Abstract
This follow-up to the STOP-BPD randomized clinical trial, which investigated the efficacy and safety of systemic hydrocortisone treatment initiated 7 to 14 days after birth in ventilator-dependent, very preterm infants in the Netherlands and Belgium at 36 weeks’ postmenstrual age, examines the prespecified composite of death and neurodevelopmental impairment among these infants at 2 years’ corrected age.
A randomized, double-blind, placebo-controlled trial investigating the efficacy and safety of systemic hydrocortisone treatment initiated 7 to 14 days after birth in ventilator-dependent, very preterm infants found no significant difference in the composite primary outcome of death or bronchopulmonary dysplasia at 36 weeks’ postmenstrual age.1 However, a preplanned exploratory analysis found a reduced risk of death at 36 weeks’ postmenstrual age in favor of hydrocortisone. We examined the prespecified follow-up of death and neurodevelopmental impairment (NDI) at 2 years’ corrected age.
Methods
The STOP-BPD study was performed in 16 neonatal intensive care units in the Netherlands and Belgium (enrollment November 15, 2011, to December 23, 2016; final follow-up June 27, 2019). Infants born at a gestational age less than 30 weeks and/or with a birth weight less than 1250 g and ventilator dependent between days 7 and 14 were randomly assigned to a 22-day course of systemic hydrocortisone (cumulative dose, 72.5 mg/kg) or placebo. The human research ethics committees approved the protocol and written parental informed consent was obtained.1
The key outcome at 2 years’ corrected age was a composite of death or NDI, with NDI defined as presence of 1 or more of the following: cognitive and/or motor composite score less than 85 on the Bayley Scales of Infant and Toddler Development Third Edition, Dutch version; cerebral palsy greater than level II in the Gross Motor Function Classification System; or hearing or visual impairment.
Crude absolute risk differences (RDs) and odds ratios (ORs) with 95% CIs were calculated for the long-term outcome and its individual components for all infants as randomized. A logistic regression and generalized linear model were used to estimate, respectively, the OR and RD for the composite outcome adjusted for stratification factors (center, gestational age). Potential confounding was examined using multivariable logistic regression models with the preselected risk factors of gestational age and small for gestational age for the composite outcome, supplemented with parental education and multilingual environment for the NDI component, and with respiratory index (product of mean airway pressure and the fraction of inspired oxygen), sex, and multiple birth for the death component.
Missing data of the key long-term outcome were not imputed because there was sufficient information to classify this outcome in more than 95% of the participants.
A 2-sided P < .05 was regarded as statistically significant. Analyses were conducted with SPSS version 26.0 (IBM Corp). The study protocol and statistical analysis plan are available in Supplement 1 and Supplement 2, respectively.
Results
At 2 years’ corrected age, 356 of 371 infants (96%) were evaluated for the outcome death or NDI; 95 infants died, and neurodevelopment assessment was performed in 95% (262/276) of survivors. Baseline characteristics showed a higher parental education level and more multiple births in the hydrocortisone group (Table 1).
Table 1. Characteristics of Surviving Infants (N = 262), Assessed at 2 Years’ Corrected Age, and Their Parents.
| Characteristics | No. (%) | |
|---|---|---|
| Hydrocortisone (n = 133) | Placebo (n = 129) | |
| Parental characteristicsa | ||
| Maternal ethnic originb | ||
| Caucasian | 101 (75.9) | 93 (72.1) |
| Mediterranean | 8 (6.0) | 7 (5.4) |
| African | 10 (7.5) | 17 (13.2) |
| Asian | 4 (3.0) | 7 (5.4) |
| Latin American | 4 (3.0) | 1 (0.8) |
| Unknown | 6 (4.5) | 4 (3.1) |
| Parental education | ||
| Low level | 25 (18.8) | 39 (30.2) |
| Middle level | 38 (28.6) | 39 (30.2) |
| High level | 58 (43.6) | 41 (31.8) |
| Unknown | 12 (9.0) | 10 (7.8) |
| Dutch as main language spoken at home | 119 (89.5) | 105 (82.0) |
| Multilingual environment | 24 (18.2) | 25 (19.4) |
| Any household members smoking at home | 11 (8.3) | 13 (10.1) |
| Infant birth characteristics | ||
| Gestational age at birth, median (IQR), wk | 25.6 (24.9-26.4) | 25.7 (24.7-26.6) |
| Birth weight, median (IQR), g | 790 (660-875) | 720 (653-852) |
| Male sex | 69 (51.9) | 77 (59.7) |
| Small for gestational agec | 17 (12.8) | 19 (14.7) |
| Multiple birth | 52 (39.1) | 35 (27.1) |
| Neonatal morbidities at 36 weeks’ postmenstrual age | ||
| Moderate and severe bronchopulmonary dysplasiad | 86 (64.7) | 82 (63.6) |
| Severe brain injurye | 11 (8.3) | 11 (8.5) |
| Severe retinopathy of prematurity grade >2 | 34 (25.6) | 36 (27.9) |
| Infectionf | 47 (35.3) | 53 (41.1) |
Abbreviation: IQR, interquartile range.
