Objective: The fragility index (FI), a minimum number of events in 1 arm of a clinical trial required to revert the statistically significant result to nonsignificant, has recently been developed as an easy-to-understand novel metric to evaluate the robustness of randomized controlled trials (RCTs). Here, we evaluated the FI of RCTs in the field of neurology, particularly in studies of ischemic stroke.
Methods: Previous literature published between June 1, 2012 and May 31, 2018 were reviewed from the MEDLINE database by the authors. The original article reporting the significant RCT result, of which a dichotomous outcome was set as its primary outcome measure, was included to evaluate the robustness of the result by calculating the FI. In addition, recent studies examining FI in other clinical fields were reviewed and summarized.
Results: In the 25 eligible RCT studies, the median total number of study participants was 206 (inter quartile range: 144-450) and the median FI was 7 (inter quartile range: 4-15.0). The FI showed a strong negative correlation with the observed P value. There was no significant difference in the FI between RCTs with and without acute settings. Our median FI was higher than the median FI of 2.5 of previous studies examining FI in other clinical fields, as only 20% (5 of 25) of studies included in our study had an FI less than 2.5.
Conclusion: Our results suggest that many RCTs in the field of ischemic stroke have a fair robustness, when compared to those in other clinical fields.
Keywords: Fragility index; RCT; ischemic stroke; statistical robustness.
Copyright © 2019 Elsevier Inc. All rights reserved.