1) In “Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1”, authors show that … (pick all correct answers) Select one or more: a) On plastic, no viable SARS-CoV-2 was measured a

1) In “Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1”, authors show that … (pick all correct answers) Select one or more: a) On plastic, no viable SARS-CoV-2 was measured a
1 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 b f masks and face coverings in controlling outward aerosol particle emission from expiratory activities Sima Asadi 1, Christopher D. Cappa 2, Santiago Barreda 3, Anthony S. Wexler 2,4,5,6 , Nicole M. Bouvier 7,8 & William D. Ristenpart 1* The COV f w transmission. f homemade masks as acceptable alternatives to surgical masks and N95 respirators. Although mask wearing is intended, in part, to protect others from exhaled, virus-containing particles, few studies have examined particle emission by mask-wearers into the surrounding air. Here, we measured outward emissions of micron-scale aerosol particles by healthy humans performing various expiratory activities while wearing di erent types of medical-grade or homemade masks. Both surgical masks and unvented KN95 respirators, even without t-testing, reduce the outward particle emission rates by 90% and 74% on average during speaking and coughing, respectively, compared to wearing no mask, corroborating their e ectiveness at reducing outward emission. These masks similarly decreased the outward particle emission of a coughing superemitter, who for unclear reasons emitted up to two orders of magnitude more expiratory particles via coughing than average. f shedding of non-expiratory micron-scale particulates from friable cellulosic bers in homemade cotton-fabric masks confounded explicit determination of their e cacy at reducing expiratory particle emission. Audio analysis of the speech and coughing intensity con rmed that people speak more loudly, but do not cough more loudly, when wearing a mask. Further work is needed to establish the e cacy of cloth masks at blocking expiratory particles for speech and coughing at varied intensity and to assess whether virus-contaminated fabrics can generate aerosolized fomites, but the results strongly corroborate the e cacy of medical-grade masks and highlight the importance of regular washing of homemade masks. Airborne transmission of infectious respiratory diseases involves the emission of microorganism-containing aerosols and droplets during various expiratory activities (e.g., breathing, talking, coughing, and sneezing). Transmission of viruses in emitted droplets and aerosols to susceptible individuals may occur via physical contact aer deposition on surfaces, reaerosolization aer deposition, direct deposition of emitted droplets on mucosal surfaces (e.g., mouth, eyes), or direct inhalation of virus-laden aerosols 1,2. Uncertainty remains regarding the role and spatial scale of these dierent transmission modes (contact, droplet spray, or aerosol inhalation) for specic respiratory diseases, including for COVID-19 3 – 7, in particular settings, but airborne transmission stems from the initial expiratory emission of aerosols or droplets. Consequently, the wearing of masks—in addition to open 1Department of Chemical b University of California Davis, 1 Shields Ave, Davis, CA 95616, USA. 2Department of Civil and b b University of California Davis, 1 Shields Ave, Davis, CA 95616, USA. 3Department of Linguistics, University of California Davis, 1 Shields Ave, Davis, CA 95616, USA. 4Department of Mechanical and Aerospace b w CA 95616, USA. 5Air Quality Research Center, University of California Davis, 1 Shields Ave, Davis, CA 95616, USA. 6Department of Land, Air and Water Resources, University of California Davis, 1 Shields Ave, Davis, CA 95616, USA. 7Department of Medicine, Division of f fcahn School of Medicine at Mount Sinai, 1 Gustave L. Levy Place, New York, NY 10029, USA. 8Department of Microbiology, f w Gustave L. Levy Place, New York, NY 10029, USA. *email: [email protected] Vol.:(0123456789 f www.nature.com/scientificreports 2 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 vigilant hand hygiene—has been put forth as a means to mitigate disease transmission, especially in healthcare settings 8 – 11 . Much research has indicated that masks can provide signicant protection to the wearer, although proper mask tting is critical to realizing such benets 12–15 . Alternatively, masks can potentially reduce outward transmission by infected individuals, providing protection to others 7,16,17 . ere have been indications of asymp- tomatic carriers of COVID-19 infecting others 18–20 , leading to increasing, albeit inconsistent 21–24 , calls for more universal wearing of masks or face coverings by the general public to help control disease transmission during pandemics. It is therefore important to understand the ecacy of masks and face coverings of dierent types in reducing outward transmission of aerosols and droplets from expiratory activities. Results from epidemiological and clinical studies assessing the eectiveness of masks in reducing disease transmission suggest that mask wearing can provide some benets 10,11 , especially with early interventions, but oen the results lack statistical signicance 25–31 . Laboratory studies provide another means to assess or infer mask eectiveness. Measurement of material ltration eciencies can provide initial guidance on potential mask eectiveness for preventing outward transmission 15,32– 35 , but do not directly address mask performance when worn. Early photographic evidence indicates masks can limit the spread of cough-generated particles 36. Meas- urements using simulated breathing with an articial test head showed the concentration of particles between 0.02  m-1  m decreases across masks of dierent types 37. Also using simulated breathing, Green et al. 38 found surgical masks eectively reduced outward transmission of endospores and vegetative cells, with seemingly greater reduction of particles > 0.7  m compared to smaller particles. Using volunteers, Davies et al. 32 found that surgical and home-made cotton masks substantially reduce emission of culturable microorganisms from coughing by healthy volunteers, with similar reduction observed over a range of particle sizes (from 0.65  m to > 7  m). Milton et al. 16 found that surgical masks substantially reduced viral copy numbers in exhaled “ne” aerosol (μ 5  m) and “coarse” droplets (> 5  m) from volunteers having in≤uenza, with greater reduction in the coarse fraction. is result diers somewhat from very recent measurements by Leung et al. 13, who showed a statistically signicant reduction in shedding of in≤uenza from breathing in coarse but not ne particles with participants wearing surgical masks. ey did, however, nd that masks reduced shedding of seasonal corona – virus from breathing for both coarse and ne particles, although viral RNA was observed in less than half of the samples even with no mask, complicating the assessment. e above studies all indicate a strong potential for masks to help reduce transmission of respiratory illnesses. To date, however, none have investigated the eectiveness of masks across a range of expiratory activities, and limited consideration has been given to dierent mask types. Furthermore, no studies to date have considered the masks themselves as potential sources of aerosol particles. It is well established that brous cellulosic mate – rials, like cotton and paper, can release large quantities of micron-scale particles (i.e., dust) into the air 39–42 . Traditionally, these particles have not been considered a potential concern for respiratory viral diseases like in≤uenza or now COVID-19, since these diseases have been thought to be transmitted via expiratory particles emitted directly from the respiratory tract of infected individuals 43. Early work in the 1940s indicated, however, that infectious in≤uenza virus could be collected from the air aer vigorously shaking a contaminated blanket 44. Despite this nding, over the next 70 years little attention focused on the possibility of respiratory virus trans – mission via environmental dust; one exception was a study by Khare and Marr, who investigated a theoretical model for resuspension of contaminated dust from a ≤oor by walking 45. Most recently, work by Asadi et al. with in≤uenza virus experimentally established that “aerosolized fomites,” non-respiratory particles aerosolized from virus-contaminated surfaces such as animal fur or paper tissues, can also carry in≤uenza virus and infect susceptible animals 46. is observation raises the possibility that masks or other personal protective equip- ment (PPE), which have a higher likelihood of becoming contaminated with virus, might serve as sources of aerosolized fomites. Indeed, recent work by Liu et al. demonstrated that some of the highest counts of airborne SARS-CoV-2 (the virus responsible for COVID-19) occurred in hospital rooms where health care workers doed their PPE, suggesting that virus was potentially being aerosolized from virus-contaminated clothing or PPE, or resuspended from virus-contaminated dust on the ≤oor 47. It remains unknown what role aerosolized fomites play in transmission of infectious respiratory disease between humans, and it is unclear whether certain types of masks are simultaneously eective at blocking emission of respiratory particles while minimizing emission of non-expiratory (cellulosic) particles. Here, we report on experiments assessing the ecacy of unvented KN95 respirators, vented N95 respirators, surgical masks, and homemade paper and cloth masks at reducing aerosol particle emission rates from breath- ing, speaking, and coughing by healthy individuals. Two key ndings are that (i) the surgical masks, unvented KN95 respirators, and, likely, vented N95 respirators all substantially reduce the number of emitted particles, but that (ii) particle emission from homemade cloth masks—likely from shed ber fragments—can substantially exceed emission when no mask is worn, a result that confounds assessment of their ecacy at blocking expiratory particle emission. Although no direct measurements of virus emission or infectivity were performed here, the results raise the possibility that shed ber particulates from contaminated cotton masks might serve as sources of aerosolized fomites. MethodsHuman subjects. We recruited 10 volunteers (6 male and 4 female), ranging in age from 18 to 45 years old. e University of California Davis Institutional Review Board approved this study (IRB# 844,369–4), and all research was performed in accordance with relevant guidelines and regulations of the Institutional Review Board. Written informed consent was obtained from all participants prior to the tests, and all participants were asked to provide their age, weight, height, general health status, and smoking history. Only participants who self-reported as healthy non-smokers were included in the study. Vol:.(1234567890) www.nature.com/scientificreports/ 3 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 b setup. e general experimental setup used was similar to that in previous work 48,49 . In brief, an aerodynamic particle sizer (APS, TSI model 3321) was used to count the number of particles between 0.3 to 20  m in aerodynamic diameter; the APS counting eciency falls o below ~ 0.5 µm, and thus the particles counted between 0.3 and 0.5 µm likely underestimate the true number. e APS was placed inside a HEPA- ltered laminar ≤ow hood that minimizes background particle concentration (Fig. 1a). Study participants were asked to sit so that their mouth was positioned in front of a funnel attached to the APS inlet via a conductive sili- cone tube. ey then performed dierent expiratory activities while wearing no mask or one of the masks shown in Fig. 1b and described in more detail below. A microphone was placed immediately on the side of the funnel to record the duration and intensity of talking and coughing activities (Fig. 1c). e participants were positioned with their mouth approximately 1 cm away from the funnel entrance; the nose rest used in our previous setup 48,49 was removed to prevent additional particle generation via rubbing of the mask fabric on the nose rest surface. e air was pulled in by the APS at 5 L/min, with 1 L/min (20%) focused into the detector to count and size the cumulative number of particles at 1-s intervals (Fig. 1d). Note that the funnel is a semi-conned environment, and not all expired particles were necessarily captured by the APS. e wearing of masks may redirect some of the expired air≤ow in non-outward directions (e.g., out the top or sides of the mask 50). Accordingly, we use the terminology “outward emission” when referring the to the particle emissions measured here. erefore, the measurements reported here do not represent the absolute number of emitted particles and may underestimate contributions from particles that escape out the sides of the masks, but do allow relative comparisons between dierent conditions. e particle emission rates reported here from the APS are likely smaller than the total expiratory particle emission rates by, approximately, the ratio of the exhaled volumetric ≤owrate that enters the funnel to the APS sample rate. All experiments were performed with ambient temperature between 22 to 24 °C. e relative humidity ranged from 30 to 35% for most experiments; a second round of testing, comparing washed vs. unwashed homemade masks, was performed at 53% relative humidity. Given the approximately 3-s delay between entering the funnel and reaching the detector within the APS, under all these conditions the aqueous components of micron-scale respiratory droplets had more than sucient time (i.e., more than ~ 100 ms) to evaporate fully to their dried Figure 1. (a ) Schematic of the experimental setup showing a participant wearing a mask in front of the funnel connected to the APS. (b ) Photographs of the masks used for the experiments. (c ) Microphone recording for a participant (F3) coughing into the funnel while wearing no mask. (d ) e instantaneous particle emission rate of all detected particles between 0.3 and 20 µm in diameter. Surg.: surgical; KN95: unvented KN95 respirator; SL-P: single-layer paper towel; SL-T: single-layer cotton t-shirt; DL-T: double-layer cotton t-shirt; N95: vented N95 respirator. e subject gave her written informed consent for publication of the images in (b ). Vol.:(0123456789) www.nature.com/scientificreports/ 4 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 residual (so-called “droplet nuclei” 51); see gure S3 of Asadi et al. 48 for direct experimental evidence of complete drying under these conditions. Although large droplets (> 20 µm) can require substantially more than 1 s to evaporate 52, as shown here the vast majority of particles are less than 5 µm and thus unlikely to have originated at sizes larger than 20 µm. e size distributions presented here are based on the diameter as observed at the APS detector. b Participants were asked to complete four distinct activities for each mask or respira- tor type: (i) Breathing : gentle breathing in through the nose and out through the mouth, for 2 min at a pace comfortable for the participant. e particle emission rate was calculated as the total number of particles emitted over the entire 2-min period, divided by two minutes to obtain the average particles per second. (ii) Talking : reading aloud the Rainbow Passage (Fairbanks 53 and Supplementary Text S1), a standard 330- word long linguistic text with a wide range of phonemes. Participants read this passage aloud at an interme – diate, comfortable voice loudness. Since participants naturally read at a slightly dierent volume and pace, the microphone recording was used to calculate the root mean square (RMS) amplitude (as a measure of loudness) and duration of vocalization (excluding the pauses between the words). e particle emission rate was calculated as the total number of particles emitted over the entire reading (approximately 100 to 150 s), divided by the cumulative duration of vocalization excluding pauses. Excluding the pauses accounts for person-to-person dierences in the fraction of time spent actively vocalizing while speaking (approximately 82% ± 5%) so that individuals who simply pause longer between words are not characterized with an articially low emission rate due to vocalization. (iii) Coughing : Successive, forced coughing for 30 s at a comfortable rate and intensity for the participant. Similar to the talking