jurnal bagus obgyn

7/25/2019 jurnal bagus obgyn

1/12

RESEARCH

1thebmj BMJ2015;350:g217 doi: 10.1136/bmj.h217

Centre or Womens HealthResearch, University oLiverpool and LiverpoolWomens Hospital, Liverpool LSS, UKSchool o Social andCommunity Medicine, Universityo Bristol, Bristol BS PS, UK

Correspondence to:Z Alfirevic [email protected]

Additional material is publishedonline only. To view please visitthe journal online ( http://dx.doi.org/./BMJ.h)

Cite this as: BMJ;:hdoi: ./bmj.h

Accepted: December

Labour induction with prostaglandins: a systematic review and

network meta-analysis

Zarko Alfirevic,Edna Keeney,Therese Dowswell,Nicky J Welton,Sofia Dias,Leanne V Jones,Kate Navaratnam,Deborah M Caldwell,

ABSTRACT

OBJECTIVES

To assess the effectiveness and saety o

prostaglandins used or labour induction.

DESIGN

Systematic review with Bayesian network meta-

analysis

DATA SOURCES

The Cochrane Pregnancy and Childbirth Groups

Database o Trials (which incorporates the results o a

broad generic search or all pregnancy and postpartum

trials). Sources included are CENTRAL, Medline,Embase, NHS Economic Evaluation Database, CINAHL,

relevant journals, conerence proceedings, and

registries o ongoing trials.

ELIGIBILITY CRITERIA FOR SELECTING STUDIES

Randomised clinical trials o prostaglandin or

prostaglandin analogues used or third trimester

cervical ripening or labour induction versus placebo or

no treatment, alternative prostaglandin dose or

administration, or a different type o prostaglandin. We

included studies recruiting women with a viable etus,

but had no other restrictions relating to indication or

labour induction or language o publication. Outcomes

assessed were serious neonatal morbidity (trialist

defined) or perinatal death; serious maternal

morbidity (trialist defined) or death; vaginal delivery

not achieved within hours, caesarean section, and

uterine hyperstimulation with etal heart rate changes.

RESULTS

randomised clinical trials were included (

women) in the review. Maternal and neonatal mortality

and serious morbidity were rarely reported and are

summarized narratively. Unresolved inconsistency was

observed or the hyperstimulation outcome. Relative

to placebo, the odds o ailing to achieve a vaginal

delivery were lowest or vaginal misoprostol ( g)

(odds ratio . (% credible interval . to .)),

with a % absolute probability o event (% credible

interval % to %). Compared with placebo, odds o

caesarean section were lowest or titrated oral

misoprostol solution (


2/12

RESEARCH

2 doi: 10.1136/bmj.h217 BMJ2015;350:g217 thebmj

evidence, both direct and indirect, a network meta-anal-

ysis produces estimates o the relative effects o each

treatment compared with all others in the network,

even i not all treatments have been directly compared

with each other. It is then possible to calculate the prob-

ability o one treatment being the best or a specific out-

come; in this way different treatment options can be

ranked rom best to worst or each outcome.

In this paper we present a network meta-analysis to

quantiy the effects and saety o different prostaglan-

dins used or labour induction. Originally conceived as

a standalone investigation, it is now also part o a larger

project looking at all methods or labour induction

(PROSPERO :CRD).

Methods

Search strategy and selection criteria

To identiy potentially eligible trials, we searched the

Cochrane Pregnancy and Childbirth Groups Database

o Trials (which incorporates the results o a broad

generic search or all pregnancy and postpartum trials).

Sources searched were CENTRAL, Medline, Embase,

NHS Economic Evaluation Database, CINAHL, relevant

journals, conerence proceedings, and registries o

ongoing trials. The ull text o every relevant trial report

was obtained and assigned to a topic depending on the

intervention beore adding to the database. This

approach leads to a more specific search. We then

screened all reports assigned to the induction o

labour topic. The detailed search strategy, along with

reerences or all reports identified by the search, are in

appendices . Inormation relating to the characteris-

tics o the included studies and different doses and reg-

imens used are outlined in appendices and . Thelatest search was completed in March .

