ON THE UNCERTAINTY ABOUT HERD IMMUNITY LEVELS REQUIRED TO STOP COVID-19 EPIDEMICS

COVID-19 evolved into a pandemic in 2020 affecting more than 150 countries. Given the absence of a vaccine, discussion has taken place on the strategy of allowing the virus to spread in a population, to increase population “herd immunity”. Knowledge of the minimum proportion of a population required to have recovered from COVID-19 infection in order to attain “herd” immunity, P _crit , is important for formulating epidemiological policy. A method for measuring uncertainty about P _crit based on a widely used package, EpiEstim, is derived. The procedure is illustrated using data from twelve countries at two early times during the COVID-19 epidemic. It is shown that simple plug-in measures of confidence on estimates of P _crit are misleading, but that a full characterization of statistical uncertainty can be derived from EpiEstim, which reports percentiles only. Because of the important levels of uncertainty, it is risky to design epidemiological policy based on guidance provided by a single point estimate.