experiment, the microphone data was used to determine the RMS amplitude of each cough, the number of coughs, and cumulative duration of coughing (excluding the pauses between the coughs). e particle emission rate was calculated as the total number of particles emitted during the 30 s of measurement, divided either by the number of coughs (to obtain particles/cough) or by the cumulative duration of the coughs (to obtain particles/s). (iv) Jaw movement: moving the jaw as if chewing gum, without opening the mouth, for 1 min, while nose breathing, to test whether facial motion in the absence of more extreme expiration caused signicant particle emission. is activity technically counts as an expiratory activity since the participant was nose breath- ing, but the main intent was to assess whether facial motion appreciably alters particle emission, due either to gentle friction between the skin and the facemask yielding enhanced particle emission, or variable gap distances between the mask and skin allowing more or less particles to escape. e particle emission rate was calculated as the total number of particles emitted over the 1-min period, divided by 60 s to obtain the average particles per second. Mask types. Participants completed each of the four expiratory activities when they wore no mask or one of the 6 dierent mask or respirator types: (i) A surgical mask (ValuMax 5130E-SB) denoted as “Surg.”, tested by 10 participants. (ii) An unvented KN95 respirator (GB2626-2006, manufacturer Nine Five Protection Technology, Dongguan, China), tested by 10 participants. (iii) A homemade single-layer paper towel mask (Kirkland, 2-PLY sheet, 27.9 cm × 17.7 cm) denoted as “SL- P” and tested by 10 participants. (iv) A homemade single-layer t-shirt mask, “SL-T”, made from a new cotton t-shirt (Calvin Klein Men’s Liquid Cotton Polo, 100% cotton, item #1341469), tested by 10 participants. (v) A homemade double-layer t-shirt mask, “DL-T”, made from the same t-shirt material as the SL-T mask, and tested by 10 participants. (vi)A vented N95 respirator (NIOSH N95, Safety Plus, TC-84A-7448)) tested by 2 participants; shortages at the time of testing precluded a larger sample size. e primary dierence between an N95 and KN95 respira- tor is where the mask is certied, in the US. (N95) or China (KN95). e homemade cloth masks (SL-T, and DL-T) were made according to the CDC do-it-yourself instructions for single- and double-layer t-shirt masks 54. e homemade paper towel masks were made according to do-it- yourself instructions 55. Photographs of all mask types are shown in Fig. 1b. Prior to wearing each mask, participants were verbally given general guidance on how to put on each mask. In the case of surgical masks and KN95 respirators they were instructed to pinch the metal bar to conform the mask to the nose. No t-testing of respirators, as mandated by OSHA standard (29 CFR Part 1910) 56, was performed, with the intent of obtaining representative particle emission rates for untrained individuals without access to professional tting assistance. Mask washing. To test whether washing of the homemade cloth masks had any eect on the particle emis- sion rate, a subset of 4 participants were asked to bring their double-layer t-shirt mask home and to hand-wash it with water and soap, rinse it thoroughly, and let it air-dry. ese participants then returned and repeated the four activities with a brand-new DL-T mask and their washed DL-T mask to provide a direct comparison of washed versus unwashed fabric. Vol:.(1234567890) www.nature.com/scientificreports/ 5 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 Particle emission via hand-rubbing. Besides the above experiments to measure the particle emission associated with dierent mask fabrics, we also performed a qualitative test of the friability of the masks by rub- bing each mask by hand in front of the APS, using a procedure similar to that performed previously with paper tissues (cf. Figure 4 of Asadi et al. 46). Specically, the mask was folded over on itself between thumb and index nger, and the mask material was rubbed against itself. A sample of each mask type was rubbed by hand by the same individual for 10 s in front of the APS, using to the best of their ability the same amount of force each time. e test was repeated 3 times for each mask type. e particle emission rate was calculated as the total number of particles emitted divided by the duration of rubbing (10 s). Note that this procedure does not preclude possible particle shedding from the skin of the experimentalist 57; the observed particle emission rates for dierent mask materials therefore represent only qualitative indications of the relative friability. Statistical analysis. Box-and-whisker plots show the median (red line), interquartile range (blue box), and range (black whiskers). Stata/IC 15 was used to perform Shapiro–Wilk normality test on the particle emis- sion rates for each activity. Aer log-transformation of the data, mixed-eects linear regression was performed to account for person-level correlations. Considering that we had only one primary random eect (person- to-person variability), all variances were set equal with zero covariances. Post hoc pairwise comparisons were performed and adjusted for multiple comparisons using Schee’s method. Schee groups are indicated with green letters below each box plot; groups with no common letter are considered signicantly dierent (p < 0.05). ResultsParticle emission rates for the four expiratory activities are shown in Fig. 2. Focusing rst on breathing (Fig. 2a), when participants wore no mask, the median particle emission rate was 0.31 particles/s, with one participant (M6) as high as 0.57 particles/s, and another participant (F3) as low as 0.05 particles/s. is median rate and person-to-person variability are both broadly consistent with previous studies 48,51 . In contrast, wearing a sur – gical mask or a KN95 respirator signicantly reduced the outward number of particles emitted per second of breathing. e median outward emission rates for these masks were 0.06 and 0.07 particles/s, respectively, representing an approximately sixfold decrease compared to no mask. Wearing a homemade single layer paper towel (SL-P) mask yielded a similar decrease in outward emission rate, although not as statistically signicant as the medical-grade masks. Surprisingly, wearing an unwashed single layer t-shirt (U-SL-T) mask while breathing yielded a signicant increase in measured particle emission rates compared to no mask, increasing to a median of 0.61 particles/s. e rates for some participants (F1 and F4) exceeded 1 particle/s, representing a 384% increase from the median no-mask value. Wearing a double-layer cotton t-shirt (U-DL-T) mask had no statistically signicant eect on the particle emission rate, with comparable median and range to that observed with no mask. Turning to speech (Fig. 2b), the overarching trend observed is that vocalization at an intermediate, comfort – able voice loudness (Figure S1a and Table S1) yielded an order of magnitude more particles than breathing. When participants wore no mask and spoke, the median rate was 2.77 particles/s (compared to 0.31 for breath- ing). e general trend of the mask type eect on the particle emission was qualitatively similar to that observed for breathing. Wearing surgical masks and KN95 respirators while talking signicantly decreased the outward emission by an order of magnitude, to median rates of 0.18 and 0.36 particles/s, respectively. Likewise, wearing the paper towel mask reduced the outward speech particle emission rate to 1.21 particles/s, lower than no mask but representing a less pronounced decrease compared to surgical masks and KN95 respirators. In contrast, the homemade cloth masks again yielded either no change or a signicant increase in emission rate during speech compared to no mask. e outward particle emissions when participants wore U-SL-T masks exceeded the no- mask condition by an order of magnitude with a median value of 16.37 particles/s. Wearing the U-DL-T mask had no signicant eect. e third expiratory activity – coughing – again yielded qualitatively similar trends with respect to mask type (Fig. 2c). We emphasize that participants coughed at dierent paces, and therefore the number of coughs, cumulative cough duration, and acoustic power varied between participants (Figure S1b, Figure S2, and Table S2). Nonetheless, we observe that coughing with no mask produced a median of 10.1 particles/s, with most partici – pants in the range of 3 to 42 particles/s. For comparison, given a coughing rate of 6 times per minute, the median outward particle count due specically to coughing over that minute is slightly smaller than that from breathing, and an order of magnitude smaller than talking over a minute (see Fig. S3 for equivalent numbers of particles per cough). Similar general trends as for breathing and speaking were observed for coughing when wearing the dierent mask types. e surgical mask decreased the median outward emission rate to 2.44 particles/s (75% decrease), while the KN95 yielded an apparent but not statistically signicant decrease to 6.15 particles/s (39% decrease). e SL-P mask yielded no statistically signicant dierence compared to no mask. In contrast, the homemade U-SL-T and U-DL-T masks however yielded a signicant increase in outward particle emission per second (or per cough) compared to no mask, with median emission rates of 49.2 and 36.1 particles/s, respectively. Notably, one individual, M6, emitted up to two orders of magnitude more aerosol particles while cough- ing than the others, emitting 567 particles/s with no mask. Even when M6 wore a surgical mask he emitted 19.5 particles/s while coughing, substantially above the median value for no mask, although still a substantial decrease compared to no mask for this individual. Acoustic analysis of the coughing, both in terms of the root mean square amplitude (Figure S1b) and the ltered power density, indicate that the coughs by M6 were not particularly louder nor more energetic than the others (see Figure S2 and Table S2). It is unclear what caused this individual to emit a factor of 100 more aerosol particles than average while coughing, although qualitatively, the coughs of M6 appeared to originate more from the chest, compared to other participants for whom coughs generally appeared to originate more from the throat; notably, this individual emitted a much closer to average Vol.:(0123456789) www.nature.com/scientificreports/ 6 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 amount of particles while speaking and breathing. Furthermore, the signicantly higher aerosol particle emission compared to average during coughing for M6 persisted regardless of the mask type. Finally, Fig. 2d shows the particle emission rate when participants moved their jaw, similar to chewing gum with their mouth closed, while only breathing through their nose. In general, jaw movement with nose breathing and no mask produced slightly fewer particles per second than the breathing activity (breathing in through nose and out through mouth), with a median rate of 0.12 particles/s for no mask. As participants were still breath- ing with closed mouth during the jaw movement, the lower particle production likely results from participants exhaling through the nose rather than through the mouth 48,51 . Wearing a surgical mask or KN95 respirator had no statistically signicant eect on particle emission from jaw movement compared to no mask. In contrast, wearing all other types of homemade masks (SL-P, U-SL-T, and U-DL-T) substantially increased the particle emission rate, with the single-layer mask yielding the most at 1.72 particles/s. All of the above experiments were also repeated with vented N95 respirators, albeit with only 2 participants (due to shortages at the time of testing). e small sample size precludes signicance testing, but in general the particle emission rates of the two tested were comparable to the surgical mask and unvented KN95 in terms of reduction in the overall emission rates. e emission rates presented in Fig. 2 represent the total for all particles in the size range 0.3 to 20 µm. We also measured the corresponding size distributions in terms of overall fraction for all trials (Fig. 3). In general, all size distributions observed here were lognormal, with a peak somewhere near 0.5 µm and decaying rapidly to negli- gible fractions above 5 µm. Breathing while wearing no mask emitted particles with a geometric mean diameter of 0.65 µm (Fig. 3a), with 35% of the particles in the smallest size range of 0.3 to 0.5 µm. Regardless of the mask Figure 2. Particle emission rates associated with (a ) breathing, (b) talking, (c) coughing, and (d ) jaw movement when participants wore no mask or when they wore one of the six mask types considered. Schee groups are indicated with green letters; groups with no common letter are considered signicantly dierent (p < 0.05). Surg.: surgical; KN95: unvented KN95; SL-P: single-layer paper towel; U-SL-T: unwashed single-layer cotton t-shirt; U-DL-T: unwashed double-layer cotton t-shirt; N95: vented N95. Note that the scales are logarithmic and the orders of magnitude dier in each subplot. Vol:.(1234567890) www.nature.com/scientificreports/ 7 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 type, wearing masks while breathing signicantly increased this fraction of particles in the smallest size range (e.g., to as high as 60% for KN95 respirator), shiing the geometric mean diameter toward smaller sizes. Talking with no mask yielded slightly larger particles compared to breathing, with mean diameter of 0.75 µm (Fig. 3b). Wearing a mask while talking aected the size distribution in a qualitatively similar manner to that observed with breathing, in that a higher fraction of particles were in the smallest size range. Unlike breathing however, the U-SL-T and U-DL-T masks released the highest fractions of small particles (47% and 51%, respectively). e eect of wearing a mask was more pronounced on the size distribution of the particles produced by coughing (Fig. 3c). For no mask, the mean diameter of cough-generated particles was 0.6 µm. e majority of particles emitted were in the smallest size range (up to 57%) during coughing while wearing homemade masks (SL-P, U-SL-T, and U-DL-T). We also note that for coughing, which produced the highest rates of particle emis – sion for of all expiratory activities tested, wearing homemade masks considerably reduced the fraction of large particles (> 0.8 µm). Finally, for jaw movement the overall size distributions for no-mask and with-mask cases were similar except that the fraction of smallest particles was lowest for no-mask and the surgical mask (Fig. 3d). To provide a direct comparison of the ecacy of medical-grade and homemade masks in mitigating the emission of particles of dierent sizes, we divided the entire size range measured by APS (0.3 – 20 µm) into three sub-ranges (smallest, 0.3 – 0.5 µm; intermediate, 0.5 – 1 µm; and largest, 1 – 20 µm), and calculated the corresponding percent change in the median particle emission rate of each sub-range during breathing, talking, and coughing compared to no mask (Fig. 4). For the smallest particles, Fig. 4a shows that up to a 92% reduction in 0.3 – 0.5 µm particle emission rate occurred while wearing surgical and KN95 masks for breathing, talking, and coughing, with the KN95 yielding a smaller decrease of 20.5% in this size range. e SL-P mask caused a 60% reduction in 0.3 – 0.5 µm particle emission for talking and breathing, but yielded a 77% increase for coughing. e least eective masks in terms of minimizing emissions of the smallest particles were the U-SL-T and U-DL-T masks, with U-SL-T substantially increasing the emission of 0.3 – 0.5 µm particles by almost 600% for speech, and the U-DL-T mask yielding very slight changes for talking and breathing and an almost 300% increase for coughing. Qualitatively similar trends were observed for intermediate size particles in the range of 0.5 – 1 µm (Fig. 4b), with the medical-grade masks yielding signicant reductions. e main dierence for this size range is Figure 3. Observed particle size distributions, normalized by particles/s per bin, associated with (a ) breathing, ( b ) talking, (c ) coughing, and (d ) jaw movement when participants wore no mask or one of the ve mask types considered. Each curve is the average over all 10 participants. e solid lines represent the data using a 5 -point smoothing function. Data points with horizontal error