Chronic Kidney Disease Prediction in Different Populations Using Routine Urine Test: A Multi-Center Study

Objectives This study aimed to assess the application value of routine urine tests for predicting chronic kidney disease (CKD) across diverse populations. By developing and validating a predictive model using samples from a multi-center dataset, the study aimed to improve early CKD screening. Methods Urine samples were collected from 3,000 patients at West China Hospital and an external multicenter cohort of 3,856 individuals. Routine dipstick and microscopic urine tests were conducted, and key predictors were identified using LASSO regression. A multivariate logistic regression model was developed, and its performance was assessed through receiver operating characteristic (ROC) curve analysis, calibration curves, and decision curve analysis. Results The study identified 11 significant predictors of CKD: RBCs, Mucus threads, UBG, KET, BLD, PRO, LEU, PH, MALB, CA, and Osmolality. The predictive model demonstrated excellent performance with an AUC of 0.849, indicating high discriminatory power. Validation results confirmed the model's robustness and generalizability across diverse populations with the AUC = 0.854 for internal validation cohorts and AUC = 0.848 for external validation cohorts. Conclusion The developed predictive model provides a reliable, non-invasive tool for CKD screening using routine urine tests. Its high accuracy and scalability make it suitable for integration into clinical workflows, facilitating early intervention and improving patient outcomes.