Health-system fragmentation and maternal mortality in Mexico, 2008–2022: a spatio-temporal spillover analysis

In Mexico, health-system fragmentation (HSF) coexists with persistent geographical inequalities in maternal mortality. Although both are spatially patterned, evidence on their relationship across time and territories remains limited. We assessed whether HSF was associated with total, direct, and indirect maternal mortality and whether these associations extended across spatially connected territories. We did a nationwide ecological spatio-temporal analysis of 2,457 Mexican municipalities from 2008 to 2022. To improve the stability of municipal maternal mortality ratios (MMRs), municipalities were grouped into 244 spatially contiguous clusters using Spatial ‘K’luster Analysis by Tree Edge Removal. The unit of analysis was the cluster–social-security group–year, distinguishing women with and without employment-based social security. HSF was defined as the proportion of live births not completing a maternal-care trajectory comprising any antenatal care, at least five antenatal contacts, first-trimester initiation, institutional delivery, and skilled birth attendance. We estimated non-spatial fixed-effects models and spatial lag and spatial Durbin panel models. Formal spatial-impact decomposition was used to estimate average direct, indirect, and total associations. Maternal mortality was spatially clustered and concentrated in marginalised regions. In 2022, direct MMR was higher among women without employment-based social security than among those with such security (53 vs. 46 deaths per 100 000 live births). A 10-percentage-point higher HSF was associated with 7.4 additional total maternal deaths per 100 000 live births ( p < 0.01) and 6.6 additional direct maternal deaths per 100 000 live births ( p < 0.01). The association with indirect MMR was smaller and less precise (0.8 additional deaths per 100 000 live births; p = 0.08). Indirect spatial associations accounted for 19.2% of the total estimated association with MMR, and spatial dependence was substantial across models (ρ = 0.16, p < 0.01). Greater discontinuity along maternal-care trajectories was associated with higher maternal mortality, particularly from direct obstetric causes, and part of this association extended across territorial boundaries. These findings have relevance beyond Mexico: expanding formal entitlement is unlikely to protect women when maternal care remains fragmented across institutions, territories, and referral networks. Advancing equitable universal health coverage requires not only broader coverage but also continuous maternal-care pathways, interoperable information systems, and coordination across providers, institutions, and territories.