Clinical interventions that influence vaginal birth after cesarean delivery rates: Systematic Review & Meta-Analysis

Background To systematically review the literature on clinical interventions that influence vaginal birth after cesarean (VBAC) rates. Methods We searched Ovid Medline, Ovid Embase, Wiley Cochrane Library, CINAHL via EBSCOhost; and Ovid PsycINFO. Additional studies were identified by searching for clinical trial records, conference proceedings and dissertations. Limits were applied for language (English and French) and year of publication (1985 to present). Two reviewers independently screened comparative studies (randomized or non-randomized controlled trials, and observational designs) according to a priori eligibility criteria: women with prior cesarean sections; any clinical intervention or exposure intended to increase the VBAC rate; any comparator; and, outcomes reporting VBAC, uterine rupture and uterine dehiscence rates. One reviewer extracted data and a second reviewer verified for accuracy. Meta-analysis was conducted using Mantel-Haenszel (random effects model) relative risks (VBAC rate) and risk differences (uterine rupture and dehiscence). Two reviewers independently conducted methodological quality assessments using the Mixed Methods Appraisal Tool (MMAT). Results Twenty-nine studies (six trials and 23 cohorts) examined different clinical interventions affecting rates of vaginal deliveries among women with a prior cesarean delivery (CD). Methodological quality was good overall for the trials; however, concerns among the cohort studies regarding selection bias, comparability of groups and outcome measurement resulted in higher risk of bias. Interventions for labor induction, with or without cervical ripening, included pharmacologic (oxytocin, prostaglandins, misoprostol, mifepristone, epidural analgesia), non-pharmacologic (membrane sweep, amniotomy, balloon devices), and combined (pharmacologic and non-pharmacologic). Single studies with small sample sizes and event rates contributed to most comparisons, with no clear differences between groups on rates of VBAC, uterine rupture and uterine dehiscence. Conclusions This systematic review evaluated clinical interventions directed at increasing the rate of vaginal delivery among women with a prior CD and found low to very low certainty in the body of evidence for cervical ripening and/or labor induction techniques. There is insufficient high-quality evidence to inform optimal clinical interventions among women attempting a trial of labor after a prior CD.


Background
Over 103,000 cesarean deliveries (CDs) occurred in 2017 within Canadian hospitals [1]. In Canada, CDs continue to be the leading inpatient surgery with elective/scheduled CD as a main contributor. Since 1997, the rate of CD has increased from 18.7 to 28.2% in 2017, and frequency of this delivery method continues on an upward trend [1,2]. Globally, rates of cesarean sections are considered high at an estimated 21% of livebirths in 2015, based on data from 169 countries [3].
A number of factors influencing the increase of this surgical delivery method include changes in healthcare practice styles, patient preferences, pressures of malpractice and demographic influences (e.g., social, economic, cultural) [4][5][6][7][8][9][10]. These influences can affect delivery options/choice and may result in complex pregnancies that ultimately require a CD [11,12]. Short-and long-term morbidity risks for the infant and mother are further influenced by the etiology or indication of their CD; however, overall risk of morbidity and mortality is more positively associated with CD compared to vaginal delivery [13][14][15][16]. This risk warrants careful consideration of potential post-operative complications before scheduled CD, a major abdominal surgery. Recent 'Early Recovery After Cesarean' (ERAS) for CD guidelines (Part 1-3) have been published to reduce maternal and neonatal morbidity and mortality [17].
For women who have undergone a prior CD, there is uncertainty regarding the choice of a repeat/scheduled CD or attempting a vaginal delivery for a successive pregnancy as both modes of birth have risks. The Society of Obstetricians and Gynaecologists of Canada (SOGC), the American College of Obstetricians and Gynecologists (ACOG) and the Royal College of Obstetricians and Gynaecologists (RCOG) recommend that a trial of labor be offered to women with one previous transverse low-segment CD [18][19][20]. Vaginal birth after cesarean (VBAC) may be desired by some women, but the patient-level benefits associated with VBAC including avoiding repeat abdominal surgery and risk of complications in future pregnancies must be considered against the potential risks of a failed trial of labor after cesarean (TOLAC) with subsequent maternal and neonatal morbidity, including an unplanned repeat CD [18]. While the risk for uterine rupture of the previous cesarean incision scar is low (single CD 0.72%; double CD 1.59%) there is maternal and neonatal risk [21].
Many studies have examined factors that are associated with a greater likelihood of a successful TOLAC, commonly identifying a history of successful vaginal delivery [22][23][24] and women who present in spontaneous labor [19,22] as significant predictors.
Due to high global cesarean rates, the promotion of VBAC may be one option to reduce the overall number of cesarean deliveries. Clinical interventions that positively impact the rate of vaginal deliveries for women choosing a VBAC need to be examined. This systematic review aimed to synthesize and evaluate the research on clinical interventions that could be directed at or used by patients, families, healthcare providers, and hospitals/ health systems to influence the success of VBAC. A systematic review of 'adjunct' clinical interventions that influence the uptake and success of VBAC has been completed and published [4].

