DOI: 10.1111/1471-0528.

14513 Systematic review

www.bjog.org

Preterm birth prevention in twin pregnancies

with progesterone, pessary, or cerclage: a
systematic review and meta-analysis
A Jarde,a O Lutsiv,a CK Park,b J Barrett,c J Beyene,b S Saito,d JM Dodd,e PS Shah,f JL Cook,g
AB Biringer,h L Giglia,i Z Han,j K Staub,k W Mundle,l C Vera,m L Sabatino,n SK Liyanage,a
SD McDonalda
a
Department of Obstetrics and Gynecology, McMaster University, Hamilton, ON, Canada b Department of Clinical Epidemiology and
Biostatistics, McMaster University, Hamilton, ON, Canada c Sunnybrook Health Sciences Centre, Toronto, ON, Canada d Department of
Obstetrics and Gynecology, University of Toyama, Toyama, Japan e Department of Obstetrics and Gynecology, University of Adelaide,
Adelaide, Australia f Department of Paediatrics, University of Toronto, Toronto, ON, Canada g The Society of Obstetricians and
Gynaecologists of Canada, Ottawa, ON, Canada h Department of Family and Community Medicine, University of Toronto, Toronto, ON,
Canada i Department of Pediatrics, McMaster University, Hamilton, ON, Canada j The First Affiliated Hospital of Xian Jiaotong University,
Xian, Shaanxi Province, China k Canadian Premature Babies Foundation, Sherwood Park, AB, Canada l Maternal Fetal Medicine Clinic,
Windsor Regional Hospital, Windsor, ON, Canada m Division of Obstetrics and Gynecology, Escuela de Medicina, Pontificia Universidad
Cat
olica de Chile, Santiago, Chile n Midwifery Education Program, McMaster University, Hamilton, ON, Canada
Correspondence: Dr A Jarde, Department of Obstetrics and Gynecology, McMaster University, 1280 Main Street West, Hamilton, Ontario,
L8S 4K1, Canada. Email jarde@mcmaster.ca

Accepted 28 November 2016. Published Online 8 February 2017.

Background About half of twin pregnancies deliver preterm, and
it is unclear whether any intervention reduces this risk.
Objectives To assess the evidence for the effectiveness of
progesterone, cerclage, and pessary in twin pregnancies.
Search strategy We searched Medline, EMBASE, CINAHL,
Cochrane Central Register of Controlled Trials, and ISI Web of
Science, without language restrictions, up to 25 January 2016.
Selection criteria Randomised controlled trials of progesterone,
cerclage, or pessary for preventing preterm birth in women with
twin pregnancies, without symptoms of threatened preterm
labour.
Data collection and analysis Two independent reviewers extracted
data using a piloted form. Study quality was appraised with the
Cochrane Risk of Bias tool. We performed pairwise inverse
variance random-effects meta-analyses.
Main results We included 23 trials (all but three were considered
to have a low risk of bias) comprising 6626 women with twin
pregnancies. None of the interventions significantly reduced the
risk of preterm birth overall at <34 or <37 weeks of gestation, or
neonatal death, our primary outcomes, compared to a control
group. In women receiving vaginal progesterone, the relative risk
(RR) of preterm birth <34 weeks of gestation was 0.82 (95% CI
0.641.05, seven studies, I2 36%), with a significant reduction in
some key secondary outcomes, including very low birthweight
(<1500 g, RR 0.71, 95% CI 0.520.98, four studies, I2 46%) and
mechanical ventilation (RR 0.61, 95% CI 0.450.82, four studies,
I2 22%).
Conclusion In twin gestations, although no overarching
intervention was beneficial for the prevention of preterm birth
and its sequelae, vaginal progesterone improved some important
secondary outcomes.
Keywords Cerclage, pessary, preterm birth, progesterone,
randomised controlled trials, twin.
Tweetable abstract Vaginal progesterone may be beneficial in
twin pregnancies, but not 17-OHPC, cerclage, or pessary.

Please cite this paper as: Jarde A, Lutsiv O, Park CK, Barrett J, Beyene J, Saito S, Dodd JM, Shah PS, Cook JL, Biringer AB, Giglia L, Han Z, Staub K,
Mundle W, Vera C, Sabatino L, Liyanage SK, McDonald SD. Preterm birth prevention in twin pregnancies with progesterone, pessary, or cerclage: a
systematic review and meta-analysis. BJOG 2017; DOI: 10.1111/1471-0528.14513.

