Abstract
Introduction
Persistent racial disparities in US infant mortality may reflect both differences in the distribution of gestational age (GA) at birth (GA-distribution) and differences in mortality within GA strata (GA-specific mortality). We decomposed the Black–White infant mortality gap into components attributable to GA-distribution and GA-specific mortality.
Methods
We analyzed the US live births from 2016 to 2023 using the National Center for Health Statistics Linked Birth/Infant Death restricted file, constrained to maternal race non-Hispanic Black (NHB) or non-Hispanic White (NHW) single-race and gestational age 20 weeks or greater. The cohort includes 19,540,960 births to NHW (n = 15,274,847) and NHB (n = 4,266,113) individuals, with 107,233 infant deaths. A symmetric two-factor Kitagawa/Das Gupta decomposition separated the NHB–NHW infant mortality gap into a GA-distribution component and a GA-specific mortality component. Additionally, we examined stratum-specific contributions and temporal changes in the two components.
Results
The NHB–NHW infant mortality gap was 5.50 per 1,000 births. Of this difference, 4.22 per 1,000 (76.7%; 95% CI, 74.6–79.0) was attributable to differences in GA-distribution, whereas 1.28 per 1,000 (23.3%) was attributable to differences in GA-specific mortality. GA-distribution dominated every GA stratum below 28 weeks; the 22–23-week stratum alone contributed 1.23 per 1,000 to the overall gap. Contrastingly, GA-specific mortality dominated at term, contributing 0.480 per 1,000 at 37–38 weeks and 0.768 per 1,000 at 39–40 weeks.
Notably, NHB infant mortality was lower than NHW mortality at the most immature gestational ages but higher at term, with the mortality curves crossing between 28 and 31 weeks gestation. However, this should not be interpreted as evidence of a survival advantage or as measures of care quality because conditioning on gestational age can induce collider bias.
Temporal analyses revealed that the share of NHB–NHW infant mortality gap attributable to GA-distribution declined by 8.40% in the 2021 birth cohort (95% CI, 6.41–10.39; P = .0002). Furthermore, the two components exhibited distinct temporal shifts: GA-distribution went down in 2020 by 0.404 per 1,000 whereas GA-specific mortality went up in 2021 by 0.507 per 1,000.
Conclusion
Most of the US Black–White infant mortality gap during 2016–2023 was attributable to differences in the distribution of gestational age rather than differences in mortality within GA categories. However, the decomposition captured an important shift in the disparity: GA-distribution dominated very preterm births, whereas GA-specific mortality dominated at term births. The distinct 2020 GA-distribution decrease and 2021 GA-specific mortality increase further indicated that the two components followed different temporal trajectories. These findings support treating GA-distribution and GA-specific mortality as distinct dimensions of the disparity rather than as interchangeable explanations.