaced persons in the sustainable
development goals era: challenges for equity in universal health care. Int J
Equity Health. 2017;16(14)

17. Hernández A, Ruano AL, Marchal B, San Sebastian M, Flores W. Engaging
with complexity to improve the health of indigenous people: a call for the
use of systems thinking to tackle health inequity. International Journal for
Equity in Health. 2017;16(XX).

18. Cerón A, Ruano AL, Sánchez S, Chew AS, Díaz D, Hernández A, et al. Abuse
and discrimination towards indigenous people in public health care
facilities: experiences from rural Guatemala. Int J Equity Health. 2016;15(1):1.

19. Daniels K, Loewenson R, George A, Howard N, Koleva G, Lewin S, et al. Fair
publication of qualitative research in health systems: a call by health policy
and systems researchers. Int J Equity Health. 2016;15(1):98.

20. Devia C, Baker E, Shanchez-Youngman S, Barnidge E, Golub M, Motton F, et
al. Advancing system and policy changes for social and racial justice:
comparing a rural and urgan community-based participatory reserach
partnership in the U.S. Int J Equity Health. 2017;16(17)

21. Flores G, Mendoza FS, Fuentes-Afflick E, Mendoza JA, Pachter L, Espinoza J,
et al. Hot topics, urgent priorities, and ensuring success for racial/ethnic
minority young investigators in academic pediatrics. Int J Equity Health.
2016;15(1):201.

22. Chang W, Fraser J. Cooperate! A paradigm shift for heatlh equity. Int J
Equity Health. 2017;16(12)

23. Kousoulis AA, Ioakeim-Ioannidou M, Economopoulos KP. Access to
health for refugees in Greece: lessons in inequalities. Int J Equity
Health. 2016;15(1):122.

Ruano et al. International Journal for Equity in Health (2017) 16:35 Page 3 of 3

Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

  • c.ICABMDD_EQHL_20170101_1C6571588948A96B5FB1_25478.pdf
    • Abstract
    • Acknowledgements
    • Funding
    • Availability of data and materials
    • Authors’ contributions
    • Competing interests
    • Consent for publication
    • Ethics approval and consent to participate
    • Author details
    • References

CDC Health Disparities and Inequalities Report —
United States, 2013

Supplement / Vol. 62 / No. 3 November 22, 2013

U.S. Department of Health and Human Services
Centers for Disease Control and Prevention

Morbidity and Mortality Weekly Report

Supplement

The MMWR series of publications is published by the Center for Surveillance, Epidemiology, and Laboratory Services, Centers for Disease Control and Prevention (CDC),
U.S. Department of Health and Human Services, Atlanta, GA 30333.
Suggested Citation: Centers for Disease Control and Prevention. [Title]. MMWR 2013;62(Suppl 3):[inclusive page numbers].

Centers for Disease Control and Prevention
Thomas R. Frieden, MD, MPH, Director

Harold W. Jaffe, MD, MA, Associate Director for Science
Joanne Cono, MD, ScM, Acting Director, Office of Science Quality

Chesley L. Richards, MD, MPH, Deputy Director for Public Health Scientific Services

MMWR Editorial and Production Staff
Ronald L. Moolenaar, MD, MPH, Editor, MMWR Series

Christine G. Casey, MD, Deputy Editor, MMWR Series
Teresa F. Rutledge, Managing Editor, MMWR Series

David C. Johnson, Lead Technical Writer-Editor
Jeffrey D. Sokolow, MA, Catherine B. Lansdowne, MS,

Denise Williams, MBA, Project Editors
Martha F. Boyd, Lead Visual Information Specialist

Maureen A. Leahy, Julia C. Martinroe,
Stephen R. Spriggs, Terraye M. Starr

Visual Information Specialists
Quang M. Doan, MBA, Phyllis H. King

Information Technology Specialists

MMWR Editorial Board
William L. Roper, MD, MPH, Chapel Hill, NC, Chairman

Matthew L. Boulton, MD, MPH, Ann Arbor, MI
Virginia A. Caine, MD, Indianapolis, IN
Barbara A. Ellis, PhD, MS, Atlanta, GA

Jonathan E. Fielding, MD, MPH, MBA, Los Angeles, CA
David W. Fleming, MD, Seattle, WA

William E. Halperin, MD, DrPH, MPH, Newark, NJ
King K. Holmes, MD, PhD, Seattle, WA

Timothy F. Jones, MD, Nashville, TN
Rima F. Khabbaz, MD, Atlanta, GA
Dennis G. Maki, MD, Madison, WI

Patricia Quinlisk, MD, MPH, Des Moines, IA
Patrick L. Remington, MD, MPH, Madison, WI

William Schaffner, MD, Nashville, TN

Asthma Attacks Among Persons with Current Asthma —
United States, 2001–2010 ……………………………………………………………….. 93

Diabetes — United States, 2006 and 2010 …………………………………….. 99
Health-Related Quality of Life — United States, 2006 and 2010 …. 105
HIV Infection — United States, 2008 and 2010 ……………………………. 112
Obesity — United States, 1999–2010 ……………………………………………. 120
Periodontitis Among Adults Aged ≥30 Years —

United States, 2009–2010 ……………………………………………………………… 129
Preterm Births — United States, 2006 and 2010 …………………………. 136
Potentially Preventable Hospitalizations — United States,

2001–2009 ………………………………………………………………………………………. 139
Prevalence of Hypertension and Controlled Hypertension —

United States, 2007–2010 ……………………………………………………………… 144
Tuberculosis — United States, 1993–2010 …………………………………… 149

Health Outcomes: Mortality ……………………………………………………..155
Coronary Heart Disease and Stroke Deaths —

United States, 2009 ………………………………………………………………………… 157
Drug-Induced Deaths — United States, 1999–2010 …………………… 161
Homicides — United States, 2007 and 2009 ……………………………….. 164
Infant Deaths — United States, 2005–2008 …………………………………. 171
Motor Vehicle–Related Deaths — United States, 2005 and 2009 .. 176
Suicides — United States, 2005–2009 …………………………………………… 179

Conclusion and Future Directions: CDC Health Disparities and
Inequalities Report — United States, 2013 ………………………………184

Foreword ……………………………………………………………………………………… 1

Introduction: CDC Health Disparities and Inequalities Report
— United States, 2013 …………………………………………………………………. 3

Social Determinants of Health …………………………………………………….. 7
Education and Income — United States, 2009 and 2011 ……………….. 9
Access to Healthier Food Retailers — United States, 2011 ………….. 20
Unemployment — United States, 2006 and 2010 ………………………… 27

Environmental Hazards ……………………………………………………………… 33
Nonfatal Work-Related Injuries and Illnesses —

United States, 2010 ………………………………………………………………………….. 35
Fatal Work-Related Injuries — United States, 2005–2009 …………… 41
Residential Proximity to Major Highways — United States, 2010 …. 46

Health-Care Access and Preventive Services …………………………….. 51
Colorectal Cancer Incidence and Screening —

United States, 2008 and 2010 ………………………………………………………… 53
Health Insurance Coverage — United States, 2008 and 2010 …….. 61
Seasonal Influenza Vaccination Coverage —

United States, 2009–10 and 2010–11 …………………………………………… 65

Behavioral Risk Factors ………………………………………………………………. 69
Pregnancy and Childbirth Among Females Aged 10–19 Years —

United States, 2007–2010 ……………………………………………………………….. 71
Binge Drinking — United States, 2011 …………………………………………… 77
Cigarette Smoking — United States, 2006-2008 and 2009-2010 …. 81

Health Outcomes: Morbidity ……………………………………………………… 85
Expected Years of Life Free of Chronic Condition–Induced

Activity Limitations — United States, 1999–2008 ………………………. 87

CONTENTS

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 1

CDC works 24 hours a day, seven days a week protecting
people in the United States from health threats in order to
save lives, promote health, and reduce costs. Achieving health
equity, eliminating health disparities, and improving health in
the United States are overarching goals to improve and protect
our nation’s health.

Over the past 50 years, the United States has made significant
progress toward these important goals. Life expectancy
increased from just under 70 years in 1960 to approximately
79 years in 2011 (1,2). People are living longer, healthier, and
more productive lives. However, this upward trend is neither
as rapid as it should be — we lag behind dozens of other
nations (3) – nor is it uniformly experienced by people in the
United States.

In fact, these two shortcomings of our health system
are distinct but related. Our overall health status does not
achieve our potential. An important part of this — even
though preventable illness, injury, disability, and death affect
all segments of society — is that life expectancy and other
key health outcomes vary greatly by race, sex, socioeconomic
status, and geographic location. In the United States, whites
have a longer healthy life expectancy than blacks, and women
live longer than men (4). There are also marked regional
differences, with much lower life expectancy among both white
and black Americans who live in the Southeast (4).

CDC Health Disparities and Inequalities Report — United
States, 2013 is the second agency report examining some of
the key factors that affect health and lead to health disparities
in the United States. Four findings bring home the enormous
personal tragedy of health disparities:
• Cardiovascular disease is the leading cause of death in the

United States. Non-Hispanic black adults are at least 50%
more likely to die of heart disease or stroke prematurely
(i.e., before age 75 years) than their non-Hispanic white
counterparts (5).

• The prevalence of adult diabetes is higher among
Hispanics, non-Hispanic blacks, and those of other or
mixed races than among Asians and non-Hispanic whites.
Prevalence is also higher among adults without college
degrees and those with lower household incomes (6).

• The infant mortality rate for non-Hispanic blacks is more
than double the rate for non-Hispanic whites. Rates also
vary geographically, with higher rates in the South and
Midwest than in other parts of the country (7).

Foreword
Thomas R. Frieden, MD, MPH

Director, CDC

• Men are far more likely to commit suicide than women,
regardless of age or race/ethnicity, with overall rates nearly
four times those of women. For both men and women,
suicide rates are highest among American Indians/Alaska
Natives and non-Hispanic whites (8).

CDC and its partners work to identify and address the
factors that lead to health disparities among racial, ethnic,
geographic, socioeconomic, and other groups so that barriers
to health equity can be removed. The first step in this process
is to shine a bright light on the problem to be solved. Providing
accurate, useful data on the leading causes of illness and death
in the United States and across the world is a foundation of
CDC’s mission and work.

In 1966, Martin Luther King said that “Of all the forms of
inequality, injustice in health care is the most shocking and
inhumane” (9). Nearly a half century after Reverend King made
this observation, we have made some but not nearly enough
progress in reducing the barriers to equitable health care and to
health equity. We should work with what he called “the fierce
urgency of now” to eliminate this form of inequality wherever
and whoever it affects.

As Secretary of Health and Human Services Kathleen G.
Sebelius has said, “Health equity benefits everyone” (10). Every
person who dies young, is avoidably disabled, or is unable to
function at their optimal level represents not only a personal
and family tragedy but also impoverishes our communities and
our country. We are all deprived of the creativity, contributions,
and participation that result from disparities in health status.

Eliminating the burden of racial and ethnic health disparities
is not easy, but it can be done. For example, 20 years ago the
Vaccines for Children (VFC) program was created to provide
vaccines at no cost to eligible children. It is now one of our
country’s most successful public health initiatives (11). By
removing cost barriers associated with vaccines, the VFC
program has protected millions of children from diseases
— both those who participated directly in the program and
others — and has helped to virtually eliminate disparities in
childhood vaccination rates. More recently, the Affordable
Care Act (12), with its provisions to require insurer coverage
of preventive services without cost to patients and to increase
health insurance access for millions of previously uninsured
Americans (13,14), provides a powerful opportunity to further
reduce health disparities.

Supplement

2 MMWR / November 22, 2013 / Vol. 62 / No. 3

Achieving health equity requires the hard work of many
people and organizations. CDC’s many partners can use
the information in this report to stimulate action to further
reduce health disparities. The future health of our nation will
be determined, to a large extent, by how effectively federal,
state, and local agencies and private organizations work with
communities to eliminate health disparities among populations
that continue to experience a disproportionate burden of
disease, disability, injury, and death.

References
1. Arias E. United States life tables, 2008. Natl Vital Stat Rep 2012;61(3).
2. Hoyert DL, Xu J. Deaths: preliminary data for 2011. Natl Vital Stat

Rep 2012;61(6).
3. World Health Organization. World health statistics 2013. Geneva,

Switzerland: World Health Organization; 2013.
4. CDC. State-specific healthy life expectancy at age 65 years—United

States, 2007–2009. MMWR 2013;62:561–6.
5. CDC. Coronary heart disease and stroke deaths—United States, 2009.

In: CDC health disparities and inequalities report—United States, 2013.
MMWR 2013;62(No. Suppl 3):155-8.

6. CDC. Diabetes—United States, 2006 and 2010. In: CDC health
disparities and inequalities report—United States, 2013. MMWR
2013;62(No. Suppl 3):97-102.

7. CDC. Infant deaths—United States, 2005-2008. In: CDC health
disparities and inequalities report—United States, 2013. MMWR
2013;62(No. Suppl 3):169-72.

8. CDC. Suicides—United States, 2005-2009. In: CDC health disparities
and inequalities report—United States, 2013. MMWR 2013;
62(No. Suppl 3):177-81.

9. Families USA. Martin Luther King Jr.: A civil rights icon’s thoughts on
health care. Available at http://www.standupforhealthcare.org/blog/
martin-luther-king-jr-a-civil-rights-icon-s-thoughts-on-health-care.

10. U.S. Department of Health and Human Services. HHS action plan to
reduce racial and ethnic health disparities. Washington, DC: US
Department of Health and Human Services; 2012. Available at http://
minorityhealth.hhs.gov/npa/files/Plans/HHS/HHS_Plan_complete.pdf.

11. CDC. Vaccines for children program (VFC). Atlanta, GA: US
Department of Health and Human Services, CDC; 2012. Available at
http://www.cdc.gov/vaccines/programs/vfc/index.html.

12. US Department of Health and Human Services. Read the Law: The
Affordable Care Act, Section by Section. Washington, DC: US
Department of Health and Human Services; 2012. Available at http://
www.hhs.gov/healthcare/rights/law/index.html.

13. Congressional Budget Office.   CBO’s February 2013 estimate of the
effects of the Affordable Care Act on health insurance coverage.
Washington, DC: Congressional Budget Office; 2013. http://cbo.gov/
sites/default/files/cbofiles/attachments/43900_ACAInsurance
CoverageEffects.pdf.

14. US Centers for Medicare and Medicaid Services. How does the health-
care law protect me? Baltimore, MD: US Centers for Medicare and
Medicaid Services; 2012. Available at https://www.healthcare.gov/how-
does-the-health-care-law-protect-me/#part8=undefined/part=1.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 3

Public Health Importance of
Health Disparities

The burden of illness, premature death, and disability
disproportionately affects certain populations. During the
past decade, documented disparities have persisted for
approximately 80% of the Healthy People 2010 objectives and
have increased for an additional 13% of the objectives (3). Data
from the REACH U.S. Risk Factor Survey of approximately
30 communities in the United States indicate that residents
in mostly minority communities continue to have lower
socioeconomic status, greater barriers to health-care access,
and greater risks for, and burden of, disease compared with
the general population living in the same county or state (4).
Both the 2012 National Healthcare Disparities Report (5) and
the 2012 National Healthcare Quality Report (6) found that
almost none of the disparities in access to care are improving.
In addition, quality of care varies not only across types of
care but also across parts of the country (5,6). Disparities in
health care access and quality can result in unnecessary direct
and indirect costs. According to a 2009 study by the Joint
Center for Political and Economic Studies, eliminating health
disparities for minorities would have reduced direct medical
care expenditures by $229.4 billion and reduced indirect costs
associated with illness and premature death by approximately
$1 trillion during 2003–2006 (7).

Introduction: CDC Health Disparities and Inequalities Report —
United States, 2013

Pamela A. Meyer, PhD1
Paula W. Yoon, ScD2

Rachel B. Kaufmann, PhD2
1Office for State, Tribal, Local and Territorial Support, CDC

2Center for Surveillance, Epidemiology, and Laboratory Services, CDC

Disparities in Health Outcomes and
Health Determinants

Health is influenced by many factors. Poor health status,
disease risk factors, and limited access to health care are often
interrelated and have been reported among persons with social,
economic, and environmental disadvantages. The conditions
and social context in which persons live can explain, in part,
why certain populations in the United States are healthier than
others and why some are not as healthy as they could be (1).
The World Health Organization (WHO) defines the social
determinants of health as the conditions in which persons
are born, grow, live, work, and age, including the health-care
system (2). According to WHO, “the social determinants
of health are mostly responsible for health inequities—the
unfair and avoidable differences in health status seen within
and between countries” (2). The social determinants of health
as well as race and ethnicity, sex, sexual orientation, age, and
disability all influence health. Identification and awareness of
the differences among populations regarding health outcomes
and health determinants are essential steps towards reducing
disparities in communities at greatest risk.

Disparities exist when differences in health outcomes or
health determinants are observed between populations. The
terms health disparities and health inequalities are often
used interchangeably. This supplement uses the terms health
disparities and inequalities to refer to gaps in health between
segments of the population.

Summary

This supplement is the second CDC Health Disparities and Inequalities Report (CHDIR). The 2011 CHDIR was the first CDC
report to assess disparities across a wide range of diseases, behavioral risk factors, environmental exposures, social determinants, and
health-care access (CDC. CDC Health Disparities and Inequalities Report—United States, 2011. MMWR 2011;60[Suppl;
January 14, 2011]). The 2013 CHDIR provides new data for 19 of the topics published in 2011 and 10 new topics. When data
were available and suitable analyses were possible for the topic area, disparities were examined for population characteristics that
included race and ethnicity, sex, sexual orientation, age, disability, socioeconomic status, and geographic location. The purpose of
this supplement is to raise awareness of differences among groups regarding selected health outcomes and health determinants and
to prompt actions to reduce disparities. The findings in this supplement can be used by practitioners in public health, academia
and clinical medicine; the media; the general public; policymakers; program managers; and researchers to address disparities and
help all persons in the United States live longer, healthier, and more productive lives.

Supplement

4 MMWR / November 22, 2013 / Vol. 62 / No. 3

About This Report
This supplement is the second in a series of reports that

address health disparities. The 2011 CHDIR was the first
CDC report to assess health disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access (8). The 2013
CHDIR includes more current data for 19 of the topics
published in 2011. Two 2011 topics, housing and air quality,
are not included in 2013 because there were no new data
to report. There are 10 new topics. The new topics include:
access to healthier food retailers, unemployment, nonfatal
work-related injuries and illnesses, fatal work-related injuries,
residential proximity to major highways, activity limitations
due to chronic diseases, asthma attacks, health-related quality
of life, periodontitis in adults, and tuberculosis. In the 2011
CHDIR, the prevalence of asthma (i.e., the percentage of
persons who have ever been diagnosed with asthma and
still have asthma) was reported, whereas in this report, the
characteristics of persons who experienced an asthma attack
during the preceding 12 months are discussed. Although
the focus of these reports is on the measurement of health
disparities, most also mentioned existing evidence-based
interventions or strategies.

Criteria for Topic Selection
Selection of new topics for this supplement was done in

consultation with CDC’s Associate Directors for Science. The
primary prerequisites for selection of topics were that data be of
high quality and appropriate for developing national estimates.
In addition, the topic had to meet one or more of the following
criteria: 1) leading cause of premature death, higher disease
burden, or lower life expectancy at birth for certain segments of
the U.S. population as defined by sex, race/ethnicity, income or
education, geography, sexual orientation, and disability status;
2) known determinant of health (e.g., social, demographic, and
environmental) where disparities have been identified; and 3)
health outcome for which effective and feasible interventions
exist where disparities have been identified.

Analysis
Most of the analyses in this supplement are descriptive and

did not control for potential confounders or adjust for age;
therefore, caution should be used in comparing these findings
to findings from studies with different analytical approaches.
When data were available and suitable for analysis, disparities
were examined for characteristics that included race and

ethnicity, sex, age, household income, educational attainment,
and geographic location. Other characteristics that were
analyzed included place of birth, language spoken at home,
disability status, and sexual orientation. Consistent definitions
were used as a guide to promote standardization of analyses
across the reports. However, readers should be attentive to the
definitions used in each report. There are some similarities and
some differences in definitions across reports because there are
multiple ways to categorize these variables. For certain variables,
the most appropriate categorization depends on the topic being
studied (e.g., age groups). For other variables, the Office of
Management and Budget (OMB) and the U.S. Department of
Health and Human Services have set rules that are to be used
in federal surveys (i.e., race, ethnicity). To the extent possible,
OMB standards were used in the analyses. However, some
data sources did not collect or report information with the
granularity recommended by OMB because the numbers of
some racial and ethnic groups were small and their estimates
would not be meaningful. Subject matter experts across CDC
participated in identifying appropriate definitions.

Analyses focused on the estimated prevalence of a risk factor or
health outcome or on the estimated rate of a health outcome in
the population. Also, in certain reports, change in the estimated
prevalence or rate over time in recent years was calculated.
Analytic methods used in the reports varied; therefore, it is
important to read the methods description for each report. Most
authors calculated absolute or relative difference in prevalence
or rate, or both, between segments of the U.S. population. The
absolute difference is the arithmetic difference between two
groups. For example, if the prevalence of a certain condition is
1% among women and 5% among men, the absolute difference
is 4 percentage points. The relative difference is the absolute
difference divided by the value for the referent group; the result
is multiplied by 100% to create a percentage. In the above
example, the relative difference for men compared with women
is 400% ([4%/1%]*100%). In other words, men have an excess
prevalence that is four times the prevalence of what occurs among
women. This example illustrates that the relative difference can
be far larger than the absolute difference, especially when the
overall prevalence of the condition is low. Conversely, the relative
difference can be smaller when the overall prevalence is high.
For example, if the prevalence is 91% among women and 95%
among men, the absolute difference is still 4 percentage points
but the relative difference is only 4%. To gain a more complete
understanding of the population’s health status and the impact
of disparities, it is instructive to look at both measures.

In most analyses, the statistical significance of observed
differences was assessed using formal significance testing with
alpha=0.05. If statistical testing was not done, differences were

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 5

assessed by calculating and comparing 95% confidence intervals
(CIs) around the estimated prevalence or rate. In this approach,
CIs were used as a measure of variability, and nonoverlapping
CIs were considered statistically different. While using CIs in
this way is a conservative evaluation of significance differences,
infrequently this approach might lead to a conclusion that
estimates are similar when the point estimates do differ. Because
of analytical constraints, neither statistical significance nor 95%
CIs were calculated for three reports (9–11).

Use of This Report
The findings and conclusions in this supplement are intended

for practitioners in public health, academia and clinical
medicine; the media; general public; policymakers; program
managers; and researchers to address disparities and help all
persons in the United States live longer, healthier, and more
productive lives. The information on disparities can be used to
help select interventions for specific subgroups or populations
and support community actions to address disparities.

References
1. U.S. Department of Health and Human Services. Healthy People 2020:

Disparities. Available at http://healthypeople.gov/2020/about/
DisparitiesAbout.aspx.

2. World Health Organization. Social determinants of health. Available at
http://www.who.int/social_determinants/sdh_definition/en/index.html.

3. US Department of Health and Human Services. Healthy people 2010
final review. Washington, DC.: U.S. Government Printing Office. 2011.

4. CDC. Surveillance of health status in minority communities—Racial and
Ethnic Approaches to Community Health Across the U.S. (REACH U.S.)
Risk Factor Survey, United States, 2009. MMWR 2011;60 (No. SS-6).

5. US Department of Health and Human Services. National healthcare
disparities report, 2012. AHRQ Publication No. 12-0006. March 2012,
Rockville, MD. Available at http://www.ahrq.gov/research/findings/
nhqrdr/nhdr11/key.html.

6. US Department of Health and Human Services. National healthcare
quality report, 2012. AHRQ Publication No. 13-0003. May 2013,
Rockville, MD. Available at http://www.ahrq.gov/research/findings/
nhqrdr/nhdr12/nhdr12_prov.pdf.

7. Joint Center for Political and Economic Studies. The economic burden
of health inequalities in the United States. 2009. Washington, DC.
Available at http://www.jointcenter.org/research/the-economic
-burden-of-health-inequalities-in-the-united-states.

8. CDC. CDC Health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

9. CDC. Residential proximity to major highways—United States, 2010.
In: CDC Health disparities and inequalities report—United States,
2013. MMWR 2013;62(No. Suppl 3).

10. CDC. Potentially preventable hospitalizations—United States, 2001–
2009. In: CDC Health disparities and inequalities report—United States,
2013. MMWR 2013;62(No. Suppl 3).

11. CDC. Tuberculosis—United States, 1993–2010. In: CDC Health
disparities and inequalities report—United States, 2013. MMWR 2013;
62(No. Suppl 3).

Supplement

6 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 7

Social Determinants of Health

Supplement

8 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 9

Education and Income — United States, 2009 and 2011
Gloria L. Beckles, MD1

Benedict I. Truman, MD2
1National Center for Chronic Disease Prevention and Health Promotion, CDC

2National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC

Corresponding author: Gloria L. Beckles, Division of Diabetes Translation, National Center for Chronic Disease Prevention and Health Promotion, CDC.
Telephone: 770-488-1272; E-mail: [email protected].

Introduction
The factors that influence the socioeconomic position of

individuals and groups within industrial societies also influence
their health (1,2). Socioeconomic position has continuous and
graded effects on health that are cumulative over a lifetime.
The socioeconomic conditions of the places where persons
live and work have an even more substantial influence on
health than personal socioeconomic position (3,4). In the
United States, educational attainment and income are the
indicators that are most commonly used to measure the effect
of socioeconomic position on health. Research indicates
that substantial educational and income disparities exist
across many measures of health (1,5–8). A previous report
described the magnitude and patterns of absolute and relative
measures of disparity in noncompletion of high school and
poverty in 2005 and 2009 (9). Notable disparities defined by
race/ethnicity, socioeconomic factors, disability status, and
geographic location were identified for 2005 and 2009, with
no evidence of a temporal decrease in racial/ethnic disparities,
whereas socioeconomic and disability disparities increased
from 2005 to 2009.

The analysis and discussion of educational attainment
and income that follow are part of the second CDC Health
Disparities and Inequalities Report (CHDIR) and update
information on disparities in the prevalence of noncompletion
of high school and poverty presented in the first CHDIR (8).
The 2011 CHDIR (9) was the first CDC report to describe
disparities across a wide range of diseases, behavioral risk
factors, environmental exposures, social determinants, and
health-care access. The topic presented in this report is based
on criteria that are described in the 2013 CHDIR Introduction
(10). The purposes of this analysis are to discuss and raise
awareness about group differences in levels of noncompletion
of high school and poverty and to motivate actions to reduce
these disparities.

Methods
To monitor progress toward eliminating health disparities

in the prevalence of noncompletion of high school and

poverty, CDC analyzed 2009 and 2011 data from the Current
Population Survey (CPS), using methods described previously
(8). The CPS is a cross-sectional monthly household survey
of a representative sample of the civilian, noninstitutionalized
U.S. household population that is conducted jointly by the
U.S. Census Bureau and the Bureau of Labor Statistics (11).
Data on the continuous income-to-poverty ratio (IPR) in
the 2009 and 2011 National Bureau of Economic Research
(NBER) data sets based on the March CPS were merged with
the March supplement files from the 2009 and 2011 Integrated
Public Use Microdata Series — Current Population Surveys
(IPUMS-CPS) (12,13).

Self-reported data were collected on various characteristics,
including demographic, socioeconomic, and geographic
characteristics and place of birth. Group disparities in age-
standardized prevalence of noncompletion of high school and
poverty were assessed according to sex, race/ethnicity, age,
educational attainment, poverty status, disability status, place
of birth, world region (country) of birth, U.S. census region
of residence, and metropolitan area of residence.

Race/ethnicity categories included non-Hispanic white, non-
Hispanic black, American Indian/ Alaska Native, Asian/Pacific
Islander, Hispanic, and multiple races. Age groups included
25–44, 45–64, 65–79, and ≥80 years. Educational attainment
categories included less than high school, high school graduate
or equivalent, some college, and college graduate. Poverty
status was derived from the IPR, which is based on family
income relative to federally established poverty thresholds that
are revised annually to reflect changes in the cost of living as
measured by the Consumer Price Index (14).

Disability status was defined by the national data collection
standards released by the U.S. Department of Health and
Human Services (HHS) in 2011 (15). World region of birth
was aggregated to approximate the regions of the world from
which the foreign born now originate (16). Absolute and
relative disparities in noncompletion of high school were
assessed separately for adults aged ≥25 years and 18–24 years;
for poverty, disparities were assessed for the total population
aged ≥18 years.

Disparities between groups were measured as deviations from
a referent category rate. Referent categories were usually those

Supplement

10 MMWR / November 22, 2013 / Vol. 62 / No. 3

that had the most favorable group estimates for most variables;
for racial/ethnic comparisons, white males and females were
selected because they were the largest group (17,18). Absolute
difference was measured as the simple difference between a
group estimate and the estimate for its respective reference
category, or referent group. Relative difference, a percentage,
was obtained by dividing the absolute difference by the value
in the referent category and multiplying by 100. To evaluate
changes in disparity over time, relative differences for the
groups in 2009 were subtracted from relative differences in
2011 (17,18). The z statistic and a two-tailed test at p<0.05
with Bonferroni correction for multiple comparisons were
used to test for the statistical significance of the observed
absolute and relative differences and for changes over time.
To calculate the standard errors for testing the change over
time, a previously described method was used (19), modified
to account for the parameter being compared (i.e., relative
difference). Statistically significant increases and decreases in
relative differences from 2009 to 2011 were interpreted as
increases and decreases in disparity, respectively. CDC used
statistical software to account for the complex sample design
of the CPS and to produce point estimates, standard errors,
and 95% confidence intervals. Estimates were age standardized
by the direct method to the year 2000 age distribution of the
U.S. population (20). Estimates with relative standard error
≥30% were not reported.

Results
In the 2011 population aged ≥25 years, statistically

significant absolute disparities in noncompletion of high school
were identified for all the characteristics studied (Table 1).
Noncompletion of high school increased with age; the absolute
differences between the age-specific percentages in the referent
group (45–64 years) and the age groups 65–79 years and
≥80 years were 6.6 and 14.8 percentage points, respectively.
The absolute racial/ethnic difference between non-Hispanic
whites and each of the other racial ethnic groups was highest
for Hispanics (30.4 percentage points), lowest for the multiple
races group (4.0 percentage points), and intermediate for non-
Hispanic American Indian/Alaska Natives (11.6 percentage
points), and non-Hispanic blacks (8.8 percentage points). This
pattern was similar in both sexes, except that among women,
the absolute difference for the multiple races group (3.1
percentage points) was not statistically significant. Absolute
differences between the age-standardized percentages of adults
who had not completed high school in each poverty status
group and the referent group (high income, IPR ≥4) were
statistically significant overall and in both men and women.

Noncompletion of high school increased with increasing
poverty; the absolute difference for the poorest group was
approximately three times the absolute difference for the
middle-income group (6.4 versus 1.7 percentage points).
Significant absolute differences between adults with and
without a disability in noncompletion of high school also
were found (total: 9.8 percentage points; men: 9.5 percentage
points; women: 10.1 percentage points).

Among adults aged ≥25 years in 2011, noncompletion of
high school was generally more common among foreign-
born than U.S.-born adults (Table 1). Significant absolute
differences from the U.S. born were observed in the total
population (24.9 percentage points), among non-Hispanic
whites (3.1 percentage points), A/PIs (9.0 percentage points),
and Hispanics (27.7 percentage points). Disparities in
noncompletion of high school also were found according to
world region (countries) of birth. In 2011, significant absolute
differences were found between persons born in the United
States (referent group) and those born in Latin American and
Caribbean countries (46.1 percentage points) or in countries
in Asia and the Pacific (6.1 percentage points). In 2011,
significant absolute differences were also found between
residents of the U.S. census regions of the Midwest, South,
or West and the referent group (the Northeast). The absolute
difference in age-standardized noncompletion of high school
between residents who lived inside metropolitan areas and
those who lived outside metropolitan areas (referent group)
also was significant. In 2009 and 2011, the magnitude and
pattern of age, poverty status, and disability differences were
similar in men and women. No significant differences were
identified in the relative differences of any these characteristics
from 2009 to 2011.

Among younger adults aged 18–24 years in 2011,
significant disparities in place of birth and in demographic,
socioeconomic, disability, and geographic characteristics were
found in the age-standardized percentages of adults who did
not complete high school (Table 2). Unlike adults aged ≥25
years, the absolute difference between the percentages of young
adults who did not complete high school in the younger age
group (18–19 years) and older referent group (20–24 years)
was significant (33.1 percentage points). The relative difference
between persons aged 18–19 years and the referent group
increased significantly by 61.6 percentage points from 2009
to 2011, whereas no change occurred from 2009 to 2011 in
age-specific disparities in the older population (≥25 years)
(Table 1). Among racial/ethnic groups, absolute differences
from non-Hispanic whites were only significant among
non-Hispanic blacks (7.2 percentage points) and Hispanics
(12.4 percentage points), with the magnitude and pattern

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 11

TABLE 1. Age-standardized* percentage of adults aged ≥25 years who did not complete high school, by selected characteristics — Integrated
Public Use Microdata Series, Current Population Survey, United States, 2009 and 2011

Characteristic

2009 2011

Change in
relative

difference
from 2009 to

2011
(percentage

points)

% who did
not complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

% who did
not complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 14.1 (0.2) 1.4† 11.0 13.2 (0.2) 1.4† 11.9 0.9
Female 12.7 (0.2) Ref. Ref. 11.8 (0.2) Ref. Ref. Ref.

Age group (yrs)§
Both sexes

25–44 11.7 (0.2) 0.7 6.2 11.0 (0.2) 0.4 4.1 -2.1
45–64 11.0 (0.2) Ref. Ref. 10.5 (0.2) Ref. Ref. Ref.
65–79 19.5 (0.4) 8.5† 77.3 17.2 (0.4) 6.6† 63.1 -14.2

≥80 27.6 (0.8) 16.6† 150.7 25.3 (0.7) 14.8† 140.6 -10.0
Male

25–44 13.0 (0.3) 1.3† 11.5 12.1 (0.2) 1.0† 9.0 -2.5
45–64 11.7 (0.3) Ref. Ref. 11.1 (0.3) Ref. Ref. Ref.
65–79 18.5 (0.6) 6.8† 58.0 16.6 (0.5) 5.4† 48.7 -9.4

≥80 27.2 (1.1) 15.5† 132.1 26.0 (1.1) 14.9† 133.5 1.4
Female

25–44 10.3 (0.2) 0 -0.3 9.8 (0.2) -0.2 -1.7 -1.4
45–64 10.4 (0.2) Ref. Ref. 9.9 (0.2) Ref. Ref. Ref.
65–79 20.4 (0.5) 10.0† 96.7 17.7 (0.5) 7.7† 77.8 19.0

≥80 27.9 (0.9) 17.5† 169.0 24.9 (0.8) 15.0† 150.6 -18.4
Race/Ethnicity

Both sexes
White, non-Hispanic 8.0 (0.1) Ref. Ref. 7.3 (0.1) Ref. Ref. Ref.
Black, non-Hispanic 17.0 (0.4) 9.0† 112.6 16.1 (0.4) 8.8† 121.2 8.9
Asian/Pacific Islander 12.7 (0.7) 4.7† 59.1 12.1 (0.6) 4.9† 66.8 7.7
American Indian/Alaska Native 20.0 (1.8) 12.0† 149.8 18.8 (1.9) 11.5† 158.5 8.8
Multiple races 13.4 (1.0) 5.4† 67.8 11.3 (1.0) 4.0† 55.4 -12.4
Hispanic¶ 40.1 (0.6) 32.1† 400.8 37.7 (0.5) 30.4† 419.3 18.5

Male
White, non-Hispanic 8.6 (0.2) Ref. Ref. 7.9 (0.2) Ref. Ref. Ref.
Black, non-Hispanic 17.6 (0.6) 9.1† 106.0 17.3 (0.6) 9.4† 120.1 14.1
Asian/Pacific Islander 10.3 (0.8) 1.8 20.9 10.2 (0.7) 2.3† 29.5 8.6
American Indian/Alaska Native 21.0 (2.2) 12.4† 145.0 20.8 (2.4) 13.0† 165.1 20.1
Multiple races 13.3 (1.4) 4.7† 55.1 12.7 (1.5) 4.8† 61.6 6.5
Hispanic 41.1 (0.7) 32.5† 380.4 38.4 (0.7) 30.5† 388.2 7.8

Female
White, non-Hispanic 7.4 (0.2) Ref. Ref. 6.7 (0.2) Ref. Ref. Ref.
Black, non-Hispanic 16.6 (0.5) 9.2† 123.3 15.2 (0.5) 8.5† 127.1 3.8
Asian/Pacific Islander 14.6 (0.8) 7.2† 96.1 13.6 (0.7) 7.0† 104.2 8.1
American Indian/Alaska Native 19.6 (2.1) 12.2† 163.8 16.9 (2.1) 10.2† 152.8 -11.0
Multiple races 13.1 (1.2) 5.7† 76.3 9.8 (1.2) 3.1 46.3 -30.1
Hispanic 38.7 (0.6) 31.3† 420.0 36.8 (0.6) 30.1† 450.9 30.9

Income-to-poverty ratio**
Both sexes

Poor, <1.00 18.4 (0.5) 7.4† 66.6 16.6 (0.4) 6.4† 61.9 -4.7
Near poor, 1.00–1.9 15.6 (0.3) 4.6† 41.7 14.7 (0.3) 4.4† 42.9 1.2
Middle income, 2.00–3.9 13.2 (0.2) 2.1† 19.4 12.0 (0.2) 1.7† 16.9 -2.5
High income, ≥4.0 11.0 (0.2) Ref. Ref. 10.3 (0.2) Ref. Ref. Ref.

Male
Poor, <1.00 18.9 (0.6) 7.5† 66.5 17.3 (0.6) 6.5† 60.1 -6.4
Near poor, 1.00–1.9 17.0 (0.5) 5.7† 50.2 16.0 (0.5) 5.2† 48.1 -2.1
Middle income, 2.00–3.9 14.1 (0.3) 2.8† 24.3 12.7 (0.3) 1.9† 17.6 -6.7
High income, ≥4.0 11.3 (0.3) Ref. Ref. 10.8 (0.3) Ref. Ref. Ref.

Female
Poor, <1.00 17.9 (0.5) 7.2† 67.1 15.9 (0.5) 6.2† 63.0 -4.1
Near poor, 1.00–1.9 14.3 (0.3) 3.6† 34.1 13.5 (0.4) 3.7† 37.9 3.8
Middle income, 2.00–3.9 12.3 (0.3) 1.6† 14.9 11.3 (0.3) 1.5† 15.8 0.9
High income, ≥4.0 10.7 (0.2) Ref. Ref. 9.8 (0.2) Ref. Ref. Ref.

See table footnotes on the next page.

Supplement

12 MMWR / November 22, 2013 / Vol. 62 / No. 3

TABLE 1. (Continued) Age-standardized* percentage of adults aged ≥25 years who did not complete high school, by selected characteristics
— Integrated Public Use Microdata Series, Current Population Survey, United States, 2009 and 2011

Characteristic

Change in
relative

difference
from 2009 to

2011
(percentage

points)

2009 2011

% who did
not complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

% who did
not complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Disability status
Both sexes

Disability 23.3 (0.5) 11.4† 95.5 21.1 (0.5) 9.8† 87.3 -8.3
No disability 11.9 (0.2) Ref. Ref. 11.2 (0.1) Ref. Ref. Ref.

Male
Disability 23.3 (0.8) 10.4† 80.8 21.5 (0.7) 9.5† 78.9 -1.9
No disability 12.9 (0.2) Ref. Ref. 12.0 (0.2) Ref. Ref. Ref.

Female
Disability 23.2 (0.6) 12.2† 111.2 20.6 (0.7) 10.1† 96.3 -14.9
No disability 11.0 (0.2) Ref. Ref. 10.5 (0.2) Ref. Ref. Ref.

Place of birth
All racial/ethnic groups

United States or U.S. territory 9.6 (0.2) Ref. Ref. 8.7 (0.1) Ref. Ref. Ref.
Foreign country 35.6 (0.6) 26.0† 270.0 33.7 (0.6) 24.9† 286.0 16.0

White, non-Hispanic
United States or U.S. territory 8.0 (0.2) Ref. Ref. 7.2 (0.1) Ref. Ref. Ref.
Foreign country 10.6 (0.7) 2.6† 33.1 10.3 (0.7) 3.1† 43.3 10.3

Black, non-Hispanic
United States or U.S. territory 17.3 (0.5) Ref. Ref. 16.1 (0.4) Ref. Ref. Ref.
Foreign country 15.7 (1.3) -1.6 -9.2 16.3 (1.4) 0.2 1.1 10.3

Asian/Pacific Islander
United States or U.S. territory 4.7 (1.1) Ref. Ref. 4.7 (1.0) Ref. Ref. Ref.
Foreign country 14.2 (0.8) 9.5† 199.8 13.7 (0.7) 9.0† 191.3 -8.5

American Indian/Alaska Native
United States or U.S. territory 20.1 (1.8) Ref. Ref. 19.0 (2.0) Ref. Ref. Ref.
Foreign country —†† — NA NA 20.0 (5.7) 1.0 5.2 NA

Multiple races
United States or U.S. territory 14.4 (1.2) Ref. Ref. 12.3 (1.2) Ref. Ref. Ref.
Foreign country — — NA NA — — NA NA NA

Hispanic
United States or U.S. territory 22.3 (1.0) Ref. Ref. 20.5 (0.9) Ref. Ref. Ref.
Foreign country 50.5 (0.7) 28.1† 125.9 48.2 (0.7) 27.7† 135.3 9.3

World region (country) of birth
United States 9.8 (0.2) Ref. Ref. 8.8 (0.1) Ref. Ref. Ref.
Canada, Europe, Australia, or

New Zealand
5.3 (1.0) -4.5† -45.9 6.0 (1.3) -2.8 -32.2 13.7

Mexico, South America, Central
America, or Caribbean

57.2 (0.9) 47.4† 485.8 54.9 (0.9) 46.1† 522.9 37.1

Africa or the Middle East 11.6 (2.3) 1.8 18.5 9.1 (2.3) 0.3 2.9 -15.6
Asia or the Pacific Islands 17.3 (1.7) 7.5† 77.3 14.9 (1.6) 6.1† 69.3 -8.0

U.S. census region§§
Northeast 9.4 (0.4) Ref. Ref. 8.7 (0.4) Ref. Ref. Ref.
Midwest 10.7 (0.4) 1.3 13.9 10.2 (0.4) 1.5† 17.4 3.5
South 12.2 (0.4) 2.8† 30.4 11.3 (0.3) 2.6† 29.5 -0.8
West 12.6 (0.6) 3.3† 34.8 11.0 (0.5) 2.3† 26.8 -8.0

Area of residence
Inside metropolitan area 16.4 (0.4) 1.4† 9.1 15.7 (0.4) 1.4† 9.5 0.4
Outside metropolitan area 15.0 (0.5) Ref. Ref. 14.3 (0.5) Ref. Ref. Ref.

Abbreviations: FPL = federal poverty level; NA = not applicable; Ref. = referent; SE = standard error.
* Age standardized to the 2000 U.S. standard population.
† Difference between a group estimate and the estimate for its respective referent group is significant (p<0.05, two-tailed z test with Bonferroni correction for multiple

comparisons).
§ Age-specific estimates are not age standardized.
¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** On the basis of the U.S. FPL. FPL was based on U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html).
†† Estimate is statistically unreliable because relative SE ≥30%.
§§ Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 13

See table footnotes on the next page.

TABLE 2. Age-standardized* percentage of adults aged 18–24 years who did not complete high school, by selected characteristics — Integrated
Public Use Microdata Series, Current Population Survey, United States, 2009 and 2011

Characteristic

2009 2011
Change in

relative
difference

from 2009 to
2011

(percentage
points)

% who
did not

complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

% who
did not

complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 22.4 (0.5) 4.6† 25.8 22.4 (0.5) 3.6† 18.9 -6.9
Female 17.8 (0.4) Ref. Ref. 18.8 (0.5) Ref. Ref. Ref.

Age group (yrs)§
Both sexes

18–19 40.1 (0.7) 28.7† 252.8 43.6 (0.8) 33.1† 314.4 61.6§§
20–24 11.4 (0.4) Ref. Ref. 10.5 (0.4) Ref. Ref. Ref.

Male
18–19 43.9 (1.0) 30.9† 238.9 47.0 (1.1) 35.4† 305.2 66.3
20–24 13.0 (0.5) Ref. Ref. 11.6 (0.5) Ref. Ref. Ref.

Female
18–19 36.2 (1.0) 26.4† 271.3 40.2 (1.1) 30.8† 327.1 55.8
20–24 9.7 (0.4) Ref. Ref. 9.4 (0.4) Ref. Ref. Ref.

Race/Ethnicity
Both sexes

White, non-Hispanic 16.3 (0.4) Ref. Ref. 17.2 (0.5) Ref. Ref. Ref.
Black, non-Hispanic 24.4 (1.1) 8.2† 50.2 24.3 (1.0) 7.2† 41.7 -8.5
Asian/Pacific Islander 13.8 (1.3) -2.5 -15.3 16.2 (2.0) -1.0 -5.9 9.4
American Indian/Alaska Native 25.1 (3.2) 8.9 54.4 26.0 (3.7) 8.9 51.7 -2.7
Multiple races 19.9 (2.2) 3.6 22.2 18.8 (2.2) 1.6 9.4 -12.8
Hispanic¶ 31.5 (1.0) 15.2† 93.7 29.5 (0.9) 12.4† 72.0 -21.7

Male
White, non-Hispanic 18.2 (0.6) Ref. Ref. 18.4 (0.6) Ref. Ref. Ref.
Black, non-Hispanic 26.7 (1.8) 8.6† 47.2 26.7 (1.5) 8.3† 45.0 -2.2
Asian/Pacific Islander 14.6 (1.8) -3.6 -19.7 17.1 (2.5) -1.3 -7.2 12.5
American Indian/Alaska Native 28.6 (4.7) 10.4 57.4 31.0 (5.9) 12.6 68.4 11.0
Multiple races 22.5 (3.1) 4.3 23.6 16.0 (2.5) -2.4 -13.0 -36.6
Hispanic 35.1 (1.4) 17.0† 93.3 32.0 (1.2) 13.6† 73.8 -19.5

Female
White, non-Hispanic 14.3 (0.5) Ref. Ref. 15.9 (0.6) Ref. Ref. Ref.
Black, non-Hispanic 22.2 (1.3) 7.9† 55.4 22.0 (1.5) 6.1 38.5 -16.9
Asian/Pacific Islander 12.7 (1.6) -1.6 -11.3 15.2 (2.3) -0.7 -4.5 6.9
American Indian/Alaska Native 21.9 (4.1) 7.6 53.4 23.2 (5.0) 7.4 46.3 -7.0
Multiple races 17.0 (2.8) 2.7 18.9 21.3 (3.4) 5.4 34.1 15.2
Hispanic 27.6 (1.2) 13.3† 92.8 26.6 (1.1 10.7† 67.4 -25.4

Income-to-poverty ratio**
Both sexes

Poor, <1.00 23.7 (1.1) 5.9† 32.9 23.5 (0.9) 4.7† 25.1 -7.8
Near poor, 1.00–1.9 22.8 (0.8) 5.0† 27.8 22.3 (0.9) 3.6† 19.0 -8.7
Middle income, 2.00–3.9 19.8 (0.6) 1.9 10.8 20.4 (0.6) 1.6 8.5 -2.3
High income, ≥4.0 17.9 (0.5) Ref. Ref. 18.8 (0.5) Ref. Ref. Ref.

Male
Poor, <1.00 25.0 (1.4) 5.4† 27.4 24.9 (1.3) 4.4 21.3 -6.1
Near poor, 1.00–1.9 26.0 (1.2) 6.3† 32.0 24.6 (1.3) 4.1† 19.9 -12.1
Middle income, 2.00–3.9 22.5 (0.9) 2.9 14.6 22.0 (0.9) 1.5 7.2 -7.4
High income, ≥4.0 19.7 (0.8) Ref. Ref. 20.5 (0.7) Ref. Ref. Ref.

Female
Poor, <1.00 22.5 (1.4) 6.5† 40.6 22.0 (1.1) 5.0† 29.6 -11.0
Near poor, 1.00–1.9 19.5 (1.0) 3.5† 21.6 20.0 (1.1) 3.0 17.9 -3.7
Middle income, 2.00–3.9 17.0 (0.7) 1.0 6.3 18.6 (0.8) 1.6 9.2 3.0
High income, ≥4.0 16.0 (0.7) Ref. Ref. 17.0 (0.7) Ref. Ref. Ref.

Disability status
Both sexes

Disability 32.4 (2.0) 12.7† 64.5 35.5 (2.3) 15.4† 76.3 11.8
No disability 19.7 (0.4) Ref. Ref. 20.1 (0.4) Ref. Ref. Ref.

Supplement

14 MMWR / November 22, 2013 / Vol. 62 / No. 3

TABLE 2. (Continued) Age-standardized* percentage of adults aged 18–24 years who did not complete high school, by selected characteristics
— Integrated Public Use Microdata Series — Current Population Survey, United States, 2009 and 2011

Characteristic

2009 2011
Change in

relative
difference

from 2009 to
2011

(percentage
points)

% who
did not

complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

% who
did not

complete
high school (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Male
Disability 29.5 (2.4) 7.3† 32.7 38.0 (2.8) 16.1† 73.8 41.1§§
No disability 22.2 (0.5) Ref. Ref. 21.8 (0.5) Ref. Ref. Ref.

Female
Disability 36.0 (3.4) 18.8† 109.4 32.0 (3.3) 13.6† 73.7 -35.7
No disability 17.2 (0.4) Ref. Ref. 18.4 (0.5) Ref. Ref. Ref.

Place of birth
All racial/ethnic groups

United States or U.S. territory 18.5 (0.4) Ref. Ref. 19.1 (0.4) Ref. Ref. Ref.
Foreign country 32.7 (1.4) 14.2† 77.1 31.4 (1.4) 12.0† 63.0 -14.1

White, non-Hispanic
United States or U.S. territory 16.5 (0.4) Ref. Ref. 17.3 (0.5) Ref. Ref. Ref.
Foreign country 13.2 (2.3) -3.4 -20.3 16.6 (3.2) -0.7 -4.1 16.2

Black, non-Hispanic
United States or U.S. territory 24.9 (1.2) Ref. Ref. 24.7 (1.1) Ref. Ref. Ref.
Foreign country 22.1 (4.1) -2.8 -11.4 23.3 (4.4) -1.4 -5.6 5.7

Asian/Pacific Islander
United States or U.S. territory —†† — Ref. Ref. — — Ref. Ref. Ref.
Foreign country 18.1 (2.3) NA NA 22.1 (3.6) NA NA NA

American Indian/Alaska Native
United States or U.S. territory 24.7 (3.2) Ref. Ref. 26.8 (3.9) Ref. Ref. Ref.
Foreign country 0 (0) -24.7 -100.0 — — NA NA NA

Multiple races
United States or U.S. territory 21.1 (2.7) Ref. Ref. 20.4 (2.6) Ref. Ref. Ref.
Foreign country — — NA NA — — NA NA NA

Hispanic
United States or U.S. territory 24.6 (1.5) Ref. Ref. 24.5 (1.4) Ref. Ref. Ref.
Foreign country 44.9 (1.9) 20.4† 82.9 40.6 (1.8) 16.0† 65.4 -17.4

World region (country) of birth
United States 18.4 (0.3) Ref. Ref. 19.2 (0.4) Ref. Ref. Ref.
Canada, Europe, Australia, or New

Zealand
— — NA NA — — NA NA NA

Mexico, South America, Central
America, or the Caribbean

46.5 (2.1) 28.1† 153.1 42.5 (2.1) 23.4† 121.9 -31.3

Africa or the Middle East — — NA NA 30.5 (0) 11.3 58.9 NA
Asia or the Pacific Islands 20.1 (4.0) 1.7 9.3 24.8 (3.8) 5.6 29.2 19.9

U.S. census region¶¶
Northeast 15.5 (1.1) Ref. Ref. 18.0 (1.2) Ref. Ref. Ref.
Midwest 18.6 (0.9) 3.1 20.0 19.2 (0.8) 1.2 6.8 -13.2
South 20.8 (0.9) 5.3† 34.1 21.1 (0.9) 3.1 17.3 -16.8
West 22.6 (1.2) 7.1† 46.0 20.4 (1.0) 2.4 13.4 -32.6

Residence in metropolitan area
Inside metropolitan area 21.3 (0.8) 0.1 0.3 22.1 (0.7) -1.3 -5.3 -5.6
Outside metropolitan area 21.3 (1.0) Ref. Ref. 23.4 (1.0) Ref. Ref. Ref.

Abbreviations: FPL = federal poverty level; NA = not applicable; Ref. = referent; SE = standard error.
* Age standardized to the 2000 U.S. standard population.
† Difference between a group estimate and the estimate for its respective referent group is significant (p<0.05, two-tailed z test with Bonferroni correction for multiple

comparisons).
§ Age-specific estimates are not age standardized.
¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** On the basis of the U.S. FPL. FPL was based on U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html).
†† Estimate is statistically unreliable because relative SE ≥30%.
§§ Difference between the relative differences in 2011 and 2009 is significant (p<0.05, two-tailed z test with Bonferroni correction for multiple comparisons).
¶¶ Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 15

similar in men and women. Overall, absolute differences in
noncompletion of high school between the referent group
(high income) and those who lived in poor (4.7 percentage
points) or near-poor families (3.6 percentage points) were
significant; however, absolute differences were only significant
for men in middle-income families and women in poor
families. Significant absolute differences in noncompletion
of high school also were found among young adults with a
disability (15.4 percentage points); however, unlike men aged
≥25 years, the disparity among younger adult men worsened
from 2009 to 2011 by 41.1 percentage points. No temporal
change in disability disparity was observed among young
adult females (Table 2). In 2011, absolute differences in the
age-standardized percentage of persons who did not complete
high school among those who were foreign born and U.S. born
(referent group) were significant in the total population (12.0
percentage points) and among Hispanics (16.0 percentage
points). In addition, absolute differences were only significant
between U.S.-born young adults and young adults born in
Latin American and Caribbean countries (23.4 percentage
points). No significant differences were found by U.S. census
region or metropolitan area. No significant changes in the
U.S. census region disparities occurred from 2009 to 2011.

In 2011, overall and for men and women, significant absolute
differences in the age-standardized percentages of adults in poor
families (IPR <1.00) were found among the youngest adults,
non-Hispanic blacks, and Hispanics; all groups that had not
completed college; and adults with disabilities (Table 3). In 2009
and 2011, disparities in poverty increased with decreasing level of
educational attainment, with the greatest disparity experienced
by the group with the lowest level of educational attainment.
Significant absolute differences in the age-standardized
percentages in poor families were found between persons of
either sex with a disability and those with no disability (referent
group) (men: 3.2 percentage points; women 3.5 percentage
points). In 2009 and 2011, the absolute differences between
persons who were foreign born and U.S. born (referent group)
in age-standardized percentages of adults in poor families were
significant in the total population (1.7 and 1.6 percentage
points, respectively) but not by race/ethnicity. In addition,
significant absolute differences also were found between adults
born in Latin American and Caribbean countries and those born
in the United States. In 2009 and 2011, significant absolute
differences in the percentages of adults who lived in poverty were
found between residents of the U.S. census regions of the West,
South, or Midwest and the referent group (Northeast region)
but not between residents who lived inside compared with
outside metropolitan areas. From 2009 to 2011, no statistically
significant changes in the relative differences in poverty by any
characteristic were found (Table 3).

Discussion
The findings in this report indicate that racial/ethnic,

socioeconomic, and geographic disparities in noncompletion
of high school and poverty persist in the U.S. adult population;
little evidence of improvement from 2009 to 2011 was
identified. Within each year studied to date, significant
absolute and relative differences were found; however, between
years, these differences were not statistically different. The
pattern of disparities is consistent with sociodemographic and
geographic differences reported by several national surveys
(6–8,16,21–25). The findings also reveal that young racial/
ethnic, foreign-born, and poor adults might be especially
vulnerable to early onset and progression of poor health as
evidenced by marked disparities in noncompletion of high
school among these subgroups.

Educational attainment and income provide psychosocial
and material resources that protect against exposure to health
risks in early and adult life (1–3). Persons with low levels of
education and income generally experience increased rates of
mortality, morbidity, and risk-taking behaviors and decreased
access to and quality of health care (1,6–8). This report
confirms that the lowest levels of education and income
are most common and persistent among subgroups that
systematically exhibit the poorest health. For example, two
out of five Hispanics and nearly one out of five non-Hispanic
blacks or American Indian/Alaska Natives had not completed
high school, and at least one out of 10 of these racial/ethnic
groups had incomes less than the official poverty threshold.
However, substantial empirical evidence from the United
States and elsewhere consistently shows no thresholds in the
relationships between education or income and health. Among
children and adults in the overall population and within racial/
ethnic groups, rates of mortality, morbidity, and poor health
behaviors decrease in a continuous and graded manner with
increasing levels of education and income (6,7,23–25).

Health-promotion efforts have emphasized racial/ethnic
disparities in health as part of an approach to risk reduction that
focuses on groups at high risk, with little or no improvement
in disparities (24,26). The patterns described in this report
suggest that interventions and policies that are also designed to
take account of the influence of educational attainment, family
income, and other socioeconomic conditions on health risks
in the entire population might prove to be more effective in
reducing health disparities (27,28).

Supplement

16 MMWR / November 22, 2013 / Vol. 62 / No. 3

See table footnotes on the next page.

TABLE 3. Age-standardized* percentage of adults aged ≥18 years with incomes less than the federal poverty level, by selected characteristics
— Integrated Public Use Microdata Series — Current Population Survey, United States, 2009 and 2011

Characteristic

2009 2011
Change in relative

difference from
2009 to 2011

(percentage points)
% with IPR

<1.00 (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Percentage
with IPR

<1.00 (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 11.4 (0.2) Ref. Ref. 13.2 (0.2) Ref. Ref. Ref.
Female 11.9 (0.1) 0.5† 4.0 13.5 (0.2) 0.4 2.9 -1.1

Age group (yrs)§
Both sexes

18–24 12.8 (0.3) 1.9† 17.1 15.1 (0.4) 2.4† 18.6 1.5
25–44 11.9 (0.2) 0.9 8.5 13.5 (0.2) 0.8 6.2 -2.3
45–64 11.0 (0.2) Ref. Ref. 12.8 (0.2) Ref. Ref. —
65–79 11.1 (0.3) 0.1 1.1 12.5 (0.3) -0.2 -1.6 -2.7
≥80 12.1 (0.6) 1.1 10.0 12.8 (0.5) 0 0 -10

Male
18–24 12.3 (0.4) 1.7† 15.8 14.8 (0.5) 2.1† 16.2 0.4
25–44 11.7 (0.2) 1.1† 10.2 13.5 (0.3) 0.8 6.1 -4.1
45–64 10.6 (0.3) Ref. Ref. 12.7 (0.3) Ref. Ref. Ref.
65–79 11.1 (0.5) 0.4 4.2 11.6 (0.4) -1.1 -8.7 -12.9
≥80 11.4 (0.9) 0.8 7.5 13.0 (0.9) 0.3 2.1 -5.4

Female
18–24 13.3 (0.4) 2.1† 18.6 15.5 (0.5) 2.7† 21 2.4
25–44 12.0 (0.2) 0.8 6.9 13.6 (0.2) 0.8 6.3 -0.6
45–64 11.2 (0.2) Ref. Ref. 12.8 (0.3) Ref. Ref. Ref.
65–79 11.1 (0.4) -0.2 -1.7 13.3 (0.4) 0.5 4.0 5.7
≥80 12.4 (0.6) 1.2 10.4 12.6 (0.7) -0.2 -1.5 -11.9
Race/Ethnicity

Both sexes
White, non-Hispanic 10.7 (0.2) Ref. Ref. 12.4 (0.2) Ref. Ref. Ref.
Black, non-Hispanic 14.4 (0.4) 3.7† 34.8 16.4 (0.4) 4.1† 32.8 -2.0
Asian/Pacific Islander 10.7 (0.5) 0 0.1 11.5 (0.5) -0.8 -6.8 -6.9
American Indian/Alaska Native 15.3 (1.9) 4.7 43.6 18.9 (3.5) 6.6 53.4 9.7
Multiple races 11.2 (0.9) 0.5 4.6 12.4 (1.0) 0 0.2 -4.4
Hispanic¶ 14.5 (0.4) 3.9† 36.3 16.0 (0.4) 3.7† 29.7 -6.6

Male
White, non-Hispanic 10.5 (0.2) Ref. Ref. 12.2 (0.2) Ref. Ref. Ref.
Black, non-Hispanic 14.0 (0.5) 3.5† 33.4 15.6 (0.5) 3.4† 28.1 -5.3
Asian/Pacific Islander 10.8 (0.6) 0.3 3.0 12.0 (0.7) -0.2 -1.6 -4.6
American Indian/Alaska Native 12.9 (1.9) 2.4 22.5 18.6 (3.1) 6.4 52.1 29.6
Multiple races 9.4 (1.2) -1.1 -10.3 12.6 (1.4) 0.4 3.3 13.6
Hispanic 14.3 (0.5) 3.8† 36.2 15.8 (0.5) 3.5† 29.0 -7.2

Female
White, non-Hispanic 10.8 (0.2) Ref. Ref. 12.5 (0.2) Ref. Ref. Ref.
Black, non-Hispanic 14.7 (0.5) 3.9† 35.6 17.0 (0.6) 4.5† 36.1 0.5
Asian/Pacific Islander 10.5 (0.6) -0.3 -3.0 11.1 (0.5) -1.4 -11.1 -8.1
American Indian/Alaska Native 17.7 (2.4) 6.9 63.8 19.3 (4.3) 6.8 54.2 -9.5
Multiple races 12.6 (1.2) 1.8 16.9 12.1 (1.2) -0.4 -3.4 -20.4
Hispanic 14.9 (0.4) 4.1† 37.6 16.2 (0.4) 3.8† 30.1 -7.5

Educational attainment
Both sexes

Less than high school 15.8 (0.4) 5.6† 55.0 17.6 (0.4) 6.1† 53.1 -1.8
High school graduate or equivalent 11.8 (0.2) 1.7† 16.2 13.8 (0.3) 2.3† 20.3 4.2
Some college 10.9 (0.2) 0.7 6.8 12.9 (0.3) 1.4† 12.5 5.8
College graduate 10.2 (0.2) Ref. Ref. 11.5 (0.3) Ref. Ref. Ref.

Male
Less than high school 15.1 (0.5) 4.6† 43.5 17.1 (0.5) 5.6† 48.4 4.9
High school graduate or equivalent 11.2 (0.3) 0.7 7.1 13.2 (0.3) 1.6† 14.1 7.1
Some college 10.8 (0.3) 0.3 2.5 12.8 (0.3) 1.3† 11.5 9.0
College graduate 10.5 (0.3) Ref. Ref. 11.5 (0.4) Ref. Ref. Ref.

Female
Less than high school 16.6 (0.5) 6.7† 66.9 18.0 (0.5) 6.7† 58.4 -8.5
High school graduate or equivalent 12.5 (0.3) 2.6† 26.0 14.6 (0.3) 3.2† 27.6 1.6
Some college 11 (0.2) 1.0† 10.5 13.0 (0.3) 1.6† 13.7 3.3
College graduate 10 (0.3) Ref. Ref. 11.5 (0.3) Ref. Ref. Ref.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 17

TABLE 3. (Continued) Age-standardized* percentage of adults aged ≥18 years with incomes less than the federal poverty level, by selected
characteristics — Integrated Public Use Microdata Series — Current Population Survey, United States, 2009 and 2011

Characteristic

2009 2011
Change in relative

difference from
2009 to 2011

(percentage points)
% with IPR

<1.00 (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Percentage
with IPR

<1.00 (SE)

Absolute
difference

(percentage
points)

Relative
difference (%)

Disability status
Both sexes

Disability 14.7 (0.5) 3.4† 29.8 16.4 (0.6) 3.4† 25.8 -4.0
No disability 11.4 (0.1) Ref. Ref. 13.0 (0.2) Ref. Ref. Ref.

Male
Disability 14.1 (0.7) 2.9† 26.0 16.1 (0.7) 3.2† 24.9 -1.1
No disability 11.2 (0.2) Ref. Ref. 12.9 (0.2) Ref. Ref. Ref.

Female
Disability 15.4 (0.7) 3.8† 33.2 16.7 (0.8) 3.5† 26.8 -6.4
No disability 11.5 (0.2) Ref. Ref. 13.2 (0.2) Ref. Ref. Ref.

Place of birth
All racial/ethnic groups

United States or U.S. territory 11.4 (0.2) Ref. Ref. 13.1 (0.2) Ref. Ref. Ref.
Foreign country 13 (0.3) 1.7† 15.1 14.6 (0.3) 1.6† 12.4 -2.7

White, non-Hispanic
United States or U.S. territory 10.6 (0.2) Ref. Ref. 12.3 (0.2) Ref. Ref. Ref.
Foreign country 11.1 (0.6) 0.5 4.7 13.4 (0.7) 1.1 8.7 3.8

Black, non-Hispanic
United States or U.S. territory 14.4 (0.4) Ref. Ref. 16.5 (0.5) Ref. Ref. Ref.
Foreign country 13.3 (1.0) -1.1 -7.8 16.8 (1.3) 0.3 1.5 9.3

Asian/Pacific Islander
United States or U.S. territory 10.2 (1.4) Ref. Ref. 12.6 (1.8) Ref. Ref. Ref.
Foreign country 10.8 (0.6) 0.7 6.7 12.0 (0.6) -0.6 -4.8 -11.5

American Indian/Alaska Native
United States or U.S. territory 15.4 (1.9) Ref. Ref. 19.7 (3.7) Ref. Ref. Ref.
Foreign country —** — NA NA — — NA NA NA

Multiple races
United States or U.S. territory 11.8 (1.0) Ref. Ref. 12.9 (1.1) Ref. Ref. Ref.
Foreign country — — NA NA — — NA NA NA

Hispanic
United States or U.S. territory 13.7 (0.7) Ref. Ref. 16.0 (0.8) Ref. Ref. Ref.
Foreign country 14.8 (0.5) 1.1 7.8 16.3 (0.5) 0.2 1.5 -6.3

World region (country) of birth
United States 11.4 (0.1) Ref. Ref. 13.1 (0.2) Ref. Ref. Ref.
Canada, Europe, Australia, or

New Zealand
12.2 (1.8) 0.8 7.0 11.7 (1.7) -1.4 -10.6 -17.6

Mexico, South America, Central
America, or the Caribbean

15.6 (0.6) 4.2† 37.0 16.5 (0.6) 3.4† 26.0 -11.0

Africa or the Middle East 7.1 (1.2) -4.3 -37.5 14.0 (3.1) 0.9 6.6 44.1
Asia or the Pacific Islands 8.9 (1.1) -2.4 -21.4 12.6 (1.3) -0.5 -4.1 17.3

U.S. census region††
Northeast 9.4 (0.5) Ref. Ref. 9.8 (0.5) Ref. Ref. Ref.
Midwest 11.1 (0.4) 1.7† 18.2 12.7 (0.4) 2.9† 29.1 10.9
South 11.3 (0.3) 1.9† 20.3 13.8 (0.4) 4.0† 40.8 20.5
West 11.7 (0.5) 2.3† 24.5 12.9 (0.6) 3.1† 31.6 7.1

Residence in metropolitan area
Inside metropolitan area 12.7 (0.3) 0.2 1.2 14.4 (0.3) 0.4 2.5 1.3
Outside metropolitan area 12.6 (0.4) Ref. Ref. 14.1 (0.4) Ref. Ref. Ref.

Abbreviations: IPR = income-to-poverty ratio; NA = not applicable; Ref. = referent; SE = standard error.
* Age standardized to the 2000 U.S. standard population.
† Difference between a group estimate and the estimate for its respective referent group is significant.
§ Age-specific estimates are not age standardized.
¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** Estimate is statistically unreliable because relative SE ≥30%.
†† Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

18 MMWR / November 22, 2013 / Vol. 62 / No. 3

Limitations
The findings in this report are subject to at least two

limitations. First, all data were self-reported and therefore are
subject to recall and social desirability bias. Second, CDC
used cross-sectional data for the analyses; therefore, no causal
inferences can be drawn from the findings. The limited
findings for disparities in place of birth among racial/ethnic
groups might reflect small sample sizes in single years of data,
as suggested by unstable estimates in the foreign-born strata
of several racial/ethnic groups.

Conclusion
The U.S. Department of Education’s Institute of Education

Sciences recommends effective evidence-based interventions
to prevent or reduce the dropout rates among middle school
and high school students (29). The U.S. Task Force on
Community Preventive Services recommends interventions
that promote healthy social environments for low-income
children and families and to reduce risk-taking behaviors
among adolescents (30). Since 2011, HHS has released several
complementary initiatives to eliminate health disparities
(26,31). The 2011 HHS action plan focuses specifically on
reduction of racial/ethnic disparities but includes education
and social and economic conditions among its major strategic
areas (26). The 2012 National Prevention Council action
plan will implement strategies of the National Prevention
Strategy by targeting communities at greatest risk for health
disparities, disparities in access to care, and the capacity of the
prevention workforce; research to identify effective strategies;
and standardization and collection of data to better identify
and address disparities. CDC proposes increasing its efforts
to eliminate health disparities by focusing on surveillance,
analysis, and reporting of disparities and identifying and
applying evidence-based strategies to achieve health equity
(31). Integration of these efforts across federal departments;
among federal, state, and local levels of government; and with
nongovernment organizations could increase understanding
of how socioeconomic disparities in health arise and persist
and provide information on how best to design effective
interventions for populationwide and targeted approaches.

References
1. Lynch J, Kaplan G. Socioeconomic position. In: Berkman LF, Kawachi

I, eds. Social epidemiology. 1st ed. New York, NY: Oxford University
Press; 2000:13–35.

2. Adler NE, Stewart J. Health disparities across the lifespan: meaning,
methods, and mechanisms. In: Adler NE, Stewart J, eds. The biology
of disadvantage. New York, NY: New York Academy of Sciences;
2010;1186:5–23.

3. Ross CE, Mirowsky J. Neighborhood socioeconomic status and health:
context or composition? City Community 2008;7:163–79.

4. Macintyre S, Ellaway A, Cummins S. Place effects on health: how can
we conceptualise, operationalise and measure them? Soc Sci Med 2002;
55:125–39.

5. Krieger N, Williams DR, Moss NE. Measuring social class in public
health research: concepts, methodologies, and guidelines. Annu Rev
Public Health 1997;18:341–78.

6. CDC, National Center for Health Statistics. Health, United States,
2011: with special feature on socioeconomic status and health.
Hyattsville, MD: CDC, National Center for Health Statistics; 2012.

7. Singh GK, Siahpush M. Trends and disparities in socioeconomic and
behavioral characteristics, life expectancy, and cause-specific mortality
of native-born and foreign-born populations in the United States,
1979–2003. Int J Epidemiol 2006;35:903–19.

8. CDC. Education and income—United States, 2005 and 2009. In: CDC
health disparities and inequalities report—United States, 2011.MMWR
2011;60(Suppl; January 14, 2011).

9. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

10. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013 MMWR 2013;62(No. Suppl 3).

11. US Census Bureau. Current population survey. Design and methodology.
Technical paper 66. Washington, DC: US Census Bureau; October
2006. Available at http://www.census.gov/prod/2006pubs/tp-66.pdf.

12. National Bureau of Economics Research. NBER CPS supplements.
Cambridge, MA. Available at http://www.nber.org/data/current-
population-survey-data.html.

13. Minnesota Population Center. Integrated public use microdata series—Current
Population Survey [Online database]. Version 3.0. Minneapolis, MN:
Minnesota Population Center; 2010. Available at http://usa.ipums.org/usa.

14. DeNavas-Walt C, Proctor BD, Smith JC. Current Population Reports.
P60-239. Income, poverty and health insurance coverage in the United
States: 2010. Washington, DC: US Census Bureau; 2011.

15. US Department of Health and Human Services. Implementation
guidance on data collection standards for race, ethnicity, sex, primary
language, and disability status. Washington, DC: US Department of
Health and Human Services; 2011. Available at http://aspe.hhs.gov/
datacncl/standards/ACA/4302/index.pdf.

16. Kandel WA. The U.S. foreign-born population: trends and selected
characteristics. Washington, DC: Congressional Research Service; 2011.
Available at http://www.fas.org/sgp/crs/misc/R41592.pdf.

17. Keppel K, Pamuk E, Lynch J, et al. Methodological issues in measuring
health disparities. Hyattsville, MD: US Department of Health and
Human Services, CDC, National Center for Health Statistics. Vital
Health Stat 2005;2(141).

18. Keppel KG, Pearcy JN, Klein RJ. Measuring progress in Healthy People
2010. Statistical Notes, No. 25. Hyattsville, MD: National Center for
Health Statistics; 2004. Available at http://www.cdc.gov/nchs/data/
statnt/statnt25.pdf.

19. US Department of Health and Human Services. Healthy people 2010.
Final review. Appendix A: technical appendix. Washington, DC: US
Department of Health and Human Services; 2012:A-9–10. Available at
http://www.cdc.gov/nchs/data/hpdata2010/hp2010_final_review_
technical_appendix.pdf.

20. Anderson RN, Rosenberg HM. Age standardization of death rates:
implementation of the year 2000 standard. Natl Vital Stat Rep
1998;47:1–16, 20.

21. Ryan CL, Siebens J. Educational attainment in the United States: 2009.
Washington, DC: US Census Bureau; 2012. Current Population Reports
Series P20, No. 566.

22. Everett BG, Rogers RG, Hummer RA, Krueger PM. Trends in
educational attainment by race/ethnicity, nativity, and sex in the United
States, 1989–2005. Ethn Racial Stud 2011;34:1543–66.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 19

23. Ohlshansky SJ, Antonucci T, Berkman L, et al. Differences in life
expectancy due to race and educational differences are widening, and
may not catch up. Health Aff 2012;31:1803–13.

24. Braveman PA, Cubbin C, Egerter S, Williams DR, Pamuk E.
Socioeconomic disparities in health in the United States: what the
patterns tell us. Am J Public Health 2010;100(Suppl 1):S186–96.
DOI:10.2105/AJPH.2009.166082.

25. Harper S, Lynch J. Trends in socioeconomic inequalities in adult health
behaviors among U.S. states, 1990–2004. Public Health Rep 2007;
122:177–89.

26. Koh HK, Graham G, Giled SA. Reducing racial and ethnic disparities:
the action plan from the Department of Health and Human Services.
Health Aff 2011;30:1822–9.

27. Marmot MC, Bell R. Action on health disparities in the United States.
JAMA 2009;301:1169–71.

28. Woolf SH, Johnson RE, Phillips RL, Philipsen M. Giving everyone the
health of the educated: an examination of whether social change would
save more lives than medical advances. Am J Public Health
2007;97:679–83.

29. Dynarski M, Clarke L, Cobb B, Finn J, Rumberger R, Smink J. Dropout
prevention: a practice guide (NCEE 2008–4025). Washington, DC:
National Center for Education Evaluation and Regional Assistance,
Institute of Education Sciences, U.S. Department of Education; 2008.
Available at http://ies.ed.gov/ncee/wwc.

30. Task Force on Community Preventive Services. Recommendations to promote
healthy social environments. Am J Prev Med 2003;24(3S):S21–4.

31. CDC. Fact sheet: health disparities in education and income. Atlanta,
GA: US Department of Health and Human Services, CDC; 2011.
Available at http://www.cdc.gov/minorityhealth/reports/CHDIR11/
FactSheets/Educationincome.pdf.

Supplement

20 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
According to the Dietary Guidelines for Americans, persons

in the United States aged ≥2 years should increase their intake
of certain nutrient-rich foods, including fruits and vegetables
(1). Fruits and vegetables contribute important nutrients that
are underconsumed in the United States (1). Higher intake of
fruits and vegetables might reduce the risk for many chronic
diseases including heart disease (2), stroke (3), diabetes (4), and
some types of cancer (5). In addition, replacing high-calorie
foods with fruits and vegetables can aid in weight management
(1,6,7). However, most persons in the United States do not
consume the recommended amounts of fruits and vegetables
and other healthier food groups (e.g., whole grains or fat-free
or low-fat dairy foods) (1,8).

Persons who live in neighborhoods with better access to
retailers such as supermarkets and large grocery stores that
typically offer fruits and vegetables and other healthy foods
might have healthier diets (9,10). However, in 2009, the
U.S. Department of Agriculture estimated that 40% of all
U.S. households do not have easy access (i.e., access within
1 mile of residence) to supermarkets and large grocery stores
(11). Although few national studies examining disparities in
access exist (11–13), research suggests that access is often lower
among residents of rural, lower-income, and predominantly
minority communities than among residents of other
communities (9,12). Because of positive associations between
the retail environment and diet (9,10), a Healthy People 2020
developmental objective (14) is to increase the percentage of
persons in the United States who have access to a retailer that
sells the various foods recommended in the Dietary Guidelines
for Americans, including fruits and vegetables, whole-grain
foods, and low-fat milk, which are referred to as healthier foods
in this report. Improving access to healthier food retailers has
also been adopted as a promising strategy to improve dietary
quality by philanthropic and governmental entities (11,15,16).

Access to healthier foods includes not only proximity to
retail locations that offer these types of foods but also the
variety, cost, and quality of foods (17). However, in this report
and in most other studies, access refers to the proximity of
food retailers because of the inherent challenges and resource
needs in measuring variety, cost and quality of food. Access

to supermarkets, supercenters, and large grocery stores is
frequently measured because these types of stores tend to offer
a wider selection and larger quantity of fruits and vegetables
and other healthy foods at affordable prices than other retailers,
such as convenience stores and small grocery stores (18).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (19) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (20). This report provides information
concerning disparities in access to healthier food retailers, a
topic that was not discussed in the 2011 CHDIR (19). The
purposes of this report on access to healthier food retailers are to
discuss and raise awareness of differences in the characteristics
of areas with access to healthier food retailers across census
tracts and to prompt actions to reduce disparities.

Methods
To estimate access to healthier food retailers across the United

States and regionally (i.e., places persons live and might shop),
CDC analyzed 2011 data from various sources using census
tracts as the unit of analysis. In this report, the term access refers
to potential access to healthier food retailers, which is where
consumers can shop, rather than actual access, which is where
consumers actually do shop. Access to healthier food retailers
by area demographics of the census tracts also was compared.
Access to a retailer was estimated by calculating the percentage
of census tracts that did not have at least one healthier food
retailer located within the tract or within ½ mile of the tract
boundary (21). Census tracts are small, relatively permanent
subdivisions of counties designed to be similar in population
characteristics, economic status, and living conditions. The
median tract area size and population was 1.9 square miles
and 4,022 people.

A list of 54,666 healthier food retailers was developed from
two national directories of retail food stores. One directory
was purchased in June 2011 from the commercial data
provider InfoUSA (available at http://www.infousa.com).

Access to Healthier Food Retailers — United States, 2011
Kirsten A. Grimm, MPH
Latetia V. Moore, PhD
Kelley S. Scanlon, PhD

National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Kirsten A. Grimm, National Center for Chronic Disease Prevention and Health Promotion, CDC. Telephone: 770-488-5041;
E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 21

The other directory was from a list of authorized stores
that accept Supplemental Nutrition Assistance Program
(SNAP) benefits as of January 2012 (available at http://www.
snapretailerlocator.com). Two independent data sources were
used to reduce inaccuracies in store operational status and store
misclassification (22–28). Evidence suggests that secondary
data might only capture 55%–68% of food outlets that truly
exist in an area (24,26,27), and store misclassification is
common (24).

Healthier food retailers are defined as supermarkets, large
grocery stores, supercenters and warehouse clubs, and fruit and
vegetable specialty stores (21). These retailers were identified
from the InfoUSA directory by using several criteria, including
2007 North American Industry Classification System (NAICS)
codes (available at http://www.census.gov/eos/www/naics/),
annual sales volume, annual employees on payroll, and chain
store name lists. Large grocery stores and supermarkets were
defined as retailers with the appropriate NAICS code (NAICS
445110: grocery stores/supermarkets) with either ≥10 annual
payroll employees or ≥$2 million in annual sales or whose
company name matches a chain name list (21). This list of 228
national and regional supermarket, supercenter, and warehouse
club chain stores was developed from 2000 and 2005 data from
the commercial data provider Nielsen TDLinx (29) and 2011
InfoUSA data and includes stores that have at least eight to
10 locations nationwide and were verified as having a full line
of groceries. Supercenters and warehouse clubs were defined
as retailers with the appropriate NAICS codes (NAICS 445,
452112, 452910: supercenters and warehouse clubs) or included
if their company name matched the national chain name list.
Fruit and vegetable specialty food stores were defined as retailers
with the appropriate NAICS codes (NAICS 445230: fruit and
vegetable specialty food stores).

The second directory of stores included retailers who had
actively processed SNAP benefits as recently as January 3, 2012,
and had store classifications through the SNAP application
process consistent with the definition of healthier food retailers
as described in this report (30). The healthier food retailers
included from SNAP were those categorized as supermarkets,
supercenters/warehouse clubs, large grocery stores, or fruit and
vegetable specialty stores (30).

To estimate national and regional percentages of census
tracts that had at least one healthier food retailer, stores from
the two directories were assigned to one or more tracts if they
were located within the tract’s boundaries or within ½ mile of
the boundary using geocodes provided by InfoUSA or SNAP
and ArcGIS 10 (available at http://www.esri.com/software/
arcgis/index.html). Boundaries for the 72,531 census tracts in
the 50 U.S. states and the District of Columbia (DC) with a
population of >0 were obtained from 2010 U.S. census TIGER/

Line shapefiles (available at http://www.census.gov/geo/maps-
data/data/tiger-line.html). Sixty-three percent (n = 22,359) of
the healthier food retailers identified in InfoUSA were also in
SNAP. Name, address, location, and store classification type
matched in these two sources for this subset of stores (referred
to as verified retailers). The remaining 32,307 stores appeared
only in one data source (7,549 InfoUSA stores and 19,418
SNAP stores) or appeared in both but store classification types
were inconsistent (n = 5,340). Previous evidence indicates that
if a store is open, the probability that a secondary data source
lists it as operational ranges from 55% to 89% (24,27,31,32).
The use of secondary data to accurately classify store type (e.g.,
grocery store, supermarket, or supercenter) has been estimated
to be 49%–85% (24). One study estimates that if a store is in
the InfoUSA list, the likelihood that the store is operational
and correctly classified as a supermarket, grocery store, or
specialty store is 34.4%–44.5% (32). Because the operational
status, store presence, and store type of the retailers that only
appeared in one directory could not be verified by a second
data source, tracts that only contained two or more of these
stores were counted as having a healthier food retailer. If a
tract has two or more unverified stores, evidence indicates that
it is reasonable to assume that at least one is operational and
appropriately classified (24,27). Nine percent of tracts (n =
6,563) were counted as having a healthier food retailer because
two or more unverified stores were present. Twelve percent of
tracts (n = 8,343) had only one unverified store from either
source and therefore were counted as not having any verifiable
healthier food retailers. Nineteen percent of tracts did not have
stores from either directory present (n = 13,761 tracts).

To estimate percentages of access to healthier food retailers by
area demographics, CDC obtained demographic information
on educational attainment and per capita income at the
census tract level from the 2006–2010 American Community
Survey. Information on age and race/ethnicity were obtained
from the 2010 U.S. census. Tracts were categorized into two
groups (low and high) for each demographic characteristic by
dichotomizing at the mean of the distribution. A census tract
was considered urban if the geographic centroid of that tract
was located in an area designated by the 2010 U.S. census as
an urbanized area or urban cluster (available at http://www.
census.gov/geo/www/ua/2010urbanruralclass.html). All other
tracts were classified as rural. Median tract size and population
density for urban tracts was 1 square mile and 3,852 persons
per square mile versus 42 square miles and 100 persons per
square mile in rural tracts.

Comparisons of percentages by demographics among national
and U.S. Census regions (available at http://www.census.gov/
geo/maps-data/maps/pdfs/reference/us_regdiv.pdf ) were
assessed using chi-square tests, with significance set at p<0.05.

Supplement

22 MMWR / November 22, 2013 / Vol. 62 / No. 3

Odds ratios and 95% confidence intervals (CIs) were estimated
using logistic regression to characterize national and region-
specific odds of not having access to a healthier food retailer
by each demographic characteristic separately. Significant
differences in access to healthier food retailers described in this
report are those in which the 95% CIs do not include 1.0; thus,
the odds of access are significantly higher or lower. Tracts that
had either no sample observations or too few sample observations
for computing demographic estimates were excluded (n = 404;
0.6%).

Disparities were measured as the deviations from a referent
category rate or prevalence. Referent groups in all analyses were
as follows: tracts with a low proportion of youths (≤23.4% of
the population aged ≤18 years), a low proportion of seniors
(≤13.6% of the population aged ≥65 years), a high per capita
income (>$27,269 per capita income adjusted to 2010
dollars), a high proportion of non-Hispanic whites (>63.9%
non-Hispanic white population), and a high proportion of
college-educated persons (>27.0% of the population with a
college degree or higher). Absolute difference was measured as
the simple difference between a population subgroup estimate
and the estimate for its respective reference group.

Results
In 2011, 30.3% of census tracts did not have at least one

healthier food retailer within the tract or within ½ mile of tract
boundaries. This represents 83.6 million persons, representing
approximately 27% of the 2010 continental U.S. population.
The percentage of census tracts without at least one healthier
food retailer ranged from 24.1% in the West to 36.6% in the
Midwest. Overall, access to healthier food retailers varied by
each of the demographic characteristics examined, although
these disparities were not always consistent by region (Tables 1,
2, and 3). Persons in rural census tracts were approximately 4
times as likely to lack access to a healthier food retailer than
persons in urban tracts. This pattern was consistent across
regions. Sensitivity analyses using national models stratified

by urban status found similar relationships only for race/
ethnicity. Other associations were mixed. For example, persons
in urban areas with a youth population of >23.4% had a higher
odds of lacking access than those in rural areas with the same
proportion of youth. Education was significantly associated
with access in rural areas but not in urban areas.

Overall, tracts where seniors comprised >13.6% of the
population were 1.3 times as likely not to have a healthier food
retailer than tracts with a lower proportion of seniors, a pattern
that was similar across regions. Nationwide, tracts with <64%
of non-Hispanic whites were about half as likely to lack access
to a healthier food retailer than tracts with a higher percentage
of non-Hispanic whites. This pattern was also similar across
regions, with up to an approximately 75% reduction in the
odds of no access among tracts in the Northeast with a low
versus high percentage of non-Hispanic whites.

Other associations were not as consistent across regions.
Nationwide, persons in tracts with an income of ≤$27,269 were
1.2 times as likely to lack access to a healthier food retailer than
tracts with higher income. This association differed by region,
with no association in the Midwest and a stronger association
in the South. However, in the Northeast and West, persons
in low-income tracts had a lower odds of lacking access to a
healthier food retailer (OR: 0.91 [95% CI: 0.85–0.98]) and
0.88 [95% CI: 0.82–0.94], respectively). Similarly, nationwide,
persons in tracts where ≤27.0% had a college education were
significantly more likely to lack access to a healthier food
retailer than persons in a tract with a higher proportion of
college-educated persons; the association was not significant
in the Northeast and West.

Nationwide, persons living in tracts where youths comprised
>23.4% of the population had slightly higher odds of lacking
access to a healthier food retailer than persons living in tracts
with low proportions of youths (OR: 1.06 [95% CI: 1.03–
1.09]). Regionally, persons living in tracts in the Midwest with
a higher proportion of youths were 1.2 times as likely to lack
access as persons in tracts with a low proportion of youths,
with no additional associations by region.

TABLE 1.  Percentage of census tracts* without at least one healthier food retailer within the tract or within ½ mile of the tract, by geographic
region† — United States, 2011

United States Northeast Midwest South West

Total no. of tracts 72,127 13,333 16,924 25,948 15,922
Tracts without at least one healthier food retailer (%) 30.3 27.3 36.6 31.6 24.1 

* N = 72,531 census tracts in the 50 U.S. states and the District of Columbia per the 2010 U.S. census. A total of 404 (0.6%) census tracts were excluded because either
no sample observations or too few sample observations were available to calculate demographic estimates.

† Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,
Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 23

Discussion
The analyses in this report reveal that persons in 30.3%

of census tracts in the U.S. do not have access to at least one
healthier food retailer. The most substantial disparities were
associated with urbanization; persons in rural tracts were four
times as likely to lack access than persons in more urban tracts.
Persons living in tracts with a high percentage of non-Hispanic
whites and those with a high percentage of seniors also had
consistently worse access across regions. Access to healthier food
retailers among youths and by income and education varied by
region. Some of the findings in this study are similar to those
of other national studies, including those that assess urban and
rural areas, whereas other findings, such as those that assess
access to food retailers according to income, are not consistent
with previous studies (11,12). However, findings related to
race/ethnicity and access vary substantially among studies.
One national study found no differential access to healthy
food retailers among racial/ethnic groups (11), whereas another
national study found a lack of access in minority neighborhoods
(12). After controlling for demographic characteristics, one
study found fewer chain supermarkets in non-Hispanic black
neighborhoods than in non-Hispanic white neighborhoods and
fewer chain supermarkets in Hispanic neighborhoods than in

non-Hispanic white neighborhoods. However, non-Hispanic
black neighborhoods were found to have more nonchain
supermarkets and grocery stores than white neighborhoods
(12). The definition of healthier food retailers in this particular
study was chain vs. nonchain supermarkets. This distinction
was used because chain supermarkets tend to have more
healthy, affordable foods than nonchain supermarkets. CDC
conducted a sensitivity analysis of the data in this report to
explore access to chain supermarkets only among tracts with
predominantly (>50%) non-Hispanic black residents compared
with predominantly non-Hispanic white residents, adjusting
for region and urbanization. This sensitivity analysis revealed
that access to chain supermarkets was lower in census tracts
with predominantly non-Hispanic black residents than in tracts
with predominantly non-Hispanic white residents, results that
are similar to those of another study (12).

Limitations
The findings in this report are subject to at least four

limitations. First, the estimates of access to food retailers reflect
potential access, which indicates retailers where consumers
are able to shop, but do not reflect actual access, which is

where consumers actually decide to shop, or
other aspects of access, such as affordability,
selection, and quality of foods within
stores or modes of transportation to stores.
Neighborhoods identified as not having at
least one healthier food retailer might still
have access to healthier foods if their local
convenience stores and corner stores provide
a wide selection and adequate quantity of
affordable produce and other items. Although
some studies have shown these types of retailers
typically do not stock healthier foods (9,18),
others have reported improved food selection
because of recent changes implemented in the
Special Supplemental Nutrition Program for
Women, Infants, and Children (WIC) that
require that healthy foods be stocked at stores
that accept vouchers (33). However, because
no systematic way exists at a national level
to identify small retailers offering healthier
foods, they are not counted as a healthier
food retailer. In addition, although residents
might have additional access to produce
in their neighborhoods through farmers
markets and farm stands, these venues are
not included in this analysis. Second, only

TABLE 2.  Percentage of census tracts* without at least one healthier food retailer within
the tract or within ½ mile of the tract, by census tract demographic characteristics —
United States, 2011

Demographic characteristics† %
Absolute difference
(percentage points) OR§ (95% CI)§

Urbanization
Rural 51.5 30.9 4.10 (3.96–4.24)
Urban¶ 20.6 Ref. — —

Youths aged ≤18 yrs (%)
High: >23.4% of population 30.9 1.2 1.06 (1.03–1.09)
Low: ≤23.4% of population 29.7 Ref. — —

Adults aged ≥65 yrs (%)
High: >13.6% of population 33.6 6.0 1.33 (1.29–1.37)
Low: ≤13.6% of population 27.6 Ref. — —

Whites, non-Hispanic (%)
Low: ≤63.9% of population 21.2 15.0 0.48 (0.46–0.49)
High: >63.9% of population 36.2 Ref. — —

Per capita income in 2010 dollars (%)
Low: ≤$27,269 31.4 2.9 1.15 (1.11–1.18)
High: >$27,269 28.5 Ref. — —

Persons with college degree (%)
Low: ≤27.0% of population 33.3 7.5 1.43 (1.38–1.48)
High: >27.0% of population 25.8 Ref. — —

Abbreviations: 95% CI = 95% confidence interval; OR = odds ratio; Ref. = referent.
* N = 72,531 census tracts in the 50 U.S. states and the District of Columbia per the 2010 U.S. census. A

total of 404 (0.6%) were excluded because either no or too few sample observations were available
to calculate demographic estimates.

