Interpretation
Antenatal corticosteroids for preterm birth have shown to reduce neonatal morbidity and mortality10 are cost-effective85, and are routinely recommended worldwide.7,86,87The Cochrane review,10and our review are aligned with these studies and recommendations.
The other Cochrane review,9 comparing both corticosteroids, did not include recent evidence. The new trial ASTEROID12, almost doubled the number of included women under direct comparisons, provided information for our main outcomes, improved the precision for neurosensory disability estimates, and unlike that it was believed it found that dexamethasone may have a beneficial effect on chorioamnionitis.
The potential beneficial effect of dexamethasone on IVH suggested by very low-CE, was reduced with the inclusion of the ASTEROID trial12 and completely disappeared when excluding Elimian 2007.46 This post-hoc sensitivity analysis, based on the very high risk of attrition bias of this study, provided more consistent results with the indirect evidence. Additionally, a meta-analyses found an increased risk of neurodevelopmental impairment in children with periventricular/intraventricular haemorrhage88, mainly driven by cerebral palsy89. Since we did not find a differential effect of dexamethasone on neurosensory disability it would be unlikely a favourable effect on IVH. Additionally, even if a reduction in IVH was true, it is more important the observed absence of differences on long-term disability for the quality of life of survivors.90
Roberts 200611 assessed indirect estimations favouring betamethasone for chorioamnionitis. This was consistent with our indirect estimation but opposite to the ASTEROID trial12 findings that were considered the most reliable estimation for this outcome.
Our NMA improved the precision and certainty of most previous estimations. We identified a another NMA that evaluated antenatal maternal administration dexamethasone, betamethasone and ambroxol to prevent RDS.91 Compared with placebo, all interventions reduced RDS and neonatal death, but no significant difference in the incidence of bronchopulmonary dysplasia. They also suggest that ambroxol seems to be the most effective treatment for reducing the incidence of RDS and neonatal death based on its SUCRA values. This conclusion was not consistent with a Cochrane review92 or the relevant preterm birth management guidelines.7,86
A wise choice should consider all factors besides evidence, including local availability, costs and cost-utility.93,94A full course of betamethasone costs around US$35 while dexamethasone $1 (3% of the cost of betamethasone).94 The cost-effectiveness of the administration of betamethasone based in individual trials is controversial, and it should be based in the best estimation of effectiveness.95,96Mainly LMICs still have significant challenges to provide safe and effective antenatal corticosteroid use, including ensuring accurate gestational age determination, establishing clear treatment guidelines, strengthening provider capacity, incorporating corticosteroid in national essential medicines lists, and monitoring use and outcomes.97