DOI: 10.1111/1471-0528.

13456 Systematic review

www.bjog.org

A systematic review and network meta-analysis

comparing the use of Foley catheters,
misoprostol, and dinoprostone for cervical
ripening in the induction of labour
W Chen,a J Xue,b MK Peprah,c SW Wen,d,e M Walker,d,e Y Gao,f Y Tangg
a
Department of Nephropathy, Xiangya Hospital, Central South University, Changsha, Hunan, China b Department of Medical Records
Information, Xiangya Hospital, Central South University, Changsha, Hunan, China c Department of Epidemiology and Community Medicine,
University of Ottawa, Ottawa, ON, Canada d OMNI Research Group, Department of Obstetrics and Gynecology, University of Ottawa,
Ottawa, ON, Canada e Ottawa Hospital Research Institute Clinical Epidemiology Program, Ottawa, ON, Canada f Department of Obstetrics
and Gynaecology, Southern Medical University, Guangzhou, Guangdong, China g Department of Urology, The Third Xiangya Hospital of
Central South University, Changsha, Hunan, China
Correspondence: Dr Y Tang, Department of Urology, The Third Xiangya Hospital of Central South University, 138 Tongzipo Road, 410013
Changsha, China. Email yuxintang1874@126.com

Accepted 24 March 2015. Published Online 4 November 2015.

Background Various methods are used for cervical ripening
during the induction of labour. It is still debatable which of these
methods of treatment is optimal.
Objective To compare treatment techniques for cervical ripening
in the induction of labour.
Search strategy Medline, Embase, and the Cochrane
Collaboration databases were searched using the keywords
‘cervical ripening’, ‘labour induced’, ‘misoprostol’, ‘dinoprostone’,
and ‘Foley catheter’.
Selection criteria Randomised controlled trials (RCTs) of cervical
ripening during the induction of labour, evaluating rates of failure
to achieve vaginal delivery within 24 hours, incidence of uterine
hyperstimulation with fetal heart rate (FHR) changes, and rates of
caesarean section. Studies including women with prelabour
rupture of membranes were excluded.
Data collection and analysis Outcome data were collected and
analysed through pairwise meta-analysis and network meta-
analysis within a Bayesian framework.
Main results A total of 96 RCTs (17 387 women) were included in
the meta-analysis. Vaginal misoprostol was the most effective
cervical ripening method to achieve vaginal delivery within
24 hours, but had the highest incidence of uterine hyperstimulation
with FHR changes. The use of a Foley catheter to induce labour was
associated with the lowest rate of uterine hyperstimulation
accompanied by FHR changes. The caesarean section rate was
lowest using oral misoprostol for the induction of labour.
Author’s conclusions No method of labour induction demonstrated
overall superiority when considering all three clinical outcomes.
Decisions regarding the choice of induction method will depend
upon the relative preference for effecting vaginal delivery within
24 hours, minimising the incidence of uterine hyperstimulation with
adverse FHR changes and avoiding caesarean section.
Keywords Cervical ripening, dinoprostone, Foley catheter,
induction of labour, misoprostol.
Tweetable abstract Oral misoprostol for the induction of labour
is safer than vaginal misoprostol and has the lowest rate of
caesarean section.
Linked article This article is commented on by K Hemming and
M Price, p.355 in this issue. To view this mini commentary visit
http://dx.doi.org/10.1111/14710528.13533.

Please cite this paper as: Chen W, Xue J, Peprah MK, Wen SW, Walker M, Gao Y, Tang Y. A systematic review and network meta-analysis comparing the
use of Foley catheters, misoprostol, and dinoprostone for cervical ripening in the induction of labour. BJOG 2016;123:346–354.

when the benefits of induction outweigh the risk to con-

Introduction
Induction of labour is defined as the use of artificial meth-
ods resulting in labour after the age of fetal viability and
before the spontaneous onset of labour.1 It is considered
tinue with a pregnancy.2 In 2012, about 23.3% of singleton
births were induced in the USA, which is more than double
the incidence of 9.5% in 1990.3 The use of oxytocin or arti-
ficial rupture of the membranes (ARM) is less likely to

