Documente Academic
Documente Profesional
Documente Cultură
MSc,
Tetsuya Isayama,
MD,
MD, MSc
TABULATION, INTEGRATION, AND RESULTS: We performed duplicate independent assessment of the title
and abstracts, full-text screening, inclusion of articles,
and data abstraction. We performed meta-analyses using
random-effects models and quality assessment with the
Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system. There were 17
observational studies, and our primary outcome, mortality to discharge in neonates receiving active intensive
treatment, had a total of 3,626 neonates. The adjusted
odds of mortality to discharge were reduced by 52% in
the antenatal corticosteroid group compared with the
control group (crude adjusted odds ratio [OR] 0.45, 95%
confidence interval [CI] 0.360.56; adjusted OR 0.48, 95%
CI 0.380.61; mortality to discharge 58.1% [intervention]
compared with 71.8% [control]) with a moderate quality of evidence based on the GRADE system. There were
no significant differences between the groups for severe
morbidity.
Financial Disclosure
Ms. Park is supported by a Canadian Institutes of Health Research (CIHR)
Training Program in Reproduction, Early Development and the Impact on
Health (TGF-96122). Dr. McDonald is supported by a CIHR Canada Research
Chair (950229920). Dr. Isayama did not report any potential conflicts of
interest.
715
SOURCES
We followed the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Version 5.1.0)14 and the Preferred Reporting Items for
Systematic Reviews and Meta-Analyses statement15
(Appendix 1, available online at http://links.lww.
com/AOG/A786). We drafted the protocol for this
systematic review before the literature search (Appendix 2, available online at http://links.lww.com/AOG/
A786) with the exception of the decision to use the
Newcastle-Ottawa Scale for individual study risk of
bias assessment, which we made before commencement of data extraction.
We searched four databases: MEDLINE, EMBASE,
Cumulative Index to Nursing and Allied Health Literature (CINAHL), and the Cochrane Central Register of
Controlled Trials. We developed separate search strategies with the aid of an experienced research librarian for
each database, consisting of a combination of MeSH
headings and multipurpose terms (.mp) (Appendix 3,
available online at http://links.lww.com/AOG/A786).
We searched ClinicalTrials.gov to find completed
trials that were unpublished using the keywords antenatal corticosteroid. We consulted experts in maternalfetal medicine and early preterm birth for
knowledge on studies published in this area. We
searched the references of included studies for additional articles of interest.
We imported all citations into bibliographic
software (Endnote X7),16 removed duplicates, and assigned each article an identification number.
716
Park et al
STUDY SELECTION
We included all published observational studies, and
randomized controlled trials (RCTs) and quasi-RCTs,
with no language restrictions. We selected a starting
point of 1990, because there have been substantial
advances in neonatal resuscitation in recent years.17
The population included preterm neonates born
before 24 weeks of gestation. The intervention was
the receipt of a single course of antenatal corticosteroids in utero before 24 weeks of gestation for women
at risk of preterm birth within 7 days before birth. The
comparator group included those who received no
treatment or a placebo treatment.
Our primary outcome was neonatal mortality up
until discharge from the hospital, the period when the
vast majority of deaths occur and allows for robust
rates of follow-up.18 For the adjusted analysis for mortality to discharge, we included only the studies with
neonates who were given active intensive postnatal
treatment (resuscitation) and not all liveborn neonates
because the inclusion of neonates who did not receive
treatment may bias the results to overestimate the
benefit of antenatal corticosteroid.19 If information
for this population was not available in the primary
studies, we contacted the study authors.
The main secondary outcomes were respiratory
distress syndrome (RDS), severe intraventricular hemorrhage (grades III and IV), necrotizing enterocolitis
(NEC), chronic lung disease at 36 weeks postmenstrual
age, retinopathy of prematurity, and neurologic impairment. These outcomes were selected for the quality of
evidence assessment using the Grading of Recommendations Assessment, Development, and Evaluation
(GRADE) system. Other secondary outcomes focused
on common major morbidities and long-term outcomes.
We excluded other types of publications (eg,
reviews, editorials, commentaries, case studies, conference proceedings, studies published only as abstracts), studies without sufficient data, and duplicated
studies or data.
