OBJECTIVE: To evaluate the efficacy of maintenance ization to delivery. Excluded studies used progestogens
tocolysis with progestogens compared with placebo or as prevention in asymptomatic women at risk. Risk of
no treatment in women with singleton pregnancies and bias assessment, subgroup analysis on type of proges-
arrested preterm labor. togens used, and sensitivity analysis by high-quality
DATA SOURCES: Studies without language restrictions studies were performed.
were identified from MEDLINE, EMBASE, PubMed, Sco- TABULATION, INTEGRATION, AND RESULTS: Sixteen
pus, the Cochrane Pregnancy and Childbirth Groups Tri- randomized controlled trials consisting of 1,917 partic-
als Register, the Cochrane Central Register of Controlled ipants were included. Study characteristics and quality
Trials, and ClinicalTrials.gov from inception to June 2015. were recorded. Preterm delivery at less than 37 weeks of
MeSH headings for progestogens were combined with gestation was decreased (38.2% compared with 44.3%;
terms regarding labor, tocolysis, or preterm birth. Refer- relative risk 0.79, 95% confidence interval [CI] 0.650.97)
ence lists of included studies and GoogleSearch were and pregnancy was prolonged (mean difference 8.1 days;
also reviewed. 95% CI 3.812.4) when women treated with progesto-
METHODS OF STUDY SELECTION: Randomized con- gens were compared with placebo or no treatment.
trolled trials that compared progestogens as a mainte- There were no differences in the outcome of delivery
nance treatment after arrested preterm labor in singleton at less than 34 weeks of gestation (15.6% compared with
pregnancies with placebo or no treatment were identi- 18.3%; relative risk 0.77, 95% CI 0.531.12). However,
fied. Selected studies evaluated delivery before 37 or 34 sensitivity analysis including five high-quality studies
weeks of gestation or the latency period from random- showed no significant differences for preterm delivery
at less than 37 weeks of gestation (37.2% compared with
From the BCNatalBarcelona Center for Maternal-Fetal and Neonatal Medicine 36.9%; relative risk 0.91, 95% CI 0.671.25) or latency
(Hospital Clnic and Hospital Sant Joan de Deu), Fetal i+D Fetal Medicine
Research Center, IDIBAPS, University of Barcelona, Barcelona, and the Centre period (mean difference 0.6 days; 95% CI 23.7 to 4.9).
for Biomedical Research on Rare Diseases (CIBER-ER), Spain; the Department CONCLUSION: There is insufficient high-quality data to
of Obstetrics and Gynaecology, Mount Sinai Hospital, University of Toronto,
Toronto, Ontario, Canada; and the Department of Obstetrics and Gynecology,
inform clinicians and patients about the use of proges-
Division of Maternal-Fetal Medicine, University of Florida Health Science togens as maintenance treatment after arrested preterm
Center, Jacksonville, Florida. labor to reduce the incidence of preterm birth or
Montse Palacio was supported by Instituto de Salud Carlos III (BA15/00028 pregnancy prolongation.
and EC07/90023) and Ministerio de Sanidad y Poltica Social (TRA-096).
(Obstet Gynecol 2016;128:9891000)
The authors thank Daphne Horn, Research Librarian at Mount Sinai Hospital, DOI: 10.1097/AOG.0000000000001676
for help with the literature search, and Josie Chundamala, Scientific Grant
M
Editor in the Department of Obstetrics and Gynaecology at Mount Sinai Hos-
pital, for editorial assistance. ore than 1 in 10 of the worlds children born in
Corresponding author: Montse Palacio, MD, PhD, BCNatal, Barcelona Centre for 2010 were born preterm, defined as before 37
Maternal-Fetal and Neonatal Medicine, Hospital Clnic, Sabino de Arana 1, 08028 weeks of gestation.1 Current tocolytic agents appear to
Barcelona, Spain; e-mail: mpalacio@clinic.cat.
be superior to placebo at delaying delivery at both 48
Financial Disclosure
The authors did not report any potential conflicts of interest.
hours and 7 days but do not reduce the incidence of
preterm birth or improve perinatal outcome.2 Main-
2016 by The American College of Obstetricians and Gynecologists. Published
by Wolters Kluwer Health, Inc. All rights reserved. tenance tocolytic therapy has not been proven to
ISSN: 0029-7844/16 improve the outcomes in women treated because of
also performed consisting of high-quality trials based lication bias statistically, Beggs test for small study
on Jadads criteria.25 effects was used (P,.05).