Methods
This study followed standardized methods and guidelines for systematic reviews [25,26], and used an a priori protocol (available from authors).

Eligibility criteria
The study population was women who had a previous CD including women with more than one prior CD. Births attended by any healthcare provider (e.g., family physician, midwife, obstetrician/gynecologist) were eligible. Any clinical intervention or exposure that was intended to achieve a successful VBAC among women with a prior CD were eligible for inclusion. Studies had to report on at least one of the following pre-determined outcomes: our primary outcome was rate of VBAC among women who attempted a vaginal delivery; secondary outcomes included uterine rupture rates and uterine dehiscence rates, whenever these were reported, as these are considered as being significantly associated with increased likelihood of maternal and neonatal morbidity. Studies that examined deliveries in any setting (e.g., hospitals, primary care centers, birthing units, home births) were eligible. All study designs (randomized [RCT] and non-randomized controlled trials [NRCT], and observational studies) with a comparison group were eligible for inclusion. Studies were not considered eligible if: all women had three or more prior cesareans; multiple births of three or more fetuses were explicitly included; there was an absence of an exposure or intervention, or an inappropriate exposure/intervention was used (e.g., prediction models, pelvimetry, nonclinical interventions such as guidelines for providers); there was absence of a comparator, or an inappropriate comparator was used (e.g., no data for comparison groups in before-after study designs, women without a previous CD); VBAC rates were not reported; or, they were not primary research (e.g., letter, editorial, commentary). Systematic reviews were not included; reference lists therein were screened for potentially relevant studies.

Study selection
Two reviewers (CJ and AW) independently screened titles and abstracts using a priori eligibility criteria. Full texts of potentially relevant publications were retrieved and independently reviewed in duplicate for inclusion; disagreements were resolved through discussion or third-reviewer (MS) consultation.

Data extraction
One reviewer extracted data and another verified data from each included study using a pre-specified and piloted form. Data were extracted for relevant study characteristics (design features), population (number of previous cesarean deliveries, parity), intervention, comparator, outcome (VBAC rate [the number of women with a previous CD who undergo a successful vaginal delivery]; uterine rupture rate [the number of women who experience a uterine rupture among those who attempt a vaginal delivery]; and, uterine dehiscence rate [the number of women who experience a uterine dehiscence among those who attempt a vaginal delivery]), funding source, and setting.
Intention-to-treat results were extracted from individual studies whenever possible. For dichotomous data (rates of VBAC, uterine rupture and uterine dehiscence), we reported counts or proportions, and sample size, by study arm. Results of statistical tests (e.g., p-values) or summary statistics (e.g., odds ratio [OR], risk ratio [RR], with confidence intervals [CI]) were extracted whenever these were reported within the studies.

Assessment of methodological quality
Two reviewers (CJ and AW) independently assessed the methodological quality of included studies; disagreements were resolved via consensus. All studies were assessed using the Mixed Methods Appraisal Tool (MMAT) [27].

Data synthesis
For rates of VBAC, we reported a relative risk and statistically pooled these using the DerSimonian and Laird random effects model with Mantel-Haenszel weights and corresponding 95% CIs. Risk difference (RD) was used for rare outcomes with small event rates (uterine rupture and uterine dehiscence). Statistical heterogeneity was quantified using the I-squared statistic.
Decisions to pool studies were based on comparability of clinical (e.g., treatment) and methodological (i.e., study design) characteristics across studies.
Analyses were performed using Review Manager Version 5.3 [28].

Assessment of overall certainty of evidence
Two reviewers (MS and AW) assessed the certainty of the body of evidence for each outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) [29], with disagreements resolved through discussion or consultation with a third reviewer (LH). Certainty was assigned initially as high for evidence from trials and low for evidence from observational studies. Each of five domains was then assessed for potential downgrading: study limitations/risk of bias, inconsistency, indirectness, imprecision, and publication bias. An overall score was determined for each outcome using the GRADE certainty of evidence categories: high, moderate, low or very low.

