Statistical Significance in Trauma Research: Too Unstable to Trust?

Abstract

Objectives: To evaluate the stability of statistical findings in the fracture care literature based on minor changes in event rates and to determine the utility of applying both the Fragility Index (FI) and Fragility Quotient (FQ) to comparative orthopaedic trauma trials. Methods: All fracture care studies from 1991 to 2013 in the Journal of Bone and Joint Surgery and the Journal of Orthopaedic Trauma were screened. The FI was determined by altering the number of reported outcome events, a single event at time, until a reversal of statistical significance was determined. The associated FQ was determined by dividing the FI by the total sample size. Results: Of the 4040 studies evaluated, 198 comparative studies met inclusion criteria with a reported 253 primary and 522 secondary outcome events. There were 118 randomized controlled trials and 80 retrospective comparative studies. Of the 775 total outcome events, 235 were initially reported as significant. The median FI for the entire study was only 5 with an associated FQ of 0.046. This represents just 3.8% of the total study population. Conclusions: The robustness of comparative trials in the orthopaedic trauma literature may not be as stable as previously thought with only a few event reversals required to alter trial significance. We therefore recommend triple reporting of a P value, FI, and FQ to aid in the evaluation and interpretation of statistical stability and quantitative significance in comparative orthopaedic trauma trials.

The Statistical Fragility of Hamstring Versus Patellar Tendon Autografts for Anterior Cruciate Ligament Reconstruction: A Systematic Review of Comparative Studies

“…For this systematic review, the overall FI was found to be 3.77 and the overall FQ to be 0.040, which are consistent with previous orthopaedic literature reporting an average median FI 12,18,22,23,30,48 of 2.5 and mean FQ 32,48 of 0.031. Our findings demonstrate that statistical significance may be altered by the reversal of \4 outcome events or the reversal of 4% of outcome events.…”

“…The authors of a recent evaluation of 20 years of orthopaedic trauma literature recommended ''triple reporting,'' consisting of P values, the FI, and the FQ. 32 Establishing guidelines that encourage the reporting of the FI and FQ together will improve how physicians and surgeons perceive the strength of study findings.…”

“…In contrast to their FI calculation, this estimated FQ is consistent with our findings (mean FQ, 0.040) and previous orthopaedic literature (mean FQ, 0.031). 32,48 This example illustrates the need to include FQ with FI reporting to provide clarity of statistical strength.…”

Dismantling the Fragility Index: A demonstration of statistical reasoning

“…The FI has been applied in review studies of anticancer medicines, clinical heart failure, anesthesiology, and several other areas of study, in order to characterize robustness of results amid concern over reproducibility of research. [2][3][4][5][6][7][8][9][10][11][12][13][14] The Fragility Index has been criticized for using Fisher's exact test as its basis when many trials use survival analysis, which is more powerful than Fisher's exact test and can demonstrate differences in time to event as well as the probability of an event. One review of oncology studies found several trials with P-values <.001 and FI values of zero (implying a P-value >.05 by Fisher's exact test).…”

“…As such, at least part of what the index is intended to capture is the fragility of results due to bias from missing data. The FI has been applied in review studies of anticancer medicines, clinical heart failure, anesthesiology, and several other areas of study, in order to characterize robustness of results amid concern over reproducibility of research 2‐14 …”