Parental characteristics at baseline are at the infant level (ie, parents were counted multiple times if they had multiple infants).
Because racial and ethnic background has been associated with the risk of death, these data were collected as reported by attending physicians based on fixed categories. The ethnic group terms listed in this table are the exact terms that were used during the study for data collection.
Defined as less than the 10th percentile on the Fenton growth chart.
Defined according to the National Institutes of Health definition published in 2001.
Includes infants with intraventricular hemorrhage greater than grade 2, cystic periventricular leukomalacia, and posthemorrhagic ventricular dilation during the neonatal period.
Includes infants with culture-proven sepsis and necrotizing enterocolitis greater than stage 2a according to the Bell classification during the neonatal period.
Occurrence of death or NDI was 56.7% (97/171) in the hydrocortisone group and 62.7% (116/185) in the placebo group (adjusted RD, –5.2% [95% CI, –15.1% to 4.6%]; adjusted OR, 0.79 [95% CI, 0.51 to 1.22]; P = .28). The rate of death was 21.5% (39/181) in the hydrocortisone group and 29.5% (56/190) in the placebo group (RD, –7.9% [95% CI, –16.6% to 1.0%]; OR, 0.66 [95% CI, 0.41 to 1.05]; P = .08). NDI occurred in 43.9% (58/132) of the hydrocortisone group and 46.5% (60/129) of the placebo group (RD, –2.6% [95% CI, –14.4% to 9.4%]; OR, 0.90 [95% CI, 0.55 to 1.47]; P = .68). The individual NDI domains were not significantly different between the groups (Table 2). Adjustment for preselected risk factors did not change these findings.
Table 2. Composite Key Long-term Outcome and Its Individual Components at 2 Years’ Corrected Age.
| No./total (%) | Difference, % (95% CI)a | Odds ratio (95% CI)b | P value | ||
|---|---|---|---|---|---|
| Hydrocortisone | Placebo | ||||
| Composite key long-term outcome (n = 356) | |||||
| Death or neurodevelopmental impairment at 2 years’ corrected age | 97/171 (56.7) | 116/185 (62.7) | |||
| Crude analysis | –6.0 (–16.0 to 4.2) | 0.78 (0.51-1.19) | |||
| Adjusted analysisc | –5.2 (–15.1 to 4.6) | 0.79 (0.51-1.22) | .28 | ||
| Adjusted analysis for preselected risk factorsd | 0.82 (0.53-1.26) | .36 | |||
| Components of key long-term outcome | |||||
| Death at 2 years’ corrected age (n = 371) | 39/181 (21.5) | 56/190 (29.5) | |||
| Crude analysis | –7.9 (–16.6 to 1.0) | 0.66 (0.41-1.05) | .08 | ||
| Adjusted analysis for preselected risk factorse | 0.63 (0.39-1.04) | .07 | |||
| Neurodevelopmental impairment at 2 years’ corrected age in survivors (n = 261) | 58/132 (43.9) | 60/129 (46.5) | |||
| Crude analysis | –2.6 (–14.4 to 9.4) | 0.90 (0.55-1.47) | .68 | ||
| Adjusted analysis for preselected risk factorsf | 1.07 (0.63-1.82) | .81 | |||
| Components of neurodevelopmental impairment | |||||
| Bayley-III-NLg | |||||
| Cognitive composite score <85 | 31/112 (27.7) | 38/111 (34.2) | –6.6 (–18.4 to 5.5) | 0.74 (0.42-1.30) | .29 |
| Mean score (SD) | 93.4 (14.4) | 93.3 (16.8) | 0.04 (–4.3 to 4.4)h | .99i | |
| Motor composite score <85 | 23/103 (22.3) | 29/99 (29.3) | –7.0 (–18.8 to 5.1) | 0.69 (0.37-1.31) | .26 |
| Mean score (SD) | 95.0 (16.1) | 92.4 (16.6) | 2.6 (–1.7 to 6.9)h | .23i | |
| Neurologic | |||||
| Cerebral palsy with gross motor function level >II | 2/129 (1.6) | 4/126 (3.2) | –1.6 (–6.5 to 2.8) | .44j | |
| Sensory | |||||
| Severe visual impairment | 3/128 (2.3) | 7/126 (5.6) | –3.2 (–8.9 to 2.0) | .22j | |
| Hearing loss requiring hearing aids or deafness | 2/130 (1.5) | 3/124 (2.4) | –0.9 (–5.5 to 3.3) | .68j | |
Abbreviation: Bayley-III-NL, Bayley Scales of Infant and Toddler Development, Third Edition, Dutch version.