2017 Royal College of Obstetricians and Gynaecologists 1

 Jarde et al.
Introduction
Preterm birth, i.e. birth before 37 completed weeks of ges-
tation, is the leading cause of mortality and a major cause
of morbidity in children.1,2 Although twin pregnancies rep-
resent 13% of all pregnancies in Europe and the USA,
they account for 1720% of all preterm births,3,4 with
about half of twin pregnancies delivering before 37 weeks
of gestation.3,4
Unlike in singleton gestations, where meta-analyses of
individual interventions, i.e. progesterone, pessary, or cer-
clage, each showed some benefit,57 previously published
meta-analyses of progesterone or cerclage have not found a
reduction in rates of preterm birth for twins.5,811 The
search strategies for progesterone were executed in 2013 or
earlier, but recent trials have been published. For cerclage,
the search strategies were executed in 2014 or earlier. Addi-
tionally, to our knowledge, no meta-analysis of pessary in
twins has been performed, despite the publication of ran-
domised trials, and the fact that it is reasonable to hypoth-
esise that such a mechanical intervention could provide
benefit in twin gestations, given that the higher risk of pre-
term birth in twins may partly arise from the stretching of
the uterus and pressure on the cervix.12
Our objective was to complete an overarching synthesis
of the current evidence in twin gestations of the effective-
ness of the main methods of preventing preterm birth, pro-
gesterone, cerclage, or pessary, and compare these
interventions using pairwise and network meta-analysis,
data permitting.
Methods
We registered our protocol in the PROSPERO database
(CRD42015016166).
Information sources and search strategy
We executed our search strategy in five electronic databases
from their inceptions until 25 January 2016 (Medline,
EMBASE, CINAHL, Cochrane CENTRAL, and ISI Web of
Science), without any language restriction (for the complete
search strategy, see Appendix S1). We screened reference
lists of previous systematic reviews and all included studies.
In addition, we contacted experts in the area to inquire for
additional studies that we might have missed.
Eligibility criteria
We included studies in which women with a twin preg-
nancy were randomised to an intervention for the preven-
tion of preterm birth (any type of progesterone, cervical
cerclage, cervical pessary, or any combination of these) or
to a control group (e.g. placebo or treatment as usual) or
another of the mentioned interventions (as we had planned
2 2017 Royal College of Obstetricians and Gynaecologists
to do a network meta-analysis if enough data were to be
identified). We excluded any other study designs including
cluster-randomised trials, non-peer reviewed literature,
studies published only as abstracts, or studies assessing the
prevention of preterm birth in women with contractions,
i.e. tocolytics. We contacted authors of studies where
results of singleton, twins, and/or higher order pregnancies
were not stratified to ask for the stratified data of the origi-
nally published outcomes.
Our primary outcomes were preterm birth at <34 and
<37 weeks of gestation, separating them into spontaneous
preterm birth or induced preterm birth, where possible. In
addition, after publishing the protocol, but before analysing
the data, we decided to add neonatal death to our primary
outcomes.
Our infant secondary outcomes were: mortality (i.e. peri-
natal death, miscarriage, and stillbirth), preterm birth (<24,
<28, <30, and <32 weeks of gestation), gestational age at
birth, low birthweight (<2500 g), small for gestational age
(below tenth, fifth, and third percentiles), birthweight,
admission and length of stay in the neonatal intensive or
special care unit (NICU), morbidity related to prematurity
(i.e. respiratory distress syndrome, bronchopulmonary dys-
plasia, intraventricular haemorrhage, periventricular leuko-
malacia, necrotising enterocolitis, sepsis), congenital
anomalies, masculinisation of female fetuses, umbilical cord
pH < 7.1, and low Apgar score at 5 minutes. We decided to
record very low birthweights of <1500 g and any other mea-
sure of preterm birth shortly after we started collecting data.
Our maternal secondary outcomes were: mortality, pre-
term premature rupture of membranes (PPROM), inter-
vention side effects (e.g. infection, vaginal discharge,
bleeding, fever, repositioning or removal of pessary), length
of inpatient antepartum stay, number of outpatient visits,
and caesarean section.
Data extraction and assessment of risk of bias
Two independent reviewers (AJ and OL) screened titles,
abstracts, and the full text of potentially relevant papers,
with a third assessor (SM) available when disagreements
could not be resolved by discussion. If necessary, we con-
tacted the authors of the original studies to confirm the
inclusion criteria and to provide clarifications on the pub-
lished data.
Two independent reviewers (AJ and either OL or CP)
extracted data on study characteristics, potential effect
modifiers, and outcomes using a piloted data collection
form. For binary outcomes we extracted raw data (e.g.
2 9 2 tables) or effect sizes (e.g. odds ratios, ORs) and
confidence intervals. For continuous outcomes, we
extracted raw data for each group (mean, standard devia-
tion, and group size) or effect sizes (e.g. mean difference)
and confidence intervals.

The same reviewers assessed risk of bias using the
Cochrane Risk of Bias tool.13 This tool assesses seven
domains: random sequence generation, allocation conceal-
ment, blinding of participants and personnel, blinding of
outcome assessment, incomplete outcome data, selective
reporting, and other biases. Each domain was assessed as
having a low, unclear, or high risk of bias. Although the
Cochrane Risk of Bias tool does not provide an overall
risk of bias assessment, we applied the following algo-
rithm (as used in previous reviews): the overall risk of
bias was considered low if four or more domains were
rated as low risk (not counting other biases), with at
least one of them being sequence generation or allocation
concealment.14
Data synthesis and statistical analyses
We had planned to carry out, data permitting, Bayesian
random-effects network meta-analyses. After considering
the results of the standard pairwise inverse variance
random-effects meta-analyses of each intervention versus
a control group, however, we considered that a
network meta-analysis would be uninformative and
decided to perform separate pairwise random-effects
meta-analyses only (DerSimonian and Laird),15 for
each intervention compared with a control group, using
REVIEW MANAGER 5.3. Relative risks (RRs) and mean
differences (MDs) were calculated, with their correspond-
ing 95% confidence intervals (95% CIs). We did not
test for heterogeneity (i.e. v2 or Q tests), but quantified
this using the I2 statistic.16 Publication bias was
assessed using funnel plots and Duval and Tweedies
trim-and-fill method (when the funnel plot showed
asymmetry).
Unit of analysis
Given that several of our outcomes are assessed at the level
of the infant, and not the pregnancy (such as respiratory
distress syndrome), these outcomes should ideally be anal-
ysed by taking into account the non-independence, that is
the clustering of the data, in twins.17 Given that most of
the primary studies did not do this, however, we were also
unable to do so.
Subgroup and sensitivity analyses
We planned the following subgroup analyses for the pri-
mary outcomes (data permitting): route of administration
of progesterone and type of cerclage. We additionally
planned subgroup analyses by cervical length 25 mm,
previous preterm birth, parity, cervical insufficiency, and
infant sex. A sensitivity analysis examining studies with a
low risk of bias was planned. Post hoc we decided not
to limit subgroup analyses to primary outcomes alone.
2017 Royal College of Obstetricians and Gynaecologists
Preterm birth prevention in twins
Results
Our search strategy yielded 12 280 results, from which we
deleted 4371 duplicates and excluded 7715 references based
on their titles and abstracts. Screening references yielded 99
additional entries and an additional study was located after
contacting experts in the field. After assessing the full text
of the remaining 294 references, 23 studies met our inclu-
sion criteria, with 6626 women (Figure S1).
Study characteristics
Sixteen studies compared progesterone with placebo or no
intervention/treatment as usual,1833 four assessed cer-
clage,3437 and three assessed pessary (Table 1).3840 No
study compared different interventions directly with each
other or allocated women randomly to a combination of
two interventions (e.g. progesterone and cerclage) or more.
In total, we identified five studies that had not been
included in a meta-analysis before: two assessing proges-
terone and three assessing pessary.32,33,3840
All studies of intramuscular progesterone used 17-alpha-
hydroxyprogesterone caproate (17-OHPC). Four studies
included women with a short cervix (variously defined).
According to the World Bank classification,41 20 studies
were conducted in high- or upper middle-income coun-
tries, and two were conducted in lower middle-income
countries. One multi-centre, multinational study recruited
28 women from a low-income country; however, it is
unclear how many of these bore a twin pregnancy. There-
fore, less than 0.4% of the women in the studies assessing
progesterone and none of the women receiving other inter-
ventions were from low-income countries. All but three
studies were considered to have a low risk of bias (Fig-
ure S2; details provided as Appendix S2). Across the
domains of the risk of bias tool, the mean proportion of
agreement between raters was 0.85 between AJ and OL,
and 0.68 between AJ and CP (proportions are presented as
there are known problems of the Kappa statistic leading to
low Kappa values despite high agreement).42
Primary outcomes
In separate pairwise meta-analyses, compared with control,
none of the interventions significantly reduced the risk of
any of our primary outcomes: preterm birth at <34 or
<37 weeks of gestation, and overall neonatal death
(Table 2). Similar results were found for spontaneous pre-
term birth (Table 3). Given these negative results we
deemed that network meta-analyses would be uninforma-
tive, and therefore did not perform them. Following Sterne
et al.s recommendations,43 publication bias was assessed in
meta-analyses with more than ten studies. No publication
bias was suspected (Figure S3).
3
 Jarde et al.