346 ª 2015 Royal College of Obstetricians and Gynaecologists


induce labour successfully in the absence of a favourable
cervix. In such circumstances cervical ripening methods
that soften, thin, and dilate the cervix are often required to
induce labour.4 Techniques employed are mechanical or
pharmacological.
The insertion of a Foley catheter into the cervical canal
is one of the more commonly used mechanical methods.
The technique was first described for the induction of
labour in 1967,5 and is more cost-effective compared with
other mechanical methods.6 Pharmacological methods
include the use of prostaglandins, oxytocin, estrogens, and
mifepristone. Prostaglandins, which are cyclopentane
derivatives of arachadonic acid, are widely used in obstet-
rics and gynaecology. Prostaglandin E2, also known as
dinoprostone, is the only prostaglandin approved by the
US Food and Drug Administration (FDA) for cervical
ripening in labour induction;7 However, dinoprostone is
expensive and requires cold storage. Misoprostol, a prosta-
glandin E1 analogue approved for treating gastric ulcers
caused by non-steroidal anti-inflammatory drugs, is often
used as an off-label drug for inducing labour. The effective-
ness of misoprostol in cervical ripening has been demon-
strated, but several case reports have suggested that the rate
of serious complications, such as excessive uterine contrac-
tions and rupture, may be increased compared with other
methods.8 In addition, there are a variety ways to adminis-
ter misoprostol that include oral, vaginal, sublingual, or
buccal routes,9 although the latter two routes are not cur-
rently recommended for labour induction.10,11
Research studies comparing the safety and effectiveness
of different methods of cervical ripening are inconsistent,
such that the optimal method of induction of labour
remains unclear. We therefore conducted a network meta-
analysis comparing the five most commonly used methods
for cervical ripening in labour induction – Foley catheter,
vaginal misoprostol, oral misoprostol, vaginal dinopros-
tone, and intracervical dinoprostone – among pregnant
women with intact membranes. We aimed to provide a
comprehensive summary of the existing evidence to further
inform clinical practice and aid in the design of future
trials.12–14
Methods
Data sources
We searched the Embase (1947–20 May 2014), Medline
(1946–20 May 2014), and Cochrane Collaboration
(searched up to 20 May 2014) databases for randomised
controlled trials (RCTs) and reviews of cervical ripening or
labour induction. The search terms used were ‘cervical
ripening’, ‘labour induced’, ‘misoprostol’, ‘dinoprostone’,
and ‘Foley catheter’. We also hand-searched the reference
lists from meta-analyses and review articles about this topic
ª 2015 Royal College of Obstetricians and Gynaecologists
Methods of cervical ripening: a systematic review
for all relevant eligible articles. The search strategy was
restricted to English language studies. A protocol was regis-
tered at the Centre for Reviews and Dissemination, Univer-
sity of York, for the review (reg. no. CRD42014009755).
Inclusion criteria and study selection
Studies included were of RCT design and reported on com-
parisons between different interventions for cervical ripen-
ing in women with an unfavourable cervix, and with intact
membranes. The interventions included in this review were
Foley catheter, vaginal misoprostol, oral misoprostol, vagi-
nal dinoprostone, and intracervical dinoprostone. The out-
comes we focused on were failure to achieve vaginal
delivery within 24 hours, uterine hyperstimulation with
fetal heart rate (FHR) changes, and caesarean section.
Hyperstimulation with FHR changes was defined as exces-
sive uterine activity (tachysystole or hypersystole) with a
non-reassuring FHR pattern, including tachycardia, persis-
tent decelerations, and decreased short-term variability.
Trials were excluded if they included the following cases:
women whose pregnancies were not more than 28 weeks of
gestational age; non-cephalic presentations; multiple preg-
nancies; and women with previous caesarean sections. Pub-
lished RCTs that studied combination therapies and those
that reported non-relevant outcomes were also excluded, as
were conference papers, abstracts, letters, communications,
or supplements. The authors of studies that provided insuf-
ficient information to make inclusion or exclusion deci-
sions (for example, trials that did not clearly report the
membrane status of participants) were contacted for fur-
ther information. We did not exclude studies in which oxy-
tocin or ARM was administered for the augmentation of
labour.
Two reviewers (WC and JX) reviewed potentially eligible
studies independently to determine whether studies met the
inclusion criteria. Discordance was resolved by discussion,
and if not a third author (MP) was involved.
Data abstraction and quality assessment