Data are percentages unless otherwise indicated. Crude data are reported unless otherwise indicated.
Logistic regression analysis. Crude data are reported unless otherwise indicated.
For the key long-term outcome, absolute risk reduction and odds ratio are adjusted for the randomization stratification variables of gestational age and study center using a logistic regression model.
Adjusted for the preselected risk factors of gestational age (<27 weeks, ≥27 weeks) and small for gestational age (yes, no) using a multivariable logistic regression model.
Adjusted for the preselected risk factors of gestational age (<27 weeks, ≥27 weeks), small for gestational age (yes, no), respiratory index at randomization (<median, ≥median), sex (male, female), and multiple birth (yes, no) using a multivariable logistic regression model.
Adjusted for the preselected risk factors of gestational age (<27 weeks, ≥27 weeks), small for gestational age (yes, no), parental education (low, middle, and high), and multilingual environment (yes, no) using a multivariable logistic regression model. This analysis is performed despite missing data in the preselected risk factor of parental education for 22 (8.4%) of 261 infants.
Normed mean of Bayley-III-NL of 100 and an SD of 15.
Crude mean difference.
Linear regression analysis.
Fisher exact test when fewer than 20 (non)events.
Discussion
This placebo-controlled trial found no significant difference in the composite outcome death or NDI at 2 years’ corrected age in ventilated very preterm infants treated with systemic hydrocortisone initiated in the second week of life to reduce the risk of death or bronchopulmonary dysplasia. This finding is consistent with randomized trials of prophylactic hydrocortisone treatment.2,3 The effect size of the reduction in mortality in favor of hydrocortisone at 2-year follow-up was similar to the previous report at 36 weeks’ postmenstrual age (OR, 0.59) but no longer reached statistical significance.
The high proportion of open-label hydrocortisone treatment in the placebo group may have diluted a possible effect of hydrocortisone, although the modulating effect of open-label steroids on long-term outcome is probably limited.4 The possible beneficial hydrocortisone effect on survival needs further investigation.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
Original Protocol and Amendments for the STOP-BPD Study
Statistical Analysis Plan
Nonauthor Collaborators
Data Sharing Statement
References
- 1.Onland W, Cools F, Kroon A, et al. ; STOP-BPD Study Group . Effect of hydrocortisone therapy initiated 7 to 14 days after birth on mortality or bronchopulmonary dysplasia among very preterm infants receiving mechanical ventilation: a randomized clinical trial. JAMA. 2019;321(4):354-363. doi: 10.1001/jama.2018.21443 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Baud O, Trousson C, Biran V, Leroy E, Mohamed D, Alberti C; PREMILOC Trial Group . Association between early low-dose hydrocortisone therapy in extremely preterm neonates and neurodevelopmental outcomes at 2 years of age. JAMA. 2017;317(13):1329-1337. doi: 10.1001/jama.2017.2692 [DOI] [PubMed] [Google Scholar]
- 3.Doyle LW, Ehrenkranz RA, Halliday HL. Postnatal hydrocortisone for preventing or treating bronchopulmonary dysplasia in preterm infants: a systematic review. Neonatology. 2010;98(2):111-117. doi: 10.1159/000279992 [DOI] [PubMed] [Google Scholar]
- 4.Onland W, van Kaam AH, De Jaegere AP, Offringa M. Open-label glucocorticoids modulate dexamethasone trial results in preterm infants. Pediatrics. 2010;126(4):e954-e964. doi: 10.1542/peds.2010-0597 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Original Protocol and Amendments for the STOP-BPD Study
Statistical Analysis Plan
Nonauthor Collaborators
Data Sharing Statement