When the p Value Doesn't Cut It: The Fragility Index Applied to Randomized Controlled Trials in Colorectal Surgery Article

Full Text via DOI: 10.1097/DCR.0000000000002146 Web of Science: 000793966800030

Cited authors

  • Nelms DW, Vargas HD, Bedi RS, Paruch JL


  • BACKGROUND: The American Statistical Association, among others, has called for the use of statistical methods beyond p <= 0.05. The fragility index is a statistical metric defined as the minimum number of patients for whom if an event rather than a nonevent occurred, then the p value would increase to >= 0.05. Previous reviews have demonstrated that many randomized controlled trials have a low fragility index, suggesting they may not be robust.OBJECTIVE: The purpose of this study was to review the fragility indices of randomized controlled trials in colorectal surgery.DATA SOURCES: A PubMed search was performed.STUDY SELECTION: Colorectal surgery randomized controlled trials with a dichotomous primary outcome p <= 0.05 and publication between 2016 and 2018 were systematically identified.INTERVENTIONS: All procedural interventions related to colorectal surgery were included.MAIN OUTCOME MEASURES: The main measures were the fragility index and the number of patients lost to follow-up for each trial. The percentage of trials with the number of patients lost to follow-up greater than the fragility index was calculated. RESULTS: In total, 712 abstracts were reviewed, with 90 trials meeting the inclusion criteria. The median fragility index was 3 (interquartile range of 1 to 10). In 51 of the 90 trials (57%), the number of patients lost to follow-up was greater than the fragility index.LIMITATIONS: The fragility index is only one measure of the robustness of a randomized clinical trial.CONCLUSIONS: Most colorectal surgery randomized controlled trials have a low fragility index. In 57% of trials, more patients were lost to follow-up than would be required to change the outcome of the trial from "significant" to "nonsignificant" based on the p value. This emphasizes the importance of assessing the robustness of clinical trials when considering their clinical application, rather than relying solely on the p value.

Publication date

  • 2022

Published in

International Standard Serial Number (ISSN)

  • 0012-3706

Number of pages

  • 8

Start page

  • 276

End page

  • 283


  • 65


  • 2