To test for the presence of publication bias, All analyses were conducted and summary results
a visual inspection of the funnel plots displaying generated as forest plots using Review Manager
individual study log RR with the standard error RR (RevMan) 5.3.
for binary outcomes and mean difference with stan-
dard error of mean difference for continuous out- RESULTS
comes was performed. Asymmetry in such funnel Sixteen randomized controlled trials met the inclu-
plots is usually caused by small trials that report sion criteria and were included in this meta-analysis
greater effects, on average, than large trials, which (Fig. 1).1318,20,2936 One trial32 had three arms: one using
suggests publication or other biases.28 To assess pub- progesterone, one using 17-hydroxyporgesterone, and
VOL. 128, NO. 5, NOVEMBER 2016 Palacio et al Progestogens as Maintenance Treatment 991
Country Years M vs Control) Route Control Primary Tocolytic Agent Medication (wk)
Borna and Sahabi,17 20042005 S 70 (37 vs 33) 400 OD, VAG No T MgSO4 Arrested, tocolysis discontinued 2434 6/7
2008, Iran
El-Abidin et al,35 20072008 S 40 (20 vs 20) 200 BID, VAG No T MgSO4 Arrested, tocolysis discontinued 2434
2009, Egypt
Sharami et al,33 2010, 20072009 S 163 (80 vs 83) 200 OD, VAG Placebo MgSO4 Arrested, tocolysis not specified 2834 6/7
Iran
Arikan et al,18 2011, NA S 83 (43 vs 40) 200 OD, VAG No T Ritodrine Acute preterm labor 2434 6/7
Turkey
Saleh Gargari et al,30 20072010 S 144 (72 vs 72) 400 OD, VAG No T MgSO4 Arrested, tocolysis not specified 2434 6/7
2012, Iran
Areia et al,15 2013, 20082010 S 52 (26 vs 26) 200 OD, VAG No T Atosiban Arrested, tocolysis discontinued 2434 6/7
Portugal
Lotfakizadeh et al,32 2010 S 73 (37 vs 36) 400 OD, VAG No T MgSO4 or nifedipine Arrested, on tocolysis 2636 6/7
2013, Iran
Mishra and Inamdar,36 20122014 S 100 (50 vs 50) 400 OD, VAG No T Isoxsuprine hydrochloride Arrested, tocolysis discontinued 2836
2014, India
Martinez de Tejada 20062012 M 379 (193 vs 186) 200 OD, VAG Placebo B-mimetic, atosiban, or Ca+ Arrested, tocolysis not specified 2433 6/7
et al,20 2015, channel blockers
Switzerland and
Argentina
Palacio et al,23 2016, 20082012 M 258 (126 vs 132) 200 OD, VAG Placebo B-mimetic, atosiban, nifedipine Arrested, tocolysis discontinued 2433 6/7
Spain
Noblot et al,29 1991, 1987 S 44 (22 vs 22) 300 TID, PO Placebo Ritrodrine Acute preterm labor ,35
France
Choudhary et al,16 20102012 S 90 (45 vs 45) 200 OD, PO Placebo Nifedipine Arrested, on tocolysis 2434 6/7
2014, India
Facchinetti et al,31 20042006 S 60 (30 vs 30) 341 bwk, IM No T Atosiban Not specified if arrested, on tocolysis 2533 6/7
2007, Italy
Rozenberg et al,13 20062008 M 188 (94 vs 94) 500 bwk, IM No T Nifedipine, nicardipine Arrested, tocolysis not specified 2431 6/7
OBSTETRICS & GYNECOLOGY
Mean GA at
Randomization Definition Prior PTB Progestogens vs PTB at Less Than 37 Wk of PTB at Less Than 34 Wk Latency (d;
(wk) of PTL Control (%) Study Primary Outcomes Gestation, n (%) of Gestation, n (%) mean)
31/32 .6 ctxs/309+cervical changes 5/37 (13.5) vs 4/33 (12.1) Latency period, recurrent PTL NA NA 36.1 vs 24.5
(manual)
27/28 .2 ctxs/109+cervical changes 9/20 (45) vs 7/20 (35) Latency period and recurrent NA 2/20 (10), 11/20 (55) 8.3 vs 5.0
(manual) PTL
33/34 $6 ctxs/309 for 30sec+cervical 1/80 (1.3) vs 3/83 (3.6) Latency period and PTB less 33/80 (41), 45/83 (54) 8/80 (10), 9/83 (11) 23.9 vs 16.7
changes (manual) than 37 and 34 wk of
gestation
32/32 .6 ctxs/309+cervical changes 4/43 (9.3) vs 3/40 (7.5) Latency period, GA at delivery 20/43 (47), 28/40 (70) 20/43 (47), 28/40 (70) 32.1 vs 21.2
(manual) and PTB less than 37 wk of
gestation
32/32 $4 ctxs/209 or $8 ctxs/ NA Latency period, GA at delivery NA NA 28.0 vs 9.8
609+cervical changes
(manual)
28/29 4 ctxs/209 or 8 ctxs/609 9/26 (34.6) vs 9/26 (34.6) Latency period 9/26 (35), 13/26 (50) 3/26 (12), 6/26 (23) 55 vs 38
+cervical changes (TVUS or
fFN+)
34/33 4 ctxs/209+2 cm cervical NA Rate of recurrent PTL NA NA 31 vs 26
