Health authorities are often alerted to suspected cancer clusters near
the vicinity of potential point sources by members of the public. A s
urveillance system, where administrative regions around the potential
point sources are regularly monitored for high disease rates, would al
low for responses which are easier to obtain, timelier, and less expen
sive than individual thorough investigations. The monitoring could be
done by using the so-called 'focused' tests for detecting disease clus
tering. However, these tests, generally designed to detect clusters of
a fixed size around the foci are not particularly effective when deal
ing with administrative regions with substantial differences in popula
tions. In this work, an approach which overcomes the problem to a cert
ain extent is described Here the selected cluster sizes are based on t
he populations of the administrative regions under examination. The ap
proach is used to investigate whether cancer clustering appears in the
vicinity of the pulp and paper mills in British Columbia for the year
s 1983-1989. The results indicate that the approach performs reasonabl
y well in identifying cancer sites for which elevated risks have also
been suggested in the epidemiologic literature. Consequently, this met
hodology could be utilized to provide guidance for further investigati
on even in the absence of local reports. Similarly, it could be readil
y utilized to provide timely responses to local reports.