Population health impacts of national health insurance coverage for neonatal intensive care unit admission in South Korea: evidence from regression discontinuity analyses

Background: Neonatal intensive care units (NICUs) are among the most resource-intensive pediatric services. While benefits for extremely preterm or low-birth-weight infants are well established, evidence is mixed for moderately at-risk groups. Objective: To evaluate the impact of South Korea's staged expansions in public insurance coverage for NICU admission on infant mortality (IM) and under-five mortality (U5M). Method: We analyzed 3461,281 birth–death–linked records (2010–2017) using two quasi-experimental designs. A local-randomization regression discontinuity (RD) design exploited clinical thresholds in birthweight (≤1750 g; 1751–2000 g) and gestational age (≤33; 34–35 weeks) to compare full, partial, and no coverage. A regression discontinuity in time (RDiT) design examined births around the March 1, 2013 policy reform. Analyses used exact matching within ±250 g or ±2 weeks, with narrower windows for sensitivity checks. Results: Full coverage had no measurable effect on mortality compared to no coverage. However, full coverage was associated with higher mortality than partial coverage at the 1,750g/34-week cutoff (IM: +6.1; U5M: +7.1). Partial coverage was associated with lower mortality than no coverage at the 2,000g/35-week threshold (IM: –9.3; U5M: –8.7), but these effects did not persist under narrower bandwidths. RDiT analyses found no significant effect of the policy change on either outcome. Conclusion: NICU coverage showed no mortality benefit for the most vulnerable infants and may have led to overuse near full-coverage thresholds. Targeted policies and rigorous evaluation are needed to maximize benefits and avoid harm.