Validation of new ultrasound algorithm for estimating prevalence of anal sphincter trauma in a urogynecological population

Objectives: To estimate the prevalence of major perineal trauma in a urogynecological population, to test the predictive value of sonographic tear grading (Gillor algorithm) for anal incontinence (AI), AI bother score and St Mark's score, and to compare the predictive power of the Gillor algorithm with that of the residual-defect method. Methods: This was a retrospective study of 721 women attending a tertiary urogynecology unit between February 2019 and May 2021. All women underwent a standardized interview, including determination of St Mark's score and visual analog scale (VAS) bother score for AI, as well as exoanal (translabial) ultrasound with later offline analysis. Results were reported as the presence of a residual defect of the external anal sphincter (EAS), i.e. a discontinuity of ≥ 30° in ≥ 4/6 tomographic slices, and according to the Gillor algorithm (normal, Grade 3a, Grade 3b or Grade 3c/4). Results: Mean age at assessment was 57 (range, 19–93) years and mean body mass index was 30 (range, 17–57) kg/m2. Six hundred and thirty-six (88.2%) women were vaginally parous and 161 (22.3%) had undergone at least one forceps delivery. AI was reported by 186/721 (25.8%) women, with a median St Mark's score of 10 (interquartile range (IQR), 6–14) and a median VAS score of 6.3 (IQR, 3.9–10). EAS defects were detected in 261 (36.2%) women, with a residual defect diagnosed in 88 (12.2%). On sonographic grading according to the Gillor algorithm, we identified 532 (73.8%) women with a normal sphincter, 66 (9.2%) with Grade-3a tear, 87 (12.1%) with Grade-3b tear and 36 (5.0%) with Grade-3c/4 tear. In total, the Gillor algorithm classified 189 (26.2%) women as having suffered a major perineal tear. The two grading systems were in moderate agreement (κ, 0.537 (95% CI, 0.49–0.56); P <0.001). There were weak, albeit significant, associations between EAS defects and measures of AI (P = 0.009 to P = 0.047), both for residual defect as well as the Gillor algorithm. Conclusion: Neither the Gillor algorithm nor the residual-defect method of quantifying sphincter trauma on imaging is clearly superior in terms of predicting AI.