Table 1. Main characteristics of studies included in systematic review and meta-analyses of prevention of preterm birth in twin pregnancies with
progesterone, pessary, and cerclage.

First author, year, country Intervention group: Control group: sample size Gestational age at randomisation*
sample size

Progesterone versus control

Hartikainen-Sorri, 1980,18 17-OHCP (IM): 39 Placebo: 38 29.2  1.9 (onset of medication)
Finland
Fonseca, 2007,19 Progesterone (PV): 11 Placebo: 13 23.6 (22.724.0)
Multinational** (All with cervical (All with cervical
length 15 mm) length 15 mm)
Rouse, 2007,20 USA***,**** 17-OHCP (IM): 325 Placebo: 330 19.2  1.5
Briery, 2009,21 USA 17-OHPC (IM): 16 Placebo: 14 24.7  3.3
Norman, 2009,22 UK*** Progesterone (PV): 247 Placebo: 247 Intervention started at
24 weeks of gestation
Cetingoz, 2011,23 Turkey Progesterone (PV): 39 Placebo: 28 Randomisation started at
24 weeks of gestation
Combs, 2011,24 USA*** 17-OHCP (IM): 160 Placebo: 78 20  2.4
Lim, 2011,25 Netherlands*** 17-OHCP (IM): 327 Placebo: 326 16.7  1.5
Cervical length 25 mm: 9 Cervical length 25 mm: 3
Rode, 2011,26 Progesterone (PV): 334 Placebo: 341 20.9 (19.922.4)
Multinational*****,**** Previous preterm birth or Previous preterm birth or
miscarriage >12 weeks: 10 miscarriage >12 weeks: 18
Aboulghar, 2012,27 Egypt**** Progesterone (PV): 49 Placebo: 42 20.1  NR
Wood, 2012,28 Canada***,**** Progesterone (PV): 40 Placebo: 41 19.3 (range 16.421.1)
Senat, 2013,29 France*** 17-OHCP (IM): 82 (All with NI/TAU: 83 (All with 27.9 (25.329.9)
cervical length 25 mm) cervical length 25 mm)
Serra, 2013,30 Spain***,**** Progesterone (PV): 194 Placebo: 96 Randomisation and onset of
intervention at 20 weeks of gestation
Awwad, 2014,31 Lebanon 17-OHPC (IM): 194 Placebo: 94 19  2.1
Brizot, 2016,32 Brazil**** Progesterone (PV): 189 Placebo: 191 19.07  1.06
Cervical length 25 mm: 22 Cervical length 25 mm: 13
El-Refaie, 2016,33 Egypt*** Progesterone (PV): 116 NI/TAU: 108 (All with Randomised at 2024 weeks of gestation
(All with cervical length cervical length = 2025 mm)
= 2025 mm)
Cerclage versus control
Dor, 1982,34 Israel McDonald: 22 NI/TAU: 23 Intervention performed at
13th week of gestation
MacNaughton, 1993, 35
McDonald or Shirodkar: 12 NI/TAU: 16 14.6  5.1 (at study entry)
Multinational******
Rust, 2001,36 USA**** McDonald: 7 (All with NI/TAU: 4 (All with 21.77  1.88
cervical length 25 mm) cervical length 25 mm)
Berghella, 2004,37 McDonald: 3 NI (Bed rest): 1 19.6  2.4
USA***,**** Cervical length 25 mm: 2 Cervical length 25 mm: 1
Pessary versus Control
Liem, 2013,38 Arabin: 394 NI/TAU: 401 16.9  2.0
Netherlands***,****
Goya, 2015,39 Spain***,**** Arabin: 68 NI/TAU: 66 22.1  0.8
(All with cervical (All with cervical
length 25 mm) length 25 mm)
Nicolaides, 2016,40 Arabin: 588 NI/TAU: 589 22.6 (21.423.9)
Multinational******* Cervical length <25 mm: 106 Cervical length <25 mm: 108

17-OHPC, 17a-hydroxyprogesterone caproate; IM, intramuscular; NI/TAU, no intervention/treatment as usual; NR, not reported; PO, per oral; PV,
per vagina.
*Mean weeks  standard deviation or median (interquartile range) of group receiving the intervention.
**Centres in the UK, Chile, Brazil, and Greece.
***Multi-centre study.
****Stratified data for published results and/or clarifications on outcome definitions provided by authors through personal communications.
*****Centres in Denmark and Austria. Stratified data for previous preterm birth or miscarriage >12 weeks of gestation reported in a separate
article.
******Centres in Belgium, Canada, France, Hungary, Iceland, Ireland, Italy, Netherlands, Norway, South Africa, UK, and Zimbabwe.
*******Centres in Albania, Austria, Belgium, Brazil, Chile, China, Germany, Italy, Portugal, Slovenia, Spain, and UK.

