Systematic Review

Dexamethasone versus betamethasone for

preterm birth: a systematic review and
network meta-analysis
Agustı́n Ciapponi, MD, MSc, PhD; Karen Klein, MD, MSc; Daniela Colaci, MD, MSc; Fernando Althabe, MD, MSc;
José M. Belizán, MD, PhD; Allie Deegan; Areti Angeliki Veroniki, MSc, PhD; Ivan D. Florez, MD, MSc, PhD

OBJECTIVE: This study aimed to evaluate the comparative clinical effectiveness and safety of dexamethasone vs betamethasone for
preterm birth.
DATA SOURCES: The sources searched were MEDLINE, EMBASE, Cochrane Library, LILACS, ClinicalTrials.gov, and International Clinical
Trials Registry Platform without language restrictions until October 2019 in addition to the reference lists of included studies. Field experts
were also contacted.
STUDY ELIGIBILITY CRITERIA: Randomized or quasi-randomized controlled trials comparing any corticosteroids against each other or
against placebo at any dose for preterm birth were included in the study.
METHODS: Three researchers independently selected and extracted data and assessed the risk of bias of the included studies by using
Early Review Organizing Software and Covidence software. Random-effects pairwise meta-analysis and Bayesian network meta-analysis
were performed. The primary outcomes were chorioamnionitis, endometritis or puerperal sepsis, neonatal death, respiratory distress
syndrome, and neurodevelopmental disability.
RESULTS: A total of 45 trials (11,227 women and 11,878 infants) were included in the study. No clinical or statistical difference was found
between dexamethasone and betamethasone in neonatal death (odds ratio, 1.05; 95% confidence interval, 0.62e1.84; moderate-
certainty evidence), neurodevelopmental disability (odds ratio, 1.03; 95% confidence interval, 0.80e1.33; moderate-certainty evi-
dence), intraventricular hemorrhage (odds ratio, 1.04; 95% confidence interval, 0.56e1.78); low-certainty evidence), or birthweight
(þ5.29 g; 95% confidence interval, 49.79 to 58.97; high-certainty evidence). There was no statistically significant difference, but a
potentially clinically important effect was found between dexamethasone and betamethasone in chorioamnionitis (odds ratio, 0.70; 95%
confidence interval, 0.45e1.06; moderate-certainty evidence), fetal death (odds ratio, 0.81; 95% confidence interval, 0.24e2.41; low-
certainty evidence), puerperal sepsis (odds ratio, 2.04; 95% confidence interval, 0.72e6.06; low-certainty evidence), and respiratory
distress syndrome (odds ratio, 1.34; 95% confidence interval, 0.96e2.11; moderate-certainty evidence). Meta-regression, subgroup,
and sensitivity analyses did not reveal important changes regarding the main analysis.
CONCLUSION: Corticosteroids have proven effective for most neonatal and child-relevant outcomes compared with placebo or no
treatment for women at risk of preterm birth. No important difference was found on neonatal death, neurodevelopmental
disability, intraventricular hemorrhage, and birthweight between corticosteroids, and there was no statistically significant dif-
ference, but a potentially important difference was found in chorioamnionitis, fetal death, endometritis or puerperal sepsis, and
respiratory distress syndrome. Further research is warranted to improve the certainty of evidence and inform health policies.
Key words: antenatal corticosteroids, betamethasone, dexamethasone, network meta-analysis, preterm birth, systematic review

Instituto de Efectividad Clínica y Sanitaria, Buenos Aires, Argentina (Drs Ciapponi, Klein, Colaci, and Belizán); Department of Reproductive Health and
Research, World Health Organization, Geneva, Switzerland (Dr Althabe); School of Global Public Health, New York University, New York, NY (Ms Deegan);
Department of Primary Education, School of Education Sciences, University of Ioannina, Ioannina, Greece (Dr Veroniki); Knowledge Translation Program,
Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, Canada (Dr Veroniki); Department of Pediatrics, University of Antioquia, Medellín,
Colombia (Dr Florez); and Department of Health Research Methods, Evidence and Impact, McMaster University, Hamilton, Canada (Dr Florez).
Received Nov. 8, 2020; revised Jan. 8, 2021; accepted Jan. 12, 2021.
The authors declare no conflict of interest.
All authors have completed the International Committee of Medical Journal Editors uniform disclosure form at www.icmje.org/coi_disclosure.pdf and
declare no support from any organization for the submitted work, no financial relationships with any organizations that might have an interest in the
submitted work in the previous 3 years, and no other relationships or activities that seem to have influenced the submitted work.
The manuscript’s guarantor affirmed that the manuscript is an honest, accurate, and transparent account of the study being reported, that no important
aspect of the study has been omitted, and that any discrepancies from the study as planned have been explained.
Cite this article as: Ciapponi A, Klein K, Colaci D, et al. Dexamethasone versus betamethasone for preterm birth: a systematic review and network meta-
analysis. Am J Obstet Gynecol MFM 2021;3:100312.
Corresponding author: Agustín Ciapponi, MD, MSc, PhD. [email protected]
2589-9333/$36.00
ª 2021 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.ajogmf.2021.100312

Click Supplemental Materials under article title in Contents at

MAY 2021 AJOG MFM 1

Systematic Review

Objectives
AJOG MFM at a Glance
This study aimed to evaluate the
Why was this study conducted? comparative clinical effectiveness and
This study was conducted to further analyze the clinical advantages of 2 corti- safety of dexamethasone vs betametha-
costeroids routinely used during preterm delivery, with the ultimate goal of sone for preterm birth.
determining if one is more advantageous over the other.
Methods
Key findings This systematic review and NMA are based
This network meta-analysis included recently published evidence that almost on the Preferred Reporting Items for Sys-
doubled the number of participants involved in the direct comparison between tematic Reviews and Meta-Analysis
corticosteroids. (PRISMA) extension statement for
This analysis not only showed no difference in neonatal death, neuro- NMA.15 The study protocol was registered
developmental disability, intraventricular hemorrhage, or birthweight but also in PROSPERO (CRD42017078006).
showed no statistically significant but potentially important differences in cho-
rioamnionitis, fetal death, puerperal sepsis, and respiratory distress syndrome. Eligibility criteria
To be eligible, studies had to be pub-
What does this add to what is known? lished or unpublished randomized
These mixed results justify shared decision-making with patients and warrant controlled trials (RCT) or quasi-RCT
further research to improve the certainty of most results to inform health policies that included women at risk of pre-
regarding preterm birth. term birth and comparing any cortico-
steroids against each other or against
Introduction placebo regardless of the dose or
Preterm birth (<37 weeks’ gestation) schedule (Table B.1 and Table B.2
accounts for around 11% of all live births Medicines List and in WHO’s “Man- describe the doses by study). The pop-
worldwide, poses risks of adverse out- aging complications in pregnancy and ulation was women with a singleton or
comes, and can be attributed to 35% of childbirth guide.”8 multiple pregnancy expected to give
deaths among newborns.1e3 Preterm In addition, 2 Cochrane systematic birth before term (before 37 weeks’
birth represents a substantial health reviews have synthesized the effects of gestation) as a result of either sponta-
burden worldwide, mainly in low- and corticosteroids. Brownfoot et al9 and neous preterm labor, preterm prelabor
middle-income countries (LMICs). Roberts et al10 compared any cortico- rupture of membranes, medical indi-
Respiratory distress syndrome (RDS) steroids for preterm birth with each cation for delivery, or elective preterm
is a serious complication of preterm other or with placebo, respectively. birth.
birth and the primary cause of early Although Brownfoot et al9 focused on The primary outcomes (defined by
neonatal death, lifelong disability, and direct comparisons, the authors also the study authors) were chorioamnio-
poor quality of life. RDS affects up to half assessed indirect comparisons of corti- nitis and endometritis or puerperal
of babies born before 28 weeks’ gestation costeroids with placebo for some out- sepsis for the mother and neonatal death
and a third of babies born before 32 comes based on Roberts et al’s study.11 and RDS for the fetus or neonate. The
weeks’ gestation.4 Antenatal corticoste- Although the indirect estimates suggest child-relevant outcome was neuro-
roids for preterm birth prevent RDS and no considerable difference between cor- developmental disability at follow-up
neonatal mortality.5 However, there still ticosteroids for puerperal sepsis, an (blindness, deafness, moderate or se-
persists doubt about the applicability in important difference favored betame- vere cerebral palsy), development delay
LMICs,6 and there is no consensus thasone for chorioamnionitis.11 or intellectual impairment (defined as
regarding the type of corticosteroid to Direct comparisons in Brownfoot developmental quotient or intelligence
use or the dose, frequency, timing of use, et al’s9 study showed that dexametha- quotient less than 2 standard deviation
or route of administration. Currently, sone may have some benefits compared below population mean16), or other
either betamethasone or dexamethasone with betamethasone, such as less intra- neurologic disorders.
is the recommended corticosteroid for ventricular hemorrhage. Roberts et al10 The secondary outcomes were (1)
clinical practice. The World Health Or- suggested that dexamethasone may also maternal death, (2) perinatal death, (3)
ganization (WHO) guideline7 states that be associated with a higher rate of fetal death, (4) chronic lung disease (need
there is no conclusive evidence that chorioamnionitis. New additional for continuous supplemental oxygen at 28
would support a recommendation of published trials,12e14 which almost days postnatal age or 36 weeks’ post-
one over the other. It should be noted doubled the previous number of par- menstrual age, whichever was later), (5)
that dexamethasone has an advantage ticipants involved in direct compari- intraventricular hemorrhage (IVH) (diag-
over betamethasone in terms of lower sons, warranted a network meta- nosed by ultrasound, diagnosed by au-
cost and wider availability, and it is analysis (NMA) to urgently define this topsy), (6) mean birthweight (measured in
currently listed on the WHO Essential hot topic. grams), and (7) low birthweight.

