[Model predicting failure in surgical repair of obstetric vesicovaginal fistula].

Introduction: there are more than 2 million cases of genital fistula in sub-Saharan Africa and in Asia. They occur in people living in areas where the access to medical care during pregnancy and childbirth are limited or of poor quality and where few hospitals can provide adequate surgical repair. The purpose of this study was to develop a score to predict the factors influencing failure in surgical repair of obstetric vesicovaginal fistula (VVF) in the southeastern part of the province of the Upper Katanga. Methods: we conducted a cross-sectional analytical study of 384 women with vesicovaginal fistula who had undergone surgery. We performed a multivariate and then a univariate analysis. The discrimination of the score was assessed using the ROC curve and the C-index while the calibration of the score using Hosmer-Lemeshow's test. Results: surgical repair of obstetric vesicovaginal fistula failed in 1719% of cases (66/384). After logistic modelling, four criteria emerged as predictors of failure in surgical repair of vesicovaginal fistula; the presence of a fibrotic scar (OR=15,22; CI 95%; 7.34-31.58), the presence of 2 fistulas or more (OR=7,41; CI 95%; 3.05-17.97), transvescical approach (OR=4,26; CI 95%; 1.92-9.44) and urethral involvement (OR=3,93; CI 95%; 1.99-7.77). The area under the ROC curve for the score was 0.8759, with a sensitivity of 57.58%, a specificity of 91.82% and a positive predictive value of 91,25%. Conclusion: this study demonstrated that the number of fistulas, the presence of fibrotic scar, urethral involvement and transvescical approach are predictors of failure in surgical repair of vesicovaginal fistula.