4 2017 Royal College of Obstetricians and Gynaecologists

 Preterm birth prevention in twins

Table 2. Primary outcomes of systematic review and meta-analyses of prevention of preterm birth in twin pregnancies with progesterone,
pessary, and cerclage, compared with placebo or treatment as usual

Outcome Intervention Studies n s2 I2 RR (95% CI) P*

Preterm birth <34 weeks Progesterone 10 2180 0.09 55% 0.91 (0.701.18)
of gestation Vaginal 7 1751 0.04 36% 0.82 (0.641.05)
0.26
Intramuscular 3 429 0.16 60% 1.18 (0.662.12)
Cerclage 3 43 0.53 39% 1.21 (0.344.31)
Pessary 2 1311 0.36 87% 0.71 (0.291.71)
Preterm birth <37 weeks Progesterone 13 3686 0.00 30% 1.01 (0.951.08)
of gestation Vaginal 6 1584 0.01 41% 0.94 (0.831.07)
0.15
Intramuscular 7 2102 0.00 8% 1.04 (0.981.12)
Cerclage 3 78 0.00 0% 1.11 (0.751.65)
Pessary 2 929 0.00 0% 0.96 (0.861.07)
Neonatal death Progesterone 12 8006 0.12 21% 1.16 (0.761.79)
Vaginal 5 3847 0.00 0% 1.38 (0.782.45)
0.53
Intramuscular 7 4159 0.34 43% 1.03 (0.512.11)
Cerclage 1 8 5.57 (0.4470.55)
Pessary 3 4210 0.00 2% 0.89 (0.571.38)

*Test for subgroup differences.

Secondary outcomes significantly increased the risk of NICU admission

Similar to the primary outcomes, none of the interventions
studied significantly reduced the risk of most of our sec-
ondary outcomes (Tables 3 and 4). With progesterone
there was less risk of requiring mechanical ventilation
(RR 0.68, 95% CI 0.530.88, six studies, 3630 women,
I2 44%), and one study found a significant risk reduction
of early neonatal death (RR 0.49, 95% CI 0.330.73, one
study, 439 women), but infants receiving progesterone also
had slightly lower Apgar scores in another study (MD
0.60, 95% CI 1.14 to 0.06, one study, 60 women).
Pessary use was associated with an increased risk of cae-
sarean section (RR 1.16, 95% CI 1.011.34, two studies,
929 women, I2 0%) and, in one study, retinopathy of pre-
maturity (RR 4.00, 95% CI 1.1314.12, one study, 2293
women; Tables 3 and 4).
Subgroup analyses
In the subgroup of women receiving vaginal progesterone,
the relative risk of preterm birth <34 weeks of gestation
was 0.82 (95% CI 0.641.05, seven studies, 1751 women,
I2 36%), with significant decreases in some key secondary
outcomes that are sequelae of preterm birth, including very
low birthweights of <1500 g (RR 0.71, 95% CI 0.520.98,
four studies, 3079 women, I2 46%), and mechanical venti-
lation (RR 0.61, 95% CI 0.450.82, four studies, 1792
women, I2 22%). Although overall neonatal death was not
found to be affected, one study reported a reduction in early
neonatal death (RR 0.49, 95% CI 0.330.73, one study, 439
women). The risk of caesarean section was reduced slightly
(RR 0.94, 95% CI 0.900.99, six studies, 2140 women,
I2 0%). In contrast, intramuscular progesterone (17-OHPC)
(RR 1.35, 95% CI 1.131.60, two studies, 1619 women,
I2 0%), and was associated with slightly lower Apgar scores
in one study (MD 0.60, 95% CI 1.14 to 0.06, one
study, 60 women). There was a significant reduction in the
risk of retinopathy of prematurity, however (RR 0.30,
95% CI 0.100.90, three studies, 1320 women, I2 6%;
Table S1).
In the subgroup of women with a cervical length of
25 mm or less, including one study of cervical length
<15 mm, nine studies (five progesterone, two cerclage, and
two pessary) reported data, but none of our primary out-
comes were significantly affected (Table S2). Among the
secondary outcomes reported for this subpopulation, gesta-
tional age was increased by progesterone (per vagina in all
studies, MD 1.00 weeks, 95% CI 0.361.64 weeks, two
studies, 259 women, I2 0%) and pessary (MD 2.20 weeks,
95% CI 1.033.37 weeks, one study, 134 women). Proges-
terone (per vagina in all studies) also reduced the risk of
very low birthweights of <1500 g (RR 0.45, 95% CI 0.32
0.63, two studies, 500 women, I2 0%), mechanical ventila-
tion (RR 0.47, 95% CI 0.320.69, one study, 439 women),
and early neonatal death (RR 0.49, 95% CI 0.330.73, one
study, 439 women).
A single study in women with previous preterm birth or
miscarriage found no significant effect of progesterone on
preterm birth at <37 or <32 weeks of gestation, or on other
secondary outcomes.26
We were unable to perform subgroup analyses according
to type of cerclage because McDonald cerclage was used in
all cases except for a minority of Shirodkar procedures
reported in MacNaughton et al. (1993).35

2017 Royal College of Obstetricians and Gynaecologists 5

 Jarde et al.

Table 3. Pregnancy level secondary outcomes of prevention of preterm birth in twin pregnancies with progesterone, pessary, and cerclage

Outcome Intervention Studies n s2 I2 RR/MD (95% CI)

Spontaneous preterm birth <34 weeks of gestation Progesterone 3 1293 0.01 12% 0.96 (0.711.29)
Cerclage 0
Pessary 2 1311 0.38 86% 0.69 (0.271.72)
Spontaneous preterm birth <37 weeks of gestation Progesterone 4 609 0.00 0% 1.11 (0.861.43)
Cerclage 0
Pessary 1 134 0.95 (0.771.18)
Preterm premature rupture of membranes Progesterone 5 1321 0.00 0% 1.16 (0.821.65)
Cerclage 1 45 0.70 (0.133.78)
Pessary 2 929 1.04 64% 0.54 (0.102.87)
Gestational age (weeks) Progesterone 12 4080 0.09 45% MD 0.07 (0.34 to 0.20)
Cerclage 1 11 MD 0.86 (8.56 to 6.84)
Pessary 2 929 1.60 88% MD 1.17 (0.68 to 3.03)
Preterm birth <35 weeks of gestation Progesterone 3 766 0.08 57% 0.86 (0.561.30)
Cerclage 1 4 1.00 (0.412.42)
Pessary 0
Preterm birth <32 weeks of gestation Progesterone 9 3564 0.24 73% 1.06 (0.721.56)
Cerclage 1 4 3.50 (0.3139.71)
Pessary 2 1972 0.00 0% 0.91 (0.691.19)
Preterm birth <28 weeks of gestation Progesterone 8 3493 0.00 0% 1.08 (0.771.51)
Cerclage 3 54 0.00 0% 1.37 (0.375.12)
Pessary 3 2106 0.07 30% 0.84 (0.491.44)
Maternal mortality Progesterone 0* 0 Not estimable
Cerclage 0
Pessary 1 795 3.05 (0.1274.72)
Caesarean section Progesterone 11 4132 0.00 0% 0.97 (0.931.01)
Cerclage 1 45 1.34 (0.612.98)
Pessary 2 929 0.00 0% 1.16 (1.011.34)

Significant results are set in bold. Cells with no results are empty.
*One study reported this outcome, but the effect size was not estimable (Norman 200922: 0/247 versus 0/247 cases in the progesterone and
placebo group, respectively).