Two reviewers (WC and JX) extracted data on the first
author’s surname, geographic location of study, year of
publication, patient characteristics (gestational age and
Bishop scores at entry into the study), number of partici-
pants, interventions, and relevant outcomes. The outcome
data were extracted by intention-to-treat principle.
Study quality was evaluated using the Cochrane evidence
quality assessment tool. Seven specific domains were
assessed, including random sequence generation, allocation
concealment, blinding, incomplete outcome data, selective
reporting, and other sources of bias.15 Assessment of each
specific bias was rated as low risk, high risk, or unclear.
Disagreements among reviewers were discussed, and agree-
ment was reached by consensus.
347
 Chen et al.
Statistical analysis
We chose three outcomes for the network meta-analysis:
vaginal delivery not achieved in 24 hours; uterine hyper-
stimulation with FHR changes; and caesarean section. Pair-
wise meta-analyses were performed to estimate direct
comparisons between cervical ripening methods. Bayesian
network meta-analysis (NMA) was then performed to com-
bine the results from direct comparisons and indirect com-
parisons.
Pairwise meta-analysis was performed using REVIEW MAN-
AGER 5.0 (Cochrane Collaboration, Copenhagen, Den-
mark).16 The random-effects model was adopted rather than
the fixed-effects model because not all studies were func-
tionally equivalent or shared a common effect size. We used
the I2 statistical method to evaluate study heterogeneity. A
value greater than 50% was considered to be substantial.
The overall risk ratio (RR) and 95% confidence intervals
(95% CIs) were estimated for the various treatments.15
The network meta-analysis was conducted within a Baye-
sian framework using the GeMTC GUI statistical package.17
The network meta-analysis combines all information from
treatment arms to form a single, integrated, consistent set
of estimates, while fully preserving the randomisation in
the trials. The rank of each intervention and its rank prob-
ability were calculated to show the comparative efficacy. In
our Bayesian analysis, we used non-informative uniform
prior distribution for treatment-effect parameters, as infor-
mative priors were not available with these data. Four Mar-
kov chains ran simultaneously in our analysis, and the
simulations were repeated 20 000 times and discarded to
allow for model convergence. Additional simulations were
repeated 100 000 times to produce the posterior probabili-
ties. The Gelman–Rubin–Brooke method was used to assess
Potenally relevant randomized controlled trials idenfied
through database searching (n = 3272)
Embase (n = 1043) Medline (n = 950) Cochrane libarary (n = 1279)
Screened records (n = 1984)
Full text arcle assessed for eligibility
(n = 342)
Randomized controlled trials included in network meta analysis (n = 96)
Two arms (n = 92) Three arms(n = 4)
Figure 1. Flow chart of study selection.
348
the convergence.18 When three treatments were connected
as a loop, the inconsistency in network meta-analysis could
be assessed through node-splitting analysis. Two posterior
distributions were generated independently from direct evi-
dence and indirect evidence on a specific node (compar-
ison). Then the compatibility of the two posterior
distributions was measured to assess whether direct and
indirect evidence was in agreement. A Bayesian P value for
the node-splitting analysis was calculated, as a large P value
indicated no inconsistency.19
Sensitivity analysis was carried out to assess the robust-
ness of the results by the omission of specific studies, speci-
fic treatment arms, or by considering gestational age,
Bishop’s score at the commencement of induction, the
dosage of prostaglandins, and the volume of the Foley
catheter. We assessed the impact of different dosage by
dividing it into low dosage and high dosage. We considered
doses of >0.5 mg intracervical dinoprostone, >3 mg vaginal
dinoprostone, and >50 micrograms oral misoprostol or
vaginal misoprostol as high doses.
Results
Study selection and characteristics
The flow diagram of the electronic search details and selec-
tion process are showed in Figure 1. A total of 96 RCTs
(17 387 women) with usable outcome data were eligible to
be included in our network meta-analysis. The trials were
published between 1983 and 2014. Most trials were from
the USA (25/96), followed by India (12/96). Table S1
details the study characteristics and our evaluation of the
risk of bias. In general, the included participants were sin-
gleton pregnancies with an unfavourable cervix, intact
Excluded duplicates
(n = 1288)
Records excluded based on tle or
abstract (n = 1642)
No relevant populaon (n = 178)
No relevant intervenons (n = 48)
No relevant outcomes (n = 11)
Not a ramdomized controlled trail (n = 9)
ª 2015 Royal College of Obstetricians and Gynaecologists
 Methods of cervical ripening: a systematic review

