Spatial Clustering and Autocorrelation of Health Events

Abstract

Since the mid-nineteenth century, scientists have sought ways to quantify observed spatial patterns of disease incidence and prevalence in order to identify clusters of high risk. We review popular methods for identifying clusters and clustering of disease in geographically referenced epidemiologic data. We identify the questions of interest and illustrate how the combination available data and the choice of analytic method often answer a more specific question, i.e., each method tends to focus on specific types, shapes, and scales of clusters and clustering. Recognizing the specification implicit in the choice of data and method provides a critical context for interpreting the results of a spatial