Professional Documents
Culture Documents
14545
www.bjog.org
Objective To explore the association between administration-to- infants unexposed to ACS [hazard ratio (HR) = 0.26; 95%
birth interval of antenatal corticosteroids (ACS) and survival in confidence interval 0.15–0.43], in infants born <24 h [HR = 0.53
extremely preterm infants. (0.33–0.87)] and >7 days after ACS [HR = 0.56 (0.32–0.97)], but
not in infants born 24–47 h after ACS [HR = 1.60 (0.73–3.50)],
Design Population-based prospective cohort study.
as compared with infants born 48 h to 7 days after
Setting All obstetric and neonatal units in Sweden from 1 April administration. The findings were similar for neonatal survival.
2004 to 31 March 2007. Survival without major neonatal morbidity among live-born
infants was 14% in unexposed infants and 30–39% in steroid-
Population All live-born infants (n = 707) born at 22–26
exposed groups, indicating that any ACS exposure was valuable.
completed weeks of gestation.
Conclusions Administration of ACS 24 h to 7 days before
Methods The relationship between time from first administration
extremely preterm birth was associated with significantly higher
of ACS to delivery and survival was investigated using Cox
survival than in unexposed infants and in infants exposed to ACS
proportional hazards regression analysis.
at shorter or longer administration-to-birth intervals.
Main outcome measures Neonatal (0–27 days) and infant (0–
Keywords Administration-to-birth interval, antenatal
365 days) survival, and infant survival without major neonatal
glucocorticoids, extremely preterm birth, major neonatal
morbidity (intraventricular haemorrhage grade ≥ 3, retinopathy of
morbidity, mortality.
prematurity stage ≥ 3, periventricular leukomalacia, necrotising
enterocolitis, or severe bronchopulmonary dysplasia). Tweetable abstract Timing of antenatal corticosteroids is
important for extremely preterm infants’ survival.
Results Five-hundred and ninety-one (84%) infants were exposed
to ACS. In the final adjusted model, infant survival was lower in
Please cite this paper as: Norberg H, Kowalski J, Marsal K, Norman M. Timing of antenatal corticosteroid administration and survival in extremely preterm
infants: a national population-based cohort study. BJOG 2017; DOI: 10.1111/1471-0528.14545.
gestation with an administration-to-birth interval of ACS gestational age (SGA). IVH was graded in accordance with
shorter or longer than 1–7 days and, if anything, these data from Papile et al.,17 and periventricular leukomalacia
findings were more pronounced in infants <29 weeks of (PVL) was defined as proposed by de Vries et al.18
gestation.9 Retinopathy of prematurity (ROP) was defined in accor-
Given that pregnant women delivering high-risk infants dance with the International Classification for Retinopathy
at extremely short gestations may deliver before ACS have of Prematurity19 and NEC according to Bell et al., where
been administered, or deliver at various time intervals after stage two or more were considered as diagnostic.20 Severe
ACS, a better understanding of the effects and importance bronchopulmonary dysplasia (BPD) was defined as the
of the administration-to-birth interval of ACS for outcome need for ≥ 30% oxygen at a postmenstrual age of 36
in extremely preterm infants is needed. The objective of weeks.21
this prospective population-based cohort study was to A complete course with ACS to the pregnant mothers
investigate the impact of the administration-to-birth inter- consisted of two doses of 12 mg betamethasone 24 h apart.
val of ACS on survival of extremely preterm infants. In the EXPRESS-database, time from the first ACS dose to
delivery in hours had been prospectively collected. We cate-
gorised all live-born infants into four categories according
Methods
to the time interval from the first ACS dose to delivery:
The Extremely Preterm Infants in Sweden Study (EXPRESS) <24 h; 24–47 h; 48 h to 7 days; and >7 days. The
included all pregnant women residing in Sweden and deliver- EXPRESS-database also provided information whether an
ing extremely preterm infants, born at 22–26 completed incomplete (one dose) or complete course (two doses) had
weeks of gestation, from 1 April 2004 to 31 March 2007. The been administered. The category <24 h comprised infants
cohort studied herein consisted of all live-born infants that had received either an incomplete (one dose) or a
(n = 707), including multiple births and infants with malfor- complete course (two doses) most likely reflecting that the
mations, whereas stillborn infants (n = 304) and infants study was descriptive with no attempts in the study frame-
born outside Sweden and transferred after birth to Sweden work to standardise treatment or compliance to treatment
for neonatal care were excluded. Detailed characteristics of protocols. Additional categories consisted of infants unex-
EXPRESS have been reported elsewhere.12–14 Among live- posed to ACS and those exposed to ACS but with
born infants, 520 survived the neonatal period and 497 were unknown timing.
