Effect of exposure to conflicts on childhood mortality: analyses of pooled cross-sectional data from 105 surveys from 52 countries.

Abstract

This study examines the relationship between armed conflict exposure and child mortality using a large-scale, multi-country dataset. We pool cross-sectional data from 105 Demographic and Health Surveys across 52 countries between 1998 and 2018, linking georeferenced conflict events from the Uppsala Conflict Data Program (UCDP) with individual-level records for 943,139 preschool-age children. A Cox proportional hazards model is used to estimate the association between conflict exposure and under-five mortality, controlling for socio-economic, demographic, and provincial-level characteristics. Results show a significant and robust association between conflict exposure and excess child mortality. In the baseline model, exposed children face a higher log hazard of death (β = 0.186, p < 0.01), corresponding to roughly a 20% increase in mortality hazard. The association holds regionally, with a 24% higher hazard in Asia and a 14% increase in Africa, and remains significant in fully adjusted specifications. Mediation analyses reveal that conflict exposure reduces maternal healthcare utilization (4+ ANC visits: β = -0.202, p < 0.01) and institutional delivery (β = -0.188, p < 0.01). Findings are robust across alternative conflict measures, geo-coded sub-samples, and non-migrant populations. The evidence highlights the urgent need to protect maternal and child health services in conflict-affected settings.