2 AJOG MFM MAY 2021

 Systematic Review

Information sources and search Any disagreement at any of the afore- classification, global RoB, and RoB in
strategy mentioned steps was resolved by blinding. To check the assumption of
A literature search strategy was estab- consensus. In the case that consensus coherence (also called network consis-
lished, developed by an experienced could not be reached by 2 reviewers, a tency) in the entire network, the design-
librarian, without language restriction. third author resolved the disagreements by-treatment interaction model was
Considering the exhaustive strategic (I.D.F). used as described by Higgins et al.23 To
searches provided by the previous evaluate the presence of local incoher-
Cochrane reviews,9,10 the evidence gath- Data synthesis and statistical analysis ence, the loop-specific approach was
ered by these reviews up to January 2013 The statistical analyses were conducted used.24,25
was used (the oldest search date reported in accordance with the guidelines A common-within network hetero-
in the study of Brownfoot et al9 was developed by Cochrane.20 geneity standard deviation (swN [0, 1];
February 13, 2013), and searches were A pairwise random-effects meta- s>0) was used, given that many treat-
added from this point to October 2019 in analysis was performed for each available ment comparisons were informed by a
PubMed, MEDLINE, EMBASE, LILACS, direct comparison. Treatment effects single study. After discarding the first
Cochrane Library, ClinicalTrials.gov, and were estimated using odds ratio (OR) for 10,000 iterations, series of 100,000 burn-
the International Clinical Trials Registry dichotomous outcomes and mean dif- in simulations with thinning of 10 values
Platform for ongoing trials search. The ference (MD) for continuous outcomes were used. The model convergence was
MeSH search terms included premature in addition to the 95% confidence in- checked by visual inspection of the
birth, betamethasone, dexamethasone, tervals (CIs). Heterogeneity was quanti- evaluation of the mixing of 2 chains. The
and glucocorticoids (full search strategy fied for all the direct comparisons with assessment of statistical heterogeneity of
in Appendix A). The reference lists of the the I2 statistic, which expresses the per- the entire network was based on the
included studies were also searched, and centage of variability that is because of magnitude of the heterogeneity variance
field experts were contacted for additional the true differences among studies rather parameter (s2) estimated from the NMA
evidence. than sampling error.21 Values of I2 be- models. For dichotomous outcomes, the
tween 30% and 60% were considered to magnitude of the heterogeneity variance
Study selection, data extraction be an indication of “moderate” hetero- was compared with the empirical dis-
assessment of risk of bias geneity. Anticipating clinical and statis- tribution as derived by Turner et al.26
Here, 3 authors (K.K., D.C., and A.C.) tical heterogeneity, a meta-analysis was The prespecified subgroup analyses
independently screened the titles and carried out using a random-effects were gestational age at trial entry (24e28
abstracts and reviewed the full texts of model to synthetize results. Subgroup weeks, 29e34 weeks, and 35e37 weeks),
the potentially eligible studies by using tests were performed to determine dif- intact vs ruptured membranes, and
the Early Review Organizing Software ferences by corticosteroid type. A P<.05 country income level: LMIC vs high-
(Instituto de Efectividad Clínica y Sani- or an I2 >30% was considered suggestive income countries (HIC), based on the
taria [IECS], Buenos Aires, Argentina)17 of subgroup differences. World Bank’s classification.27 The study
and Covidence (Veritas Health Innova- In addition, Bayesian random-effects included sensitivity analysis for the over-
tion, Melbourne, Australia).18 Multiple NMA was performed for each outcome all quality of the studies (low or moderate
publications of the same trial (or sam- to estimate the overall treatment effects, if vs high risk of overall bias) and for the use
ple) were considered as unique studies, the between-study homogeneity, transi- of placebo or mask treatment vs no
and the larger sample sizes were selected. tivity, and coherence assumption across treatment or unmasked intervention. To
The same authors independently, and in treatment comparisons were judged to be see the impact on the results, a fixed-
duplicate, evaluated the risk of bias justifiable.22 The network geometry and effect NMA (in a frequentist frame-
(RoB) domains of included studies using connectivity were explored using network work) was conducted for outcomes with
the Cochrane tool.19 Data were extracted diagrams, and results were presented in rare events and fewer than 5 studies in the
from each of the included studies, and league tables and forest plots. network. A network meta-regression
the information was entered into a data Treatment effects were estimated us- was performed on the basis of gesta-
extraction form designed and piloted for ing OR for dichotomous outcomes and tional age at entry, country GNI per capita
this purpose. A summary RoB was clas- MD for continuous outcomes, along (current international dollar) based on
sified as high risk for a study if at least 1 with the 95% credible intervals (CrI). the World Bank at the beginning of
domain is classified as high risk, whereas The study assessed the validity of the the study (https://data.worldbank.org/
a summary was low or moderate risk if transitivity assumption by comparing country) and the year of publication.
there is no domain classified as high risk. the distribution of potential effect The study assessed small-study effects
Information was obtained regarding modifiers across treatment comparisons, and publication bias that may affect the
publication details, sources of support, including mean gestational age, mean cumulative evidence through compari-
trial methods, characteristics of partici- gross national income (GNI), multiple son adjusted funnel plots for the main 8
pants, intervention, and comparators or single corticosteroid doses, intact or outcomes and when at least 10 studies
and outcomes. ruptured membranes, country income were available per outcome.28

MAY 2021 AJOG MFM 3

Systematic Review

included 45 RCTs12e14,37e78 involving

FIGURE 1
11,227 women and 11,878 infants
Study flowchart (Figure 1). Here, 13 RCTs compared
dexamethasone vs betamethasone (2903
women and 3170 infants), and 32 trials
compared corticosteroids vs placebo or
no treatment (8324 women and 8708
infants). In addition to the 12
trials28,29,31,32,39,44,47,48,54,56,59,62 included
by Brownfoot et al9 and the 30
studies21e27,30,33e38,40e43,45,46,49e53,55,57,
58,60,61
included by Roberts et al,10 this
study included 3 additional studies12e14
and identified 4 references from 3
ongoing trials.6,79e81

Study characteristics
The included trials had heterogeneity
regarding settings, baseline population
characteristics, and intervention schemes
(Table B.2.1 and Table B.2.2 in Appendix
B). The studies were published between
1972 and 2019. A total of 16 studies were
conducted in the United States, 4 studies
were conducted in Iran, and 2 studies
each were conducted in the United
Kingdom, the Netherlands, Finland,
France, Israel, and Brazil. In addition, 37
studies were conducted in HICs, 5 in
upper MICs, and only 3 in lower MICs
(median GNI per capita was $20,170).
RCT, randomized controlled trial.
The sample size varied from 18 to 2831
Ciapponi. Corticosteroids for preterm birth: a network meta-analysis. AJOG MFM 2021.
women (median, 118 women). Mem-
branes were intact, ruptured, and mixed
(both intact and ruptured) in 7, 12, and
Treatments were ranked from best to criteria for NMA, such as potential 26 studies, respectively. Regarding the
worst using the Summary Under the intransitivity (based on the variables recruitment gestational age, 30 RCTs were
Cumulative Ranking (SUCRA) curve considered potential effect modifiers that set at 23 to 28 weeks’ gestation as the
and presented SUCRA values with their were described in the subgroup analysis) lower limit, and 33 studies were set at 34
CrI29,30 across all outcomes in a rank- and potential incoherence (based on the to 37 weeks’ gestation as the upper limit
heat plot.31 All analyses were conducted statistical consistency assessment).34,35 (median gestational age, 30.44 weeks).
in OpenBugs (version 3.2.3; OpenBUGS Here, 2 authors (A.C. and I.D.F.) inde- The most common doses used were
Foundation, http://openbugs.net/w/ pendently graded the certainty of the ev- 24 mg of dexamethasone and 24 mg of
FrontPage).32 For secondary outcomes idence, and differences were resolved by betamethasone in different regimens
with scarce direct evidence, pairwise meta- consensus. The Cochrane recommenda- (full description schemes in Table B.2
analyses were performed using RevMan tions were followed for reporting results and a descriptive summary in the foot-
(version 5.3; The Cochrane Collaboration, according to the certainty of evidence and note of the table).
Copenhagen, Denmark).33 As a sensitivity the magnitude of effect.36 In addition, a Table B.3 describes the full character-
analysis, a fixed-effect model was applied focus group was conducted to reflect the istics of the ongoing studies.6,79e81 The
for pairwise meta-analysis to assess the patients’ perspectives in the discussion Antenatal Corticosteroids for Improving
robustness of the findings. (Appendix C). Outcomes in Preterm Newborns (AC-
The confidence in the estimates for TION-I)6 is a parallel, double-blind,
each reported outcome were assessed Results placebo-controlled RCT of antenatal
using the Grading of Recommendations Study selection dexamethasone that recruited 6018
Assessment, Development and Evalua- The search identified 765 records, after women at risk of preterm birth in hos-
tion (GRADE) approach and specific removing duplicates, and ultimately pitals in low-resource countries. The

