Evaluating the effectiveness of social distancing interventions against COVID-19

Laura Matrajt; Tiffany Leung

To quantify the uncertainty around our results, we performed 1,000 simulations where we varied three parameters: the basic reproduction number R 0 , the latent period and the duration of infectiousness while keeping the other parameters fixed. Based on current estimates [1, 22, 23, 24] , we sampled values of R 0 from a truncated normal distribution with mean 2.26 ranging from 1.6 to 3. We assumed that the latent period was similar to the incubation period, and assumed a gamma distribution with mean 5.1 days and standard deviation 0.7 [17] . We sampled the duration of infectiousness from a truncated normal distribution with mean 5 days and standard deviation 0.7 ranging from 3 to 9 [18, 25] . For each R 0 , infectious period and latent period considered, we used the next-generation matrix approach [26] to calculate the transmission coefficient β. For each statistic presented in the results, we computed the error bars by removing the top and bottom 2.5% of the simulations.