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

Lucas Böttcher; Mingtao Xia; Tom Chou

In Fig. 3 of the main text, we observe that the population-level mortality ratio M 0 p (t) approaches a plateau during the initial exponential growth phase of an epidemic (i.e., for S(t) ≈ S 0 ). If the number of new infections decreases (e.g., due to quarantine measures), M 0 p (t) starts growing until it reaches its asymptotic value M 0 p (∞). Interestingly, the pre-asymptotic values of M 0 p (t) are smaller for larger infection rates β 1 (see Fig. S5(a) ). This counter-intuitive effect arises because larger values of β 1 generate relatively larger numbers of new infected which have a lower chance of dying before τ inc (see Eq. (4) in the main text). A similar effect occurs for non-delayed transmission (i.e., τ β ≈ 0). We observe a similar effect for non-delayed transmissions (i.e., τ β ≈ 0). As long as S(t) ≈ S0, smaller transmission delays τ β lead to larger relative numbers of new infections and smaller M 0 p (t).