Identification of risk factors and construction of a predictive model for postoperative new-onset stress urinary incontinence in patients with pelvic organ prolapse: A single-center retrospective study.

This study was to identify independent risk factors for new-onset stress urinary incontinence (SUI) after pelvic organ prolapse (POP) surgery and to develop and validate optimized prediction models, thereby providing an evidence-based tool for clinical decision-making. A single-center retrospective cohort study was conducted, including 213 patients who underwent POP surgery. Data on baseline characteristics, preoperative assessments, surgical details, and 12-month postoperative follow-up were collected via the electronic medical record system. Multiple imputation was used to handle missing data, least absolute shrinkage and selection operator (LASSO) regression for feature selection, and both logistic regression and classification and regression tree (CART) decision tree models were developed. LASSO regression identified preoperative POP-Q stage, abnormal urodynamics with prolapse reduction, postmenopausal duration >10 years, and parity as core risk factors. The logistic regression model achieved an internally validated area under curve (AUC) of 0.78 and accuracy of 83.0%, with external validation AUC of 0.75. After calibration, Hosmer-Lemeshow test yielded P=0.65. The CART decision tree model demonstrated an internally validated AUC of 0.84 and accuracy of 87.4%, with external validation AUC of 0.81 and accuracy of 85.2%. Preoperative POP-Q stage and abnormal urodynamics with prolapse reduction are the strongest predictors of new-onset SUI after POP surgery.