Food swamp vs. food desert: Analysis of geographic disparities in obesity and diabetes in North Carolina using GIS and spatial regression

Abstract

This study examined how two different types of food environments––food deserts and food swamps––were associated with geographic disparities in obesity and diabetes in North Carolina using geographic information systems and spatial regression. To better identify the association between food swamps and health disparities, this study incorporated socioeconomic dimensions into the food swamp measure and considered dollar stores and gas stations with convenience stores as unhealthy food retailers, which were often overlooked in previous studies. It found that food deserts were concentrated in eastern North Carolina (ENC) and western North Carolina and food swamps in ENC, while obesity/diabetes hot spots were clustered in ENC. The results indicated that obesity/diabetes remained associated with food swamps even after spatial autocorrelation was controlled for, but the associations with food deserts––statistically significant in a nonspatial regression––became no longer significant when food swamps and spatial autocorrelation were controlled for. These results demonstrate that a food swamp may better explain health disparities in North Carolina than a food desert. Identifying ENC as a region containing both food deserts and food swamps illustrates the need for state and local governments to focus their efforts in this region to mitigate food insecurity and health disparities.