Development and External Validation of a Risk Prediction Model for Hyperuricemia in Adolescents: A Multicenter Clinical Study

Hyperuricemia is increasingly prevalent among adolescents and is associated with metabolic disorders and future cardiovascular risk, yet risk prediction tools for this population remain limited. We developed and externally validated a multivariable prediction model for adolescent hyperuricemia using routinely collected health examination data. A total of 903 adolescents aged 12–18 years from two centers in Chongqing, China, were included, with 747 individuals for model development and 156 for external validation. Candidate variables were screened by univariate analysis and incorporated into a multivariable logistic regression model using Akaike information criterion -based selection. Model performance was evaluated in terms of discrimination, calibration, Brier score, and clinical utility using decision curve analysis. Eight predictors were retained, including body mass index, alkaline phosphatase, urea, red blood cell count, hemoglobin, γ-glutamyl transferase, absolute monocyte count, and absolute neutrophil count. Discrimination was maintained across datasets, with AUCs of 0.831 in development and 0.804 in external validation, while Brier scores were 0.142 and 0.194, respectively, indicating overall adequate model performance. Decision curve analysis further suggested net clinical benefit across low-risk thresholds. The model provides individualized risk estimates and may facilitate early identification and risk stratification of hyperuricemia in adolescents.