Evaluating the effectiveness of social distancing interventions against COVID-19

Laura Matrajt; Tiffany Leung

Our work has several limitations and has to be interpreted accordingly. First, deterministic mathematical models tend to overestimate the final size of an epidemic. Further, deterministic models will always predict a rebound in the epidemic once the intervention is lifted provided that the number of infected people (exposed or infectious) is positive. To avoid that problem, we forced our infected compartments to be zero if they had less than one person infected at any given time. Second, we considered the latent period to be equal to the incubation period, but it has been suggested that pre-symptomatic transmission is occurring [22] , and the virus SARS-CoV-2 is shed for a prolonged time after symptoms end [33] . It is unclear at this point however, if virus shed by convalescent individuals can infect others. Further, we considered that mild and severe cases would be equally infectious. In this sense, our results are conservative. We used mortality and hospitalizations data from China to assess hospitalizations and deaths rates, but these rates will presumably be country-dependent. New information about the epidemiological characteristics of SARS-CoV-2 is continuously arising, and incorporating such information in mathematical models such as ours is key to provide public health officials with the best tools to make decisions in uncertain times. Taken together, our results suggest that more aggressive approaches need to be taken to mitigate the transmission of SARS-CoV-2. Social distancing interventions need to occur in tandem with testing and contact tracing to minimize the burden of COVID-19.