Several studies have shown that postnatal weight gain is a significant predictor for retinopathy of prematurity (ROP) in preterm infants. Using a cohort of 1,301 infants from a single-center ROP registry, we investigated whether incorporation of changes in Fenton preterm growth curve z scores (ie, deviation from the population average) provides improved predictive ability for developing ROP compared to weight gain alone. Three logistic regressions were fit to severe ROP: (1) baseline model that included gestational age and birth weight, (2) the baseline model adding weight gain, and (3) the baseline model adding change in z score. The area under the receiver operating characteristic curve (C index) was used to compare models. Both weight gain and change in z scores were significant predictors after adjusting for birth weight (P = 0.01) and gestational age (P < 0.01). The C indices were not significantly improved by including weight gain or z score to the baseline model; however, for a subset of subjects, change in weight z score may be a more useful measure compared to simple weight gain with regards to assessing risk for severe ROP.
Copyright © 2019 American Association for Pediatric Ophthalmology and Strabismus. Published by Elsevier Inc. All rights reserved.