key: cord-0891460-19xbno94
authors: Coyle, J. P.; Derk, R. C.; Lindsley, W. G.; Boots, T.; Blachere, F. M.; Reynolds, J. S.; McKinney, W. G.; Sinsel, E. W.; Lemons, A. R.; Beezhold, D. H.; Noti, J. D.
title: Reduction of exposure to simulated respiratory aerosols using ventilation, physical distancing, and universal masking
date: 2021-09-22
journal: nan
DOI: 10.1101/2021.09.16.21263702
sha: df6abd52c28490809476e56e67a98fec0fea0432
doc_id: 891460
cord_uid: 19xbno94

To limit community spread of SARS-CoV-2, CDC recommends universal masking indoors, maintaining 1.8 m of physical distancing, adequate ventilation, and avoiding crowded indoor spaces. Several studies have examined the independent influence of each control strategy in mitigating transmission in isolation, yet controls are often implemented concomitantly within an indoor environment. To address the influence of physical distancing, universal masking, and ventilation on very fine respiratory droplets and aerosol particle exposure, a simulator that coughed and exhaled aerosols (the source) and a second breathing simulator (the recipient) were placed in an exposure chamber. When controlling for the other two mitigation strategies, universal masking with 3-ply cotton masks reduced exposure to 0.3-3 m coughed and exhaled aerosol particles by > 77% compared to unmasked tests, whereas physical distancing (0.9 or 1.8 m) significantly changed exposure to cough but not exhaled aerosols. The effectiveness of ventilation depended upon the respiratory activity, i.e., coughing or breathing, as well as the duration of exposure time. Our results demonstrate that a combination of administrative and engineering controls can reduce personal inhalation exposure to potentially infectious very fine respiratory droplets and aerosol particles within an indoor environment.

• Universal masking provided the most effective strategy in reducing inhalational exposure 26 to simulated aerosols. 27

• Physical distancing provided limited reductions in exposure to small aerosol particles. 28

• Ventilation promotes air mixing in addition to aerosol removal, thus altering the exposure 29 profile to individuals. 30

• A combination of mitigation strategies can effectively reduce exposure to potentially 31 infectious aerosols. 32

for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

The association between human respiratory infection transmission by respiratory droplets and 52 aerosols is well-established for several known pathogens. 1 Given that the average individual 53 spends > 90% of their day indoors 2 , there has been intense focus on factors associated with 54 indoor transmission of SARS-CoV-2, the virus that causes 4 Epidemiological 55 investigations highlight the role of congested, poorly ventilated spaces with high levels of 56 secondary attack rates and community transmission. 5,6 While the specific contribution of 57 respiratory droplets and aerosols remains a topic of active research , increasing evidence of 58 asymptomatic and pre-symptomatic individuals 7 contributing to community COVID-19 59 transmission suggests that very fine respiratory droplets and aerosol particles can spread SARS-60

CoV-2. 8 To minimize exposure risks, the Centers for Disease Control and Prevention (CDC) 61 recommends several mitigation strategies to limit COVID-19 transmission, including wearing 62 masks, maintaining physical distances, and avoiding crowded indoor and outdoor spaces, among 63 other strategies. 9,10 64 opening of the recipient simulator to allow for measurement in the personal breathing zone 117 outside of a mask affixed to the simulator. All OPCs were controlled and data logged using a 118 custom program in LabVIEW v. 2009 (National Instruments; Austin, TX) . 119

In addition to particle removal, the HEPA system provided ventilation, with a variable 120 transformer (Staco Energy Products, Co.; Miamisburg, OH) used to set the HEPA system flow 121 rate. Air exchange rates were determined via single-point measurement of the linear air flow at 122 the inlet duct using a Model 5725 VelociCalc rotating vane anemometer (TSI, Inc.; ; Shoreview, 123 MN) equipped with a tapered air cone (TSI, Inc.). The inlet duct was straightened for a length of 124 > 10 diameters from the inlet to minimize turbulent flow during anemometer readings for air 125 changes per hour (ACH) derivation. The HEPA system was set to 0 ACH, 4 ACH (15.3 m 3 /min 126 flow), 6 ACH (22.9 m 3 /min), and 12 ACH (45.9 m 3 /min); calculations assumed zero leakage into 127 the chamber. Effective air filtration rates were derived empirically. Briefly, the chamber was 128 saturated with particles using a stand-alone TSI Model 8026 generator until the 0.3-0.4 um 129 particle size channel reached 10 5 particles per liter under constant mixing using a household fan. 130

