We study transmission of COVID-19 using five well-documented case studies : a Washington state church choir, a Korean call center, a Korean exercise class, and two different Chinese bus trips. In all cases the likely index patients were pre-symptomatic or mildly symptomatic, which is when infective patients are most likely to interact with large groups of people. An estimate of N 0 , the characteristic number of COVID-19 virions needed to induce infection in each case, is found using a simple physical model of airborne transmission. We find that the N 0 values are similar for five COVID-19 superspreading cases (~300-2,000 viral copies) and of the same order as influenza A. Consistent with the recent results of Goyal et al , these results suggest that viral loads relevant to infection from presymptomatic or mildly symptomatic individuals may fall into a narrow range, and that exceptionally high viral loads are not required to induce a superspreading event [1,2] . Rather, t he accumulation of infective aerosols exhaled by a typical pre-symptomatic or mildly symptomatic patient in a confined, crowdedspace (amplified by poor ventilation, particularly activity like exercise or singing, or lack of masks) for exposure times as short as one hour are sufficient. We calculate that talking and breathing release ~460 N 0 and ~10 N 0 (quanta)/hour, respectively, providing a basis to estimate the risks of everyday activities. Finally, we provide a calculation which motivates the observation that fomites appear to account for a smallpercentage of total COVID-19 infection events.

We probed the transmission of COVID-19 by applying an airborne transmission model to five well-documented case studies—a Washington state church choir, a Korean call center, a Korean exercise class, and two different Chinese bus trips. For all events the likely index patients were pre-symptomatic or mildly symptomatic, which is when infective patients are most likely to interact with large groups of people. Applying the model to those events yields results that suggest the following: (1) transmission was airborne; (2) superspreading events do not require an index patient with an unusually high viral load; (3) the viral loads for all of the index patients were of the same order of magnitude and consistent with experimentally measured values for patients at the onset of symptoms, even though viral loads across the population vary by a factor of >108. In particular we used a Wells-Riley exposure model to calculate q, the total average number of infectious quanta inhaled by a person at the event. Given the q value for each event, the simple airborne transmission model was used to determined Sq, the rate at which the index patient exhaled infectious quanta and N0, the characteristic number of COVID-19 virions needed to induce infection. Despite the uncertainties in the values of some parameters of the superspreading events, all five events yielded (N0∼300–2,000 virions), which is similar to published values for influenza. Finally, this work describes the conditions under which similar methods can provide actionable information on the transmission of other viruses.