Background: Antenatal betamethasone is recommended before preterm delivery to accelerate fetal lung maturation. However, reports of growth and neurodevelopmental dose-related side-effects suggest that the current dose (12 mg plus 12 mg, 24 h apart) might be too high. We therefore investigated whether a half dose would be non-inferior to the current full dose for preventing respiratory distress syndrome.

Methods: We designed a randomised, multicentre, double-blind, placebo-controlled, non-inferiority trial in 37 level 3 referral perinatal centres in France. Eligible participants were pregnant women aged 18 years or older with a singleton fetus at risk of preterm delivery and already treated with the first injection of antenatal betamethasone (11·4 mg) before 32 weeks' gestation. We used a computer-generated code producing permuted blocks of varying sizes to randomly assign (1:1) women to receive either a placebo (half-dose group) or a second 11·4 mg betamethasone injection (full-dose group) 24 h later. Randomisation was stratified by gestational age (before or after 28 weeks). Participants, clinicians, and study staff were masked to the treatment allocation. The primary outcome was the need for exogenous intratracheal surfactant within 48 h after birth. Non-inferiority would be shown if the higher limit of the 95% CI for the between-group difference between the half-dose and full-dose groups in the primary endpoint was less than 4 percentage points (corresponding to a maximum relative risk of 1·20). Four interim analyses monitoring the primary and the secondary safety outcomes were done during the study period, using a sequential data analysis method that provided futility and non-inferiority stopping rules and checked for type I and II errors. Interim analyses were done in the intention-to-treat population. This trial was registered with ClinicalTrials.gov, NCT02897076.

Findings: Between Jan 2, 2017, and Oct 9, 2019, 3244 women were randomly assigned to the half-dose (n=1620 [49·9%]) or the full-dose group (n=1624 [50·1%]); 48 women withdrew consent, 30 fetuses were stillborn, 16 neonates were lost to follow-up, and 9 neonates died before evaluation, so that 3141 neonates remained for analysis. In the intention-to-treat analysis, the primary outcome occurred in 313 (20·0%) of 1567 neonates in the half-dose group and 276 (17·5%) of 1574 neonates in the full-dose group (risk difference 2·4%, 95% CI -0·3 to 5·2); thus non-inferiority was not shown. The per-protocol analysis also did not show non-inferiority (risk difference 2·2%, 95% CI -0·6 to 5·1). No between-group differences appeared in the rates of neonatal death, grade 3-4 intraventricular haemorrhage, stage ≥2 necrotising enterocolitis, severe retinopathy of prematurity, or bronchopulmonary dysplasia.

Interpretation: Because non-inferiority of the half-dose compared with the full-dose regimen was not shown, our results do not support practice changes towards antenatal betamethasone dose reduction.