We performed a metaanalysis of publications comparing the efficacy and safety of cervical ripening and labor induction by Foley catheter balloon (FCB) vs locally applied prostaglandins (LAPG) in the third trimester of pregnancy. Twenty-seven randomized controlled trials (1966-2008; 3532 participants) were selected from MEDLINE, EMBASE, and CENTRAL searches. There was no significant difference between FCB and LAPG in cesarean delivery rates. LAPG had a significantly increased risk of excessive uterine activity ( P = .001). FCB had a significantly higher risk of oxytocin induction/augmentation during labor ( P = .0002). Cervical prostaglandin-E2 was less effective ( P = .04), and vaginal prostaglandin-E1 bore a significantly higher risk of excessive uterine activity ( P < .0001) and meconium staining ( P = .04). We concluded that FCB and LAPG result in similar cesarean delivery rates, that FCB bears a higher risk of oxytocin use for labor induction and/or augmentation, and that LAPG carries a higher risk of contraction abnormalities.
Induction of labor is used frequently when the risks to the mother and/or fetus with pregnancy continuation outweigh the risks that are involved with the intervention. The unripe cervix is a major impediment for the success of labor induction and vaginal delivery. Various pharmacologic and mechanical methods are used to achieve cervical ripening; locally applied prostaglandins (LAPG) and extraamniotic catheter balloons are used the most frequently. A considerable number of clinical trials that compared the efficacy and safety of these 2 major categories of cervical ripening methods have been published over the last 3 decades, but few attempts have been made to review and summarize them.
In an earlier review of cervical ripening by extraamniotic balloon catheter, we reported that 6 prospective randomized trials and 2 case-controlled studies reported that the ripening success rate and/or mean change in cervical score after balloon ripening ranged from significantly higher to similar, compared with the rates that were obtained after vaginal or cervical prostaglandin-E2 (PGE2) or prostaglandin-F2–alpha. Additionally, ripening-/induction-delivery intervals with the catheter balloon were shorter in 5 studies, similar in 2 studies, and longer in 1 study. Balloon ripening was associated with a higher rate of oxytocin use but a lower rate of side-effects, compared with prostaglandins use. Most studies, however, did not find any significant difference in the mode of delivery or perinatal outcome between the 2 ripening methods; however, this review was semiquantitative, and no metaanalysis was performed to extrapolate pooled outcome estimates.
A metaanalysis of 45 randomized controlled studies that was published in the Cochrane Database of Systematic Reviews compared various mechanical methods (laminaria tents, various types of balloon catheters, and extraamniotic infusion of fluids) of cervical ripening to placebo/no treatment, prostaglandins or oxytocin in women in the third trimester of pregnancy with an unfavorable cervix. They found no difference in the risk of delivery by cesarean delivery, but there was an increased risk of not achieving vaginal delivery within 24 hours when mechanical methods were compared with vaginal or cervical PGE2 (relative risk [RR], 1.74 and 1.70, respectively) and a similar risk when mechanical methods were compared with PGE1. Because these comparisons were based on a limited number of studies and included relatively small numbers of patients, the authors concluded that there was insufficient evidence to compare the effectiveness of the various methods in terms of likelihood of vaginal delivery within 24 hours. They noted that the use of mechanical methods reduced the risk of uterine hyperstimulation with fetal heart rate changes when compared with vaginal or cervical PGE2 or PGE1; there was no significant difference in other measures of safety. Similar results were obtained when only balloon catheter (Foley or Atad double balloon device; Willy Rusch AG, Kernen, Germany) ripening was compared with prostaglandins, but balloon ripening with extraamniotic saline solution infusion (EASI) was associated with a higher risk than prostaglandins for both caesarean section delivery and not achieving vaginal delivery within 24 hours, without a reduction in the risk of uterine hyperstimulation. Despite the small number of studies, the authors of the Cochrane review suggested that EASI should be abandoned. A later review concluded that EASI is both safe and effective (compared with misoprostol).