4 AJOG MFM MAY 2021

 Systematic Review

antenatal corticosteroids in twin late increased the rates of chorioamnionitis betamethasone, dexamethasone has no
preterm neonates (ACTWIN) RCT 79,80 but on one side of the CI could also effect on neonatal death, but the CrI limits
compared betamethasone with placebo reduce it: OR, 1.46 (95% CI, 0.81e2.66). could also be compatible with beneficial
and included 1616 neonates (808 twin On the contrary, betamethasone reduced or detrimental effect: OR, 1.05 (95% CrI,
pregnancies) from South Korea. The the rates of chorioamnionitis: OR, 0.63 0.62e1.84), moderate-certainty evidence.
other ongoing RCT81 compared 150 (95% CI, 0.41e0.95). In addition, the
women at risk of preterm birth from test for subgroup differences by corti- Fetal death (3857 patients, 13 stud-
Nigeria divided to 3 arms, dexametha- costeroid type showed this disparity ies). Dexamethasone reduced the rates of
sone, betamethasone, and placebo. (P¼.010; I2, 84.2%). fetal death, and betamethasone has little
Table B.4 presents the reasons of excluded Because there was a serious incoherence or no effect compared with control: OR,
studies initially included by full text.82e84 between direct and indirect evidences 0.86 (95% CI, 0.32e2.16) and 1.05 (95%
(ratio of OR [ROR], 3.18; 95% CI, CI, 0.58e2.15), respectively.
Risk of bias assessment of included 1.26e8.02) and the differences in the The test for subgroup differences by
studies certainty of the evidence (direct and in- corticosteroid type showed no
The RoB was considered low in 26 direct: moderate), the direct evidence was disparity (P¼.70; I2, 0%). There was
studies (58%) for random sequence considered to be the most reliable esti- no report of direct evidence regarding
generation, 17 (38%) for allocation mation for this outcome. Compared with this outcome. Indirect evidence have
concealment, 22 (49%) for blinding of betamethasone, dexamethasone reduced suggested that compared with beta-
participants and personnel, 15 (33%) for the rates of chorioamnionitis but on one methasone, dexamethasone reduced
blinding of outcome assessment, 25 side of the CrI could also be slightly the rates of fetal death, but the CrI
(56%) for incomplete outcome data, 31 detrimental: OR, 0.70 (95% CrI, limits could also be compatible with
(69%) for selective reporting, 16 (36%) 0.45e1.06), moderate-certainty evidence. large beneficial or detrimental effect:
for other bias, and 21 (47%) for global OR, 0.81 (95% CrI, 0.24e2.41), low-
RoB (Appendix D). Endometritis or puerperal sepsis (4030 certainty evidence.
patients, 10 studies). Dexamethasone
Synthesis of results increased the rates of endometritis or Respiratory distress syndrome (9784 pa-
The composition of the networks and the puerperal sepsis, and betamethasone has tients, 30 studies). Both dexamethasone
direct, indirect, and NMA (mixed) effect little or no effect compared with control: and betamethasone reduced the rates of
estimations for the main 8 outcomes of OR, 1.93 (95% CI, 0.85e34.41) and 0.94 neonatal death compared with control:
the comparison between dexamethasone (95% CI, 0.47e1.87), respectively. The OR, 0.64 (95% CI, 0.47e0.90) and 0.47
and betamethasone are presented in test for subgroup differences by corti- (95% CI, 0.35e0.60), respectively. The
Figure 2, Figure 3, and the Table. In costeroid type suggested disparities test for subgroup differences by corti-
addition, Appendix E presents the direct (P¼.16; I2, 49.8%). costeroid type suggested disparities
comparison between dexamethasone and There was no report of direct evidence (P¼.11; I2, 54.7%).
betamethasone using forest plots, regarding this outcome. Indirect evi- There was no serious incoherence
Appendix F summarizes the findings of dence have suggested that compared found between direct and indirect evi-
the tables of corticosteroid vs control, and with betamethasone, dexamethasone dence (ROR 1.14; 95% CI, 0.71e2.75);
Appendix G displays the pairwise meta- increased the rates of endometritis or therefore, the NMA evidence was
analysis forest plots by type of cortico- puerperal sepsis with respect to beta- considered to be the most reliable esti-
steroid compared with placebo. Because methasone but on one side of the CrI mation. Compared with betamethasone,
there were only 3 nodes for this network could also be protective: OR, 2.04 (95% dexamethasone increased the rates of
and as the focus was on the comparison CrI, 0.72e6.06), low-certainty evidence. RDS, but the CrI limits could also be
between dexamethasone and betametha- compatible with a small protective effect:
sone, the ranking made less sense. How- Neonatal death (8697 patients, 23 stud- OR, 1.38 (95% CrI, 0.96e2.11),
ever, the SUCRA values for the main ies). Both dexamethasone and betame- moderate-certainty evidence.
outcomes are reported in Appendix H. thasone reduced the rates of neonatal
The analysis found no statistically death compared control: OR, 0.60 (95% Neurodevelopmental disability (2628
significant difference for any outcome. CI, 0.37e0.94) and OR, 0.57 (95% CI, patients, 3 studies)
Furthermore, following the Cochrane 0.39e0.80), respectively. The test for No direct evidence was found for beta-
guideline,36 each central estimate were subgroup differences by corticosteroid methasone vs placebo. Dexamethasone
highlighted as the most probable result type showed no disparity (P¼.81; I2, 0%). may reduce the rates of neuro-
while describing the CI as shown below. There was no incoherence found between developmental disability compared with
direct and indirect evidence (ROR, 1.15; control: OR, 0.39 (95% CI, 0.01e8.08).
Chorioamnionitis (6698 patients, 15 95% CI, 0.44e2.96); therefore, the NMA Compared with betamethasone, dexa-
studies). Compared with placebo or no evidence was considered to be the most methasone has no effect on neuro-
treatment (control), dexamethasone reliable estimation. Compared with developmental disability, but the CrI

MAY 2021 AJOG MFM 5

Systematic Review

FIGURE 2
Network composition by outcome

IVH, intraventricular hemorrhage; RDS, respiratory distress syndrome.

Ciapponi. Corticosteroids for preterm birth: a network meta-analysis. AJOG MFM 2021.

