A Bayesian Re-analysis of the Steroids to Reduce Systemic Inflammation after Infant Heart Surgery (STRESS) Trial
Medrxiv : the Preprint Server for Health Sciences
|February 24, 2025
Summary
Prophylactic methylprednisolone shows a high probability of benefit for infants undergoing heart surgery, with a low risk of harm. This Bayesian re-analysis of the STRESS trial provides deeper insights into its efficacy.
Area of Science:
- Pediatric Cardiac Surgery
- Pharmacology
- Biostatistics
Background:
- Prophylactic steroids, specifically methylprednisolone, are commonly administered to mitigate systemic inflammatory responses during cardiopulmonary bypass in pediatric heart surgery.
- The original STRESS trial indicated no significant difference in worse outcomes between methylprednisolone and placebo groups in primary analyses.
- However, secondary analyses suggested potential benefits, prompting further investigation into methylprednisolone's efficacy.
Purpose of the Study:
- To re-analyze the STRESS trial data using Bayesian analytics to precisely assess the probability of benefit from prophylactic methylprednisolone.
- To determine the likelihood of benefit versus harm, ensuring potentially effective therapies are not overlooked.
Main Methods:
- Bayesian analysis was applied to the STRESS trial's primary composite outcome (death, transplant, major complications, post-operative length of stay).
- Markov Chain Monte Carlo simulations (10 simulations, 2000 iterations, 1000 burn-in) were used to evaluate posterior distributions and probability of benefit (OR<1).
- Sensitivity analyses explored various prior beliefs (pessimistic, neutral, optimistic) and prior belief strengths (weak, moderate, strong).
Main Results:
- The primary Bayesian analysis yielded a 91% probability of benefit and a 9% probability of harm for methylprednisolone.
- The composite outcome of death or major complication occurred in 18.8% of subjects, with an absolute risk difference of -2% favoring methylprednisolone.
- Nine sensitivity analyses consistently showed a higher probability of benefit than harm, with 8 demonstrating >80% probability of benefit and ≥1% absolute risk reduction.
Conclusions:
- The re-analysis indicates a high probability of benefit and low likelihood of harm associated with prophylactic methylprednisolone in this patient population.
- These findings offer a more comprehensive understanding of methylprednisolone's role, expanding upon the initial STRESS trial evaluation.
