Prediction of severe retinopathy of prematurity using the screening algorithm WINROP in preterm infants

Abstract

PURPOSE To assess the sensitivity and specificity of weight gain, insulin-like growth factor 1 (IGF-1), and neonatal retinopathy of prematurity (WINROP) algorithm to predict proliferative retinopathy of prematurity (ROP), in a Turkish population of preterm infants. METHODS The medical records of infants screened and monitored for ROP from 2007 to 2014 were analyzed retrospectively. Birth weights of infants born before 32 weeks' gestation were recorded on the WINROP online database system weekly until postmenstrual week 36. The sensitivity, specificity, and positive and negative predictive values of the WINROP algorithm were analyzed. RESULTS A total of 223 infants were included. WINROP yielded a low-risk result in 106 infants (48%) and a high-risk result (red alarm) in the remaining 117 infants (53%). The sensitivity of the WINROP online system was found to be 84.3% (27/32), whereas its specificity was found to be 52.8% (101/191). The time between the first alarm and treatment was 8.59 +/- 3.92 (2-15) weeks. Using this algorithm, 106 infants would not have needed eye examinations, possibly resulting in a 40% decrease in the total number of examinations. CONCLUSIONS The WINROP online system is a valuable and easy-to-use monitoring system that could decrease the number of infant ROP examinations.