Hierarchical Bayesian modeling of the space‐time diffusion patterns of cholera epidemic in Kumasi, Ghana

Abstract

This study analyses the joint effects of the two transmission routes of cholera on the space‐time diffusion dynamics. Statistical models are developed and presented to investigate the transmission network routes of cholera diffusion. A hierarchical Bayesian modelling approach is employed for a joint analysis of nonlinear effects of continuous covariates, spatially structured variation, and unstructured heterogeneity. Proximity to primary case locations and population density serve as continuous covariates. Reference to communities is modelled as a spatial effect. The study applied to the Kumasi area in Ghana shows that communities proximal to primary case locations are infected relatively early during the epidemics, with more remote communities infected at later dates. Similarly, more populous communities are infected relatively early and less populous communities at later dates. The rate of infection increases almost linearly with population density. A non systematic relation occurs between the rate of infection and proximity to primary case locations. It is discussed how these findings could serve as significant information to help health planners and policy makers in making effective decisions to limit cholera epidemics.