Diabetes mortality in Mexico: socioeconomic and contextual gaps

Background Type 2 diabetes mellitus (T2DM) is one of the leading causes of mortality in Mexico and exhibits marked territorial disparities. While socioeconomic gradients in diabetes outcomes are well documented at the individual level, limited evidence from the sociology of stratification quantifies the extent to which mortality inequalities are structured by state-level institutional and governance contexts beyond municipal socioeconomic composition. This study examines territorial heterogeneity in T2DM mortality using a multilevel analytical framework grounded in social epidemiology and Latin American social medicine. Methods An ecological study with cross-sectional structural covariates was conducted using municipal-level data from Mexico. Mortality data correspond to deaths registered in 2024 (ICD-10 E11–E14, INEGI); structural municipal indicators derive from the most recent available the most recent available CONEVAL municipal poverty measurement (2020). The analytical sample comprised 2,454 municipality observations nested within 32 federal states, selected as the second level of analysis given their constitutional autonomy over health system organization, expenditure allocation, and social protection policy. The standardized mortality ratio (SMR) of diabetes mellitus (ICD-10 E11–E14) was calculated by indirect standardization by age, as the ratio of observed to expected deaths in each municipality, using the Mexican population of 2024 as the reference population. Following the social epidemiology framework, the design distinguishes compositional effects (municipal socioeconomic characteristics) from contextual effects (unobserved state-level institutional environments). Mortality counts among individuals aged ≥20 years were modeled using multilevel mixed-effects Poisson and negative binomial regressions with state-level random intercepts and a log population offset. Structural municipal indicators included income vulnerability, educational marginalization, food vulnerability, health service access, mean age, and demographic composition. Collinearity was assessed through variance inflation factors (VIF; all <5). Contextual heterogeneity was quantified using the intraclass correlation coefficient (ICC), median rate ratio (MOR), and proportional change in variance (PCV). Results The null model showed substantial contextual heterogeneity (state-level variance = 0.134), with 51.6% of unexplained variability attributable to differences between states. The median rate ratio (MOR = 1.42) indicated that municipalities with identical characteristics would differ by 42% in mortality solely due to state context. Adjustment for municipal structural variables reduced contextual variance by 38.1% (PCV; MOR = 1.32), yet substantial residual heterogeneity persisted. In the negative binomial model, food vulnerability (IRR = 1.006; p < 0.001) and mean age (IRR = 1.020; p < 0.001) remained independently associated with mortality across all specifications. State-level variance remained significant (σ² = 0.098; MOR = 1.36). Conclusion Diabetes mortality in Mexico is territorially structured and partially mediated by municipal material conditions, yet substantial residual inequality persists at the state level after rigorous adjustment. These findings, interpreted through the lens of the sociology of stratification and Latin American social medicine, suggest that institutional and governance contexts, including differential health system organization, social infrastructure investment, and historical patterns of territorial inequality, operate as structural determinants of chronic disease mortality beyond local socioeconomic composition.