Targeted Cycloplegia in Refractive Screening of Children and Adolescents: Development and Internal Validation of a Pre-Cycloplegia Risk Prediction Model - 10 PLADs, China, 2020-2024.

Introduction: Cycloplegic refraction is the gold standard for accurate refractive assessment in children and adolescents, but its use is limited by operational complexity and low acceptance. Because effective tools for identifying individuals who require cycloplegia are lacking, this study aimed to develop a risk prediction model for refractive misclassification based on pre-cycloplegia parameters to assist targeted cycloplegic refraction. Methods: A total of 43,760 children and adolescents aged 5-18 years were enrolled from 10 provincial-level administrative divisions in China using stratified cluster probability-proportional-to-size (PPS) sampling. Pre-cycloplegia data, including individual characteristics, ocular indicators, and lifestyle factors, were collected. Binary logistic regression was used to develop the model, and receiver operating characteristic (ROC) curves were used to determine the optimal cutoff value. Results: The final model included six core variables and achieved an overall classification accuracy of 79.4% and an area under the ROC curve (AUC) of 0.730. The optimal cutoff score was 18.4. Moderate or low uncorrected visual acuity and abnormal axial length were major risk factors, whereas a higher school stage was protective. Conclusion: The prediction model demonstrated favorable performance and identified individuals at high risk of refractive misclassification. Individuals with a risk score ≥18.4 may be prioritized for cycloplegic refraction.