Investigating causation in cancer clusters.

Questions concerning clusters of cancer cases frequently arise in public health practice. The process of investigating any such cluster requires awareness that such case groupings can easily occur by chance and that any search for biologically meaningful causes will be severely constrained by various methodologic difficulties. These include (1) the long and probably variable latent periods between causative events and cancer diagnosis, (2) the limited numbers of cases available for study in any given cluster situation, and (3) the clinical non-specificity of cancer cases whereby no readily available means are at hand to identify the specific causes for any particular case. Evaluation of any given cluster should involve careful consideration of such limitations, together with a preliminary assessment of the specific cases involved and their community or workplace setting, before more intensive study is undertaken.