| Public health investigators are often called upon to address community concerns over emerging patterns or "clusters" of disease in local areas. Traditional epidemiological study designs are labor intensive and often inadequate for investigations of small local areas.; Under an ATSDR grant for the development of methods for small areas (ATPM TS1067), I explored a simple but realistic situation: a single-incident exposure of a population (e.g. railway tanker accident) resulting in an acute disease outbreak. I developed study methods using software available to community investigators, including GIS technology to facilitate spatial data handling and visualization.; This spatial study design uses matched analysis of actual case or health event data paired with counterfactual referent data. The study method compares each case's observed exposure to a counterfactual referent exposure derived from the exposure prevalence in that case's local area. Simulation testing under highly controlled theoretical conditions and with realistic census-based populations demonstrates that this spatial study design controls or mitigates several types of spatial confounding present in many real-world study situations.; I conducted field testing with two very different scenarios. The first involved hydrogen sulfide plumes from a large slaughterhouse waste pool in Nebraska, previously linked to rises in hospital visits for respiratory distress. Feasibility investigation involved all the steps necessary to set up a field study. The second scenario analyzed self reports of smoke exposure from New York City area women who were pregnant at the time of the World Trade Center attack. Maps of respondent locations and responses were compared with a model for smoke emanating from the site in the first few days after the attack.; Data analyzed using three levels of geographic areas for referent exposure assignment (full area, zip code, and neighborhood) showed divergent results, from a null finding at the zip code level, to a matched odds ratio (MOR) of 18.7 for the full study area referents. After rigorous simulation testing, pilot study, and field study, the generalized matched spatial study methods developed in this dissertation add to existing options for the study of disease clusters and environmental exposures. |
