Development and internal validation of a prediction model to evaluate the risk of severe hemorrhage following mini-percutaneous nephrolithotomy.

PURPOSE

To develop a new prediction model for assessing the severe hemorrhage events in post mini-percutaneous nephrolithotomy (mini-PCNL) patients and internally validate it, thus to guide decision making in clinical practice.

METHODS

The patients who underwent mini-PCNL were retrospectively reviewed. Potential risk factors were included as prediction variables for multivariate logistic regression analysis to identify independent risk factors, and prediction model was constructed. The predictive ability of the model was evaluated using the Concordance index (C-index) and Brier score. Bootstrapping resampling technique was used to perform internal validation. The related packages in R were used to generate the web application based on the prediction model.

RESULTS

Multiple-tract was the strongest predictor of severe hemorrhage following mini-PCNL. Other risk factors were none or mild hydronephrosis, congenital anomalies of urinary system, urinary tract infection, operation time and stone peak Hounsfield unit. A prediction model was constructed to assess the probability of severe hemorrhage after mini-PCNL. The C-index and Brier score were 0.731 and 0.093, respectively after correcting for optimism, which signified the excellent discrimination and calibration.

CONCLUSION

A new prediction model was developed to estimate risk of severe hemorrhage after mini-PCNL. It had been internally validated with good discrimination and calibration. The prediction model might be beneficial for endourologists in surgical decision-making and risk aversion.