Association between gestational age at birth, antenatal corticosteroids, and outcomes at 5 years: multiple courses of antenatal corticosteroids for preterm birth study at 5 years of age (MACS-5)

Women were enrolled in the MACS if they were between 25 and 32 weeks of gestation, remained pregnant 14 to 21 days after an initial course of antenatal corticosteroid therapy (either betamethasone or dexamethasone) therapy, and continued to be viewed as at high risk of preterm birth by their clinicians. Women were not eligible if they had a contraindication to corticosteroid use, needed chronic treatment with these drugs, demonstrated evidence of chorioamnionitis, carried a fetus with a known lethal congenital anomaly, had received an initial course of prenatal corticosteroid therapy before 23 weeks of gestation, or had participated previously in the MACS trial. In multiple-fetus pregnancies, if a fetus was thought to have died prior to 13 weeks, that fetus was not considered part of the pregnancy for the purposes of this study. Women assigned to the multiple courses arm received two doses of 12 mg betamethasone intramuscularly 24 hours apart; those assigned to the single course arm received a similarly appearing placebo injection. The study medication was given every two weeks until 33 weeks of gestation or birth, whichever occurred first. Randomization took place from April 9, 2001 to August 31, 2006.

All children alive at 5 years of age underwent the 5-year assessment which included a neurologic assessment to determine the presence of cerebral palsy and any hearing/visual difficulties, and the completion of two parent questionnaires. The 5-year study and all secondary analyses were approved by the Research Ethics Board at the Sunnybrook Health Sciences Centre. The institutions were encouraged to contact the families of all surviving children even if no contact had been made at 18–24 months of age for the 18–24 month follow-up assessment. The target date for the visit was the child’s 5th birthday; completing the assessments within 4 months of the target date was encouraged but efforts to locate and assess the children continued beyond this age when necessary. Following ethics approval at each institution, written informed consent was obtained from a parent or guardian. The 5-year follow-up began in June 2006 and was complete by May 2012.The primary outcome of the 5-year follow-up was a composite of death or survival with a neurodevelopmental disability in at least one of the following domains: neuromotor (non-ambulatory cerebral palsy), neurosensory (blindness, deafness or a need for visual or hearing aids), or neurocognitive/neurobehavioral (abnormal attention, memory or behaviour) function. Non-ambulatory cerebral palsy was present if the child had a non-progressive motor impairment characterized by abnormal muscle tone and decreased range of movements, with a gross motor function of 3–5 as defined in the Gross Motor Function Classification System (GMFCS) [4]. Neurosensory disability was defined as blindness, deafness, or need for visual or hearing aids based on local criteria. Neurocognitive/neurobehavioral disability was defined as an abnormally elevated score (>1.5standard deviations above the normative control sample) on either one of two parent-administered questionnaires: the Behavior Rating Inventory of Executive Function- Preschool version (BRIEF-P) and the Child Behavior Checklist-1½-5 (CBCL-1½-5) [5, 6].

The primary outcome and its components were analysed using a general linear model for a binary response (i.e. logistic regression) with repeated measures for children from the same pregnancy. The model included ACS treatment group (multiple, single), gestational age at birth group (<30, 30–36, ≥37 weeks), plus its interaction with ACS treatment group. The model included the stratification variable gestational age at randomization (26–27 vs. 28+ weeks) and, if the variables were significant at the two-sided level of 0.1, the following covariates from Table 1: preterm pre-labour rupture of membranes (yes/no), multiple pregnancy (yes/no), maternal smoking (yes/no), parity (0, 1+), sex of the infant and the country’s perinatal mortality rate (<=10/1000, >10-20/1000, >20/1000). Because there were very few neuromotor outcomes, the model would not converge and no results for this component of the primary outcome are given. Model estimation was facilitated by generalized estimating equations, using PROC GENMOD in SAS. Odds ratios plus 95% confidence intervals were determined. This research adhered to the STROBE guidelines for observational studies.

Figure 1 illustrates the study group for MACS-5. For the main 5-year follow-up, 413 children were unable to be followed because they could not be located or parents declined participation, 1 child lost for the neonatal and 2 year follow-up phase was found, leaving 1728 children (80.7% of the 2141 eligible children) to contribute to the outcomes of the 5-year follow-up. Nine children had insufficient information to contribute to the primary outcome. In total, 1719 children contributed to the primary outcome of MACS-5 and it is these children, broken down into groupings of <30, 30–36, and ≥ 37 weeks gestation at birth, who contribute to this secondary analysis. The median age for the 5 year assessments was 5.2 years.

The baseline maternal characteristics and co-interventions after randomization are outlined in Table 1 by gestational ages at birth. A summary of the neonatal, 18–24 months and 5-year neurodevelopmental outcomes of the study cohort are outlined in Table 2. The baseline neonatal outcomes of the infants followed compared to those not followed are outlined in the Additional file 1: Table S1.

The primary outcome and its components at 5 years of age are given in Table 3 by ACS treatment groups and gestational age at birth. The covariate results for the primary outcome and its components are given in Table 4, and include the odds ratios, confidence intervals and levels of significance. The odds ratios, confidence limits and significance levels for the primary outcome and its components are given by gestational age at birth in Tables 5 and 6; the results were similar when the covariates were omitted.

Gestational age at birth was strongly associated with primary outcome as defined for 5 years of age, p <0.001. The interaction between ACS group and gestational age at birth was not significant, p =0.064 (Table 5). In the two preterm groups, there was no difference seen in the primary outcome or its components between ACS groups. However, in children born at term, there was a statistically significant increase in the risk of the primary outcome in the multiple ACS group compared to the single ACS group, 48/213 (22.5%) vs. 38/249 (15.3%); OR = 1.69 [95% CI: 1.04, 2.77]; p = 0.037 (Table 5). In addition, there was a significant interaction between ACS group and gestational age at birth for neurosensory disability (p = 0.021), and for the children born at term there was a significant increase in the risk of neurosensory disability in the multiple ACS group: 19/212 (9.0%) vs. 7/247 (2.8%); OR = 3.70 [95% CI: 1.57, 8.75]; p = 0.004. There was no interaction between ACS group and gestational age at birth with respect to neurocognitive/neurobehavioral disability or death.