External validation of the de novo stress urinary incontinence prediction model after pelvic organ prolapse surgery in Korean women: a retrospective cohort study

Background De novo stress urinary incontinence (SUI) may develop following pelvic organ prolapse surgery. Performing prophylactic continence surgery may reduce the risk of de novo SUI and subsequent continence surgery; however, it may increase the risk of complications. Therefore, many surgeons try to identify women at high risk for de novo SUI and perform continence surgery selectively. Recently, a model for predicting the risk of de novo SUI after prolapse surgery was developed using data from the Outcomes following vaginal Prolapse repair and midUrethral Sling (OPUS) trial; its prediction accuracy was significantly better than that of the stress test alone. However, few studies have verified its prediction accuracy in discrete populations. The aim of this study was to externally validate the prediction model for de novo SUI after prolapse surgery in Korean women. Methods This retrospective cohort study included 320 stress-continent women who underwent prolapse surgery for pelvic organ prolapse quantification stage 2–4 anterior or apical prolapse and who completed a 1-year follow-up. Predicted probabilities by the de novo SUI online risk calculator were compared with observed outcomes and quantitated using the model’s area under the curve and calibration plot. Subgroup analyses were also performed by the type of prolapse surgery. Results The de novo SUI prediction model showed moderate discrimination in our study cohort; area under the curve (95% confidence interval) = 0.73 (0.67–0.78) in the whole cohort, 0.69 (0.61–0.78) in women who underwent native tissue repair or colpocleisis, and 0.74 (0.65–0.82) in those who underwent sacrocolpopexy. Calibration curves demonstrated that the model accurately predicted the observed outcomes of de novo SUI in women who underwent native tissue repair or colpocleisis but underestimated outcomes in those who underwent sacrocolpopexy. The predicted probability cutoff points corresponding to an actual risk of 50% were 40% in women who underwent native tissue repair or colpocleisis and 30% in those who underwent sacrocolpopexy. Conclusions The de novo SUI prediction model is acceptable for use in Korean women and may aid in shared decision-making regarding prophylactic continence procedure at the time of prolapse surgery.