Cesarean delivery to prevent anal incontinence: a systematic review and meta-analysis.

Abstract

Cesarean delivery (CD), is increasingly recommended as a mode of delivery that prevents the anal incontinence (AI) that arises in some women after vaginal delivery (VD). The assessment of the efficacy of CD in this regard was the subject of this systematic review. Searches were conducted in Medline, EMBASE and the Cochrane Library. Both randomized (RCTs) and non-randomized trials (NRTs) comparing the risk of sustained fecal and/or flatus incontinence after VD or CD were sought from 1966 to 1 January, 2019. Studies were eligible if they assessed AI more than 6months after birth, and had statistical adjustment for at least one of the three major confounders for AI: age, maternal weight or parity. In addition, each study was required to contain more than 250 participants, more than 50 CDs and more than 25 cases of AI. Data after screening and selection were abstracted and entered into Revman for meta-analysis. Analyses were done for combined fecal and flatus incontinence (comAI), fecal incontinence (FI), gas incontinence (GI), CD before or during labor, time trend of incontinence after delivery, assessment of both statistical and clinical heterogeneity, parity and late incident AI. Out of the2526 titles and abstracts found, 24 eligible studies were analyzed, 23 NRTs and one RCT. These included women with 29,597 VDs and women with 6821 CDs. Among the primary outcomes, VD was found not to be a significant predictor of postpartum comAI compared to CD in 6 studies, incorporating 18,951 deliveries (OR = 0.74; 0.54-1.02). VD was also not a significant predictor of FI in 14 studies, incorporating 29,367 deliveries, (OR = 0.89; 0.76-1.05). VD was not a significant predictor of GI in six studies, incorporating 6724 deliveries (OR = 0.96; 0.79-1.18). The strength of the grading of recommendations, assessment, development and evaluations (GRADE) evidence for each of these was low for comAI and moderate for FI and GI (upgrade for lack of expected effect). Time trend FI showed incontinence at 3months often resolved at 1year. Other secondary analyses assessing parity, delayed incidence of FI, clinical and statistical heterogeneity, spontaneous VD only, late risk of incidence of AI, and CD in or prior to labor all had similar results as in the primary outcomes. There are three components of pelvic floor dysfunction that are thought to be caused by VD and hopefully prevented by CD: AI, urinary incontinence and pelvic floor prolapse. Of these, AI was not found to be reliably prevented by CD in this review.