With the lack of stratified data, we could not perform
the other subgroup analyses in the protocol regarding par-
ity, cervical insufficiency, and infant sex.
Sensitivity analyses
Examining studies with a low risk of bias yielded very simi-
lar results for preterm birth at <34 and <37 weeks of gesta-
tion, and neonatal death. Similar results were also found
for secondary outcomes (Table S3).
Individual study data for primary and secondary out-
comes, as well as subgroup and sensitivity analyses, are
reported in Appendix S3.
Discussion
Main findings
In the first over-arching synthesis to our knowledge in twin
pregnancies, none of the overall interventions, i.e. proges-
terone, cerclage, or pessary, prevented preterm birth,
neonatal death, or most secondary outcomes; however, in
the subgroup of women receiving vaginal progesterone
there was a trend towards a reduction in preterm birth at
<34 weeks of gestation, and there were significant decreases
in some secondary outcomes that are sequelae of preterm
birth, including very low birthweights of <1500 g. In con-
trast, intramuscular progesterone (17-OHPC) increased the
risk of some adverse outcomes. In women with a short cer-
vix of 25 mm or less, vaginal progesterone appeared to sig-
nificantly decrease some secondary outcomes. Pessary
improved gestational age at birth by 2 weeks in this popu-
lation.
Strengths and limitations
This is a comprehensive synthesis of the literature on the
prevention of preterm birth in women with a twin preg-
nancy, a group with approximately a 50% chance of pre-
term birth.3,4 It includes the first systematic review and
meta-analysis of the efficacy of pessary in twins, and an
updated synthesis of progesterone and cerclage, which
extended previous findings. Although we did not perform

6 2017 Royal College of Obstetricians and Gynaecologists

 Preterm birth prevention in twins

Table 4. Infant secondary outcomes of systematic review and meta-analyses of prevention of preterm birth in twin pregnancies with
progesterone, pessary, and cerclage

Outcome Intervention Studies n s2 I2 RR/MD (95% CI)

Congenital anomalies Progesterone 5* 5070 0.04 21% 0.86 (0.581.28)

Cerclage 0
Pessary 1 1590 0.70 (0.411.19)
Miscarriage Progesterone 1 162 5.12 (0.25105.08)
Cerclage 0
Pessary 1 1590 0.44 (0.111.68)
Stillbirth Progesterone 8 4094 0.00 0% 0.78 (0.471.27)
Cerclage 0
Pessary 1 1590 0.70 (0.301.64)
Early neonatal death Progesterone 1 439 0.49 (0.330.73)
Cerclage 1 90 1.19 (0.473.02)
Pessary 0
Perinatal death Progesterone 4 1966 0.36 56% 0.82 (0.351.90)
Cerclage 1 22 0.57 (0.152.19)
Pessary 1 2354 0.91 (0.551.49)
Birthweight (grams) Progesterone 8 5227 6367.86 74% MD 4.47 (63.52 to 72.46)
Cerclage 0
Pessary 0
Low birthweight <2500 g Progesterone 8 6706 0.01 66% 0.98 (0.911.05)
Cerclage 0
Pessary 3 4210 0.00 53% 0.95 (0.871.05)
Very low birthweight <1500 g Progesterone 8 6706 0.18 79% 0.92 (0.651.29)
Cerclage 0
Pessary 3 4210 0.00 0% 0.97 (0.801.17)
Small for gestational age Progesterone 2 1205 0.00 0% 1.12 (0.861.44)
(<tenth percentile) Cerclage 0
Pessary 0
Apgar score <7 Progesterone 6 5672 0.00 0% 0.87 (0.671.11)
Cerclage 0
Pessary 1 1590 0.84 (0.571.22)
Apgar score (continuous) Progesterone 1 60 MD 0.60 (1.14 to 0.06)
Cerclage 0
Pessary 0
Umbilical cord pH <7.2 Progesterone 1 321 1.36 (0.593.13)
Cerclage 0
Pessary 0
Neonatal intensive care unit admission Progesterone 7 5671 0.02 72% 1.03 (0.901.17)
Cerclage 2 18 2.84 69% 0.45 (0.037.21)
Pessary 1 1590 0.82 (0.641.06)
Length of stay in neonatal intensive Progesterone 1 325 MD 3.30 (3.55 to 10.15)
care unit (days) Cerclage 0
Pessary 1 206 MD 2.81 (1.99 to 7.62)
Respiratory distress syndrome Progesterone 11 6701 0.10 74% 1.02 (0.811.29)
Cerclage 0
Pessary 2 2559 0.00 0% 1.08 (0.841.39)
Mechanical ventilation Progesterone 6 3630 0.04 44% 0.68 (0.530.88)
Cerclage 0
Pessary 1 2293 1.17 (0.911.52)
Bronchopulmonary dysplasia Progesterone 5 3623 0.22 45% 0.98 (0.521.83)
Cerclage 0
Pessary 1 1590 0.34 (0.071.68)
Intraventricular haemorrhage (IIV) Progesterone 4 2307 0.00 0% 1.52 (0.743.10)
Cerclage 0

2017 Royal College of Obstetricians and Gynaecologists 7

 Jarde et al.

