Abstract
Rift Valley fever (RVF) is a vector-borne viral disease of major animal and public health importance. In 2018-19, it caused an epidemic in both livestock and human populations of the island of Mayotte. Using Bayesian modelling approaches, we assessed the spatio-temporal pattern of RVF virus (RVFV) infection in livestock and human populations across the island, and factors shaping it. First, we assessed if (i) livestock movements, (ii) spatial proximity from communes with infected animals, and (iii) livestock density were associated with the temporal sequence of RVFV introduction into Mayotte communes' livestock populations. Second, we assessed whether the rate of human infection was associated with (a) spatial proximity from and (b) livestock density of communes with infected animals. Our analyses showed that the temporal sequence of RVFV introduction into communes' livestock populations was associated with livestock movements and spatial proximity from communes with infected animals, with livestock movements being associated with the best model fit. Moreover, the pattern of human cases was associated with their spatial proximity from communes with infected animals, with the risk of human infection sharply increasing if livestock in the same or close communes were infected. This study highlights the importance of understanding livestock movement networks in informing the design of risk-based RVF surveillance programs.