Interpreting Non-Inferiority Across Effect Scales: A Bayesian Perspective with Clinical Translation.
Danila Azzolina, Silvia Bressan, Mohd Rashid Khan et al.
✦ AI-curated · Sources linked
In 30 seconds
This study conducted a secondary Bayesian reanalysis of a randomized controlled trial comparing high-flow nasal cannula (HFNC) and non-invasive ventilation (NIV) in infants with acute bronchiolitis. The analysis revealed that a non-inferiority margin of 0.15 in absolute risk reduction could imply a doubling of intubation odds, with a Number Needed to Harm of approximately 6.7.
Key findings
- The Bayesian reanalysis indicated that a 0.15 non-inferiority ARR margin might imply a doubling of the odds of intubation.
- The Number Needed to Harm (NNH) for HFNC compared to NIV was approximately 6.7.
- The original analysis concluded non-inferiority using a fixed absolute margin.
Why it matters
Understanding the implications of non-inferiority margins is crucial for clinicians when interpreting trial results, as these margins can significantly affect treatment decisions and patient outcomes.