Estimation of unreported novel coronavirus (Sars-coV-2) infections from reported deaths: A susceptible–exposed–infectious–recovered–dead model

Abstract

In the midst of the novel coronavirus (SARS-CoV-2) epidemic, examining reported case data could lead to biased speculations and conclusions. Indeed, estimation of unreported infections is crucial for a better understanding of the current emergency in China and in other countries. In this study, we aimed to estimate the unreported number of infections in China prior to the 23 January 2020 restrictions. To do this, we developed a Susceptible–Exposed–Infectious–Recovered–Dead (SEIRD) model that estimated unreported infections from the reported number of deaths. Our approach relied on the fact that observed deaths were less likely to be affected by ascertainment biases than reported infections. Interestingly, we estimated that the basic reproductive number (R0) was 2.43 (95%CI = 2.42–2.44) at the beginning of the epidemic and that 92.9% (95%CI = 92.5%–93.1%) of total cases were not reported. Similarly, the proportion of unreported new infections by day ranged from 52.1% to 100%, with a total of 91.8% (95%CI = 91.6%–92.1%) of infections going unreported. Agreement between our estimates and those from previous studies proves that our approach is reliable for estimating the prevalence and incidence of undocumented SARS-CoV-2 infections. Once it has been tested on Chinese data, our model could be applied to other countries with different surveillance and testing policies.