The violence against women risk index: Design and application of a synthetic measure for subnational surveillance in Mexico

OBJECTIVES: To develop a reproducible public health index for monitoring violence against women at the municipal level that addresses statistical instability in small populations and supports equitable resource allocation. STUDY DESIGN: Cross-sectional ecological study using 2023 national administrative and vital statistics records. METHODS: The Violence Against Women Risk Index was constructed for Mexico's 2455 municipalities using 411,631 officially reported incidents. The index integrates hybrid burden-risk scoring with adaptive Empirical Bayes smoothing calibrated to population size (α = 200-8000). Six violence categories (domestic, sexual, harassment, exploitation, gender-specific, and fatal) were weighted by severity and aggregated into a composite measure. RESULTS: The mean national Violence Against Women Risk Index was 0.27 (range 0.00-0.98). Eleven per cent of municipalities, predominantly rural and Indigenous, reported zero incidents. Urban municipalities averaged 17 times higher risk (0.84) than small communities (0.05). Within-state inequality (SD ≈ 0.18) exceeded between-state variation (SD ≈ 0.11), indicating systematic underreporting and institutional capacity gaps in marginalized jurisdictions. CONCLUSIONS: The Violence Against Women Risk Index demonstrates that municipal-level gender-based violence can be measured reliably from administrative data using transparent, statistically stable methods. By exposing geographic inequalities and reporting heterogeneity in marginalized jurisdictions, the index provides public health systems with actionable surveillance infrastructure for evidence-based prevention planning, equitable resource allocation, and accountability. The framework is transferable to other decentralized or federated systems seeking subnational violence surveillance without requiring survey infrastructure or additional data collection costs.