Fragility of randomized clinical trials of treatment of clavicular fractures

Abstract

Background: Statistical significance, as reported by the P value, has traditionally been the most commonly reported way to determine whether a difference exists between clinical interventions. Unfortunately, P values alone confer little about the robustness of a study's conclusions. An emerging metric, the fragility index (FI), helps to address this challenge by quantifying the number of events per outcome group that would need to be reversed to the alternative outcome in order to raise the P value above the 0.05 threshold.

Methods: Using systematic search strategy, we identified randomized controlled trials (RCTs) pertaining to clavicular fractures published in the last decade (2007-2017). Studies included for analysis involved 2 parallel arms, were published in English, allocated patients to treatment and control arms in a 1:1 ratio, and reported statistical significance (P < .05) for dichotomous variables. The FI was determined based on the Fisher exact test, using previously published methods.

Results: Fifteen RCTs were included. The median FI was 2 (range, 0-17). Eleven studies (73.3%) had an FI of 2 or less. Seven of the trials (46.7%) reported that the number of patients lost to follow-up exceeded the FI.

Conclusions: The median FI reported in the recent literature on clavicular fractures is only 2. The FI is a useful metric to analyze the robustness of study conclusions that should complement other methods of critical data evaluation, including the P value or effect sizes. Future efforts are needed to increase institutional collaboration and patient recruitment to strengthen the robustness of RCT conclusions, especially in the realm of clavicular fracture management.

Keywords: Evidence-based medicine; P value; clavicular fractures; dichotomous outcome variables; fragility index; randomized controlled trial.