Environmental Surveillance and Transmission Risk Assessments for SARS-CoV-2 in a Fitness Center

Fitness centers are considered high risk for SARS-CoV-2 transmission due to their high human occupancy and the type of activity taking place in them, especially when individuals pre-symptomatic or asymptomatic for COVID-19 exercise in the facilities. In this study, air (N=21) and surface (N=8) samples were collected at a fitness center through five sampling events from August to November 2020 after the reopening restrictions were lifted in Florida. The total attendance was ~2500 patrons during our air and environmental sampling work. Air samples were collected using stationary and personal bioaerosol samplers. Moistened flocked nylon swabs were used to collect samples from high-touch surfaces. We did not detect SARS-CoV-2 by rRT-PCR analyses in any air or surface sample. A simplified infection risk model based on the Wells-Riley equation predicts that the probability of infection in this fitness center was 1.77% following its ventilation system upgrades based on CDC guidelines, and that risk was further reduced to 0.89% when patrons used face masks. Our model also predicts that a combination of high ventilation, minimal air recirculation, air filtration, and UV sterilization of recirculated air reduced the infection risk up to 94% compared to poorly ventilated facilities. Amongst these measures, high ventilation with outdoor air is most critical in reducing the airborne transmission of SARS-CoV-2. For buildings that cannot avoid air recirculation due to energy costs, the use of high filtration and/or air disinfection devices are alternatives to reducing the probability of acquiring SARS-CoV-2 through inhalation exposure. In contrast to the perceived ranking of high risk, the infection risk in fitness centers that follow CDC reopening guidance, including implementation of engineering and administrative controls, and use of personal protective equipment, can be low, and these facilities can offer a relatively safe venue for patrons to exercise.


Section S1. Limit of detection of air samples
The LOD of the rRT-PCR detection system is ~1.5 SARS-CoV-2 genome equivalents per 25 µL rRT-PCR assay.Accordingly, the number of virus genome equivalents in 140 μL (volume of collection media from which RNA was purified) of the detection limit, N 140, is: As an example, for BioSpot-VIVAS and VIVAS air samples, the total amount of virus genome equivalents in the collection medium is: The concentration of virus in the air samples is: Results of LOD for other samples are shown in Table S1.

Section S3. Determination of the quanta emission rate
There are a few studies that have reported the estimated values of quanta emission rate.Miller et al. (2020) assessed the emission rate of 970 quanta h -1 for singing activity, based on the infection rate in a choir rehearsal outbreak.Based on viral load in the sputum, Buonanno et al. (2020) derived a function to estimate the emission rate from <1 quanta h -1 to >100 quanta h -1 , at different activity levels.Harrichandra et al. (2020) applied that model with a modified breathing rate and used the emission rate of 142 quanta h -1 to estimate the infection risk in nail salons.Dai and Zhao (2020) fitted a curve between emission rate and airborne disease reproductive number, including Tuberculosis, MERS, SARS, Influenza, and Measles, to obtain 14-48 quanta h -1 at light activity levels.Kriegel et al. (2020) implemented the previous model with measurements of particle emission rate and breathing rate at different activity levels, and they reported 139 quanta h -1 for light exercise.Assuming it depends on aerosol emission rates and breathing rates at different activity levels (Kriegel et al., 2020a), the average quanta emission rate can be written as where E 0 (quanta h -1 ) is the reference quanta emission rate and can be found in the aforementioned studies; Nj and N0 (particles cm -3 ) are the emitted particle number concentrations at unmodulated vocalization (i.e., assumed expiratory activity as baseline conditions in the fitness center) and reference expiratory activity (e.g., voice counting, beathing, and singing), respectively (Morawska et al., 2009); Vi (cm 3 ) is the volume of a single particle and i refers to four particle sizes (Morawska et al., 2009); Qb and Qb,0 (m 3 h -1 ) are the breathing rates at heavy exercise, 3.3 m 3 h -1 , and reference activity levels, respectively (Morawska et al., 2009).
Applying the estimated values from previous studies to Equation 4, the average emission rate of 896 quanta h -1 is obtained, and details are shown in Table S2.If more than one person exercising in the site is infectious, the inferred quanta emission rates would be the sum of all individual emission rates.b .Calculated from the median values of particle generations for singing and voice counting as 1923/133=14.4(Kriegel et al., 2020b); and based on the ratio of particle generations for voice counting and unmodulated vocalization (i.e., 6.41), calculated the ratio of particle generations for singing and unmodulated vocalization as 6.41/14.4=0.45 (Morawska et al., 2009).

