Applicability of the ROPScore as a predictive algorithm for early detection of retinopathy of prematurity

Abstract

Introduction and purpose:

ROPScore is a scoring system that was proposed to predict severe retinopathy of prematurity (ROP) using easily obtainable parameters as predictive variables. The aim of this study is to assess the accuracy of the ROPScore algorithm as a predictor of ROP by the second week of life.

Materials and Methods:

Retrospective cohort study of 239 preterm infants with a gestational age (GA) ≤ 32 weeks and/or birthweight (BW) ≤ 1500 g. No ROP, any stage of ROP and severe ROP requiring treatment were categorized. ROPScore was calculated in the second week of life using the following parameters: GA, BW, weight by the second week of life, use of oxygen in mechanical ventilation and use of blood transfusions. Sensitivity, specificity and positive (PPV) and negative (NPV) predictive values were calculated. The best cut-offs of the algorithm were calculated and, whenever possible, a sensitivity of 100% was chosen.

Results:

Mean BW was 1241.6 ± 310.0g and mean GA was 29.8 ± 3.4 weeks. Of the 239 infants, ROP was identified in 101 (42.3%) and 12 (11.9%) had severe ROP requiring treatment. Mean ROPScore was 15.39 ± 1.94 in the any stage of ROP cohort and 17.52 ± 1.80 in the severe ROP cohort. The sensitivity of the algorithm was 96.0% and 100% for any stage of ROP and severe ROP, respectively. NPV was 92.3% and 100% for any stage of ROP and severe ROP, respectively.

Conclusion:

ROPScore is a simple public domain algorithm that can be applied as early as by the second week of life. It may be used as a reliable screening tool for ROP and optimize examination timings, but further validation is required.