Enhanced Screening Model for Retinopathy of Prematurity
JAMA Ophthalmology, September 2018
Current screening criteria for retinopathy of prematurity (ROP) are predominantly based on birth weight (BW) and gestational age (GA) and have low predictive value for identifying infants who are at risk for severe cases. To improve ROP screening specificity and sensitivity, Binenbaum et al. proposed a new set of criteria combining BW, GA, and postnatal weight gain.
The retrospective Postnatal Growth and ROP Study included 6-year data on 7,483 premature infants from the United States and Canada who were examined for ROP and had a known ROP outcome. The researchers developed a hybrid predictive model to apply to this data, which combined common BW and GA thresholds with a comparison with expected growth from infants without ROP, an assessment of multiple growth intervals, and a consideration of nonphysiological weight gain.
Their final model consisted of the following 6 screening criteria:
- a BW of <1,051 g;
- a GA of <28 weeks;
- a weight gain of <120 g during days 10-19 after birth, <180 g during days 20-29, or <170 g during days 30-39; and
- hydrocephalus diagnosed on brain imaging study.
Applied in this fashion to the 6-year data, the researchers’ model accurately predicted 459 of 459 infants with type 1 ROP, 466 of 472 with type 2 ROP, and 524 of 524 treated for ROP. It also reduced the number of unnecessary examinations by 2,269.
According to the authors, these criteria predict the development of severe ROP with a greater specificity and sensitivity than do current screening methods. And because the model uses routinely collected data and requires minimal calculation, it would have a minimal impact on workflow in neonatal intensive care units.
The original article can be found here.