Work

Development and Internal Validation of a Clinical Prediction Model for Rapid Renal Function Decline After Urinary Stone Surgery: A Retrospective Cohort Study

Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background Urinary stone disease and its surgical management may adversely affect renal function, yet validated tools for predicting postoperative rapid renal function decline remain lacking. We aimed to develop and internally validate a clinical prediction model for 1-year postoperative rapid renal function decline in patients undergoing urinary stone surgery. Methods This retrospective cohort study included 3,726 adult patients who underwent urinary stone surgery between January 1 and December 31, 2024, at Renji Hospital, Shanghai Jiao Tong University School of Medicine, with follow-up through December 31, 2025. The primary outcome was rapid renal function decline at approximately 1 year, defined as > = 30% reduction in eGFR from baseline. Candidate predictors were identified a priori based on clinical relevance and prior literature; LASSO logistic regression was used for data reduction, and variables with stable contributions together with clinically established predictors were incorporated into the final multivariable logistic model. Internal validation used 1,000 bootstrap resamples. Model performance was assessed by discrimination (AUC), calibration (intercept and slope), Brier score, and decision curve analysis. Results At 1 year, 306 patients (8.2%) experienced rapid renal function decline. The final model included seven predictors: age, baseline eGFR, diabetes, gout, hyperuricemia, hydronephrosis, and proteinuria. Proteinuria was the strongest predictor (OR = 2.762, 95% CI 2.028–3.761). The apparent AUC was 0.912, with an optimism-corrected AUC of 0.911 (95% CI 0.910–0.912). Bootstrap-corrected calibration intercept was − 0.017 and slope was 0.987 (95% CI 0.900–1.084); Brier score was 0.048. Sensitivity analysis showed the full model provided a modest but statistically significant improvement over baseline eGFR alone (ΔAUC = 0.014). A five-variable parsimonious model (excluding diabetes and gout) achieved comparable discrimination (AUC = 0.912). Conclusion We developed and internally validated a clinical prediction model and nomogram that accurately predicts 1-year postoperative rapid renal function decline after urinary stone surgery. The model may facilitate preoperative risk stratification and guide postoperative surveillance. External validation is warranted before clinical implementation. Trial registration Not applicable.
Language: English
Subfield: Urology
Field: Medicine
Open Access status: green
OpenAlex ID: W7236750437
Cites: 30
Cited by:
Related to: 10