ABSTRACT
Exploring spatial-temporal patterns of disease incidence identifies areas of significantly elevated risk and can lead to discoveries of disease risk factors. One popular way to investigate patterns in risk over space and time is spatial-temporal cluster detection analysis. The identification of significant clusters could lead to etiological hypotheses to explain the pattern of elevated risk and to additional epidemiologic studies to explore these hypotheses. Several methodological issues and data challenges that arise in space–time cluster analysis of chronic diseases, such as cancer, include poor spatial precision of residence locations, long disease latencies, and adjustment for known risk factors. This article reviews the key challenges faced when performing cluster analyses of chronic diseases and presents a spatial-temporal analysis of non-Hodgkin lymphoma (NHL) risk addressing these challenges. Residential histories, collected as part of a population-based case-control study of NHL (the National Cancer Institute [NCI] Surveillance, Epidemiology, and End Results [SEER] NHL study) in four SEER centers (Detroit metropolitan area, Los Angeles, Seattle metropolitan area, and Iowa) were geocoded. In this analysis, we explored previously detected spatial-temporal clusters and adjusted for exposure to polychlorinated biphenyls (PCBs) and genetic polymorphisms in four genes, previously found to be associated with NHL, using a generalized additive model framework. We found that the genetic factors and PCB exposure did not fully explain previously detected areas of elevated risk.
