How accurate are effective reproduction number estimates for mosquito-borne diseases in urban settings?

The time-varying effective reproduction number, Rt, characterizes the dynamics of infectious disease transmission at a given point in time. For mosquito-borne diseases, there has been growing interest in estimating Rt for spatially targeted vector control interventions. However, systematic evaluations of Rt estimation accuracy for mosquito-borne diseases are lacking: a concerning phenomenon given that errors in Rt estimation at fine spatial scales could be amplified by human mobility-driven transmission and high rates of asymptomatic cases. We systematically evaluated the accuracy of popular Rt estimation tools, EpiEstim and EpiFilter, against ground truth transmission pair data generated by agent-based simulations of dengue epidemics involving humans and mosquitoes in Singapore. We applied data collection artefacts to simulate real-world operational challenges, including spatial and temporal aggregation of incidence data, under-reporting of cases and reporting delays. The findings suggest that EpiEstim and EpiFilter are not well-suited for estimating Rt for mosquito-borne diseases at fine spatial scales because they assume localized disease transmission, whereas human mobility is a significant driver of transmission across wider geographic scales. Consequently, public health officials should be cautious when interpreting Rt estimates at fine spatial scales. Our findings highlight an urgent need for innovations in Rt estimation tools for mosquito-borne diseases.