Randomized controlled trials: significant results-fragile, though

Abstract

In their systematic review, Shochet et al. question the robustness of the statistical significance of the results of randomized controlled trials (RCTs) in nephrology reported in high-impact journals (2005-2014). They hypothesized that a high proportion of RCTs in nephrology would be fragile. They calculated a Fragility Index that is a tool for testing robustness of RCTs, based on 107 RCT reports. The observation of an alarming median Fragility Index equal to 3 suggests that in half of the trials, the sole additional occurrence of 3 events would compromise the significance of nominally significant results.