bars show the small particles ranging from 0.3 to 0.5  m in diameter detected by APS with no further information about their size distribution in this range. Surg.: surgical; KN95: unvented KN95; SL-P: single-layer paper towel; U-SL-T: unwashed single-layer cotton t-shirt; U-DL-T: unwashed double-layer cotton t-shirt; N95: vented N95. Vol.:(0123456789) www.nature.com/scientificreports/ 8 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 that the SL-P mask yielded a 15.7% decrease in particle emissions for coughing, and the U-DL-T mask provided up to 34.1% reduction in particle emissions for breathing and talking. As for the largest particle sizes (1 – 20 µm), the observed trends were again qualitatively similar to the intermediate particles (Fig. 4c), with the medical-grade masks yielding large reductions. Notably, the U-DL-T mask emitted much fewer large particles for breathing and talking with approximately 60% reductions, but still a sizable 160% increase for coughing. e percent change in median particle emission over the entire size range of 0.3 – 20 µm is presented in Fig. 4d, which shows that the homemade masks in general yielded more particles in total for coughing, and had mixed ecacy in reducing particle emissions for breathing and talking. e key point is that the surgical and KN95 masks eectively decreased the particle emission for all expiratory activities tested here over the entire range of particle sizes measured by the APS. To help interpret our ndings we also quantied the particles emitted from manual rubbing of mask fabrics. e results (Fig. 5a) show that, in the absence of any expiratory activity, rubbing a surgical mask fabric generated on average 1.5 particles per second, while KN95 and N95 respirators produced fewer than 1 particle per second. In contrast, rubbing the homemade paper and cotton masks aerosolized signicant number of particles, with the highest values for SL-P (8.0 particles/s) and U-SL-T (7.2 particles/s) masks. Intriguingly, we found that the size distribution of the particles aerosolized from homemade mask fabrics via manual rubbing (Fig. 5b) was qualitatively dierent from when participants wore the same masks to perform expiratory activities. An extra Figure 4. Percent change in median particle emission rate (N) for 10 participants compared to no-mask median, while wearing dierent mask types and while breathing (blue points), talking (red points), or coughing (green points), for particles in the following size ranges: (a ) smallest, 0.3–0.5 µm; (b) intermediate, 0.5–1 µm; ( c ) largest, 1–20 µm; and (d ) all sizes, 0.3–20 µm. e dashed lines are to guide the eye. Surg.: surgical; KN95: unvented KN95; SL-P: single-layer paper towel; U-SL-T: unwashed single-layer cotton t-shirt; U-DL-T: unwashed double-layer cotton t-shirt. Vol:.(1234567890) www.nature.com/scientificreports/ 9 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 peak appeared at approximately 6 µm and the fraction of small particles dropped to below 27%, suggesting that the frictional forces of bers against bers helped fragment and dislodge larger particulates into the air. Impor – tantly, however, manual rubbing produced a sizeable number of particulates in the size range of 0.3 to 2 µm, commensurate with the range observed while the masks were worn during expiratory activities. Note that the coarse skin particulates (> 2 µm) released from hand during the mask fabric rubbing experiments could have contributed to the observed particle counts 57. However, since this factor was the same in all the manual rubbing experiments, and only facemask fabrics diered, it is dicult to explain the observed trends solely in terms of friction between skin and mask fabrics. Moreover, although in these experiments the applied tribological force was not strictly controlled or quantied, the presented results strongly suggest that cotton fabric masks have much more friable material, consistent with our observation that more particles are emitted when participants perform expiratory activities in those cotton fabric masks. Since the cotton masks were all prepared from fabric that was brand new and unwashed, as a nal test we hypothesized that perhaps washing the masks would remove surface-bound dust and otherwise friable material and decrease the emission rate. Our experiments do not corroborate this hypothesis. Handwashing the double- layer t-shirt mask with soap and water followed by air-drying yielded no signicant change in the particle emission rate as compared to the original unwashed masks (Fig. 6). Moreover, manual rubbing of a washed double-layer cotton mask aerosolized slightly more particles than the unwashed mask. ese results suggest that a single washing has little impact on the presence of aerosolizable particulate matter in standard cotton fabrics. Note also that the ranges observed here accord qualitatively with the prior measurements taken with the same 4 participants on a previous day (compare the results for each category in Fig. 6 versus the U-DL-T columns for the respective expiratory activities in Fig. 2). is observation suggests that day-to-day variability for a given individual is less than the person-to-person variability observed for all expiratory activities and mask types tested. DiscussionOur results clearly indicate that wearing surgical masks or unvented KN95 respirators reduce the outward particle emission rates by 90% and 74% on average during speaking and coughing, respectively, compared to wearing no mask. However, for the homemade cotton masks, the measured particle emission rate either remained unchanged (DL-T) or increased by as much as 492% (SL-T) compared to no mask for all of the expiratory activities. For jaw movement, the particle emission rates for homemade paper and cloth masks were an order of magnitude larger than that of no mask (Fig. 2d). ese observations, along with our results from manual mask rubbing experi- ments (Fig. 5), provide strong evidence of substantial shedding of non-expiratory micron-scale particulates from friable cellulosic bers of the paper and cloth masks owing to mechanical action 40. e higher particle emission rate for jaw movement than for breathing is an indication of greater frictional shedding of the paper towel and cotton masks during jaw movement compared to breathing, at least as tested here. Likewise, the dierence in the size distributions of mask rubbing and with-mask expiratory activities is likely due to the vigorous frictional force applied by hand on the masks. Regardless of the larger particles (> 5 µm), rubbing mask fabrics generates a considerable number of particles in the range of 0.3–5 µm similar to that observed for the expiratory activities. Figure 5. (a ) Number of particles emitted per second of manual rubbing for all masks tested. Each data point is time-averaged particle emission rate over 10 s of rubbing. (b ) Corresponding size distribution for homemade paper and cotton masks for a total of 30 s of manual rubbing in front of the APS. e solid lines represent the data using a 5 -point smoothing function. Data points with horizontal whiskers show the small particles ranging from 0.3 to 0.5  m in diameter detected by APS. Surg.: surgical; KN95: unvented KN95; SL-P: single-layer paper towel; U-SL-T: unwashed single-layer cotton t-shirt; U-DL-T: unwashed double-layer cotton t-shirt; N95: vented N95. Vol.:(0123456789) www.nature.com/scientificreports/ 10 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 is nding corroborates the interpretation that some proportion of the particulates observed during expiration were particulates aerosolized from the masks themselves. Another factor to consider is that masks can reduce the intelligibility of the speech signal 58, and can reduce the intensity of sounds passed through them by a signicant amount (e.g., > 10 dB in Saedi et al.59). Likely as a response to this, people will speak louder and otherwise adjust their speech when wearing masks. Mendel et al. 60 found that the measured intensity of speech was approximately the same for a group of speakers with and with- out surgical masks, suggesting that speakers increased the actual intensity of their speech when wearing masks. Fecher 61 found that speakers will actually produce louder output through some types of masks in cases where they overestimate the dampening eects of the mask. It is also possible that speakers may produce Lombard speech when wearing certain types of masks 62. Lombard speech is louder, has a higher fundamental frequency, and tends to have longer vowel durations, all characteristics that may contribute to an increase in the emission of aerosols 48, 49 . Our results showed that the root mean square amplitude of speech, as measured externally when participants wore any type of mask, equaled or exceeded that of the no-mask condition (Figure S1a), suggesting that participants were indeed talking louder with the mask. Although an increase in the intensity of the speech signal when wearing masks would result in greater output of particles in these conditions 48, the dierence in the intensity of speech across the dierent conditions was not very large (Figure S1a). As a result, this mechanism alone cannot explain the increased particle output in some of the masked conditions. Intriguingly, the root mean square amplitude of coughing decreased for most of the participants aer they wore a type of mask (Figure S1b), suggesting that they do not cough louder when they wear a mask, i.e., there is no Lombard eect for coughing. e substantial particle shedding by the cloth masks confounds determination of the cloth mask ecacy for reducing outward emission of particles produced from the expiratory activity. Measured material ltration eciencies vary widely for dierent cloth materials 32,34,35,63 . e in≤uence of particle shedding on such deter – minations has not been previously considered; our results raise the possibility that particle shedding has led to underestimated material ltration eciencies for certain materials. While the material eciency of the cotton masks was not determined here, we note that the use of the double-layer cotton masks reduced the emission of larger particles (both on a normalized and absolute basis), indicating some reasonable ecacy towards reduction of the expiratory particle emission. Further work dierentiating between expiratory and shed particles, possibly based on composition, can help establish the specic ecacy of the cloth masks towards expiratory particles. at the masks shed bers from mechanical stimulation indicates care must be taken when removing and cleaning (for reusable masks) potentially contaminated masks so as to not dislodge deposited micro-organisms. We also note that the emission reduction due to surgical masks was greater than the corresponding reduction due to KN95 respirators, although this dierence was only signicant for coughing (p < 0.05). at the surgical masks appear to provide slightly greater reduction than the KN95 respirators is perhaps surprising, as KN95s are commonly thought of as providing more protection than surgical masks for inhalation. Both surgical masks and KN95 respirators typically have high material ltration eciencies (> 95%) 63, although the quality of surgi- cal masks can vary substantially 64. e t of surgical and KN95 respirators diers substantially. Here, no t tests were performed to ensure good seals of the KN95 respirators. It may be that imperfect tting of KN95 respirators allows for greater escape of particles from the mask-covered environment compared to the more ≤exible surgical Figure 6. Particle emission rate from breathing, talking, coughing and jaw movement for 4 participants wearing unwashed or washed double-layer t-shirt masks (U-DL-T vs. W-DL-T). Last column shows the particles emission rates for manual rubbing of washed and unwashed masks (three 10-s trials for each mask). Vol:.(1234567890) www.nature.com/scientificreports/ 11 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 masks. Regardless, all surgical masks, KN95 and N95 respirators tested here provided substantial reduction of particle emission compared to no mask. A particularly important observation was the existence of a coughing superemitter, who for unknown reasons emitted two orders of magnitude more particles during coughing than average (Fig. 2c, red points for M6). is huge dierence persisted regardless of mask type, with even the most eective mask, the surgical mask, only reducing the rate to a value twice the median value for no mask at all. Although the underlying mechanism lead- ing to such enhanced particle emission is unclear, these observations nonetheless conrm that some people act as superemitters during coughing, similar to “speech superemitters” 48, and “breathing high producers” 65. is observation raises the possibility that coughing superemitters could serve as superspreaders who are dispropor – tionately responsible for outbreaks of airborne infectious disease. Notably, the coughing superemitter was not a breathing superemitter or speaking superemitter, indicating that testing only one type of expiratory activity might not necessarily identify superemitters for other expiratory activities. As a nal comment, we emphasize that here we only measured the physical dynamics of outward aerosol particle emission for dierent expiratory activities and mask types. Redirected expiratory air≤ow, involving exhaled air moving up past the nose or out the side of the mask, were not measured here but should be consid- ered in future work. Likewise, more sophisticated biological techniques are necessary to gauge mask ecacy at blocking emission of viable pathogens. Our work does raise the possibility, however, that virus-contaminated masks could release aerosolized fomites into the air by shedding ber particulates from the mask fabric. Since mask ecacy experiments are typically only conducted with fresh, not used, masks, future work assessing emis – sion of viable pathogens should consider this possibility in more detail. Our work also raises questions about whether homemade masks using other fabrics, such as polyester, might be more ecient than cotton in terms of blocking expiratory particles while minimizing shedding of fabric particulates, and whether repeated washings might aect homemade masks. Future experiments using controlled bursts of clean air through the masks will help to resolve the source of these non-expiratory particles. Nonetheless, as a precaution, our results suggest that individuals using homemade fabric masks should take care to wash or otherwise sterilize them on a regular basis to minimize the possibility of emission of aerosolized fomites. Conclusionsese observations directly demonstrate that wearing of surgical masks or KN95 respirators, even without t- testing, substantially reduce the number of particles emitted from breathing, talking, and coughing. While the ecacy of cloth and paper masks is not as clear and confounded by shedding of mask bers, the observations indicate it is likely that they provide some reductions in emitted expiratory particles, in particular the larger particles (> 0.5  m). We have not directly measured virus emission; nonetheless, our results strongly imply that mask wearing will reduce emission of virus-laden aerosols and droplets associated with expiratory activities, unless appreciable shedding of viable viruses on mask bers occurs. e majority of the particles emitted were in the aerosol range (< 5  m). As inertial impaction should increase as particle size increases, it seems likely that the emission reductions observed here provide a lower bound for the reduction of particles in the droplet range (> 5  m). Our observations are consistent with suggestions that mask wearing can help in mitigating pandem – ics associated with respiratory disease. Our results highlight the importance of regular changing of disposable masks and washing of homemade masks, and suggests that special care must be taken when removing and cleaning the masks. Data availabilitye datasets generated during and/or analyzed during the current study are available in the Dryad Digital Reposi- t o r y, https ://doi.org/10.25338 /B87C9 V. Received: 9 June 2020; Accepted: 23 August 2020 References 1. Jones, R. M. & Brosseau, L. M. Aerosol transmission of infectious disease. J. Occup. Environ. Med. 57, 501–508 (2015). 2. Tellier, R., Li, Y., Cowling, B. J. & Tang, J. W. Recognition of aerosol transmission of infectious agents: a commentary. BMC Infect. Dis. 19, 101 (2019). 