We included all randomised clinical trials comparing

a prostaglandin or prostaglandin analogue used or

third trimester cervical ripening or labour induction

with placebo or no treatment, with the same prosta-

glandin administered by a different route or dose, or

with a different type o prostaglandin. We included only

studies recruiting women with a viable etus but had no

other restrictions relating to indication or labour

induction, language, or date o publication.

We included different types o prostaglandin or

prostaglandin analogue: vaginal prostaglandin Eas

tablets, gel, pessary, or sustained release pessary; intra-cervical prostaglandin E; prostaglandin F gel; vagi-

nal misoprostol tablet (dose


3/12

RESEARCH


treatment effect estimates to inormally assess agree-

ment. Studies with or % events in all arms were

excluded rom the analysis because these studies pro-

vide no evidence on relative effects.For studies with

a or % events in one arm only, we planned to

analyze the data without continuity corrections where

computationally possible. To avoid double counting o

events, multi-arm trials were analysed in their original

orm without the need to combine treatment arms.

Both fixed and random effects models (accounting

or the correlations induced between trial-specific

effects in multi-arm trials) were considered on the basis

o model fit. Goodness o fit was measured using the

posterior mean o the residual deviance, the degree o

between-study heterogeneity, and the deviance inor-

mation criterion. In a well fitting model the posterior

mean residual deviance should be close to the number

o data points.Heterogeneity was reported as the

posterior median between-trial standard deviation ()

with its % credible interval. Differences o points

or deviance inormation criterion were considered

meaningul.

Consistency between the different sources o evi-

dence was explored statistically by comparing the fit o

a model assuming consistency with a model which

allowed or inconsistency.I the inconsistency model

had the smallest posterior mean residual deviance, het-

erogeneity, or deviance inormation criterion value then

this indicates potential inconsistency in the data.

Where model fit was indicative o inconsistency, we

firstly planned to restrict analysis to those trials with

adequate allocation concealment. I this did not resolve

apparent inconsistency we planned urther subgroup

analyses using potential treatment effect modifiersidentified as being unevenly distributed across the

treatments. The impact o removing placebo and

no-treatment nodes rom the network was also exam-

ined in sensitivity analyses, because o the possibility

that such trials may have included lower risk popula-

tions than head-to-head comparisons o active treat-

ments.

Full details o priors and convergence checks are

given in appendix , but they are briefly summarized

here. Vague prior distributions were specified or treat-

ment effect and heterogeneity parameters. Convergence

was assessed using the Brooks-Gelman-Rubin diagnos-

tic plotsand was satisactory by simulationsor all outcomes. A urther simulation sample o at least

iterations post-convergence was obtained on

which all reported results were based.

Relative treatment effects are reported as posterior

median odds ratios and % credible intervals. We cal-

culated the probability o each treatment being first,

second, third, etc, most effective or each outcome. As

this metric can be unstable, the posterior median o the

ranking o each treatment (and % credible intervals)

are also reported, with the convention that the lower

the rank the better the treatment. The absolute proba-

bility o an event on each treatment was also calculated

by applying the log odds ratios to the log odds esti-mated rom a synthesis o reerence treatment (placebo)

arms.We used this to inorm the calculation o the

number needed to treat to harm (all events considered

here are undesirable). For the interested reader, poste-

rior median risk ratios are also reported in appendix

(table ).

Results

The search identified studies ( reports) poten-

tially eligible or inclusion in the review. O these,

studies were excluded (see fig or reasons or exclu-

sion). We included studies in the systematic review,

and data were available or at least one o our outcomes

or women. The data used in the analyses are

reported in appendix (tables ).

The complete comparison networksincluding

randomised controlled trials o prostaglandins exam-

ining different regimens, no treatment, and pla-

ceboare presented in figure or ailure to achieve

vaginal delivery in hours, caesarean section, uter-

ine hyperstimulation, serious neonatal morbidity or

perinatal death, and serious maternal morbidity or

death.

Vaginal delivery not achieved within -hours

Afer the exclusion o trials with zero events in all arms,

trials were available or network analysis. There were

no trials remaining that compared prostaglandin F.

