An ultrasound model for stress urinary incontinence at six weeks postpartum
Xia Zhang, Yang Long, Shan Liu, Limin Liu, Fenghuang Zhou, Wenjie Liu, Bing ZhouAbstract
Objectives
Stress urinary incontinence (SUI) after childbirth has both clinical risk factors and measurable changes in urethral support. Whether integrating these complementary features can identify affected women at six weeks postpartum remains uncertain.
Methods
This retrospective, single-centre study included 454 women, divided chronologically into development (n=351) and temporal internal-validation (n=103) cohorts. LASSO selected ultrasound predictors. Ultrasound-only and combined clinical–ultrasound models were evaluated for discrimination, calibration and clinical utility.
Results
SUI prevalence was 17.7 % and 26.2 % in the development and validation cohorts, respectively. LASSO retained bladder-neck mobility (BNM) and urethral funneling; the combined model additionally included age, vaginal delivery and overweight. The combined model outperformed the ultrasound-only model in development (C-statistic, 0.780 [95 % CI, 0.717–0.843] vs. 0.703 [0.634–0.771]; p=0.002). In validation, the corresponding C-statistics were 0.805 (0.716–0.894) and 0.737 (0.633–0.841), but the difference was not significant (p=0.141). The validation calibration intercept, slope and Brier score were 0.704, 1.054 and 0.161, respectively. Decision-curve analysis indicated potential net benefit at selected thresholds; continuous-predictor analysis yielded similar discrimination.
Conclusions
Combining clinical and transperineal ultrasound features improved concurrent SUI identification over ultrasound features alone. External validation is required before clinical implementation.