membranes, and with the fetus in cephalic presentation in
the third trimester of pregnancy. Figure S1 indicates the
evidence of the network that shows the interventions and
numbers of direct comparisons.
Assessment of risk of bias
We assessed the risk of bias of the included trials according
to the Cochrane evidence quality assessment tool
(Table S1). The generation of the random sequence and
allocation concealment were reasonable in most trials (67
and 65%, respectively). The lack of blinding in most of the
study trials was likely to be the main source of bias; only
25 trials reported an adequate description of blinding of
the study subjects. Incomplete outcome data and selective
reporting were considered a low risk of bias in more than
half of the included trials.
Vaginal delivery not achieved in 24 hours
A total of 51 studies (n = 10 305) contributed to the
analysis of vaginal delivery not achieved within 24 hours.
A total of 1486 women (14.4%) were assigned to
Foley catheter, 2557 (24.8%) were assigned to vaginal
misoprostol, 1528 (14.8%) were assigned to oral miso-
prostol, 3193 (31.0%) were assigned to vaginal dinopros-
tone, and 1541 (15.0%) were assigned to intracervical
dinoprostone.
Table 1 shows the pooled estimates of network meta-
analysis and traditional pairwise meta-analysis. In the
network meta-analysis, vaginal misoprostol showed a sig-
nificant difference in reducing the number of vaginal
deliveries not achieved in 24 hours, compared with vagi-
nal dinoprostone (RR 0.62, 95% CI 0.49–0.79), Foley
catheter (RR 0.48, 95% CI 0.35–0.67), oral misoprostol
(RR 0.44, 95% CI 0.33–0.58), and intracervical dinopros-
tone (RR 0.43, 95% CI 0.34–0.54). Vaginal dinoprostone
demonstrated a reduction in the number of vaginal
deliveries not achieved in 24 hours, compared with oral
misoprostol (RR 0.70, 95% CI 0.52–0.95) and intracervi-
cal dinoprostone (RR 0.69, 95% CI 0.53–0.89). The com-
parisons among Foley catheter, intracervical dinoprostone,
and oral misoprostol showed no statistical significance in
the outcome of vaginal delivery not achieved in
24 hours. Figure S2 shows the rank probabilities for each
intervention for vaginal delivery not achieved in
24 hours. Vaginal misoprostol has the highest probability
(100%) of being ranked the best treatment option. Vagi-
nal dinoprostone has the highest probability (95%) of
being the second-best treatment method, followed by the

