Association of Antenatal Corticosteroids With Mortality and Neurodevelopmental Outcomes Among Infants Born at 22 to 25 Weeks' Gestation FREE

Data from large neonatal databases document that infants born around the limits of viability have high mortality and morbidity rates,^1- 3 which have not improved during recent years.⁴ Broad application of interventions proven to be effective at more advanced gestational ages, such as antenatal corticosteroids, may improve outcomes in these infants. A single course of antenatal corticosteroids given 24 hours to 7 days before birth to women in preterm labor at less than 34 weeks' gestation improves lung maturity and reduces neonatal problems, including respiratory distress syndrome, necrotizing enterocolitis, severe intraventricular hemorrhage, and death.⁵

However, there are limited data from high-quality studies on the effects of antenatal corticosteroids at 22 to 25 weeks' gestation. Two recent meta-analyses confirmed that published data do not demonstrate that antenatal corticosteroids improve outcomes at less than 26 weeks' gestation.^5- 6 However, the small number of infants born at gestational ages around the limits of viability and included in published trials limits these results.^5- 6 Furthermore, antenatal corticosteroid trials include almost no data on neurodevelopmental outcomes even though these immature infants are at high risk of severe neurodevelopmental impairment.⁷

In 1995, National Institutes of Health Consensus Conference⁸ participants concluded that “Because data from randomized controlled trials did not include infants below 24 weeks, the current recommendations are to administer antenatal corticosteroids as early as but not before 24 weeks.” The recent opinion of the Committee on Obstetric Practice of the American College of Obstetrics and Gynecologists⁹ states that sparse data exist on the efficacy of corticosteroid use “before the age of viability,” and thus such use before 24 weeks' gestation was not recommended. Similar recommendations are used worldwide.^10- 11 The limited data on the effectiveness of antenatal corticosteroids in very preterm deliveries has led to restriction in their use^{1,4,10,12- 13} and international¹⁰ and regional¹ practice variability, even though infants as immature as 22 and 23 weeks' gestation frequently receive intensive care.

This study was designed to determine if antenatal corticosteroid exposure in infants born at each gestational age from 22 to 25 weeks is associated with improvement in important outcomes, including the primary outcome of death or childhood neurodevelopmental impairment, using a large cohort of infants.

Infants born at any of the 23 National Institute of Child Health and Human Development Neonatal Research Network centers between January 1, 1993, and December 31, 2009, were included if they were 22 to 25 weeks' gestation with a birth weight between 401 g and 1000 g. Gestational age was determined according to standard Neonatal Research Network definitions using the hierarchy of best obstetrical estimate¹⁴ over the best neonatal data estimate. Infants with congenital anomalies were included. Infants who died during the first 12 hours after birth without receiving delivery room resuscitation (no ventilation, intubation, or medications) were excluded from the primary analysis to make sure the results were not affected by postnatal restriction of care. Infants who died in the delivery room were included in a secondary analysis.

Infants were categorized as having been exposed to antenatal corticosteroids if their mother received 1 or more doses of dexamethasone or betamethasone or not exposed if their mother did not receive antenatal corticosteroids. Data on full or partial course of antenatal corticosteroids were collected but data on timing and dose were not available.

Trained research coordinators prospectively collected maternal and neonatal data according to common definitions.¹ Sociodemographic and clinical data were collected until death, discharge home, or 120 days, whichever occurred first, including transfers to other hospitals, using hospital chart reviews. Race/ethnicity, which can alter the response to antenatal steroids, was recorded as self-selected by the mother from options defined by federally funded study guidelines. Maternal education level and insurance data were available except for children born between 1993 and 2001 and those who were not followed up. Data on language spoken were available only on infants who were followed up. Data on type of corticosteroid were collected starting in 2002. Chorioamnionitis data were collected starting in 2006.

All survivors were scheduled at 18 to 22 months' corrected age for a standardized comprehensive neurodevelopmental assessment that was performed by certified examiners,¹⁵ who were unaware of exposure to antenatal corticosteroids. Only neurodevelopmental assessments on infants born between 2003 and 2008 were included in this study.

