Spatial analysis of food environment quality and chronic disease patterns in Detroit: repeated cross-sectional study (2014–2024)

Source abstract

Background: Food environment interventions assume stable protective associations with health outcomes. This study contributes to the Cities & Health Special Issue on visualizing health inequalities and health equity for urban planning. We show how objective spatial visualization methods, including GIS mapping and hot spot analysis, can reveal temporal dynamics in food environment inequities. We tested this assumption during rapid urban change in Detroit, Michigan (2014–2024). Methods: Repeated cross-sectional analysis of all Detroit census tracts at four timepoints (2014, 2018, 2020, 2024). Nutrition Environment Measures Survey for Stores (NEMS-S) assessments of full-line grocery stores (n = 80 in 2014 declining to n = 60 in 2024) examined availability, price, and quality. Linear regression tested associations controlling for Social Vulnerability Index. Getis-Ord Gi* analysis identified spatial clusters (hot and cold spots) of food environment quality. Results: Food environment quality improved (NEMS-S: 25.5 to 29.3, p 0.05), emerged in 2020, and reversed by 2024 when better scores were paradoxically associated with higher chronic disease prevalence. Adding economic controls (income, SNAP participation) attenuated associations by 20–37%, but significant positive associations persisted for obesity (β = 0.21, p = 0.035) and blood pressure (β = 0.39, p = 0.020). Stratified analysis revealed these paradoxical associations were concentrated in hot spots (improving food environments) while cold spots showed null associations, consistent with residential selection bias in gentrifying areas. Conclusions: Food environment-health relationships proved temporally unstable, shifting from null to harmful associations. Price improvements paradoxically predicted worse health outcomes by study end. These findings challenge assumptions of consistent protective food environment effects and suggest that unmeasured contextual factors may attenuate or obscure associations between grocery store quality and urban health patterns in ecological analyses.