Why estimating population-based case fatality rates during epidemics may be misleading

Lucas Böttcher; Mingtao Xia; Tom Chou

FIG. S4. Fractional testing. An example of fractional testing in which a fixed fraction f of the real total infected population is assumed to be tested. The remaining 1 − f proportion of infecteds are untested. Equivalently, if the total tested fraction has unit population, then the total population of the untested pool is 1/f − 1. (a) At short times after an outbreak, the known tested infected population has not yet resolved and is composed of deaths (gray), recovereds (green), and infecteds (red). We assume that the untested fraction of infecteds (red) have mild or no symptoms, do not die, and can only recover (green). (b) At longer times, the infecteds further resolve. The Mp(t) and CFR metrics that are based on only the tested fraction will overestimate the true mortality fraction of all infected cases.