External Validation of a Prediction Model to Assess the Risk of Severe Hemorrhage After Mini Percutaneous Nephrolithotomy: A Retrospective Cohort Study

Objective: Predicting hemorrhage risk after mini-percutaneous nephrolithotomy (mini-PCNL) is of significant clinical importance. This study aimed to externally validate an existing prediction model developed to estimate the risk of severe hemorrhage following mini-PCNL. Material and Methods: Data of patients who underwent mini-PCNL between 2013 and 2023 were retrospectively analyzed. Demographic data, renal anatomy, stone characteristics, operative data, and complications were recorded. Severe hemorrhage was defined as a hemoglobin decrease greater than 34.5 g/L or bleeding that required transfusion or angioembolization. Patients were categorized according to hemorrhage status, and intergroup comparisons were performed. Hemorrhage risk for each patient was calculated using the prediction model. The model's performance was evaluated using receiver operating characteristic (ROC) analysis and calibration curves. Results: Among 826 patients, postoperative hemorrhage occurred in 87 (10.5%). The mean stone size was significantly larger in the hemorrhage group (p=0.036). The rate of urinary tract infection was higher in the hemorrhage group compared to the non-hemorrhage group (59.8% vs. 47.5%, p=0.030). Hydronephrosis was more frequent in the non-hemorrhage group (29.0% vs. 18.4%, p=0.038). The mean operative time was significantly longer in the hemorrhage group (125.6±41.5 vs. 104.3±41.6 minutes, p=0.001). The mean number of accesses was also higher in the hemorrhage group (p=0.003). The prediction model demonstrated acceptable discriminatory performance, with an AUC of 0.703 (p=0.001, 95% CI: 0.645-0.762). Conclusion: This study externally validates the predictive accuracy of the current model for estimating severe hemorrhage risk after mini-PCNL and supports its potential utility in clinical practice.