Objective: To measure the extent to which the evidence underpinning prenatal corticosteroids for preterm fetal lung maturity is at risk of bias due to selective outcome reporting.
Study design: Five biomedical databases (Medline, ClinicalTrials.gov, Embase, OVID and HMIC) were searched for trials published between February 2016 and September 2017. Randomised trials of prenatal steroids for women at risk of preterm birth were identified. For each, we recorded the registration status and timing, and whether the registered primary outcome had been reported. For unregistered trials, we estimated the potential for selective outcome reporting indirectly by tabulating all reported outcomes and by comparing the first outcome reported in the methods to the first in the results section.
Results: Twenty seven trials were identified. Only three (11%) trials had been registered. All three reported their pre-specified primary outcome. Among the unregistered trials, thirteen (54%) had different first reported outcomes in the methods and results sections, and among all trials many different outcomes were reported suggesting considerable potential for selective reporting. However, the single outcome of respiratory distress syndrome, albeit defined in different ways, was reported in all but two trials. This is reassuring evidence that the beneficial effect of steroids on this outcome has not been exaggerated by selective outcome reporting.
Conclusion: We conclude that the evidence that steroids reduce respiratory distress syndrome is secure.
Keywords: Antenatal corticosteroids; Clinical trials; Respiratory distress syndrome.
Copyright © 2019 Elsevier B.V. All rights reserved.