3. Asadi, S., Bouvier, N., Wexler, A. S. & Ristenpart, W. D. e coronavirus pandemic and aerosols: does COVID-19 transmit via expiratory particles?. Aerosol. Sci. Technol. 54, 635–638 (2020). 4. Bourouiba, L. Turbulent gas clouds and respiratory pathogen emissions: potential implications for reducing transmission of COVID-19. JAMA 323, 1837–1838 (2020). 5. Dancer, S.J., Tang, J.W., Marr, L.C., Miller, S., Morawska, L., Jimenez, J.L. Putting a balance on the aerosolization debate around SARS-CoV-2. J. Hosp. Infect. (2020). 6. Peters, A., Parneix, P., Otter, J., Pittet, D. Putting some context to the aerosolization debate around SARS-CoV-2. J. Hosp. Infect. (2020). 7. Allen, J., Marr, L., Re-thinking the potential for airborne transmission of SARS-CoV-2. preprints 2020, 2020050126. https ://doi. org/10.20944 /prepr ints2 02005 .0126.v1 . 8. Gralton, J. & McLaws, M. L. Protecting healthcare workers from pandemic in≤uenza: N95 or surgical masks?. Crit. Care Med. 38, 657–667 (2010). 9. Bartoszko, J. J., Farooqi, M. A. M., Alhazzani, W. & Loeb, M. Medical masks vs N95 respirators for preventing COVID-19 in health- care workers: a systematic review and meta-analysis of randomized trials. Inuenza Other Respir. Virus. https ://doi.org/10.1111/ irv.12745 (2020). 10. MacIntyre, C. R. & Wang, Q. Physical distancing, face masks, and eye protection for prevention of COVID-19. e Lancet 20, 31183 (2020). Vol.:(0123456789) www.nature.com/scientificreports/ 12 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 11. Chu, D. K. et al. Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-CoV-2 and COVID-19: a systematic review and meta-analysis. e Lancet 20, 31142 (2020). 12. Bałazy, A. et al. Do N95 respirators provide 95% protection level against airborne viruses, and how adequate are surgical masks?. Am. J. Infect. Control 34, 51–57 (2006). 13. Leung, N. H. L. et al. Respiratory virus shedding in exhaled breath and ecacy of face masks. Nat. Med. 26, 676–680 (2020). 14. Bowen, L. E. Does that face mask really protect you?. Appl. Biosaf. 15, 67–71 (2010). 15. Rengasamy, S., Eimer, B. & Shaer, R. E. Simple respiratory protection—evaluation of the ltration performance of cloth masks and common fabric materials against 20–1000 nm size particles. Ann. Occup. Hyg. 54, 789–798 (2010). 16. Milton, D.K., Fabian, M.P., Cowling, B.J., Grantham, M.L., McDevitt, J.J. In≤uenza virus aerosols in human exhaled breath: particle size, culturability, and eect of surgical masks. Plos Pathog. 9, (2013). 17. Prather KA, Wang CC, Schooley RT. Reducing transmission of SARS-CoV-2. Science , eabc6197 (2020). 18. He, X. et al. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nat. Med. 26, 672–675 (2020). 19. Rothe C , et al. Transmission of 2019-nCoV infection from an asymptomatic contact in Germany. N. Engl. J. Med. 382, (2020). 20. Wölfel R , et al. Virological assessment of hospitalized patients with COVID-2019. Nature (2020). 21. Cheng, K.K., Lam, T.H., Leung, C.C. Wearing face masks in the community during the COVID-19 pandemic: altruism and soli – darity. e Lancet (2020). 22. Feng, S. et al. Rational use of face masks in the COVID-19 pandemic. e Lancet Respirat. Med. 8 , 434–436 (2020). 23. U. S. Center for Disease Control, Recommendation Regarding the Use of Cloth Face Coverings, Especially in Areas of Signicant Community-Based Transmission: https ://www.cdc.gov/coron aviru s/2019-ncov/preve nt-getti ng-sick/cloth -face-cover .html , access: 19 May 2020, 2020. 24. World Health Organization: Advice on the use of masks in the context of COVID-19, WHO/2019-nCov/IPC_Masks/2020.3, 2020. 25. Jeerson T , et al. Physical interventions to interrupt or reduce the spread of respiratory viruses. Cochrane Database System. Rev. (2011). 26. Wong, V. W. Y., Cowling, B. J. & Aiello, A. E. Hand hygiene and risk of in≤uenza virus infections in the community: a systematic review and meta-analysis. Epidemiol. Infect. 142, 922–932 (2014). 27. Lau, M. S. Y., Cowling, B. J., Cook, A. R. & Riley, S. Inferring in≤uenza dynamics and control in households. Proc. Natl. Acad. Sci. 112, 9094 (2015). 28. MacIntyre CR, Chughtai AA. Facemasks for the prevention of infection in healthcare and community settings. BMJ: Br. Med. J. 350, h694 (2015). 29. Smith, S. M. S. et al. Use of non-pharmaceutical interventions to reduce the transmission of in≤uenza in adults: a systematic review. Respirology 20, 896–903 (2015). 30. Saunders-Hastings, P., Crispo, J. A. G., Sikora, L. & Krewski, D. Eectiveness of personal protective measures in reducing pandemic in≤uenza transmission: a systematic review and meta-analysis. Epidemics 20, 1–20 (2017). 31. Dharmadhikari, A. S. et al. Surgical face masks worn by patients with multidrug-resistant tuberculosis: impact on infectivity of air on a hospital ward. Am. J. Respir. Crit. Care Med. 185, 1104–1109 (2012). 32. Davies, A. et al. Testing the ecacy of homemade masks: would they protect in an in≤uenza pandemic?. Disaster Med. Public Health Preparedness 7 , 413–418 (2013). 33. Harnish, D. A. et al. Challenge of N95 ltering facepiece respirators with viable H1N1 in≤uenza aerosols. Infect. Control Hosp. Epidemiol. 34, 494–499 (2013). 34. Mueller, W. et al. e eectiveness of respiratory protection worn by communities to protect from volcanic ash inhalation. Part I: Filtration eciency tests. Int. J. Hyg. Environ. Health 221, 967–976 (2018). 35. Jung, H. et al. Comparison of ltration eciency and pressure drop in anti-yellow sand masks, quarantine masks, medical masks, general masks, and handkerchiefs. Aerosol. Air Qual. Res. 14, 991–1002 (2014). 36. Shah, M., Crompton, P. & Vickers, M. D. e ecacy of face masks. Ann. R. Coll. Surg. Engl. 65, 380–381 (1983). 37. van der Sande, M., Teunis, P. & Sabel, R. Professional and home-made face masks reduce exposure to respiratory infections among the general population. PLoS ONE 3, e2618 (2008). 38. Green, C. F. et al. Eectiveness of selected surgical masks in arresting vegetative cells and endospores when worn by simulated contagious patients. Infect. Control Hosp. Epidemiol. 33, 487–494 (2012). 39. Licina, D., Tian, Y. & Nazaro, W. W. Emission rates and the personal cloud eect associated with particle release from the perihu- man environment. Indoor Air 27, 791–802 (2017). 40. Licina, D. & Nazaro, W. W. Clothing as a transport vector for airborne particles: chamber study. Indoor Air 28, 404–414 (2018). 41. McDonagh, A. & Byrne, M. A. A study of the size distribution of aerosol particles resuspended from clothing surfaces. J. Aerosol. Sci. 75, 94–103 (2014). 42. S Mehta SPHPATAGB. Evaluating textiles and apparel for controlling contamination in cleanrooms performance of protective clothing: fourth volume. ASTM International (1992). 43. Killingley, B. & Nguyen-Van-Tam, J. Routes of in≤uenza transmission. Inuenza Other Respir. Viruses 7, 42–51 (2013). 44. Edward, D. G. F. Resistance of in≤uenza virus to drying and its demonstration on dust. Lancet 2 , 664–666 (1941). 45. Khare, P. & Marr, L. C. Simulation of vertical concentration gradient of in≤uenza viruses in dust resuspended by walking. Indoor Air 25, 428–440 (2015). 46. Asadi, S. et al. In≤uenza A virus is transmissible via aerosolized fomites. Nat. Commun. 11, 4062 (2020). 47. Liu, Y . et al. Aerodynamic analysis of SARS-CoV-2 in two Wuhan hospitals. Nature , (2020). 48. Asadi, S. et al. Aerosol emission and superemission during human speech increase with voice loudness. Sci. Rep. 9 , 2348 (2019). 49. Asadi, S. et al. Eect of voicing and articulation manner on aerosol particle emission during human speech. PLoS ONE 15, e0227699–e0227699 (2020). 50. Tang, J. W., Liebner, T. J., Craven, B. A. & Settles, G. S. A schlieren optical study of the human cough with and without wearing masks for aerosol infection control. J. R. Soc. Interface 6 , S727–S736 (2009). 51. Morawska, L. et al. Size distribution and sites of origin of droplets expelled from the human respiratory tract during expiratory activities. J. Aerosol. Sci. 40, 256–269 (2009). 52. Su, Y. Y., Miles, R. E. H., Li, Z. M., Reid, J. P. & Xu, J. e evaporation kinetics of pure water droplets at varying drying rates and the use of evaporation rates to infer the gas phase relative humidity. Phys. Chem. Chem. Phys. 20, 23453–23466 (2018). 53. Fairbanks, G. Voice and articulation drillbook Vol. 2 (Harper & Row, New York, 1960). 54. U. S. Center for Disease Control, How to Make Cloth Face Coverings: https ://www.cdc.gov/coron aviru s/2019-ncov/preve nt-getti ng-sick/how-to-make-cloth -face-cover ing.html. Accessed 24 May 2020. 55. Cray Journal-Imaginative cra ideas and tip, DIY paper towel face mask: s https ://cra yjour nal.com/diy-paper -towel -face-mask/. Accessed 28 May 2020. 56. Occupational Safety and Health Administration, Respiratory Fit Testing: https ://www.osha.gov/video /respi rator y_prote ction /tte sting .html , Accessed 28 May 2020. 57. Bhangar, S. et al. Chamber bioaerosol study: human emissions of size-resolved ≤uorescent biological aerosol particles. Indoor Air 26, 193–206 (2016). 58. Palmiero, A. J., Symons, D., Morgan, J. W. & Shaer, R. E. Speech intelligibility assessment of protective facemasks and air-purifying respirators. J. Occup. Environ. Hyg. 13, 960–968 (2016). Vol:.(1234567890) www.nature.com/scientificreports/ 13 Scientific RepoRtS | (2020) 10:15665 | https://doi.org/10.1038/s41598-020-72798-7 59. Saeidi, R., Huhtakallio, I., Alku, P. Analysis of face mask eect on speaker recognition. In INTERSPEECH, 1800–1804, (2016). 60. Mendel, L. L., Gardino, J. A. & Atcherson, S. R. Speech understanding using surgical masks: a problem in health care?. J. Am. Acad. Audiol. 19, 686–695 (2008). 61. Fecher, N. Eects of forensically-relevant facial concealment on acoustic and perceptual properties of consonants, Doctoral dis- sertation, University of York (2014). 62. Bond, Z. S., Moore, T. J. & Gable, B. Acoustic-phonetic characteristics of speech produced in noise and while wearing an oxygen mask. J. Acoust. Soc. Am. 85, 907–912 (1989). 63. Konda, A., Prakash, A., Moss, G.A., Schmoldt, M., Grant, G.D., Guha, S. Aerosol ltration eciency of common fabrics used in respiratory cloth masks. ACS Nano (2020). 64. Oberg, T. & Brosseau, L. M. Surgical mask lter and t performance. Am. J. Infect. Control 36, 276–282 (2008). 65. Edwards, D. A. et al. Inhaling to mitigate exhaled bioaerosols. Proc. Natl. Acad. Sci. U.S.A. 101, 17383–17388 (2004). Acknowledgementsis research was supported by National Institute of Allergy and Infectious Diseases of the National Institutes of Health (NIAID/NIH), Grant R01 AI110703. Author contributionsS.A. performed the APS measurements of airborne particulates, fabricated the homemade masks, conducted the statistical analyses, and prepared gures. S.A., C.D.C, and W.D.R. analyzed the APS data and wrote the manu – script. S.A. and S.B. analyzed and interpreted the acoustic data. All authors reviewed and revised the manuscript for accuracy and intellectual content. Competing interests e authors declare no competing interests. Additional informationSupplementary information is available for this paper at https ://doi.org/10.1038/s4159 8-020-72798 -7. Correspondence and requests for materials should be addressed to W.D.R. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional aliations. Open Access is article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. e images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. © e Author(s) 2020 Vol.:(0123456789) www.nature.com/scientificreports/
1) In “Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1”, authors show that … (pick all correct answers) Select one or more: a) On plastic, no viable SARS-CoV-2 was measured a
Virologic features of SARS -CoV -2 infection in children 1 2 Lael M. Yonker 1,2,3 *, Julie Boucau 4*, James Regan 5, Manish C. Choudhary 3,5, M adeleine D. 3 Burns 1, Nicola Young 1, Eva J . Farkas 1, Jameson P. Davis 1, Peter P. Moschovis 2,3, T. Bernard 4 Kinane 2,3, Alessio Fasano 1,2,3 , Anne M. Neilan 2,3,6 , Jon athan Z. Li 3,5*, Amy K. Barczak 3,4,6 * 5 6 1. Massachusetts General Hospital, Mucosal Immunology and Biology Research Center, 7 Boston, MA, USA 8 2. Massachusetts General Hospital, Department of Pediatrics, Boston, MA, USA 9 3. Ha rvard Medical School, Boston, MA, USA 10 4. Ragon Institute of MGH , MIT and Harvard, Cambridge, MA, USA 11 5. Brigham and Women’s Hospital, Department of Medicine , Boston, MA, USA 12 6. Massachusetts General Hospital, Department of Medicine, Boston, MA, USA 13 14 *Authors contributed equally 15 16 Correspondence : 17 Lael Yonker: Massachusetts General Hospital, 55 Fruit St, Jackson 14, Boston, MA 02114. 18 617 -724 -2890. [email protected] 19 20 Word c o u nt a b str a ct: 200 21 Wo rd c o u nt m ain t e x t: 2 89 4 22 23 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. Abstract 24 Background : Data on pediatric COVID -19 has lagged behind adults throughout the pandemic . 25 An understanding of SARS -CoV -2 viral dynamics in children would enable data -driven public 26 health guidance. 27 Methods : Respiratory swabs were collected from children with COVID -19 . Viral load was 28 quantified by RT -PCR ; viral culture was assessed by direct observation of cytopathic effects and 29 semiquantitative viral titers . C orrelation s with age , symptom duration , and disease severity were 30 analyzed . SARS -CoV -2 whole genome sequences were compared with contemporaneous 31 sequences . 32 Results : 110 children with COVID -19 (median age 10 years , range 2 weeks -21 years) were 33 included in this study. Age did not impact SARS -CoV -2 viral load . Children were most infectious 34 within the first five days of illness , and severe disease did not correlate with increased viral 35 loads. Pediatric SARS -CoV -2 sequences were representative of those in the community and 36 novel variants were identified . 37 Conclusions : Symptomatic and asymptomatic c hildren can carry high quantities of live, 38 replicating SARS -CoV -2, creating a potential reservoir for transmission and evolution of genetic 39 variants. As guidance around social distancing and masking evolves following vaccine uptake in 40 older populations , a clear understanding of SARS -CoV -2 infection dynamics in children is critical 41 for rational development of public health policies and vaccination strategies to mitigate the 42 impact of COVID -19. 43 44 45 Keywords: SARS -CoV -2, Pediatric COVID -19, Viral dynamics 46 47 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Background 48 Since the SARS -CoV -2 virus ignited the COVID -19 global pandemic, the impact of the virus on 49 children and the role that children pla y in this pandemic has been understudied. I nitially , 50 epidemiology reports suggested that children may have been relatively spared from infection , 51 however, as COVID -19 testing became more available, it has been increasingly recognized that 52 children can be infected with SARS -CoV -2 at rates comparable to adults [1, 2] . To date , over 53 4.1 million children in the Unites States have been reported as testing positive for COVID -19 [3] . 54 Since the winter of 20 20 -20 21, children under 19 years of age have represented one of the age 55 groups with the highest rates of infection [4] , which likely reflects a combination of increased 56 number of infections among children plus increased vaccination rates amongst adults . Most 57 children generally have milder symptoms when infected with SARS -CoV -2 [5] , although a small 58 subset of individuals develop severe diseas e. In the US, over 16,000 children have been 59 hospitalized for acute COVID -19 with over 300 deaths reported [3] . A baseline understanding of 60 the viral characteristics of SARS -CoV -2 infection in children is a necessary prerequisite to 61 understanding the pathogenesis of severe presentations of COVID -19 [6] . 62 At a population perspective, the role that children play in viral transmission remains poorly 63 unders tood. Epidemiologic studies suggest that children exhibit lower transmission rates than 64 adults [7] , however, these findings are potentially confounded by higher rates of asymptomatic 65 or pauci -symptomatic infection in children, increased social isolation by children early in the 66 pandemic, and reduced COVID -19 testing in children. To date, one small study demonstrated 67 that live virus can be cultured from children [8] . However, t he types of systematic studies that 68 ha ve informed our understanding of the viral dynamics of SARS -CoV -2 in adult populations [9- 69 11] have not similarly been carried out in children . As vaccination has become available for 70 adults and adolescents in many places in the world and our understanding of transmission 71 dynamics ha ve evolved , masking and distancing policies are being relaxed [12] . Policy changes 72 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint have necessarily been made despite the paucity of data providing insight into the role that 73 pediatric disease might play in ongoing transmission. As viral variants that enhance the potential 74 for transmission and/ or reduce vaccine efficacy emerge [13 -15] , the importance of identifying 75 potential reservoirs of viral replication and transmission has been brought into the spotlight . 