Two reviewers (C.K.P. and T.I.) independently
screened the titles and abstracts of all retrieved citations
with a low threshold for selecting studies for full-text
review. We independently screened the full texts to
determine inclusion and recorded reasons for exclusion.
We assessed interreviewer agreement study inclusion
using k statistics. We considered a k statistic of less than
0.00, 0.000.20, 0.210.40, 0.410.60, 0.610.80, and
0.811.00 as poor, slight, fair, moderate, substantial, and almost perfect agreement, respectively.20
We developed a data collection form and each
reviewer piloted the form before data abstraction of
included studies. We extracted the following characteristics from the included studies: first author and year of
publication, study design, country or region, years of
study, final sample size, inclusion and exclusion criteria,
description of intervention, and outcome measures. We
resolved discrepancies between reviewers through
discussion and consensus. A third reviewer (S.D.M.)
resolved any disagreements or uncertainties. For the
primary outcome, we included only neonates who
received active treatment to avoid a potential overestimation of the effect caused by including those who
did not receive active treatment. Authors of primary
studies were contacted and the study was excluded if it
was not possible to obtain these data.
We summarized the data quantitatively. Because
between-study heterogeneity was anticipated, we
performed meta-analyses using random-effects models, implementing an inverse variance method, which
adjusts study weights according to the magnitude of
variation among the intervention effects and could be
considered more conservative.14 We planned separate
analyses for RCTs and observational studies. We generated summary effect estimates (relative risks or odds
ratios and standardized mean differences) with associated 95% confidence intervals (CIs) for each outcome.
We pooled unadjusted (the raw number of events and
total) and adjusted analyses separately, if available.
The adjusted analyses were preferred because they
are an indication of the independent effect of antenatal
corticosteroids after accounting for confounders.
We performed all analyses using Review Manager 5.3.21 We estimated statistical significance using
a two-sided P value of .05. As is typical in metaanalyses, we did not adjust the P value for multiple
secondary outcomes.
We assessed heterogeneity using I2 statistics. I2
values of 040%, 3060%, 5090%, and 75100%
were considered low, moderate, substantial, and considerable heterogeneity, respectively.14 The a priori
hypotheses of sources of significant heterogeneity
were gestational age or multiple birth. Therefore,
where possible, we performed subgroup analyses for
gestational age and singleton or multiple pregnancies.
Two of the authors (C.K.P. and T.I.) independently
assessed risk of bias for each study using the Cochrane
Collaborations tool for assessing risk of bias for RCTs
and quasi-RCTs and a modified version of the
Newcastle-Ottawa Scale for observational studies.22,23
We quantitatively assessed publication bias with a funnel plot for outcomes with 10 or more studies.24
The Newcastle-Ottawa Scale assesses selection,
comparability, and outcome. One of the items to
assess selection bias, demonstration that outcome of
Park et al
RESULTS
The searches yielded 6,435 articles (MEDLINE51,929,
EMBASE53,887, Cochrane Central Register of Controlled Trials5173, CINAHL5446; March 13, 2015;
Fig. 1) with 12 additional articles from ClinicalTrials.
gov. After removing duplicates, 4,479 records
717
718
Park et al
Fig. 2. Summary of adjusted pooled odds ratios (95% confidence interval [CI]) for the association between mortality to
discharge and antenatal corticosteroid (ANCS) therapy compared with not receiving antenatal corticosteroid therapy. IV,
inverse variance; df, degrees of freedom.
Park. Antenatal Corticosteroid Therapy Before 24 Weeks. Obstet Gynecol 2016.
Fig. 3. Summary of adjusted pooled odds ratios (95% confidence interval [CI]) for the association between respiratory
distress syndrome and antenatal corticosteroid (ANCS) therapy compared with not receiving antenatal corticosteroid
therapy. IV, inverse variance; df, degrees of freedom.
Park. Antenatal Corticosteroid Therapy Before 24 Weeks. Obstet Gynecol 2016.
Park et al
719
Fig. 4. Summary of adjusted pooled odds ratios (95% confidence interval [CI]) for the association between severe intraventricular hemorrhage (grades III and IV) and antenatal corticosteroid (ANCS) therapy compared with not receiving
antenatal corticosteroid therapy. IV, inverse variance; df, degrees of freedom.
