Understanding Disparities in Obesity Prevalence in Appalachia Region: A County-Level Analysis Using an Ecological Approach

Introduction: The Appalachian region has historically experienced economic deprivation. Economic deprivation is itself a factor often related to increased rates of obesity with attendant effects on health. Although studies have examined contextual influences on obesity prevalence, their influence on adult obesity prevalence in Appalachia remains unknown. This study examined, for all counties in the Appalachia region, the association between adult obesity prevalence and physical inactivity prevalence, poverty rate, the percentage of rural areas, the number of fast food restaurants, the percentage of African American population, and the percentage of non-Hispanic white population in Appalachia. 

Method: In this study, we included 420 counties based on the inclusion criteria from the Appalachian Regional Commission (ARC). We collected fast food restaurant geo-locations in Appalachia from Yelp website Application Programming Interface (API) and obtained the remaining county-level data from the Centers for Disease Control and Prevention and the U.S. Census Bureau. We then aggregated the number of fast food restaurants to county-level based on their geo-location. To examine bivariate relationships between each predictor and obesity prevalence, we estimated Moran’s I statistic, a measure of spatial autocorrelation, and applied a geographically weighted regression model using a spatial error term, which corrects for spatial dependence. Statistical and spatial analyses were performed in GeoDa. Mapping was performed in QGIS 2.10.1.

Results: The average age-adjusted obesity rate of Appalachian region counties was 33 % in 2012, among which Breathitt County, Kentucky was the highest (43%). The Moran’s I statistic indicated a positive spatial autocorrelation for the obesity prevalence by county-level (Moran’s I = .45). Significant bivariate predictors of the obesity prevalence were: poverty rate (r² = .35, β (SE) .10 (.62), p < .001), the percentage of rural areas (r² = .36, β (SE) =2.28 (.65), p < .001), the percentage of non-Hispanic white population (r² =35, β (SE) = -4.74 (1.8), p < .001), the percentage of African American population (r² = .36, β (SE) = 7.56 (1.96), p < .001), the physical inactivity prevalence (r² =45, β (SE) = .33, p < .001), and the number of fast food restaurants (r² = .35, β (SE) = -.25, p < .05).

Implication: This study highlighted the disparities in obesity prevalence in the Appalachian region by race, poverty level, and community. By highlighting patterns of obesity and searching for causal relationships throughout the region, we intend to provide useful data upon which fundable targeted health interventions and obesity prevention and reduction efforts may be based.