Results
The literature search yielded 5833 unique records. After screening titles and abstracts, 339 potentially relevant articles were identified. Full text screening identified 29 relevant studies . The screening process is illustrated in Fig. 1. Table 1 provides a summary of the included studies; detailed study characteristics are in Additional file 1: Appendix 2.
All studies included patients who delivered in a healthcare setting.
Induction of labor is the artificial stimulation of labor before its spontaneous onset to achieve vaginal delivery, taking into account the status/readiness of the cervix (termed "favorable" or "ripe") prior to initiating the labor process [59]. Cervical ripening or induction of labor can be achieved using non-pharmacologic methods, pharmacologic agents, or some combination of both techniques, each with advantages and disadvantages. Mechanical and surgical methods are forms of non-pharmacologic induction, including membrane stripping, balloon catheters, or amniotomy (artificial rupture of membranes) [60]. Pharmacologic agents commonly used for cervical ripening or induction of labor include prostaglandin (PGE2 analog, in gel or pessary form), misoprostol (PGE1 analog), or mifepristone [60].
All studies examined some manner of cervical ripening and/or labor induction in the intervention group, usually compared to a group of women undergoing spontaneous labor or expectant pregnancy management.
Many cervical ripening and labor induction methods were represented by the included studies.
Methodological quality of included studies (Table 2 and Additional file 1: Appendix 3) All of the studies reported a clear research question or objective and collected data that addressed the intended research question.

VBAC rates
Effect estimates for VBAC rates are summarized in Table 3. There was low to very low certainty of evidence for all comparisons involving pharmacologic, nonpharmacologic, and combined (pharmacologic and nonpharmacologic) induction methods.

Pharmacologic versus non-pharmacologic
One cohort study compared pharmacologic with mechanical induction methods [54].  11). No significant differences between groups were found in other RCTs [40,45,56] and cohort studies [44,47,58] (various comparisons) that stratified vaginal deliveries as assisted or unassisted (very low certainty of evidence).

Uterine rupture rates
Of the studies that reported on uterine rupture, most had zero cases among study arms. Effect estimates are summarized in Table 4. All comparisons for pharmacologic, non-pharmacologic, combined (pharmacologic and non-pharmacologic) induction methods provided low to very low certainty of evidence.

Uterine dehiscence rates
Effect estimates for uterine dehiscence rates are summarized in Table 5. All comparisons involving pharmacologic, non-pharmacologic and combined (pharmacologic and non-pharmacologic) induction methods provided low to very low certainty of evidence.

Pharmacologic and non-pharmacologic (combined)
One cohort study found no significant differences in rates of uterine dehiscence between women with (any of PGE2, oxytocin, amniotomy with or without Foley catheter) and without induction (RD 0.00, 95% CI − 0.01 to 0.01) [54].

Discussion
Overall, there is low to very low certainty of evidence for clinical interventions that influence rates of VBAC, owing to downgrading mainly for the GRADE domains of risk of bias, inconsistency and imprecision. Many of the comparisons were based on single studies with small sample sizes and event rates. The interventions were heterogeneous and focused on pharmacologic and/or mechanical methods of induction. Most of the evidence showed no significant differences between groups on VBAC rates, with very low certainty of evidence. There was some evidence of higher VBAC rates among women who undergo spontaneous labor, when compared to women whose labors are induced, regardless of method or agent used for induction; however, there was very low certainty in this body of evidence.
Other systematic reviews of clinical interventions have been reported in the literature. Catling-Paull et al examined the effect of clinical interventions on the uptake or success of VBACs and found that inductions of labor by amniotomy, prostaglandins, or oxytocin (or a combination of these methods) were associated with lower rates of vaginal deliveries (six RCTs and 28 cohort studies) [61]. The review also examined prognostic criteria/factors and concluded that pelvimetry may adversely affect women's chances of a VBAC, scoring systems are not clinically useful, and there is lack of sufficient data to inform the value of assessing methods of cesarean closures for predicting successful vaginal deliveries [61]. A Cochrane systematic review compared women (with a prior CD) undergoing cervical ripening and/or labor induction with placebo, no treatment or other methods, and found overall moderate to low certainty of evidence for the varied interventions represented among the small number of included studies, concluding insufficient availability and high-quality of evidence to determine the optimal method of labor induction [62].