† Tracts were categorized into low and high groups for each demographic characteristic by
dichotomizing at the mean of the distribution.

§ ORs and 95% CIs were estimated using logistic regression.
¶ A census tract was considered urban if the centroid of that tract was located in a 2010 U.S. census–

designated urbanized area or urban cluster. All other tracts were considered rural.

Supplement

24 MMWR / November 22, 2013 / Vol. 62 / No. 3

tracts that had at least one store that was verified by two
independent data sources (60% tracts) or at least two stores
that appeared in either directory of stores (9% of tracts) were
counted as having a healthier food retailer. Not including tracts
with only a single store listed in only one source might have
overestimated lack of access if that one store was operational
and appropriately classified. A sensitivity analysis showed
that demographic estimates using stores identified in either
source (not just those that were verified by two sources and
those where two or more unverified stores were present) were
similar to results shown in this report, with the exception of
urbanization. In general, odds ratios were attenuated, although
the direction of the associations remained unchanged. Third,
only secondary data were available for this national and regional
analysis. Secondary data sources have been show to misclassify

store type and operational status and both undercount and
overcount stores in comparison with direct field assessments
(22–28). However, the analyses in this report included two
sources of secondary data to reduce these inaccuracies. Finally,
a national and regional analysis might mask various local and
state disparities in access.

Conclusion
This report describes one of the few national studies assessing

disparities in access to healthier food retailers by demographic
characteristics nationwide and by region. Because the data cannot
fully account for the heterogeneity of the U.S. food environment,
a more in-depth evaluation is required to determine whether
interventions are needed in specific neighborhoods.

TABLE 3.  Percentage of census tracts* without at least one healthier food retailer within the tract or within ½ mile of the tract, by census tract
demographic characteristics and region†— United States, 2011

Demographic
characteristics§

No. of
tracts

Northeast Midwest South West

%

Absolute
differ ence

(percentage
points) OR¶ (95% CI)¶ %

Absolute
differ ence

(percentage
points) OR (95% CI) %

Absolute
differ ence

(percentage
points) OR (95% CI) %

Absolute
differ ence

(percentage
points) OR (95% CI)

Urbanization
Rural 11,675 52.1 32.2 4.37 (4.01–4.76) 53.4 26.4 3.10 (2.90–3.31) 50.6 30.5 4.06 (3.84–4.29) 50.3 34.1 5.26 (4.85–5.70)
Urban**,†† 10,186 19.9 Ref. — — 27.0 Ref. — — 20.1 Ref. — — 16.2 Ref. — —

Youths aged ≤18 yrs (%)
High: >23.4% of

population
11,535 26.7 0.9 0.95 (0.88–1.03) 38.4 3.9 1.18 (1.11–1.26) 31.8 0.4 1.02 (0.97–1.07) 24.1 0.1 1.00 (0.93–1.08)

Low: ≤23.4% of
population††

10,326 27.6 Ref. — — 34.5 Ref. — — 31.4 Ref. — — 24 Ref. — —

Adults aged ≥65 yrs (%)
High: <13.6% of

population
10,879 30.8 7.2 1.44 (1.33–1.56) 38.2 3.2 1.15 (1.08–1.22) 34.4 5.1 1.27 (1.20–1.34) 28.5 6.8 1.44 (1.33–1.55)

Low: ≤13.6% of
population††

10,982 23.6 Ref. — — 35..0 Ref. — — 29.3 Ref. — — 21.7 Ref. — —

Whites, non-Hispanic (%)
Low: ≤63.9% of

population
6,029 11.3 23.6 0.24 (0.22–0.26) 27.7 11.4 0.60 (0.55–0.65) 25.2 11.7 0.58 (0.55–0.61) 18.0 13.2 0.48 (0.45–0.52)

High: >63.9% of
population††

15,832 34.9 Ref. — — 39.1 Ref. — — 36.9 Ref. — — 31.2 Ref. — —

Per capita income in
2010 dollars (%)
Low: ≤$27,269 of

population
13,990 26.4 1.8 0.91 (0.85–0.98) 37.5 2.7 1.13 (1.05–1.20) 33.6 6.2 1.34 (1.27–1.42) 23.0 2.4 0.88 (0.82–0.94)

High: >$27,269 of
population††

7,871 28.2 Ref. — — 34.8 Ref. — — 27.4 Ref. — — 25.4 Ref. — —

Persons with college
degree (%)
Low: ≤27.0% of

population
14,471 27.8 1.1 1.05 (0.98–1.14) 40.0 9.9 1.55 (1.45–1.66) 35.5 11.2 1.72 (1.62–1.82) 24.5 0.9 1.05 (0.98–1.13)

High: >27.0% of
population††

7390 26.7 Ref. — — 30.1 Ref. — — 24.3 Ref. — — 23.6 Ref. — —

Abbreviations: 95% CI = 95% confidence interval; OR= odds ratio; Ref. = referent.
* N = 72,531 census tracts in the 50 U.S. states and the District of Columbia per the 2010 U.S. census. A total of 404 (0.6%) were excluded because either no or too few sample observations

were available to calculate demographic estimates.
† Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota,

Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi,
North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon,
Utah, Washington, and Wyoming.

§ Tracts were categorized into low and high groups for each demographic characteristic by dichotomizing at the mean of the distribution.
¶ ORs and 95% CIs were estimated using logistic regression.
** A census tract was considered urban if the centroid of that tract was located in a 2010 U.S. census designated urbanized area or urban cluster. All other tracts were considered rural.
†† Significant difference in percentage across regions using chi-square tests (p<0.001)

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 25

Several strategies might improve community access to
retailers that sell healthier foods. Such strategies include
incentives to bring healthier food retailers into underserved
areas, transportation improvements so that residents in
underserved areas can reach the food retailers, and upgrading
facilities to enable stocking of all forms of fruits and vegetables
and to increase shelf space dedicated to fruits and vegetables,
ultimately increasing the availability of high-quality, affordable
fruits and vegetables in existing venues (15).

An example of efforts at the national level to bring healthier
food retailers into underserved areas is a collaboration
among the U.S. Department of Agriculture (USDA), U.S.
Department of Health and Human Services (HHS), and
the U.S. Department of Treasury to support projects that
increase access to healthier, affordable food and encourage
the purchase and consumption of healthier food (available at
http://apps.ams.usda.gov/fooddeserts). The state-level pioneer
effort called the Pennsylvania Fresh Food Financing Initiative
has provided funding for 88 fresh-food retail projects in 34
Pennsylvania counties and improved access to healthier food
for approximately 500,000 persons (34). Similar efforts have
been expanding rapidly across states.

Changes in WIC-authorized stores improve access to
healthy food in existing stores. Stores authorized to accept
WIC benefits must maintain on their shelves at all times a
minimum variety of healthy foods, including fruits, vegetables,
and whole grains that align with the 2005 Dietary Guidelines
for Americans and the American Academy of Pediatrics infant
feeding practice guidelines (35). Studies have demonstrated
that WIC-authorized stores are providing more healthy
foods than stores that are not WIC authorized (33,35,36).
Additional ways to bring healthier foods to persons living in
underserved areas without changing existing retailers include
establishing farmers markets, farm stands, and green carts
(15). For example, in New York, the New York City Green
Cart Initiative provides fruits and vegetables to underserved
neighborhoods (information available at http://www.nyc.gov/
html/doh/html/diseases/green-carts.shtml), and the Veggie
Mobile delivers fruits and vegetables to low-income seniors in
upstate New York (information available at http://www.cdcg.
org/programs/veggie/veggie). Fruits and vegetables also can
be delivered through drop-off boxes to churches, community
centers, and other central locations (15).

Although the precise number of healthy food retailers that
need to be in a particular area to allow adequate access to fruits
and vegetables and other healthy foods is not known, ensuring
that all persons in the United States have access to at least one
retail venue that offers healthier foods is an important step
toward supporting healthy choices and diets in communities.
Improving access to healthy food retailers is important but

unlikely to be sufficient to improve overall diet quality. Even in
communities that have sufficient access, strategies such as store
promotions and shelf labeling that help consumers identify
healthy options, education on health benefits of particular
foods, and information about preparation, storage, and cooking
skills can encourage persons to purchase healthy foods in retail
venues and might improve diet quality. The combined efforts
of interventions that improve knowledge and skills, as well
as increase the affordability, selection, and quality of foods in
many settings are needed to encourage healthier choices among
persons in the United States.

References
1. US Department of Agriculture, US Department of Health and Human

Services. Dietary guidelines for Americans, 2010. Washington, DC: US
Government Printing Office; 2010.

2. He FJ, Nowson CA, Lucas M, MacGregor GA. Increased consumption
of fruit and vegetables is related to a reduced risk of coronary heart
disease: meta-analysis of cohort studies. J Hum Hypertens 2007;
21:717–28.

3. He FJ, Nowson CA, MacGregor GA. Fruit and vegetable consumption
and stroke: meta-analysis of cohort studies. Lancet 2006;367:320–6.

4. Montonen J, Knekt P, Jarvinen R, Reunanen A. Dietary antioxidant
intake and risk of type 2 diabetes. Diabetes Care 2004;27:362–6.

5. World Cancer Research Fund. Food, nutrition, physical activity, and
the prevention of cancer: a global perspective. Washington DC: American
Institute for Cancer Research; 2007.

6. Rolls BJ, Ello-Martin JA, Tohill BC. What can intervention studies tell
us about the relationship between fruit and vegetable consumption and
weight management? Nutr Rev 2004;62:1–17.

7. Tohill BC, Seymour J, Serdula M, Kettel-Khan L, Rolls BJ. What
epidemiologic studies tell us about the relationship between fruit and
vegetable consumption and body weight. Nutr Rev 2004;62:365–74 .

8. Krebs-Smith SM, Guenther PM, Subar AF, Kirkpatrick SI, Dodd KW.
Americans do not meet federal dietary recommendations. J Nutr
2010;140:1832–8.

9. Larson NI, Story MT, Nelson MC. Neighborhood environments:
disparities in access to healthy foods. Am J Prev Med 2009;36:74–81.

10. Moore LV, Diez Roux AV, Nettleton JA, Jacobs DR Jr. Associations of
the local food environment with diet quality—a comparison of
assessments based on surveys and geographic information systems: the
multi-ethnic study of atherosclerosis. Am J Epidemiol 2008;
167:917–24.

11. US Department of Agriculture. Access to affordable and nutritious food:
measuring and understanding food deserts and their consequences.
Washington DC: US Department of Agriculture Economic Research
Service; 2009. Available at http://www.ers.usda.gov/media/242675/
ap036_1_.pdf.

12. Powell LM, Slater S, Mirtcheva D, Bao Y, Chaloupka FJ. Food store
availability and neighborhood characteristics in the United States. Prev
Med 2007;44:189–95.

13. CDC. Children’s food environment state indicator report, 2011. Atlanta,
GA: US Department of Health and Human Services, CDC; 2011. Available
at http://www.cdc.gov/obesity/downloads/childrensfoodenvironment.pdf.

14. US Department of Health and Human Services. Healthy people 2020
topics and objectives. Washington, DC: US Department of Health and
Human Services; 2011. http://www.healthypeople.gov.

15. CDC. Strategies to prevent obesity and other chronic diseases: the CDC
guide to strategies to increase the consumption of fruits and vegetables.
Atlanta, GA: US Department of Health and Human Services; 2011.

Supplement

26 MMWR / November 22, 2013 / Vol. 62 / No. 3

16. Institute of Medicine. Accelerating progress in obesity prevention: solving
the weight of the nation. Washington, DC: The National Academies
Press; 2012.

17. CDC. Healthier food retail: beginning the assessment process in your
state or community. Atlanta, GA: US Department of Health and Human
Services, CDC; 2011. Available at http://www.cdc.gov/obesity/
downloads/hfrassessment.pdf.

18. Franco M, Diez Roux AV, Glass TA, Caballero B, Brancati FL.
Neighborhood characteristics and availability of healthy foods in
Baltimore. Am J Prev Med 2008;35:561–7.

19. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

20. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

21. CDC. State indicator report on fruits and vegetables. Atlanta, GA:
CDC; 2009.

22. Bader MD, Ailshire JA, Morenoff JD, House JS. Measurement of the
local food environment: a comparison of existing data sources. Am J
Epidemiol 2010;171:609–17.

23. Cummins S, Macintyre S. Are secondary data sources on the
neighbourhood food environment accurate? Case-study in Glasgow,
UK. Prev Med 2009;49:527–8.

24. Han E, Powell LM, Zenk SN, Rimkus L, Ohri-Vachaspati P, Chaloupka
FJ. Classification bias in commercial business lists for retail food stores
in the U.S. Int J Behav Nutr Phys Act 2012;9:46.

25. Hoehner CM, Schootman M. Concordance of commercial data sources
for neighborhood—effects studies. J Urban Health 2010;87:713–25.

26. Lake AA, Burgoine T, Greenhalgh F, Stamp E, Tyrrell R. The foodscape:
classification and field validation of secondary data sources. Health Place
2010;16:666–73.

27. Liese AD, Colabianchi N, Lamichhane AP, et al. Validation of 3 food
outlet databases: completeness and geospatial accuracy in rural and urban
food environments. Am J Epidemiol 2010;172:1324–33.

28. Paquet C, Daniel M, Kestens Y, Leger K, Gauvin L. Field validation of
listings of food stores and commercial physical activity establishments
from secondary data. Int J Behav Nutr Phys Act 2008;5:58.

29. Auchincloss AH, Moore K, Moore L, Diez Roux AV. Improving
characterization of the food environment for a large region in the United
States during a historic time period. Health Place 2012;18:1341–7.

30. US Department of Agriculture. Benefit redemption patterns in the
Supplemental Nutrition Assistance Program. Washington, DC: US
Department of Agriculture; 2011.

31. Fleischhacker SE, Rodriguez DA, Evenson KR, et al. Evidence for validity
of five secondary data sources for enumerating retail food outlets in seven
American Indian communities in North Carolina. Int J Behav Nutr
Phys Act 2012;9:137.

32. Powell LM, Han E, Zenk SN, et al. Field validation of secondary
commercial data sources on the retail food outlet environment in the
U.S. Health Place. 2011;17:1122–31.

33. Gleason S, Morgan R, Bell L. Impact of the revised WIC food package
on small WIC vendors: insight from a four-state evaluation: Altarum
Institute: Systems Research for Better Health; 2011.

34. Giang T, Karpyn A, Laurison HB, Hillier A, Perry RD. Closing the
grocery gap in underserved communities: the creation of the Pennsylvania
Fresh Food Financing Initiative. J Public Health Manag Pract
2008;14:272–9.

35. US Department of Agriculture, Food and Nutrition Service. 7 CFR Part
246. Special supplemental nutrition program for women, infants, and
children (WIC): revisions in the WIC food packages; interim rule. Fed
Regist 2007;72:68966–9032.

36. Tester JM, Yen IH, Pallis LC, Laraia BA. Healthy food availability and
participation in WIC (Special Supplemental Nutrition Program for
Women, Infants, and Children) in food stores around lower- and higher-
income elementary schools. Public Health Nutr 2011;14:960–4.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 27

Introduction
The association between unemployment and poor physical

and mental health is well established (1–7). Unemployed
persons tend to have higher annual illness rates, lack health
insurance and access to health care, and have an increased risk
for death (1,2,8,9). Several studies indicate that employment
status influences a person’s health; however, poor health also
affects a person’s ability to obtain and retain employment (10).
Poor health predisposes persons to a more uncertain position in
the labor market and increases the risk for unemployment (5,6).

According to the Bureau of Labor Statistics (BLS), the
unemployment prevalence in the United States increased from
4.7% in 2006 to 9.4% in 2010, yielding an estimated 14.5 million
unemployed persons (11). Both the prevalence of unemployment
and the health status of populations vary widely among and
within communities by age, sex, and race/ethnicity. In 2010, the
unemployment prevalence both for males and females was twice
as high in the black and Hispanic populations as in the white
population (11). The disparities in unemployment prevalence
extend across the country and have increased from January 2008
to December 2010 (12). Because unemployment has historically
been substantially higher in black and Hispanic populations
during past decades and because unemployment has increased
substantially from the start of the recession in December 2007
(13,14), associations between unemployment and health and
between unemployment and minority status need to be further
studied.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (15) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (16). This report is the first assessment of
unemployment and health status in a CHDIR. The purposes of
this unemployment and health analysis are to discuss and raise
awareness of differences in the characteristics of persons who
are unemployed and differences in health status by employment
status and to prompt actions to reduce these disparities.

Methods
To assess changes in unemployment rates by several

population characteristics, CDC analyzed 2006 and 2010 data
from the Behavioral Risk Factor Surveillance System (BRFSS).
The association between unemployment and self-reported
health status, physical health, and mental health in 2010
also was examined. The 2010 state-specific unemployment
prevalences were calculated and shown on a U.S. map using
statistical software; prevalences were shown for men and
women, non-Hispanic blacks, and non-Hispanic whites (17).
All analyses were limited to persons aged 18–64 years.

BRFSS is a state population-based, telephone survey of
noninstitutionalized U.S. adults aged ≥18 years collected in
all states and selected territories. The BRFSS median response
rate* for 2006 was 51.4% and for 2010 was 54.6%; the median
cooperation rate† for 2006 was 74.5% and for 2010 was 76.9%
(18,19). The same question from the BRFSS survey was used
to assess employment status and unemployment status by
asking participants whether they are currently 1) employed for
wages; 2) self-employed; 3) out of work for >1 year; 4) out of
work for <1 year; 5) a homemaker; 6) a student; 7) retired; or
8) unable to work. Persons who did not respond to this question
were excluded from the analysis. The employment question
responses were recategorized into the following groups: 1)
employed (including employed for wages and self-employed),
2) unemployed (out of work for <1 year and out of work for
>1 year), and 3) other (homemaker, student, retired, or unable
to work).

Data were analyzed to assess disparities in unemployment
prevalence for 2006 and 2010. To examine the association
between unemployment and health status, data for the
following three health outcomes were collected from the 2010
BRFSS data set: 1) health status, 2) number of physically
unhealthy days, and 3) number of mentally unhealthy days.

Unemployment — United States, 2006 and 2010
Heba M. Athar, MD

Man-Huei Chang, MPH
Robert A. Hahn, PhD

Eric Walker, MPH
Paula Yoon, ScD

Center for Surveillance, Epidemiology, and Laboratory Services, CDC

Corresponding author: Heba M. Athar, Division of Epidemiology, Analysis and Library Services, Center for Surveillance, Epidemiology, and Laboratory
Services, CDC. Telephone: 404-498-2216; E-mail: [email protected].

* The perentage who completed interviews among all eligible persons, including
those who were not contacted successfully.

† The perentage who completed interviews among all eligible persons who were
contacted.

Supplement

28 MMWR / November 22, 2013 / Vol. 62 / No. 3

The related BRFSS questions were as follows: 1) “Would you
say that in general your health is excellent, very good, good,
fair, or poor?” 2) “Thinking about your physical health, which
includes physical illness and injury, for how many days during
the past 30 days was your physical health not good?” and 3)
“Thinking about your mental health, which includes stress,
depression, and problems with emotions, for how many days
during the past 30 days was your mental health not good?” In
this analysis, responses to the BRFSS health status question
were recategorized into the following groups: excellent/very
good, good, and fair/poor. Participants with the responses
“do not know/not sure” or “refused to respond” to both the
physical health and mental health questions were categorized
as having missing values and were excluded from the analysis.
Physically unhealthy days and mentally unhealthy days were
categorized separately as 0 days, 1–15 days, and 16–30 days.

Disparities were measured as the deviations from a referent
group, which was the group that had the most favorable
estimate for the variables used to assess disparities during
the time reported. Absolute difference was calculated by
subtracting the unemployment prevalence for the group of
interest from the referent group. The relative difference, a
percentage, was calculated by dividing the absolute difference
by the value in the referent category and multiplying by 100.
All state and national estimates were weighted by BRFSS
sample weights using statistical software to account for the
complex design. For unemployment prevalence and health
status prevalence, 95% confidence intervals (CIs) were
calculated for the point estimates. CIs were used as measure of
variability, and nonoverlapping CIs were considered statistically
different. Using CIs in this way is a conservative evaluation
of significance differences; infrequently, this might lead to a
conclusion that estimates are similar when the point estimates
do differ. All reported differences in this report are significant
based on the CI comparison.

Results
Unemployment prevalence increased from 2006 to 2010 for

all adults aged 18–64 years, particularly among adults aged
25–44 years (Table 1). In general, unemployment prevalence
increased among both males and females (referent group);
however, males reported higher unemployment prevalence than
females in both 2006 and 2010, and this difference gradually
increased to 2010. The highest unemployment prevalence
among racial/ethnic groups was among non-Hispanic blacks
(10.4% in 2006 and 16.5% in 2010), which was almost
twofold that of non-Hispanic whites (4.7% in 2006 and

8.3% in 2010). The unemployment prevalence for American
Indians/Alaska Natives increased substantially from 8.8% in
2006 to 15.8% in 2010. In both years, the unemployment
prevalence among persons with no health insurance was
approximately 4 times higher than that for persons with health
insurance. The unemployment prevalence decreased as levels of
education and income increased in both 2006 and 2010. The
greatest change in unemployment prevalence for education and
income from 2006 to 2010 occurred among those who did not
graduate from high school and in households with an annual
income of <$25,000 per year. In 2006 and 2010, persons with
a disability had an unemployment prevalence (60%) that was
higher than that of persons without a disability (40%).

In 2010, the highest prevalence of unemployment among
men was in the Northeast and West (Figure 1) and among
women was in the South and West (Figure 2). The Midwest
region had the lowest unemployment prevalence for both sexes.
The West region (e.g., Nevada, California, and Oregon) had
the highest prevalence of unemployment both among non-
Hispanic blacks and non-Hispanic whites (Figures 3, and 4),
and the Midwest region (e.g., North Dakota and South
Dakota) had the lowest prevalence of unemployment for both
these groups.

In 2010, unemployed persons were less likely than employed
persons to report their health as excellent or very good
(Table 2). A higher percentage of employed persons reported
that they were in excellent or very good health (62.7%) than
did persons who were unemployed for <1 year (49.2%) or
unemployed for >1 year (39.7%). Persons who were employed
were more likely to report no physically unhealthy days
(70.3%) and no mentally unhealthy days (67.3%) in the past
30 days than were persons who were unemployed for <1 year
(no physically unhealthy days: 63.1%; no mentally unhealthy
days: 54.2%). Persons who were unemployed for >1 year were
even less likely to report having had no physically or mentally
unhealthy days in the past 30 days.

Discussion
BRFSS defines unemployment differently from BLS, the

agency that monitors unemployment in the United States. BLS
defines an unemployed person as someone who does not have a
job, has been actively looking for work in the past 4 weeks, and
is currently available for work (12,20). This might contribute
to the slight difference in unemployment estimates between
BLS and BRFSS. BRFSS was selected for this analysis because
the data set includes variables of interest that enable health
status assessment and report on the disparities by employment.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 29

The analysis in this report found an association between
unemployment and overall health status and between
unemployment and number of physically and mentally
unhealthy days and also found that disparities by employment
increased from 2006 to 2010 for certain population groups,
including persons with less than a high school education,
household income <$25,000, and no health insurance
coverage, as well as for American Indians/Alaska Natives. The
disparities in unemployment among demographic groups
reported in this study are consistent with findings from several
other national surveys (11,14,21,22).

Limitations
The findings in this report are subject to at least six limitations.

First, the 2006 and 2010 surveys excluded certain populations,
such as persons without landlines, persons in institutions, and
homeless persons; therefore, the results of this study might not
be generalizable to the entire U.S. adult population. Second,
the low BRFSS median state response rates for 2006 and
2010 (18,19) increase the possibility of nonresponse bias in
the results. Third, BRFSS health status data are self-reported
and are therefore subject to recall bias and measurement error
(23,24). Fourth, BRFSS is cross-sectional, and the timeframe

TABLE 1. Unemployment* prevalence among adults aged 18–64 years, by selected demographic characteristics — Behavioral Risk Factor
Surveillance System, United States, 2006 and 2010

Characteristic

2006 2010

% (95% CI)
Absolute

difference
Relative

difference (%) % (95% CI)
Absolute

difference
Relative

difference (%)

Age group (yrs)
18–24 10.5 (9.6–11.6) 6.5 162.5 14.4 (13.5–15.5) 6.7 87.0
25–34 5.8 (5.3–6.3) 1.8 45.0 11.4 (10.8–12.1) 3.7 48.1
35–44 4.7 (4.4–5.1) 0.7 17.5 9.5 (9.0–10.0) 1.8 23.4
45–54 5.2 (4.8–5.5) 1.2 30.0 9.2 (8.8–9.5) 1.5 19.5
55–64 4.0 (3.7–4.3) Ref. Ref. 7.7 (7.4–8.0) Ref. Ref.

Sex†
Male 5.9 (5.5–6.2) 0.1 1.7 11.3 (10.9–11.6) 2.5 28.4
Female 5.8 (5.5–6.1) Ref. Ref. 8.8 (8.6–9.1) Ref. Ref.

Race/Ethnicity§
White, non-Hispanic 4.7 (4.5–4.9) Ref. Ref. 8.3 (8.1–8.6) Ref. Ref.
Black, non-Hispanic 10.4 (9.5–11.3) 5.7 121.3 16.5 (15.6–17.4) 8.2 98.8
Hispanic 7.2 (6.4–8.1) 2.5 53.2 12.4 (11.7–13.3) 4.1 49.4
Asian American/Pacific Islander 6.2 (4.8–8.1) 1.5 31.9 9.7 (8.5–11.1) 1.4 16.9
American Indian/Alaska Native 8.8 (6.9–11.2) 4.1 87.2 15.8 (13.2–18.9) 7.5 90.4

Health insurance¶
Yes 4.0 (3.8–4.2) Ref. Ref. 6.7 (6.5–6.9) Ref. Ref.
No 13.8 (13.0–14.7) 9.8 245.0 25.1 (24.3–26.0) 18.4 274.6

Educational attainment
Some high school 9.7 (8.8–10.6) 6.4 193.9 16.9 (15.9–18.0) 11.0 186.4
High school graduate or equivalent 8.0 (7.5–8.5) 4.7 142.4 13.7 (13.2–14.3) 7.8 132.2
Some college 5.2 (4.8–5.7) 1.9 57.6 9.8 (9.4–10.2) 3.9 66.1
College graduate 3.3 (3.0–3.6) Ref. Ref. 5.9 (5.6–6.2) Ref. Ref.

Income**
<$25,000 12.7 (12.0–13.4) 10.9 605.6 21.6 (20.9–22.3) 18.3 554.5
$25,000–$50,000 5.0 (4.6–5.5) 3.2 177.8 9.8 (9.3–10.3) 6.5 197.0
$50,000–$75,000 2.8 (2.4–3.2) 1.0 55.6 5.7 (5.3–6.2) 2.4 72.7
>$75,000 1.8 (1.6–2.1) Ref. Ref. 3.3 (3.0–3.5) Ref. Ref.

Disability††
Yes 8.4 (7.9–9.0) 3.2 61.5 12.8 (12.3–13.3) 3.5 37.6
No 5.2 (5.0–5.5) Ref. Ref. 9.3 (9.1–9.6) Ref. Ref.

Abbreviations: 95% CI = 95% confidence interval.
* Includes persons unemployed for <1 year and unemployed for >1 year.
† Sex is assessed by the interviewer, and the question is only asked if necessary
§ Race and ethnicity are two separate questions, and the data from both was merged. Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Health insurance includes health insurance plans, prepaid plans such as health maintenance organizations, or government plans such as Medicare or Indian Health Services.
** Income is annual household income from all sources at intervals of $25,000.
†† Respondents were asked if they were limited in any activities because of physical, mental, or emotional problems and if they had any health problems that require

use of special equipment, such as a cane, a wheelchair, a special bed, or a special telephone. If they answered “no” to one question and either said “don’t know” or
didn’t repond to the other question, their disability status was coded as missing.

Supplement

30 MMWR / November 22, 2013 / Vol. 62 / No. 3

used to define unemployment is limited to two categories
(<1 year and >1 year); therefore, the directionality of the
relations between short-term unemployment, longer-term
unemployment, and health outcomes cannot be assessed. Fifth,
the categorization of the BRFSS health status, physical health,
and mental health responses is subjective. Finally, this analysis
could neither address the reason that unemployment and health
status are related nor determine whether these pathways altered
substantially between 2006 and 2010.

Conclusion
This study supports existing findings on unemployment and

health status (1–7). The relation between unemployment and
health status is multifactorial and complex. Studies of health
and unemployment disparities have identified associated
intermediary factors, including health insurance coverage
and access to health care (14,25–27). Similarly, the National
Prevention Strategy notes that health disparities are often

FIGURE 1. Unemployment prevalence among men aged 18–64
years, by state — Behavioral Risk Factor Surveillance System, United
States, 2010

12.2%–15.1%

8.1%–10.3%
10.3%–12.2%

5.9%–8.1%

FIGURE 2. Unemployment prevalence among women aged 18–64
years, by state — Behavioral Risk Factor Surveillance System, United
States, 2010

8.1%–9.9%
6.1%–8.1%

9.9%–12.6%

4.1%–6.1%

FIGURE 3. Unemployment prevalence among non-Hispanic white
adults aged 18–64 years, by state — Behavioral Risk Factor
Surveillance System, United States, 2010

7.5%–9.0%
5.7%–7.5%

9.0%–11.5%

3.7%–5.7%

FIGURE 4. Unemployment prevalence among non-Hispanic black
adults aged 18–64 years, by state — Behavioral Risk Factor
Surveillance System, United States, 2010

18.3%–23.4%
13.5%–18.3%

23.4%–38.1%

0.0%–13.5%
Insu�cient data

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 31

linked to social, economic, and environmental disadvantages
(28). The federal government is implementing the Affordable
Care Act and, beginning in 2014, the law will expand access to
health insurance coverage for millions of previously uninsured
persons in the United States (29).

References
1. Rosenthal L, Caroll-Scott A, Earnshaw VA, et al. The importance of

full-time work for urban adults’ mental and physical health. Soc Sci Med
2012;75:1692–6.

2. Dooley D, Fielding J, Levi L. Health and unemployment. Annu Rev
Public Health 1996;17:449–65.

3. Leach LS, Butterworth P, Strazdins L, et al. The limitations of
employment as a tool for social inclusion. BMC Public Health
2010;10:621.

4. Harris E, Harris MF. Reducing the impact of unemployment on health:
revisiting the agenda for primary health care. Med J Aust 2009;
191:119–22.

5. Jefferis BJ, Nazareth I, Marstone L, et al. Associations between
unemployment and major depressive disorder: evidence form an
international, prospective study (the predict cohort). Soc Sci Med
2011;73:1627–34.

6. Tefft N. Insights on unemployment, unemployment insurance, and
mental health. J Health Econ 2011;30:258–64.

7. Kroll LE, Lampert T. Unemployment, social support and health
problems—results of the GEDA study in Germany, 2009. Dtsch Arztebl
Int 2011;108:47–52.

8. Freyer-Adam J, Gaertner B, Tobschall S, et al. Health risk factors and
self-rated health among job-seekers. BMC Public Health 2011;11:659.

9. Idler EL, Benyamini Y. Self-rated health and mortality; A review of
twenty seven studies community studies. J Health Soc Behav
1997;38:21–37.

10. Shavers VL. Measurement of socioeconomic status in health disparities
research. J Natl Med Assoc 2007;99:1013–23.

11. US Department of Labor, Bureau of Labor Statistics. Unemployment
rates by race and ethnicity, 2010. Washington, DC: Bureau of Labor
Statistics, US Department of Labor; 2011. Available at http://www.bls.
gov/opub/ted/2011/ted_20111005.htm.

12. US Department of Labor, Bureau of Labor Statistics. Unemployment
in December 2010. Washington, DC: Bureau of Labor Statistics, US
Department of Labor; 2011. Available at http://www.bls.gov/opub/
ted/2011/ted_20110111.htm.

13. Austin A. Unequal unemployment—racial disparities by state will worsen
in 2010. Economic Policy Institute 2009;257.

14. Austin A. Uneven pain: unemployment by metropolitan area and race.
Economic Policy Institute 2010;278.

15. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

16. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

17. ArcGIS software version 10.1. ESRI. Available at http://www.esri.com.
18. 2006 Behavioral Risk Factor Surveillance System summary data quality

report, 2007. Division of Behavioral Surveillance, Public Health
Surveillance and Informatics Program. Available at ftp://ftp.cdc.gov/
pub/Data/Brfss/2006SummaryDataQualityReport.pdf.

19. Behavioral Risk Factor Surveillance System 2010 Survey Data and
Documentation Division of Behavioral Surveillance, Public Health
Surveillance and Informatics Program. Available at http://www.cdc.gov/
brfss/annual_data/annual_2010.htm.

20. US Department of Labor, Bureau of Labor Statistics Labor Force.
Statistics from Current Population Survey. Washington, DC: Bureau of
Labor Statistics, US Department of Labor; 2012. Available at http://
www.bls.gov/cps.

21. Deaton A. The financial crisis and the well-being of Americans. Oxf
Econ Pap 2012;64:1–26.

22. Driscoll AK, Bernsetein AB. Health and access to care among employed
and unemployed adults: U.S. NCHS Data Brief 2009–2010;2012:83.
Available at http://www.cdc.gov/nchs/data/databriefs/db83.htm.

23. CDC. Behavioral surveys. Atlanta, GA: CDC; 2007. Available at http://
www.cdc.gov/hiv/topics/surveillance/resources/guidelines/epi-guideline/
la_supp/appendix-a_behavioral.htm.

24. US Department of Health and Human Services, Agency for Healthcare
Research and Quality. Data. Understanding the foundation of quality
improvement. Rockville, MD; US Department of Health and Human
Services, Agency for Healthcare Research and Quality; 2008. Available
at http://www.ahrq.gov/qual/diabqual/diabqguidemod2.htm.

25. Jin RL, Shah PC, Svoboda TJ. The impact of unemployment on health:
a review of the evidence. Can Med Assoc J 1995;153:529–40.

TABLE 2. Overall health status and number of physically and mentally unhealthy days in past 30 days among adults aged 18–64 years, by
employment status — Behavioral Risk Factor Surveillance System, United States, 2010

Health status

Unemployed

Employed*>1 year <1 year

% (95% CI) % (95% CI) % (95% CI)

Overall health status
Excellent or very good† 39.7 (38.1–41.4) 49.2 (47.7–51.2) 62.7 (62.2–63.1)
Good† 35.1 (33.5–36.7) 33.9 (32.3–35.6) 29.1 (28.7–29.5)
Fair or poor† 25.2 (23.8–26.7 16.6 (15.4–18.0) 8.2 (8.0–8.5)

No. of physically unhealthy days in past 30 days
0† 55.9 (54.2–57.5) 63.1 (61.4–64.9) 70.3 (69.9–70.7)
1–15† 31.4 (29.8–33.0) 28.6 (27.0–30.2) 26.1 (25.7–26.5)
16–30† 12.8 (11.7–13.9) 8.3 (7.3–9.4) 3.6 (3.5–3.8)
No. of mentally unhealthy days in past 30 days
0† 50.6 (48.9–52.3) 54.2 (52.4–56.0) 67.3 (66.9–67.8)
1–15† 32.6 (31.0–34.3) 32.9 (31.2–34.6) 27.3 (26.8–27.7)
16–30† 16.8 (15.6–18.1) 12.9 (11.9–14.1) 5.4 (5.2–5.6)

Abbreviation: 95% CI = 95% confidence interval.
* Employed for wages or self-employed.
† Statistically significant (p<0.05).

Supplement

32 MMWR / November 22, 2013 / Vol. 62 / No. 3

26. Bambra C, Eikemo TA. Welfare state regimes, unemployment, and
health: a comparative study of the relationship between unemployment
and self-reported health in 23 European countries. J Epidemiol
Community Health 2009;63:92–8.

27. Harris JR, Huang Y, Hannon PA, et al. Low-socioeconomic status
workers: their health risks and how to reach them. J Occup Environ
Med 2011;53:132–8.

28. US Department of Health and Human Services, National Prevention
Council. National prevention strategy, Washington, DC: US Department
of Health and Human Services; 2011. Available at http://www.healthcare.
gov/prevention/nphpphc/strategy/report.pdf.

29. HealthCare.gov. What is the health insurance marketplace? Baltimore,
MD: Healthcare.gov. Available at https://www.healthcare.gov/
what-is-the-health-insurance-marketplace.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 33

Environmental Hazards

Supplement

34 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 35

Introduction
In 2012, the U.S. civilian labor force comprised an estimated

155 million workers (1). Although employment can contribute
positively to a worker’s physical and psychological health, each
year, many U.S. workers experience a work-related injury or
illness. In 2011, approximately 3 million workers in private
industry and 821,000 workers in state and local government
experienced a nonfatal occupational injury or illness (2).
Nonfatal workplace injuries and illnesses are estimated to cost
the U.S. economy approximately $200 billion annually (3).
Identifying disparities in work-related injury and illness rates can
help public health authorities focus prevention efforts. Because
work-related health disparities also are associated with social
disadvantage, a comprehensive program to improve health equity
can include improving workplace safety and health.

This report and a similar study (4) are part of the second
CDC Health Disparities and Inequalities Report (CHDIR).
The 2011 CHDIR (5) was the first CDC report to assess
disparities across a wide range of diseases, behavior risk factors,
environmental exposures, social determinants, and health-care
access. The topic presented in this report is based on criteria
that are described in the 2013 CHDIR Introduction (6). This
report provides information concerning disparities in nonfatal
work-related injury and illness, a topic that was not discussed
in the 2011 CHDIR. A separate report providing information
on disparities in fatal work-related injuries and homicides
across industry and occupation categories also is included in
this second CHDIR (4). The purposes of this report are to
discuss and raise awareness of differences in the characteristics
of workers employed in high-risk occupations and to prompt
actions to reduce these disparities.

Methods
To examine disparities in nonfatal work-related injury and

illness by selected characteristics, CDC used two sources of
data. Health outcomes were identified by using the Bureau
of Labor Statistics (BLS) Survey of Occupational Injuries and
Illnesses (SOII) (available at http://www.bls.gov/iif ). Data
on selected worker characteristics (i.e., race/ethnicity, place

of birth, sex, age, educational attainment, income level, and
geographic region of residence) were derived from the 2010
Current Population Survey (CPS) microdata files (available
at http://thedataweb.rm.census.gov/ftp/cps_ftp.html). CPS
(available at http://www.census.gov/cps) is the primary source
of U.S. workforce statistics and is based on monthly household
surveys conducted by the U.S. Census Bureau.

Race and ethnicity were combined into seven groups:
Hispanic, non-Hispanic white, non-Hispanic black, American
Indian/Alaska Native, Asian, Hawaiian or Pacific Islander,
or multiple races. Persons of Hispanic ethnicity can be of
any race or combination of races. Educational attainment
was defined as either 1) no education beyond high school,
including those with less than a first-grade education to
those who received a high school diploma or its equivalent or
2) education beyond high school, including enrollment in an
occupational/vocational program, completion of some college,
or receipt of a college degree or an advanced degree. Place of
birth was defined as the United States, a U.S. territory, or a
foreign country. Persons born in a foreign country include
U.S. citizens born abroad (one or both of whose parents were
U.S. citizens), naturalized citizens, and noncitizens. Income
level was defined as low wage or nonlow wage; a low wage was
defined as an income of ≤$435 per week (which is equivalent
to the wage earned by a person working 40 hours a week at
or less than 1.5 times the minimum wage of $7.25 per hour).
Geographic region of residence was defined using the four U.S.
Census Bureau regions: Northeast, Midwest, South and West.

SOII is a collaborative federal/state survey program
administered by BLS that includes reports from a nationally
representative sample of approximately 220,000 private-
sector employers. The survey excludes workers on farms with
<11 employees, private household workers, self-employed
persons, and federal government workers. Data for employees
covered by certain specific federal safety and health legislation
are provided to BLS to be included in SOII by the Mine
Safety and Health Administration of the U.S. Department of
Labor and the Federal Railroad Administration of the U.S.
Department of Transportation. Employers are required to
report workplace injuries and illnesses that meet recordkeeping

Nonfatal Work-Related Injuries and Illnesses — United States, 2010
Sherry L. Baron, MD
Andrea L. Steege, PhD

Suzanne M. Marsh, MPA
Cammie Chaumont Menéndez, PhD

John R. Myers, MS
National Institute for Occupational Safety and Health

Corresponding author: Sherry L. Baron, National Institute for Occupational Safety and Health, CDC. Telephone: 513-458-7159; E-mail: [email protected].

Supplement

36 MMWR / November 22, 2013 / Vol. 62 / No. 3

requirements established by the Occupational Safety and
Health Administration (OSHA), including those that result
in loss of consciousness, restriction of work or motion,
transfer to another job, or medical treatment other than first
aid. Information about the survey methodology is available
at http://www.bls.gov/opub/hom/pdf/homch9.pdf. In 2008,
BLS began including state and local government employees,
with that portion of the survey conducted separately from the
private-sector survey. SOII data presented in this report are
limited to private-sector workers.

In SOII, for those persons whose cases result in ≥1 day
away from work (DAFW), employers provide additional
information, including the affected worker’s occupation coded
according to the Standard Occupational Classification Manual
(SOC) (7). For each of approximately 800 occupations, BLS
then estimates the rate of DAFW cases per 10,000 full-time
equivalents (FTEs) using the formula one FTE = 2,000 hours
worked per year. BLS derives occupation-specific denominator
data from the Occupational Employment Statistics program,
which produces employment and wage estimates for
approximately 800 occupations at the state and national level
(available at http://www.bls.gov/oes). In addition to detailed
SOC occupation rates, BLS also provides injury and illness
rates for all higher-level SOC categories (i.e., injury and illness
rates at the two- through six-digit level of SOC).

CDC used the SOII occupation-specific DAFW injury and
illness rates from 2008 to categorize all private-sector occupations
into two groups: high-risk occupations and all other occupations.
The list of high-risk occupations was obtained from the Council
of State and Territorial Epidemiologists (CSTE) Occupational
Health Indicator activity, indicator no. 15, workers employed
in occupations with a high risk for occupational morbidity
(8). A high-risk occupation was defined as one with a DAFW
rate of at least twice the national DAFW rate of 113.3 cases of
injury and illness per 10,000 FTEs. The CSTE Occupational
Health Indicator activity used Census Bureau occupation
codes, which are a condensed version of the SOC code set
that includes approximately 500 occupation codes (9). Injury
and illness rates for Census Bureau occupation codes were
determined by matching to the corresponding hierarchical
SOC occupation code injury and illness rates released by BLS
in SOII. Of all Census Bureau occupation codes, 61 were
classified as high risk (i.e., having at least twice the national
average DAFW injury and illness rate). Employment estimates
and demographic characteristics were obtained from the 2010
CPS for private-sector wage and salary workers aged ≥16 years
who were employed in the group of 61 high-risk occupations
and for all occupations.

Disparities within high-risk occupations presented in this
report were measured as the absolute differences from a referent

prevalence within each demographic category examined. The
relative difference was calculated by dividing the absolute
difference by the value of the referent category and multiplying
by 100. Statistical significance was assessed based on whether
the 95% confidence intervals (CIs) for each absolute or relative
measure overlapped with the comparison value selected for
each demographic variable. Of the 61 high-risk occupations,
six occupations in which more than 1 million workers were
employed (health aides; janitors and cleaners; maids and
housekeepers; miscellaneous production workers; drivers: sales
and trucks; and hand laborers: freight, stock, material movers)
were examined more closely. Demographic characteristics
were calculated for each specific occupation. Differences were
assessed by calculating and comparing 95% CIs around the
percentage of workers experiencing a nonfatal work-related
injury or illness. In this approach, CIs were used as a measure
of variability and nonoverlapping CIs were considered
statistically different. Using CIs in this way is a conservative
way to evaluate significance differences; infrequently this might
lead to a conclusion that estimates are similar when the point
estimates do differ.

Results
In 2010, approximately 16,679,000 wage and salary workers,

or 16% of all private-sector workers in the United States, were
employed in high-risk occupations. The proportion of workers
employed in high-risk occupations differed significantly by
demographic category, with 21% of males, 24% of Hispanics,
21% of non-Hispanic blacks, 20% of American Indians/Alaska
Natives, 22% of foreign-born workers, and 26% of workers with
no more than a high school education employed in high-risk
occupations, compared with 9% of women, 9% of Asians, 13%
of non-Hispanic whites, and 14% of persons born in the United
States. A higher percentage of workers receiving low wages
worked in high-risk occupations compared with those receiving
higher wages (18% vs. 14%), and the proportion of workers
employed in high-risk occupations was higher in the Midwest
and the South than in the West (16% versus 14%) (Table 1).

In 2010, the six high-risk occupations in which more
than 1 million workers were employed (in each occupation)
accounted for 61% of private-sector wage and salary workers
employed in a high-risk job (Table 2). When the demographic
profiles of each of these six occupations were compared with
those of all U.S. private-sector wage and salary workers, two
demographic characteristics were found consistently to be
statistically elevated in all six occupations: the proportion of
non-Hispanic black workers and that of workers with a high
school education or less. More than half of the workers in four

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 37

of the six occupations earned low wages, a proportion that
exceeded the national average: health aides (64%), janitors
and cleaners (64%), maids and housekeepers (78%), and hand
laborers (58%) (Table 2).

Whereas overall almost three quarters of those employed
in any high-risk occupation were males, two of the six largest
high-risk occupations employed predominately females: maids
and housekeepers (89%) and health aides (88%). Maids
and housekeepers had the highest proportion of Hispanics
(42%) among the six high-risk occupations but a much lower
proportion of non-Hispanic black workers (16%). A reverse
pattern is apparent among health aides, who had the highest
percentage of non-Hispanic black workers (34%) and the

lowest percentage of Hispanics (15%). Foreign-born workers
make up a significantly higher proportion of the workforce
compared with all private-sector workers in four of the six
high-risk occupations: maids and housekeepers (52%), janitors
and cleaners (36%), miscellaneous production workers (25%),
and health aides (25%). With the exception of miscellaneous
production workers, these occupations also had the highest
proportions of low-wage workers found among the six high-
risk/high employment occupations. Compared with all private
sector workers, a higher proportion of maids and housekeepers
(40%) and drivers (40%) were employed in the South, and
a higher proportion of health aides (26%), miscellaneous

TABLE 1. Estimated number and percentage of workers employed in high-risk* occupations, by selected characteristics — United States, 2010

Characteristic

Workers employed in high-risk occupations Absolute difference
Relative

difference†
%No. % (95% CI) 

Percentage
points (95% CI)

Sex
Male 12,240,312 21.1 (20.7–21.5) 12.2§ (11.7–12.7) 137.1§
Female 4,438,820 8.9 (8.6–9.2) Ref. —¶ Ref.

Race/Ethnicity
Hispanic** 4,009,024 24.4 (23.6–25.2) 15.2§ (14.1–16.3) 165.2§
White, non-Hispanic 9,584,598 13.0 (12.7–13.3) 3.8§ (3.0–4.6) 41.3§
Black, non-Hispanic 2,277,643 20.8 (19.9–21.7) 11.6§ (10.4–12.8) 126.1§
American Indian/Alaska Native 97,197 20.2 (15.9–24.5) 11.0§ (6.7–15.3) 119.6§
Asian 494,505 9.2 (8.4–10.0) Ref. — Ref.
Hawaiian or Pacific Islander 47,318 17.8 (13.0–22.6) 8.6§ (3.8–13.4) 93.5§
Multiple races 168,847 15.1 (12.7–17.5) 6.0§ (3.4–8.6) 64.1§

Educational attainment
No education beyond high school†† 11,095,990 25.6 (25.2–26.0) 16.9§ (16.4–17.4) 197.7§
Education beyond high school§§ 5,583,142 8.6 (8.4–8.8) Ref. — Ref.

Place of birth
United States 12,253,418 13.9 (13.6–14.2) Ref. — Ref.
U.S. territory 110,365 19.9 (16.1–23.7) 6.1§ (2.3–9.9) 43.2§
Foreign country¶¶ 4,315,349 22.1 (21.4–22.8) 8.2§ (7.5–8.9) 59.0§

Income level
Low-wage earner *** 7,275,060 18.3 (17.7–18.9) 4.5§ (3.7–5.3) 32.6§
Nonlow-wage earner 9,421,506 13.8 (13.4–14.2) Ref. — Ref.

Geographic region†††
Northeast 3,034,789 14.8 (14.2–15.4) 0.6 (-0.2–1.4) 4.1
Midwest 3,941,498 15.8 (15.3–16.3) 1.6§ (0.9–2.3) 11.2§
South 6,272,961 16.2 (15.8–16.6) 2.0§ (1.3–2.7) 14.3§
West 3,429,884 14.2 (13.7–14.7) Ref. — Ref.

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
Source: U.S. Department of Labor, Bureau of Labor Statistics, Current Population Survey microdata files (available at http://thedataweb.rm. census.gov/ftp/cps_ftp.html).
* Occupations for which having a “days away from work” nonfatal injury and illness rate of ≥226.6 cases per 10,000 full-time equivalents based on U.S. Department

of Labor, Bureau of Labor Statistics 2008 Survey of Occupational Injuries and Illnesses (available at http://www.cste.org/resource/resmgr/OccupationalHealth/
OHIGuidanceMarch2013.pdf?hhSearchTerms=%22Occupational+and+Health+and+Indicator%22).

† Compared with the lowest category.
§ Significantly different when assessed by comparison of nonoverlapping 95% CIs.
¶ Confidence intervals are not provided for the reference category.
** Persons of Hispanic ethnicity might be of any race or combination of races.
†† Includes those with less than a first-grade education to those who received a high school diploma or its equivalent.
§§ Includes enrollment in an occupational/vocational program, completion of some college, or receipt of a college degree or an advanced degree.
¶¶ Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens.
*** Worker whose wage is ≤$435 per week.
††† Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

38 MMWR / November 22, 2013 / Vol. 62 / No. 3

TABLE 2. Estimated percentage of private sector wage and salary workers employed in six high-risk* injury and illness occupations† (each with
>1 million workers), by selected characteristics— United States, 2010

Characteristic

All occupations
(N = 108,216,000;

rate§: 113.3)

Health aides
(n = 1,656,000;

rate: 320.7)

Janitors and
cleaners

(n = 1,561,000;
rate: 243.0)

Maids and
housekeepers
(n = 1,198,000;

rate: 277.7)

Misc. production
workers

(n = 1,047,000;
rate: 462.4)

Drivers: sales
and trucks

(n = 2,721,000;
rate: 329.4)

Hand laborers:
freight, stock, and
material movers
(n = 1,616,000

rate: 440.3)

% (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI)

Sex
Male 53.7 (53.4–54.0) 11.7¶ (9.9–13.5) 67.5¶ (64.8–70.2) 11.1¶ (9.1–13.1) 72.9¶ (69.8–76.0) 95.6¶ (94.7–96.5) 82.5¶ (80.4–84.6)
Female 46.3 (46.0–46.6) 88.3¶ (86.6–90.0) 32.5¶ (29.9–35.1) 88.9¶ (86.9–90.9) 27.1¶ (24.1–30.1) 4.4¶ (3.5–5.3) 17.6¶ (15.5–19.7)

Race/Ethnicity
Hispanic** 15.2 (14.9–15.5) 14.8 (12.8–16.8) 34.0¶ (31.2–36.8) 41.7¶ (38.4–45.0) 23.7¶ (20.6–26.8) 17.8¶ (16.1–19.5) 21.2¶ (18.8–23.6)
White,

non-Hispanic
68.0 (67.7–68.3) 45.4¶ (42.7–48.1) 46.2¶ (43.4–49.0) 35.4¶ (32.3–38.5) 56.7¶ (53.3–60.1) 66.4 (64.4–68.4) 57.6¶ (54.8–60.4)

Black,
non-Hispanic

10.1 (9.9–10.3) 33.5¶ (30.7–36.3) 14.9¶ (12.7–17.1) 16.3¶ (13.7–18.9) 14.0¶ (11.4–16.6) 12.6¶ (11.1–14.1) 16.3¶ (14.1–18.5)

American
Indian/Alaska
Native

0.4 (0.4–0.4) 1.0 (0.4–1.6) 0.7 (0.2–1.2) 0.5 (0.0–1.0) 0.6 (0.0–1.2) 0.4 (0.1–0.7) 0.6 (0.1–1.1)

Asian 5.0 (4.9–5.1) 4.0 (3.0–5.0) 3.4¶ (2.5–4.3) 5.2 (3.9–6.5) 3.9 (2.7–5.1) 1.5¶ (1.0–2.0) 2.8¶ (2.0–3.6)
Hawaiian or

other Pacific
Islander

0.3 (0.3–0.3) 0.3 (0.0–0.6) 0.2 (0.0–0.4) 0.1 (-0.1–0.3) 0.4 (0.0–0.8) 0.2 (0.0–0.4) 0.4 (0.1–0.7)

Multiple races 1.0 (0.9–1.1) 1.1 (0.5–1.7) 0.7 (0.2–1.2) 0.9 (0.3–1.5) 0.9 (0.2–1.6) 1.1 (0.6–1.6) 1.1 (0.5–1.7)
Educational attainment

No education
beyond high
school††

40.1 (39.8–40.4) 53.8¶ (51.3–56.3) 75.4¶ (73.2–77.6) 81.4¶ (79.1–83.7) 72.2¶ (69.4–75.0) 68.9¶ (67.1–70.7) 69.3¶ (67.0–71.6)

Education
beyond high
school§§

59.9 (59.6–60.2) 46.2¶ (43.7–48.7) 24.6¶ (22.4–26.8) 18.6¶ (16.5–20.7) 27.8¶ (25.0–30.6) 31.1¶ (29.3–32.9) 30.7¶ (28.4–33.0)

Place of birth
United States 81.5 (81.2–81.8) 74.4¶ (72.0–76.8) 62.7¶ (59.9–65.5) 47.6¶ (44.3–50.9) 74.3¶ (71.2–77.4) 82.9 (81.3–84.5) 81.3 (79.1–83.5)
U.S. territory 0.5 (0.5–0.5) 0.8 (0.3–1.3) 1.7¶ (1.0–2.4) 0.6 (0.1–1.1) 0.8 (0.2–1.4) 0.5 (0.2–0.8) 0.6 (0.2–1.0)
Foreign

country¶¶
18.0 (17.7–18.3) 24.8¶ (22.4–27.2) 35.6¶ (32.8–38.4) 51.8¶ (48.5–55.1) 24.9¶ (21.9–27.9) 16.6 (15.0–18.2) 18.1 (15.9–20.3)

Income level
Low-wage

earner***
36.8 (36.3–37.3) 63.7¶ (59.9–67.5) 64.1¶ (60.2–68.0) 78.0¶ (74.1–81.9) 34.7 (30.0–39.4) 28.5¶ (25.7–31.3) 57.5¶ (53.5–61.5)

Nonlow-wage
earner

63.2 (62.7–63.7) 36.4¶ (32.6–40.2) 35.9¶ (32.0–39.8) 22.0¶ (18.1–25.9) 65.3 (60.6–70.0) 71.5¶ (68.7–74.3) 42.5¶ (38.5–46.5)

Geographic region†††
Northeast 19.0 (18.7–19.3) 25.1¶ (22.7–27.5) 19.5 (17.2–21.8) 16.8 (14.3–19.3) 16.2 (13.6–18.8) 15.2¶ (13.6–16.8) 16.6 (14.5–18.7)
Midwest 23.1 (22.8–23.4) 26.2¶ (23.7–28.7) 23.8 (21.3–26.3) 17.6¶ (15.1–20.1) 31.2¶ (27.9–34.5) 24.7 (22.8–26.6) 27.9¶ (25.4–30.4)
South 34.7 (34.4–35.0) 34.7 (32.0–37.4) 34.0 (31.3–36.7) 40.2¶ (37.0–43.4) 35.7 (32.3–39.1) 40.4¶ (38.3–42.5) 35.7 (33.0–38.4)
West 22.3 (22.0–22.6) 14.0¶ (12.1–15.9) 22.7 (20.3–25.1) 25.4 (22.5–28.3) 17.0¶ (14.4–19.6) 19.7¶ (18.0–21.4) 19.8 (17.5–22.1)

Abbreviation: 95% CI = 95% confidence interval.
Source: U.S. Department of Labor, Bureau of Labor Statistics, Current Population Survey microdata files (available at http://thedataweb.rm.census.gov/ftp/cps_ftp.html).
* Occupations for which having a “days away for work” nonfatal injury and illness rate of 226.6 cases per 10,000 full time equivalents or greater based on U.S.

Department of Labor, Bureau of Labor Statistics 2008 Survey of Occupational Injuries and Illnesses (available at http://www.cste.org/resource/resmgr/
OccupationalHealth/OHIGuidanceMarch2013.pdf?hhSearchTerms=%22Occupational+and+Health+and+Indicator%22).

† 2002 Census Occupation codes are as follows: health aides (3600); janitors and cleaners (4220); maids and housekeepers (4230); miscellaneous production workers
(8850–8960); drivers: sales and trucks (9130); and hand laborers: freight, stock, and material movers (9620).

§ Injury and illness rate/10,000 full-time equivalents.
¶ Significantly different than all occupations percentage when assessed by comparison of nonoverlapping 95% CIs.
** Persons of Hispanic ethnicity can be of any race or combination of races.
†† Includes those with less than a first-grade education to those who received a high school diploma or its equivalent.
§§ Includes enrollment in an occupational/vocational program, completion of some college, or receipt of a college degree or an advanced degree.
¶¶ Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens.
*** Worker whose wage is ≤$435 per week.
††† Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 39

production workers (31%), and hand laborers (28%) were
employed in the Midwest.

Discussion
Work-related injuries and illnesses are common and

preventable. On average each year, one of every 100 workers
suffers a work-related injury or illness that is severe enough to
result in a missed day of work. These injuries and illnesses are
costly to workers, their families, and society at large. Compared
with U.S.-born white non-Hispanic workers, in 2010, a
higher percentage of workers of all other races, ethnicities,
and places of birth (other than Asian workers) worked in a
job that had at least twice the national average DAFW injury
and illness rate. Overall, a higher percentage of males worked
in a high-risk occupation, but in certain high-risk and high-
employment occupations, workers are predominately female.
A higher percentage of workers who had no more than a high
school education or who earned a weekly wage of ≤$435
worked in a higher-risk job compared with workers who had
a higher education level or who earned more. The burden of
a work-related injury or illness for these workers might be
compounded further by other sources of health inequalities.
For example, in 2010, among working-age adults with an
income of 100%–200% of the federal poverty level, 43%
did not have access to health insurance for at least part of the
previous year (10). In addition, because a greater proportion
of workers in high-risk occupations are foreign-born and have
lower levels of educational attainment than other workers, as
one element of a comprehensive workplace safety and health
program, training and education materials are needed that
focus on addressing the needs of persons with low English
proficiency and literacy levels (11). Intervention priorities
can be informed by employment patterns (i.e., geographic
concentration of workers employed in high-risk occupations)
such as programs to promote better workplace safety for maids
and housekeepers employed in the southern states.

Limitations
The findings in this report are subject to at least three

limitations. First, the private-sector SOII data exclude workers
on farms with <11 employees, private household workers,
and persons who are self-employed (12). If these excluded
workers have higher or lower injury and illness rates than
other private-sector wage and salary workers employed in the
same occupation, then the workplace injury and illness rates
for that occupation might be under- or overestimated. Second,
inclusion of cases in SOII is dependent on identifying cases as

work-related; such determinations can be difficult for certain
types of incidents for which the work relationship might not
be clear or recordkeeping requirements are misinterpreted
(13). Also, the work relationship might be underreported
by some workers, especially those who perceive their jobs
as being insecure, which might affect minority and lower-
income workers differentially (14). Finally, underreporting of
work-related illnesses is especially problematic because many
work-related illnesses (e.g., cancer and chronic obstructive
lung diseases) take years to develop and might be difficult to
attribute to the workplace (15).

Conclusion
The findings provided in this report highlight the importance

of preventing work-related injuries and illnesses. The
Occupational Safety and Health Act affords equal protection
to all workers, regardless of race, ethnicity, or immigrant
status. Furthering a culture in which occupational safety and
health is recognized and valued as a fundamental component
of economic growth and prosperity can play an important
role in promoting health equity. Identifying disparities in
work-related injury and illness rates can help public health
authorities focus prevention efforts. Because work-related
health disparities also are associated with social disadvantage
(i.e., workers with low socioeconomic status are those workers
who had no more than a high school education or who earned
a weekly wage of ≤$435), a comprehensive program to improve
health equity should include improving workplace safety and
health. The data presented in this report can be used to help
focus prevention efforts on those workers in the highest-risk
jobs. This information can be used to improve intervention
efforts by developing programs that better meet the needs of
the increasing diversity of the U.S. workforce. The National
Institute for Occupational Safety and Health’s Occupational
Health Disparities program has prioritized research projects to
improve outreach to eliminate health disparities. Prevention
recommendations and publications that discuss common
injury and illness concerns for these workers are available
in English and Spanish; topics include safe patient lifting,
chemical use, eye protection, motor vehicle safety, and manual
materials handling (available at http://www.cdc.gov/NIOSH).

References
1. US Bureau of Labor Statistics. Current Population Survey Table 1.

Employment status of the civilian noninstitutional population, 1942 to
date. Washington, DC: US Bureau of Labor Statistics; 2013. Available
at http://www.bls.gov/cps/cpsaat01.pdf.

2. US Bureau of Labor Statistics. Economic news release: workplace injury
and illness summary. Washington, DC: US Department of Labor, Bureau
of Labor Statistics; 2012.

Supplement

40 MMWR / November 22, 2013 / Vol. 62 / No. 3

3. Leigh JP. Economic burden of occupational injury and illness in the
United States. Millbank Q 2011;89:728–72.

4. CDC. Fatal work-related injuries—United States, 2005–2009: In: CDC.
CDC health disparities and inequalities report—United States, 2013.
MMWR 2013;62(No. Suppl 3).

5. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

6. CDC. Introduction: In: CDC. CDC health disparities and inequalities
report—United States, 2013. MMWR 2013;62(No. Suppl 3).

7. US Bureau of Labor Statistics. Standard occupational classification
(SOC) user guide. Washington, DC: US Bureau of Labor Statistics;
2000. Available at http://www.bls.gov/soc/2000/socguide.htm.

8. Council of State and Territorial Epidemiologists. Occupational health
indicators: a guide for tracking occupational health conditions and their
determinants; Atlanta, GA: Council of State and Territorial
Epidemiologists; 2012.

9. US Bureau of Labor Statistics. Occupational and industry classification
systems used in the Current Population Survey. Available at http://www.
bls.gov/cps/cpsoccind.htm.

10. Cohen RA, Ward BW, Schiller JS. Health insurance coverage: early
release of estimates from the National Health Interview Survey, 2010.
Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics; 2010. Available at http://www.
cdc.gov/nchs/data/nhis/earlyrelease/insur201106.htm#Footnotes4.

11. O’Connor T. White paper on reaching Spanish-speaking workers and
employers with occupational safety and health information. In:
Committee on Communicating Occupational Safety and Health
Information to Spanish-speaking Workers; Committee on Earth
Resources; Board on Earth Sciences and Resources (BESR); Division on
Earth and Life Studies (DELS); National Research Council. Safety is
seguridad: a workshop summary. Washington, DC: National Academies
Press; 2003.

12. US Bureau of Labor Statistics. Occupational safety and health statistics.
In: US Bureau of Labor Statistics. BLS handbook of methods.
Washington, DC: US Bureau of Labor Statistics; 2009. Available at
http://www.bls.gov/opub/hom/pdf/homch9.pdf.

13. US Government Accounting Office. Workplace safety and health:
enhancing OSHA’s records audit process could improve the accuracy of
worker injury and illness data. Washington, DC: US Government
Accounting Office; 2009.

14. McGreevy K, Lefkowitz D, Valiante D, Lipsitz S. Utilizing hospital
discharge data (HD) to compare fatal and non-fatal work-related injuries
among Hispanic workers in New Jersey. Am J Ind Med 2010;
53:146–52.

15. Ruser J. Examining evidence on whether BLS undercounts workplace
injuries and illnesses. Monthly Labor Review 2008;131:20–32.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 41

Introduction
In 2012, the U.S. civilian labor force comprised an estimated

155 million workers (1). Although employment can contribute
positively to a worker’s physical and psychological health, each
year, many U.S. workers are fatally injured at work. In 2011,
a total of 4,700 U.S. workers died from occupational injuries
(2). Workplace deaths are estimated to cost the U.S. economy
approximately $6 billion annually (3). Identifying disparities
in work-related fatality rates can help public health authorities
focus prevention efforts. Because work-related health disparities
also are associated with social disadvantage, a comprehensive
program to improve health equity should include improving
workplace safety and health.

This report and a similar study (4) are part of the second
CDC Health Disparities and Inequalities Report (CHDIR).
The 2011 CHDIR (5) was the first CDC report to assess
disparities across a wide range of diseases, behavior risk factors,
environmental exposures, social determinants, and health-care
access. The topic presented in this report is based on criteria
that are described in the 2013 CHDIR Introduction (6). This
report provides information on disparities in work-related
death and homicide rates across industry and occupation
categories, a topic that was not discussed in the 2011 CHDIR.
A separate report providing information on disparities in
nonfatal work-related injuries and illnesses also is included
in this second CHDIR (4). The purposes of this report are to
discuss and raise awareness of differences in the characteristics
of work-related fatal injuries and to prompt actions to reduce
these disparities.

Methods
To characterize work-related death and homicide rates by

selected characteristics, CDC used two sources of data. Fatalities
were identified by using the Census of Fatal Occupational
Injuries (CFOI),* and employment data were derived from
the Current Population Survey (CPS) microdata files.

For CFOI, BLS collects data on occupational injury deaths
from multiple federal, state, and local sources, including
death certificates, police reports, and workers’ compensation
reports. To be included in CFOI, the decedent must have been
employed at the time of the incident, working as a volunteer
in the same functions as a paid employee, or present at a site
as a job requirement (7). Public- and private-sector civilian
workers are included. CFOI excludes deaths that occurred
during a worker’s normal commute to and from work and
deaths related to occupational illnesses (e.g., lung disease or
cancer). CFOI uses its fatality source documents to extract
and code demographic information and place of birth as well
as information related to the event or exposure that directly
caused the death and the occupation and industrial sector in
which the decedent was employed.

Race and ethnicity were combined into four broad groups:
non-Hispanic white, non-Hispanic black, American Indian/
Alaska Native/Asian/Pacific Islander (AI/AN/A/PI), and
Hispanic. Persons of Hispanic ethnicity might be of any race
or combination of races. Place of birth was defined as either the
United States or its territories (including Puerto Rico, Guam,
and the U.S. Virgin Islands) or a foreign country. Persons born
in a foreign country include U.S. citizens born abroad (one or
both of whose parents were U.S. citizens), naturalized citizens,
and noncitizens. Legal immigrants, legal nonimmigrants, and
undocumented workers were included in the foreign-born
population if their deaths were confirmed as work-related.
Information on educational attainment was not available
from the CFOI data. Information on geographic region, while
available in the data, were not included in the analysis.

To calculate injury-related fatality rates, CDC derived labor
force denominator estimates from the CPS microdata files (8).
CPS is the primary source of U.S. labor force statistics and is
based on monthly household surveys conducted by the U.S.
Census Bureau. Demographic and employment characteristics
in CPS were grouped to match categories in CFOI. CPS
uses the Census Bureau definition of “foreign-born,” which
is slightly different than the definition used by CFOI. Along
with including persons who were born in the United States and
its territories, CPS, unlike CFOI, also identifies persons born

Fatal Work-Related Injuries — United States, 2005–2009
Suzanne M. Marsh, MPA

Cammie Chaumont Menéndez, PhD
Sherry L. Baron, MD
Andrea L. Steege, PhD

John R. Myers, MS
National Institute for Occupational Safety and Health, CDC

Corresponding author: Suzanne M. Marsh, National Institute for Occupational Safety and Health, CDC. Telephone: 304-285-6009; E-mail: [email protected].

* Analysis was conducted using restricted CFOI data that the National Institute for
Occupational Safety and Health receives through a Memorandum of Understanding.
Results might differ from those released by the Bureau of Labor Statistics.

Supplement

42 MMWR / November 22, 2013 / Vol. 62 / No. 3

abroad to a U.S. citizen as “native-born.” Rates were calculated
per 100,000 workers aged ≥15 years.

Poisson regression was used to estimate injury-fatality rates
and 95% confidence intervals (CIs) for selected categorical
groups (sex, age group, selected events, industry division, and
occupation) stratified by demographic variables (race/ethnicity
and place of birth). The injury-fatality category rate was
considered elevated if it was >1.5 times the U.S. rate (3.7 per
100,000 workers for all fatalities and 0.4 per 100,000 workers
for homicides) and also was considered significantly different
if it did not contain the U.S. rate (3.7 for all fatalities and
0.4 for homicides). The injury-fatality rate for each category
was further stratified by certain demographic variables (race/
ethnicity and place of birth). The demographic-specific rate was
considered elevated if it also was >1.5 times the corresponding
U.S. rate for that particular category (i.e., sex, age group,
selected events, industry division, and occupation) and also
was considered significantly different if its confidence interval
did not contain the overall category rate. No statistical testing
was done for this analysis.

Results
During 2005–2009, U.S. workers died from an injury while

at work at a rate of 3.7 per 100,000 workers. Hispanics and
foreign-born workers had the highest work-related fatal injury
rates (4.4 and 4.0 per 100,000 workers, respectively) (Table 1).
For all races, ethnicities, and places of birth, males had work-
related fatality rates that were 9 to 14 times higher than the rates
for females. Fatal injury rates increased with age for all races,
ethnicities, and nativities, with non-Hispanic whites, non-
Hispanic blacks, and workers born in the United States or its
territories having the most dramatic increases. Hispanics of all
age groups <65 years had the highest fatality rates, particularly
Hispanics aged 15–24 years. Similarly, foreign-born workers of
all age groups <65 years had higher fatality rates than workers
who were born in the United States or its territories.

The greatest differences in work-related injury fatality rates
were across industry sectors, with the rates in agriculture,
mining, construction, and transportation/warehousing/utilities
being three to almost eight times higher than the overall U.S.
rate (Table 1). Although fatality rates by industry sector were
similar across most races/ethnicities, non-Hispanic blacks
had either the highest or second highest fatality rate for every
industry sector, and in agriculture, forestry, and fishing, their
rate was just over 1.5 times the U.S. rate for that industry. AI/
AN/A/PI and foreign-born workers in the trade sector had rates
that were 1.5 to 2.0 times the U.S. rate.

Transportation incidents at work resulted in the highest
work-related fatality rates for workers of all races, ethnicities,
and nativities (Table 1). Rates for assaults and violent acts,
particularly homicides, showed the greatest disparity across
race, ethnicity, and place of birth and were highest among
non-Hispanic blacks, AI/AN/A/PIs, and foreign-born workers.

During 2005–2009, a total of 2,803 workers were homicide
victims (rate: 0.4 per 100,000 workers) (Table 2). Homicide
rates for non-Hispanic black and AI/AN/A/PI workers were
three times those of non-Hispanic white workers. The homicide
rate for foreign-born workers was more than twice that of all
other workers. The majority of workplace homicide victims
among non-Hispanic blacks were not foreign-born (83%),
whereas the majority of such victims among Hispanic workers
(61%) and AI/AN/A/PI workers (89%) were foreign-born.

Male workers experienced at least triple the homicide rate
that women experienced regardless of race/ethnicity or place of
birth (Table 2). Most notably, non-Hispanic black, AI/AN/A/
PI, and foreign-born men experienced the highest homicide
rates. Hispanic women had the highest rate among women.
Overall, workers aged 15–19 years experienced the lowest
rates, and workers aged ≥65 years experienced the highest rates.
Non-Hispanic black and AI/AN/A/PI workers experienced
significantly higher rates for every age group.

Sales and related occupations (e.g., store managers, clerks,
and cashiers) and transport and material moving occupations
(e.g., taxi drivers and truck drivers) had the highest work-related
homicide rates (Table 2). AI/AN/A/PI workers in sales and
transportation occupations experienced the highest homicide-
related fatality rates. Non-Hispanic blacks consistently had at
least double the work-related homicide rates compared with
non-Hispanic whites for every industry and occupation group.

To further understand the circumstances of these workplace
homicides, CDC explored specific characteristics of the victims
(data not presented). Among the 1,483 (55%) homicides for
which the type of perpetrator was specified, 1,039 (70%) were
committed by suspected robbers, 292 (20%) by a coworker or
former coworker, 109 (7%) by a relative of the homicide victim,
and 43 (3%) by miscellaneous “others.” Men and women were
generally victims of different types of workplace violence. Of
those homicides that occurred during a suspected robbery or that
were perpetrated by a coworker/former coworker, 1,119 (84%)
victims were men, whereas of the 109 homicides perpetrated by
a relative of the victim, 84 (77%) victims were women.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 43

TABLE 1. Number and rate* of fatal occupational injuries — Census of Fatal Occupational Injuries, United States, 2005–2009

Characteristic

Total

Race/Ethnicity Place of birth†

White,
non-Hispanic

Black,
non-Hispanic AI/AN/A/PI Hispanic§ Foreign born

U.S. or U.S.
territories

No. Rate No. Rate No. Rate No. Rate No. Rate No. Rate No. Rate

Total¶ 26,996 3.7 18,682 3.7 2,707 3.5 982 2.6 4,367 4.4 4,665 4.0 22,331 3.6

Sex
Male 24,995 6.4** 17,227 6.4 2,485 7.0 893 4.4 4,151 6.9 4,416 6.3 20,579 6.4
Female 2,001 0.6 1,455 0.6 222 0.5 89 0.5 216 0.5 249 0.5 1,752 0.6

Age group (yrs)
15–19 572 1.8 342 1.5 45 1.5 20 2.3 163 3.6†† 141 6.1†† 431 1.4
20–24 1,845 2.6 1,049 2.2 183 2.3 69 2.2 524 4.3†† 445 4.8†† 1,400 2.2
25–34 4,603 2.9 2,744 2.8 493 2.7 153 1.6 1,162 3.9 1,054 3.4 3,549 2.8
35–44 5,720 3.4 3,660 3.3 689 3.6 212 2.0 1,089 4.2 1,179 3.5 4,541 3.3
45–54 6,696 3.9 4,802 3.8 722 4.1 284 3.4 836 4.8 1,034 4.0 5,662 3.9
55–64 4,603 4.6 3,569 4.5 395 4.7 182 4.0 430 5.7 610 4.8 3,993 4.6
≥65 2,882 10.0** 2,472 10.5 179 8.8 56 5.0 159 8.4 198 5.8 2,684 10.6

Selected events§§

Contact with object and equipment¶¶ 4,596 0.6 3,210 0.6 375 0.5 108 0.3 881 0.9†† 809 0.7 3,787 0.6
Falls 3,789 0.5 2,561 0.5 219 0.3 107 0.3 877 0.9†† 879 0.7 2,910 0.5

Fall to lower level*** 3,279 0.5 2,175 0.4 172 0.2 85 0.2 826 0.8†† 817 0.7 2,462 0.4
Exposure to harmful substances/

environs†††
2,388 0.3 1,593 0.3 220 0.3 56 0.1 495 0.5†† 446 0.4 1,942 0.3

Transportation incidents 11,228 1.5 8,263 1.6 1,134 1.5 309 0.8 1,394 1.4 1,358 1.1 9,870 1.6
Highway incident 6,407 0.9 4,669 0.9 728 1.0 176 0.5 780 0.8 730 0.6 5,677 0.9

Fires and explosions 800 0.1 588 0.1 79 0.1 12 <0.1 118 0.1 104 0.1 696 0.1
Assaults and violent acts 4,097 0.6 2,403 0.5 666 0.9†† 390 1.0†† 582 0.6 1,052 0.9†† 3,045 0.5

Assaults and violent acts by person 2,803 0.4 1,354 0.3 605 0.8†† 334 0.9†† 459 0.5 876 0.7†† 1,927 0.3

Industry division§§§

Agriculture/Forestry/Fishing 3,236 29.2** 2,576 31.0 130 46.9†† 64 31.4 426 19.7 399 18.7 2,837 31.7
Mining 810 22.6** 612 22.1 38 22.3 16 16.8 140 27.1 57 19.1 753 23.0
Construction 5,674 10.2** 3,661 9.8 390 13.6 117 9.9 1,473 10.9 1,338 10.1 4,336 10.2
Manufacturing 1,984 2.5 1,373 2.5 214 2.9 65 1.4 327 2.8 321 2.1 1,663 2.6
Trade 2,725 2.6 1,813 2.4 275 2.8 260 5.0†† 346 2.5 644 4.3†† 2,081 2.3
Transportation/Warehousing/Utilities 4,484 11.9** 3,125 12.8 679 11.5 151 9.0 469 9.2 663 11.6 3,821 12.0
Services, excluding health care 7,388 2.1 4,995 2.0 889 2.5 283 1.4 1,141 2.6 1,189 2.2 6,199 2.1
Health care and social services 664 0.7 509 0.9 90 0.6 23 0.4 39 0.4 49 0.4 615 0.8

Occupation group¶¶¶

Management, business, and finance 2,896 2.7 2,614 3.1 84 1.1 81 1.4 106 1.4 192 1.5 2,704 2.9
Professional and related 1,274 0.8 1,031 0.9 93 0.7 57 0.5 75 0.7 144 0.7 1,130 0.9
Service 3,456 2.8 2,153 3.0 504 2.8 104 1.6 673 2.8 607 2.3 2,849 3.0
Sales and related 1,532 1.9 982 1.6 154 2.1 214 5.0 158 1.7 448 4.1†† 1,084 1.5
Office and administrative support 516 0.5 359 0.5 66 0.5 20 0.5 63 0.5 71 0.7 445 0.5
Farming, fishing, and forestry 1,405 27.5** 844 31.0 98 43.6 54 44.0 376 19.3 351 18.6 1,054 32.9
Construction and extraction 5,445 12.3** 3,412 12.4 380 14.2 113 12.7 1,502 11.7 1,338 11.0 4,107 12.8
Installation, maintenance, and repair 1,896 7.2 1,466 7.6 145 7.2 39 4.1 234 6.4 198 5.5 1,698 7.5
Production 1,289 2.9 824 3.0 143 2.7 51 2.0 265 2.8 260 2.4 1,029 3.0
Transport and material moving 7,057 16.0** 4,837 18.0 1,020 14.8 238 17.2 889 10.6 1,039 12.6 6,018 16.8

Abbreviation: AI/AN/A/PI = American Indian/Alaska Native/Asian/Pacific Islander.
* Per 100,000 workers aged ≥15 years. Rates were calculated by CDC based on the number of fatalities from restricted data from the Bureau of Labor Statistics (BLS) Census of Fatal

Occupational Injuries during 2005–2009 and might differ from estimates published by the BLS; the estimated number of employed workers was obtained from the BLS Current Population
Survey, 2005–2009. Per BLS publication requirements, numbers of deaths are reported for workers of all ages whereas rates are for workers aged ≥15 years.

† For CFOI, persons born in a foreign country include U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens. For CPS, persons
born in the U.S. or its territories include U.S. citizens born abroad (one or both of whose parents were U.S. citizens).

§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Totals include workers of other/unknown race and ethnicity.
** Indicates that the overall rate for the certain categories (sex, age, selected events, industry division, and occupation group) is >1.5 times the U.S. injury-fatality rate of 3.7 per 100,000

workers. Also indicates that the overall category rate is significantly different from the U.S. rate because the confidence interval for the category rate does not contain 3.7 (the U.S. rate).
†† Indicates that demographic-specific (race/ethnicity and place of birth) rate is considered elevated because it is >1.5 times the corresponding U.S. rate for that particular category (i.e.,

sex, age group, selected events, industry division and occupation) rate. Also indicates that this demographic-specific rate is considered significantly different because its confidence
interval does not contain the corresponding overall category rate.

§§ Event or exposure according to the BLS Occupational Injury and Illness Classification System (available at http://www.bls.gov/iif/oiics_manual_2007.pdf ). Totals for major events or
exposures include subcategories not shown separately.

¶¶ Examples include being struck by a falling object such as a tree, being crushed during a cave-in while digging ditches, or getting caught in running machinery.
*** Examples include falling from a ladder, roof, or scaffold; falling down stairs or steps; or falling through a floor or roof.
††† Examples include heat stroke or hypothermia, poisoning through inhalation or ingestion of harmful substances, insect stings and animal bites, and non-transportation-related drownings.
§§§ Industry in which the decedent worked was coded according to the 2002 North American Industry Classification System (NAICS) (available at http://www.census.gov/eos/www/naics).

The detailed codes from the 20 NAICS sectors were combined into eight industry sectors according to the similarity of their occupational safety and health risks.
¶¶¶ Occupation in which the decedent worked was coded according to the 2000 Standard Occupational Classification Manual (SOC) (available at http://www.bls.gov/soc). The detailed codes

from the 22 civilian SOC groups were combined into ten occupation groups according to the similarity of their work and their occupational safety and health risks.

Supplement

44 MMWR / November 22, 2013 / Vol. 62 / No. 3

Discussion
On average, each day, 12–13 workers in the United States

die from injuries sustained at work. Hispanic and foreign-
born workers are at higher risk compared with other workers,
primarily because of the type of work that they do. Workers of
all races, ethnicities, and places of birth working in construction,
agriculture, mining, and transportation face a similar and
higher risk for a work-related fatal injury than workers in other
industries. Approximately 10% of injury-related fatalities at

TABLE 2. Number and rate* of homicide deaths — Census of Fatal Occupational Injuries, United States, 2005–2009

Characteristic

Total

Race/Ethnicity Place of birth†

White,
non-Hispanic

Black,
non-Hispanic AI/AN/A/PI Hispanic§ Foreign born

U.S. or U.S.
territories

No. Rate No. Rate No. Rate No. Rate No. Rate No. Rate No. Rate

Total¶ 2,803 0.4 1,354 0.3 605 0.8** 334 0.9** 459 0.5 876 0.74** 1,927 0.3

Sex
Male 2,291 0.6†† 1,054 0.4 523 1.5** 299 1.5** 368 0.6 772 1.1** 1,519 0.5
Female 512 0.2 300 0.1 82 0.2 35 0.2 91 0.2** 104 0.2 408 0.1

Age group (yrs)
15–19 62 0.2 23 0.1 14 0.5** 8 0.9** 17 0.4** 23 1.0** 39 0.1
20–24 207 0.3 81 0.2 51 0.7** 23 0.8** 50 0.4 60 0.6** 147 0.2
25–34 589 0.4 247 0.3 163 0.9** 44 0.5 123 0.4 157 0.5 432 0.4
35–44 714 0.4 324 0.3 155 0.8** 80 0.8** 136 0.5 251 0.8** 463 0.3
45–54 633 0.4 326 0.3 125 0.7** 89 1.1** 82 0.5 209 0.8** 424 0.3
55–64 415 0.4 243 0.3 60 0.7** 72 1.6** 34 0.5 137 1.1** 278 0.3
≥65 178 0.6†† 110 0.5 36 1.8** 16 1.4** 16 0.9 38 1.1** 140 0.6

Industry division§§

Agriculture/Forestry/Fishing 40 0.4 24 0.3 —¶¶ — — — 13 0.6** 15 0.7** 25 0.3
Mining 5 0.1 — — — — — — — — — — — —
Construction 99 0.2 51 0.1 13 0.5** — — 28 0.2 26 0.2 73 0.2
Manufacturing 80 0.1 — — — — — — 23 0.2** 24 0.2** 56 0.1
Trade 790 0.8†† 342 0.5 141 1.5** 190 3.7** 98 0.7 389 2.6** 401 0.5
Transportation/Warehousing/

Utilities
267 0.7†† 99 0.4 96 1.6** 18 1.0 39 0.8 91 1.6** 176 0.6

Services, excluding health care 1,412 0.4 731 0.3 312 0.9** 107 0.6 247 0.6** 314 0.6 1,098 0.4
Health care and social services 109 0.1 58 0.1 32 0.2** 9 0.2 — — 16 0.1 93 0.1

Occupation group***
Management, business, and finance 267 0.3 154 0.2 42 0.6** 39 0.7** 31 0.4** 73 0.6** 194 0.2
Professional and related 162 0.1 111 0.1 29 0.2** 11 0.1 8 0.1 29 0.1 133 0.1
Service 841 0.7†† 419 0.6 216 1.2** 37 0.6 161 0.7 157 0.6 684 0.7
Sales and related 773 0.9†† 344 0.6 129 1.7** 179 4.2** 103 1.1 373 3.4** 400 0.6
Office and administrative support 133 0.1 74 0.1 26 0.2** — — 21 0.2 27 0.3** 106 0.1
Farming, fishing, and forestry 26 0.5 10 0.4 — — — — 12 0.6 13 0.7 13 0.4
Construction and extraction 82 0.2 36 0.1 — — — — 31 0.2 23 0.2 59 0.2
Installation, maintenance, and repair 74 0.3 37 0.2 19 0.9** 5 0.5** 13 0.4 19 0.5** 55 0.3
Production 66 0.2 26 0.1 12 0.2** — — 23 0.3** 26 0.2** 40 0.1
Transport and material moving 365 0.8†† 135 0.5 120 1.7** 40 2.9** 54 0.6 134 1.6** 231 0.7

Abbreviation: AI/AN/A/PI = American Indian/Alaska Native/Asian/Pacific Islander.
* Per 100,000 workers aged ≥15 years. Rates were calculated by CDC based on the number of fatalities from restricted data from the Bureau of Labor Statistics (BLS) Census of Fatal

Occupational Injuries during 2005–2009 and might differ from estimates published by the BLS; the number of employed workers from the BLS Current Population Survey, 2005–2009.
Per BLS publication requirements, numbers of deaths are reported for workers of all ages whereas rates are for workers aged ≥15 years.

† For CFOI, persons born in a foreign country include U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens. For CPS, persons born in
the U.S. or its territories include U.S. citizens born abroad (one or both of whose parents were U.S. citizens).Persons of Hispanic ethnicity might be of any race or combination of races.

§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Totals include workers of other/unknown race and ethnicity.
** Indicates that demographic-specific (race/ethnicity and place of birth) rate is considered elevated because it is >1.5 times the corresponding U.S. rate for that particular category (i.e.,

sex, age group, industry division and occupation) rate. Also indicates that this demographic-specific rate is considered significantly different because its confidence interval does not
contain the corresponding overall category rate.

†† Indicates that the overall rate for the certain categories (sex, age, industry division, and occupation group) is >1.5 times the U.S. injury-fatality homicide rate of 0.4 per 100,000 workers.
Also indicates that the overall category rate is significantly different from the U.S. rate because the confidence interval for the category rate does not contain 0.4 (the U.S. rate).

§§ Industry in which the decedent worked was coded according to the 2002 North American Industry Classification System (NAICS) (available at http://www.census.gov/eos/www/naics/).
The detailed codes from the 20 NAICS sectors were combined into eight industry sectors according to the similarity of their occupational safety and health risks.

¶¶ Data do not meet confidential BLS publication criteria.
*** Occupation in which the decedent worked was coded according to the 2000 Standard Occupational Classification Manual (SOC) (available at http://www.bls.gov/soc/). The detailed codes

from the 22 civilian SOC groups were combined into ten occupation groups according to the similarity of their work and their occupational safety and health risks.

work are homicides, which occur most frequently during a
robbery. Customer service workers who handle money and
who often work alone (e.g., cashiers and taxi drivers) are at
highest risk. AI/AN/A/PI workers in transportation and sales
occupations were at especially high risk. However for every
type of occupation, black non-Hispanic workers were twice
as likely as white non-Hispanic workers to be a homicide
victim. Efforts to prevent robbery-related homicides include
establishing workplace policies and procedures that engage
management and employees; providing appropriate worksite

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 45

analysis and safety and health training; ensuring that minimal
cash is kept on hand; and enhancing and securing the physical
environment with alarm systems, surveillance cameras, mirrors,
and adequate lighting and barriers (9).

Women were more likely to be the victim of a homicide
perpetrated by a relative. In these instances, the violence not only
affects the worker but may also affect co-workers and/or customers
who may be present during the incident. Multidisciplinary
workplace violence prevention programs that incorporate training
and perpetrator-specific prevention strategies should be made
available and implemented widely (10).

Limitations
The findings in this report are subject to at least four

limitations. First, inclusion of cases in CFOI is dependent
upon identifying work-relatedness. This determination can
be difficult for certain types of incidents for which the work
relationship might not be clear. Second, work-related deaths
enumerated in CFOI are limited to fatal injuries and do not
include work-related deaths attributable to chronic illnesses
such as cancer or lung disease. It is estimated that approximately
49,000 deaths each year can be attributed to work-related
illnesses (11). Third, CFOI includes fatalities to volunteers.
However, volunteers are not included in the CPS denominator,
potentially resulting in an overestimation of fatality rates
presented in this report by CDC. Finally, CFOI and CPS use
different approaches to defining place of birth, which might
result in an underestimate of injury rates for some categories.

Conclusion
These findings highlight the importance of preventing

work-related deaths. All workers, regardless of their race,
ethnicity, or immigrant status are afforded equal protection
under the Occupational Safety and Health Act. Furthering a
culture in which occupational safety and health is recognized
and valued as a fundamental component of economic growth
and prosperity can play an important role in promoting health
equity. The fatality data presented in this report provide
important information to focus prevention efforts. These
findings highlight priority industries and occupations of

workers in the highest risk jobs for all occupational fatalities
and for homicides specifically. This information can be used
to improve intervention efforts by developing programs
that better meet the needs of the increasing diversity of the
U.S. workforce. NIOSH’s Occupational Health Disparities
program has prioritized research projects to improve outreach
to eliminate health disparities and NIOSH’s National
Occupational Research Agenda addresses high priority needs in
individual industry sectors through research and partnerships.
Prevention recommendations and publications that focus
on the most serious concerns for these workers are available
in English and Spanish; topics include workplace violence
prevention, motor vehicle safety, and machine safety (available
at http://www.cdc.gov/NIOSH/injury).

References
1. US Bureau of Labor Statistics. Current population survey. Table 1.

Employment status of the civilian noninstitutional population, 1942 to
date . Washington, DC: US Bureau of Labor Statistics; 2013. Available
at http://www.bls.gov/cps/cpsaat01.pdf.

2. US Bureau of Labor Statistics. Census of Fatal Occupational Injuries
charts, 1992–2011 (revised data), number of fatal work injuries,
1992–2011. Washington, DC: US Bureau of Labor Statistics; 2013.
Available at http://www.bls.gov/iif/oshwc/cfoi/cfch0010.pdf.

3. Leigh JP. Economic burden of occupational injury and illness in the
United States. Milbank Q 2011;89:728–72.

4. CDC. Nonfatal work-related injuries and illnesses—United States, 2010.
In: CDC. CDC health disparities and inequalities report—United States,
2013. MMWR 2013;62(No. Suppl 3).

5. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

6. CDC. Introduction. In: CDC. CDC health disparities and inequalities
report—United States, 2013. MMWR 2013;62(No. Suppl 3).

7. US Bureau of Labor Statistics. Occupational safety and health statistics.
In: BLS handbook of methods. Washington, DC: US Bureau of Labor
Statistics; 2013.

8. US Bureau of Labor Statistics. Labor force data derived from the Current
Population Survey. In: BLS handbook of methods. Washington, DC:
US Bureau of Labor Statistics; 2013. Available at http://www.bls.gov/
opub/hom/pdf/homch1.pdf.

9. Occupational Safety and Health Administration. Recommendations for
workplace violence prevention programs in late-night retail establishments.
DOL (OSHA) Publication No. 3153-12R 2009. Available at http://
www.osha.gov/Publications/osha3153.pdf.

10. National Institute for Occupational Safety and Health. Workplace
violence prevention strategies and research needs. DHHS (NIOSH)
Publication No. 2006-144. Available at http://www.cdc.gov/niosh/
docs/2006-144/pdfs/2006-144.pdf.