This study was a secondary analysis of the Neonatal Research Network Hospital Generic Database and Follow-up protocols, which were approved by each center's institutional review board. Written informed consent was obtained for the follow-up protocol. In 3 centers, written or oral informed consent was obtained for the hospital database protocol, while in the other centers, waiver of consent was approved by the institutional review boards.

Hospital death was defined as death of an infant before discharge or by 120 days in those with longer initial hospitalizations. Mortality before follow-up was documented at 18 to 22 months' corrected age.¹⁶ Bronchopulmonary dysplasia was defined as continuous use of supplemental oxygen at 36 weeks' postmenstrual age. The physiological definition for bronchopulmonary dysplasia¹⁷ was not used. Intraventricular hemorrhage grade 3-4 was determined according to the classification by Papile et al.¹⁸ For infants born up to 2005, severe neurodevelopmental impairment at 18 to 22 months' corrected age was defined as 1 or more of the following: a Bayley II Mental Developmental Index score of less than 70 (>2 SDs below the mean for normal infants), a Bayley II Psychomotor Developmental Index score of less than 70, moderate to severe cerebral palsy, blindness (no useful vision in either eye), or deafness (functional hearing impairment with aids in both ears).¹⁵ For infants born after 2005, the Bayley III was used instead, and severe neurodevelopmental impairment was defined as 1 or more of the following: Bayley III cognitive composite score of less than 70, gross motor function level of 2 or greater, blindness (some or no useful vision in either eye), or deafness (functional hearing impairment).

Outcomes were analyzed by exposure or nonexposure to antenatal corticosteroids. The primary outcome measure was death or neurodevelopmental impairment at follow-up (18-22 months). Secondary outcome measures and subgroup analyses were prespecified. Secondary outcomes at hospital discharge included death, bronchopulmonary dysplasia, intraventricular hemorrhage grade 3-4 or periventricular leukomalacia (or both), necrotizing enterocolitis, and death plus each morbidity. Secondary outcomes at 18 to 22 months' corrected age included death, neurodevelopmental impairment, and the individual components of neurodevelopmental impairment. Subgroups were selected as potential effect modifiers based on published meta-analyses.^5- 6 Missing data were excluded; imputation was not used. Differences in categorical measures were analyzed with χ² analyses. Within each week of gestational age, logistic regression models were used to estimate the relationship between antenatal corticosteroid administration and outcomes with adjustments made for maternal variables (age, marital status, race, diabetes, hypertension or preeclampsia, rupture of membranes >24 hours, antepartum hemorrhage, and delivery mode), multiple birth, sex, and center. A cohort effect was included in the follow-up data models to indicate Bayley II vs Bayley III assessments. To allow for changes in neonatal care over time and to assess whether the effect of antenatal corticosteroids was stable over time, additional logistic regression models were run adding adjustment for epoch (1993-1998, 1999-2003, 2004-2009) and statistical interaction between antenatal corticosteroids and epoch.

The study was designed to use all the data available in the Neonatal Research Network database to explore the association between antenatal corticosteroid use and outcomes. Formal power analysis and sample size estimates were not performed. SAS software version 9.2 (SAS Institute Inc) was used for all statistical analyses. Odds ratios (ORs) and 95% confidence intervals were estimated for binary outcomes. Two-sided P values of less than .05 indicated statistical significance. Adjustments for multiple comparisons were not done. A Kaplan-Meier survival analysis for age at death was performed. Interaction terms between antenatal corticosteroid use and gestational age were added to the models fitted to the entire data set to examine differences in the effect of antenatal corticosteroids by gestational age.

A total of 1848 infants (941 born at 22 weeks’ gestation, 684 at 23 weeks, 179 at 24 weeks, and 44 at 25 weeks) died in the first 12 hours after birth without receiving delivery room resuscitation and were excluded from the primary analyses. The study population consisted of 10 541 infants (range, 63-971 infants per center), of whom 7808 (74.1%) were born to mothers who received antenatal corticosteroids (Table 1). Of the 5691 infants born between 1993 and 2008 who survived to 18 to 22 months, 4924 (86.5%) had neurodevelopmental assessments. The highest percentage of missing data for hospitalization outcomes was for bronchopulmonary dysplasia (4.0%). For the primary outcome, the percentage of missing data was 10.3%. There was no difference in the birth weight of infants whose mothers received antenatal corticosteroids and in those whose mothers did not receive them but there were other baseline differences; the women who did not receive corticosteroids were more likely to be black, be 19 years old or younger, have a lower income, and have Medicaid coverage and were less likely to be married, have completed high school, or have a cesarean delivery (Table 1). The percentage of infants whose mothers received antenatal corticosteroids increased in the early to mid-1990s as research on antenatal corticosteroids in early gestation became known and the 1995 National Institutes of Health consensus was published. This occurred in all 4 gestational age subgroups but remained lower for the infants born at 22 and 23 weeks' gestation (Figure).