Meaningul differences were observed in posterior

mean residual deviance and deviance inormation

Studies identified through searching Cochrane Pregnanyand Childbirth Group database of trials (n=; reports)

No of arms relating to different treatments (n=): No treatment (n=) Placebo (n=) Vaginal prostaglandin Etablet (n=) Vaginal prostaglandin Egel (n=) Vaginal prostaglandin Epessary (slow release) (n=) Prostaglandin Fgel (n=) Intracervical prostaglandin E(n=) Vaginal prostaglandin Epessary (normal release) (n=)

Vaginal misoprostol (dose


4/12

RESEARCH


VagPGEtablet = Vaginal prostaglandin Etablet; VagPGEgel = Vaginal prostaglandin Egel; VagPGEpessarySR = Vaginal prostaglandin Epessary (slow release);cxPGE = Intracervical prostaglandin E; VagPGEpessary = Vaginal prostaglandin Epessary (normal release); VagPGF gel = Vaginal prostaglandin Fgel;VagMiso mcg

OMisomcg

VagMiso>mcg

VagMiso


5/12

RESEARCH


criterion values, suggesting that, or the ull network,

the inconsistency model is preerred over the consis-

tency model (appendix , table ). To explore the source

o the observed inconsistency, we conducted a pre-spec-

ified sensitivity analysis examining the effect o remov-

ing trials at high risk o bias rom the network. The

random effects model, excluding these trials, provided

an adequate fit to the data (appendix , table ), and

reported results are based on this model, with trials

(Table, and fig ).

Despite the observation o high between-trials het-

erogeneity ( =. (% credible interval . to .))

(relative to the size o the treatment effect estimates as

measured on the log odds scale), there was strong

evidence that all prostaglandins increased the odds o

achieving a vaginal birth within hours when com-

pared with placebo (Table ). When active treatments

were compared, out o comparisons reached con-

ventional level o statistical significance (fig ). The

absolute probabilities o ailing to achieve a vaginal

birth within hours are shown in table , alongside the

odds ratios relative to placebo. We note that the abso-

lute probability o an event or vaginal misoprostol tab-

let g was % (% credible interval % to %),

closely ollowed by that or titrated (low dose) oral

misoprostol solution (% probability (% to %)).

Figure shows the distribution o the rankings or

each o the prostaglandin treatments. The x axis

reports each o the possible ranks, where position

means the treatment is ranked the highest and position

the lowest. The y axis shows the probability with

which each treatment has been ranked at each o the

possible positions and thereore ully encapsulates the

uncertainty in the treatment rankings. For example,consider vaginal prostaglandin E tablet in the top

lef-hand corner o the figure. For the outcome achiev-

ing vaginal delivery within hours o induction, the

probability o vaginal prostaglandin E tablet being

ranked first is %, the probability o being ranked sec-

ond is around %, peaking at a % probability o

being ranked sixth. We can conclude rom figure that

there is considerable uncertainty in the estimation o

any rank or vaginal prostaglandin Etablet. Note that

this uncertainty is also reflected in the credible interval

around the median rank, which spans rom st to th

best treatment or this outcome.

The best treatment or achieving a vaginal birth

within hours was vaginal misoprostol tablet g,

with a probability o being best o %. The probability

o being ranked in the top three treatments was % or

vaginal misoprostol tablet g and % or titrated

(low dose) oral misoprostol solution. The remaining

treatments all rated lower than a % probability o

being in the top three treatments. The probability o

being ranked in the bottom three (that is, poorest in

terms o achieving a vaginal birth within hours) was

% or vaginal prostaglandin E pessary (normal

release), % or oral misoprostol


6/12

RESEARCH


Vaginal prostaglandin Etablet v

Vaginal prostaglandin Egel

Vaginal prostaglandin Epessary (slow release)

Intracervical prostaglandin E Vaginal prostaglandin Epessary (normal release)

Vaginal misoprostol (dose


7/12

RESEARCH


included). We note that the results are not materially

different between the ull network ( trials) and the

sensitivity analysis network ( trials).

Table reports the posterior median odds ratios (%credible intervals) or each treatment relative to placebo

(the ull results are reported in appendix , table ). We

note that, or all but vaginal prostaglandin Etablet ver-

sus placebo (Table), the direct and network meta-anal-

ysis results are similar. However, this is consistent with

chance (rom statistical tests we might expect . to

be significant by chance), and we did not observe any

meaningul difference in residual deviance or deviance

inormation criterion values between the inconsistency

and consistency models or caesarean section (appendix

, table ). Relative to the size o the treatment effect esti-

mates, between-trial heterogeneity was observed or this

outcome ( =. (% credible interval . to .)).

