Discussion

Main findings

Preventative value of CS for AI after OASI

This analysis does not demonstrate a protective benefit for routine elective CS in preventing new or worsening symptoms of AI after OASI in a group of women who are both symptomatic and asymptomatic after their index OASI. This was analysed from either a deterioration on a validated scoring system or a description of new or worsening symptoms in women by study authors after a subsequent birth after OASI. There is no evidence of difference in deterioration of AI between women who deliver by CS compared to women who deliver by VB after OASI overall, in the shorter-term or longer-term (Figure 3A-C). For two studies in which an asymptomatic group was analysed, there was also no evidence of a difference in AI symptoms in the short-term; however there are significant limitations to both studies (Figure 3D).
Regarding the certainty of the results, all included studies were at high risk of bias for at least one outcome (Figure 1). This was due to: limitations with the population recruited; non-randomisation to treatment outcomes; inclusion of symptomatic women causing systemic bias; retrospective analysis; failure to analyse by intention-to-treat (emergency CS group included with elective CS group or excluded) and a lack of validated outcome data measurement (Figure 2). We do not have data to counsel specific groups; i.e. symptomatic women, asymptomatic women, women with recurrent OASI.

Long term data

In the longer term, there is evidence from a larger population-based study that women subsequently delivering by VB after OASI may experience worsening of AI symptoms 5-12y after subsequent birth compared to CS; however, this was not significant on the author’s multivariate analysis (12). The study relied on retrospective recall of patient symptoms after index birth. EAUS and ARMS were not used to routinely counsel women regarding subsequent birth; it was national practice for symptomatic women to undergo CS and asymptomatic women to undergo subsequent VB.
The other study providing long-term data outcomes compared two different control groups of 50 patients however this study was not powered to investigate long term AI symptoms between these groups (27). Therefore, both studies in our meta-analysis have significant limitations and we need higher quality data to investigate differences in long-term AI symptoms in asymptomatic women after OASI who deliver by VB or CS.

Recurrence of OASI

Women who sustain an OASI recurrence have poorer long term incontinence outcomes than women who deliver without an OASI (5); studies included in this review did not provide data for this specific outcome. Rates of index OASI are higher in teaching and university hospitals compared to population-based studies and district-general hospitals (DGHs). However, rates of recurrent OASI are highest in DGHs. There is an extremely wide variation in OASI recurrence rates reported across studies, representing a 25-fold difference. This may relate to training in OASI detection and participation in research in some units, differences in OASI detection rates and classification over time, regional differences in obstetric practice and care to avoid recurrence e.g. by using OASI care bundles in some units.

Strengths and limitations

This comprehensive review summarises the current evidence for whether elective cesarean section is protective against the development of AI after OASI. It uses a robust prospectively published systematic methodology to double-extract data and appraise the risk of bias of this data by two independent study authors. This is an extremely important area clinically as AI has a profound negative effect on the lives of young women, OASI is common, and AI is very common after OASI. This area is under-researched; current guidelines utilise level 4 evidence and there is consequently widespread variation in clinical practice.
This study meta-analyses both (1) the total incidence of AI after subsequent VB compared to subsequent CS after OASI and (2) the presence of new or worsening AI after subsequent VB compared to subsequent CS after OASI. To our knowledge, it is the first study to investigate both outcomes. It highlights challenges appraising evidence from non-randomised mode of birth studies due to significant confounding caused by including women with pre-existing symptoms of AI. It is therefore difficult to provide clinically interpretable information on the preventative value of CS against worsening AI from existing published data.

Asymptomatic vs symptomatic women

This review is limited by the methodology of the published data, particularly the ability to extract values for just women who are symptomatic or asymptomatic after index OASI. When counselling women clinically, this is an important factor. Most studies are service evaluations which include both symptomatic and asymptomatic women as one group. Additionally, most published data do not used a validated incontinence score to pair data pre- and post- birth after OASI therefore it is difficult to extract data for women who have ‘new’ or ‘worse’ AI after birth.

Quality of the data

The vast majority of included studies were at high risk of bias for at least one outcome. This was due to: limitations with the population recruited; non-randomisation to treatment outcomes; inclusion of symptomatic women causing systemic bias; retrospective analysis; failure to analyse by intention-to-treat (emergency cesarean group included with elective cesarean group or excluded); and a lack of validated outcome data assessment (Figure 1). All studies except two included women with pre-existing AI symptoms.

Data heterogeneity

One difference between studies is the use of EAUS, ARMS and 3D-TPUS to assess anal sphincter defects and counsel women alongside symptoms to undergo planned VB or planned CS after OASI (Table S1, S4). Defects on EAUS are significantly correlated with symptoms of AI both in women with a history of OASI diagnosed at birth (55% have a persistent sphincter defect and 38% have AI) and in those without (13% have sphincter defects and 14% have AI) (83). We include all women in these meta-analyses however it would be valuable to appreciate the role of EAUS, ARMs and 3D-TPUS in the long term in guiding practice, specifically whether CS is valuable in preventing AI in asymptomatic women with defects/abnormal physiology who undergo subsequent birth.

Deviations from study protocol

Although our protocol aim was to stratify AI by follow up time period (short-term (≤1y), medium-term (>1y <5y), long-term (≥5y)) this was not due to the ranges in follow-up time period employed by particular studies (S4) (12, 31, 33, 34). Instead, we meta-analyses two time periods; 3-24 months (short term) and ≥5y (long term).

Interpretation of results

This systematic review has not demonstrated a protective effect of CS over VB to prevent new or worsening AI symptoms after OASI in the short-term. Data presented in this meta-analysis represents a mixed group of symptomatic and asymptomatic women and nearly all studies were at high risk of bias and short follow up duration. It therefore does not reliably assess an effect of subsequent MOB on long-term AI outcomes, or provide individualised risk assessment for groups of women who are symptomatic or asymptomatic after OASI.
Evidence to support counselling women after OASI is limited. There is a paucity of data on long-term outcomes and knowledge about women’s preferences and decision-making (84). Factors predicting worse AI with subsequent birth after OASI include transient or ongoing AI symptoms after index birth (13), fourth-degree tear (5, 24) and recurrent OASI (4). Factors predicting better AI outcomes with subsequent birth after OASI include absence of AI after index birth (13), normal anorectal physiology (29), lesser degree of injury (24) and absence of injury at subsequent birth (4).

Practical and research recommendations

The evidence included in this review is limited and comes mostly from service evaluations undertaken on mixed groups of women (17, 18, 19, 29, 31, 32, 70, 82). Higher-quality evidence is required to enable clinicians to personalise the risk of future AI based on index birth factors, maternal risk and current pregnancy details in women who deliver after OASI. This should differentiate the risk for asymptomatic and symptomatic women. Research to ascertain the value of anorectal physiology investigations in predicting longer term AI outcomes would be beneficial and enable standardisation of antenatal counselling pathways.