Nomogram for Predicting Myopia Risk in Chinese Children and Adolescents Aged 5-19 Years - 10 PLADs, China 2020-2024.

INTRODUCTION

This study aimed to identify the determinants of myopia in children and adolescents across 10 provincial-level administrative divisions (PLADs) and develop a predictive nomogram to aid risk identification.

METHODS

From November 2020 to July 2024, 32,075 students aged 5-19 years were selected using stratified cluster random sampling across 10 PLADs. Myopia screening and questionnaires were used to collect demographic, behavioral, and environmental data. To account for the hierarchical structure of the data, multilevel logistic regression was employed to identify the independent predictors of myopia. A nomogram was constructed based on the regression coefficients, following the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) guidelines. The model performance was evaluated using receiver operating characteristic (ROC) curves, area under the curve (AUC), calibration plots, and decision curve analysis (DCA).

RESULTS

The prevalence of myopia was 29.49% (9,460/32,075). Sex, daily frequency of school eye exercises, homework/reading and writing time after school, time spent on computers, 33-centimeter eye-to-book distance during reading and writing, 3-meter eye-to-TV distance, lying or prone position for reading/screen, parent-controlled game time, and parental myopia were identified as significant predictors. The nomogram demonstrated strong predictive ability, with AUCs of 0.80 (training) and 0.81 (validation), good calibration, and a high net clinical benefit in DCA.

CONCLUSION

The developed nomogram proved effective for identifying high-risk children, supporting personalized risk stratification, and targeting early intervention strategies.