6 AJOG MFM MAY 2021

 Systematic Review

FIGURE 3
Network and direct forest plot for dexamethasone vs betamethasone

IVH, intraventricular hemorrhage; NDD, neurodevelopmental disability; RDS, respiratory distress syndrome.
Ciapponi. Corticosteroids for preterm birth: a network meta-analysis. AJOG MFM 2021.

limits could also be compatible with betamethasone in a frequentist meta- IVH: OR, 0.473 (95% CI, 0.281e0.738)
large beneficial or detrimental effect: analytical framework estimated and 0.381; (95% CI, 0.191e0.668),
OR, 1.14 (95% CrI, 0.24e13.86). s2¼0.00, which was equivalent to a respectively. The test for subgroup dif-
Notably, 2 of the included studies had fixed-effect meta-analysis model. The ferences by corticosteroid type showed
rare events, and a random-effects model frequentist analysis suggested a more no disparity (P¼.88; I2, 0%).
added extra variability to the already precise and reliable estimation (OR, There was no serious incoherence
uncertain treatment effect estimates. 1.03; 95% CI, 0.80e1.33; moderate- found between direct and indirect evi-
Allowing for the same between-study certainty evidence). dence (ROR, 1.54; 95% CI, 0.57e4.16).
variance across comparisons (s2¼0.45; The NMA evidence suggested that
95% CrI, 0.00e4.88) increased CrIs. A Intraventricular hemorrhage (7449 pa- compared with betamethasone, dexa-
sensitivity analysis estimating the treat- tients, 17 studies). Both dexamethasone methasone reduces the rates of IVH, but
ment effect for dexamethasone vs and betamethasone reduced the rates of the CI limits could also be compatible

MAY 2021 AJOG MFM 7

 Systematic Review
8 AJOG MFM MAY 2021

TABLE
Summary of finding: dexamethasone vs betamethasone
Direct evidence Indirect evidence NMA evidence Absolute effects for NMA estimate
Risk with Risk with
Outcome OR (95% CrI)a Certaintyb OR (95% CrI)a Certaintyb OR (95% CrI)a Certainty controlc betac Difference (95% CI)
Chorioamnionitis 0.695 (0.451e1.055) 444 d
2.321 (1.152e4.943) 444 d
0.149 (0.003e0.878) 44 e
59 per 82 per 23 fewer per 1000
moderate moderate low 1000 1000 (43 fewer or 5 more)
Endometritis — — 2.043 (0.715e6.058) 44d,f 2.043 (0.715e6.058) 44d — — 73 more per 1000 (22
or puerperal sepsis low low fewer or 270 more)
Neonatal death 0.941 (0.550e1.646) 444d 1.078 (0.629e1.848) 4d,g 1.053 (0.619e1.836) 444d 270 per 32 per 2 more per 1000 (12
moderate very low moderate 1000 1000 fewer or 25 more)
Fetal death — — 0.812 (0.241e2.406) 44d,f 0.812 (0.241e2.406) 44d — — 6 fewer per 1000 (24
low low fewer or 42 more)
Respiratory distress 1.040 (0.818e1.323) 444d 1.449 (0.990e2.128) 4d,g 1.377 (0.959e2.114) 444d 326 per 260 per 66 more per 1000 (8
Syndrome moderate very low moderate 1000 1000 fewer or 166 more)
Neurodevelopmental 1.031 (0.799e1.329) 444h — — — — 295 per 289 per 6 more per 1000 (44
disability moderate 1000 1000 fewer or 62 more)
Intraventricular 0.706 (0.297e1.581) 4d,i 1.128 (0.509e2.497) 44d,f 1.036 (0.555e1.783) 44j 53 per 52 per 2 more per 1000 (22
hemorrhage very low low low 1000 1000 fewer or 37 more)
Mean MD þ17.81 4444 MD 0.008 4444 MD þ5.29 4444 — — þ5.29 g(49.79
birthweight (g) (62.98 to 98.18) high (0.072 to 0.056) high (49.79 to 58.97) high or þ58.97 g)
GRADE working group grades of evidence are as follows:
High certainty: very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: limited confidence in the effect estimate: the true effect may be substantially different from the estimate of the effect.
Very low certainty: very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.
The shadowed cells are more reliable estimations that were used to calculate the absolute differences presented in the last column.
CI, confidence interval; CrI, credible interval; GRADE, Grading of Recommendations Assessment, Development and Evaluation; MD, mean difference; NMA, network meta-analysis; OR, odds ratio.
a
MD for mean birthweight; b Presented here is the certainty of evidence considering imprecision, but this domain was omitted to directly assess imprecision for the NMA estimation35; c Estimated from direct comparisons and if not available from most similar
outcome (ie, for endometritis, the risks of chorioamnionitis were used); d Downgraded 1 level because of serious imprecision; e Downgraded 2 levels, because of very serious incoherence; f The lowest certainty of evidence (moderate) comes from the
dexamethasone vs control; thus, the final indirect assessment in the absence of intransitivity was moderate; g The lowest certainty of evidence (low) comes from the dexamethasone vs control; thus, the final indirect assessment in the absence of intransitivity was
low; h Downgraded 1 level because of serious imprecision; i Downgraded 2 levels because of very serious methodological limitations (high risk of attrition bias); j Both direct and indirect evidence showed moderate certainty of evidence without considering
imprecision and was downgraded 1 level because of serious imprecision for the NMA estimation.
Ciapponi. Corticosteroids for preterm birth: a network meta-analysis. AJOG MFM 2021.
 Systematic Review

with beneficial or detrimental effect: OR, 0.62 (95% CI, 0.33e1.18) and 0.66 (95% betamethasone were chorioamnionitis
0.812 (95% CrI, 0.420e1.427), low- CI, 0.48e0.91), respectively. The test for in studies only including women with
certainty evidence. However, there was subgroup differences by corticosteroid ruptured membranes; endometritis or
mild heterogeneity (I2, 31%) and type showed no disparity (P¼.86; I2, puerperal sepsis in studies with multiple
important subgroup differences by 0%). Dexamethasone increased the rates doses, with intact or ruptured mem-
corticosteroid type (I2, 63.5%). of chronic lung disease, and betametha- branes or only ruptured membranes, in
Considering the very high risk of sone reduced this outcome, but the CI HICs or adjusted by GNI per capita or
attrition bias (43% of nonanalyzed in- limits could also be compatible with for missing data; and RDS in the sub-
fants), the unique marked effect favoring large beneficial or detrimental effect: group of upperemiddle income
dexamethasone of the study by Elimian OR, 1.30 (95% CI, 0.57e2.96) and 0.75 countries.
et al,48 and the incoherence that this study (95% CI, 0.22e2.62), respectively. The
had generated particularly before the in- test for subgroup differences by corti- Comment
clusion of A Study to Evaluate the Effect costeroid type showed no disparity Principal findings
of Rosuvastatin on Intravascular (P¼.47; I2, 0%). Both corticosteroids have been proven
Ultrasound-Derived Coronary Atheroma There was no direct evidence found effective for women at risk of preterm
Burden (ASTEROID) trial,12 a posthoc that compared dexamethasone with birth on most neonatal and child-
sensitivity analysis was performed while betamethasone regarding maternal relevant outcomes compared with pla-
excluding it and rerunning the analyses. death and perinatal death; scarce evi- cebo or no treatment. As expected, there
Both the heterogeneity and the test for dence was found about low birthweight was no effect on birthweight found
subgroup differences changed to an I2 of and chronic lung disease (need for considering the short time frame be-
0%. The new estimation, still low- continuous supplemental oxygen at 28 tween the intervention and delivery.
certainty evidence, OR, 1.04 (95% CrI, days postnatal age or 36 weeks’ post- No statistically significant effect was
0.56e1.78), was more consistent with the menstrual age, whichever was later); and found suggesting that compared with
indirect evidence (ROR was reduced to these evidence were assessed only by betamethasone, dexamethasone reduced
1.14 [95% CI, 0.51e2.57]), and therefore, pairwise meta-analysis. the rates of chorioamnionitis by 30%
it was considered as the most reliable One study (105 participants) assessed and fetal death by 20% but increased
estimation (Appendix E). low birthweight, and another study rates of puerperal sepsis by 100% and
(1489 participants) assessed chronic RDS by 40%. There was no difference in
Mean birthweight (8645 patients, 23 lung disease. Compared with betame- neonatal death and neurodevelopmental
studies). Both dexamethasone and beta- thasone, dexamethasone may reduce disability and no difference in IVH and
methasone have no effect on birthweight these outcomes; however, the CI limits in birthweight. Except for neuro-
compared control: MD, 17.04 g (95% could also be compatible with large developmental disability and birth-
CI, 75.48 to 41.41) and 9.74 g (95% beneficial or detrimental effect: OR, 0.75 weight, these effects were imprecise.
CI, 43.11 to 23.63), respectively. The (95% CI, 0.33e1.71; low-certainty evi-
test for subgroup differences by corti- dence) and 0.92 (95% CI, 0.62e1.37), Strengths and limitations
costeroid type showed no disparity respectively (Appendix E). Among the strengths of this work, it is
(P¼.80; I2, 0%). No serious incoherence The effect and the test for subgroup worth mentioning that Cochrane
was found between direct and indirect differences by corticosteroid type for the guidelines20 were followed and the
evidence (ROR, 1.15; 95% CI, main 8 outcomes of the comparison PRISMA-NMA extension15 (Appendix
0.44e2.96), and both direct and indirect between corticosteroid and placebo or J), and the study protocol was regis-
evidence were considered as high- no treatment by type of corticosteroid tered in advance. This work is the most
certainty evidence; therefore, the NMA and related certainty of evidence are updated and complete systematic review
evidence was considered to be the most presented in the summary of table find- assessing clinical effectiveness and safety
reliable estimation. Dexamethasone has ings in Appendix F. of corticosteroids for preterm birth. The
no effect on mean birthweight: MD, 5.29 Appendix H presents several NMA exhaustive search strategy used in this
g (95% CrI, 49.79 to 58.97), high- outputs, and Appendix I presents the study, included clinical trials registries
certainty evidence. meta-regression, subgroup, and sensi- and contacted experts for additional
tivity analyses. Meta-regression for the 4 relevant evidence. Although the strategy
Other outcomes. Regarding corticoste- primary outcomes did not find statisti- did not hand-search conference pro-
roids vs control, there was no direct ev- cally significant differences. The sub- ceedings, it is unlikely that the search
idence found about low birthweight, and group and sensitivity analysis did not strategy missed RCTs not included in
there may be no difference between reveal important changes regarding the biomedical databases or the trial
betamethasone and control on maternal main analysis. All the CI limits were registries.
death: OR, 0.98 (95% CI, 0.06e15.90). compatible with beneficial or detri- This NMA added 2 small trials13,14
Dexamethasone and betamethasone mental effects. The ORs of >2.0 for and 1 large, good quality trial that
reduced the rates of perinatal death: OR, dexamethasone compared with compared dexamethasone directly with

