County-level Variation in Infant Mortality Reporting at Early Previable Gestational Ages

Background: Infant mortality rate (IMR), or number of infant deaths per 1000 livebirths, varies widely across the US While fetal deaths are not included in this measure, reported infant deaths do include those delivered at previable gestations, or ≤20 weeks gestation. Variation in reporting of these events may have a significant impact on IMR estimates. Methods: This retrospective analysis used US National Center for Health Statistics 2007–2013 data from 2391 US counties. Counties were categorised by US region, demographic characteristics, and state-level fetal death reporting requirements. County percentage of fetal deaths among all 17–20 week fetal and infant deaths was evaluated using multivariable linear regression. County-level characteristics were then included in multivariable linear regression to determine the associated change in county IMR. Results: County percentage of deaths at 17–20 weeks reported as fetal ranged from 0% to 100% (mean 63.7%). Every 1 point increase in this percentage was associated with a 0.02 point decrease in county IMR (95% confidence interval (CI) 0.02, 0.03). When county IMRs were recalculated holding the percentage of fetal vs. infant deaths at 17–20 weeks constant at 63.7%, results suggest that the predicted gap in county IMR between Northeast and Midwest regions would narrow by 0.45 points. Conclusions: Variable reporting of previable fetal and infant deaths may compromise the validity of county IMR comparisons. Improved consistency and accuracy of fetal and infant death reporting is warranted.