Fragility Index Analysis of the 2018 Clinical Practice Guidelines on Tranexamic Acid Use in Total Joint Arthroplasty.

ABSTRACT

BACKGROUND: The 2018 American Association of Hip and Knee Surgeons clinical practice guideline (CPG) 'Tranexamic Acid Use in Total Joint Arthroplasty' evaluated the efficacy and safety of tranexamic acid in primary total joint arthroplasty. The following review assessed the statistical fragility of the randomized controlled trial (RCT) outcomes on which the CPG recommendations were based using a fragility analysis. METHODS: All dichotomous outcomes from the randomized controlled trials used to guide the CPG from its associated network and direct meta-analyses were analyzed. Fragility and reverse fragility indices (FI and rFI) and quotients (FQ and rFQ) were calculated for each outcome. Mean indices and quotients were calculated for each guideline question, outcome category, and comparison of tranexamic dose, formulation, and administration timing. RESULTS: This review evaluated 403 dichotomous outcomes on transfusion and complication rates associated with tranexamic acid (TXA) administration. The mean FI of significant outcomes of the CPG was 5.23, and the mean rFI of nonsignificant outcomes was 5.80. Outcomes assessing complication rates had a mean rFI of 6.48. Most outcomes on transfusion in categories comparing TXA to placebo administration had higher mean FIs than rFIs, and all outcomes comparing transfusion risk associated with different TXA formulations and doses had higher mean rFIs than FI or no associated significant outcomes. CONCLUSION: The rFI and FIs calculated for this CPG are comparable to or higher than mean values reported across orthopaedic literature, indicating the relative statistical stability of its included outcomes. As we learn more about fragility analyses and their potential applications, this type of statistical analysis shows promise as a useful tool to incorporate into future guidelines to assess the quality of RCTs and evaluate the strength of recommendations.