Hypothesis / aims of study
This is the first study to derive a validated clinical algorithm for uncomplicated vesicovaginal fistula (VVF) repair route selection using prospectively-defined anatomical predictors. Traditional teaching mandates transabdominal repair for supratrigonal fistulas, yet no evidence-based algorithm exists [1]. We aimed to: (i) identify measurable anatomical determinants of surgical route selection, prioritising vaginal accessibility over cystoscopic fistula location; (ii) derive a data-driven algorithm using CART analysis; and (iii) evaluate interposition flap use as a predictor of surgical outcome — an association not previously examined in comparative VVF literature [2,3].
Study design, materials and methods
A retrospective cohort study of 84 consecutive patients undergoing uncomplicated VVF repair (January 2018–June 2025) was conducted; complicated cases were excluded so route selection reflected anatomical accessibility. Route selection followed a structured access-first protocol: vaginal space (2-finger test: tight vs loose) and distance from the vaginal introitus were primary determinants; cystoscopic fistula location was secondary. Transabdominal repair (n=56) used robotic-assisted modified O'Conor technique; transvaginal repair (n=28) used the Latzko technique. Demographics, fistula characteristics (location, Goh classification [2], size, number, introitus distance, ureteric orifice distance), operative details, and flap use were recorded. The primary outcome was absence of urinary leakage at four weeks post-catheter removal. CART analysis derived the algorithm; ROC analysis established the distance cutoff. Univariate predictors were assessed by Fisher's exact test and logistic regression. With 8 leakage events, statistical power was limited (~18%), acknowledged as a limitation. Ethical approval and informed consent were obtained.
Results
Groups were comparable across all baseline variables including age (38.2 vs 38.0 years; p=0.342), fistula size (p=0.180), and Goh classification (p=0.722). Vaginal space (tight: 89.3% transabdominal vs 7.1% transvaginal; p<0.001) and distance from introitus (6.72 vs 5.21 cm; p<0.001; Cohen's d=1.69) were the dominant determinants of route selection. The CART algorithm achieved AUC 0.927 (accuracy 91.4%), with vaginal space as the primary split and distance from introitus as secondary (cutoff 6.25 cm; standalone AUC=0.883). Seven supratrigonal fistulas were repaired transvaginally with 85.7% success (6/7), refuting that supratrigonal location mandates transabdominal repair. Four transvaginal repairs involved multiple fistulas, all successful — confirming multiplicity is not a contraindication when access is adequate (p=0.242). Overall four-week success was 90.5% (transabdominal 92.9% vs transvaginal 85.7%; p=0.431). Interposition flap use was the strongest modifiable predictor of success: 0.0% leakage (0/33) with flap vs 15.7% (8/51) without (p=0.020). Younger age (OR=0.878; p=0.046) and loose vaginal space (OR=0.173; p=0.049) were also associated with leakage. All 84 patients had Clavien-Dindo Grade 0 complications.
Interpretation of results
The two-step access-first algorithm — vaginal space first, then distance from introitus (cutoff 6.25 cm) for tight cases, with cystoscopic location as contextual only — achieves 91.4% accuracy (AUC=0.927), directly challenging two dogmas: that supratrigonal location and fistula multiplicity mandate transabdominal repair. Both were refuted. The zero leakage rate with interposition flaps across both routes is novel and clinically significant, suggesting flaps may confer complete protection against failure regardless of approach [3]. Study limitations include retrospective design, 2:1 group imbalance, and only 8 leakage events precluding multivariable analysis. In uncomplicated VVF, interposition flap use should be considered universally, and prospective multicentre validation of both the algorithm and flap-outcome association is warranted.