alive at 1 year old (Figure S1). Outcomes were neonatal (0–28 days) and infant survival
The procedures for data collection and other method- (0–365 days). Infant survival without major neonatal mor-
ological details have been previously described.13 In brief, bidity, i.e. survival without IVH grade ≥ 3, ROP stage ≥ 3,
during the study period, Sweden had seven healthcare PVL, NEC or severe BPD, was also included as an out-
regions, each served by a regional level III hospital. The come.
general policy was to centralise extremely preterm deliveries The study was approved by the Regional Research Ethics
to these regional hospitals. Data on mothers were collected Board, Lund University, Lund, Sweden (date of approval 1
at the time of delivery. Data on live-born infants were col- July 2004; reference number 42/2004). The parents pro-
lected prospectively during the first 180 days of hospitalisa- vided oral informed consent for data acquisition.
tion or until discharge or death. Information on infant
deaths after discharge home until 1 year was obtained from Statistical analyses
the National Population Registry. Data are presented as numbers and proportions (percent-
Maternal hypertensive disease was defined as essential ages) if not stated otherwise. The differences across ACS
hypertension, preeclampsia or eclampsia. Information on administration-to-birth interval categories were presented
maternal smoking was obtained at the first antenatal visit, using descriptive statistics. Cox proportional hazards
and the information was dichotomised (smoking/non- regression analysis was used to estimate hazard ratios (HR)
smoking). Preterm prelabour rupture of membranes for survival across the ACS administration-to-birth interval
(PPROM) was defined as spontaneous rupture of the mem- categories, including one unexposed category and one
branes at least 1 h before the onset of contractions. dummy category representing unknown ACS timing. Based
Live birth and perinatal mortality were defined according on available knowledge on optimal timing for ACS admin-
with WHO.15 Gestational age was recorded in completed istration, the category 48 h to 7 days was considered as ref-
weeks and based on ultrasound in 95% of pregnancies. erence category, i.e. HR = 1.00. HRs for neonatal and
Birth weight was evaluated in accordance with the national infant survival in relation to ACS administration-to-birth
standard for normal fetal growth16 and expressed in stan- intervals were evaluated in a multiple Cox proportional
dard deviation (SD) scores. Infants with birth weight more hazards regression model adjusting for factors known to
than 2 SD below the mean were classified as small for affect infant mortality: maternal smoking;22 maternal blood
pressure disease;23 placenta previa; placental abruption; Statistical calculations were performed using JMP 11.0
PPROM;24 regionalisation of care with birth at level III (SAS Institute, Cary, NC, USA) and the IBM SPSS Statistics
hospital;23,24 gestational age;13,23 SGA;23,24 infant gender;23 V22 or higher (IBM Corporation).
and surfactant therapy within 2 h after birth. The rationale
for including surfactant therapy within 2 h after birth in
Results
the model was that this intervention has been shown to be
associated with significantly reduced infant mortality in Among 707 infants born alive at 22–26 weeks of gestation,
randomised controlled trials and in the EXPRESS-study, 591 (84%) had been exposed to ACS, whereas 85 (12%)
independently of ACS exposure.13,25 Nonetheless, because were unexposed to ACS, and in 31 infants (4%) data
early surfactant therapy may be in the causal pathway, we regarding ACS exposure were missing. Of the infants
ran a sensitivity analysis without surfactant in the model exposed to ACS, 149 (25%) were born less than 24 h after
for infant mortality. Cumulative neonatal and infant sur- the first ACS dose, 66 (11%) were born within 24–47 h,
vival adjusted for covariates was also determined using 171 (29%) were born within 48 h to 7 days, and 103
Kaplan–Meier survival analyses. The odds for surviving (17%) infants were born more than 7 days after the first
without major neonatal morbidity were analysed using a ACS dose. One-hundred and two (17%) infants have been
multiple logistic regression model adjusting for the same exposed to ACS without any information on the adminis-
factors as for the analysis in the Cox proportional hazards tration-to-birth time interval.
