Methodological problems and the role of statistics in cluster response studies: A framework

More and more citizens urge public health authorities to investigate report s of disease excess in their neighbourhood. These environmental concerns ar e legitimate and it is part of good public health practice to respond to th ese complaints. However, the methodological and practical problems are seve re and a lot of controversy exists about the usefulness of these investigat ions. To clarify the possibilities and limitations in this situation, this paper proposes a typology of cluster studies. According to this framework, cluster response is distinguished from two other types of cluster studies: Cluster monitoring, screening proactively for clusters to act as an early w arning system, and cluster research, scrutinizing clustering to generate an d test aetiological hypotheses. To each of these three types of cluster stu dies corresponds a different public health context; respectively public hea lth action, public health surveillance and public health research. Probably , part of the controversy mentioned stems from not acknowledging sufficient ly the corresponding intrinsic differences in rationality and practical con straints. Cluster response is crisis management and not scientific research . In a relatively short time, an informed decision should be taken by a mul tidisciplinary team of experts using readily available information and know ledge. In accordance with this point of view, cluster reports should be han dled stepwise and the role of statistics is to quantify a cluster exploring different points of view as an input to the decision process.