The detection of areas in which the risk of a particular disease is significantly elevated, leading to an excess of cases, is an important enterprise in spatial epidemiology. A number of frequentist approaches have been suggested for the detection of clusters, based on a hypothesis testing framework. Unfortunately, these suffer from a number of drawbacks including the difficulty in specifying a p-value threshold at which to call significance, the inherent multiplicity problem, and the possibility of multiple clusters. In this talk I will describe a Bayesian approach in which the study region is partitioned into, possibly multiple, ``zones'' within which the risk is either at a null, or non-null, level. Computation is carried out using Markov chain Monte Carlo, tuned to the cluster model. The method is applied to leukemia data in Upstate New York State, and Washington SEER data.
This is joint work with Albert Kim.