For the cohort as a whole, of the hospitalization outcomes, death (35.5% with exposure to antenatal corticosteroids vs 56.0% without exposure; adjusted OR [AOR], 0.58 [95% CI, 0.52-0.65]), intraventricular hemorrhage grade 3-4 or periventricular leukomalacia (or both) (19.2% with exposure to antenatal corticosteroids vs 27.6% without exposure; AOR, 0.67 [95% CI, 0.57-0.79]), and many of the composite outcomes were significantly lower in the infants whose mothers had received antenatal corticosteroids (Table 2). The lower hospital death rate was partially offset by a higher rate of bronchopulmonary dysplasia (60.3% with exposure to antenatal corticosteroids vs 54.0% without exposure; AOR, 1.43 [95% CI, 1.23-1.67]). Hospital death was lower in infants exposed to antenatal corticosteroids who were born at 23, 24, and 25 weeks' gestation (Table 2). The AOR for reduction in hospital death in the infants born at 22 weeks' gestation was in the range of those born at other weeks but the sample size was smaller and the confidence level wider, overlapping 1.0. The Kaplan-Meier survival analyses revealed that there was a significant association between exposure to antenatal corticosteroids and survival (eFigure). Death, intraventricular hemorrhage grade 3-4, or periventricular leukomalacia (or any of the 3) also was lower in those exposed to antenatal corticosteroids who were born at 23, 24, and 25 weeks' gestation but was not significantly lower in those born at 22 weeks' gestation. The rate of death or bronchopulmonary dysplasia (or both) was lower in the infants born at 23 weeks' gestation but bronchopulmonary dysplasia in survivors was higher in infants born at 24 and 25 weeks' gestation. Death or necrotizing enterocolitis was lower at all gestational ages, including at 22 weeks' gestation.

At the 18- to 22-month follow-up, the primary outcome of death or neurodevelopmental impairment was less frequent in those exposed to antenatal corticosteroids (AOR, 0.60; 95% CI, 0.53-0.69) as well as death (37.0% with exposure to antenatal corticosteroids vs 57.2% without exposure; AOR, 0.59 [95% CI, 0.53-0.65]), neurodevelopmental impairment (38.8% with exposure to antenatal corticosteroids vs 49.2% without exposure; AOR, 0.83 [95% CI, 0.70-0.99]), moderate to severe cerebral palsy (8.6% with exposure to antenatal corticosteroids vs 12.0% without exposure; AOR, 0.76 [95% CI, 0.59-0.98]), and psychomotor developmental index score of less than 70 (27.3% with exposure to antenatal corticosteroids vs 32.8% without exposure; AOR, 0.79 [95% CI, 0.65-0.96]) (Table 3). Intact survival (no death or neurodevelopmental impairment by follow-up) at 18 to 22 months was higher in infants whose mothers received antenatal corticosteroids (35.8% with exposure to antenatal corticosteroids vs 18.5% without exposure; AOR, 1.66 [95% CI, 1.46-1.90]). Death or neurodevelopmental impairment was less frequent in those who had been exposed to antenatal corticosteroids and were born at 23 weeks' gestation (83.4% with exposure to antenatal corticosteroids vs 90.5% without exposure; AOR, 0.58 [95% CI, 0.42-0.80]), at 24 weeks' gestation (68.4% with exposure to antenatal corticosteroids vs 80.3% without exposure; AOR, 0.62 [95% CI, 0.49-0.78]), and at 25 weeks' gestation (52.7% with exposure to antenatal corticosteroids vs 67.9% without exposure; AOR, 0.61 [95% CI, 0.50-0.74]) but not in those born at 22 weeks' gestation (90.2% with exposure to antenatal corticosteroids vs 93.1% without exposure; AOR, 0.80 [95% CI, 0.29-2.21]) (Table 3). Interaction analyses performed for both hospital and follow-up outcomes provided no evidence that the associations of antenatal corticosteroids with outcomes differed across the 23, 24, and 25 weeks' gestational age groups.