Table 1. Results of network meta-analysis of vaginal delivery not achieved in 24 hours

Comparators (numbers of cases/participants) Head-to- Risk ratio (95% CI)* I2** P***
head trials

Vaginal misoprostol Vaginal dinoprostone (1322/3193) 8 0.62 (0.49–0.79) 62% 0.88

(771/2557) 0.77 (0.65–0.92)
Foley catheter (718/1486) 4 0.48 (0.35–0.67) 0% 0.20
0.81 (0.64–1.03)
Oral misoprostol (650/1528) 8 0.44 (0.33–0.58) 35% 0.26
0.63 (0.54–0.73)
Intracervical dinoprostone (622/1541) 13 0.43 (0.34–0.54) 0% 0.93
0.68 (0.61–0.75)
Vaginal dinoprostone Foley catheter 7 0.77 (0.57–1.03) 88% 0.05
0.79 (0.61–1.02)
Oral misoprostol 4 0.70 (0.52–0.95) 30% 0.43
0.89 (0.78–1.00)
Intracervical dinoprostone 8 0.69 (0.53–0.89) 31% 0.62
0.87 (0.75–1.02)
Foley catheter Oral misoprostol 0 0.91 (0.61–1.35) NA NA
NA
Intracervical dinoprostone 1 0.89 (0.63–1.28) NA 0.14
0.58 (0.34–0.99)
Oral misoprostol Intracervical dinoprostone 3 0.99 (0.73–1.34) 82% 0.92
0.98 (0.65–1.46)

*Results of network meta-analysis and pairwise meta-analysis, respectively.

**Heterogeneity of pairwise meta-analysis.
***P value from node-splitting method.

ª 2015 Royal College of Obstetricians and Gynaecologists 349

 Chen et al.

Foley catheter, with a 59% probability of being ranked
in the third position.
Uterine hyperstimulation with fetal heart rate
changes
A total of 58 studies (n = 10 342) contributed to the analy-
sis of uterine hyperstimulation with FHR changes: 1578
women (15.3%) were allocated to receive Foley catheter,
3020 (29.1%) were allocated to vaginal dinoprostone, 3172
(30.7%) were allocated to vaginal misoprostol, 1234
(12.0%) were allocated to oral misoprostol, and 1338
(12.9%) were allocated to intracervical dinoprostone.
Table 2 shows the pooled estimates for the rate of uter-
ine hyperstimulation with FHR changes. In the network
meta-analysis, the Foley catheter demonstrated a significant
reduction in the rate of uterine hyperstimulation compared
with vaginal misoprostol (RR 0.15, 95% CI 0.07–0.30).
Intracervical dinoprostone had a lower risk of uterine
hyperstimulation with FHR changes compared with vaginal
dinoprostone (RR 0.46, 95% CI 0.25–0.87) and vaginal
misoprostol (RR 0.26, 95% CI 0.15–0.44). Compared with
vaginal misoprostol, oral misoprostol (RR 0.44, 95% CI
0.23–0.88) and vaginal dinoprostone (RR 0.57, 95% CI
0.36–0.88) both revealed a reduction in uterine hyperstim-
ulation with FHR changes. Figure S3 shows rank probabili-
ties for each intervention for uterine hyperstimulation with
FHR changes. Foley catheter ranked the best (rank 1), with
a probability of 90%. Vaginal misoprostol had the highest
probability (99%) of being the worst ranked treatment with
regard to uterine hyperstimulation with FHR changes.
Caesarean section
A total of 95 studies (n = 16 311 women) contributed to the
analysis of caesarean section: 1691 women (10.4%) were
assigned to oral misoprostol, 4144 (25.4%) were assigned to
vaginal misoprostol, 2961 (18.2%) were assigned to intracer-
vical dinoprostone, 4841 (29.7%) were assigned to vaginal
dinoprostone, and 2674 (16.3%) were assigned to Foley
catheter.
Table 3 shows the pooled estimates for the outcomes of
caesarean section in network meta-analysis and pairwise
meta-analysis. In the network meta-analysis, compared
with intracervical dinoprostone, oral misoprostol yielded
the most significant effect on reducing caesarean section
(RR 0.74, 95% CI 0.60–0.93), followed by vaginal miso-
prostol (RR 0.82, 95% CI 0.70–0.95). Figure S4 shows the
rank probabilities for each intervention for the rate of cae-
sarean section. Oral misoprostol ranked the highest on
reducing the likelihood of caesarean section, with a proba-
bility of 83%, followed by vaginal misoprostol, which
showed the highest probability (80%) for the second
ranked position.

