Although SARS-CoV-2 RNA was present in 9% of all samples, no positive samples were found in the vicinity of CW1 or CW2 while they were used for patient care. The only positive sample in CW1 occurred while it was closed for cleaning. During this time, the negative pressure exhaust system was no longer in use, the ward doors were open, and cleaning crews were passing by the sampler to sanitize the ward. Other studies have documented positive air samples collected in COVID-19 patient rooms [13,14,15]. Our finding of no viral RNA outside of the COVID-19 wards while they were active (and only one positive sample outside the ICU) suggests that the negative pressure units were effective in limiting airborne exposure outside the units. Previous studies have shown that inside COVID-19 wards, including inside patient rooms, airborne SARS-CoV-2 RNA is detectable [13,14,15].
Unexpectedly, the nurses’ station on the non-COVID 19 ward (NCW) had the greatest number of positive samples. We observed frequent congregation of staff and consultants at this location. Although it was policy for all hospital personnel to wear masks, it is possible that the positive samples were due to breaches of mask-wearing. The lack of association between the number of patients in the ED and probability of a positive sample in the ED may be due to the fact that the patients in the ED were not predominantly COVID-19 patients, and the positive samples may instead reflect staff activity and patient flow near the ED workstation. The finding of greater positive rates in non-COVID-19 locations, in conjunction with the positive association between probability of a positive sample across all locations and the number of COVID-19 patients in the hospital, suggests that presence of SARS-CoV-2 RNA in the hospital reflects the disease burden more broadly in the community. This conclusion is supported by the strong positive association between the number of COVID-19 patients in the hospital and average daily new cases in Massachusetts.
The fact that we found concentrations in all particle size fractions suggests that virus-containing particles are from sources at different proximities to the sampler or produced by different mechanisms. SARS-CoV-2 RNA on larger particles, such as those in the NCW, may have been due to a cough by someone located near the sampler. Coughing generates larger particles than speaking . Viral RNA that was associated with smaller particles, such as that found in the ED, may reflect a greater distance between the sampler and source, formation of smaller aerosols from larger droplets (e.g., by evaporation), or production by processes emitting smaller particles (e.g., speaking as opposed to coughing). Finding positive samples during the cleaning period in CW1 may be due to resuspension caused by cleaning. We found these particles in the coarse (10.0–2.5 µm) size fraction. Liu et al. (2020) suggested that lofting of coarse particles may be caused by resuspension of particles from floors and hard surfaces .
This is the first study to document the presence of SARS-CoV-2 RNA in size-fractionated air samples in non-COVID-19 areas in a U.S. hospital. Previous efforts to study COVID-19 have been focused in COVID-19 patient care areas, with samplers located close to the source (infected patients) and collected onto a single bulk filter to analyze total suspended particulate (TSP). Santarpia et al. (2020), Ong et al. (2020), and Ding et al. (2020) all collected only TSP [15, 21, 22]. Liu et al. (2020) collected samples in Wuhan, China that were mostly TSP, with only three size-segregated samples . Chia et al. (2020) also collected only three size-segregated samples . We detected maximum concentrations on the same order of magnitude as the size segregated samples of Liu et al. (51 copies m−3 in our study; 42 copies m−3 for Liu et al., 2020) . Liu et al. (2020) found higher concentrations of SARS-CoV-2 RNA in the fine PM fraction than in larger sizes . Chia et al. (2020) had positive samples in the 4–1 µm size fractions . These results support our finding of SARS-CoV-2 RNA associated with fine particles that are capable of long-distance transport.
We detected a greater percentage of positive samples compared to some previous studies. The percent of positive samples was greater in our study (9%) than in the study by Ong et al. (2020) (0%) and Ding et al. (2020) (2%), despite the fact that these studies were conducted in COVID-19 patient care areas [21, 22]. Potential explanations for our higher positive sample rate may be related to a greater viral load in the air or methodological differences. For instance, our study collected a greater volume of air per sample (14.4 m3) compared to these studies (1.2 and 1.0 m3, respectively) and more samples.
The percentage of positive samples was smaller in our study compared to Chia et al. (2020) (67%), Liu et al. (2020) (77%), and Santarpia et al. (2020) (58–63%).[13, 14, 15]. A possible explanation is the proximity to the source: Chia et al. collected only in airborne infection isolation rooms of COVID-19 patients , and Santarpia collected only inside the Nebraska Biocontainment/National Quarantine Unit . Liu et al. sampled under conditions of higher disease prevalence (in Wuhan in February and March, 2020) . Differences in extraction efficiency from the collection substrate, variability in RNA degradation rates, or differences in PCR sensitivity among studies may also explain the differences in rates of positives samples and air concentrations.
The estimation of airborne virus concentrations (copies m−3) assumes that there is a continuous emission source. However, it is more likely that the emissions of the virus occurred as isolated events (e.g., a sneeze, cough, or speaking) from infected people rather than as a continuous flux over the entire 48-h sampling period. Since the calculated concentrations are time-weighted averages, someone exposed at the time of emission would likely receive a larger dose over a shorter time period than those implied based on the calculated concentrations.
While the present study detected SARS-CoV-2 RNA in hospital air samples, it did not determine whether the airborne virus was viable (capable of causing infection). Lednicky et al. (2020) recently reported that SARS-CoV-2 in hospital air is infectious . Santarpia et al. (2020) found viable SARS-CoV-2 in particles < 1 µm . Laboratory-generated aerosols containing SARS-CoV-2 were found to remain infectious for three  to 16 h . The infectious dose of SARS-CoV-2 is still unknown. It is possible that the infectious dose of SARS-CoV-2 is similar to that of SARS-CoV-1 , which was estimated to require 280 viral particles to cause illness in 50% of people . The concentrations measured in the present study are likely underestimated, which may be attributable to losses during extraction from the substrates, RNA degradation, and the sensitivity of PCR, as we detected SARS-CoV-2 RNA by shotgun sequencing in a sample near the lower limit of PCR detection (see section “Shotgun Sequencing” in Additional file 1).