Nurse-Led Antenatal Care in Conflict-Affected Yemen: A Prospective Cohort Study with Difference-in-Differences Analysis.

BACKGROUND: The growth of nurse-delivered care is often required in unstable and conflict-affected health systems due to physician shortages. Nevertheless, there is still a lack of research on the potential benefits of organized nurse-led prenatal care (ANC) in these kinds of environments. We evaluated the clinical outcomes and implementation characteristics associated with a protocol-driven nurse-led ANC model in conflict-affected Yemen. METHODS: We conducted a prospective cohort study across 12 public primary healthcare facilities in Al-Dhalea, Yemen. Pregnant women aged 15-49 years with a gestational age of ≤ 20 weeks were enrolled in structured nurse-led ANC (n = 224) or routine ANC (n = 226) according to facility capacity. Primary outcomes included follow-up hemoglobin levels and anemia (Hb < 11 g/dL). Secondary outcomes included low birth weight, preterm birth, and NICU admission. Multivariable regression, propensity score matching, difference-in-differences, mediation, and E-value analyses were performed. RESULTS: Follow-up was completed for 435 women (96.7%). Nurse-led ANC was associated with higher follow-up hemoglobin (adjusted β = +1.05 g/dL, 95% CI 0.82 to 1.28), lower odds of anemia (adjusted OR = 0.32, 95% CI 0.21 to 0.48), and reduced low birth weight (adjusted OR = 0.62, 95% CI 0.41 to 0.94). The difference-in-differences estimator was +1.12 g/dL (p < 0.001), with parallel trends satisfied (p = 0.42). E-value analysis (2.41) suggested robustness to unmeasured confounding. The model demonstrated characteristics consistent with adaptive nursing leadership across clinical, administrative, and community dimensions. CONCLUSION: Structured nurse-led prenatal care has been linked to better outcomes for mothers and newborns in Yemen affected by war. These results encourage funding for workforce resilience and nurse-led approaches in order to enhance the health system in humanitarian contexts. To prove causation, more studies using cluster-randomized designs are required.