Park. Antenatal Corticosteroid Therapy Before 24 Weeks. Obstet Gynecol 2016.
and not for all neonates including those who died. Neonates with intraventricular hemorrhage, periventricular
leukomalacia, and NEC have a high risk of death after
developing these complications,49,50 and thus including
only neonates who survived could lead to bias.
The outcomes of all but four studies had a low
risk of bias according to the modified NewcastleOttawa Scale (Appendix 5, available online at
Fig. 5. Summary of adjusted pooled odds ratios (95% confidence interval [CI]) for the association between necrotizing
enterocolitis and antenatal corticosteroid (ANCS) therapy compared with not receiving antenatal corticosteroid therapy. IV,
inverse variance; df, degrees of freedom.
Park. Antenatal Corticosteroid Therapy Before 24 Weeks. Obstet Gynecol 2016.
720
Park et al
Fig. 6. Summary of adjusted pooled odds ratios (95% confidence interval [CI]) for the association between chronic lung
disease and antenatal corticosteroid (ANCS) therapy compared with not receiving antenatal corticosteroid therapy. IV,
inverse variance; df, degrees of freedom.
Park. Antenatal Corticosteroid Therapy Before 24 Weeks. Obstet Gynecol 2016.
No
Corticosteroids
Corticosteroids
Adjusted OR
(95% CI)
No. of Participants
(No. of Studies)
Quality of
Evidence
718/1,000
555/1,000 (505600)
0.48 (0.380.61)
3,646 (4)
Moderate
NA
NA
1.09 (0.691.73)
739 (2)
Very low
NA
NA
0.82 (0.551.21)
737 (2)
Very low
NA
785/1,000
NA
562/1,000
NA
756/1,000 (694809)
NA
587/1,000 (490680)
1.46
0.85
0.76
1.11
(0.862.48)
(0.621.16)
(0.115.43)
(0.751.66)
917
1,184
NA
601
(3)
(3)
(1)
(1)
Very low
Very lowk
Very low#
Very low#
CI, confidence interval; OR, odds ratio; NA, not available; ROP, retinopathy of prematurity.
* The risk in the intervention group is based on the assumed risk in the comparison group and the relative effect of the intervention.
Statistically significant.
Optimal information sizes (OIS) was met, and CI overlapped no effect and failed to exclude appreciable benefit or harm.
OIS was not met, and CI overlapped no effect and failed to exclude appreciable benefit.
k
OIS was not met, and CI overlapped no effect and failed to exclude appreciable benefit or harm.
OIS was met, but CI overlapped no effect and failed to exclude appreciable benefit.
#
Not enough information to calculate OIS, and CI overlapped no effect and failed to exclude appreciable benefit or harm.
Park et al
721
722
Park et al
DISCUSSION
In our analysis, which consisted exclusively of
observational data, we found a halving of the odds
of mortality to discharge in neonates before 24 weeks
of gestation who received antenatal corticosteroids
compared with those who did not. The subgroup
analysis at 22 compared with 23 weeks of gestation
found no difference in mortality to discharge. However, more data are needed, because the sample size
for 22 weeks of gestation was small and the CIs
around the adjusted, but not the unadjusted, odds
included the null value. At 22 weeks of gestation, the
fetus has limited but detectable alveolar development.51 The fetal lung at 2223 weeks of gestation is
at the end stage of the canalicular stage (1624 weeks
of gestation) of lung development followed by the
saccular stage (2440 weeks of gestation). Because
these critical lung developments start during the
end of the canalicular stage (2224 weeks of gestation),52 neonates at 2223 weeks of gestation are considered potentially viable.
No significant reduction in secondary outcomes
in the steroid group was contrary to a hypothesis that
the reduction of morbidities would lead to lower odds
of mortality. This may be because the optimal
information sizes were not met for these secondary
outcomes, unlike the primary outcome, indicating
insufficient sample sizes. Another possible explanation may be the misdiagnosis of RDS. It may be
difficult to ascertain RDS at 2223 weeks of gestation
if neonates immediately intubated and administered
surfactant as would occur in many institutions.
Slightly more mature neonates are often managed
without intubation, making the diagnosis of RDS
potentially easier. A third possibility is that corticosteroids are not effective at lowering morbidity in neonates delivered before 24 weeks of gestation.
