The Statistical Fragility of Tranexamic Acid in Spinal Surgery

STUDY DESIGN: Systematic review. OBJECTIVE: To evaluate the statistical robustness of TXA use in spine surgery as a potential contributor to controversies in this field. SUMMARY OF BACKGROUND DATA: Tranexamic acid (TXA) is an antifibrinolytic medication administered during spinal surgery to limit blood loss. Existing randomized controlled trials (RCTs) on the efficacy of TXA contain varied results, particularly when reporting outcomes related to blood transfusion rates and thromboembolic events. By calculating the fragility index (FI), reverse fragility index (rFI), and fragility quotient (FQ), statistical robustness was quantified and compared across all included RCTs. METHODS: PubMed, Embase, and MEDLINE were systematically searched for recent RCTs (January 1, 2000-August 1, 2023) assessing TXA use in patients undergoing spine surgery. The FI and rFI were calculated for each outcome, representing the number of event reversals required to alter statistical significance for significant and nonsignificant outcomes, respectively. The FQ was determined by dividing the FI/rFI by the study sample size. RESULTS: Of the 297 RCTs screened, 31 studies were included for analysis, yielding 80 dichotomous outcomes. Across these outcomes, the median FI (mFI) was 5.0, with an associated median FQ (mFQ) of 0.060. Nine outcomes were statistically significant (mFQ=0.018), and 71 were nonsignificant (mFQ=0.064). The most common outcome categories included blood/platelet transfusions (38 outcomes), thromboembolic events (15 outcomes), and other adverse events (27 outcomes), resulting in mFQs of 0.056, 0.049, and 0.064, respectively. CONCLUSIONS: Outcomes examining TXA in spinal surgery demonstrated statistical fragility, with significant and thromboembolic outcomes proving the most fragile. Among all outcomes, there was a lack of significant results. To better guide future research on TXA use in spine surgery, this study recommends RCTs report fragility statistics along with P values and include these metrics when proposing clinical implications. LEVEL OF EVIDENCE: Level III.