初产妇会阴二至四度裂伤后产后1年肛门失禁的多变量预测模型开发与验证

原文

ABSTRACT Objective To develop and validate a prediction model estimating the individual risk of anal incontinence 1 year postpartum following second-, third- and fourth-degree perineal tears. Design Development of a prognostic prediction model. Setting Data were retrieved from three Swedish health and quality registers. Population 8909 primiparous women with second-, third- and fourth-degree perineal tears who delivered vaginally at ≥ 34 weeks' gestation between 2014 and 2019. Methods Seventeen predictors were identified as candidate variables. A prediction model was built using logistic regression to predict anal incontinence at 1 year postpartum. Additional models were built for gas and faecal incontinence. Model performance was assessed using the Efron-Gong optimism bootstrap. Temporal validation was conducted in a separate cohort of women who delivered between 2020 and 2022. Main Outcome Measure Anal incontinence (gas and/or faecal incontinence) at 1 year postpartum. Results Seven variables remained in the final model for anal incontinence. A high Wexner score was the strongest predictor of anal incontinence at 1 year postpartum. The models accurately predicted anal incontinence (AUC 0.78, 95% CI: 0.76–0.80), faecal incontinence (AUC 0.80, 95% CI: 0.77–0.83), and gas incontinence (AUC 0.78, 95% CI: 0.76–0.80). Calibration curves demonstrated that all models were accurate, particularly for women with third-degree tears. Temporal validation confirmed the high accuracy (AUC 0.78; 95% CI: 0.75–0.81). Conclusions The article presents an accurate prediction model for anal incontinence at 1 year after vaginal delivery among women with second-, third- and fourth-degree perineal tears, based on perineal laceration register data collected at 8 weeks postpartum.