11. CDC. Workers Memorial Day—April 28, 2012. MMWR 2012;61:281.

Supplement

46 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
Traffic-related air pollution is a main contributor to

unhealthy ambient air quality, particularly in urban areas
with high traffic volume. Within urban areas, traffic is a
major source of local variability in air pollution levels, with
the highest concentrations and risk of exposure occurring near
roads. Motor vehicle emissions represent a complex mixture
of criteria air pollutants, including carbon monoxide (CO),
nitrogen oxides (NOx), and particulate matter (PM), as well
as hydrocarbons that react with NOx and sunlight to form
ground-level ozone. Individually, each of these pollutants is
a known or suspected cause of adverse health effects (1–4).
Taking into consideration the entire body of evidence on
primary traffic emissions, a recent review determined that there
is sufficient evidence of a causal association between exposure
to traffic-related air pollution and asthma exacerbation
and suggestive evidence of a causal association for onset of
childhood asthma, nonasthma respiratory symptoms, impaired
lung function, all-cause mortality, cardiovascular mortality, and
cardiovascular morbidity (5).

The mixture of traffic-related air pollutants can be difficult
to measure and model. For this reason, many epidemiologic
studies rely on measures of traffic (e.g., proximity to major
roads, traffic density on nearest road, and cumulative traffic
density within a buffer) as surrogates of exposure (6–8). These
traffic measures typically account for both traffic volume (i.e.,
number of vehicles per day), which is a marker of the type
and concentration of vehicle emissions, and distance, which
addresses air pollution gradients near roads. Traffic emissions
are highest at the point of release and typically diminish to near
background levels within 150 to 300 meters of the roadway
(7,9,10); however, the potential exposure zone around roads can
vary considerably depending on the pollutant, traffic volume,
ambient pollution concentrations, meteorologic conditions,
topography, and land use (5). Traffic exposure metrics in the
published literature have used a variety of different density and
distance cut-points (6). Nevertheless, numerous epidemiologic
studies have consistently demonstrated that living close to
major roads or in areas of high traffic density is associated with

adverse health effects, including asthma, chronic obstructive
pulmonary disease, and other respiratory symptoms (11–15);
cardiovascular disease risk and outcomes (16–20); adverse
reproductive outcomes (21,22); and mortality (23–25).
Some studies have observed a dose-response gradient such
that living closer to major roads is associated with increased
risk (13,14,16–18). In terms of traffic density, several studies
have reported adverse health effects associated with residential
proximity to roads with average daily traffic volume as low as
10,000 vehicles per day (6,11,15–17).

In the United States, it is widely accepted that economically
disadvantaged and minority populations share a disproportionate
burden of air pollution exposure and risk (26,27). A growing body
of evidence demonstrates that minority populations and persons of
lower socioeconomic status experience higher residential exposure
to traffic and traffic-related air pollution than nonminorities and
persons of higher socioeconomic status (5,28–31). Two recent
studies have confirmed that these racial/ethnic and socioeconomic
disparities also exist on a national scale (32,33).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (34) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (35). This report provides descriptive
data on residential proximity to major highways, a topic that
was not discussed in the 2011 CHDIR. The purposes of this
report are to discuss and raise awareness of the characteristics of
persons exposed to traffic-related air pollution and to prompt
actions to reduce disparities.

Methods
To characterize the U.S. population living close to major

highways, CDC examined data from several sources using
Geographical Information Systems (GIS). Three data
sources were used for this assessment: 1) the 2010 U.S.
census (available at http://www.census.gov/2010census),

Residential Proximity to Major Highways — United States, 2010
Tegan K. Boehmer, PhD1

Stephanie L. Foster, MPH2
Jeffrey R. Henry, BA2

Efomo L. Woghiren-Akinnifesi2
Fuyuen Y. Yip, PhD1

1National Center for Environmental Health, CDC
2Agency for Toxic Substances and Disease Registry

Corresponding author: Tegan K. Boehmer, Division of Environmental Hazards and Health Effects, CDC. Telephone: 770-488-3714; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 47

2) 2006–2010 American Community Survey (ACS) 5-year
estimates (available at http://www.census.gov/acs), and 3) 2010
(Quarter 3) road network data from NAVTEQ, a commercial
data source that provides comprehensive road information for
the United States (available at http://www.navteq.com). Seven
sociodemographic variables were examined. Data on age, sex,
and race/ethnicity were obtained from the 2010 census; data
on nativity, language spoken at home, educational attainment,
and poverty status were obtained from the ACS.

The U.S. Census Bureau collects data on race and ethnicity
(i.e., Hispanic origin) as two separate questions. For this
analysis, persons of non-Hispanic ethnicity were classified as
white, black, Asian/Pacific Islander, American Indian/Alaska
Native, other race, and multiple races. Persons of Hispanic
ethnicity, who might be of any race or combination of races,
were grouped together as a single category. Educational
attainment was defined as less than high school, high school
graduate or equivalent, some college, or college graduate. For
the variable nativity, “native born” includes U.S. citizens born
abroad (one or both of whose parents were citizens at the
time of birth) and anyone born in the United States or a U.S.
territory; “foreign-born” denotes persons who were not U.S.
citizens at birth. Poverty status was categorized by using the
ratio of income to the federal poverty level (FPL), in which
“poor” is <1.0 times FPL, “near poor” is 1.0–2.9 times FPL,
and “nonpoor” is ≥3.0 times FPL.

Major highways were defined as interstates (Class 1) or as
other freeways and expressways (Class 2) based on the Federal
Highway Administration (FHWA) Functional Classification
system. These road types represent the most heavily-trafficked,
controlled-access highways in the United States. Although traffic
volume is not factored directly into the Functional Classification
system, FHWA statistics indicate that the majority of major
highways have average daily traffic volumes exceeding 10,000
vehicles per day (i.e., 77% of rural interstates have >10,000
vehicles per day and >72% of urban interstates and other
freeways and expressways have >30,000 vehicles per day) (36).

The census tract is the smallest geographic unit of analysis
available for the variables of interest in the ACS data. ESRI
ArcGIS v10 GIS software was used to create circular buffers of
150 meters around all major highways, and the proportion of
each census tract included within the buffer area was calculated.
This area proportion was then applied to the census tract-level
data from the 2010 census and ACS to estimate the number
of persons living within 150 meters of a major highway for
the total population and by sociodemographic characteristics.
Census tract count estimates were summed to obtain state and
national estimates. The proportion of the population living
within 150 meters of a major highway was calculated for
each category of the seven sociodemographic variables, using

category-specific denominators derived from the 2010 census
and ACS. No sampling error is associated with the 100%
population counts obtained from the 2010 census. Standard
errors were not calculated for the estimated population counts
derived from the ACS because of the complexity of the
GIS analysis used to generate these data. Therefore, for this
descriptive analysis, no statistical testing or calculation of 95%
confidence intervals was conducted, and it was not possible
to determine if the observed differences across population
subgroups are statistically significant.

Results
Approximately 11.3 million persons (or 3.7% of the 308.7

million U.S. population) live within 150 meters of a major
highway. State-level estimates ranged from 1.8% in Maine
to 5.6% in New York (Figure). Regional patterns, based on
U.S. Census Bureau groupings, indicate that the estimated
proportion of the population living within 150 meters of a
major highway ranged from 3.1% in the Midwest and 3.3%
in the South to 4.3% in the Northeast and 4.4% in the West.
The proportion of the population living near a major highway
did not differ by sex (Table). By age group, the estimated
proportion of persons living close to a major highway varied
from 3.4% among those aged 45–79 years to ≥4.0% among
those aged 18–34 years.

The greatest disparities were observed for race/ethnicity,
nativity, and language spoken at home; the populations with
the highest estimated percentage living within 150 meters of a
major highway included members of racial and ethnic minority
communities, foreign-born persons, and persons who speak a
language other than English at home (Table). The estimated
percentage of the population living within 150 meters of
a major highway ranged from a low of 2.6% for American
Indians/Alaska Natives and 3.1% for non-Hispanic whites
to a high of 5.0% for Hispanics and 5.4% for Asians/Pacific
Islanders. Likewise, the estimated proportion of the population
living near a major highway was 5.1% for foreign-born persons,
5.1% for persons who speak Spanish at home, and 4.9% for
persons who speak another non-English language at home.

Disparities by educational attainment and poverty status
were less pronounced (Table). The estimated percentage of
the population living near a major highway varied from 3.4%
for high school graduates to 4.1% for those with less than a
high school diploma. A more consistent pattern was observed
for poverty status; the estimated proportion of the population
living near a major highway was 4.2% for those in the poor
category, 3.7% for those in the near-poor category, and 3.5%
for those in the nonpoor category.

Supplement

48 MMWR / November 22, 2013 / Vol. 62 / No. 3

Discussion
Overall, approximately 4% of the total U.S. population lives

within 150 meters of a major highway, suggesting increased
exposure to traffic-related air pollution and elevated risk for
adverse health outcomes. Estimates of residential proximity to
major roads are influenced by the number and type of roads
and the distance or buffer size used. In terms of quantifying the
total U.S. population exposed to traffic-related air pollution,
the estimate of 11.3 million people derived from this analysis
should be considered conservative because only interstates,
freeways, and expressways were included and a relatively small
buffer distance of 150 meters was used. These conditions
were selected to capture persons who are at the highest risk
for exposure to traffic-related air pollution. In addition, this
estimate is based on distance to a single road and does not
account for cumulative exposure to traffic from multiple roads.

The percentage of the population exposed to traffic-related
air pollution is expected to be larger in urban areas because
of higher population density, more roads, and higher traffic
volume. A case study of two North American cities (Los Angeles
County and Toronto, Canada) estimated that 30%–45% of
the population in these urban areas lives within 500 meters
of a highway or 50–100 meters of a major road (5). Although
this report does not address urban/rural differences directly,
an additional state-level analysis of these data indicated that
the percentage of the population living within 150 meters of
a major highway was correlated positively (R = 0.65) with the
percentage of the population living in urban areas. Additional

studies are needed to understand potential sociodemographic
disparities among populations living near major highways
across levels of urbanization.

This analysis suggests that social and demographic disparities
exist with respect to residential proximity to major highways.
Larger disparities were observed for indicators of minority

TABLE. Number and percentage of population living within 150
meters of a major highway, by selected characteristics — United
States, 2010

Characteristic No. (%)*

Total† 11,337,933 (3.7)
Sex†

Male 5,547,223 (3.7)
Female 5,790,844 (3.7)

Age group (yrs)†
0–4 766,603 (3.8)
5–9 727,279 (3.6)
10–17 1,168,995 (3.5)
18–24 1,219,887 (4.0)
25–34 1,714,903 (4.2)
35–44 1,523,607 (3.7)
45–64 2,808,121 (3.4)
65–79 977,948 (3.4)
≥80 412,215 (3.7)
Race/Ethnicity†

Non-Hispanic
White 6,030,811 (3.1)
Black 1,676,225 (4.4)
Asian/Pacific Islander 800,723 (5.4)
American Indian/Alaska Native 59,378 (2.6)
Other 27,239 (4.5)
Multiple race 235,995 (4.0)

Hispanic§ 2,502,616 (5.0)
Nativity¶

Native born** 9,172,481 (3.5)
Foreign born†† 1,966,763 (5.1)

Language spoken at home (≥5 yrs)¶
English only 7,513,304 (3.3)
Spanish 1,805,261 (5.1)
Other 1,059,572 (4.9)

Educational attainment (≥25 years)¶
Less than high school 1,225,735 (4.1)
High school graduate or equivalent 1,988,228 (3.4)
Some college 1,977,261 (3.5)
College graduate 2,092,232 (3.8)

Poverty status¶,§§
Poor (<1.0 times FPL) 1,733,031 (4.2)
Near-poor (1.0–2.9 times FPL) 3,882,694 (3.7)
Nonpoor (≥3.0 times FPL) 5,227,274 (3.5)

Abbreviation: FPL = federal poverty level.
* Denominator for overall population is 308,745,348. Percentages for all other

rows were calculated by using category-specific denominators.
† Source: U.S. Census Bureau, 2010 census (available at http://www.census.

gov/2010census).
§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Source: U.S. Census Bureau, 2006–2010 American Community Survey

(available at http://www.census.gov/acs).
** Includes U.S. citizens born abroad (one or both of whose parents were citizens

at the time of birth) and anyone born in the United States or a U.S. territory.
†† Persons who were not U.S. citizens at birth.
§§ Additional information is available at http://aspe.hhs.gov/poverty/figures-

fed-reg.cfm.

3.8–5.6
3.1–3.7
2.7–3.1
1.8–2.7

FIGURE. Percentage* of population living within 150 meters of a
major highway, by state — United States, 2010

* Calculated by dividing the population within 150 meters of a major highway
by the total population per state and multiplying by 100. The percentages are
displayed using quartiles.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 49

status (i.e., race/ethnicity, nativity, and language spoken
at home) than for traditional indicators of socioeconomic
status (i.e., poverty and educational attainment). Two
other national studies have reported similar findings using
alternative approaches. A study that examined the distribution
of sociodemographic variables across various traffic exposure
metrics assessed at the residential address found that race,
ethnicity, poverty status, and education all were associated
with one or more traffic exposure metrics (32). Another study
demonstrated that the correlation between traffic exposure
metrics and sociodemographic variables across all U.S. census
tracts was stronger for race and ethnicity than it was for
poverty, income, and education and that the magnitude of the
correlations varied spatially by region and state (33).

The environmental justice literature suggests that socially
disadvantaged groups might experience a phenomenon
known as “triple jeopardy” (37). First, poor and minority
groups are known to suffer negative health effects from social
and behavioral determinants of health (e.g., psychosocial
stress, poor nutrition, and inadequate access to health care).
Second, as suggested in this analysis, certain populations (e.g.,
members of minority communities, foreign-born persons, and
persons who speak a non-English language at home) might be
at higher risk for exposure to traffic-related air pollution as a
result of residential proximity to major highways. Third, there
is evidence suggesting a multiplicative interaction between
the first two factors, such that socially disadvantaged groups
experience disproportionately larger adverse health effects from
exposure to air pollution (37–39).

Limitations
The findings in this report are subject to at least three

limitations. First, the area-proportion technique used
assumes a homogeneous population density and population
distribution by sociodemographic characteristics within each
census tract, which might result in erroneous count estimates.
The direction of the bias (overestimate or underestimate)
could differ across population subgroups. For example, if
socioeconomic disparities associated with residential proximity
to major highways exist within census tracts, then the calculated
percentages for minority subgroups might be underestimated
and those for nonminority subgroups might be overestimated.
Second, living within 150 meters of a major highway is only a
surrogate for exposure to traffic-related air pollution. This study
did not address the following factors that could affect exposure
to traffic-related air pollution: number and type of vehicles
traveling on major highways, cumulative effect of living near
multiple roads, individual time-activity patterns (e.g., time

spent at home vs. away, time spent inside vs. outside),
meteorologic conditions, topography, and land-use patterns.
Finally, it was not possible to perform testing to determine if
the differences in the estimated percentages across population
subgroups were statistically significant. However, the findings
are consistent with other published research (32,33).

Conclusion
Primary prevention strategies to reduce traffic emissions

include improving access to alternative transportation options
(e.g., transit, rideshare programs, walking, and cycling),
financial incentives to reduce vehicle miles traveled and
congestion, diesel retrofitting, and promoting the use of
electric and low emission vehicles. In addition, secondary
prevention strategies to reduce exposure to traffic emissions
include mitigation techniques for existing homes and
buildings (e.g., roadside barriers and improved ventilation
systems) and land-use policies that limit new development
close to heavily-trafficked roads. For example, a recent study
of roadside barriers suggests that solid barriers (i.e., noise
barriers) might be more effective at mitigating traffic-related
air pollution than vegetative barriers (i.e., tree stands) (41). In
California, public health law has been used to restrict siting
of new schools near major highways and busy traffic corridors
(California Education Code §7213.c.2.C). Implementation
of these strategies can help reduce exposures to traffic-related
air pollution and health risks associated with these exposures.

Focusing prevention and mitigation interventions in urban
areas, where there is a higher concentration of traffic-related
air pollution and a greater proportion of the population
residing near major roads, and in areas with the most socially
disadvantaged populations will likely result in larger health
benefits (37). Future and ongoing efforts to address disparities
in residential proximity to major highways and traffic-related
air pollution exposures will require an interdisciplinary
collaboration between transportation, urban planning, and
public health specialists.

References
1. US Environmental Protection Agency. Integrated science assessment for

carbon monoxide. Report No. EPA/600/R-09/019F. Research Triangle
Park, NC: US Environmental Protection Agency; 2010.

2. US Environmental Protection Agency. Integrated science assessment for
particulate matter. Report No. EPA/600/R-08/139F. Research Triangle
Park, NC: US Environmental Protection Agency; 2009.

3. US Environmental Protection Agency. Integrated science assessment for
oxides of nitrogen—health criteria. Report No. EPA/600/R-08/071.
Research Triangle Park, NC: US Environmental Protection Agency; 2008.

4. US Environmental Protection Agency. Air quality criteria for ozone and
related photochemical oxidants. Report No. EPA/600/R-05/004aF.
Research Triangle Park, NC: US Environmental Protection Agency; 2006.

Supplement

50 MMWR / November 22, 2013 / Vol. 62 / No. 3

5. HEI Panel on the Health Effects of Traffic-Related Air Pollution. Traffic-
related air pollution: a critical review of the literature on emissions,
exposure, and health effects. HEI Special Report 17. Boston, MA: Health
Effects Institute; 2010.

6. Boothe VL, Shendell DG. Potential health effects associated with
residential proximity to freeways and primary roads: review of scientific
literature, 1999–2006. J Environ Health 2008;70:33–41.

7. Brugge D, Durant JL, Rioux C. Near-highway pollutants in motor
vehicle exhaust: a review of epidemiologic evidence of cardiac and
pulmonary health risks. Environ Health 2007;6:23.

8. Rioux CL, Gute DM, Brugge D, Peterson S, Parmenter B. Characterizing
urban traffic exposures using transportation planning tools: an illustrated
methodology for health researchers. J Urban Health 2010;87:167–88.

9. Gilbert NL, Woodhouse S, Stieb DM, Brook JR. Ambient nitrogen
dioxide and distance from a major highway. Sci Total Environ 2003;
312:43–6.

10. Zhu YF, Hinds WC, Kim S, Sioutas C. Concentration and size
distribution of ultrafine particles near a major highway. J Air Waste
Manag Assoc 2002;52:1032–42.

11. Garshick E, Laden F, Hart JE, et al. Residence near a major road and
respiratory symptoms in U.S. Veterans. Epidemiology 2003;14:728–36.

12. Gauderman WJ, Vora H, McConnell R, et al. Effect of exposure to
traffic on lung development from 10 to 18 years of age: a cohort study.
Lancet 2007;369(9561):571–7.

13. Kim JJ, Huen K, Adams S, et al. Residential traffic and children’s
respiratory health. Environ Health Perspect 2008;116:1274–9.

14. McConnell R, Berhane K, Yao L, et al. Traffic, susceptibility, and
childhood asthma. Environ Health Perspect 2006;114:766–72.

15. Schikowski T, Sugiri D, Ranft U, et al. Long-term air pollution exposure
and living close to busy roads are associated with COPD in women.
Respir Res 2005;6:152.

16. Hoffmann B, Moebus S, Mohlenkamp S, et al. Residential exposure to
traffic is associated with coronary atherosclerosis. Circulation
2007;116:489–96.

17. Hoffmann B, Moebus S, Stang A, et al. Residence close to high traffic and
prevalence of coronary heart disease. Eur Heart J 2006;27:2696–702.

18. Kan H, Heiss G, Rose KM, Whitsel EA, Lurmann F, London SJ.
Prospective analysis of traffic exposure as a risk factor for incident
coronary heart disease: the Atherosclerosis Risk in Communities (ARIC)
study. Environ Health Perspect 2008;116:1463–8.

19. Tonne C, Melly S, Mittleman M, et al. A case-control analysis of exposure
to traffic and acute myocardial infarction. Environ Health Perspect
2007;115:53–7.

20. Van Hee VC, Adar SD, Szpiro AA, et al. Exposure to traffic and left
ventricular mass and function: the Multi-Ethnic Study of Atherosclerosis.
Am J Respir Crit Care Med 2009;179:827–34.

21. Wilhelm M, Ritz B. Residential proximity to traffic and adverse birth
outcomes in Los Angeles county, California, 1994–1996. Environ Health
Perspect 2003;111:207–16.

22. Yorifuji T, Naruse H, Kashima S, et al. Residential proximity to major
roads and preterm births. Epidemiology 2011;22:74–80.

23. Gehring U, Heinrich J, Kramer U, et al. Long-term exposure to ambient
air pollution and cardiopulmonary mortality in women. Epidemiol
2006;17:545–51.

24. Jerrett M, Finkelstein MM, Brook JR, et al. A cohort study of traffic-
related air pollution and mortality in Toronto, Ontario, Canada. Environ
Health Perspect 2009;117:772–7.

25. Medina-Ramon M, Goldberg R, Melly S, et al. Residential exposure to
traffic-related air pollution and survival after heart failure. Environ
Health Perspect 2008;116:481–5.

26. Bell ML, O’Neill MS, Cifuentes L A, et al. Challenges and
recommendations for the study of socioeconomic factors and air
pollution health effects. Environ Sci Pol 2005;8:525–33.

27. O’Neill MS, Jerrett M, Kawachi L, et al. Health, wealth, and air
pollution: advancing theory and methods. Environ Health Perspect
2003;111:1861–70.

28. Apelberg BJ, Buckley TJ, White RH. Socioeconomic and racial disparities
in cancer risk from air toxics in Maryland. Environ Health Perspect
2005;113:693–9.

29. Brender JD, Maantay JA, Chakraborty J. Residential proximity to
environmental hazards and adverse health outcomes. Am J Public Health
2011;101:S37–52.

30. Chakraborty J. Automobiles, air toxics, and adverse health risks:
environmental inequities in Tampa Bay, Florida. Ann Assoc Am Geogr
2009;99:674–97.

31. Gunier RB, Hertz A, Von Behren J, Reynolds P. Traffic density in
California: socioeconomic and ethnic differences among potentially
exposed children. J Expo Anal Environ Epidemiol 2003;13:240–46.

32. Parker JD, Kravets N, Nachman K, Sapkota A. Linkage of the 1999–2008
National Health and Nutrition Examination Surveys to traffic indicators
from the National Highway Planning Network. National health statistics
reports; No. 45. Hyattsville, MD: US Department of Health and Human
Services, CDC, National Center for Health Statistics; 2012.

33. Tian N, Xue J, Barzyk TM. Evaluating socioeconomic and racial
differences in traffic-related metrics in the United States using a GIS
approach. J Expo Sci Environ Epidemiol 2013;23:215–22.

34. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

35. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

36. US Department of Transportation. Highway statistics 2008. Table
HM-57: length by average daily traffic volume, arterials and collectors.
Washington, DC: Federal Highway Administration; 2009. Available at
http://www.fhwa.dot.gov/policyinformation/statistics/2008.

37. Jerrett M, Burnett RT, Kanaroglou P, et al. A GIS- environmental justice
analysis of particulate air pollution in Hamilton, Canada. Environment
and Planning A 2001;33:955–73.

38. Cakmak S, Dales RE, Rubio MA, Vidal CB. The risk of dying on days
of higher air pollution among the socially disadvantaged elderly. Environ
Res 2011;111:388–93.

39. Ou C-Q, Hedley AJ, Chung RY, et al. Socioeconomic disparities in air
pollution-associated mortality. Environ Res 2008;107:237–44.

40. Panel on the Functionality and Usability of Data from the American
Community Survey. Using the American Community Survey: benefits
and challenges. Washington, DC: National Academies Press; 2007.

41. Hagler GSW, Lin MY, Khlystov A, et al. Field investigation of roadside
vegetative and structural barrier impact on near-road ultrafine particle
concentrations under a variety of wind conditions. Sci Total Environ
2012;419:7–15.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 51

Health-Care Access and Preventive Services

Supplement

52 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 53

Introduction
Colorectal cancer (CRC) is the second leading cause of

cancer-related deaths in the United States among cancers that
affect both men and women (1). Screening for CRC reduces
incidence and mortality (2). In 2008, the U.S. Preventive
Services Task Force (USPSTF) recommended that persons aged
50–75 years at average risk for CRC be screened for the disease
by using one or more of the following methods: fecal occult
blood testing (FOBT) every year, sigmoidoscopy every 5 years
(with high-sensitivity FOBT every 3 years), or colonoscopy
every 10 years (2).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (3) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (4). This report updates information
regarding CRC screening provided in the 2011 CHDIR (5).
The purposes of this report are to discuss and raise awareness
of differences in colorectal cancer incidence, mortality, and
screening and to prompt actions to reduce these disparities.

Methods
To characterize disparities for CRC incidence, CRC death

rates, and CRC screening test use by test type, CDC analyzed
data from multiple sources and years. Different analytic
approaches were used to characterize disparities depending
on the data source (i.e., deviation from referent group or
comparison of weighted estimates and confidence intervals
[CIs]).

To describe CRC incidence and death rates, CDC analyzed
2008 CRC incidence and mortality data from U.S. Cancer
Statistics (USCS) (1). Demographic characteristics analyzed
included sex, age, race and ethnicity. Data on household
income and educational attainment are not collected by
cancer registries. Race was classified as non-Hispanic white,

non-Hispanic black, Asian/Pacific Islander, or American
Indian/Alaska Native. Ethnicity was classified as Hispanic or
non-Hispanic; persons of Hispanic ethnicity might be of any
race or combination of races. Incidence data were drawn from
CDC’s National Program of Cancer Registries (NPCR) and
the National Cancer Institute’s Surveillance, Epidemiology, and
End Results Program (SEER) registries that met U.S. Cancer
Statistics publication criteria for the diagnosis year 2008, and
mortality data were derived from the National Vital Statistics
System. In 2008, all 50 states and the District of Columbia had
high-quality incidence and mortality data available, and thus
100% of the U.S. population is represented for both. Incident
CRCs were coded* according to the International Classification
of Diseases for Oncology, Third Edition (ICD-O-3). All death
certificates with CRC identified as the underlying cause of
death according to the International Classification of Diseases,
Tenth Revision (ICD-10) during 2008 were included in this
analysis. Incidence and death rates were calculated for all age
groups using SEER*Stat software (version 7.04); rates were
reported per 100,000 population. Data were age-adjusted
to the 2000 U.S. standard population by the direct method;
corresponding 95% CIs were calculated as modified gamma
intervals (6).

Disparities were measured as the deviations from a “referent”
category rate. Absolute difference was measured as the simple
difference between a population subgroup estimate and the
estimate for its respective reference group. The relative difference,
a percentage, was calculated by dividing the difference by the
value in the referent category and multiplying by 100.

To assess disparities in CRC screening test use by test
type, CDC analyzed 2010 survey data from the Behavioral
Risk Factor Surveillance System (BRFSS). BRFSS is a
state-based, random digit-dialed telephone survey of the
noninstitutionalized, U.S. civilian population aged ≥18 years
(7). Survey data were available for all 50 states and the District

* Malignant behavior, ICDO3 site codes 18.0–18.9, 19.9, 20.9, and 26.0;
excludes histology codes for lymphomas, mesothelioma, and Kaposi Sarcoma
(9050–9055, 9140, and 9590–9989).

Colorectal Cancer Incidence and Screening —
United States, 2008 and 2010

C. Brooke Steele, DO
Sun Hee Rim, MPH

Djenaba A. Joseph, MD
Jessica B. King, MPH
Laura C. Seeff, MD

National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: C. Brooke Steele, Division of Cancer Prevention and Control, CDC. Telephone: 770-488-4261; E-mail: [email protected].

Supplement

54 MMWR / November 22, 2013 / Vol. 62 / No. 3

of Columbia. In 2010, the median response rate was 54.6%,
and the median cooperation rate was 76.9% (7). Respondents
who refused to answer, had a missing answer, or did not know
the answer to a question were excluded from analysis of that
specific question. Of the 226,205 persons aged 50–75 years
who responded in 2010, approximately 4.2% were excluded
from the analyses.

Demographic characteristics from BRFSS that were analyzed
included sex, age, race, ethnicity, educational attainment,
income level, disability status, health insurance status, and
geographic location. Race was classified as non-Hispanic white,
non-Hispanic black, Asian/Pacific Islander, American Indian/
Alaska Native, or other non-Hispanic. Ethnicity was classified
as Hispanic or non-Hispanic; persons of Hispanic ethnicity
might be of any race or combination of races. Educational
attainment was classified as less than high school, some high
school, high school graduate or equivalent, some college/
technical school, or college graduate. Income level was classified
as <$15,000, $15,000–$34,999, $35,000–$49,999, $50,000–
$74,999, and ≥$75,000.The median response rate† and the
median cooperation rate§ are based on Council of American
Survey and Research Organizations guidelines (available at
http://www.cdc.gov/brfss/annual_data/annual_2010.htm).

BRFSS respondents aged 50–75 years, the age group for which
USPSTF recommends CRC screening, were asked if they had
ever used a “special kit at home to determine whether the stool
contains blood (FOBT),” whether they had ever had “a tube
inserted into the rectum to view the colon for signs of cancer
or other health problems (sigmoidoscopy or colonoscopy),” and
when these tests were last performed. To allow assessment of
up-to-date screening according to current USPSTF guidelines,
the measure of overall screening prevalence used in the 2011
CHDIR (5) was modified. Percentages were estimated for
persons aged 50–75 years who reported receiving an FOBT
within 1 year, a sigmoidoscopy within 5 years with FOBT
within 3 years, or a colonoscopy within 10 years preceding
the survey. Data for the three recommended test options were
combined to estimate overall prevalence of up-to-date CRC
screening. States were categorized into four poverty quartiles by
using data from the 2010 Current Population Survey (available
at http://cps.ipums.org/cps), and composite screening rates per
quartile were calculated. Composite percentages and 95% CIs
were calculated by selected characteristics. Data were weighted
according to the sex, racial/ethnic, and age distribution of the
adult population of each state by using intercensal estimates

and were age standardized to the 2010 BRFSS population
aged 50–75 years.

Results
Compared with women, men had higher CRC incidence

rates (51.6 versus 38.7 per 100,000 population) and death rates
(19.7 vs. 13.8 per 100,000 population) in 2008 (Table 1). CRC
incidence and mortality increased with advancing age (Figure).
Incidence and death rates were highest among persons aged
≥75 years. Non-Hispanic blacks had higher CRC incidence and
death rates than non-Hispanic whites, Asians/Pacific Islanders,
and American Indians/Alaska Natives. Incidence and death
rates were higher among non-Hispanics than among Hispanics.

In 2010, among respondents aged 50–75 years, 64.5%
reported being up-to-date with CRC screening (Table 2).
The proportion of respondents who reported having had any
of the test options was greater among persons aged 65–75
years compared with those aged 50–64 years, among non-
Hispanics compared with Hispanics, among persons with
a disability compared with those with no disability, and
among persons with health insurance compared with those
with no health insurance. This disparity in reported test use
by health insurance status was evident for all three test types
(FOBT, sigmoidoscopy with FOBT, and colonoscopy). The
proportions for colonoscopy use and for overall CRC screening
were slightly greater among women than among men. Reported
rates of test use increased with increasing education level
and household income, with the greatest increases occurring
among those who reported having had a colonoscopy within
10 years preceding the survey. The prevalence of respondents
who were up-to-date with CRC screening was highest
among non-Hispanic whites (66.4%), followed closely by
non-Hispanic blacks (64.8%). Non-Hispanic whites had the
greatest proportion of respondents reporting having had a
colonoscopy within 10 years preceding the survey compared
with all other races; non-Hispanic blacks had the greatest
proportion of respondents reporting having had FOBT within
the year preceding the survey.

By composite state poverty quartiles, the relationship
between reported screening rates and poverty varied by
test type. No consistent relationship was observed between
poverty and reported use of FOBT; however, the number of
respondents who reported use of FOBT testing was small
(Table 3). An inverse relationship was observed for reported
use of colonoscopy and poverty, with reported colonoscopy use
generally decreasing with increasing levels of poverty (Table 3).

† The percentage of persons who completed interviews among all eligible persons,
including those who were not contacted successfully.

§ The percentage of persons who completed interviews among all eligible persons
who were contacted.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 55

Discussion
CRC incidence and death rates were

higher among older, male, and non-Hispanic
populations. CRC incidence and death rates
for many of these groups exceeded Healthy
People 2020 targets of 38.6 new CRC cases
per 100,000 population and 14.5 CRC
deaths per 100,000 population (8). Progress
in reducing deaths from CRC has been
achieved through a combination of primary
prevention, early detection, and treatment (9).
In 2010, approximately two thirds of the U.S.
population aged 50–75 years met USPSTF
criteria for up-to-date CRC screening. The
proportion screened in a timely manner varied
by race and other demographic characteristics.

Although estimates of the overall prevalence
of up-to-date CRC screening in this report
and in the 2011 report were computed
differently, certain patterns were similar.
The 2011 report analyzed BRFSS data
for 2002–2008. During that time period,
non-Hispanic whites had the highest overall
prevalence of CRC screening, followed closely
by non-Hispanic blacks (3). The same finding

TABLE 1. Colorectal cancer incidence and death rates,* by selected demographic characteristics — United States, 2008†

Characteristic

Incidence Absolute
difference

(Rate)

Relative
difference

(%)

Death Absolute
difference

(Rate)

Relative
difference

(%)Rate (95% CI) Rate (95% CI)

Sex 
Male 51.6 (51.2–52.0) 12.9 33.3 19.7 (19.4–19.9) 5.9 42.8
Female 38.7 (38.4–39.0) Ref. Ref. 13.8 (13.6–14.0) Ref. Ref.

Age group (yrs)
<50 6.7 (6.6–6.8) Ref. Ref. 1.7 (1.7–1.8) Ref. Ref.
50–54 55.7 (54.7–56.7) 49.0 731.3 13.4 (12.9–13.9) 11.7 688.2
55–59 72.7 (71.5–73.9) 66.0 985.1 21.6 (20.9–22.3) 19.9 1,170.6
60–64 101.2 (99.6–102.9) 94.5 1,410.4 32.7 (31.8–33.6) 31.0 1,823.5
65–69 152.2 (149.9–154.5) 145.5 2,171.6 48.7 (47.4–50.0) 47.0 2,764.7
70–74 199.0 (196.0–201.9) 192.3 2,870.1 70.1 (68.3–71.8) 68.4 4,023.5
≥75 283.5 (281.1–285.9) 276.8 4,131.3 134.0 (132.4–135.7) 132.3 7,782.4
Race

White, non-Hispanic 43.8 (43.6–44.1) Ref. Ref. 16.1 (15.9–16.2) Ref. Ref.
Black, non-Hispanic 53.9 (53.0–54.7) 10.1 23.1 23.5 (22.9–24.1) 7.4 46.0
Asian/Pacific Islander 34.5 (31.8–37.3) -9.3 -21.2 15.9 (14.1–17.9) -0.2 -1.2
American Indian/Alaska Native 35.1 (34.1–36.2) -8.7 -19.9 11.5 (10.9–12.1) -4.6 -28.6

Ethnicity
Hispanic§ 37.8 (37.0–38.6) Ref. Ref. 12.1 (11.7–12.6) Ref. Ref.
Non-Hispanic 45.0 (44.8–45.2) 7.2 19.0 16.7 (16.5–16.8) 4.6 38.0

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Per 100,000 population
† Rates are age-adjusted to the U.S. Census Bureau’s 2000 US Standard Population for 19 age groups (available at http//seer.cancer.gov/stdpopulations/stdpop.19ages.

html). Incidence data come from CDC’s National Program of Cancer Registries (NPCR) and the National Cancer Institute’s Surveillance, Epidemiology, and End Results
Program (SEER) registries that met U.S. Cancer Statistics publication criteria for diagnosis year 2008 and cover 100% of the U.S. population. Underlying mortality
data are provided by the National Vital Statistics System and cover 100% of the U.S. population.

§ Persons of Hispanic ethnicity might be of any race or combination of races.

FIGURE. Colorectal cancer incidence and mortality rates per 100,000 population, by age
group — United States, 2008*

* Rates are age-adjusted to the 2000 U.S.Census Bureau Standard Population for 19 age groups (available
at http://seer.cancer.gov/stdpopulations/stdpop.19ages.html). Incidence data come from from CDC’s
National Program of Cancer Registries (NPCR) and the National Cancer Institute’s Surveillance,
Epidemiology, and End Results Program (SEER) registries that met U.S. Cancer Statistics publication
criteria for diagnosis year 2008 and cover 100% of the U.S. population. Underlying mortality data are
provided by the National Vital Statistics System and cover 100% of the U.S. population.

Incidence

0

50

100

150

200

250

300

<50 50–54 55–59 60–64 65–69 70–74 ≥75

R
at

e

Age group (yrs)

Mortality

Supplement

56 MMWR / November 22, 2013 / Vol. 62 / No. 3

was observed in 2010. American Indians/Alaska Natives and
Hispanics had lower CRC screening rates in 2002–2008
than non-Hispanic blacks. This disparity persisted in 2010.
The pattern was less consistent for the Asian/Pacific Islander
population; in 2002 and 2004, their overall prevalence of
up-to-date CRC screening was substantially lower than the
prevalence for non-Hispanic whites. The gap narrowed in
2006 and 2008 but widened in 2010.

In 2010, CRC screening test use increased with age,
educational level, and household income level. The demographic
disparities were greater for colonoscopy than for sigmoidoscopy
and FOBT. Similar findings were reported in the 2011 report
(3) and in other previous studies (10–12). Having health
insurance is also a strong predictor of screening for colorectal

cancer (11,13,14). Disparities in the overall prevalence of
up-to-date CRC screening by health insurance status were
observed in 2008 (5) and 2010. Screening rates among insured
respondents were 66.6 in 2008 and 67.5 in 2010. Rates
among uninsured respondents were 37.5 in 2008 and 35.4 in
2010. Medicare has covered CRC screening for enrollees since
2001. Although this expansion of cancer screening coverage
has increased CRC screening among older persons, persistent
racial/ethnic, socioeconomic, and geographic disparities in test
use have been reported among persons aged ≥65 years (15–17).
Among younger adults, those with lower incomes and less
than a high school education are less likely to have health-care
insurance than those with higher incomes and at least some
college education (18). For many patients, implementation of

TABLE 2. Percentage* of respondents aged 50–75 years who reported being up-to-date with colorectal cancer screening, by selected
characteristics and test type — Behavioral Risk Factor Surveillance System, United States, 2010

Characteristic

FOBT within 1 yr

Flexible sigmoidoscopy
within 5 yrs with FOBT

within 3 years
Colonoscopy
within 10 yrs Total CRC screening†

% (95% CI) % (95% CI) % (95% CI) % (95% CI)

Sex
Male 12.4 (12.0–12.8) 1.4 (1.3–1.5) 59.6 (59.0–60.2) 64.0 (63.4–64.6)
Female 11.1 (10.9–11.4) 1.2 (1.1–1.3) 60.9 (60.4–61.3) 64.9 (64.5–65.4)

Age group (yrs)
50–64 10.3 (10.1–10.6) 1.0 (0.9–1.1) 55.4 (55.0–55.9) 59.7 (59.2–60.1)
65–75 15.1 (14.7–15.6) 1.9 (1.8–2.1) 71.9 (71.3–72.4) 76.1 (75.6–76.7)

Race
White, non-Hispanic 11.3 (11.1–11.6) 1.2 (1.2–1.4) 62.5 (62.1–62.9) 66.4 (66.0–66.8)
Black, non-Hispanic 15.1 (14.2–16.1) 1.4 (1.1–1.7) 59.8 (58.5–61.1) 64.8 (63.6–66.1)
Asian/Pacific Islander 12.5 (10.5–14.7) 1.6 (0.9–2.7) 49.3 (45.9–52.6) 54.4 (51.0–57.8)
American Indian/Alaska Native 14.6 (12.1–17.6) 0.9 (0.5–1.9) 48.9 (45.0–52.8) 55.2 (51.3–59.1)
Other, non-Hispanic 13.5 (11.9–15.4) 2.1 (1.5–3.0) 55.1 (52.4–57.7) 61.3 (58.7–63.8)

Ethnicity
Non-Hispanic 11.8 (11.6–12.1) 1.3 (1.2–1.4) 61.6 (61.2–61.9) 65.7 (65.3–66.1)
Hispanic§ 10.7 (9.6–11.8) 1.2 (0.9–1.7) 45.4 (43.6–47.3) 51.0 (49.1–52.9)

Educational attainment
Less than high school 8.3 (7.1–9.7) 0.7 (0.4–1.2) 34.6 (32.2–37.0) 39.2 (36.7–41.7)
Some high school 10.4 (9.5–11.5) 0.9 (0.6–1.3) 44.3 (42.7–46.0) 49.4 (47.7–51.1)
High school graduate or equivalent 11.0 (10.6–11.5) 0.9 (0.8–1.1) 54.9 (54.2–55.6) 59.3 (58.6–60.0)
Some college/technical school 12.3 (11.9–12.8) 1.4 (1.3–1.6) 61.2 (60.5–61.9) 65.7 (65.0–66.3)
College graduate 12.5 (12.1–12.9) 1.7 (1.5–1.8) 68.3 (67.7–68.9) 72.0 (71.4–72.6)

Income level
<$15,000 11.2 (10.4–12.0) 0.9 (0.7–1.2) 42.3 (41.0–43.6) 47.7 (46.4–49.0)
$15,000–$34,999 11.6 (11.1–12.1) 1.1 (1.0–1.3) 50.9 (50.2–51.7) 56.0 (55.2–56.8)
$35,000–$49,999 12.0 (11.4–12.7) 1.5 (1.3–1.8) 60.5 (59.5–61.5) 65.0 (64.0–65.9)
$50,000–$74,999 12.0 (11.4–12.6) 1.3 (1.1–1.6) 65.1 (64.2–66.0) 68.9 (68.0–69.7)
≥$75,000 12.1 (11.7–12.6) 1.7 (1.5–1.9) 69.9 (69.2–70.7) 73.4 (72.7–74.1)

Disability status
Has a disability 12.5 (12.1–12.9) 1.4 (1.2–1.6) 61.7 (61.1–62.4) 66.3 (65.7–67.0)
Does not have a disability 11.5 (11.2–11.7) 1.2 (1.2–1.4) 59.7 (59.2–60.2) 63.8 (63.3–64.3)

Health insurance status
Has health insurance 12.2 (11.9–12.4) 1.4 (1.3–1.5) 63.3 (62.9–63.7) 67.5 (67.2–67.9)
Does not have health insurance 7.9 (6.8–9.1) 0.4 (0.3–0.6) 31.6 (29.7–33.5) 35.4 (33.5–37.5)

Total 11.7 (11.5–12.0) 1.3 (1.2–1.4) 60.2 (59.9–60.6) 64.5 (64.1–64.8)

Abbreviations: 95% CI = 95% confidence interval; CRC = colorectal cancer; FOBT = fecal occult blood testing.
* Percentages standardized to age distribution in the 2010 Behavioral Risk Factor Surveillance System.
† Home FOBT within the past year, flexible sigmoidoscopy within the past 5 years with FOBT within the past 3 years, or colonoscopy within the past 10 years.
§ Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 57

the Affordable Care Act has removed financial barriers to CRC
screening by mandating that nongovernmental health plans
cover certain preventive health services without cost-sharing
requirements (19).

FOBT and sigmoidoscopy screening rates were low in
2010. Previous studies have noted a continued decline in use
of these tests (12,20–22), despite the fact that screening with
each has been reported to be associated with reduced mortality
from CRC (23,24). Some primary care physicians perceive
FOBT and sigmoidoscopy to be less effective in reducing
CRC mortality than colonoscopy, which might influence
which tests they recommend to their patients (25–27). Studies
indicate that some patients prefer FOBT, and discordance
between physician and patient preferences might affect uptake
of CRC screening (28–30). Discussing multiple options
for CRC screening with patients and acknowledging their
preferences when recommendations are made could contribute
to improved completion of testing.

State-level poverty percentages were less clearly associated
with use of FOBT compared with colonoscopy. During
2002–2008, screening with annual FOBT or lower endoscopy

within 10 years was related to both income and state poverty
levels (3). In previous studies of trends in CRC screening,
changes in FOBT and colonoscopy use varied substantially by
income level, health insurance status, race/ethnicity, and other
demographic characteristics (31,32). Additional studies are
needed to investigate the effects of poverty and sociocultural
indicators on test use, independent of insurance status.

CDC has funded activities to improve CRC screening
rates, including efforts to increase access to screening for
underserved populations. The Colorectal Cancer Control
Program (CRCCP) was established in 2009 following the
successful implementation of a CRC screening demonstration
program in five sites across the country (33). CRCCP funds
25 states and four tribes, with the goal of increasing screening
rates among those aged 50–75 years to 80% in funded states
(http://www.cdc.gov/cancer/colorectal). Approximately one
third of funds are used to provide direct screening services and
follow-up care to low-income men and women aged 50–64
years who are underinsured or uninsured. The majority of
funds are used to promote and implement evidence-based
strategies recommended by the Task Force on Community

See table footnotes on the next page.

TABLE 3. Percentage of persons reporting having had colorectal cancer screening, by state poverty-level quartiles* — Behavioral Risk Factor
Surveillance System, United States, 2010

State/Area

Population
in poverty

(%)

FOBT within past 1 yr Colonoscopy within past 10 yrs Overall CRC screening†

% (95% CI) % (95% CI) % (95% CI)

Quartile 1
New Hampshire 8.2 10.2 (9.1–11.5) 72.2 (70.3–74.1) 75.1 (73.2–76.9)
Connecticut 9.1 11.5 (10.1–13.0) 72.5 (70.3–74.5) 74.8 (72.8–76.8)
Maryland 9.2 14.7 (13.3–16.1) 68.5 (66.6–70.4) 72.2 (70.3–73.9)
Wyoming 9.4 8.3 (7.3–9.4) 52.9 (50.9–54.9) 56.4 (54.4–58.4)
Vermont 9.8 8.3 (7.4–9.3) 68.8 (67.1–70.5) 71.6 (69.9–73.2)
New Jersey 10.1 11.6 (10.5–12.9) 60.4 (58.6–62.2) 64.6 (62.8–66.3)
Nebraska 10.4 8.8 (7.9–9.8) 56.4 (54.6–58.2) 60.0 (58.2–61.7)
Virginia 11.0 12.7 (10.9–14.6) 63.2 (60.5–65.8) 67.1 (64.4–69.6)
Utah 11.2 4.6 (3.9–5.4) 65.6 (63.9–67.3) 67.0 (65.3–68.7)
Pennsylvania 11.2 8.9 (8.0–9.8) 63.2 (61.4–64.9) 66.3 (64.6–68.0)
Iowa 11.3 10.8 (9.6–12.2) 60.3 (58.3–62.3) 63.3 (61.3–65.3)
Wisconsin 11.4 8.7 (7.4–10.1) 64.3 (61.8–66.7) 68.1 (65.6–70.5)
Massachusetts 11.6 11.9 (10.9–12.9) 72.3 (70.7–73.8) 75.1 (73.6–76.6)
Composite§ 10.3 10.7 (10.3–11.1) 64.8 (64.1–65.5) 68.2 (67.5–68.8)

Quartile 2
North Dakota 11.7 11.1 (9.9–12.6) 51.8 (49.6–54.1) 56.7 (54.4–58.9)
Minnesota 11.9 6.3 (5.4–7.4) 67.5 (65.4–69.5) 69.3 (67.3–71.3)
Maine 12.2 11.6 (10.6–12.7) 70.7 (69.0–72.2) 73.7 (72.1–75.2)
Hawaii 12.4 16.5 (14.8–18.2) 51.1 (48.8–53.4) 60.4 (58.1–62.6)
Washington 12.4 13.9 (13.1–14.7) 66.7 (65.5–67.9) 70.9 (69.7–72.0)
Alaska 12.5 8.1 (5.6–11.5) 57.4 (52.6–62.1) 58.8 (54.0–63.4)
Delaware 13.2 8.7 (7.4–10.2) 68.7 (66.2–71.2) 70.6 (68.1–73.0)
Colorado 13.3 12.0 (11.0–13.0) 59.4 (57.8–61.0) 65.1 (63.6–66.7)
Oklahoma 13.7 9.3 (8.3–10.4) 50.8 (49.0–52.7) 54.4 (52.6–56.3)
Ohio 13.7 11.7 (10.7–12.9) 58.0 (56.2–59.7) 62.6 (60.9–64.4)
Kansas 13.8 11.3 (10.4–12.4) 58.9 (57.2–60.5) 63.0 (61.4–64.6)
Louisiana 14.0 12.6 (11.4–14.0) 55.0 (53.1–56.9) 59.9 (58.0–61.8)
Illinois 14.0 7.3 (6.1–8.6) 55.9 (53.3–58.4) 58.3 (55.7–60.8)
Composite 13.0 10.4 (10.0–10.8) 59.1 (58.3–59.9) 63.0 (62.2–63.8)

Supplement

58 MMWR / November 22, 2013 / Vol. 62 / No. 3

Preventive Services (http://www.thecommunityguide.org/
index.html) to increase population-level CRC screening.
Funded states and tribes are encouraged to partner with health-
care systems, insurers, worksites, and others to maximize the
impact of implemented interventions. To date, all funded
states and tribes have implemented at least one evidence-
based intervention, with the majority implementing two or
more. Grantees have partnered with federally qualified health
centers (i.e., organizations that receive grants under Section
330 of the Public Health Service act) and private and nonprofit
health-care systems to implement patient navigation programs
and interventions to reduce structural barriers to screening,
with private health insurers and state Medicaid offices to
implement provider and patient reminder systems, and with
comprehensive cancer control coalitions and local health
departments to implement small media campaigns.

CDC also funds the National Comprehensive Cancer
Control Program (NCCCP), which provides support to all
50 states and the District of Columbia, seven tribes/tribal
organizations, and seven U.S.-associated Pacific Islands/
Territories to establish partnerships, determine priorities, and
create and implement cancer plans to reduce the burden of
cancer in their communities (34). Activities that have been
implemented successfully by selected NCCCP programs
to reduce the burden of CRC have included initiation of
professional education and practice improvement initiatives
for primary care providers and collaboration with community-
based organizations to promote CRC prevention (35). Many
grantees also have made the elimination of health disparities a
priority. Some include goals and objectives to improve cancer
prevention, early detection, treatment, and survivorship care
among disparate populations in their cancer plans (http://
cancercontrolplanet.cancer.gov).

TABLE 3. (Continued) Percentage of persons reporting having had colorectal cancer screening, by state poverty-level quartiles* — Behavioral
Risk Factor Surveillance System, United States, 2010

State/Area

Population
in poverty

(%)

FOBT within past 1 yr Colonoscopy within past 10 yrs Overall CRC screening†

% (95% CI) % (95% CI) % (95% CI)

Quartile 3
Nevada 14.6 10.0 (8.1–12.2) 53.3 (49.8–56.7) 57.5 (54.1–60.9)
Oregon 14.6 11.5 (10.2–12.9) 57.9 (55.7–60.1) 63.8 (61.6–65.9)
Michigan 14.6 11.5 (10.5–12.7) 65.8 (64.0–67.4) 69.3 (67.6–70.9)
Florida 14.7 13.5 (12.5–14.7) 61.3 (59.6–63.1) 65.7 (63.9–67.4)
South Carolina 14.8 9.3 (8.2–10.6) 61.7 (59.5–63.8) 64.9 (62.7–67.0)
Rhode Island 15.1 9.6 (8.4–10.9) 71.7 (69.7–73.6) 74.1 (72.2–76.0)
Idaho 15.2 8.1 (7.2–9.2) 52.5 (50.4–54.5) 55.9 (53.9–57.9)
Montana 15.3 8.8 (7.8–10.0) 53.8 (51.8–55.8) 57.9 (55.8–59.9)
South Dakota 15.5 10.1 (8.9–11.4) 60.9 (58.8–63.0) 63.9 (61.8–66.0)
California 15.9 19.3 (18.3–20.4) 52.6 (51.1–54.1) 62.1 (60.7–63.6)
Missouri 16.3 8.5 (7.0–10.2) 60.1 (57.4–62.7) 63.4 (60.7–65.9)
Tennessee 16.8 12.8 (11.3–14.4) 56.7 (54.2–59.1) 61.0 (58.6–63.4)
New York 17.0 9.8 (8.8–10.9) 66.7 (64.9–68.4) 69.2 (67.4–70.8)
Composite 15.6 13.5 (13.1–14.0) 59.6 (58.9–60.3) 64.8 (64.1–65.5)

Quartile 4
Indiana 17.3 10.0 (8.9–11.1) 57.6 (55.8–59.4) 61.2 (59.4–63.0)
West Virginia 17.4 12.9 (11.4–14.5) 49.9 (47.6–52.2) 54.7 (52.4–57.0)
District of Columbia 17.5 16.5 (14.6–18.6) 65.9 (63.2–68.5) 71.1 (68.5–73.5)
North Carolina 17.6 14.1 (12.9–15.3) 65.0 (63.2–66.7) 68.7 (66.9–70.4)
Kentucky 17.7 8.6 (7.5–9.9) 59.2 (57.0–61.4) 61.7 (59.5–63.9)
Alabama 17.8 10.2 (9.0–11.5) 58.2 (56.0–60.3) 62.4 (60.2–64.5)
Texas 18.4 8.7 (7.8–9.8) 55.1 (53.1–57.2) 58.7 (56.7–60.8)
Georgia 19.0 14.1 (12.6–15.7) 62.5 (60.2–64.7) 66.3 (64.1–68.5)
Arkansas 19.2 10.2 (8.8–11.9) 54.6 (51.9–57.2) 58.9 (56.2–61.5)
New Mexico 19.7 9.9 (8.7–11.2) 55.2 (53.1–57.2) 59.2 (57.2–61.2)
Arizona 22.5 11.7 (10.1–13.6) 60.3 (57.3–63.3) 64.1 (61.1–67.0)
Mississippi 23.4 11.0 (9.9–12.2) 52.8 (50.8–54.7) 57.1 (55.1–59.0)
Composite 18.7 11.0 (10.6–11.5) 58.4 (57.6–59.2) 62.6 (61.3–63.0)

Abbreviations: 95% CI = 95% confidence interval; CRC = colorectal cancer; FOBT = fecal occult blood testing.
* Quartiles were determined by calculating the state-level percent of residents living at or below the poverty level, and the range (low to high); the range was divided

evenly into three groups by ranking the states from lowest to highest for the percentage living in poverty. Source: Current Population Survey 2010 file (available at
http://www.census.gov/hhes/www/poverty/data).

† Home FOBT within the past year, flexible sigmoidoscopy within the past 5 years with FOBT in the last 3 years, or colonoscopy within the last 10 years.
§ The weighted number of persons who received a test divided by the estimated population total of all states within the quartile.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 59

Limitations
The findings in this report are subject to at least six limitations.

First, cancer registries have an interval of approximately 24
months after the close of the diagnosis year to submit cases to
NPCR and SEER, which affects the timely calculation of cancer
incidence rates. The mose recent year for which incidence data
were available for this report was 2008. Second, variation in the
quality of race and ethnicity information in medical records
and death certificates (36,37) could influence the accuracy of
surveillance data. Third, BRFSS results might underestimate
or overestimate actual CRC screening test rates because BRFSS
does not determine the indication for the test (screening versus
diagnostic use) or whether the tests are conducted according to
timelines recommended in CRC screening guidelines. Fourth,
because BRFSS does not collect information from persons in
institutions, nursing homes, long-term–care facilities, military
installations, and correctional institutions, the results cannot
be generalized to these populations. Fifth, BRFSS responses
are self-reports and not validated by medical record or claims
data review. Finally, participation rates for random-digit-dialed
health surveys have been decreasing. However, although
BRFSS has a low median response rate, the BRFSS weighting
procedure partially corrects for nonresponse.

Conclusion
Disparities in CRC incidence, mortality, and screening

persist. CRC incidence and death rates have decreased among
adults in the United States since 1999 (38). However, men have
higher rates of both incidence and mortality than women, and
non-Hispanic blacks have higher rates than other racial and
ethnic groups (1). Although increased screening could reduce
mortality from CRC by an estimated 50% (9), the prevalence
of up-to-date screening according to USPSTF guidelines
among Asians/Pacific Islanders and American Indians/Alaska
Natives remains lower than the prevalence for other racial
and ethnic groups. Coordinated and systems-focused efforts
by CDC and other federal agencies, state and local health
departments, and the medical community to address barriers
to end disparities in CRC screening should continue so that
the incidence and mortality associated with this disease can
be reduced among all populations.

Acknowledgment

Kristine Gerdes, EdS, MPH, Division of Public Affairs, Office of
the Associate Director of Communication, CDC, provided assistance
with this report.

References
1. US Cancer Statistics Working Group. United States cancer statistics:

1999–2008 incidence and mortality web-based report. Atlanta, GA: US
Department of Health and Human Services, CDC, National Cancer
Institute; 2011. Available at http://www.cdc.gov/uscs.

2. Zauber AG, Landsdorp-Vogelaar I, Knudson AB, et al. Evaluating test
strategies for colorectal cancer screening: a decision analysis for the U.S.
Preventive Services Task Force. Ann Int Med 2008;14:659–69.

3. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

4. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

5. CDC. Colorectal cancer screening—United States, 2002, 2004, 2006,
2008. In: CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

6. Tiwari RC, Clegg LX, Zou Z. Efficient interval estimation of age-adjusted
cancer rates. Stat Methods Med Res 2006;15:547–69.

7. CDC. Behavioral Risk Factor Surveillance System. Atlanta, GA: US
Department of Health and Human Services, CDC; 2010. Available at
http://www.cdc.gov/brfss.

8. US Department of Health and Human Services. Healthy people 2020
topics and objectives: cancer. Washington, DC: US Department of Health
and Human Services; 2011. Available at http://www.healthypeople.
gov/2020/topicsobjectives2020/objectiveslist.aspx?topicId=5.

9. Edwards BK, Ward E, Kohler BA, et al. Annual report to the nation on
the status of cancer, 1975–2006, featuring colorectal cancer trends and
impact of interventions (risk factors, screening, and treatment) to reduce
future rates. Cancer 2010;116:544–73.

10. CDC. Cancer screening—United States, 2010. MMWR 2012;61:41–5.
11. Shapiro JA, Seeff LC, Thompson, et al. Colorectal cancer test use from

the 2005 National Health Information Survey. Cancer Epidemiol
Biomarkers Prev 2008;17:1623–30.

12. CDC. Prevalence of colorectal cancer screening among adults—
Behavioral Risk Factor Surveillance System, United States, 2010.
MMWR 2012;61:51–6.

13. Doubeni CA, Laiyemo AO, Young AC, et al. Primary care, economic
barriers to health care, and use of colorectal cancer screening tests among
Medicare enrollees over time. Ann Fam Med 2010;8:299–307.

14. Beydoun HA, Beydoun MA. Predictors of colorectal cancer screening
behaviors among average-risk older adults in the United States. Cancer
Causes Control 2008;19:339–59.

15. White A, Vernon SW, Franzini L, Du XL. Racial and ethnic disparities
in colorectal cancer screening persisted despite expansion of Medicare’s
screening reimbursement. Cancer Epidemiol Biomarkers Prev
2011;20:811–7.

16. Semrad JJ, Tancredi DJ, Baldwin LM, Green P, Fenton JJ. Geographic
variation of racial/ethnic disparities in colorectal cancer testing among
Medicare enrollees. Cancer 2011;117:1755–63.

17. Doubeni CA, Laivemo AO, Reed G, Field TS, Fletcher RH.
Socioeconomic and racial patterns of colorectal cancer screening among
Medicare enrollees in 2000 to 2005. Cancer Epidemiol Biomarkers Prev
2009;18:2170–5.

18. Agency for Healthcare Research and Quality. National healthcare
disparities report: 2011. Washington, DC: US Department of Health
and Human Services, Agency for Healthcare Research and Quality;
2012. Available at http://www.ahrq.gov/qual/nhdr11/chap9.htm.

19. Skopec L, Sommers BD. Seventy-one million additional Americans are
receiving preventive services coverage without cost-sharing under the
Affordable Care Act. Washington, DC: US Department of Health and
Human Services; 2013. Available at http://aspe.hhs.gov/health/
reports/2013/PreventiveServices/ib_prevention.cfm.

Supplement

60 MMWR / November 22, 2013 / Vol. 62 / No. 3

20. Klabunde CN, Lanier D, Nadel MR, et al. Colorectal cancer screening
by primary care physicians: recommendations and practices, 2006–2007.
Am J Prev Med 2009;37:8–16.

21. Meissner HI, Breen N, Klabunde CN, Vernon SW. Patterns of colorectal
cancer screening uptake among men and women in the United States.
Cancer Epidemiol Biomarkers 2006;15:389–94.

22. Schenck AP, Peacock S, Klabunde CN, et al. Trends in colorectal cancer test
use in the Medicare population, 1998–2005. Am J Prev Med 2009;37:1–7.

23. Whitlock EP, Lin J, Liles E, et al. Screening for colorectal cancer: an
updated systematic review. Rockville, MD: US Department of Health and
Human Services, Agency for Healthcare Research and Quality; 2008.

24. Schoen RE, Pinksky PF, Weissfeld JL, et al. Colorectal-cancer incidence
and mortality with screening flexible sigmoidoscopy. New Engl J Med
2012;366:2345–57.

25. Feldstein AC, Perrin N, Liles EG, et al. Primary care colorectal cancer
screening recommendation patterns: associated factors and screening
outcomes. Med Decis Making 2012;32:198–208.

26. Yabroff KR, Klabunde CN, Yuan G, et al. Are physicians’ recommendations
for colorectal cancer screening guideline-consistent? J Gen Intern Med
2010;26:177–84.

27. Zapka JM, Klabunde CN, Arora NK, et al. Physicians’ colorectal cancer
screening discussion and recommendation patterns. Cancer Epidemiol
Biomarkers Prev 2011;20:509–21.

28. Inadomi JM, Vijan S, Janz NK, et al. Adherence to colorectal cancer
screening: a randomized clinical trial of competing strategies. Arch Intern
Med 2012;172:575–82.

29. Hawely ST, McQueen A, Bartholomew LK, et al. Preferences for
colorectal cancer screening tests and screening test use in a large
multispecialty primary care practice. Cancer 2012;118:2726–34.

30. Hawley ST, Volk RJ, Krishnamurthy P, et al. Preferences for colorectal
cancer screening among racially/ethnically diverse primary care patients.
Med Care 2008;46:S10–6.

31. Bandi P, Cokkinides V, Smith RA, Jemal A. Trends in colorectal cancer
screening with home-based fecal occult blood tests in adults ages 50 to
64 years, 2000 to 2008. Cancer 2012;118:5092–9.

32. Klabunde CN, Cronin KA, Breen N, et al. Trends in colorectal cancer
test use among vulnerable populations in the United States. Cancer
Epidemiol Biomarkers Prev 2011;20:1611–21.

33. CDC. Colorectal Cancer Control Program. Atlanta, GA: US Department
of Health and Health Services, CDC; 2010. Available at http://www.
cdc.gov/cancer/crccp/about.htm.

34. CDC. National Comprehensive Cancer Control Program. Atlanta, GA:
US Department of Health and Human Services, CDC; 2012. Available
at http://www.cdc.gov/cancer/ncccp/about.htm.

35. CDC. National Comprehensive Cancer Control Program. Success
stories. Atlanta, GA: US Department of Health and Human Services,
CDC; 2012. Available at http://www.cdc.gov/cancer/ncccp/state.htm.

36. Clegg LX, Reichman ME, Hankey BF, et al. Quality of race, Hispanic
ethnicity, and immigrant status in population-based cancer registry data:
implications for health disparity studies. Cancer Causes Control
2007;18:177–87.

37. Arias E, Schauman WS, Eschbach K, Sorlie P, Backlund E. The validity
of race and Hispanic origin on death certificates in the United States.
Vital Health Stat 2008;148:1–23.

38. Eheman C, Henley SJ, Ballard-Barbash R, et al. Annual report to the
nation on the status of cancer, 1975–2008, featuring cancers associated
with excess weight and lack of sufficient physical activity. Cancer
2012;118:2338–66.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 61

Introduction
One out of four adults aged 19–64 years reported not

having health insurance at some time during 2011, with a
majority remaining uninsured for ≥1 year (1). In the first
quarter of 2010, an estimated 59.1 million persons had no
health insurance for at least part of the year, an increase from
58.7 million in 2009 and 56.4 million in 2008 (2). The
unemployment rate increased from 5.8% to 9.3% from 2008
to 2009, the largest 1-year increase on record (3). Losing or
changing jobs was the primary reason persons experienced a
gap in health insurance (1). Employment-based coverage for
persons aged <65 years continued to erode for the ninth year
in a row, falling 3.0 percentage points from 61.9% in 2008 to
58.9% in 2009 (3). Persons aged 18–64 years with no health
insurance during the preceding year were seven times as likely
as those continuously insured to forgo needed health care
because of cost (2).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (4) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the
2013 CHDIR Introduction (5). This report provides updated
information that complements the health insurance coverage
data published in the 2011 CHDIR (6). This report on health
insurance coverage discusses and raises awareness of differences
in the characteristics of persons who lack health insurance
coverage, and prompts actions to reduce these disparities.

Methods
To identify disparities in the lack of health insurance

coverage for adults aged 18–64 years for different demographic
and socioeconomic groups over time, CDC analyzed data
from the 2008 and 2010 National Health Interview Survey
(NHIS). NHIS is a cross-sectional survey of a representative
sample of the civilian, noninstitutionalized U.S. household
population. NHIS includes various questions on family health

insurance (ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/
Survey_Questionnaires/NHIS/2010/English/qfamily.pdf ).
The 2010 NHIS included 27,157 respondents, of whom a
total of 80 were excluded because of unknown insurance status.
The overall response rate was 60.8%. The questionnaire begins
with the question “are you/is anyone in the family covered
by any kind of health insurance or some kind of health-care
plan?” Respondents were considered uninsured if they did
not have any private health insurance, Medicare, Medicaid,
State Children’s Health Insurance Program coverage, state-
sponsored or other government-sponsored health plan, or a
military health-care plan at the time of the interview. Persons
also were considered uninsured if they reported having only
Indian Health Service coverage or a private plan that paid for
only one type of service (e.g., unintentional injuries or dental
care). Rate of uninsured is the percentage of adults aged 18–64
who did not have health insurance.

Disparities were examined by characteristics that included
race and ethnicity, sex, age (adults aged 18–64 years), household
income, disability status, and educational attainment. Poverty
status was defined by using the ratio of income to the federal
poverty level (FPL), in which “poor” is <1.0 times FPL, “near
poor” is 1.0–2.9 times FPL, and “nonpoor” is ≥3.0 times FPL.
Educational attainment was defined as less than high school,
high school graduate or equivalent, some college, and college
graduate or higher. Disability was defined as limitations in a
person’s activity because of a health condition or impairment.
Race was defined as white, black, American Indian/ Alaska
Native, and other and multiple race. Ethnicity was defined as
Hispanic or non-Hispanic.

Disparities were measured as deviations from a “referent”
category for an uninsured rate; defined as the lowest percentage
for a population group-specific without health insurance.
Absolute difference was measured as the simple difference
between an estimate for a population subgroup and the
estimate for the referent category rate. The relative difference,
a percentage, was calculated by dividing the absolute difference
by the value in the referent category and multiplying by 100.
The 95% confidence intervals for uninsured rates were estimated
using statistical software (7). Pair-wise differences by sex, age

Health Insurance Coverage — United States, 2008 and 2010
Ramal Moonesinghe, PhD1
Man-huei Chang, MPH2
Benedict I. Truman, MD3

1Office of Minority Health and Health Equity, Office of the Director, CDC
2Center for Surveillance, Epidemiology, and Laboratory Services, CDC

3National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC
Corresponding author: Ramal Moonesinghe, PhD, Office of Minority Health and Health Equity, Office of the Director, CDC. Telephone: 770-488-8203;
E-mail: [email protected].

Supplement

62 MMWR / November 22, 2013 / Vol. 62 / No. 3

group, race/ethnicity, disability status, educational achievement,
and differences between 2008 and 2010 were tested by the
z-statistic (one-tailed) at the 0.05 level of significance. A
covariance of zero between estimates in conducting these tests
was assumed. When testing differences within demographic
groups, the Bonferroni method was used to account for multiple
comparisons. If κ comparisons existed within a group, the level of
significance was set to 0.05/κ. Estimates with a relative standard
error of >20% were considered unreliable.

Results
During 2010, substantial disparities persisted in uninsured

rates for all demographic and socioeconomic groups. Statistically
significant disparities by sex (p<0.001) also continued during
2010, with a higher percentage of males (24.1%) than females
(18.8%) being uninsured (Table 1). The uninsured rate for
young adults aged 18–34 years was approximately double the

uninsured rate for adults aged 45–64 years (aged 18–34 years,
28.5%; aged 45–64 years, 15.4%). Uninsured rates for all the
age groups analyzed were significantly higher (p<0.001) than
with adults aged 45–64 years.

During 2010, among adults aged 18–64 years, approximately
two out of five persons of Hispanic ethnicity and one out of
four non-Hispanic blacks were classified as uninsured. Both
these groups had significantly higher (p<0.001) uninsured
rates (41.0% and 26.2%, respectively), compared with
Asians/Pacific Islanders and non-Hispanic whites (17.3% and
16.1%, respectively). No significant difference in uninsured
rates existed between non-Hispanic whites and Asians/Pacific
Islanders. During 2010, approximately half of uninsured adults
were non-Hispanic whites (Table 2). Hispanics accounted for
29.3% of the uninsured population. The estimate of uninsured
rate for non-Hispanic American Indian and Alaska Native
persons was not reliable enough to make comparisons with
estimates from other subpopulations.

TABLE 1. Percentage* of adults aged 18–64 years without health insurance, by selected demographic characteristics — National Health Interview
Survey, United States, 2008 and 2010

Characteristic

2008 2010

% (95% CI)

Absolute
difference

(percentage points)

Relative
difference

(%) % (95% CI)

Absolute
difference

(percentage points)

Relative
difference

(%)

Sex
Male 22.2 (21.0–23.5) 4.9 28.3 24.1 (23.0–25.2) 5.3 28.5
Female 17.3 (16.2–18.3.) Ref. — 18.8 (17.8–19.7) Ref. Ref.

Age group (yrs)
18–24 27.9 (25.4–30.4) 14.3 105.1 29.8 (27.6–31.9) 14.4 93.5
25–34 26.6 (24.7–28.5) 13.0 95.6 27.2 (25.6–28.9) 11.8 76.6
35–44 18.7 (17.3–20.2) 5.1 37.5 21.4 (20.1–22.7) 6.0 39.0
45–64 13.6 (12.6–14.6) Ref. — 15.4 (14.5.–16.2) Ref. Ref.
Poverty status†

Poor 37.0 (34.0–40.0) 28.1 315.1 41.2 (38.9–43.5) 33.1 410.5
Near poor 30.5 (28.8–32.2) 21.6 242.2 34.2 (32.8–35.6) 26.1 323.6
Nonpoor 8.9 (8.1–9.8) Ref. — 8.1 (7.4–8.7) Ref. Ref.

Race/Ethnicity
Hispanic§ 41.6 (38.8–44.4) 27.6 197.1 41.0 (39.0–43.0) 24.9 154.2
White, non-Hispanic 14.6 (13.7–15.5) 0.6 4.3 16.1 (15.3–17.0) Ref. —
Black, non-Hispanic 22.1 (20.3–23.9) 8.1 57.9 26.2 (24.2–28.3) 10.1 62.6
American Indian/Alaska Native 33.7¶ — — 33.5¶ — — —
Asian/Pacific Islander 14.0 (11.2–16.9) Ref. — 17.3 (14.7–19.8) 1.2 7.1
Other, non-Hispanic other, and

multiple race
20.1¶ — — — 21.5¶ — — —

Disability status
Persons with a disability 17.7 (16.4–19.0) Ref. — 19.6 (18.4–20.7) Ref. —
Persons without a disability 20.5 (19.4–21.5) 2.8 15.8 22.3 (21.4–23.1) 2.7 13.7

Educational attainment
Less than high school 40.5 (37.6–43.3) 32.4 400.0 42.8 (40.6–45.0) 34.8 432.2
High school graduate or equivalent 24.4 (22.8–26.1) 16.3 201.2 27.5 (26.1–28.9) 19.5 242.5
Some college 16.6 (15.4–17.7) 8.5 104.9 20.0 (18.8–21.2) 12.0 148.8
College graduate or higher 8.1 (7.1–9.0) Ref. Ref. 8.0 (7.2–8.8) Ref. —

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Rate of uninsured is the percentage of adults aged 18–64 who did not have health insurance.
† Poor = ≤1.0 times the federal poverty level (FPL), near poor = 1.0–2.9 times FPL, and nonpoor = ≥3.0 times FPL. FPL was based on U.S. Census Bureau poverty

thresholds, available at http://www.census.gov/hhes/www/poverty/html.
§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Estimates are considered unreliable because the relative standard errors are >20%.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 63

From 2008 to 2010, uninsured rates increased significantly
(p<0.05) for all groups considered in this report with the
exception of persons with less than a high school diploma,
college graduates, highest income group considered, Hispanics,
and persons in age group 18–34 years. However, those with less
than a high school diploma and Hispanics were groups with
the highest uninsured rates. Chronically ill patients without
insurance are more likely than those with coverage 1) not to
have visited a health-care professional, and 2) either not to have
a standard site for care or to identify their standard site of care
as an emergency department (8). Because minority populations
and the poor have high uninsured rates, these populations tend
to visit the emergency department for nonurgent health care.
Costly emergency department care could be saved if primary
care were available to these populations (9).

Limitations
The findings in this report are subject to at least two

limitations. First, health insurance coverage information
in NHIS is self-reported and subject to recall bias. Second,
because NHIS does not include institutionalized persons, the
results are not generalizable to segments of the population
that include prison inmates, military personnel, and adults in
nursing homes and other long-term care facilities.

Conclusion
Disparities in health insurance coverage continue among all

demographic and socioeconomic groups. Coverage expansion
resulting from current or future reform of health insurance
policies is likely to reduce disparities in uninsured rates. For
example, after implementation of the 2010 Affordable Care Act,
an estimated 6.6 million adults aged 19–25 years who might
have been uninsured stayed on or joined their parents’ health
plans between November 2010 and November 2011 (10).

References
1. Collins SR, Robertson R, Garber T, Doty MM. Gaps in health insurance:

why so many Americans experience breaks in coverage and how the
Affordable Care Act will help. Issue Brief (Commonwealth Fund) 2012
April. Available at http://www.commonwealthfund.org/Publications/
Issue-Briefs/2012/Apr/Gaps-in-Health-Insurance.aspx.

2. CDC. Vital signs: health insurance coverage and health care utilization—
United States, 2006—2009 and January-March 2010. MMWR 2010;
59:1448–54.

3. Gould E. Decline in employer-sponsored health coverage accelerated three times
as fast in 2009. Economic Policy Institute. Available at http://www.epi.org/
publication/decline_in_employer-sponsored_health_coverage_accelerated.

4. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011, 60;(Suppl; January 14, 2011).

5. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013; 62 (No. Suppl 3).

During 2010, among persons aged 18–64 years, uninsured
rates for poor (those living at the federal poverty level [FPL])
and near poor persons (those at <3.0 times FPL) ranged from
34.2% to 41.2%, and these rates were significantly higher
(p<0.001) than the uninsured rate among the nonpoor
(those at ≥3.0 FPL) (Table 1). Approximately half (50.7%)
of uninsured adults were near poor (Table 2). During 2010,
income for the near poor ranged from $22,314 to $66,942
per year for a family of four. Uninsured rates for persons in the
poor and near poor categories increased significantly (p<0.014)
from 2008 (37.0% and 30.5%, respectively) to 2010 (41.2%
and 34.2%, respectively). The uninsured rate for non-Hispanic
blacks also increased significantly (p<0.002) from 22.1% in
2008 to 26.2% in 2010. No significant difference existed in
the uninsured rate between 2008 (41.6%) and 2010 (41.0%)
for the Hispanic population (Table 1).

Regarding educational attainment, when compared with
college graduates, all groups continued to have significantly higher
uninsured rates (p<0.001). From 2008 to 2010, uninsured rates
for graduates from high school and with some college education
increased significantly (p<0.003). The uninsured rate for persons
without a disability (22.3%) also remained significantly higher
(p<0.001) than persons with a disability (19.6%).

Discussion
During 2010, similar to the disparities observed in 2004

and 2008 (6), substantial disparities persisted in uninsured
rates for all demographic and socioeconomic groups.

TABLE 2. Number and percentage of adults aged 18–64 years without
health insurance, by poverty status and race/ethnicity — National
Health Interview Survey, United States, 2010

Characteristic No. (% of total)

% without
health

insurance

Poverty status*
Poor 11,078,526 (30.8) 41.2
Near poor 18,246,425 (50.7) 34.2
Nonpoor 6,641,720 (18.5) 8.1
Total 35,966,671§ (100.0) —

Race/Ethnicity
Hispanic† 11,957,253 (29.3) 41.0
White, non-Hispanic 20,130,159 (49.5) 16.1
Black, non-Hispanic 6,097,277 (15.0) 26.2
American Indian/Alaska Native 363,140 (0.9) 33.5
Asian/Pacific Islander 1,575,972 (3.9) 17.3
Other, non-Hispanic and

multiple race
560,459 (1.4) 21.5

Total 40,684,260§ (100.0) —

* Poor = ≤1.0 time the federal poverty level (FPL), near poor = 1.0 2.9 times FPL,
and nonpoor = ≥3.0 times FPL. FPL was based on U.S. Census Bureau poverty
thresholds, available at http://www.census.gov/hhes/www/poverty/html.

† Persons of Hispanic ethnicity might be of any race or combination of races.
§ Totals are different because of unknown poverty status.

Supplement

64 MMWR / November 22, 2013 / Vol. 62 / No. 3

6. CDC. Health insurance coverage—United States, 2004 and 2008. In:
CDC health disparities and inequalities report—United States, 2011.
MMWR 2011; (Suppl; January 14, 2011).

7. SAS Institute, Inc. SAS version 9.02. Cary, NC: SAS Institute, Inc.; 2.
8. Wilper AP, Woolhandler S, Lasser KE, et al. A national study of chronic

disease prevalence and access to care in uninsured US adults. Ann Intern
Med 2008;149:170–6.

9. Gill JM, Arch GM, Musa N. The effect of continuity of care on
emergency department use. Arch Fam Med 2000;9:333–8.

10. Collins SR, Robertson R, Garber T, Doty MM. Young, uninsured, and in
debt: why young adults lack health insurance and how the Affordable Care
Act is helping. Issue Brief (Commonwealth Fund) 2012. Available at http://
www.commonwealthfund.org/Publications/Issue-Briefs/2012/Jun/Young-
Adults-2012.aspx.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 65

Introduction
Infection with influenza viruses can cause severe morbidity

and mortality among all age groups. Children, particularly
those aged <5 years (1–3), have the highest incidence of
infection during epidemic periods; however, the highest rates of
influenza-associated hospitalizations and deaths are among the
elderly (aged ≥65 years), children aged <2 years, and those of
any age with underlying medical conditions (1,4,5). Each year,
influenza-related complications are estimated to result in more
than 226,000 hospitalizations (6). During 1976–2006, estimates
of influenza-associated deaths in the United States ranged from
approximately 3,000 to an estimated 49,000 persons (7,8)
(http://www.cdc.gov/flu/keyfacts.htm#howserious). Annual
vaccination is the most effective strategy for preventing influenza
virus infection and its complications (9).

Racial and ethnic disparities in seasonal influenza vaccination
coverage have been observed in previous influenza seasons among
children and adults (10). This summary updates the evaluation
of these disparities among all persons aged ≥6 months, previously
reported for the 2000–01 through the 2009–10 season (10),
with findings from the 2010–11 influenza season and compares
coverage in 2009–10 and 2010–11. For the 2010–11 influenza
season, the Advisory Committee on Immunization Practices
(ACIP) expanded its recommendations to include annual
influenza vaccination of all persons aged ≥6 months (11). For
the first time, the 2010–11 ACIP flu season recommendations
included healthy adults aged 18–49 years.

This report is part of the second CDC Health Disparities and
Inequalities Report (CHDIR) (12). The 2011 CHDIR (13) was
the first CDC report to assess disparities across a wide range of
diseases, behavioral risk factors, environmental exposures, social
determinants, and health-care access. The criteria for inclusion
of topics that are presented in the 2013 CHDIR are described
in the 2013 CHDIR Introduction (14). This report provides
an update on the progress of influenza vaccination coverage in
the United States, by age, race/ethnicity, and risk status. The
purposes of this report on influenza vaccination are to discuss and
raise awareness of differences in the characteristics of populations

who received influenza vaccination, and to prompt actions to
reduce disparities.

Methods
To estimate the progress of influenza vaccination coverage

in the United States, by age, race/ethnicity, and risk status,
various data sources were used. Age groups were defined as
aged <45 years, 45–74 years, <75 years, ≥75 years, and ≥85
years. Race was defined as white, black, Asian/Pacific Islander
(A/PI), American Indian/Alaska Native (AI/AN), and other
and multiple race. Ethnicity was defined as Hispanic or non-
Hispanic. Race/ethnicity categories are mutually exclusive.
For the 2009–10 season, high risk conditions included
asthma, other lung problems, diabetes, heart disease, kidney
problems, anemia, and weakened immune system caused by a
chronic illness or by medicines taken for a chronic illness. For
the 2010–11 season, high risk conditions included asthma,
diabetes, and heart disease. Other medical conditions that place
persons at increased risk for complications from influenza (11)
were not under surveillance for this report. For this update,
vaccination by household income, educational attainment,
poverty status, disability status, and geographic location were
not analyzed. Data on place of birth was not available.

To estimate the proportion of persons aged ≥6 months
who received influenza vaccination during the 2009–10
influenza season, combined data from the National 2009
H1N1 Flu Survey (NHFS) and the Behavioral Risk Factor
Surveillance System (BRFSS) were used. The NHFS included
children identified from the National Immunization Survey
(NIS) and from a stand-alone telephone survey. To estimate
the proportion of children aged 6 months through 17 years
who received influenza vaccination during the 2010–11
influenza season, data from the NIS were used. To estimate
the proportion of adults aged ≥18 years who received influenza
vaccination during the 2010–11 influenza season, BRFSS data
were used. Both NIS and BRFSS collected monthly data on
vaccinations reported during August 2010 through May 2011
for all 50 states and the District of Columbia.

Seasonal Influenza Vaccination Coverage —
United States, 2009–10 and 2010–11

Anne F. McIntyre, PhD
Amparo G. Gonzalez-Feliciano, MPH

Leah N. Bryan, MPH,
Tammy A. Santibanez, PhD

Walter W. Williams, MD
James A. Singleton, PhD

National Center for Immunization and Respiratory Diseases, CDC

Corresponding author: Anne F. McIntyre, PhD, National Center for Immunization and Respiratory Diseases, CDC. Telephone: 404-639-8284; E-mail:
[email protected].

Supplement

66 MMWR / November 22, 2013 / Vol. 62 / No. 3

Comparisons between the 2010–11 and 2009–10 seasons
used estimates for the recommended trivalent seasonal
vaccines (11,15), and all 2009–10 estimates in this report are
for trivalent seasonal vaccination, although for the 2009–10
seasons, two vaccines were recommended: the trivalent seasonal
vaccine (15), along with the influenza A(H1N1)pdm09
monovalent vaccine to provide immunity against the pandemic
strain that emerged in 2009 (16). Coverage estimates for all
persons aged ≥6 months were determined using combined
state-level monthly estimates weighted by the age-specific
populations of each state. In 2009–10, the unweighted sample
sizes for children aged 6 months through 17 years and persons
≥18 years were 149,872 and 361,485, respectively (http://www.
cdc.gov/flu/professionals/vaccination/coverage_0910estimates.
htm). For 2010–11, the unweighted sample size for children
aged 6 months through 17 years was 116,799 and 377,569
for persons ≥18 years (http://www.cdc.gov/flu/professionals/
vaccination/coverage_1011estimates.htm).

Disparities were measured as the deviations from a “referent”
category cumulative proportion. Absolute difference was
measured as the simple difference between a population
subgroup estimate and the estimate for its respective reference
group. A description of the methods for estimating national
influenza vaccination coverage and comparing coverage by
age group and race/ethnicity has been published previously
(10). The same statistical methods were used for both seasons
(2009–10 and 2010–11). Estimates were suppressed if the
sample size was <30 or the relative standard error was >0.3.
Student t tests were used to determine statistical significance in
differences between groups and between 2009–10 and 2010–11
vaccination coverage levels with significance defined as p<0.05.
Only statistically significant results are highlighted in this report.

Results
Overall, influenza vaccination coverage was two percentage

points higher for the 2010–11 season versus the 2009–10
season (43.0% versus 41.2%, respectively), primarily because
of an increase in vaccine coverage among children aged 6
months–17 years (51.0% versus 43.7%, respectively) (Table).
Vaccine coverage increased significantly among four groups
of children: Hispanic and non-Hispanic whites, blacks, and
those of other/multiple races. During the 2010–11 seasons,
compared with non-Hispanic white children, coverage among
Hispanic, Asian/Pacific Islander, and children of other and
multiple races was higher (Table).

Overall, influenza vaccination coverage among adults aged ≥18
years remained relatively stable, at 40.4% during 2009–10 and
40.5% during the 2010–11 influenza season (Table). Among
those aged 18–49 years (regardless of risk status) and 50–64

years, coverage was similar in both seasons. However, among
adults aged ≥65 years, coverage decreased from 69.6% to 66.6%.

During 2010–11, among all adults, including persons aged
18–49 overall, 50–64, and ≥65 years, coverage remained lower
among non-Hispanic blacks (28.1%, 38.4%, and 56.1%,
respectively) than among non-Hispanic whites (31.6%, 45.7%,
67.7%, respectively). Coverage also was lower among Hispanic
adults aged 18–49 and 50–64 years (27.1% and 41.9%,
respectively) than among non-Hispanic whites (31.6% and
45.7%, respectively). During 2010–11, coverage was similar
among Hispanics and non-Hispanic whites aged ≥65 years;
however, compared with 2009–10, coverage decreased by 4.0
percentage points among non-Hispanic whites and increased
by 10.7 percentage points among Hispanics (Table).

Discussion
Overall, influenza vaccination coverage estimates were

significantly higher during the 2010–11 season than during the
2009–10 season because of an increase in vaccinations among
children. Coverage among non-Hispanic black and Hispanic
children has improved, and is either similar to, or slightly higher
than, coverage among non-Hispanic white children. Efforts to
improve coverage are ongoing. The federally funded Vaccines
for Children program provides vaccines at no cost to children
who might not otherwise be vaccinated because of inability to
pay (17). Community demand for influenza vaccination can be
increased by client reminder and recall systems (18). Provider
and systems-based interventions (e.g., provider assessment and
feedback, and use of immunization information systems) also can
increase vaccination coverage (http://www.thecommunityguide.
org/vaccines/universally/index.html) (17,18).

Among adults aged ≥65 years, influenza vaccination coverage
was lower among non-Hispanic blacks than all other racial/ethnic
groups, suggesting that additional efforts to reach this population
are needed. Interventions (18) to provide all ACIP recommended
vaccinations throughout the lifespan could be a step toward
increasing coverage and addressing disparities among adults.

The revised ACIP recommendations to vaccinate all persons
aged ≥6 months were in place for the entire 2010–11 influenza
season, but did not appear to have an effect on coverage
among those aged 18–49 years (regardless of risk status)
compared with the previous season. Additional promotion
of or education about the expanded recommendations might
increase coverage in this age group. Promising strategies might
include 1) expanding access through nontraditional settings
(e.g., pharmacy, workplace, and school venues) for vaccination
to reach persons who might not visit a traditional provider
during the flu season; 2) improving the use of evidence-based
practices at medical sites (e.g., standing orders, reminder/recall

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 67

TABLE. Seasonal influenza vaccination coverage,* by race/ethnicity† — Behavorial Risk Factor Surveillance System, National 2009 H1N1 Flu
Survey, and National Immunization Survey, United States, 2009–2010 and 2010–2011.

Race/Ethnicity by age group

2009–2010 2010–2011

Coverage difference
from 2009–10 to

2010–11
(percentage points)% (95% CI)

Absolute
difference§

(percentage
points) % (95% CI)

Absolute
difference§

(percentage
points)

≥6 mos
Total 41.2 (40.8–41.6) 43.0¶ (42.6–43.4) 1.8††

White, non-Hispanic 43.9 (43.5–44.3) Ref. 44.3 (43.9–44.7) Ref. 0.4
Black, non-Hispanic 33.7 (32.5–34.9) -10.2†† 39.0 (37.5–40.5) -5.3†† 5.3††
Hispanic 33.6 (32.4–34.8) -10.3†† 40.0 (38.6–41.4) -4.3†† 6.4††
Asian/Pacific Islander 44.3 (42.0–46.6) 0.4 43.1 (40.3–45.9) -1.2 -1.2
American Indian/Alaska Native 46.3 (43.7–48.9) 2.4 42.1 (38.1–46.1) -2.2 -4.2
Other and multiple race 38.6 (36.6–40.6) -5.3†† 42.9 (40.4–45.4) -1.4 4.3††

6 mos–17 yrs
Total 43.7 (42.8–44.6) 51.0¶ (50.1–51.9) 7.3††

White, non-Hispanic 43.2 (42.3–44.1) Ref. 48.5¶ (47.5–49.5) Ref. 5.3††
Black, non-Hispanic 37.0 (34.4–39.6) -6.2†† 50.8¶ (47.9–53.7) 2.3 13.8††
Hispanic** 46.9 (44.3–49.5) 3.7†† 55.1¶ (52.5–57.7) 6.6†† 8.2††
Asian/Pacific Islander 56.1 (52.4–59.8) 12.9†† 59.4 (54.7–64.1) 10.9†† 3.3
American Indian/Alaska Native 51.7 (47.0–56.4) 8.5†† 55.7 (47.7–63.7) 7.2 4.0
Other and multiple race 49.7 (45.7–53.7) 6.5†† 55.6¶ (51.5–59.7) 7.1†† 5.9††

≥18 yrs
Total 40.4 (40.0–40.8) 40.5 (40.1–40.9) 0.1

18–49 yrs
All, including high risk 29.9 (29.4–30.4)   30.5 (29.9–31.1)   0.6

White, non-Hispanic 31.9 (31.3–32.5) Ref. 31.6 (30.8–32.4) Ref. -0.3
Black, non-Hispanic 25.3 (23.6–27.0) -6.6†† 28.1 (25.7–30.5) -3.5†† 2.8
Hispanic 24.7 (23.3–26.1) -7.2†† 27.1 (25.1–29.1) -4.5†† 2.4
Asian/Pacific Islander 35.5 (32.2–38.8) 3.6†† 33.4 (29.5–37.3) 1.8 -2.1
American Indian/Alaska Native 39.3 (35.3–43.3) 7.4†† 31.3 (25.2–37.4) -0.3 -8.0††
Other and multiple race 27.9 (25.0–30.8) -4.0†† 32.1 (27.8–36.4) 0.5 4.2

High risk only§§ 38.2 (36.9–39.5) 39.0 (36.8–41.2) 0.8
White, non-Hispanic 39.9 (38.3–41.5) Ref. 39.2 (36.8–41.6) Ref. -0.7
Black, non-Hispanic 34.8 (31.5–38.1) -5.1†† 37.1 (30.2–44.0) -2.1 2.3
Hispanic 35.9 (32.0–39.8) -4.0 37.3 (30.8–43.8) -1.9 1.4
Asian/Pacific Islander 42.9 (32.3–3.5)§§ 3.0 34.0 (21.5–6.5)¶¶ -5.2 -8.9
American Indian/Alaska Native 45.8 (38.1–53.5) 5.9 40.3 (25.8–54.8)¶¶ 1.1 -5.5
Other and multiple race 36.8 (30.7–42.9) -3.1 45.5 (35.7–55.3) 6.3 -8.9

50–64 yrs
Total 45.0 (44.4–45.6) 44.5 (43.9–45.1) -0.5

White, non-Hispanic 46.5 (45.9–47.1) Ref. 45.7 (44.9–46.5) Ref. -0.8
Black, non-Hispanic 40.3 (38.3–42.3) -6.2†† 38.4 (36.0–40.8) -7.3†† -1.9
Hispanic 40.3 (37.5–43.1) -6.2†† 41.9 (38.6–45.2) -3.8†† 1.6
Asian/Pacific Islander 48.8 (42.6–55.0) 2.3 49.3 (43.6–55.0) 3.6 0.5
American Indian/Alaska Native 48.6 (44.2–53.0) 2.1 44.6 (37.9–51.3) -1.1 -4.0
Other and multiple race 39.2 (35.7–42.7) -7.3†† 40.5 (36.2–44.8) -5.2†† 1.3

≥65 yrs
Total 69.6 (69.0–70.2) 66.6¶ (66.0–67.2) -3.0††

White, non-Hispanic 71.7 (71.2–72.2) Ref. 67.7¶ (67.1–68.3) Ref. -4.0††
Black, non-Hispanic 55.1 (52.8–57.4) -16.6†† 56.1 (52.8–59.4) -11.6†† 1.0
Hispanic 56.1 (52.8–59.4) -15.6†† 66.8¶ (63.1–70.5) -0.9 10.7††
Asian/Pacific Islander 70.7 (65.1–76.3) -1.0 67.9 (61.6–74.2) 0.2 -2.8
American Indian/Alaska Native 61.6 (56.1–67.1) -10.1†† 68.7 (60.7–76.7) 1.0 7.1
Other and multiple race 64.2 (60.1–68.3) -7.5†† 60.7 (56.4–65.0) -7.0†† -3.5

Abbreviations: 95% CI = 95% confidence interval; Ref = referent.
* Coverage estimates for 2010–2011 are for persons with reported vaccination during August 2010–May 2011 who were interviewed during September 2010–June

2011. Coverage estimates for 2009–2010 are for persons with reported vaccination during August 2009–May 2010 who were interviewed during October 2009–June
2010; estimates for 2009–2010 included data from NHFS; season estimates for 2010–2011 use NIS only for children and BRFSS only for adults.