MAY 2021 AJOG MFM 9

Systematic Review

betamethasone12 to the body of evi- about which corticosteroids they would phosphate alone as a third arm (finding
dence. In addition, the NMA provided choose because the trade-off between no difference in fetal heart rate); there-
new indirect estimations and increased risk and benefits were very complex. The fore, this study was not able to assess the
the precision of the estimations, still low women agreed that it would be a deci- differential effect of both schemes.
for most outcomes, by combining direct sion that they would share or delegate to
and indirect evidence. The prespecified a professional with whom they estab- Comparison with existing literature
meta-regression, subgroup and sensi- lished a bond of trust. It is noteworthy The new evidence provided by this study
tivity analyses reinforced the robustness that the health risk for women had more does not seem to contradict previous
of the results. importance for those participants who studies. Antenatal corticosteroids
The certainty of the evidence was already had children, because they administered to women who are at risk
assessed by the GRADE-NMA considered the family impact of their of preterm birth have shown to reduce
approach,34,35 the validity of the transi- own health. neonatal morbidity and mortality10 and
tivity assumption by comparing the The evidence showed limitations, are cost-effective88; thus, they are
distribution of potential effect modifiers regarding its generalizability to lower- routinely recommended world-
across comparisons and the coherence resource countries, as only 314,44,49 of wide.7,89,90 The pairwise meta-analyses
assumption by the design-by-treatment 45 included RCTs (only 5.5% of the results of both corticosteroids
interaction model and loop-specific included infants) were from lower MICs compared with placebo agree with the
approaches.23,25 and none from LICs. Trials have been results of the last Cochrane review by
The results of the NMA were mostly largely conducted in tertiary hospitals Roberts et al.10 The 2 additional studies
coherent, except for chorioamnionitis, and recruited highly selected pop- found13,14 for this study only slightly
which may be because of differences ulations.85 Concerns about safety and improved the precision of the results for
between populations included in indi- efficacy in low-resource settings were RDS and birthweight outcomes.
rect and direct evidence and differences supported by the adverse findings in The second Cochrane review by
in RoB. The indirect evidence came neonatal deaths and maternal infection Brownfoot et al9 compared both corti-
mostly from mothers with ruptured of the Antenatal Corticosteroid Trial costeroids and different regimes against
membranes,38,39,43,49,51,53,54,57e59,62,67, (ACT), a community-based, cluster RCT each other, but the authors could not
69,71,74
whereas the direct evidence was conducted in 6 LMICs.86 However, no consider recent evidence. The inclusion
from a mix of mothers with intact and important difference was found by of the new trial ASTEROID12 almost
ruptured membranes.12 However, meta- country income classification or by GNI doubled the number of included women
regression, subgroup, and sensitivity per capita. Hopefully, the ongoing AC- and provided information for the main
analyses did not explain this incoher- TION study in 5 low-resource countries outcomes, except puerperal sepsis, fetal
ence. Therefore, for chorioamnionitis, will answer this question regarding the death, and chronic lung disease.
following the GRADE approach,34 the effects of dexamethasone on 6018 Furthermore, this recent evidence
direct evidence was considered the most women at risk of preterm birth.6 An improved the precision of the previous
reliable estimation of 23 fewer cases (43 additional placebo-controlled ongoing estimates for neurosensory disability,
fewer or 5 more) per 1000 women RCT to be conducted in Nigeria will where this study found no important
treated with dexamethasone. include separate arms for each cortico- difference between corticosteroids,
Generally, the included trials were steroid, allowing a direct comparison in whereas for chorioamnionitis it was
heterogeneous in terms of clinical set- a low-resource setting. The placebo- found that dexamethasone may have a
tings, baseline population characteris- controlled ongoing ACTWIN RCT79,80 beneficial effect.
tics, and intervention schemes and will evaluate the effect of betametha- The potential beneficial effect of
doses. Studies conducted in a range of 50 sone on 808 twin pregnancies in South dexamethasone on IVH, suggested by
years and healthcare advances were Korea. This body of evidence will in- very low-certainty direct evidence, was
included, specifically in neonatology, crease the precision of estimations in a reduced with the inclusion of the
adding an extra source of heterogeneity future update of our NMA. However, ASTEROID trial12 and completely dis-
that could partially explain the contra- future studies are needed to explore the appeared when excluding the trial of
dictory direction of effect for some out- differential effect of doses, because the Elimian et al.48 This posthoc sensitivity
comes, but the effect modifiers or RoB multiple schemes used across the analysis was key mainly because of the
did not provide a solid explanation of the included studies precluded this analysis. very high risk of attrition bias of this
effects. The mixed beneficial or detri- In addition, many studies used a com- study and provided more consistent re-
mental effect of different outcomes bination of betamethasone phosphate sults with the indirect evidence. In
warranted the decision to explore the and acetate to maximize the drug’s effi- addition, systematic reviews and meta-
patients’ perspectives about the analyses’ ciency. Both formulations have different analyses found an increased risk of
findings comparing corticosteroids pharmacokinetic and pharmacodynamic neurodevelopmental impairment in
through a focus group (Appendix C). characteristics,87 but only Subtil et al75 children with periventricular or intra-
Briefly, women failed to make a decision compared acetate and phosphate with ventricular hemorrhage (PIVH),91