Table 2. Results of network meta-analysis of hyperstimulation with FHR changes

Comparators (numbers of cases/participants) Head-to-head Risk ratio (95% CI)* I2** P***
trials

Foley catheter (21/1578) Intracervical dinoprostone (27/1338) 0 0.58 (0.24–1.33) NA NA

NA
Oral misoprostol (26/1234) 0 0.34 (0.13–0.82) NA NA
NA
Vaginal dinoprostone (95/3020) 6 0.27 (0.12–0.52) 23% 0.18
0.26 (0.09–0.77)
Vaginal misoprostol (193/3172) 8 0.15 (0.07–0.29) 0% 0.11
0.37 (0.20–0.67)
Intracervical dinoprostone Oral misoprostol 2 0.59 (0.25–1.30) 0% 0.80
0.44 (0.11–1.72)
Vaginal dinoprostone 7 0.46 (0.25–0.87) 0% 0.61
0.64 (0.26–1.59)
Vaginal misoprostol 18 0.26 (0.15–0.44) 0% 0.70
0.36 (0.23–0.58)
Oral misoprostol Vaginal dinoprostone 3 0.78 (0.39–1.59) 0% 0.97
0.84 (0.39–1.83)
Vaginal misoprostol 7 0.44 (0.23–0.88) 0% 0.64
0.48 (0.20–1.14)
Vaginal dinoprostone Vaginal misoprostol 14 0.57 (0.36–0.88) 0% 0.42
0.72 (0.49–1.05)

*Results of network meta-analysis and pairwise meta-analysis, respectively.

**Heterogeneity of pairwise meta-analysis.
***P value from node-splitting method.

350 ª 2015 Royal College of Obstetricians and Gynaecologists

 Methods of cervical ripening: a systematic review

Table 3. Results of network meta-analysis of caesarean section

Comparators (numbers of cases/participants) Head-to-head Risk ratio (95% CI)* I2** P***
trials

Oral misoprostol (350/1691) Vaginal misoprostol (930/4144) 9 0.91 (0.74–1.11) 0% 0.93

0.94 (0.79–1.13)
Vaginal dinoprostone (1108/4841) 6 0.79 (0.66–0.96) 0% 0.83
0.83 (0.71–0.98)
Foley catheter (689/2674) 1 0.76 (0.61–0.95) NA 0.87
0.86 (0.33–2.20)
Intracervical dinoprostone (620/2961) 3 0.74 (0.60–0.93) 0% 0.73
0.84 (0.58–1.22)
Vaginal misoprostol Vaginal dinoprostone 16 0.87 (0.76–1.00) 1% 0.54
0.98 (0.86–1.12)
Foley catheter 12 0.84 (0.71–0.99) 30% 0.17
0.80 (0.65–0.98)
Intracervical dinoprostone 21 0.82 (0.70–0.95) 12% 0.38
0.90 (0.77–1.06)
Vaginal dinoprostone Foley catheter 10 0.96 (0.82–1.13) 16% 0.54
0.98 (0.85–1.14)
Intracervical dinoprostone 20 0.94 (0.81–1.10) 0% 0.46
0.96 (0.81–1.14)
Foley catheter Intracervical dinoprostone 5 0.98 (0.81–1.18) 0% 0.12
0.75 (0.56–1.00)

*Results of network meta-analysis and pairwise meta-analysis, respectively.

