Power analysis

A power analysis was conducted a priori, with an effect size of 1, to determine the sample size needed to result in 80% power when using a paired t test with an α set at the 0.05 using a normal distribution. We performed the calculations using R Core Team 2020 (R: A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria). The determined sample size was 16.7 for each group; therefore, at mimunm, 17 trials were conducted.

Coughing machine virus aerosolization

A coughing machine was designed to simulate a patient coughing viral aerosolized particles and exposing HCPs (article currently under review). The coughing machine operates by pushing 4.1 L of air, which is mixed with aerosolized virus through a mannequin’s mouth, producing a coughing plume. The resulting plume is like a human cough. A simulated HCP was placed 0.41 m in front of the cough to simulate exposure in a hospital or other care facility. A face shield was placed on the simulated HCP to mimic part of the PPE worn. MS2, the bacteriophage used as a surrogate for SARS-CoV-2 was placed in the liquid reservoir of the coughing machine and was aerosolized during the cough. Three consecutive simulated coughs were performed in each trial to mimic a “coughing fit.”

Virus and particle sampling

Virus and particle concentrations were sampled across the disposable plastic face shield (placed on a simulated HCP) while the coughing machine was operated. Two SKC biosamplers (no. 225-9595, SKC Inc, Eighty-Four, PA) were used to sample the MS2 (virus concentration) and 2 optical particle counters (OPCs) (GRIMM 11-D Aerosol Counters, nos. 619012 and 619025, Durag Group, Germany) were used to sample the particle concentration. The Biosamplers and OPCs were placed on the inside and the outside of the simulated HCP face shield (Fig. 1). The liquid media used in the biosampler was 20 mL phosphate-buffered saline (PBS, Gibco, 1x, pH 7.4 -/-, TX). The biosamplers were each connected to a pump that sampled at 12.5 liters per minute (LPM). The OPCs used a 1-second interval and sampled number concentration information across 16 channel (bin) sizes (range, 0.253–3.515 µm). The biosamplers and OPCs were started 1 minute prior to the first cough and operated nonstop throughout the 3 coughs (∼28 seconds).

Figure 1.

Sampling setup. The disposable plastic face shield is placed on the simulated healthcare personnel with the coughing mannequin representing a patient with COVID-19. Biosamplers and optical particle counters (OPCs) are placed on the inside and the outside of the face shield to evaluate the healthcare personnel’s exposure to bioaerosols produced by coughing.

The sampling was conducted in an undisturbed chamber (8.5 m × 3.7 m × 2.4 m). Relative humidity and temperature were also monitored to ensure consistency across trials using a direct reading instrument (Extech Instruments Hygro-Thermometer Model SDL500, serial no. Z335535). After sampling, the chamber-like room and the coughing machine were cleaned with sodium hypochlorite wipes and 70% ethanol.

Virus sample analysis

Viability of the biosampler liquid media was assessed using a plaque assay. The sample was concentrated from ∼20 mL to <1,000 µL using an Amicon Ultra 15 centrifugal filter (Cork, Ireland), which was spun in the centrifuge at 4,000 rotations per minute (RPM) in 1-minute intervals. The concentrated sample volume was measured and recorded. A serial dilution was then conducted using 100 µL of the concentrated sample and 900 µL of MS2 broth (Appendix A). Petri dishes (Fisher Scientific, 100 mm × 15 mm, Polystyrene, MA) for the plaque assay were prepared using 10 mL bottom agar (Appendix A), which was left to rest for 30 minutes before any overlay was added. Top agar was prepared and placed in aliquots in 3-mL tubes (FalconBrand, 14-mL polystyrene round bottom tubes, Mexico), which were then incubated for 20 minutes at 45°C. After the incubation period, 10 µL liquid Escherichia coli (10⁷ PFU/mL, ATCC15597) and 100 µL serial dilution sample were added to the 3 mL of top agar, which was mixed and poured over the solidified-bottom agar plates. Each serial dilution was plated in triplicate. The plates were left to solidify for 20 minutes and then sealed with a thermoplastic film (Parafilm wrapping film, Bemis, RPI, WI). The sealed plates were placed upside down in a 37°C incubator for 14–24 hours. After the incubation period, the MS2 plaques formed. The serial dilution plate that produced between 20 and 250 plaques were then counted. The arithmetic mean was calculated using the triplicate sample.

Viruses were quantified by counting plaque-forming units (PFU) per unit volume (eg, PFU/mL). ^{Reference Langlet, Gaboriaud and Gantzer10,Reference Baer and Kehn-Hall11} Viable virus was the primary concern because it is needed to infect other individuals. ^{Reference Letizia, Smith and Ge12} In the case of nosocomial infection, a patient must shed viable virus for the HCP to be infected. ^{Reference Letizia, Smith and Ge12} The viable virus concentration was calculated from the PFU. The plaque count and dilution factor were used to calculate the plaque concentration (Eq. 1). The plaque concentration, concentrated volume, and air volume sampled were then used to determine the airborne concentration of MS2 (Eq. 2).

(1) $$\eqalign{{\rm{Plaque\;}}\,{\rm{Concentration\;}}\left( {{\rm{PFU}}} \right) =\ ({{{\rm{Average\;}}\,{\rm{plaque}}\,{\rm{forming\;}}\,{\rm{units}}}})/ \cr ({{{\rm{Total\;}}\,{\rm{volume}}\,{\rm{plated}}}})}$$

(2) $$\eqalign{{\rm{Sample\;Concentration}}\,{\rm{\;}}({\rm{PFU/}}{{\rm{m}}^3}) =\ {{{\rm{Plaque\;concentration}}\,{\rm{\;}}({\rm{PFU}})}}/\cr {{{\rm{Volume\;of\;air\;sampled\;}}\,{\rm{(}}{{\rm{m}}^3}{\rm{)}}}} }$$

After all 18 trials were completed, the normality of the samples was tested using the Shapiro-Wilks test. A box plot showing the mean, median, outliers, minimum, and maximum was computed across the 18 trials for the inside and outside of the face shield. A paired t test was performed to compare the 2 groups, with a critical P set to 0.05.

Particle sample analysis

The OPCs were connected to 2 computers operating the OPC Software (Spectrometer, GRIMM Control, 1179, V1-0-6, 25-10-2018) where particle data were recorded, stored, and exported to Microsoft Excel (Microsoft Office Professional Plus 2016, version 16.0.5278.1000, Redmond, WA) for analysis. The inside and outside particle concentration data for all the trials were compiled for each channel size. The Shapiro-Wilk test was used to assess for normality, after the data were log transformed. Any censored data (data that were recorded as zero) were assigned a value below the limit of detection (LOD, 5 particles/m³). ¹³ All OPC zero values were replaced with log (2.5) or 0.39794 (count/m³). Descriptive statistics were calculated using box plots to identify the median of the first quartile, the median of third quartile, standard deviations, and outliers for each experimental condition. For each variable, outliers were identified as values greater than 1.5 interquartile range (IQR) of the third quartile or less than 1.5 IQR of the first quartile. Paired t tests were also performed to examine the differences in number particle concentration between the inside and outside of the face shield. The paired t test was conducted using an unadjusted α of .05 as well as an α that was adjusted for false discovery rate using the Benjamini Hochberg procedure. ^{Reference Benjamini and Hochberg14}