Section S4. Sensitivity analysis of infection risk
The infection risk Pi is based on the estimated quanta emission rate when patrons vocalize in unmodulated voice.Among the four expiratory activities (i.e., breathing, whispered counting, voiced counting, and unmodulated vocalization) that generate different size distribution of particles (Morawska et al., 2009), unmodulated vocalization was conservatively used for fitness center patron expiratory activities.Buonanno et al. (2020) assumed the expiratory activity for patrons in a gym at the breathing level, resulting in the median quanta emission rate of 2.5 quanta h -1 person -1 .If this quanta value were used, the Pi is calculated as 0.005% at the visited fitness center, which is much lower than the Pi when the infected patron conducts unmodulated vocalization activities (E = 896 quanta h -1 person -1 ).
As a baseline assessment in the fitness center, the βdep= 0.3 h -1 measured by Thatcher et al.
(2002) was used, although a higher value of 1.5 h -1 was reported by Diapouli et al. (2013).The βina = 0.63 h -1 measured by van Doremalen et al. ( 2020) was used for the estimations, corresponding to the 1.1 h half-life of aerosolized virus, whereas a minimal decay for virus-laden aerosols after 16 hours reported by Fears et al. (2020) resulting in βina = 0 h -1 would contribute to Pi =1.87%, which is slightly higher than that calculated by βina = 0.63 h -1 (van Doremalen et al., 2020).As this assessment aims to investigate the effectiveness of ventilation in reducing the airborne SARS-CoV-2 transmission, the quanta emission rate at the unmodulated voice level and decay rate of 0.63 h -1 (van Doremalen et al., 2020) were adopted for the baseline examinations.

Section S5. Virus inactivation efficiency of the UVGI unit in the fans
Figure S2 displays the schematic diagram of the UVGI unit installed in the upper room with its dimensions.The UVC lamp is 9 watts, and air is drawn by a fan through the UVGI unit at a flow rate, Qr, of 930 m 3 h -1 .Based on Figure S2, the air velocity υr (m s -1 ) at the cross-section is: where Across-section (m 2 ) is the cross-sectional area of the air flow; Dr = 0.325 m is the diameter of the cross-sectional area.The residence time τr (s) of room air passing through the UVGI is: where Lr = 0.311 m is the length of the UVC lamp.
The median intensity can be approximated at the circular side at Rr,50 (Figure S2), which is the radius corresponding to half of the cross-sectional area defined as: Accordingly, Rr,50 is determined as: The median intensity Ir,50 of the UVC lamp is calculated by: where Pr = 9 watts is the power of UVC lamp; Ar,50 is the tubular area at radius of Rr,50.

Section S6. Virus inactivation efficiency of the UVGI unit in the AHU
There are 12 AHUs in the HVAC system at the visited fitness center, and details of AHU dimensions are shown in Table S2.The calculation of the virus inactivation efficiency in Main area 1 is shown here as an example, at 10 ACH with 36% air recirculation (conditions of the visited fitness center).Calculations of other areas can be obtained using the same method with the corresponding air flow rate listed in Table S3.For the Main area 1, the air velocity υA (m s -1 ) at the AHU inlet surface (H A×WA) (Figure S2), which is the same for all 12 areas, is: The median intensity IA,50 is calculated by: If one air stream enters AHU at point i (Figure S2), with degree of θ to the center horizontal surface, the residence time τA,i is: The weighted residence time τA for total air flow is: The median virus inactivation efficiency ηA,50 (unitless) can be described as (Tseng et al., 2005): The values of ηA,50 corresponding to the three k, obtained in Section S5 are 0.999, 0.974, and 0.995, respectively.Therefore, the calculated average virus inactivation efficiency in Main area unit 1 is  ,50 ������ = 98.9%.
On the basis of the averaged virus inactivation efficiency in each AHU, the summed efficiency of UVGI in AHU can be obtained by:

Section S7. Infection risk assessment when wearing a face mask
If a patron wears a cloth mask in the fitness center, the Wells-Riley equation can be defined as (Bazant and Bush, 2021): where pm is the face mask penetration factor.Konda et al. (2020) measured the aerosol filtration efficiency FE= 54-95% for a wide range of cloth masks using common fabrics (e.g., cotton quilt, hybrid-cotton, and double-layer fabrics).As pm=1-FE, the values of pm using cloth mask would be 5-46%.Using a conservative penetration factor of 50% would provide a baseline assessment for using masks in reducing the airborne transmission of SARS-CoV-2.

Figure S2 .
Figure S2.Schematic figure of UVC dose in upper room UVGI Figure S3.Schematic figure of UVC dose in the UVGI inside AHU

Table S3 .
Details of 12 AHU dimensions

Table S4 .
Calculations of the virus inactivation efficiency for UVGI inside AHU at different ventilation rates *Calculated by the measured flow rate divided by the cross-sectional area. *

Table S5 .
Detailed characteristics of the visited fitness center