**Heterogeneity of pairwise meta-analysis.
***P value from node-splitting method.

Heterogeneity, inconsistency, and sensitivity
analysis
The results of traditional pairwise meta-analysis are presented
along with results of network meta-analysis in Tables 1–3.
Heterogeneity evaluated by I2 showed some statistical signifi-
cance in the analysis of vaginal delivery not achieved in
24 hours, whereas we did not find any significant heterogene-
ity in the analysis of uterine hyperstimulation with FHR
changes and caesarean section. The node-splitting method
showed no significant inconsistency except in the comparison
between vaginal dinoprostone versus Foley catheter in the
analysis of vaginal delivery not achieved in 24 hours.
Sensitivity analysis was undertaken to explore the impact
of different levels of bias. The comparisons among studies
were complicated because of the variation in dosages used
for misoprostol and dinoprostone. In the vaginal misopros-
tol group, more than half of the trials studied misoprostol
at a dose of 50 micrograms (34/59), whereas a dose of
25 micrograms of misoprostol was used in 22 trials (22/
59). In the oral misoprostol group, a dose of 50 micro-
grams of misoprostol was the most commonly used (12/
19). In the intracervical dinoprostone group, almost all tri-
als used a dose of 0.5 mg of dinoprostone (49/50). In the
vaginal dinoprostone group, the dosage of dinoprostone
ranged from 1 to 10 mg (1 mg, nine studies; 3 mg, 12
studies; 2 mg, nine studies; 10 mg, 14 studies).
After excluding studies with ‘high dose’ regimes, there
was no significant change compared with the original
results. We also reanalysed the data after excluding studies
that had large sample sizes, which may dominate a specific
treatment effect, or by excluding specific treatment arms.
The results did not change significantly after this sensitivity
analysis. The significance of some results changed after sen-
sitivity analyses according to gestational age, Bishop scores,
and Foley catheter volume, although the rank of each treat-
ment remained the same (see Tables S2–S4).
Discussion
Main findings
This study reviewed five different methods for cervical
ripening in labour induction through network meta-analy-
sis. It showed that vaginal misoprostol was the most effec-
tive treatment for achieving vaginal delivery in 24 hours.
Foley catheter was the method of induction least likely to
cause uterine hyperstimulation with FHR changes, followed
by intracervical dinoprostone. The use of oral misoprostol
resulted in the lowest rate of caesarean section compared
with the rest of the techniques, except for vaginal miso-
prostol.
Although vaginal misoprostol performed well in achiev-
ing vaginal delivery within 24 hours, it was at the cost of

ª 2015 Royal College of Obstetricians and Gynaecologists 351

 Chen et al.
having the highest occurrence of uterine hyperstimulation
with FHR changes. Compared with vaginal misoprostol,
oral misoprostol had a reduced incidence of uterine hyper-
stimulation with FHR changes, but at the expense of signif-
icantly fewer vaginal deliveries within 24 hours. Compared
with all other cervical ripening methods, oral misoprostol
ranked first in terms of avoidance of delivery by caesarean
section, even though the difference was not significant
compared with vaginal misoprostol. Although many
women may appreciate a rapid birth, the primary aim of
induction is to achieve a safe, vaginal delivery. Thus, the
use of a Foley catheter or oral misoprostol may be prefer-
able to vaginal misoprostol for the induction of labour in
women with intact membranes and an unfavourable
cervix.20,21
Strengths and limitations
As far as we know, this is the first network meta-analysis
comparing different cervical ripening interventions used for
the induction of labour. Our network meta-analysis pro-
vides comprehensive comparisons between five commonly
used cervical ripening techniques, and provides estimates of
comparisons that are rarely reported in head-to-head trials.
Through a Bayesian approach, we reported the probabilities
of ranking for these cervical ripening methods, even when
pairwise meta-analysis detected no significant differences
between them. We tried to minimise potential bias by
reducing the variation in the characters of included women
by applying several restrictions for inclusion in the review.
For instance, we excluded studies that included women
with ruptured membranes or who were in the second tri-
mester. Thus, the results obtained in this review should be
interpreted in light of the populations studied. The results
of sensitivity analysis indicated that the overall outcome
effects were stable and justified.
Our review has limitations. We may have missed studies
by reviewing those published in the English literatures only.
We did not assess other clinical outcomes such as maternal
death, perinatal death, and uterine rupture because these
variables were not commonly reported. Three significant
heterogeneities were detected in the analysis of vaginal
delivery not achieved in 24 hours. As a result, we
attempted to perform a subgroup analysis stratified by
Bishop score, but heterogeneities among studies in the
pairwise meta-analysis remained. Subtle differences in treat-
ment, such as the use of oxytocin or time of amniotomy
for augmentation, could have contributed to clinical
heterogeneity. Through the node-splitting method we
found no inconsistency except in the comparison between
vaginal dinoprostone and Foley catheter in vaginal delivery
not achieved in 24 hours. The discrepant results between
direct comparison and indirect comparison may arise from
the heterogeneity among the studies.22 We were unable to
352
stratify for the dosage of drugs, as this varied among study
trials, but as part of a sensitivity analysis we divided drugs
into low-dose and high-dose regimes. This approach did
not account for the varying dosing intervals (usually
between 2 and 6 hours).
Interpretation (in light of other evidence)
Our results are mostly compatible with previous pairwise
meta-analysis. A recent Cochrane review found that women
induced with intact membranes using oral misoprostol had
a lower rate of caesarean section, compared with vaginal
dinoprostone (RR 0.81, 95% CI 0.70–0.93).20 There was no
significant difference with regards to delivery by caesarean
section between oral misoprostol with intracervical dino-
prostone (RR 0.81, 95% CI 0.65–1.00) or vaginal misopros-
tol (RR 1.00, 95% CI 0.89–1.14), respectively. Recent
research compared Foley catheter induction with 25 micro-
grams vaginal misoprostol in women with intact mem-
branes.23 They found no difference in caesarean section
rates, but found a 61% reduction in the incidence of uter-
ine hyperstimulation with FHR changes in the Foley cathe-
ter group. Another comprehensive review in 2012
concluded that the incidence of hyperstimulation with FHR
changes was lower in transcervical balloon catheter groups,
compared with any prostaglandins (RR 0.19, 95% CI 0.08–
0.43), regardless of the status of the membranes.24 Hofmeyr
and colleagues found that vaginal misoprostol achieved
more vaginal deliveries in 24 hours than vaginal dinopros-
tone (RR 0.78, 95% CI 0.67–0.91) or intracervical dinopro-
stone (RR 0.64, 95% CI 0.56–0.73), but led to more
hyperstimulation with FHR changes than vaginal dinopros-
tone (RR 1.88, 95% CI 1.29–2.72) and intracervical dino-
prostone (RR 3.62, 95% CI 2.22–5.90).21
In most countries misoprostol is not authorized to be
used for labour induction because of the concern that it
may be associated with a high rate of uterine hyperstimula-
tion. Our results suggest that oral misoprostol is safer than
vaginal misoprostol in this regard, and it performed better
than vaginal dinoprostone in reducing the likelihood of
caesarean section. In addition to being effective and cheap,
the administration of misoprostol by the oral route can be
more easily measured and precise compared with other
routes, and may be more acceptable to women. In develop-
ing countries or poorly resourced areas, we suggest oral
misoprostol can be the first choice for cervical ripening, as
it offers the best combination of cost, safety, and effective-
ness.
The Foley catheter has similar effects as oral misoprostol
in achieving vaginal delivery in 24 hours. With regards to
caesarean section, the use of the Foley catheter decreases
the risk of caesarean section compared with oxytocin
alone.25 As our results showed, the Foley catheter ranked
fourth in the comparisons with prostaglandins. We found
ª 2015 Royal College of Obstetricians and Gynaecologists