† Race/ethnicity categories are mutually exclusive; Native Hawaiians, Pacific Islanders, and persons of other or multiple races were classified in the “Other and multiple race” group.
§ Absolute difference (percentage points): (percentage racial/ethnic group of interest) – (percentage white only, non-Hispanic).
¶ Estimated vaccination coverage for the 2010–2011 season is significantly different from the 2009–2010 season (referent) at (p<0.05).
** Persons of Hispanic ethnicity might be of any race or combination of races.
†† Estimated vaccination coverage is significantly different from the white only, non-Hispanic population (referent) within age group at (p<0.05).
§§ For the 2010–2011 season, high risk conditions included asthma, diabetes, and heart disease. For the 2009–2010 season, high risk conditions included asthma, other

lung problems, diabetes, heart disease, kidney problems, anemia, and weakened immune system caused by a chronic illness or by medicines taken for a chronic illness.
¶¶ Estimates might be unreliable because the confidence interval half-width is >10.

Supplement

68 MMWR / November 22, 2013 / Vol. 62 / No. 3

notification, and provider recommendation) to ensure that all
persons who visit a health-care provider during the flu season
receive a vaccination recommendation and offer; or 3) using
immunization information systems, also known as registries,
at the point of clinical care and at the population level to guide
clinical and public health vaccination decisions (18).

Limitations
The findings in this report are subject to at least five

limitations. First, children aged 6 months to <9 years are
recommended for up to 2 doses of vaccine depending on past
vaccination history (11); however, this report only measured
receipt of at least 1 dose for children of all ages. Second, the
estimates are made on the basis of self-report for adults and
parental-report for children, and were not validated by medical
record reviews. Racial/ethnic disparities also might differ on the
basis of parent versus provider report, child’s age, and whether
receipt of 1 dose or full vaccination status is measured; previous
studies have shown racial/ethnic disparities in influenza
vaccination coverage of children aged 6–23 months on the basis
of provider-reported data for full vaccination; most children
in this age group would need 2 doses to be considered fully
vaccinated (19,20). Third, the sample might not be nationally
representative because of incomplete sample frames (e.g., NIS
and BRFSS surveys miss households without phones), and
selection bias from survey nonresponse might remain after
weighting adjustments (1,17,21,22). Fourth, misclassification
of 2009 H1N1 vaccine for seasonal influenza vaccine, unique
to this season, might have contributed to some overreporting.
Finally, comparisons of estimates during 2009–10 and 2010–
11 might be affected by different data sources used: NHFS and
BRFSS for both children and adults for 2009–10, and NIS for
children and BRFSS for adults in 2010–11.

Conclusion
Compared with the 2009–10 season, estimates for 2010–11

suggest that progress was made in increasing coverage among
non-Hispanic white, black, Hispanic, and other and multiple
race children. In contrast with the past, in which non-Hispanic
white children generally had the highest coverage, estimates
for both seasons indicated that Hispanic and A/PI children
and those of other/multiple races had better coverage than
non-Hispanic white children. Despite these improvements
in coverage among historically underserved groups, Healthy
People 2020 targets for influenza vaccination of children and
adults—to increase the percentage of children aged 6 months
through 17 years and adults aged ≥18 years vaccinated to
70%—were not achieved. Efforts are needed to continue
improving coverage for all persons (18–23).

References
1. Monto AS, Kioumehr F. The Tecumseh study of respiratory illness. IX.

Occurence of influenza in the community, 1966–1971. Am J Epidemiol
1975;102:553–63.

2. Glezen WP, Couch RB. Interpandemic influenza in the Houston area,
1974–76. N Engl J Med 1978;298:587–92.

3. Glezen WP, Greenberg SB, Atmar RL, et al. Impact of respiratory virus
infections on persons with chronic underlying conditions. JAMA
2000;283:499–505.

4. Barker WH. Excess pneumonia and influenza associated hospitalization
during influenza epidemics in the United States, 1970–78. Am J Public
Health 1986;76:761–5.

5. Barker WH, Mullooly JP. Impact of epidemic type A influenza in a
defined adult population. Am J Epidemiol 1980;112:798–811.

6. Thompson WW, Shay DK, Weintraub E, et al. Influenza-associated
hospitalizations in the United States. JAMA 2004;292:1333–40.

7. Thompson WW, Shay DK, Weintraub E, et al. Mortality associated
with influenza and respiratory syncytial virus in the United States. JAMA
2003;289:179–86.

8. CDC. Estimates of deaths associated with seasonal influenza—United
States, 1976–2007. MMWR 2010;59:1057–62.

9. Cox NJ, Subbarao K. Influenza. Lancet 1999;354:1277–82.
10. CDC. Influenza vaccination coverage—United States, 2000–2010. In:

CDC health disparities and inequalities report—United States, 2011.
MMWR 2011;60(Suppl; January 14, 2011):38–41.

11. CDC. Prevention and control of influenza with vaccines: recommendations
of the Advisory Committee on Immunization Practices (ACIP), 2010.
MMWR 2010;59(No. RR-8):1–62.

12. CDC. CDC health disparities and inequalities report—United States,
2013. MMWR 2013; 62(No. Suppl 3).

13. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011; 60(Suppl; January 14, 2011).

14. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

15. CDC. Prevention and control of seasonal influenza with vaccines:
recommendations of the Advisory Committee on Immunizations
Practices (ACIP), 2009. MMWR 2009;58(No. RR-8):1–52.

16. CDC. Use of influenza A (H1N1) 2009 monovalent vaccine:
recommendations of the Advisory Committee on Immunizations Practices
(ACIP), 2009. MMWR 2009;58(No. RR-10):1–8.

17. CDC. National, state, and local area vaccination coverage among children
aged 19–35 months—United States, 2009. MMWR 2010;59:1171–7.

18. Task Force on Community Preventive Services. Recommendations
regarding interventions to improve vaccination coverage in children,
adolescents, and adults. Am J Prev Med 2000;18:92–6.

19. Santibanez TA, Santoli JM, Bridges CB, Euler GL. Influenza vaccination
coverage of children aged 6 to 23 months: the 2002–2003 and 2004
influenza seasons. Pediatrics 2006;118:1167–75.

20. Yoo BK, Berry A, Kasajima M, Szilagyi PG. Association between
Medicaid reimbursement and child influenza vaccination rates. Pediatrics
2010;126:e998–1010.

21. CDC. Influenza vaccination coverage among children and adults—
United States, 2008–09 Influenza Season. MMWR 2009;58:1091–5.

22. Lavrakas PJ, Blumberg S, Battaglia M, et al. New considerations for
survey researchers when planning and conducting RDD telephone
surveys in the US with respondents reached via cell phone numbers.
Deerfield, IL: American Association for Public Opinion Research; 2010.
Available at http://aapor.org/Cell_Phone_Task_Force.htm.

23. US Department of Health and Human Services. Healthy People 2020.
Washington, DC: US Department of Health and Human Services.
Available at http://www.healthypeople.gov/2020/.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 69

Behavioral Risk Factors

Supplement

70 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 71

Introduction
Pregnancy and childbirth among females aged <20 years

have been the subject of long-standing concern among the
public, the public health community, and policy makers (1–3).
Teenagers who give birth are much more likely than older
women to deliver a low birthweight or preterm infant, and
their babies are at higher risk for dying in infancy (4–6). The
annual public costs associated with births among teenage girls
are an estimated $10.9 billion (7). According to the 2006–2010
National Survey of Family Growth (NSFG), an estimated 77%
of births to teenagers aged 15–19 years were unintended (8).

The 2010 U.S. birth rate among females aged 15–19 was
34.2 births per 1,000. This is a 10% decrease from 2009
(37.9) and an 18% decrease from 2007 (41.5) (9). A long-
term decrease that began in 1991 was continuous except for
a brief increase during 2005–2007; the birth rate among
females aged 15–19 years decreased by 45% from 1991 (61.8
per 1,000) to 2010 (9,10). An analysis found that if the 1991
birth rates for females aged 15–19 years had remained the same
during 1992–2010, an additional 3.4 million births would
have occurred among women aged 15–19 years in the United
States (11). Significant decreases in birth rates for females aged
15–19 years occurred among all race and Hispanic ethnicity
groups from 2007 to 2010, including non-Hispanic whites,
non-Hispanic blacks, American Indian/Alaska Natives (AI/
ANs), Asians or Pacific Islanders (A/PIs), and Hispanics. Rates
also decreased for certain Hispanic groups, including those
of Mexican and Puerto Rican origin. Despite the widespread
decreases, disparities persist (9,11), and the U.S. birth rate for
females aged 15–19 years remains one of the highest among
industrialized countries (12).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR) and updates information
presented in the first CHDIR (13). The 2011 CHDIR (14) was
the first CDC report to assess disparities across a wide range of
diseases, behavioral risk factors, environmental exposures, social
determinants, and health-care access. The topics presented in
this report are based on criteria that are described in the 2013
CHDIR Introduction (15). The purposes of this pregnancy

and childbirth analysis and discussion are to highlight and raise
awareness of differences in the characteristics of females aged
<20 years (including 10–14, 15–19, 15–17, and 18–19 years)
who become pregnant and give birth and to prompt actions
to reduce these disparities.

Methods
To analyze recent trends and variations in birth rates and

pregnancy rates by selected characteristics among females aged
10–19 years, CDC examined final 2007 and 2010 natality
data from the National Vital Statistics System (NVSS) and
comparable data for earlier years. Characteristics analyzed
varied by rate calculated and included four age groups (10–14,
15–19, 15–17, and 18–19 years), race, ethnicity, and state,
including the District of Columbia. Household income and
educational attainment were not analyzed because income
information is not collected on the birth certificate, and data
on educational attainment are collected in different ways across
the states. Thus, national data on educational attainment are
not available.

Data by maternal race and Hispanic ethnicity are based
on information reported by the mother during the birth
registration process. Race and ethnicity are reported separately
on birth certificates, and persons of Hispanic origin might be of
any race. Race categories are consistent with the 1977 Office of
Management and Budget standards (4,9). In 2010, a total of 38
states and the District of Columbia reported multiple-race data
that were bridged to the single-race categories for comparability
with other states (9). Population estimates with bridged-race
categories for the rates in this report were produced under a
collaborative arrangement with the U.S. Census Bureau. Rates
for 2010 are based on the 2010 U.S. census, and rates for earlier
years are based on intercensal estimates (9,16,17). Rates are
not shown when the number of births in a given group is <20
or, for specified Hispanic groups, if <50 females are in the
denominator in the census year 2010 or <75,000 females are
in the denominator for all other years.

Birth rates were calculated as the number of births to females
per 1,000 female population in the specified age, race, and

Pregnancy and Childbirth Among Females Aged 10–19 Years —
United States, 2007–2010

Stephanie J. Ventura, MA
Brady E. Hamilton, PhD

T.J. Mathews, MS
National Center for Health Statistics, CDC

Corresponding author: Stephanie J. Ventura, Division of Vital Statistics, National Center for Health Statistics, CDC. Telephone: 301-458-4547;
E-mail: [email protected].

Supplement

72 MMWR / November 22, 2013 / Vol. 62 / No. 3

Hispanic origin group. The rate in 2010 was compared with
2007 and previously published rates for 1991. The change in
birth rate over time (i.e., relative difference) was calculated
by subtracting the rate at the end of the period from the rate
at the beginning of the period, dividing the difference by the
rate at the beginning of the period, and then multiplying by
100. Percentage change for birth rates in 2007 and 2010 was
assessed for statistical significance by using the z test at the
95% confidence level (18). Disparities were measured as the
deviations from a referent category rate. Absolute difference
was measured as the simple difference between a population
group estimate and the estimate for its respective reference
group. The relative difference, a percentage, was calculated by
dividing the absolute difference by the value in the referent
category and multiplying by 100.

Pregnancy rates are presented by pregnancy outcome (live
births, induced abortions, and fetal losses), race, and Hispanic
ethnicity. Data on live births are based on complete counts
of births provided by every state to CDC through the Vital
Statistics Cooperative Program of NVSS (18). Abortion
estimates are from abortion surveillance information on the
characteristics of women who have abortions collected from
most states by CDC; these estimates are adjusted to national
totals by the Guttmacher Institute (19–22). Fetal loss rates
are derived from pregnancy history data collected from several
cycles of the National Survey of Family Growth (NSFG)
conducted by CDC (22,23). Data regarding pregnancy are
not as current, complete, or comprehensive as NVSS data
regarding births. The most recent pregnancy estimates that
include data on live births, induced abortions, and fetal losses
are for 2008 (22).

Results
In 2010, birth rates for females aged 15–19 years varied

considerably by race and Hispanic origin (Table 1). The rates
for Hispanics (55.7 births per 1,000 females aged 15–19 years)
and non-Hispanic blacks (51.5 births) were approximately five
times the rate for A/PIs (10.9 births) and approximately twice
the rate for non-Hispanic whites (23.5 births). The rate for AI/
ANs aged 15–19 years was intermediate (38.7 births per 1,000
females aged 15–19 years). Rates varied considerably across
specified Hispanic groups. The rate in 2010 was highest for
“other” Hispanics aged 15–19 years (65.4 births per 1,000),
followed by Mexican (55.5 births), Puerto Rican (43.7 births),
and Cuban (24.4 births).

From 2007 to 2010, birth rates for females aged 15–19
years decreased significantly for all race groups and for nearly
all specified Hispanic groups (Table 1). Decreases for females

aged 15–19 years ranged from 14% to 17% for non-Hispanic
whites and non-Hispanic blacks, respectively, to 32% for
Mexicans. Among females aged 15–17 years, significant
decreases from 2007 to 2010 ranged from 16% to 21% among
non-Hispanic whites and non-Hispanic blacks, respectively, to
35% for Mexicans. Among females aged 18–19 years, decreases
ranged from 16% to 19% among non-Hispanic whites and
non-Hispanic blacks, respectively, to 30% for Mexicans. The
trends cannot be reliably analyzed for Cubans because the
numbers of births were too few and for Puerto Ricans aged
18–19 years because the estimated number of females in this
age group in 2007 was <75,000.

In 2010, birth rates among females aged 15–19 years by
state ranged from <20 per 1,000 females aged 15–19 years
in four states (New Hampshire [15.7 births], Massachusetts
[17.2 births], Vermont [17.9 births], and Connecticut [18.7
births]) to 50 per 1,000 or more in five states (Oklahoma
[50.4 births], Texas [52.2 births], Arkansas [52.5 births], New
Mexico [53.0 births], and Mississippi [55.0 births]) (Table 2)
(9,11). From 2007 to 2010, rates decreased significantly in all
but three states (Montana, North Dakota, and West Virginia).
Decreases in 16 states ranged from 20% to 30%.

The pregnancy rate for teenagers aged 15–19 years was 69.8
per 1,000 in 2008. The rates by pregnancy outcome were
40.2 for live births, 17.8 for induced abortions, and 11.8 for
fetal losses; substantial demographic differences were found
in these rates (Table 3) (22). Within each age group from
10–14 years through 18–19 years, pregnancy rates among
non-Hispanic black and Hispanic females were two to three
times higher than rates for non-Hispanic white females (22).
The rate decreased by 3% from 71.9 pregnancies per 1,000
females aged 15–19 years in 2007 to 69.8 per 1,000 in 2008
(Table 3) (22). Pregnancy rates have decreased for females
aged 10–14, 15–17, and 18–19 years and for non-Hispanic
white, non-Hispanic black, and Hispanic females aged 15–19,
15–17, and 18–19 years.

Discussion
In 2010, the U.S. birth rate for females aged 15–19 years

had decreased 45% since the 1991 peak (from 61.8 in 1991
to 34.2 in 2010) (9–11) Trends in birth rates by age and by
race and Hispanic ethnicity group indictae that the long-term
reductions since 1991 have been experienced by all population
groups but were somewhat greater for certain groups (9–11).
The birth rate for females aged 10–14 years decreased 71%,
from 1.4 per 1,000 in 1991 to 0.4 in 2010; rates for those
aged 10–14 years decreased approximately 60% in each racial
and Hispanic origin group. The birth rate for all females aged

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 73

TABLE 1. Birth rates for females aged 10–19 years, by age, race/ethnicity, and Hispanic origin of mother — National Vital Statistics System,
United States, 2007 and 2010

Characteristic

2007 2010

Change in rate
from 2007 to

2010
(%)†Birth rate*

Absolute
difference

(percentage
points)

Relative
difference

(%) Birth rate

Absolute
difference

(percentage
points)

Relative
difference

(%)

Ages 10–14 yrs
All races/ethnicities§ 0.6 — — 0.4 — — -33
White, non-Hispanic 0.2 Ref. Ref. 0.2 Ref. Ref. 0
Black, non-Hispanic 1.4 1.2 600 1.0 0.8 400 -29
Asian/Pacific Islander 0.2 0.0 0 0.1 -0.1 -50 -50
American Indian/Alaska Native 0.7 0.5 250 0.5 0.3 150 -29
Hispanic 1.2 1.0 500 0.8 0.6 300 -33
Mexican 1.2 1.0 500 0.8 0.6 300 -33
Puerto Rican 0.8 0.6 300 0.6 0.4 200 -25
Cuban NA¶ NA NA NA NA NA NA
Other Hispanic** 1.2 1.0 500 1.0 0.8 400 -17

Ages 15–19 yrs
All races/ethnicities 41.5 — — 34.2 — — -18
White, non-Hispanic 27.2 Ref. Ref. 23.5 Ref. Ref. -14
Black, non-Hispanic 62.0 34.8 128 51.5 28.0 119 -17
Asian/Pacific Islander 14.8 -12.4 -46 10.9 -12.6 -54 -26
American Indian/Alaska Native 49.4 22.2 82 38.7 15.2 65 -22
Hispanic 75.3 48.1 177 55.7 32.2 137 -26
Mexican 86.6 54.5 200 55.5 32.0 136 -32
Puerto Rican 61.8 34.6 127 43.7 20.2 86 -29
Cuban NA NA NA 24.4 0.9 4 NA
Other Hispanic 68.1 40.9 150 65.4 41.9 178 -4

Ages 15–17 yrs
All races/ethnicities 21.7 — — 17.3 — — -20
White, non-Hispanic 11.9 Ref. Ref. 10.0 Ref. Ref. -16
Black, non-Hispanic 34.6 22.7 191 27.4 17.4 174 -21
Asian/Pacific Islander 7.4 -4.5 -38 5.1 -4.9 -49 -31
American Indian/Alaska Native 26.2 14.3 120 20.1 10.1 101 -23
Hispanic 44.4 32.5 273 32.3 22.3 223 -27
Mexican 49.9 38.0 319 32.4 22.4 224 -35
Puerto Rican 32.8 20.9 176 24.2 14.2 142 -26
Cuban NA NA NA 8.7 -1.3 -13 NA
Other Hispanic 38.8 26.9 226 38.6 28.6 286 -1

Ages 18–19 yrs
All races/ethnicities 71.7 — — 58.2 — — -19
White, non-Hispanic 50.4 Ref. Ref. 42.5 Ref. Ref. -16
Black, non-Hispanic 105.2 54.8 109 85.6 43.1 101 -19
Asian/Pacific Islander 24.9 -25.5 -51 18.7 -23.8 -56 -25
American Indian/Alaska Native 86.4 36.0 71 66.1 23.6 56 -23
Hispanic 124.7 74.3 147 90.7 48.2 113 -27
Mexican 130.6 80.2 159 91.5 49.0 115 -30
Puerto Rican NA NA NA 69.7 27.2 64 NA
Cuban NA NA NA 57.8 15.3 36 NA
Other Hispanic 113.4 63.0 125 101.3 58.8 138 -11

Abbreviations: NA = not available; Ref. = referent.
* Per 1,000 females in specified age, race, and ethnicity group. Reliable birth rates cannot be computed for Cuban women in these age groups except in U.S. census years.
† Statistical testing for significance was assessed by using the z test at the 95% confidence level. All changes are significant (p<0.05) except for the “other Hispanics”

category of girls aged 15–17 years.
§ Data for persons of Hispanic origin are included in the data for each racial group according to the mother’s reported race. Race and Hispanic origin are reported

separately on birth certificates. Race categories are consistent with the 1977 Office of Management and Budget standards (available at http://www.whitehouse.
gov/omb/fedreg_race-ethnicity). Persons of Hispanic ethnicity might be of any race or combination of races. Thirty-eight states and the District of Columbia
reported multiple-race data on the birth certificate in 2010. The multiple-race data for these states were bridged to the single-race categories of the 1977 standards
for comparability with other states.

¶ Data do not meet standards of reliability or precision because 1) <20 births are in the numerator or 2) for persons of specified Hispanic origin, <75,000 females
were in the denominator in 2007.

** Includes Central American and South American as well as other and unknown Hispanic women.

Supplement

74 MMWR / November 22, 2013 / Vol. 62 / No. 3

15–17 years decreased more by approximately half, from 38.6
per 1,000 in 1991 to 17.3 in 2010; decreases ranged from
53% for Hispanics aged 15–17 years to 68%–69% for non-
Hispanic blacks and A/PIs (9). The overall decrease in the rate
for females aged 18–19 years was 38%, from 94.0 per 1,000 in
1991 to 58.2 in 2010; decreases ranged from 40% and 42%
for non-Hispanic white and Hispanic females aged 18–19
years, to 47% to 56% for non-Hispanic black, AI/AN, and
A/PI females aged 18–19 years (9,11).

The recent decreases in birth rates have coincided with
decreases in pregnancy, abortion, and fetal loss rates among
females aged 10–14, 15–19, 15–17, and 18–19 years (22). The
pregnancy rate in 2008 for females aged 15–19 years was the
lowest ever in the more than 3 decades for which a national
series of rates is available (22,24). However, disparities in rates
by race and ethnicity have changed little since 1990.

The findings in this report and a recent overview of state-level
birth rates both have documented the persistent large variation
across states (Table 2) (11). Birth rates for females aged 15–19
years tend to be highest in the South and Southwest and
lowest in the Northeast and upper Midwest, a pattern that has
persisted for many years (25,26). Some of the variation among
states reflects differences in the racial/ethnic composition of
the population within states (26). More in-depth analysis of
trends and variations in state-specific rates by race and Hispanic
ethnicity is forthcoming with the recent availability of revised
intercensal population estimates (9).

Limitations
The findings in this report are subject to at least four

limitations. First, a full assessment of disparities in childbearing
among females aged 10–19 years depends on having complete
data on patterns of pregnancies in this age group. The downward
trend since 1991 in abortions among females aged <20 years
has been more substantial than the downward trend among
births. For example, the abortion rate for females aged 15–19
years decreased 52% from 1991 to 2008, whereas the birth
rate decreased 35% during this period. A full understanding of
patterns in pregnancy among females aged <20 years requires
timely data on abortions and fetal losses as well as live births.
The birth rate decreased 15% during 2008–2010. The extent
to which the downward trends in abortions continued from
2008 to 2010 is not yet known. Second, the components of
the pregnancy estimates and pregnancy outcome estimates
vary in quality and completeness. Birth data are complete
counts, whereas the abortion estimates are based on incomplete
surveillance and survey data, and the fetal loss estimates are
based on pregnancy histories collected from survey data

TABLE 2. Number of births and birth rates for females aged 15–19
years, by state — National Vital Statistics System, United States, 2007
and 2010

State

No. of births

Change in
rate from

2007 to 2010
(%)†

Birth rate*

2010 2007 2010 2007

United States 367,678 444,899 34.2 41.5 -18
Alabama 7,343 8,696 43.6 52.2 -16
Alaska 956 1,117 38.3 43.0 -11
Arizona 9,389 12,868 41.9 59.5 -30
Arkansas 5,229 5,926 52.5 60.1 -13
California 43,149 53,417 31.5 39.6 -20
Colorado 5,474 6,737 33.4 41.5 -20
Connecticut 2,274 2,837 18.7 23.0 -19
Delaware 974 1,244 30.5 39.2 -22
District of Columbia 951 1,053 45.4 50.4 -10
Florida 19,127 25,693 32.0 42.9 -25
Georgia 14,378 18,085 41.4 53.3 -22
Hawaii 1,347 1,610 32.5 38.9 -16
Idaho 1,863 2,257 33.0 40.0 -18
Illinois 14,798 18,089 33.0 40.1 -18
Indiana 8,665 9,948 37.3 42.9 -13
Iowa 3,017 3,529 28.6 32.8 -13
Kansas 3,865 4,271 39.3 42.4 -7
Kentucky 6,684 7,547 46.2 52.4 -12
Louisiana 7,689 8,974 47.7 55.1 -13
Maine 917 1,172 21.4 26.0 -18
Maryland 5,396 6,892 27.3 34.3 -20
Massachusetts 3,909 4,949 17.2 21.4 -20
Michigan 10,835 12,497 30.1 33.5 -10
Minnesota 4,035 5,193 22.5 27.9 -19
Mississippi 6,077 7,811 55.0 69.9 -21
Missouri 7,669 9,244 37.1 44.0 -16
Montana 1,128 1,200 35.0 35.3 -1
Nebraska 1,958 2,280 31.1 35.4 -12
Nevada 3,421 4,351 38.6 51.6 -25
New Hampshire 722 924 15.7 19.3 -19
New Jersey 5,793 7,255 20.1 24.9 -19
New Mexico 3,872 4,720 53.0 63.9 -17
New York 15,126 17,621 22.7 26.1 -13
North Carolina 12,309 15,079 38.3 47.9 -20
North Dakota 659 696 28.8 29.3 -2
Ohio 13,752 16,362 34.1 39.9 -15
Oklahoma 6,496 7,543 50.4 58.5 -14
Oregon 3,496 4,343 28.2 34.6 -18
Pennsylvania 11,959 13,841 27.0 30.7 -12
Rhode Island 891 1,192 22.3 29.3 -24
South Carolina 6,849 8,329 42.6 52.0 -18
South Dakota 975 1,191 34.9 41.3 -15
Tennessee 9,254 11,260 43.2 53.3 -19
Texas 47,751 54,281 52.2 61.7 -15
Utah 3,049 3,775 27.9 35.5 -21
Vermont 401 492 17.9 21.1 -15
Virginia 7,374 9,200 27.4 34.2 -20
Washington 6,002 7,430 26.7 33.4 -20
West Virginia 2,608 2,714 44.8 46.0 -3
Wisconsin 5,100 6,243 26.2 31.2 -16
Wyoming 723 921 39.0 50.1 -22

* Births per 1,000 females aged 15–19 years living in each state.
† Statistical testing for significance was assessed by using the z test at the 95%

confidence level. All changes are significant (p<0.05) except for Montana,
North Dakota, and West Virginia.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 75

(9,19,21,22,24). Third, data on teen pregnancy are available
only for the largest population groups: non-Hispanic white,
non-Hispanic black, and Hispanic. The necessary information
on abortions and fetal losses is not available for other race
groups (i.e., A/PI or AI/AN) or for specific Hispanic groups

(19–21,23,24). Finally, evaluating trends and disparities in
state-specific birth rates depends on having accurate population
estimates by age, race, and Hispanic ethnicity. Recently released
revised intercensal population estimates provide detailed data
at the state and county level by single year of age. In years
going forward, these newly released population estimates and
estimates from the American Community Survey for Hispanic
population groups will be used to improve the precision of the
estimated rates.

Conclusion
Data from the 2006–2010 NSFG conducted by CDC have

shown little change in the proportion of males and females
aged 15–19 years who have ever had sex (27). This finding
was corroborated in data released from the 2011 Youth Risk
Behavior Surveillance report (28). For the period from 2002
to 2006–2010, NSFG found a significant decrease only
among non-Hispanic black females aged 15–19 years in the
percentage of those who were sexually experienced; changes
for other groups were not significant. However, virtually all
race and Hispanic origin groups have experienced significant
long-term decreases in the proportion of those who are sexually
experienced (27). The 2006–2010 NSFG also documents
increased use, compared with 2002 and earlier rounds of the
NSFG, of contraception at first intercourse and increased use
of two methods of contraception (i.e., condoms and hormonal
methods) among sexually active male and female teenagers (27).
The recent NSFG data show fewer differences than in previous
years by race and Hispanic origin in overall contraceptive use
at first and last sex, largely reflecting increasing condom use
among all groups (27). Various other factors contribute to the
observed variations in teenage birth rates, including differences
in education and income and in attitudes among teenagers
toward pregnancy and childbearing; these factors in turn
affect sexual activity and contraceptive use (27). The impact
of strong and consistent pregnancy prevention messages and
programs directed toward teenagers aged <20 years has been
credited with the long-term decline in teenage birth rates.
These programs were implemented in the aftermath of rapid
increases in teenage birth rates from 1986 to 1991. Studies
have shown that to be effective, programs must be designed
to meet the specific needs of different groups of teenagers, and
continually evaluating interventions and programs to assess
their effectiveness is important (1–3,29–32).

References
1. Hoffman SD, Maynard RA, editors. Kids having kids: economic and

social consequences of teen pregnancy. 2nd ed. Washington, DC: Urban
Institute Press; 2008.

TABLE 3. Pregnancy rates and rates of pregnancy outcomes (live
births, induced abortions, and fetal losses) among females aged
10–19 years, by age, race, and Hispanic origin of female — National
Vital Statistics System, National Survey of Family Growth, CDC
Abortion Surveillance System, and Guttmacher Institute surveys,*
United States, 2008

Characteristic
Pregnancy

rate†

Pregnancy outcome

Live birth
rate§

Induced
abortion

rate§
Fetal loss

rate§

Aged 10–14 yrs
All races/ethnicities¶ 1.4 0.6 0.6 0.2
White, non-Hispanic 0.5 0.2 0.2 0.1
Black, non-Hispanic 3.8 1.4 2.0 0.5
Hispanic** 2.2 1.1 0.6 0.4

Aged 15–19 yrs
All races/ethnicities 69.8 40.2 17.8 11.8
White, non-Hispanic 44.8 26.7 10.4 7.7
Black, non-Hispanic 121.6 60.4 43.4 17.8
Hispanic 111.5 70.3 20.1 21.1

Aged 15–17 yrs
All races/ethnicities 39.5 21.1 10.4 7.9
White, non-Hispanic 21.6 11.6 5.7 4.3
Black, non-Hispanic 72.8 33.6 26.7 12.6
Hispanic 69.7 42.2 11.7 15.8

Aged 18–19 yrs
All races/ethnicities 114.2 68.2 28.6 17.5
White, non-Hispanic 78.0 48.6 17.0 12.4
Black, non-Hispanic 193.8 100.0 68.2 25.6
Hispanic 176.4 114.0 33.2 29.2

* Birth data are from the National Vital Statistics System. Abortion estimates
are from abortion surveillance information collected from most states by
CDC on the characteristics of females who have abortions; these estimates
are adjusted to national totals by the Guttmacher Institute (Sources: CDC.
Abortion surveillance—United States, 2008. MMWR 2011;60[No. SS-15];
Henshaw SK. Unpublished tabulations. The Guttmacher Institute. 2000, 2011,
2012; Jones RK, Kooistra K. Abortion incidence and access to services in the
United States, 2008. Perspect Sex Reprod Health 2011;43:41–50; Ventura SJ,
Curtin SC, Abma JC, Henshaw SK. Estimates pregnancy rates and rates of
pregnancy outcomes for the United States, 1990-2008. Natl Vital Stat Rep
2012;60[7]. Available at http://www.cdc.gov/nchs/data/nvsr/nvsr60/
nvsr60_07.pdf ). Fetal loss rates are derived from pregnancy history data
collected from several cycles of the National Survey of Family Growth (NSFG)
conducted by CDC (Sources: Ventura SJ, Curtin SC, Abma JC, Henshaw SK.
Estimated pregnancy rates and rates of pregnancy outcomes for the United
States, 1990–2008. Natl Vital Stat Rep 2012;60[7]. Available at http://www.
cdc.gov/nchs/data/nvsr/nvsr60/nvsr60_07.pdf; Lepkowski JM, Mosher WD,
Davis KE, et al. The 2006–2010 National Survey of Family Growth: sample
design and analysis of a continuous survey. Vital Health Stat 2010;2[150].
Available at http://www.cdc.gov/nchs/data/series/sr_02/sr02_150.pdf ).

† Per 1,000 females in specified age, race, and Hispanic origin group. Rates
cannot be calculated for other population groups because the necessary
data for abortions and fetal losses are not available.

§ Per 1,000 females in specified age, race, and Hispanic origin group.
¶ Rates for “all races/ethnicities” include other races not shown separately and

origin not stated.
** Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

76 MMWR / November 22, 2013 / Vol. 62 / No. 3

2. National Campaign to Prevent Teen and Unplanned Pregnancy. National
data. Washington, DC: National Campaign to Prevent Teen Pregnancy;
2010. Available at http://www.thenationalcampaign.org/national-data/
default.aspx.

3. CDC. Adolescent reproductive health, teen pregnancy. Atlanta, GA: US
Department of Health and Human Services, CDC; 2012. Available at
http://www.cdc.gov/TeenPregnancy/index.htm.

4. Martin JA, Hamilton BE, Ventura SJ, et al. Births: final data for 2009.
Natl Vital Stat Rep 2011;60. Available at http://www.cdc.gov/nchs/data/
nvsr/nvsr60/nvsr60_01.pdf.

5. Martin JA, Osterman MJK, Sutton PD. Are preterm births on the decline
in the United States? Recent data from the National Vital Statistics
System. NCHS data brief no 39. Hyattsville, MD: CDC, National
Center for Health Statistics; 2010. Available at http://www.cdc.gov/
nchs/data/databriefs/db39.pdf.

6. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2008
period linked birth/infant death data set. Natl Vital Stat Rep 2012;60.
Available at http://www.cdc.gov/nchs/data/nvsr/nvsr60/nvsr60_05.pdf.

7. National Campaign to Prevent Teen and Unplanned Pregnancy.
Counting it up: the public costs of teen childbearing. Key data.
Washington, DC: National Campaign to Prevent Teen and Unplanned
Pregnancy; 2012. Available at http://www.thenationalcampaign.org/
costs/pdf/counting-it-up/key-data.pdf.

8. Mosher WD, Jones J, Abma JC. Intended and unintended births in the
United States: 1982–2010. Natl Health Stat Rep 2012;55. Hyattsville,
MD: National Center for Health Statistics.

9. Martin JA, Hamilton BE, Ventura SJ, et al. Births: Final data for 2010.
Natl Vital Stat Rep 2012;61. Available at http://www.cdc.gov/nchs/data/
nvsr/nvsr61/nvsr61_01.pdf.

10. Ventura SJ, Hamilton BE. US teenage birth rate resumes decline. NCHS
data brief no. 58. Hyattsville, MD: National Center for Health Statistics;
2011. Available at http://www.cdc.gov/nchs/data/databriefs/db58.pdf.

11. Hamilton BE, Ventura SJ. Birth rates for U.S. teenagers reach historic
lows for all and ethnic groups. NCHS data brief no. 89. Hyattsville,
MD: National Center for Health Statistics. 2012. Available at http://
www.cdc.gov/nchs/data/databriefs/db89.pdf.

12. United Nations Statistics Division. Demographic yearbook 2009–2010.
New York, NY: United Nations; 2011. Available at http://unstats.un.org/
unsd/demographic/products/dyb/dyb2009-2010.htm.

13. CDC. Adolescent pregnancy and childbirth—United States, 1991–2008.
In: CDC health disparities and inequalities report—United States, 2011.
MMWR 2011;60(Suppl; January 14, 2011).

14. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60 (Suppl; January 14, 2011).

15. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

16. CDC, National Center for Health Statistics. Estimates of the April 1,
2010, resident population of the United States, by county, single year
of age, bridged race, Hispanic origin, and sex. Hyattsville, MD: CDC,
National Center for Health Statistics; 2011. Available at http://www.
cdc.gov/nchs/nvss/bridged_race/data_documentation.htm#april2010.

17. Hamilton BE, Martin JA, Ventura SJ. Births: Preliminary data for 2010.
Natl Vital Stat Rep 2011;60. Available at http://www.cdc.gov/nchs/data/
nvsr/nvsr60/nvsr60_02.pdf.

18. CDC, National Center for Health Statistics. User guide to the 2010 natality
public use file. Hyattsville, MD: CDC, National Center for Health
Statistics; 2010. Available at ftp://ftp.cdc.gov/pub/Health_Statistics/
NCHS/Dataset_Documentation/DVS/natality/UserGuide2010.pdf.

19. CDC. Abortion surveillance—United States, 2008. MMWR 2011;
60(No. SS-15).

20. Henshaw SK. Unpublished tabulations. The Guttmacher Institute. 2000,
2011, 2012.

21. Jones RK, Kooistra K. Abortion incidence and access to services in the
United States, 2008. Perspect Sex Reprod Health 2011;43:41–50.

22. Ventura SJ, Curtin SC, Abma JC, Henshaw SK. Estimated pregnancy
rates and rates of pregnancy outcomes for the United States, 1990–2008.
Natl Vital Stat Rep 2012;60: Available at http://www.cdc.gov/nchs/data/
nvsr/nvsr60/nvsr60_07.pdf.

23. Lepkowski JM, Mosher WD, Davis KE, et al. The 2006–2010 National
Survey of Family Growth: sample design and analysis of a continuous
survey. Vital Health Stat 2010;2(150). Available at http://www.cdc.gov/
nchs/data/series/sr_02/sr02_150.pdf.

24. Ventura SJ, Mosher WD, Curtin SC, et al. Trends in pregnancies and
pregnancy rates by outcome: Estimates for the United States, 1976–96.
Vital Health Stat 2000;21(56). Available at http://www.cdc.gov/nchs/
data/series/sr_21/sr21_056.pdf.

25. Ventura SJ, Mathews TJ, Hamilton BE. Births to teenagers in the United
States, 1940–2000. Natl Vital Stat Rep 2001;49: Available at http://
www.cdc.gov/nchs/data/nvsr/nvsr49/nvsr49_10.pdf.

26. Mathews TJ, Sutton PD, Hamilton BE, Ventura SJ. State disparities in
teenage birth rates in the United States. NCHS data brief no. 46.
Hyattsville, MD: National Center for Health Statistics; 2010. Available
at http://www.cdc.gov/nchs/data/databriefs/db46.pdf.

27. Martinez G, Copen CE, Abma JC. Teenagers in the United States: sexual
activity, contraceptive use, and childbearing, 2006–2010 National Survey
of Family Growth. National Center for Health Statistics. Vital Health
Stat 2011;23(31). Available at http://www.cdc.gov/nchs/data/series/
sr_23/sr23_031.pdf.

28. CDC. Youth risk behavior surveillance—United States, 2011. MMWR
2012;61(No. SS-4).

29. Oringanje C, Meremikwu MM, Eko H, Esu E, Meremikwu A, Ehiri
JE. Interventions for preventing unintended pregnancies among
adolescents. Cochrane Database Syst Rev 2009;4:CD005215.

30. Suellentrop K. What works 2011–2012: curriculum-based programs
that help prevent teen pregnancy. Washington, DC: National Campaign
to Prevent Teen and Unplanned Pregnancy; 2011. Available at http://
www.thenationalcampaign.org/resources/pdf/pubs/WhatWorks.pdf.

31. Kirby D. Emerging answers 2007: research findings on programs to
reduce teen pregnancy and sexually transmitted diseases. Washington,
DC: National Campaign to Prevent Teen and Unplanned Pregnancy;
2007. Available at http://www.thenationalcampaign.org/EA2007/
EA2007_full.pdf.

32. CDC. Guide to community preventive services. Youth development
behavioral interventions coordinated with community service to reduce
sexual risk behaviors in adolescents. Atlanta, GA: CDC; 2007. Available
at http://www.thecommunityguide.org/hiv/supportingmaterials/
IS-youthdev-community.html.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 77

Binge Drinking — United States, 2011
Dafna Kanny, PhD

Yong Liu, MS
Robert D. Brewer, MD

Hua Lu, MS
National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Dafna Kanny, Division of Population Health, National Center for Chronic Disease Prevention and Health Promotion, CDC.
Telephone: 770-488-5411; E-mail: [email protected].

Introduction
During 2001–2005, excessive alcohol use accounted for an

estimated average of 80,000 deaths and 2.3 million years of
potential life lost (YPLL) in the United States each year, and an
estimated $223.5 billion in economic costs in 2006 (1). Binge
drinking, defined as consuming four or more alcoholic drinks
on one or more occasions for women and five or more drinks
on one or more occasions for men, was responsible for more
than half of these deaths, two-thirds of the YPLL (2), and three
quarters of the economic costs (1). Reducing the prevalence of
binge drinking among adults is also a leading health indicator
in Healthy People 2020 (objective SA-14.3) (3).

The binge drinking prevalence, frequency, and intensity
analysis, and discussion that follows is part of the second
CDC Health Disparities and Inequalities Report (CHDIR)
(4). The 2011 CHDIR (5) was the first CDC report to
assess disparities across a wide range of diseases, behavioral
risk factors, environmental exposures, social determinants,
and health-care access. The topic presented in this report is
described in the criteria for the 2013 CHDIR Introduction
(6). This report provides more current information on binge
drinking measures, and updates information on the status
of evidence-based strategies recommended to prevent binge
drinking presented in the 2011 CHDIR. The purposes of
this report are to discuss and raise awareness of differences in
the characteristics of people who binge drink, and to prompt
actions to reduce these disparities.

Methods
To examine sociodemographic disparities in binge drinking

nationwide and by state, CDC analyzed 2011 data from the
Behavioral Risk Factor Surveillance System (BRFSS). BRFSS
is a state-based, random-digit–dialed landline and cellular
telephone survey of the noninstitutionalized civilian U.S.
adults that is conducted monthly in all states, the District of
Columbia (DC), and three U.S. territories. BRFSS collects
data on leading health conditions and health risk behaviors,
including binge drinking. For this report, responses to questions

regarding the prevalence, frequency, and largest number of
drinks consumed by binge drinkers (a measure of the intensity
of binge drinking) were analyzed, beginning with the question,
“Considering all types of alcoholic beverages, how many times
during the past 30 days did you have X [X = 5 for men; X = 4
for women] or more drinks on an occasion?” Respondents
then were asked, “During the past 30 days, what is the largest
number of drinks you had on any occasion?” Responses to this
question were assessed for binge drinkers only. A more detailed
description of BRFSS methods has been published (4,7). In
2011, the median survey response rate* was 49.7%, ranging
from 33.8% to 64.1%. After excluding 48,912 persons who
reported ‘don’t know/not sure’ or ‘refused,’ those with missing
information, and respondents from the U.S. territories, data
from 457,555 respondents in the 50 states and DC were used
for analysis.

This report describes binge drinking prevalence, frequency
(i.e., the average number of binge drinking episodes), and
intensity (i.e., the average largest number of drinks consumed
by binge drinkers). Sociodemographic characteristics analyzed
included sex, age group, race/ethnicity, education level,
income level, and disability status. Race was defined as white,
black, Asian/Pacific Islander, and American Indian/Alaska
Native. Ethnicity was defined as Hispanic or non-Hispanic.
Annual household income was defined as follows: <$25,000,
$25,000–$49,999, $50,000–$74,999, and ≥$75,000.
Educational attainment was defined as follows: less than high
school, high school or equivalent, some college, and college
graduate. Disability status was defined as respondents reporting
limited activities in any way because of physical, mental, or
emotional problems.

Binge drinking prevalence was calculated by dividing
the total number of respondents who reported at least one
binge drinking episode during the preceding 30 days by
the total number of BRFSS respondents in all 50 states and

* Response rates for BRFSS are calculated using standards set by the American
Association of Public Opinion Research (AAPOR) response rate formula no. 4,
available at http://www.aapor.org/standard_definitions2.htm. The response
rate is the number of respondents who completed the survey as a proportion
of all eligible and likely eligible persons. Additional information is available at
http://cdc.gov/brfss/pdf/2011_Summary_Data_Quality_Report.pdf.

Supplement

78 MMWR / November 22, 2013 / Vol. 62 / No. 3

DC. Frequency of binge drinking (i.e., the number of binge
drinking episodes) was calculated by averaging the number of
episodes reported by all binge drinkers during the preceding 30
days. Intensity of binge drinking was calculated by averaging
the largest number of drinks consumed by binge drinkers
during the past 30 days. BRFSS data were weighted to adjust
for several demographic variables (e.g., educational attainment,
marital status, home ownership, and telephone source) (7).
Data were age- and sex-adjusted to the 2000 U.S. Census
standard population to provide estimates for race/ethnicity,
educational attainment, annual household income level, and
disability status. We calculated 95% confidence intervals for
binge drinking prevalence. Two-tailed t-tests were used to
determine differences between subgroups.

Results
In 2011, the overall prevalence of binge drinking among

adults in the 50 states and DC was 18.4% (Table). On average,
binge drinkers reported a frequency of 4.1 binge drinking
episodes during the preceding 30 days and an intensity of 7.7
drinks per occasion during the past 30 days. Binge drinking
prevalence was significantly higher among persons aged 18–24
years (30.0%) and 25–34 years (29.7%) than among those in
older age groups. Similarly, the intensity of binge drinking
was highest among binge drinkers aged 18–24 and 25–34
(8.9 and 8.2 drinks, respectively); however, the frequency of
binge drinking was highest among binge drinkers aged ≥65
years (4.9 episodes). The prevalence of binge drinking was also
significantly higher among non-Hispanic whites (21.1%) than
among all other race/ethnicity categories, but the intensity of
binge drinking was highest among American Indians/Alaska
Natives (8.4 drinks). Those with household incomes ≥$75,000
had significantly higher binge drinking prevalence (22.2%)
than those with lower household incomes. In contrast, binge
drinkers with household incomes <$25,000 reported the
highest frequency (4.3 episodes) and intensity (7.1 drinks) of
binge drinking.

Respondents who did not graduate from high school
reported significantly lower binge drinking prevalence (16.8%)
than those with high school or higher education. However,
binge drinkers with less than high school education had the
highest frequency (4.7 episodes) and intensity (7.4 drinks)
of binge drinking. Respondents with disabilities also had a
significantly lower prevalence of binge drinking (16.9%), but
those who binge drank had a higher frequency (4.5 episodes)
and intensity (7.2 drinks) of binge drinking, compared with
those without disabilities.

Overall, areas with the highest age- and sex-adjusted
prevalence of binge drinking were states in the Midwest, as well
as DC and Hawaii (Figure 1). States with the highest intensity
of binge drinking were generally located in the Midwest, and
included some states (e.g. Oklahoma, Arkansas, Kentucky,
West Virginia, and Utah) that had a lower prevalence of binge
drinking (Figure 2).

Discussion
Binge drinking is a risk factor for many adverse health

and social outcomes, including unintentional injuries (e.g.,
motor vehicle crashes); violence; suicide; hypertension;
acute myocardial infarction; sexually transmitted diseases;
unintended pregnancy; fetal alcohol syndrome; and sudden
infant death syndrome (8). This report indicates that in
2011 binge drinking was common among U.S. adults, and
persons who binge drank tended to do so frequently (average
of four times per month) and with high intensity (average of
eight drinks on occasion), placing themselves and others at a
significantly greater risk for alcohol-attributable harms (8). In
a number of states with a lower prevalence of binge drinking,
those who binge drank did so with high intensity.

The groups at highest risk for binge drinking (i.e., persons
aged 18–34 years, males, whites, non-Hispanics, and persons
with higher household incomes), and those who reported
the highest binge drinking frequency (i.e., binge drinkers
aged ≥65 years) and intensity (i.e., persons aged 18–24
years) are consistent with previous reports (4,9), and might
reflect differences in state and local laws on the marketing of
alcoholic beverages (e.g., price and availability) (10), as well
as other cultural and religious factors (11). These differences
are reflected in state measures of the prevalence and intensity
of binge drinking, and highlight that states with a lower
prevalence of binge drinking might still include subgroups
that binge drink with high intensity. Furthermore, unlike other
leading health risks (e.g., smoking and obesity) binge drinking
has not been widely recognized as a health risk or subjected to
intense prevention efforts (12).

Limitations
The findings in this analysis are subject to at least three

limitations. First, BRFSS data are self-reported; alcohol
consumption, generally, and excessive drinking, in particular,
is underreported in surveys because of recall bias, social
desirability response bias, and nonresponse bias (13). A recent
study reported that BRFSS identifies 22%–32% of presumed
alcohol consumption in states when compared with alcohol

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 79

sales data (14). However, a strong correlation existed between
BRFSS estimates of alcohol consumption and per capita
alcohol sales in states, suggesting that BRFSS data are still a
useful measure of alcohol consumption even after taking into
account known underreporting (14). Second, response rates
for BRFSS were low, which can increase response bias. Third,
BRFSS does not collect information from persons living in
institutional settings (e.g., on college campuses), so findings
might not be representative of those populations.

Conclusion
Binge drinking is common among U.S. adults, and persons

who binge drink tend to do so frequently and with high
intensity. The Community Preventive Services Task Force

has recommended several population-level, evidence-based
strategies to reduce binge drinking and related harms (15).
These include 1) limiting alcohol outlet density (http://
www.thecommunityguide.org/alcohol/outletdensity.html)
(i.e., the concentration of retail alcohol establishments,
including bars and restaurants and liquor or package stores,
in a given geographic area), 2) holding alcohol retailers liable
for harms related to the sale of alcoholic beverages to minors
and intoxicated patrons (dram shop liability), 3) measures
increasing the price of alcohol, 4) maintaining existing limits
on the days and hours when alcohol is sold, 5) avoiding further
privatization of alcohol sales in states with government-operated
or contracted liquor stores, and 6) electronic screening and brief
interventions (eSBI), including interventions delivered using
computers, telephones, or mobile devices in the clinical setting.

TABLE. Prevalence, frequency, and intensity of binge-drinking, by sex, age group, race/ethnicity, education, and disability — Behavioral Risk
Factor Surveillance System, United States,* 2011

Characteristic

Prevalence† Frequency§ Intensity¶

No.
Weighted

% (95% CI) No.
No. of

episodes (95% CI) No.
No. of
drinks (95% CI)

Total 457,555 18.4 (18.1–18.6) 59,553 4.1 ( 4.0– 4.2) 55,929 7.7 ( 7.6– 7.7)
Sex**

Men 179,224 24.6 (24.2–25.0) 34,859 4.6 ( 4.0 – 4.7) 32,564 8.7 ( 8.6– 8.8)
Women 278,331 12.5 (12. 2–12.8) 24,694 3.2 ( 3.1– 3.3) 23,365 5.7 ( 5.6– 5.8)

Age group (yrs)**
18–24 20,016 30.0 (28.9-31.1) 6,210 4.4 ( 4.1- 4.6) 5,792 8.9 ( 8.7- 9.1)
25–34 44,441 29.7 (28.9-30.5) 12,167 3.8 ( 3.7- 4.0) 11,493 8.2 ( 8.0- 8.4)
35–44 58,980 21.1 (20.5-21.8) 11,781 3.9 ( 3.8- 4.1) 11,158 7.4 ( 7.2- 7.5)
45–64 187,811 14.1 (13.8-14.5) 23,710 4.2 ( 4.1- 4.3) 22,293 6.6 ( 6.5- 6.7)
≥65 146,307 4.3 ( 4.1- 4.5) 5,685 4.9 ( 4.5- 5.3) 5,193 5.6 ( 5.5- 5.7)
Race/Ethnicity††

White, non-Hispanic 363,127 21.1 (20.7-21.4) 47,879 4.1 ( 4.0- 4.2) 45,255 6.8 ( 6.8- 6.9)
Black, non-Hispanic 35,919 14.2 (13.4-15.0) 3,446 3.8 ( 3.5- 4.1) 3,111 6.1 ( 5.9- 6.3)
Hispanic§§ 28,275 17.7 (16.9-18.4) 4,338 3.3 ( 3.0- 3.6) 3,978 6.8 ( 6.6- 7.0)
Asian/Pacific Islander 8,746 10.3 ( 9.1-11.4) 885 3.4 ( 2.5- 4.3) 839 6.1 ( 5.7- 6.5)
American Indian/Alaska Native 6,248 18.2 (16.1-20.4) 992 4.5 ( 3.7- 5.3) 906 8.4 ( 7.8- 9.1)

Educational attainment††
Less than high school 39,348 16.8 (15.9–17.6) 3,888 4.7 ( 4.3– 5.1) 3,482 7.4 ( 7.2– 7.7)
High school or equivalent 133,510 18.7 (18.2–19.1) 16,670 4.2 ( 4.1– 4.4) 15,455 7.2 ( 7.0– 7.3)
Some college 124,124 20.1 (19.6–20.6) 17,353 4.0 ( 3.8 –4.1) 16,344 6.6 ( 6.5– 6.7)
College graduate 159,762 20.4 (20.0–20.9) 21,593 3.3 ( 3.2– 3.4) 20,611 6.2 ( 6.1– 6.3)

Annual household income ($)††
<25,000 118,636 17.6 (17.0–18.1) 12,656 4.3 ( 4.1 –4.5) 11,733 7.1 ( 7.0– 7.2)
25,000–49,999 107,486 19.4 (18.8–19.9) 13,748 4.2 ( 4.0– 4.4) 12,945 6.9 ( 6.8– 7.0)
50,000–74,999 63,510 19.8 (19.0–20.5) 9,370 3.7 ( 3.5– 3.9) 8,916 6.7 ( 6.6– 6.9)
≥75,000 107,907 22.2 (21.6–22.9) 18,820 3.6 ( 3.4 –3.7) 17,998 6.5 ( 6.4– 6.6)

Disability status††
Yes 131,816 16.9 (16.2–17.5) 11,592 4.5 ( 4.3– 4.7) 10,833 7.2 ( 7.0– 7.4)
No 323,525 19.6 (19.3–19.9) 47,763 3.8 ( 3.7– 3.9) 44,933 6.7 ( 6.6 6.7)

Abbreviation: 95% CI = 95% confidence interval.
* Respondents were from all 50 states and the District of Columbia.
† Prevalence = total number of respondents who reported at least one binge drinking episode during the past 30 days divided by the total number of respondents.
§ Frequency = average number of binge-drinking episodes reported by all binge drinkers during the past 30 days.
¶ Intensity = average largest number of drinks consumed by binge drinkers on any occasion during the past 30 days.
** Unadjusted estimates.
†† Age- and sex-adjusted to the 2000 U.S. Census standard population.
§§ Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

80 MMWR / November 22, 2013 / Vol. 62 / No. 3

FIGURE 1. Prevalence* of binge drinking† — Behavioral Risk Factor
Surveillance System, United States,§ 2011

* Total number of respondents who reported at least one binge drinking episode
during the preceding 30 days divided by the total number of respondents.

† Consuming ≥4 alcoholic drinks on ≥1 occasion for women and ≥5 drinks on
≥1 occasion for men.

§ States are divided into tertiles.

20.5–25.1
17.9–20.4
10.9–17.8

DC

Prevalence (%)

Despite the array of strategies that have been recommended,
efforts are needed to implement them to a point of measurable
success toward reducing binge drinking. The frequency and
intensity of binge drinking also should be monitored routinely
to support the implementation and evaluation of Community
Guide recommendations for reducing binge drinking and to
monitor changes in this behavior among groups at greater risk.

References
1. Bouchery EE, Harwood HJ, Sacks JJ, Simon CJ, Brewer RD. Economic

costs of excessive alcohol consumption in the United States, 2006. Am
J Prev Med 2011;41:516–24.

2. CDC. Alcohol attributable deaths and years of potential life lost—United
States, 2001. MMWR 2004;53:866–70.

3. US Department of Health and Human Services. SA–14.3 Reduce the
proportion of persons engaging in binge drinking during the past 30
days—adults aged 18 years and older. Healthy People 2020. Washington,
DC: US Department of Health and Human Services; 2011. Available
at http://healthypeople.gov/2020/lhi/substanceabuse.aspx.

4. CDC. Binge drinking—United States, 2009. MMWR 2011;60(Suppl;
January 14, 2011).

5. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011; 60(Suppl; January 14, 2011).

6. CDC. Introduction: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013; 62(Suppl; No. 3).

7. CDC. Methodologic changes in the behavioral risk factor surveillance
system in 2011 and potential effects on prevalence estimates. MMWR
2012;61;410–3.

8. National Institute of Alcohol Abuse and Alcoholism. Tenth special report
to the US Congress on alcohol and health. Bethesda, MD: National
Institutes of Health; 2000.

9. CDC. Vital signs: binge drinking prevalence, frequency, and intensity
among adults—United States, 2010. MMWR 2012;61:14–9.

10. National Institute on Alcohol Abuse and Alcoholism. Alcohol policy
information system. Rockville, MD: US Department of Health and
Human Services, National Institutes of Health. Available at http://www.
alcoholpolicy.niaaa.nih.gov.

11. Holt JB, Miller JW, Naimi TS, Sui DZ. Religious affiliation and alcohol
consumption in the United States. Geog Rev 2006;96:523–42.

12. Brewer RD, Swan MH. Binge drinking and violence. JAMA 2005;
294:616–8.

13. Stockwell T, Donath S, Cooper-Stanbury M, et al. Under-reporting of alcohol
consumption in household surveys: a comparison of quantity-frequency,
graduated-frequency and recent recall. Addiction 2004; 99:1024–33.

14. Nelson DE, Naimi TS, Brewer RD, Roeber J. US state alcohol sales
compared to survey data, 1993–2006. Addiction 2010;105:1589–96.

15. Task Force on Community Preventive Services. Preventing excessive
alcohol use. Atlanta, GA: Task Force on Community Preventive Services;
Available at http://www.thecommunityguide.org/alcohol/index.html.

FIGURE 2. Intensity* of binge drinking† — Behavioral Risk Factor
Surveillance System, United States,§ 2011

* Average largest number of drinks consumed by binge drinkers on any occasion
during the past 30 days.

† Consuming ≥4 alcoholic drinks on ≥1 occasion for women and ≥5 drinks on
≥1 occasion for men.

§ States are divided into tertiles.

7.1–7.8
6.8–7.0
6.0–6.7

DC

Most drinks
consumed on
any occasion

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 81

Introduction
Cigarette smoking is the leading cause of preventable disease

and death in the United States, resulting in approximately
443,000 deaths and $193 billion in direct health-care
expenditures and productivity losses each year (1). Declines
in smoking prevalence would significantly impact the health-
care and economic costs of smoking. Efforts to accelerate the
decline in cigarette smoking include reducing cigarette smoking
disparities among specific population groups. Findings from
the previous report on cigarette use in the first CDC Health
Disparities and Inequalities Report (CHDIR) indicated that
progress has been achieved in reducing disparities in cigarette
smoking among certain racial/ethnic groups (2). However,
little progress has been made in reducing disparities in cigarette
smoking among persons of low socioeconomic status (SES)
and low educational attainment.

This report on cigarette smoking and the analysis and
discussion that follows is part of the second CHDIR. The
2011 CHDIR (3) was the first CDC report to take a broad
view of disparities across a wide range of diseases, behavioral
risk factors, environmental exposures, social determinants, and
health-care access. The topic presented in this report is based
on criteria that are described in the 2013 CHDIR Introduction
(4). The report that follows provides more current information
to what was presented in the 2011 CHDIR (2). The purposes
of this report are to discuss and raise awareness of differences
in the smoking prevalence of current smokers and to prompt
actions to reduce disparities.

Methods
To assess the changes in disparities in smoking prevalence

by selected sociodemographic characterisitcs during 2006-
2008 and 2009-2010, CDC analyzed aggregated data from
the National Survey on Drug Use and Health (NSDUH),
which is sponsored by the Substance Abuse and Mental
Health Services Administration (SAMHSA) and provides
annual data on alcohol, tobacco, and illegal drug use among
the noninstitutionalized U.S. household population aged ≥12
years (http://www.sahmhsa.gov/data/NSDUH.aspx). Smoking

prevalence was determined for youths and adults (aged ≥12
years). Current smokers include persons who reported smoking
at least one cigarette during the 30 days before the survey.

Aggregated data were analyzed for two survey cycles. The
2006–2008 survey cycle included 42,693 respondents with
response rates of 74.0%, 73.9%, and 74.2%, respectively. The
2009–2010 survey cycle included 27,636 respondents with
response rates of 75.7% and 74.4%, respectively. Demographic
characteristics analyzed included race and ethnicity, sex, age,
household income, employment status, and educational
attainment. Geographic location was not analyzed because of
limited data for this variable. Race and ethnicity were defined as
non-Hispanic white, non-Hispanic black, Hispanic, American
Indian/Alaska Native, Native Hawaiian/Other Pacific Islander,
Asian, and multirace. Household income was reported by
poverty status, which is based on U.S. Census Bureau thresholds
for federal poverty levels (FPL) (http://www.census.gov/hhes/
www/poverty/html). Employment status was defined as fulltime,
parttime, unemployed, and other. Educational attainment
was defined as less than high school, high school diploma or
equivalent, some college, and college graduate. For adults,
low-SES was defined as those persons with less than a high
school diploma unemployed or living at, near, or below the
U.S. FPL. Disparities were measured as the absolute difference
between rates. Population-weighted prevalence estimates and
95% confidence intervals (CIs) were calculated using statistical
software to account for the multistage probability designs of
NSDUH. No statistical testing was done for this analysis. In
this approach, CIs were used as measure of variability and
nonoverlapping CIs were considered statistically different. Using
CIs in this way is a conservative evaluation of significance
differences; infrequently, this might lead to a conclusion that
estimates are similar when the point estimates do differ.

Results
Some progress in reducing smoking prevalence among

certain racial/ethnic groups was observed; however, disparities
among persons with low-SES persisted. For both youth and
adults, little to no changes in smoking prevalence for those

Cigarette Smoking — United States, 2006-2008 and 2009-2010
Bridgette E. Garrett, PhD

Shanta R. Dube, PhD
Cherie Winder, MSPH

Ralph S. Caraballo, PhD
National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Bridgette E. Garrett, PhD, Office on Smoking and Health, National Center for Chronic Disease Prevention and Health Promotion, CDC.
Telephone: 770-488-5715; E-mail: [email protected].

Supplement

82 MMWR / November 22, 2013 / Vol. 62 / No. 3

below FPL was observed from 2006–2008 to 2009–2010;
however, decreases were observed for youth and adults who
were above FPL (Tables 1 and 2). During 2009–2010, the
prevalence of smoking was 46.4% among 12th-grade–aged
youth who had dropped out of school compared with 21.9%
among youth who were still in the 12th grade (Table 1). Among
adults, smoking prevalence was 34.6% for those who did not
graduate from high school compared with 13.2% among
those with a college degree (Table 2). From 2006–2008 to
2009–2010, smoking declined from 44.7% to 40.9% among
adults who were unemployed (Table 2). Among racial/ethnic
groups, smoking prevalence was lowest among black and
Asian youth aged 12–17 years during both survey cycles
(Table 1). Although smoking prevalence remained highest
among American Indian/Alaska Native youth and adults,
smoking declined from 17.2% to 13.6% in youth and from
42.2% to 34.4% in adults (Table 1 and 2).

Discussion
Prevalence of smoking is highest for persons aged ≥18 years

who do not have high school diploma. Assessing and reporting
the prevalence of smoking among youth aged <18 years who
drop out of school is critical because this is the period when
problems with academic achievement occur. The findings in
this report indicate that during 2009–2010, approximately
half of youth who dropped out of school were smokers. These
findings underscore the need to address tobacco use early in the
life span, particularly among school-aged youth, who might
be more vulnerable, to eliminate tobacco-related disparities.
Implementing the key effective strategies known to prevent
and reduce tobacco use among youth are needed, including
reducing tobacco industry influences towards minors,
particularly those in low SES communities (5).

To make progress toward reducing the persistent higher
prevalence of smoking among low-SES populations, current
tobacco-control interventions should be targeted toward these

TABLE 1. Prevalence of current smoking* among persons aged 12–17 years, by selected characteristics — National Survey on Drug Use and
Health, United States, 2006–2010†

Characteristic

2006–2008 2009–2010 Absolute difference from
2006–2008 to 2009–2010

(percentage points)% (95% CI) % (95% CI)

Sex
Male 9.7 (9.2–0.2) 8.9 (8.3–9.5) -0.8
Female 9.9 (9.4–0.4) 8.4 (7.8–8.9) -1.5

Race/Ethnicity
White, non-Hispanic 11.8 (11.4–2.3) 10.2 (9.6–10.8) -1.6
Black, non-Hispanic 5.9 (5.2–6.5) 5.0 (4.2–5.7) -0.9
Hispanic§ 7.4 (6.7–8.2) 7.7 (6.9–8.4) 0.3
American Indian/Alaska Native 17.2 (13.2–1.2) 13.6 (9.6–17.7) -3.6
Native Hawaiian/Other Pacific Islander 5.2 (1.7–8.8) 7.9 (0.515.2) 2.7
Asian 4.1 (3.0–5.3) 3.0 (1.4–4.5) -1.1
Multirace 12.1 (9.5–14.7) 11.2 (8.5–13.9) -0.9

Grade
≤5 1.2 (0.8–1.6) 1.2 (0.6–1.7) 0
6 1.8 (1.4–2.2) 1.2 (0.8–1.6) -0.6
7 4.6 (3.85.4) 3.5 (2.7–4.2) -1.1
8 8.0 (7.3–8.7) 7.3 (6.6–8.0) -0.7
9 12.1 (11.1–13.0) 10.8 (9.8–11.8) -1.3
10 16.3 (15.3–17.2) 14.0 (12.715.3) -2.3
11 18.8 (17.1–20.4) 16.5 (14.7–18.3) -2.3
12 19.0 (14.024.0) 21.9 (15.9–27.9) 2.9

High school dropout 45.7 (40.7–50.7) 46.4 (39.4–53.5) 0.7
Poverty status¶

<100% (below threshold) 10.4 (9.4–11.3) 9.6 (8.6–10.6) -0.8
100%–199% (at or near threshold) 10.7 (10.0–11.5) 9.6 (8.8–10.3) -1.1
≥200% (above threshold) 9.3 (8.9–9.7) 7.9 (7.4–8.5) -1.4

Abbreviation: 95% CI = 95% confidence interval.
* Current smokers include all persons who smoked at least one cigarette during the 30 days before the survey.
† N = 42,693 for 2006–2009; N = 27,636 for 2009–2010.
§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Based on self-reported family income or imputed family income and poverty thresholds published by the U.S. Census Bureau, 2005-2009. Available at http://www.

census.gov/hhes/www/poverty.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 83

more vulnerable smokers. Educating the public about the
harms of tobacco use through mass media campaigns is an
effective strategy for raising awareness and decreasing smoking
prevalence in the general population (6). Advertisements that
are emotionally provocative and contain personal testimonies are
especially effective in reaching low-SES populations (7). CDC
recently implemented its first paid national media campaign
to encourage smokers to quit (www.cdc.gov/quitting/tips).
Mass media campaigns can be most effective in reaching all
populations when they are part of a comprehensive tobacco-
control program that includes comprehensive smoke-free policies
that make all indoor public places 100% smoke-free, increase
tobacco price, counter tobacco industry marketing activities,
and increase the availability and accessibility of evidence-based
cessation services (6,8).

Limitations
The findings in this report are subject to at least five

limitations. First, data were based on self-reports and were
not validated biochemically. However, studies have indicated
that self-reported smoking status validated by measured
serum cotinine levels yield similar prevalence estimates (9).
Second, the NSDUH questionnaire is administered only in
English and Spanish; therefore, estimates for certain racial/
ethnic populations might be underestimated if neither English
nor Spanish is the primary language spoken. Moreover,
race/ethnicity was not adjusted by socioeconomic status.
Third, because NSDUH does not include institutionalized
populations and persons in the military, these results might not
be generalizable to these groups. Fourth, although smoking

TABLE 2. Prevalence of current smoking* among persons aged ≥18 years, by selected characteristics — National Survey on Drug Use and
Health, United States, 2006–2010†

Characteristic

2006–2008 2009–2010 Absolute difference from
2006–2008 to 2009–2010

(percentage points)% (95% CI) % (95% CI)

Age group (yrs)
18–25 36.8 (36.3–37.4) 35.0 (34.2-35.8) -1.8
26–34 33.7 (32.8–34.7) 33.6 (32.4-34.9) -0.1
35–49 28.1 (27.5–28.8) 26.1 (25.1-27.1) -2.0
50–64 22.9 (21.8–23.9) 22.4 (21.1-23.7) -0.5
≥65 9.4 (8.5–10.4) 9.2 (8.1–10.3) -0.2
Sex
Male 29.2 (28.6–29.8) 27.5 (26.8–28.3) -1.7
Female 23.0 (22.5–23.5) 22.4 (21.7–23.1) -0.6
Race/Ethnicity

White, non-Hispanic 26.9 (26.4–27.3) 25.8 (25.1–26.6) -1.1
Black, non-Hispanic 26.9 (25.6–28.1) 25.4 (23.9–27.0) -1.5
Hispanic§ 22.9 (21.7–24.1) 22.9 (21.3–24.5) 0
American Indian/Alaska Native 42.2 (35.5–48.8) 34.4 (27.9–40.9) -8.0
Native Hawaiian/Other Pacific Islander 28.5 (20.9–36.1) 18.6 (11.5–25.8) -9.9
Asian 14.7 (13.0–16.4) 11.8 (9.9–13.6) -2.9
Multirace 35.2 (31.4–39.0) 33.2 (29.1–37.2) -2.0

Educational attainment
Less than high school 34.3 (33.0–35.6) 34.6 (33.3–35.9) 0.3
High school graduate or equivalent 31.1 (30.3–32.0) 30.4 (29.4–31.4) -0.7
Some college 27.1 (26.3–28.0) 25.6 (24.6–26.5) -1.5
College graduate 14.1 (13.4–14.8) 13.2 (12.4–13.9) -0.9

Employment status
Full-time 27.8 (27.2–28.4) 25.4 (24.7–26.1) -2.4
Part-time 24.5 (23.5–25.4) 24.2 (23.1–25.4) -0.3
Unemployed 44.7 (42.3–47.2) 40.9 (39.2–42.7) -3.8
Other (including not in work force) 20.9 (20.2–21.7) 20.7 (19.6–21.8) -0.2

Poverty status¶
<100% (below threshold) 36.5 (35.1–37.8) 37.9 (36.4–39.4) 1.4
100%–199% (at or near threshold) 32.8 (31.8–33.8) 31.5 (30.3–32.7) -1.3
≥200% (above threshold) 22.5 (21.9–23.0 20.5 19.9–21.0 -2.0

Abbreviation: 95% CI = 95% confidence interval.
* Current smokers include all persons who smoked at least one cigarette during the 30 days before the survey.
† N = 42,693 for 2006–2008; N = 27,636 for 2009–2010.
§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶ Based on self-reported family income or imputed family income and poverty thresholds published by the U.S. Census Bureau, 2005-2009. Available at http://www.

census.gov/hhes/www/poverty.

Supplement

84 MMWR / November 22, 2013 / Vol. 62 / No. 3

prevalence was determined to be lowest among Asian and
Hispanic women, variations in smoking prevalence have been
observed with specific Asian and Hispanic groups (e.g., Korean
and Vietnamese men and Puerto Rican men and women) (10).
Finally, because of limited sample sizes for certain population
groups (e.g., AI/AN), single-year estimates might have resulted
in imprecise estimates.

Conclusion
Comprehensive tobacco-control strategies should be

implemented in an equitable manner to be effective in
addressing tobacco-related disparities. These strategies should
ensure that all populations are covered by comprehensive
smoke-free policies, including workplaces, restaurants, and
bars; prices are increased on all tobacco products and coupled
with access to evidence-based cessation services; exposure to
industry advertising, promotions, and sponsorship are reduced
among all populations; and the availability, accessibility, and
effectiveness of tailored cessation services are increased for all
populations (11).

The findings in this report underscore conclusions from the
2011 CHDIR that efforts to reduce future tobacco-related
disparities associated with low SES should take a lifespan
approach (2). Specifically, continuing population-based
strategies that target youth, particularly among those with
low academic achievement and drop-outs, will be critical in
preventing future tobacco-related disparities. Coordinated,
multicomponent interventions that combine mass media
campaigns, price increases including those that result from tax
increases, school-based policies and programs, and statewide
or community-wide changes in smoke-free policies and
norms are effective in reducing the initiation, prevalence, and
intensity of smoking among youth and young adults (5,6).
Finally, addressing the social determinants of health (e.g.,
socioeconomic status, cultural characteristics, acculturation,
stress, targeted advertising, price of tobacco products, and
varying capacities of communities to mount effective tobacco-
control initiatives) will be necessary to disrupt the cycle of
smoking among low-SES populations (2,12,13).

References
1. CDC. Smoking-attributable mortality, years of potential life lost, and

productivity losses—United States, 2000–2004. MMWR 2008;
57:1226-8.

2. CDC. Cigarette smoking—United States, 1965-2008. In: CDC health
disparities and inequalities report—United States, 2011. MMWR
2011;60(Suppl; January 14, 2011).

3. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

4. CDC. Introduction: CDC Health Disparities and Inequalities Report—
United States, 2013. In: CDC Health Disparities and Inequalities
Report—United States, 2013. MMWR 2013;62(No. Suppl 3)

5. US Department of Health and Human Services, Office of the Surgeon
General. Preventing tobacco use among youth and young adults; a report
of the Surgeon General. Atlanta, GA: US Department of Health and
Human Services, CDC; 2012.

6. Institute of Medicine. Ending the tobacco problem: a blueprint for the
nation. Washington, DC: Institute of Medicine, 2007.

7. Niederdeppe J, Kuang X, Crock B, Skelton A. Media campaigns to
promote smoking cessation among socioeconomically disadvantaged
populations; what do we know, what do we need to learn, and what
should we do now? Social Science & Medicine 2008;67:1343-55.

8. National Cancer Institute. The role of media in promoting and reducing
tobacco use. Tobacco Control Monograph No. 19. Bethesda, MD: US
Department of Health and Human Services, National Institutes of
Health, National Cancer Institute, NIH Pub. No. 07-6242, June 2008.

9. Caraballo RS, Giovino GA, Pechacek TF, Mowery PD. Factors associated
with discrepancies between self-reports on cigarette smoking and
measured serum cotinine levels among persons aged 17 years or older:
Third National Health and Nutrition Examination Survey, 1988-1994.
Am J Epidemiol 2001;53:807-14.

10. Caraballo RS, Yee SL, Gfroerer J, Mirza SA. Adult tobacco use among
racial and ethnic groups living in the United States, 2002–2005. Prev
Chronic Dis 5. Available at http://www.cdc.gov/pcd/issues/2008/
jul/07_0116.htm.

11. CDC. Best practices for comprehensive tobacco-control programs.
Atlanta, GA: U.S. Department of Health and Human Services, CDC;
2007. Available at http://www.cdc.gov/tobacco/tobacco_control_
program/stateandcommunity/best_practices.

12. Koh HK, Piotrowski JJ, Kumanyika S, Fielding JE. Healthy People a
2020 vision for the social determinants approach. Health Education and
Behavior 2011;38:1207-12.

13. U.S. Department of Health and Human Services, Office of the Surgeon
General. Tobacco use among U.S. racial/ethnic minority groups—
African Americans, American Indians and Alaska Natives, Asian
Americans and Pacific Islanders, and Hispanics: a report of the Surgeon
General. Atlanta, GA: U.S. Department of Health and Human Services,
CDC; 1998.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 85

Health Outcomes: Morbidity

Supplement

86 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 87

Introduction
Over the 20th century, the U.S. population has witnessed

major changes in fatal and nonfatal health outcomes. Mortality
has declined, and life expectancy has increased continuously;
chronic conditions have replaced acute diseases as leading causes
of both illness and death (1). During 1900–2008, average life
expectancy at birth for the total U.S. population increased from
47.3 years in 1900 to 78.1 years in 2008 (2), a gain of 30.8 years.
In addition, an increasing proportion of the U.S. population is
aged >65 years. According to the U.S. Census Bureau estimates,
at the beginning of the 20th century, the U.S. population aged
>65 years constituted only 4.1 percent of the total population;
by 2008, the percentage of the total U.S. population aged
>65 years was 12.8% (3,4). However, declines in mortality
are not necessarily associated with declines in morbidity or
the consequences of chronic conditions on life activities. The
possibility that longer life might be accompanied by poor
health makes it essential to develop measures that account for
both mortality and morbidity at the same time. Hence, over
the past 40 years, a new set of health measures (e.g., “healthy
life expectancies”) have been developed that account for both
mortality and life spent free of the consequences of ill health.
One of these newly developed set of measures (called “active
life expectancy”) is the average number of years expected to be
lived without activity limitations.

In general, being “active” entails the continuing participation
of a person in social, economic, cultural, spiritual, and civic
affairs (5). In health studies, the context in which “being
active” has been used has varied depending on the population
group under study (6–9). In this report, “active” is used to
differentiate between a person with limitations in social roles
and one without such limitations. This analysis focuses on
activity limitations caused by chronic conditions. Active life
expectancy or active life at any age is defined as the remaining
years of life free of activity limitations (YFAL) caused by
chronic conditions.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (10) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in

this report is based on criteria that are described in the 2013
CHDIR Introduction (11). This report provides information
on disparities in YFAL as a result of chronic conditions, a topic
that was not discussed in the 2011 CHDIR. The purposes of
this report are to discuss and raise awareness of differences
in the characteristics of persons who experience chronic
condition–induced physical activity limitations and to prompt
actions to reduce these disparities.

Methods
To assess disparities in YFAL as a result of chronic conditions,

CDC analyzed data from the National Vital Statistics System
(NVSS) and the National Health Interview Survey (NHIS).
Demographic variables analyzed included sex and race. Period
life tables for males and females and for the white and black
populations of the United States for each year from 1999
through 2008 come from CDC’s National Center for Health
Statistics (NCHS). Expected years free of chronic condition–
induced activity limitations by ethnicity, environmental, or
behavior risk factors and socioeconomic determinants of
access to health care were not included in this analysis because
officially released NCHS life expectancy estimates by these and
other similar factors for the years 1999 through 2008 were
not available. Hence, because officially released annual life
tables by ethnicity were not available for all the 10 years of the
study period, the expected YFAL for Hispanics, non-Hispanic
whites, and non-Hispanic blacks were not analyzed separately.
The racial category “white” includes person of Hispanic origin
who identified themselves as white, and the racial category
“black” includes persons of Hispanic origin who identified
themselves as black.

Data on activity status come from NHIS, which defines an
activity limitation as a limit on a person’s ability to perform
activities normally expected of someone of his or her age.
Depending on how they answered questions on activity status,
survey respondents were classified into four categories: 1) not
limited, 2) unable to perform major activity, 3) limited in kind
or amount of major activity, and 4) limited in other activities.
Whenever any form of activity limitation was identified,
NHIS survey participants also were asked the health condition
causing the limitation, and the cause was classified as chronic

Expected Years of Life Free of Chronic Condition–Induced Activity
Limitations — United States, 1999–2008

Michael T. Molla, PhD
National Center for Health Statistics, CDC

Corresponding author: Michael T. Molla, National Center for Health Statistics, CDC. Telephone: 301-458-4379; E-mail: [email protected].

Supplement

88 MMWR / November 22, 2013 / Vol. 62 / No. 3

or nonchronic. Conditions were considered chronic if they
cannot be cured once acquired or had existed continuously
for >3 months after onset (12).

Expected years of life with and without chronic condition–
induced activity limitations were estimated by using a
demographic-epidemiologic model (13–16) that combined
average life expectancy and the prevalence of being with and
without chronic condition–induced activity limitation. The
expected years without any activity limitations then were used
to calculate the percentage of remaining life expected to be
lived without such limitations. Expected years of life with and
without chronic condition–induced activity limitations were
estimated separately for four population subgroups: males,
females, whites, and blacks. The definition of standard error of
expected YFAL and the statistical test and level of significance
used have been summarized (Appendix).

Disparities were measured as the deviations from a “referent”
category rate or prevalence. Absolute difference was measured
as the simple difference between the rate for a population
subgroup and the rate for its respective reference group. The
relative difference, a percentage, was calculated by dividing the
absolute difference by the value in the referent category and
multiplying by 100. Whether a disparity in expected YFAL
between a male and a female or between a white person and
a black person of the same age was statistically significant was
tested by using a 2-tailed test at the 95% level of significance.
A hypothesis of equality was rejected if the value of the absolute
value of the z-score exceeded 1.96.

Results
During 1999–2008, total life expectancy improved (Table 1).

During this 10-year period, total life expectancy at birth for
males increased by 1.7 years, from 73.9 years in 1999 to 75.6
years in 2008, and female life expectancy at birth increased by 1.2
years, from 79.4 years in 1999 to 80.6 years in 2008. Expectation
of life at birth for the white population increased by 1.2 years,
from 77.3 years in 1999 to 78.5 years in 2008. Life expectancy
for the black population increased by 2.6 years, from 71.4 years
in 1999 to 74.0 years in 2008.

The percentage of total life expectancy that was estimated to
be spent free of chronic condition–induced activity limitations
fluctuated over the 10-year period, with the percentage of life
expectancy spent free of activity limitations slightly lower in 2008
than in 1999. For males, the percentage of life expected to be
free of activity limitations was 86.5% in 1999 and declined to
86.1% in 2008. In 1999, blacks expected to spend about 82.9%

of their total expected life free of activity limitations compared
with 82.6% in 2008. For whites, the percentage of life expected
to be spent free of activity limitations declined from 85.6% in
1999 to 85.4% in 2008, while it remained almost the same for
females (84.5% in 1999 and 84.4% in 2008).

In 1999, males would expect to spend 63.9 of their 73.9 years
of life expectancy free of chronic condition–induced activity
limitations compared with 67.1 years out of 79.4 years of
total expected years of life free of such limitations for females.
The white and black populations would expect to spend 66.2
years of 77.3 years of total life expectancy and 59.2 years of
71.4 years of total life expectancy respectively free of chronic
condition–induced activity limitations.

By 2008, males would expect to live 65.1 years out of total
expected years of life of 75.6 years free of activity limitations.
Females would expect to live 68.0 years of the total life
expectancy of 80.6 years limitation-free. The white and black
populations would expect to live 67 activity limitation–free
years (out of a total life expectancy of 78.5 years) and 61.1
activity limitation–free years (out of a total life expectancy of
74.0 years) respectively.

Over the 10 years, improvements in the expected YFAL were
observed (Table 2). The increase in expected YFAL caused by
chronic conditions during the 10-year period was 1.2 years for
males (from 63.9 years in 1999 to 65.1 years in 2008), 0.9 years
for females (from 67.1 years in 1999 to 68.0 years in 2008),
0.8 years for whites (from 66.2 years in 1999 to 67.0 years in
2008), and 1.9 years for blacks (from 59.2 in 1999 to 61.1
years in 2008). In the 10-year period, the black population
had the largest increase both in life expectancy at birth as well
as in expected YFAL caused by chronic conditions.

The changes in the differences in YFAL caused by chronic
conditions between males and females and between whites
and blacks from 1999 to 2008 have been calculated (Figure).
In 1999, the difference in expected YFAL at birth between
the white and the black populations was 7 years (Figure),
and the difference between males and females was 3.2 years.
After 10 years, the difference between the white and the black
populations had decreased to 5.9 years, and the difference
between males and females had dropped to 2.9 years.

During the 10-year period, the disparity between the white
and the black populations declined by 1.1 years, and the
disparity between males and females dropped by slightly more
than 0.3 years. However, the observed disparities in expected
YFAL caused by chronic conditions between men and women
as well as that between whites and blacks remained statistically
significant at the 5% level throughout the 10-year period
(Table 2).

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 89

Discussion
Between 1999 and 2008, life expectancy and expected YFAL

at birth increased for all the population subgroups indicating
the expected improvements both in mortality and quality of
life. For males, life expectancy at birth increased from 73.9
in 1999 to 75.6 years in 2008. For females, life expectancy
increased from 79.4 years in 1999 to 80.4 years in 2008. For
the white population, life expectancy increased from 77.3 years
in 1999 to 78.5 years in 2008, and for the black population,
it increased from 71.4 years in 1999 to 74.0 years in 2008.
During the same 10-year period, expected male YFAL at birth
increased from 63.9 years in 1999 to 65.1 years in 2008. For
females, expected YAFL at birth increased from 67.1 years in
1999 to 68.0 years in 2008. For the white population, expected
YFAL at birth increased from 66.2 years in 1999 to 67.0 years
in 2008, and for the black population, it increased from 59.2
years in 1999 to 61.1 years in 2008.

Significant disparities in expected YFAL existed between
males and females as well as between the white and the black

populations throughout the 10-year period. However, the
disparities in expected YFAL have been declining throughout
the course of the 10-year period. Disparities in YFAL between
males and females decreased by 0.3 year, from 3.2 years in 1999
to 2.9 years in 2008. In the same period, disparities between
the white and the black populations decreased by more than
one year from 7.0 years in 1999 to 5.9 years in 2008. These
results are consistent with results of other similar studies and
federal government health reports (17–18).

The 10-year health initiative Healthy People 2010 had as its
two overarching goals when it was launched in 2000 increasing
the quality and years of healthy life of the U.S. population
and eliminating health disparities. The final assessment of
this initiative has concluded that during the 10-year period
2000–2010, life expectancy improved for the populations that
could be assessed; women had longer life expectancy than men,
and the white population had a longer life expectancy than
the black population; and differences were observed both by
race and sex in life expectancy measure (at birth) and expected
YFAL. On the basis of data from 2006–2007, on average, the

TABLE 1. Life expectancy at birth and expected years free of activity limitations caused by chronic conditions, by sex and race — United States,
1999–2008

Year

Male Female White Black

LE YFAL
YFAL as
% of LE LE YAFL

YFAL as
% of LE LE YAFL

YFAL as
% of LE LE YAFL

YFAL as
% of LE

1999 73.9 63.9 86.5 79.4 67.1 84.5 77.3 66.2 85.6 71.4 59.2 82.9
2000 74.1 64.3 86.8 79.3 67.7 85.4 77.3 66.6 86.2 71.8 60.2 83.8
2001 74.2 64.0 86.3 79.4 67.5 85.0 77.4 66.4 85.8 72.0 59.5 82.6
2002 74.3 64.1 86.3 79.5 67.2 84.5 77.4 65.3 84.4 72.1 59.4 82.4
2003 74.5 64.5 86.6 79.5 67.2 84.5 77.6 66.5 85.7 72.3 59.6 82.4
2004 74.9 64.8 86.5 79.9 67.8 84.9 77.9 66.9 85.9 72.8 60.0 82.4
2005 74.9 64.9 86.6 79.9 68.1 85.2 77.9 67.0 86.0 72.8 61.1 83.9
2006 75.1 65.1 86.7 80.2 68.4 85.3 78.2 67.3 86.1 73.2 61.2 83.6
2007 75.4 65.1 86.3 80.4 68.0 84.6 78.4 67.1 85.6 73.6 60.2 81.8
2008 75.6 65.1 86.1 80.6 68.0 84.4 78.5 67.0 85.4 74.0 61.1 82.6

Abbreviations: LE = life expectancy at birth; YFAL = years free of activity limitations.
Source: National Vital Statistics System and National Health Interview Survey, 1999–2008.

TABLE 2. Difference in expected years free of activity limitations caused by chronic conditions, by sex and race — United States, 1999–2008

Year

Expected YFAL

Difference (yrs)* 

Expected YFAL

Difference (yrs)* Male Female White Black

1999 63.9 67.1 3.2 66.2 59.2 7.0
2000 64.3 67.7 3.4 66.6 60.2 6.4
2001 64.0 67.5 3.5 66.4 59.5 6.9
2002 64.1 67.2 3.1 65.3 59.4 5.9
2003 64.5 67.2 2.7 66.5 59.6 6.9
2004 64.8 67.8 3.0 66.9 60.0 6.9
2005 64.9 68.1 3.2 67.0 61.1 5.9
2006 65.1 68.4 3.3 67.3 61.2 6.1
2007 65.1 68.0 2.9 67.1 60.2 6.9
2008 65.1 68.0 2.9 67.0 61.1 5.9

Abbreviation: YFAL = years free of activity limitations.
Source: National Vital Statistics System and National Health Interview Survey, 1999–2008.
* Implies statistically significant difference in expected years without activity limitations at p<0.05.

Supplement

90 MMWR / November 22, 2013 / Vol. 62 / No. 3

U.S. population was expected to spend 66.2 years of their
entire lives free of activity limitations (19).

During 1999–2008, expected YFAL caused by chronic
conditions increased for both males and females and for
both blacks and whites. During the entire 10 years, although
disparities in expected YFAL existed between males and females
as well as between whites and blacks, the extent of these
disparities declined during the 10-year period.

Limitations
The findings provided in this report are subject to at least two

limitations. First, estimates of expected YFAL caused by chronic
conditions are based on current life expectancy estimates and
the prevalence of activity limitations. Annual life expectancy
estimates are based on the total U.S. population whereas
prevalence rates on activity limitations come from NHIS,
which does not include the institutionalized population of the
United States. However, because the size of this population
is very small compared with the total household population,
the effect of the exclusion of the group on the comparison of
estimates over time is assumed to be minimal. Second, estimates
in this analysis might have been sensitive to the operational
definition of expected YFAL caused by chronic conditions.
Activity limitation is part of a larger continuum process known

as the “disablement process.” Hence, whenever measures such
as activity limitations induced by chronic conditions (which are
discrete in nature) are used, cut-off points on the continuum
have to be determined to differentiate those with and without
limitations. These cut-off points are functions of the operational
definitions and might vary from one study to another. Although
the estimates could be sensitive to these operational definitions,
the effect of such definitions on the comparison of estimates
over time is assumed to be minimal.

Conclusion
The findings provided in this report indicate that during

the 10-year period 1999–2008, while disparities in expected
years free of chronic condition caused activity limitations still
existed between males and females as well as between the white
and the black populations, expected YFAL increased for all
four population subgroups studied, and disparities decreased.
Increasing the length of life, improving the quality of life, and
eliminating health disparities among population groups have
been the major health goals of all the Healthy People initiatives
since the decade-long health programs were initiated with
the publication of the Surgeon General ‘s Report on Health
Promotion and Disease Prevention in 1979 (20).

FIGURE. Disparities in expected (at birth) years free of activity limitations caused by chronic conditions, by race and sex — United States, 1999
and 2008

Source: National Vital Statistics System and National Health Interview Survey, 1999 and 2008.

2.9 years

5.9 years

3.2 years

7.0 years

0 1 2 3 4 5 6 7 8

Female vs. male

White vs. black

1999
2008

Year

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 91

The overarching goals of the first Healthy People initiative
were to decrease mortality and increase independence among
older adults by 1990. The first two of the three major goals of
the second 10-year health initiative (Healthy People 2000) were
to increase the span of healthy life and to reduce disparities
in health status (21). Healthy People 2010 focused on two
major goals: to increase quality of years of healthy life and to
eliminate health disparities. Healthy People 2020, the fourth
10-year national health initiative, has four major goals, one
of which focuses on achieving health equity by eliminating
health disparities.

