Research on the retail food environment commonly treats nearby food retailers as determinants of diet and obesity risk. Over the past quarter-century in the United States, grocery stores have declined in many rural and small-town areas while limited-assortment discount or dollar stores have proliferated. Typical food-environment indices classify retailers into healthy and less-healthy groups but often omit dollar stores, which are among the fastest-growing food-retail formats. The classification decision—whether to count dollar stores as less-healthy outlets—can materially change measured exposure.
This study develops a dollar-aware food-environment index, quantifies how including dollar stores alters measurement over time, and derives a 27-year trajectory typology across Mississippi census tracts to support behavioural-nutrition and childhood-obesity research.
The analysis used establishment-level business records licensed from Data Axle for all 878 Mississippi census tracts covering 1997–2024. Food retailers were classified into five mutually exclusive categories using a previously validated approach; the source article does not list the five categories in detail. The authors calculated the modified Retail Food Environment Index (mRFEI) in two forms: the standard index and a dollar-aware mRFEI, with dollar stores counted as less-healthy outlets in the latter.
For longitudinal analysis, the tract-level dollar-aware index was modelled with Nagin-style group-based trajectory models to identify distinct temporal classes. The authors related trajectory-class membership to the Centers for Disease Control and Prevention Social Vulnerability Index (SVI) and to urbanicity using multinomial regression. They also characterised spatial patterns using Getis-Ord Gi* and join-counts and described grocery access across tracts. Detailed model specifications, variable coding, and statistical parameters were not reported in the summary.
Across the study period, the count of grocery stores in Mississippi fell from 1,616 to 716 establishments, while dollar stores increased from 315 to 1,005. The two series intersected in 2018. As dollar stores were counted as less-healthy outlets in the dollar-aware mRFEI, the index values were lower than the standard mRFEI, and the negative margin widened over time.
A growing number of census tracts contained only dollar-store retail; such tracts were undefined under the standard mRFEI but were captured by the dollar-aware formulation. The changing composition of retailer types therefore altered the distribution and interpretation of the food-environment exposure over time.
Group-based trajectory modelling of the dollar-aware index identified six longitudinal classes at the census-tract level:
Only the stable adequate class retained adequate measured food environments throughout the 27-year period; this class represented 5.2% of children in the state according to the authors’ population linkage. The persistently constrained and chronic desert classes together comprised a majority of tracts, indicating widespread sustained limitations in healthy retail access when dollar stores are counted as less-healthy outlets.
Multinomial regression relating trajectory-class membership to the Social Vulnerability Index (SVI) and urbanicity found strong gradients for some classes. Membership in the persistently constrained and steady-decline classes increased steeply with higher SVI, with reported relative risk ratios of 11.7 and 9.9 respectively, indicating much greater likelihood of being in these disadvantaged trajectories for more vulnerable tracts.
The chronic desert class was associated with urbanicity (relative risk ratio 5.2), a pattern the authors describe as consistent with a food-swamp configuration where less-healthy retail concentrates in urban areas. By contrast, the early- and late-collapse classes did not exhibit a clear cross-sectional social signature by SVI or urbanicity in the reported summary.
The authors characterised spatial clustering using Getis-Ord Gi* statistics and join-count analyses and described patterns of grocery access across tracts. Specific spatial results, maps, and numeric cluster statistics were not detailed in the source summary. However, the narrative indicates clustering consistent with inequities: higher vulnerability tracts were more likely to follow constrained or declining trajectories, and urban tracts were overrepresented in the chronic desert class.
In Mississippi—the US state with the highest adult obesity prevalence—the decision to classify dollar stores as less-healthy outlets substantially changes measurement of the retail food environment. Grocery-store decline alongside dollar-store growth led to widening differences between the standard and dollar-aware mRFEI over time, created tracts with only dollar-store retail that were undefined under the standard index, and produced six distinct longitudinal trajectories of food-retail exposure.
The authors present the dollar-aware mRFEI and the 27-year trajectory typology as transferable, time-varying exposure constructs suitable for behavioural-nutrition and childhood-obesity research. These measures provide a measurement foundation for future studies linking changing retail environments to diet and obesity outcomes among children.
The establishment-level Data Axle business records analysed in this study were obtained under commercial licence and are not publicly redistributable; equivalent historical files can be obtained from Data Axle under licence or subscription. All secondary datasets cited by the authors are publicly available from their sources. The authors state that the R analysis code will be made public on acceptance.
The summary does not report detailed model specifications, covariate definitions beyond SVI and urbanicity, or numeric spatial-statistic outputs. Where the source does not provide such details, they are not reported here. The authors declared no competing interests and confirmed ethical approvals and participant-consent procedures as applicable.