JMIR Publications

ABSTRACT

Background:

National surveillance of commercial retail environments is constrained by data sources that are updated infrequently and capture narrow dimensions of food access. The Google Places application programming interface (API) provides continuously updated, programmatically accessible data on US business locations, but its validity as a population-level exposure measure has not been systematically evaluated.

Objective:

To develop and evaluate a Google Places-derived US Built Environment Retail (UBER) Index as a scalable measure of county-level commercial retail infrastructure in the contiguous United States, and to characterize its spatial association with age-adjusted diabetes prevalence.

Methods:

We conducted a cross-sectional ecological study of contiguous US counties in accordance to STROBE guidelines. We attempted complete enumeration rather than probability sampling. Counts of alcohol outlets, fast-food and convenience stores, grocery and health-food stores, and fitness and recreation facilities were extracted from the Google Places API in February 2026. Principal component analysis of four standardized indicators produced one composite index. Construct validity was assessed against USDA Food Access Research Atlas and County Health Rankings benchmarks, with adequate convergence prespecified as absolute r ≥ 0.40. We estimated associations with age-adjusted diabetes prevalence from CDC PLACES 2025 using a spatial error model adjusted for Area Deprivation Index, urbanicity, and census division. Quantile regression, county-versus-tract comparisons, outcome-specificity analyses using obesity and coronary heart disease, and included-versus-excluded comparisons assessed robustness, scale sensitivity, specificity, and selection bias. All tests were two-sided at alpha = .05.

Results:

The analytic sample comprised 1,701 of 2,957 contiguous-US counties (57.5%). The first principal component explained 92.9% of variance with near-equal loadings (0.492 to 0.506). Convergent validity was weak (strongest Pearson r = -0.20; none reached an absolute r of 0.40). Each 1-SD increase in the UBER Index was associated with 0.24 percentage points higher diabetes prevalence (95% CI, 0.16 to 0.32; P < .001); area deprivation was the strongest predictor (0.086 per percentile; 95% CI, 0.081 to 0.091). The association was stable across quantiles. The index was not associated with obesity and was weakly and inversely associated with coronary heart disease. Excluded counties were more rural (85.0% vs 45.4%). The positive county-level association reversed at the tract level, indicating scale-dependent ecological confounding.

Conclusions:

Google Places data can be transformed into a reliable, nationally scalable index of commercial retail infrastructure. The index captures a general commercial-density dimension not represented in existing food-environment surveillance tools; its scale-dependent sign reversal and weak convergent validity indicate that the diabetes association is ecological rather than causal. The primary contribution is methodological: a proof-of-concept that API-based platform data can yield standardized built-environment exposure surfaces supporting chronic-disease surveillance and the commercial-determinants-of-health agenda, pending validation of temporal stability, finer-scale validity, and health relevance.