Measure of clusters

Clark W. Heath describes eight community clusters of leukemia and lymphoma, which are very suggestive for infectious causation of the disease. In these clusters malignities occured often in several children attending same school or parish or in children with siblings enrolled in a same school. All that usually, but not necessarily, in areas with recent population abruption.

I might be interesting investigate for such clusters in Czech Republic (general incidence is growing after socioeconomic transformation in 1989), but unfortunately, there isn't residence information in my data set, although it should be available with some effort. And are there such communities presenting considerable population increase or movement?

Nevertheless, we're considering a search for ALL clusters in a little bit larger measure - is it worth to look for season/annual clustering in general children population? For example, identifying a virus season association with ALL diagnosis might ease investigation of potential underlying infectious agents. But uncertain illness latency distorts the clustering... Tricky puzzle.