10 AJOG MFM MAY 2021

 Systematic Review

mainly driven by cerebral palsy.92 As cesarean delivery with betamethasone large, well-designed RCTs are warranted
there was no differential effect of dexa- observed, but it was not sustained in an to improve the certainty of this evidence.
methasone on neurosensory disability to adjusted posthoc analyses. Ideally, these trials should represent low-
be found, it would be unlikely a potential A wise choice between dexamethasone resource countries and address the best
favorable effect on IVH. In addition, and betamethasone should consider all schemes for administration in different
even if a reduction in IVH was true, the factors besides evidence, including local subgroups. Individual participant data
observed absence of differences on long- availability, costs, and cost-utility.96,97 A meta-analysis could help to answer these
term disability for the quality of life of full course of betamethasone costs questions. In the meantime, monitoring
survivors is more important.93 around $35, whereas dexamethasone short-term and long-term health out-
Roberts and Dalziel11 assessed indirect costs $1 (3% of the cost of betametha- comes, including neurodevelopmental
estimations favoring betamethasone for sone).97 The cost-effectiveness of the disability, will be important to obtain
chorioamnionitis. This was consistent administration of betamethasone in the real-life data.
with the indirect estimation but opposite late preterm period is controversial, and Because there is no robust evidence on
to the ASTEROID trial12 findings that it should be based on the best estimation which corticosteroid should be pre-
were considered the most reliable esti- of effectiveness instead of individual scribed, decisions should be based on
mation for this outcome. trials.98,99 Mainly LMICs still have availability, costs, opportunity, and fa-
This NMA improved the precision marked challenges to provide safe and cilities. Shared decision-making would
and certainty of most previous estima- effective antenatal corticosteroid use, help patients to make their choices when
tions. The study identified a very recent including ensuring accurate gestational facing this scenario. -
NMA that evaluated antenatal maternal age determination, establishing clear
administration of dexamethasone and treatment guidelines, strengthening ACKNOWLEDGMENTS
betamethasone and ambroxol.94 Authors provider capacity, incorporating corti- We acknowledge Daniel Comandé for his sup-
reported that compared with placebo, all costeroid in national essential medicines port as librarian and Ioannis Gallos and Argyro
interventions demonstrated better effi- lists, and monitoring use and out- Papadopoulou for their exhaustive revision of the
cacy in terms of preventing RDS and comes.100 Hopefully, the findings of this manuscript. In addition, we acknowledge Mss
María Belizán and Natali Ini for coordinating and
neonatal death and no significant dif- study contribute to an informed
analyzing the focus group that explored the
ference in the assessment of the inci- decision-making process and to improve women’s perspectives of our findings.
dence of bronchopulmonary dysplasia. maternal and newborn care.
In addition, authors suggest that
ambroxol seems to have the potential to Conclusions and implications REFERENCES
be the most effective treatment for This comprehensive NMA showed that 1. Chawanpaiboon S, Vogel JP, Moller AB, et al.
reducing the incidence of RDS and corticosteroids were mostly effective for Global, regional, and national estimates of levels
neonatal death based on its SUCRA neonatal and child-relevant outcomes of preterm birth in 2014: a systematic review and
modelling analysis. Lancet Glob Health 2019;7:
values. This conclusion was not consis- compared with placebo or no treatment. e37–46.
tent with a Cochrane systematic review There was no clinical or statistical dif- 2. Platt MJ. Outcomes in preterm infants. Public
showing insufficient evidence to support ference between corticosteroids on Health 2014;128:399–403.
or refute the practice of giving ambroxol neonatal death, neurodevelopmental 3. United Nations Department of Economic and
to women at risk of preterm birth.95 disability, intraventricular hemorrhage, Social Affairs, United Nations Children’s Fund,
World Health Organization, World Bank Group.
Furthermore, ambroxol is not even or birthweight, but both corticosteroids Levels and trends in child mortality: 2017 report.
mentioned in relevant guidelines for the have different effects on the remaining 2017. Available at: https://www.un.org/en/
management of preterm birth.7,89 outcomes. There was no statistically development/desa/population/publications/
Unlike this study and the previous significant difference, but a potentially mortality/child-mortality-report-2017.asp.
systematic review, the authors included clinically important effect was found Accessed December 20, 2020.
4. Glass HC, Costarino AT, Stayer SA,
studies with repeated exposure to ante- between dexamethasone and betame- Brett CM, Cladis F, Davis PJ. Outcomes for
natal corticosteroids, and owing to the thasone. Considering the central esti- extremely premature infants. Anesth Analg
search date, relevant recent trials could mations, low- to moderate-certainty 2015;120:1337–51.
not be included. The largest and most evidence suggest that dexamethasone 5. Crowley P. Prophylactic corticosteroids for
preterm birth. Cochrane Database Syst Rev
recent trial12 reported that children may reduce chorioamnionitis and fetal
2000:CD000065.
exposed to dexamethasone were less death but may increase puerperal sepsis 6. WHO ACTION Trials Collaborators. The
likely to be hypertensive at the age of 2 and RDS. However, the 95% CI indi- World Health Organization ACTION-I (Antenatal
years than those exposed to betametha- cated both beneficial and detrimental corticosteroids for Improving Outcomes in pre-
sone (adjusted OR, 0.78; 95% CI, effects against betamethasone for these term Newborns) Trial: a multi-country, multi-
0.64e0.95), but there was incomplete outcomes. The opposing direction of centre, two-arm, parallel, double-blind, placebo-
controlled, individually randomized trial of ante-
outcome data and the clinical signifi- these outcomes does not allow for deri- natal corticosteroids for women at risk of immi-
cance of this finding is still uncertain. In vation of recommendations about what nent birth in the early preterm period in hospitals
addition, there was higher use of corticosteroid should be used. A few very in low-resource countries. Trials 2019;20:507.

MAY 2021 AJOG MFM 11

Systematic Review

7. World Health Organization. WHO recom- www.training.cochrane.org/handbook 2011. rating the quality of treatment effect estimates
mendations on interventions to improve preterm Accessed December 20, 2020. from network meta-analysis. BMJ 2014;349:
birth outcomes. 2015. Available at: https:// 20. Chaimani A, Caldwell DM, Li T, Higgins JP, g5630.
www.who.int/reproductivehealth/publications/ Salanti G. Chapter 11. Undertaking network 35. Brignardello-Petersen R, Bonner A,
maternal_perinatal_health/preterm-birth- meta-analyses. In: Higgins J, Thomas J, Alexander PE, et al. Advances in the GRADE
guideline/en/ 2015. Accessed December 20, Chandler J, et al, eds. Cochrane handbook for approach to rate the certainty in estimates from
2020. systematic reviews of interventions version 6. a network meta-analysis. J Clin Epidemiol
8. World Health Organization, United Nations Chichester, United Kingdom: John Wiley & Sons; 2018;93:36–44.
Population Fund, United Nations Children’s 2016. 36. Schünemann H, Vist G, Higgins J, et al.
Fund, The World Bank. Managing complications 21. Higgins JP, Thompson SG. Quantifying Interpreting results and drawing conclusions. In:
in pregnancy and childbirth: a guide for midwives heterogeneity in a meta-analysis. Stat Med Higgins J, Thomas J, Chandler J, et al, eds.
and doctors. 2017. Available at: https:// 2002;21:1539–58. Cochrane handbook for systematic reviews of
www.mcsprogram.org/resource/managing- 22. Baker SG, Kramer BS. The transitive fallacy interventions version 6.0. Chichester, United
complications-pregnancy-childbirth-guide- for randomized trials: if A bests B and B bests C Kingdom: John Wiley & Sons; 2019.
midwives-doctors/ 2017. Accessed in separate trials, is A better than C? BMC Med 37. Effect of antenatal dexamethasone admin-
December 20, 2020. Res Methodol 2002;2:13. istration on the prevention of respiratory distress
9. Brownfoot FC, Gagliardi DI, Bain E, 23. Higgins JP, Jackson D, Barrett JK, Lu G, syndrome. Am J Obstet Gynecol 1981;141:
Middleton P, Crowther CA. Different cortico- Ades AE, White IR. Consistency and inconsis- 276–87.
steroids and regimens for accelerating fetal lung tency in network meta-analysis: concepts and 38. Amorim MMR, Santos LC, Faúndes A.
maturation for women at risk of preterm birth. models for multi-arm studies. Res Synth Corticosteroid therapy for prevention of respi-
Cochrane Database Syst Rev 2013:CD006764. Methods 2012;3:98–110. ratory distress syndrome in severe preeclamp-
10. Roberts D, Brown J, Medley N, Dalziel SR. 24. Veroniki AA, Vasiliadis HS, Higgins JP, sia. Am J Obstet Gynecol 1999;180:1283–8.
Antenatal corticosteroids for accelerating fetal Salanti G. Evaluation of inconsistency in net- 39. Attawattanakul N, Tansupswatdikul P. Ef-
lung maturation for women at risk of preterm works of interventions. Int J Epidemiol 2013;42: fects of antenatal dexamethasone on respiratory
birth. Cochrane Database Syst Rev 2017;3: 332–45. distress in late preterm infant: a randomized
CD004454. 25. Dias S, Welton NJ, Caldwell DM, Ades AE. controlled trial. J Obstet Gynaecol 2015;23:
11. Roberts D, Dalziel S. Antenatal corticoste- Checking consistency in mixed treatment com- 25–33.
roids for accelerating fetal lung maturation for parison meta-analysis. Stat Med 2010;29: 40. Balci O, Ozdemir S, Mahmoud AS, Acar A,
women at risk of preterm birth. Cochrane 932–44. Colakoglu MC. The effect of antenatal steroids
Database Syst Rev 2006:CD004454. 26. Turner RM, Davey J, Clarke MJ, on fetal lung maturation between the 34th and
12. Crowther CA, Ashwood P, Andersen CC, Thompson SG, Higgins JP. Predicting the extent 36th week of pregnancy. Gynecol Obstet Invest
et al. Maternal intramuscular dexamethasone of heterogeneity in meta-analysis, using empir- 2010;70:95–9.
versus betamethasone before preterm birth ical data from the Cochrane Database of Sys- 41. Block MF, Kling OR, Crosby WM. Antenatal
(ASTEROID): a multicentre, double-blind, rand- tematic Reviews. Int J Epidemiol 2012;41: glucocorticoid therapy for the prevention of
omised controlled trial. Lancet Child Adolesc 818–27. respiratory distress syndrome in the premature
Health 2019;3:769–80. 27. The World Bank. World Bank country and infant. Obstet Gynecol 1977;50:186–90.
13. Mirzamoradi M, Hasani Nejhad F, Jamali R, lending groups. 2019, https://datahelpdesk. 42. Cararach V, Botet F, Sentis J,
Heidar Z, Bakhtiyari M. Evaluation of the effect of worldbank.org/knowledgebase/articles/906519- Carmona F. A multicenter, prospective and
antenatal betamethasone on neonatal respira- world-bank-country-and-lending-groups 2019. randomized study in premature rupture of
tory morbidities in late preterm deliveries (34-37 Accessed December 20, 2020. membranes (PROM). Maternal and perinatal
weeks). J Matern Fetal Neonatal Med 2020;33: 28. Chaimani A, Higgins JP, Mavridis D, complications. Int J Gynaecol Obstet
2533–40. Spyridonos P, Salanti G. Graphical tools for 1991;36(Suppl):267–9.
14. Ontela V, Dorairajan G, Bhat VB, network meta-analysis in STATA. PLoS One 43. Carlan SJ, Parsons M, O’Brien WF,
Chinnakali P. Effect of antenatal steroids on 2013;8:e76654. Krammer J. Pharmacologic pulmonary matura-
respiratory morbidity of late preterm newborns: 29. Trinquart L, Attiche N, Bafeta A, Porcher R, tion in preterm premature rupture of mem-
a randomized controlled trial. J Trop Pediatr Ravaud P. Uncertainty in treatment rankings: branes. Am J Obstet Gynecol 1991;164:371.
2018;64:531–8. reanalysis of network meta-analyses of random- 44. Chen CY, Wang KG, Chang TY, Chen CP,
15. Hutton B, Salanti G, Caldwell DM, et al. The ized trials. Ann Intern Med 2016;164:666–73. Loo JH. Effects of antenatal betamethasone and
PRISMA extension statement for reporting of 30. Veroniki AA, Straus SE, Rücker G, dexamethasone in preterm neonates. Taiwan J
systematic reviews incorporating network meta- Tricco AC. Is providing uncertainty intervals in Obstet Gynecol 2005;44:247–51.
analyses of health care interventions: checklist treatment ranking helpful in a network meta- 45. Danesh A, Janghorbani M, Khalatbari S.
and explanations. Ann Intern Med 2015;162: analysis? J Clin Epidemiol 2018;100:122–9. Effects of antenatal corticosteroids on maternal
777–84. 31. Veroniki AA, Straus SE, Fyraridis A, serum indicators of infection in women at risk for
16. Binder LM, Iverson GL, Brooks BL. To err is Tricco AC. The rank-heat plot is a novel way to preterm delivery: a randomized trial comparing
human: “abnormal” neuropsychological scores present the results from a network meta-analysis betamethasone and dexamethasone. J Res
and variability are common in healthy adults. including multiple outcomes. J Clin Epidemiol Med Sci 2012;17:911–7.
Arch Clin Neuropsychol 2009;24:31–46. 2016;76:193–9. 46. Doran TA, Swyer P, MacMurray B, et al.
17. Ciapponi A, Glujovsky D, Bardach A, García 32. Lunn DJ, Thomas A, Best N, Results of a double-blind controlled study on the
Martí S, Comande D. EROS: a new software for Spiegelhalter D. WinBUGS - a Bayesian use of betamethasone in the prevention of res-
early stage of systematic reviews 2011;14: modelling framework: concepts, structure, and piratory distress syndrome. Am J Obstet
PA564. extensibility. Stat Comput 2000;10:325–37. Gynecol 1980;136:313–20.
18. Covidence systematic review software. 33. Review Manager (RevMan) [Computer pro- 47. Egerman RS, Walker RA, Mercer BM,
Melbourne, Commonwealth of Australia: Veritas gram] version 53. Copenhagen: The Nordic Doss JL, Sibai BM, Andersen RA. Comparison
Health Innovation. Cochrane Centre, The Cochrane Collaboration; between oral and intramuscular dexamethasone
19. Higgins J, Green S, Chandler J, et al. 2014. in suppressing unconjugated estriol levels during
Cochrane handbook for systematic reviews of 34. Puhan MA, Schünemann HJ, Murad MH, the third trimester. Am J Obstet Gynecol
interventions version 5.1. 2011. Available at: et al. A GRADE Working Group approach for 1998;179:1234–6.

