Strategies for evaluating the environment-public health interaction of long-term latency disease: the quandary of the inconclusive case-control study

Abstract

Environmental links to disease are difficult to uncover because environmental exposures are variable in time and space, contaminants occur in complex mixtures, and many diseases have a long time delay between exposure and onset. Furthermore, individuals in a population have different activity patterns (e.g., hobbies, jobs, and interests), and different genetic susceptibilities to disease. As such, there are many potential confounding factors to obscure the reasons that one individual gets sick and another remains healthy. An important method for deducing environmental associations with disease outbreak is the retrospective case-control study wherein the affected and control subject cohorts are studied to see what is different about their previous exposure history. Despite success with infectious diseases (e.g., food poisoning, and flu), case-control studies of cancer clusters rarely have an unambiguous outcome. This is attributed to the complexity of disease progression and the long-term latency between exposure and disease onset. In this article, we consider strategies for investigating cancer clusters and make some observations for improving statistical power through broader non-parametric approaches wherein sub-populations (i.e., whole towns), rather than individuals, are treated as the cases and controls, and the associated cancer rates are treated as the dependent variable. We subsequently present some ecological data for tungsten and cobalt from studies by University of Arizona researchers who document elevated levels of tungsten and cobalt in Fallon, NV. These results serve as candidates for future hybrid ecologic case-control investigations of childhood leukemia clusters.