This study reports the first quantification of infectivity of SARS-CoV-2 in aerosols sampled directly from exhaled air. Aerosol samples were culturable from three of 16 individuals with detectable SARS-CoV-2 RNA in exhaled air. From these three individuals, five of six culturable aerosol samples were successfully quantified. The highest infectivity was found for samples collected close to symptom onset and during singing. Based on the culture results, we calculated the emission rate from the three individuals during singing and for two of them also for talking. The emission rates were thereafter implemented in an indoor air model for calculating the time needed for a susceptible person to inhale one infectious dose when being in the same room as the infectious person. This time can be as short as 6 min when a highly infectious individual enters the room or only 1 min if the infected person already has been in the room long enough to reach steady-state concentration of viruses in the air.
Previous studies have quantified infectivity of SARS-CoV-2 sampled from room air, but not directly from exhaled air6,7,8. Vass et al. found an infectivity of 132 and 192 plaque-forming units (PFU)/L air in two of five virus positive air samples in a residential setting. Kitagawa et al. detected RNA in the majority (12 of 18) of their air samples, and found five culturable samples with 0.58–10 TCID50/L of sampled air. Lednicky et al. reported viable virus in four of six air samples collected in a patient room, ranging from 6 to 74 TCID50/L of air. In comparison, the steady-state concentration in a room in our study would be 2.0, 0.6 and 0.06 TCID50/L air with the three individuals, respectively, at enhanced ventilation and a half-life time of 30 min (Table 2); thus, similar concentrations as those measured by Kitagawa et al.8 but lower values than those in Lednicky et al.7. However, they both had their aerosol samplers within one meter from the source where the concentration is likely enhanced, while our model assumes equal mixing in the room. For our simulated normal ventilation, which is more applicable for comparison with the measurement in Vass et al.6, the steady-state concentration would be 4.5, 1.3 and 0.14 TCID50/L air (Table 2). Although the room sizes, respiratory activities, experimental procedures and analysis differ between these previous measurements, it seems likely that the steady-state concentrations we simulated can be reached in a room with an infectious individual.
To verify the results, aerosol samples from singing and NPH samples were cultivated twice in the qualitative assay. The genome sequences showed high agreement between the supernatant of culture-positive aerosol samples and NPH samples, which suggests that air and nasal viruses originated from the same individuals. Moreover, the isolated virus types represented those circulating in the region at the time of sample collection (Feb–Mar 2021). CPE also matched with genotypes, where alpha variants are more prone to result in syncytia formation than pre-alpha variants13. The samples in the current study were transported for a few hours in outdoor temperature (5–10 °C) before storage at − 80 °C for 1 year, and were freeze-thawed at least once before cultivation. Thus, due to suboptimal sample handling there is a risk that we underestimated the infectivity of the culture-positive samples and the total number of culture-positive samples.
Although the RNA concentration of a sample is not directly related to its infectivity, RNA concentration has often been used as a proxy of infectivity or transmissibility in clinical settings. Remarkably, the culture-positive samples in our study all had relatively low levels of SARS-CoV-2 RNA (Ct-value range: 32–38) and the sample showing the highest infectivity was not the sample with the highest RNA concentration. In this study, the successful cultivation is partly attributed to the early phase of the infection14,15. The aerosol samples from individual 1 and 2, which had the highest TCID50 values, were collected on the day of symptom onset, which is when peak infectiousness is reached16, yet also when higher concentrations of SARS-CoV-2 RNA have been found in aerosol samples8,11,12. Transmission before and around symptom onset has been an important factor driving the covid-19 pandemic17,18, and a good predictor of infectivity is likely a combination of viral load and days from symptom onset.
Individual emission rates have strong influence over the calculated time to inhale one infectious dose (from 6 to 37 min in the transient scenario, Fig. 2). The enhanced ventilation (3 ACH instead of 0.5) is of less importance in the modelled indoor setting. Our indoor air model is based on an assumption of instant complete mixing of room air, i.e. that the concentration of airborne virus is similar at all places in the room. This assumption is a reasonable approximation for room sizes up to a few dozens of cubic meters. Still, on shorter time scales of seconds to minutes, the concentration is higher close to the source.
The time airborne viruses remain infectious is difficult to measure, and it is altered by the local environmental conditions such as temperature and humidity19,20. However, from the sensitivity analysis (Fig. 3) we can see that the half-life time of viruses is of less importance in well-ventilated rooms, as aerosol particles are physically removed prior to virus inactivation. Oswin et al. found that a substantial part of the infectivity is lost within the very first seconds in the air, presumably in the environmental transition from exhaled breath to room air conditions20,21. The aerosol samples cultivated in this study were collected after about 20 s from emission and after drying. Thus, we estimate that the large initial loss of infectivity had already happened before the point where we measure the TCID50.
We used the ID50 of 10 TCID50 that was identified in a human challenge study on SARS-CoV-2 for unvaccinated people10. The infectious dose in Killingley et al. was derived from cultivation in Vero E6 cells that did not express the transmembrane protease 2 (TMPRSS2). VeroE6/TMPRSS2 cells have been shown to have about ten-fold increased entry efficiency of SARS-CoV-2 caused by the TMPRSS222,23. However, in the challenge study, SARS-CoV-2 was pipetted in the nose and hence, not inhaled via aerosols. For many viruses the infectious dose can be orders of magnitude lower via the aerosol route24,25. Our sensitivity analysis of the infectious dose (Fig. S1) shows that individual 1 and 2 are likely to transmit one infectious dose within 20–50 min also for an ID50 of 100 TCID50.
Our study includes the first model calculations based on measured source emission rates of exhaled viruses. However, previous studies have modelled well-documented superspreading events and estimated emission rates based on the number of people that became infected, by assuming aerosols as the only route of transmission26,27. Prentiss et al. analyzed six superspreading events with attack rates between 15 and 87% for which they assumed emission rates in the range 7.2 × 104 to 1.1 × 106 virions/h. This is in the same range as found here, if assuming 1 virion equals 1 TCID50 (4.6 * 105, 1.3 × 105 and 0.14 × 105 TCID50/h for individuals 1–3, respectively). Their calculated ID50 (notated N0) was in the range of 300–2000. Reichert et al. measured particle emission rates from two index cases in choir superspreader events and calculated an ID50 of 12 virions26, which is similar to what was found in the human challenge study. They also predicted that one person would have inhaled one infectious dose within 8 min.
This study presents experimentally measured emission rates of infectious SARS-CoV-2 aerosols during breathing, talking and singing. When applying the measured emission rates in an indoor air particle transmission model, we found that an infectious dose is inhaled within a few minutes in a typical room with normal or enhanced ventilation. These findings demonstrate the potential of rapid aerosol transmission of SARS-CoV-2 in indoor environments.
