CLUSTER - A SOFTWARE SYSTEM FOR EPIDEMIOLOGIC CLUSTER-ANALYSIS

A software package for cluster analysis was developed that incorporate s 12 methods to analyse residence and time data, and morbidity or mort ality rates. The Knox, Barton, Chen and Ohno methods of the software a re illustrated with data for 123 cancer cases reported by a citizen re questing a cluster investigation. Longitude and latitude for each addr ess were assigned by a geographic information system for analyses with the Barton (p > 0.05) and Knox methods (p > 0.05). The data were also analysed for time of the event with the Chen method, comparing the ob served incidence to expected county disease rates (p = 0.61), and for temporal patterns within the geographically defined population with th e Ohno method (p > 0.05). It was concluded that no statistically signi ficant disease cluster existed.