Cancer clusters: a myth or a method.

Historically, cluster reports have been considered to be only marginally productive and generally discouraged as statistical artifacts. This report presents the results of a three year study involving in excess of 100 cancer cluster reports. The majority of these investigations represented statistical events; there were, however, some exceptions. Several of these reports were the basis for larger and more productive studies. A survey of the types of reports received, their sources and outcomes is presented. Examples of productive and nonproductive studies are highlighted to indicate decision points for selecting the most promising reports. Investigators are encouraged to consider cancer cluster reports from lay individuals as well as from more authoritative sources (i.e., tumor registries, physicians, medical-social workers). Investigators are also urged to perform actual field investigations of those reports choosing statistically promising and/or provocative circumstances for the greatest likelihood of identifying new environmental risk factors for cancer.