An examination of three spatial disease clustering methodologies.

Patterns of disease in space are often analysed to determine whether a relationship exists between a disease outcome and environmental exposures. This report examines the performance of three cluster analytical methods when applied to a single data set. These methods, designed to assess the purely spatial variation of events, have been examined to assess their ability to detect clustering in an area where disease rates have previously been shown to be significantly elevated. The ability of these methods to detect spatial clustering was also examined using simulation techniques. All three methods were found to be poor at detecting spatially localized disease rates which were approximately three time the expected rate, as measured by the relative risk.