Reconstructing the course of the COVID-19 epidemic over 2020 for US states and counties: results of a Bayesian evidence synthesis model

Chitwood, Melanie H.; Russi, Marcus; Gunasekera, Kenneth; Havumaki, Joshua; Pitzer, Virginia E.; Salomon, Joshua A.; Swartwood, Nicole; Warren, Joshua L.; Weinberger, Daniel M.; Cohen, Ted; Menzies, Nicolas A.

Estimating the true magnitude of the United States (US) SARS-CoV-2 epidemic is crucial for understanding disease dynamics and, ultimately, for determining the effectiveness of interventions intended to interrupt transmission. We developed a Bayesian evidence synthesis model that explicitly accounts for reporting delays and secular variation in case ascertainment to generate estimates of incident COVID-19 infections on the basis of reported cases and deaths. We estimate time trends in COVID-19 epidemiology for every US state and county, from the first reported case (January 13, 2020) through January 1, 2021. Across counties, we estimate considerable variability in the level and pattern of incidence, producing major differences in the estimated proportion of the population infected by the end of 2020. Our estimates of COVID-19 deaths are consistent with independent estimates of excess mortality, and our estimates of cumulative incidence of infection are consistent with seroprevalence estimates from available antibody testing studies.