Strengths and limitations of study
This systematic review of clinical interventions for influencing VBAC rates encompassed a broad range of study designs and strategies for cervical ripening and/or labor induction. The nature of the labor process and difficulty blinding women or healthcare providers to interventions lends few studies to an RCT design. Unsurprisingly, nearly 80% of the studies were cohort designs. However, observational studies have inherent biases due to confounding and there were concerns among the included cohorts regarding comparability of population characteristics between groups. Few studies adequately reported maternal characteristics. Some studies included women with only one CD while others enrolled women with multiple cesarean deliveries, a documented risk of increased uterine rupture among women attempting a VBAC. Additionally, few studies reported parity, type of previous uterine incision, prior vaginal delivery, prior VBAC, or indication(s) for previous CD. The concept of spontaneous labor among most studies is erroneously represented in that women in a comparator group (as opposed to the intervention group) are often managed expectantly, where they may or may not then progress to spontaneous labor [18]. Further, spontaneous onset of labor is often used synonymously with spontaneous delivery. One outcome of induction of labor (versus expectant management) is induction success (or failed induction) distinguished from VBAC success. Ambiguous operationalization of these concepts may result in important differences in outcomes. Among women who were induced, studies often did not provide data for the number of women whose labor was augmented. While VBAC was reported by all studies, other perinatal outcomes were unreported or undefined despite following up women to delivery. Four (14%) studies did not report whether there were cases of uterine ruptures in their study population. Among studies that reported on uterine ruptures and/or uterine dehiscence, authors defined these outcomes inconsistently (and without a cited reference) or they were not explicitly differentiated from each other, or they were undefined altogether. Contextual and mediating factors, such as induction-to-labor interval or co-interventions were sparsely reported for women in all study arms.

Implications for practice
The ability for the maternity care provider to use this evidence to counsel women with a previous CD so they have the information to make the best choice for their situation is a goal of this systematic review. Ananth et al completed a population based study, in the USA, to show the impact of scheduled/repeat CD on the total number of unscheduled/emergency and scheduled/repeat CD [63]. From 1979 to 2010, unscheduled and scheduled CD increased 68% (95% CI 67-69) and 178% (95% CI 176-179), respectively [63]. For the total number of deliveries in 2010, 18.5 and 14.4% were unscheduled and scheduled CDs, respectively [63]. Maternal factors attributed to the unscheduled cesarean increases were increased maternal age, obesity and chronic hypertension [63]. Clinical impact for the scheduled CD group of 14.4% (likely higher in 2019) requires better VBAC information for maternal education and choice [63].
One of the aspects of this VBAC choice is that women cannot choose to be in spontaneous labor at any given gestational age [64]. In their VBAC choice, women are faced with the choice of expectant pregnancy management (waiting for labor to start) or consider an induction of labor protocol acceptable for VBAC [65] or having the scheduled CD. Cervical status/Bishop score is an important clinical factor for induction of labor. The induction of labor protocol for women with a prior uterine scar reports that oxytocin is the only induction agent that can be used other than mechanical methods [65]. Oxytocin has limited effectiveness for cervical ripening. Adequate cervical dilatation allows for amniotomy and oxytocin use to assist in induction process [59]. There are no 'real' expectant management VBAC cohorts. For non-VBAC cohorts, there are studies comparing risks of elective labor induction with expectant management. For multiparous women a large retrospective cohort study showed that induction of labor is associated with decreased pregnancy-related hypertension and increased time in labor and delivery [66]. A Cochrane review compared induction of labor at or beyond term with expectant management in uncomplicated, singleton pregnancies, with no parity factor identified; they reported that induction was associated with lower risk of perinatal death, stillbirth and fewer cesarean section, but more operative vaginal birth and no change for perineal trauma, postpartum hemorrhage, stillbirth, low Apgar score, or neonatal distress [67].
While the studies show heterogeneity and certainty of evidence is low to very low, the uterine rupture and dehiscence rates among studies are very clinically equivalent to rates reported in the literature (ruptures, pharmacological/non-pharmacological/combined for each comparison group 0.56, 0.70%; 0, 0; 1.05, 0.65%, respectively and dehiscence, pharmacological/nonpharmacological/combined for each comparison group 2.10, 0.76%; 1.44, 0.72%; 1.21, 0.87%, respectively). These findings can support the safety of oxytocin use for ripening alone and in combination with mechanical methods to allow amniotomy as commented above.