76 Defining the virologic features of SARS -CoV -2 infection in children and the potential for children 77 to transmit virus will facilitate rational public health decision -making for pediatric populations. 78 In this work, we sought to define fundamental virologic features of SARS -CoV -2 in a pediatric 79 population across a range of disease severity . W e analyzed respiratory swabs from children 80 presenting to urgent care clinics or the hospital with symptomatic and asymptomatic COVID -19 81 infection . Clinical factors, such as age, C OVID -19 risk factors, and disease severity were 82 compared with viral features including SARS -CoV -2 viral load, isolation of replication -competent 83 virus, and whole viral sequenc ing . Our data indicate that age , from infancy through adulthood, is 84 not a predict or of viral infection dynamics, and that children of all ages can have high SARS – 85 CoV -2 viral loads of replication -competent virus, including variants, displaying comparable 86 dynamics to those seen in adults. 87 88 Methods 89 Sample collection 90 Infants, children and adolescents <21 years of age presenting to Massachusetts General 91 Hosp it a l u rg en t c a re c lin ic s o r th e h o sp it a l wi th ei ther sym ptom s concer ning for or know n 92 exposur e to CO VID -19 (4/202 0 -4/ 2021) were pr ospect ivel y offer ed enr ollm ent in t he Inst it ut ional 93 Revie w Bo ard -appr oved MGH P edi atric CO VID -19 Bior eposi tor y (IR B # 2020P 000955) [16] . 94 Af te r in fo rm ed c o nse nt, a nd a sse nt w he n a pp ro p ria te , w as o bta in ed v e rb al ly, a re se arc h – 95 desi gnat ed swab of the nasophar yn x, oroph ary n x and/ or ant erior nar es was o b ta in ed and 96 pl aced in phosphat e buf fer ed saline. Sample s we re a liq uote d a nd s to re d a t -80 oC. Sa mple s 97 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint from patients who tested positive for COVID -19 on clinical SARS -CoV -2 RT -PCR testing were 98 analyzed . Nasal samples from adults hospitalized with acute COVID -19 [10] (4/2020 -8/2020; 99 enrolled in Institutional Review Board -approved MGH COVID -19 Biorepository, IRB # 100 2020P000804) with duration of symptoms equal to the hospitali zed pediatric cohort were 101 selected for comparative studies. 102 103 Clinical data collection 104 Demographic and clinical factors were recorded through a combination of manual chart reviews 105 and data extraction from electronic health records (EHR) , then collected in a R EDC ap database 106 [17] through the Partners Electronic Health Record Registry of Pediatric COVID -19 Disease 107 (IRB # 2020P003588) . Trained reviewers collected demographics, SARS -CoV -2 risk factors, 108 comorbid conditions, medications, COVID -19 related symptoms, and laboratory tests. Outcome 109 of initial presentation to care, admission status, and complications of COVID -19 disease were 110 also extracted by manual review. 111 112 SARS -CoV -2 viral load quantification 113 Virions were pelleted from anterior nasal, oropharyngeal, and nasopharyngeal swab fluids by 114 centrifugation at approximately 21,000 x g for 2 hours at 4°C. RNA was extracted using Trizol – 115 LS (Thermofisher) according to the manufacturer’s instructions. RNA wa s then concentrated by 116 isopropanol precipitation , and SARS -CoV -2 RNA was quantified using the CDC N1 primers and 117 probe [ https://www.cdc.gov/coronavirus/2019 -ncov/ lab/rt -pcr -panel -primer -probes.html ] as 118 previ ousl y descr ibed [1 0] . As t h e re wa s n o s ig n if ic a nt d if fe re nce in v ir a l lo a d fr o m r e sp ir a to ry 119 secr etions obt ained fr om t he anterior nar es, nasophar ynx or or ophar ynx of par tici pant s 120 (Suppl em ent al Fi gur e 1 ), s am ple s w ere analy ze d to ge th er, re ga rd le ss o f c oll e cti o n s ite . 121 122 Vi ra l Cul tur e 123 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Vero -E6 cells (ATCC) were maintained in D10+ media [Dulbec co’s modified Eagle’s media 124 (DMEM) ( Corning ) supplemented with HEPES ( Corning ), 1X Penicillin/Streptomycin ( Corning ), 125 1X Glutamine (Glutamax, Thermo Fisher Scientific ) and 10% Fetal Bovine serum (FBS) (Sigma)] 126 in a humidified incubator at 37 oC in 5% CO2. Vero -E6 cells were passaged every 3 -4 days, 127 detached using Trypsin -EDTA (Fisher Scientific) and seeded at 150,000 cells per wells in 24 128 well plates for culture experiments and 20,000 cells per well in 96w plates the day before 129 inoculation for median tissue cultur e infectious dose ( TCID50 ) experiments. 130 131 After thawing, each specimen was filtered through a Spin -X 0.45 µm filter (Corning) at 10,000 x g 132 for 5min. 50 µL of the supernatant was then diluted in 450 µL of D + media [DMEM supplemented 133 with HEPES, 1X Penicillin/Streptomycin and 1X Glutamine]. The viral culture experiments were 134 performed as previously reported [18] with the following modifications: 100 µL of the solution was 135 used to inoculate wells in a 24 well plate and 1mL of D 2+ media [D+ media with 2% FBS] was 136 added to each well after 1 hour of incubation . The plates were then placed in a 5% CO2 137 incubator at 37 oC. For TCID50 measurements conducted in parallel , 25 µL of the Spin -X flow – 138 through was used to inoc ulate Vero -E6 cells in a 96 well plate in the presence of 5 µg/mL of 139 polybrene (Santa Cruz Biotechnology) using 5 -fold dilutions (5 -1 to 5 -6) and 4 repeats for each 140 sample. The plates were centrifuged for 1 hour at 2,000 x g at 37 oC before being placed in a 5% 141 CO2 incubator at 37 oC. The SARS -CoV -2 isolate USA -WA1/2020 strain (BEI Resources) was 142 used as a positive control for CPE in both culture and TCID50 experiments. 143 144 Vira l c u lt u re a nd TCI D50 p la te s we re o bse rv e d a t 3 – and 6 -days post -infe ctio n w it h a lig ht 145 mi cro sc o p e a nd w ell s s h ow in g C PE w ere c o unte d . Th e TCI D50 tit e rs we re c a lc u la te d u sin g t h e 146 Sp ea rm an -Ka rb er m eth o d. Fo r th e c u lt u re p la te s, th e s u pe rn ata nt o f t h e we ll s d is p la yin g CPE 147 wa s h a rv e ste d 1 0 -14 days post -infe ctio n a n d R N A w as is ola te d usi ng a QIA am p Vir al R NA M ini 148 kit ( Q IA G EN ) for conf ir m at ion of t he viral sequence. 149 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint 150 SARS -CoV -2 sequencing 151 cDNA synthesis was performed using Superscript IV reverse transcriptase (Invitrogen). Whole 152 viral amplification was performed with the Artic protocol using multiplexed primer pools designed 153 with Primal Scheme generating 400 -bp tiling amplicons [19, 20] . PCR products were pooled and 154 Illumina library construction was performed using the Nextera XT Library Prep Kit (Illumina). The 155 comparison dataset included 183 representative contemporaneous SARS -CoV -2 genomes from 156 Massachusetts present in GISAID to assess for local clustering. Nucleotide sequence alignment 157 was performed with MAFFT (multiple alignment using fast Fourier transform) [21] . Best -fit 158 nucleotide substitution GTR+G+I was used for the datasets using model selection in IQ -Tree 159 followed by maximum likelihood phylogenetic tree construction usi ng IQ -Tree web server with 160 1000 -bootstrap replicates [22] . 161 162 Analysis 163 All statistical analyses were perfo rmed using p arametric comparisons in GraphPad Prism 164 (Version 9.1.1), including Pearson correlation, ANOVA with multiple comparisons , and unpaired 165 t test. 166 167 Results 168 Clinical cohort 169 One -hundred -ten children diagnosed with COVID -19 with a mean age of 10 years (range 0-21 170 years ) were included in the study (Table 1 ). There were slightly more boys (56%) than girls 171 (44%) with SARS -CoV -2 infection included in our analyses . One third of the participants were 172 White ( 33 %), 10% African American/Black, and 4% were Asian; one third (38%) reported their 173 ethnicity as Hispanic. Past medical history in children is reported in Supplemental Table 1 . 174 Thirty children were asymptomatic but were identified as having COVID -19: twenty -six children 175 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint (27%) presented to urgent care /COVID -19 testing sites because of a COVID -19 exposure , while 176 four children ( 4%) were identified on routine screening during hospital admission. Eight (7%) 177 presented with COVID -19 symptoms but had no known COVID -19 contact. The majority of 178 participants with COVID -19 did not require hospitalization (72 children, 65%) . Thirty -six children 179 (33%) were hospitalized with COVID -19, although only 1 8 children (1 6%) required supplemental 180 oxygen and/or invasive or non -invasive respir atory suppor t (referred to as “moderate/severe 181 COVID -19”) . 182 183 Age did not impact SARS -CoV -2 viral load or recovery of replication competent virus 184 Age is a well -established risk factor for developing severe COVID -19. Accordingly, 185 asymptomatic patients were significantly younger than patients with mild disease, and pediatric 186 patients who were hosp italized with hypoxemia were significantly older than asymptomatic 187 children or children with mild disease (Figure 1A ). However, viral load was not increased in 188 more severe disease: asymptomatic children and children with mild disease displayed 189 significantly higher viral loads than adults hospitalized with COVID -19 with comparable duration 190 of symptoms (less than 10 days) (Figure 1B ). However, there were no differences in viral load 191 between pediatric patients hospitalized with moderate/severe dis ease and hospitalized adults of 192 similar duration of illness (Figure 1B ) (Adult demographics are detailed in Supplemental Table 193 2). 194 195 The a ge of each infected child was analyzed to determine whether age impacted viral load . 196 There was no significant correlation of age with viral load ( Figure 1C ), nor were there significant 197 differences between ages when grouped by school levels : 0-4 years (infant through pre -school), 198 5-10 years (elementary school), 11 -16 years (middle school), 17 and older (high school and 199 higher education) (ANOVA, P = 0.12) (Figure 1 D). Thus , a child’s age did not appear to impact 200 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint viral load: all children , from 2 weeks through 2 1 years of age , were equally capable of carrying a 201 high viral load. 202 203 As SARS -CoV -2 RNA detection by RT -PCR does not specify whether replication -competent 204 virus is being shed, we next sought to ascertain risk factors for shedding live virus by performing 205 viral culture assays for recoverable SARS -CoV -2 in parallel with viral load testing . From the 110 206 participants , we collected 126 samples ; live virus was cultured from 33 samples coming from 32 207 participants . Of note, eight of these children with cultur able SARS -CoV -2 were asymptomatic. 208 Higher viral load was significantly predictive of shedding of live virus (t test, P < 0.0001) ( Figure 209 2A ). Consistent with the results for viral load, age was not correlated with viral culture results; 210 virus was recovered from children ages 1 month t hrough 21 years (Figure 2B ). Semi – 211 quantitative assessment of the amount of virus shed by an individual participant was assessed 212 by media n tissue culture infectious dose (TCID50). TCID50 for culture -positive specimens 213 correlated strongly with viral load (Pearson correlation r = 0. 7, P < 0.0001) ( Figure 2C ) b ut did 214 not correlate with age across all pediatric participants (Figure 2D ). 215 216 Children with COVID -19 were most infectious within first five days of illness 217 To define the likely period of infectiousness in our pediatric population, we analyzed viral load, 218 culturability , and TCID50 in comparison with duration of symptoms. Of note, duration of 219 symptoms does not necessarily indicate duration of infection, as time infection was acquired 220 cannot be confirmed. Consistent with prior reports in adults [9] , viral load s in children w ere the 221 highest earliest in the course of illness and declined over time after symptom onset (Pearson, r 222 = -0.4, P <0.001) ( Figure 3A). Viral load was highest in the first two days of symptoms , with 223 significant decrease after 5 days of symptoms and further decline after 10 days of sympt oms (P 224 < 0.0001) ( Figure 3B). Analysis of pediatric viral culture results demonstrated that children 225 tested early after symptom onset were more likely to shed replication competent virus (P = 226 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint 0.004) ( Figure 3C). Correspondingly , semi -quantitative assessment of infectious vir al shed ding 227 in children showed that the TCID50 was higher early after symptom onset and decreased over 228 time . When grouped by days of symptoms, children in days 0 -2 of their symptoms had the 229 highest infectivity, while children with greater tha n six days of illness shed less virus (P = 0.00 4) 230 (Figure 3D). 231 232 We then sought to assess whether COVID -19 severity impacted the relationship between viral 233 load and age in pediatric cohorts of varying severity : asymptomatic, mildly symptomatic, and 234 moderate/severe pediatric COVID -19 patients. None of these COVID -19 seve rity groups 235 revealed any correlation of age with viral load (Figure 4A ). Further, there were no differences in 236 viral clearance over the duration of illness, not only when comparing mild pediatric COVID -19 237 with moderate/severe COVID -19, but also when compar ing pediatric COVID -19 with adults 238 hospitalized with COVID -19 (Figure 4B ). Duration of symptoms did not affect the finding that 239 viral load does not correlate with age of the infected child ( Supplemental Figure 2 ). 240 241 Pediatric SARS -CoV -2 sequences were representative of those found in the community 242 We successfully performed whole -viral sequencing of 57 respiratory samples from 54 children. 243 Phylogenetic analysis of these pediatric sequences with contemporaneous Massachusetts 244 sequences from GISAID showed that they were represent ative of the spectrum of sequences 245 found in the community ( Figure 5). Notable variants identified in the p ediatric samples included 246 four Alpha ( B.1.1.7 ) and three lota (B.1.526.2) variants. To validate our culture results on a 247 subset of culture -positive samples , we sequenced virus isolated from the supernatant from 8 248 positive samples. Sequences from the supernatant and respiratory specimens were identical in 249 7 cases and demonstrated only 1 nucleotide change in the last case. 250 251 Discussion 252 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint As the global COVID -19 pandemic took hold , infected older a dults suffer ed high rates of 253 hospitalization and death while infected children typically experienced pauci symptomatic or 254 asymptomatic infection. While it is now clear that children can become infected with and 255 transmit SARS -CoV -2, viral dynamics in children have been understudied , and a full 256 understanding of the dynamics of infection in children is needed to inform public health polic ies 257 specific to the pediatric population. Here, we show that pediatric patients of all ages, from 258 infancy to young adulthood, can carry a high SARS -CoV -2 viral load in their upper airways , 259 particularly early in the course of infection, and an elevated viral load correspond s with high 260 levels of viable , replicating virus. Pediatric sequences were largely reflective of those found in 261 the general community and the presence of novel variants was identified. 