Diagnostic Accuracy of Adjusted Growth Curves to Predict Cognitive Scores at 24 Months in Extremely Preterm Infants (P11-054-19)

Abstract

ObjectivesTo compare the diagnostic accuracy of adjusted and unadjusted growth curves to define postnatal growth and predict neurodevelopment at age 2 years in extremely preterm infants. MethodsWe performed a retrospective cohort study assessing infant growth at 36 weeks post-menstrual age (PMA) in 350 extremely preterm infants born ≤ 26 6/7 weeks gestational age (GA) between 01/01/2006-12/31/2014 at University of Alabama at Birmingham Regional Neonatal Intensive Care Unit. Postnatal growth was defined as below, within, or above target using adjusted and unadjusted growth curves. Linear regression models were used to compare adjusted and standard growth trajectories at 36 weeks PMA. The primary outcome was Cognitive Composite Score (CCS) of the Bayley Scales of Infant Development-III (Bayley-III) at 24 months. ResultsMean birthweight (BW) was 750 ± 138 g and median GA was 25 weeks (interquartile range: 24 to 26). A multivariate analysis of postnatal growth defined with adjusted curves and eight covariates (GA, BW, weight-Z-score at birth, sex, race, antenatal steroid use, singleton birth, and corrected age at follow-up assessment) predicted higher CCS-Bayley-III scores at 24 months in infants with postnatal growth within target (adjusted mean ± standard error: 89 ± 3) and lower scores in infants with postnatal growth below and above target (85 ± 3 vs. 83 ± 6, respectively) at 36 weeks PMA (P = 0.04). A multivariate analysis of postnatal growth defined with unadjusted curves and the same covariates did not predict significant differences in scores of infants with postnatal growth below (86 ± 3), within (86 ± 3), or above target (86 ± 6) at 36 weeks PMA (P = 0.99). ConclusionsAdjusted growth curves identified an inverted “U-shaped” association between postnatal growth and CCS-Bayley-III scores at 24 onths. Individualized growth trajectories adjusted for physiologic weight loss may predict cognitive impairment more accurately than traditionally defined growth standards. Additional well-powered studies are needed to validate the diagnostic ability of adjusted growth curves in routine clinical practice. Funding SourcesNone.