Method
Systematic review and meta-analysis were performed according to our
prospectively published protocol (PROSPERO CRD42022372442) using PRISMA
guidelines (20). Authors were contacted to provide data in a suitable
format for meta-analysis if not supplied. Core outcome sets for OASI
have not been published, but there are published protocols in
development seeking to establish this for OASI (21), perineal trauma
(22) and the effects of episiotomy and OASI (23). Patients and the
public were not directly involved in conducting this review.
Eligibility criteria
Participants/Population:
All women >18y who have had an OASI and a subsequent birth
by any mode. OASI is defined as a 3rd or 4th degree tear by the Sultan
classification (24). Inclusion and exclusion criteria are described
(S2).
Intervention:
Planned elective CS in a subsequent birth after an OASI, analysed by
intention-to treat. Observational studies comparing vaginal birth to all
CS (emergency and elective) included with risk of bias assessment.
Comparison:
Planned vaginal birth. Emergency CS was analysed with vaginal birth
outcomes via intention-to-treat when possible. When data were not
provided (i.e. when emergency CS was included with planned CS) the study
is included with risk of bias assessed.
Outcomes:
Our primary outcome is AI (solid, liquid, flatus) after subsequent birth
by any mode; including but not limited to; St Marks Score, Wexner score,
EPAQ, clinical description. AI outcome is stratified by time if
available: ≤2y and ≥5y.
Subgroup analysis for patients asymptomatic after index birth.
Our secondary outcomes are: QOL, regret with MOB choice, satisfaction
with MOB choice, fecal urgency, obstructive defecation, change in AI
pre- and post- subsequent birth, AI in women after subsequent birth vs
no subsequent birth, repeat OASI rate and adverse events (return to
theatre, organ injury).
Measures of effect
Dichotomous outcomes are presented as risk ratio/odds ratios with 95%
CI. Continuous outcomes presented as mean difference with 95% CI.
Search strategy
MeSH subject heading and database-specific truncated search terms were
used to avoid excluding potential studies. Three concepts were
implemented for the search strategy: (1) identification of all types of
anal sphincter or anal canal injury, (2) identification of all
subsequent birth by any mode and (3) identification of all types of anal
incontinence. The search terms and strategy are detailed (S1).
Information sources
Search strategy was executed in Ovid MEDLINE/PubMed 1946-2022 (Ovid),
Embase 1974-2022 (Ovid), CINAHL(EBSCOhost) 1937-2022, Cochrane Combined
1996-2022, Clinical trials and Google Scholar from inception to
1st December 2022.
Study selection
Setting
We included randomised-controlled trials and non-randomised studies
including observational, cohort, cross-sectional, case-control and
retrospective reviews in any publication format which provided data for
our outcomes of interest for a birth after OASI (Table S1).
Data extraction (selection and
coding)
All title and abstract screening (EC, RH), full text review (EC, RH),
data extraction (EC, KA, RH) and risk of bias assessments (EC, RH), were
completed by two blinded independent reviewers according to standard
Cochrane methodology. Screening took place using Rayyan review software
(25). Discrepancies were resolved by a third reviewer (RK). We describe
characteristics of included studies in Table S1.
Risk of bias assessment
Risk of bias was undertaken using the ROBINS-I tool (26).
Strategy for data
synthesis
Dichotomous outcomes were meta-analysed in RevMan 5.4 and are presented
as risk ratio/odds ratios with 95% confidence intervals. Continuous
outcomes are presented as mean difference with 95% confidence
intervals. When significant statistical heterogeneity was demonstrated
on meta-analysis (including but not limited to I²>40%)
random effects modelling is used for meta-analyses. Studies providing
data on incidence of AI are included in the ‘total AI’ meta-analysis,
whilst studies demonstrating a change in AI score or the development of
new AI symptoms in previously asymptomatic women after a subsequent
birth after OASI are included in the ‘new or worsening AI’
meta-analysis.
Subgroup analysis
Women who were symptomatic of anal incontinence after the first birth vs
those who were asymptomatic.