Table 4. (Continued)

Outcome Intervention Studies n s2 I2 RR/MD (95% CI)

Pessary 3 4149 0.09 21% 0.99 (0.492.00)

Intraventricular haemorrhage (IIIIV) Progesterone 4 2479 0.00 0% 0.86 (0.421.77)
Cerclage 0
Pessary 0
Periventricular leukomalacia Progesterone 3** 2049 0.00 0% 0.90 (0.322.52)
Cerclage 0
Pessary 0
Necrotising enterocolitis Progesterone 6** 4048 0.00 0% 0.75 (0.381.51)
Cerclage 0
Pessary 3 4149 0.00 0% 1.05 (0.512.16)
Retinopathy of prematurity Progesterone 5 3265 0.30 32% 0.62 (0.261.48)
Cerclage 0
Pessary 1*** 2293 4.00 (1.1314.12)
Sepsis Progesterone 7 5878 0.52 76% 1.16 (0.602.23)
Cerclage 0
Pessary 3 4149 0.00 0% 0.92 (0.691.22)

Significant results are set in bold. Cells with no results are empty.
*One additional study could not be pooled because the effect size was not estimable (Norman 200922: 0/494 versus 0/494 cases in the
progesterone and placebo group, respectively).
**One additional study could not be pooled because the effect size was not estimable (Awwad 201431: 0/382 versus 0/188 cases in the
progesterone and placebo group, respectively).
***One additional study could not be pooled because the effect size was not estimable (Goya 201639: 0/136 versus 0/130 cases in the pessary
and no intervention group, respectively).

network meta-analyses because of the lack of benefit of the
interventions in standard pairwise meta-analyses, this over-
arching synthesis of the three interventions will enable clin-
icians to understand their efficacy in relation to each other.
A limitation of our systematic review is the lack of
robust primary data on cerclage: only 44 women received
cerclage in four studies, compared with thousands of
women receiving progesterone (2362) and pessary (1050).
In addition, only one of the four studies assessing cerclage
focused only on twins, whereas the other three enrolled
both singletons and twins without stratifying the randomi-
sation process; hence, it is possible that some baseline
characteristics were not evenly distributed in some out-
comes, and a multivariate analysis with individual partici-
pant data (if available) would be more appropriate.
Although preterm birth is a worldwide problem, few pri-
mary data are available from low-income countries,
although we do not anticipate markedly different results
there. As the primary studies generally did not report data
adjusted for the correlated nature of twins, we could not
take this into account in our meta-analyses. This is likely
to have led to an underestimation of the standard errors,
yielding confidence intervals that are artificially narrow for
infant outcomes; however, given that confidence intervals
were only marginally wider in a study assessing the effect
of ignoring the correlation between twins,17 we do not
think that our results would markedly change. In
addition, this would not apply to preterm birth or other
outcomes assessed at the level of the pregnancy (Tables 2
and 3). Given the reported data we could study the sub-
population of women with a short cervix only, but not
with previous preterm birth. Schuits individual partici-
pant data meta-analyses on progesterone did not find
markedly different results in this subgroup, however (nei-
ther did Romero et al. in their subgroup analyses,
although they included only women with a short cervix in
their study).8,9 Finally, many of our meta-analyses
included a small number of studies, a scenario that
remains challenging for random-effects meta-analysis (both
using the DerSimonianLaird and the KnappHartung
methods), and for which no clear guidance exists.44 We
therefore advise caution in interpreting these results.
Interpretation
To our knowledge, this is the first systematic review that
presents the meta-analyses of the effect of progesterone,
cerclage, and pessary in a way that allows for an easy com-
parison, which is much more challenging when results
of different reviews, with their own particularities in terms
of inclusion and exclusion criteria, and definitions of out-
comes and subgroups, have to be compared. This is also
the first systematic review to synthesise the effect of pessary
in twin gestations, which has been assessed in a pooled
sample of over 2000 women. As in previous reviews,

