Résumé
Background: Reducing child mortality rates is a unifying goal of the global health and international development communities. In Africa, unambiguous empirical evidence on how health system interventions can drive such reductions has been elusive. This gap in the literature is due to challenges in implementing system-level changes on a scale and pace to have measurable impacts on mortality, and the challenges of collecting adequate data on the population and programs over sufficient time with plausible counterfactuals. This study aimed to assess the population health impact of the first decade of implementation of a health system strengthening (HSS) intervention in a rural district of Madagascar.Methods and findings: The study is a prospective quasi-experiment using a district-representative cohort of over 1,500 households (five waves of survey collection), in combination with patient data collected across different levels of care (community health workers and health facilities), geographic information systems, and programmatic data to assess changes in mortality, healthcare coverage and utilization from 2014 to 2023. The HSS intervention integrates support to clinical programs with strengthened health system building blocks and social protection at all levels of care of a district health system (community health, primary care centers, and hospital). Under-five, infant and neonatal mortality were estimated at the population level using the synthetic life-table method for DHS surveys. Impact of the HSS intervention on healthcare coverage and utilization was assessed through interrupted time-series analyses. Changes in geographic and financial inequalities in coverage indicators were studied via the relative concentration index and slope index of inequality. Our results show that trends in child mortality rates (neonatal, infant, under-five) decreased in the initial HSS intervention area from 2014 to 2023, but increased in the comparison area as well as the rest of the country over the same period. The HSS intervention was associated with statistically significant increases in service coverage and primary care utilization for a wide range of maternal and child health indicators, as well as reductions in geographic and financial barriers to care. The main limitations of this study were that the intervention was not randomized, and that changes in child mortality were estimated from 5-year averages from repeated cross sections, with overlapping time windows that prevented formal integration into the statistical modeling framework used for coverage indicators.Conclusions: By measuring both indirect and direct impacts of HSS on population health in a context where health and economic indicators are not otherwise improving, these results provide converging evidence on how strengthening health systems, from community health to hospitals, in low-resource settings increases overall utilization of services, reduces inequities in access to those services, and corresponds with reductions in mortality.