262 263 Our findings have significant implications for both public health policy and the potential role of 264 universal vaccination of pediatric populations in fully curbing the COVID -19 pandemic . As 265 vaccination has rolled out in adult populations, public health poli cies are being adjusted to 266 account for changes in risk that result from vaccination. Our results emphasize the importance 267 of considering and clarifying how these policy changes relate to children. As adult populations 268 have been vaccinated, pediatric cases have represented a growing proportion of infections, 269 currently accounting for up to 25% of all COVID -19 cases across different regions of the United 270 States [3] . Our results suggest that the low rates of transmission in settings such as schools an d 271 daycares cannot be attributed to low viral loads, low rates of viral shedding, or rapid clearance 272 of virus in younger patient populations. As changes in masking and distancing policies are 273 implemented for vaccinated adults, consideration of how and whether policies changes will be 274 applied for children will be critical for ongoing reduction of new COVID -19 cases. 275 276 Our results additionally suggest that pediatric populations have the potential to serve as a 277 community res ervoir of actively replicating virus , with implications for both new waves of 278 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint infection and the evolution of viral variants. The duration of natural and vaccine -induced 279 immunity for each vaccine in clinical use are not yet known. If a community reservoir of actively 280 replicating virus is maintained and transmitted within unvaccinated pediatric populations, that 281 population could then serve as a source of new infections as vaccine -induced immunity wanes 282 in vaccinated adult populations . In addition, viral genomic variants were readily identified in the 283 pediatric samples and t hese variants have the potential to impact viral transmission [23 -25] , 284 disease severity [26, 27] , and vaccine efficacy [28] . Ongoing viral replication within pediatric 285 populations has the potential to serve as a source of existing and new viral variants that 286 interfere with eradication efforts. 287 288 Our study has several limitations. First, the data collected here represent a single medical 289 center and affiliated pediatric urgent care /COVID -19 screening clinic s. However, these were 290 amongst the few pediatric testing centers encompassing a large catchmen t area during the 291 duration of this study, and patients enrolled spanned a wide range of symptoms . Additionally, 292 many of these samples were collected early in the pandemic and SARS -CoV -2 variants of 293 interest have shifted over time . Ongoing studies analyzing shifts in virologic features of SARS – 294 CoV -2 infection in children alongside studies of infection in adults are needed to better 295 understand the full reach of the COVID -19 pandemic. 296 297 Ultimately , our data suggest that although age is generally protective a gainst severe disease, 298 children, especially early in the infection course, carry high viral loads of SARS -CoV -2, which 299 can include viral variants. Our results underline the importance of defining public health policy 300 with viral dynamics in children in mind and of including pediatric populations in vaccine efforts 301 aimed at eradication. 302 303 304 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Acknowledgements : 305 The viral culture work was performed in the Ragon Institute BSL3 core, which is supported in 306 part by the NIH -funded Harvard University Center for AIDS Research (P30 AI060354). This 307 research was supported by the National Heart, Lung, and Blood Institute (5K08HL143183 to 308 LMY), the Department of Pediatrics at Massachusetts General Hospital for Children (to LMY) , 309 the Massachusetts Consortium for Pathogen Readiness and a gift from Mark and Lisa Schwartz 310 (to JZL) . 311 312 Conflicts of Interest 313 The authors do not report any conflicts of interest. 314 315 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint 316 1. Mehta NS, Mytton OT, Mullins EWS, et al. SARS -CoV -2 (COVID -19): What Do We Know About 317 Children? A Systematic Review. Clin Infect Dis 2020 ; 71:2469 -79. 318 2. Heald -Sargent T, Muller WJ, Zheng X, Rippe J, Patel AB, Kociolek LK. Age -Related Differences 319 in Nas opharyngeal Severe Acute Respiratory Syndrome Coronavirus 2 (SARS -CoV -2) Levels in 320 Patients With Mild to Moderate Coronavirus Disease 2019 (COVID -19). JAMA Pediatr 2020 ; 321 174:902 -3. 322 3. https://services.aap.org/en/pages/2019 -novel -coronavirus -covid -19 -infections/children -and – 323 covid -19 -state -leve l -data -report/ . 324 4. https://www.mass.gov/info -details/covid -19 -res po nse -reporting . 325 5. Yonker LM, Neilan AM, Bartsch Y, et al. Pediatric Severe Acute Respiratory Syndrome 326 Coronavirus 2 (SARS -CoV -2): Clinical Presentation, Infectivity, and Immune Responses. J Pediatr 327 2020 ; 227:45 -52 e5. 328 6. Fialkowski A, Gernez Y, Arya P, Weinacht K G, Kinane TB, Yonker LM. Insight into the pediatric 329 and adult dichotomy of COVID -19: Age -related differences in the immune response to SARS – 330 CoV -2 infection. Pediatr Pulmonol 2020 . 331 7. Soriano -Arandes A, Gatell A, Serrano P, et al. Household SARS -CoV -2 transmission and 332 children: a network prospective study. Clin Infect Dis 2021 . 333 8. L’Huillier AG, Torriani G, Pigny F, Kaiser L, Eckerle I. Culture -Competent SARS -CoV -2 in 334 N asopharynx of Sym ptomatic N eonates, Children, and Adolescents. Emerg Infect Dis 2020 ; 335 26:2494 -7. 336 9. Weiss A, Jellingso M, Sommer MOA. Spatial and temporal dynamics of SARS -CoV -2 in COVID – 337 19 patients: A systematic review and meta -analysis. EBioMedicine 2020 ; 58:102916. 338 10. Fajnzylber J, Regan J, Coxen K, et al. SARS -CoV -2 viral load is associated with increased 339 disease severity and mortality. N at Commun 2020 ; 11:5493. 340 11. Wolfel R, Corman VM, Guggemos W, et al. Virological assessment of hospitalized patients 341 with COVID -2019. Nature 2020 ; 581:465 -9. 342 12. https://www.cdc.gov/coronavirus/2019 -ncov/index.html . 343 13. Madhi SA, Baillie V, Cutland CL, et al. Efficacy of the ChAdOx1 nCoV -19 Covid -19 Vacc ine 344 against the B.1.351 Variant. N Engl J Med 2021 . 345 14. Liu Y, Liu J, Xia H, et al. BN T162b2 -Elicited N eutralization against N ew SARS -CoV -2 Spike 346 Variants. N Engl J Med 2021 . 347 15. Shinde V, Bhikha S, Hoosain Z, et al. Efficacy of N VX -CoV2373 Covid -19 Vaccin e against the 348 B.1.351 Variant. N Engl J Med 2021 . 349 16. Lima R, Gootkind EF, De la Flor D, et al. Establishment of a pediatric COVID -19 biorepository: 350 unique considerations and opportunities for studying the impact of the COVID -19 pandemic on 351 children. BMC M ed Res Methodol 2020 ; 20:228. 352 17. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N , Conde JG. Research electronic data 353 capture (REDCap) –a metadata -driven methodology and workflow process for providing 354 translational research informatics support. J Biome d Inform 2009 ; 42:377 -81. 355 18. Choi B, Choudhary MC, Regan J, et al. Persistence and Evolution of SARS -CoV -2 in an 356 Immunocompromised Host. N Engl J Med 2020 ; 383:2291 -3. 357 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint 19. Gonzalez -Reiche AS, Hernandez MM, Sullivan MJ, et al. Introductions and early sprea d of 358 SARS -CoV -2 in the New York City area. Science 2020 ; 369:297 -301. 359 20. Quick J. nCoV -2019 sequencing protocol v1. Available at: 360 https://www.protocols.io/view/ncov -2019 -sequencing -protocol -bbmuik6w . 361 21. Katoh K , Rozewicki J, Yamada KD. MAFFT online service: multiple sequence alignment, 362 interactive sequence choice and visualization. Brief Bioinform 2019 ; 20:1160 -6. 363 22. Trifinopoulos J, Nguyen LT, von Haeseler A, Minh BQ. W -IQ -TREE: a fast online phylogenetic 364 tool for maximum likelihood analysis. N ucleic Acids Res 2016 ; 44:W232 -5. 365 23. Zhou B, Thao TTN, Hoffmann D, et al. SARS -CoV -2 spike D614G change enhances replication 366 and transmission. N ature 2021 ; 592:122 -7. 367 24. Deng X, Garcia -Knight MA, Khalid MM, et al. Transmission, infectivity, and neutralization of 368 a spike L452R SARS -CoV -2 variant. Cell 2021 . 369 25. Davies N G, Abbott S, Barnard RC, et al. Estimated transmissibility and impact of SARS -CoV -2 370 lineage B.1.1.7 in England. Sc ienc e 2021 ; 372. 371 26. Davies N G, Jarvis CI, Group CC -W, et al. Increased mortality in community -tested cases of 372 SARS -CoV -2 lineage B.1.1.7. N ature 2021 ; 593:270 -4. 373 27. Funk T, Pharris A, Spiteri G, et al. Characteristics of SARS -CoV -2 variants of concern B.1.1.7, 374 B.1.351 or P.1: data from seven EU/EEA countries, weeks 38/2020 to 10/2021. Euro Surveill 375 2021 ; 26. 376 28. Garcia -Beltran WF, Lam EC, St Denis K , et al. Multip le SARS -CoV -2 variants escape 377 neutralization by vaccine -induced humoral immunity. Cell 2021 ; 184:2372 -83 e9. 378 379 380 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Table 1 : Participant Demographics, past medical history , reason for presenting for SARS -CoV -2 381 RT -PCR testing, and disease severity of children with COVID -19 (n= 110 ). 382 383 384 385 COVID + ch ild re n (N =1 1 0) Ag e, a vera g e ( m ax, m in ) 10 (0, 21) Se x, n ( % ) Ma le 62 (56) Fe male 48 (44) Ra ce, n ( % ) Wh it e 36 (33) Bl ack 11 (10) As ia n 4 (4) Ot her 45 (41) Et hnic it y , n ( % ) Hi sp an ic 42 (38) Pa st M ed ic a l H is to ry , n ( % ) Ob esit y 28 (25) As th m a 11 (10) Other 39 (35) Re aso n f o r p re se n tin g f o r C O VID -1 9 t e stin g, n ( % ) As ym pto m atic , e xp osu re 30 (27) Sy mpto m atic , e xp osu re 72 (65) Sy mpto m atic , n o e xp osu re 8 (7) CO VID -1 9 S everit y , n ( % ) Ho sp it a liz e d 36 (33) Su pple m en ta l o xygen 18 (16) Ou tp atie n t 75 (68) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Figure 1 : COVID -19 disease severity and SARS -CoV -2 viral load across age groups. A. Age of 386 pediatric patients with SARS -CoV -2 infection, stratified by disease severity: asymptomatic 387 (n=2 7), mild disease, outpatient (n=4 8), moderate/severe COVID -19 , hospitalized (n=1 8). 388 Analyzed by ordinary one -way ANOVA. B. SARS -CoV -2 viral load was quantified across a 389 range of disease severit ies . Patients presenting with <10 days of symptoms were compared, 390 including asymptomatic pediatric outpatients (n=2 7), mildly symptomatic pediatric outpatients 391 (n= 44 ), moderate/severe pediatric hospitalized patients with oxygen requirement (n=1 8), and 392 moderate/severe adult hospitalized patients (n=2 9). Analyzed by ordinary one -way ANOVA. C. 393 Vira l lo a d f o r e ach s p e cim en ( n = 110 ) w as d ete rm in ed b y q PC R, p lo tte d a gain st p arti cip an t a ge 394 and analyzed by Pear son correl at ion. D. Vira l l o ad le ve ls re porte d b y s ch oo l a ge g ro up : 0 -4 395 year s ol d (yo) – in fa nt t h ro ugh pr e -school ( n= 32 ), 5 -10yo – el em ent ary school ( n=23) , 11 -16yo – 396 mi dd le s c h oo l ( n = 3 3), 1 7yo a nd o ve r – hi gh school and higher educat ion (n= 22 ). A na ly ze d b y 397 or dinar y one -wa y ANO VA. Dott e d li n es d epic t lim it o f d e te ctio n . * P <0.0 5, *** P <0.0 01, **** P < 398 0. 0001 , ns = not si gni ficant 399 400 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Figure 2 : SARS -CoV -2 culture results across age groups and viral load. A-B. Samples with 401 observable CPE (culture +) (n= 31 ) or without observable CPE (culture -) (n= 95 ) plotted against 402 viral load ( A) and participant age ( B) and compared using t test. C-D. Semiquantitative viral titer 403 expressed as TCID50/mL for culture positive samples plotted against corresponding viral load 404 (C) or participant age ( D), Analyzed using Pearson correlation. Dotted line depict s limit of 405 detecti on. **** P < 0.0001 406 407 408 409 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Figure 3 : Culture positivity and duration of symptoms. A. Viral load for each specimen was 410 determined by qPCR and plotted against the duration of symptoms (in days). Analysis by 411 Pearson correlation . B. Viral load reported by binned duration of symptoms . Ordinary one -way 412 ANOVA used for analysis. C. Duration o f symptoms for samples with observable CPE (culture 413 +, n=29 ) and without observable CPE (culture -, n=85 ). Analysis by t test. D. Semiquantitative 414 viral titer reported by binned duration of symptoms , analyzed by ordinary one -way ANOVA . 415 Dotted lines depict limit of detection. ** P < 0.01, * *** P < 0. 0001 416 417 418 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Figure 4 : Correlation of viral load with age and duration of illness based on disease severity. 419 A. Correlation of viral load and age, stratified by asymptomatic (n=30) , mild outpatient (n=48) 420 and moderate/severe hospitalized (n=18) cohorts. B. Viral load of hospitalized adult (n=29), 421 hospitalized pediatric participants requiring respiratory support (n=18) and pediatric outpatients 422 with mild disease (n=48), plotted again st duration of symptoms . Dotted lines depict limit of 423 detection. 424 425 426 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Figure 5. Phylogenetic analysis of pediatric and community SARS -CoV -2 sequences. Maximum 427 likelihood tree generated from pediatric sequen ces (red) and 183 contemporaneous 428 Massachusetts sequences from GISAID. 429 430 431 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Supplemental Table 1 : Past medical history of children infected with SARS -CoV -2. 432 433 Past Medical History – Pediatric Patients Acute rec urr en t p an crea tit is AD HD Alopec ia As th m a Au tis m Cardi ac conducti on disorde r Chroni c ki dne y di se ase Coats Plus Syndrom e Crohn’s di se ase Cysti c Fibrosi s Eczema Epi le psy G6PD Defic ien cy Gl uta ric a cid em ia , t yp e 1 Live r tr anspl ant Mi to ch o ndria l c o mp le x 1 d efic ie n cy Ob esit y Ob str u ctive sleep a p nea Pr em atu rit y Si ck le c e ll tra it Spe ech de la y Subg lotti c ste nosi s Trache obronchom alaci a 434 435 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Supplemental Table 2 : Demographics, past medical history and disease severity of adults with 436 COVID -19 (n= 29 ) included in analysis of SARS -CoV -2 viral load. 437 438 439 440 COVID+ adults (n=2 9 ) Age, average (m ax , m i n ) 62 (26,93) Sex, number (%) Mal e 13 (45) Female 16 (55) Race, number (%) White 20 (69) Black 5 (17) Asi an 0 (0) Ot her 4 (14) Et h n i c i t y, n u m b er (%) Hispanic 4 (14) COVID-19 Severity Hospitalized, number (%) 21 (100) Respiratory support 22 (78) Out pat i ent , number (%) 0 (0) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Supplemental Figure 1: Viral load by sample collection location. Viral load of samples collected 441 from nasopharynx (n= 60 ), anterior nares (n=47) , or oropharynx (n=1 9) were compared and 442 analyzed by ordinar y one -wa y ANO VA. n s = non -si gni ficant . 443 444 445 446 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint Supplemental Figure 2 : Viral load plotted against pediatric patient age, for patients with 0 -2 447 days symptoms (n= 67 ), 3 -5 days symptoms (n= 30 ), and >6 days of symptoms (n=2 9). No 448 signi ficant cor rel at ion in any groupi ng, w hen analyzed by Pear son correl at ion. Do tt e d li n es 449 depi ct lim it of det ection. ns = not si gni ficant . 450 451 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 17, 2021. ; https://doi.org/10.1101/2021.05.30.21258086 doi: medRxiv preprint
1) In “Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1”, authors show that … (pick all correct answers) Select one or more: a) On plastic, no viable SARS-CoV-2 was measured a
