The four county study of childhood cancer: clusters in context.

Observations of childhood 'cancer clusters' in small communities in central California prompted us to examine the distribution of childhood cancer in communities throughout the region to see if the overall cancer rate or the distribution of 'cancer clusters' was unusual for agricultural towns where pesticide exposure might be elevated. The distribution of rates was evaluated using a variety of methods: comparison of rates to the regional average, evaluation of the empirical observed versus expected Poisson distribution of events, and multivariate modelling using Poisson regression. These analyses suggest that there were no previously undiscovered communities with excess rates, although the index community which prompted the initial investigation does stand out as unusual. We discuss the impact of a range of forces of morbidity on the likelihood of 'cancer clusters' and the distributions of observed and expected numbers of cancers in a population of locales.