A stochastic epidemiological model to estimate the size of an outbreak at the first case identification.

The identification of a first case (e.g. by a disease-related death or hospitalization event) raises the question of the actual size of a local outbreak. Quick estimates of the outbreak size are required to assess the necessary testing, contact tracing and potential containment effort. Using general branching processes and assuming that epidemic parameters (including the basic reproductive number) are constant over time, we characterize the distribution of the first hospitalization time and of the epidemic size at this random time. We find that previous estimates either overestimate or largely underestimate the actual epidemic size. In addition, we provide upper and lower bounds for the number of infectious individuals of the local outbreak over time. The upper bound is the cumulative epidemic size, and the lower bound is a constant fraction of it. Lastly, we compute the number of detectable cases if one were to test the whole local outbreak at a single point in time. In a growing epidemic, most individuals have been infected recently, which can strongly limit the detection of infected individuals when there is a delay between an infection and its potential detection. Overall, our analysis provides new analytical estimates about the epidemic size at identification of a first disease-related case. This piece of information is important to inform policy makers during the early stages of an epidemic outbreak.