regression. Results are presented as odds compared with
the reference category of ACS 48 h to 7 days before birth, Maternal characteristics
odds ratio (OR) = 1.00. We made no adjustments for Maternal smoking in pregnancy did not differ significantly
multiple births as our analyses are to be interpreted as between ACS exposure categories. Pregnant women with
exploratory. All tests were two-sided, and a P-value of less hypertensive disease had more often received ACS treat-
than 0.05 was considered as statistically significant. ment than women without such disease (Table 1).
no ACS <24 h 24–47 h 48 h–7 days >7 days ACS but unknown timing
n = 85 n = 149 n = 66 n = 171 n = 103 n = 102
Maternal characteristics
Smoker 9 (10.6) 15 (10.1) 5 (7.6) 27 (15.8) 12 (11.7) 16 (15.7)
Hypertensive disease* 3 (3.5) 12 (8.1) 9 (13.6) 29 (17.0) 12 (11.7) 21 (20.6)
Obstetric characteristics
Born at level III hospital 26 (30.6) 107 (71.8) 62 (93.9) 157 (91.8) 98 (95.2) 85 (83.3)
Tocolytic treatment 23 (27.1) 119 (79.9) 42 (63.6) 111 (64.9) 61 (59.2) 55 (53.9)
PPROM 12 (14.1) 11 (7.4) 11 (16.7) 31 (18.1) 20 (19.4) 18 (17.7)
Placenta previa 4 (4.7) 2 (1.3) 3 (4.6) 6 (3.5) 7 (6.8) 5 (4.9)
Placenta abruption 13 (15.3) 9 (6.0) 5 (7.6) 14 (8.2) 13 (12.6) 13 (12.8)
Infant characteristics
Gestational age at birth (weeks)
22 29 (34.1) 11 (7.4) 1 (1.5) 3 (1.8) 0 5 (4.9)
23 14 (16.5) 28 (18.8) 14 (21.2) 26 (15.2) 4 (3.9) 13 (12.8)
24 7 (8.2) 32 (21.5) 13 (19.7) 38 (22.2) 25 (24.3) 22 (21.6)
25 22 (25.9) 40 (26.9) 22 (33.3) 52 (30.4) 29 (28.2) 32 (31.4)
26 13 (15.3) 38 (25.5) 16 (24.2) 52 (30.4) 45 (43.7) 30 (29.4)
SGA 6 (7.1) 12 (8.1) 13 (20.0) 41 (24.0) 20 (19.4) 22 (21.6)
Male gender 44 (51.8) 78 (52.4) 32 (48.5) 92 (53.8) 58 (56.3) 67 (65.7)
Multiplets 15 (17.6) 41 (27.5) 18 (27.3) 30 (17.5) 23 (22.3) 26 (25.5)
Surfactant <2 h after birth 27 (31.8) 94 (63.1) 32 (48.5) 84 (49.1) 57 (55.3) 60 (58.8)
ACS, antenatal corticosteroids; PPROM, preterm prelabour rupture of membranes; SGA, small for gestational age (birthweight more than 2 SD
below the mean of a national fetal weight-based growth standard).
Data are numbers (percentages).
*Essential hypertension and/or preeclampsia and/or eclampsia.
Table 2. Survival among live-born infants according to the administration-to-birth interval of ACS
no ACS <24 h 24–47 h 48 h–7 days >7 days ACS but unknown P-value
n = 85 n = 149 n = 66 n = 171 n = 103 timing
n = 102
Neonatal survival 28 days 33 (38.8) 105 (70.5) 56 (84.9) 143 (83.6) 85 (82.5) 77 (75.5) <0.0001
Infant survival 365 days 30 (35.3) 105 (70.5) 52 (78.8) 137 (80.1) 79 (76.7) 74 (72.6) <0.0001
Infant survival 365 days without major morbidity* 12 (14.1) 47 (31.5) 20 (30.3) 67 (39.2) 37 (35.9) 33 (32.4) 0.0018
1.0
0.6
No ACS
0.5
0 7 14 21 28
Age, days
Figure 1. Survival curve adjusted for covariatesa for extremely preterm infants according to antenatal corticosteroid (ACS) treatment and
administration-to-birth intervals.