The findings of this report as well as reports of the first
three Healthy People programs demonstrate that expected
years of life are getting longer, health-related quality of life is
improving, and health disparities between population groups
are decreasing. However, group comparisons also demonstrate
that disparities in mortality (as measured by expectation of
life at birth) and health-related quality of life (as measured by
expected YFAL caused by chronic conditions) still exist.

Reference
1. Fries JF. Aging, natural death, and the compression of morbidity. N Engl

J Med 1980;303:130–5.
2. Miniño AM, Murphy SL, Xu J, Kochanek KD. Death: final data for

2008. Natl Vital Stat Rep 2011;59(10).
3. US Census Bureau. Resident population estimates of the United States

by age and sex: April 1, 1990 to July 1, 1999, with short-term projection
to November 1, 2000. Washington, DC: US Census Bureau; 2001.

4. US Census Bureau. Resident population—estimates by age, sex, and
race: July 1, 1900 (excluding Alaska and Hawaii). Washington, DC: US
Census Bureau; 2004.

5. World Health Organization. Active ageing: a policy framework: a
contribution of the second United Nations World Assembly on Ageing.
Geneva, Switzerland: World Health Organization; 2002.

6. Katz S, Branch LG, Branson MH, Papsidero JA, Beck JC, Greer DS.
Active life expectancy. N Engl J Med 1983;309:1218–24.

7. Manton KG, Stallard E. Cross-sectional estimates of active life expectancy
for the U.S. elderly and oldest-old population. J Gerontol 1991;
46(Suppl):S170–82.

8. Crimmins EM, Hayward MD, Saito Y. Differentials in active life
expectancy in the older population of the United States. J Gerontol
1996;51(Suppl):S111–20.

9. Wolf DA, Laditka SB, Laditka JN. Patterns of active life among older
women: differences within and between groups. J Women Aging
2002;14:9–26.

10. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

11. CDC. Introduction: In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

12. CDC. 2008 National Health Interview Survey (NHIS): public use
data release. NHIS survey description. Hyattsville, MD: US
Department of Health and Human Services, CDC, National Center
for Health Statistics; 2009.

13. Sanders BS. Measuring community health levels. Am J Public Health
1964;54:1063–70.

14. Sullivan DF. A single index of mortality and morbidity. Health Services
and Mental Health Administration. Health Reports 1971;86:347–54.

15. Rogers A, Rogers RG, Belanger A. Longer life but worse health?
Measurement and dynamics. Gerontologist 1990;30:640–9.

16. Molla MT, Wagener DK, Madans JH. Summary measures of population
health: methods for calculating healthy life expectancy. Healthy People
2010 Statistical Notes No 21. Hyattsville, MD: US Department of Health
and Human Services, CDC, National Center for Health Statistics; 2001.

17. Manton KG, Gu X, Lamb VL. Long-term trends in life expectancy and active
life expectancy in the United States. Popul Dev Rev 2006;32:81–105.

18. Molla MT, Madans JH. Life Expectancy free of chronic condition
induced activity limitations among white and black Americans, 2000–
2006. Vital Health Stat 2010;3(34).

19. CDC. Healthy people 2010: final review. Atlanta, GA: US Department
of Health and Human Services, US Public Health Service, CDC; 2011.
Available at http://www.cdc.gov/nchs/data/hpdata2010/hp2010_final_
review.pdf.

20. Department of Health, Education, and Welfare, Public Health Service.
Healthy people: the Surgeon General’s report on health promotion and
disease prevention. Washington, DC: Government Printing Office; 1979.

21. US Department of Health and Human Services. Healthy people 2010:
understanding and improving health. Washington, DC: US Department
of Health and Human Services; 2000.

Supplement

92 MMWR / November 22, 2013 / Vol. 62 / No. 3

Standard Errors for Expected Years
Free of Activity Limitations

Expected years free of activity limitations (YFAL) ( ) at
age x is defined as the remaining years of life that is free of
limitations caused by chronic conditions and is given by:

[1]
where

is the remaining years free of activity limitation due to
chronic conditions for persons who have reached age x;

lx is the number of survivors at age x;

(1 – nπx) represents the age-specific health state free of
activity limitations due to chronic conditions;

nLx is the total number of years lived by a cohort in the
age interval (x, x+n); and

ω is the oldest age category.

The variance and standard errors of the estimated YFAL
can be calculated based on the variances of the prevalence
rates of the different health states. Within each age group,
the prevalence of each health state is a proportion with an
associated standard error. Since there are only two health states,
the variance of the health state with activity limitations equals
to the variance of its complement. The variance S2 of (nπx )
or (1- nπx) is given by the variance of a binomial distribution
as follows:

xxxnxnxn NS /)1([)(
2 πππ −= ]. [2]

where nNx is the number of persons in the interval (x, x+n)
of the sample from which the prevalence rates were computed.

Equation 2 can then be used to calculate the variance of
expected YFAL, ‘xe using the following formula:

)].1([
1

)( 22
2


in

xi
in

x
x SLl

eVAR π
ϖ

−= ∑
=

[3]

The standard error of the expected YFAL caused by chronic
conditions at age x is simply the square root of its variance.

Test of Significance
Disparities between the expected years of life free of activity

limitations of two population subgroups of the same age group
can be tested by using a statistical method commonly used
for testing the significance of a difference between two means
using the following formula:

)( ‘ 2,

1,
2


2,


1,

xx

xx

eeS

ee
z


=

, [4]

where, ‘ 1,xe and

2,xe are the expected YFAL of two different
population subgroups of the same age x.

The critical value of a z-score for a 2-tailed test at the 95%
level of significance is 1.96, i.e., the hypothesis of equality is
rejected if the absolute value of z exceeds 1.96.

Appendix
Definition of Terms Used in This Report


xe

,)1(
1′

inin
x

x Ll
e ∑ −= π


xe

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 93

Introduction
Asthma is a chronic inflammatory disorder of the airways that

is characterized by episodic and reversible airflow obstruction,
airway hyper-responsiveness, and underlying inflammation.
Common asthma symptoms include wheezing, coughing, and
shortness of breath (1). With correct treatment and avoidance
of exposure to environmental allergens and irritants that are
known to exacerbate asthma, the majority of persons who have
asthma can expect to achieve optimal symptom control (2).

Multiple reports published previously provide detailed
surveillance information on asthma (1,3–8). A 1987 report
that included asthma surveillance data for 1965–1984
identified differences among certain demographic groups by
age, sex, and race/ethnicity (5). Subsequent asthma surveillance
reports confirmed these differences and documented that the
differences have persisted over time (1,3,4,6). These reports
indicate that population-based asthma prevalence rates,
emergency department visit rates, and hospitalization rates
were higher among blacks than among whites, higher among
females than among males, higher among children (aged 0–17
years) than among adults (aged ≥18 years), and higher among
males aged 0–17 years than among females in the same age
group. In addition, more detailed analysis of ethnicity data
demonstrated that asthma health outcomes differed among
Hispanic groups. Hispanics of Puerto Rican descent (origin or
ancestry) had higher asthma prevalence and death rates than
other Hispanics (e.g., Hispanics of Mexican descent), non-
Hispanic blacks, and non-Hispanic whites (7,8).

Current asthma prevalence rates among the demographic
groups for the years covered in this report were similar to those
in previous CDC reports (1,3,4,6). During 2006–2010, an
estimated 8.0% of the U.S. population had current asthma.
Asthma prevalence varied by demographic group: 6.9% among
males, 9.0% among females, 9.4% among children, 7.6%
among adults, 7.9% among whites, 10.5% among blacks,
10.8% among American Indians/Alaska Natives, 5.0% among
Asians, 14.4% among multi-race/other-race persons, 15.9%
among Puerto Ricans, and 5.4% among Mexicans.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (9) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (10). This report provides information
regarding asthma attacks among persons with current asthma
that supplements information about current asthma prevalence
provided in the 2011 CHDIR (4). The purposes of this report
are to discuss and raise awareness of differences in asthma
attacks among persons with current asthma and to prompt
actions to reduce these disparities.

Methods
To examine whether disparities in asthma attacks exist

among persons with current asthma by selected demographic
characteristics, CDC analyzed data from the 2001–2010
National Health Interview Survey (NHIS). NHIS is an annual,
in-person survey of the civilian, noninstitutionalized U.S.
population based on a multistage sampling of households
(11). An adult family member is selected to act as a proxy
respondent for children aged 0–17 years. NHIS includes several
questions about asthma. The first question, “Have you ever
been told by a doctor or other health professional that you
had asthma?” has been used as a lifetime prevalence measure
for asthma since 1997. A second question, “Do you still have
asthma?” was added in 2001. Respondents are considered
to have current asthma if they answer “yes” to both of these
questions. A response of “yes” to a third question, “During
the past 12 months, have you had an episode of asthma or an
asthma attack?” indicates an attack in the past year and was used
in this analysis as an indicator of symptom control (1,3,4,6).

The percentage of persons with current asthma who reported
an asthma attack in the past year, crude prevalence ratios,
and adjusted prevalence ratios were estimated for selected
demographic characteristics: race/ethnicity, sex, age (children
aged 0–17 years, adults aged ≥18 years, and eight age groups),

Asthma Attacks Among Persons with Current Asthma —
United States, 2001–2010

Jeanne E. Moorman, MS
Cara J. Person, MPH

Hatice S. Zahran, MD
Division of Environmental Hazards and Health Effects, National Center for Environmental Health, CDC

Corresponding author: Jeanne Moorman, Division of Environmental Hazards and Health Effects, National Center for Environmental Health, CDC.
Telephone: 770-488-3726; E-mail: [email protected].

Supplement

94 MMWR / November 22, 2013 / Vol. 62 / No. 3

educational attainment for adults, place of birth, geographic
region, and federal poverty level (FPL). Estimated percentages
with standard errors and prevalence ratios with 95% confidence
intervals (CIs) are presented for 2006–2010. Estimated
percentages with standard errors for more limited demographic
groups are presented for both 2001–2004 and 2006–2010 for
a historical comparison. Race/ethnicity is categorized on the
basis of the respondents’ self-reported race and ethnicity. Non-
Hispanic race groups include white, black, American Indian/
Alaska Native, Asian, and other or multiple races. Persons of
Hispanic ethnicity might be of any race or combination of
races. Hispanic subgroups include Puerto Rican, Mexican, and
other Hispanic. FPL is based on U.S. Census Bureau poverty
thresholds. The poverty threshold is based on the size of the
family and the ages of family members. Income divided by the
poverty threshold is called the “income-to-poverty ratio” (12).

Multiple years of survey data were combined to provide
stable estimates for relatively small respondent groups. If
the relative standard error was >30%, or if the sample size
(denominator) was <50, estimates were considered unreliable
and were suppressed. Analysis software was used to account for
the complex survey design, and sample weights were used to
produce national estimates. A multivariate (binary response)
logistic regression model was used to determine the association
(adjusted prevalence ratio [APR]) between reporting an asthma
attack in the past year and demographic variables including
age, sex, race/ethnicity, educational attainment, FPL, place of
birth, and geographic region. A univariate logistic regression
model was used to determine the association (crude prevalence
ratio) between an asthma attack and each variable separately.
The Wald chi-square test statistic was used for all logistic
regression models to test for an association between the
dependent variable (asthma attack status) and independent
variables of interest. Chi-square tests and z-tests were used to
test for demographic group and time period differences. All
statistical tests were 2-sided, with p<0.05 denoting statistical
significance. Comparative terms used to describe findings in
this report (e.g., “higher” and “similar”) indicate the results of
statistical testing at p<0.05.

Results
During 2006–2010, reported attacks among those with

current asthma were higher for females (53.5%) than for males
(48.8%) (Table 1). The difference in reporting an asthma attack
by sex was not statistically significant among children (adjusted
prevalence ratio [APR]: 1.1) but was significant among adults
(APR: 1.4) after adjusting for age, race/ethnicity, educational
attainment, federal poverty level, place of birth, and geographic

region (Table 2). Overall, asthma attacks were reported more
frequently for children (56.1%) than for adults (49.6%)
(Table 1). Asthma attacks were reported more frequently for
children aged 0–4 years (APR: 1.9) and 5–11 years (APR: 1.3)
than for children aged 12–17 years (Table 2). Among adults,
persons aged 18–34 (APR: 1.4), 35–44 (APR: 2.0), 45–54
(APR: 1.9), and 55–64 years (APR: 1.6) were more likely to
report asthma attacks than persons aged ≥65 years. Regardless
of age, persons with asthma living in the West (54.5%) and
the South (53.1%) were more likely to report asthma attacks
than persons living in the Midwest (49.4%) and the Northeast
(47.8%) (Table 1). The differences in reporting an asthma
attack between the South and West regions compared with the
reference region (Northeast) remained statistically significant
after adjusting for covariates (Table 2). No significant
interaction terms or multicollinearity effects were identified
among any of the variables in the final model.

For children, reporting an asthma attack did not differ
significantly by poverty level (range: 56.3% [FPL <100%]–
57.8% [FPL ≥450%]) (Table 1). However, for adults, reporting
an asthma attack did differ significantly by poverty level. Adults
with incomes <100% of FPL (53.9%; APR: 1.4) and adults
with incomes of 100%–249% of FPL (50.1%; APR: 1.2) were
more likely to report asthma attacks than adults with incomes
≥450% of FPL (48.9%) (Tables 1 and 2). Among persons
with incomes <100% of FPL, asthma attacks did not differ
significantly by race/ethnicity, sex, age, level of education, place
of birth, or geographic region (Table 1). However, subgroup
differences in reported asthma attacks were observed among
persons with higher income levels (100%–249% of FPL,
250%–449% of FPL, and ≥450% of FPL). In the three higher
income groups, asthma attacks were reported more frequently
among females, children, and persons living in the West than
among males, adults, and persons living in the Northeast,
respectively (Table 1). Notable changes in reporting an asthma
attack in the past year were observed between 2001–2004 and
2006–2010 (Table 3). For many demographic groups (whites,
blacks, Puerto Ricans, males, females, children, adults, male
children, persons with incomes <450% of FPL, and those living
in the Northeast and Midwest), reporting an asthma attack
decreased significantly. Between 2001–2004 and 2006–2010,
the disparity between the various Hispanic subgroups and that
between male children and female children were eliminated, and
the disparity between adults and children and that among the
FPL groups decreased. However, more women than men now
report having had an asthma attack in the past year, and persons
in the West and the South now report having had an attack
more often than persons in the Midwest and the Northeast.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 95

TABLE 1. Percentage of persons with current asthma* who reported an asthma attack in the past year,† by selected characteristics — National
Health Interview Survey, United States, 2006–2010

Characteristic

Total

FPL§

<100% FPL 100%–249% FPL 250%–449% FPL ≥450% FPL

Weighted
% SE

Sample
size¶

Weighted
% SE

Sample
size

Weighted
% SE

Sample
size

Weighted
% SE

Sample
size

Weighted
% SE

Sample
size

Race/Ethnicity
Non-Hispanic 51.1 (0.6) 11,586 54.3 (1.3) 2,533 52.0 (1.2) 3,476 48.8 (1.2) 2,843 50.3 (1.2) 2,734

White** 51.1 (0.7) 7,552 56.3 (1.8) 1,229 52.1 (1.5) 2,157 48.6 (1.4) 1,982 50.0 (1.3) 2,184
Black** 49.4 (1.1) 3,015 52.1 (2.1) 1,066 48.1 (1.9) 1,042 48.4 (2.7) 606 45.9 (3.9) 138
American Indian/

Alaska Native**
61.6 (7.8) 92 —††  (—) 36 — (—) 28 — (—) 17 — ( — ) 11

Asian** 53.7 (3.1) 474 50.1 (7.9) 72 46.8 (5.9) 94 55.1 (5.3) 138 57.3 (5.1) 170
Other§§ 56.3 (3.3) 436 50.2 (5.8) 126 68.7 (4.4) 147 42.4 (6.3) 97 60.6 (8.0) 66

Hispanic¶¶ 53.8 (1.5) 2,644 57.0 (2.4) 899 52.1 (2.3) 935 49.3 (3.0) 497 58.1 (3.8) 313
Puerto Rican 55.6 (3.4) 663 61.0 (4.4) 299 54.1 (4.8) 102 47.8 (8.1) 52 53.6 (9.9) 64
Mexican 52.6 (2.1) 1,338 54.0 (3.4) 396 52.3 (3.5) 521 49.6 (4.2) 266 55.8 (5.3) 155
Other*** 54.6 (2.6) 643 57.9 (4.2) 204 50.1 (4.3) 214 49.8 (5.7) 131 66.6 (6.8) 94

Sex
Male 48.8 (0.9) 5,699 52.7 (2.0) 1,242 50.0 (1.5) 1,754 46.1 (1.8) 1,388 47.0 (1.8) 1,316
Female 53.5 (0.7) 8,531 56.3 (1.4) 2,190 53.4 (1.3) 2,657 50.9 (1.4) 1,952 53.9 (1.5) 1,731

Age
Child (aged 0–17 yrs) 56.1 (0.9) 4,739 56.3 (1.9) 1,208 56.1 (1.7) 1,544 54.3 (2.1) 1,111 57.8 (2.2) 876
Adult (aged ≥18 yrs) 49.6 (0.7) 9,491 53.9 (1.4) 2,224 50.1 (1.2) 2,867 46.9 (1.4) 2,228 48.9 (1.4) 2,171
Child        

Male 55.7 (1.4) 2,740 54.4 (2.7) 717 57.1 (2.3) 884 54.3 (2.7) 631 56.7 (3.0) 508
Female 56.7 (1.5) 1,999 59.4 (2.7) 491 54.8 (3.0) 660 54.3 (3.1) 480 59.5 (3.5) 368

Adult
Male 44.1 (1.2) 2,959 50.5 (2.8) 525 44.8 (2.1) 870 41.6 (2.5) 757 42.8 (2.2) 807
Female 52.6 (0.8) 6,532 55.2 (1.5) 1,699 53.0 (1.4) 1,997 50.0 (1.7) 1,472 52.9 (1.1) 1,364

Educational attainment
(aged ≥18 yrs)
Less than high school 51.2 (1.5) 1,889 55.5 (2.2) 889 49.2 (2.3) 712 47.4 (4.6) 215 46.4 (7.5) 73
High school graduate or

equivalent
44.6 (1.3) 2,349 50.4 (2.7) 584 44.5 (2.3) 864 40.6 (2.7) 570 44.8 (3.3) 331

Some college 50.3 (1.3) 2,090 53.4 (3.0) 464 53.1 (2.4) 680 46.7 (2.4) 552 48.8 (3.3) 394
College graduate or higher 52.1 (1.2) 3,099 59.1 (3.6) 273 56.7 (2.6) 586 51.0 (2.2) 876 50.2 (1.8) 1,364

Place of birth
U.S. and U.S. territories 51.2 (0.6) 13,277 54.6 (1.2) 3,177 52.0 (1.1) 4,110 48.5 (1.2) 3,148 50.3 (1.2) 2,842
Outside U.S. and

U.S. territories†††
55.5 (2.1) 950 58.4 (4.4) 255 51.8 (3.6) 300 54.8 (5.4) 192 58.2 (4.8) 203

Geographic region§§§

Northeast 47.8 (1.3) 2,731 52.9 (3.0) 699 47.1 (2.2) 781 45.4 (2.9) 608 47.1 (2.4) 642
Midwest 49.4 (1.1) 3,263 51.9 (2.2) 810 50.2 (2.4) 986 45.4 (2.4) 774 50.5 (2.7) 693
South 53.1 (1.0) 4,957 56.1 (1.9) 1,255 53.8 (1.7) 1,669 51.5 (2.0) 1,113 50.7 (2.1) 920
West 54.5 (1.3) 3,279 58.3 (2.9) 669 55.1 (2.2) 974 51.3 (2.5) 844 54.9 (2.2) 792

Total 51.5 (0.6) 14,230 54.8 (1.2) 3,432 52.0 (1.0) 4,411 48.8 (1.2) 3,340 50.8 (1.2) 3,047

Abbreviations: FPL = federal poverty level; SE = standard error.
* Persons who answered “yes” to the questions, “Have you ever been told by a doctor or other health professional that you had asthma?” or “Has a doctor or other health professional ever

told you that (sample child) had asthma?” and “yes” to the question, “Do you (does sample child) still have asthma?”
† Persons who answered “yes” to the question, “During the past 12 months, have you (has sample child) had an episode of asthma or an asthma attack?”
§ FPL was based on U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html). Imputed income values were used when income was not

reported.
¶ Unweighted pooled sample size, 2006–2010. Because of item nonresponse, individual characteristic categories might not sum to total.
** Includes persons who indicated only a single race.
†† If the relative SE is >30%, or if the sample size (denominator) is <50, estimates are considered unreliable and are suppressed.
§§ Includes Native Hawaiians and Other Pacific Islanders, persons reporting more than one race, and persons reporting their race as something other than those listed.
¶¶ Persons of Hispanic ethnicity can be of any race or combination of races.
*** Includes persons reporting Cuban, Dominican, Central or South American, Spanish, multiple, and unspecified Hispanic ancestry.
††† Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens) as well as naturalized citizens and noncitizens.
§§§ Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan,

Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana,
Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada,
New Mexico, Oregon, Utah, Washington, and Wyoming. 

Supplement

96 MMWR / November 22, 2013 / Vol. 62 / No. 3

TABLE 2. Multivariate association between reporting an asthma attack in the past year* and selected characteristics of persons with current
asthma† — National Health Interview Survey, United States, 2006–2010

Characteristic

Child (aged 0–17 yrs) Adult (aged ≥18 yrs)

Crude Adjusted Crude Adjusted

PR (95% CI) PR (95% CI) PR (95% CI) PR (95% CI)

Race/Ethnicity
Non-Hispanic, white§ 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.
Non-Hispanic, black§ 0.8 (0.7–1.0) 0.8 (0.7–1.0) 0.9 (0.8–1.0) 0.8 (0.7–0.9)
Non-Hispanic, other¶ 1.0 (0.8–1.4) 0.9 (0.7–1.3) 1.2 (1.0–1.5) 1.1 (0.9–1.4)
Hispanic** 0.9 (0.5–1.1) 0.8 (0.7–1.0) 1.1 (1.0–1.3) 1.0 (0.8–1.2)

Puerto Rican 1.3 (0.9–1.8) 1.6 (1.2–2.3) 1.0 (0.7–1.5) 1.2 (0.8–1.7)
Mexican/Other†† 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.

Sex
Male 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.
Female 1.0 (0.9–1.3) 1.1 (0.9–1.3) 1.4 (1.3–1.6) 1.4 (1.2–1.5)

Age group (yrs)
0–4 1.9 (1.5–2.4) 1.9 (1.5–2.5) NA NA NA NA
5–11 1.3 (1.1–1.5) 1.3 (1.1–1.6) NA NA NA NA
12–17 1.0 Ref. 1.0 Ref. NA NA NA NA
18–34 NA NA NA NA 1.3 (1.1–1.5) 1.4 (1.2–1.6)
35–44 NA NA NA NA 1.9 (1.6–2.3) 2.0 (1.7–2.4)
45–54 NA NA NA NA 1.8 (1.5–2.2) 1.9 (1.6–2.3)
55–64 NA NA NA NA 1.5 (1.3–1.8) 1.6 (1.4–1.9)
≥65 NA NA NA NA 1.0 Ref. 1.0 Ref.
Educational attainment (aged ≥18 years)

Less than high school education NA NA NA NA 1.0 (0.8–1.1) 0.9 (0.8–1.1)
High school graduate or equivalent NA NA NA NA 0.7 (0.7–0.9) 0.7 (0.6–0.9)
Some college NA NA NA NA 0.9 (0.8–1.1) 0.9 (0.8–1.1)
College graduate or higher NA NA NA NA 1.0 Ref. 1.0 Ref.

Federal poverty level (FPL)§§
<100% FPL 0.9 (0.8–1.2) 1.0 (0.8–1.2) 1.2 (1.1–1.4) 1.4 (1.2–1.6)
100%–249% FPL 0.9 (0.4–1.2) 0.9 (0.8–1.2) 1.1 (0.9–1.2) 1.2 (1.1–1.4)
250%–449% FPL 0.9 (0.7–1.1) 0.9 (0.7–1.1) 0.9 (0.8–1.1) 1.0 (0.8–1.2)
≥450% FPL 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.

Place of birth
U.S. and U.S. territories 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.
Outside U.S. and U.S. territories¶¶ 1.1 (0.7–1.8) 1.3 (0.8–2.0) 1.3 (1.1–1.5) 1.2 (1.0–1.4)

Geographic region***
Northeast 1.0 Ref. 1.0 Ref. 1.0 Ref. 1.0 Ref.
Midwest 1.2 (1.0–1.6) 1.2 (1.0–1.6) 1.0 (0.9–1.2) 1.0 (0.9–1.2)
South 1.3 (1.1–1.7) 1.3 (1.1–1.7) 1.2 (1.0–1.4) 1.2 (1.1–1.4)
West 1.5 (1.1–1.9) 1.5 (1.2–2.0) 1.3 (1.1–1.5) 1.3 (1.1–1.5)

Abbreviations: 95% CI = 95% confidence interval; NA = not applicable; PR = prevalence ratio; Ref. = Referent.
* Persons who answered “yes” to the question, “During the past 12 months, have you (has sample child) had an episode of asthma or an asthma attack?
† Persons who answered “yes” to the questions, “Have you ever been told by a doctor or other health professional that you had asthma?” or “Has a doctor or other

health professional ever told you that (sample child) had asthma?” and “yes” to the question, “Do you (does sample child) still have asthma?”
§ Includes persons who indicated only a single race group.
¶ Includes Asians, American Indians/Alaska Natives, Native Hawaiians and Other Pacific Islanders, persons reporting more than one race, and persons reporting

their race as something other than those listed.
** Persons of Hispanic ethnicity can be of any race or combination of races.
†† Includes persons reporting Mexican, Cuban, Dominican, Central or South American, Spanish, multiple, and unspecified Hispanic ancestry.
§§ FPL was based on U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html). Imputed income values were used when

income was not reported.
¶¶ Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens.
*** Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming. 

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 97

Discussion
Findings from this report are comparable to those of previous

reports (1,3,4,6). Within the U.S. population, the percentage
of persons with current asthma reporting an asthma attack in
the past year varies by demographic and economic groups.
Similar to current asthma prevalence, asthma attacks were
more prevalent among females, children, the poor, persons
of multiple races, and Puerto Ricans. However, while current
asthma prevalence was higher in the Northeast, attacks were
reported more frequently in the South and in the West than
in other regions.

Demographic and socioeconomic characteristics associated
with more frequent reporting of asthma attacks (e.g.,
females, children, persons living in the South and the West,
Puerto Ricans, and the poor) were identified. However,
causality cannot be determined from cross-sectional survey
data. Surveillance data on asthma attacks cannot be used to
determine the reasons for the observed differences among
the demographic and economic subgroups examined in this
report. The differences might be attributable to differing levels
of exposure to environmental irritants and allergens (e.g.,
environmental irritants such as tobacco smoke or air pollutants
and environmental allergens such as house dust mites,
cockroach particles, and cat and dog dander) (7,9,13) or they
might be attributable to differences in disease self-management
or medical treatment. The reasons for the differences in the
frequency of reported asthma attacks can be addressed only by
research studies designed to determine the effect of a specific
exposure or a specific disease management protocol.

Limitations
The results of this analysis are subject to at least three

limitations. First, the asthma attack prevalence estimates in
this report rely on self-reported data and thus are subject to
recall bias. The respondent must recall a physician’s diagnosis
of asthma correctly, which in turn requires that the physician’s
diagnosis was correct and conveyed successfully to the person.
Because no definitive test exists for asthma, the diagnosis
and self-report cannot be validated; however, a 1993 review
of asthma questionnaires documented a mean sensitivity
of 68% and a mean specificity of 94% when self-reported
data on an asthma diagnosis were compared with a clinical
diagnosis (1). Second, because NHIS includes only the civilian,
noninstitutionalized population of the United States, results
might not be representative of other populations. Finally,
because NHIS is conducted only in English and Spanish,
results might not be representative of households whose
residents have other primary languages.

TABLE 3. Percentage of persons with current asthma* who reported
an asthma attack in the past year,† by selected characteristics —
National Health Interview Survey, United States, 2001–2004 and
2006–2010

Characteristic

2001–2004 2006–2010

% (SE) % (SE)

Race/Ethnicity
Non-Hispanic 55.2 (0.6) 51.1 (0.6)

White§ 54.6 (0.7) 51.1 (0.7)
Black§ 56.6 (1.2) 49.4 (1.1)
Other¶ 59.2 (2.3) 55.9 (2.2)

Hispanic** 55.7 (1.3) 53.8 (1.5)
Puerto Rican 64.2 (2.3) 55.6 (3.4)
Mexican 50.3 (1.9) 52.6 (2.1)

Sex
Male 54.3 (0.8) 48.8 (0.9)
Female 56.0 (0.7) 53.5 (0.7)

Child (aged 0–17 yrs) 62.6 (1.0) 56.1 (0.9)
Male 63.9 (1.2) 55.7 (1.4)
Female 60.8 (1.6) 56.7 (1.5)

Adult (aged ≥18 yrs) 52.0 (0.6) 49.6 (0.7)
Male 47.2 (1.2) 44.1 (1.2)
Female 54.6 (0.8) 52.6 (0.8)

Federal poverty level (FPL)††
<100% FPL 58.9 (1.3) 54.8 (1.2)
100%–249% FPL 56.4 (1.0) 52.0 (1.0)
250%–449% FPL 57.3 (1.2) 48.8 (1.2)
≥450% FPL 53.2 (1.1) 50.8 (1.2)

Geographic region§§
Northeast 54.3 (1.1) 47.8 (1.3)
Midwest 55.7 (1.1) 49.4 (1.1)
South 55.5 (0.9) 53.1 (1.0)
West 55.3 (1.2) 54.5 (1.3)

Total 55.3 (0.5) 51.5 (0.6)

Abbreviation: SE = standard error.
* Persons who answered “yes” to the questions, “Have you ever been told by a

doctor or other health professional that you had asthma?” or “Has a doctor
or other health professional ever told you that (sample child) had asthma?”
and “yes” to the question, “Do you (does sample child) still have asthma?”

† Persons who answered “yes” to the question, “During the past 12 months,
have you (has sample child) had an episode of asthma or an asthma attack?

§ Includes persons who indicated only a single race.
¶ Includes Asians, American Indians/Alaska Natives, Native Hawaiians and

Other Pacific Islanders, persons reporting more than one race, and persons
reporting their race as something other than those listed.

** Persons of Hispanic ethnicity can be of any race or combination of races.
†† FPL was based on U.S. Census Bureau poverty thresholds (available at http://

www.census.gov/hhes/www/poverty.html). Imputed income values were
used when income was not reported.

§§ Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire,
New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana,
Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio,
South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District
of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi,
North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and
West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho,
Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming. 

Supplement

98 MMWR / November 22, 2013 / Vol. 62 / No. 3

Conclusion
With the exception of a few specific occupational exposures,

the exact cause of asthma is unknown. Once diagnosed, asthma
requires ongoing comprehensive management on a long-term
basis. Appropriate management requires both access to the
health system and appropriate use of that system. Financial
resources and social support are instrumental in long-term
management of all chronic conditions, not just asthma
(13–15). Acquiring self-management knowledge and skills
and limiting exposure to environmental allergens and irritants
are necessary to improve health and quality of life for persons
with asthma, and especially for those with uncontrolled asthma
(2,13–15).

Although surveillance data can identify disproportionately
affected groups, research is needed into the role of self-
management factors, environmental exposures, health-care
system factors, and financial factors to understand better how
their interrelations affect individual asthma management
and control. Identifying the population-specific factors that
contribute to asthma exacerbations among disproportionately
affected demographic and socioeconomic groups can lead to
targeted interventions. Strategies for asthma self-management
and medical treatment protocols for asthma that are culturally
appropriate and take into consideration population-specific
characteristics can reduce the occurrence and severity of
asthma exacerbations (14). For example, an intervention for
children with asthma that included the use of multitrigger,
multicomponent environmental factors resulted in improved
symptom control and reduced the number of school days
missed (16). Similar effective interventions are needed to
address other disproportionately affected demographic and
economic groups identified in this report.

References
1. CDC. National surveillance for asthma—United States, 1980–2004.

MMWR 2007;56(No. SS-8).
2. National Institutes of Health. National Heart, Lung, and Blood Institute.

Expert panel report 3: guidelines for the diagnosis and management of
asthma. Bethesda, MD: US Department of Health and Human Services,
National Institutes of Health, National Heart, Lung, and Blood Institute,
National Asthma Education Program; 2007.

3. Moorman JE, Akinbami LJ, Bailey CM, et al. National surveillance of
asthma—United States, 2001–2010. Vital Health Stat 2012;3(35).

4. CDC. Current asthma prevalence—United States, 2006–2008. In: CDC
health disparities and inequalities report—United States, 2011. MMWR
2011;60(Suppl; January 14, 2011).

5. Evans R 3rd, Mullally DI, Wilson RW, et al. National trends in the
morbidity and mortality of asthma in the US. Prevalence, hospitalization
and death from asthma over two decades: 1965–1984. Chest
1987;91(Suppl 6):S65–74.

6. Mannino DM, Homa DM, Akinbami LJ, Moorman JE, Gwynn C,
Redd SC. Surveillance for asthma—United States, 1980–1999. MMWR
2002;51(No. SS-1):1–13.

7. Homa DM, Mannino DM, Lara M. Asthma mortality in U.S. Hispanics
of Mexican, Puerto Rican, and Cuban heritage, 1990–1995. Am J Respir
Crit Care Med 2000;161:504–9.

8. Rose D, Mannino DM, Leaderer BP. Asthma prevalence among US
adults, 1998–2000: role of Puerto Rican ethnicity and behavioral and
geographic factors. Am J Public Health 2006:96:880–8.

9. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

10. CDC. Introduction: In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

11. CDC. National Health Interview Survey (NHIS), 2007 data release.
Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics; 2010. Available at http://www.
cdc.gov/NCHS/nhis/nhis_2007_data_release.htm.

12. US Census Bureau. How the Census Bureau measures poverty. Available at
http://www.census.gov/hhes/www/poverty/about/overview/measure.html.

13. Institute of Medicine. Clearing the air: asthma and indoor air exposures.
Washington, DC: National Academies Press; 2000.

14. US Department of Health and Human Services. Healthy people 2010:
understanding and improving health. 2nd ed. Washington, DC: US
Government Printing Office; 2000.

15. Wade S, Weil C, Holden G, et al. Psychosocial characteristics of inner-
city children with asthma: a description of the NCICAS psychosocial
protocol. National Cooperative Inner-City Asthma Study. Pediatr
Pulmonol 1997;24:263–76.

16. US Task Force on Community Prevention Services. Asthma control:
home-based multi-trigger, multicomponent environmental interventions.
Av a i l a b l e a t h t t p : / / w w w. t h e c o m m u n i t y g u i d e . o r g / a s t h m a /
multicomponent.html.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 99

Introduction
In 2011, an estimated 26 million persons aged ≥20 years

(11.3% of the U.S. population) had diabetes (1). Both the
prevalence and incidence of diabetes have increased rapidly
since the mid-1990s, with minority racial/ethnic groups and
socioeconomically disadvantaged groups experiencing the
steepest increases and most substantial effects from the disease
(2–5).

This analysis and discussion of diabetes is part of the second
CDC Health Disparities and Inequalities Report (CHDIR).
The 2011 CHDIR (6) was the first CDC report to assess
disparities across a wide range of diseases, behavioral risk
factors, environmental exposures, social determinants, and
health-care access. The 2011 CHDIR report discussed the
magnitude and patterning of absolute and relative measures
of disparity in the prevalence and incidence rate of medically
diagnosed diabetes during 2004 and 2008 and identified
marked disparities in terms of race/ethnicity, socioeconomic
status, disability status, and geography (7). The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (8). This report updates information
on disparities in prevalence and incidence rates of diagnosed
diabetes presented in the 2011 CHDIR. The purposes of this
report are to discuss and raise awareness about group differences
in the level of diagnosed diabetes and to prompt actions to
reduce these disparities.

Methods
To monitor progress toward eliminating health disparities

in the prevalence and incidence rate of medically diagnosed
diabetes, CDC used data from the 2006 and 2010 National
Health Interview Survey (NHIS). NHIS is an ongoing, cross-
sectional, in-person household interview survey of a probability
sample of the civilian, noninstitutionalized U.S. population.
Household interviews were completed for 75,716 persons
in 2006 and 89,976 persons in 2010, with response rates of
87.3% and 79.5%, respectively (9,10).

The methods used to assess prevalence and incidence rates of
medically diagnosed diabetes have been described previously
(7). Analyses were repeated to assess disparities in each year and

changes in disparity over time (11), according to the selected
characteristics of age, sex, race/ethnicity, socioeconomic status,*
geographic region as defined by the U.S. Census Bureau,† and
disability status. Because of the association between place of
birth and diabetes, the data also were examined by place of
birth, defined as U.S.-born or not U.S.-born§ (12,13).

Prevalence (cases of diabetes of any duration per 100
population) was calculated for adults aged ≥18 years. Incidence
rate (cases of diabetes ≤1 year’s duration per 1,000 population)
was calculated for adults aged 18–79 years. Estimates were
standardized by the direct method to the age distribution of the
U.S. 2000 Census adult population (14). Age-specific estimates
were not age-standardized. CDC used software to account for
the complex sample design of NHIS and to produce point
estimates, standard errors, and 95% confidence intervals (CIs).

Disparities were measured as the deviations from a referent
category incidence rate or prevalence. Absolute difference
was measured as the simple difference between a population
subgroup estimate and the estimate for its respective reference
group. The relative difference, a percentage, was calculated by
dividing the difference by the value in the referent category and
multiplying by 100. To assess change in disparities over time,
CDC calculated change in relative difference by subtracting the
relative difference in the ending time period from the relative
difference for the beginning period. To test for the statistical
significance of the observed absolute and relative differences,
CDC used the z statistic and a 2-tailed test at p<0.05 with

* Measured as educational attainment and household income by income-to-
poverty ratio [IPR]. Following the Office of Management and Budget’s Statistical
Policy Directive 14, the U.S. Census Bureau uses a set of money income
thresholds that vary by family size and composition to determine who is in
poverty. IPR is the total family income expressed as a ratio or percentage of the
family’s official poverty threshold. An IPR <1.00 or <100% of poverty denotes
a family in poverty; an IPR ≥1.00 or ≥100% of the poverty threshold denotes
family income equal to or higher than poverty. Official poverty thresholds are
corrected for inflation using the Consumer Price Index.

† Northeast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey,
New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana,
Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio,
South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District
of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi,
North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and
West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho,
Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

§ Includes U.S. citizens born abroad (one or both of whose parents were U.S.
citizens), naturalized citizens, and noncitizens.

Diabetes — United States, 2006 and 2010
Gloria L. Beckles, MD

Chiu-Fang Chou, DrPH
Division of Diabetes Translation, National Center for Chronic Disease Prevention and Health Promotion

Corresponding author: Gloria L. Beckles, Division of Diabetes Translation, National Center for Chronic Disease Prevention and Health Promotion, CDC.
Telephone: 770-488-1272; E-mail: [email protected].

Supplement

100 MMWR / November 22, 2013 / Vol. 62 / No. 3

Bonferroni correction for multiple comparisons; 95% CIs
were calculated. Statistically significant increases and decreases
in relative differences from 2009 to 2011 were interpreted as
increases and decreases in disparity, respectively. Estimates with
relative standard error ≥30% were not reported.

Results
Racial/ethnic and socioeconomic disparities were identified

in the age-standardized prevalence and incidence rate of
medically diagnosed diabetes in 2006 and 2010 (Tables 1
and 2). In both years, overall, and for both males and females,
significant absolute differences for race and ethnicity were
present between non-Hispanic whites and non-Hispanic blacks
or Hispanics (p<0.05 for all comparisons). The only significant
temporal decline in disparity was found for age-standardized
prevalence of diagnosed diabetes among non-Hispanic black
females (change: -23.3 percentage points; p<0.05). Temporal
increases in disparities from 2006 to 2010 were identified
for prevalence of diagnosed diabetes among Hispanics, with
increases greater among Hispanic females (change: 34.6
percentage points; p<0.05) than among Hispanic males
(change: 11.6 percentage points; p<0.05). Temporal increases
in disparities in incidence rates were greater among Hispanics
(change: 79.0 percentage points; p<0.05) than among non-
Hispanic blacks (change: 12.6 percentage points; p<0.05).

In 2006 and 2010, the groups with the lowest levels of
education and income continued to experience the greatest
socioeconomic disparity in age-standardized prevalence and
incidence rate of diagnosed diabetes (Tables 1 and 2). Among
these disadvantaged groups, no significant change in the
relative difference in prevalence occurred from 2006 to 2010,
but the disparity in the incidence rate worsened over time. In
addition, a significant decline in disparity in the prevalence
and incidence of diagnosed diabetes occurred among persons
with a high school education (p<0.05 for both comparisons).
From 2006 to 2010, age disparities in the age-standardized
prevalence and age-standardized incidence rate of diagnosed
diabetes worsened, and no significant change occurred in
the geographic and disability disparities in age-standardized
prevalence. However, for the age-standardized incidence rate,
disparities between the Northeast and each of the other U.S.
Census Bureau regions worsened significantly while disability
disparities improved (Table 2). No significant disparities
between U.S.-born and not U.S.-born persons were identified
in the total population or in any racial/ethnic population.

Discussion
From 2006 to 2010, a decline occurred in the disparity

between the prevalence of diagnosed diabetes among non-
Hispanic black women and that among white women; among
men, no evidence of a decline in racial/ethnic disparities in
diagnosed diabetes was identified. In addition, during the
survey years, socioeconomic disparities in the incidence of
diagnosed diabetes worsened among the groups with the lowest
level of education and income.

Although racial/ethnic and socioeconomic disparities in the
prevalence and incidence rate of diagnosed diabetes persist
in the U.S. adult population, some improvements occurred
from 2006 to 2010. Significant improvements were noted for
prevalence of diagnosed diabetes among non-Hispanic black
women compared with non-Hispanic white women, among
those with a high school diploma or some college compared
with those with a college degree or higher, and among the poor
(IPR <1.0 federal poverty level [FPL]) and middle income (IPR
2.0–2.9 FPL) groups compared with persons whose incomes
were high (IPR ≥4.0 FPL). A significant improvement also
occurred in the disparity in the diabetes incidence rate by
disability status.

Although improvements are noted for disparities in
prevalence of diagnosed diabetes, the annual incidence of
diagnosed diabetes among the U.S. population is increasing
(2,4), and mortality is declining among age, racial/ethnic,
socioeconomic, and disabled subgroups in the adult diabetic
population (15,16). If these circumstances continue, then
the prevalence of diabetes among the U.S. population is
projected to increase to as high as 33% by 2050 (15), posing
major challenges for U.S. public health. Diabetes is the
principal cause of kidney failure, nontraumatic lower extremity
amputation, and new cases of blindness, and it is a major cause
of cardiovascular disease among U.S. adults (1). The economic
costs of diagnosed diabetes reflect the substantial burden
imposed on the U.S. society (18). Between 2007 and 2012, the
total estimated annual cost increased by 41% (in 2007 dollars)
to $245 billion, including $69 billion in reduced productivity
(18). Medical expenditure among persons with diabetes is two
to three times that of persons without diabetes, and the largest
component (43%) of total medical expenditures attributed to
diagnosed diabetes is hospital inpatient care.

Limitations
The findings presented in this report are subject to at least

two limitations. First, all data are self-reported and therefore
subject to recall and social desirability bias. However, self-
reported diabetes data have been reported to have high

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 101

reliability (18,19). Second, differences were not assessed for
total prevalence of diabetes (i.e., diagnosed and undiagnosed);
therefore, the findings might underestimate the extent of
the disparities in prevalence and incidence among the U.S.
population. The percentage of persons with undiagnosed
diabetes is estimated to range from 24% to 40% of the total
prevalence of diabetes (1,20). However, the racial/ethnic,
socioeconomic, geographic, disability, and change over time
of the disparities in prevalence and incidence of medically
diagnosed diabetes provided in this report are consistent with
data provided in previously published reports on diabetes risk
among U.S. adults (2–5,12,20,21).

Conclusion
Obesity and lack of physical activity are major risk factors for

diabetes (22,23). The Community Preventive Task Force has
recommended several effective evidence-based interventions
that communities, policy makers, and public health authorities
can use to delay or prevent onset of diabetes by reducing
obesity and increasing physical activity. Strategies to increase
physical activity and physical fitness include communitywide
campaigns, school-based physical education, and creation
of or enhanced access to places for physical activity (22).
Interventions to prevent or control obesity include behavior
interventions to reduce screen time, multicomponent

TABLE 1. Age-standardized prevalence* of medically diagnosed diabetes among adults aged ≥18 years, by selected characteristics — National
Health Interview Survey, United States, 2006 and 2010

Characteristic

2006 2010 Change in
relative

difference from
2006 to 2010
(percentage

points)

Age-
standardized

prevalence
(%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
standardized

prevalence
(%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 7.0 (6.4–7.6) 0.3 4.5 8.6 (8.1–9.2) 1.5† 21.1 16.6
Female 6.7 (6.2–7.3) Ref. Ref. 7.1 (6.7–7.6) Ref. Ref.

Age group (yrs)§
18–44 2.7 (2.3–3.0) Ref. Ref. 2.7 (2.4–3.1) Ref. Ref.
45–64 11.4 (10.3–12.6) 8.8† 330.2 12.3 (11.5–13.1) 9.6† 350.5 20.3¶
65–74 18.9 (16.9–20.9) 16.3† 613.2 21.8 (19.9–23.8) 19.1† 698.5 85.3¶
≥75 18.2 (16.1–20.2) 15.6† 586.8 21.7 (19.8–23.5) 19.0† 694.9 108.1¶

Race/Ethnicity
Both sexes

White, non-Hispanic 6.0 (5.6–6.5) Ref. Ref. 6.8 (6.4–7.2) Ref. Ref.
Black, non-Hispanic 10.9 (9.8–12.1) 4.9† 81.7 11.3 (10.4–12.2) 4.5† 66.2 -15.5
Asian 7.4 (5.7–9.5) 1.4 23.3 7.9 (6.6–9.5) 1.1 16.2 -7.1
Mixed race/Other 10.6 (8.3–10.9) 4.6† 60.0 14.0 (9.2–20.8) 7.2† 105.9 45.9¶
Hispanic** 9.0 (7.9–10.2) 3.0† 50.0 11.5 (10.8–13.0) 4.7† 69.1 19.1¶

Male
White, non-Hispanic 6.3 (5.7–7.0) Ref. Ref. 7.8 (7.1–8.5) Ref. Ref.
Black, non-Hispanic 9.9 (8.4–11.7) 3.6† 57.1 12.4 (10.9–14.0) 4.6† 59.0 1.9
Asian 8.8 (6.3–12.2) 2.5 39.7 10.2 (8.1–12.7) 2.4 30.8 -8.9
Mixed race/Other —†† — NA NA 16.3 (11.5–22.7) 8.5† 109.0 NA
Hispanic 8.4 (6.9–10.2) 2.1† 33.3 11.3 (9.9–12.9) 3.5† 44.9 11.6¶

Female
White, non-Hispanic 5.8 (5.3–6.4) Ref. Ref. 6.0 (5.4–6.5) Ref. Ref.
Black, non-Hispanic 11.6 (10.0–13.4) 5.8† 100.0 10.6 (9.4–11.9) 4.6† 76.7 -23.3¶
Asian 6.3 (4.4–8.9) 0.5 8.6 6.1 (4.6–8.0) 0.1 1.7 -6.9
Mixed race/Other 8.3 (4.2–15.6) 2.5 43.1 13.5 (7.9–22.0) 7.5† 125.0 81.9
Hispanic 9.4 (8.0–10.9) 3.6† 62.1 11.8 (10.6–13.2) 5.8† 96.7 34.6¶

Educational attainment (aged ≥25 years)
Less than high school 9.1 (8.3–10.1) 4.5† 97.8 11.6 (10.6–12.8) 5.8† 100.0 2.2
High school or

equivalent
7.7 (6.8–8.7) 3.1† 67.4 8.5 (7.7–9.3) 2.7† 46.6 -20.8¶

Some college 8.0 (7.3–8.8) 3.4† 74.1 8.8 (8.1–9.6) 3.0† 51.7 -22.4¶
College degree or

higher
4.6 (3.9–5.3) Ref. Ref. 5.8 (5.1–6.5) Ref. Ref.

Income-to-poverty ratio§§
Poor 10.1 (8.9–11.4) 4.6† 83.1 10.6 (9.6–11.6) 4.6† 71.5 -11.6¶
Near poor 8.1 (7.2–9.0) 2.6† 46.3 9.6 (8.8–10.5) 3.4† 53.9 7.7¶
Middle income 6.8 (6.1–7.4) 1.2† 22.5 7.6 (7.0–8.2) 1.2† 18.6 -4.0¶
High income 5.5 (4.9–6.2) Ref. Ref. 6.4 (5.7–7.1) Ref. Ref.

See table footnotes on the next page.

Supplement

102 MMWR / November 22, 2013 / Vol. 62 / No. 3

counseling intended to reduce weight and maintain weight
loss, and worksite programs intended to reduce weight among
employees by improving diet and physical activity (23).
The CDC-led National Diabetes Prevention Program (24)
is designed to bring to communities strategies for adopting
evidence-based lifestyle changes known to prevent or delay the
onset of type 2 diabetes among adults at high risk for diabetes,
including modest weight loss, increased physical activity, and

reduced fat and calorie intake. Widespread implementation of
these and similar interventions to prevent obesity and promote
physical activity might reduce future incidence and prevalence
of diabetes and reduce disparities in diabetes risk.

TABLE 1. (Continued) Age-standardized prevalence* of medically diagnosed diabetes among adults aged ≥18 years, by selected characteristics
— National Health Interview Survey, United States, 2006 and 2010

Characteristic

2006 2010 Change in
relative

difference from
2006 to 2010
(percentage

points)

Age-
standardized

prevalence
(%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
standardized

prevalence
(%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Place of birth
All racial/ethnic groups

U.S.-born 6.9 (6.5–7.3) Ref. Ref. 7.7 (7.4–8.1) Ref. Ref.
Not U.S.-born¶¶ 7.2 (6.3–8.2) 0.3 4.3 8.6 (7.8–9.5) 0.9 11.7 7.4

White, non-Hispanic
U.S.-born 6.1 (5.6–6.6) Ref. Ref. 6.8 (6.4–7.2) Ref. Ref.
Not U.S.-born 4.7 (3.5–6.3) -1.4 -23.0 6.8 (5.2–8.9) 0.0 0.0 -23.0

Black, non-Hispanic
U.S.-born 11.3 (10.1–12.6) Ref. Ref. 11.6 (10.6–12.7) Ref. Ref.
Not U.S.-born 7.3 (4.6–11.4) -4.0 -35.4 8.9 (6.6–12.0) -2.7 -23.3 12.1

Asian/Pacific Islander
U.S.-born 5.9 (2.9–11.6) Ref. Ref. 8.7 (5.7–13.0) Ref. Ref.
Not U.S.-born 8.4 (6.6–10.7) 2.5 42.4 7.8 (7.5–8.1) -0.9 -10.3 -27.2

Hispanic
U.S.-born 9.7 (8.0–11.6) Ref. Ref. 13.1 (11.5–15.0) Ref. Ref.
Not U.S.-born 8.3 (6.9–10.0) -1.4 -14.4 10.3 (9.1–11.8) -2.8† -21.4 -7.0

Geographic region***
Northeast 6.2 (5.3–7.3) Ref. Ref. 6.3 (5.4–7.4) Ref. Ref.
Midwest 7.1 (6.3–8.1) 0.9 14.5 7.9 (7.3–8.6) 1.6† 25.4 10.9
South 7.1 (6.5–7.8) 0.9 14.5 8.8 (8.3–9.4) 2.5† 39.7 25.2
West 6.6 (5.9–7.4) 0.4 6.5 7.3 (6.7–8.0) 1.0† 15.9 9.4

Disability status
Has a disability 10.8 (9.9–11.8) 6.4† 145.5 12.1 (11.2–13.0) 7.2† 160.0 14.5
Does not have a

disability
4.4 (4.0–4.8) Ref. Ref. 4.9 (4.6–5.3) Ref. Ref.

Abbreviations: 95% CI = 95% confidence interval; NA = not available; Ref. = Referent.
* Cases of diabetes of any duration per 100 population. Estimate standardized by the direct method to the U.S. Census 2000 population.
† Simple difference between group estimate and Referent category significant at p<0.05 by z statistic and a 2-tailed test with Bonferroni correction for multiple

comparisons.
§ Age-specific estimates are not age-standardized.
¶ Difference between the relative differences in 2010 and 2006 significant at p<0.05 by z statistic and 2-tailed test with Bonferroni correction for multiple comparisons.
** Persons of Hispanic ethnicity might be of any race or combination of races.
†† Unstable estimate; relative standard error ≥30%.
§§ Following the Office of Management and Budget’s Statistical Policy Directive 14, the U.S. Census Bureau uses a set of money income thresholds that vary by family size

and composition to determine who is in poverty. The Income-to-Poverty Ratio (IPR) is the total family income expressed as a percentage of the family’s official poverty
threshold. An IPR <100% of poverty denotes a family in poverty; an IPR ≥100% of the poverty threshold denotes a family income equal to or higher than poverty. Official
poverty thresholds are corrected for inflation using the Consumer Price Index. Additional information is available at http://www.census.gov/hhes/www/poverty/methods/
definitions.html. Poor = <1.0 times the federal poverty level (FPL), near-poor = 1.0–1.9 times FPL, middle income = 2.0–3.9 times FPL, and high income = ≥4.0 times FPL.
FPL was calculated on the basis of U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html).

¶¶ Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens.
*** Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, New Mexico, Nevada, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 103

See table footnotes on the next page.

TABLE 2. Age-standardized incidence rate* of medically diagnosed diabetes among adults aged 18–79 years, by selected characteristics —
National Health Interview Survey, United States, 2006 and 2010.

Characteristic

2006 2010 Change in
relative

difference
from 2006 to

2010
(percentage

points)

Age-
standardized
incidence rate (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
standardized

incidence rate (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 7.6 (5.8–9.9) 0.1 0.8 8.9 (7.0–11.3) 2.8† 47.0 46.2§
Female 7.5 (6.1–9.3) Ref. Ref. 6.0 (4.8–8.0) Ref. Ref.

Age group (yrs)¶
18–44 5.1 (3.9–6.7) Ref. Ref. 3.8 (2.8–5.2) Ref. Ref.
45–64 10.8 (8.7–13.5) 5.7† 112.0 11.5 (9.4–14.0) 7.7† 200.0 88.0§
65–74 10.7 (7.0–16.3) 5.6† 109.4 14.7 (10.1–21.3) 10.9† 283.9 174.5§
≥75 14.4 (8.1–25.4) 9.3† 182.3 16.4 (9.4–28.4) 12.6† 328.3 146.0§

Race/Ethnicity
Both sexes

White, non-Hispanic 6.8 (5.5–8.5) Ref. Ref. 6.0 (4.7–7.5) Ref. Ref.
Black, non-Hispanic 9.6 (6.7–13.8) 2.8† 41.1 9.2 (6.6–12.6) 3.2† 53.7 12.6§
Hispanic** 8.6 (5.7–13.0) 1.8† 25.6 12.2 (8.8–17.0) 6.2† 104.6 79.0§

Male
White, non-Hispanic 6.7 (4.7–9.6) Ref. Ref. 7.2 (5.1–10.0) Ref. Ref.
Black, non-Hispanic 10.0 (6.1–16.5) 3.3† 49.6 11.8 (7.3–19.1) 4.6† 64.7 15.1§
Hispanic 6.6 (3.6–11.9) -0.1 -1.8 15.4 (9.4–25.1) 8.2† 114.6 116.4§

Female
White, non-Hispanic 7.0 (5.4–9.2) Ref. Ref. 4.9 (3.5–6.7) Ref. Ref.
Black, non-Hispanic 9.3 (5.6–15.5) 2.3† 32.0 7.1 (4.9–10.3) 2.2† 45.7 13.7§
Hispanic 10.5 (6.0–18.4) 3.5† 49.8 9.5 (6.1–14.7) 4.6† 94.0 44.2§

Educational attainment (aged ≥25 yrs)
Less than high school 10.2 (7.1–14.8) 5.9† 136.4 13.7 (9.5–19.6) 9.4 219.9 83.6§
High school or

equivalent
10.7 (7.6–15.1) 6.4† 147.1 8.3 (5.8–11.8) 4.0 93.9 -53.2§

Some college 9.9 (7.0–13.8) 5.5† 127.9 10.0 (7.5–13.3) 5.7 133.3 5.4§
College degree or

higher
4.3 (2.8–6.8) Ref. Ref. 4.3 (2.7–6.8) Ref. Ref.

Income-to-poverty ratio††
Poor 10.9 (6.9–17.1) 4.9† 82.5 11.5 (7.5–16.3) 7.1† 113.7 31.2§
Near poor 8.4 (6.1–11.1) 2.5† 41.4 8.2 (5.9–11.4) 2.2 35.5 -5.9
Middle income 8.4 (6.1–11.3) 2.4† 40.2 8.0 (6.0–10.7) 1.5 24.7 -15.6
High income 6.0 (3.9–9.2) Ref. Ref. 6.2 4.2–9.2) Ref. Ref.

Place of birth
All races/ethnicities

U.S.-born 7.6 (6.3–9.2) Ref. Ref. 7.3 (6.1–8.8) Ref. Ref.
Not U.S.-born§§ 6.6 (4.5–9.8) -1.0 11.7 7.6 (5.1–11.5) 0.3 11.7 0.0

White, non-Hispanic
U.S.-born 6.9 (5.5–8.6) Ref. Ref. 6.1 (4.8–7.8) Ref. Ref.
Not U.S.-born —¶¶ — NA NA — — NA NA NA

Black, non-Hispanic
U.S.-born 9.9 (6.8–14.4) Ref. Ref. 10.3 (7.4–14.2) Ref. Ref.
Not U.S-born — — NA NA —†† —†† NA NA NA

Hispanic
U.S.-born — — Ref. Ref. 14.8 (8.6–25.4) Ref. Ref.
Not U.S.-born 6.8 (4.1–11.2) NA NA 10.8 (7.2–16.3) -4.0† -26.9 8.0

Geographic region***
Northeast 7.4 (5.2–10.6) Ref. Ref. 6.3 (3.9–10.1) Ref. Ref.
Midwest 6.7 (4.8–9.6) -0.7 -9.0 7.2 (4.9–10.4) 0.8† 13.4 22.5§
South 8.4 (6.3–11.2) 1.0 13.8 8.3 (6.5–10.4) 2.0† 31.1 17.3§
West 6.6 (4.5–9.9) -0.8 -10.5 7.2 (4.9–10.5) 0.9 13.7 24.2§

Disability status
Has a disability 14.1 (10.8–18.4) 9.2† 187.2 12.0 (9.4–15.2) 6.7† 125.5 -61.6§
Does not have a

disability
4.9 (3.8–6.3) Ref. Ref. 5.3 (3.2–7.3) Ref. Ref.

Supplement

104 MMWR / November 22, 2013 / Vol. 62 / No. 3

References
1. CDC. National Diabetes Surveillance System: national diabetes fact

sheet, 2011. Atlanta, GA: US Department of Health and Human
Services, CDC; 2011. Available at http://www.cdc.gov/diabetes/pubs/
pdf/ndfs_2011.pdf.

2. CDC. National Diabetes Surveillance System. Atlanta, GA: US
Department of Health and Human Services, CDC; 2010. Available at
http://www.cdc.gov/diabetes/statistics/index.htm.

3. Kanjilal S, Gregg EW, Cheng YJ, et al. Socioeconomic status and trends
in disparities in 4 major risk factors for cardiovascular disease among
US adults, 1971–2002. Arch Intern Med 2006;166:2348–55.

4. Narayan KM, Boyle JP, Geiss LS, Saaddine JB, Thompson TJ. Impact
of recent increase in incidence on future diabetes burden: U.S., 2005–
2050. Diabetes Care 2006;29:2114–6.

5. Geiss LS, Pan L, Cadwell B, Gregg EW, Benjamin SM, Engelgau MM.
Changes in incidence of diabetes in U.S. adults, 1997–2003. Am J Prev
Med 2006;30:371–7.

6. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

7. CDC. Diabetes—United States, 2004 and 2008. In: CDC health
disparities and inequalities report—United States, 2011. MMWR
2011;60(Suppl; January 14, 2011).

8. CDC. Introduction: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

9. Adams PF, Lucas JW, Barnes PM. Summary health statistics for the U.S.
population: National Health Interview Survey, 2006. Vital Health Stat
2008(10).

10. Adams PF, Martinez ME, Vickerie JL, Kirzinger WK. Summary health
statistics for the U.S. population: National Health Interview Survey,
2010. Vital Health Stat 2011(10).

11. Keppel K, Pamuk E, Lynch J, et al. Methodological issues in measuring
health disparities. Vital Health Stat 2 2005;141:1–16.

12. Cunningham SA, Ruben JD, Venkat Narayan KM. Health of foreign-
born people in the United States: a review. Health & Place 2008;
14:623–35.

13. Huh J, Prause JA, Dooley CD. The impact of nativity on chronic diseases,
self-rated health, and comorbidity status of Asian and Hispanic
immigrants. J Immig Minority Health 2008;10:103–18.

14. Anderson RN, Rosenberg HM. Age standardization of death rates:
implementation of the year 2000 standard. Natl Vital Stat Rep 1998;
47:1–16, 20.

15. Boyle JP, Thompson TJ, Gregg EW, Barker LE, Williamson DF. Projection
of the year 2050 burden of diabetes in the US adult population: dynamic
modeling of incidence, mortality, and prediabetes prevalence. Popul Health
Metrics 2010;8:29. DOI:10.1186/1478-7954-8-29.

16. Gregg EW, Cheng YJ, Saydah S, et al. Trends in death rates among U.S.
adults with and without diabetes between 1997 and 2006. Diabetes
Care 2012;35:1252–7.

17. American Diabetes Association. Economic costs of diabetes in the U.S. in
2012. Diabetes Care 2013;36. March 6, 2013. DOI: 10.2337/dc12-2625.

18. Tisando DM, Adams JL, Liu H, et al. What is the concordance between
the medical record and patient self-report as data sources for ambulatory
care? Med Care 2006;44:132–40.

19. Newell SA, Girgis A, Sanson-Fisher RW, Savolainen NJ. The accuracy
of self-reported health behaviors and risk factors relating to cancer and
cardiovascular disease in the general population: a critical review. Am J
Prev Med 1999;17:211–29.

20. Cowie CC, Rust KF, Ford ES, et al. Full accounting of diabetes and
pre-diabetes in the U.S. population in 1988–1994 and 2005–2006.
Diabetes Care 2009;32:287–94.

21. Volpato S, Maraldi C, Fellin R. Type 2 diabetes and risk for functional decline
and disability in older persons. Curr Diabetes Rev 2010; 6:134–43.

22. Community Preventive Services Task Force. Increasing physical activity.
Atlanta, GA: Community Preventive Services Task Force; 2012. Available
at http://www.thecommunityguide.org/pa/index.html.

23. Community Preventive Services Task Force. Obesity prevention and
control: interventions in community settings. Atlanta, GA: Community
Preventive Services Task Force; 2012. Available at http://www.
thecommunityguide.org/obesity/communitysettings.html.

24. CDC. National Diabetes Prevention Program. Atlanta, GA: US
Department of Health and Human Services, CDC; 2012. Available at
http://www.cdc.gov/diabetes/prevention/about.htm.

TABLE 2. (Continued) Age-standardized incidence rate* of medically diagnosed diabetes among adults aged 18–79 years, by selected
characteristics — United States, National Health Interview Survey, 2006 and 2010.

Abbreviations: 95% CI = 95% confidence interval; NA = not available; Ref. = Referent.
* Cases of diabetes of ≤1 year duration per 1,000 population. Estimates standardized by the direct method to the US Census Bureau 2000 population.
† Difference between group estimate and referent group estimate statistically significant at p<0.05 by z statistic and a 2-tailed test with Bonferroni correction for

multiple comparisons.
§ Difference between the group relative differences in 2010 and 2006 statistically significant at p<0.05 by z statistic and a 2-tailed test with Bonferroni correction

for multiple comparisons.
¶ Age-specific estimates are not age-standardized.
** Persons of Hispanic ethnicity might be of any race or combination of races.
†† Following the Office of Management and Budget’s Statistical Policy Directive 14, the U.S. Census Bureau uses a set of money income thresholds that vary by family size

and composition to determine who is in poverty. The Income-to-Poverty Ratio (IPR) is the total family income expressed as a percentage of the family’s official poverty
threshold. An IPR <100% of poverty denotes a family in poverty; an IPR ≥100% of the poverty threshold denotes a family income equal to or higher than poverty. Official
poverty thresholds are corrected for inflation using the Consumer Price Index. Additional information is available at http://www.census.gov/hhes/www/poverty/methods/
definitions.html. Poor = <1.0 times the federal poverty level (FPL), near-poor = 1.0–1.9 times FPL, middle income = 2.0–3.9 times FPL, and high income = ≥4.0 times FPL.
FPL was calculated on the basis of U.S. Census Bureau poverty thresholds (available at http://www.census.gov/hhes/www/poverty.html).

§§ Includes U.S. citizens born abroad (one or both of whose parents were U.S. citizens), naturalized citizens, and noncitizens.
¶¶ Unstable estimate; relative standard error ≥30%.
*** Northeast: Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, New York, Pennsylvania, Rhode Island, and Vermont; Midwest: Illinois, Indiana, Iowa,

Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, New Mexico, Nevada, Oregon, Utah, Washington, and Wyoming.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 105

Introduction
Health-related quality of life is physical and mental health,

as perceived by a person or group of people, during a period
of time (1,2). This measure complements traditional public
health measures of mortality and morbidity. Fair or poor self-
rated health, physically unhealthy days, and mentally unhealthy
days are reported by higher percentages of women, older
persons, minority racial/ethnic groups (except Asian/Pacific
Islanders), and persons with less education, with lower annual
household incomes, who are unemployed, with a disability
or a chronic disease, and who are widowed, separated, or
divorced than, respectively, men, younger persons, and non-
Hispanic whites, and those with more education, with higher
annual household incomes, who are employed by others or
self-employed, without a disability or a chronic disease, and
who are married (1).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (3) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (4). This report provides information
concerning disparities in health-related quality of life, a topic
that was not discussed in the 2011 CHDIR. The purposes
of this health-related quality of life report are to describe and
raise awareness of how different kinds of disparities affect
health-related quality of life among adults in the United States,
whether and how these effects changed from 2006 to 2010
and to prompt actions to reduce disparities.

Methods
To examine health-related quality of life disparities by

selected characteristics among adults (aged ≥18 years) in the
United States, CDC analyzed 2006 and 2010 data from the
Behavioral Risk Factor Surveillance System (BRFSS). BRFSS
is a continuous, random-digit–dialed telephone survey of
noninstitutionalized adults aged ≥18 years in the 50 states, the
District of Columbia (DC), Puerto Rico, the U.S. Virgin Islands,
Guam (5,6) (available at http://www.cdc.gov/brfss/index.htm).
This analysis compares health-related quality of life measures

stratified by specific characteristics in respondents from the 50
states and DC in 2006 (N = 347,790) and 2010 (N = 444,927).

Two indicators of BRFSS survey quality are its cooperation
rate and its overall response rate (7,8). The cooperation
rate is the proportion of all respondents interviewed of all
eligible units in which a respondent was selected and actually
contacted. In 2006, the cooperation rate ranged from 56.9%
in California to 83.5% in Minnesota; in 2010, the cooperation
rate ranged from 56.8% in California to 86.1% in Minnesota.
The overall response rate is an outcome rate with the number
of complete and partial interviews in the numerator and an
estimate of the number of eligible units in the sample in the
denominator that assumes that more unknown records are
eligible, specifically, that all likely households are households
and that 98% of known or probable households contain an
adult who uses the telephone number. In 2006, the overall
response rate ranged from 20.5% in Georgia to 58.4% in
Utah, and in 2010, from 19.2% in Oregon to 57.4% in Utah.

The three health-related quality of life measures represented
in BRFSS are 1) self-rated health status, 2) number of physically
unhealthy days, and 3) number of mentally unhealthy days.
The related BRFSS questions were as follows: 1) “Would you
say that in general your health is excellent, very good, good,
fair, or poor?” 2) “Now thinking about your physical health,
which includes physical illness and injury, for about how many
days during the past 30 days was your physical health not
good?” and 3) “Now thinking about your mental health, which
includes stress, depression, and problems with emotions, for
about how many days during the past 30 days was your mental
health not good?” CDC calculated the percentage reporting
fair or poor self-rated health, mean number of physically
unhealthy days, and mean number mentally unhealthy days
as the primary health-related quality of life outcome measures.
Respondents with the responses “do not know/not sure” or
“refused to respond” were excluded from the analysis on a
question-by-question basis.

Health-related quality of life disparities were assessed by
stratifying results by sex, age group (18–24, 25–34, 35–44,
45–64, 65–79, and ≥80 years), race/ethnicity (non-Hispanic
white, non-Hispanic black, Hispanic [might be of any race
or any combination of races], non-Hispanic Asian/Pacific
Islander [A/PI], non-Hispanic American Indian/Alaska Native
[AI/AN], and other), educational attainment at the time of

Health-Related Quality of Life — United States, 2006 and 2010
Matthew M. Zack, MD

Division of Population Health, National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Matthew M. Zack, Division of Population Health, National Center for Chronic Disease Prevention and Health Promotion, CDC.
Telephone: 770-488-5460; E-mail: [email protected].

Supplement

106 MMWR / November 22, 2013 / Vol. 62 / No. 3

the survey (less than high school, high school graduate or
equivalent, some college, and college graduate), the primary
language spoken at home (English, Spanish, or other), and
disability status, which was defined as an affirmative answer
to either or both of the following questions (9): “Are you
limited in any way in any activities because of physical, mental,
or emotional problems?” and “Do you now have any health
problem that requires you to use special equipment, such as
a cane, a wheelchair, a special bed, or a special telephone?”)
Each outcome measure was analyzed separately. Household
income was not examined because educational attainment
was considered a sufficient indicator of socioeconomic status
for examination of disparities and because approximately
14% of BRFSS respondents did not know or refused to report
household income, but <2% did not know or refused to report
their educational attainment.

CDC used statistical software for the analyses to account
for the stratified, complex sampling design of BRFSS (10).
Data were weighted using the respondents’ sampling weights
based on the population of noninstitutionalized adults aged
≥18 years in their states of residence and aggregated across the
50 states and DC. Because age is associated with the health-
related quality of life measures and because the age composition
differs among the various categories analyzed, CDC adjusted
the health-related quality of life measures by using age
group categories in the specific survey year as covariates
in logistic regression (for fair or poor self-rated health) and
linear regression (for number of physically and mentally
unhealthy days). No formal statistical testing was conducted
for this analysis. Differences were assessed by calculating and
comparing the 95% confidence intervals (CIs) around the
age-adjusted percentages and means. In this approach, CIs
were used as measure of variability, and nonoverlapping CIs
were considered statistically different. Using CIs in this way
is a conservative evaluation of significance differences; this
might lead to a conclusion that estimates are similar when the
point estimates differ at a significance level of 0.05. CIs were
assessed before rounding for the tables.

Disparities were measured as the deviations from a referent
group, which was the group that had the most favorable
estimate for the variables used to assess disparities during
the time reported. Absolute difference was measured as the
simple difference between a population subgroup estimate
and the estimate for its respective reference group. The
relative difference, a percentage, was calculated by dividing the
difference by the value in the referent category and multiplying
by 100. Change in percentage and mean from 2006 to 2010
was calculated by subtracting the estimate for 2010 from the
estimate for 2006. The significance of changes over time was
assessed by comparing CIs as described in this section.

Results
Overall, the age-adjusted percentage of respondents rating

their health as fair or poor did not change significantly from
2006 (16.3%) to 2010 (16.1%) (Table 1). A higher percentage
of women than men reported fair or poor health in both years.
However, neither of the groups experienced a significant
change from 2006 to 2010. A higher percentage of persons
in older age groups than younger groups rated their health as
fair or poor in both years. The percentage of persons aged
≥65 years reporting fair or poor health significantly decreased
approximately 2 percentage points from 2006 to 2010. Both
in 2006 and 2010, a significantly lower percentage of non-
Hispanic whites rated their health as fair or poor than all
other racial/ethnic groups except A/PIs. However, only two
of these racial/ethnic groups experienced a significant change
in self-rated health from 2006 to 2010: the percentage of
non-Hispanic blacks reporting fair or poor health increased
by 2 percentage points, and that of Hispanics decreased
approximately 3 percentage points. In both 2006 and 2010,
a higher percentage of those who had not graduated from
high school reported fair or poor health than did high school
graduates, and a lower percentage of college graduates reported
fair or poor health than did high school graduates. From 2006
to 2010, the percentage of high school graduates who reported
fair or poor self-rated health increased by 1.2 percentage points,
and the percentage of persons with some college education
who reported fair or poor self-rated health decreased by 1.6
percentage points. A higher percentage of persons who spoke
a language other than English at home reported fair or poor
health than those who spoke English at home. However, the
percentage of those who spoke Spanish at home and reported
fair or poor health decreased by 7 percentage points from 2006
to 2010. A higher percentage of persons with a disability rated
their health as fair or poor than did those without a disability
both in 2006 and 2010. Nonetheless, the percentage of
persons without a disability who rated their health as fair or
poor decreased by 0.8 percentage points from 2006 to 2010.

From 2006 to 2010, the overall age-adjusted mean number
of physically unhealthy days in the last 30 days increased
by approximately 0.1 days (2006: 3.6 days; 2010: 3.7 days
(Table 2). A higher mean number of physically unhealthy
days were reported by women than men in 2006 and 2010.
However, only men experienced a statistically significant
increase in mean number of days (0.2 days) over time. A higher
mean number of physically unhealthy days was reported by
older respondents than younger respondents. From 2006 to
2010, only persons aged 25–34 years reported a statistically
significant increase in mean number of physically unhealthy
days (0.3 days). In both 2006 and 2010, the fewest physically

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 107

unhealthy days were reported by A/PIs (2006: 2.4 days; 2010:
2.5 days) and the most were reported by AI/ANs, (2006:
6.2 days; 2010: 6.3 days). Hispanics showed an increase in
mean number of physically unhealthy days from 2006 to
2010 (0.6 days). Compared with high school graduates, more
physically unhealthy days were reported by those who had
not graduated from high school than by those with at least
some college. However, all but college graduates experienced
an increase in physically unhealthy days from 2006 to 2010,
with the least educated showing the largest increase (0.8 days).
More physically unhealthy days were reported by those who
spoke a language other than English at home than by those
who spoke English at home. From 2006 to 2010, those who
spoke Spanish at home had an 0.8-day increase in physically
unhealthy days, compared with an almost 10-day decrease

among those who spoke languages other than English and
Spanish. Approximately 8 more physically unhealthy days were
reported by persons with a disability (10 days) than by those
without a disability (1.8 days). Neither group had a significant
change in number of days from 2006 to 2010.

From 2006 to 2010, the overall age-adjusted mean number
of mentally unhealthy days in the last 30 days increased by
approximately 0.1 days (2006: 3.4 days; 2010: 3.5 days)
(Table 3). The mean number of mentally unhealthy days for
women exceeded those for men by approximately 1 day in both
years. However, only men showed a significant increase from
2006 to 2010 (by 0.2 days). A higher percentage of younger
respondents reported a mean number of mentally unhealthy
days than older respondents. Only those aged 35–79 years
experienced an increase in mean number of days from 2006

TABLE 1. Estimated percentage of adults aged ≥18 years who rated their health as fair or poor, by selected characteristics — Behavioral Risk
Factor Surveillance System, United States, 2006 and 2010

Characteristic

2006 2010

Change from
2006 to 2010
(percentage

points)

Age-
adjusted

percentage (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
adjusted

percentage (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 16.3 (15.9–16.6) — — 16.1 (15.9–16.4) — — -0.2

Sex
Male 15.8 (15.3–16.3) -0.8* -5.3* 15.4 (15.0–15.8) -1.4* -8* -0.4
Female 16.7 (16.3–17.0) Ref. Ref. 16.8 (16.5–17.1) Ref. Ref. 0.1

Age group (yrs)
18–24 9.3 (8.3–10.2) -2.4* -21* 7.6 (6.8–8.4) -4.0* -34* -1.7
25–34 9.9 (9.2–10.7) -1.7* -15* 9.9 (9.3–10.5) -1.6* -14* 0.0
35–44 11.7 (11.0–12.3) Ref. Ref. 11.5 (11.0–12.1) Ref. Ref. -0.2
45–64 18.8 (18.3–19.3) 7.2*. 61* 19.0 (18.6–19.4) 7.5* 65* 0.2
65–79 27.4 (26.6–28.2) 15.8* 135* 25.1 (24.6–25.6) 13.6* 118* -2.3*
≥80 33.2 (31.9–34.5) 21.6* 185* 31.1 (30.3–31.9) 19.6* 170* -2.1*
Race/Ethnicity

White, non-Hispanic 13.1 (12.8–13.4) Ref. Ref. 13.3 (13.0–13.5) Ref. Ref. 0.1
Black, non-Hispanic 21.3 (20.3–22.2) 8.1* 62* 23.3 (22.5–24.1) 10.0* 76* 2.0*
Hispanic† 31.0 (29.5–32.5) 17.9* 137* 28.1 (27.1–29.1) 14.8* 112* -2.9*
Asian/Pacific Islander 11.8 (9.8–13.7) -1.3 -10 11.9 (10.7–13.1) -1.4 -10 0.1
American Indian/

Alaska Native
26.7 (23.9–29.6) 13.6* 104* 30.8 (28.0–33.6) 17.6* 133* 4.1

Other 22.0 (18.5–25.5) 8.9* 68* 18.6 (15.9–21.3) 5.3* 40* -3.4
Educational attainment

Less than high school 39.1 (37.8–40.5) 20.0* 104* 38.4 (37.4–39.5) 18.1* 89* -0.7
High school graduate

or equivalent
19.2 (18.6–19.7) Ref. Ref. 20.3 (19.8–20.8) Ref. Ref. 1.2*

Some college 13.8 (13.3–14.3) -5.3* -28* 15.5 (15.0–15.9) -4.9* -24* 1.6*
College graduate 7.1 (6.8–7.4) -12.0* -63* 7.3 (7.0–7.6) -13.0* -64* 0.2

Language spoken at home
English 14.8 (14.5–15.1) Ref. Ref. 15.0 (14.8–15.3) Ref. Ref. 0.2
Spanish 44.8 (42.3–47.3) 30.0* 203* 37.6 (35.9–39.3) 22.6* 151* -7.2*
Other language 41.9 (18.2–65.6) 27.1* 183* 40.7 (31.3–50.1) 25.7* 171* -1.2

Disability status
With disability 38.7 (37.9–39.6) 29.3* 312* 39.4 (38.7–40.0) 30.7* 356* 0.6
Without disability 9.4 (9.1–9.7) Ref. Ref. 8.6 (8.4–8.9) Ref. Ref. -0.8*

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Difference considered statistically significantly different by comparison of nonoverlapping 95% CIs. Unrounded CIs do not overlap.
† Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

108 MMWR / November 22, 2013 / Vol. 62 / No. 3

to 2010 (0.2–0.3 days). A/PIs reported the fewest mentally
unhealthy days, and AI/ANs reported the most. However, from
2006 to 2010, only Hispanics showed a significant increase
(0.6 days). The number of mentally unhealthy days in 2006
and 2010 was higher for persons with less education than for
those with more education. However, all groups without a
college degree experienced a significant increase in the number
of days from 2006 to 2010. Similar to the change among
Hispanic respondents, who experienced an increase of 0.6
mentally unhealthy days from 2006 to 2010, the mean number
of mentally unhealthy days increased among those who spoke
Spanish at home by 0.9 days. The mean number of mentally
unhealthy days among persons with a disability (7 days) was
approximately five more than among persons without a disability
(2 days). Nonetheless, only persons with a disability showed a
statistically significant increase from 2006 to 2010 (0.3 days).

Discussion
The patterns of the health-related quality of life measures

among the various groups in this report are similar to previous
findings (1,11). Groups with higher percentages of fair or
poor health and who report more physically unhealthy days
and more mentally unhealthy days are usually women, older
persons (with respect to physical health), younger persons (with
respect to mental health), minority racial/ethnic groups (except
for A/PIs), those with less education, those who speak another
language besides English at home, and those with a disability.

Groups that had statistically significant changes in health-
related quality of life from 2006 to 2010 differ from groups
with statistically significant differences from the reference
groups during the individual years. Although minimal change
occurred overall, statistically significant changes occurred in

TABLE 2. Mean number of physically unhealthy days in the past 30 days among adults aged ≥18 years, by selected characteristics — Behavioral
Risk Factor Surveillance System, United States, 2006 and 2010

Characteristic

2006 2010

Change from
2006 to 2010
(percentage

points)

Age-
adjusted

mean no. of
days (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
adjusted

mean no. of
days (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 3.6 (3.5–3.6) — — 3.7 (3.6–3.7) — — 0.1*

Sex
Male 3.2 (3.1–3.3) -0.7* -19* 3.4 (3.3–3.5) -0.6* -14* 0.2*
Female 3.9 (3.8–4.0) Ref. Ref. 4.0 (3.9–4.0) Ref. Ref. 0.1

Age group (yrs)
18–24 2.1 (1.9–2.3) -0.7* -24* 2.0 (1.9–2.2) -0.8* -29* -0.1
25–34 2.2 (2.1–2.3) -0.6* -20* 2.5 (2.4–2.6) -0.4* -13* 0.3*
35–44 2.8 (2.7–2.9) Ref. Ref. 2.9 (2.8–3.0) Ref. Ref. 0.1
45–64 4.3 (4.2–4.4) 1.6* 55* 4.3 (4.3–4.4) 1.5* 51* 0.1
65–79 5.3 (5.2–5.5) 2.6* 92* 5.1 (5.0–5.2) 2.3* 79* -0.2
≥80 6.6 (6.3–6.9) 3.8* 127* 6.2 (6.0–6.4) 3.3* 116* -0.4
Race/Ethnicity

White, non-Hispanic 3.4 (3.4–3.5) Ref. Ref. 3.5 (3.4–3.6) Ref. Ref. 0.1
Black, non-Hispanic 4.0 (3.9–4.2) 0.6* 18* 4.3 (4.1–4.4) 0.8* 22* 0.2
Hispanic† 3.8 (3.6–4.0) 0.4* 10* 4.4 (4.2–4.5) 0.9* 25* 0.6*
Asian/Pacific Islander 2.4 (2.2–2.7) -1.0* -30* 2.5 (2.3–2.7) -1.0* -29* 0.1
American Indian/

Alaska Native
6.2 (5.5–6.9) 2.8* 82* 6.3 (5.7–6.9) 2.8* 79* 0.0

Other 5.2 (4.3–6.1) 1.8* 52* 4.3 (3.9–4.8) 0.9* 24* -0.9
Educational attainment

Less than high school 5.7 (5.4–5.9) 1.6* 40* 6.5 (6.3–6.7) 2.2* 51* 0.8*
High school graduate

or equivalent
4.0 (3.9–4.2) Ref. Ref. 4.3 (4.2–4.4) Ref. Ref. 0.3*

Some college 3.7 (3.6–3.8) -0.4* -9* 3.9 (3.8–4.0) -0.4* -9* 0.2*
College graduate 2.2 (2.2–2.3) -1.8* -44* 2.3 (2.2–2.3) -2.1* -48* 0.0

Language spoken at home
English 3.5 (3.5–3.6) Ref. Ref. 3.6 (3.6–3.7) Ref. Ref. 0.1
Spanish 4.0 (3.6–4.4) 0.5* 13* 4.8 (4.5–5.1) 1.2* 33* 0.8*
Other language 12.7 (4.4–21.0) 9.2* 260* 2.9 (1.8–4.0) -0.7 20 -9.8*

Disability status
With disability 10.0 (9.8–10.2) 8.3* 464* 10.2 (10.0–10.3) 8.4* 471* 0.1
Without disability 1.8 (1.7–1.8) Ref. Ref. 1.8 (1.7–1.8) Ref. Ref. 0.0

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Difference considered statistically significantly different by comparison of nonoverlapping 95% CIs. Unrounded CIs do not overlap.
† Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 109

specific groups. Men (but not women) reported an increase
in the number of physically and mentally unhealthy days over
time. Persons aged ≥65 years rated their overall health better in
2010 than in 2006. Hispanics and those who spoke Spanish
at home also rated their overall health better in 2010 than in
2006, despite reporting increases in numbers both of physically
and mentally unhealthy days. Numbers of physically and
mentally unhealthy days increased from 2006 to 2010 among
persons without a college degree. The number of mentally
unhealthy days but not of physically unhealthy days increased
among persons with a disability, although persons without a
disability rated their overall health better.

Reasons for particular changes in health-related quality are
unclear. Differences in risky and protective health behaviors,
in socioeconomic circumstances such as employment status
and household income, and in disease status have been
associated with differences in the measures used in this analysis
to assess health-related quality of life (1,11). Hispanics and
those without a college degree reported more physically and
mentally unhealthy days in 2010 than in 2006; however,
others in similar socioeconomic circumstances (e.g., non-
Hispanic blacks and AI/ANs) did not. What accounted for
these differences is unclear. Additional analyses that adjust for
changes in employment status, the effects of housing loss, and
the recent increase in enforcement against illegal immigrants
might clarify these differences.

TABLE 3. Mean number of mentally unhealthy days in the past 30 days among adults aged ≥18 years, by selected characteristics — Behavioral
Risk Factor Surveillance System, United States, 2006 and 2010

Characteristic

2006 2010

Change from
2006 to 2010
(percentage

points)

Age-
adjusted

mean no. of
days (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age-
adjusted

mean no. of
days (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 3.4 (3.3–3.5) — — 3.5 (3.5–3.6) — — 0.1*

Sex
Male 2.7 (2.7–2.8) -1.3* -32 3.0 (2.9–3.0) -1.2* -28* 0.2*
Female 4.0 (4.0–4.1) Ref. Ref. 4.1 (4.0–4.2) Ref. Ref. 0.1

Age group (yrs)
18–24 4.3 (4.0–4.5) 0.9* 26* 4.0 (3.8–4.2) 0.4* 10* -0.3
25–34 3.7 (3.5–3.8) 0.3* 9* 3.8 (3.7–4.0) 0.2 6 0.2
35–44 3.4 (3.3–3.5) Ref. Ref. 3.6 (3.5–3.8) Ref. Ref. 0.3*
45–64 3.6 (3.5–3.7) 0.2 6 3.8 (3.8–3.9) 0.2* 6* 0.3*
65–79 2.1 (2.0–2.3) -1.2* -37* 2.3 (2.3–2.4) -1.3* -36* 0.2*
≥80 2.0 (1.9–2.2) -1.3* -40* 2.0 (1.9–2.1) -1.7* -46* -0.1
Race/Ethnicity

White, non-Hispanic 3.4 (3.3–3.4) Ref. Ref. 3.5 (3.4–3.5) Ref. Ref. 0.1
Black, non-Hispanic 3.8 (3.6–4.0) 0.5* 13* 4.0 (3.8–4.2) 0.5* 15* 0.2
Hispanic† 3.2 (3.0–3.5) -0.1 4 3.8 (3.6–4.0) 0.3* 10* 0.6*
Asian/Pacific Islander 2.1 (1.9–2.4) -1.2* -37* 2.0 (1.7–2.3) -1.5* -42* -0.1
American Indian/

Alaska Native
5.7 (5.0–6.3) 2.3* 68* 5.7 (5.1–6.4) 2.3* 65* 0.1

Other 5.1 (4.3–6.0) 1.7* 52* 3.9 (3.4–4.5) 0.5 14 -1.2
Educational attainment

Less than high school 4.9 (4.6–5.1) 1.1* 28* 5.6 (5.4–5.8) 1.6* 40* 0.7*
High school graduate

or equivalent
3.8 (3.7–3.9) Ref. Ref. 4.0 (3.9–4.1) Ref. Ref. 0.2*

Some college 3.6 (3.5–3.8) -0.2 -5 3.9 (3.8–4.0) -0.1 -3 0.3*
College graduate 2.3 (2.2–2.4) -1.5* -39* 2.3 (2.3–2.4) -1.7* -42* 0.0

Language spoken at home
English 3.5 (3.4–3.5) Ref. Ref. 3.5 (3.5–3.6) Ref. Ref. 0.1
Spanish 2.9 (2.6–3.3) -0.5* -15* 3.8 (3.6–4.1) 0.3 9 0.9*
Other language 8.9 (0.0–17.9) 5.5 159 3.9 (2.3–5.5) 0.4 11 -5.0

Disability status
With disability 7.2 (7.1–7.4) 4.9* 207* 7.5 (7.4–7.7) 5.2* 220* 0.3*
Without disability 2.4 (2.3–2.4) Ref. Ref. 2.4 (2.3–2.4) Ref. Ref. 0.0

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Difference considered statistically significantly different by comparison of nonoverlapping 95% CIs. Unrounded CIs do not overlap.
† Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

110 MMWR / November 22, 2013 / Vol. 62 / No. 3

Limitations
The findings in this report are subject to at least four

limitations. First, although the BRFSS health-related quality
of life questions have been shown to be reliable in predicting
30-day and 1-year hospitalization and mortality (12,13),
because the health-related quality of life data are self-reported,
they might be misclassified because they are not objectively
verifiable and are subject to recall bias and measurement
error. Second, although BRFSS uses poststratification
to adjust respondent sampling weights for non-response
(7,8), this adjustment assumes that nonrespondents would
have answered in similar ways to respondents with similar
demographic characteristics; such poststratification might not
have fully adjusted for differences between nonrespondents
and respondents, given the low, state-specific overall response
rates. Third, BRFSS data are cross-sectional; therefore,
changes in the composition of the BRFSS sample from 2006
to 2010 that affect responses to the health-related quality of
life questions might affect measured differences from 2006 to
2010. Finally, the results were adjusted for age only; therefore,
other confounding variables also might have affected measured
differences from 2006 to 2010.

Conclusion
Although direct interventions to improve health-related

quality of life are not possible, indirect interventions to change
characteristics associated with health-related quality of life
might result in improvements. For example, risky health
behaviors can decrease health-related quality of life. Persons
who smoke cigarettes have worse health-related quality of life
than former smokers or never smokers (14), and smoking is
more prevalent among those with certain health conditions
such as epilepsy (15).

Cigarette smoking is a well-known cause of multiple types
of cancer (16). Persons with epilepsy (17) and cancer (18)
have worse health-related quality of life than those without
these conditions. Moreover, protective health behaviors can
increase health-related quality of life. For example, persons
who engage in physical activity have better health-related
quality of life than those who are sedentary (19). Physical
activity also reduces obesity (20) and its complications and has
been associated both with reduced colon cancer rates (20) and
reduced complications from different kinds of arthritis (21).
Persons who are obese (22), have cancer (18), or have arthritis
(23) have worse health-related quality of life than those without
these conditions. Therefore, interventions to eliminate risky
behaviors, promote protective behaviors, and delay or prevent

complications from diseases and other conditions would
probably improve health-related quality of life.

References
1. CDC. Measuring healthy days: population assessment of health-related

quality of life. Atlanta, GA: US Department of Health and Human Services,
CDC; 2000. Available at http://www.cdc.gov/hrqol/pdfs/mhd.pdf.

2. Moriarty DG, Zack M, Kobau R. The Centers for Disease Control and
Prevention’s Healthy Days Measures—population tracking of perceived
physical and mental health over time. Health Qual Life Outcomes
2003;1:37 .

3. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

4. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

5. Li C, Balluz LS, Okoro CA, et al. Surveillance of certain health behaviors
and conditions among states and selected local areas—Behavioral Risk Factor
Surveillance System, United States, 2009. MMWR 2011;60(No. SS-9).

