Multi-level modeling of early COVID-19 epidemic dynamics in French regions and estimation of the lockdown impact on infection rate

Abstract

AbstractWe developed a multi-level model of the French COVID-19 epidemic at the regional level. We rely on a global extended Susceptible-Exposed-Infectious-Recovered (SEIR) mechanistic model as a simplified representation of the average epidemic process, with the addition of region specific random effects. Combining several French public datasets on the early dynamics of the epidemic, we estimate region-specific key parameters conditionally on this mechanistic model through Stochastic Approximation Expectation Maximization (SAEM) optimization usingMonolixsoftware. We thus estimate the basic reproductive numbers by region before lockdown (with a national average at 2.81 with 95% Confidence Interval [2.58; 3.07]), attack rates (i.e. percentages of infected people) over time per region which range between 1.9% and 9.9% as of May 11th, 2020, and the impact of nationwide lockdown on the infection rate which decreased the transmission rate by 76% towards reproductive numbers ranging from 0.63 to 0.73 at the end of lockdown across regions. These results confirm the low population immunity, the strong effect of the lockdown on the dynamics of the epidemics and the need for further intervention when lifting the lockdown to avoid an epidemic rebound.