Table 3. Association between the administration-to-birth interval of ACS and survival among live-born infants
no ACS <24 h 24–47 h 48 h–7 days >7 days ACS but unknown
n = 85 n = 149 n = 66 n = 171 n = 103 timing
n = 102
ACS, antenatal corticosteroids; CI, confidence interval; HR, hazard ratio; OR, odds ratio.
*Adjusted for maternal smoking, maternal blood pressure disease, placenta previa, placental abruption, PPROM, regionalisation of care,
gestational age, SGA, infant gender and surfactant therapy within 2 h after birth.
**Survival without any of the following: IVH ≥ 3, ROP ≥ 3, severe BPD, PVL or NEC.
difference in neonatal survival without morbidity in any of associated with a doubled risk for infant mortality. The
the ACS groups (Table 3). In the fully adjusted model, the lowest survival was found in infants unexposed for ACS –
OR for infant survival without major neonatal morbidity their adjusted infant mortality risk was five times higher
was significantly lower in infants unexposed to ACS than in infants in the reference group exposed to ACS 48 h
(OR = 0.33; 95% CI = 0.16–0.69; Table 3). to 7 days before birth. Similar results were found for
neonatal survival and infant survival without major neona-
tal morbidity.
Discussion
Main findings Strengths and limitations
We found clear associations between timing of ACS and The strengths of this study include prospective enrollment
survival in extremely preterm infants. Shorter or longer of all extremely preterm births – including delivery room
administration-to-birth intervals than 24 h to 7 days were deaths – in Sweden over a 3-year period. The study is one
of the largest7,9 and the only population-based cohort eval- fetal–infant survival among pregnant women exposed to
uated for effects of ACS, and of timing of ACS, in all live- ACS before 27 weeks of gestation but who, for various rea-
born infants born at 22–26 weeks of gestation. The ACS sons, did not deliver before this upper limit for inclusion.
administration-to-birth interval was prospectively collected
in hours. The outcomes were robust and clinically highly Interpretation
relevant. The sample size allowed for adjusted analyses of Antenatal corticosteroid treatment has previously been
survival in relation to timing of ACS-administration and associated with reduced mortality in extremely preterm
several potential confounders. The case-mix of our infants born after 22–26 weeks of gestation.7,8,26 The mag-
cohort12,13 is similar to what has been reported in other nitude of an effect of ACS in those studies varied between
contemporary cohorts of extremely preterm infants. an adjusted HR of 0.38 and 0.72, which is similar to our
We excluded stillborn infants. Among stillborn infants findings. In contrast, a meta-analysis concluded that there
alive at onset of labour (n = 66/304; 22% of all stillborns), were no effects from ACS on neonatal mortality and mor-
a minority (n = 22/66; 30% of stillborns alive at onset of bidity in extremely preterm infants.6 Adding the data pro-
labour) had been treated with ACS, meaning that these vided in our study is likely to change that conclusion.
fetuses had been proactively managed. Excluding intra- Animal studies have shown the greatest benefits from
partum mortality may therefore have introduced some bias, ACS on surfactant production when delivery occurs
overestimating the protective effects of ACS at the time it between 24 h and 7 days after a complete course of
was administered. However, under the premise that the ACS.27,28 However, human studies are more contradictory.
infant was live-born, our risk estimates of no or untimely Some studies have found a decline in the effectiveness of
ACS are valid. ACS over time,9,10,29,30 whereas others failed to do so.11,31–
33
During the study period, a restricted treatment policy The conflicting results may be attributed partly to the
was – if parents agreed – used in some but not all of the use of diverse outcome measures.