6. Mokdad AH. The Behavioral Risk Factor Surveillance System: past,
present, and future. Annu Rev Public Health 2009;30:43–54.

7. CDC. 2006 Behavioral Risk Factor Surveillance System summary data
quality report. Atlanta, GA: CDC; 2007. Available at http://www.cdc.
gov/brfss/annual_data/2006/2006SummaryDataQualityReport.docx.

8. CDC. 2010 Behavioral Risk Factor Surveillance System summary data
quality report. Atlanta, GA: CDC; 2011. Available at http://www.cdc.gov/
brfss/annual_data/2010/2010_Summary_Data_Quality_Report.docx.

9. CDC. Healthy people 2010 operational definition: 6–1. Atlanta, GA:
CDC. Available at http://ftp.cdc.gov/pub/health_statistics/nchs/datasets/
data2010/focusarea06/O0601.pdf.

10. Research Triangle Institute. SUDAAN language manual, release 10.0.
Research Triangle Park, NC: Research Triangle Institute; 2010.

11. Zahran HS, Kobau R, Moriarty DG, et al. Health-related quality of life
surveillance—United States, 1993–2002. MMWR 2005;54(No. SS-4).

12. Andresen EM, Catlin TK, Wyrwich KW, Jackson-Thompson J. Retest
reliability of surveillance questions on health related quality of life. J
Epidemiol Community Health 2003;57:339–43.

13. Dominick KL, Ahern FM, Gold CH, Heller DA. Relationship of health-
related quality of life to health care utilization and mortality among
older adults. Aging Clin Exp Res 2002;14:499–508.

14. Mody RR, Smith MJ. Smoking status and health-related quality of life:
Findings from the 2001 Behavioral Risk Factor Surveillance System. Am
J Health Promot 2006;20:251–8.

15. CDC. Epilepsy surveillance among adults—19 states, Behavioral Risk
Factor Surveillance System, 2005. MMWR 2008;57(No. SS-6).

16. US Department of Health and Human Services. How tobacco smoke
causes disease: the biology and behavioral basis for smoking-attributable
disease: a report of the Surgeon General. Atlanta, GA: US Department
of Health and Human Services, CDC; 2010.

17. Kobau R, Zahran H, Grant D, et al. Prevalence of active epilepsy and
health-related quality of life among adults with self-reported epilepsy in
California: California Health Interview Survey, 2003. Epilepsia 2007;
48:1904–13.

18. Richardson LC, Wingo PA, Zack MM, Zahran HS, King JB. Health-
related quality of life (HRQOL) in cancer survivors between 20 and 64:
Population-based estimates from the Behavioral Risk Factor Surveillance
System (BRFSS). Cancer 2008;112:1380–9.

19. Brown DW, Balluz LS, Heath GW, et al. Associations between
recommended levels of physical activity and health-related quality of
life. Findings from the 2001 Behavioral Risk Factor Surveillance System
(BRFSS) survey. Prev Med 2003;37:520–8 Available at http://www.
sciencedirect.com/science/article/pii/S0091743503001798.

20. US Department of Health and Human Services. Physical activity and
health: a report of the Surgeon General. Atlanta, GA: US Department
of Health and Human Services, CDC; 1996.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 111

21. Loew L, Brosseau L, Wells GA, et al. Ottawa panel evidence-based clinical
practice guidelines for aerobic walking programs in the management of
osteoarthritis. Arch Phys Med Rehabil 2012;93:1269–85.

22. Ford ES, Moriarty DG, Zack MM, Mokdad AH, Chapman DP. Self-
reported body mass index and health-related quality of life: findings
from the Behavioral Risk Factor Surveillance System. Obes Res
2001;9:21–31.

23. Furner SE, Hootman JM, Helmick CG, Bolen J, Zack MM. Health-
related quality of life of U.S. adults with arthritis: analysis of data from
the Behavioral Risk Factor Surveillance System, 2003, 2005, and 2007.
Arthritis Care Res (Hoboken) 2011;63:788–99.

Supplement

112 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
At the end of 2009, approximately 1.1 million persons in the

United States were living with human immunodeficiency virus
(HIV) infection (1), with approximately 50,000 new infections
annually (2). The prevalence of HIV continues to be greatest
among gay, bisexual, and other men who have sex with men
(MSM), who comprised approximately half of all persons with
new infections in 2009 (2). Disparities also exist among racial/
ethnic minority populations, with blacks/African Americans
and Hispanics/Latinos accounting for approximately half of
all new infections and deaths among persons who received an
HIV diagnosis in 2009 (2,3). Improving survival of persons
with HIV and reducing transmission involve a continuum of
services that includes diagnosis, linkage to and retention in HIV
medical care, and ongoing HIV prevention interventions (4).

The HIV analysis and discussion that follows is part of
the second CDC Health Disparities and Inequalities Report
(CHDIR) and updates information presented in the first
CHDIR (5). The 2011 CHDIR (6) was the first CDC report
to assess disparities across a wide range of diseases, behavioral
risk factors, environmental exposures, social determinants, and
health-care access. The topic presented in this report is based
on criteria that are described in the 2013 CHDIR Introduction
(7). The purposes of this HIV infection report are to discuss
and raise awareness of differences in the characteristics of
people with HIV infection and to prompt actions to reduce
these disparities

Methods
To estimate the number of adults aged ≥18 years who received

a diagnosis of HIV infection during 2008 and 2010, CDC
analyzed data reported through June 2011 to the National
HIV Surveillance System (NHSS). CDC funds and assists
state and local health departments to collect case information
on persons with an HIV diagnosis. Health departments

report deindentified data to CDC, which are compiled for
national analyses. Analysis of HIV case surveillance data was
limited to the 46 states that had reported HIV cases since at
least January 2007 to allow for estimation of diagnoses rates:
Alabama, Alaska, Arizona, Arkansas, California, Colorado,
Connecticut, Delaware, Florida, Georgia, Idaho, Illinois,
Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Michigan,
Minnesota, Mississippi, Missouri, Montana, Nebraska,
Nevada, New Hampshire, New Jersey, New Mexico, New York,
North Carolina, North Dakota, Ohio, Oklahoma, Oregon,
Pennsylvania, Rhode Island, South Carolina, South Dakota,
Tennessee, Texas, Utah, Virginia, Washington, West Virginia,
Wisconsin, and Wyoming. Rates per 100,000 population were
calculated for 2008 and 2010 by age, sex, and race/ethnicity,
with population denominators based on postcensal estimates
for 2009 from the U.S. Census Bureau (8). Household income
and educational attainment were not calculated because these
data are not collected by NHSS. Geographic location was
not calculated because estimates of HIV diagnoses among
persons in all 50 states and the District of Columbia were
unable to be calculated at the time of this analysis. Analysis
of transmission categories was limited to all men and MSM
because denominator data for transmission categories other
than MSM were unavailable (9); the category of all men
was used as the referent group. To compute estimated MSM
population denominators used for calculating HIV diagnosis
rates, CDC applied the estimated proportion of men in
the United States who reported ever having male-to-male
sex (6.9%; 95% confidence interval [CI]: 5.1%–8.6%) to
postcensal estimated populations for men (9). Analyses were
adjusted for reporting delays (i.e., the time between diagnosis
and report) and for missing risk factor information but not
for underreporting (3).

Data from the Medical Monitoring Project (MMP) were
used to estimate percentages of adults aged ≥18 years receiving
outpatient medical care whose medical record documented that
they 1) were prescribed antiretroviral therapy (ART) during

HIV Infection — United States, 2008 and 2010
Anna Satcher Johnson, MPH

Linda Beer, PhD
Catlainn Sionean, PhD

Xiaohong Hu, MS
Carolyn Furlow-Parmley, PhD

Binh Le, MD
Jacek Skarbinski, MD
H. Irene Hall, PhD
Hazel D. Dean, ScD

National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC

Corresponding author: Anna Satcher Johnson, Division of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention,
CDC. Telephone: 404-639-2050; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 113

the past 12 months, 2) had a suppressed viral load (defined as
undetectable or ≤200 copies/mL at their most recent test), and
3) reported receiving prevention counseling in a clinical setting
during the 12 months preceding the interview. Nationally
representative percentages and associated standard errors were
estimated for patients in care in 2009 and interviewed during
2009–2010. MMP collects behavioral and clinical information
from a nationally representative sample of adults receiving
medical care for HIV infection in outpatient facilities in the
United States and Puerto Rico (10–12). A total of 23 project
areas were funded to conduct data collection activities for the
2009 MMP data collection cycle: California; Chicago, Illinois;
Delaware; Florida; Georgia; Houston, Texas; Illinois; Indiana;
Los Angeles County, California; Michigan; Mississippi; New
Jersey; the state of New York; New York City, New York; North
Carolina; Oregon; Pennsylvania; Philadelphia, Pennsylvania;
Puerto Rico; San Francisco, California; Texas; Virginia; and
Washington. Patients who received medical care during
January–April 2009 at an MMP participating facility were
interviewed once during June 2009–April 2010 regarding
the 12 months preceding the interview. In addition, patients’
medical records were abstracted for documentation of medical
care (including prescription of ART and HIV viral load) for
the 12 months preceding the interview. All percentages were
weighted for the probability of selection and adjusted for
nonresponse bias. Standard errors were calculated and account
for weighting and complex sample survey design. Associations
between variables were assessed using Rao-Scott chi-square
tests, with significance set at p<0.05. Detailed methods for
MMP have been described previously (10–12).

Data from the 2008 MSM cycle of the National HIV
Behavioral Surveillance System (NHBS)§ were used to estimate
percentages of MSM aged 18–64 years who 1) engaged
in unprotected anal sex with a casual partner, 2) reported
testing for HIV during the previous 12 months, and 3) who
participated in a behavioral intervention. Men who reported
being infected with HIV or who had no male sex partners
during the 12 months before interview were excluded from
this analysis. NHBS monitors HIV-associated behaviors and
HIV positivity within selected metropolitan statistical areas
(MSAs) with a high prevalence of acquired immunodeficiency
syndrome (AIDS) among three populations at high risk for HIV
infection: MSM, injection-drug users, and heterosexual adults at
increased risk for HIV infection. Data for NHBS are collected
in annual rotating cycles. All NHBS participants must be aged
≥18 years, live in a participating MSA, and be able to complete
a behavioral survey in English or Spanish. MSM participants
were recruited using venue-based sampling. Detailed methods
for NHBS have been described previously (13).

Disparities were measured as the deviations from a referent
category rate or prevalence. Absolute difference was measured as
the simple difference between a population subgroup estimate
and the estimate for its respective reference group. The percentage
relative difference was calculated by dividing the difference by
the value in the referent category and multiplying by 100 (14).

Results
In the 46 states for which HIV case surveillance data from the

NHSS were analyzed, a total of 46,379 adults aged ≥18 years
received a diagnosis of HIV in 2008, and 46,381 received an
HIV diagnosis in 2010. During 2010, the relative difference
in the HIV diagnosis rate among blacks/African Americans
compared with whites was eightfold and for Hispanics/Latinos,
persons of multiple races, and Native Hawaiians/other Pacific
Islanders (NH/OPI), the relative difference was twofold
compared with whites (Table 1).

Although the racial/ethnic disparities in rates of HIV diagnoses
among men were similar to the disparities observed for the
racial/ethnic groups overall, larger differences occurred among
women. In 2010, among women, the relative difference in HIV
diagnosis rates among black/African American women was
twentyfold compared with whites, among women of multiple
races was fourfold compared with whites, among Hispanic/
Latino women was threefold compared with whites, and
among AI/ANs was twofold compared with whites (Table 1).
From 2008 to 2010, the relative differences increased for all
racial/ethnic groups of women except NH/OPIs and women
of multiple races compared with whites. The largest relative
difference was observed for MSM compared with all other men
(an approximate 46-fold difference) in 2010, as well as the largest
change from 2008 to 2010 (763 percentage points).

Among adults aged ≥18 years in MMP, representing persons
receiving medical care in 2009, assessment of the data by age
group indicated that the percentages of persons who were
prescribed ART increased as age increased. Compared with
adults aged ≥55 years, a lower percentage of young persons (aged
18–24 years and 25–34 years) were prescribed ART (relative
difference: -18% and -16%, respectively). By race/ethnicity,
lower percentages of blacks/African Americans were prescribed
ART than were whites (relative difference: -7%) (Table 2).
A higher percentage of men were prescribed ART than were
women, with a relative difference of 5%. Among men, lower
percentages of blacks/African Americans were prescribed ART
than were whites, with a relative difference of -6%. A similar
pattern was observed in the percentage of women prescribed
ART, with a lower percentage of blacks/African Americans
prescribed ART than whites, (relative difference: -7%).

Supplement

114 MMWR / November 22, 2013 / Vol. 62 / No. 3

Among persons prescribed ART in 2009, persons aged 25–34
years and 35–44 years accounted for the lowest percentages of
persons with a suppressed viral load. Compared with persons
aged ≥55 years, relative differences in viral suppression were
-18% for persons aged 25–34 years and -15% for persons aged
35–44 years. By race/ethnicity, lower percentages of blacks/
African Americans and Hispanics/Latinos had a suppressed
viral load than whites, with relative differences of -15% and
−5%, respectively (Table 2). A higher percentage of men had
a suppressed viral load at their most recent test than women
(relative difference: 10%). Among men, lower percentages

of blacks/African Americans and Hispanics/Latinos had a
suppressed viral load than whites, with relative differences
of -16% and -6%, respectively. Examination of other
demographic characteristics indicated that a higher percentage
of persons who spoke Spanish with friends and family had a
suppressed viral load at their most recent test than English-
speaking persons, with a relative difference of 6%. A higher
percentage of persons who self-identified as homosexual had
a suppressed viral load than persons who self-identified as
heterosexual, with a relative difference of 11%. The percentage

TABLE 1. Estimated rate* of HIV infection diagnoses among adults aged ≥18 years† — National HIV Surveillance System, 46 states,§ 2008 and 2010

Characteristic 2008 rate

Absolute
difference

(percentage
points)

Relative difference
(%) 2010 rate

Absolute
difference

(percentage
points)

Relative difference
(%)

Age group (yrs)
18–24 27.7 21.9 377.6 32.0 26.2 451.7
25–34 32.2 26.4 455.2 32.3 26.5 456.9
35–44 31.7 25.9 446.6 28.5 22.7 391.4
45–54 21.9 16.1 277.6 21.2 15.4 265.5
≥55 5.8 Ref. Ref. 5.8 Ref. —
Race/Ethnicity

American Indian/Alaska Native 13.3 4.1 44.6 13.5 4.4 48.4
Asian 8.1 -1.1 -12.0 8.4 -0.7 -7.7
Black/African American 86.0 76.8 834.8 84.0 74.9 823.1
Hispanic/Latino¶ 31.1 21.9 238.0 30.9 21.8 239.6
Native Hawaiian/Other Pacific Islander 26.9 17.7 192.4 27.0 17.9 196.7
White 9.2 Ref. Ref. 9.1 Ref. Ref.
Multiple races 34.7 25.5 277.2 28.4 19.3 212.1

Sex
Male 33.3 23.6 243.3 34.0 25.4 295.3
Female 9.7 Ref. Ref. 8.6 Ref. Ref.
Male

American Indian/Alaska Native 21.3 5.1 31.5 20.2 3.7 22.4
Asian 14.6 -1.6 -9.9 14.8 -1.7 -10.3
Black/African American 125.4 109.2 674.1 128.4 111.9 678.2
Hispanic/Latino 49.7 33.5 206.8 49.9 33.4 202.4
Native Hawaiian/Other Pacific Islander 46.8 30.6 188.9 49.2 32.7 198.2
White 16.2 Ref. Ref. 16.5 Ref. Ref.
Multiple races 52.1 35.9 221.6 46.9 30.4 184.2

Female
American Indian/Alaska Native 5.7 3.1 119.2 7.1 4.9 222.7
Asian 2.2 -0.4 -15.4 2.6 0.4 18.2
Black/African American 51.8 49.2 1,892.3 45.3 43.1 1,959.1
Hispanic/Latino 11.0 8.4 323.1 10.2 8.0 363.6
Native Hawaiian/Other Pacific Islander 7.2 4.6 176.9 5.0 2.8 127.3
White 2.6 Ref. Ref. 2.2 Ref. Ref.
Multiple races 18.7 16.1 619.2 11.4 9.2 418.2

Transmission category
Men who have sex with men** 359.1 349.9 3,803.3 382.6 374.4 4,565.9
All other men 9.2 Ref. Ref. 8.2 Ref. Ref.

Abbreviations: HIV = human immunodeficiency virus; Ref. = referent.
* Per 100,000 population.
† A total of 46,379 adults aged ≥18 years received a diagnosis of HIV in 2008; 46,381 received a diagnosis in 2010.
§ Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine,

Michigan, Minnesota, Mississippi, Missouri, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio,
Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Washington, West Virginia, Wisconsin, and Wyoming.

¶ Persons of Hispanic/Latino ethnicity might be of any race or combination of races.
** Denominator calculated by applying the estimated proportion of men in the United States who reported ever having male-to-male sex (6.9%; 95% confidence

interval: 5.1%–8.6%) to postcensal estimated populations for men.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 115

TABLE 2. Percentage of adults aged ≥18 years receiving care for HIV infection who were prescribed antiretroviral therapy, had viral load
suppression at their most recent HIV viral load test, and received prevention counseling during the past 12 months, by selected characteristics
— Medical Monitoring Project, United States, 2009*

Characteristic

Prescribed ART

Prescribed ART and most recent HIV
viral load test was undetectable or

≤200 copies/mL
Received prevention counseling

from a health-care provider†

% (SE)§

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Age group (yrs)
18–24 75.8 (5.8)¶ -16.4 -17.8 77.8 (4.6) -7.5 -8.8 73.3 (5.4)¶ 37.5 104.7
25–34 77.6 (2.8)¶ -14.6 -15.8 70.1 (2.5)¶ -15.2 -17.8 59.0 (3.4)¶ 23.2 64.8
35–44 88.3 (1.4)¶ -3.9 -4.2 72.8 (2.3)¶ -12.5 -14.7 46.7 (2.6)¶ 10.9 30.4
45–54 91.4 (0.7) -0.8 -0.9 79.2 (1.6)¶ -6.1 -7.2 41.6 (2.9)¶ 5.8 16.2
≥55 92.2 (1.1) Ref. Ref. 85.3 (1.4) Ref. Ref. 35.8 (2.6) Ref. Ref.
Race/Ethnicity

Black/African American 86.0 (1.3)¶ -6.2 -6.7 71.4 (1.8)¶ -12.7 -15.1 54.2 (2.7)¶ 25.5 88.9
Hispanic/Latino** 89.2 (1.4) -3.0 -3.3 79.8 (1.8)¶ -4.3 -5.1 51.9 (2.2)¶ 23.2 80.8
White 92.2 (0.8) Ref. Ref. 84.1 (1.7) Ref. Ref. 28.7 (1.8) Ref. Ref.
Other 85.7 (3.5)¶ -6.5 -7.0 76.7 (2.9)¶ -7.4 -8.8 47.9 (4.0)¶ 19.2 66.9

Sex
Male 89.9 (0.9)¶ 4.1 4.8 79.7 (1.6)¶ 6.9 9.5 42.6 (2.7)¶ -7.3 -14.6
Female 85.8 (1.5) Ref. Ref. 72.8 (1.8) Ref. Ref. 49.9 (2.2) Ref. Ref.
Male

Black/African American 87.1 (1.5)¶ -5.4 -5.8 71.7 (2.4)¶ -13.8 -16.1 55.5 (3.1)¶ 27.9 101.1
Hispanic/Latino 90.7 (1.4) -1.8 -1.9 80.4 (1.9)¶ -5.1 -6.0 50.8 (2.6)¶ 23.2 84.1
White 92.5 (0.9) Ref. Ref. 85.5 (1.7) Ref. Ref. 27.6 (1.9) Ref. Ref.
Other 85.3 (4.1)¶ -7.2 -7.8 79.4 (2.8) -6.1 -7.1 45.3 (4.5)¶ 17.7 64.1

Female
Black/African American 84.4 (1.7)¶ -6.3 -6.9 71.0 (2.1) -4.1 -5.5 52.2 (2.5)¶ 17.1 48.7
Hispanic/Latino 85.2 (3.3) -5.5 -6.1 77.9 (3.6) 2.8 3.7 55.3 (4.0)¶ 20.2 57.5
White 90.7 (2.0) Ref. Ref. 75.1 (3.2) Ref. Ref. 35.1 (3.8) Ref. Ref.
Other 87.2 (5.2) -3.5 -3.9 66.1 (11.8) -9.0 -12.0 58.5 (6.6)¶ 23.4 66.7

Place of birth
United States or U.S. territory 88.9 (0.9) Ref. Ref. 77.3 (1.5) Ref. Ref. 43.6 (2.8) Ref. Ref.
Other 87.5 (2.1) -1.4 -1.6 81.7 (2.4) 4.4 5.7 51.1 (3.1)¶ 7.5 17.2

Language most comfortable
speaking with family and friends
English 88.6 (1.0) Ref. Ref. 77.3 (1.5) Ref. Ref. 43.1 (2.8) Ref. Ref.
Spanish 90.1 (1.7) 1.5 1.7 81.7 (1.9)¶ 4.4 5.7 54.9 (2.9)¶ 11.8 27.4
Other 88.9 (4.1) 0.3 0.3 79.9 (4.3) 2.6 3.4 54.3 (7.4) 11.2 26.0

Sexual identity
Heterosexual (straight) 88.6 (1.1) Ref. Ref. 74.4 (1.5) Ref. Ref. 49.0 (2.5) Ref. Ref.
Homosexual (gay or lesbian) 89.7 (0.9) 1.1 1.2 82.2 (1.6)¶ 7.8 10.5 38.5 (2.5)¶ -10.5 -21.4
Bisexual 85.1 (2.3) -3.5 -4.0 76.2 (2.9) 1.8 2.4 49.3 (4.1) 0.3 0.6

Educational attainment
Less than high school 90.1 (1.2) 0.7 0.8 70.9 (2.0)¶ -15.6 -18.0 53.4 (2.5)¶ 18.7 53.9
High school graduate or equivalent 89.1 (1.0) -0.3 -0.3 75.0 (1.9)¶ -11.5 -13.3 47.8 (3.0)¶ 13.1 37.8
Some college 87.2 (1.4) -2.2 -2.5 80.3 (1.8)¶ -6.2 -7.2 41.3 (2.5)¶ 6.6 19.0
College graduate 89.4 (1.4) Ref. Ref. 86.5 (1.8) Ref. Ref. 34.7 (2.6) Ref. Ref.

Abbreviations: ART = antiretroviral therapy; HIV = human immunodeficiency virus; MMP = Medical Monitoring Project; Ref. = referent; SE = standard error.
* A total of 23 project areas were funded to conduct data collection activities for the 2009 MMP data collection cycle: California; Chicago, Illinois; Delaware; Florida;

Georgia; Houston, Texas; Illinois; Indiana; Los Angeles County, California; Michigan; Mississippi; New Jersey; the state of New York; New York City, New York; North
Carolina; Oregon; Pennsylvania; Philadelphia, Pennsylvania; Puerto Rico; San Francisco, California; Texas; Virginia; and Washington. Information regarding prescription
of ART and HIV viral load was abstracted from the patient’s medical record. Patients who received medical care during January–April 2009 at an MMP participating
facility were interviewed once during June 2009–April 2010 regarding all medical visits during the 12 months before the interview. In addition, patients’ medical
records were abstracted for documentation of medical care for the 12 months before the interview.

† Based on self-reported information from the patient interview about discussions with a physician, nurse, or other health-care provider. Topics might have included
condom negotiation, how to practice safer sexual behavior or injection use, or how to talk with partners about safe sex. Discussion occurring during sessions that
were part of HIV testing and counseling encounters were not included.

§ All percentages are weighted for probability of selection and nonresponse bias adjustment.
¶ Significant difference between group estimate and referent category, with significance set at p<0.05 by Rao-Scott chi-square test.
** Persons of Hispanic/Latino ethnicity might be of any race or combination of races.

Supplement

116 MMWR / November 22, 2013 / Vol. 62 / No. 3

of persons with a suppressed viral load increased as educational
attainment increased.

Of persons receiving HIV care in the United States in 2009,
persons in younger age groups reported higher percentages of
receipt of HIV prevention counseling than those aged ≥55
years. Higher percentages of blacks/African Americans and
Hispanics/Latinos had received HIV prevention counseling
from a health-care provider during the 12 months before
their interview than whites, with relative differences of 89%
and 81%, respectively (Table 2). A lower percentage of men
received HIV prevention counseling from a health-care
provider than women (relative difference: -15%). Among men,
higher percentages of blacks/African Americans and Hispanics/
Latinos received HIV prevention counseling than whites, with
relative differences of 101% and 84%, respectively. Findings
were similar among women; higher percentages of blacks/
African Americans and Hispanics/Latinas had received HIV
prevention counseling than whites, with relative differences of
49% and 58%, respectively. A higher percentage of persons
born outside the United States received HIV prevention
counseling than persons born in the United States, with a
relative difference of 17%. A higher percentage of persons
who spoke Spanish with friends and family had received HIV
prevention counseling than English-speaking persons, with a
relative difference of 27%. A lower percentage of persons who
self-identified as homosexual had received HIV prevention
counseling than persons who self-identified as heterosexual,
with a relative difference of -21%. The percentage of persons
receiving prevention counseling increased as educational
attainment decreased. Compared with college graduates,
relative differences in the percentage of persons who received
HIV prevention counseling were 54% for persons with less
than a high school education, 38% for high school graduates,
and 19% for persons with some college or the equivalent.

Among MSM in NHBS in 2008, unprotected anal sex with a
casual male partner was most common in younger age groups,
with relative differences of 38% among MSM aged 25–34
years and 26% among those aged 35–44 years, compared with
MSM aged ≥55 years (Table 3). By race/ethnicity, Hispanic/
Latino MSM and MSM of multiple races accounted for the
largest percentages of MSM who engaged in unprotected anal
sex with a casual partner, with relative differences of 14% and
17%, respectively, compared with whites.

The percentages of MSM who had been tested for HIV
infection in the 12 months before the interview were higher
among younger than older MSM and those who identified
as homosexual than those who did not, similar among racial
and ethnic groups, and increased with educational attainment
(Table 3). Specifically, HIV testing in the 12 months before
interview was highest among MSM aged 18–24 and 25–34

years, with relative differences of 37% and 36%, respectively,
compared with men aged ≥55 years. The percentage of MSM
who reported HIV testing in the 12 months before interview
was lowest among MSM with less than a high school education,
with a relative difference of -27% compared with MSM who
were college graduates.

The percentages of MSM who reported participation in a
behavioral HIV intervention in the 12 months before interview
were higher among younger than older MSM and among
MSM of minority racial/ethnic groups than whites (Table 3).
MSM aged 18–24 years accounted for the highest percentage
of MSM who participated in a behavioral intervention, with
a relative difference of 148% compared with men aged ≥55
years. The percentage of MSM who participated in a behavioral
intervention varied by level of educational attainment.
Compared with MSM who had graduated from college, the
percentage of MSM who had participated in a behavioral
intervention was higher among MSM with lower levels of
educational attainment, with relative differences of 17%, 29%,
and 46% for less than high school, high school graduate, and
some college or technical school, respectively.

Discussion
Although the relative difference in HIV infection diagnoses

between whites and blacks/African Americans decreased from
2008 to 2010, all racial/ethnic minorities, except Asians,
continue to experience higher rates of HIV diagnoses than
whites. These differences might reflect HIV incidence, testing
patterns, or both. Compared with whites, lower percentages
of blacks/African Americans were prescribed ART and lower
percentages of both blacks/African Americans and Hispanics/
Latinos had suppressed viral loads. Differences in rates of ART
prescription and viral suppression might reflect differences
in insurance coverage, prescription drug costs, health-care
providers’ perceptions of patients, or other factors associated
with adherence (4). Rates of HIV infection are increasing among
MSM, particularly young black/African American MSM (2).
However, among MSM, similar percentages of blacks/African
American and Hispanic/Latino MSM reported HIV testing
compared with white MSM, and higher percentages reported
receipt of behavioral interventions than white MSM.

Limitations
The NHSS data presented in this report are subject to at

least three limitations. First, data were not available from
all states. According to the cumulative estimated number of
AIDS diagnoses through 2010, the 46 states with confidential

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 117

name-based reporting since at least 2007 for which data were
used represent approximately 92% of AIDS diagnoses in the
50 states and the District of Columbia. Second, adjustments
made to HIV case surveillance data for reporting delays and
missing transmission category information are subject to a
degree of uncertainty that might result in less stable rates for
the most recent years. Finally, although postcensal estimates

were used to determine population denominators for women,
estimated population denominators were calculated for MSM
and other men by applying the estimated proportion of men in
the United States who reported ever having male-to-male sex
(6.9%; 95% CI: 5.1%–8.6%) to the 2009 postcensal estimated
population for men (9). Population denominators for other
men were calculated by subtracting the MSM population

TABLE 3. Percentage of men aged 18–64 years who have sex with men, who are at risk for acquiring HIV infection,* and who engaged in selected
HIV-related risk behaviors during the 12 months before the interview — National HIV Behavioral Surveillance System, 21 U.S. cities,† 2008

Characteristic

Unprotected anal sex with a
casual partner§ Received an HIV test

Participated in a behavioral
intervention¶

No. of
participants%

Absolute
difference

(percentage
points)

Relative
difference

% %

Absolute
difference

(percentage
points)

Relative
difference

% %

Absolute
difference

(percentage
points)

Relative
difference

%

Age group (yrs)
18–24 24.2 4.3 21.6 67.5 18.4 37.4 26.2 15.6 147.7 1,997
25–34 27.4 7.5 37.7 66.6 17.5 35.7 17.8 7.2 67.6 2,737
35–44 25.1 5.2 26.1 58.4 9.3 18.9 13.2 2.7 25.0 2,076
45–54 24.4 4.5 22.8 52.1 3.1 6.2 11.2 0.7 6.2 978
55–64 19.9 Ref. Ref. 49.1 Ref. Ref. 10.6 Ref. Ref. 387
Race/Ethnicity

American Indian/Alaska Native 20.5 -4.2 -17.2 63.6 1.2 1.9 20.5 7.5 58.1 44
Asian 20.6 -4.1 -16.6 60.3 -2.1 -3.4 13.1 0.1 1.0 199
Black/African American 23.9 -0.8 -3.1 62.0 -0.4 -0.7 22.8 9.9 76.3 1,938
Hispanic/Latino** 28.0 3.3 13.5 61.7 -0.8 -1.3 20.2 7.3 56.2 2,019
Native Hawaiian/Other Pacific Islander 23.7 -1.0 -3.9 64.4 2.0 3.1 30.5 17.6 135.8 59
White 24.7 Ref. Ref. 62.4 Ref. Ref. 12.9 Ref. Ref. 3,579
Multiple races 28.9 4.2 16.9 62.7 0.2 0.4 19.7 6.8 52.4 284
Other single race 19.1 -5.6 -22.5 66.0 3.5 5.6 29.8 16.9 130.3 47

Place of birth
United States or U.S. territory 24.9 Ref. Ref. 62.7 Ref. Ref. 16.9 Ref. Ref. 6,741
Other 27.5 2.6 10.5 59.8 -2.8 -4.5 20.9 4.1 24.1 1,434

Sexual identity
Heterosexual (straight) 26.3 Ref. Ref. 40.4 Ref. Ref. 17.2 Ref. Ref. 99
Homosexual (gay) 25.4 -0.8 -3.2 64.0 23.6 58.4 17.6 0.4 2.4 6,553
Bisexual 24.9 -1.3 -5.1 55.6 15.2 37.6 17.6 0.4 2.4 1,513

Educational attainment
Less than high school 33.6 10.1 43.2 48.8 -17.7 -26.6 16.8 2.4 16.9 512
High school graduate or equivalent 25.5 2.0 8.6 57.3 -9.2 -13.9 18.5 4.1 28.6 1,868
Some college 25.8 2.4 10.2 63.0 -3.6 -5.4 20.9 6.6 45.7 2,627
College graduate 23.5 Ref. Ref. 66.5 Ref. Ref. 14.4 Ref. Ref. 3,167

Total 25.3 62.2 17.6 8,175

Abbreviations: HIV = human immunodeficiency virus; MSA = metropolitan statistical area; MSM = men who have sex with men; Ref. = referent.
* Participants at risk for acquiring HIV infection were defined as those who reported having never had an HIV test or that their most recent HIV test result was negative,

indeterminate, or unknown. This group includes those who did not know they were HIV positive before the interview but tested positive during the interview.
Analyses were limited to men who reported oral or anal sex with another man during the 12 months before interview and did not report a previous positive HIV
test result or diagnosis.

† Data were collected in the following 21 MSAs; if a metropolitan division is listed, sampling was conducted within that specific division of that MSA: Atlanta-Sandy
Springs-Marietta, Georgia; Baltimore-Towson, Maryland; Boston-Quincy, Massachusetts; Chicago-Naperville-Joliet, Illinois; Dallas-Plano-Irving, Texas; Denver-Aurora-
Broomfield, Colorado; Detroit-Livonia-Dearborn, Michigan; Houston-Sugar Land-Baytown, Texas; Los Angeles-Long Beach-Glendale, California; Miami-Miami
Beach-Kendall, Florida; Nassau-Suffolk, New York; Newark-Union, New Jersey-Pennsylvania; New Orleans-Metairie-Kenner, Louisiana; New York-White Plains-Wayne,
New York-New Jersey; Philadelphia, Pennsylvania; San Diego-Carlsbad-San Marcos, California; San Francisco-San Mateo-Redwood City, California; San Juan-Caguas-
Guaynabo, Puerto Rico; Seattle-Bellevue-Everett, Washington; St. Louis, Missouri-Illinois; and Washington-Arlington-Alexandria, DC-Virginia-Maryland-West Virginia.

§ Unprotected sex was defined as insertive or receptive anal sex without a condom. A casual partner was defined as a man with whom the participant did not feel
committed, whom he did not know very well, or with whom he had sex in exchange for something such as money or drugs.

¶ Includes behavioral interventions received as an individual or as part of a group. An individual intervention was defined as a one-on-one conversation with an
outreach worker, a counselor, or a prevention program worker about ways to protect against HIV infection or other sexually transmitted diseases. This excludes
conversations that took place solely as part of obtaining HIV testing (e.g., pretest or posttest counseling). A group behavioral intervention was defined as a small-
group discussion about ways to protect against HIV or other sexually transmitted diseases.

** Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

118 MMWR / November 22, 2013 / Vol. 62 / No. 3

denominators from the 2009 postcensal estimated population
for men.

The MMP data presented in this report are subject to at least
three limitations. First, MMP estimates are not representative
of all persons with HIV in the United States because only HIV-
infected persons in care during the first 4 months of 2009 were
eligible for selection into the MMP sample. Second, MMP
data might include persons more likely to be retained in care or
adhere to ART, leading to overestimation of certain measures.
For example, measures might be overestimated because persons
in MMP are more engaged in care and adherent to ART use.
Finally, documentation of a recent suppressed viral load might
not indicate persistent viral suppression over time.

The NHBS data presented in this report are subject to at
least two limitations. First, participants in the MSM cycle of
NHBS were recruited from venues, primarily bars and clubs,
within 21 MSAs with a high AIDS prevalence and might not
be representative of MSM who do not attend such venues or
of MSM in other areas. Second, NHBS data regarding risk
behaviors and use of prevention services are self-reported. Social
desirability might lead to underreporting of risk behaviors and
overreporting of recent HIV testing and participation in HIV
behavioral interventions.

Conclusion
The findings in this report highlight a need for continued

expansion of effective HIV prevention efforts for racial/
ethnic minorities and MSM. In 2007, CDC initiated the
Expanded HIV Testing Initiative, Expanded and Integrated
HIV Testing for Populations Disproportionately Affected
by HIV, which was expanded in 2010 to include MSM.
(Additional information is available at http://www.cdc.gov/
hiv/topics/funding/ps10-10138/index.htm.) In addition,
the 2010 national HIV/AIDS strategy has goals that include
reducing HIV incidence, increasing access to care, improving
health outcomes for persons living with HIV, and reducing
HIV-related disparities and health inequities. These goals are
interdependent (4,15) and also consistent with the Healthy
People 2020 goal of achieving health equity, eliminating
disparities, and improving the health of all groups. (Additional
information available at http://healthypeople.gov/2020/
about/default.aspx.) Reducing HIV incidence and improving
individual health outcomes require increased access to care
and elimination of disparities in the quality of care received
(4). CDC is working with health departments throughout
the United States to expand efforts in using local data (in

accordance with privacy and confidentiality policies, laws, and
regulations) to 1) identify HIV-infected persons who are not
receiving care and to facilitate efforts to ensure they receive
appropriate care and 2) identify populations within their
local areas at greatest risk for HIV and with greatest need for
prevention services. CDC will continue using its national HIV
surveillance systems to monitor HIV incidence and diagnosis in
the population and to monitor receipt of ART, risk behaviors,
and receipt of prevention services among HIV-infected persons
in care to identify opportunities for improvement. Information
will be shared with grantees, partners, health-care providers,
and other federal agencies (e.g., the Health Resources and
Services Administration) to improve delivery of care, treatment,
and prevention services for those with HIV infection (4).
Behaviors of populations at high risk for HIV infection also
will be monitored as part of CDC’s comprehensive approach
to reducing the spread of HIV infection in the United States.

To reduce the number of new HIV infections, CDC
has devoted HIV resources to High-Impact Prevention, a
combination of scientifically proven, cost-effective, and scalable
interventions that have demonstrated the potential to reduce
new HIV infections in the relevant populations and geographic
areas to yield a greater reduction in HIV incidence (16).
Optimally scaled implementation of the most cost-effective
interventions will have the greatest impact on reducing the
spread of HIV in the United States.

The progress in HIV prevention since the beginning of the
U.S. epidemic is a result of a multisectoral approach to HIV
from governmental, non governmental, and community-based
organizations, academia, and the business sector. Reducing the
higher prevalence of HIV infection in racial/ethnic minority
groups and MSM also will require public health interventions
and societal actions as a whole that address social, economic,
health system, and other environmental factors that play a
role in HIV prevalence in these communities (17). These
factors might include poverty, which can limit access to health
care and HIV testing; stigma and discrimination, which can
discourage individuals from seeking testing, prevention, and
treatment services; barriers to timely access and use of medical
and social services; and higher rates of incarceration, which
can disrupt social and safe sexual networks. The results of this
report underscore the need for high-priority, carefully targeted
HIV prevention efforts in these communities to ensure that
individual, social, health system, and other environmental
determinants of health are considered in the design and
implementation of HIV prevention and care programs.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 119

References
1. CDC. Monitoring the national HIV indicators through surveillance of

HIV infection in the United States and 6 U.S. dependent areas—2010.
HIV Surveillance Supplemental Report 2012;17(No. 3).

2. Prejean J, Song R, Hernandez A, et al. Estimated HIV incidence in the United
States, 2006–2009. PLoS ONE 2011;6:e17502. Epub August 3, 2011. Available
at http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.
pone.0017502.

3. CDC. HIV surveillance report, 2010. Vol. 22. Atlanta, GA: CDC; 2010.
Available at http://www.cdc.gov/hiv/topics/surveillance/resources/reports.

4. CDC. Vital signs: HIV prevention through care and treatment—United
States. MMWR 2011;60:1618–23.

5. CDC. HIV infection—United States, 2005 and 2008. In: CDC health
disparities and inequalities report—United States, 2010. MMWR 2011;
60(Suppl; January 14, 2011).

6. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

7. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

8. US Census Bureau. Population estimates [entire data set]. Washington,
DC: US Census Bureau; 2010. Available at http://www.census.gov/
popest/estbygeo.html.

9. Purcell DW, Johnson CH, Lansky A, et al. Estimating the population
size of men who have sex with men in the United States to obtain HIV
and syphilis rates. Open AIDS J 2012;6:98–107.

10. McNaghten AD, Wolfe MI, Onorato I, et al. Improving the
representativeness of behavioral and clinical surveillance for persons with
HIV in the United States: the rationale for developing a population-based
approach. Epub June 20, 2007. PLoS ONE 2007;2:e550 10.1371/
journal.pone.0000550.

11. CDC. Clinical and behavioral characteristics of adults receiving medical
care for HIV infection: Medical Monitoring Project, United States, 2007.
MMWR 2011;60(No. SS-11).

12. Frankel MR, McNaghten A, Shapiro MF, et al. A probability sample for
monitoring the HIV-infected population in care in the U.S. and in
selected states. Open AIDS J 2012;6:67–76.

13. MacKellar DA, Gallagher KM, Finlayson T, Sanchez T, Lansky A,
Sullivan PS. Surveillance of HIV risk and prevention behaviors of men
who have sex with men—a national application of venue-based, time-
space sampling. Public Health Rep 2007;122(Supp 1):39–47.

14. Keppel K, Pamuk E, Lynch J, et al. Methodological issues in measuring
health disparities. Vital Health Stat 2 2005;141:1–16.

15. White House Office of National AIDS Policy. National HIV/AIDS
strategy for the United States. Washington, DC: White House Office
of National AIDS Policy; 2010. Available at http://www.aids.gov/federal-
resources/policies/national-hiv-aids-strategy/nhas.pdf.

16. CDC. High-impact HIV prevention. Atlanta, GA: US Department of
Health and Human Services; 2011. Available at http://www.cdc.gov/
hiv/strategy/hihp.

17. CDC. Establishing a holistic framework to reduce inequities in HIV,
viral hepatitis, STDs, and tuberculosis in the United States. Atlanta,
GA: US Department of Health and Human Services, CDC; 2010.
Available at http://www.cdc.gov/socialdeterminants/docs/SDH-White-
Paper-2010.pdf.

Supplement

120 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
Obesity is a major public health problem affecting adults

and children in the United States. Since 1960, the prevalence
of adult obesity in the United States has nearly tripled, from
13% in 1960–1962 to 36% during 2009–2010 (1,2). Since
1970, the prevalence of obesity has more than tripled among
children, from 5% in 1971–1974 (3) to 17% in 2009–2010
(4,5). Although the prevalence of obesity is high among all
U.S. population groups, substantial disparities exist among
racial/ethnic minorities and vary on the basis of age, sex, and
socioeconomic status.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (6) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the
2013 CHDIR Introduction (7). This report provides more
current information regarding what was presented in the 2011
CHDIR (8). The purposes of this report are to discuss and raise
awareness of differences in the characteristics of persons who
are obese and to prompt actions to reduce these disparities.

Methods
To assess disparities and trends over time in obesity prevalence

among adults aged ≥18 years and children and adolescents aged
2–17 years, CDC analyzed data from the National Health and
Nutrition Examination Survey (NHANES) between 1999 and
2008 that were included in the 2011 CDC Health Disparities
and Inequities Report (CHDIR) (8) and data from NHANES
for 2009–2010. To assess disparities and trends over time in
obesity prevalence among adults aged ≥18 years and children
and adolescents aged 2–17 years, CDC analyzed data from
the National Health and Nutrition Examination Survey
(NHANES) between 1999 and 2008 that were included in the
2011 CDC Health Disparities and Inequities Report (CHDIR)
(8) and data from NHANES for 2009–2010. CDC examined
obesity prevalence by sex, age, and race/ethnicity and by the
following variables that were not included in the previous

report: educational attainment, disability status, country of
birth, and language spoken at home. Geographic location
was not examined because this information was not available
in the publicly available datasets, and educational attainment
was analyzed rather than family income because a smaller
number of participants had missing data for educational
attainment than for income. In addition, for many persons,
income was categorized into very broad ranges (e.g., <$20,000
and ≥$20,000). The highest income category was ≥$75,000.

NHANES is a complex, multistage probability sample of
the noninstitutionalized population of the United States.
Information regarding the survey’s methodology has been
published previously (9). Data for NHANES 2-year samples
were collected from 1999–2000 through 2009–2010 (10)
using a stratified, multistage cluster design. The sample
was representative of the U.S. civilian, noninstitutionalized
population. Weight and height were measured using
standardized techniques and equipment, and body mass index
(BMI) (weight [kg]/height [m]2) was calculated (11). Persons
aged ≥20 years were classified as obese if they had a BMI
≥30 kg/m2 (12). Persons aged 2–17 years were considered
obese if they had a BMI ≥95th (sex- and age-specific) percentile
of the 2000 CDC growth charts (13). Persons aged 18–19
years were classified as obese if they had a BMI ≥30 kg/m2
or ≥95th percentile of the CDC growth charts. This age
classification differs from that used in other studies of obesity
using NHANES data (2,5), which grouped persons aged
18–19 years with children and adolescents. Information on
race/ethnicity was self-reported for persons aged ≥16 years.
For persons aged <16 years, race/ethnicity was reported by a
family member. Respondents reported race/ethnicity from a
list provided to them that included an open-ended response.
Analyses that focused on race/ethnicity were restricted to
non-Hispanic whites, non-Hispanic blacks, and Mexican-
Americans because of insufficient numbers of persons in other
racial/ethnic groups. However, all race/ethnicity groups were
included in analyses of other characteristics (e.g., educational
attainment). Non-Hispanic blacks and Mexican-Americans
were oversampled to improve the precision of estimates for
these race/ethnicity groups (14).

Obesity — United States, 1999–2010
Ashleigh L. May, PhD
David Freedman, PhD
Bettylou Sherry, PhD

Heidi M. Blanck, PhD
Division of Nutrition, Physical Activity, and Obesity, National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Ashleigh L. May, Division of Nutrition, Physical Activity, and Obesity, National Center for Chronic Disease Prevention and Health
Promotion, CDC. Telephone: 770-488-8062. E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 121

During each household interview, adult respondents were
asked to report the highest level of school completed or
the highest degree that they had received. For children and
adolescents, this information was collected for the adult head
of household. This information was summarized into four
categories of education attainment: less than high school, high
school graduate or equivalent, some college, and college degree
or higher. Because adults aged 18–22 years were unlikely to
have completed college, analyses of educational attainment
among adults were restricted to subjects who were aged ≥23
years. For approximately 2% of the sample, information on
education attainment was missing, and these persons were
excluded from the analyses that focused on this characteristic.

Disability status was determined by self-reports among
persons aged ≥60 years and was based on responses to 11
questions concerning problems in memory (confusion) and
hearing, along with ambulatory difficulties (e.g., difficulties in
walking, going up steps, and standing) and self-care difficulties
(e.g., dressing, eating, and getting out of bed). A response
of “some difficulty” or “much difficulty” was considered
to indicate that the activity was difficult. The number of
positive responses was then summed, and this variable was
categorized into three groups: no difficulties or problems
(reported by 50% of adults aged ≥60 years), difficulties in
one to three activities (33%), and difficulties in four or more
activities (17%). Although this type of classification is based
on various assumptions, it allows for the assessment of whether
a dose-response relationship is evident. Standard disability
classifications also include vision problems (15), but these data
were not available for NHANES 2009–2010.

Information also was collected on country of birth and,
on the basis of a question in the acculturation data file, the
language usually spoken at home. Because few non-Hispanic
white or non-Hispanic black subjects were born outside the
United States or spoke a language other than English at home,
analyses of country of birth and language spoken at home were
restricted to Mexican-Americans. Of the Mexican-American
adults for whom information was collected, 57% reported that
they were born in Mexico, and 56% reported that they usually
spoke Spanish at home.

Trends in obesity prevalence over the 2-year study cycles were
examined, with year coded as a six-level interval variable. To
decrease the variability of the estimates of the prevalence of
obesity within categories of the examined characteristics, this
report presents estimates for three 4-year periods: 1999–2002,
2003–2006, and 2007–2010. Within these three 4-year
periods, the number of children aged 2–17 years ranged from
6,081 to 7,293, and the number of adults ranged from 9,630 to
12,067. All estimates of the prevalence of obesity among adults
have been age adjusted to the 2000 Standard U.S. Population.

The estimated prevalence is considered to be unstable if the
relative SE (SE ÷ prevalence) is ≥30%. Estimates that have a
relative SE ≥40% are not presented.

The overall (12-year) prevalence of obesity was examined
across categories of the various characteristics (race/ethnicity,
educational attainment, number of disabilities, country
of birth, and language spoken at home). Separate analyses
were performed by age and sex (i.e., for men, women, boys,
and girls). Differences in obesity prevalence across these
characteristics were examined, using an interaction term in
logistic regression models. Educational attainment (four levels),
number of disabilities (three levels), and 2-year study cycle
(six levels) were coded as ordinal variables in these models. All
analyses accounted for the examination sampling weights and
for the complex sampling design. All estimate comparisons
represent absolute differences. Statistical significance (p<0.05)
was assessed in logistic regression models, with various models
including age and study period as covariates. All analyses were
performed with the survey package in R (16,17).

Results
Between 1999–2002 and 2007–2010, the age-adjusted

prevalence of obesity among adults aged ≥18 years increased
from 26.5% to 33.0% among men and from 32.4% to 34.9%
among women (Table 1). Controlling for age and race/
ethnicity in regression models indicated that the increase in
the prevalence of obesity over the study period was statistically
significant among men but not among women.

The prevalence of obesity differed substantially across
categories of various demographic characteristics (Table 1).
Among men, there was little difference in the prevalence
of obesity by race/ethnicity, but among women, the overall
(1999–2010) prevalence among non-Hispanic blacks (51%)
was 10 percentage points higher than that among Mexican-
Americans and 20 percentage points higher than that among
non-Hispanic white women.

Inverse associations were identified between the prevalence
of obesity and educational attainment that were statistically
significant among both men and women; differences were much
greater among women (Table 1). These associations appeared
to be nonlinear. For example, among men, the prevalence was
lowest (25%) among college graduates but highest (35%) among
those who had completed some college. Among women, the
overall prevalence of obesity among those who had completed
college was 13–16 percentage points lower than in other groups,
but there was little difference in obesity prevalence between those
who had not finished high school and those who had completed
some college. The analysis of disability status of adults aged ≥60

Supplement

122 MMWR / November 22, 2013 / Vol. 62 / No. 3

years indicated that the overall prevalence of obesity among those
who reported having difficulties with four or more activities
was substantially higher than obesity prevalence among those
without a disability (men: 16 percentage points higher; women:
27 percentage points higher).

In contrast to these differences, which were larger among
women, the association of obesity with country of birth and
language spoken at home was stronger among men (Table
1). Mexican-American men who were born in the United
States had 13 percentage points higher overall prevalence of
obesity than men born in Mexico (39% versus. 26%), but
the equivalent difference among Mexican-American women
was only 3 percentage points. Similarly, Mexican-American
men who spoke mostly English at home had a 12 percentage
points higher overall prevalence of obesity compared with those
who spoke mostly Spanish at home (38% versus 26%), while
there was no significant difference among Mexican-American
women. As assessed by an interaction term (each characteristic
x study period) in sex-specific regression models, there was no

indication that disparities in obesity prevalence varied across
the 12-year study period among either men or women.

Between 1999–2002 and 2007–2010, the prevalence of
obesity among children and adolescents aged 2–17 years
increased from 15.4% to 18.6% among boys and from 13.8%
to 15.1% among girls (Table 2). After adjustment for age and
race/ethnicity in regression models, the increase over the six
2-year study cycles was statistically significant among boys but
not among girls.

Differences in the prevalence of obesity among children and
adolescents over the 12-year study period across categories
of the various characteristics were somewhat similar to those
among adults (Tables 1 and 2). Substantial differences existed
in the prevalence of obesity by race/ethnicity; among boys,
prevalence was highest among Mexican-Americans (24%),
whereas among girls, prevalence was highest among non-
Hispanic blacks (22%). Educational attainment of the adult
head of household was associated inversely with obesity among
both boys and girls. Overall, the prevalence of obesity among

TABLE 1. Prevalence of obesity* among adults aged ≥18 years, by selected characteristics — National Health and Nutrition Examination Survey,
United States, 1999–2010

Characteristic

Prevalence in males Prevalence in females

Total 1999–2002 2003–2006 2007–2010 Total 1999–2002 2003–2006 2007–2010

% (SE) % (SE) % (SE) % (SE) % (SE) % (SE) % (SE) % (SE)

Total† 30 (1) 26 (1) 31 (1) 33 (1) 34 (1) 32 (1) 33 (1) 35 (1)
Race/Ethnicity

White, non-Hispanic 31 (1) 27 (1) 31 (1) 33 (1) 31§ (1) 30 (1) 31 (1) 32 (1)
Black, non-Hispanic 33 (1) 27 (1) 35 (2) 37 (2) 51§ (1) 47 (2) 53 (2) 53 (2)
Mexican-American 31 (1) 26 (2) 29 (2) 35 (2) 41§ (1) 37 (2) 41 (2) 44 (2)

Educational attainment¶
Less than high school 29** (1) 26 (2) 29 (2) 32 (2) 40§ (1) 39 (2) 40 (2) 41 (2)
High school graduate or

equivalent
33** (1) 30 (2) 35 (2) 35 (2) 38§ (1) 36 (2) 38 (2) 41 (2)

Some college 35** (1) 28 (2) 35 (2) 41 (2) 37§ (1) 35 (2) 36 (2) 38 (1)
College graduate 25** (1) 23 (2) 26 (2) 26 (2) 24§ (1) 22 (2) 24 (2) 27 (2)

No. of disabilities††
0 31§ (1) 30 (2) 29 (4) 34 (2) 30§ (1) 31 (3) 26 (2) 32 (2)
1–3 41§ (2) 35 (4) 40 (5) 46 (3) 45§ (3) 40 (4) 42 (5) 53 (4)
4–11 47§ (3) 50 (5) 48 (8) 44 (5) 57§ (3) 53 (4) 59 (5) 60 (4)
Country of birth§§

United States 39§ (2) 37 (4) 35 (3) 44 (3) 43** (1) 38 (4) 45 (2) 44 (2)
Mexico 26§ (1) 19 (1) 25 (2) 31 (2) 40** (1) 36 (3) 39 (3) 43 (2)

Language spoken at home§§
English 38§ (2) 36 (4) 34 (3) 41 (3) 41 (2) 35 (3) 42 (2) 44 (2)
Spanish 26§ (1) 20 (1) 25 (2) 32 (2) 41 (1) 37 (3) 41 (3) 43 (2)

Abbreviation: SE = standard error.
* All estimates have been age adjusted to the 2000 Standard U.S. Population.
† The increase in obesity prevalence over the 12-year study period was statistically significant among men (p<0.001) but not among women (p = 0.09).
§ p<0.001. P-values assess whether the overall prevalence of obesity differed across categories of each characteristic (i.e., a main effect). These p-values were calculated

from sex-specific regression models that included year of study (a six-level ordinal variable) and age as covariates. Educational attainment and the number of
disabilities were coded as ordinal variables in these models.

¶ Asked of persons aged ≥23 years.
** p<0.05.
†† Asked of persons aged ≥60 years. Disabilities were classified on the basis of responses to 11 questions concerning having memory and hearing problems and some

or much difficulty in walking, carrying, preparing meals, standing, getting out of bed, eating, dressing, and going out.
§§ Asked of Mexican-Americans.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 123

children and adolescents whose adult head of household had
completed college was approximately half that of prevalence
among children whose adult head of household did not
complete high school. In contrast to the differences among
adults, the prevalence of obesity among Mexican-American
children did not differ significantly according to either country
of birth or language spoken at home.

As assessed by an interaction term (each characteristic x study
period, which was coded as 1–6) in sex-specific regression
models, there was little indication that differences in the
prevalence of obesity across most of the characteristics analyzed
varied over the 12-year study period among children (Table 2).
However, the prevalence of obesity among girls whose adult head
of household had not finished high school increased (17% for
1999–2002 versus 23% for 2007–2010) while the prevalence
decreased among girls whose adult head of household had
completed college (11% for 1999–2002 versus 7% for 2007–
2010). There was not a comparable interaction among boys.
Because education attainment differs substantially across race/
ethnicity groups, the associations between education attainment
and obesity prevalence were examined.

The relation of educational attainment to obesity varied
significantly by sex and race/ethnicity among both adults

(Figure 1) and children (Figure 2). Among non-Hispanic
white women (Figure 1), in each 4-year period, the prevalence
of obesity was approximately 15% lower among those who
had completed college than it was among those who had not
completed high school. Although the prevalence of obesity
among non-Hispanic white men and non-Hispanic black
women was also lowest among those who had completed college,
the trend over the four educational attainment categories was
not consistent in these two groups. Furthermore, there was
no evidence that educational attainment was associated with
obesity among non-Hispanic black men or among Mexican-
Americans. For example, Mexican-American men who did not
complete high school had the lowest prevalence of obesity in
1999–2002 and in 2003–2006.

Associations between obesity and adult head of household
education attainment among children and adolescents were also
less consistent after stratifying for race/ethnicity (Figure 2). (The
prevalence of obesity among Mexican-American children and
adolescents is not shown because many of the estimates were
unstable). Although the lowest prevalence of obesity among non-
Hispanic white children and adolescents was observed among
those whose adult head of household had completed college,
this was not the case among non-Hispanic black children.

TABLE 2. Prevalence of obesity among children and adolescents aged 2–17 years, by selected characteristics — National Health and Nutrition
Examination Survey, United States, 1999–2010

Characteristic

Prevalence in males Prevalence in females

Total 1999–2002 2003–2006 2007–2010 Total 1999–2002 2003–2006 2007–2010

% (SE) % (SE) % (SE) % (SE) % (SE) % (SE) % (SE) % (SE)

Total* 17 (1) 15 (1) 17 (1) 19 (1) 15 (1) 14 (1) 15 (1) 15 (1)
Race/Ethnicity

White, non-Hispanic 15† (1) 13 (1) 15 (1) 16 (1) 13† (1) 11 (1) 14 (2) 13 (1)
Black, non-Hispanic 18† (1) 16 (1) 17 (1) 21 (2) 22† (1) 20 (1) 23 (1) 23 (2)
Mexican-American 24† (1) 24 (1) 24 (2) 25 (2) 18† (1) 17 (1) 19 (1) 18 (2)

Educational attainment§
Less than high school 21† (1) 21 (2) 19 (2) 24 (2) 19† (1) 17 (1) 18 (2) 23 (2)
High school graduate or

equivalent
18† (1) 15 (2) 19 (2) 19 (1) 18† (1) 15 (2) 20 (2) 19 (2)

Some college 17† (1) 17 (2) 16 (2) 19 (2) 14† (1) 12 (1) 14 (2) 14 (2)
College graduate 11† (1) 11 (2) 12 (2) 12 (2) 9† (1) 11 (2) 10 (2) 7 (1)

Country of birth¶
United States 24 (1) 24 (1) 24 (2) 24 (2) 18 (1) 18 (2) 19 (2) 18 (2)
Mexico 24 (2) 24 (3) 22 (3) 27 (4) 15 (2) 13 (3) 16 (3) 17 (5)

Language spoken at home¶
English 26 (2) 27 (3) 24 (3) 27 (4) 22 (2) 23 (3) 23 (2) 20 (4)
Spanish 18 (4) 26 (3) 19 (3) 41 (11) 16 (3) 21 (3) 16 (4) —** —

Abbreviation: SE = standard error.
* The increase in obesity prevalence over the 12-year study period was statistically significant among boys (p<0.01) but not girls (p = 0.34).
† p<0.001. P-values assess whether the overall prevalences of obesity differed across categories of each characteristic (i.e., a main effect). These p-values were

calculated from sex-specific logistic regression models that included year of study (an ordinal variable with six levels) and age as covariates. Adult head of household
educational attainment was coded as an ordinal variable in these models; other characteristics were considered to be categorical.

§ Asked of adult head of household. As assessed by an interaction term in logistic regression models, disparities in obesity prevalence across educational attainment
of the head of household increased over the study period among girls (p = 0.01).

¶ Asked of Mexican-Americans.
** Not shown because SE was ≥40% of the estimated prevalence. During 2007–2010, data were available only for 50 Mexican-American girls whose families usually

spoke Spanish at home.

Supplement

124 MMWR / November 22, 2013 / Vol. 62 / No. 3

Furthermore, there was no evidence of any association between
obesity and adult head of household educational attainment
among non-Hispanic black boys, and the trend in the prevalence
of obesity across the lower three categories of adult head of
household educational attainment was not consistent in any
of the sex-race groups. The observed interaction between study
period and educational attainment among girls (Table 2) was

largely attributable to the trend among non-Hispanic white girls,
and the 2007–2010 prevalence of obesity among non-Hispanic
white girls varied from 6% (± 2) to 28% (± 5) across the four
groups of educational attainment by the adult head of household
more than it did in previous years.

FIGURE 1. Prevalence of obesity among adults aged ≥23 years,* by sex, race/ethnicity, and educational attainment — National Health and
Nutrition Examination Survey, United States, 1999–2010

* Adults aged 18–22 years were unlikely to have completed their education and were excluded from this analysis.

20

30

40

50

60

20

30

40

50

60

1999–2002 2003–2006 2007–2010 1999–2002 2003–2006 2007–2010 1999–2002 2003–2006 2007–2010

Pr
ev

al
en

ce
(%

)

Non-Hispanic white Non-Hispanic black Mexican-American

M
ale

Fem
ale

Educational attainment

Less than high school

High school graduate
or equivalent

Some college

College graduate

Survey years

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 125

Discussion
Recent trends suggest that although increases in obesity

prevalence have slowed or even stopped for some subgroups,
the prevalence remains high (2,5). This report highlights
the persistence of substantial disparities among certain
population groups, all of which further complicate the efforts

to understand, control, and prevent obesity. Although the
specific causes of these disparities have not been identified,
it is likely that they are associated with complex social and
cultural factors that affect obesity-related behaviors. One
possible contributing factor is that rates of breastfeeding
are lower among non-Hispanic black women compared
with non-Hispanic white women (18). In addition, greater

FIGURE 2. Prevalence of obesity among children and adolescents aged 2–17 years, by sex, race/ethnicity, and educational attainment of adult
head of household — National Health and Nutrition Examination Survey, United States 1999–2010*

* Prevalences are not shown for Mexican-American children because many of the estimates were unstable (standard error [SE] >30% of the prevalence) with the
relative SE reaching a maximum of 49%. An asterisk in the figure indicates that the relative SE is between 30% and 40% of the prevalence; this was seen during
2003–2006 among white non-Hispanic boys and girls from a household in which the adult head did not complete high school.

*

*

Educational attainment

Less than high school

High school graduate
or equivalent

Some college

College graduate

5

10

15

20

25

10

15

20

25

1999–2002 2003–2006 2007–2010 1999–2002 2003–2006 2007–2010

Survey years

Pr
ev

al
en

ce
(%

)

Non-Hispanic white Non-Hispanic black

M
ale

Fem
ale

5

Supplement

126 MMWR / November 22, 2013 / Vol. 62 / No. 3

satisfaction in body size among racial/ethnic minority women
(19), preferences for larger body types (20,21), and previous
threats of, or experiences with, undernutrition (22) also
might promote obesogenic behaviors. Further, racial/ethnic
differences in physical activity levels among adults (23) and
children (24) and differential preferences for specific types of
physical activity (25) also might play a role. These and other
behaviors occur within a broader context of obesity-promoting
environments that limit opportunities for physical activity,
encourage excess television viewing and passive screen time,
and provide easy access to high-calorie, low-nutrient foods
and beverages, including those high in added sugars and solid
fats (26,27).

Limitations
The findings presented in this report are subject to at least

two limitations. First, NHANES does not sample an adequate
number of persons who are members of racial/ethnic minority
communities other than non-Hispanic blacks and Mexican-
Americans to permit estimating obesity prevalence in these
communities; however, previous research has reported high
prevalence levels among American Indians/Alaska Natives
(28,29). Second, the data presented, although age adjusted,
do not allow for assessment of covarying issues or stratification
to further assess independent effects.

Conclusion
The data provided in this report can be used to help identify

high-priority groups (e.g., those with low levels of educational
attainment, Mexican-American boys, and non-Hispanic black
girls and women) for intervention. Because high-priority
groups frequently are defined by nonmodifiable characteristics
(e.g., race/ethnicity and sex), designers of effective interventions
should consider which dietary or physical activity behaviors
contribute to the differences as well as how those behaviors are
influenced by social and cultural factors and by the settings in
which persons spend their time. For example, because studies
have found that access to healthy foods is more limited in
low-income communities and communities of color than
in other communities (30), interventions could focus on
neighborhood walkability (e.g., sidewalks), Complete Streets
(31), and community design in these communities. Further,
opportunities to reduce disparities related to nutrition are also
present. For example, many low-income household members,
who also tend to have low educational attainment, consume
higher amounts of sugary drinks and fewer fruits and vegetables

than persons in higher income households (32,33). Effective
interventions that focus on increasing access to healthy food
outlets, initiatives for local businesses to provide healthier
foods and beverages such as fruits and vegetables (e.g., Healthy
Food Financing Initiative [34]), and education combined
with vouchers for low-income families (e.g., the Special
Supplemental Nutrition Program for Women, Infants, and
Children [WIC; 35] participants) that can use to purchase
healthy foods might help reduce this disparity.

Having a sustainable impact on reducing disparities
associated with obesity includes making healthy choices easily
assessable and available to all persons. Environmental strategies
that support healthy eating and active living opportunities
within communities can help provide healthy choices for
persons. In addition, such changes can help provide ongoing
training and support of public health practitioners with tools
to implement effective responses to obesity in populations
that are facing health disparities (36,37). CDC provides
funding and support to multiple public health programs to
improve access to healthy foods and beverages in underserved
communities (38,39), including increased access to markets
and convenience stores that offer healthier food and beverage
choices; expanding programs that promote food affordability
such as WIC farmers’ markets; assisting persons through green
carts and mobile vans in inner-city neighborhoods (Farm-
to-Where-You-Are) (40,41); and promotion of food policy
councils that include diverse stakeholders that often consider
both food security and improvements of the food environment
at the state and local levels.

Certain early child care education initiatives promote
active play and healthier beverage and food offerings such
as drinking water and fruits and vegetables. These initiatives
can address disparities by providing age-appropriate health
curricula, parental outreach, increased healthier foods and
beverages served, and training and technical support for staff on
menu planning and food preparation (42) for children of low
socioeconomic status and children who hold immigrant and/
or refugee status, among other high-priority groups (Adrienne
Dorf, Child Care Health Program Public Health Seattle and
King County, personal communication, 2012). Strategies such
as promoting physical activity early in child care and school,
increasing low- or no-cost physical activity opportunities,
building and enhancing trails and parks, developing shared-
use agreements with public venues such as schools, improving
sidewalks, and other initiatives that promote physical activity
to prevent and reduce obesity have been implemented to help
all persons and communities to become physically active (43).

Although the rate of obesity has plateaued in recent years
for some groups, the overall prevalence of the condition

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 127

remains high for all U.S. residents, and disparities persist in
the prevalence of obesity. Continued monitoring of obesity
prevalence and further research are needed to identify and
understand the factors that influence individual behaviors,
especially among high-priority groups, and to augment current
population-based approaches with interventions that are
tailored to their needs.

References
1. Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL. Overweight and

obesity in the United States: prevalence and trends, 1960–1994. Int J
Obes Relat Metab Disord 1998;22:39–49.

2. Flegal KM, Carroll MD, Kit BK, Ogden CL. Prevalence of obesity and
trends in the distribution of body mass index among US adults, 1999-
2010. JAMA 2012;307:491–7.

3. Ogden C, Carroll M. Prevalence of obesity among children and
adolescents: United States, trends 1963–1965 through 2007–2008.
Atlanta, GA: US Department of Health and Human Services, CDC,
National Center for Health Statistics; 2010.

4. Ogden CL, Carroll MD, Curtin LR, Lamb MM, Flegal KM. Prevalence
of high body mass index in US children and adolescents, 2007–2008.
JAMA 2010;303:242–9.

5. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and
trends in body mass index among US children and adolescents, 1999–
2010. JAMA 2012;307:483–90.

6. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

7. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(Suppl 3).

8. CDC. Obesity—United States, 1988–2008. In: CDC health disparities
and inequalities report—United States, 2011. MMWR 2011;60(Suppl;
January 14, 2011).

9. CDC. National Health and Nutrition Examination survey data.
Hyattsville, MD: US Department of Health and Human Services, CDC;
2010. Available at http://www.cdc.gov/nchs/nhanes.htm.

10. CDC. National Health and Nutrition Examination Sur vey:
questionnaires, datasets, and related documentation. Hyattsville, MD:
US Department of Health and Human Services, CDC, National Center
for Health Statistics; 2012. Available at http://www.cdc.gov/nchs/
nhanes/nhanes_questionnaires.htm.

11. CDC. National Health and Nutrition Examination Survey: anthropometry
procedures manual. Hyattsville, MD: US Department of Health and
Human Services, CDC, National Center for Health Statistics; 2009.

12. Expert Panel on the Identification, Evaluation, and Treatment of
Overweight in Adults. Clinical guidelines on the identification,
evaluation, and treatment of overweight and obesity in adults: executive
summary. Am J Clin Nutr 1998;68:899–917.

13. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC growth charts for the
United States: methods and development. Vital Health Stat 11 2002;(246).

14. CDC. Analytic note regarding 2007–2010 survey design changes and
combining data across other survey cycles. Available at http://www.cdc.
gov/nchs/data/nhanes/analyticnote_2007–2010.pdf.

15. US Department of Health and Human Services. Implementation
guidance on data collection standards for race, ethnicity, sex, primary
language, and disability status. Available at http://aspe.hhs.gov/datacncl/
standards/ACA/4302/index.shtml.

16. Lumley T. Complex surveys: a guide to analysis using R. Hoboken, NJ:
Wiley; 2010.

17. R Development Core Team. R: a language and environment for statistical
computing (version 2.15.0). Vienna, Austria: R Foundation for Statistical
Computing; 2012. Available at http://www.r-project.org.

18. CDC. Racial and ethnic differences in breastfeeding initiation and
duration, by state—National Immunization Survey, United States,
2004–2008. MMWR 2010;59:327–34.

19. Millstein RA, Carlson SA, Fulton JE, et al. Relationships between body
size satisfaction and weight control practices among US adults. Medscape
J Med 2008;10:119.

20. Kumanyika SK. Environmental influences on childhood obesity: ethnic
and cultural influences in context. Physiol Behav 2008;94:61–70.

21. Sherry B, Scanlon K, Barden E, Kallio J. Do race and ethnicity influence
parents’ feeding strategies, perceptions of and concerns about child weight
and intervention techniques? In: Birch L, Dietz W, eds. Eating behaviors
of the young child: prenatal and postnatal influences on healthy eating. Elk
Grove Village, IL: American Academy of Pediatrics; 2008:157–79.

22. Mintz S. Tasting food, tasting freedom: excursions into eating, power,
and the past. Boston, MA: Beacon Press; 1997.

23. CDC. Adult participation in aerobic and muscle-strengthening physical
activities—United States, 2011. MMWR 2013;62:326–330.

24. Whitt-Glover MC, Taylor WC, Floyd MF, Yore MM, Yancey AK,
Matthews CE. Disparities in physical activity and sedentary behaviors
among US children and adolescents: prevalence, correlates and
intervention implications. J Public Health Policy 2009;30(Suppl):309–34.

25. Saint Onge JM, Krueger PM. Education and racial-ethnic differences in types
of exercise in the United States. J Health Soc Behavior, 2011;52:197–211.

26. French SA, Story M, Jeffery RW. Environmental influences on eating
and physical activity. Annu Rev Public Health 2001;22:309–35.

27. Candib LM. Obesity and diabetes in vulnerable populations: reflection
on proximal and distal causes. Ann Fam Med 2007;5:547–56.

28. Anderson SE, Whitaker RC. Prevalence of obesity among US preschool
children in different racial and ethnic groups. Arch Pediatr Adolesc Med
2009;163:344–8.

29. Redwood DG, Lanier AP, Johnston JM, Asay ED, Slattery ML. Chronic
disease risk factors among Alaska Native and American Indian people,
Alaska, 2004–2006. Prev Chronic Dis 2010;7:A85. Available at http://
www.cdc.gov/pcd/issues/2011/nov/10_0268.htm.

30. PolicyLink and The Food Trust. The grocery gap: who has access to
health food and why it matters. Philadelphia, PA: The Food Trust; 2010.
Available at http://thefoodtrust.org/uploads/media_items/grocerygap.
original.pdf.

31. National Complete Streets Coalition. Benefits of Complete Streets:
complete streets promote good health 2010. Available at http://www.
smartgrowthamerica.org/documents/cs/factsheets/cs-health.pdf.

32. Dubowitz T, Heron M, Bird CE, et al. Neighborhood socioeconomic
status and fruit and vegetable intake among whites, blacks, and Mexican
Americans in the United States. Am J Clin Nutr2008;87:1883–91.

33. Han E, Powell LM. Consumption patterns of sugar-sweetened beverages
in the United States. J Acad Nutr Diet 2013;113:43–53.

34. The Food Trust. Healthy Food Financing Initiative: an innovative
approach to improve health and spark economic development.
Philadelphia, PA: The Food Trust; 2012. Available at http://thefoodtrust.
org/uploads/media_items/hffi-one-pager.original.pdf.

35. U.S. Department of Agriculture Food and Nutrition Service. WIC Farmers’
Market Nutrition Program. Available at http://www.fns.usda.gov/fmnp.

36. Glickman D, Parker L, Sim LJ, Del Valle Cook H, Miller EA, eds.
Accelerating progress in obesity prevention: solving the weight of the
nation. Washington, D.C.: The National Academies Press; 2012.

37. CDC. CDC health equity resource toolkit for state practitioners:
addressing obesity disparities toolkit. Atlanta, GA: US Department of
Health and Human Services, CDC; 2012. Available at http://www.cdc.
gov/obesity/stateprograms/resources.html.

38. CDC. CDC’s state-based nutrition and physical activity program to
prevent obesity and other chronic diseases. Atlanta, GA: US Department
of Health and Human Services, CDC; 2010. Available at http://www.
cdc.gov/obesity/stateprograms/index.html.

39. CDC. Recommended community strategies and measurements to
prevent obesity in the United States. MMWR 2009;58(No. RR-7).

Supplement

128 MMWR / November 22, 2013 / Vol. 62 / No. 3

40. CDC. DNPAO state program highlights: farm-to-where-you-are programs.
Atlanta, GA: US Department of Health and Human Services, CDC; 2012.
Available at http://www.cdc.gov/obesity/downloads/farmtowhereyouare.pdf.

41. CDC. Strategies to prevent obesity and other chronic diseases: the CDC
guide to strategies to increase the consumption of fruits and vegetables.
Atlanta, GA: US Department of Health and Human Services; 2011.
Available at http://www.cdc.gov/obesity/downloads/fandv_2011_web_
tag508.pdf.

42. CDC. Spectrum of opportunities for obesity prevention in the early care
and education setting. Atlanta, GA: US Department of Health and
Human Services, CDC; 2012. Available at http://www.cdc.gov/obesity/
downloads/Spectrum-of-Opportunities-for-Obesity-Prevention-in-
Early-Care-and-Education-Setting_TAbriefing.pdf.

43. CDC. State program highlights: urban design and transportation policies
and practices. Atlanta, GA; US Department of Health and Human
Services, CDC; 2011. Available at http://www.cdc.gov/obesity/
stateprograms/highlights.html.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 129

Introduction
Periodontal disease, or gum disease, is a chronic infection

of the hard and soft tissue supporting the teeth (1) and is a
leading cause of tooth loss in older adults (2). Tooth loss impairs
dental function and quality of life in older adults (2). The
chronic infections associated with periodontitis can increase
the risk for aspiration pneumonia in older adults and has
been implicated in the pathogenesis of chronic inflammation
that impairs general health (3,4). The severity of periodontal
disease can be categorized as mild, moderate, or severe on the
basis of multiple measurements of periodontal pocket depth,
attachment loss, and gingival inflammation around teeth (5).

At the national level, monitoring the reduction of moderate
and severe periodontitis in the adult U.S. population is part
of the health-promotion and disease-prevention activities of
Healthy People 2020 (6). Approximately 47% of adults aged
≥30 years in the United States (approximately 65 million
adults) have periodontitis: 8.7% with mild periodontitis,
30.0% with moderate, and 8.5% with severe periodontitis (7).
Periodontitis increases with age; adults aged ≥65 years have
periodontitis at rates of 5.9%, 53.0%, and 11.2% for mild,
moderate, and severe forms, respectively (7). As the U.S. adult
population ages and is more likely to retain more teeth than
previous generations, the prevalence of periodontitis is expected
to increase and consequently could increase the need for
expenditures for preventive care and periodontal treatment (8).

Periodontitis is directly associated with lower levels of
education and higher levels of poverty, both of which influence
the use of dental services by adults (9–12). Educational
attainment and poverty might mediate significant differences in
the prevalence of periodontal disease between different racial/
ethnic populations. Smoking and some chronic diseases such as

diabetes are important modifiable risk factors for periodontitis
(13). Since the early 1960s, U.S. national surveys have assessed
the periodontal status of adults (14). However, the validity
of estimates from these surveys has been limited by the use
of partial-mouth periodontal examination protocols, which
significantly underestimate the prevalence of periodontitis
(15–17). The 2009–2010 National Health and Nutrition
Examination Survey (NHANES) cycle is the first to include
a full-mouth periodontal examination for U.S. adults (aged
≥30 years) and provides the most direct evidence for the true
prevalence of periodontitis in this population.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (18) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (19). This report provides information
concerning disparities in periodontitis, a topic that was not
discussed in the 2011 CHDIR (18). The purposes of this
periodontitis in adults report are to discuss and raise awareness
of differences in the characteristics of people with periodontal
disease and to prompt actions to reduce these disparities.

Methods
To examine racial/ethnic disparities in the estimated percentage

of adults aged ≥30 years with periodontitis by age, sex, education,
poverty levels, and smoking status, CDC analyzed data from the
2009-2010 NHANES cycle. NHANES is a cross-sectional survey
designed to monitor the overall health and nutritional status of
civilian, noninstitutionalized U.S. population. NHANES uses
a stratified multistage probability sampling design. For 2-year

Periodontitis Among Adults Aged ≥30 Years —
United States, 2009–2010

Gina Thornton-Evans, DDS1
Paul Eke, PhD2
Liang Wei, MS3
Astrid Palmer1

Refilwe Moeti, MA4
Sonja Hutchins, MD5

Luisa N. Borrell, DDS6
1Division of Oral Health, National Center for Chronic Disease Prevention and Health Promotion, CDC

2Division of Population Health, National Center for Chronic Disease Prevention and Health Promotion, CDC
3DB Consulting Group, Inc.

4Division of Heart Disease and Stroke Prevention, National Center for Chronic Disease Prevention and Health Promotion, CDC
5Office of Minority Health and Health Equity, CDC

6Department of Health Sciences, Lehman College, City University of New York, Bronx, New York

Corresponding author: Paul Eke, Division of Population Health, National Center for Chronic Disease Prevention and Health Promotion, CDC. Telephone:
770-488-6092; E-mail: [email protected].

Supplement

130 MMWR / November 22, 2013 / Vol. 62 / No. 3

data cycles, NHANES surveys a national representative sample.
The technical details of the survey, including sampling design,
periodontal data collection protocols, and data, are available
online (http://www.cdc.gov/nchs/nhanes.htm). A total of 5,037
adults aged ≥30 years participated in the survey, and 951 were
excluded for medical reasons or incomplete oral examinations. In
this analysis, 343 edentulous participants were excluded, leaving
a total of 3,743 participants, representing a weighted population
of approximately 137.1 million civilian noninstitutionalized
U.S. adults. The findings in this report cannot be compared with
those of previous studies using NHANES data (9,10) because
the case definitions and age range used in this analysis differed.

All periodontal examinations were conducted in a mobile
examination center by dental hygienists registered in at least
one U.S. state. Gingival recession was defined as the distance
between the free gingival margin and the cementoenamel
junction; pocket depth was defined as the distance from free
gingival margin to the bottom of the sulcus or periodontal
pocket. These measurements were made at six sites per
tooth (mesiobuccal, midbuccal, distobuccal, mesiolingual,
midlingual, and distolingual) for all teeth except third molars.
For measurements at each tooth site, a periodontal probe
(Hu-Friedy PCP 2) with graduations of 2 mm, 4 mm, 6 mm,
8 mm, 10 mm, and 12 mm was positioned parallel to the long
axis of the tooth at each site. Each measurement was rounded
to the lowest whole millimeter. Data were recorded directly
into an NHANES oral health data management program that
instantly calculated attachment loss as the difference between
probing depth and gingival recession. Bleeding from probing
and the presence of dental furcations were not assessed.

Periodontal measurements were used to classify participants
as having mild, moderate, or severe disease by using standard
case definitions for surveillance of periodontitis (4); total
prevalence of periodontitis in the population was calculated
by combining prevalence of mild, moderate, and severe
periodontitis. Severe periodontitis was defined as having two or
more interproximal sites with ≥6 mm attachment (not on the
same tooth) and one or more interproximal sites with ≥5 mm
pocket depth. Moderate periodontitis was defined as two or
more interproximal sites with ≥4 mm clinical attachment
(not on the same tooth) or two or more interproximal sites
with pocket depth of ≥5 mm (not on the same tooth). Mild
periodontitis was defined as two or more interproximal sites
with ≥3 mm attachment and two or more interproximal sites
with ≥4 mm pocket depth (not on the same tooth) or one site
with ≥5 mm.

Race/ethnicity was self-reported; for this analysis, three
race/ethnicity groups, each with a sample size large enough
to ensure statistically reliable estimates, were used: non-
Hispanic white, non-Hispanic black, and Mexican-American.

Poverty status categories, or percentage of poverty relative to
the federal poverty level (FPL), was based on family income,
family size, and number of children in the family, for families
with two or fewer adults, and on the age of the adults in the
household. Families or individuals with income below their
appropriate income thresholds, as determined by family size
and composition, were classified as living below the FPL. The
income thresholds are updated annually by the U.S. Census
Bureau (available at http://aspe.hhs.gov/poverty/11poverty.
shtml). Education was classified as less than high school, high
school graduate or equivalent, and greater than high school.
Smoking status was determined by responses to two questions:
1) “Have you smoked at least 100 cigarettes in your life?”
and 2) “Do you now smoke cigarettes?” Participants who
answered yes to both questions were categorized as current
smokers, participants who answered yes to the first question
and no to the second were categorized as former smokers,
and participants who answered no to both questions were
categorized as never smokers. Geographic regions were not
analyzed because NHANES is not designed to be representative
at regional (or lower) levels.

Disparities were assessed by age group, sex, race/ethnicity,
education, FPL, and smoking status for the total population
and by race/ethnicity. Referent groups for each category had
the best overall periodontal health for the category. Disparities
were measured as deviations from a referent group, which
was the group that had the most favorable estimate for the
variables used to assess disparities during the time reported.
Absolute difference was measured as the simple difference
between the periodontitis prevalence for the group of interest
and the referent group. The relative difference, a percentage,
was calculated by dividing the absolute difference by the value
in the referent category and multiplying by 100. The z test
was used to assess significant differences between absolute
differences from the referent group, with significance set at
p<0.05. Data (using mobile examination center weights) were
analyzed using statistical software to adjust for the effects of
the sampling design, including the unequal probability of
selection, and to determine standard errors (SEs).

Results
During 2009–2010, an estimated 47.2% of adults aged

≥30 years in the United States had periodontitis (Table 1).
The prevalence of total and moderate periodontitis increased
with increasing age among all adults. However, the prevalence
of mild and severe periodontitis remained relatively steady at
<15% across all age groups (Figure).

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 131

The prevalence of periodontitis was significantly higher in
non-Hispanic blacks (58.6%) and Mexican-Americans (59.7%)
compared with non-Hispanic whites (42.6%). Among all
racial/ethnic groups, the prevalence of periodontitis increased
with age (24.4%–70.1%), with the largest relative difference in
prevalence within age groups occurring among non-Hispanic
whites (range: 16.6%–68.0%). The prevalence of periodontitis
was significantly higher among men (56.4%) than women
(38.4%) overall, and this finding was consistent among racial/
ethnic groups. By education level, periodontitis was highest
among persons with less than a high school education (66.9%),
and the relative difference between those with greater than a
high school education and those with less education was largest
in Mexican-Americans (73.8%) and smallest in non-Hispanic
blacks (28.8%). The prevalence of periodontitis increased as
FPL percentage decreased, with an estimate of 65.4% of persons
in the poorest families (<100% FPL), representing an 85%
relative increase compared with families at ≥400% FPL. The
relative difference in prevalence between these categories of FPL
was largest among non-Hispanic whites (82.8%) and smallest
among non-Hispanic blacks (35.5%). Periodontitis was more
prevalent among current smokers (64.2%) than nonsmokers

(39.8%) and significantly higher among non-Hispanic black
current smokers (79.1%) than non-Hispanic white (60.8%)
and Mexican-American current smokers (69.1%) (Table 1).

During 2009–2010, an estimated 8.7% of the U.S. adult
population had mild periodontitis. The prevalence of moderate
periodontitis was 30.0% (Table 2). Prevalence of moderate
periodontitis increased with age and peaked at age ≥65
years. Overall, prevalence was higher in men (33.8%) than
women (26.4%) and higher among non-Hispanic black men
(42.7%) than men in other racial/ethnic groups. Increasing
prevalence was associated with lower education and poverty
levels. Specifically, the prevalence of moderate periodontitis
at the lowest levels of education and poverty were higher
among non-Hispanic whites and Mexican-Americans than
non-Hispanic blacks. Prevalence of moderate periodontal
disease was higher among current smokers (36.5%) and former
smokers (35.6%) than among nonsmokers (25.6%). However,
this pattern was not consistent among non-Hispanic blacks
and Mexican-Americans, among whom the highest prevalence
of moderate periodontitis was among former smokers. The
relative difference in prevalence between poverty levels was
smallest among non-Hispanic blacks, suggesting that income

TABLE 1. Prevalence of periodontitis among adults aged ≥30 years, by race/ethnicity and selected characteristics — National Health and Nutrition
Examination Survey, United States, 2009–2010

Characteristics

Total* (N = 3,743)
White, non-Hispanic

(N = 1,792)
Black, non-Hispanic

(N = 673) Mexican-American (N = 1,076)

No. of
adults

Weighted
no. of

adults (in
millions) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 3,743 137 47.2 (2.1) — — 42.6 (3.0) — — 58.6 (3.1) — — 59.7 (2.2) — —

Age group (yrs)
30–34 435 16.7 24.4 (2.7) Ref. — 16.6 (3.3) Ref. — 37.9 (8.2) Ref. — 43.7 (3.0) Ref. —
35–49 1,352 54.0 36.6 (1.6) 12.2† 50.0 28.5 (2.3) 11.9† 71.7 51.0 (3.7) 13.1 34.6 56.9 (2.9) 13.2† 30.2
50–64 1,128 43.4 57.2 (2.6) 32.8† 134.4 51.2 (3.7) 34.6† 208.4 72.8 (3.2) 34.9† 92.1 72.9 (3.6) 29.2† 66.8
≥65 828 22.9 70.1 (3.0) 45.7† 187.3 68.0 (3.9) 51.4† 309.6 72.7 (4.7) 34.8† 91.8 78.4 (4.1) 34.7† 79.4

Sex
Female 1,871 69.6 38.4 (2.4) Ref. — 34.6 (3.4) Ref. — 47.1 (3.5) Ref. — 47.4 (2.5) Ref. —
Male 1,872 67.5 56.4 (2.1) 18.0† 46.9 50.7 (3.0) 16.1† 46.5 72.9 (3.9) 25.8† 54.8 70.8 (2.3) 23.4† 49.4

Education
Less than high

school
1,030 23.8 66.9 (2.4) 27.6† 70.2 59.6 (4.6) 21.7† 57.3 64.9 (4.1) 14.5† 28.8 71.8 (2.2) 30.5† 73.8

High school
graduate or
equivalent

815 29.6 53.5 (3.2) 14.2† 36.1 49.3 (4.2) 11.4† 30.1 67.3 (4.0) 16.9† 33.5 59.5 (4.6) 18.2† 44.1

Greater than
high school

1,889 83.3 39.3 (2.3) Ref. — 37.9 (3.0) Ref. — 50.4 (4.1) Ref. — 41.3 (3.2) Ref. —

Poverty level
<100% FPL 625 13.5 65.4 (2.5) 30.0† 84.7 62.7 (6.0) 28.4† 82.8 58.8 (5.7) 15.4 35.5 69.4 (2.9) 28.3† 68.9
100%–199% FPL 901 22.7 57.4 (3.0) 22.0† 62.1 52.6 (6.0) 18.3† 53.4 65.6 (4.1) 22.2† 51.2 59.9 (4.1) 18.8† 45.7
200%–499% FPL 905 37.7 50.2 (2.5) 14.8† 41.8 48.0 (3.0) 13.7† 39.9 62.0 (4.1) 18.6† 42.9 54.8 (3.9) 13.7† 33.3
≥400% FPL 960 52.4 35.4 (3.0) Ref. — 34.3 (3.4) Ref. — 43.4 (6.1) Ref. — 41.1 (4.4) Ref. —

Smoking status
Current smoker 728 23.2 64.2 (2.6) 24.4† 61.3 60.8 (3.0) 26.8† 78.8 79.1 (5.0) 32.6† 70.1 69.1 (3.6) 13.3† 23.8
Former smoker 957 35.7 52.5 (3.1) 12.7† 31.9 48.8 (4.1) 14.8† 43.5 67.1 (5.4) 20.6† 44.3 64.1 (5.7) 8.3 14.9
Nonsmoker 2,058 78.1 39.8 (2.1) Ref. — 34.0 (3.0) Ref. — 46.5 (3.5) Ref. — 55.8 (2.6) Ref. —

Abbreviations: FPL = federal poverty level; Ref. = referent; SE = standard error.
* The 202 respondents in “other” race group are not included.
† Significant at p<0.05 by z test.

Supplement

132 MMWR / November 22, 2013 / Vol. 62 / No. 3

had the least impact on moderate periodontitis in this racial/
ethnic group. Significant absolute differences were found
in moderate periodontitis among current smokers, former
smokers, and nonsmokers and was significantly higher among
non-Hispanic blacks.

Severe periodontitis was estimated to occur in 8.5% of U.S.
adults aged ≥30 years (Table 3). Severe periodontitis was twice
as common among non-Hispanic blacks (13.2%) and Mexican-
Americans (13.3%) as among non-Hispanic whites (6.3%).
Severe periodontitis increased with age and peaked at age 50 years
among all racial/ethnic groups. Overall, severe disease was almost
three times higher among men (12.5%) than women (4.2%) and
approximately two times higher among non-Hispanic black men
(19.3%) and Mexican-American men (18.8%) than among non-
Hispanic white men (9.4%). Severe periodontitis among persons
with less than a high school education was an estimated 17.3%
and decreased with increasing levels of education. Among racial/
ethnic groups, the smallest relative differences by level of education
occurred among non-Hispanic blacks. Similarly, the prevalence
of severe periodontitis increased with increasing poverty levels,
with an estimated 16.3% of adults in families living at <100%
FPL having severe disease. The relative difference in prevalence
by poverty level (across all racial/ethnic groups) was smallest
among non-Hispanic blacks, suggesting that income had the least
influence on severe periodontitis in this racial/ethnic group. The

prevalence of severe periodontitis was approximately two times as
common among smokers at 17.7% than among former smokers
(9%) and nonsmokers (5.4%) and was significantly higher among
non-Hispanic blacks (24.4%) and Mexican-Americans (24.5%)
than among non-Hispanic white smokers (13.9%)

Discussion
Overall, significant disparities exist in the prevalence of

periodontitis by race/ethnicity, education and poverty level.
These results suggest that non-Hispanic blacks and Mexican-
Americans have similar prevalences of periodontitis but higher
prevalences than non-Hispanic whites. In addition, the relative
differences in the prevalence of total periodontitis (i.e., mild,
moderate, and severe combined) among non-Hispanic blacks
varied the least by poverty and education levels, possibly
suggesting that poverty and education have less of an effect
than other factors on the higher prevalence of periodontitis
among non-Hispanic Blacks and Mexican-Americans. The
highest prevalence of periodontitis was found among adults
aged ≥65 years. By 2030, the number of adults aged ≥65 years
in the U.S. will double to 71 million adults, or one in every
five Americans (8), with significant changes in the distribution
of demographic and socioeconomic groups.