dilatation or progressive
cervical changes (manual)
Stratified for 4 ctxs/209 or 8 ctxs/609 Latency, birth weight, NICU 17/50 (34), 34/50 (68) 4/50 (8), 13/50 (26) 23.4 vs 11.7
+cervical dilatation .1 cm
Palacio et al
range of GA admission
and eff .80%
29/29 $2 ctxs/109 for 309+cervical 46/193 (23.8) vs 39/186 (21.0) PTB less than 37 wk of 82/193 (43), 66/186 (36) 38/193 (20), 24/186 (13) 45 vs 52
changes (TVUS or manual, gestation
or fFN+)
32/32 $2 ctxs/109 for 309+cervical 11.9 vs 12.1 PTB less than 37 and 34 wk of 36/126 (29), 29/132 (22) 9/126 (7), 10/132 (8) NA
changes (TVUS and manual) gestation
32/31 Regular ctxs /109 for 1h NA NA 8/22 (36), 6/22 (27) 44 vs 42
Progestogens as Maintenance Treatment
and latency to delivery was reported in 11, 8, and 16 no differences in the outcome of delivery at less than
out of 16 studies, respectively. 34 weeks of gestation (15.6% compared with 18.3%;
The forest plots in Figures 35 provide study- RR 0.77, 95% CI 0.531.12). Significant heterogeneity
specific details regarding evaluation of pooled pre- among studies was noted for preterm delivery at less
term delivery at less than 37 weeks of gestation, less than 37 weeks of gestation (I2 65%, P5.001), less than
than 34 weeks of gestation, and latency period 34 weeks of gestation (I2 56%, P5.03), and latency
between preterm labor and delivery. Preterm delivery period (I2 96%, P,.001).
at less than 37 weeks of gestation was decreased Only five studies fulfilled all of Jadads criteria.
(38.2% compared with 44.3%; RR 0.79, 95% CI Sensitivity analysis of these studies showed no signif-
0.650.97) and pregnancy was prolonged (mean dif- icant differences in preterm delivery at less than 37
ference 8.1 days; 95% CI 3.812.4) when women trea- weeks of gestation (36.9% compared with 37.2%; RR
ted with progestogens were compared with women 0.91, 95% CI 0.671.25), preterm delivery at less than
who received placebo or no treatment. There were 34 weeks of gestation (14.2% compared with 11.7%;
VOL. 128, NO. 5, NOVEMBER 2016 Palacio et al Progestogens as Maintenance Treatment 995
RR 1.20, 95% CI 0.861.69), or for latency period necrotizing enterocholitis, or sepsis) were inconsis-
(mean difference 0.6 days; 95% CI 23.7 to 4.9). Sen- tently reported (Appendix 2, http://links.lww.com/
sitivity analysis did not explain the source of hetero- AOG/A869).
geneity for preterm delivery at less than 37 weeks of Table 2 summarizes the overall results and the
gestation (I2 65%, P5.02) or the latency period (I2 results of the sensitivity analysis on the five high-
86%, P,.001), but did explain the heterogeneity for quality studies (Jadad score of 5) for all outcomes.
preterm delivery at less than 34 weeks of gestation (I2 Use of progestogens did not improve the primary or
0%, P5.55). secondary outcomes. Appendix 3, available online at
There were limited data reporting on the second- http://links.lww.com/AOG/A869, summarizes the re-
ary outcomes of gestational age at delivery (13 of 16 sults for the secondary outcomes of the subgroup anal-
studies), recurrence of preterm labor (8 of 16 studies), ysis when the type of progestogen used was
neonatal admission to the neonatal intensive care unit considered. Subgroup analysis did not show addi-
(12 of 16 studies), or neonatal respiratory distress tional benefit for any of the progestogens used.
syndrome (10 of 16 studies). Other outcomes related Analysis of publication bias based on Eggers test
to neonatal morbidity (intraventricular hemorrhage, showed publication bias for the outcome of latency to
delivery (17 studies, P5.010) (Fig. 6) and suggests bias no differences in preterm birth at less than 37 weeks
for preterm birth at less than 37 weeks of gestation (11 of gestation, less than 34 weeks of gestation, or in
studies, P5.095) (Appendix 4, available online at the latency period to delivery between the pro-
http://links.lww.com/AOG/A869). gestogen and placebo arms.
