Single versus Double-Balloon Transcervical Catheter for Labor Induction: A Systematic Review and Meta-Analysis

Abstract

Objective

To evaluate for difference in outcomes between single- and double-balloon catheters for labor induction.

Study Design

We searched CINAHL, Embase, Cochrane Register, MEDLINE, ISI Web of Sciences, LILACs, and Google Scholar and retrieved studies through May 2017. Selection criteria included randomized controlled trials comparing single- versus double-balloon catheters. The primary outcome was time from catheter insertion to delivery. Heterogeneity of the results among studies was tested with the quantity I². For I² values ≥50%, a random effects model was used to pool data across studies. Summary measures were reported as adjusted odds ratios (aORs) or as a mean difference (MD) with 95% confidence interval (CI).

Results

Four trials including a total of 682 patients were included: 340 patients were randomized to induction with a single-balloon catheter and 342 to induction with a double-balloon catheter. There was no significant difference between groups with respect to time to delivery (18.8 vs. 19.6 hours; MD: 0.40; 95% CI: −1.56 to 0.76), vaginal delivery rate (65.3 vs. 62.3%; aOR: 1.04; 95% CI: 0.56–1.92), cesarean delivery rate (25.6 vs. 27.5%; aOR: 0.98; 95% CI: 0.55–1.73), or epidural use (58.4 vs. 62%; aOR: 0.81; 95% CI: 0.56–1.18).

Conclusion

Double-balloon catheters have no apparent advantage over single-balloon catheters for labor induction.

Labor induction is one of the most common procedures in modern obstetrics, affecting up to a quarter of women with singleton births in the United States. Methods for labor induction include topical or systemic administration of prostaglandins, oxytocin infusion, mechanical modalities, or a combination of each of these.^1-3 Mechanical modalities are used to provide local pressure to the cervix, leading to cervical dilation and stimulating local prostaglandin release to help with cervical remodeling. Mechanical dilation is often performed with either a single-balloon catheter (Foley catheter) or a larger transcervical double-balloon catheter (Cook catheter).⁴

When choosing a method, consideration should be paid to cost, patient comfort, and patient safety.⁵ Studies of mechanical methods of induction vary by instrument used as well as concomitant use of pharmacological agents such as prostaglandins and oxytocin. There are only a small number of studies comparing mechanical methods alone without the inclusion of pharmacological agents, but these studies are limited due to a small sample size, lack of statistical significance, and contradicting results.^5-8

When comparing mechanical methods, the single-balloon Foley catheter is easy to insert, has a low cost, and has been associated with minimal side effects.⁹ The double-balloon (Cook) catheter has been reported to have an improved mechanism of cervical dilation with the application of pressure on both the internal and external cervical os compared with the unilateral pressure of a single-balloon catheter.¹⁰ However, the decision as to which catheter to use to affect the timing of delivery has been difficult to determine with the available published evidence.

Labor outcomes comparing a single-balloon catheter with a double-balloon catheter have not yet been studied by way of systematic review and meta-analysis. This study was performed to compare efficacy, labor outcomes, and analgesia use as a marker of discomfort in women induced using a single- versus double-balloon catheter.

Materials and Methods

Our study was reported following the Preferred Reporting Item for Systematic Reviews and Meta-Analysis (PRISMA) statement.¹¹ Searches were performed in the CINAHL, Embase, Cochrane Register, MEDLINE, ISI Web of sciences, LILACs, and Google Scholar from their inceptions through May 2017. Combination of keywords and text words related to “Foley bulb, induction of labor, cook catheter, double-balloon catheter, single-balloon catheter” were used during the search process. Additional publications were found by reviewing proceedings of the American College of Obstetricians and Gynecologists. Bibliographies of identified studies and review articles were also reviewed. For studies with many publications, the data from the most complete report were used and supplemented if additional information, including secondary analysis of data, appeared in other publications.

Study Selection

We included prospective randomized controlled studies if the outcomes of single- versus double-balloon catheter were studied. Studies were excluded in the following circumstances: (1) the studies compared single- or double-balloon catheter to a pharmacological agent, (2) case reports and case series, and (3) published abstracts alone were excluded if additional information on methodological issues and results could not be obtained. All published studies that were deemed suitable were retrieved and reviewed independently by two authors (A. Eke and S. de los Reyes) to determine inclusion eligibility. Disagreements were resolved through discussion between the authors. Authors of any selected studies were contacted to complement data on trial methods and or outcomes.