When active treatments were compared, six out o

comparisons reached the conventional level o statisti-

cal significance (fig ). With placebo used as the reer-

ence (Table ), five active treatment regimens resulted insignificant reduction in caesarean section, namely vagi-

nal prostaglandin E (gel), vaginal misoprostol tab-

let


8/12

RESEARCH


or oral misoprostol tablet g. All other treatments

had a lower probability o being among the top three.

The probability o being ranked in the bottom three

(poorest in terms o risk o caesarean section) was %

or vaginal prostaglandin Epessary (normal release)

and % or oral misoprostol tablet


9/12

RESEARCH


Vaginal prostaglandin Etablet v Vaginal prostaglandin E(gel) Vaginal prostaglandin Epessary (slow release) Prostaglandin Fgel Intracervical prostaglandin E Vaginal prostaglandin Epessary (normal release) Vaginal misoprostol (dose


10/12

RESEARCH


prostaglandin Epessary (normal release) arm excluded

women with a previous caesarean section (appendix ,

table ). A similar distribution o characteristics was

observed or ailure to achieve vaginal delivery within

hours (appendix , table ).

It is possible that these differences were partly due to

poor reporting (that is, inclusion and exclusion criteria

were not always well reported). Nevertheless, subgroup

analyses were conducted or both vaginal delivery and

caesarean section outcomes, in which we removed

trials which had randomized women with previous

caesarean section (either all or some participants). The

findings were robust when these trials were removed

rom the analyses, with titrated (low dose) oral miso-

prostol solution still being the highest ranking treat-

ment or the outcome o caesarean section, and vaginal

misoprostol g remaining the highest ranking

treatment or achieving vaginal delivery within

hours (see appendix , table ). The one noticeable

change in results was the median rank o oral misopros-

tol tablet


11/12

RESEARCH


to have contributed to the difficulties in model fitting

and statistical inconsistency observed or this outcome.

In practice, caesarean section and uterine hyperstimu-

lation are clearly relatedone would expect that clini-

cally persistent, clinically important uterine

hyperstimulation will eventually result in caesarean

section.

Conclusions and implications for practice

The best saety profile o low dose misoprostol solution

with reasonable efficacy (median rank ) may have

important implications or clinical practice. It is note-

worthy that this method is currently not recommended

by the World Health Organization, while low dose

( g) oral tablets every hours are recommended

despite the worst overall ranking in our network

meta-analysis. In clinical situations where less re-

quent vaginal administration o misoprostol may be

preerable, particularly in settings with intensive moni-

toring acilities, g vaginal misoprostol tablets may

be a reasonable treatment o choice. Most o the studies

in our network meta-analysis used hourly treatment

regimens, and some opted or less requent administra-

tion ( hourly).

Overall, misoprostol may be the best prostaglandin

or labour induction, as titrated low dose oral solution

seems to be the saest in terms o caesarean section risk,

while vaginal misoprostol tablets ( g) are the most

effective in achieving vaginal delivery within hours

o induction. These findings have important implica-

tions or national and international guidelines or

induction o labour and uture research in this area.

We thank Lynn Hampson, Helen West, and Nancy Medley,Cochrane Pregnancy and Childbirth Group, or their support inpreparing the final draf o this paper.

Contributors : ZA and DMC conceived and designed the originalstudy, and are the study guarantors. ZA contributed to planning thereview and network meta-analysis, data interpretation and writingthe report. EK conducted the statistical analyses and revised thepaper. TD contributed to planning the review, data collection,quality assessment, and writing the report. NJW supervised thestatistical analyses and drafed and revised the manuscript. SDcontributed to and supervised the statistical analyses and revisedthe manuscript. LVJ contributed to data collection, qualityassessment, and writing the report. KN contributed to datacollection and quality assessment and commented on the report.DMC conducted and supervised the statistical analyses and drafedand revised the manuscript.

