Design and Registration
We conducted the review according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.
Inclusion and exclusion criteria
The inclusion criteria in this meta-analysis were: 1)participants:women with singleton pregnancy after 37 weeks of gestation and a previous low transverse cesarean section (CS), who voltunteered to accept a trial of labor after CS (TOLAC) and did not have a contraindication to TOLAC such as previous uterine body incision, placenta previa or abnormal pelvimetry with a breech presentation; 2) intervention: oxytocin induction or augmentation during a TOLAC; 3) outcomes: the success rate of TOLAC (VBAC), the usage rate of oxytocin in TOLAC and the risk of uterine rupture; 4) type of studies: designed with case-control study or cross-sectional study.
The exclusion criteria were: 1)women with previous classical CS, history of more than one CS, major fetal anomalies, active labor before rupture of membranes, scheduled elective CS, contraindications for spontaneous delivery (placenta previa, breech presentation, etc.), history of uterine rupture, and lack of information regarding the previous delivery; 2)duplication of previous publication(s). Two independent investigators finished the procedure, solving the dispute by discussion.
Search strategies
A literature search with no date restrictions was conducted in PubMed, EMBASE,Web of Science, Clinicaltrial and Google Scholar. The following key words were used: "vaginal birth after a cesarean section" OR "VBAC" AND "a trial of labor after cesarean section" OR "TOLAC" OR "trial of labor" AND "oxytocin" OR "oxytocin infusion" OR "induction of labor" AND "uterine rupture" OR "maternal morbidity". Final literature searches were performed in June 2019. The hits were reviewed, and duplicates were eliminated. Then, inclusion and exclusion criteria were set for including records. Finally, the titles, abstracts, key words, and whole texts of retrieved studies were checked to exclude irrelevant ones. Also, the reference lists of the retrieved studies and recent reviews were checked manually to avoid missing any studies meeting the inclusion criteria.
Data Extraction
Necessary data from eligible studies were extracted in this meta-analysis, including first author's name, publication year, sample size, maternal age, maternal BMI, gestational age, dose of oxytocin, the number of spontanteous delivery, the number of induction labor, the number of patients using oxytocin, the number of VBAC and the number of uterine rupture. Two independent reviewers double-extracted the data and cross-checked the results for discrepancy, which were discussed for correction. A third independent reviewer assessed the coding for accuracy by randomly selecting and recoding five articles and examining potential outliers in the data. The authors were contacted by e-mail and requested for relevant data if these values were not reported, to collect the most complete dataset. In the case of no feedback, the studies with missing information were abandoned.
Statistics Analysis
In order to evaluate the effect of oxytocin directly, studies were grouped by women using oxytocin and not using oxytocin. To exclude the possible influence, studies were grouped by women with spontaneous labor and induction labor. We calculated the usage rate of oxytocin, rate of VBAC and rate of uterine rupture by the number of TOLAC, the number of spontaneous delivery, the number of induction labor, the number of VBAC, the number of patients using oxytocin and the number of uterine rupture. Due to the anticipated heterogeneity, we planned to choose the random-effects model or fixed-effects model to calculate overall effect size. For each measure, we calculated Cohen's d in line with the general systematic approach and performed with the associated website (http://www.campbellcollaboration.org/ resources/ effect size input.php), using means and standard deviations or standard errors where possible[19]; occasionally, F, t or p values were used with sample size to estimate the effect size. In order to correct for overestimation of the effect size associated with small sample sizes, we applied Hedge's correction to each effect size, and calculated inverse variance weights for each study using the corrected effect size. In addition, we used independent-sample T test to compare the difference of synthetic rates between groups. Statistical significance was considered at a P value of .05.
Heterogeneity was tested using the I2 statistic and Q test. We considered statistical heterogeneity to be low for I2 ≤40%, moderate for I2 =30%–60%, substantial for I2 =50%–90% and considerable for I2=75%–100%. Sensitivity analyses and meta-regression were used to explore the potential sources of heterogeneity. Publication bias was checked using the funnel plot and Egger's tests. Besides, the critical evaluation of the bias risk of the included studies was conducted by two independent reviewers using the Newcastle-Ottawa Scale (NOS) [20].
All statistical analyses were conducted using Stata 14.0.
Assessment of Evidence in Cumulative Evidence
We evaluated the quality of evidence for each outcome across studies using four levels (high, moderate, low or very low confidence) according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria [21].