Journal of Exposure Science & Environmental Epidemiology (2021) 31:953–960 https://doi.org/10.1038/s41370-021-00337-1 ARTICLE Assessing the effect of beard hair lengths on face masks used as personal protective equipment during the COVID-19 pandemic Steven E. Prince 1●Hao Chen 2●Haiyan Tong 3●Jon Berntsen 4●Syed Masood 5●Kirby L. Zeman 6● Phillip W. Clapp 6,7 ●William D. Bennett 6,8 ●James M. Samet 3 Received: 25 January 2021 / Revised: 14 April 2021 / Accepted: 22 April 2021 / Published online: 18 May 2021 This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply 2021 Abstract BackgroundGlobally, a large percentage of men keep a beard at least occasionally. Workplace regulations prohibit beards with N95 respirators, but there is little information on the effect of beards with face masks worn by the public for protection against SARS-CoV-2. Methods andfindingsWe examined thefittedfiltration efficiency (FFE) offive commonly worn protective face masks as a function of beard length following the US Occupational Safety and Health Administration Quantitative Fit Test: N95 (respirator), KF94 and KN95, surgical/procedure, and cloth masks. A comparison using N95 respirators was carried out in shaven and bearded men. A detailed examination was conducted for beard lengths between 0 and 10 mm (0.5 mm incre- ments). The effect of an exercise band covering the beard on FFE was also tested. Although N95 respirators showed considerable variability among bearded men, they had the highest FFE for beard lengths up to 10 mm. KF94 and KN95 masks lost up to 40% of their FFE. Procedure and cotton masks had poor performance even on bare skin (10–30% FFE) that did not change appreciably with beard length. Marked performance improvements were observed with an exercise band worn over the beard. ConclusionsThough variable, N95 respirators offer the best respiratory protection for bearded men. While KF94 and KN95 FFE is compromised considerably by increasing beard length, they proved better options than procedure and cotton face masks. A simple exercise band improves FFE for face masks commonly used by bearded men during the COVID-19 pandemic. Keywordsmask ●respiratory protection ●SARS-CoV-2 ●beard ●intervention ●particles Introduction The broad variety of face masks of varying efficiencies currently in use by the public represent a primary preventive measure to control the transmission of the SARS-CoV-2 These authors contributed equally: Steven E. Prince, Hao Chen *Steven E. Prince [email protected] 1 Public Health and Environmental Systems Division, Center for Public Health and Environmental Assessment, Office of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC, USA 2 Oak Ridge Institute for Science Education, Oak Ridge, TN, USA 3 Public Health and Integrated Toxicology Division, Center for Public Health and Environmental Assessment, Office of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC, USA 4 TRC, Raleigh, NC, USA 5 Curriculum in Toxicology and Environmental Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA 6 Center for Environmental Medicine, Asthma and Lung Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA 7 Department of Pediatrics, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA 8 Department of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA Supplementary informationThe online version contains supplementary material available athttps://doi.org/10.1038/s41370- 021-00337-1. 1234567890();,: 1234567890();,: virus as an intervention [1]. While the availability of high- efficiency N95 respirators has been largely limited to health care professionals due to supply shortages, Chinese (KN95) and Korean (KF94) designed face masks, surgical-style disposable masks, and reusable cloth masks of widely varying material, design, and construction are currently in use by the general public. A number of intrinsic factors influence thefiltration performance of a face mask. Properties such as the porosity and surface charge of the face mask material can be seen as representing the theoretical limit of the efficiency of a face mask. On the other hand, design features influence the extent to which this limit can be approximated by deter- mining the integrity of the seal that can be achieved and maintained between the face mask material and the facial skin. Extrinsic factors linked to the wearer that are known to affect face mask performance include variations in facial morphology [2] and the presence of facial hair [3]. According to a 2017 survey, approximately a third of adult males in the US report keeping a beard, with another 27% responding that they do so sporadically [4] and in a 2016 British survey, 37% of all males reported beards or stubble, with 61% of 18- to 39-year olds reporting some form of facial hair [5]. Relative to bare skin, facial hair potentially alters thefit and performance of a face mask by changing the texture of the skin as well as topology at the points of contact. With the exception of a few styles, reg- ulations established by the US Occupational Safety and Health Administration (OSHA) disallow facial hair for the workplace use of respirators such as N95filtering face pieces [3]. However, information on the effect of facial hair on the efficacy of other types of face masks is lacking. Guidance and regulations from government agencies, including the UK Health and Safety Executive [6], US Centers for Disease Control and Prevention/National Institute for Occupational Safety and Health [7], and the US OSHA [3] describe facial hair, stubble, and beards in par- ticular, as a contraindication to the workplace use of tight- fitting respirator masks. OSHA respiratory standards,fina- lized in 1998, describe beard growth as one condition that “interferes with the face-to-facepiece seal.”Since the 1980s, several studies that examined the impacts of beard hair on standardized respirator masks provided the basis for agency guidelines. Although one study examining workplace pro- tection factors found no clear relationship, multiple studies provided evidence for beards causing an impaired seal, reducedfit factors and detrimental impacts on respirator performance. This led agencies to advise against bearded workers using these respirator masks (in a workplace set- ting) and in the US to prohibit their use andfit testing for bearded workers. Suggested remedies for this include the use of loose-fitting hoods or helmets that would accom- modate facial hair. Current guidance on the issue of beardhair is highly specific to the use of tight-fitting respirators in the workplace and there is a lack of information on cost- effective face masks that are available to the general public during the COVID-19 pandemic or practical solutions for usage scenarios in the general public. Given the importance of face masks as a primary pre- ventive measure in controlling the spread of the SARS- CoV-2 virus, understanding the factors that both harm and help theirfiltering performance is critical to public health. In the current study, we examine in detail the relationship between varying lengths of beard hair and thefittedfiltra- tion efficiency (FFE) of a variety of face masks used to prevent SARS-CoV-2 infection during the COVID-19 pandemic, and identify a potential solution to overcome decrements in FFE caused by the presence of beard hair. Methods Testing procedure All face masks werefitted on adult male staff members (n=10) who volunteered to be tested using the US Occu- pational Safety and Health Administration Quantitative Fit Testing Protocol (Modified Ambient Aerosol CNC Quan- titative Fit Testing Protocol For Filtering Facepiece Table A–2—RESPIRATORS). Conditions related to the spread of SARS-CoV-2 necessitated modified recruitment and testing of volunteers. The aerosolfiltration efficiency tests were conducted in a custom-built 10 × 10 foot stainless-steel exposure chamber (US EPA Human Studies Facility, Cha- pel Hill, NC) as recently described [8]. Briefly, a TSI 8026 Particle Generator (TSI, Inc., Shoreview, MN) was used to supplement the chamber with NaCl particles with a median diameter of 0.05μm, as measured by a scanning mobility particle sizer (MODEL 3938 TSI, Inc). Prior to testing, the test atmosphere was allowed to stabilize for 30 min in an atmosphere in which the temperature and humidity during testing ranged from 21.5 to 25 °C and 32 to 55%, respec- tively. A steel sampling port was installed in each mask using a TSI model 8025-N95 Fit Test Probe Kit to allow sampling behind the mask. A pair of TSI 3775 condensation particle counters (CPCs) was run in single particle analysis mode to continuously monitor ambient particles in the size range 0.02–3μm in the chamber air just outside the face mask and particles in the breathing space behind the face mask at a sampling rate of 1 s. Ten feet of 0.25-inch con- ductive rubber tubing was used for each sampling line and was connected to the face mask through metal Luer con- nectorfittings. The ambient sampling line and masks sam- pling lines were made identical to reduce variability in the system. Chamber particle counts/cc (cubic centimeter) reported by the CPCs were typically in the range of 954S. E. Prince et al. 6000–9000, meaning that 95% reduction results in counts of 300–450 or less inside the mask, with enough instrument dynamic range to accurately detectfiltration performance in excess of 99%. The overall meanfilter efficiency was averaged from start to end of the testing period, and the average standard deviation over the period of sampling was computed. The FFE of the face mask was measured while the subject performed a series of repeated movements of the torso, head, and facial muscles (i.e., bending at the waist, reading aloud, looking left and right, and looking up and down) designed to simulate workplace activities, as pre- scribed by the OSHA Fit Test. Although related metrics such as particle penetration orfit factors are also found in the literature, FFE is reported here as a commonly under- stood value of mask or respirator performance. The FFE (%) is calculated as [1–(mask count/ambient count)] × 100 (the concentration of particles behind the mask divided by the particle concentration in the chamber atmosphere). Data were collected over the duration of each test described in the OSHA protocol (approximately 140 s plus 10 s between exercises). Face masks tested Five face masks representative of respiratory protection options available to the public during the COVID-19 pan- demic were selected for testing: an N95 respirator (Model #8210 TC 84A-0007, 3M St. Paul, MN), a KN95 earloop mask (Lei Shi De, EN149-2001+A1:2009, CIRS Garments, Shandong, PRC), a KF94 mask worn with supplied clip affixed to ear loops in the back of the head (Dr Puri, KM Corporation, Gyeonggi-do, Republic of Korea), an earloop procedure mask (Medline Industries, Northfield, IL), and a reusable 3-ply 100% cotton fabric face mask with ear loops and adjustable nosepiece (Hanesbrands, Winston- Salem, NC). Testing procedures All bearded volunteers (n=5) had full facial hair that covered the jaw and cheeks, with starting lengths varying between 9 and 30 mm. Beards tested all featured hair that extended into the respirator sealing surface area and, therefore, are contraindicated for respirator use by the National Institute of Occupational Safety and Health gui- dance. Two volunteers additionally trimmed their beards to varying lengths. Volunteer 1 (head circumference 55.9 cm), who repeated testing 2–3 times for each length (in both ascending and descending growth across multiple testing days), used a Phillips Norelco QT4018 (Norelco, Andover, MA) beard trimmer equipped with length adjustments between 0.5 and 10 mm, with 0.5 mm increments.Volunteer 2, with a head circumference of 57.4 cm, used a multifunction trimmer (Model SH1970, GOLEEN, Guang- dong, PRC)fitted with 9-, 6-, and 3-mm combs in a des- cending order in a single day (data from three face masks shown in Supplementary Fig. 1). Volunteers with no facial hair (n=5) used disposable razors and shaving cream to shave within 6 h of testing. Volunteer 1 also tested an intervention strategy using two different yoga style non- latex exercise resistance bands (“light”green color 60” length × 4”width (Theraband, Akron, OH), and“medium” blue color 60”length × 6”width (A Azurelife, Hong Kong)) worn over the beard. Prior to testing, the band was wrapped around the beard and knotted at the top and middle of the head, leaving access to properly affix all masks. Figure1 shows Volunteer 1 shaven, with a full beard and with the exercise bands tested; see Supplementary Fig. 2 for com- plete photos of all beard lengths. Informed consent was obtained to publish these images in an online open access publication. Results A preliminaryfit test survey of N95 respirator performance showed overall high FFE when worn by bearded men who were trained in proper donning and were supervised during the process. As shown in Table1, the presence of a beard with an average hair length between 9 and 30 mm appeared to have only minor impact overall, when compared to FFE Fig. 1Shaven, bearded and beard-covered conditions tested in this study.Volunteer 1 shown within 1 hour of a shave (top left panel), with the maximal beard length tested (top right panel), and wearing non-latex exercise (yoga) bands of light (bottom left) and medium (bottom right) resistance as a beard hair cover. Assessing the effect of beard hair lengths on face masks used as personal protective equipment during. . . 955 in shaven men. Volunteer 1, the bearded staff member who showed the lowest FFE with an N95 respirator, underwent additionalfit testing at varying beard lengths, and with additional face mask types that are available to the public during the COVID-19 pandemic. Figure2shows FFE (%), sampled at 1-s intervals, during different phases offit testing (i.e., bending at the waist, reading aloud, turning of head, raising, and lowering of head) with an N95 respirator, KF94, KN95, a procedure style mask, and a 3-ply 100% cotton mask, as a function of beard lengths between 0 (shaven), 1.5, 3, 6, 9, and >10 mm. Results show averaged values across three unique tests. A repeating pattern, conserved across nearly all face masks tested, was observed in which FFE was highest during reading aloud (i.e., while exerting a positive pressure inside the mask). Interestingly, exceptions to this pattern were for the N95 respirator between 0 and 2.5 mm of beard length (Fig.2A–E and Supplementary Tables 1–5). The rapid inhalation and facial movements associated with speech likely have an effect of pulling a looserfitting mask or respirator toward the face, versus challenging the seal when a more rigid respiratorfits tightly. To generate a more detailed view of beard length as determinant of face mask performance, we retested the face masks’FFE for Volunteer 1 at 0.5 mm increments over a range of 0–10 mm (Fig.3). Three additional measurements at an average hair length greater than 10 mm (approximately 16 mm) were also made (Supplementary Tables 1–5). The overall FFE of the N95 respirator was the most resistant to beard length, remaining at or above material specifications (i.e., 95% efficiency) at lengths up to 2.5 mm, with mod- erate reductions in performance observed with increasinglength out to 10 mm (≥80% FFE). The KF94 and KN95 masks showed a lower starting FFE compared to the N95 respirator, with the KF94 performing slightly better than the KN95 on average at all beard lengths tested (Supplementary Tables 1–5). However, both KF94 and KN95 masks showed proportionately similar decrements in overall FFE from their respective baselines as a function of beard length when compared to the N95 respirator. The plotted data support this observation, showing a rank order of overall FFE performance of N95 > KF94 > KN95 with increasing beard hair length up to 10 mm (Fig.3) and beyond to approximately 16 mm (Supplementary Tables 1–5). The procedure mask and the cotton cloth mask overall FFEs were substantially lower than the N95 respirator, KF94, and KN95 at baseline and were notably unaffected by increasing beard length throughout the range tested (Figs.2and3). For Volunteer 1, the slopes of linear regression analyses indi- cate the steepest negative impact on FFE for KN95 ( 3.0), similar negative impacts for the KF94 and N95 (approxi- mately 2), and values closer to zero for procedure and cloth masks. Data from Volunteer 2 showed similar results for KF94 ( 2.4) and procedure masks (0.1), but a slope closer to zero for the N95 respirator ( 0.1, see Supple- mentary Fig. 1), suggesting a generalizable pattern for two of the three tested masks. A previous report [9] showed that a significant improvement in FFP3 (European standard similar to N95) respiratorfi ltration efficiency could be achieved in bearded individuals by covering the beard hair with an elastomeric band (e.g., a resistance exercise band). As shown in Table2, when tested with >8.5-mm beard length, the band improved the FFE of the N95 respirator significantly, raising it above its rated performance level (i.e.,≥95%). Use of the band also produced marked improvements in the FFE of the KF94 and KN95 masks (approximately 30 and 20% improvements, respectively). The FFE of the procedure mask increased by a similar amount with the use of the band, while that of the cloth mask was raised only mar- ginally (Table2). Lower variance (SD) of obtained mea- surements, found for all except the cloth mask, further demonstrated the benefit of this intervention. Discussion Recent surveys found high percentages of British and American men (42% and approximately 60%, respec- tively) reported having or keeping a beard at least some of the time with even higher prevalence of facial hair among 18- to 39-year-old British men (61%) [4,5]. Given these rates of beard hair in the population, thefindings of this study suggest a significant limitation on the efficacy of face masks as personal protective equipment against Table 1N95 respiratorfittedfiltration efficiency (FFE) against submicron particles for shaved and unshaven men. Subject Average beard length (mm) % FFE (SD) a N95 1 0 99.4 (0.5) 2 0 99.3 (0.4) 3 0 99.1 (0.3) 4 0 99.0 (0.5) 5 0 97.6 (1.8) Average 98.9 6 9 96.9 (1.1) 7 9.8 97.1 (1.0) 8 11.2 98.4 (0.4) 9 15.7 85.0 (3.0) 10 30.4 98.8 (0.4) Average 95.2 aThe efficiency is calculated as [1–(mask count/ambient count)] × 100 across the length of the test. The FFE ± SD are shown. 