Subgroup analyses indicated that exposure to antenatal corticosteroids was associated with lower hospital mortality, lower mortality at 18 to 22 months, and lower mortality or neurodevelopmental impairment at 18 to 22 months in singleton and multiple births, partial and full antenatal corticosteroid treatment groups, betamethasone and dexamethasone treatment groups, infants of mothers with and without diabetes, all durations of rupture of membrane subgroups, infants of mothers with and without antepartum hemorrhage, those delivered vaginally and by cesarean section, males and females, infants not small for gestational age, and all racial/ethnic subgroups (Table 4; eTable 1 and eTable 2). These outcomes did not significantly differ by exposure to antenatal corticosteroids in the infants born small for gestational age. In infants born to mothers with hypertension or preeclampsia or eclampsia, exposure to antenatal corticosteroids was associated with lower hospital death and death by 18 to 22 months but there was no difference in death or neurodevelopmental impairment by 18 to 22 months. Neither epoch nor interaction between antenatal corticosteroids and epoch was significant in the models for the primary outcome of death or neurodevelopmental impairment, indicating the robustness of the results with regard to any temporal trends (eTable 3 and eTable 4).

The association of exposure to antenatal corticosteroids and lower mortality was even stronger when the 1848 early deaths during the first 12 hours after birth without receiving delivery room resuscitation were included in the analysis (eTable 5). In this analysis, exposure to antenatal corticosteroids in infants born at 22, 23, 24, and 25 weeks' gestation was associated with a statistically significant lower rate of death.

This multicenter observational study cohort is larger than all other reported cohorts of infants born at 22 to 25 weeks’ gestation combined and documents that exposure to antenatal corticosteroids was associated with lower mortality or neurodevelopmental impairment at 18 to 22 months in infants born at each week from 23 to 25 weeks’ gestation, even after adjustment for multiple potential confounders. A significant association was not found at 22 weeks' gestation; however, the 95% confidence intervals were wide and the power was limited. However, even though intact survival doubled with the administration of antenatal steroids in the entire cohort, it remained relatively low (36%).

The outcome of neurodevelopmental impairment and death after discharge in infants is unique because most previous studies have focused on hospital outcomes.

Multiple secondary outcomes measured during hospitalization and at 18 to 22 months also were significant, and no evidence was found to suggest that these associations with antenatal corticosteroids differed across the range from 23 through 25 weeks' gestation. Subgroup analyses indicated that these associations were significant in all subgroups at this early gestational age range, except in infants small for gestational age and in infants of mothers with hypertension, preeclampsia, or eclampsia. The lower mortality rate during hospitalization in infants exposed to antenatal corticosteroids was partially offset by a higher rate of bronchopulmonary dysplasia at the early gestational ages as reported by others,⁶ which may be due to the high rate of bronchopulmonary dysplasia in infants who would be expected to die if they were not exposed to antenatal corticosteroids.

These data addressed the most important neonatal outcomes using a large prospective database with a high follow-up rate and adjusted for multiple maternal and neonatal variables. However, there are limitations of the study that should be noted. Potential biases include that mothers who received antenatal corticosteroids and their infants were different on a number of characteristics than those who did not. Adjustments were made using logistic regression models for many maternal, infant, and perinatal center variables but not for all variables. In addition, we did not have data on some potential confounders such as whether women presenting late in labor who would not have been eligible for antenatal corticosteroids were more commonly in the group that did not receive them. It is unlikely that the results are due to confounding, but it is possible that there is some residual or unmeasured bias in the results due to baseline differences between the study groups.

Another limitation is that other aspects of obstetrical care are likely to affect neonatal survival in pregnancies around the limits of viability. Willingness to perform a cesarean section for fetal indications was associated with a lower rate of neonatal mortality in a multicenter prospective observational study that included 713 singleton infants with birth weights of 1000 g or less.¹⁹ We cannot exclude the possibility that willingness to perform a cesarean section or use of other therapies may have influenced outcomes. The cesarean section rate was higher in the group exposed to antenatal corticosteroids than in the group without exposure. However, it should be noted that more than one-third of the infants in the group without exposure to antenatal corticosteroids were delivered by cesarean section.