The particles for effective air changes per hour were generated using a 1% solution of NaCl in 131 distilled water formulated from 100 mg tablets provided with the TSI Model 8026 generator as 132 per manufacturer's instructions. After a 15-minute air settling period, the HEPA filtration system 133 was set to the desired ACH based on anemometer measurements. Particle concentrations were 134 measured for 20 minutes using five of the six OPCs to derive particle exponential decay curves 135 spatially throughout the chamber. Theoretical particle exponential decay curves were modeled 136 from the three smallest size bins (0.3-0.4 µm, 0.4-0.5 µm, and 0.5-0.65 µm) assuming 137 negligible loss to chamber surfaces and aerosol agglomeration using MATLAB v. 9.6 138 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 tests) or 15 l/min (breathing tests), and neutralization by an ionizer (Model HPX-1, 162 Electrostatics, Inc.; Hatfield, PA). The coughing modality was performed by loading the 163 simulator elastomeric bellows with test aerosol, followed by a single 4.2 l rapid exhalation at a 164 peak flow rate of 11 l/min; 31 the simulator did not breathe following the cough. For breathing 165 tests, the simulator breathing rate was 12 breaths/min with a tidal volume of 1.25 l and 166 ventilation rate of 15 l/min. The breathing parameters correspond to the ISO standard for females 167 performing light work. 32 For the breathing modality, the nebulizer was cycled 10 seconds on and 168 50 seconds off continuously throughout the test duration. Tests were conducted for a duration of 169 15 minutes, except for a limited subset of testing conditions which were conducted for 60 170 minutes. As an additional examination of the time-dependency of ventilation in reducing 171 recipient exposure, additional tests were conducted using a modified aerosol generation cadence 172 during the breathing action. During these tests, the nebulizer generated aerosol continuously for 173 the first 3 minutes of the test, after which the nebulizer was turned off, and are henceforth 174 designated short-term aerosol generation tests. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

After mask fitting and distance configuration, the environmental chamber was sealed, and 202 the HEPA filtration system run at maximal rate to minimize background airborne particles. 203

Thereafter, the HEPA filtration system was either turned off (0 ACH) or set to the desired air 204 exchange rate (4-12 ACH) and allowed to run for 15 minutes, during which time all OPCs were 205 initialized to begin particle concentration data collection and the recipient simulator activated to 206 begin breathing. After the air exchange stabilized, the source simulator was initiated to cough or 207 breathe, and aerosol concentrations were measured for 15 minutes. The chamber was allowed to 208 cool to 22 °C between experiments to reduce the inter-test temperature variability. Three 209 independent experimental replicates were conducted for each unique experimental condition. 210 211

The background aerosol concentration was determined based on the mean particle 213 concentration during the 3 minutes prior to cough or exhalation. The bin-specific particle counts 214 per cubic meter of air were converted to volume based on the mean bin diameter (assuming 215 spherical particles) and then to mass concentration by multiplying by the density of KCl (1.984 216 g/cm 3 ). The total mass concentration was calculated by summing the bin-specific mass 217 concentrations for all size bins. The mean mass concentration was calculated as the average mass 218 concentration over the test duration and served as the exposure metric in these simulations. OPC 219 data were processed using the R statistical environment v. 4.0.2 (R Project for Statistical 220

Computing; Vienna, Austria). All point estimates are presented as the arithmetic mean ± 1 221 standard deviation of the measured mean mass concentration. 222

for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Across all experiments, the mean chamber temperature was 24.1 ± 1.1 °C with a relative 237 humidity of 26.0% ± 2.4%. Particle clearance by the ventilation system followed first-order 238 exponential decays, with overall clearance rates 74.1% ± 4.4% of decay rates estimated by 239 anemometer readings (Range: 73.1%-76.7%; Figure 2A ). Particle decay rates throughout the 240 chamber, as measured by the five OPCs, were largely homogeneous (Supplemental Figure S1 ). 241

