The beneficial effect of prophylactic hydrocortisone treatment in extremely preterm infants improves upon adjustment of the baseline characteristics

Abstract

Background: Prophylactic low-dose hydrocortisone (HC) was found to improve survival without bronchopulmonary dysplasia (BPD) in extremely preterm infants. However, appropriately adjusting for baseline risks of BPD or death might substantially increase the precision of the HC effect size.

Methods: We conducted a secondary analysis of the PREMILOC trial. The treatment effect was evaluated on the primary endpoint through a covariance analysis ANCOVA, adjusting for the baseline covariates using a mixed linear model. Several sensitivity analyses were conducted to assess the potential heterogeneity of the treatment effect across centers and subpopulations.

Results: The interaction between treatment group and baseline risk for BPD or death was not statistically significant (p = 0.498). After adjusting for the patient's probability of BPD-free survival using baseline predictors alone, the HC treatment exhibited a highly significant effect (OR [95% CI] = 2.053 [1.602-2.501], p = 0.002), with a number needed to treat NNT [95% CI] = 5.8 [4.1-23.0]. Despite a weak interaction with sex, we found a lack of heterogeneity in the treatment effect across specific subpopulations.

Conclusions: In the PREMILOC trial, the beneficial effect of prophylactic HC versus placebo on BPD-free survival in extremely preterm neonates was found to be greater when adjusted to baseline risks of BPD or death.

Registration numbers: EudraCT number 2007-002041-20, ClinicalTrial.gov number NCT00623740.

Impact: Prophylactic low-dose hydrocortisone (HC) provided past evidence of a beneficial effect in improving survival without BPD in infants born extremely preterm. Adjustment for baseline risks of BPD or death might substantially increase the precision of the HC effect size. The beneficial effect of prophylactic HC vs placebo on BPD-free survival in extremely preterm neonates was found to be greater when adjusted to baseline risks of BPD or death. We evidenced a lack of heterogeneity in the treatment effect in specific subpopulations despite some weak interaction with sex.