Since the publication of the Cochrane metaanalysis in 2001, additional randomized controlled trials (RCTs) that have compared ripening with catheter balloons and cervical/vaginal prostaglandins have appeared in the literature. Of the 45 studies included in the earlier metaanalysis, only 17 studies compared Foley catheter balloon (FCB) to LAPG of the E series, and that subgroup was not analyzed separately. Given that the mechanism of action of mechanical ripening devices is not clear and that there is no good reason to consider them as being a homogenous group, we decided to focus on comparative randomized controlled studies on the use of FCB and LAPG (vaginal or cervical) to induce delivery in the third trimester of pregnancy.
Materials and Methods
Search strategy
We conducted a systematic review of the English literature to identify all published randomized clinical trials that evaluated the use of an extraamniotic catheter balloon, with or without EASI, and/or intravenous oxytocin (IVO) vs LAPG for cervical ripening and labor induction in women with unfavorable cervices during the third trimester of pregnancy. An electronic literature search was conducted with MEDLINE (PubMed; National Institutes of Health, Bethesda, MD), Ovid Medline (Ovid Technologies, New York, NY), EMBASE (Elsevier BV, New York, NY) and CENTRAL (The Cochrane Central Register of Controlled Trials; Wiley InterScience, Indianapolis, IN) databases from January 1966 through April 2008. The search strategy included the following terms: ( labor, induced or cervical ripening or cervix uteri ) and ( balloon dilatation or catheterization or Foley or extraamniotic saline ) and prostaglandin/s . Bibliographies of all relevant articles that were identified by the database searches and review papers were searched manually to locate additional studies.
Institutional review board approval was not requested because this is a metaanalysis that was based on published articles from the English literature and no new patients were enrolled.
Selection criteria
Only RCTs with results that appeared in English-language publications and that compared extraamniotic FCB and LAPG for cervical ripening and labor induction in the third trimester of pregnancy were included in this analysis. Two of the study investigators (Z.V., D.S.) reviewed the selected publications for study design suitability and for meeting inclusion/exclusion criteria. The study investigators, who used a standardized data abstraction form, extracted study characteristics and outcomes independently. Any disagreements were resolved by consensus. Inclusion criteria for the studies were singleton pregnancies with a live fetus in vertex presentation and intact membranes who was presenting with an unfavorable cervix (Bishop score, ≤6) at ≥28 weeks gestation. The included studies used prostaglandins preparations of the E series in tablets, gel, or pessary. The FCB was used either alone or in combination with EASI and/or IVO with gentle traction on the catheter. Studies that combined FCB and prostaglandins or studies that used other balloon devices were excluded. Studies were excluded if they appeared only as abstracts or if they did not report the mode of delivery for all participants who received the allocated treatment.
Treatment outcome measures
The outcomes that were included in our analysis addressed the efficacy and safety (side-effects) of each method. We chose 2 main outcomes: the proportion of subjects who were delivered by cesarean section and the incidence of excessive uterine activity ( tachysystole , defined as at least 6 uterine contractions per 10-minute period, or hyperstimulation , if combined with fetal heart rate abnormalities, or hypertonus defined as single contraction of >2-minute duration). We also included 4 secondary efficacy outcomes: the proportion of subjects who delivered vaginally within 24 hours from the beginning of the ripening or within 12 hours from the beginning of induction (ie, vaginal delivery within 12-24 hours), the proportion of subjects with unfavorable/unchanged Bishop score 12-24 hours after ripening (ie, failed ripening at 12-24 hours), the ripening-delivery time interval and the need for oxytocin induction and/or augmentation of labor (studies that combined FCB with IVO were excluded for the latter). In addition, there were 4 secondary safety variables: the proportion of subjects who experienced intrapartum or postpartum fever, the incidence of meconium staining, the proportion of neonates with 5-minute Apgar scores of <7, and the number of neonates who were admitted to the neonatal intensive care unit. We also recorded the number of cases of serious maternal and/or neonatal morbidity or death (including uterine rupture, maternal admission to the intensive care unit, maternal septicemia, placental abruption, hemorrhage that required blood transfusion, hysterectomy, neonatal seizures, birth asphyxia, meconium aspiration syndrome, or neonatal sepsis) when they were reported. Not all trials included in this metaanalysis evaluated each of those outcomes; therefore, specific outcome analyses were based on a variable number of studies.
