This body of work applies an interdisciplinary approach to methodological issues in environmental epidemiology research. Exposure assessment, precision, confounding and scale are addressed by integrating knowledge from the fields of environmental health sciences, epidemiology and spatial statistics. Environmental health sciences provide myriad tools for the improvement of exposure assessment in environmental epidemiology. Similar contributions from the rapidly growing field of spatial statistics have not yet been fully explored, particularly with regard to bridging environmental health science exposure measurement and epidemiology research design. Using studies of air pollution and asthma as an example of the wider field of environmental epidemiology, this work reviews recent approaches to exposure assessment and areas in which spatial statistics can provide methodological improvements. Precision and exposure assessment are addressed through a power evaluation of hypothesis tests and models to detect focused clustering, disease clustering associated with a known pollution source. Environmental health science tools describe the spatial dispersion of environmental contaminants. Human exposure, and thus disease, should follow these spatial dispersion patterns. The power evaluations in this work investigate the ability of focused cluster hypothesis tests and models to detect disease cluster shapes that are based on pollution dispersion principles. Findings show the importance of mathematical functions that reflect the particular pollution dispersion factors in a given cluster investigation. Finally, a study of the association of asthma with ozone, PM10 and PM2.5 is conducted as an applied example of methodologically innovative interdisciplinary research. Bayesian kriging and spatial prediction of environmental variables are used in this investigation. Data are analyzed with generalized linear mixed models that include random effects addressing individual frailty and geographic area. Not only is the feasibility of these techniques demonstrated but significant associations are evident between PM 10 and asthma ED admissions among children in South Carolina. In sum, this body of work provides evidence that the crossing of traditional disciplinary boundaries can advance methodological issues in environmental epidemiology research. Challenges include the time and effort involved in conducting studies that are interdisciplinary from design to completion. Yet, these collaborations can prove as rewarding to the researchers as to the quality of the research.
展开▼
