External validation of a model predicting de novo stress urinary incontinence after pelvic organ prolapse surgery

De novo stress urinary incontinence (SUI) may develop after surgical correction of advanced pelvic organ prolapse (POP) in otherwise continent women. Prediction of which women with POP will develop SUI after the prolapse is corrected is difficult. We aimed to externally validate a previously described prediction model for de novo SUI after performing vaginal surgery for POP and to assess its clinical performance when used as a diagnostic test.

This retrospective cohort study included all continent women with ≥ stage 2 POP according to the POP-Quantification System who underwent reconstructive surgery for symptomatic POP. Surgical correction for prolapse of the anterior and/or apical compartment was performed using native tissue or vaginal mesh repair. Seven parameters of the prediction model including age at surgery, number of vaginal births, body mass index, preoperative stress test, previous continence procedure history, urine leakage associated with a feeling of urgency, and diagnosis of diabetes for each patient was provided from the medical records, and the predicted probability of de novo SUI after POP surgery was calculated. The primary outcome used to validate the prediction model was the presence of SUI 1 year after surgery. A receiver operating characteristic (ROC) curve was generated to evaluate the predictive accuracy. A cut-off point of ≥ 50% was used to evaluate its clinical performance as a diagnostic test.

Two hundred twenty-five women were suitable for analysis. The rate of de novo SUI was 5.3%. The predictive accuracy of the model in our population using the area under the ROC curve was 0.56 (95% confidence interval = 0.35-0.77). Its performance as a diagnostic test was poor (positive likelihood ratio = 1.20 and negative likelihood ratio = 0.89).

Our clinical validation of this model showed that it did not have good clinical performance. We need future prospective studies to identify and incorporate additional markers of de novo SUI to improve the prediction capacity.