Abstract
Objective Lack of standardization of infant mortality rate (IMR) calculation between regions
in the United States makes comparisons potentially biased. This study aimed to quantify
differences in the contribution of early previable live births (<20 weeks) to U.S.
regional IMR.
Study Design Population-based cohort study of all U.S. live births and infant deaths recorded
between 2007 and 2014 using Centers for Disease Control and Prevention's (CDC's) WONDER
database linked birth/infant death records (births from 17–47 weeks). Proportion of
infant deaths attributable to births <20 vs. 20 to 47 weeks, and difference (ΔIMR)
between reported and modified (births ≥20 weeks) IMRs were compared across four U.S.
census regions (North, South, Midwest, and West).
Results Percentages of infant deaths attributable to birth <20 weeks were 6.3, 6.3, 5.3,
and 4.1% of total deaths for Northeast, Midwest, South, and West, respectively, p < 0.001. Contribution of < 20-week deaths to each region's IMR was 0.34, 0.42, 0.37,
and 0.2 per 1,000 live births. Modified IMR yielded less regional variation with IMRs
of 5.1, 6.2, 6.6, and 4.9 per 1,000 live births.
Conclusion Live births at <20 weeks contribute significantly to IMR as all result in infant
death. Standardization of gestational age cut-off results in more consistent IMRs
among U.S. regions and would result in U.S. IMR rates exceeding the healthy people
2020 goal of 6.0 per 1,000 live births.
Keywords
infant mortality rate - previable - regional
