Assessment of the Children's Hospital of Philadelphia Score Algorithm for Screening Retinopathy of Prematurity in Premature Infants in Saudi Arabia

Abstract

Background: Retinopathy of prematurity (ROP) causes blindness among children, particularly preterm infants. While several screening models exist, their performance in developing countries is not well established. This study assessed the accuracy of Children's Hospital of Philadelphia-Retinopathy of Prematurity (CHOP-ROP) in Saudi Arabia and compared it with other models. Objective: To assess the diagnostic performance of the CHOP-ROP model in Saudi preterm infants and compare it with other ROP screening tools. Methods: This retrospective study involving 524 preterm infants was conducted in two tertiary hospitals in Jeddah. ROP risk was assessed using the CHOP-ROP and compared with four ROP risk models. Predictive values, specificity, sensitivity, and receiver operating characteristic curves were analyzed. Logistic regression identified type 1 ROP predictors. Results: Any-stage ROP was detected in 22.3% of infants; 9.1% required treatment. CHOP-ROP had the highest specificity (62.5%), accuracy (64.2%), and area under the curve (AUC) (0.71), indicating better ability to distinguish infants needing treatment, though its sensitivity was lower (81.4%) than other models. Lower birth weight, longer neonatal intensive care unit stay, and early blood transfusion were significant type 1 ROP predictors (all, P < 0.05). Conclusion: Although CHOP-ROP demonstrated strong specificity and accuracy, its lower sensitivity raises concerns about missed cases requiring treatment. Compared to more sensitive local tools, its performance was suboptimal. Region-specific model validation and threshold adjustments are needed to enhance its predictive value while minimizing false negatives.