The clinical performance and population health impact of birth weight-for-gestational age indices with regard to adverse neonatal outcomes in term infants

Abstract

Background Despite the recent creation of several birth weight-for-gestational age references and standards, none has proven superior. We identified birth weight-for-gestational age cut-offs, and corresponding United States population-based, Intergrowth 21 st and World Health Organization centiles associated with higher risks of adverse neonatal outcomes, and evaluated their ability to predict serious neonatal morbidity and neonatal mortality (SNMM). Methods and findings The study population comprised singleton live births at 37-41 weeks’ gestation in the United States, 2003-2017. Birth weight-specific SNMM, which included 5-minute Apgar score<4, neonatal seizures, assisted ventilation and neonatal death, was modeled by gestational week using penalized B-splines. We estimated the birth weights at which SNMM odds was minimized (and higher by 10%, 50% and 100%), and identified the corresponding population, Intergrowth 21 st and World Health Organization (WHO) centiles. We then evaluated the individual- and population-level performance of these cut-offs for predicting SNMM. The study included 40,179,663 live births at 37-41 weeks’ gestation and 991,486 SNMM cases. Among female singletons at 39 weeks’ gestation, SNMM odds was lowest at 3,203 g birth weight (population, Intergrowth and WHO centiles 40, 52 and 46, respectively), and 10% higher at 2,835 g and 3,685 g (population centiles 11 th and 82 nd , Intergrowth centiles 17 th and 88 th and WHO centiles 15 th and 85 th ). SNMM odds were 50% higher at 2,495 g and 4,224 g and 100% higher at 2,268 g and 4,593 g. Birth weight cut-offs were poor predictors of SNMM. For example, the birth weight cut-off associated with 10% higher odds of SNMM among female singletons at 39 weeks’ gestation resulted in a sensitivity of 12.5%, specificity of 89.4% and population attributable fraction of 2.1%, while the cut-off associated with 50% higher odds resulted in a sensitivity of 2.9%, specificity of 98.4% and population attributable fraction of 1.3%. Conclusions Birth weight-for-gestational age cut-offs and centiles perform poorly when used to predict adverse neonatal outcomes in individual infants, and the population impact associated with these cut-offs is also small.