Spatio-temporal modeling of sudden infant death syndrome data

Sudden infant death syndrome (SIDS) is a classification of death for apparently healthy infants under one year old. However, its etiology is still largely a mystery. In this research, we analyze a spatio-temporal data set that contains yearly SIDS information from 1979 to 1984 for the counties of North Carolina. Cressie and Chan (1989) [10] used a purely spatial model to analyze the aggregated version of this data set. In this article, we present a spatio-temporal model from which optimal smoothing of SIDS rates can be derived. We use a Bayesian hierarchical statistical model (BHM) with a hidden dynamical Markov random field and extra-Poisson variability. Potential confounding of sources of variability is avoided by calibrating the extra-Poisson variability with the microscale variation in an approximate Gaussian model.