that the Foley catheter was associated with the lowest risk
of hyperstimulation with FHR changes, compared with
prostaglandin. In women at high risk of fetal hypoxaemia,
such as those with post-term pregnancy, sickle-cell disease,
pre-eclampsia, or intrauterine growth restriction, labour
induction with the Foley catheter may lead to a reduction
in fetal acidosis. The use of Foley catheters is low in cost
and easy for storage, compared with dinoprostone. Once a
Foley catheter is inserted, women wouldn’t need further
clinical intervention before the expulsion of the catheter
and less stringent monitoring of uterine contractions may
be needed during cervical ripening. Thus Foley catheters
could be the most appropriate method for home induction
of labour.
There is a major concern that Foley catheters may be
associated with higher rates of chorioamnionitis, however.
We did not perform network meta-analysis to compare the
rate of chorioamnionitis as this outcome was rarely
reported, especially in RCTs. According to the Cochrane
review by Jozwiak and colleagues, there is no evidence of
increased infection with a mechanical method.24 The
authors suggested that this should be interpreted cautiously
in view of the limited evidence available. The satisfaction of
women receiving the Foley catheter appears comparable
with those induced with dinoprostone, but women with
Foley catheters endure less pain.26
Conclusion
Vaginal misoprostol followed by vaginal dinoprostone are
the most effective methods for induction of labour after
28 weeks of gestation in women with intact membranes, in
terms of achieving delivery within 24 hours; however, these
methods are associated with higher rates of uterine hyper-
stimulation with adverse FHR changes. In contrast,
mechanical induction using Foley catheters was the least
effective method of induction, along with oral misoprostol
and intracervical dinoprostone, but had the lowest inci-
dence of uterine hyperstimulation with FHR changes. Oral
misoprostol was the best method of induction in terms of
reducing the likelihood of delivery by caesarean section,
and caused less uterine hyperstimulation with FHR changes
than vaginal misoprostol.
Disclosure of interests
None declared. Completed disclosure of interests form
available to view online as supporting information.
Contribution to authorship
WHC, YXT, and JX conceived and designed the study.
WHC and JX performed the literature search and data
extraction. MKP checked the data. WHC and JX performed
the data analysis. WHC, JX, and MKP wrote the article,
ª 2015 Royal College of Obstetricians and Gynaecologists
Methods of cervical ripening: a systematic review
and WHC, SWW, MW, and YFG were responsible for the
revision.
Details of ethics approval
There is no need for ethical approval for a systematic
review.
Funding
None.
Acknowledgement
None.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Evidence of network formed by interventions
and numbers of direct comparisons.
Figure S2. Rankings for vaginal delivery not achieved in
24 hours.
Figure S3. Rankings for hyperstimulation with FHR
changes.
Figure S4. Rankings for caesarean section.
Table S1. Characteristics of the included trials and qual-
ity assessment.
Table S2. Results of network meta-analysis for vaginal
delivery not achieved in 24 hours in sensitivity analyses.
Table S3. Results of network meta-analysis for uterine
hyperstimulation with FHR changes in sensitivity analyses.
Table S4. Results of network meta-analysis for caesarean
section in sensitivity analyses. &
References
1 Justus Hofmeyr G. Induction of labour with an unfavourable cervix.
Best Pract Res Clin Obstet Gynaecol 2003;17:777–94.
2 Jagani N, Schulman H, Fleischer A, Mitchell J, Randolph G. Role
of the cervix in the induction of labor. Obstet Gynecol
1982;59:21–6.
3 Osterman MJK, Martin JA. Recent declines in induction of labor by
gestational age. NCHS Data Brief 2014;155:1–8.
4 Xenakis EM, Piper JM, Conway DL, Langer O. Induction of labor in
the nineties: conquering the unfavorable cervix. Obstet Gynecol
1997;90:235–9.
5 Embrey MP, Mollison BG. The unfavourable cervix and induction of
labour using a cervical balloon. J Obstet Gynaecol Br Commonw
1967;74:44–8.
6 Lim CED, Ng RWC, Xu K. Non-hormonal methods for induction of
labour. Curr Opin Obstet Gynecol 2013;25:441–7.
7 Church S, Van Meter A, Whitfield R. Dinoprostone compared with
misoprostol for cervical ripening for induction of labor at term. J
Midwifery Womens Health 2009;54:405–11.
8 Sanchez-Ramos L, Kaunitz AM, Wears RL, Delke I, Gaudier FL.
Misoprostol for cervical ripening and labor induction: a meta-
analysis. Obstet Gynecol 1997;89:633–42.
353
 Chen et al.
9 Allen R, O’Brien BM. Uses of misoprostol in obstetrics and
gynecology. Rev Obstet Gynecol 2009;2:159–68.
10 Muzonzini G, Hofmeyr GJ. Buccal or sublingual misoprostol for
cervical ripening and induction of labour. Cochrane Database Syst
Rev 2004;(4):Cd004221.
11 Schaff EA, DiCenzo R, Fielding SL. Comparison of misoprostol
plasma concentrations following buccal and sublingual
administration. Contraception 2005;71:22–5.
12 Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of
multiple treatments: combining direct and indirect evidence. BMJ
2005;331:897–900.
13 Lu G, Ades AE. Combination of direct and indirect evidence in
mixed treatment comparisons. Stat Med 2004;23:3105–24.
14 Salanti G, Higgins JP, Ades AE, Ioannidis JP. Evaluation of networks
of randomized trials. Stat Methods Med Res 2008;17:279–301.
15 Higgins JPT, Green S, editors. Cochrane Handbook for Systematic
Reviews of Interventions Version 5.1.0 [updated March 2011]. The
Cochrane Collaboration, 2011. [www.cochrane-handbook.org].
16 Manager R. Copenhagen: The Nordic Cochrane Centre, The
Cochrane Collaboration. 2012.
17 van Valkenhoef G, Lu G, de Brock B, Hillege H, Ades AE, Welton
NJ. Automating network meta-analysis. Res Synth Methods
2012;3:285–99.
18 Brooks SP, Gelman A. General methods for monitoring convergence
of iterative simulations. J Comput Graph Stat 1998;7:434–55.
354
19 Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in
mixed treatment comparison meta-analysis. Stat Med 2010;29:932–
44.
20 Alfirevic Z, Aflaifel N, Weeks A. Oral misoprostol for induction of
labour. Cochrane Database Syst Rev 2014;(6):Cd001338.
21 Hofmeyr JG, Gulmezoglu MA, Pileggi C. Vaginal misoprostol for
cervical ripening and induction of labour. Cochrane Database Syst
Rev 2010;3:CD000941.
22 Ioannidis JP. Integration of evidence from multiple meta-analyses: a
primer on umbrella reviews, treatment networks and multiple
treatments meta-analyses. CMAJ 2009;181:488–93.
23 Jozwiak M, ten Eikelder M, Oude Rengerink K, de Groot C, Feitsma H,
Spaanderman M, et al. Foley catheter versus vaginal misoprostol:
randomized controlled trial (PROBAAT-M study) and systematic review
and meta-analysis of literature. Am J Perinatol 2014;31:145–56.
24 Jozwiak M, Bloemenkamp WMK, Kelly AJ, Mol WB, Irion O,
Boulvain M. Mechanical methods for induction of labour. Cochrane
Database Syst Rev 2012;3:CD001233
25 Gelber S, Sciscione A. Mechanical methods of cervical ripening and
labor induction. Clin Obstet Gynecol 2006;49:642–57.
26 Pennell CE, Henderson JJ, O’Neill MJ, McChlery S, Doherty DA,
Dickinson JE. Induction of labour in nulliparous women with an
unfavourable cervix: a randomised controlled trial comparing double
and single balloon catheters and PGE2 gel. BJOG 2009;116:
1443–52.
ª 2015 Royal College of Obstetricians and Gynaecologists