Detecting space-time cancer clusters using residential histories

Methods for analyzing geographic clusters of disease typically ignore the space-time variability inherent in epidemiologic datasets, do not adequately account for known risk factors (e.g., smoking and education) or covariates (e.g., age, gender, and race), and do not permit investigation of the latency window between exposure and disease. Our research group recently developed Q-statistics for evaluating space-time clustering in cancer case-control studies with residential histories. This technique relies on time-dependent nearest neighbor relationships to examine clustering at any moment in the life-course of the residential histories of cases relative to that of controls. In addition, in place of the widely used null hypothesis of spatial randomness, each individual's probability of being a case is instead based on his/her risk factors and covariates. Case-control clusters will be presented using residential histories of 220 bladder cancer cases and 440 controls in Michigan. In preliminary analyses of this dataset, smoking, age, gender, race and education were sufficient to explain the majority of the clustering of residential histories of the cases. Clusters of unexplained risk, however, were identified surrounding the business address histories of 10 industries that emit known or suspected bladder cancer carcinogens. The clustering of 5 of these industries began in the 1970's and persisted through the 1990's. This systematic approach for evaluating space-time clustering has the potential to generate novel hypotheses about environmental risk factors. These methods may be extended to detect differences in space-time patterns of any two groups of people, making them valuable for security intelligence and surveillance operations.