Effects of latency on estimates of the COVID-19 replication number

It is not currently known how long it takes a person infected by the COVID-19 virus to become infectious. Models of the spread of COVID-19 use very different lengths for this latency period, leading to very different estimates of the replication number R , even when models work from the same underlying data sets. In this paper we quantify how much varying the length of the latency period affects estimates of R , and thus the fraction of the population that is predicted to be infected in the first wave of the pandemic. This variation underscores the uncertainty in our understanding of R and raises the possibility that R may be considerably greater than has been assumed by those shaping public policy.