Evaluation of the quality of the studies
Methodologic quality assessment of each of the included trials was carried out with a validated scale in which identifiers and conclusions were concealed by means of a black marker. The criteria that were used to assess the methodologic quality included the randomization scheme, allocation concealment, and reporting of dropouts and withdrawals after randomization. The original scale also includes the method of masking (double blinding), which is impossible in these trials. The maximal score was 4; high-quality studies were defined as a score of ≥3. None of the trials were excluded on the basis of quality assessment.
Methods of the metaanalyses
All metaanalyses were performed with the Review Manager software package (RevMan, version 4.2; The Nordic Cochrane Center, Copenhagen, Denmark). For each study with binary outcomes, we calculated the RR and 95% CI of the selected outcomes. Estimates of RRs for dichotomous outcomes were calculated with the fixed-effects (Mantel–Haenszel) or random-effects (Der Simonian & Laird) models. The weighted mean difference (WMD) and CIs were calculated for continuous variables. Pooled estimates for WMD were derived from fixed-effects model (inverse variance method) and random-effects model (Der Simonian & Laird). Heterogeneity between trials was assessed by the χ 2 and I 2 -statistic tests. A probability value of < .10 or an I 2 of >50% indicated statistical significance in the analysis of heterogeneity. For all analyses, results from fixed effect model are presented when both χ 2 and I 2 -statistic tests noted no heterogeneity between the RCTs. Results from random effects model are presented whenever statistically significant heterogeneity was present. To explore potential sources of study results heterogeneity, we performed subgroup analyses to assess differences in treatment effects separately for (1) trials that included term pregnancies only and those that included preterm gestations, (2) trials that excluded and those that included patients with previous cesarean delivery, (3) trials that included only primiparous patients and those that included patients of all parities, (4) trials that compared FCB to PGE1, cervical PGE2, and vaginal PGE2, (5) trials that compared LAPG with FCB alone and FCB combined with EASI and/or IVO. For outcomes that had adequate data, we performed a sensitivity analysis by comparing the findings of the metaanalysis of high- and low-quality studies together with only those studies that had been ranked as high quality.
Publication bias was examined with the Egger test and by visual inspection of funnel plots of ORs against study sample size. A probability value of <.05 indicated statistical significance in the analysis of publication bias. Additionally, sensitivity analyses were conducted to assess the influence of each individual study on the pooled estimate and to ensure that no metaanalytic estimate was dominated by 1 single trial.
Only 2 trials performed an intent-to-treat analysis. There were no outcome data for patients who were excluded from the trial in the remaining studies; therefore, our analysis relates to participants who received the treatment as allocated by study group (“per protocol” analysis).
Statistical significance was defined as a probability value of < .05.
Results
A total of 111 citations were identified by our literature search strategy, with the use of the broad search terms that were cited earlier and by manual retrieval from related reference lists. Of these, 45 articles were excluded based on title and abstract review ( Figure 1 ). The remaining 66 articles of comparative clinical trials were retrieved for detailed evaluation, of which 12 articles were excluded as not being RCTs. Of the 54 potentially appropriate RCTs ( Figure 1 ), we excluded 27 because they did not assess safety and efficacy in terms of labor and delivery outcomes (n = 6), they used the FCB in combination with prostaglandins (n = 13), they used the FCB with significant traction (n = 1), or they included second trimester gestations, nonviable fetuses, or fetal deaths (n = 7). The remaining 27 randomized trials were included in this metaanalysis. Sixteen studies (n = 1793 women) compared FCB alone with LAPG ( Table 1 ). Four studies (n = 376 women) compared FCB combined with EASI alone with LAPG, and 7 studies (n = 1363 women) compared FCB combined with EASI and/or IVO with LAPG ( Table 2 ). PGE1 tablets were applied vaginally in 13 trials (n = 921 women); PGE2 gel was applied cervically in 7 trials (n = 412 women), and PGE2 was applied vaginally as tablets, gel, or pessary in another 7 trials (n = 381 women; Tables 1 and 2 ). Sample sizes ranged from 22-169 participants in each study group, with 44-279 participants in both study groups combined.
