USING A GIS TO DESIGN A CANCER EPIDEMIOLOGY STUDY

描述（申请人描述）拟建项目的目标是

DESCRIPTION (Applicant's Description) The goal of the proposed project is to pilot test a new technique for designing epidemiologic studies of spatially clustered rare exposures and/or rate diseases. Typically, studies of rare exposures and/or rare diseases identify a cohort with a specific exposure and either use a follow up or a nested case-control approach. For spatially clustered, rare exposures, such as proximity to point or small area sources (e.g., hazardous waste sites, factories) and line sources (e.g., power lines, roadways, rivers), this can be extremely difficult. In this project, we will determine if a Geographic Information System (GIS) can be used effectively to define such a population, estimate the population's demographics and quantify its exposure. To show the utility of the methods, we will apply quality assurance and validation procedures. Results from these procedures should enable other researchers to assess the adequacy of these and similar data and methods for their own investigations. Given the reluctance of many researchers to use new, untested and invalidated methodology, we view these efforts as potentially providing substantial encouragement for the use of new GIS tools in cancer epidemiology. Finally, we will use the results to conduct a preliminary ecologic study. The substantive public health question we will investigate is the reported association between residential exposure to magnetic fields from overhead high voltage electric power transmission lines and the incidence of childhood leukemia. Previous studies have investigated populations with exposures toward the low end of the observed range of exposures. By using the GIS to enrich the study population with heavily exposed subjects, we expect to increase the exposure gradient leading to increased effect sizes and increased statistical power. The main strength of this proposed study is that it will develop and validate new computerized methods for the identification and characterization of populations for epidemiologic investigations. These methods will be far more cost effective than traditional, in-the-field assessments in that then will use existing databases encompassing populations over a much larger geographic region. We believe that our study has a high likelihood of success because much of the methodology has been developed and tested on a pilot basis, although generalizations from pilot to full scale will likely require substantial modifications.