The statistical fragility of randomised controlled trials evaluating tranexamic acid against a saline control in total hip arthroplasty.

Abstract

This study applied the fragility index (FI), reverse fragility index (rFI), and fragility quotient (FQ) to evaluate the statistical fragility of outcomes reported in RCTs evaluating TXA against a saline control in total hip arthroplasty (THA). PubMed, Embase, and MEDLINE were searched for RCTs on TXA in THA compared to a saline control. FI and rFI were calculated for dichotomous outcomes, representing the number of outcome reversals required to alter significance. 32 studies were included. Across 69 total outcomes, the median FI was 3.0 (IQR 2.0-6.0) with an associated median FQ of 0.031 (IQR 0.020-0.053), suggesting that 3 outcome event reversals would alter significance for 50% of all outcomes. 18 outcomes were statistically significant, with a median FI of 3.0 (IQR 1.0-5.0) and a median FQ of 0.026 (IQR 0.017-0.036). 51 outcomes were non-significant, with a median rFI of 4.0 (IQR 2.5-7.0) and a median FQ of 0.032 (IQR 0.020-0.066). Results designated as primary outcomes were more fragile (median FQ = 0.025, <i>n</i> = 20) than non-primary outcomes (median FQ = 0.032, <i>n</i> = 49). The efficacy of TXA use during THA from RCTs is statistically fragile, particularly among significant outcomes and those related to blood or platelet transfusion rates. We recommend combined reporting of <i>p</i>-values with FI and FQ metrics to help interpret clinical findings in RCT involving TXA use during THA.