Funding: The work was supported in part by National Institute or

Health Research, UK, HTA project // (www.netscc.ac.uk). Theviews and opinions expressed therein are those o the authors and donot necessarily relect those o the NIHR, NHS, or Department oHealth. DMC is supported by an MRC Population Health Scientistpostdoctoral ellowship (G). The unders had no role in studydesign, data collection and analysis, decision to publish, orpreparation o the manuscript.

Competing interests: All authors have completed the ICMJE uniormdisclosure orm at www.icmje.org/coi_disclosure.pd (available onrequest rom the corresponding author) and declare: support romNIHR or the submitted work; no financial relationships with anyorganisations that might have an interest in the submitted work in theprevious three years; no other relationships or activities that couldappear to have inluenced the submitted work.

Data sharing: All data and statistical code are available in theappendices provided by the authors.

Transparency: The manuscripts guarantors affirm that the manuscriptis an honest, accurate, and transparent account o the study beingreported; that no important aspects o the study have been omitted;

and that any discrepancies rom the study as planned (and, i relevant,registered) have been explained.

This is an Open Access article distributed in accordance with theCreative Commons Attribution Non Commercial (CC BY-NC .) license,

which permits others to distribute, remix, adapt, build upon this worknon-commercially, and license their derivative works on different terms,

provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/./.

American College o Obstetricians and Gynecologists. ACOGpractice bulletin. Clinical Management Guidelines or Obstetrician-Gynecologists. No. . Induction o L abor. Obstet Gynecol;:.

Chauhan SP, Ananth CV. Induction o labor in the United States:a critical appraisal o appropriateness and reducibility. SeminPerinatol;:.

National Institute or Health and Care Excellence. Induction olabour. (Clinical guideline ). NICE, . www.nice.org.uk/guidance/cg.

Tang J, Kapp N, Dragoman M, de Souza JP. WHO recommendationsor misoprostol use or obstetric and gynecologic indications.Int J Gynaecol Obstet;:.

Austin SC, Sanchez-Ramos L, Adair CD. Labor induction withintravaginal misoprostol compared with the dinoprostone vaginalinsert: a systematic review and metaanalysis.Am J Obstet Gynecol;:.e.

Facchinetti F, Fontanesi F, Del Giovane C. Pre-induction o labour:

comparing dinoprostone vaginal insert to repeated prostaglandinadministration: a systematic review and meta-analysis.J Matern FetalNeonatal Med;:.

Mozurkewich EL, Chilimigras JL, Berman DR, Perni UC, Romero VC,King VJ, et al. Methods o induction o labour: a systematic review.BMC Pregnancy Childbirth;:.

Caldwell DM, Ades AE, Higgins JPT. Simultaneous comparison omultiple treatments: combining direct and indirect evidence.BMJ;:.

Lee AW. Review o mixed treatment comparisons in publishedsystematic reviews shows marked increase since .

J Clin Epidemiol;:. Higgins JPT, Whitehead A. Borrowing strength rom external trials in a

meta-analysis. Stat Med;:. Lu G, Ades AE. Combination o direct and indirect evidence in mixed

treatment comparisons. Stat Med;:. Wells GA, Sultan SA, Chen L, Khan M, Coyle D. Indirect evidence:

indirect treatment comparisons in meta-analysis. Canadian Agency

or Drugs and Technologies in Health, . National Institute or Health and Care Excellence.Guide to the methodsof technology appraisal. NICE, . www.nice.nhs.uk/aboutnice/howwework/devnicetech/GuideToMethodsTechnologyAppraisal.jsp.

Cochrane Collaboration. Comparing Multiple Interventions Methods Group.Comparing multiple interventions in Cochrane reviews. . http://cmimg.cochrane.org/comparing-multiple-interventions-cochrane-reviews.

Haas DM, Caldwell DM, Kirkpatrick P, McIntosh JJ, Welton NJ. Tocolytictherapy or preterm delivery: systematic review and networkmeta-analysis. BMJ;:e

Stettler C, Wandel S, Allemann S, Kastrati A, Morice MC, Schmig A,et al. Outcomes associated with drug-eluting and bare-metal stents:a collaborative network meta-analysis. Lancet ;:.

