Peer-reviewed article

Effect of Continuum of Maternal Healthcare on Neonatal Mortality in Sub-Saharan Africa: A Pooled DHS-8 Analysis

Samba Camara, Duah Dwomoh, Prudence Tettey, Amadou Barrow · 21 Aug 2026

Abstract

Abstract Background Neonatal mortality remains disproportionately high in sub-Saharan Africa (SSA), accounting for nearly half of all deaths in children under five. Although antenatal care, skilled birth attendance, and postnatal care are individually associated with improved newborn survival, few studies have examined whether their integrated receipt as a continuum of care (CoC) confers additional protection against neonatal death, particularly after accounting for sociodemographic confounding and heterogeneity across country contexts. Methods A pooled cross-sectional analysis was conducted using DHS-8 Births Recode files from five countries: Nigeria (2024), Mali (2023–2024), Congo DRC (2023–2024), Kenya (2022), and Lesotho (2023–2024). The analytical sample comprised 37,351 births within the 36-month postnatal care reference window. Complete CoC was defined as receipt of adequate antenatal care (≥4 visits with first-trimester initiation), skilled birth attendance, and postnatal care within 48 hours for the mother or newborn. Neonatal mortality was defined as death within 27 days of birth. Survey-weighted logistic and log-Poisson regression models estimated adjusted odds ratios (aOR) and adjusted prevalence ratios (aPR). G-computation quantified the population attributable fraction (PAF). Country-specific heterogeneity was examined through random-effects meta-analysis (DerSimonian-Laird method) and a two-level multilevel logistic regression model. Results The overall neonatal mortality rate was 29.3 per 1,000 live births (95% CI: 27.6–31.0). Complete CoC prevalence was 19.2% (95% CI: 18.5–19.9%), ranging from 7.8% in Congo DRC to 47.8% in Lesotho. In unadjusted analysis, complete CoC was associated with a 24% reduction in neonatal death odds (cOR: 0.764, 95% CI: 0.583–1.000, p = 0.050). After adjustment for wealth, education, residence, parity, maternal age, child sex, and country, the association was substantially attenuated and non-significant (aOR: 0.961, 95% CI: 0.717–1.289; aPR: 0.962, 95% CI: 0.722–1.282). The PAF under universal complete CoC was 3.2%. The pooled meta-analytic estimate was aOR 0.718 (95% CI: 0.447–1.152), with moderate heterogeneity (I² = 38.9%; τ² = 0.089). Country-specific estimates ranged from aOR 0.455 (95% CI: 0.256–0.810) in Kenya to 1.447 (95% CI: 0.496–4.220) in Lesotho. Conclusion Complete continuity of maternal healthcare was not independently associated with reduced neonatal mortality after full adjustment, suggesting that the unadjusted benefit was attributable to sociodemographic selection rather than a direct causal pathway. These findings underscore the insufficiency of service utilisation metrics in isolation and highlight the need to address the structural and contextual determinants that simultaneously constrain both care access and neonatal survival. Country-level heterogeneity in the CoC-mortality relationship points to the importance of tailored, context-specific interventions.