Gestational age, birthweight, and maternal factors associated with neonatal mortality in the United States.

BACKGROUND: Neonatal mortality remains a key indicator of maternal and infant health and continues to show substantial variation across demographic and clinical groups in the United States. A clearer understanding of nonlinear associations and effect modification may help characterize epidemiologic patterns of neonatal mortality and inform perinatal prevention efforts. METHODS: We conducted a population-based retrospective cohort study using the U.S. Linked Birth-Infant Death Database from 2015 to 2020. Neonatal mortality was defined as death occurring from day 0 through day 27 after live birth, inclusive. The study included 22,184,639 live births, among which 77,523 neonatal deaths were identified. Multivariable logistic regression was used to estimate adjusted odds ratios and 95% confidence intervals. Restricted cubic splines were used to characterize nonlinear associations, and subgroup analyses were conducted to evaluate potential effect heterogeneity. RESULTS: Among 22,184,639 live births, 77,523 neonatal deaths occurred, corresponding to a neonatal mortality rate of 0.35%. Preterm birth was strongly associated with neonatal mortality (aOR, 4.81; 95% CI, 4.62-5.04), as was low birthweight (aOR, 5.21; 95% CI, 5.04-5.45). Late or absent prenatal care was associated with higher odds of neonatal mortality (aOR, 1.70; 95% CI, 1.60-1.82), as were non-Hispanic Black race/ethnicity (aOR, 1.45; 95% CI, 1.38-1.53), maternal smoking (aOR, 1.35; 95% CI, 1.28-1.43), hypertensive disorders of pregnancy (aOR, 1.22; 95% CI, 1.15-1.31), and male infant sex (aOR, 1.12; 95% CI, 1.08-1.16). Restricted cubic spline analyses showed nonlinear associations for both gestational age and birthweight, with the lowest mortality odds near 39-40 weeks and approximately 3,200-3,600 g. CONCLUSION: In this nationwide observational cohort, gestational age, birthweight, and several maternal and infant characteristics were associated with neonatal mortality. The identified nonlinear patterns and subgroup differences may help characterize population-level mortality gradients and identify groups with comparatively elevated risk, although causal inferences cannot be made.