Total periodontitis
Mild periodontitis
Moderate periodontitis
Severe periodontitis

0

10

20

30

40

50

60

70

80

90

30 40 50 60 70 80

Age (yrs)

Pe
rc

en
ta

g
e

FIGURE. Prevalence of total, mild, moderate, and severe periodontitis among adults aged ≥30 years, by age — National Health and Nutrition
Examination Survey, United States, 2009–2010

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 133

Limitations
The findings in this report are subject to at least

four limitations, all of which might have resulted in an
underestimation of the prevalence of periodontitis cases. First,
the case definitions for periodontitis used measures from four
interproximal sites, and not all six of the sites were measured.
Second, estimates did not include persons with gingivitis.
Gingivitis is a form of periodontal disease that was not assessed
in the NHANES 2009–2010 data cycle. Third, NHANES does
not sample institutionalized persons such as older adults in
nursing homes, which might have resulted in an underestimate
for older adults. Fourth, NHANES does not collect data from
third molars. This exclusion of third molars is consistent with
previous NHANES data cycles; third molars are difficult to
assess clinically because of their alignment in the mouth, and
some are partially impacted.

Conclusion
Preventive dental care programs should be an integral part

of preventive health services for all ages and should include
strategies to make dental care programs accessible to all racial/
ethnic groups to promote health and preserve health-related
quality of life in older adults. Adults aged ≥65 years do not
have dental coverage through Medicare, and approximately
70% of U.S. adults in this age group have no dental coverage
(20). Management of diabetes and smoking is an important
component of prevention and treatment of adult periodontitis
(13). The findings in this report indicate that current smokers
had a much higher prevalence of severe periodontitis; smoking
is categorized as a major modifiable risk factor for periodontitis.
This is consistent with the 2004 Surgeon General’s Report on
the Health Consequences of Smoking, which infers a causal
relationship between smoking and periodontitis (21). Because
the prevalence of severe periodontitis is higher among current
smokers, tobacco cessation programs are a potential strategy to
address disparities in periodontitis in the U.S. population (22).

TABLE 2. Prevalence of moderate periodontitis among adults aged ≥30 years, by race/ethnicity and selected characteristics — National Health
and Nutrition Examination Survey, United States, 2009–2010

Characterisitcs

Total* (N = 3,743)
White, non-Hispanic

(N = 1,792) Black, non-Hispanic (N = 673)
Mexican-American

(N = 1,076)

No. of
adults

Weighted
no. of

adults (in
millions) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (SE)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 3,743 137 30.0 (1.6) —  — 28.5 (2.3) —  — 33.6 (2.1) —  — 32.8 (1.7) —  —

Age groups (yrs)
30–34 435 16.7 13.0 (1.7) Ref. —  9.7 (1.6) Ref. —  18.6 (5.7) Ref. —  21.4 (3.4) Ref. — 
35–49 1,352 54.0 19.4 (1.7) 6.4† 49.2 15.5 (2.2) 5.8† 59.8 24.3 (3.6) 5.7 30.6 29.7 (2.4) 8.3† 38.0
50–64 1,128 43.4 37.7 (2.5) 24.7† 190.0 34.2 (3.4) 24.5† 252.6 46.4 (3.0) 27.8† 149.5 42.9 (1.6) 21.5† 100.5
≥65 828 22.9 53.0 (2.3) 40.0† 307.7 52.8 (2.6) 43.1† 444.3 50.2 (4.4) 31.6† 169.9 51.9 (4.3) 30.5† 142.5

Sex
Female 1,871 69.6 26.4 (2.2) Ref. —  24.8 (3.0) Ref. —  26.2 (2.7) Ref. —  29.2 (2.2) Ref. — 
Male 1,872 67.5 33.8 (1.4) 7.4† 28.0 32.3 (2.0) 7.5† 30.2 42.7 (3.3) 16.5† 63.0 36.1 (2.1) 6.9† 23.6

Education
Less than

high school
1,030 23.8 40.6 (2.6) 15.2† 59.8 42.2 (5.4) 17.1† 68.1 33.0 (3.6) 4.0 13.8 40.1 (2.4) 17.5† 77.4

High school
graduate or
GED certificate

815 29.6 34.2 (2.2) 8.8† 34.6 32.6 (3.1) 7.5 29.9 41.8 (3.8) 12.8† 44.1 30.6 (4.2) 8.0 35.4

Greater than
high school

1,889 83.3 25.4 (1.8) Ref. —  25.1 (2.3) Ref. —  29.0 (3.2) Ref. —  22.6 (2.4) Ref.  —

Poverty level
<100% FPL 625 13.5 37.8 (3.9) 14.3† 60.9 39.2 (7.9) 16.4† 71.9 30.8 (4.4) 4.1 15.4 38.4 (3.0) 16.2† 73
100%–199% FPL 901 22.7 32.9 (2.2) 9.4† 40.0 31.2 (4.6) 8.4 36.8 40.9 (2.1) 14.2† 53.2 30.5 (3.6) 8.3 37.4
200%–499% FPL 905 37.7 34.4 (2.3) 10.9† 46.4 33.9 (3.0) 11.1† 48.7 35.9 (5.3) 9.2 34.5 32.6 (3.1) 10.4 46.8
≥400% FPL 960 52.4 23.5 (2.0) Ref. —  22.8 (2.4) Ref. —  26.7 (5.4) Ref. —  22.2 (4.6) Ref. — 

Smoking status
Current smoker 728 23.2 36.5 (2.1) 10.9† 42.6 38.2 (2.8) 15.1† 65.4 41.5 (3.8) 15.7† 60.9 30.2 (2.6) -0.7 2.3
Former smoker 957 35.7 35.6 (3.1) 10.0† 39.1 33.3 (4.0) 10.2† 44.2 47.0 (3.3) 21.2† 82.2 41.0 (3.8) 10.1† 32.7
Nonsmoker 2,058 78.1 25.6 (1.4) Ref. —  23.1 (2.1) Ref. —  25.8 (1.9) Ref. —  30.9 (2.1) Ref. — 

Abbreviations: Ref. = referent; SE = standard error.
* The 202 respondents in “other” race group are not included.
† Significant at p<0.05 by z test.

Supplement

134 MMWR / November 22, 2013 / Vol. 62 / No. 3

Two related Healthy People 2020 objectives are currently being
monitored. One focuses on dental professionals providing
tobacco cessation counseling in a dental setting, and another
monitors consumers’ self-report of tobacco cessation counseling
in a dental office (9). Overall, this study demonstrates disparities
in periodontitis by age, race, education, and income, and risk
factors such as smoking status in the U.S. adult population.
The capacity of oral health programs within state and local
health agencies can be broadened to capture this subset of the
population. The program activities might include efforts to
1) reduce tobacco use, particularly smoking; 2) educate persons
on the benefits of regular dental care; and 3) facilitate health
communication efforts to make key groups aware of effective
preventive interventions.

References
1. Page RC, Eke PI. Case definitions for use in population-based

surveillance of periodontitis. J Periodontol 2007;78:1387–99.
2. Martin JA, Page RC, Kaye EK, Hamed MT, Loeb CF. Periodontitis severity

plus risk as a tooth loss predictor. J Periodontol 2009;80:202–9.
3. Pace CC, McCullough GH. The association between oral microorganisms

and aspiration pneumonia in the institutionalized elderly: review and
recommendations. Dysphagia 2010;25:307–22.

4. Lamster IB, DePaola DP, Oppermann RV, Papapanou PN, Wiler RS.
The relationship of periodontal disease to diseases and disorders at distant
sites: communication to health care professionals and patients. J Am
Dent Assoc 2008;139:1389–97.

5. Eke PI, Page RC, Wei L, Thornton-Evans GO, Genco RJ. Update of
the case definitions for population-based surveillance of periodontitis.
J Periodontol 2012;83:1449–54.

6. US Department of Health and Human Services. Healthy people 2020.
Washington, DC: US Department of Health and Human Services; 2011.
Available at http://www.healthypeople.gov/2020.

7. Eke PI, Dye Ba, Wei L, Thornton-Evans GO, Genco RJ. Prevalence of
periodontitis in adults in the United States: 2009 and 2010. J Dent Res
2012;91:914–20.

8. CDC. Trends in aging—United States and worldwide. MMWR 2003;
52:101–6.

9. Borrell LN, Crawford ND. Socioeconomic position indicators and
periodontitis: examining the evidence. Periodontol 2000 2012;58:69–83.

10. Borrell LN, Burt BA, Taylor GW. Prevalence and trends in periodontitis
in the USA: from the NHANES III to the NHANES 1988 to 2000. J
Dent Res 2005;84:924–30.

11. Borrell LN, Crawford ND. Social disparities in periodontitis among
United States adults 1999–2004. Community Dent Oral Epidemiol
2008;36:383–91.

12. Gibson RM, Fisher CR. Age differences in health care spending fiscal
year 1977. Soc Secur Bull 1979;42:3–16.

13. Genco RJ. Current view of risk factors for periodontal diseases. J
Periodontol 1996;67:1041–9.

TABLE 3. Prevalence of severe periodontitis among adults aged ≥30 years, by race/ethnicity and selected characteristics — National Health
and Nutrition Examination Survey, United States, 2009–2010

Characteristics

Total* (N = 3,743)
White, non-Hispanic

(N = 1,792)
Black, non-Hispanic

(N = 673)
Mexican-American

(N = 1,076)

No. of
adults

Weighted
no. of

adults (in
millions) % (SE) Difference

Relative
difference

(%) % (SE) Difference

Relative
difference

(%) % (SE) Difference

Relative
difference

(%) % (SE) Difference

Relative
difference

(%)

Total 3,743 137 8.5 (0.9) — — 6.3 (1.1) — — 13.2 (1.8) — — 13.3 (1.8) — —

Age groups (yrs)
30–34 435 16.7 1.9 (0.6) Ref. — 0.8 (0.6) Ref. — 3.1 (2.0) Ref. — 4.6 (2.3) Ref. —
35–49 1,352 54.0 6.7 (0.8) 4.8† 252.6 4.8 (0.9) 4.0† 500.0 10.4 (1.6) 7.3† 235.5 10.2 (1.8) 5.6 121.7
50–64 1,128 43.4 11.7 (1.6) 9.8† 515.8 8.3 (1.4) 7.5† 937.5 19.8 (3.3) 16.7† 538.7 25.2 (3.6) 20.6† 447.8
≥65 828 22.9 11.2 (2.2) 9.3† 489.5 8.8 (2.6) 8.0† 1000.0 17.8 (4.4) 14.7† 474.2 21.3 (3.0) 16.7† 363.0

Sex
Female 1,871 69.6 4.5 (0.7) Ref. — 3.3 (0.8) Ref. — 8.3 (2.0) Ref. — 7.3 (1.7) Ref. —
Male 1,872 67.5 12.6 (1.3) 8.1† 180.0 9.4 (1.7) 6.1† 184.8 19.3 (2.1) 11.0† 132.5 18.8 (2.0) 11.5† 157.5

Education
Less than

high school
1,030 23.8 17.3 (2.1) 11.8† 214.5 12.0 (2.3) 7.2† 150 21.0 (5.0) 10.7 103.9 18.2 (2.1) 12.5† 219.3

High school
graduate or
equivalent

815 29.6 9.8 (1.6) 4.3† 78.2 8.5 (1.9) 3.7 77.1 11.5 (2.6) 1.2 11.7 14.1 (3.3) 8.4† 147.4

Greater than
high school

1,889 83.3 5.5 (1.0) Ref. — 4.8 (1.1) Ref. — 10.3 (2.3) Ref. — 5.7 (1.2) Ref. —

Poverty level
<100% FPL 625 13.5 16.3 (2.2) 12.2† 297.6 14.2 (3.4) 10.6† 294.4 16.8 (3.5) 7.7 84.6 16.7 (3.5) 10.4† 165.1
100%–199% FPL 901 22.7 14.1 (1.8) 10.0† 243.9 11.9 (2.1) 8.3† 230.6 13.0 (3.5) 3.9 42.9 15.3 (3.1) 9.0† 142.9
200%–499% FPL 905 37.7 7.9 (1.3) 3.8† 92.7 7.0 (1.6) 3.4 94.4 13.0 (1.8) 3.9 42.9 10.0 (1.9) 3.7 58.7
≥400% FPL 960 52.4 4.1 (0.9) Ref. — 3.6 (1.0) Ref. — 9.1 (2.7) Ref. — 6.3 (2.8) Ref. —

Smoking status
Current smoker 728 23.2 17.7 (2.4) 12.3† 227.8 13.9 (3.2) 10.3† 286.1 24.4 (3.3) 17.0† 229.7 24.5 (4.9) 14.2† 137.9
Former smoker 957 35.7 9.0 (1.3) 3.6† 66.7 7.4 (1.6) 3.8† 105.6 14.9 (4.2) 7.5 101.4 13.8 (2.0) 3.5 34.0
Nonsmoker 2,058 78.1 5.4 (0.9) Ref. — 3.6 (0.9) Ref. — 7.4 (1.6) Ref. — 10.3 (1.5) Ref. —

Abbreviations: Ref. = referent; SE = standard error.
* The 202 respondents in “other” race group are not included.
† Significant at p<0.05 by z test.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 135

14. Dye BA, Thornton-Evans GO. A brief history of national surveillance
efforts for periodontal disease in the United States. J Periodontol 2007;
78:1373–9.

15. Susin C, Kingman A, Albandar JM. Effect of partial recording protocols
on estimates of prevalence of periodontal disease. J Periodontol
2005;76:262–7.

16. Hunt RJ, Fann SJ. Effect of examining half the teeth in a partial
periodontal recording of older adults. J Dent Res 1991;70:1380–5.

17. Eke PI, Thornton-Evans GO, Wei L, Borgnakke WS, Dye BA. Accuracy
of NHANES periodontal examination protocols. J Dent Res 2010;
89:1208–13.

18. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

19. CDC. Introduction: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3)

20. Manski RJ, Brown E. Dental use, expenses, private dental coverage and
changes, 1996 and 2004. MEPS chartbook No. 17. Rockville MD:
Agency for Healthcare Research and Quality; 2007.

21. US Department of Health and Human Services, CDC. The health
consequences of smoking: a report of the Surgeon General. Atlanta, GA:
US Department of Health and Human Services, CDC; 2004. Available
at http://www.cdc.gov/tobacco/data_statistics/sgr/2004/complete_
report/index.htm.

22. Carr AB, Ebbert J. Interventions for tobacco cessation in the dental
setting. Cochrane Database Syst Rev 2012;6:CD005084.

Supplement

136 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
Approximately one third of all infant deaths in the U.S. are

related to preterm birth (1). Infants who survive a preterm
birth are at greater risk than those born later in pregnancy
for early death and lifelong effects such as neurologic and
cognitive difficulties (1–4). The rate of preterm births (i.e.,
<37 completed weeks’ gestation) increased approximately
30% during 1981–2006 (5). In 2007, this trend began to
reverse; the U.S. preterm birth rate decreased for the fourth
consecutive year in 2010, decreasing from the 2006 high of
12.8% to 12.0% in 2010 (5). A total of 4,265,555 births were
reported for 2006, including 542,893 preterm births, and
3,999,386 births were reported for 2010, including 478,790
preterm births. Although most of the recent decrease in this
rate was among infants born at 34 to 36 weeks’ gestation
(i.e., late preterm), with a decrease from 9.15% to 8.49%
during 2006–2010, the rate of infants born at <34 weeks’
gestation (i.e., early preterm) also decreased from 3.66%
in 2006 to 3.50% in 2010 (5). Despite improvements in
the rate of preterm births, the total number of infants born
preterm remains higher than any year during 1981–2001 (5).
Substantial differences in preterm birth rates by race/ethnicity
persist; additional examination of these differences can provide
insight into potential areas for interventions.

The preterm birth analysis and discussion that follows is
part of the second CDC Health Disparities and Inequalities
Report (CHDIR) and updates information presented in the
2011 CHDIR (6). The 2011 CHDIR (7) was the first CDC
report to assess disparities across a wide range of diseases,
behavioral risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (8). The purposes of this preterm birth
report are to raise awareness of racial/ethnic differences among
women giving birth to preterm infants and to motivate actions
to reduce disparities.

Methods
To assess differences in preterm birth rates by race/ethnicity,

CDC analyzed final 2006 and 2010 birth certificate data
from the National Vital Statistics System (9). Birth certificates

provide demographic and health information on the mother
and newborn such as sex, race, ethnicity, gestational age,
and geographic region. Geographic region was not analyzed
independently because this variable is related to demographic
characteristics that can influence preterm birth rates.
Comparable information on educational attainment of the
mother is not available for the entire national reporting area.

Gestational age measurement is based primarily on the
interval between the date of the last normal menses, or last
menstrual period (LMP), and the date of birth. The preterm
birth rate is defined as births at <37 completed weeks of
gestation per 100 total births in a given category; early preterm
birth rate is defined as <34 weeks, and late preterm as 34–36
weeks. Race/ethnicity of the mother was self-reported in five
categories; white, black, American Indian/Alaska Native (AI/
AN), Asian/Pacific Islander (A/PI), and Hispanic. In this
report, references to whites, blacks, AI/ANs, and A/PIs refer
to non-Hispanic women. Women of Hispanic ethnicity might
be of any race or combination of races.

Disparities were measured as the deviations from a referent
category rate. Births to non-Hispanic white mothers were used
as the referent group for racial/ethnic comparisons. Absolute
difference was measured as the simple difference between the
rate for a population subgroup and the rate for its respective
reference group. The relative difference, a percentage, was
calculated by dividing the difference by the value in the referent
category and multiplying by 100. The statistical significance of
the differences was determined by using the z test at the 95%
confidence level (10).

Results
Decreases in preterm births occurred for each of the race/

ethnicity groups; white, black, Hispanic, AI/AN, and A/PI
from 2006 to 2010 (Table). From 2006 to 2010, the preterm
birth rate for black infants decreased by 8% to 17.1%, the
lowest level ever reported (5). Despite the decrease, the 2010
preterm rate for black infants (17.1%) was approximately 60%
higher than that for white infants (10.8%). AI/AN (13.6%)
and Hispanic (11.8%) infants were also at a higher risk for
preterm birth in 2010 than white and A/PI infants.

Preterm Births — United States, 2006 and 2010
Joyce A. Martin, MPH

Michelle J.K. Osterman, MHS
National Center for Health Statistics, CDC

Corresponding author: Joyce A. Martin, National Center for Health Statistics, CDC. Telephone 301-458-4362; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 137

The largest relative differences among the race/ethnicity
groups are in early preterm births. Decreases in early preterm
births occurred from 2006 to 2010 for white, black, and
Hispanic infants. Despite an 8% decrease in the early preterm
rate for black infants from 2006 to 2010, the 2010 early
preterm birth rate among black infants (6.1%) was double the
rate among white (2.9%) and A/PI (2.9%) infants.

The rate of late preterm births declined among each of
the race/ethnicity groups during 2006–2010. In 2010, black
infants were approximately 40% more likely to be born late
preterm than white and A/PI infants. AI/AN and Hispanic
infants also were more likely than white and A/PI infants to
be born late preterm.

Discussion
Decreases occurred from 2006 to 2010 in preterm birth

rates overall and in all racial/ethnic groups examined; however,
substantial disparities persisted among racial ethnic groups in
2010. The greatest absolute difference by race/ethnicity in
total preterm, early preterm, and late preterm birth rates was
among black infants. Black infants have had the highest risk
for preterm birth since comparable data on gestational age have
been available (1981). The causes of preterm births are not well
understood (2). However, disparities among groups might be

related to differences in socioeconomic status, prenatal care,
maternal risk behaviors, infection, nutrition, preconception
stress, and genetics (2).

Limitations
The findings in this report are subject to at least one limitation.

The date of the LMP is subject to error from imperfect
maternal recall, transcription error, or misidentification of
LMP because of postconception bleeding, delayed ovulation,
or intervening early miscarriage (5).

Conclusion
Continued reduction in the preterm birth rate is important

because approximately one out of every eight infants was born
too early in 2010. If the preterm rate continues to decrease
at the pace observed from 2006 to 2010, the Healthy People
2020 objective to reduce the rate to 11.4% (objective no.
MICH 9-1) (11) will be achieved for the nation overall and
for some racial/ethnic groups (i.e., white and A/PI). The 2020
goal for preterm birth rates is further from reach for others;
the 2010 rate among blacks (17.1%) must decrease by 50%
percent for 2020 (or 5% per year), and the 2010 rate among
AI/ANs (13.6%) must decrease by approximately 20% (2%

TABLE. Total, early, and late preterm birth rates,* by race/ethnicity of mother — National Vital Statistics System, United States, 2006 and 2010

Weeks of gestation at birth and race/ethnicity of mother

2006 2010

Rate
%

Absolute
difference

(percentage
points)

Relative
difference

(%)
Rate

%

Absolute
difference

(percentage
points)

Relative
difference

(%)†

Total preterm births (<37 weeks’ gestation) 12.8 — — 12.0 — —
White, non-Hispanic 11.7 Ref. Ref. 10.8 Ref. Ref.
Black, non-Hispanic 18.5 6.8 58 17.1 6.3 58
Hispanic§ 12.2 0.5 4 11.8 1.0 9
Asian/Pacific Islander 10.9 -0.8 -7 10.7 -0.1 -1
American Indian/Alaska Native 14.2 2.5 21 13.6 2.8 26

Total early preterm births (<34 weeks’ gestation) 3.7 — — 3.5 — —
White, non-Hispanic 3.1 Ref. Ref. 2.9 Ref. Ref.
Black, non-Hispanic 6.6 3.5 113 6.1 3.2 110
Hispanic 3.4 0.3 10 3.3 0.4 14
Asian/Pacific Islander 2.8 -0.3 -10 2.9 0.0 0
American Indian/Alaska Native 4.0 0.9 29 4.0 1.1 38

Total late preterm births (34–36 weeks’ gestation) 9.1 — — 8.5 — —
White, non-Hispanic 8.6 Ref. Ref. 7.8 Ref. Ref.
Black, non-Hispanic 11.9 3.3 38 11.0 3.2 41
Hispanic 8.8 0.2 2 8.5 0.7 9
Asian/Pacific Islander 8.1 -0.5 -6 7.8 0.0 0
American Indian/Alaska Native 10.2 1.6 19 9.6 1.8 23

Abbreviation: Ref. = referent.
* Per 100 total births in a given category.
† Statistical significance was determined by using the z test at the 95% confidence level. All differences between the reference group and other groups are significant

(p<0.05) except for Asian/Pacific Islander.
§ Persons of Hispanic ethnicity might be of any race or any combination of races.

Supplement

138 MMWR / November 22, 2013 / Vol. 62 / No. 3

per year). Additional research is needed to clarify the causes of
preterm delivery and to develop policies for a future in which
preterm birth is a rare event for all populations.

References
1. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2008

period linked birth/infant death data set. Natl Vital Stat Rep 2010;60.
2. Behrman RE, Butler AS, editors. Preterm birth: causes, consequences,

and prevention. Washington, DC: National Academies Press; 2007.
3. Pitcher JB, Schneider LA, Drysdale JL, et al. Motor system development of

the preterm and low birthweight infant. Clin Perinatol 2011; 38:605–25.
4. Engle WA. Morbidity and mortality in late preterm and early term

newborns: a continuum. Clin Perinatol 2011;38:493–516.
5. Martin JA, Hamilton BE, Ventura SJ, et al. Births: final data for 2010.

Natl Vital Stat Rep 2012;61.

6. Martin JA. Preterm births—United States, 2007. In: CDC health
disparities and inequalities report—United States, 2011. MMWR
2011;60(Suppl; January 14, 2011).

7. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

8. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3)

9. CDC. Natality public use files: 2006–2010 and CD-ROM. Hyattsville,
MD: National Center for Health Statistics. Available at http://www.cdc.
gov/nchs/data_access/VitalStatsOnline.htm.

10. National Center for Health Statistics. User guide to the 2010 natality
public use file. Hyattsville, MD. Available from: ftp://ftp.cdc.gov/pub/.

11. US Department of Health and Human Services. Healthy People 2020
topics and objectives. Washington, DC: US Department of Health and
Human Services; 2011. Available at http://www.healthypeople.
gov/2020/topicsobjectives2020/default.aspx.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 139

Introduction
Potentially preventable hospitalizations are admissions

to a hospital for certain acute illnesses (e.g., dehydration)
or worsening chronic conditions (e.g., diabetes) that might
not have required hospitalization had these conditions been
managed successfully by primary care providers in outpatient
settings. Although not all such hospitalizations can be avoided,
admission rates in populations and communities can vary
depending on access to primary care, care-seeking behaviors,
and the quality of care available (1,2). Because hospitalization
tends to be costlier than outpatient or primary care, potentially
preventable hospitalizations often are tracked as markers of
health system efficiency. The number and cost of potentially
preventable hospitalizations also can be calculated to help
identify potential cost savings associated with reducing these
hospitalizations overall and for specific populations.

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (2) was
the first CDC report to assess disparities across a wide range
of diseases, behavior risk factors, environmental exposures,
social determinants, and health-care access. The topic
presented in this report is based on criteria that are described
in the 2013 CHDIR Introduction (3). This report updates
information on potentially preventable hospitalizations that
was presented in the first CHDIR (4). The purposes of this
report are to discuss and raise awareness of differences in the
race/ethnicity and income of persons with excess potential
preventable hospitalizations and to prompt actions to reduce
these disparities.

Methods
To examine trends in a composite measure of potentially

preventable hospitalizations among adults aged ≥18 years in the
United States, the Agency for Healthcare Research and Quality
(AHRQ) analyzed data for 2001–2009 from the Healthcare
Cost and Utilization Project (HCUP) databases (available at
http://www.ahrq.gov/research/data/hcup/index.html). HCUP
databases combine the data-collection efforts of state data

organizations, hospital associations, private data organizations,
and the federal government to create a national information
resource of discharge-level health-care data. HCUP includes
the largest collection of longitudinal hospital care data in the
United States with all-payer, encounter-level information,
beginning with 1988.

Numbers of potentially preventable hospitalizations in 2009
were estimated by race/ethnicity and income quartile for the
following eight conditions: diabetes, hypertension, congestive
heart failure, angina without procedure, asthma, dehydration,
bacterial pneumonia, and urinary infections. Hospitalizations
include all inpatient stays with these conditions listed as
the principal diagnosis regardless of admitting source (e.g.,
admissions through an emergency room, transfers from
other facilities, and direct admissions by a provider). Because
coding of race/ethnicity varies across state hospital databases,
analyses by race/ethnicity used a specially created 40%
sample of hospitals from states that contribute comparable
race/ethnicity data to HCUP (concerning approximately 16
million discharges from 2,000 hospitals in 36 states in 2009)
(5). Race was classified as non-Hispanic white, non-Hispanic
black, Asian/Pacific Islander (A/PI), and other. Ethnicity was
classified as Hispanic and non-Hispanic. Persons of Hispanic
ethnicity might be of any race or combination of races. Area
income, based on the income of the neighborhood in which
a patient lives, was used as a proxy for socioeconomic status.
Area income was divided into quartiles on the basis of the
mean household income by the patient’s ZIP Code. Quartile
1 refers to the lowest income communities, and quartile 4
refers to the wealthiest communities. Analyses by area income
used the Nationwide Inpatient Sample, a nationally stratified
20% sample of hospitals from states that contribute data to
HCUP (concerning approximately 8 million discharges from
1,000 hospitals in 44 states in 2009). Data regarding patients’
educational attainment or disability status were unavailable
or insufficient to provide estimates for certain populations
(i.e., American Indians/Alaska Natives, Native Hawaiians and
Other Pacific Islanders, and persons of multiple races). Data
on disparities related to sex and geographic location are not
presented but are available at http://www.ahrq.gov/research/
findings/nhqrdr/nhqrdr11/index.html#Efficiency.

Potentially Preventable Hospitalizations — United States, 2001–2009
Ernest Moy, MD1
Eva Chang, MPH2

Marguerite Barrett, MS3
1Agency for Healthcare Research and Quality, Rockville, Maryland

2Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
3M. L. Barrett, Inc., Del Mar, California

Corresponding author: Ernest Moy, Agency for Healthcare Research and Quality. Telephone: 301-427-1329; E-mail: [email protected].

Supplement

140 MMWR / November 22, 2013 / Vol. 62 / No. 3

Disparities in hospital admission rates per 100,000
population for 2001–2009 were estimated, and potential cost
savings related to income and race/ethnicity were examined.
Hospitalization rates for the potentially preventable conditions
were calculated by using the AHRQ Prevention Quality
Indicators (PQIs) modified version 4.2, which were adjusted
by age and gender on the basis of the 2000 U.S. standard
population (5). Excess potentially preventable hospitalizations
by area income were estimated by comparing the 2009 AHRQ
PQI composite rate of hospitalizations for residents of the
neighborhoods in the highest income quartile, the group
with the lowest rate, with the composite rate for residents of
neighborhoods in lower income quartiles. Similarly, excess
potentially preventable hospitalizations by race/ethnicity were
estimated by comparing the 2009 AHRQ PQI composite rate
of hospitalizations for A/PIs, the group with the lowest rate,
with the composite rate for other racial/ethnic groups. Total
charges included on hospital claims were converted to costs by
using hospital-level cost-to-charge ratios based on the Centers
for Medicare and Medicaid Services’ (CMS) hospital cost
report data (5). Costs associated with potentially preventable

hospitalizations were estimated by multiplying numbers of
excess hospitalizations for a group by the average cost per
hospitalization for that group. Costs are for the hospital cost
of producing the services and do not include physician costs
associated with hospital stay.

Results
During 2001–2009, the AHRQ PQI composite rate

decreased from 1,635 to 1,395 per 100,000 adults. Declines
in potentially preventable hospitalization rates were observed
across all income quartiles between 2001 and 2009 (Figure 1).
In all years, rates of hospitalizations were higher among residents
of neighborhoods in the three lower income quartiles (quartiles
1–3) compared with residents of neighborhoods in the highest
income quartile (quartile 4).

During 2009, if residents of the lowest income neighborhoods
(quartile 1) had the same rate of hospitalizations as residents
of the highest income neighborhoods (quartile 4), they would
have had approximately 500,000 fewer hospitalizations and

FIGURE 1. Rate* of potentially preventable hospitalizations† among adults aged ≥18 years, by income quartile§ — United States, 2001–2009

Source: Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project, Nationwide Inpatient Sample, 2001–2009.
* Per 100,000 population.
† For diabetes, hypertension, congestive heart failure, angina without procedure, asthma, dehydration, bacterial pneumonia, and urinary infections.
§ Area income was divided into quartiles based on the mean household income by the patient’s ZIP Code. Quartile 1 refers to the lowest income communities, and

quartile 4 refers to the wealthiest communities.

0

500

1,000

1,500

2,000

2,500

2001 2002 2003 2004 2005 2006 2007 2008 2009

Ra
te

Year

Quartile 1

Quartile 2

Quartile 3

Quartile 4

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 141

saved $3.6 billion in hospitalization costs
(Figure 2). If residents of income quartiles 2
and 3 had had the same hospitalization rate
as residents of income quartile 4, they would
have had approximately 220,000 and 90,000
fewer hospitalizations and saved $1.7 billion
and $700 million, respectively, in 2009.

Significant declines in hospitalization rates
also were observed across all race/ethnicity
groups during 2001–2009 (Figure 3). In
general, non-Hispanic blacks and Hispanics
had higher rates of hospitalizations than non-
Hispanic whites, and A/PIs had lower rates
than non-Hispanic whites.

During 2009, if non-Hispanic whites had had
the same rate of hospitalizations as A/PIs, they
would have had 700,000 fewer hospitalizations
and saved $7.7 billion in hospitalization
costs (Figure 4). If non-Hispanic blacks and
Hispanics had the same hospitalization rate
as A/PIs, they would have had 540,000 and
240,000 fewer hospitalizations, respectively,
and saved $3.7 billion and $700 million,
respectively, in 2009.

Discussion
The findings in this report are consistent with previous

studies showing decreasing rates of potentially preventable
hospitalizations for specific conditions such as congestive
heart failure (6,7) and for these hospitalizations in aggregate
(8). Because rates of all groups defined by race/ethnicity and
neighborhood income decreased at a similar pace, disparities
that were present in 2001 persisted through 2009. These
findings extend previous work by demonstrating that these
disparities accounted for a considerable share of costs associated
with potentially preventable hospitalizations.

Reducing hospitalization rates is a key to controlling
health-care costs. For many chronic conditions, inpatient
costs are the dominant expense. For example, approximately
half of the expenditures of persons with diabetes are spent on
hospital inpatient care, compared with 12% spent on diabetes
medications and supplies and 9% spent on physician office
visits (9). Disease management programs typically incur higher
outpatient and pharmacy costs that are offset by lower inpatient
costs (10). Programs to prevent chronic diseases also generate
savings by lowering rates of hospitalizations. Patient-centered
medical homes generate most of their savings by reducing
hospitalizations, and it is anticipated that the success of

accountable care organizations also will depend on their ability
to hold down inpatient costs (11).

Populations with the highest rates of potentially preventable
hospitalizations have the largest potential for lower rates and
inpatient costs. Communities with high rates of potentially
preventable hospitalizations might see the benefit of investing in
primary care, care coordination, and community health worker
strategies that can lower inpatient costs. The national Million
Hearts initiative (http://millionhearts.hhs.gov/index.html)
aims to prevent one million heart attacks and strokes by
focusing in primary care settings on the “ABCS”: aspirin when
appropriate, blood pressure control, cholesterol management,
and smoking cessation. Community-based prevention efforts
are focused on reducing sodium in the food supply, eliminating
consumption of transfats, and promoting tobacco cessation.
Implementation of these interventions can reduce potentially
preventable hospitalizations and decrease disparities for
hypertension, congestive heart failure, and angina.

Limitations
The data provided in this report are subject to at least

three limitations. First, hospital administrative data might be
incomplete. Data from only two thirds of the 50 states were
usable in this analysis, and individual states might differ in how
conditions and race/ethnicity are coded. Coding of conditions

FIGURE 2. Number of potentially preventable hospitalizations* among adults aged ≥18
years, by income quartile† — United States, 2009

Source: Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project, Nationwide
Inpatient Sample, 2009.
* For diabetes, hypertension, congestive heart failure, angina without procedure, asthma, dehydration,

bacterial pneumonia, and urinary infections.
† Area income was divided into quartiles based on the mean household income by the patient’s ZIP

Code. Quartile 1 refers to the lowest income communities, and quartile 4 refers to the wealthiest
communities.

0

200

400

600

800

1,000

1,200

1 (lowest) 2 3 4 (highest)

N
o

. (
in

1
,0

00
s)

Income quartile

Excess

Expected at best rate

Supplement

142 MMWR / November 22, 2013 / Vol. 62 / No. 3

FIGURE 3. Rate* of potentially preventable hospitalizations† among adults aged ≥18 years, by race/ethnicity — United States, 2001–2009

Source: Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project, State Inpatient Databases disparities analytic file, 2001–2009.
* Per 100,000 population.
† For diabetes, hypertension, congestive heart failure, angina without procedure, asthma, dehydration, bacterial pneumonia, and urinary infections.
§ Persons of Hispanic ethnicity can be of any race or combination of races

0

500

1,000

1,500

2,000

2,500

3,000

3,500

2001 2002 2003 2004 2005 2006 2007 2008 2009

Ra
te

Year

Non-Hispanic white

Non-Hispanic black

Asian/Paci�c Islander

Hispanic§

by hospitals also might change over time.
Second, cost estimates capture only hospital
facility costs during the inpatient stay and do
not include costs of inpatient or outpatient
physician visits, including follow-up outpatient
care. Although incomplete, because hospital
facility costs are the largest type of health
expenditure, analyses over time and across
populations might be informative. Finally,
these analyses cannot address causality. Rather
than residence in low-income neighborhoods
contributing to poorer health and thus
requiring hospitalization, for some patients,
poorer health might lead to residence in low-
income neighborhoods. Moreover, whether
providing improved primary care to residents
of low-income neighborhoods can reduce their
rates of potentially preventable hospitalizations
to the rates experienced by residents of high-
income neighborhoods is unclear.

FIGURE 4. Number of potentially preventable hospitalizations* among adults aged ≥18
years, by race/ethnicity — United States, 2009

Source: Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project, State
Inpatient Databases disparities analytic file, 2009.
* For diabetes, hypertension, congestive heart failure, angina without procedure, asthma, dehydration,

bacterial pneumonia, and urinary infections.
† Persons of Hispanic ethnicity can be of any race or combination of races.

Excess

Expected at best rate

0

500

1,000

1,500

2,000

2,500

Non-Hispanic white Non-Hispanic black Hispanic† Asian/Paci�c Islander

N
o

. (
in

1
,0

00
s)

Race/Ethnicity

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 143

Conclusion
Potentially preventable hospitalizations are common and

costly. Identification of disparities in potentially preventable
hospitalizations rates is necessary to address communities
and groups that would benefit the most. Because residents
of low-income neighborhoods have the highest rates or
preventable hospitalizations, providing interventions among
low-income neighborhoods might yield the largest reductions
in hospitalizations.

Acknowledgment

This report is based in part on data provided by the partner
organizations that participated in the HCUP Nationwide Inpatient
Sample and the State Inpatient Databases. A list of these organizations
is available at http://www.hcup-us.ahrq.gov/db/hcupdatapartners.jsp.

References
1. Bindman AB, Grumbach K, Osmond D, et al. Preventable

hospitalizations and access to health care. JAMA 1995;274:305–11.
2. CDC. CDC Health disparities and inequalities report—United States,

2011. MMWR 2011;60 (Suppl; January 14, 2011).
3. CDC. Introduction: In: CDC health disparities and inequalities report—

United States, 2013. MMWR 2013;62(No. Suppl 3).

4. CDC. Potentially preventable hospitalizations—United States, 2004–
2007. In: CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

5. Coffey R, Barrett M, Houchens R, et al. Methods applying AHRQ
quality indicators to Healthcare Cost and Utilization Project (HCUP)
data for the tenth (2012) National Healthcare Quality Report (NHQR)
and National Healthcare Disparities Report (NHDR). Rockville, MD:
Agency for Healthcare Research and Quality. Available at http://www.
hcup-us.ahrq.gov/reports/methods/methods.jsp.

6. Chen J, Normand SL, Wang Y, Krumholz HM. National and regional
trends in heart failure hospitalization and mortality rates for Medicare
beneficiaries, 1998–2008. JAMA 2011;306:1669–78.

7. Will JC, Valderrama AL, Yoon PW. Preventable hospitalizations for
congestive heart failure: establishing a baseline to monitor trends and
disparities. Prev Chronic Dis 2012;9:110260. DOI: http://dx.doi.
org/10.5888/pcd9.110260.

8. Agency for Healthcare Research and Quality. 2011 national healthcare
quality report. Rockville, MD: Agency for Healthcare Research and
Quality; 2012. Available at http://www.ahrq.gov/research/findings/
nhqrdr/nhqr11/index.html.

9. American Diabetes Association. Economic costs of diabetes in the U.S.
in 2007. Diabetes Care 2008;31:596–615.

10. Dall TM, Roary M, Yang W, et al. Health care use and costs for
participants in a diabetes disease management program, United States,
2007–2008. Prev Chronic Dis 2011;8:A53.

11. Longworth DL. Accountable care organizations, the patient-centered
medical home, and health care reform: what does it all mean? Cleve Clin
J Med 2011;78:571–82.

Supplement

144 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
Hypertension is a major risk factor for heart disease and

stroke. As the first and fourth leading causes of death in the
United States, heart disease and stroke occur in approximately
30% of adults aged ≥18 years in the United States (1).
Disparities in the prevalence of hypertension among racial/
ethnic groups have persisted at least since 1960, with the
prevalence remaining highest among non-Hispanic black
adults (1–4). Blood pressure control among those with
hypertension can reduce the risk of subsequent cardiovascular
diseases (5). Among adults with hypertension, Mexican-
American persons born outside the United States, and persons
without health insurance had lower rates of blood pressure
control in 2005–2008 (3). Not only do non-Hispanic black
adults have higher rates of hypertension, but among those
with hypertension they also have lower rates of blood pressure
control than non-Hispanic white adults (2,3).

Healthy People 2020 includes objectives to reduce the
prevalence of hypertension among adults to 26.9% (objective
HDS-5.1) and to increase the prevalence of blood pressure
control among adults with hypertension to 61.2% (objective
HDS-12) (6). Further, in 2011, the U.S. Department of
Health and Human Services launched the Million Hearts
initiative, which is intended to bring together communities,
health systems, nonprofit organizations, federal agencies, and
private-sector partners from across the country to prevent
1 million heart attacks and strokes over the course of 5 years
(available at http://millionhearts.hhs.gov/index.html). Blood
pressure control is a part of the initiative in the prevention
of these adverse events. Therefore, hypertension prevalence
and blood pressure control among those with hypertension
are important indicators to monitor over time to identify
improvements or persistent challenges in vulnerable segments
of the U.S. population.

This analysis of hypertension and the discussion that follows
are part of the second CDC Health Disparities and Inequalities
Report (2013 CHDIR) (3).The 2011 CHDIR was the first
CDC report to assess disparities across a wide range of diseases,
behavioral risk factors, environmental exposures, social

determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (7). This report provides more current
information on the prevalence of hypertension and blood
pressure control among adults aged ≥18 years. The purposes of
this report on hypertension and controlled hypertension are to
discuss and raise awareness of differences in the characteristics
of persons with hypertension and controlled hypertension, and
to prompt actions to reduce disparities.

Methods
To estimate the age-adjusted prevalence of hypertension and

blood pressure control among adults aged ≥18 years by selected
demographic and health characteristics, CDC analyzed data
from the National Health and Nutrition Examination Survey
(NHANES) aggregated from two survey cycles: 2007–2008
and 2009–2010. NHANES is a national survey representative
of the U.S. civilian noninstitutionalized population. Details of
the NHANES survey methodology are available at http://www.
cdc.gov/nchs/nhanes/about_nhanes.htm. During 2007–2010,
the response rate among persons screened was 76.3%. Data
were analyzed for 11,782 participants who had adequate data
from the interview and examination components of the survey
necessary to determine hypertension status. Blood pressure
was determined by an average of up to three measurements
taken during a single examination. Hypertension was defined
as an average systolic blood pressure (SBP) ≥140 mmHg, an
average diastolic blood pressure (DBP) ≥90 mmHg, or if the
participant reported the current use of blood pressure lowering
medication. Blood pressure control was defined as an average
SBP <140 mmHg and an average DBP <90 mmHg among
persons with hypertension. Pregnant women were excluded.

Hypertension prevalence and control estimates were analyzed
by selected demographic and health characteristics: sex, age
group (18–44, 45–64, and ≥65 years), race and ethnicity, marital
status, educational attainment, country of birth, family income
to federal poverty threshold, health insurance status (for persons
aged 18–64 years), veteran status, diagnosed diabetes status,

Prevalence of Hypertension and Controlled Hypertension —
United States, 2007–2010

Cathleen D. Gillespie, MS1
Kimberly A. Hurvitz, MHS2

1National Center for Chronic Disease Prevention and Health Promotion, CDC
2National Center for Health Statistics, CDC

Corresponding author: Cathleen D. Gillespie, MS, Division for Heart Disease and Stroke Prevention, National Center for Health Statistics, CDC. Telephone:
770-488-5855; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 145

obesity status, and disability. Race was defined as white, black,
and Mexican American. Ethnicity was defined as Hispanic or
non-Hispanic. Educational attainment among adults aged ≥25
years was defined as follows: less than high school, high school
graduate or equivalent, some college or Associate of Arts (AA)
degree, and college graduate or above. Household income was
defined as family income to federal poverty threshold, as defined
by the Department of Health and Human Services poverty
guidelines (8), specific to family size and appropriate year and
state. Health insurance status was defined as having either private
or public insurance, or being uninsured. Obesity among adults
aged ≥20 years is defined as a body mass index ≥30 kg/m2
based on measured weight and height. Veteran status, diagnosed
diabetes,* and disability† status were self-reported (Table).

Disparities were measured as the deviations from a “referent”
category prevalence. The referent group was the group that
had the most favorable estimate for the variables used to assess
disparities during the period reported. Absolute difference
was measured as the simple difference between a population
subgroup estimate and the estimate for its respective reference
group. The relative difference, a percentage, was calculated by
dividing the absolute difference by the value in the referent
category and multiplying by 100.

Statistical analyses were weighted to account for the complex
survey design. Prevalence estimates, except those by age group,
were age adjusted to the 2000 U.S. standard population using
the direct method. Estimates of hypertension control that are
age-adjusted to the 2000 U.S. standard population tend to be
lower than those adjusted to the population with hypertension
because of the difference between the age distribution of
the general population and that of the population with
hypertension (9).

Results
During 2007–2010, the overall age-adjusted prevalence

of hypertension among persons aged ≥18 years was 29.6%
(Table). Among persons aged ≥18 years with hypertension,
the overall age-adjusted prevalence of blood pressure control
was 48.0%. Substantial differences (relative difference >10%)

in the prevalence of hypertension were indicated by age group,
race/ethnicity, educational attainment, country of birth, family
income, health insurance, diabetes, obesity, and disability status.
The highest rates of hypertension were observed among those
aged ≥65 years (71.6%), adults with diabetes (59.4%), and
non-Hispanic black adults (41.3%). Although the difference
in hypertension prevalence by sex was statistically significant,
the difference was not substantial. Hypertension prevalence
increased with age and decreased with increasing income level,
but no significant trend was observed by educational attainment.
Non-Hispanic blacks had a higher rate of hypertension
(41.3%) than non-Hispanic whites (28.6%) and Hispanics
(27.7%). Adults born in the United States had a higher rate
of hypertension (30.6%) than non-U.S.-born adults (25.7%).
Adults aged <65 years with public insurance had a higher rate
of hypertension (28.3%) than those with private insurance
(20.0%) and those with no insurance (20.4%). Persons with
diabetes had a significantly higher rate of hypertension than
those without diabetes (59.4% versus 27.7%), as did those who
were obese compared with those who were not (40.5% versus
25.0%) and those with a disability compared with those with
no disability (40.2% versus 29.0%).

Substantial differences in the prevalence of blood pressure
control were obser ved among all population groups
except veteran status. Among persons aged ≥18 years with
hypertension, rates of blood pressure control were lowest
among those without health insurance (27.9%), Mexican-
Americans (30.3%), those who were never married (34.9%),
and those born outside the United States (38.9%). Men, adults
aged 18–44 years, Hispanics, Non-Hispanic blacks, those who
were never married, non-U.S.-born, persons without health
insurance had a lower prevalence of hypertension control
than their counterparts. Men had a lower rate of hypertension
control than women (42.7%). Adults aged 18–44 years had a
lower rate of hypertension control (40.9%) than adults aged
45–64 years (53.3%) and 64 years and over (51.4%). The
rate of controlled blood pressure was lower among Hispanics
(34.4%) and non-Hispanic blacks (42.5%) than non-Hispanic
whites (52.6%). Non-U.S.-born adults had a lower rate of
hypertension control (38.9%) than U.S.-born adults (49.3%).
Adults aged <65 years with no insurance had a lower rate
of hypertension control (27.9%) than those with public
insurance (60.2%) or private insurance (50.6%). Controlled
hypertension was also lower among those classified as not
obese compared with those who were obese (41.4% versus
54.0%), persons without diabetes compared with those with
diabetes (45.4% versus 63.6%), and persons with no disability
compared with those with a disability (45.0% versus 59.3%).
Controlled hypertension was not linearly associated with age,
educational attainment, or income level.

* Persons with diagnosed diabetes are defined as those who have ever been told
by a health-care professional that they have diabetes. Persons without diabetes
are defined as those who have never been told by a health-care professional that
they have diabetes or have never been told that they have borderline diabetes.

† Persons classified as having a disability provided the answer ‘Yes’ to any of four
questions:

• Unable to work at a job or business because of a physical, mental, or emotional
problem

• Limited because of difficulty remembering or because of periods of confusion
• Limited in any activity because of a physical, mental or emotional problem
• Uses special equipment, such as a cane, a wheelchair a special bed, or a special

telephone.

Supplement

146 MMWR / November 22, 2013 / Vol. 62 / No. 3

Discussion
The prevalence of hypertension has remained consistent

over the past 10 years, at an overall rate of approximately 30%
(1,13). During 2007–2010, the prevalence of hypertension
by the analyzed demographic characteristics was highest
among those aged ≥65 years (71.6%) and among non-
Hispanic blacks (41.3%), two population groups known to
be disproportionately affected (1,3,12). Although the overall
prevalence of blood pressure control has improved over the
past 10 years, non-Hispanic blacks and Hispanics continue
to have lower prevalence of control than their non-Hispanic
white counterparts (8,12). Also consistent with other research,
the prevalence of hypertension was higher among those with

diagnosed diabetes, obese persons, and persons with disabilities
(3). However, all three of these groups had higher rates of
blood pressure control than their counterparts in 2007–2010
(63.6%, 54.0%, and 59.3%, respectively, versus 45.4%,
41.4%, and 45.0% among those without diagnosed diabetes,
obesity, and disabilities, respectively). This difference is likely
because of higher rates of treatment with medication among
these groups (5,14). In contrast, although the prevalence of
hypertension was lowest among those aged 18–44 years (9.8%),
the prevalence of blood pressure control was significantly lower
among this group than their older counterparts. This is most
likely because of lower rates of hypertension awareness and
treatment with medication among younger adults (13,15).

See table footnotes on the next page.

TABLE. Age-adjusted* prevalence of hypertension and blood pressure control among adults aged ≥18 years, by selected demographic and
health characteristics — National Health and Nutrition Examination Survey, United States, 2007–2010

Characteristic

Hypertension§ Blood pressure control¶

Sample
Size† % (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%) % (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Total 11,782 29.6 (28.6–30.7) 48.0 (44.6–51.4)
Sex

Male 5,854 30.5 (29.0–31.9) 1.9** 6.6 42.7 (38.3–47.2) -12.8** -23.1
Female 5,928 28.6 (27.4–29.7) Ref. Ref. 55.5 (51.8–59.3) Ref. Ref.

Age group (yrs), unadjusted††
18–44 5,051 9.8 (8.9–10.7) Ref. Ref. 40.9 (34.4–47.5) -12.4** -23.3
45–64 3,854 40.4 (37.9–43.0) 30.6** 312.2 53.3 (49.8–56.8) Ref. Ref.
≥65 2,877 71.6 (68.4–74.7) 61.8** 630.6 51.4 (48.2–54.6) -1.9 -3.6
Race/Ethnicity

White, non-Hispanic 5,559 28.6 (27.1–30.2) 0.9 3.2 52.6 (48.8–56.5) Ref. Ref.
Black, non-Hispanic 2,305 41.3 (39.1–43.5) 13.6** 49.1 42.5 (37.6–47.5) -10.1** -19.2
Hispanic§§ 3,372 27.7 (26.4–29.1) Ref. Ref. 34.4 (30.7–38.2) -18.2** -34.6
Mexican American 2,121 27.5 (25.8–29.2) NA NA 30.3 (26.1–34.5) -22.3** -42.4

Marital status (persons aged ≥20 years)
Never married 1,885 31.8 (29.4–34.3) 1.7 5.6 34.9 (29.1–40.6) -15.7** -31.0
Married or living with partner 6,678 30.1 (28.6–31.6) Ref. Ref. 50.6 (46.8–54.5) Ref. Ref.
Divorced/separated or widowed 2,656 31.1 (29.2–33.1) 1.0 3.3 50.4 (42.3–58.4) -0.2 -0.4

Educational attainment (persons aged ≥25 years)††
Less than high school 3,127 36.9 (34.5–39.4) 8.9** 31.8 41.8 (33.9–49.6) -10.8 -20.5
High school graduate or equivalent 2,422 36.3 (34.2–38.4) 8.3** 29.6 51.6 (45.6–57.7) -1.0 -1.9
Some college or AA degree 2,677 34.5 (32.7–36.4) 6.5** 23.2 49.3 (44.3–54.2) -3.3 -6.3
College graduate or higher 2,096 28.0 (25.3–30.6) Ref. Ref. 52.6 (46.0–59.2) Ref. Ref.

Country of birth
United States 8,784 30.6 (29.5–31.7) 4.9** 19.1 49.3 (45.7–52.9) Ref. Ref.
Outside of the United States 2,993 25.7 (24.2–27.1) Ref. Ref. 38.9 (32.4–45.5) -10.4** -21.1

Family income to federal poverty threshold††,¶¶ (%)
<100 2,359 32.8 (30.6–34.9) 5.2 18.8 46.2 (38.0–54.3) -6.9 -13.0
100–199 2,940 32.5 (30.9–34.1) 4.9** 17.8 42.0 (34.7–49.4) -11.1 -20.9
200–399 2,777 30.6 (28.8–32.5) 3.0 10.9 53.1 (47.1–59.2) Ref. Ref.
400–499 840 28.0 (25.0–31.0) 0.4 1.4 45.7 (35.2–56.3) -7.4 -13.9
≥500 1,773 27.6 (25.1, 30.1) Ref. Ref. 51.4 (46.6–56.2) -1.7 -3.2

Health insurance status*** (persons aged 18–64 years)
Private 4,555 20.0 (18.4–21.5) Ref. Ref. 50.6 (46.3–54.9) -9.6** -15.9
Public 1,489 28.3 (25.6–30.9) 8.3 41.5 60.2 (51.8–68.6) Ref. Ref.
Uninsured 2,829 20.4 (18.2–22.6) 0.4 2.0 27.9 (21.7–34.2) -32.3** -53.7

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 147

Limitations
The findings in this report are subject to at least four

limitations. First, NHANES data are restricted to the civilian
noninstitutionalized population; thus, results from this study
are not generalizable to persons who live in nursing homes,
long-term care facilities, or prisons, or to military personnel.
Second, reliable data were not available for persons of certain
racial/ethnic groups or sexual orientation/gender identity.
Only non-Hispanic blacks and Hispanics were oversampled;
consequently, estimates cannot be calculated for other racial/
ethnic populations (e.g., American Indians/Alaska Natives,

Asians/Pacific Islanders). Third, the cross-sectional study
design provides a one-time only assessment of blood pressure,
although blood pressure can be measured multiple times
during one visit. This one-time assessment can overestimate
or underestimate hypertension prevalence. However, the
standardized measurement of blood pressure in a mobile
examination center makes NHANES the best source of
national data on hypertension. Finally, this report does not
examine time trends in disparities to assess progress toward
eliminating disparities. Although other studies included time
trends, only a limited number of demographic characteristics
such as race/ethnicity, age, and sex were examined (10).

TABLE. (Continued) Age-adjusted* prevalence of hypertension and blood pressure control among adults aged ≥18 years, by selected demographic
and health characteristics — National Health and Nutrition Examination Survey, United States, 2007–2010

Characteristic

Hypertension§ Blood pressure control¶

Sample
Size† (%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%) (%) (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Veteran status
Yes 1,473 30.8 (26.9–34.8) 1.1 3.7 52.5 (44.4–60.5) Ref. Ref.
No (referent) 10,307 29.7 (28.7–30.7) Ref. Ref. 47.4 (43.4–51.5) -5.1 -9.7

Diagnosed diabetes status††† 
Yes 1,421 59.4 (54.1–64.7) 31.7 114.4 63.6 (56.2–71.1) Ref. Ref.
No (referent) 10,352 27.7 (26.6–28.8) Ref. Ref. 45.4 (41.7–49.0) -18.2** -28.6

Obesity status§§§ (persons aged ≥20 years)
Yes 4,197 40.5 (39.0–41.9) 15.5 62.0 54.0 (50.2–57.8) Ref. Ref.
No (referent) 6,890 25.0 (23.5–26.4) Ref. Ref. 41.4 (36.5–46.3) -12.6** -23.3

Disability¶¶¶
Yes 2,612 40.2 (37.6–42.9) 11.2 38.6 59.3 (53.2–65.3) Ref. Ref.
No (referent) 8,613 29.0 (27.8–30.1) Ref. Ref. 45.0 (41.2–48.8) -14.3** -24.1

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent; NA = not applicable.
* Age adjusted to the 2000 U.S. standard population. Age specific data are not age adjusted. Hypertension prevalence data (except those by education status, health

insurance coverage, diabetes status, and age group) are age adjusted to the following seven age groups: 18–29, 30–39, 40–49, 50–59, 60–69, 70–79, and ≥80 years.
Data by health insurance status are age adjusted using the age groups 18–29, 30–39, 40–49, 50–59, and 60–64 years. Data by diabetes status are age adjusted
using the age groups 18–49, 50–59, 60–69, 70–79, and ≥80 years. Blood pressure control data (except those by education status, health insurance coverage, and
age group) are age adjusted to the following five age groups: 18–49, 50–59, 60–69, 70–79, and ≥80 years. Data by education status are age adjusted using the age
groups 25–49, 50–59, 60–69, 70–79, and ≥80 years. Data by health insurance status are age adjusted using the age groups 18–49, 50–59, and 60–64 years.

† Pregnant women were excluded.
§ Hypertension among adults is defined as an average systolic blood pressure ≥140 mmHg, an average diastolic blood pressure ≥90 mmHg, or self-reported current

use of blood pressure lowering medication.
¶ Blood pressure control is defined as an average systolic blood pressure <140 mmHg and an average diastolic blood pressure <90 mmHg among adults with

hypertension.
** p<0.05 for absolute difference compared with referent group, with Bonferroni adjustment for demographic variables with more than two categories.
†† p<0.05, test of trend for hypertension prevalence by income and age using weighted least squares regression on the categorical variable; not significant by

education or for controlled hypertension.
§§ Persons of Hispanic ethnicity might be of any race or combination of races.
¶¶ Family income: income of all persons within a household who are related to each other by blood, marriage, or adoption. Family income to federal poverty threshold:

the ratio of family income to the federal poverty threshold as defined by the Department of Health and Human Services’ (HHS) poverty guidelines, specific to
family size, as well as the appropriate year and state.

*** Private health insurance: private health insurance or Medigap insurance. Public health insurance: Medicare, Medicaid, State Children’s Health Insurance Program,
military health care, state-sponsored health plan, or other government insurance.

††† Persons with diagnosed diabetes: those who have ever been told by a health-care professional that they have diabetes. Persons without diabetes: those who have
never been told by a health-care professional that they have diabetes or have never been told that they have borderline diabetes.

§§§ Obesity: body mass index ≥30 kg/m2 based on measured weight and height.
¶¶¶ Disability: inability to work at a job or business because of a physical, mental, or emotional problem; limitation caused by difficulty remembering or periods of

confusion; limitation in any activity because of a physical, mental, or emotional problem; or use of special equipment (e.g., a cane, wheelchair, special bed, or
special telephone).

Supplement

148 MMWR / November 22, 2013 / Vol. 62 / No. 3

Conclusion
Consistent with the 2011 CHDIR and other studies, no

change has occurred in the prevalence of hypertension over the
last decade, although the rate of hypertension control continues
to improve (2,3,10). Disparities in hypertension prevalence
and control persist among most population groups assessed
similar to what has been published elsewhere. Although rates
of control have continued to show improvement (2,10), more
time is needed to determine whether the population will meet
the Healthy People 2020 target of 61.2%. Certain subgroups
of persons with hypertension exhibit even lower rates of blood
pressure control, indicating a need for interventions that span
the population and focus on vulnerable subgroups. The United
States Preventive Services Task Force (USPSTF) recommends
blood pressure screening for all adults aged ≥18 years, and as
a result of provisions in the Patient Protection and Affordable
Care Act (ACA), Medicare now covers certain adult clinical
preventive services recommended by the USPSTF without
patient cost sharing (§4103) (11,12). The law also requires
that “nongradfathered” private health plans include these
same services without cost sharing (§1001). In addition, the
ACA ensures certain preventive and wellness services without
cost-sharing for Medicare recipients (§4103), a group most
in need of hypertension management. The national Million
Hearts initiative endeavors to increase the number of persons
in the United States whose hypertension is under control by 10
million, as part of its goal to prevent 1 million heart attacks and
strokes by the year 2017. The Guide to Community Preventive
Services Task Force recommends system interventions to
improve blood pressure control, including clinical decision
support systems, reducing out-of-pocket costs for CVD
preventive services for patients with hypertension, and team-
based care. Because the rate of blood pressure control is lowest
among persons without health insurance, compared to those
with insurance coverage, it will be important to monitor this
and other vulnerable population groups in the future.

References
1. CDC. Vital signs: prevalence, treatment, and control of hypertension—

United States, 1999–2002 and 2005–2008. MMWR 2011;60:103–8.
2. CDC. National Center for Health Statistics. Health, United States,

2012: with special feature on emergency care. Hyattsville, MD: US
Department of Health and Human Services; 2013.

3. CDC. Prevalence of hypertension and controlled hypertension—United
States, 2005–2008. MMWR 2011:60;94–97. In: CDC. CDC health
disparities and inequalities report—United States, 2011. MMWR 2011,
60(Suppl; January 14, 2011).

4. Burt VL, Cutler JA, Higgins M, et al. Trends in the prevalence, awareness,
treatment, and control of hypertension in the adult US population. Data
from the health examination surveys, 1960 to 1991. Hypertension
1995;26:60–9

5. National Heart, Lung, and Blood Institute. Seventh report of the Joint
National Committee on Prevention, Detection, Evaluation, and
Treatment of High Blood Pressure. Hypertension 2003;42:1206–52.

6. US Department of Health and Human Services. Healthy people 2020.
Available at http://www.healthypeople.gov/2020/default.aspx.

7. CDC. Introduction: In: CDC Health disparities and inequalities report—
United States, 2011. MMWR 2011;60(Suppl; January 14, 2011).

8. US Department of Health and Human Services. Poverty Guidelines,
Research, and Measurement. Available at http://aspe.hhs.gov/poverty/
index.cfm.

9. Crim MT, Yoon SS, Ortiz E, et al. National surveillance definitions for
hypertension prevalence and control among adults. Circ Cardiovasc
Qual Outcomes 2012;5:343–51.

10. Yoon S, Burt V, Louis T, Carroll MD. Hypertension among adults in
the United States, 2009–2010. NCHS data brief, no 107. Hyattsville,
MD: National Center for Health Statistics; 2012.

11. Agency for Healthcare Research and Quality: Screening for high blood
pressure. Available at http://www.uspreventiveservicestaskforce.org/
uspstf07/hbp/hbprs.htm.

12. Patient Protection and Affordable Care Act. Public. Law. 111-148 111th
Congress, March 23, 2010. Government Printing Office, 2010. Available
at http://www.gpo.gov/fdsys/pkg/PLAW-111publ148/pdf/PLAW-
111publ148.pdf.

13. Yoon S, Ostchega Y, Louis T. Recent trends in the prevalence of high blood
pressure and its treatment and control, 1999-2008. NCHS data brief, no
48. Hyattsville, MD: National Center for Health Statistics; 2010.

14. Bertoia ML, Waring ME, Gupta PS, et al. Implications of new
hypertension guidelines in the United States. Hypertension 2012;
60:639–44.

15. Egan BM, Zhoa Y, Axon RN. US trends in prevalence, awareness, treatment,
and control of hypertension, 1988-2008. JAMA 2010;303:2043–50.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 149

Tuberculosis — United States, 1993–2010
Awal D. Khan, PhD, Elvin Magee, MPH, Gail Grant

National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC

Corresponding author: Awal D. Khan, PhD, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC. Telephone 404-639-6272;
E-mail: [email protected].

Introduction
Tuberculosis (TB) is transmitted via the airborne route by

person-to-person contact. Although TB is a leading cause
of death on a global scale (1), most cases can be cured with
treatment. From 1993 to 2010, the number of TB cases
reported in the United States decreased from 25,103 to
11,182. Despite the decrease, TB continues to affect many
communities in the United States disproportionately and
unequally, especially racial/ethnic minorities and foreign-
born persons (2). TB remains one of many diseases and
health conditions with large disparities and inequalities by
income, race/ethnicity, educational attainment, and other
sociodemographic characteristics (3).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (4) was
the first CDC report to assess disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic
presented in this report is based on criteria that are described
in the 2013 CHDIR Introduction (5). This report provides
new information on TB, a topic not covered in the 2011
CHDIR. The purposes of this TB report are to discuss and
raise awareness of differences in the characteristics of people
who have TB in the United States and to prompt actions to
reduce these disparities.

Methods
Tuberculosis (TB) is a disease caused by bacteria that is spread

from person to person through the air when a TB sufferer
coughs, sneezes, speaks, sings, or laughs. TB usually affects
the lungs, but it can also affect other parts of the body, such as
the brain, the kidneys, or the spine. This analysis included all
TB cases, and no cases or latent TB infection (LTBI). To assess
disparities in newly reported cases of TB disease among persons
of all ages in the United States, CDC analyzed 1993–2010
data from the National TB Surveillance System (NTSS). TB
is a nationally notifiable disease (2). Since 1953, state and local
health departments have submitted information to CDC on
each newly reported case of TB disease in the United States.
Currently, all 50 U.S. states and the District of Columbia
(DC) as well as Puerto Rico, the U.S. Virgin Islands, and six

other jurisdictions in the Pacific region (American Samoa, the
Commonwealth of the Northern Mariana Islands, the Federated
States of Micronesia, Guam, the Republic of the Marshall
Islands, and the Republic of Palau) report information on newly
diagnosed TB cases electronically using NTSS. The Report of
a Verified Case of TB (RVCT) form (http://ftp.cdc.gov/pub/
software/tims/2009%20rvct%20documentation/rvct%20
training%20materials/rvct%20instruction%20manual.pdf ),
which was released in 1993, was expanded to collect additional
information for each case, including human immunodeficiency
virus (HIV) status, occupation, and history of substance
abuse, homelessness, and drug susceptibility test results (2).
Subsequent revisions of the RVCT form in 2009 include risk
factors (e.g., diabetes, end-stage renal disease, and contact with
a drug-resistant person), residential status, immigration status,
and reasons for longer than usual TB therapy.

This report examines the number of TB cases and rates during
2006 and 2010 by patient-reported sex at birth, race/ethnicity,
country of birth, patient primary occupation, employment
status, number of years patient has been living in the United
States, and type of health-care provider. Race was defined as
white, black, Asian/Pacific Islander, and American Indian/
Alaska Native. Ethnicity was defined as Hispanic and non-
Hispanic. A person was considered U.S.-born or foreign-born
on the basis of definitions used in the 2010 TB surveillance
report (2). For employment status, a person was considered
unemployed if not employed during the 12 months preceding
TB diagnosis. During 1993–2008, occupation was assessed for
the previous 2 years and multiple choices were accepted, but
starting in 2009, occupation was assessed for 1 year before and
multiple choice answers were no longer accepted. Geographic
region was not analyzed because, in 2010, approximately half
(49.2%) of all TB cases were concentrated in a small number
of states (California, Florida, Texas, and New York), in which
67.5% of cases occurred in foreign-born persons (2).

Trends in TB rates during 1993–2010 by race/ethnicity
and cases by country of birth are presented. TB case rates
per 100,000 population and by sex and race/ethnicity were
calculated using population estimates from the U.S. Census
Bureau’s Federated Electronic Research, Review, Extra, and
Tabulation Tool (DataFerrett version 1.3.3), which were
available during 2006–2010. The 2010 Current Population
Survey was used to obtain population estimates stratified by

Supplement

150 MMWR / November 22, 2013 / Vol. 62 / No. 3

country of birth (6). Disparities were measured as deviations
from a “referent” category rate or percentage. Referent
categories were selected because they demonstrated the most
favorable group estimates for the variables used to assess
disparities during 2006 and 2010 (7–8). For example, non-
Hispanic white was selected as the referent category for the
racial/ethnic variable. Absolute difference was measured as the
simple difference between a group estimate and the estimate
for its respective reference category, or referent group. Relative
difference, a percentage, was obtained by dividing the absolute
difference by the value in the referent category and multiplying
by 100. To evaluate changes in disparity over time, relative
differences for the groups in 2006 were subtracted from relative
differences in 2010. No statistical testing was performed.

Results
During 2006–2010, a total of 62,642 verified TB cases were

reported to CDC’s NTSS from the 50 states, DC, Puerto Rico,
the U.S. Virgin Islands, and six other jurisdictions in the Pacific
region. Of these, 13,732 were reported in 2006 and 11,182
were reported in 2010. The national TB case rate was 4.6 cases
per 100,000 population in 2006 and 3.6 cases per 100,000
population in 2010, a 20% decline over 5 years. The rate for
males was 5.8 in 2006 and 4.5 in 2010.

The relative difference between males and females in reported
TB rates was 70.6% in 2006 and 66.7% in 2010 (Table). From
2006 to 2010, the changes in relative differences for the various
age groups were as follows: persons aged 15–24 years (22.9%),
persons aged 25–44 years (40.0%), persons aged 45–64 years
(44.3%), and persons aged ≥65 years (35.7%) (Table). In 2010,
the relative difference between persons aged ≥65 years and the
referent group was 450%.

From 2006 to 2010, all racial/ethnic minorities experienced
decreases in TB case rates (Table). In 2006, Hispanics had case
rates of 9.2 per 100,000, compared with 6.5 in 2010. Asians/
Pacific Islanders had a rate of 26.1 in 2006 and 22.4 in 2010.
Compared with whites, TB rates in 2010 were approximately
seven times higher among Hispanics, eight times higher among
blacks, and 25 times higher among Asians/Pacific Islanders.

During 2006–2010, 59% of 62,642 reported TB cases
occurred among foreign-born persons. In 2006, the relative
difference among foreign-born persons compared with U.S.-
born persons was 857% and in 2010, the relative difference
in reported TB rates among foreign-born persons compared
with U.S.-born persons was 1,031%. The change in the relative
difference from 2006 to 2010 was 175% (1,031% versus
856.5%, respectively) (Table).

Although racial/ethnic relative differences in TB case rates
were similar in both U.S.-born and foreign-born persons, the
magnitude of the relative disparities varied markedly between
U.S.-born and foreign-born persons and was three-to-four
times greater among foreign-born persons. In U.S.-born
persons in 2010, the relative difference in TB rates compared
with whites was 614% for blacks, 429% for Asians/Pacific
Islanders, 286% for Hispanics, and 757% for American
Indians/Alaska Natives (Table). Among foreign-born persons
in 2010, the relative difference in TB rates compared with
whites was 2,271% for Asians/Pacific Islanders, 1,771% for
blacks, and 836% for Hispanics.

Among 6,748 foreign-born persons in the United States
during 2010 with reported TB, approximately 21% received a
diagnosis of TB disease within <2 years of arrival in the United
States, approximately 18% within 2–5 years of arrival, and 50%
in >5 years after arrival; an additional 11% had no information
on arrival dates. The relative difference in TB cases diagnosed
>5 years after arrival in the United States compared with cases
diagnosed 2–5 years after arrival was 178.8% in 2010. The
change in the relative difference between 2006 to 2010 for TB
cases diagnosed >5 years after arrival in the United States was
41% (136.3% vs. 177.8%, respectively) (Table).

The proportion of TB cases among unemployed persons was
53% (7,245 of 13,732) in 2006 and 59% (6,217 of 10,520) in
2010. During 2010, the relative difference in reported TB cases
among unemployed persons compared with those employed
in fields other than health care (referent) was 74%. a change
in the relative difference of 44.2% over time (Table).

The relative difference in reported TB cases among persons
whose primary health-care provider for TB disease was a health
department compared with persons whose primary health-care
provider for TB disease was private/other providers (referent
category) was 217% a change in the relative difference of 109%
over time (326.3% in 2006 and 216.7% in 2010) (Table). The
proportion of TB cases treated at health departments was 81%
(10,830 of 13,308) in 2006 and 76% (4,587 of 6,011) in 2010.

From 1993 to 2010, TB case rates declined by approximately
63% (Figure 1). TB rates for Asians/Pacific Islanders were
41.2 per 100,000 population in 1993 and 22.4 per 100,000
in 2010, with differences in rates of 45.6%. From 1993 to
2010, among blacks, the rates ranged from 28.5 to 7.0 per
100,000 population, among Hispanics from 19.9 to 6.5,
among American Indians/Alaska Natives from 14.0 to 6.4,
and among non-Hispanic whites from 3.6 to 0.9.

From 1993 to 2010, the proportion of TB cases among
foreign-born persons increased from 29% to 60% (Figure 2).
From 1993 to 2010, the TB case rate in the United States has
declined annually in both U.S.-born and foreign-born persons;

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 151

TABLE. Reported tuberculosis rates,* by date and selected characteristics — United States, 2006 and 2010

Characteristic

2006 2010

TB rate†

Absolute
difference§

(percentage
points)

Relative
difference

(%)¶ TB rate†

Absolute
difference§

(percentage
points)

Relative
difference

(%)¶

Sex at birth
Male 5.8 2.4 70.6 4.5 1.8 66.7
Female 3.4 Ref Ref 2.7 Ref. Ref.

Age group (yrs)
<15 1.4 Ref. Ref. 1.0 Ref. Ref.
15–24 3.6 2.2 157.1 2.8 1.8 180.0
25–44 5.6 4.2 300.0 4.4 3.4 340.0
45–64 5.4 4.0 285.7 4.3 3.3 330.0
≥65 7.2 5.8 414.3 5.5 4.5 450.0
Race/Ethnicity

White, non-Hispanic 1.2 Ref. Ref. 0.9 Ref. Ref.
Black, non-Hispanic 10.2 9.0 750.0 7.0 6.1 677.8
Hispanic** 9.2 8.0 666.7 6.5 5.6 622.2
Asian/Pacific Islander 26.1 24.9 2,075.0 22.4 21.5 2,388.9
American Indian/Alaska Native 7.2 6.0 500.0 6.4 5.5 611.1

Country of birth
Born in United States 2.3 Ref. Ref. 1.6 Ref. Ref.
Born outside United States 22.0 19.7 856.5 18.1 16.5 1,031.3

Born in the United States
White, non-Hispanic 0.9 Ref. Ref. 0.7 Ref. Ref.
Black, non-Hispanic 7.7 6.8 755.6 5.0 4.3 614.3
Hispanic 3.9 3.0 333.3 2.7 2.0 285.7
Asian/Pacific Islander 3.4 2.5 277.8 3.7 3.0 428.6
American Indian/Alaska Native†† 7.9 7.0 777.8 6.0 5.3 757.1

Born outside the United States
White, non-Hispanic 1.7 Ref. Ref. 1.4 Ref. Ref.
Black, non-Hispanic 36.3 34.6 2,035.2 26.2 24.8 1,771.4
Hispanic 17.5 15.8 929.4 13.1 11.7 835.7
Asian/Pacific Islander 38.3 36.6 2,152.9 33.2 31.8 2,271.4
American Indian/Alaska Native†† 1.3 -0.4 -23.5 1.9 0.5 35.7

Years in the United States (foreign-born)§§
<2 years 27 8 42.1 21 3 16.7
2–5 years 19 Ref. Ref. 18 Ref. Ref.
>5 years 45 26 136.8 50 32 177.8

Patient’s primary occupation¶¶
Unemployed/no occupation 53 12 29.3 59 25 73.5
Health-care worker 3 -38 -92.7 4 -30 -88.2
Other employment¶¶,*** 41 Ref. Ref. 34 Ref. Ref.
Unknown 3 3

Health-care provider type§§
Any health department 81 62 326.3 76 52 216.7
Private/other providers 19 Ref. Ref. 24 Ref. Ref.

Abbreviations: Ref. = referent group; TB = tuberculosis.
* This analysis included all TB cases, and no cases of latent TB infection (LTBI).
† Per 100,000 U.S. standard population; based on the U.S. Census Bureau’s Federated Electronic Research, Review, Extra, and Tabulation tool (DataFerrett version

1.3.3) that were available during 2006–2010.
§ Absolute difference = the simple difference between a particular group rate and the rate for its respective referent group.
¶ Obtained by expressing the value for the difference as a percentage of the estimate for its respective referent group.
** Persons of Hispanic ethnicity might be of any race or combination of races.
†† Small sample size (n = 4) and uncertainty of data quality, mostly misclassification.
§§ Includes available information for Patient’s “Month-Year” Arrived in the United States.
¶¶ Used proportions of all reported and available information for Patient’s Primary Occupation and Type of Health Care Provider. During 1993–2008, occupation was

assessed for the previous 2 years and multiple choices were accepted, but starting in 2009, occupation was assessed for 1 year before and multiple choice answers
were no longer accepted.

*** Includes migrant and seasonal workers.

Supplement

152 MMWR / November 22, 2013 / Vol. 62 / No. 3

overall, TB cases have declined 78% among U.S.-born persons
compared with 47% among foreign-born persons.

In 1993, approximately 69% of reported TB cases occurred
among U.S.-born persons (7.4 cases per 100,000) and 29%
occurred among foreign-born persons (34.0 cases per 100,000).
In comparison, during 2006–2010, on average, approximately
59% of reported TB cases occurred among foreign-born
persons and remained relatively stable, and the rates of cases
reported were 1.9 per 100,000 for U.S.-born and 22.0 for
foreign-born persons.

Discussion
The number of new TB cases reported in the United

States in 2010 represented an 87% decrease since reporting
began in 1953 and a 58% decrease since the peak resurgence
of TB reported in 1992 (2). Despite the downward trend,
TB continues to affect many U.S. racial/ethnic minorities
disproportionately, both U.S.-born and foreign-born.
Approximately half of new TB cases in the United States occur
among foreign-born persons and the TB rate in foreign-born
persons was approximately 10 times that of persons born in
the United States. This disparity has become more recognizable
since 1993, when surveillance was enhanced to include routine
collection of country of birth information (9). The foreign-
born population presents a challenge to health-care staff and

TB programs for providing diagnosis and care, and these
challenges include the unequal prevalence of TB risk factors
and barriers to access to TB care.

Several factors contribute to the disproportionate prevalence
of TB among racial/ethnic and foreign-born minorities.
Persons who were born in countries where TB morbidity
is high might have acquired TB before immigrating and
not have symptoms of active TB disease until after arrival
in the United States. Different social and environmental
living conditions create large and predictable differences in
health outcomes among nations and between population
groups within nations (10). In the United States, adjusting
for six socioeconomic indicators (i.e., crowding, income,
poverty, public assistant, education, and unemployment),
low socioeconomic status accounted for approximately half
of the increased risk for TB among blacks, Hispanics, and
Native Americans (11). Unequal prevalence of TB risk factors
(e.g., HIV infection, homelessness, incarceration, substance
use, and TB disease severity) among racial/ethnic groups
also might contribute to increased exposure to TB or to an
increased risk for developing TB once infected. Economically
disadvantaged persons, the uninsured, low-income children,
the elderly, the homeless, those with HIV, and those with other
chronic health conditions (e.g., diabetes and severe mental
illness) encounter barriers to accessing health-care services.
The effects of these barriers on TB prevention and control

* Cases per 100,000 population, from the U.S. Census Bureau’s Federated Research Review, Extra, and Tabulation too (DataFerrett version 1.3.3, available during
2006–2010 .

† All races are non-Hispanic. In 2003, the Asian/Pacific Islander category included persons who reported race as Asian only and/or Native Hawaiian or Other Pacific
Islander only. Updated on July 21, 2011.

FIGURE 1. Tuberculosis rates,* by race/ethnicity† — United States, 1993–2010

0

10

20

30

40

50

1993 1996 1999 2002 2005 2008 2010

White

Black or African-American

Hispanic

American Indian/Alaska Native

Asian/Paci�c Islander

Year

Ra
te

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 153

vary across racial/ethnic groups (12–14). In addition, poverty,
language barriers, and immigration status also can be additional
barriers to ameliorating TB disparities and inequality, jointly
or independently (15–17).

Controlling and preventing TB in the United States
necessitates addressing disparities among racial/ethnic
minorities and foreign-born persons. The continuous arrival
of new immigrants and refugees from countries with a high
prevalence of TB has impeded elimination efforts. Reduction
in TB rates among foreign born communities can be
accomplished by identification of local at-risk populations,
increased knowledge of issues affecting immigrants and foreign-
born persons and modification of existing TB programs to meet
the needs of these communities.  In particular, training and
education can aid health-care staff serving the foreign-born
community at risk for TB disease.

Limitations
The findings in this report are subject to at least four

limitations. First, certain data (e.g., race/ethnicity and years
in the United States) were incomplete and did not include
U.S. territories and the U.S.-affiliated Pacific Islands. Second,

the analysis does not assess the effects of socioeconomic risk
factors (e.g., homelessness, substance abuse, and incarceration),
HIV coinfection, and drug resistance on TB disparities. The
prevalence of certain risk factors is particularly extensive in
minority groups (e.g., persons with HIV/AIDS and diabetes).
Third, educational attainment and family or household
income, two indicators used commonly to explain health
disparities and inequalities, were not available. Finally, social
aspects that include language barriers and cultural differences
with respect to health-seeking behaviors and the ability to
access the complex U.S. health-care system were not examined.

Conclusion
Progress toward TB elimination in the United States will

require ongoing surveillance and improved TB control and
prevention activities to address persistent disparities between
U.S.-born and foreign-born persons and between whites and
racial/ethnic minorities. Disparities and inequalities among
racial/ethnic minorities are affected by many unmeasured
factors. CDC recommends improving awareness, testing, and
treatment of latent infection and TB disease in minorities and
foreign-born populations to reduce TB (9).

FIGURE 2. Number and and percentage of tuberculosis cases, by origin of birth — United States, 1993–2010

0

10

20

30

40

50

60

70

80

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

20,000

1993 1994 1995 1996 1997 1998 1999 2000

Year

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

%
o

f casesN
o

. o
f c

as
es

Foreign-born

U.S.-born

Foreign-born

U.S.-born

Supplement

154 MMWR / November 22, 2013 / Vol. 62 / No. 3

References
1. Mandell G, Bennett J, Dolin R. Mandell, Douglas, and Bennett’s

principles and practice of infectious diseases. 7th ed. Philadelphia, PA:
Churchill Livingstone; 2009.

2. CDC. Reported tuberculosis in the United States, 2010. Atlanta, GA:
US Department of Health and Human Services, CDC; 2011. Available
at http://www.cdc.gov/tb/statistics/reports/2010/default.htm.

3. Keppel KG, Pearcy JN, Wagener DK. Trends in racial and ethnic-specific
rates for the health status indicators: United States, 1990–1998. Healthy
People 2000 Stat Notes 2002; 23:1–6.

4. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

5. CDC. Introduction: In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

6. US Census Bureau. Current population survey, March supplements (2006–
2010). DataFerrett. Available at http://www.thedataweb.org/faq.html.

7. Keppel K, Pamuk E, Lynch J, et al. Methodological issues in measuring
health disparities. Hyattsville, MD: US Department of Health and
Human Services, CDC. Vital Health Stat 2 2005; 141:1–16.

8. US Census Bureau. Current population survey: design and methodology.
Washington, DC: US Census Bureau; 2006. Technical Paper No. 6.

9. CDC. Report of Verified Case of Tuberculosis (RVCT) instruction
manual. Atlanta, GA: US Department of Health and Human Services,
CDC; 2009. Available at http://ftp.cdc.gov/pub/software/tims/2009%20
rvct%20documentation/rvct%20training%20materials/rvct%20
instruction%20manual.pdf.

10. Agency for Health Care Research Quality. National health care disparities
report. Publ No. 11-0004; 2010. Available at http://www.ahrq.gov/
research/findings/nhqrdr/nhdr10/pdf/nhdr10.pdf.

11. Cantwell MF, McKenna MT, McCray E, Onorato IM. Tuberculosis and
race/ethnicity in the United States: impact of socioeconomic status. Am
J Respir Crit Care Med 1998;157: 1016–20.

12. CDC; Infectious Disease Society of America. Controlling tuberculosis
in the United States—Recommendations from the American Thoracic
Society. MMWR 2005;54(No. RR-12).

13. Nahid P, Horne DJ, Jarlsberg LG, et al. Racial differences in tuberculosis
infection in United States communities: the coronary artery risk
development in young adult study. CID 2011;53: 291–4.

14. Serpa JA, Teeter LD, Musser JM, Gravis EA. Tuberculosis disparity
between US-born backs and whites, Houston, Texas, USA. Emerg Infect
Dis 2009;15: 899–904.

15. Lonnroth K, Castro KG, Chakaya JM, et al. Tuberculosis control and
elimination 2010–50: cure, care, and social development. Lancet
2010;375:1814–29.

16. Craig GM, Booth H, Story A, The impact of social factors on tuberculosis
management. J Adv Nurs 2007;58:418–24.

17. Lonnroth K, Jaramillo E, Williams BG, Dye C, Raviglione MC. Drivers
of tuberculosis epidemics: the role of risk factors and social determinants.
Soc Sci Med 2009;68: 2240–6.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 155

Health Outcomes: Mortality

Supplement

156 MMWR / November 22, 2013 / Vol. 62 / No. 3

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 157

Introduction
Heart disease and stroke are the first and fourth leading

causes of death, respectively in the United States (1,2). In
2008, heart disease and stroke were responsible for nearly a
third of all deaths in the United States (30.4%), killing more
than three-quarters of a million people that year (1). Coronary
heart disease (CHD) is the cause of more than two-thirds of
all heart disease-related deaths (1,2). One of the Healthy People
2020 objectives includes reducing the rate of CHD deaths
by 20% from the baseline rate of 126 deaths per 100,000
population per year, to a goal of 100.8 deaths per 100,000
(objective HDS-2) (3). The objectives also include reducing
the rate of stroke deaths by 20% over the baseline of 42.2
deaths per 100,000, to a goal of 33.8 deaths per 100,000
population. Although the rates of death from both CHD and
stroke have declined continuously in recent decades and the
Healthy People 2010 goals for these two objectives were met
among the overall U.S. population in 2004, the death rates
remain high, particularly among men and blacks (4–6).

This heart disease and stroke analysis and discussion that
follows is part of the second CDC Health Disparities and
Inequalities Report (2013 CHDIR) (6). The 2011 CHDIR
(7) was the first CDC report to assess disparities across a
wide range of diseases, behavioral risk factors, environmental
exposures, social determinants, and health-care access. The
topic presented in this report is based on criteria described
in the 2013 CHDIR Introduction (8). This report provides
more current information on CHD and stroke deaths among
different age and racial/ethnic groups. The purposes of the
coronary heart disease and stroke mortality report are to
discuss and raise awareness of differences in the characteristics
of persons dying from coronary heart disease and stroke, and
to prompt actions to reduce disparities.

Methods
To examine the number and age-specific CHD and stroke

mortality rates of persons of all ages, by sex, age group, and race/
ethnicity, CDC analyzed final 2009 data from the National
Vital Statistics System (NVSS). NVSS data are described in

detail elsewhere (http://www.cdc.gov/nchs/nvss.htm). Race
was defined as white, black, American Indian/Alaska Native
(AI/AN), and Asian/Pacific Islander (A/PI). Ethnicity was
defined as Hispanic or non-Hispanic. Sociodemographic
information beyond age, sex, and race/ethnicity is not available
in the NVSS.

CDC estimated the number of deaths and the rate of
death per 100,000 population for which coronary heart
disease or stroke were the underlying cause of death (ICD-
10 codes I20–I25 for CHD, I60–I69 for stroke), and 95%
confidence intervals were calculated based on a Poisson
distribution, consistent with NCHS methodology (1). Rates
per 100,000 population were age-adjusted to the 2000 U.S.
standard population, except where stratified by age group (9).
Disparities were measured as the deviations from a “referent”
category rate and by characteristics that included sex, age,
and race/ethnicity. Absolute difference was measured as the
simple difference between a population subgroup estimate
and the estimate for its respective reference group. The
relative difference, a percentage, was calculated by dividing
the difference by the value in the referent category and
multiplying by 100. Significant differences between rates were
determined by nonoverlapping 95% confidence intervals. All
tests for differences in age-adjusted death rates were significant
compared with the indicated referent group after Bonferroni
adjustment for multiple comparisons.

Results
The age-adjusted rate of death from CHD in 2009 was 116.1

per 100,000 population (Table), and CHD was listed as the
underlying cause of death in 386,324 persons in the United
States. The age-adjusted death rate per 100,000 population
from CHD was higher among men than women (155.8 versus
86.2) and higher among non-Hispanic blacks (141.3) than
among any other racial/ethnic group. The rate of premature
death (death among persons aged <75 years) was higher
among non-Hispanic blacks than their white counterparts
(65.5 versus 43.2).

Similar differences were observed for deaths from stroke,
which was listed as the underlying cause of death in 128,842

Coronary Heart Disease and Stroke Deaths — United States, 2009
Cathleen D. Gillespie, MS
Charles Wigington, MPH

Yuling Hong, MD,
National Center for Chronic Disease Prevention and Health Promotion, CDC

Corresponding author: Cathleen D. Gillespie, MS, Division for Heart Disease and Stroke Prevention, National Center for Chronic Disease Prevention and
Health Promotion, CDC. Telephone: 770-488-5855; E-mail: [email protected].

Supplement

158 MMWR / November 22, 2013 / Vol. 62 / No. 3

persons in the United States in 2009, an age-adjusted rate of
38.9 deaths per 100,000 population. The age-adjusted death
rate per 100,000 population from stroke was higher among
non-Hispanic blacks (73.6) than among any other racial/ethnic
group. The rate of premature death (death among persons aged
<75 years) from stroke was higher among non-Hispanic blacks
than their white counterparts (25.0 versus 10.2).

Discussion
Although death rates from CHD and stroke are declining

overall (4), disparities still remain in the rate of death from
these events between racial/ethnic groups. The premature death
rate from CHD and stroke continues to be higher among black
adults than their white counterparts. The Healthy People 2020

See table footnotes on the next page.

TABLE. Number of deaths and age-adjusted death rates* from persons within coronary heart disease† and stroke§ by sex, age, and race/ethnicity
— National Vital Statistics System, United States, 2009

Coronary heart disease† No.

Age-adjusted (except where noted)

Rate* (95% CI)

Absolute
difference

(rate)

Relative
difference

(%)

Total 386,324 116.1 (115.7–116.5)
Male¶ 210,069 155.8 (155.2–156.5) Ref. Ref.
Female 176,255 86.2 (85.8–86.6) -69.6 -44.7

Age in years
<45 6,679 3.9 (3.8–4.0) -129.3 -97.1
0–24 150 0.1 (0.1–0.2)
25–44 6,529 8.4 (8.2–8.6)
45–74¶ 131,632 133.2 (132.5–133.9) Ref. Ref.
45–54 23,285 52.2 (51.5–52.9)
55–64 46,018 132.3 (131.1–133.5)
65–74 62,329 299.8 (297.4–302.2)
<75 138,311 43.5 (43.3–43.8)
≥75 247,990 1,245.80 (1,240.8–1,250.7) 1112.6 835.3
≥85 (crude) 143,204 2543.3 (2,530.1–2,556.5)
Race/Ethnicity

Hispanic** 20,228 86.5 (85.3–87.7) -31.2 -26.5
Non-Hispanic 365,119 118.2 (117.8–118.6)

White, non-Hispanic¶ 315,810 117.7 (117.3–118.1) Ref. Ref.
Black, non-Hispanic 39,956 141.3 (139.9–142.8) 23.6 20.1
American Indian/Alaska Native 1,737 92 (87.5–96.5) -25.7 -21.8
Asian/Pacific Islander 7,616 67.3 (65.8–68.8) -50.4 -42.8

Age in years/race-ethnicity
<45 years

Hispanic** 561 1.9 (1.7–2.0) -2.2 -53.7
Non-Hispanic 6,094 4.3 (4.2–4.4)

White, non-Hispanic¶ 4,459 4.1 (4.0– 4.3) Ref. Ref.
Black, non-Hispanic 1,369 6.2 (5.9–6.5) 2.1 51.2

45–74
Hispanic** 8,176 98 (95.8–100.2) -33.5 -25.5
Non-Hispanic 122,907 136.1 (135.3–136.9)

White, non-Hispanic¶ 99,389 131.5 (130.7–132.3) Ref. Ref.
Black, non-Hispanic 19,820 199.5 (196.7–202.3) 68 51.7

<75
Hispanic** 8,737 31.3 (30.7–32.0) -11.9 -27.5
Non-Hispanic 129,001 44.7 (44.5–45.0)

White, non-Hispanic¶ 103,848 43.2 (42.9–43.5) Ref. Ref.
Black, non-Hispanic 21,189 65.5 (64.6–66.4) 22.3 51.6

≥75
Hispanic** 11,490 945.7 (928.4–963.0) -331.6 -26
Non-Hispanic 236,100 1,262.9 (1,257.8–1,268.1)

White, non-Hispanic¶ 211,949 1,277.3 (1,271.8–1,282.8) Ref. Ref.
Black, non-Hispanic 18,763 1,322.8 (1,303.9–1,341.8) 45.5 3.6

≥85
Hispanic** 5,793 1,787.5 (1,741.5–1,833.5) -841.1 -32
Non-Hispanic 137,234 2,586.1 (2,572.4–2,599.8)

White, non-Hispanic¶ 125,303 2,628.6 (2,614.0–2,643.2) Ref. Ref.
Black, non-Hispanic 9,085 2,555.2 (2,502.7–2,607.7) -73.4 -2.8

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 159

objectives and goals for heart disease and stroke are intended
to reduce premature deaths by promoting prevention of these
events and reducing their recurrence. In 2011, CDC launched
the Million Hearts initiative, which is intended to bring together

communities, health systems, nonprofit organizations, federal
agencies, and private-sector partners from across the country to
prevent 1 million heart attacks and strokes over 5 years.