12 AJOG MFM MAY 2021

 Systematic Review

48. Elimian A, Garry D, Figueroa R, Spitzer A, dexamethasone: effects on the antenatal fetal 73. Shanks A, Gross G, Shim T, Allsworth J,
Wiencek V, Quirk JG. Antenatal betamethasone heart rate. Br J Obstet Gynaecol 1997;104: Sadovsky Y, Bildirici I. Administration of steroids
compared with dexamethasone (betacode trial): 1233–8. after 34 weeks of gestation enhances fetal lung
a randomized controlled trial. Obstet Gynecol 61. Mansouri M, Seyedolshohadaei F, maturity profiles. Am J Obstet Gynecol
2007;110:26–30. Company F, Setare S, Mazhari S. Effect of 2010;203:47.e1–5.
49. Fekih M, Chaieb A, Sboui H, Denguezli W, antenatal betamethasone on prevention of res- 74. Silver RK, Vyskocil C, Solomon SL, Ragin A,
Hidar S, Khairi H. [Value of prenatal cortico- piratory distress syndrome among neonates Neerhof MG, Farrell EE. Randomized trial of
therapy in the prevention of hyaline membrane with gestational age of 35-36 weeks. J Gorgan antenatal dexamethasone in surfactant-treated
disease in premature infants. Randomized pro- Univ Med Sci 2010;12:18–23. infants delivered before 30 weeks’ gestation.
spective study]. Tunis Med 2002;80:260–5. 62. Morales WJ, Angel JL, O’Brien WF, Obstet Gynecol 1996;87:683–91.
50. Gamsu HR, Mullinger BM, Donnai P, Knuppel RA. Use of ampicillin and corticosteroids 75. Subtil D, Tiberghien P, Devos P, et al. Im-
Dash CH. Antenatal administration of betame- in premature rupture of membranes: a random- mediate and delayed effects of antenatal corti-
thasone to prevent respiratory distress syn- ized study. Obstet Gynecol 1989;73:721–6. costeroids on fetal heart rate: a randomized trial
drome in preterm infants: report of a UK 63. Mulder EJ, Derks JB, Visser GH. Antenatal that compares betamethasone acetate and
multicentre trial. Br J Obstet Gynaecol 1989;96: corticosteroid therapy and fetal behaviour: a phosphate, betamethasone phosphate, and
401–10. randomised study of the effects of betametha- dexamethasone. Am J Obstet Gynecol
51. Garite TJ, Rumney PJ, Briggs GG, et al. sone and dexamethasone. Br J Obstet Gynae- 2003;188:524–31.
A randomized, placebo-controlled trial of beta- col 1997;104:1239–47. 76. Taeusch HW, Frigoletto F, Kitzmiller J, et al.
methasone for the prevention of respiratory 64. Mushkat Y, Ascher-Landsberg J, Keidar R, Risk of respiratory distress syndrome after pre-
distress syndrome at 24 to 28 weeks’ gestation. Carmon E, Pauzner D, David MP. The effect of natal dexamethasone treatment. Pediatrics
Am J Obstet Gynecol 1992;166:646–51. betamethasone versus dexamethasone on fetal 1979;63:64–72.
52. Sakamoto S, Tojo S, Nakayama T. Gyne- biophysical parameters. Eur J Obstet Gynecol 77. Teramo K, Hallman M, Raivio KO. Maternal
cology and obstetrics: proceedings of the IX Reprod Biol 2001;97:50–2. glucocorticoid in unplanned premature labor.
World Congress of Gynecology and Obstetrics, 65. Nelson LH, Meis PJ, Hatjis CG, Ernest JM, Controlled study on the effects of betametha-
Tokyo, October 25-31, 1979/Editors, Shoichi Dillard R, Schey HM. Premature rupture of sone phosphate on the phospholipids of the
Sakamoto, Shimpei Tojo, and Tetsuya membranes: a prospective, randomized evalu- gastric aspirate and on the adrenal cortical
Nakayama. Amsterdam: Excerpta Medica; ation of steroids, latent phase, and expectant function of the newborn infant. Pediatr Res
1980. Print. Available at: https://catalog.library. management. Obstet Gynecol 1985;66:55–8. 1980;14:326–9.
vanderbilt.edu/permalink/01VAN_INST/13em2a7/ 66. 8th Annual Meeting of the Society of Peri- 78. Urban R, Lemancewicz A, Przepiesc J,
alma991023098319703276. Accessed natal Obstetricians January 1988. Available at: Urban J, Kretowska M. Antenatal corticosteroid
December 20, 2019. https://els-jbs-prod-cdn.jbs.elsevierhealth.com/ therapy: a comparative study of dexamethasone
53. Gyamfi-Bannerman C, Thom EA, pb/assets/raw/Health%20Advance/journals/ and betamethasone effects on fetal Doppler flow
Blackwell SC, et al. Antenatal betamethasone ymob/1988_SMFM_Abstracts.pdf. Accessed velocity waveforms. Eur J Obstet Gynecol
for women at risk for late preterm delivery. N Engl December 20, 2019. Reprod Biol 2005;120:170–4.
J Med 2016;374:1311–20. 67. Pattinson RC, Makin JD, Funk M, et al. The 79. Hong S, Lee SM, Kwak DW, et al. Ef-
54. Kari MA, Hallman M, Eronen M, et al. Pre- use of dexamethasone in women with preterm fects of antenatal corticosteroids in twin ne-
natal dexamethasone treatment in conjunction premature rupture of membranes—a multi- onates with late preterm birth (ACTWIN
with rescue therapy of human surfactant: a centre, double-blind, placebo-controlled, rand- [Antenatal corticosteroids in TWIN late pre-
randomized placebo-controlled multicenter omised trial. Dexiprom Study Group. S Afr Med J term neonates] trial): study protocol for a
study. Pediatrics 1994;93:730–6. 1999;89:865–70. randomized controlled trial. BMC Pregnancy
55. Khandelwal M, Chang E, Hansen C, 68. Porto AMF, Coutinho IC, Correia JB, Childbirth 2019;19:114.
Hunter K, Milcarek B. Betamethasone dosing Amorim MMR. Effectiveness of antenatal corti- 80. Seoul National University Hospital. Effects of
interval: 12 or 24 hours apart? A randomized, costeroids in reducing respiratory disorders in ACS in twin with LPB: study protocol for a RCT.
noninferiority open trial. Am J Obstet Gynecol late preterm infants: randomised clinical trial. 2018. Available at: https://clinicaltrials.gov/ct2/
2012;206:201.e1–11. BMJ 2011;342:d1696. show/NCT03547791 2018. Accessed
56. Khazardoust S, Javadian P, 69. Qublan HS, Malkawi HY, Hiasat MS, et al. November 18, 2019.
Salmanian B, et al. A clinical randomized trial The effect of antenatal corticosteroid therapy on 81. Ahmadu Bello University Teaching Hospital.
on endocervical inflammatory cytokines and pregnancies complicated by premature rupture Effect of antenatal corticosteroids on neonatal
betamethasone in prime-gravid pregnant of membranes. Clin Exp Obstet Gynecol morbidity. 2018. Available at: https://
women at risk of preterm labor. Iran J 2001;28:183–6. clinicaltrials.gov/ct2/show/NCT03446937
Immunol 2012;9:199–207. 70. Rotmensch S, Liberati M, Vishne TH, 2018. Accessed December 20, 2019.
57. Lewis DF, Brody K, Edwards MS, Celentano C, Ben-Rafael Z, Bellati U. The effect 82. Cartwright R, Crowther C, Anderson P,
Brouillette RM, Burlison S, London SN. Preterm of betamethasone and dexamethasone on fetal Harding J, Doyle L, McKinlay C. Influence of fetal
premature ruptured membranes: a randomized heart rate patterns and biophysical activities. A growth restriction on neurocognitive function
trial of steroids after treatment with antibiotics. prospective randomized trial. Acta Obstet after repeat antenatal betamethasone: second-
Obstet Gynecol 1996;88:801–5. Gynecol Scand 1999;78:493–500. ary analysis of a randomised trial. J Paediatr
58. Liggins GC, Howie RN. A controlled trial of 71. Schutte MF, Treffers PE, Koppe JG, Child Health 2019;55:12–3.
antepartum glucocorticoid treatment for pre- Breur W. The influence of betamethasone and 83. Cartwright RD, Crowther CA,
vention of the respiratory distress syndrome in orciprenaline on the incidence of respiratory Anderson PJ, Harding JE, Doyle LW,
premature infants. Pediatrics 1972;50:515–25. distress syndrome in the newborn after preterm McKinlay CJD. Association of fetal growth
59. Lopez L, Rojas L, Rodriguez V, Sanchez J. labour. Br J Obstet Gynaecol 1980;87:127–31. restriction with neurocognitive function after
Use of corticoids in preterm pregnancy with 72. Senat MV, Minoui S, Multon O, repeated antenatal betamethasone treatment
premature rupture of membranes. Rev Colomb Fernandez H, Frydman R, Ville Y. Effect of vs placebo: secondary analysis of the
Obstet Ginecol 1989;40:147–51. dexamethasone and betamethasone on fetal ACTORDS randomized clinical trial. JAMA
60. Magee LA, Dawes GS, Moulden M, heart rate variability in preterm labour: a rando- Network Open 2019;2:e187636.
Redman CW. A randomised controlled com- mised study. Br J Obstet Gynaecol 1998;105: 84. Schmitz T, Alberti C, Ursino M, Baud O,
parison of betamethasone with 749–55. Aupiais C; BETADOSE Study Group and the

