Tranexamic acid for acute gastrointestinal bleeding: updated after the HALT‐IT trial

Abstract

Editors, Twum-Barimah et al recently addressed the importance of tranexamic acid in acute gastrointestinal bleeding in their systematic review.1 The HALT-IT trial reported contradictory results that is, tranexamic acid confers no additional survival benefit and is associated with increased harm.2 Karadaş et al found no statistically significant difference in mortality between patients with gastrointestinal bleeding who received tranexamic acid or placebo.3 The pooled results of the outcomes of interest after these studies remain unclear. Moreover, concerns have been raised regarding the conclusiveness of the results from a conventional meta-analysis. Due to repeated statistical testing or insufficient sample size, a conventional meta-analysis is susceptible to a type I or type II error.4 In order to investigate the pooled results after including the HALT-IT trial and the study by Karadaş et al, we added these studies and re-performed meta-analysis. We also performed trial sequential analysis (TSA) to evaluate if the results of the updated meta-analysis are conclusive. TSA was conducted using TSA software (version 0.9.5.10 Beta), with an alpha of 5%, power of 80% and a relative risk reduction of 20% given the incidence rate in the placebo group. The updated meta-analysis reveals that tranexamic acid, compared with placebo, is associated with reduced mortality risk (random effects model: risk ratio [RR], 0.82; 95% confidence interval [CI], 0.69-0.98; P = 0.025; Figure 1). Also shown in Figure 1 are 24-hour mortality, 72-hour mortality, 1-month mortality and end of follow-up mortality. In contrast, there was no statistically significant difference in rebleeding (RR, 0.86; 95% CI, 0.72-1.04; P = 0.116), need for surgery (RR, 0.78; 95% CI, 0.57-1.08; P = 0.131) or need for transfusion (RR, 0.99; 95% CI, 0.97-1.02; P = 0.525) between tranexamic acid and placebo. In the TSA of mortality, although the cumulative Z-curve passed the traditional significance boundary, it did not pass the sequential monitoring boundary, and did not reach the required information size (RIS) (mortality RIS number: 36935), indicating a false positive result. In the TSA of rebleeding, the cumulative Z-curve touched the boundary for futility before reaching the RIS (rebleeding RIS number: 24314), indicating that it will be unlikely to reach statistical significance even if more randomised control trials are included until the RIS is reached. In the TSA of need for surgery, although the cumulative Z-curve did not pass the traditional significance boundary, it did not pass the sequential monitoring boundary, and did not reach the RIS (need for surgery RIS number: 62933), indicating a false negative result. Finally, in the TSA of need for transfusion, the cumulative Z-curve reached the RIS (need for transfusion RIS number: 427), but did not pass the traditional significance boundary, indicating a true negative result. In brief, while more trials are required for a definitive result regarding the effects of tranexamic acid on mortality and need for surgery, our analyses confirm that tranexamic acid is not superior to placebo in reducing rebleeding or need for transfusion. Meanwhile, the increased risk of venous thromboembolism reported in HALT-IT cautions against the routine use of tranexamic acid in acute gastrointestinal bleeding.