Unifying the effective reproduction number, incidence, and prevalence under a stochastic age-dependent branching process
Renewal equations are a popular approach used in modelling the number of new infections (incidence) in an outbreak. A unified set of renewal equations where incidence, prevalence and cumulative incidence can all be recovered from the same stochastic process has not been attempted. Here, we derive a set of renewal equations from an age-dependent branching process with a time-varying reproduction number. Our new derivation utilises a measure-theoretic approach and yields a fully self-contained mathematical exposition. We find that the renewal equations commonly used in epidemiology for modelling incidence are an equivalent special case of our equations. We show that these our equations are internally consistent in the sense that they can be separately linked under the common back calculation approach between prevalence and incidence. We introduce a computationally efficient discretisation scheme to solve these renewal equations, and this algorithm is highly parallelisable as it relies on row sums and elementwise multiplication. Finally we present a simple simulation example in the probabilistic programming language Stan where we jointly fit incidence and prevalence under a single time-varying reproduction number and generation interval.