Evolution of spatial disease clusters via a Bayesian space-time variability modelling.

This study proposes to use exceedance posterior probabilities of a space-time random-effects model to study the temporal dynamics of clusters. The local time trends specified for each area is further smoothed over space. We modelled the common spatial and the space-varying temporal trend using a multivariate Markov Random field to incorporate within-area correlations. We estimate the model parameters within a fully Bayesian framework. The exceedance posterior probabilities are further used to classify the common spatial trend into hot-spots, cold-spots, and neutral-spots. The local time trends are classified into increasing, decreasing, and stable trends. The results is a 3×3 table depicting the time trends within clusters. As a demonstration, we apply the proposed methodology to study the evolution of spatial clustering of intestinal parasite infections in Ghana. We find the methodology presented in this paper applicable and extendable to other or multiple tropical diseases which may have different space-time conceptualizations.