8 2017 Royal College of Obstetricians and Gynaecologists


vaginal progesterone was associated with a significant
reduction in several infant secondary outcomes; however,
unlike results previously reported by Schuit et al. and
Romero et al.,8,9 this was not limited to women with a
short cervix. Additionally, we identified a trend towards a
reduction in preterm birth <34 weeks of gestation in
women receiving vaginal progesterone, not found in previ-
ous reviews. This could be explained by two recent stud-
ies,32,33 that were not included in the previous systematic
reviews of progesterone in twin pregnancies,5,8,9 in which
the literature searches ended in 2011 (Romero et al.9) and
in 2013 (Dodd et al.5 and Schuit et al.8).
We additionally compared our findings with the two
previous systematic reviews of cerclage in twin pregnancies
conducted by Rafael et al.10 and Saccone et al.11 Their liter-
ature searches ended in 2014, although there have not been
additional trials published since then, which means that no
study on the effect of cerclage in twin pregnancies has been
published since 2004. We did not find any positive or neg-
ative effects of cerclage, but with the inclusion of unpub-
lished (non peer-reviewed) data in their analysis the
previous reviews reported that neonates of women with a
short cervix were at an increased risk of very low birth-
weight and respiratory distress syndrome,10,11 as well as at
an increased risk of a composite outcome of perinatal
death and neonatal morbidity.10
Conclusion
In twin gestations, although no overarching intervention
was beneficial for the prevention of preterm birth and its
sequelae, vaginal progesterone improved some important
secondary outcomes in women overall and in women with
short cervices. There were very limited data on cerclage.
Further research is required to reduce the risks of preterm
birth and its sequelae in twin pregnancies, including the
use of combinations of therapies.
Disclosure of interests
KS reports, as executive director (until 31st of December
2015) of the Canadian Premature Babies Foundation, that
corporate sponsorship for the foundation comes from
AbbVie Canada, Prolacta Biosciences Canada, and Ikaria
Canada (2013), but not for her personally. All other
authors declare that they have no competing interests.
Contribution to authorship
SDM conceived the original study. AJ, OL, JB (Joseph),
and SDM planned and designed the study. All authors [AJ,
OL, CKP, JB (Barrett), JB (Beyene), SS, JMD, PSS, JLC,
ABB, LG, ZH, KS, WM, CV, LS, SKL, and SDM] con-
tributed to the design of the study. AJ and OL executed
the search strategy and study selection. AJ, OL, and CKP
2017 Royal College of Obstetricians and Gynaecologists
Preterm birth prevention in twins
collected the data and assessed the risk of bias and the
quality of the evidence. AJ conducted the statistical analy-
ses, supervised by JB (Beyene). AJ and SDM drafted the
manuscript. All authors [AJ, OL, CKP, JB (Barrett), JB
(Beyene), SS, JMD, PSS, JLC, ABB, LG, ZH, KS, WM, CV,
LS, SKL, and SDM] contributed to the interpretation of
the results and commented on the manuscript. JB (Barrett),
JB (Beyene), SS, JMD, PSS, JLC, ABB, LG, ZH, KS, WM,
CV, LS, and SDM contributed to obtaining funding for the
project. AJ is the guarantor for the study.
Details of ethics approval
None required, as this was secondary research.
Funding
This study was funded by a Canadian Institutes of Health
Research (CIHR) Knowledge Synthesis Grant (RN281576-
362279). JB (Beyene) holds the John D. Cameron Endowed
Chair in the Genetic Determinants of Chronic Diseases,
Department of Clinical Epidemiology and Biostatistics,
McMaster University. SDM is supported by a Tier-II
Canada Research Chair, Sponsor Award #950-229920. PS
holds and Applied Research Chair in Reproductive and
Child Health Services Research from CIHR. SS is supported
by a Japan Society for the Promotion of Science KAKENHI
grant (#JP15H04980). None of the agencies had any influ-
ence on the design, analysis, interpretation, conclusions, or
decision to publish this manuscript.
Acknowledgements
We thank Dr Ahmet Metin G ulmezoglu (World Health
Organization), Ms Tonia Occhionero (Executive Director
of the Canadian Association of Midwives), Dr Jean Cham-
berlain (Uganda Christian University and McMaster
University, Hamilton, Canada), and Dr Maite Lopez-Yarto
(Universitat Autonoma de Barcelona) for valuable contri-
butions to this study. We thank Ms Neera Bhatnagar, BSc,
MLIS, Head of Systems and Public Services, Coordinator of
Research & Graduate Education Support, Health Sciences
Library, McMaster University, for her assistance with the
creation of the literature searches. We thank Mr Jeffrey
Mah, Ms Yi Wang, Ms Julie Yu, and Ms Angela Ma,
research assistants funded by the CIHR Knowledge Synthe-
sis Grant for the study in the Department of Obstetrics and
Gynecology at McMaster University, for their administra-
tive support of the project.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Flow diagram of the study selection process.
Figure S2. Risk of bias of included studies.
9
 Jarde et al.
Figure S3. Funnel plots of meta-analyses of primary out-
comes with more than ten studies.
Table S1. Subgroup analyses by route of administration
and type of progesterone.
Table S2. Meta-analyses of prevention of preterm birth
in twin pregnancies with a short cervix (25 mm).
Table S3. Comparison of results obtained when pooling
all studies and pooling only studies with a low risk of bias.
Appendix S1. Complete search strategy.
Appendix S2. Risk of Bias assessment details.
Appendix S3. Individual study data. &
References
1 World Health Organization. Fact sheet N363 - Preterm birth
[Internet]: World Health Organization. 2014. [www.who.int/mediace
ntre/factsheets/fs363/en/.]. Accessed 25 September 2015
2 Behrman RE, Butler AS. Preterm Birth: Causes, Consequences,
and Prevention. Washington, DC: National Academies Press (US),
2007.
3 Hamilton BE, Martin JA, Osterman MJ, Curtin SC, Mathews T.
Births: final data for 2014. Natl Vital Stat Rep 2015;64:164.
4 Zeitlin J, Mohangoo A, Alexander S, Barros H, Blondel B, Bouvier-
Colle M-H, et al. European perinatal health report, 2008.
5 Dodd JM, Jones L, Flenady V, Cincotta R, Crowther CA. Prenatal
administration of progesterone for preventing preterm birth in
women considered to be at risk of preterm birth. Cochrane
Database Syst Rev. 2013;7:CD004947.
6 Alfirevic Z, Stampalija T, Roberts D, Jorgensen AL. Cervical stitch
(cerclage) for preventing preterm birth in singleton pregnancy.
Cochrane Database Syst Rev 2012;4:CD008991.
7 Abdel-Aleem H, Shaaban OM, Abdel-Aleem MA. Cervical pessary for
preventing preterm birth. Cochrane Database Syst Rev 2013;5:
CD007873.
8 Schuit E, Stock S, Rode L, Rouse DJ, Lim AC, Norman JE, et al.
Effectiveness of progestogens to improve perinatal outcome in twin
pregnancies: an individual participant data meta-analysis. BJOG
2015;122:2737.
9 Romero R, Nicolaides K, Conde-Agudelo A, Tabor A, OBrien JM,
Cetingoz E, et al. Vaginal progesterone in women with an
asymptomatic sonographic short cervix in the midtrimester
decreases preterm delivery and neonatal morbidity: a systematic
review and metaanalysis of individual patient data. Am J Obstet
Gynecol 2012;206:124 .e1-19. eng.
10 Rafael TJ, Berghella V, Alfirevic Z. Cervical stitch (cerclage) for
preventing preterm birth in multiple pregnancy. Cochrane Database
Syst Rev. 2014;9:CD009166.
11 Saccone G, Rust O, Althuisius S, Roman A, Berghella V. Cerclage for
short cervix in twin pregnancies: systematic review and meta-
analysis of randomized trials using individual patient-level data. Acta
Obstet Gynecol Scand 2015;94:3528. eng.
12 Hodgson EJ, Lockwood CJ. Preterm birth: a complex disease. In:
Berghella, V, editor. Preterm Birth: Prevention and Management
Chichester: John Wiley & Sons Ltd; 2010. pp. 816.
13 Higgins J, Green S. Cochrane handbook for systematic reviews of
interventions 5.1.0 [updated March 2011]: The Cochrane
Collaboration, 2011. [www.cochrane-handbook.org].
14 Haas DM, Caldwell DM, Kirkpatrick P, McIntosh JJ, Welton NJ.
Tocolytic therapy for preterm delivery: systematic review and
network meta-analysis. BMJ 2012;345:116.
10
15 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Trials 1986;7:17788.
16 Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring
inconsistency in meta-analyses. BMJ 2003;327:55760. PubMed
PMID: 12958120. Pubmed Central PMCID: PMC192859.
17 Gates S, Brocklehurst P. How should randomised trials including
multiple pregnancies be analysed? BJOG 2004;111:21319.
18 Hartikainen-Sorri A-L, Kauppila A, Tuimala R. Inefficacy of 17
[alpha]-Hydroxyprogesterone Caproate in the Prevention of
Prematurity in Twin Pregnancy. Obstet Gynecol 1980;56:6925.
19 Fonseca EB, Celik E, Parra M, Singh M, Nicolaides KH. Progesterone
and the risk of preterm birth among women with a short cervix.
N Engl J Med 2007;357:4629.
20 Rouse DJ, Caritis SN, Peaceman AM, Sciscione A, Thom EA, Spong
CY, et al. A trial of 17 alpha-hydroxyprogesterone caproate to
prevent prematurity in twins. N Engl J Med 2007;357:45461.
21 Briery CM, Veillon EW, Klauser CK, Martin RW, Chauhan SP,
Magann EF, et al. Progesterone does not prevent preterm births in
women with twins. South Med J 2009;102:9004.
22 Norman JE, Mackenzie F, Owen P, Mactier H, Hanretty K, Cooper S,
et al. Progesterone for the prevention of preterm birth in twin
pregnancy (STOPPIT): a randomised, double-blind, placebo-controlled
study and meta-analysis. The Lancet 2009;373:203440.
23 Cetingoz E, Cam C, Sakall M, Karateke A, Celik C, Sancak A.
Progesterone effects on preterm birth in high-risk pregnancies: a
randomized placebo-controlled trial. Arch Gynecol Obstet
2011;283:4239.
24 Combs CA, Garite T, Maurel K, Das A, Porto M, Network OCR. 17-
Hydroxyprogesterone caproate for twin pregnancy: a double-blind,
randomized clinical trial. Am J Obstet Gynecol 2011;204:221. e1-e8.
25 Lim AC, Schuit E, Bloemenkamp K, Bernardus RE, Duvekot JJ,
Erwich JJH, et al. 17a-hydroxyprogesterone caproate for the
prevention of adverse neonatal outcome in multiple pregnancies: a
randomized controlled trial. Obstet Gynecol 2011;118:51320.
26 Rode L, Klein K, Nicolaides K, Krampl-Bettelheim E, Tabor A. Prevention
of preterm delivery in twin gestations (PREDICT): a multicenter,
randomized, placebo-controlled trial on the effect of vaginal micronized
progesterone. Ultrasound Obstet Gynecol 2011;38:27280.
27 Aboulghar MM, Aboulghar MA, Amin YM, Al-Inany HG, Mansour
RT, Serour GI. The use of vaginal natural progesterone for
prevention of preterm birth in IVF/ICSI pregnancies. Reprod Biomed
Online 2012;25:1338.
28 Wood S, Ross S, Tang S, Miller L, Sauve R, Brant R. Vaginal
progesterone to prevent preterm birth in multiple pregnancy: a
randomized controlled trial. J Perinat Med 2012;40:5939.
29 Senat M-V, Porcher R, Winer N, Vayssi ere C, Deruelle P, Capelle M,
et al. Prevention of preterm delivery by 17 alpha-
hydroxyprogesterone caproate in asymptomatic twin pregnancies
with a short cervix: a randomized controlled trial. Am J Obstet
Gynecol 2013;208:194 e1-.e8.
30 Serra V, Perales A, Meseguer J, Parrilla J, Lara C, Bellver J, et al.
Increased doses of vaginal progesterone for the prevention of
preterm birth in twin pregnancies: a randomised controlled double-
blind multicentre trial. BJOG 2013;120:507.
31 Awwad J, Usta I, Ghazeeri G, Yacoub N, Succar J, Hayek S, et al. A
randomised controlled double-blind clinical trial of 17-
hydroxyprogesterone caproate for the prevention of preterm birth in
twin gestation (PROGESTWIN): evidence for reduced neonatal
morbidity. BJOG 2015;122:719.
32 Brizot ML, Hernandez W, Liao AW, Bittar RE, Francisco RP, Krebs
VL, et al. Vaginal progesterone for the prevention of preterm birth
in twin gestations: a randomized placebo-controlled double-blind
study. Am J Obstet Gynecol 2015;213:82.e1-.e9.
2017 Royal College of Obstetricians and Gynaecologists