Our results differed from two previous systematic
reviews examining outcomes of extremely preterm
birth, beginning at 24 weeks of gestation.4,53 The 2006
Cochrane review4 found that neonatal mortality was
significantly reduced with antenatal corticosteroids
before 32, 34, and 36 weeks of gestation, but not in
the two studies of 89 neonates at 24 to less than 28
weeks of gestation (relative risk 0.79 [95% CI 0.56
1.12]). A more recent 2011 systematic review,53 also
of RCTs, did not find a significant reduction in neonatal mortality in pooled analyses of two studies with
173 neonates at 2429 weeks of gestation (relative risk
0.86 [95% CI 0.481.53]). The point estimates of both
meta-analyses were similar, but the sample sizes were
modest. Further research is needed to investigate
short- and long-term morbidity.
A strength of this systematic review is the subgroup
analysis for 22 and 23 weeks of gestation, which is
clinically relevant. Studies were from diverse settings,
which increased generalizability. We performed the risk
of bias in individual studies and the quality of evidence
for outcomes. Our primary outcome was well powered,
exceeding the optimal information size. For the primary
outcome, we included only neonates who received
active treatment to avoid a potential overestimation of
the effect caused by including those who did not receive
active treatment. Including only neonates who received
active intensive treatment rather than expectant management in both antenatal corticosteroid and control
groups minimizes selection bias favoring corticosteroids. Adjusted data were pooled separately from
unadjusted data to estimate the independent effect of
antenatal corticosteroids to understand the absolute and
relative effects for survival. A limitation was that many
of the secondary outcomes lacked data with some
outcomes only containing one study, and all had very
low quality of evidence. The studies that contributed to
the adjusted analyses did not always control for the
same confounding variables. Some clinically relevant
aspects of care not reported included the time from
administration of steroids to birth, use of surfactant,
and use of magnesium sulfate for neonatal neuroprotection. Only observational studies have been published,
which are not as robust as RCTs.
Although variation exists in the literature on the
survival rates of neonates born before at 24 weeks of
gestation, it is agreed that family counseling and planning
of services are essential for these births,11,54 because advances in care have opened the possibility of survival at
progressively early gestations. Increasingly, parents,
Park et al
723
16. EndNote X7. Available at: http://endnote.com/product-details/X7. Thomson Reuters; 2014. Retrieved February 12,
2016.
17. Morley CJ, Davis PG. Advances in neonatal resuscitation: supporting transition. Arch Dis Child Fetal Neonatal Ed 2008;93:F3346.
18. Carlo WA, McDonald SA, Fanaroff AA, Vohr BR, Stoll BJ,
Ehrenkranz RA, et al. Association of antenatal corticosteroids
with mortality and neurodevelopmental outcomes among infants born at 22 to 25 weeks gestation. JAMA 2011;306:
234858.
19. Litmanovitz I, Reichman B, Arnon S, Boyko V, Lerner-Geva L,
Bauer-Rusak S, et al. Perinatal factors associated with active
intensive treatment at the border of viability: a populationbased study. J Perinatol 2015;35:70511.
20. Landis JR, Koch GG. The measurement of observer agreement
for categorical data. Biometrics 1977;33:15974.
21. Zhang J, Yu KF. Whats the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998;280:16901.
22. Higgins JP, Altman DG, Gotzsche PC, Juni P, Moher D,
Oxman AD, et al. The Cochrane Collaborations tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928.
23. Wells G, Shea B, OConnell D, Peterson J, Welch V, Losos M,
et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality
of nonrandomised studies in meta-analyses. Available at: http://
www.ohri.ca/programs/clinical_epidemiology/oxford.asp. Ottawa
Hospital Research Institute; Retrieved February 12, 2016.
24. Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J,
et al. Recommendations for examining and interpreting funnel
plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011;343:d4002.
25. The GRADE working group. GRADE working group.
Available at: http://www.gradeworkinggroup.org/index.htm.
Retrieved February 12, 2016.
26. Guyatt G, Oxman AD, Akl EA, Kunz R, Vist G, Brozek J, et al.
GRADE guidelines: 1. Introduction-GRADE evidence profiles
and summary of findings tables. J Clin Epidemiol 2011;64:38394.
724
Park et al
Park et al
rev 2/2015
725