MAY 2021 AJOG MFM 13

Systematic Review

GROG (Groupe de Recherche en Gynéco- 89. National Institute for Health and Care respiratory distress syndrome. Cochrane Data-
logie Obstétrique). Full versus half dose of Excellence. NICE guideline. Preterm labour and base Syst Rev 2014:CD009708.
antenatal betamethasone to prevent severe birth. 2015. Available at: https://www.nice.org. 96. Vogel JP, Souza JP, Gülmezoglu AM, et al.
neonatal respiratory distress syndrome uk/guidance/ng25/resources/preterm-labour- Use of antenatal corticosteroids and tocolytic
associated with preterm birth: study proto- and-birth-pdf-1837333576645 2015. drugs in preterm births in 29 countries: an anal-
col for a randomised, multicenter, double Accessed December 22, 2020. ysis of the WHO Multicountry Survey on Maternal
blind, placebo-controlled, non-inferiority trial 90. Committee on Obstetric Practice. Commit- and newborn Health. Lancet 2014;384:1869–77.
(BETADOSE). BMC Pregnancy Childbirth tee Opinion no. 713: antenatal corticosteroid 97. United Nations Commission, Born Too
2019;19:67. therapy for fetal maturation. Obstet Gynecol Soon Care Antenatal Corticosteroids Working
85. Vogel JP, Oladapo OT, Pileggi-Castro C, 2017;130:e102–9. Group. Dexamethasone versus betamethasone
et al. Antenatal corticosteroids for women at risk 91. Mukerji A, Shah V, Shah PS. Periventricular/ as an antenatal corticosteroid (ACS). 2013.
of imminent preterm birth in low-resource intraventricular hemorrhage and neuro- Available at: https://www.healthynewborn
countries: the case for equipoise and the need developmental outcomes: a meta-analysis. Pe- network.org/hnn-content/uploads/ACS-Beta-
for efficacy trials. BMJ Glob Health 2017;2: diatrics 2015;136:1132–43. vs-Dexa-130820.pdf 2013. Accessed
e000398. 92. Gotardo JW, Volkmer NFV, Stangler GP, December 15, 2020.
86. Althabe F, Belizán JM, McClure EM, et al. Dornelles AD, Bohrer BBA, Carvalho CG. Impact 98. Gyamfi-Bannerman C, Zupancic JAF,
A population-based, multifaceted strategy to of periintraventricular haemorrhage and peri- Sandoval G, et al. Cost-effectiveness of ante-
implement antenatal corticosteroid treatment ventricular leukomalacia in the neuro- natal corticosteroid therapy vs no therapy in
versus standard care for the reduction of development of preterms: a systematic review women at risk of late preterm delivery: a sec-
neonatal mortality due to preterm birth in low- and meta-analysis. PLoS One 2019;14: ondary analysis of a randomized clinical trial.
income and middle-income countries: the ACT e0223427. JAMA Pediatr 2019;173:462–8.
cluster-randomised trial. Lancet 2015;385: 93. Saigal S, Doyle LW. An overview of mortality 99. Rosenbloom JI, Lewkowitz AK,
629–39. and sequelae of preterm birth from infancy to Sondgeroth KE, et al. Antenatal corticosteroid
87. Jobe AH, Milad MA, Peppard T, adulthood. Lancet 2008;371:261–9. administration in late-preterm gestations: a cost-
Jusko WJ. Pharmacokinetics and pharmaco- 94. Zhang H, Liu J, Liu T, Wang Y, Dai W. effectiveness analysis. J Matern Fetal Neonatal
dynamics of intramuscular and oral betame- Antenatal maternal medication administration in Med 2020;33:2109–15.
thasone and dexamethasone in reproductive preventing respiratory distress syndrome of 100. Greensides D, Robb-McCord J,
age women in India. Clin Transl Sci 2020;13: premature infants: a network meta-analysis. Clin Noriega A, Litch JA. Antenatal corticosteroids
391–9. Respir J 2018;12:2480–90. for women at risk of imminent preterm birth in
88. Lee VR, Kaimal AJ, Caughey AB. Cost- 95. Gonzalez Garay AG, Reveiz L, Velasco 7 sub-Saharan African Countries: a policy and
effectiveness of antenatal late preterm steroids. Hidalgo L, Solis Galicia C. Ambroxol for women implementation landscape analysis. Glob
Am J Obstet Gynecol 2017;216:S233. at risk of preterm birth for preventing neonatal Health Sci Pract 2018;6:644–56.

14 AJOG MFM MAY 2021