participating hospitals in Sweden for births that were con- Our results are in line with a recent study where the
sidered nonviable. The EXPRESS-database does not contain authors found that ACS for preterm birth at 24–34 weeks
specific information on reasons for not treating with ACS; of gestational age had maximum beneficial impact on
however, a recommendation to withhold treatment rested neonatal morbidity and mortality when administered 1–
mainly on regional guidelines focusing on management of 7 days before birth. In that study, the decline in effective-
extremely short gestations or severe malformations. Among ness of ACS after 7 days was primarily observed among
infants born at 22 weeks of gestation, 29/49 (59%) did not neonates born at 24–28 weeks of gestation.9
receive ACS, at 23 weeks of gestation the corresponding We found decreased survival with an ACS administra-
proportion was 14/99 (14%), and in infants born at tion-to-birth interval exceeding 7 days. The NIH Consen-
24 weeks of gestation the proportion not receiving ACS sus Development Panel in 2000 recommended that a repeat
was 7/137 (5.1%). Accordingly, crude estimates of survival course of corticosteroids should be reserved for women
in relation to ACS exposure and administration-to-birth participating in controlled trials.34 This recommendation
intervals are likely to be confounded by indication. To was subsequently endorsed by the Canadian Society of
resolve this issue, we only present mortality data adjusted Obstetricians and Gynecologists in 2003. Since then, large
for gestational age. randomised controlled trials have reported conflicting
Small numbers of 22-week fetuses, with only 20 exposed results.35,36 The latest Cochrane review concluded that
to ACS and five survivors, make generalisability of our repeat courses of corticosteroids were associated with a
findings to this group limited. And although based on 707 reduction in the incidence and severity of neonatal lung
live-born infants, limitations in numbers may also con- disease and a small reduction in size at birth.37 Recent
tribute to wide confidence intervals for some of our risk work indicate no harm at follow-up to school age in those
estimates. exposed to repeat courses of ACS at 24–28 weeks and born
We cannot exclude that other confounding factors than later after on average 32 weeks of gestation.38 Observational
those included herein were unevenly distributed between data from adolescence and young adult age also suggest
categories and affected our results. We excluded tocolysis long-term safety after exposure to repeat ACS.39 Repeat or
from our multivariate analyses because, in univariate test- rescue doses of glucocorticoids may therefore be considered
ing, survival was not related to this variable, and because in women at continued risk of preterm birth 7 or more
of some missing data (n = 29) on use of tocolysis. Other days after an initial course.37
and unknown differences in case-mix between the different The included infants were born almost a decade ago.
ACS categories may have occurred. Moreover, our results Since then neonatal care has improved and the evidence
cannot be extrapolated to other drugs or dosing regimens for less invasive ventilation strategies has become clearer.40
than the ones used in Sweden. Finally, we had no data on Still, we think our findings are generalisable for the current
13 Group E, Fellman V, Hellstrom-Westas L, Norman M, Westgren M, 28 Vidaeff AC, Ramin SM, Gilstrap LC 3rd, Alcorn JL. Characterization
Kallen K, et al. One-year survival of extremely preterm infants after of corticosteroid redosing in an in vitro cell line model. Am J Obstet
active perinatal care in Sweden. JAMA 2009;301:2225–33. Gynecol 2004;191:1403–8.
14 Serenius F, Kallen K, Blennow M, Ewald U, Fellman V, Holmstrom 29 Kuk JY, An JJ, Cha HH, Choi SJ, Vargas JE, Oh SY, et al. Optimal
G, et al. Neurodevelopmental outcome in extremely preterm infants time interval between a single course of antenatal corticosteroids
at 2.5 years after active perinatal care in Sweden. JAMA and delivery for reduction of respiratory distress syndrome in
2013;309:1810–20. preterm twins. Am J Obstet Gynecol 2013;209:256.e1–7.
15 World Health Organization. Neonatal and Perinatal Mortality: 30 Waters TP, Mercer B. Impact of timing of antenatal corticosteroid
Country, Regional and Global Estimates. Geneva: WHO Press, 2006. exposure on neonatal outcomes. J Matern Fetal Neonatal Med
16 Marsal K, Persson PH, Larsen T, Lilja H, Selbing A, Sultan B. 2009;22:311–14.
Intrauterine growth curves based on ultrasonically estimated foetal 31 Peaceman AM, Bajaj K, Kumar P, Grobman WA. The interval
weights. Acta Paediatr 1996;85:843–8. between a single course of antenatal steroids and delivery and its
17 Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution association with neonatal outcomes. Am J Obstet Gynecol
of subependymal and intraventricular hemorrhage: a study of 2005;193:1165–9.
infants with birth weights less than 1,500 gm. J Pediatr 32 Sehdev HM, Abbasi S, Robertson P, Fisher L, Marchiano DA, Gerdes
1978;92:529–34. JS, et al. The effects of the time interval from antenatal
18 de Vries LS, Eken P, Dubowitz LM. The spectrum of leukomalacia corticosteroid exposure to delivery on neonatal outcome of very low
using cranial ultrasound. Behav Brain Res 1992;49:1–6. birth weight infants. Am J Obstet Gynecol 2004;191:1409–13.
