Anal incontinence (AI), a debilitating outcome of obstetric anal sphincter injuries (OASIs), significantly impacts quality of life of women affected by the latter. The early diagnosis of OASIs is particularly challenging as anorectal symptoms, in particular AI, can be subtle or not yet developed shortly after delivery, as indicated by largely retrospective literature. This discrepancy underscores the urgent need for robust, prospective data to leverage both diagnosis and management of OASIs. Addressing this critical gap, this study evaluated the correlation between early postpartum AI symptoms and results of diagnostic tests addressing the integrity of anal sphincter complex or its function, the former including the novel Machine Learning-supported electric impedance spectroscopy (ONIRYTM OASIS Diagnostic) test).

This exploratory analysis was performed within a prospective, comparative clinical study on detection of OASI conducted at four European labour wards and one outpatient maternal care clinic in 152 women shortly (up to 5 weeks) after vaginal delivery, enriched for women with OASI (n=60) as determined by endoanal ultrasound (EAUS) (the primary results thereof pending publication). AI symptoms were analysed using Wexner score, not earlier than 3 days postpartum. The correlation analysis for Wexner total score was evaluable and performed using Spearman's correlation tests for 54, 54, 53, and 46 women with OASI and 62, 62, 62, and 55 women without OASI for results of digital rectal examination (subjectively suggestive of OASI), electric impedance spectroscopy (ONIRYTM test result compatible with OASI), EAUS (grades 3 and 4 per OASIS classification), and high-resolution anorectal manometry (major finding per London Classification by International Anorectal Physiology Working Group).

The results of the correlation analysis are shown in the table 1. 
		
The main clinical study had two parts: the clinical conduct and the in silico part; during the latter the Machine Learning model was trained using 10 iterations of 10-fold cross-validation. Therefore, in case of ONIRYTM test, the correlation coefficient was calculated as a mean value obtained from 10 iterations of each cross-validation, thus there is no single corresponding p-value; the one presented in the table is the highest of the ten (still being statistically significant).

The analysis revealed significant correlations between the Wexner total score and all four diagnostic methods evaluated. Notably, the strongest correlation was with the physical digital rectal examination whereas the correlation coefficient for the other 3 methods was less strong, including the anorectal manometry. This may be due to the fact that the digital rectal examination, known for its limited sensitivity for detection of OASI, is able to reliably detect mostly the cases with markedly reduced anal sphincter tonus, which are compatible with FI. The somewhat lower correlation with anorectal manometry abnormal results, despite the latter being the gold standard method for anal sphincter dysfunction, may illustrate the latency between the actual, early dysfunction measurable by this highly sensitive and specific method and the development of clinically meaningful FI symptoms. The correlation with EAUS and ONIRYTM test, both statically evaluating the integrity of perianal structures, at a comparably lower level, may also illustrate the latency between the injury (OASI) and development of dysfunction (FI); the latter frequently not being immediate.

This exploratory study provides data indicating that the early postpartum AI symptoms are suggestive of OASI but with only moderate level of correlation. Knowing the low sensitivity of digital rectal examination for detection of OASI as such, this data underscores the necessity for a multifaceted diagnostic strategy considering both traditional and innovative methods to address the complexities of OASI detection effectively. While EAUS is of marginal accessibility at obstetric settings as a screening tool for OASI, explored for this purpose should be the novel, more accessible methods such Machine Learning-supported electric impedance spectroscopy provided with the ONIRYTM test.

Continence 12S (2024) 101649
DOI: 10.1016/j.cont.2024.101649