The possibility of postnatal treatment bias based on antenatal corticosteroid administration (eg, withholding postnatal therapies in infants whose mothers did not receive antenatal corticosteroids) cannot be excluded. To mitigate this effect, infants not receiving resuscitation who died within the first 12 hours after birth were excluded, but otherwise all live births were entered into the database, including infants who died in the delivery room. Nonetheless, obstetrical and neonatal care was provided at the discretion of clinicians and parents, and the frequency and intensity with which such care was applied varied among and within centers.

Because infants with birth weights of 400 g or less were excluded from the database, not all infants born at 22 to 25 weeks' gestation are included in this study. In addition, the primary outcome could not be determined in 10.3% of the infants, raising some concerns about validity. Data were not collected on the timing or dose of antenatal corticosteroid treatment, fetal monitoring, or length of maternal hospitalization before delivery so these could not be studied. Data were not collected on maternal complications but the large meta-analysis⁵ did not report important maternal adverse effects other than glucose intolerance in 1 study.

Data from randomized controlled trials of exposure to antenatal corticosteroids in infants born at less than 26 weeks' gestation are limited, and meta-analyses do not support or refute the benefits of exposure to antenatal corticosteroids at these early gestational ages.^5- 6 However, several reports of large series of infants exposed to antenatal corticosteroids and born at gestational ages around the limits of viability describe benefits. Treatment with antenatal corticosteroids was associated with reduced mortality to hospital discharge among 753 infants born at 22 to 23 weeks' gestation from 87 hospitals in Japan.²⁰ However, the major neonatal morbidities were not affected and the data were only analyzed for the combined weeks. A population-based series of 705 infants born at less than 26 weeks' gestation showed that exposure to antenatal corticosteroids was associated with reductions in hospital mortality,²¹ severe motor disability, and low scores on the Mental Developmental Index.²² A study of 181 infants born at 23 weeks' gestation in 3 hospitals over a 10-year period reported reductions in mortality only for the infants exposed to a full course of antenatal corticosteroids.²³ A study of 117 singleton infants exposed to a complete course of antenatal corticosteroids at less than 24 weeks' gestation and born at 23 to 25 weeks' gestation in a single hospital over a 10-year period revealed that corticosteroid administration was associated with reduction in mortality and severe intracranial hemorrhage.²⁴ Other large multicenter studies have shown that antenatal corticosteroids were associated with decreased hospital mortality^10,16,25 and severe disability^13,26- 27 in very preterm infants but have not focused the analysis on those born before 24 weeks' gestation. These patient series show that antenatal corticosteroids are associated with variable reductions in mortality and morbidities.

The current study showed large reductions in mortality with a multicenter database and extends the findings to include reductions in important in-hospital and follow-up morbidities. Furthermore, the reduction in adverse outcomes was observed in almost all patient subgroups. Thus, it is likely that the results of this study are generalizable to a broad population. Even though concern has been expressed about repeated courses of antenatal corticosteroids if started early during pregnancy,⁷ 2 recent trials of repeated courses demonstrated no increased risk for neurodevelopmental impairment.^28- 29 Thus, concerns about repeated courses of antenatal steroids should not preclude the initial administration of antenatal steroids at a gestational age around the limits of viability.

In summary, antenatal corticosteroid therapy for mothers of infants born at 23, 24, and 25 weeks' gestation was associated with lower rates of both mortality and important morbidities, including a lower rate of death or severe neurodevelopmental impairment at 18 to 22 months' corrected age. These benefits were observed across subgroups without evidence that they differed across the ranges of gestational ages around the limits of viability. Despite their potential to improve outcomes, the administration of antenatal corticosteroids is not increasing at gestational ages around the limits of viability⁴ and remains substantially lower than at later gestational ages. Controlled trials could be performed to precisely determine the benefits of antenatal corticosteroids when administered this early but such trials will be difficult to perform. Initiation of antenatal corticosteroids may be considered starting at 23 weeks' gestation and later if the infant will be given intensive care because this therapy is associated with reduced mortality and morbidity.