Samantha Gonski1, Susan R Hupp2, C Michael Cotten3, Reese H Clark4, Matthew Laughon5, Kevin Watt3,6, Christoph P Hornik3,6, Karan Kumar3, P Brian Smith3,6, Rachel G Greenberg7,8. 1. North Carolina School of Science and Mathematics, Durham, NC, USA. 2. Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas, TX, USA. 3. Department of Pediatrics, Duke University, Durham, NC, USA. 4. Pediatrix-Obstetrix Center for Research and Education, Sunrise, FL, USA. 5. School of Medicine, Division of Neonatal-Perinatal Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA. 6. Duke Clinical Research Institute, Durham, NC, USA. 7. Department of Pediatrics, Duke University, Durham, NC, USA. rachel.greenberg@duke.edu. 8. Duke Clinical Research Institute, Durham, NC, USA. rachel.greenberg@duke.edu.
Abstract
OBJECTIVE: Quantify the risk of treatment for retinopathy of prematurity (ROP) among infants meeting current U.S. screening guidelines. STUDY DESIGN: Among infants ≤1500 g birth weight or ≤30 weeks gestation screened for ROP from 2006-2015, we developed a risk prediction model to identify infants treated for ROP. We applied our model to a separate infant cohort discharged in 2016. RESULT: Seventy-five thousand eight hundred and twenty one infants met inclusion criteria; 2306 (3%) were treated for ROP. Infants with several risk factor combinations (no ventilator support or oxygen on postnatal day 28, no history of necrotizing enterocolitis, and no intraventricular hemorrhage) were at low risk of ROP. Applied to 6127 infants discharged in 2016, our model had 97.9% sensitivity, 63.3% specificity, positive predictive value of 4.0%, and negative predictive value of 99.9%. CONCLUSION: Large numbers of infants at low risk of developing ROP are required to undergo screening. Refining current ROP guidelines may reduce unnecessary examinations.
OBJECTIVE: Quantify the risk of treatment for retinopathy of prematurity (ROP) among infants meeting current U.S. screening guidelines. STUDY DESIGN: Among infants ≤1500 g birth weight or ≤30 weeks gestation screened for ROP from 2006-2015, we developed a risk prediction model to identify infants treated for ROP. We applied our model to a separate infant cohort discharged in 2016. RESULT: Seventy-five thousand eight hundred and twenty one infants met inclusion criteria; 2306 (3%) were treated for ROP. Infants with several risk factor combinations (no ventilator support or oxygen on postnatal day 28, no history of necrotizing enterocolitis, and no intraventricular hemorrhage) were at low risk of ROP. Applied to 6127 infants discharged in 2016, our model had 97.9% sensitivity, 63.3% specificity, positive predictive value of 4.0%, and negative predictive value of 99.9%. CONCLUSION: Large numbers of infants at low risk of developing ROP are required to undergo screening. Refining current ROP guidelines may reduce unnecessary examinations.