33 El-refaie W, Abdelhafez MS, Badawy A. Vaginal progesterone for
prevention of preterm labor in asymptomatic twin pregnancies with
sonographic short cervix: a randomized clinical trial of efficacy and
safety. Arch Gynecol Obstet 2016;293:617.
34 Dor J, Shalev J, Mashiach S, Blankstein J, Serr D. Elective cervical suture
of twin pregnancies diagnosed ultrasonically in the first trimester
following induced ovulation. Gynecol Obstet Invest 1982;13:5560.
35 MacNaughton M, Chalmers I, Dubowitz V, Dunn P, Grant A,
McPherson K, et al. Final report of the Medical Research Council/Royal
College of Obstetricians and Gynaecologists multicentre randomised
trial of cervical cerclage. Br J Obstet Gynaecol 1993;100:51623.
36 Rust OA, Atlas RO, Jones KJ, Benham BN, Balducci J. A randomized
trial of cerclage versus no cerclage among patients with
ultrasonographically detected second-trimester preterm dilatation of
the internal os. Am J Obstet Gynecol 2000;183:8305.
37 Berghella V, Odibo AO, Tolosa JE. Cerclage for prevention of
preterm birth in women with a short cervix found on transvaginal
ultrasound examination: a randomized trial. Am J Obstet Gynecol
2004;191:131117.
38 Liem S, Schuit E, Hegeman M, Bais J, de Boer K, Bloemenkamp K,
et al. Cervical pessaries for prevention of preterm birth in women
with a multiple pregnancy (ProTWIN): a multicentre, open-label
randomised controlled trial. The Lancet 2013;382:13419.
2017 Royal College of Obstetricians and Gynaecologists
Preterm birth prevention in twins
39 Goya M, de la Calle M, Pratcorona L, Merced C, Rodo
C, Mun~oz B,
et al. Cervical pessary to prevent preterm birth in women with twin
gestation and sonographic short cervix: a multicenter randomized
controlled trial (PECEP-Twins). Am J Obstet Gynecol 2016;214:145
52.
40 Nicolaides KH, Syngelaki A, Poon LC, de Paco Matallana C,
Plasencia W, Molina FS, et al. Cervical pessary placement for
prevention of preterm birth in unselected twin pregnancies:
a randomized controlled trial. Am J Obstet Gynecol 2016;214:3.
e19.
41 The World Bank. Country and Lending Groups [internet]. [http://da
ta.worldbank.org/about/country-and-lending-groups#High_income)
Accessed 19 November 2015.
42 Feinstein AR, Cicchetti DV. High agreement but low Kappa:
I. The problems of two paradoxes. J Clin Epidemiol 1990;1990:543
9.
43 Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J, et al.
Recommendations for examining and interpreting funnel plot
asymmetry in meta-analyses of randomised controlled trials. BMJ
2011;343:d4002.
44 Bender R, Friede T, Koch A, Kuss O, Schlattmann P, Schwarzer G,
et al. Performing meta-analyses in the case of very few studies.
Cochrane Colloquium; Seoul 2016.
11