Two recent meta-analyses21,22 have shown that
DISCUSSION progestogens, when used after arrested preterm labor,
There is insufficient high-quality evidence to eval- might be of some benefit. However, the authors were
uate the use of progestogens as maintenance treat- cautious with their conclusions because of the limited
ment after arrested preterm labor to either reduce quality of the studies published (ie, small sample size
preterm birth or increase latency to delivery. or lack of blinding). Since that publication, a subse-
Studies included in this meta-analysis were variable quent meta-analysis showed conflicting results,37
in both quality and design because they varied in recent data have been published,23 and additional
size, type of progestogen, route of administration, studies in which progestogen was introduced shortly
and whether they included a placebo arm. after preterm labor was arrested were not included in
Although differences in the primary outcomes were the earlier meta-analyses20; a review of the topic at this
found when all trials were included, sensitivity time is timely. Furthermore, previous systematic re-
analysis performed on high-quality trials showed views21,22,37 did not include small trials not cited in
VOL. 128, NO. 5, NOVEMBER 2016 Palacio et al Progestogens as Maintenance Treatment 997
PTB at less than 37 11 1,462 65 .001 0.79 (0.650.97)* 5 934 65 .02 0.91 (0.671.25)*
wk of gestation
PTB at less than 34 8 1,263 56 .03 0.77 (0.531.12) 4 890 0 .55 1.20 (0.861.69)
wk of gestation
Latency to delivery (d) 16 1,917 96 ,.001 8.1 (3.812.4)* 5 929 86 ,.001 0.6 (23.74.9)*
GA at delivery 13 1,672 87 ,.001 1.31 (0.801.81) 4 890 69 .02 0.14 (20.510.78)
Recurrent preterm 8 1,047 0 .56 0.63 (0.520.78) 3 725 20 .29 0.74 (0.531.02)
labor
Admission to NICU 12 1,672 33 .13 0.91 (0.751.10) 4 882 0 .98 1.15 (0.871.52)
Respiratory distress 10 1,057 7 .38 0.70 (0.500.97) 3 506 0 .97 0.75 (0.431.31)
syndrome
PTB, preterm birth; GA, gestational age; NICU, neonatal intensive care unit; CI, confidence interval.
* Risk ratio (IV, random, 95% CI).
PubMed.35,36 This systematic review also addresses care unit, and neonatal distress syndrome ensured
this deficiency. a complete evaluation of the intervention. In addition,
The strengths of this systematic review included a subgroup analysis was performed and allowed the
its comprehensive search strategy, methodologic comparison of the estimated effect of different proges-
design, and statistical analysis. In particular, it synthe- togens, which is relevant for different clinical practices
sized the results of existing studies in which pro- and sensitivity analysis showed results derived from
gestational agents were used as a maintenance inclusion of high-quality studies.
treatment after arrested preterm labor whether or The studies heterogeneity was this systematic
not the treatment was initiated together with acute reviews most obvious limitation and limits the val-
tocolysis. The inclusion of a range of outcomes such idity of the combined results, especially because
as preterm birth, latency period from randomization findings from larger and low risk of bias trials con-
to delivery, recurrence of preterm labor, gestational flicted with those from smaller trials. The potential
age at delivery, admissions to the neonatal intensive for selection bias exists in randomized studies when
there is no blinding or inadequate concealment,
which happened in 10 of 16 studies (Fig. 2), limiting
the quality of the data. Indeed, heterogeneity might
also be explained by the fact that inclusion criteria
among studies were very different. Evidence for that
is the high rate of preterm birth in some studies
compared with others. On the other hand, some out-
comes (ie, preterm delivery at less than 34 weeks of
gestation) were reported in some studies but not in
others. The inclusion of core outcome sets for pre-
term delivery in future research may avoid hetero-
geneity resulting from this fact.
In addition, the risk of publication bias was
assessed by visual inspection of the funnel and
asymmetric plots and suggested publication bias. This
Fig. 6. Funnel plot of studies (comparing progestational
agents and nontreatment or placebo) for outcome: latency suggests underreporting of negative trials. Thus, this
to delivery. comprehensive systematic review provides data to
Palacio. Progestogens as Maintenance Treatment. Obstet Gynecol evaluate these limitations and forms the basis for
2016. future studies.
VOL. 128, NO. 5, NOVEMBER 2016 Palacio et al Progestogens as Maintenance Treatment 999