19 International Committee for the Classification of Retinopathy of P. 33 Vermillion ST, Soper DE, Newman RB. Is betamethasone effective
The International Classification of Retinopathy of Prematurity longer than 7 days after treatment? Obstet Gynecol 2001;97:491–3.
revisited. Arch Ophthalmol 2005;123:991–9. 34 National Institutes of Health Consensus Development P. Antenatal
20 Bell MJ, Ternberg JL, Feigin RD, Keating JP, Marshall R, Barton L, corticosteroids revisited: repeat courses – National Institutes of
et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based Health Consensus Development Conference Statement, August 17-
upon clinical staging. Ann Surg 1978;187:1–7. 18, 2000. Obstet Gynecol 2001;98:144–50.
21 Jobe AH, Bancalari E. Bronchopulmonary dysplasia. Am J Respir Crit 35 Murphy KE, Hannah ME, Willan AR, Hewson SA, Ohlsson A, Kelly EN,
Care Med 2001;163:1723–9. et al. Multiple courses of antenatal corticosteroids for preterm birth
22 Cnattingius S. The epidemiology of smoking during pregnancy: (MACS): a randomised controlled trial. Lancet 2008;372:2143–51.
smoking prevalence, maternal characteristics, and pregnancy 36 Crowther CA, Haslam RR, Hiller JE, Doyle LW, Robinson JS,
outcomes. Nicotine Tob Res 2004;6(Suppl 2):S125–40. Australasian Collaborative Trial of Repeat Doses of Steroids Study G.
23 Johansson S, Montgomery SM, Ekbom A, Olausson PO, Granath F, Neonatal respiratory distress syndrome after repeat exposure to
Norman M, et al. Preterm delivery, level of care, and infant death in antenatal corticosteroids: a randomised controlled trial. Lancet
sweden: a population-based study. Pediatrics 2004;113:1230–5. 2006;367:1913–19.
24 Kallen K, Serenius F, Westgren M, Marsal K, Group E. Impact of 37 McKinlay CJ, Crowther CA, Middleton P, Harding JE. Repeat antenatal
obstetric factors on outcome of extremely preterm births in Sweden: glucocorticoids for women at risk of preterm birth: a Cochrane
prospective population-based observational study (EXPRESS). Acta Systematic Review. Am J Obstet Gynecol 2012;206:187–94.
Obstet Gynecol Scand 2015;94:1203–14. 38 Crowther CA, Anderson PJ, McKinlay CJ, Harding JE, Ashwood PJ, Haslam
25 Bahadue FL, Soll R. Early versus delayed selective surfactant RR, et al. Mid-childhood outcomes of repeat antenatal corticosteroids: a
treatment for neonatal respiratory distress syndrome. Cochrane randomized controlled trial. Pediatrics, 2016;138. DOI: 10.1542/peds.2016-
Database Syst Rev, 2012;11:CD001456. DOI: 10.1002/14651858. 0947 (Epub ahead of print).
CD001456.pub2 39 Stalnacke J, Diaz Heijtz R, Norberg H, Norman M, Smedler AC,
26 Abbasi S, Oxford C, Gerdes J, Sehdev H, Ludmir J. Antenatal Forssberg H. Cognitive outcome in adolescents and young adults
corticosteroids prior to 24 weeks’ gestation and neonatal outcome after repeat courses of antenatal corticosteroids. J Pediat
of extremely low birth weight infants. Am J Perinatol 2010;27:61–6. 2013;163:2–441.
27 Ikegami M, Polk D, Jobe A. Minimum interval from fetal 40 Morley CJ, Davis PG, Doyle LW, Brion LP, Hascoet JM, Carlin JB,
betamethasone treatment to postnatal lung responses in preterm et al. Nasal CPAP or intubation at birth for very preterm infants. N
lambs. Am J Obstet Gynecol 1996;174:1408–13. Engl J Med 2008;358:700–8.