Outcome Measures

The primary outcome measure was time from catheter insertion to delivery. Secondary outcomes were rates of vaginal delivery, cesarean delivery, and epidural use.

Data Extraction and Assessment of the Risk of Bias

Data abstraction was completed by two independent investigators (A. Eke and S. de los Reyes). Each investigator independently abstracted data from each study. Data from each eligible study were extracted without modification of original data onto custom-made data collection forms. Differences were resolved by consensus. Information of confounders adjusted and adjusted risk estimates were collected when available. Reviewers (A. Eke and S. de los Reyes) independently assessed the risk of bias of the included studies through the Methodological Index for Non-Randomized Studies (MINORS).¹² Seven domains related to the risk of bias were assessed in each study: (1) aim (i.e., clearly stated aim), (2) rate (i.e., inclusion of consecutive patients and response rate), (3) data (i.e., prospective collection of data), (4) bias (i.e., unbiased assessment of study endpoints), (5) time (i.e., follow-up time appropriate), (6) loss (i.e., loss to follow-up), and (7) size (i.e., calculation of the study size). Reviewers’ judgments were categorized as low risk, high risk, or unclear risk of bias. Discrepancies were resolved by discussion.

Assessment of the Risk of Bias of Included Studies

Two reviewers (A. Eke and S. de los Reyes) independently assessed the risk of bias in each included study using the Cochrane’s risk of bias instrument (RevMan). We assessed the risk of bias using the following domains: selection bias, information bias, attrition bias, and reporting bias. Each element of bias was categorized as “low risk of bias,” “high risk of bias,” or “unclear risk of bias” according to the following criteria:

• Selection bias: We assessed each study for the risk of bias in random sequence generation and concealment of random allocation. The reviewers recorded the method of random sequence generation described and assessed the risk of bias associated with the method used based on whether the sequence generation was truly random or not. Methods for generating a truly random sequence include, for example, use of a computer-generated randomization sequence or number table. Randomization methods that depended upon patient characteristics such as birthdays or initials were considered high risk of bias in this area. The reviewers also assessed whether concealment of random allocation was described and whether random allocation appeared to be adequately concealed based on the described methods. Only studies that described both appropriate methods of randomization and adequate concealment of random allocation were determined to be “low risk” for selection bias. We assessed the risk of selection bias to be “high risk” if the method of sequence generation was not random or if allocation was not concealed or inadequately concealed. We categorized the risk of selection bias as “unclear” if the methods of randomization or the methods of allocation concealment were not described.

• Information bias: We categorized the risk of information bias as “low” if participants, treating providers, outcomes assessors, and data analysts were masked as to the intervention group. We categorized the risk of information bias as “high” if one or more of these participants, treating providers, outcomes assessors, and data analysts were not masked or if masking was inadequate. We assessed the risk of information bias as “unclear” if masking was not described.

• Attrition bias: We categorized the risk of attrition bias as “low” if the reasons for patient dropout were described and were similar between intervention groups, if the number of patient dropouts was clearly reported, and if the methods for handling missing data were reported. We categorized the risk of attrition bias as “high” if the reasons for patient dropout were different between groups. If the number of patient dropouts was not reported or if the number randomized to each group were not reported, the risk of bias was determined to be “unclear.” Studies without a flow chart or adequately described data for patient dropout were considered “high risk of bias.” We categorized the risk of attrition bias as “unclear” if the study did not meet the aforementioned criteria for “high risk of bias” in this area and if the methods for handling missing data were unclear.

• Reporting bias: We categorized the risk of reporting bias as “low” if the study was registered in a clinical trials registry and if the primary outcomes were as described in the clinical trials registry. If the study was not registered in a clinical trials registry but if the protocol was previously published and if the primary outcomes were the same in the published results as in the protocol, the study was classified as “low risk of bias.” We categorized reporting bias as “high” if the primary outcomes did not match those described in the protocol or in the clinical trials registry. We categorized reporting bias as “unclear” if the study was not registered in a clinical trials registry and does not have a separate published protocol.

Statistical Analysis

Statistical analysis was performed according to the guidelines set by the Cochrane Collaboration.¹² If no evidence of a substantial difference in study populations, interventions, or outcome measurements was found, meta-analysis was performed. For dichotomous data, we calculated the summary adjusted odds ratio (aOR) with 95% confidence interval (CI). Heterogeneity of the results among studies was tested with the quantity I², which describes the percentage of total variation across studies that is due to heterogeneity rather than chance.¹³ A value of 0% indicates no observed heterogeneity, whereas I² values of ≥50% indicate a substantial level of heterogeneity.¹³ A fixed-effects model was used to pool data across studies if substantial statistical heterogeneity was not present. If I² values were ≥50%, a random effects model was used to pool data across studies if causes of heterogeneity could not be determined and if the average treatment effect was considered clinically meaningful. We assessed publication and related biases visually by examining the symmetry of funnel plots and statistically by using the Egger test (data not shown). A probability value of <0.1 was considered to indicate significant asymmetry. Analyses were performed using the latest Review Manager statistical software (RevMan, version 5.3.5, Nordic Cochrane Centre, Copenhagen Denmark).

