Evaluating the Methodology and Clinical Utility of Spatial Access to Health Care Measures in Appalachia.

Abstract

The Appalachia region of the U.S. has noted socioeconomic disparities, elevated rates and mortality for many cancers, and substandard cancer treatment patterns. These disparities, combined with Appalachia’s largely rural population, suggest that the region has reduced access to health care. This research investigated the methodology behind spatial access to healthcare using population-based clinical data and geographic information systems (GIS) software. The dissertation’s goal was to provide a guide of the latest spatial access methods in Appalachia and to demonstrate how those methods can be incorporated into models studying cancer disparities in the region. Accredited mammography centers and primary care physicians in 2008 from Pennsylvania, Ohio, Kentucky, and North Carolina, along with U.S. Census population data, were geocoded into GIS software. Methods compared included ratios of mammography centers and physicians to county populations, travel time to closest mammography centers and physicians, and several versions of the newer two-step floating catchment area (2SFCA) method, which has never been evaluated in Appalachia. As a measure of predictive validity, spatial access methods were then used to predict two important breast cancer clinical indicators: stage at diagnosis and receipt of adjuvant hormonal therapy after a qualifying surgery. Urban and non-Appalachia areas had consistently shorter travel times than the rural and Appalachia areas of the same states, across both travel times to mammography centers and primary care physicians. The 2SFCA measures that included variable catchment sizes appeared distinct from the original 2SFCA method and 2SFCA methods that included distance decay weights but no variable catchments. Predictors of late-stage breast cancer diagnosis included age, insurance status, county primary care to population ratio, and primary care 2SFCA score. Geographically weighted logistic regression revealed that the effect of the predictor variables varied throughout the study region. Predictors of adjuvant hormonal therapy included the presence of multiple diseases, county economic status, and mammography center 2SFCA score. Overall, the 2SFCA method with variable catchment sizes offered the greatest predictive validity of the access measures and offers theoretical improvements over the other access to care measures. Nonetheless, further research is needed to validate the 2SFCA method parameters with patient healthcare utilization data.