Predictors of Treatment-warranted Retinopathy of Prematurity in the SUNDROP Cohort: Influence of Photographic Features

Purpose: The Stanford University Network for the Diagnosis of Retinopathy of Prematurity (SUNDROP) initiative-an ongoing telemedicine-based initiative for in-hospital screening of high-risk infants for treatment-warranted ROP (TW-ROP)-has been shown to be a safe, reliable, and cost-effective supplement to the efforts of ROP specialists. We utilized data collected in the SUNDROP initiative to determine demographic (birth weight, sex, multiplicity), weight gain, and ocular imaging (media haze, peripapillary atrophy, fundus pigmentation) predictors of TW-ROP.

Methods: This was a retrospective nested case-control study. Cases and controls were selected from a cohort of 843 low birth weight, premature newborns who survived to an estimated gestational age of 31 weeks and underwent screening through the SUNDROP initiative. Infants were screened at one of six neonatal intensive care units from December 1, 2005, to April 1, 2016. Cases (n = 37) were newborns with TW-ROP who underwent retinal ablative therapy. Two controls (n = 74) without TW-ROP were matched to each case by gestational age. One reviewer graded media haze, presence of peripapillary atrophy, and fundus pigmentation in images taken at the baseline exam for each newborn. The main outcome measure was association of TW-ROP with predictive factors.

Results: In the SUNDROP trial, 37 out of 843 (4.4%) newborns developed TW-ROP. In a multivariable model, birth weight (OR, 0.32; 95% CI, 0.15-0.70) was inversely associated with TW-ROP. In contrast to prior reports, we found no significant difference in sex, multiplicity, or fundus pigmentation at baseline exam in those with TW-ROP as compared to controls. High levels of media haze (>2, scale 0 to 5) were found in the majority of cases (67.6%, 25/37) and controls (65.7%, 44/67). Presence of peripapillary atrophy did not improve prediction of the development of TW-ROP compared to birth weight and weight gain rate alone.

Conclusions: The finding of high levels of media haze at baseline ROP screening exams is novel. This study supports the current model for detection of TW-ROP using birth weight, gestational age, and weight gain rate. We found no significant difference between newborns with TW-ROP and controls in baseline presence of media haze, fundus pigmentation or peripapillary atrophy.