Results

Eligible Studies

A total of 1,259 articles were identified through our search strategy. After duplicates were removed and titles were screened for abstracts and contents, 16 articles were reviewed for inclusion. An additional 12 were excluded for content (Fig. 1).

Fig. 1.

PRISMA diagram of study selection process.

Study Characteristics and Results

The description of the studies included in this meta-analysis is summarized in Table 1. Four studies were included in this meta-analysis. One study was conducted in Israel,⁵ one in the United States,⁶ one in Egypt,⁷ and one in Australia.⁸ All studies were prospective controlled studies, and all included a comparison of single-balloon catheter and double-balloon catheter for labor induction. None of the studies used concomitant prostaglandin at the time of balloon insertion. All of the inductions were of term pregnancies. A total of 682 patients were analyzed, with 340 patients in the single-balloon group and 342 in the double-balloon group.

Table 1.

Description of studies

Study	Type	Country and duration	Catheters Compared	Single balloon, n	Double balloon, n	Exclusion criteria	Parity included	Outcome measured	Results
Salim et al, 2011	RCT	Israel (2008–2010)	Single-balloon (60 mL) vs. double-balloon (80 mL each side)	145	148	Contraindication to vaginal delivery, previous cesarean delivery, low-lying placenta, fetal malformations, IUFD, clinical amnionitis, women who were carriers of hepatitis B, C, or HIV, women with a latex allergy	Nulliparous Multiparous	Time duration from catheter insertion to delivery	No difference between groups or when analyzed nulliparous women only
Sayed Ahmed et al, 2016	RCT	Egypt (2013–2014)	Single-balloon (50 mL) vs. double-balloon (80 mL each side)	37	37	Previous cesarean delivery, macrosomic babies, nonreassuring fetal conditions, ruptured membranes, placenta previa, malpresentation	Nulliparous	Cervical ripening and changes in bishop score	Double-balloon catheter results in greater ripening, but duration from balloon insertion to delivery was significantly shorter with single-balloon catheter
Hoppe et al, 2015	RCT	United States (2010–2013)	Single-balloon (30 mL) vs. double-balloon (80 mL each side)	48	50	Contraindication to vaginal delivery, planned or received exogenous prostaglandin administration as an initial induction agent, unexplained vaginal bleeding, active HSV, more than one previous cesarean delivery, history of a classical cesarean delivery, previous attempt at induction of labor during current pregnancy, non-English speaking	Nulliparous Multiparous	Bishop score ≥ 6 at the time of catheter removal, mean change in Bishop score after removal	Double-balloon catheter more effective for preinduction ripening and achieving vaginal delivery
Pennell et al, 2009	RCT	Australia (2001–2003)	Single-balloon (30 mL) vs. double-balloon (80 mL each side) vs. PGE2 gel	110	107	Age < 16 y, previous uterine surgery, low-lying placenta, any active or purulent infection of the lower vaginal tract, or an abnormal preinduction fetal heart rate tracing	Nulliparous	Cesarean delivery rate	No difference in the rate of cesarean delivery. The induction to delivery interval was longer in the double balloon group

Abbreviations: HIV, human immunodeficiency virus; HSV, herpes simplex virus; IUFD, intrauterine fetal death; PGE2, prostaglandin E2; RCT, randomized controlled study.

One study by Salim et al⁵ instilled 60 mL into the single-balloon catheter and another by Sayed Ahmed et al⁷ instilled 50 mL. The remainder of the studies all instilled 30 mL into the single-balloon catheter and 80 mL on either side of the double-balloon catheter.^6,8 Two studies only included nulliparous women,^7,8 and the remainder included both nulliparous and multiparous women eligible for induction with mechanical methods.^5,6

For the primary outcome, all studies reported time from catheter insertion to delivery by hours and minutes. One study was excluded from this primary outcome calculation as the time from catheter insertion to delivery was reported as median and interquartile range only.⁶ A total of three studies were used in the calculation of the primary outcome, with a total of 292 patients in each arm and a total sample size of 584 patients.^5,7,8 A second study was excluded as it reported this primary outcome on a log value using a Kaplan–Meier curve, limiting statistical extraction of standard deviation.⁸

