Predictors of Increased Fragility Index Scores in Surgical Randomized Controlled Trials: An Umbrella Review

Abstract

AbstractBackgroundThe fragility index (FI) is defined as the minimum number of patients or subjects needed to switch experimental groups for statistical significance to be lost in a randomized control trial (RCT). This index is used to determine the robustness of a study's findings and recently as a measure of evaluating RCT quality. The objective of this review was to identify and describe published systematic reviews utilizing FI to evaluate surgical RCTs and to determine if there were common factors associated with higher FI values. MethodsThree databases (PubMed, MEDLINE [Ovid], Embase) were searched, followed by a subsequent abstract/title and full‐text screening to yield 50 reviews of surgical RCTs. Authors, year of publication, name of journal, study design, number of RCTs, subspecialty, sample size, median FI, patients lost to follow‐up, and associations between variables and FI scores were collected.ResultsAmong 1007 of 2214 RCTs in 50 reviews reporting FI (median sample size 100), the pooled median FI was 3 (IQR: 1–7). Most reviews investigated orthopaedic surgery RCTs (n = 32). There was a moderate correlation between FI and p value (r = 0.‐413), a mild correlation between FI and sample size (r = 0.188), and a mild correlation between FI and event number (r = 0.129). ConclusionBased on a limited sample of systematic reviews, surgical RCT FI values are still low (2–5). Future RCTs in surgery require improvement to study design in order to increase the robustness of statistically significant findings.