Spatial Disparities in Potential Access to Food Environments in Rural Texas

Abstract

PP-31-177 Background/Aims: Spatial disparities in access to food environments have emerged as important issues in public health. Of particular interest is further methodological improvement of the measures that incorporate spatial nature of the data. Spatially correlated access measures and/or indicators of neighborhood characteristics have not been adequately addressed in previous studies of food access. In this study, we explored the use of multiple spatial analysis methods that consider the spatial nature/limitations. Methods: We used data from the Brazos Valley Food Environment Project, which included direct observation and on-site GPS identification of all traditional and nontraditional food stores and comprehensive in-store observational survey of the presence and variety of fruits and vegetables in all food stores in 6 rural counties. GIS-based exploratory spatial analysis methods, including global and local measures of spatial autocorrelation (LISA), and kernel density estimation method, were used to examine spatial disparities in potential access, based on measures including distance, coverage, and the ratio of food stores to population. Results: Overall, we observed both global and local associations in access measures; spatial data are frequently characterized by positive spatial autocorrelation, and we found evidence of highly positive spatial autocorrelation across a set of food stores with different types and varying options of food availability. Kernel density surfaces depicting variations in potential access, adjusting for population density, were used to compare local associations in access measures, and LISA maps to identify those geographic areas with better or poor potential access. Results also indicate that access measures were associated with neighborhood characteristics, thus control of spatially correlated measures may be necessary in estimating the relationships between access and neighborhood characteristics. Conclusion: To reduce the potential bias and errors associated with spatial characteristics of neighborhood and food environment data, this study illustrates the need for methodological considerations in further ecologic studies of spatial disparities in potential access to food environments.