UC Irvine

Spatial methods can be used to assess environmental contaminants and analyze their effect on pregnancy outcomes. When considering space, current methods in selecting controls in case-control studies may not be adequate in selecting a geographically diverse sample of participants, particularly when the exposure varies across a study area. Unmatched spatially stratified random sampling (SSRS) designs may improve study efficiency compared to simple random samples (SRS) in studies that require controls across the study area. Typically, SRS selects controls from data dense areas, and may not sufficiently select controls along the edges or where environmental exposures may occur. In a simulation study, SSRS selected controls from evenly sized strata was found to have lower mean squared error across simulations and along edges of the study area where SRS naturally may not select controls. The SSRS method was applied to a case-control study of Texas birth defects from 1999 to 2011 examining the relationship between exposure to unconventional natural gas developments (UNGD) and birth defects. Birth defects are multifaceted diseases that can affect multiple organs throughout the body and their etiology are poorly understood. In-utero exposure to environmental contaminants, such as UNGD, may contribute to risk of birth defects in Texas, where UNGDs are spatially distributed in shales across the state. Exposure was a density measure evaluated as the number of UNGD wells within 1, 3, and 7.5km of maternal address for congenital heart defects (CHD), neural tube defects (NTD), orofacial defects, and gastroschisis. UNGD exposure was associated with some CHDs, NTDs, and some gastroschisis, particularly within 1km of maternal address. Spatial methods were also used to investigate birth defect risk factors in nested models to produce maps indicating high and low risk. Birth defects were significantly associated with maternal address location, and high-risk areas substantially decreased after adjusting for maternal, environmental and community variables. However, birth defect risk remained high in areas around North and South Texas and warrant further investigation of potential chemical exposures. This research has demonstrated the effective use of a novel control selection method to investigate spatial risk factors. Additionally, despite limitations in the research, exposure to UNGD chemicals may increase the risk of birth defects and more studies are needed to understand and prevent birth defects.