**Background:** Childhood overweight and obesity are rapidly increasing in Mexico, with prevalence reaching 36% in children aged 5–11 and 38% in adolescents in 2018. While individual dietary behaviors are well-studied, less attention has been given to contextual factors like the food environment. The authors aimed to evaluate the longitudinal association between changes in retail food store density and changes in BMI among Mexican children and adolescents, hypothesizing that increases in fruit/vegetable and specialty food stores would be associated with BMI decreases, while increases in supermarkets, convenience stores, and small food retail stores would be associated with BMI increases.
**Methods:** Data came from 7,507 participants aged 5–19 years from the Mexican Family Life Survey (MxFLS), a longitudinal, multi-thematic survey representative of the Mexican population, with three waves (2002, 2005–2006, 2009–2012). Weight and height were measured by trained personnel using a digital scale (Tanita, nearest 0.1 kg) and SECA stadiometer (nearest 0.1 cm). Implausible BMI values were excluded (<10 kg/m² and >38 kg/m² for children; <10 kg/m² and >58 kg/m² for adolescents). Food store density at the municipal level (number of stores/area in km²) was obtained from the Economic Censuses of 1999, 2004, and 2009. Stores were categorized into five mutually exclusive types: small food retail stores, specialty food stores (poultry, meat, fish/seafood, dairy, seed-grain, bakery), fruit and vegetable stores, convenience stores, and supermarkets. Density was classified into quartiles. Five separate longitudinal linear fixed-effects regression models were estimated (one per food store type), adjusted for age (linear, quadratic, and cubic terms), sex, parental education, municipal-level socioeconomic deprivation (marginality index), and population density. Analyses were stratified by urbanization (municipalities belonging to cities of ≥100,000 residents vs. smaller areas).
**Key Results:** At baseline, mean age was 10.0 years, 51.2% were female, and the combined proportion of overweight and obesity was 30.0%, increasing to 36.6% over 10 years (urban: 32.7% to 38.5%; non-urban: 27.1% to 34.4%). Small food retail and specialty stores dominated the food environment. There was a significant inverse dose-response association between fruit/vegetable store density and BMI (β = −0.455 kg/m², 95% CI: −0.704 to −0.207 in quartile 2; β = −0.733 kg/m², 95% CI: −1.072 to −0.394 in quartile 3; β = −0.838 kg/m², 95% CI: −1.291 to −0.385 in quartile 4; p-trend <0.001). In non-urban areas, children in municipalities with the highest density of small food retail stores experienced a reduction in BMI (β = −0.840 kg/m², 95% CI: −1.612 to −0.068). In urban areas, specialty food stores showed an inverse association with BMI (quartile 3: β = −0.789 kg/m², 95% CI: −1.512 to −0.066; quartile 4: β = −1.204 kg/m², 95% CI: −2.046 to −0.362). Supermarket density in the third quartile was positively associated with BMI overall (β = 0.310 kg/m², 95% CI: 0.145 to 0.476; p-trend = 0.006). Age-stratified analyses suggested stronger associations in adolescents. For example, higher fruit/vegetable store density was associated with decreasing BMI trends across all ages, but the magnitude was greater for adolescents.
**Clinical Implications:** This study provides longitudinal evidence from a middle-income country that the availability of healthy food retailers—particularly fruit and vegetable stores—may contribute to reductions in children's and adolescents' BMI. The findings suggest that policy approaches to combat childhood obesity could be tailored by type of food store, with consideration for urbanization level and age. Promoting healthy food stores (fruit/vegetable and specialty stores) should be included within strategies aimed at improving nutritional health in Mexican children. The study's limitations include lack of data on pubertal stages, potential selection bias (though only 2.4% of children moved across municipalities), inability to assess the consumer food environment within stores, and lack of spatial accessibility measures. Nonetheless, the longitudinal design with fixed-effects models strengthens causal inference by controlling for time-invariant confounders.