Please login to view this media
- Talk
- 15/09/2021
- Canada
The Fragility of Statistically Significant Findings From Randomized Controlled Trials in Hip and Knee Arthroplasty
Description
The presentation discusses the concept of fragility in clinical trials, specifically focusing on a statistical measure introduced by Feinstein in 1990 and later refined in 2014 by Walsh and colleagues, known as the fragility index. The fragility index quantifies how robust the findings of statistically significant trials are by determining the number of patient outcome events that would need to change for the results to become nonsignificant.
The speaker presents examples of two hypothetical randomized controlled trials (RCTs) with different sample sizes and outcomes to illustrate the fragility index concept. Despite significant results, they find the first trial with a small sample size has a fragility index of one, meaning that just one patient outcome flip could change the conclusions, while the second trial with a larger subject pool is less fragile.
The study subsequently focuses on arthroplasty RCTs over the past decade, aiming to assess their fragility index. They reviewed around 5,000 studies, narrowing down to 34 significant trials. The findings revealed a median fragility index of one, indicating a high fragility where numerous trials could potentially lead to a different conclusion if some patients were lost to follow-up.
The limitations of the study are acknowledged, especially that the fragility index only applies to dichotomous outcomes, and the researchers encourage increased attention to the fragility in future RCT planning and reporting. The overall conclusion emphasizes the need for larger trials while addressing shortcomings in the fragility index. The presentation ends with a call for questions and engagement from attendees.