Estimating the contagiousness ratio between two viral strains

Abstract

SummaryWe propose a new method to estimate the ratio between the basic reproduction numbers of a newly emerged variant and the one currently dominating. We use public data of two kinds: the proportions of the daily infected from each strain obtained from a random sample that has been sequenced, and the epidemic curves of total infections and recoveries. Our method is based on a new discrete-time SIR model with two strains, considered both in the deterministic and stochastic versions. In the deterministic case we use maximum likelihood, while in the stochastic setting, since we need to reconstruct the missing incidence and prevalence information of the new variant, we decided to use a hierarchical Bayesian hidden Markov model. This new methodology is applied to data from the Piedmont Italian region in December-January 2022, when the Omicron variant started to be observed and quickly became prevalent. With both approaches, we obtain an estimate of the contagiousness ratio that is consistent with other studies specifically designed for the aim.