Candidates for TOLAC
Various maternal characteristics, clinical history, and presenting labor indications associated with likelihood of VBAC have been reported in the literature. Factors associated with increased risk of maternal morbidity from a TOLAC include: prior classical cesarean section or T incision as a contraindication [18][19][20], previous low segment cesarean section where the uterus was closed in a single layer [19], less than 18-month inter-delivery interval [19], medical induction or augmentation of labor as compared to spontaneous labor [20], medical induction of labor with misoprostol [18,19] or PGE2 [19], and increasing number of prior CDs [19].
A previous successful trial of vaginal delivery is associated with greater likelihood of successful VBAC; alternatively, the history of a failed trial of vaginal delivery requires a CD [15,20,68,69]. Multiple obstetrical indications for inducing labor or considering CD, also need to be considered (e.g., post-dates, hypertension, macrosomia, intra-uterine growth restriction, unfavorable cervix) [20,70]. Different induction methods may have differential effects on progress of labor, and the impact on mode of delivery. Limited evidence from RCTs on methods of labor induction have not been able to determine which method results in greater benefits and lower risks [62].
Evidence-based guidelines comparison from the US (2010 version), Canada (2005 version) and United Kingdom (2007 version) show similarities and differences in recommendations for VBAC process [71]. Uterine rupture is identified as the most serious complication among women attempting a trial of labor; however, there is recognition that differences in the definition may contribute to the reported incidence rates [71]. With regards to induction of labor, all three guidelines agree that induction and augmentation of labor is an option for women attempting a VBAC, but there is an increased risk of uterine rupture with inappropriate use of oxytocin [71]. Both the American College of Obstetricians and Gynecologists (ACOG) and SOGC recommend against PGE1 for the same outcome, with the SOGC guideline recommending PGE2 only in rare circumstances [71]. All three guidelines agree that women with a history of one or two uncomplicated lowtransverse CD and no contraindications to VBAC should be educated, counselled and offered a TOLAC [71]. The SOGC's updated guidelines (July 2019) continue to support the previous recommendation of offering women (without contraindications) the option of a TOLAC [19].
The new guideline provides details on the likelihood of VBAC (greater among women with spontaneous labor, lower among women with factors that negatively affect this likelihood), the risk of uterine rupture (baseline risk of 0.47% among women with a prior CD; higher relative risk of uterine rupture, but low absolute risks for women attempting TOLAC compared with elective repeat cesarean section; risk greatest for women over 40 weeks of gestation; use of fetal monitoring is recommended as an indicator of the presence of uterine rupture), risk of other outcomes (higher relative risk of maternal and perinatal morbidity, but lower absolute risks for women attempting TOLAC compared with elective repeat cesarean section; higher relative risk of maternal death for elective cesarean section), and recommends against using ultrasonographic measurements of the lower uterine segment to counsel patients on the possibility of a TOLAC [19].

Shared decision-making
For women with a previous CD, the decision to proceed with a repeat cesarean section or attempt a vaginal delivery will be based on multiple factors. First steps in the decision-making process should be consultation with a healthcare provider regarding maternal and neonatal benefits and risks, for both VBAC (induction/spontaneous onset of labor) and scheduled CD options. Triaging women for VBAC (i.e., women without contraindications) must include maternal characteristics and obstetric history to assist in the prediction of a successful vaginal delivery. Family planning plays a key role. Women who have had one or more prior cesarean deliveries may opt for a trial of vaginal birth or elect to have another CD; this maternal decision impacts future pregnancies and delivery due to the increased risk of abnormal placentation secondary to the additional uterine surgeries [72]. Unexpected severe adverse maternal and neonatal outcomes (e.g., maternal mortality) are important for patient-level counselling and decisionmaking.
Each woman should be given the opportunity to be counselled early in pregnancy on her available options, with provision of written or on-line patient information literature (e.g., VBAC clinical care pathway, decision aids) to help inform decision-making throughout her pregnancy for optimal delivery outcomes and patient satisfaction.

Conclusion
This systematic review evaluated clinical interventions for safe practice directed at increasing the rate of vaginal delivery among women with a prior CD but found low to very low certainty in the body of evidence for cervical ripening and/or labor induction techniques. There is insufficient high-quality evidence to inform on optimal clinical interventions/protocols for women considering to have a VBAC.