A dollar-aware food-environment index and a 27-year trajectory typology: a measurement foundation for diet and childhood-obesity research in Mississippi, 1997–2024

Background

The retail food environment is a widely used exposure in behavioural-nutrition and obesity research, on the premise that nearby food retailers shape diet and obesity risk. Over the past quarter-century, grocery stores have declined across rural and small-town America while limited-assortment discount (“dollar”) stores have proliferated. Standard food-environment indices classify retailers as healthy or less-healthy but typically exclude dollar stores, now the fastest-growing food-retail format. As a result, a single classification decision may alter how the food environment is measured and the conclusions drawn from it. We develop a dollar-aware index, quantify how counting dollar stores changes the measured exposure, and derive a longitudinal trajectory typology.

Methods

Using establishment-level data from Data Axle for all 878 Mississippi census tracts (1997–2024), we classified food retailers into five mutually exclusive categories using a previously validated approach and calculated the modified Retail Food Environment Index (mRFEI) in both its standard and dollar-aware forms, with the latter counting dollar stores as less-healthy outlets. We fitted Nagin-style group-based trajectory models to the tract-level dollar-aware index, related class membership to the Social Vulnerability Index (SVI) and urbanicity with multinomial regression, and characterised spatial clustering (Getis-Ord Gi*, join-counts) and grocery access.

Results

Grocery stores fell from 1,616 to 716 while dollar stores rose from 315 to 1,005, intersecting in 2018. Counting dollar stores lowered the index by a margin that widened over time, and a growing number of tracts had only dollar-store retail, undefined under the standard index. Six trajectory classes emerged: stable adequate (5.6% of tracts), steady decline (13.1%), early collapse (11.1%), late collapse (6.7%), persistently constrained (34.1%) and chronic desert (29.3%); only the stable-adequate class (5.2% of children) stayed adequate throughout. Constrained and steady-decline membership rose steeply with vulnerability (RRR 11.7 and 9.9); chronic desert was urban (RRR 5.2, a food-swamp pattern); collapse classes had no cross-sectional social signature.

Conclusions

In the US state with the highest adult obesity prevalence, a single retailer-classification decision substantially changes the measured food environment. The dollar-aware index and trajectory typology offer a transferable, time-varying exposure for behavioural-nutrition and obesity research and establish a foundation for future childhood-obesity studies.

Contributions to the literature

• Standard food-environment indices omit dollar stores, the fastest-growing US food-retail format; we show this classification choice changes the measured exposure by a widening margin and, in a growing set of tracts, determines whether the food environment can be scored at all.

• We provide a dollar-aware modified Retail Food Environment Index and a 27-year, tract-level trajectory typology that converts decades of food-retail change into a small set of interpretable, time-varying exposure histories suited to life-course diet and obesity research.

• The typology distinguishes urban food swamps from rural food deserts and separates communities by the timing of food-environment loss — information a single cross-sectional measure cannot capture.

The approach transfers to any longitudinal commercial business dataset and supplies the exposure-measurement foundation for forthcoming research on childhood obesity.