A Tiered Approach to Retinopathy of Prematurity Screening (TARP) Using a Weight Gain Predictive Model and a Telemedicine System

Abstract

The Telemedicine Approaches to Evaluating Acute-Phase Retinopathy of Prematurity (e-ROP) Study telemedicine system of remote fundus image grading and The Children's Hospital of Philadelphia Retinopathy of Prematurity (CHOP-ROP) postnatal weight gain predictive model are 2 approaches for improving ROP screening efficiency. Current screening has low specificity for severe ROP. To describe a tiered approach to ROP screening (TARP) for identifying children who develop severe ROP using telemedicine and a predictive model synergistically. This investigation was a post hoc analysis of a cohort in the e-ROP Study (a multicenter prospective telemedicine study) and the Postnatal Growth and Retinopathy of Prematurity (G-ROP) Study (a multicenter retrospective cohort study). The setting was neonatal intensive care units at The Children's Hospital of Philadelphia and the Hospital of the University of Pennsylvania. Participants in the e-ROP Study were premature infants with a birth weight less than 1251 g and a known ROP outcome enrolled between May 25, 2011, and October 31, 2013. The G-ROP Study enrolled all infants undergoing ROP examinations with a known ROP outcome who were born between January 1, 2006, and December 31, 2011. The mean outcomes were the sensitivity for type 1 ROP, reductions in infants requiring imaging or examinations, numbers of imaging sessions and examinations, and total clinical encounters (imaging sessions and examinations combined). The following 4 screening approaches were evaluated: ROUTINE (only diagnostic examinations by an ophthalmologist), CHOP-ROP (birth weight and gestational age, with weekly weight gain initiating examinations when the risk cut point is surpassed), e-ROP IMAGING (trained reader grading of type 1 or 2 ROP initiates diagnostic examinations), and TARP (CHOP-ROP alarm initiates imaging, and imaging finding of severe ROP initiates diagnostic examinations). A total of 242 infants were included in the study, with a median birth weight of 858 g (range, 690-1035 g). The median gestational age was 27 weeks (range, 25-29 weeks). Fifty-one percent (124 of 242) were female, and 49% (118 of 242) were male. The race/ethnicity was 27.3% (66 of 242) white, 56.2% (136 of 242) black, 2.1% (5 of 242) Native American, 1.7% (4 of 242) Asian, and 12.8% (31 of 242) other. The sensitivity for detecting type 1 ROP (32 infants) was 100% (95% CI, 89.3%-100%) with each approach. With ROUTINE, 242 infants had 877 examinations; with CHOP-ROP, 184 infants had 730 examinations; with e-ROP IMAGING, 242 infants had 532 imaging sessions, and 94 infants had 345 examinations (877 patient encounters); and with TARP, 182 infants had 412 imaging sessions, and 87 infants had 322 examinations (734 patient encounters). The tiered approach to ROP screening was associated with a reduced number of examinations and imaging sessions compared with the other approaches. Applying a postnatal growth model and telemedicine system in a tiered approach may reduce the number of clinical ROP interventions more than either approach alone.