Cipriani A, Furukawa TA, Salanti G, Geddes JR, Higgins JPT, Churchill R,et al. Comparative efficacy and acceptability o new-generationantidepressants: a multiple-treatments meta-analysis. Lancet;:.

Barth J, Munder T, Gerger H, Nesch E, Trelle S, Znoj H, et al.

Comparative efficacy o seven psychotherapeutic interventions orpatients with depression: a network meta-analysis. PLOS Med;:e

Singh JA, Christensen R, Wells GA, Suarez-Almazor ME, Buchbinder R,Lopez-Olivo MA, et al. Biologics or rheumatoid arthritis: an overviewo Cochrane reviews. Cochrane Database Syst Rev;():CD.

Padwal R, Klarenbach S, Wiebe N, Birch D, Karmali S, Manns B, et al.Bariatric surgery: a systematic review and network meta-analysis orandomized trials. Obesity Rev;:.

Homeyr G, Alfirevic Z, Kelly T, Kavanagh J, Thomas J, Brocklehurst P,et al. Methods or cervical ripening and labour induction in laterpregnancy: generic protocol [protocol]. Cochrane Database Syst Rev;():CD.

Kelly A, Alfirevic Z, Homeyr G, Kavanagh J, Neilson J, Thomas J.Induction o labour in specific clinical situations: generic protocol[protocol]. Cochrane Database Syst Rev;():CD.

Higgins J, Green S, eds. Cochrane handbook for systematic reviews ofinterventions. Version ... Cochrane Collaboration, . www.cochrane-handbook.org; .

Wood L, Egger M, Gluud LL, Schulz KF, Juni P, Altman DG, et al.Empirical evidence o bias in treatment effect estimates in controlled


12/12

RESEARCH

trials with different interventions and outcomes: meta-epidemiological study. BMJ;:.

Glenny AM, Altman DG, Song F, Sakarovitch C, Deeks JJ, DAmico R,et al. Indirect comparisons o competing interventions. Health Technol

Assess;:. OpenBUGS. www.openbugs.net/. Dias S, Ades AE, Sutton AJ, Welton NJ. Evidence synthesis or decision

making : a generalized linear modeling ramework or pairwise andnetwork meta-analysis o randomized controlled trials. Med Decis

Making;:. The BUGS project. DIC: Deviance Inormation Criterion. How does DICdepend on the parameterisation used? www.mrc-bsu.cam.ac.uk/sofware/bugs/the-bugs-project-dic/#q.

Welton N, Sutton A, Cooper N, Abrams K, Ades A. Evidence synthesisfor decision making in healthcare. Wiley, .

Dias S, Welton NJ, Sutton AJ, Caldwell DM, Lu G, Ades AE. Evidencesynthesis or decision making : inconsistency in networks oevidence based on randomized controlled trials. Med Decis Making;:.

Wing DA, Lovett K, Paul RH. Disruption o prior uterine incisionollowing misoprostol or labor induction in women with previouscesarean delivery. Obstet Gynecol;:.

World Health Organization. WHO recommendations forinduction of labour. WHO, . http://whqlibdoc.who.int/publications//_eng.pd.

Brooks SP, Gelman A. Alternative methods or monitoringconvergence o iterative simulations.J Comp Graph Stat;:.

Gelman A, Rubin DB. Inerences rom iterative simulation usingmultiple sequences. Stat Sci;:.

Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphicaltools or network meta-analysis in STATA. PLoS One;:e.

BMJ Publishing Group Ltd

Appendix . Search strategy used to maintain the

Cochrane Pregnancy and Childbirth Groups Database

o Trials

Appendix . List o studies included in network

meta-analysis

Appendix . List o studies excluded rom network

meta-analysis

Appendix . Description o included studies

Appendix . Description o included studiesdoses

and regimens

Appendix . Tables : Characteristics o trials by

treatment and outcome. Tables : Odds ratios or out-

comes with each treatment. Tables : Model fit and

selection statistics by outcome. Tables : Datafiles

used in OpenBUGS analyses or outcomes. Table :

Posterior median rankings or sensitivity analyses.

Table : Comparison o risk ratios and odds ratios or

vaginal delivery and caesarean section

Appendix . Technical appendix

Appendix . Inconsistency plots

Documents

jurnal bagus obgyn