The experimental decay rates after single coughs were 76.1% ± 1.5% of theoretical values 242 (Range: 74.4%-77.3%). These experimental decay rate magnitudes and variances were 243 comparable to those obtained from particle decay testing, which suggests that the ventilation 244 system promoted adequate air mixing to disperse cough aerosols through the chamber volume. 245 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The time-concentration curves at the 1.8 m physical distance are shown in Figure 3A ; 268 analogous results for the 0.9 m physical distance are presented in Supplemental Figure This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Adjusting for ACH and physical distance, universal masking significantly reduced 287 aerosol exposure compared to unmasked exposures (p < 0.001 among all modalities) during the 288 15-minute tests. Fit factors of the 3-ply cloth mask were 4.1 ± 2.6 (n = 43) for the recipient and 289 1.7 ± 0.6 (n = 42) for the source simulator. The largest reduction in aerosol mass exposure was 290 observed after a single cough (90.9%; CI95%: 89.6%-91.9%), while exposure reduction was 291 comparatively lower during breathing (80.8%; CI95%: 77.5%-83.6%). The reduction in mass 292 concentration was likely due to preferential filtration of aerosols > 1 µm in diameter after having 293 traversed two masks (Figure 4 ). The differences in exposure reduction among the aerosol 294 generation modalities were likely due to specific changes in aerosol spatiotemporal dispersion 295 when the source was masked. Aerosol plumes generated during both breathing modalities and a 296 single cough escape through face seal leaks. 36 The plumes would then be deflected behind and/or 297 to the side of the source and thus effectively farther from the recipient compared with the 298 experiments with no masks. Without chamber mixing, as observed with no ventilation, the cough 299 aerosol deflected by the mask took longer to disperse throughout the chamber compared to 300 without a mask as was observed in Figure 3A . The time concentration curves for breathing 301 shifted to the right when masked, though not as much as after a single cough, showing that 302 dispersion kinetics likely played a larger role in the heterogeneity observed for exposure 303 reduction among the respiratory actions simulated here. While we cannot rule out the possibility 304 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 that differential filtration of the source's mask was influenced by aerosol generation (for 305 example, the higher expulsion velocity during coughing causing greater mask aerosol filtration 306 compared to breathing), our previous work suggests the aerosol generation modality likely does 307 not influence mask collection efficiency for this 3-ply cotton mask (51.7% ± 7.1 % for coughing 308 and 44.3% ± 14.0% for breathing). 13 Lastly, we have previously shown the recipient's breathing 309 can contribute towards airflow patterns and subsequently influence their specific exposure 37 . 310

The exposure reductions associated with the other predictor variables varied depending 311 on the respiratory action for the 15-minute tests. When controlling for masking, increasing 312 physical distance from 0.9 m to 1.8 m significantly reduced aerosol exposure from a single 313 cough by 15.4% (CI95%: 3.9%-25.5 %; p = 0.011); increasing ventilation also reduced exposure 314 by 4.3% per ACH (CI95%: 2.9%-5.7%; p < 0.001). Neither increasing ACH (p = 0.522) nor 315 increasing physical distance (p = 0.451) provided protection during breathing for the 15-minute 316 tests. When extending the test duration to 60 minutes for breathing, the mean mass concentration 317 from aerosol generation reached a dynamic equilibrium with each of the examined ACH rates 318 ( Figure 5) . Analysis for the 60-minute tests by OLS regression demonstrated increasing 319 ventilation significantly decreased mean mass concentration by 9.0% (CI95%: 7.7%-10.3%; p < 320 0.001; Table 2 ), while universal masking expectedly reduced mean mass concentration 321 significantly. When condensing the total aerosol generation period to the initial 3 minutes in the 322 short-term aerosol generation tests, increasing ACH became a significant predictor in exposure 323 reduction (5.2%; CI95%: 3.8%-6.5%; p < 0.001; supplemental table 2). The time-concentration 324 curves of the short-term aerosol generation tests demonstrated the log-linear decay similarly to 325 time-concentration profiles observed from a single cough, albeit shifted to the right to reflect the 326 longer aerosol generation period (Supplemental Figure S3 ). This result demonstrates that 327 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101/2021.09.16.21263702 doi: medRxiv preprint attainment of a dynamic equilibrium with continuous aerosol input or removal of aerosols 328 produced by an intense, short-term generation events through increasing ventilation can result in 329 significant exposure reduction for a recipient. We did not examine the extended exposure 330 duration for a single cough over 60 minutes, though we expect increasing ventilation will remain 331 a significant predictor of mean mass concentration reduction. 332

