What is it about?

I propose using the Relative Risk Index (RRI) alongside the Fragility Index (FI) when evaluating the reliability of orthopedic surgery research. While the FI shows how many patient outcomes must change to alter statistical significance, the RRI measures the percentage change needed to reach therapeutic equivalence. In tranexamic acid research, RRI provides different insights than FI, potentially showing robustness where FI suggests fragility. The RRI focuses on clinical relevance rather than just statistical significance.

Featured Image

Why is it important?

This article introduces a practical tool that could significantly improve doctors' evaluation of orthopedic research evidence. Unlike traditional measures that focus primarily on p-values (statistical significance), the Relative Risk Index emphasizes therapeutic equivalence—what matters most for patient care. This approach could help surgeons make more informed decisions about treatments like tranexamic acid by showing when research findings are truly reliable from a clinical perspective, not just a statistical one.

Perspectives

While valuable, I believe the Fragility Index doesn't tell the complete story about research reliability. My proposed Relative Risk Index addresses a fundamental problem in medical research—we've become overly fixated on p-values rather than clinical meaning. When I analyzed the tranexamic acid example, I found that changing just three patient outcomes flipped statistical significance, suggesting fragility. However, this represented a substantial 30% increase in transfusions, indicating the findings were robust in real time. The RRI captures this clinical reality better than traditional measures.

Thomas F Heston MD
University of Washington

Read the Original

This page is a summary of: Letter to the Editor: The Relative Risk Index: A Complementary Metric for Assessing Statistical Fragility in Orthopaedic Surgery Research, Journal of the American Academy of Orthopaedic Surgeons, November 2024, Wolters Kluwer Health,
DOI: 10.5435/jaaos-d-24-00473.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page