Estimating the real-time case fatality rate of COVID-19 using Poisson mixtures model

Paul Hong Lee

The case fatality rate (CFR) of a disease refers to the proportion of death within those infected from the disease. The CFR of an emerging infectious disease can only be determined after its outbreak, but the real-time CFR estimation during the disease outbreak, especially in the early stage, has to be made for public health decisions. There are two naive, simple methods to estimate CFR at a time point t using the cumulative number of deaths, recoveries, and confirmed cases at t. Denote D(t), R(t), and C(t) be the observed number of deaths, recoveries, and confirmed cases at day t. The first simple estimator equals D(t)/C(t) and the second simple estimator equals D(t)/[D(t)+R(t)]. Note that these two estimators are equivalent after the outbreak when D(t)+R(t)=C(t) [1] .