With respect to ventilation, the restricted 15-minute exposure duration contributed to the 333 lack of pronounced effect of increasing ACH for breathing and can be explained when 334 considering air flow. Ventilation not only provides contaminant removal, but also impacts the 335 overall air flow. Modeling of aerosol dispersion through central ventilation systems demonstrates 336 this complex interplay between ventilatory clearance and overall air flow patterns that can, under 337 certain situations, increase the short-term exposure during rapid, thorough mixing 38 that was 338 observed during the breathing respiratory action. Increasing ventilation reduced monotonically 339 the bulk aerosol concentration throughout the entire chamber over the total duration of the 340 ventilation testing ( Figure 3A ) but tended to decrease the time of aerosol contact at the mouth of 341 the recipient. Therefore, ventilation tended to increase the recipient aerosol exposure at the onset 342 of testing where, in extreme cases, ventilation paradoxically increased the mean mass exposure 343 ( Figure 3B ); this observation was independent of physical distance. The authors opine such 344 increases were caused by the observed thorough air mixing, as was noted in particle decay 345 studies, in conjunction with the short exposure duration of 15 minutes. As previously noted, the 346 effect of ventilation was appreciated during the 60-minute breathing test and the short-term 347 aerosol generation tests. These results demonstrate that the aerosol reduction measures by 348 ventilation must consider the air mixing, aerosol spatial dispersion, and exposure duration in 349 addition to other mitigation strategies to ascribe the degree of protection afforded. This becomes 350 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The current investigation has several noteworthy limitations that must be considered. First, the 359 mass concentration of aerosol generated during the experimental modalities, particularly 360 breathing, was higher than those produced from human exhalations. 21 The higher concentrations 361 combined with the wide dynamic range of the OPC allowed for stable and reproducible 362 measurements while assuring attainment of quantitative limits of detection among all tests. 363

Second, the simulators lack generation of body heat and the ability to generate a thermal 364 exhalation plume, which affects aerosol dispersion and inhalation exposure. 41 Given the confines 365 of the environmental chamber, the internal ventilation setup, and the high aerosol concentrations, 366 we would not expect substantial differences in mean mass exposure given the small volume of 367 the chamber. Therefore, limits must be placed on the interpretation of the results within a larger 368 indoor environment, especially considering the dispersion potential of an exhalatory thermal 369 plume. Third, the range of human respiratory aerosols can be smaller and larger than the 370 measured range of this investigation (0.3-3.0 µm). 33, 35 For droplets, the effect of physical 371 distancing may be higher than those suggested by the observed results for droplets. Fourth, the 372 study investigated the exposure reduction of a single 3-ply cotton mask. The authors recognize 373 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 the limitation of having tested a single mask, since the effectiveness of exposure reduction by 374 other masks could be either higher or lower, depending on the mask. Nonetheless, the analytics 375 of the study allow for reasonable expectation of exposure reduction of the other predictor 376 variables provided the aerosol behavior does not significantly deviate from this study with 377 another type of mask. 378

for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

The current investigation highlights the contribution of three common engineering and 381 administrative controls recommended for limiting SARS-CoV-2 exposure within an indoor 382 environment: ventilation, physical distancing, and universal masking. When controlling for the 383 other two mitigation strategies, universal masking with a 3-ply cotton mask contributed to the 384 plurality of the observed reduction in aerosol mass exposure irrespective of aerosol generation 385 modality. This reduction was due, in part, to preferential reduction of particles > 1. between 0.9 m and 1.8 m), and optical particle counters (green dots) for area measurements (S1-524 4) and personal breathing zone measurements at the mouth (M) and beside the head (B) of the 525 recipient. The HEPA system intake and exhaust are shown with the HEPA filter and blower unit 526 demarcated by the red square containing an "X". 527 24 ng for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101/2021.09.16.21263702 doi: medRxiv preprint simulated very fine respiratory droplets and aerosol particles for the examined respiratory actions 534 and ventilation rates. (C) Bin-specific particle distributions as determined by mass (bars) and by 535 number of particles (line). The median particle diameter (Dp) indicates the bin. Results are the 536 arithmetic mean ± standard deviation of three independent experiments. Error bars for the 537 number of particles (line) too small to visualize. ACH = Air changes per hour. 538

for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 22, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

