Association for Public Policy Analysis & Management

Little is known about the association between the retail food environments near schools and diet quality among adolescents. Understanding how the food environment surrounding schools is related to children’s diet and weight status is crucial to creating more effective interventions that reduce the increasing disparities in obesity prevalence among racial/ethnic minority and lower-income children in America. The main objective of this study is to explore the effect of food swamps (areas inundated with unhealthy food retailers) around middle schools on disparities in adolescents’ diet quality and weight status.

Our methodological approach includes three phases. First, we assessed 0.5-mile radius food environments surrounding 20 middle schools categorizing them as food swamps, food deserts, neither or both. To measure food swamps, we used 2016 National Establishment Time Series food store data to create a range of geospatial measures of local food environments including the Retail Food Environment Index (RFEI), the Modified Retail Food Environment Index (mRFEI), and population density measures. To measure food deserts, we extrapolated from the USDA’s criteria for food deserts which includes an area that is low income (poverty rate >20%) with limited access to a grocery store (> 0.5 miles from a supermarket/grocery store to a census tract centroid). These measures of the food environment surrounding schools were then merged with survey data on food consumption and measured BMI from 2,500 6^th graders collected in 2016.

Second, to explore the relationship between food environments and adolescent diet, we ran multinomial logistic regressions to assess the association between low, moderate, and high consumption of snack, sugary drinks, and fast food, whether or not a school was surrounded by a food swamp, food desert, or neither. Models were stratified by sociodemographic factors including race, gender, and whether or not the student was eligible for free or reduced school meals.

Third, to explore potential causal links between the presence of a food swamp or desert surrounding a school and unhealthy food consumption, as well as student weight status, we employ a propensity score matching strategy. This method estimates the average treatment effect of food swamps or food deserts by creating a statistical comparison group of non-food swamp and non-food desert students. A treatment group of “food swamp students” were matched to a comparison group based on this probability (considered a propensity score) using a caliper matching approach. Our main outcome variables for this analysis are (1) whether or not a student is overweight or obese and (2) whether or not the student is in the high consumption of unhealthy food categories.

Results suggest that students who attend schools in food swamps were more likely to be overweight or obese and more likely to be classified as a high consumer of sugary drinks, snack foods, and fast food than those students attending schools situated in non-food swamp areas (including food deserts). Zoning policies enforcing minimum distance requirements for unhealthy food retailers near schools warrants consideration as a strategy to address disparities in diet and weight status among children.