956S. E. Prince et al. SARS-Cov-2 infection. Previous studies examining the effect of beard hair on thefit and performance of respirators have reported variablefindings, with some showing significant impairment [10]. For instance, a recent survey of a cohort of Australian health care workers reported that no individuals with full beards could achieve a satisfactory respiratorfit[11]. While other studies haveshown satisfactory performance in bearded workers [12], the US OSHA prohibitions on respirator use andfitting include beard hair as a“condition that interferes with face-to-facepiece seal”[3]. The US Center for Disease Control and OSHA have published a chart showing recognized facial hair styles, most of which are contra- indicated for respirator use [7]. Fig. 2Fittedfiltration efficiency (FFE) percentage for face masks at different beard lengths measured using the Occupational Safety and Health Administration modified ambient aerosol CNC quantitativefit testing protocol.Data from Volunteer 1 show the overall FFE decreased for a NIOSH N95 respirator (A), a Korean standard KF94 mask (B), and a Chinese standard KN95 mask (C), with increasing beard hair length (0, 1.5, 3.0, 6.0, 9.0, and >10.0 mm). The FFE percentages for a reusable cloth mask (D) and a procedure mask with elastic ear loops (E)were low even with shaven skin and did not deteriorate appreciably with beard hair length. The numbers adjacent to the data “1, 2, 3, and 4”indicate the starting time of the four exercises in thefit test, (i.e., 1=bending at the waist for 50 s, 2=reading aloud for 30 s, 3=looking left and right for 30 s, and 4= looking up and down for 30 s). Data corresponding to FFE of the face mask are expressed as [1– (mask count/ambient count)] × 100 shown as percent (0–100) on they-axis. The average of three independent tests is plotted against time on thex-axis (seconds), with 10 additional seconds recorded after each exercise. Assessing the effect of beard hair lengths on face masks used as personal protective equipment during. . . 957 Information on the use of other types of face masks by individuals with beards is less clear than that available for respirators. A study by McLure et al. reported increasedshedding of bacteria in bearded men wearing surgical masks relative to shaved individuals [13]. However, disposable surgical and procedure style masks are not subject tofit testing under OSHA regulations. Tests examining surgical- style masks have found 8–12 times higher total particle penetration compared to N95 respirators and attributed particle penetration to be driven in large part by facepiece seal leakage as opposed tofilter media [14]. Recent demonstrations of improved procedure mask FFE following modification interventions support an important role of maskfit in enhancing their performance [15]. No guidelines are currently available in the US for Korean (KF94) and Chinese (KN95) face masks that are sometimes marketed to the public as high-efficiency alternatives to N95 respirators. Similarly, cloth face masks, a reusable, low cost, and environmentally friendly choice of respiratory protection, have no specific direction provided regarding their com- patibility with beards. The data shown in this study demonstrate that FFE decreased relatively linearly overall with beard length for N95 and KF94 masks, with a more variable decline observed for a KN95 mask. The N95 respirator showed the greatest tolerance of hair intrusion, with FFE remaining at or above 95% with beard length of 2.5 mm, equivalent to approximately 7 days of growth for Volunteer 1 and con- sistent with“stubble”findings in a large study by the British Government [6]. Compared to the N95 respirator, the per- formance of KF94 and KN95 masksfirst showed FFE Fig. 3The overallfittedfiltration efficiency (FFE) for face masks at different beard lengths using the Occupational Safety and Health Administration modified ambient aerosol CNC quantita- tivefit testing protocol.The overall FFE as a function of beard hair length (data from Volunteer 1) in the range of 0–10.0 mm in 0.5 mm increments decreased for a N95 respirator (slope= 2.2), a Korean standard KF94 mask (slope= 2.0), and a Chinese standard KN95 mask (slope= 3.0). The comparatively low FFE of a proceduremask with elastic ear loops (slope= 0.3) and a reusable cloth mask (slope=0.1) did not decrease further with beard hair length. The FFE is calculated as [1–(mask count/ambient count)] × 100 and presented as the average of all FFE data across the test period at each beard length. Linear regression was used to calculate the slope (dotted line) of FFE against beard lengths. The average of at least two independent tests is shown. Table 2The effect of an elastic band worn over the beard on thefitted filtration efficiency of various face masks. Mask (brand) Yoga band % FFE (SD) a N95 respirator (3M, 8210) No 85.3 (3.1) Yes 96.1 (0.6) % improvement b +12.7 KF94 micro-dust protection mask (Dr Puri)No 61.9 (6.7) Yes 80.2 (3.8) % improvement b +29.6 KN95 folding particle protection disposable mask (Lei Shi De)No 54.9 (7.9) Yes 65.7 (5.7) % improvement b +19.8 Procedure mask, earloop (Medline Industries)No 30.6 (10.4) Yes 36.6 (7.7) % improvement b +19.7 Cloth/cotton mask (Hanesbrands) No 39.4 (7.3) Yes 40.6 (8.0) % improvement b +2.9 aThe efficiency (FFE) is calculated as [1–(mask count/ambient count)] × 100 across the length of the test. The FFE ± SD are shown. The average of at least two unique tests is presented. b% improvement was calculated as the percentage increase of FFE with and without the elastic band. 958S. E. Prince et al. improvement with minimal stubble (0.5 and 1.0 mm) fol- lowed largely by impairment with increasing beard lengths. An interesting possibility is that the gritty texture of stubble hairs may have facilitated a“Velcro effect”to enable the mask layer in direct contact with the face to adhere better and improve the seal with the skin. Beyond stubble growth, these masks, worn as recommended by the manufacturer, showed marked decreases in FFE with beard length, albeit the KF94 mask was consistently more resistant than the KN95 to increasing length throughout the range tested. The decline in performance of higher efficiency respirator type masks as a function of increased beard hair growth is attributed to disruption of the seal formed between the margins of the face mask and the facial skin. From this perspective, the lack of a response to beard length shown by the procedure and cloth masks may reflect the overall low FFE performance of these face masks. Modest improvements in FFE within a narrow beard hair length range noted for Volunteer 1 could be related to facial morphology, suggesting these masks may have better con- formed to the face with added depth. Other possibilities include addedfiltration efficiency by hairfibers, through sieving and/or electrostatic mechanisms, or an improvement in adherence of the face mask relative to bare skin. Improved FFE previously demonstrated with simple mod- ifications to procedure earloop masks [15] suggests that such enhancements would also benefit bearded individuals. As demonstrated here, covering beard hair with an elastic exercise band can substantially mitigate potential perfor- mance declines of face masks, by creating a surface that more closely approximates the facial skin. This low cost, easily available intervention is especially likely to benefit masks constructed with melt-blown materials that perform best when theyfit well (good adherence and conformation to the face) and could be an option for bearded individuals for whom it is a cultural norm, religious imperative, or medical necessity (e.g., skin sensitivity) to not shave. Recent media reports [16] highlighted many males who have grown so-called quarantine beards. For this group, covering or closely trimming beards (combined with mask fit and seal enhancements), would likely provide a greater degree of protection from infectious aerosols. Given the prevalent use of a wide variety of masks during the current pandemic, results from this study and previous work provide practical solutions to improvefiltration per- formance across different types of populations as well as face masks. Evaluation and steps taken to match the shape, size, and features of faces with the best protective face masks broadly available, together withfit enhancements and train- ing for proper wearing and usage, may help to reduce viral transmission. Future research on the relationship between facial features and FFE and the balance of improvement modifications as a function of material andfitparameterswillhelp protect the health of wearers in the context of multiple airborne exposures including environmental pollution and pathogens such as the SARS-CoV-2 virus. Although a partial replication was noted in data from Volunteer 2, one limitation of the current study is that to provide a controlled point of reference, most of the data showing the incremental effect of beard hair length on FFE of thefive face masks tested and exercise bands were col- lected by repeated cycles of beard growth andfit testing of a single individual. Therefore, thefindings may not com- pletely capture the role of interindividual factors such as differences in facial morphology, beard density, and hair texture. Additional studies are needed to investigate the potential effect of these variables on face maskfitting and the potential interaction with beard hair. Higher overall FFE for bearded subjects wearing N95 respirators (Table1and Supplementary Fig. 1), relative to Volunteer 1, may be a function of the specifi c mask model tested, and/or char- acteristics of an individual’s face and facial hair. Additional testing with different shapes, styles, and material rigidities of N95 respirators (e.g., horizontal shape with fold outflaps, cup shape, cone shape, duckbill) in a broader sample of volunteers would help to address the factors contributing to variability reported here and observed by others [6,10–12]. Similarly, results we report here show the efficiency of face masks worn as personal protective equipment. The extent to which thesefindings are predictive of the performance of face masks as source control, specifically in reducing the emission of SARS-CoV-2 aerosols, is not known at present and is a focus of ongoing investigation. Finally, despite standardized procedures, individual differences in profi- ciency of maskfit testing can add variability to observed results. Taken together, this study furthers our understanding of the impact of a variety of beard lengths on the efficiency of facial coverings commonly available to the public. Detailed and repeated results in a volunteer who showed impaired performance with facial hair indicated that FFE decreased linearly with hair length for the three most effective facial coverings tested (N95, KF94, and KN95). The relatively low performance of procedure style and cloth masks was not substantially impacted by the presence of facial hair. An intervention using exercise bands as a beard cover yielded performance improvements for all face masks tested, while reducing performance variability. Face mask performance is related to the materials used and the interaction of various shapes, sizes, and sealing surfaces with a given wearer’s face. The presence of a beard is likely to impactfit para- meters for many available styles and it is worth considering limiting its length, completely shaving (if feasible), or covering beards to achieve betterfiltration performance. Combining this approach with other readily availablefit improvement modifications should help reduce exposure to Assessing the effect of beard hair lengths on face masks used as personal protective equipment during. . . 959 environmental contaminant particles, including both pollu- tion and pathogens such as the SARS-CoV-2 virus, and thereby benefit public health. AcknowledgementsThe authors gratefully acknowledge Philip A. Bromberg, MD, and the journal reviewers for their critical review of the manuscript. Author contributionsSEP, HC, HT, and JMS had access to the data and take responsibility for the integrity of the data and the accuracy of the analysis. Concept and design: SEP and JMS. Data collection, analyses, and interpretation: SEP, JMS, HC, HT, WDB, KLZ, PWC, and JB. Drafting of the manuscript: SEP, HC, and JMS. Critical revision of the manuscript: all authors. Administrative, technical, or material support: SEP, JMS, HT. Supervision: JMS, HT, and SEP. Compliance with ethical standards Conflict of interestThe authors declare no competing interests. Publisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. Sunjaya AP, Jenkins C. Rationale for universal face masks in public against COVID-19. Respirology. 2020;25:678–9. 2. Oestenstad RK, Bartolucci AA. Factors affecting the location and shape of face seal leak sites on half-mask respirators. J Occup Environ Hyg. 2010;7:332–41. 3. OSHA Respiratory protection–OSHA. Final rule; request for com- ment on paperwork requirements. Fed Regist. 1998;63:1152–1300. 4. Statista.https://www.statista.com/statistics/725005/men-wearing- beards-us/. Accessed 18 Dec 2020. 5. Smith M. YouGov data shows how beards have become more popular since 2011, especially among younger Brits. 2017.https:// yougov.co.uk/topics/politics/articles-reports/2017/03/10/beards-a re-growing-british-public. Accessed 18 Dec 2020.6. Frost S, Harding A. The effect of wearer stubble on the protection given by Filtering Facepieces Class 3 (FFO3) and Half Masks. Health and Safety Executive, 2015. 7. CDC. To Beard or not to Beard? That’s a good Question!https:// blogs.cdc.gov/niosh-science-blog/2017/11/02/noshave/. Accessed 18 Dec 2020. 8. Sickbert-Bennett EE, Samet JM, Clapp PW, Chen H, Berntsen J, Zeman KL, et al. Filtration efficiency of hospital face mask alternatives available for use during the COVID-19 pandemic. JAMA Intern Med. 2020.https://doi.org/10.1001/jamainternmed. 2020.4221. 9. Singh R, Safri HS, Singh S, Ubhi BS, Singh G, Alg GS, et al. Under-mask beard cover (Singh Thattha technique) for donning respirator masks in COVID-19 patient care. J Hosp Infect. 2020;106:782–5. 10. Skretvedt OT, Loschiavo JG. Effect of facial hair on the face seal of negative-pressure respirators. Am Ind Hyg Assoc J. 1984;45: 63–66. 11. Sandaradura I, Goeman E, Pontivivo G, Fine E, Gray H, Kerr S, et al. A close shave? Performance of P2/N95 respirators in healthcare workers with facial hair: results of the BEARDS (BEnchmarking Adequate Respiratory DefenceS) study. J Hosp Infect. 2020;104:529–33. 12. Floyd EL, Henry JB, Johnson DL. Influence of facial hair length, coarseness, and areal density on seal leakage of a tight-fitting half- face respirator. J Occup Environ Hyg. 2018;15:334–40. 13. McLure HA, Mannam M, Talboys CA, Azadian BS, Yentis SM. The effect of facial hair and sex on the dispersal of bacteria below a masked subject. Anaesthesia. 2000;55:173–6. 14. Grinshpun SA, Haruta H, Eninger RM, Reponen T, McKay RT, Lee SA. Performance of an N95filtering facepiece particulate respirator and a surgical mask during human breathing: two pathways for particle penetration. J Occup Environ Hyg. 2009;6: 593–603. 15. Clapp PW, Sickbert-Bennett EE, Samet JM, Berntsen J, Zeman KL, Anderson DJ, et al. Evaluation of cloth masks and modified procedure masks as personal protective equipment for the public during the COVID-19 pandemic. JAMA Intern Med. 2020. https://doi.org/10.1001/jamainternmed.2020.8168. 16. Lipstein RL. 2020.https://www.newyorker.com/culture/video- dept/the-rise-of-the-quarantine-beard. 960S. E. Prince et al.