Abstract
Yellow fever (YF), a vector-borne viral hemorrhagic fever, is endemic in tropical regions of Africa and South America, with large vaccination programmes being used for control. However, significant outbreaks have occurred in recent years. Data on infection rates and seroprevalence is often sparse, requiring robust mathematical models to estimate the burden of yellow fever. In particular, modelling is required to estimate the risk of outbreaks and inform policy decisions regarding the targeting of vaccination. We present a dynamic, stochastic model of YF transmission which uses environmental covariates to estimate the force of infection due to spillover from the sylvatic (non-human primate) reservoir and the basic reproduction number for human-to-human transmission. We examine the potential for targets identified by the World Health Organization EYE Strategy (50%, 60% or 80% vaccination coverage in 1-60 year olds) to achieve different threshold values for the effective reproduction number. Threshold values are chosen to reflect the potential for seasonal and/or climatic variation in YF transmission even in a scenario where vaccination lowers the median reproduction number below 1. Based on parameter estimates derived from epidemiological data, it is found that the 2022 EYE Strategy target coverage is sufficient to reduce the static averaged annual effective reproduction number R below 1 across most or all regions in Africa depending on the effectiveness of reported vaccinations, but insufficient to reduce it below 0.5 and thereby eliminate outbreaks in areas with high seasonal range. Coverage levels aligned with the 2026 targets are found to significantly decrease the proportion of regions where R is greater than 0.5.