Patients with GI bleeds present to the ED with a panoply of symptoms ranging from hematemesis and hematochezia to melena and that nice elderly patient who comes in feeling fatigued right before he crunks spectacularly in front of your eyes. Likewise, we have a number of treatment options for these patients ranging from the good (blood transfusions, antibiotic prophylaxis for cirrhotic patients) to the useless (Hi, octreotide!) to the painful and unhelpful. (See previous roasting of NG lavage: EMN. 2015;37:20;; https://bit.ly/2RUQ3Ww.)
Regardless, it's always nice to have another temporizing tool in the box for these patients as they await more definitive 'scopic management from our GI or surgical colleagues. Even better if the intervention is cheap, painless, and low on harmful side effects.
Enter tranexamic acid (TXA). Made famous by showing a 1.5 percent absolute mortality benefit for trauma patients in the CRASH-2 trial (NNT=67 to save a life), this antifibrinolytic agent (inhibits clot breakdown) looked like a natural fit for helping patients with GI bleeding. (Lancet. 2010;376:23; http://bit.ly/2CoZJB4.) I mean, both patients are bleeding on the inside; what difference should it make if the bleeding is caused by a vexed varix or an unruly ulcer as opposed to a car crash?
Plus TXA is pretty affordable, especially compared with really pricey pharmacologic interventions like prothrombin complex concentrates (Kcentra) and anti-inhibitor coagulant complexes (FEIBA), not that we are commonly giving these to patients with GI bleeds, though given how profitable they are, I'm sure that kind of intention creep is on the way.
TXA is affordable and appears effective in another group of patients with bleeding, which means it was with quite a bit of enthusiasm that I was looking forward to the results of the HALT-IT trial. (Lancet. 2020;395:1927; https://bit.ly/3iM28sg.) In the authors' words, this was “an international, multicentre, randomised, placebo-controlled trial in 164 hospitals in 15 countries” that enrolled a little more than 12,000 patients with acute GI bleeds to receive TXA or placebo.
It was a huge study and of high methodological quality: Patients, clinicians, and assessors were blinded to whether the patient received TXA, and they lost a grand total of three of 12,009 subjects (0.025%) to follow-up. Primary outcome was death due to bleeding within five days. Secondary outcomes were other important things like death at days one and 28, heart attacks, strokes, DVTs and PEs, seizures, need for blood transfusion, etc., all very reasonable and with nary a sketchy composite endpoint in sight.
You had to be an adult (16 or 18 depending on the country) and have a pretty serious bleed to be included in the trial. The diagnosis of significant bleeding was clinical; “significant was defined as a risk of bleeding to death and included patients with hypotension, tachycardia, or signs of shock, or those likely to need transfusion or urgent endoscopy or surgery,” the authors wrote. This is exactly the kind of population we want to study to determine if TXA is going to have a meaningful impact on outcomes.
This huge, high-quality, multicenter study of TXA v. placebo found that TXA didn't do a thing for patients with acute GI bleeds. No mortality difference at one day or at five or 28 days. No difference if they started TXA early or late (defined as more or less than three hours). No difference in upper or lower GI bleeds, variceal or nonvariceal, not very sick or very, very sick (as determined by the Rockall Score: http://bit.ly/MDCalcRockallScore). No difference if they gave it on a boat or with a goat, with a fox or in a box. TXA didn't work here or there, for GI bleeds, not anywhere.
It's actually surprising that in a study this large that there wasn't at least some subgroup where statistical benefit was noted and then puffed up by the authors to make it look like a positive trial. In fact, the only statistically significant outcomes were the harms. The TXA group had slightly more VTEs (PE and DVT) with a RR 1.85 (95% CI 1.15-2.98) for an NNH=250, though it should be noted this was a composite outcome and that the RR for PE or DVT was not statistically significant. TXA patients also had slightly more seizures, RR 1.73 (95% CI 1.03-2.93) for an NNH=500. Why seizures? Don't ask me; no one really knows. (Just kidding; smart people know, you can read all about it at https://bit.ly/33Gjd0v. [Ann Neurol. 2016;79(1):18].)
This outcome is a bit surprising and very disappointing, but it sends two unequivocal messages: TXA appears to have no role in managing acute GI bleeds, and this is why we do science—high-quality studies can actually answer important questions, even when the answer isn't what we were looking for.
Dr. Rundeis the assistant residency director and an assistant professor of emergency medicine at the University of Iowa Hospitals and Clinics, where he serves as co-director for the associate fellowship in medical education. He creates content for and is a member of the editorial board forwww.TheNNT.com, and is a content contributor forwww.MDCalc.com. Follow him on Twitter@Runde_MC, and read his past articles athttp://bit.ly/EMN-ReasonableDoubt.