Format

Send to

Choose Destination
Biometrics. 2006 Jun;62(2):465-70.

Spatial event cluster detection using a compound Poisson distribution.

Author information

1
Department of Pediatrics, University of Alberta, Edmonton, Alberta T6G 2J3, Canada. rhonda.rosychuk@ualberta.ca

Abstract

Geographic disease surveillance methods identify regions that have higher disease rates than expected. These approaches are generally applied to incident or prevalent cases of disease. In some contexts, disease-related events rather than individuals are the appropriate units of analysis for geographic surveillance. We propose a compound Poisson approach that detects event clusters by testing individual areas that may be combined with their nearest neighbors. The method is applicable to situations where the population sizes are diverse and the population distribution by important strata may differ by area. For example, a geographical region might have sparse population in the northern areas, and other areas which are predominantly retirement communities. The approach requires a coarse geographical relationship and administrative data for the numbers of population, cases, and events in each area. Pediatric self-inflicted injuries requiring presentation to Alberta emergency departments provide an illustration.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Wiley
Loading ...
Support Center