Establishment and Validation of a Risk Prediction Model for Stress Urinary Incontinence in Pelvic Organ Prolapse：A simple screening tool

Abstract

Abstract Background pelvic organ prolapse (POP) combined with stress urinary incontinence (SUI) has varying impacts on patients' quality of life. Due to the neglect of SUI symptoms by both patients and some doctors, as well as the diverse and complex diagnostic methods for SUI, there is significant variation in the reported incidence of SUI. This often leads to missed diagnoses and misdiagnoses of SUI, resulting in delayed treatment and future implications on patients' lives and socioeconomic factors. Therefore, improving awareness and early identification and diagnosis of SUI in POP patients is crucial. Currently, there is no clinical risk prediction model available for POP with SUI. Objective This study aimed to evaluate the general condition, pelvic floor muscle function, and quality of life in women with pelvic organ prolapse. It aimed to explore the independent influencing factors of stress urinary incontinence in women with pelvic organ prolapse and establish and validate a risk prediction model for pelvic organ prolapse accompanied by stress urinary incontinence. The goal was to provide a simple self-screening tool for SUI in women with POP. Methods A total of 1242 patients treated at the Pelvic Floor Center of the Seventh Affiliated Hospital of Sun Yat-Sen University from January 2021 to December 2021 were included in the study. Data on general information, pelvic floor electromyography, and pelvic floor questionnaires were collected. After data screening and processing, a modeling dataset comprising 1165 patients with POP-Q scores of Ⅰ-Ⅳ degrees was selected. Additionally, data from patients admitted to the same hospital from January 2022 to April 2022 were collected as an external validation dataset. SPSS 26.0 was used for clinical characteristic analysis of the modeling dataset, and univariate analysis was performed to identify independent influencing factors of POP with SUI. LASSO regression analysis in RStudio software (based on R version 4.2.2) was used to screen variables, and multivariate logistic regression analysis was conducted to establish the POP and SUI risk prediction model. Receiver operating characteristic curves (ROCs) were calculated. Based on the established risk prediction model, a nomogram was developed, and its fitting ability was evaluated using C-Statistic (AUC) for model differentiation and the Hosmer-Lemeshow test for consistency. Clinical Decision Curve Analysis (DCA) was conducted to assess the threshold probability of net income for the model. Results LASSO regression analysis identified five predictors (weight, pregnancy, vaginal delivery, I-QOL, and PFDI-20) from the 17 variables studied. The model constructed using these five predictors exhibited moderate predictive ability, with an area under the ROC of 0.755 in the training set, 0.727 in the internal validation set, and 0.833 in the external validation set. The DCA curve demonstrated that the nomogram could be applied clinically when the risk threshold ranged from 26–82%, which was validated externally as ranging from 24–97%. Conclusion SUI in POP can be accurately predicted using the number of vaginal births, the number of gravidity, weight, I – QOL and PFDI – 20 as predictors. These predictions can guide the selective implementation of SUI prevention strategies.