Aerosol transmission of infectious disease

It's about time: a comparison 413 of Canadian and American time-activity patterns

Indoor transmission of SARS-CoV-2. 416 Indoor Air

Estimation of the asymptomatic ratio of novel 418 coronavirus infections (COVID-19)

COVID-19 Outbreak Associated with Air Conditioning in 420

Probable airborne transmission of SARS-CoV-2 in a poorly 422 ventilated restaurant

Evidence Supporting Transmission of Severe Acute 424 Respiratory Syndrome Coronavirus 2 While Presymptomatic or Asymptomatic. Emerg 425 Infect Dis

A paradigm shift to combat indoor respiratory 427 infection

CDC calls on Americans to wear masks to prevent COVID-19 spread

A comparison of performance metrics 438 for cloth face masks as source control devices for simulated cough and exhalation 439 aerosols

Comparison of performance of three 441 different types of respiratory protection devices

Surgical mask filter and fit performance

Community Use Of Face Masks And COVID-19: Evidence From A 446 Natural Experiment Of State Mandates In The US. Health Aff (Millwood)

Trends in County-Level COVID-19 449 Incidence in Counties With and Without a Mask Mandate -Kansas

Association of State-Issued Mask Mandates

Allowing On-Premises Restaurant Dining with County-Level COVID-19 Case and Death 453

MMWR Morb Mortal Wkly 454 Rep

Face masks, public policies and 456 slowing the spread of COVID-19: evidence from Canada

Causal impact of masks, policies, behavior on 459 early covid-19 pandemic in the U.S

Aerosol 461 emission and superemission during human speech increase with voice loudness

Turbulent Gas Clouds and Respiratory Pathogen Emissions: Potential 464 Implications for Reducing Transmission of COVID-19

High SARS-CoV-2 Attack Rate Following 466 Exposure at a Choir Practice

Enacting national social distancing 469 policies corresponds with dramatic reduction in COVID19 infection rates

Examining the effect of social distancing on the compound growth 472 rate of COVID-19 at the county level (United States) using statistical analyses and a 473 random forest machine learning model

Social 475 distancing decreases an individual's likelihood of contracting COVID-19

Role of mechanical 478 ventilation in the airborne transmission of infectious agents in buildings

Influence of room ventilation 481 settings on aerosol clearance and distribution

Small droplet aerosols in poorly 483 ventilated spaces and SARS-CoV-2 transmission

Mathematical models for assessing the role of airflow on the risk 486 of airborne infection in hospital wards

Efficacy of an ambulance ventilation 488 system in reducing EMS worker exposure to airborne particles from a patient cough 489 aerosol simulator

International 492 Organization for Standardization

Size distribution and sites of origin of 494 droplets expelled from the human respiratory tract during expiratory activities

Modality of human expired aerosol size 497 distributions

Size distribution of 499 exhaled particles in the range from 0.01 to 2.0m

Visualizing the effectiveness of face masks in 501 obstructing respiratory jets

Efficacy of universal masking for source 503 control and personal protection from simulated cough and exhaled aerosols in a room

Investigation of potential aerosol transmission and 506 infectivity of SARS-CoV-2 through central ventilation systems

Factors involved in the aerosol 509 transmission of infection and control of ventilation in healthcare premises

Role of ventilation in airborne transmission of 512 infectious agents in the built environment -a multidisciplinary systematic review. Indoor 513 Air

How human thermal plume influences near-human transport of 515 respiratory droplets and airborne particles: a review

The authors wish to acknowledge the facilities, maintenance, and security personnel of NIOSH 394