Two studies did not specify vaginal delivery as operative or spontaneous but as “normal”⁷ or “vaginal delivery mode.”⁶ All studies reported the number of cesarean deliveries for both groups. The use of epidural, as a marker for pain, for each group was reported for three studies.^5,6,8

This meta-analysis demonstrated that there was no statistical difference in time to delivery between single- and double-balloon catheters (18.8 vs. 19.6 hours; mean difference [MD]: 0.40; 95% CI: −1.56 to 0.76; Fig. 2).

Fig. 2.

Time from catheter insertion to delivery.

There was no statistical difference between vaginal delivery (65.3 vs. 62.3%; aOR: 1.04; 95% CI: 0.56–1.92; Fig. 3), cesarean delivery (25.6 vs. 27.5%; aOR: 0.98; 95% CI: 0.55–1.73; Fig. 4), or epidural use (58.4 vs. 62%; aOR: 0.81; 95% CI: 0.56–1.18, Fig. 5) between the single- and double-balloon group (25.6 vs. 27.5%; aOR: 0.98; 95% CI: 0.55–1.73).

Fig. 3.

Comparison of vaginal delivery rate.

Fig. 4.

Comparison of cesarean delivery rate.

Fig. 5.

Comparison of epidural use.

Fig. 6 shows the risk of bias in the trials that were included in this review. No trials were assessed as having a low risk of bias across all domains. All trials had a low risk of bias in the blinding of outcome assessment. However, most of the trials had an unclear risk of bias overall. No trials were assessed as having a high risk of bias in all domains. As expected, most of the trials had a high risk of bias in allocation concealment because it is difficult to conceal a single- or a double-balloon Foley catheter from either the patient or providers.

Fig. 6.

Risk of bias of included studies. The blue circles represent low risk of bias, whereas the red and yellow circles represent high risk and unclear risk of bias, respectively.

Comment

The findings of our meta-analysis demonstrate no statistically significant difference when comparing single- versus double-balloon catheters with respect to time to delivery, rates of vaginal delivery, cesarean delivery, or epidural use for labor induction.

The strength of this meta-analysis is that it provides data regarding the use of the two mechanical methods for labor induction without concomitant prostaglandin use. The individual studies provide unclear conclusions and are limited by their small sample size and conflicting results. Some studies reported shorter delivery times with the single-balloon group⁷ and others reported longer delivery times.⁸ The same inconclusive results are also drawn across the studies in regard to rates of vaginal delivery, cesarean delivery, and analgesia use. By analyzing the data by way of meta-analysis, the sample size in our study was markedly increased compared with the largest study in our analysis.⁵ As such, our findings of no difference across outcomes reflect a composite evaluation of the data.

Additional strengths of our study are the consistency in measured primary and secondary outcomes in most of the included studies. All the studies included time to delivery measured using minute/hour increments and defined this outcome as time from catheter insertion to delivery. Cesarean delivery was reported consistently across studies, and epidural use was evaluated as a binary “yes or no” answer across studies. Hence, it was easy to perform a meta-analysis of the data.

Unfortunately, this consistency in individual study design did not apply to all outcomes. For our primary outcome of time to delivery, one study had to be excluded because of statistical limitations, limiting our analysis of the primary outcome to three studies.⁶ It should be noted that two studies did not use the standard 30- and 80-mL fill volume for single- and double-balloon catheters, respectively.^5,7 These two studies both used 80 mL for the double balloon. However, their single balloons were filled with 50 or 60 mL which is different from the other studies in this analysis. As such, this could have skewed the results.

Finally, another factor that may have contributed to the results is the stratification of nulliparous subjects versus multiparous subjects. Salim et al and Hoppe et al both accounted for this in their analysis with subanalysis and matching^5,6 but Sayed Ahmed et al did not, and Pennell et al only included nulliparous women.^7,8 The heterogeneity of the data should be considered when interpreting the results.

Our findings suggest that the type of catheter used at the time of labor induction did not affect the studied primary and secondary outcomes. Further studies should be performed using standardized labor protocols after removal of the balloon to better elucidate an isolated effect of single versus double balloons in the early stages of labor. In our institution, the cost of a single-balloon catheter is US$7.32 versus US$41 for a double-balloon catheter. Given that the findings of our study demonstrated no added advantage between the two methods, we would propose favoring the single-balloon catheter.