SGEM#301: You Can’t Stop GI Bleeds with TXA

Subscribe: RSS

Date: September 16th, 2020

Guest Skeptics: Dr. Robert Goulden and Dr. Audrey Marcotte are Chief Residents from the Royal College of Emergency Medicine Program at McGill University. Robert’s academic interests include research and evidence-based medicine. Alongside his EM residency, he is doing a PhD in epidemiology. Audrey’s academic interests include trauma and resuscitation. Outside of medicine, Audrey likes to play rugby and run.

Reference: Roberts et al. Effects of a high-dose 24-h infusion of tranexamic acid on death and thromboembolic events in patients with acute gastrointestinal bleeding (HALT-IT): an international randomised, double-blind, placebo-controlled trial. The Lancet 2020

This was an SGEM Journal Club episode recorded live at McGill University Grand Rounds. This was the third time coming to McGill University Department of Emergency Medicine to give Grand Rounds. The first visit was back in 2013 for SGEM#50: Under Pressure – Vasopressin, Steroids and Epinephrine in Cardiac Arrest. The bottom line was this was interesting, but VSE protocol was not ready for routine use.

The second visit was SGEM#176: Somebody’s Watching Me – Cardiac Monitoring for Chest Pain. We were trying to answer the question: Do all patients presenting to the emergency department with chest pain need to be placed on cardiac monitoring or could some be safely removed? The SGEM Bottom Line was that for some patients presenting with chest pain who are chest pain free and have normal/non-specific ECG findings could potentially be safely removed from cardiac monitoring using the Ottawa CPCM Rule.

Five Rules of SGEM-JC

Case:A 58-year-old man presents with hypotension, tachycardia, and pallor. He vomits a large amount of bloody emesis and has epigastric discomfort. He is not taking any anti-coagulants. He remains hemodynamically unstable despite initial resuscitation and has another episode of hematemesis in front of you. While waiting for your consultant to answer the phone, you consider treating him with tranexamic acid (TXA), but wonder if it will prevent death from gastrointestinal (GI) bleeding.

Background: We have covered the use of TXA a number of times on the SGEM. TXA is an anti-fibrinolytic agent that inhibits clot breakdown and has demonstrated mixed results in different clinical settings.

The CRASH-2 trial showed a 1.5% absolute mortality benefit with TXA in adult trauma patients compared to placebo (SGEM#80). TXA also seems to improve patient-oriented outcomes in epistaxis (SGEM#53 and SGEM#210).

However, TXA did not show a statistically significant difference for the primary outcome in post-partum hemorrhage (SGEM#214) WOMAN Trial, hemorrhagic stroke (SGEM#236) or traumatic intracranial hemorrhage (SGEM#270) CRASH-3.

A Cochrane systematic review and meta-analysis of eight smaller trials (n=1,701) using TXA in gastrointestinal bleeding suggested a large (40%) risk reduction in all-cause mortality (Bennett et al 2014). However, even a meta-analysis is prone to bias and is only as good as the quality of the included trials. When all participants in the intervention group with missing outcome data were included as treatment failures, or when the analysis was limited to trials with low risk of attrition bias the mortality benefit of TXA disappeared.

Clinical Question: Does treatment with TXA reduce the mortality of patients with upper or lower GI bleeds?

Population: Adult patients (16 years of age or 18 years of age and older depending on country) with significant upper or lower GI bleed. Significant bleed was defined clinically (judged at risk of bleeding to death, hypotension <90 mmHg systolic, tachycardia, signs of shock, needing transfusion, urgent endoscopy or surgery).
- Exclusion: Any patient whom the clinician felt had a clear indication or clear contraindication for TXA
Intervention: Intravenous TXA, 1g loading dose over 10 minutes followed by 3g maintenance over 24 hours
Comparison: Matching placebo (Sodium chloride 0.9% IV)
Outcome:
- Primary Outcome: Death due to gastrointestinal (GI) bleeding within five days
- Secondary Outcomes:
  - Death due to gastrointestinal bleeding within 24h and within 28 days
  - All-cause and cause specific mortality at 28 days
  - Rebleeding within 24h, 5 days, 28 days
  - Surgical or radiological intervention
  - Blood product transfusion
  - Thromboembolic events (deep vein thrombosis, pulmonary embolism, stroke, myocardial infarction)
  - Seizures and other complications (sepsis, pneumonia, renal and liver failure, cardiac event)
  - Days in intensive care unit
  - Functional status (Katz index of Independence in activities of daily living)

Authors’ Conclusions: “We found that tranexamic acid did not reduce death from gastrointestinal bleeding. On the basis of our results, tranexamic acid should not be used for the treatment of gastrointestinal bleeding outside the context of a randomised trial.”

Quality Checklist for Randomized Clinical Trials:

The study population included or focused on those in the emergency department. Yes
The patients were adequately randomized. Yes
The randomization process was concealed. Yes
The patients were analyzed in the groups to which they were randomized. Yes
The study patients were recruited consecutively (i.e. no selection bias). Unsure
The patients in both groups were similar with respect to prognostic factors. Yes
All participants (patients, clinicians, outcome assessors) were unaware of group allocation. Yes
All groups were treated equally except for the intervention. Yes
Follow-up was complete (i.e. at least 80% for both groups). Yes
All patient-important outcomes were considered. Yes
The treatment effect was large enough and precise enough to be clinically significant. No

Key Results: Mean age 58 years, two-thirds male, nine out of ten were upper GI bleeds, half were thought to be due to varices and about 9% were known to be on anti-coagulants.

