Discussion
Main Findings : In this randomized trial of women with asymptomatic anal sphincter lesions resulting from a first delivery, planned cesarean section for the second delivery was not protective against anal incontinence at 6 months post-partum. In addition, we failed to show any benefit of cesarean section in this population on quantitative evaluations of urinary incontinence, sexual functions or quality of life. These results should be useful for clinicians and women, avoiding numerous unnecessary cesarean sections12.
Results vis-à-vis other studies : Our findings are consistent with those from recent observational studies, including longer follow-up16,23. In a recent observational cohort study, CS for women with anal sphincter disruption at the first delivery was associated with no benefit on anal incontinence 5 years after the second delivery 16. However, most women who sustained obstetric injuries develop anal incontinence later, after their 50s. Nygaard et al 24 found that anal sphincter disruption following an index delivery was a risk factor for flatus incontinence 30 years later. Some large population-based cohort studies failed to show any difference in the incidence of flatus incontinence in women above 50 years old, according to whether they delivered vaginally or by CS17, but a recent population-based study from Sweden found that the risk of anal incontinence was lower after CS than after VD 25 . In this study, AI also was higher among women who delivered by CS compared with nulliparous and higher among nulliparous compared with men. In another study, an association has been found between ultrasound diagnosis of anal sphincter lesions and long-term fecal incontinence after a first delivery26. Because anal incontinence is multifactorial, including neurological and gastro-intestinal as well as mechanical causes, this symptom can occur without sphincter lesions and vice versa. Anal sphincter lesions are observed by ultrasound in less than half of women with postpartum anal incontinence 8. In an unselected primiparous population, anal sphincter disruption was detected by ultrasound screening after delivery in 27 % of women, most of whom had no symptoms 27. Thus, although CS can be protective from anal sphincter lesions 8, 9, ultrasound evidence is one of many factors associated with anal continence. Besides, it has been shown that the severity of the anal sphincter lesion is an important risk factor for subsequent anal incontinence, particularly the depth of the disruption of both the external and internal sphincter 28. In our trial, we did not observe a protective effect of CS in the subgroup with severe anal sphincter ruptures (defined as >90°)(data not shown). In another study, only 4th-degree tears were associated with an increased risk of anal incontinence at 10 months postpartum29, but this was an exclusion criterion in our trial.
One important difference between our study and most retrospective studies was the inclusion of women whose first delivery was by forceps, even in the absence of a diagnosis of a third-degree perineal laceration.
Some observational studies have shown that a subsequent vaginal delivery following an obstetrical anal sphincter injury may result in additional or recurrent lesions 30, which may be apparent or occult, however without any significant change in the continence score according to the mode of delivery. In our trial, the incidence of repeated clinically apparent OASIS was low, since only one woman had a repeated 3rd degree tear.
Endosonographic aggravation of external sphincter lesions occurred significantly more often in the VD group than the CS group. These findings may indicate that CS avoids some occult sphincter disruptions, but on the other hand they signify that ultrasound evidence of anal sphincter lesions is not predictive of symptoms of anal incontinence.
Clinical implications: Our results do not support a policy of systematic prophylactic CS in women with asymptomatic ultrasound anal sphincter lesions resulting from a first delivery. However, we cannot exclude a protective effect of prophylactic CS for women with symptomatic anal sphincter lesions. In a subgroup analysis, we did find a significant benefit of CS among women with mild clinical anal incontinence detected before the second delivery at the proctological visit. Since it is a post-hoc analysis, it must be interpreted with caution. Our findings underline the importance of correct diagnosis of anal incontinence.
Because of the taboos surrounding anal incontinence, it is difficult to reveal without meticulous questioning. In our study, 27 women self-reported no anal incontinence at inclusion, but had a Vaizey score ≥5 at the visit with a proctologist. Comparatively to endosonography, clinical-based diagnosis of anal incontinence is less expensive, more accessible and appears to be more predictive of functional outcome, as has been previously suggested in retrospective studies16,23. Thus, our findings do not support performing anal endosonography for women with an overt OASI or forceps instrumentation for their first delivery in order to decide on the mode of delivery.
Research Implications : Further studies are needed to determine whether CS may be useful in the long term, among women with mildly symptomatic anal lesions, and if so whether women with third- or fourth-degree perineal tears and/or forceps at their first delivery can benefit from a proctological examination in order to make a decision regarding their subsequent deliveries.8
Strengths/Limitations: This trial has several strengths. To our knowledge, it is the first randomized controlled trial addressing this issue. Also, anal incontinence was assessed with a standardized, validated and widely used score, the Vaizey score, and sphincter lesions were defined by endosonography. Third, the external validity was supported by the diversity of trial settings, including teaching hospitals and general hospitals in diverse populations from neighborhoods ranging from poor to affluent, with no center effect.
This trial also has limits. It was necessarily unblinded, and the main outcomes were patient-reported, thus we cannot exclude patient reporting bias. However, all investigators were unaware of aggregate outcomes during the study. Also, crossovers were observed (cesareans in the vaginal delivery group and vice versa), as expected in this pragmatic trial comparing strategies (planned CS or planned VD). Lastly, one fifth of the randomized women did not complete the 6 months post-partum follow-up, which could lead to attrition bias; however their characteristics did not differ between the two study groups, and multiple imputations were performed for missing variables and comparisons were based on all the patients randomized.