TABLE. (Continued) Number of deaths and age-adjusted death rates* from persons within coronary heart disease† and stroke§ by sex, age, and
race/ethnicity — National Vital Statistics System, United States, 2009

Stroke§  No.

Age-adjusted (except where noted)

Rate* (95% CI)

Absolute
difference

(rate)

Relative
difference

(%)

Total 128,842 38.9 (38.7–39.1)
Male¶ 52,073 39.7 (39.3–40.0) Ref. Ref.
Female 76,769 37.8 (37.5–38.1) -1.9 -4.8

Age in years
<45 2,914 1.6 (1.6–1.7) -33.4 -95.4
0–24 461 0.4 (0.4–0.5)
25–44 2,453 3.1 (3.0–3.2)
45–74¶ 34,264 35 (34.6–35.3) Ref. Ref.
45–54 (crude) 6,163 13.8 (13.5–14.1)
55–64 (crude) 10,523 30.2 (29.6–30.8)
65–74 (crude) 17,578 84.5 (83.3–85.7)
<75 37,178 11.9 (11.7–12.0)
≥75 91,660 460.1 (457.1–463.1) 425.1 1,214.6
≥85 (crude) 53,253 945.8 (937.8–953.8)
Race/Ethnicity

Hispanic** 7,065 29.5 (28.8–30.2) -8.3 -22.0
Non-Hispanic 121,540 39.5 (39.3–39.7)

White, non-Hispanic¶ 101,703 37.8 (37.5–38.0) Ref. Ref.
Black, non-Hispanic 15,718 55.7 (54.8–56.6) 17.9 47.4
American Indian/Alaska Native 533 29.8 (27.2–32.4) -8 -21.2
Asian/Pacific Islander 3,586 31.6 (30.6–32.7) -6.2 -16.4

Age in years/race-ethnicity
<45 years

Hispanic 498 1.5 (1.4–1.6) 0.2 1.5
Non-Hispanic 2,406 1.7 (1.6–1.7) 0.4 3.1

White, non-Hispanic¶ 1,439 1.3 (1.3–1.4) Ref. Ref.
Black, non-Hispanic 796 3.5 (3.2–3.7) 2.2 16.9

45–74
Hispanic 2,654 31.5 (30.3–32.7) 1.2 0.4
Non-Hispanic 31,506 35.2 (34.8– 35.6)

White, non-Hispanic¶ 22,699 30.3 (29.9–30.7) Ref. Ref.
Black, non-Hispanic 7,338 73.6 (71.9–75.3) 43.3 142.9

<75
Hispanic 3,152 10.7 (10.3–11.1) 0.5 0.5
Non-Hispanic 33,912 12 (11.8–12.1)

White, non-Hispanic¶ 24,138 10.2 (10.1–10.3) Ref. Ref.
Black, non-Hispanic 8,134 25 (24.4–25.5) 14.8 14.5

≥75
Hispanic 3,913 322.5 (312.4–332.6) -144 -30.9
Non-Hispanic 87,624 468.3 (465.1–471.4)

White, non-Hispanic¶ 77,562 466.5 (463.5–469.8) Ref. Ref.
Black, non-Hispanic 7,584 534.5 (522.5–546.5) 68 14.6

≥85 (crude)
Hispanic 1,901 586.6 (560.2–613.0) -387.3 -39.8
Non-Hispanic 51,292 966.6 (958.2–975.0)

White, non-Hispanic¶ 46,426 973.9 (965.0–982.8) Ref. Ref.
Black, non-Hispanic 3,619 1,017.9 (984.7–1,051.1) 44 4.5

Abbreviations: 95% CI = 95% confidence interval; Ref = referent.
* Per 100,000. Directly standardized to the 2000 U.S. standard population, except where stratified by age.
† ICD-10 codes: I20–I25
§ ICD-10 codes: I60–I69
¶ All tests for differences in age-adjusted death rates were significant compared with the indicated referent group after Bonferroni adjustment for multiple comparisons.
** Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

160 MMWR / November 22, 2013 / Vol. 62 / No. 3

Limitations
The findings in this report are subject to at least three

limitations. First, misclassification of race and ethnicity of the
decedent on the death certificate might underestimate rates
among AI/ANs, A/PIs, and Hispanics (10). Second, results
from a study in New York City, New York, indicated that
CHD is overreported as a cause of death on death certificates
(11). However, these results might be specific to New York
City. Third, the death rates reflect only the underlying cause
of death and no other contributing causes of death such as
diabetes, which vary substantially across racial/ethnic groups.

Conclusion
Risk factors for cardiovascular disease include tobacco use,

physical inactivity, poor diet, diabetes, obesity, hypertension,
and dyslipidemia. Preventing or controlling hypertension
and high low-density lipoprotein (LDL) cholesterol have
been shown to greatly reduce the risk for stroke and CHD,
respectively (12,13). In 2011, the U.S. Department of Health
and Human Services launched the Million Hearts initiative to
prevent 1 million heart attacks and strokes by the year 2017,
through focused clinical and policy strategies. The Guide
to Community Preventive Services Task Force recommends
system interventions to improve CVD risk factors, including
clinical decision support systems, reducing out-of-pocket costs
for CVD preventive services for patients with hypertension
and high cholesterol, and team-based care to improve blood
pressure control. The United States Preventive Services Task
Force (USPSTF) recommends blood pressure screening for all
adults aged ≥18 years and LDL-cholesterol screening for adults
in certain sex, age, and heart disease risk groups (14,15). As a
result of provisions in the Patient Protection and Affordable
Care Act, USPSTF-recommended clinical preventive services
covered by Medicare now have no patient cost sharing (§4104)
(16,17). The law also requires that “nongrandfathered” private
health plans include these same services without cost sharing
(§1001) and encourages Medicaid to cover them through an
increase in the federal matching rate for those services (§4006).
Because the rates of premature death from CHD and stroke are
higher among blacks, it will be important to monitor this and
other vulnerable population groups (i.e., those with limited
access to regular medical care) to determine if improvements
are evident in the future.

References
1. Miniño AM, Murphy SL, Xu J, Kochanek KD. Deaths: Final data for

2008. Hyattsville, MD: US Department of Health and Human Services,
CDC, National Center for Health Statistics; 2011. National Vital
Statistics Report, vol 59, no 10. Available at http://www.cdc.gov/nchs/
data/nvsr/nvsr59/nvsr59_10.pdf.

2. Murphy SL, Jiaquan Xu J, Kochanek KD. Deaths: Preliminary Data for
2010. Hyattsville, MD: US Department of Health and Human Services,
CDC, National Center for Health Statistics; 2011. National Vital
Statistics Report, vol 59, no 10. Available at http://www.cdc.gov/nchs/
data/nvsr/nvsr60/nvsr60_04.pdf.

3. US Department of Health and Human Services. Healthy people 2020.
Available at http://www.healthypeople.gov/2020/default.aspx.

4. Hoyert DL. 75 Years of mortality in the United States, 1935–2010.
Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics; 2012. NCHS Data Brief no 88.
Available at http://www.cdc.gov/nchs/data/databriefs/db88.pdf.

5. CDC. Healthy people data 2010: the healthy people database. Atlanta,
GA: US Department of Health and Human Services, CDC: 2010.
Available at http://wonder.cdc.gov/data2010.

6. CDC. Coronary heart disease and stroke deaths—United States, 2006.
In: CDC. CDC health disparities and inequalities report. MMWR
2011;60(Suppl; January 14, 2011).

7. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60(Suppl; January 14, 2011).

8. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3)

9. Klein RJ, Schoenborn CA. Age adjustment using the 2000 projected
U.S. population. Hyattsville, MD: US Department of Health and
Human Services, CDC, National Center for Health Statistics; 2001.
NCHS Statistical Notes no 20. Available at http://www.cdc.gov/nchs/
data/statnt/statnt20.pdf.

10. Arias E, Schauman WS, Eschbach K, Sorlie PD, Backlund E. The validity
of race and Hispanic origin reporting on death certificates in the United
States. National Center for Health Statistics. Vital Health Stat
2(148);2008.

11. Agarwal R, Norton JM, Konty K, et al. Overreporting of deaths from
coronary heart disease in New York City hospitals, 2003. Prev Chronic
Dis 2010:A47. Available at http://www.cdc.gov/pcd/issues/2010/
may/09_0086.htm.

12. Chobanian AV, Bakris GL, Black HR, et al. Seventh report of the Joint
National Committee on Prevention, Detection, Evaluation, and
Treatment of High Blood Pressure. Hypertension 2003;43:1206–52.

13. National Institutes of Health. Third report of the National Cholesterol
Education Program (NCEP) Expert Panel on Detection, Evaluation,
and Treatment of High Blood Cholesterol in Adults (Adult Treatment
Panel III). Executive Summary. Heart, Lung, and Blood Institute.
National Institutes of Health. NIH Publication No. 01–3670. May
2001. Available at http://www.nhlbi.nih.gov/guidelines/cholesterol/
atp3_rpt.htm.

14. Agency for Healthcare Research and Quality. Screening for high blood
pressure. Available at http://www.uspreventiveservicestaskforce.org/
uspstf07/hbp/hbprs.htm.

15. Agency for Healthcare Research and Quality. Screening for lipid disorders
in adults. Available at http://www.uspreventiveservicestaskforce.org/
uspstf/uspschol.htm.

16. US Department of Health and Human Services. Patient Protection and
Affordable Care Act of 2010. Pub. L. No. 114–48 (March 23, 2010),
as amended through May 1, 2010. Available at http://www.healthcare.
gov/law/full/index.html.

17. Koh HK, Sebelius KG. Promoting prevention through the Affordable
Care Act. New Engl J Med 2010;363:1296–9.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 161

Introduction
Drug-induced deaths include all deaths for which drugs are

the underlying cause (1), including those attributable to acute
poisoning by drugs (drug overdoses) and deaths from medical
conditions resulting from chronic drug use (e.g., drug-induced
Cushing’s syndrome). A drug includes illicit or street drugs
(e.g., heroin and cocaine), as well as legal prescription and
over-the-counter drugs; alcohol is not included. Deaths from
drug overdose have increased sharply in the past decade. This
increase has been associated with overdoses of prescription
opioid pain relievers, which have more than tripled in the past
20 years, escalating to 16,651 deaths in the United States in
2010 (2). Most drug-induced deaths are unintentional drug
poisoning deaths, with suicidal drug poisoning and drug
poisoning of undetermined intent comprising the majority
of the remainder (3).

This drug-induced deaths analysis and discussion that
follows are part of the second CDC Health Disparities and
Inequalities Report (CHDIR) (3). The 2011 CHDIR (4) was
the first CDC report to assess disparities across a wide range of
diseases, behavioral risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (5). This report provides more current
information to what was presented in the 2011 CHDIR (3). The
purpose of this drug-induced deaths analysis is to raise awareness
of disparities by age, gender, racial/ethnic and/or geographic
differences, and to prompt actions to reduce disparities.

Methods
To determine differences in the prevalence of drug-induced

deaths by sex, race/ethnicity, age, and geographic region in the
United States, CDC analyzed 2010 data from the mortality
component of the National Vital Statistics System (NVSS).
To examine patterns of drug-induced death rates by age group
and race/ethnicity, NVSS data from 1999 through 2010 were
aggregated because limited sample sizes are available for any
single year for certain groups.

Death certificates provide information on the decedent’s age,
sex, race, ethnicity, and geographic region. They do not provide

information on decedent income, disability, or language
spoken at home. Race is categorized as white, black, American
Indian/Alaska Native, or Asian/Pacific Islander. Ethnicity was
categorized as Hispanic or non-Hispanic. Geographic location
is categorized as Northeast, Midwest, South, and West.*
Adverse effects from drugs taken as directed and infections
resulting from drug use are not included.

The number of drug-induced deaths are presented and
unadjusted (crude) drug-induced death rates per 100,000
population are calculated for 2010 by age, racial/ethnic group,
sex, and geographic region (based on the U.S. Census 2010
population survey) (Table). The 95% confidence intervals (CIs)
for unadjusted drug-induced death rates are based on ≥100
deaths and were calculated using a normal approximation; CIs
based on <100 deaths were calculated using a gamma method.
(Additional information is available from Vital Statistics of the
United States: Mortality, 1999 Technical Appendix, available
at http://wonder.cdc.gov/wonder/sci_data/mort/mcmort/
type_txt/mcmort05/techap99.pdf ).

Results
During 2010 (the year in which the latest national NVSS

mortality data are available), a total of 40,393 drug-induced
deaths occurred in the United States. The majority of drug-
induced deaths were unintentional 74.3%; remainder: 13.1%;
suicidal drug poisoning; 7.3% drug poisoning of undetermined
intent; 5.1% mental and behavioral disorders from drug use;
<1% homicide; <1% medical conditions from chronic drug
use. Drug-induced mortality was highest among persons aged
40–49 years (25.1) (Table). Rates for males exceeded those for
females aged ≥10 years. Rates were lowest in the Northeast
region of the United States (11.6), and the largest percentage
of cases was in the South (38.2%). Non-Hispanic whites

Drug-Induced Deaths — United States, 1999–2010
Karin A. Mack, PhD

National Center for Injury Prevention and Control, CDC

Corresponding author: Karin A. Mack, Division of Analysis, Research, and Practice Integration, National Center for Injury Prevention and Control, CDC.
Telephone: 770-488-4389; E-mail: [email protected].

* Northeast (Connecticut, Maine, Massachusetts, New Hampshire, New Jersey,
New York, Pennsylvania, Rhode Island, and Vermont), Midwest (Illinois,
Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North
Dakota, Ohio, South Dakota, and Wisconsin), South (Alabama, Arkansas,
Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana,
Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee,
Texas, Virginia, and West Virginia), and West (Alaska, Arizona, California,
Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah,
Washington, and Wyoming).

Supplement

162 MMWR / November 22, 2013 / Vol. 62 / No. 3

accounted for 82.1% of all 40,393 drug-induced deaths. The
highest rates were among American Indians/Alaska Natives
(17.1) and non-Hispanic whites (16.6).

During 1999–2010, drug-induced death rates by race/
ethnicity and age group demonstrated varying patterns by
racial/ethnic group, although the highest rate occurred in the
40–49 year age group for non-Hispanic whites, American
Indians/Alaska Natives, and non-Hispanic blacks (Figure).
Rates among American Indians/Alaska Natives were highest
in the 30–39 and 40–49 year age groups and then decreased
in the older ages. Rates among non-Hispanic blacks increased
dramatically in persons aged 40–49, remained high in persons
aged 50–59, and then decreased. Rates were lowest at all ages
for Asians/Pacific Islanders.

Discussion
American Indians/Alaska Natives and non-Hispanic whites

had the highest drug-induced death rates overall. This finding is
consistent with the previous report for rates during 2003–2007
(6). However, it does reflect a change from the 1980s and
1990s, when drug-induced mortality rates were higher among
blacks than whites (3). Prescribed drugs have replaced illicit
drugs as a leading cause of drug-related overdose deaths (7).
Non-Hispanic blacks are less likely than non-Hispanic whites
to use prescription drugs, and therefore might be less likely to
misuse such drugs (8).

TABLE. Number and rate* of drug-induced deaths, by age group, race/ethnicity, and sex — National Vital Statistics System, United States, 2010

Characteristic

Female Male Total

Percentage
of total

No. of
deaths Rate (95% CI)†

No. of
deaths Rate (95% CI)†

No. of
deaths Rate (95% CI)†

Total 16,017 10.2 (10.0–10.4) 24,376 16.1 (15.9–16.3) 40,393 13.1 (13.0–13.2) 100.00
Age group (yrs)
0–9 32 0.2 (0.1–.02) 48 0.2 (0.2–0.3) 80 0.2 (0.2–0.2) 0.0
10–19 258 1.2 (1.1–1.4) 636 2.9 (2.7–3.1) 894 2.1 (2.0–2.2) 2.2
20–29 1,943 9.2 (8.8–9.6) 4,788 22.1 (21.5–22.7) 6,731 15.8 (15.4–16.1) 16.7
30–39 2,978 14.8 (14.3–15.3) 5,115 25.5 (24.8–26.2) 8,093 20.2 (19.7–20.6) 20.0
40–49 4,620 21.0 (20.4–21.6) 6,333 29.3 (28.6–30.0) 10,953 25.1 (24.7–25.6) 27.1
50–59 4,240 19.7 (19.1–20.3) 5,474 26.8 (26.0–27.5) 9,714 23.1 (22.7–23.6) 24.0
60–69 1,258 8.2 (7.8–8.7) 1,447 10.4 (9.9–10.9) 2,705 9.2 (8.9–9.6) 6.7
70–79 373 4.1 (3.7–4.5) 314 4.2 (3.8–4.7) 687 4.1 (3.8–4.4) 1.7
≥80 314 4.4 (3.9–4.9) 218 5.3 (4.6–6.0) 532 4.7 (4.3–5.1) 1.3
Geographic region§

Northeast 2,245 7.9 (7.6–8.2) 4,154 15.5 (15.0–15.9) 6,399 11.6 (11.3–11.9) 15.8
Midwest 3,480 10.2 (9.9–10.6) 5,298 16.1 (15.71–6.5) 8,778 13.1 (12.8–13.4) 21.7
South 6,243 10.7 (10.4–11.0) 9,202 16.4 (16.1–16.7) 15,445 13.5 (13.3–13.7) 38.2
West 4,049 11.2 (10.9–11.6) 5,722 16 (15.5–16.4) 9,771 13.6 (13.3–13.9) 24.2

Race/Ethnicity
White, non-Hispanic 13,456 13.2 (13.0–13.4) 19,689 20.0 (19.7–20.3) 33,145 16.6 (16.4–16.7) 82.1
Black, non-Hispanic 1,332 6.5 (6.1–6.8) 2,170 11.5 (11.0–12.0) 3,502 8.9 (8.6–9.2) 8.7
American Indian/

Alaska Native
200 15.3 (13.2–17.4) 239 19.0 (16.5–21.4) 439 17.1 (15.5–18.7) 1.1

Asian/Pacific Islander 129 1.5 (1.3–1.8) 205 2.7 (2.3–3.0) 334 2.1 (1.8–2.3) 0.8
Hispanic¶ 844 3.4 (3.2–3.6) 1,944 7.6 (7.3–7.9) 2,788 5.5 (5.3–5.7) 6.9
Unknown** 56 — — 129 — — 185 — — 0.5

Abbreviation: 95% CI = 95% confidence interval.
* Unadjusted (crude) death rates per 100,000 population.
† CIs based on ≥100 deaths were calculated using a normal approximation; CIs based on <100 deaths were calculated using a gamma method. (Additional information

available from Vital Statistics of The United States: Mortality, 1999 Technical Appendix. Available at http://wonder.cdc.gov/wonder/sci_data/mort/mcmort/type_txt/
mcmort05/techap99.pdf ).

§ Northeast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, and Vermont. Midwest: Illinois, Indiana, Iowa,
Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin. South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia. West: Alaska,
Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** Rates for persons with unknown race/ethnicity were not included because population data were unavailable.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 163

Limitations
The findings in this report are subject to at least two

limitations. First, overdose deaths are likely underestimated
because lengthy investigations are often required. This
sometimes results in a “pending manner and cause of death”
category being selected at the close of the mortality file. Second,
injury mortality data might underestimate the actual number
of deaths for American Indians/Alaska Natives and certain
other racial/ethnic populations (e.g., Hispanics) because of
the misclassification of race/ethnicity of decedents on death
certificates (9).

Conclusion
Preventing drug-induced deaths will require change at many

levels (10). Improving prescription drug monitoring programs,
which are electronic databases that track prescriptions for
opioid pain relievers and other controlled prescription drugs
in a state, can assist with identification of improper prescribing
and use of these drugs. Health insurers and pharmacy benefit
managers can develop prescription claims review programs
to identify and address improper prescribing and use of pain
relievers. Health-care providers can follow guidelines for
responsible prescribing, including screening and monitoring

for substance abuse and mental health problems. Patients
also should be encouraged to use prescription pain relievers
only as directed by a health-care provider, and store and
dispose of them properly (http://www.cdc.gov/injury/pdfs/
NCIPC_Overview_FactSheet_PPO-a.pdf ).

References
1. Miniño AM, Murphy SL, Xu J, Kochanek KD. Deaths: final data for

2008. Natl Vital Stat Rep 2011;59:1–126.
2. Jones CM, Mack KA, Paulozzi LJ. Pharmaceutical overdose deaths,

United States, 2010. JAMA 2013;309:657–9.
3. Paulozzi LJ, Annest JL. US data show sharply rising drug-induced death

rates. Inj Prev 2007;13:130–2.
4. CDC. CDC health disparities and inequalities report—United States,

2011. MMWR 2011;60 (Suppl; January 14, 2011).
5. CDC. Introduction. In: CDC health disparities and inequalities report—

United States, 2013. MMWR 2013;62(Suppl No. 3).
6. CDC. Drug-induced deaths—United States, 2003–2007. MMWR 2011;

60 (Suppl; January 14, 2011).
7. Paulozzi LJ, Budnitz DS, Xi Y. Increasing deaths from opioid analgesics

in the United States. Pharmacoepidemiol Drug Saf 2006;15:618–27.
8. CDC. QuickStats: Percentage of persons reporting use of at least one

prescription drug during the preceding month, by sex and race/
ethnicity—United States, 1999–2002. MMWR 2006;55:15.

9. Arias E, Schauman WS, Eschbach K, Sorlie PD, Backlund E. The validity
of race and Hispanic origin reporting on death certificates in the United
States. Vital Health Stat 2 2008:1–23.

10. CDC. Vital signs: overdoses of prescription opioid pain relievers—United
States, 1999–2008. MMWR 2011;60:1487–92.

FIGURE. Drug-induced death rates,* by race/ethnicity and age group — National Vital Statistics System, United States, 1999–2010

* Crude death rates per 100,000 population.

0

5

10

15

20

30

0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 ≥80

Ra
te

*

Age group (yrs)

Non-Hispanic white
Non-Hispanic black
American Indian/
Alaska Native
Asian/Paci�c Islander
Hispanic

25

Supplement

164 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
According to 1981–2009 data, homicide accounts for

16,000–26,000 deaths annually in the United States and
ranks within the top four leading causes of death among U.S.
residents aged 1–40 years (1). Homicide can have profound
long-term emotional consequences on families and friends of
victims and on witnesses to the violence (2,3), as well as cause
excessive economic costs to residents of affected communities
(1,4). For years, homicide rates have been substantially higher
among certain populations. Previous reports have found that
homicides are higher among males (5–7), adolescents and
young adults (6), and certain racial/ethnic groups, such as
non-Hispanic blacks, non-Hispanic American Indian/Alaska
Natives (AI/ANs), and Hispanics (6–9). The 2011 CDC
Health Disparities and Inequalities Report (CHDIR) described
similar findings for the year 2007 (10). For example, the 2011
report showed that the 2007 homicide rate was highest among
non-Hispanic blacks (23.1 deaths per 100,000), followed by
AI/ANs (7.8 deaths per 100,000), Hispanics (7.6 deaths per
100,000), non-Hispanic whites (2.7 deaths per 100,000), and
Asian/Pacific Islanders (A/PIs) (2.4 deaths per 100,000) (10).
In addition, non-Hispanic black men aged 20–24 years were
at greatest risk for homicide in 2007, with a rate that exceeded
100 deaths per 100,000 population (10). Other studies have
reported that community factors such as poverty and economic
inequality and individual factors such as unemployment and
involvement in criminal activities can play a substantial role
in these persistent disparities in homicide rates (11). Public
health strategies are needed in communities at high risk for
homicide to prevent violence and save lives.

The homicide analysis and discussion that follow are part of
the second CHDIR and update information presented in the
first CHDIR (10). The 2011 CHDIR (12) was the first CDC
report to take a broad view of disparities across a wide range
of diseases, behavioral risk factors, environmental exposures,
social determinants, and health-care access. The topic presented
in this report is based on criteria that are described in the 2013
CHDIR Introduction (13). The purposes of this homicide
report are to discuss and raise awareness of differences in
homicide rates by some of these characteristics and to prompt
actions to reduce these disparities.

Methods
To assess disparities in homicide rates in the United States,

CDC analyzed data from the CDC National Vital Statistics
System (NVSS), the same data source used for the 2011
CHDIR on homicides (10). In this report, NVSS data provided
as of February 2012 were used. NVSS collects death certificate
data filed in the 50 states and the District of Columbia (DC)
(14). Death certificates provide information on the decedent’s
age, sex, race, ethnicity, and geographic region. They do not
provide information on decedent income, sexual orientation,
disability, or language spoken at home.

This report summarizes the homicide data by providing the
number, proportion, and rates of homicides by age, sex, and
race/ethnicity for the year 2009; providing the homicide rates
by U.S. state for the year 2009; and comparing 2009 with
2007 data. Data in this report are based on homicides caused
by any mechanism. More details on homicide rates by age,
sex, and race/ethnicity for each state and census region can be
accessed through the CDC Web-based Injury Statistics Query
and Reporting System — Fatal (WISQARS Fatal) Injury Data
module (available at http://www.cdc.gov/injury/wisqars/index.
html). Data on individual and socioeconomic risk factors for
homicide were unavailable for analysis. In addition, sufficient
data were not available to assess disparities by certain racial/
ethnic subgroups, household income, disability status, and
sexual orientation. NVSS codes racial categories as white, black,
AI/AN, and A/PI, and ethnicity is coded separately as Hispanic
or non-Hispanic (14). In this report, references to whites,
blacks, AI/ANs, and A/PIs refer to non-Hispanic persons.
Hispanics might be of any race or combination of races. Crude
homicide rates per 100,000 population were calculated by age,
sex, and race/ethnicity, as well as by the combination of these
three variables. Crude rates per 100,000 population by state
in 2009 also are provided. Confidence intervals (CIs) of rates
were calculated in two ways: 1) groupings of annual death
counts of <100 were calculated by using a gamma estimation
method (14), and 2) groupings of annual death counts of ≥100
were calculated by using a normal approximation approach.
Rates calculated from <20 deaths were considered unreliable
and are not reported.

Homicides — United States, 2007 and 2009
Joseph E. Logan, PhD

Jeffrey Hall, PhD
Dawn McDaniel, PhD

Mark R. Stevens, MSPH, MA
National Center for Injury Prevention and Control, CDC

Corresponding author: Joseph E. Logan, National Center for Injury Prevention and Control, CDC. Telephone: 770-488-1529; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 165

Disparities were measured as deviations from a referent
category rate or prevalence. The group with the largest
population of the U.S. census data in each demographic
category was used as the referent (e.g., females, non-Hispanic
whites, or persons aged 30–49 years). Absolute difference
was measured as the simple difference between a population
subgroup estimate and the estimate for its respective reference
group. The relative difference as a percentage was calculated
by dividing the absolute difference by the value in the referent
category and multiplying by 100. Relative differences in rates
between each race/ethnic category were also stratified by
sex and age. Rate comparisons were considered significantly
different if they had nonoverlapping 95% CIs, which provide
a conservative test for statistical significance.

Results
An estimated 18,361 homicides occurred in 2007 and

16,799 occurred in 2009 (Table 1). The relative rate difference
reported for males was at least 250% higher than that of
females in both data years. In addition, in each data year, the
relative rate difference for non-Hispanic blacks was at least
650% higher than the rate reported for non-Hispanic whites.

Non-Hispanic AI/ANs and Hispanics also had rates that far
exceeded those of non-Hispanic whites in both years. Rates
were highest among persons aged 15–29 years both in 2007
and 2009 and then decreased with each subsequent age group;
however, the lowest rates reported in both years were among
children aged 0–14 years.

The homicide rate for the U.S. population in 2009 was
significantly lower than the U.S. homicide rate reported
in 2007. Differences in rates also occurred among certain
populations. Specifically, homicide rates were lower in 2009
than those reported in 2007 for males, non-Hispanic whites,
non-Hispanic blacks, Hispanics, persons aged 15–29 years, and
persons aged 30–49 years. None of the demographic groups
had significantly higher rates in 2009 compared with 2007.

Among males, the risk for homicide was greatest among
non-Hispanic blacks aged 15–29 years in both 2007 and 2009
(Table 2). Furthermore, for both years, the male homicide
rate was significantly higher among non-Hispanic blacks
than among those in other racial/ethnic groups in each age
category assessed, except among men aged 50–64 years, for
whom the 95% CIs overlapped with the rate for AI/ANs in
2009. Hispanic males had higher rates than non-Hispanic
white males in every age group among males aged ≥15 years
in both years as well.

TABLE 1. Number, percentage, and crude rate* of homicides, by sex, race/ethnicity, and age group — National Vital Statistics System, United
States, 2007 and 2009

Characteristic

2007 2009

No. of
deaths (%) Rate (95% CI)†

Absolute
difference

(percentage
points)

Relative
difference§

(%)
No. of
deaths (%) Rate (95% CI)

Absolute
difference

(percentage
points)

Relative
difference

(%)

Sex
Male 14,538 (79.2) 9.8 (9.6–10.0) 7.3 291.4 13,126 (78.1) 8.7 (8.5–8.8) 6.3 267.1
Female 3,823 (20.8) 2.5 (2.4–2.6) Ref. Ref. 3,673 (21.9) 2.4 (2.3–2.4) Ref. Ref.

Race/Ethnicity
White, non-Hispanic 5,512 (30.0) 2.7 (2.7–2.8) Ref. Ref. 5,163 (30.7) 2.6 (2.5–2.6) Ref. Ref.
Black, non-Hispanic 8,746 (47.6) 23.1 (22.6–23.6) 20.3 742.5 7,733 (46.0) 19.9 (19.5–20.3) 17.3 679.1
American Indian/Alaska Native 199 (1.1) 7.8 (6.7–8.9) 5.1 185.9 235 (1.4) 9.0 (7.9–10.2) 6.5 252.6
Asian/Pacific Islander 341 (1.9) 2.4 (2.2–2.7) -0.3 -11.9 325 (1.9) 2.2 (1.9–2.4) –0.4 -14.9
Hispanic¶ 3,466 (18.9) 7.6 (7.4–7.9) 4.9 178.3 3,179 (18.9) 6.6 (6.3–6.8) 4.0 157.1

Age group (yrs)
0–14 1,096 (6.0) 1.8 (1.7–1.9) -5.6 -75.7 998 (5.9) 1.6 (1.5–1.7) –5.2 -76.5
15–29 8,268 (45.0) 13.0 (12.8–13.3) 5.6 76.1 7,241 (43.1) 11.2 (10.9–11.4) 4.3 63.1
30–49 6,327 (34.5) 7.4 (7.2–7.6) Ref. Ref. 5,776 (34.4) 6.9 (6.7–7.0) Ref. Ref.
50–64 1,886 (10.3) 3.5 (3.4–3.7) -3.9 -52.6 1,906 (11.3) 3.4 (3.2–3.5) –3.5 -50.8
≥65 759 (4.1) 2.0 (1.9–2.1) -5.4 -73.0 860 (5.1) 2.2 (2.0–2.3) –4.7 -68.3
Total** 18,361 (100.0) 6.1 (6.0–6.2) — — 16,799 (100.0) 5.5 (5.4–5.6) — —

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Per 100,000 population.
† CIs based on <100 deaths were calculated using a gamma method, and those based ≥100 deaths were calculated using a normal approximation (Source: Xu J,

Kockanek KD, Murphy SL, Tejada-Vera B. Deaths: final data for 2007. Hyattsville, MD: US Department of Health and Human Services, CDC, National Center for Health
Statistics. Natl Vital Stat Rep 2010;58).

§ Relative differences were calculated based on rates that were estimated to five decimal places. Therefore, relative differences calculated based on the rates provided
in the table might differ from those displayed because of rounding.

¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** Total counts for 2007 include 97 deaths of persons of unknown race/ethnicity and 25 deaths of persons unknown age. Total counts for 2009 include 164 deaths of

persons of unknown race/ethnicity and 18 deaths of persons of unknown age.

Supplement

166 MMWR / November 22, 2013 / Vol. 62 / No. 3

Among females, the homicide rates also were generally
higher among racial/ethnic minorities (Table 2). For example,
in both years, female homicide rates were markedly higher
among non-Hispanic blacks than among non-Hispanic whites
in every age group <65 years. Female homicide rates also were
higher among Hispanic than among non-Hispanic whites in
every age group <30 years for both years. In 2009, the female
homicide rate was higher among AI/ANs than among non-
Hispanic whites aged 15–29 years as well.

Compared with 2007, homicide rates were significantly
lower in 2009 in certain demographic populations (Table 2).
Among non-Hispanic blacks, rates were significantly lower
among males aged 15–29, 30–49, and 50–64 years and women
aged 30–49 years. The homicide rates for each age category
among Hispanic males aged 15–49 years also were lower in
2009 than in 2007.

State-specific homicide rates for 2009 ranged from 1.1 to
12.8 deaths per 100,000 population, and rates were generally
higher in the southern states (Figure 1). Most states did not
have any significant changes in homicide rates from 2007 to
2009; however, 10 states experienced significant decreases:
Arizona, California, Florida, Georgia, Idaho, Maryland, New
Jersey, North Carolina, Ohio, and Pennsylvania. Decreases
in rates ranged from 12.4% in California to 55.8% in Idaho.
The 2009 crude homicide rate for DC was an estimated 22.8
per 100,000 population.

Discussion
Homicide rates are still particularly high among non-

Hispanic black, Hispanic, and non-Hispanic AI/AN

TABLE 2. Crude homicide rates,* by sex, age group, and race/ethnicity — National Vital Statistics System, United States, 2007 and 2009

Characteristic

2007 2009

Rate (95% CI)†
Relative difference§

(%) Rate (95% CI)
Relative difference

(%)

Male
0–14 yrs

White, non-Hispanic 1.3 (1.1–1.4) Ref. 1.2 (1.0–1.4) Ref.
Black, non-Hispanic 5.4 (4.7–6.0) 319.6 4.6 (4.0–5.2) 280.5
Hispanic¶ 1.8 (1.5–2.1) 39.8 1.5 (1.2–1.8) 21.9
American Indian/Alaska Native —** — — — — —
Asian/Pacific Islander — — — — — —

15–29 yrs
White, non-Hispanic 5.4 (5.0–5.7) Ref. 4.3 (4.0–4.6) Ref.
Black, non-Hispanic 90.1 (87.4–92.9) 1,584.5 75.3 (72.9–77.8) 1,640.6
Hispanic 27.4 (26.1–28.7) 411.2 22.7 (21.5–23.9) 424.4
American Indian/Alaska Native 20.1 (15.5–25.5) 275.0 24.7 (19.7–30.7) 471.8
Asian/Pacific Islander 6.1 (4.9–7.5) 14.4 4.5 (3.5–5.7) 4.3

30–49 yrs
White, non-Hispanic 5.1 (4.8–5.4) Ref. 4.7 (4.4–4.9) Ref.
Black, non-Hispanic 47.2 (45.3–49.1) 826.0 43.1 (41.3–44.9) 821.7
Hispanic 13.0 (12.1–13.8) 154.0 11.3 (10.5–12.0) 141.0
American Indian/Alaska Native 17.1 (13.0–22.1) 235.8 18.6 (14.3–23.7) 296.6
Asian/Pacific Islander 3.7 (2.9–4.5) -28.4 3.2 (2.5–4.0) -32.3

50–64 yrs
White, non-Hispanic 3.4 (3.2–3.7) Ref. 3.3 (3.1–3.6) Ref.
Black, non-Hispanic 20.7 (18.9–22.4) 504.0 16.5 (15.0–18.0) 392.9
Hispanic 6.7 (5.6–7.7) 95.6 6.2 (5.3–7.2) 85.9
American Indian/Alaska Native — — — 9.9 (6.1–15.3) 196.5
Asian/Pacific Islander 3.4 (2.4–4.7) -0.5 3.5 (2.5–4.8) 5.1

≥65 yrs
White, non-Hispanic 2.0 (1.8–2.3) Ref. 2.3 (2.1–2.6) Ref.
Black, non-Hispanic 11.0 (9.2–12.9) 447.6 9.1 (7.4–10.7) 293.8
Hispanic 3.7 (2.7–5.1) 84.5 4.7 (3.6–6.1) 104.7
American Indian/Alaska Native — — — — — —
Asian/Pacific Islander — — — — — —

Female
0–14 yrs

White, non-Hispanic 1.1 (0.9–1.2) Ref. 1.0 (0.8–1.1) Ref.
Black, non-Hispanic 3.5 (2.9–4.0) 216.9 3.2 (2.7–3.8) 232.3
Hispanic 1.7 (1.4–2.0) 56.1 1.4 (1.1–1.7) 44.8
American Indian/Alaska Native — — — — — —
Asian/Pacific Islander — — — — — —

See table footnotes on the next page.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 167

populations and remain highest among young, non-Hispanic
black males. The findings in this report estimate that 75 out of
100,000 non-Hispanic black males aged 15–29 years die from
homicide in a given year. Moreover, 2009 data from the 16 U.S.
states that report data on homicides to the National Violent
Death Reporting System suggest that nearly half of homicides
in this population were outcomes of escalated arguments and
conflicts; one third were precipitated by another crime such
as burglary, robbery, or assault; one fifth involved illicit drug
activity; and approximately 16% were gang related (1,6).

Homicide remains less common among women; however,
in 2009, homicide was the sixth leading cause of death
among females aged 15–49 years (1). Similar to the findings
from the 2011 CHDIR, data from this report indicate
that non-Hispanic black and non-Hispanic AI/AN females
experience death by homicide more frequently than women
in other racial/ethnic populations (10). Female homicides
are characteristically different from male homicides in that

females are more likely to be killed by a family member during
childhood or adolescence (15) and by an intimate partner
during adulthood (16). Increased equality between men and
women in regards to education, wages, and occupational
status might increase women’s access to services that prevent
intimate partner homicide, such as protective orders, shelters,
and advocacy services (17).

Although the findings in this report do not indicate whether
a long-term decrease in homicide rates is occurring, rates
were noticeably lower in 2009 than in 2007. The decrease
in homicide rates between these 2 years was considerable,
particularly among males aged 15–29 years, which is consistent
with the long-term decreasing trend in homicide rate that
has been observed among this demographic population since
the early 1990s (1) (Figure 2). Possible explanations for this
decreasing homicide rate among young males are reductions
in drug trade and sales, increases in police response to youths
who carry firearms, and increases in incarceration (17). Despite

TABLE 2. (Continued) Crude homicide rates,* by sex, age group, and race/ethnicity — National Vital Statistics System, United States, 2007 and 2009

Characteristic

2007 2009

Rate (95% CI)†
Relative difference

(%) Rate (95% CI)
Relative difference

(%)

Female
15–29 yrs

White, non-Hispanic 2.3 (2.1–2.5) Ref. 2.0 (1.8–2.2) Ref.
Black, non-Hispanic 9.8 (8.9–10.7) 326.2 8.7 (7.8–9.5) 332.6
Hispanic 3.6 (3.1–4.2) 58.7 3.4 (2.9–3.9) 68.3
American Indian/Alaska Native — — — 6.7 (4.2–10.2) 235.3
Asian/Pacific Islander 1.8 (1.2–2.7) -20.0 1.5 (0.9–2.2) -26.5

30–49 yrs
White, non-Hispanic 2.5 (2.3–2.7) Ref. 2.4 (2.2–2.6) Ref.
Black, non-Hispanic 8.8 (8.0–9.5) 252.4 7.2 (6.5–7.9) 199.5
Hispanic 2.9 (2.5–3.3) 15.8 2.9 (2.5–3.3) 19.8
American Indian/Alaska Native 6.8 (4.4–10.2) 174.6 — — —
Asian/Pacific Islander 1.8 (1.3–2.4) -28.3 1.8 (1.3–2.4) -26.6

50–64 yrs
White, non-Hispanic 1.4 (1.3–1.6) Ref. 1.5 (1.3–1.7) Ref.
Black, non-Hispanic 3.5 (2.9–4.2) 149.2 3.4 (2.7–4.0) 125.1
Hispanic 1.6 (1.1–2.2) 12.8 1.8 (1.3–2.3) 18.3
American Indian/Alaska Native — — — — — —
Asian/Pacific Islander 1.6 (1.0–2.5) 15.1 1.9 (1.2–2.8) 27.9

≥65 yrs
White, non-Hispanic 1.3 (1.1–1.4) Ref. 1.5 (1.3–1.7) Ref.
Black, non-Hispanic 2.7 (2.0–3.5) 112.6 2.2 (1.6–3.0) 48.5
Hispanic 1.0 (0.5–1.6) -23.7 1.3 (0.8–2.0) -16.0
American Indian/Alaska Native — — — — — —
Asian/Pacific Islander — — — — — —

Abbreviations: 95% CI = 95% confidence interval; Ref. = referent.
* Per 100,000 population.
† CIs based on <100 deaths were calculated using a gamma method, and those based ≥100 deaths were calculated using a normal approximation (Source: Xu J,

Kockanek KD, Murphy SL, Tejada-Vera B. Deaths: final data for 2007. Hyattsville, MD: US Department of Health and Human Services, CDC, National Center for Health
Statistics. Natl Vital Stat Rep 2010;58).

§ Relative differences were calculated based on rates that were estimated to five decimal places. Therefore, relative differences calculated based on the rates provided
in the table might differ from those displayed because of rounding.

¶ Persons of Hispanic ethnicity might be of any race or combination of races.
** Rates unreliable (calculated from <20 deaths).

Supplement

168 MMWR / November 22, 2013 / Vol. 62 / No. 3

the decreases, the disparity in homicide rates between non-
Hispanic black males and non-Hispanic white males is still
pronounced. Although the rate among non-Hispanic black
males aged 15–29 years in 2009 is half the rate reported
in 1993 (75 vs. 158 per 100,000, respectively) (1), similar
decreases have been reported for males of similar age and of
other races/ethnicities.

Socioeconomic factors play a substantial role in homicide
disparities by race/ethnicity, sex, age, and geographic area.
For example, racial/ethnic minorities are more likely to live
in disadvantaged neighborhoods (11). Residential areas with
high levels of poverty, unemployment, and jobs with low wages
can increase risk of income-generating crimes such as burglary
and robbery, stress and conflict, and substance abuse among
residents (18,19), all factors that increase risk for homicide and
violence (11,20). One longitudinal study reported that after
controlling for similar socioeconomic factors, such as living
in a disadvantaged community, being on welfare, and having
a young or single parent, race was not predictive of being a
homicide offender (21). Similar risk factors might explain
the differences in homicide rates by age and geographic area.
Future studies controlling for socioeconomic factors might
offer additional support for this conclusion.

Prevention strategies that can change the characteristics of
communities, relationships, and persons that are associated
with violence perpetration might reduce violence rates not only

for all persons in the United States but also among groups with
the highest rates of violence (22). Certain promising strategies
have been developed that use multicomponent approaches,
involve coordinated efforts by numerous relevant stakeholders,
and include an appropriate mix of both universal interventions
and interventions that address the needs of groups at highest
risk for violence (22). Communities That Care (23,24),
Promoting School-community-university Partnerships to
Enhance Resilience (PROSPER) (25,26), Striving to Reduce
Youth Violence Everywhere (STRYVE) (available at http://
www.safeyouth.gov), and Urban Networks to Increase Thriving
Youth (UNITY) (available at http://preventioninstitute.org/
unity.html) are examples of coalition-based operating systems
and violence prevention initiatives that can assist communities
in developing the type of tailored, broad strategies described.
Promising multicomponent programs such as CeaseFire (27)
and Safe Streets (28), which work to change community norms
regarding violence, cultivate skills for using alternatives to
violence, and interrupt escalating tensions, also are promising
strategies for preventing general violence, shootings, and
shooting-related homicides.

The high homicide rates among youths in late adolescence
and young adulthood suggest that the school years are an
important developmental point for intervention. Creating
a positive school environment is an example of one way to
improve youths’ access to a safe, stable, nurturing setting;
promote norms of nonviolence; facilitate the formation of
supportive and positive social relationships; and maximize the
development of social and problem-solving skills (29). The
influences, experiences, and socialization provided by such
school environments could help youths become more adept at
navigating problematic interactions and adapting to challenges
and difficulties that could lead to serious violence (29). Such
school environments might particularly be important for
youths who lack other positive influences (29).

Limitations
These findings in this report are subject to at least three

limitations. First, small numbers of homicides precluded stable
rate estimations among some populations. Second, data on
individual and environmental risk factors for homicide were
unavailable, which precluded closer examination of possible
sources of disparities by age, sex, race/ethnicity, and geography.
Third, racial misclassification might result in overestimated
homicide rates for non-Hispanic blacks and non-Hispanic
whites and underestimated rates for AI/ANs, A/PIs, and
Hispanics (8).

FIGURE 1. Crude homicide rates,* by state — National Vital Statistics
System, United States, 2009†

* Number of deaths per 100,000 population.
† Ten states experienced a significant decrease from 2007 to 2009: Arizona

(29.3%), California (12.4%), Florida (15.5%), Georgia (19.5%), Idaho (55.6%),
Maryland (21.1%), New Jersey (19.3%), North Carolina (19.1%), Ohio (15.5%),
and Pennsylvania (13.9%).

7.51–22.51
5.51–7.50
3.01–5.50
0–3.00
Suppressed (≤1.0 death)

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 169

Conclusion
Effective evidence-based strategies to reduce violence are

available (30); however, additional work is needed to build
organizational and community capacity to make best use
of these programs, policies, and strategies. Many health-
related disparities can be reduced by altering influential
socially embedded conditions such as 1) neighborhood
living conditions, 2) opportunities for learning and capacity
for development, and 3) employment opportunities and
community development (31,32). Because these outcomes
mediate the effects of social determinants of health, they
might be viable mechanisms for changing or eliminating
social influences that create or increase disparities in homicide
rates. Promising strategies such as implementing business or

community improvement districts might help decrease levels
of violent crimes by increasing employment opportunities for
local residents and creating physical or cultural environments
that are more aesthetically and economically attractive
(33,34). These community-level strategies might reduce or
offset the effects of poverty, improve the social environments
of communities, and implement safety measures (33,34). To
eliminate homicide disparities, more research is needed to
understand the scope of the problem and the risk and protective
factors implicated in these violent events, evaluate programs
that prevent and reduce violence, and better understand how
to adapt, disseminate, and implement these strategies in the
communities and populations in greatest need.

Non-Hispanic, white
Non-Hispanic, black
American Indian/Alaska Native
Asian/Paci�c Islander
Hispanic§

0

20

40

60

80

100

120

140

160

180

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

C
ru

d
e

h
o

m
ic

id
e

ra
te

Year

FIGURE 2. Crude homicide rates* among males aged 15–29 years, by racial/ethnic group† and year — National Vital Statistics System, United
States, 1990–2009

Abbreviations: ICD-10 = International Classification of Diseases, 10th Revision; ICD-9 = International Classification of Diseases, 9th Revision.
* ICD-10 to ICD-9 comparability ratio for homicides = 0.998.
† Number of deaths per 100,000 population.
§ Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

170 MMWR / November 22, 2013 / Vol. 62 / No. 3

Acknowledgment

The findings in this report are based, in part, on contributions by
Nimeshkumar Patel.

References
1. CDC. Web-based Injury Statistics Query and Reporting System

(WISQARS). Atlanta, GA: US Department of Health and Human
Services, CDC; 2012. Available at http://www.cdc.gov/injury/wisqars/
index.html.

2. Rheingold AA, Zinzow H, Hawkins A, Saunders BE, Kilpatrick DG.
Prevalence and mental health outcomes of homicide survivors in a
representative U.S. sample of adolescents: data from the 2005 National
Survey of Adolescents. J Child Psychol Psychiatry 2012;53:687–94.

3. Schiavone DB. The effects of exposure to community violence on aspects
of adolescent identity development. J Child Adolesc Psychiatr Nurs 2009;
22:99–105.

4. Hemenway D. Measuring the cost of injury: underestimating the costs
of street violence. Inj Prev 2011;17:289–90.

5. Shahpar C, Li G. Homicide mortality in the United States, 1935–1994:
age, period, and cohort effects. Am J Epidemiol 1999;150:1213–22.

6. CDC. Surveillance for violent deaths—National Violent Death
Reporting System, 16 states, 2007. MMWR 2010;59(No. SS-4).

7. Miniño AM. Mortality among teenagers aged 12–19 years: United States,
1999–2006 Hyattsville, MD: US Department of Health and Human Services,
CDC, National Center for Health Statistics; 2010. NCHS data brief no. 37.
Available at http://www.cdc.gov/nchs/data/databriefs/db37.htm.

8. CDC. Fatal injuries among children by race and ethnicity—United
States, 1999–2002. MMWR 2007;56(No. SS-5).

9. Najem GR, Aslam S, Davidow AL, Elliot N. Youth homicide racial
disparities: gender, years, and cause. J Natl Med Assoc 2004;96:
558–66.

10. CDC. Homicides—United States, 1999–2007. In: CDC health
disparities and inequalities report—United States, 2011. MMWR 2011;
60(Suppl; January 14, 2011).

11. Krueger PM, Bond Huie SA, Rogers RG, Hummer RA. Neighbourhoods
and homicide mortality: an analysis of race/ethnic differences.
J Epidemiol Community Health 2004;58:223–30.

12. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011; 60(Suppl; January 14, 2011).

13. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

14. Xu J, Kockanek KD, Murphy SL, Tejada-Vera B. Deaths: final data for
2007. Hyattsville, MD: US Department of Health and Human Services,
CDC, National Center for Health Statistics. Natl Vital Stat Rep 2010;58.

15. Muftić LR, Moreno RD. Juvenile homicide victimization: differences
and similarities by gender. Youth Violence Juv Justice 2010;8:386–98.

16. Vieraitis LM, Britto S, Morris RG. Assessing the impact of changes in
gender equality on female homicide victimization: 1980–2000. Crime
Delinq. Epub November 8, 2011. Available at http://cad.sagepub.com/
content/early/2011/09/10/0011128711420100.

17. Blumstein A, Rivara FP, Rosenfeld R. The rise and decline of homicide—
and why. Annu Rev Public Health 2000;21:505–41.

18. Gould ED, Weinberg BA, Mustard DB. Crime rates and local labor
market opportunities in the United States: 1979–1997. Rev Econ Stat
2002;84:45–61.

19. Hill TD, Angel RJ. Neighborhood disorder, psychological distress, and
heavy drinking. Soc Sci Med 2005;61:965–75.

20. Loeber R. The prediction of violence and homicide in young men.
J Consult Clin Psychol 2005;73:1074.

21. Farrington DP, Loeber R, Berg MT. Young men who kill. Homicide
Stud 2012;16:99–128.

22. Krug EG, Dahlberg LL, Mercy JA, Zwi AB, Lozano R, eds. World report
on violence and health. Geneva, Switzerland: World Health Organization;
2002.

23. Hawkins JD. Preventing crime and violence through communities that
care. Eur J Crim Policy Res 1999;7:443–58.

24. Hawkins JD, Catalano RF, Arthur MW. Promoting science-based
prevention in communities. Addict Behav 2002;27:951–76.

25. Spoth R, Greenberg M, Bierman K, Redmond C. PROSPER
community—university partnership model for public education systems:
capacity-building for evidence-based, competence-building prevention.
Prev Sci 2004;5:31–9.

26. Spoth RL, Greenberg MT. Toward a comprehensive strategy for effective
practitioner–scientist partnerships and larger-scale community health
and well-being. Am J Community Psychol 2005;35:107–26.

27. Skogan WG, Hartnett SM, Bump N, Dubois J. Evaluation of
CeaseFire—Chicago. Chicago, IL: Northwestern University; 2008.

28. Webster DW, Vernick JS, Mendel J. Interim evaluation of Baltimore’s Safe
Streets Program. Baltimore, MD: Johns Hopkins Bloomberg School of
Public Health, Center for the Prevention of Youth Violence; 2009.

29. Hahn R, Fuqua-Whitley D, Wethington H, et al. Effectiveness of
universal school-based programs to prevent violent and aggressive
behavior: a systematic review. Am J Prev Med 2007;33(Suppl):S114–29.

30. Mihalic S, Irwin K, Elliot D, Fagan A, Hansen D. Blueprints for violence
prevention. Washington, DC: US Department of Justice, Office of Justice
Programs, Office of Juvenile Justice and Delinquency Prevention; 2001.

31. Anderson LM, Scrimshaw SC, Fullilove MT, Fielding JE. The
Community Guide’s model for linking the social environment to health.
Am J Prev Med 2004;24:12–20.

32. Williams DR, Costa MV, Odunlami AO, Mohammed SA. Moving
upstream: how interventions that address the social determinants of
health can improve health and reduce disparities. J Public Health Manag
Pract 2008;14:S8–17.

33. MacDonald JM, Blumenthal RN, Golineli D, et al. Neighborhood
effects on crime and youth violence: the role of business improvement
districts in Los Angeles. Santa Monica, CA: Rand; 2009.

34. MacDonald J, Golinelli D, Stokes RJ, Bluthenthal R. The effect of
business improvement districts on the incidence of violent crimes. Inj
Prev 2010;16:327–32.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 171

Introduction
Infant mortality rates are associated with maternal health,

quality of and access to medical care, socioeconomic
conditions, and public health practices, which makes infant
mortality an increasingly important public health concern
(1,2). After large declines throughout the twentieth century, the
U.S. infant mortality rate did not decline significantly during
2000–2005 (3). Analysis of 2000–2004 infant mortality in the
United States indicated considerable disparities by race and
Hispanic origin (4). Race and ethnic disparities in U.S. infant
mortality have been apparent since vital statistics data began to
be collected more than 100 years ago. These disparities have
persisted over time, and research indicates that not all groups
have benefited equally from social and medical advances (5–7).

The infant mortality analysis and discussion that follows is
part of the second CDC Health Disparities and Inequalities
Report (CHDIR) (4). The 2011 CHDIR (8) was the first
CDC report to assess disparities across a wide range of diseases,
behavioral risk factors, environmental exposures, social
determinants, and health-care access. The criteria for inclusion
of topics that are presented in the 2013 CHDIR are based on
criteria that are described in the 2013 CHDIR Introduction
(9). This report provides more current information on infant
mortality rates on the basis of race/ethnicity, mother’s place
of birth, and by state and region. The purposes of this infant
mortality analysis are to raise awareness of differences in infant
mortality by selected maternal and infant characteristics, and
to prompt actions to reduce these disparities.

Methods
To estimate disparities in infant mortality rate by selected

characteristics and specified group, CDC analyzed data from
the United States linked birth/infant death data sets (linked
files) for 2005 through 2008 (the latest year for which accurate
race/ethnicity data are available) (5). In these data sets,
information from the birth certificate is linked to information
from the death certificate for each infant (aged <1 year) who
dies in the United States. Characteristics analyzed included sex,
maternal race/ethnicity, maternal place of birth, and the state

of residence of the mother at the time of birth. Household
income and educational attainment were not analyzed because
they were either not collected or not collected consistently on
birth certificates. Maternal race was defined as white, black,
Asian/Pacific Islander, and American Indian/Alaska native.
Ethnicity is defined as Hispanic or non-Hispanic. Hispanic
data were further subdivided into Mexican, Puerto Rican,
Cuban, and Central and South American. Place of birth was
defined as born in the 50 states and DC, or born outside of
the 50 states and DC.

Infant mortality rates were calculated as the number of
infant deaths per 1,000 live births in the specified group (i.e.,
by maternal race/Hispanic origin, maternal birthplace, state
of residence, and infant gender). Ratios of non-Hispanic black
to non-Hispanic white infant mortality rates were computed
to assess the magnitude of the disparity in non-Hispanic
black and non-Hispanic white infant mortality rates by state.
Data from 2006–2008 were aggregated to obtain statistically
reliable state-specific rates by race and Hispanic origin; rates
are not shown for cells with <20 infant deaths. Rates based
on <20 infant deaths are not shown separately as they do not
meet standards of reliability or precision. Differences between
infant mortality rates were assessed for statistical significance
by using the z test (p<0.05).

Disparities were measured as the deviations from a “referent”
category rate. Absolute difference was measured as the simple
difference between a population subgroup mortality rate and
the rate for its respective reference group. The relative difference,
a percentage, was calculated by dividing the difference by the
value in the referent category and multiplying by 100.

Results
The U.S. infant mortality rate declined 10% from 2005 to

2010, from 6.86 infant deaths per 1,000 live births in 2005
to a preliminary estimate of 6.14 in 2010 (5,13). In 2008,
the overall U.S. infant mortality rate was 6.61 infant deaths
per 1,000 live births, with differences by race and Hispanic
origin (Table 1). The highest infant mortality rate was for non-
Hispanic black women (12.67), with a rate 2.3 times that for
non-Hispanic white women (5.52) (Table 1). Compared with

Infant Deaths — United States, 2005–2008
Marian F. MacDorman, PhD, T. J. Mathews, MS

National Center for Health Statistics, CDC

Corresponding author: Marian F. MacDorman, PhD, Division of Vital Statistics, National Center for Health Statistics, CDC. Telephone: 301-458-4356;
E-mail: [email protected].

Supplement

172 MMWR / November 22, 2013 / Vol. 62 / No. 3

non-Hispanic white women, infant mortality rates were 53%
higher for American Indian/Alaska Native* women (8.42) and
32% higher for Puerto Rican women (7.29). Infant mortality
rates for Asian/Pacific Islanders* (4.51) and Central or South
American women (4.76) were lower than those for non-
Hispanic white women. From 2005 to 2008, infant mortality
rates declined approximately 4% for the total population and
for non-Hispanic white women, approximately 7% for non-
Hispanic black women, and 12% for Puerto Rican women;
changes for other racial/ethnic groups were not statistically
significant. When examined by place of birth of the mother,
the 2008 infant mortality rate was 38% higher for women
born in the 50 states and DC than for women born elsewhere
(Table 1). The infant mortality rate was 21% higher for male
than for female infants.

Differences also exist in infant mortality rates between
various states, with a twofold or greater difference in rates
between the states with the highest and lowest rates for the total
population and for each race/ethnic group studied. Across the
United States, infant mortality rates are generally higher in the
South and Midwest and lower in other parts of the country.

During 2006–2008, total infant mortality rates ranged from
a high of 11.97 per 1,000 live births for DC and Mississippi
10.16 to a low of 4.94 for Massachusetts and Utah. However,
because DC has high concentrations of high-risk women, its
rate is more appropriately compared with rates for other large
U.S. cities. For non-Hispanic white women, Alabama had the
highest rate (7.67) and New Jersey the lowest rate (3.78). For
non-Hispanic black women, the rate was highest in Hawaii
(18.54) and lowest in Washington (7.66). For Hispanic
women, the rate was highest in Pennsylvania (7.94) and lowest
in Louisiana (3.92).

Ratios of non-Hispanic black to non-Hispanic white infant
mortality rates were computed to assess the magnitude of the
disparity in non-Hispanic black and non-Hispanic white infant
mortality rates by state (Figure). Although the average rate ratio
in the United States was 2.35, seven areas (Connecticut, DC,
Hawaii, Massachusetts, New Jersey, New York, and Wisconsin)
had rate ratios of 2.60 or greater. In contrast, seven other
states (Arkansas, Alabama, Kentucky, Mississippi, Oklahoma,
Oregon, and Washington) had ratios <2.10. Rate ratios are not
shown for states with <20 non-Hispanic black infant deaths.

* Includes Hispanic and non-Hispanic women.

TABLE 1. Infant mortality rates* by selected characteristics — United States, 2005 and 2008

Characteristic

2005 2008

Infant
mortality rate

Absolute
difference

(rate)
Relative

difference (%)
Infant

mortality rate

Absolute
difference

(rate)
Relative

difference (%)

Total 6.86 6.61
Sex

Male 7.56 1.4 23.5 7.22 1.3 20.9
Female 6.12 Ref. Ref. 5.97 Ref. Ref.

Race/Ethnicity†
White, non-Hispanic 5.76 Ref. Ref. 5.52 Ref. Ref.
Black, non-Hispanic 13.63 7.9 136.6 12.67 7.2 129.5
Asian/Pacific Islander§ 4.89 -0.9 -15.1 4.51 -1.0 -18.3
American Indian/Alaska Native 8.06 2.3 39.9 8.42 2.9 52.5
Hispanic¶ 5.62 -0.1 -2.4 5.59 0.1 1.3

Mexican 5.53 -0.2 -4.0 5.58 0.1 1.1
Puerto Rican 8.30 2.5 44.1 7.29 1.8 32.1
Cuban 4.42 -1.3 -23.3 4.90 -0.6 -11.2
Central and South American 4.68 -1.1 -18.8 4.76 -0.8 -13.8

Place of birth
Born in the 50 states and DC 7.26 2.2 42.9 6.99 1.9 38.4
Born outside the 50 states and DC 5.08 Ref. Ref. 5.05 Ref. Ref.

Abbreviation: Ref. = Referent.
Source: CDC. Period linked birth/infant death public-use data files (Downloadable data files). Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics. Available at http://www.cdc.gov/nchs/data_access/VitalStatsOnline.htm.
* Infant mortality rate = number of deaths among infants aged <1 year per 1,000 live births in a specific group.
† Race and Hispanic origin are reported separately on birth certificates. Race categories are consistent with the 1977 Office of Management and Budget standards.

Thirty states reported multiple-race data on the birth certificate in 2008. For the <2% of events in these states that reported multiple race data, the multiple-race
data were bridged to the single race categories of the 1977 standards for compatibility with other states.

§ Includes persons of Hispanic or non-Hispanic origin.
¶ Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 173

Discussion
The U.S. infant mortality rate has declined 10% from 2005

(6.86) to 2010 (preliminary estimate: 6.14) (5,13); however,
disparities have persisted. Higher infant mortality rates for
male than for female infants have persisted for many years and
occur among most world populations, and have been explained
in part by differences in genetic susceptibility to disease (14).
Differences in infant mortality rates by race/ethnicity, maternal
birthplace, and geographic area might reflect in part different
population profiles, with regard to sociodemographic and
behavioral risk factors. For example, infant mortality rates
are higher than the U.S. average for adolescents, women aged
≥35 years, unmarried mothers, smokers, those with lower
educational levels, or inadequate prenatal care (5). Substantial
differences between groups in income and access to health care
also might contribute to differences in infant mortality (15).
Population groups with the lowest infant mortality rates tended
to have a smaller percentage of births to women with some
or all of these characteristics, whereas groups with the highest
infant mortality rates tended to have a higher percentage of
births in women with some or all of these characteristics. Other
factors that might contribute to racial/ethnic differences in
infant mortality include differences in maternal preconception
health, infection, stress, racism, and social and cultural

FIGURE. Ratio of non-Hispanic black and non-Hispanic white infant
mortality rates,* by state — United States, 2006–2008

Source: National Vital Statistics System, NCHS, CDC.
* Infant mortality rate = number of deaths among infants aged <1 year per 1,000

live births in a specific group.

≥2.60
2.35–2.59

DC

U.S. average
rate ratio = 2.35

Ratio not available;
<20 infant deaths
for either group

2.10–2.34
<2.10

TABLE 2. Infant mortality rates,* by race and Hispanic origin of mother
and by state — United States, 2006–2008

Total

White,
non-

Hispanic

Black,
non-

Hispanic Hispanic

United States 6.68 5.58 13.11 5.50

Alabama 9.47 7.67 13.73 7.50
Alaska 6.54 4.10 † †
Arizona 6.54 6.04 14.85 6.13
Arkansas 7.89 6.70 13.53 5.71
California 5.12 4.51 10.72 4.88
Colorado 6.04 5.13 11.97 6.96
Connecticut 6.27 4.80 13.11 6.35
Delaware 8.03 5.89 13.46 7.10
District of Columbia 11.97 4.46 17.68 †
Florida 7.21 5.71 12.83 5.38
Georgia 8.02 5.87 12.70 5.06
Hawaii 6.04 4.58 18.54 4.98
Idaho 6.46 5.95 † 7.91
Illinois 7.10 5.70 13.45 5.91
Indiana 7.44 6.47 15.36 6.28
Iowa 5.43 5.06 11.10 6.61
Kansas 7.50 6.94 14.62 7.15
Kentucky 7.04 6.62 12.13 5.07
Louisiana 9.38 6.62 13.88 3.92
Maine 6.04 5.90 † †
Maryland 7.98 5.50 12.98 5.33
Massachusetts 4.94 4.04 10.90 6.08
Michigan 7.56 5.87 14.70 7.09
Minnesota 5.55 4.77 11.33 4.64
Mississippi 10.16 7.07 13.82 6.64
Missouri 7.34 6.18 14.49 5.12
Montana 6.47 5.89 † †
Nebraska 5.93 5.33 12.98 5.21
Nevada 6.10 5.29 12.54 5.69
New Hampshire 5.10 5.00 † †
New Jersey 5.35 3.78 12.06 5.12
New Mexico 5.81 6.12 † 5.60
New York 5.57 4.29 11.29 5.01
North Carolina 8.29 6.17 14.62 6.32
North Dakota 6.44 5.63 † †
Ohio 7.74 6.25 15.03 6.88
Oklahoma 7.85 7.52 13.91 5.09
Oregon 5.41 5.22 10.16 5.36
Pennsylvania 7.52 5.78 14.04 7.94
Rhode Island 6.47 4.28 10.56 7.77
South Carolina 8.30 6.04 12.97 5.87
South Dakota 7.15 5.59 † †
Tennessee 8.37 6.54 15.36 6.47
Texas 6.22 5.48 11.69 5.61
Utah 4.94 4.73 † 5.03
Vermont 5.12 4.95 † †
Virginia 7.24 5.48 13.40 5.97
Washington 5.01 4.33 7.66 5.28
West Virginia 7.38 7.11 14.93 †
Wisconsin 6.57 5.37 15.14 6.34
Wyoming 7.05 6.32 † 7.90

* Infant mortality rate = number of deaths among infants aged <1 year per 1,000
live births in a specific group.

† Does not meet standards of reliability or precision; based on <20 deaths in
the numerator.

Supplement

174 MMWR / November 22, 2013 / Vol. 62 / No. 3

differences (7,16–21). However, the influence of an individual
risk factor can vary considerably between population groups,
indicating different medical profiles and life experiences for
women of different backgrounds (7,16,21–22).

The risk factors for infant mortality discussed earlier (e.g.,
maternal age, tobacco use, lower income or educational levels,
and inadequate prenatal care) are very similar to the risk factors
for preterm or low birthweight delivery, and these risk factors
can affect infant mortality either directly or through the
mechanism of preterm or low birthweight delivery. In 2008,
the percentage of infants born preterm (<37 completed weeks’
gestation) was higher for non-Hispanic black (17.5%), Puerto
Rican (14.1%), and American Indian/Alaska Native (13.6%)
mothers, than for non-Hispanic white mothers (11.1%) (5).
Infant mortality rates are substantially higher for preterm
and low birthweight infants, and even limited changes in the
percentages of preterm or low birthweight births can have a
major impact on infant mortality (5,6). In fact, the recent
decline in U.S. infant mortality is linked to a recent decline in
the percentage of preterm births, from a high of 12.8% in 2006
to 12.0% in 2010 (5,22). Still the U.S. infant mortality rate
was higher than for the majority of other developed countries,
in part because of a substantially higher percentage of preterm
births, a critical risk factor for infant mortality (23–24).

Limitation
The findings in this report are subject to at least one

limitation. Differences in infant mortality rates for smaller
states and certain race/ethnic groups (e.g., American Indians/
Alaska Natives, Asians/Pacific Islanders, and Cubans) should
be interpreted with caution, as small numbers of infant deaths
(i.e., <20) in specific subcategories might lead to a lack of
statistical precision.

Conclusion
Infant mortality remains a complex and multifactorial

problem that will continue to challenge researchers and
policymakers in the years ahead. Despite recent declines in
the overall infant mortality rate, the longstanding disparities
in infant mortality by racial/ethnic group, mother’s birthplace,
and geographic area persist. One of the Healthy People 2020
objectives is to achieve an infant mortality rate of 6.0 for the
total population and for each race/ethnic group. Although the
U.S. infant mortality rate of 6.14 in 2010 approximates the
Healthy People 2020 objective, rates for several racial/ethnic
groups are substantially higher than the goal (25). Prevention
of preterm birth is critical to both lowering the overall infant
mortality rate and to reducing racial/ethnic disparities (5,6).

References
1. CDC. Achievements in public health, 1900–1999: healthier mothers

and babies. MMWR 1999;48:849–58.
2. Guyer B, Freedman MA, Strobino DM, Sondik EJ. Annual summary

of vital statistics: trends in the health of Americans during the 20th
century. Pediatrics 2000;106:1307–17.

3. MacDorman MF, Mathews TJ. Recent trends in infant mortality in the
United States. Hyattsville, MD: US Department of Health and Human
Services, CDC, National Center for Health Statistics; 2008. NCHS Data
Brief no. 9. Available at http://www.cdc.gov/nchs/data/databriefs/db09.pdf.

4. CDC. Infant deaths—United States, 2000–2007. In: CDC health
disparities and inequalities report—United States, 2011. MMWR 2011;
60(Suppl; January 14, 2011).

5. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2008
period linked birth/infant death data set. National Vital Statistics Reports
vol 60 no 5. Hyattsville, MD: US Department of Health and Human
Services, CDC, National Center for Health Statistics; 2012.

6. MacDorman MF, Mathews TJ. Understanding racial and ethnic
disparities in US infant mortality rates. NCHS data brief, No. 74.
Hyattsville, MD: National Center for Health Statistics. 2011. Available
at http://www.cdc.gov/nchs/data/databriefs/db74.pdf.

7. Krieger N, Rehkopf DH, Chen JT, et al. The fall and rise of US inequities
in premature mortality: 1960–2002. PLoS Med 2008;5:e46.

8. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011; 60(Suppl; January 14, 2011).

9. CDC. Introduction. In: CDC health disparities and inequalities report—
United States, 2013. MMWR 2013;62(No. Suppl 3).

10. CDC. Period linked birth-infant death public-use data files
[Downloadable data files]. Hyattsville, MD: US Department of Health
and Human Services; 2012. Available at http://www.cdc.gov/nchs/
data_access/VitalStatsOnline.htm.

11. CDC. User’s guide for the 2008 period linked birth/infant death data
set. 2012. Available at http://www.cdc.gov/nchs/data_access/
VitalStatsOnline.htm.

12. Rosenberg HM, Maurer JD, Sorlie PD, et al. Quality of death rates by
race and Hispanic origin: a summary of current research, 1999. Vital
Health Stat 2 1999:1–13.

13. Murphy SL, Xu J, Kochanek KD. Deaths: preliminary data for 2010.
In: National Vital Statistics Reports. Hyattsville, MD: US Department
of Health and Human Services, CDC, National Center for Health
Statistics. NCHS Data Brief no 4; 2012. Available at http://www.cdc.
gov/nchs/data/nvsr/nvsr60/nvsr60_04.pdf.

14. Pongou R. Why is infant mortality higher in boys than in girls? A new
hypothesis based on preconception environment and evidence from a
large sample of twins. Demography [Epub ahead of print].

15. DeNavas-Walt C, Proctor BD, Smith JC. Income, poverty, and health
insurance coverage in the United States: 2010. Washington, DC: US
Census Bureau; 2011. Current Population Reports no. P60–239.
Available at http://www.census.gov/prod/2011pubs/p60-239.pdf.

16. Geronimus AT. Black/white differences in the relationship of maternal
age to birthweight: a population-based test of the weathering hypothesis.
Soc Sci Med 1996;42:589–97.

17. Fiscella K. Racial disparity in infant and maternal mortality: confluence
of infection, and microvascular dysfunction. Matern Child Health J
2004;8:45–54.

18. Hogan VK, Njoroge T, Durant TM, Ferre CD. Eliminating disparities
in perinatal outcomes: lessons learned. Matern Child Health J 2001;
5:135–40.

19. Collins JW Jr, David RJ, Handler A, Wall S, Andes S. Very low
birthweight in African American infants: the role of maternal exposure
to interpersonal racial discrimination. Am J Public Health 2004;
94:2132–8.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 175

20. Martin JA, Hamilton BE, Sutton PD, et al. Births: final data for 2008.
Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics; 2010. National Vital Statistics
Reports Vol. 59, no. 1. Available at http://www.cdc.gov/nchs/data/nvsr/
nvsr59/nvsr59_01.pdf.

21. Hummer RA, Powers DA, Pullum SG, Gossman GL, Frisbie WP. Paradox
found (again): infant mortality among the Mexican-origin population
in the United States. Demography 2007;44:441–57.

22. Hamilton BE, Martin JA, Ventura SJ. Births: preliminary data for 2010.
Hyattsville, MD: US Department of Health and Human Services, CDC,
National Center for Health Statistics; National Vital Statistics Report
vol 60 no 2. Available at http://www.cdc.gov/nchs/data/nvsr/nvsr60/
nvsr60_02.pdf.

23. MacDorman MF, Mathews TJ. Behind international rankings of infant
mortality: how the United States compares with Europe. Hyattsville,
MD: US Department of Health and Human Services, CDC, National
Center for Health Statistics; 2009. NCHS Data Brief no 23. Available
at http://www.cdc.gov/nchs/data/databriefs/db23.htm.

24. Organization for Economic Cooperation and Development. OECD
Health Data 2013. Available at http://www.oecd.org/health/health-
systems/oecdhealthdata.htm.

25. CDC. HealthyPeople.gov; 2020 topics and objectives—maternal, infant
and child health. Available at http://www.healthypeople.gov/2020/
topicsobjectives2020/objectiveslist.aspx?topicId=26.

Supplement

176 MMWR / November 22, 2013 / Vol. 62 / No. 3

Introduction
Motor vehicle crashes are a leading cause of death for

children, teenagers, and young adults in the United States (1).
In 2009, approximately 36,000 persons were killed in motor
vehicle crashes, and racial/ethnic minorities were affected
disproportionally (1,2). Approximately 4.3% of all American
Indian/Alaska Native (AI/AN) deaths and 3.3% of all Hispanic
deaths were attributed to crashes, whereas crashes were the
cause of death for <1.7% of blacks, whites, and Asian/Pacific
Islanders (A/PI) (1).

The motor vehicle–related death rate analysis and discussion
that follows is part of the second CDC Health Disparities and
Inequalities Report (CHDIR). The 2011 CHDIR (3) was the
first CDC report to take a broad view of disparities across a
wide range of diseases, behavioral risk factors, environmental
exposures, social determinants, and health care access. The topic
presented in this report is based on criteria that are described in
the CHDIR Introduction (4). The report that follows provides
more current information to what was presented in the 2011
CHDIR (2). The purposes of this motor vehicle–related death
report are to discuss and raise awareness of differences in the
characteristics of persons who die from motor vehicle–related
crashes and to prompt actions to reduce disparities.

Methods
To assess disparities in motor vehicle–related death rates by

race/ethnicity and sex, CDC analyzed data from the National
Vital Statistics System (NVSS). NVSS does not collect data
on other variables such as education and income. Race/
ethnicity was divided into five mutually exclusive categories:
non-Hispanic whites, non-Hispanic blacks, non-Hispanic AI/
ANs, non-Hispanic A/PIs, and Hispanics of all races.

Bridged-race postcensal population estimates from the U.S.
Census Bureau were used to calculate death rates. Death rates
and corresponding 95% confidence intervals were calculated
and age-adjusted to the 2000 standard U.S. population.
Absolute and relative differences in rates were calculated by sex
and race/ethnicity. Disparities were measured as the deviations
from a “referent” category rate. The absolute difference was
measured as the simple difference between a population

subgroup estimate and the estimate for its respective reference
group. The relative difference, a percentage, was calculated by
dividing the difference by the value in the referent category and
multiplying by 100. Differences between age-adjusted death
rates in 2005 and 2009 were compared using the z statistic
based on a normal approximation, and p values ≤0.05 were
considered statistically significant.

Results
The overall motor vehicle–related age-adjusted death rate

was 11.7 deaths per 100,000 population in 2009 (Table 1).
The death rate for males was 2.5 times that for females (16.8
vs. 6.8). In 2009, AI/ANs consistently had the highest motor
vehicle–related death rates among both males and females
(Table). Among males, the AI/AN death rate (33.6) was
approximately 2–5 times the rates of other races/ethnicities.
Black males had the second-highest death rate (18.5), followed
by whites (17.3), Hispanics (14.7), and A/PIs (6.3). Among
females, the AI/AN motor vehicle–related death rate (17.3)
was approximately 2-4 times the rates of other races/ethnicities.
White females had the second-highest death rate (7.1),
followed by blacks (6.4), Hispanics (5.7), and A/PIs (4.0).

Between 2005 and 2009, age-adjusted death rates showed
statistically significant declines by sex among all race/ethnicities
with the exception of AI/AN women (Table). The greatest
decrease in rates for males occurred among AI/AN, from a
death rate of 42.7 per 100,000 population in 2005 to 33.6 in
2009 (absolute rate change: -9.1). Among females, the greatest
decrease occurred among whites, from a death rate of 9.4 in
2005 to 7.1 in 2009 (absolute rate change: -2.3).

Discussion
Evidence-based strategies to reduce overall motor vehicle–

related deaths and injuries include primary seat belt laws (i.e.,
legislation allowing police to stop a vehicle solely for a safety
belt violation), age- and size-appropriate child safety seat and
booster seat use laws, focused child restraint distribution plus
education programs, ignition interlock devices (i.e., devices
that disable a vehicle’s ignition after detection of alcohol in the

Motor Vehicle–Related Deaths — United States, 2005 and 2009
Bethany A. West, MPH

Rebecca B. Naumann, MSPH
National Center for Injury Prevention and Control, CDC

Corresponding author: Bethany A. West, MPH, Division of Unintentional Injury Prevention, National Center for Injury Prevention and Control, CDC.
Telephone: 770-488-0602; E-mail: [email protected].

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 177

driver’s breath), sobriety checkpoints, minimum drinking age
laws (21 years), and 0.08 g/dL blood alcohol concentration
laws (5). Tailoring these strategies to the unique cultures of
different racial/ethnic groups can help reduce disparities in
motor vehicle–related mortality (6,7).

To address the disparities in motor vehicle–related death
and injury among AI/AN, CDC funded four American Indian
tribes during 2004–2009 to tailor, implement, and evaluate
evidence-based interventions to reduce motor vehicle–related
injury and death in their communities. These pilot programs
were successful at increasing seat belt use, increasing child
safety seat use, and decreasing motor vehicle crashes (6,7).
Across the four pilot programs, relative increases in drivers’
observed seat belt use ranged from a 38% increase to a 315%
increase and child safety seat use increases ranged from a 45%
increase to an 85% increase in use. Declines in motor vehicle
crashes ranged from a 29% decrease to a 36% decrease in the
number of motor vehicle crashes and the number of motor
vehicle crashes in which someone was injured, respectively. As
a result, CDC has expanded the tribal programs and is funding
eight new tribes during 2010–2014.

Limitations
The findings in this report are subject to at least one

limitation. Because NVSS data are extracted from death
certificates and not self-reported, some racial misclassification
is likely, particularly for AI/AN (8).

Conclusion
Despite the recent declines in motor vehicle–related death

rates noted in this report, the need remains for increased use of
evidence-based strategies to reduce disparities. More translational
research is warranted on the scalability of interventions that have
successfully been tailored to communities of different racial/
ethnic and cultural backgrounds.

References
1. CDC. Web-based Injury Statistics Query and Reporting System

(WISQARS) [Online database]. Atlanta, GA: US Department of Health
and Human Services, CDC, National Center for Injury Prevention and
Control; 2010. Available at http://www.cdc.gov/injury/wisqars/index.html.

TABLE. Age-adjusted rates* of motor vehicle–related deaths, by race/ethnicity, sex, and year — National Vital Statistics System, United States,
2005 and 2009

Characteristic

2005 2009
Absolute

change in rate
from 2005 to

2009

P-value for
difference

between 2005
and 2009

Age-
adjusted

death rate (95% CI)

Absolute
difference

(rate)

Relative
difference

(%)

Age-
adjusted

death rate (95% CI)

Absolute
difference

(rate)

Relative
difference

(%)

Total 15.2 (15.1–15.4) — — 11.7 (11.6–11.8) — — -3.55 <0.0001
Sex

Male 21.8 (21.6–22.1) 12.9 143.4 16.8 (16.6–17.0) 10.0 148.2 -5.0 <0.0001
Female 9.0 (8.8–9.1) Ref. Ref. 6.8 (6.6–6.9) Ref. Ref. -2.2 <0.0001

Race/Ethnicity
White 15.6 (15.41–5.8) Ref. Ref. 12.1 (11.9–12.2) Ref. Ref. -3.5 <0.0001
Black 14.9 (14.5–15.4) -0.7 -4.2 12.0 (11.6–12.3) -0.1 -0.9 -3.0 <0.0001
Hispanic† 14.8 (14.4–15.3) -0.8 -4.9 10.4 (10.1–10.7) -1.7 -14.0 -4.5 <0.0001
American Indian/

Alaska Native
30.6 (28.3–32.8) 15.0 96.0 25.2 (23.2–27.2) 13.2 109.0 -5.4 0.0004

Asian/Pacific Islander 7.7 (7.2–8.2) -7.9 -50.7 5.1 (4.85–.5) -6.9 -57.5 -2.6 <0.0001
Race/Ethnicity (Males)

White 22.1 (21.8–22.4) Ref. Ref. 17.3 (17.0–17.5) Ref. Ref. -4.9 <0.0001
Black 23.3 (22.6–24.1) 1.2 5.5 18.5 (17.8–19.1) 1.2 6.9 -4.9 <0.0001
Hispanic 21.4 (20.7–22.2) -0.7 -3.1 14.7 (14.2–15.3) -2.6 -14.8 -6.7 <0.0001
American Indian/

Alaska Native
42.7 (38.9–46.5) 20.6 93.1 33.6 (30.3–36.9) 16.3 94.6 -9.1 0.0004

Asian/Pacific Islander 9.7 (8.9–10.5) -12.4 -56.2 6.3 (5.7–7.0) -11.0 -63.4 -3.4 <0.0001
Race/Ethnicity (Females)

White 9.4 (9.2–9.6) Ref. Ref. 7.1 (6.9–7.2) Ref. Ref. -2.3 <0.0001
Black 7.9 (7.5–8.3) -1.5 -16.2 6.4 (6.1–6.8) -0.6 -9.1 -1.5 <0.0001
Hispanic 7.9 (7.5–8.4) -1.5 -15.6 5.7 (5.4–6.1) -1.3 -19.0 -2.2 <0.0001
American Indian/

Alaska Native
18.9 (16.4–21.4) 9.5 100.7 17.3 (15.0–19.5) 10.2 144.6 -1.6 0.3469

Asian/Pacific Islander 5.9 (5.3–6.5) -3.5 -37.2 4.0 (3.6–4.5) -3.0 -42.9 -1.9 <0.0001

Abbreviation: 95% CI = 95% confidence interval; Ref. = Referent.
* Age adjusted death rates per 100,000 population.
† Persons of Hispanic ethnicity might be of any race or combination of races.

Supplement

178 MMWR / November 22, 2013 / Vol. 62 / No. 3

2. CDC. Motor vehicle-related deaths—United States, 2003-2007. In: CDC
health disparities and inequalities report—United States, 2011. MMWR
2011;60(Suppl; January 14, 2011):52-55.

3. CDC. CDC health disparities and inequalities report—United States,
2011. MMWR 2011;60 (Suppl; January 14, 2011).

4. CDC. Introduction: CDC health disparities and inequalities report-
United States, 2013. MMWR 2013;62(No. Suppl 3).

5. Task Force on Community Preventive Services. Motor vehicle–related
injury prevention. Atlanta, GA: Task Force on Community Preventive
Services; 2010. Available at http://www.thecommunityguide.org/mvoi/
index.html.

6. Reede C, Piontkowski S, Tsatoke G. Using evidence-based strategies to
reduce motor vehicle injuries on the San Carlos Apache Reservation. IHS
Primary Care Provider 2007;32:209–12.

7. Letourneau RJ, Crump CD, Thunder N, Voss R. Increasing occupant
restraint use among Ho-Chunk Nation members: tailoring evidence-based
strategies to local context. IHS Primary Care Provider 2009;34:212–7.

8. US Department of Health and Human Services; Westat. Data on health
and well-being of American Indians, Alaska Natives, and other Native
Americans: data catalog. December 2006. Available at http://aspe.hhs.
gov/hsp/06/Catalog-AI-AN-NA/index.htm.

Supplement

MMWR / November 22, 2013 / Vol. 62 / No. 3 179

Introduction
Injury from self-directed violence, which includes suicidal

behavior and its consequences, is a leading cause of death
and disability. In 2009, suicide was the 10th-leading cause
of death in the United States and the cause of 36,909 deaths
(1). In 2005, the estimated cost of self-directed violence (fatal
and nonfatal treated) was $41.2 billion (including $38.9
billion in productivity losses and $2.2 billion in medical costs)
(2). Suicide is a complex human behavior that results from
an interaction of multiple biological, psychological, social,
political, and economic factors (3). Although self-directed
violence affects members of all racial/ethnic groups in the
United States, it often is misperceived to be a problem affecting
primarily non-Hispanic white males (4).

This report is part of the second CDC Health Disparities
and Inequalities Report (CHDIR). The 2011 CHDIR (5) was
the first CDC report to assess disparities across a wide range of
diseases, behavior risk factors, environmental exposures, social
determinants, and health-care access. The topic presented in
this report is based on criteria that are described in the 2013
CHDIR Introduction (6). This report updates information
that was presented in the 2011 CHDIR (7) by providing more
current data on suicide in the United States. The purposes of
this report are to discuss and raise awareness of differences in
the characteristics of suicide decedents and to prompt actions
to reduce these disparities.

Methods
To determine differences in the prevalence of suicide by

sex, race/ethnicity, age, and educational attainment in the
United States, CDC analyzed 2005–2009 data from the Web-
based Injury Statistics Query and Reporting System — Fatal
(WISQARS Fatal) (8) and the National Vital Statistics System
(NVSS). In this report, NVSS data provided as of February
2012 were used. The 2009 data were used to describe the overall
patterns in suicides. The aggregate 2005–2009 reporting period
was used to describe patterns for the combined age group and
race/ethnicity because sample sizes for any single year were

limited. Mortality data were drawn from CDC’s National
Vital Statistics System (NVSS), which collects death certificate
data filed in the 50 states and the District of Columbia (1).
Data in this report include suicides from any cause during
2005–2009. The WISQARS database contains mortality data
based on NVSS and population counts for all U.S. counties
based on U.S. Census data. Counts and rates of death can be
obtained by underlying cause of death, mechanism of injury,
state, county, age, race, sex, year, injury cause of death (e.g.,
firearm, poisoning, or suffocation) and by manner of death
(e.g., suicide, homicide, or unintentional injury) (8).

NVSS codes racial categories as non-Hispanic white, non-
Hispanic black, American Indian/Alaska Native (AI/AN), and
Asian/Pacific Islander (A/PI); ethnicity is coded separately as
Hispanic or non-Hispanic (1). Persons of Hispanic ethnicity
might be of any race or combination of races. Absolute
differences in rates between two populations were compared
using a test statistic, z, based on a normal approximation at a
critical value of α = 0.05 (9).

Educational attainment is recorded by two methods on
death certificates. In 28 states* and the District of Columbia
(DC), the 2003 version of the standard certificate of death is
used (which collects the highest degree completed), whereas 20
states† use the 1989 version of the certificate (which collects
the number of years of education completed). For this reason,
these two groups of states were analyzed separately. Death rates
by educational attainment were based on population estimates
from the U.S. Census Bureau’s 2009 American Community
Survey (ACS) (10). Data for Georgia and Rhode Island were
excluded because educational attainment was not recorded on
their death certificates. Rates are presented only for persons
aged ≥25 years because persons aged <25 years might not have
completed their formal education (9).

Suicides — United States, 2005–2009
Alex E. Crosby, MD1

LaVonne Ortega, MD2
Mark R. Stevens, MSPH, MA1

1National Center for Injury Prevention and Control, CDC
2Center for Surveillance, Epidemiology, and Laboratory Services, CDC

Corresponding author: Alex E. Crosby, Division of Violence Prevention, National Center for Injury Prevention and Control, CDC. Telephone: 770-488-4272;
E-mail: [email protected].

* Arkansas, California, Connecticut, Delaware, Florida, Idaho, Illinois, Indiana,
Kansas, Michigan, Montana, Nebraska, Nevada, New Hampshire, New Jersey,
New Mexico, New York, North Dakota, Ohio, Oklahoma, Oregon, South
Carolina, South Dakota, Texas, Utah, Vermont, Washington, and Wyoming.

† Alabama, Alaska, Arizona, Colorado, Hawaii, Iowa, Kentucky, Louisiana, Maine,
Maryland, Massachusetts, Minnesota, Mississippi, Missouri, North Carolina,
Pennsylvania, Tennessee, Virginia, Wisconsin, and West Virginia.

Supplement

180 MMWR / November 22, 2013 / Vol. 62 / No. 3

Unadjusted (crude) suicide rates were based on resident
population data from the U.S. Census Bureau (10). Rates based
on <20 deaths were considered unreliable and not included
in the analysis. Confidence intervals were calculated in two
ways: 1) groupings of <100 deaths were calculated by using
the gamma method (9), and 2) groupings of ≥100 deaths were
calculated by using a normal approximation (9).

Results
In 2009, a total of 36,909 suicides occurred in the United

States, 83.5% of which were among non-Hispanic whites,
7.0% among Hispanics, 5.5% among non-Hispanic blacks,
2.5% among A/PIs, and 1.1% among AI/ANs (Table).
Although AI/ANs represented the smallest proportion of
suicides of all racial/ethnic groups, they shared the highest rates
with whites. Overall, the crude suicide rate for males (19.2 per
100,000 population) was approximately four times higher than
the rate for females (5.0 per 100,000 population). In each of
the racial/ethnic groups, suicide rates were higher for males
than for females, but the male-female ratio for suicide differs
among these groups. Among non-Hispanic whites, the male-
female ratio was 3.8:1; among Hispanics it was 4.5:1; among
non-Hispanic blacks it was 4.7:1; among A/PIs it was 2.3:1;
and among AI/ANs it was 2.8:1. These male-female ratios did
not change significantly from those reported previously (7).

Overall, suicide rates varied by the level of educational
attainment. Persons with the highest educational attainment
had the lowest rates, those with the lowest educational
attainment had intermediate rates, and those who had
completed only the equivalent of high school (or 12 years of
education) had the highest rates. This pattern was consistent
for males, but the pattern of educational inequalities was
different among females. Females with a lower educational
level had the lowest suicide rates followed by those with the
highest educational level, while those females with a high
school education (12 years of education) had the highest suicide
rates. For each version of the death certificate, whether overall
or by sex, suicide rates differed significantly between levels of
educational attainment, except that rates for females did not
differ significantly between the lowest and highest educational
attainment levels in the states on the basis of data from the
1989 death certificate version.

Suicide rates by race/ethnicity and age group demonstrated
different patterns by racial/ethnic group, with the highest rates
occurring among AI/AN adolescents and young adults aged
15–34 years (Figure). Rates among AI/ANs and non-Hispanic
blacks were highest among adolescents and young adults, then
declined or leveled off with increasing age, respectively. Among

A/PIs and Hispanics, rates were highest among young adults in
their early 20s, then leveled off among other adults but increased
for those aged ≥65 years. In contrast, rates among non-Hispanic
whites were highest among those aged 40–54 years. Although
the 2009 overall rates for AI/ANs are similar to those of non-
Hispanic whites, the 2005–2009 rates among adolescent and
young adult AI/ANs aged 15–29 years were substantially higher.

Discussion
The burden of suicide among AI/AN youths is considerably

higher than that among other racial/ethnic groups. In 2009,
suicide ranked as the fourth leading cause of years of potential
life lost (YPLL) for AI/AN