No statistically significant difference in mortality from GI bleed.

Primary Outcome: Mortality from GI bleed within five days
- 3.7% (TXA) vs. 3.8% (placebo), RR 0·99 (0·82–1·18)
Secondary Outcomes:
- No significant difference in all-cause mortality at 28 days, 9.5% (TXA) vs. 9.2% (Placebo), RR 1·03 (95% CI; 0·92–1·16)
- Statistically significant increase in venous thromboembolism (VTE), 0.8% (TXA) vs. 0.4% (Placebo). RR 1.85 (95% CI; 1.15-2.98). This effect was more marked in those with suspected variceal bleeding than in those with non-variceal bleeding (p=0.035 for heterogeneity).
- No significant difference in all other secondary outcomes including rebleed and death due to rebleed at multiple time points, need for additional interventions, and other safety outcomes

1) Consistency in Results: This trial is definitively and unambiguously “negative” – it’s actually rare that we get a trial result that is so clear cut. All of the different outcomes lined up as showing no statistical benefit, and there were no subgroups in which that differed. In contrast, think of CRASH-3 (TXA for intracranial hemorrhage) or PARAMEDIC-2 (epinephrine for out-of-hospital cardiac arrest), other well-conducted large RCTs, where the fact that some outcomes were “positive” and others “negative” have left people still debating how to interpret them.

2) Changing the Primary Outcome: Switching the outcome midway through a trial is a red flag for potential statistical shenanigans, as there is a risk that investigators are aware of partial results and are switching from a non-significant to a significant outcome. However, in this case the decision to switch was made and published with a justification before unblinding.

However, they did shift from a reliable, unambiguous, patient-centred outcome – all-cause mortality at 28 days – to a much less patient and clinically important outcome – mortality from GI bleeding at five days. Patients (and clinicians) don’t usually care what they die of – they care whether they die or not. It’s also much more challenging to reliably determine the cause of death as compared to determining ifsomeone is dead or not. Imagine if TXA decreased GI bleeding death by 2% but increased VTE death by 4%. Should that be considered a positive trial?! In the end, it doesn’t matter as the original primary outcome of all-cause mortality at 28 days – which I think should be the outcome we are most interested in – was also not significantly different.

“The sample size calculation was initially based on all cause mortality as the primary outcome since we expected that most deaths would be due to bleeding. However, while the trial was underway, we observed that over half of all deaths were due to nonbleeding causes. Accumulating evidence from other large trials of tranexamic acid showed no apparent effect on nonbleeding deaths.”
“The primary outcome was therefore changed to death due to bleeding within 5 days of randomisation on Nov 21, 2018. Based on the amended primary outcome, assuming a risk of death due to bleeding of 4%, a study with 12000 patients has about 85% power (two-sided α of 5%) to detect a clinically important 25% relative reduction in death due to bleeding from 4% to 3%”.

3) Secondary Outcomes: The finding of increased VTE risk is interesting. It may be a chance finding since there were lots of secondary analyses and by random chance alone, we would expect a few to be statistically different. However, this was a safety outcome for which there was a reasonable prior belief in potential harm, and their point estimate (RR 1.85) was almost identical to the point estimate from the prior Cochrane review (RR 1.86). Arguably a 0.4% absolute increase in VTE (0.8% vs. 0.4%, NNH = 250) is not that clinically important, but given there is no evidence of benefit here, it strengthens the case against using TXA for this indication.

4) Selection Bias: As with CRASH-3 and the WOMAN trials, patients whom clinicians felt there was a clear indication or contraindication to TXA were excluded. We don’t have a clear idea how many patients were excluded for these reasons. As such, we must consider the possibility of selection bias.

5) Large RCT vs. SRMA of Small Studies: The prior Cochrane meta-analysis of RCTs showed a statistically significant 40% relative reduction in mortality with TXA in GI bleeds. This trial refutes the finding of a large reduction and suggests we should be careful in putting too much faith in meta-analyses of methodologically flawed small trials. However, it does not rule out a small or modest reduction.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: We agree with the authors conclusion, though we would emphasize that all-causemortality was also unaffected by TXA.

SGEM Bottom Line: The latest evidence does not support the use of TXA in GI bleeds.

Case Resolution: You decide not to give TXA based on the results of HALT-IT. You continue to resuscitate the patient and organize for urgent endoscopy. He has two esophageal varices that are successfully ligated.

Clinical Application: When it comes to GI bleeding, it is time to put aside TXA from your treatment toolbox and focus on basic volume resuscitation while liaising with your consultants for definitive hemorrhage control.

What Do I Tell My Patient? We have no specific medication that directly stops the bleeding, but we will stabilize you with blood transfusions to prepare you for definitive treatment with endoscopy.

Keener Kontest: Last weeks’ winner was Sam Johnson a student in neuroscience at the university of Aberdeen in Scotland. He knew the Likert scale was named after Rensis Likert to measure attitudes and feelings.

Listen to the SGEM podcast to hear this weeks’ question. Send your answer to TheSGEM@gmail.com with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.

Other FOAMed:

First10EM – TXA for GI bleeds: No benefit (The HALT-IT trial)
The Bottom Line – HALT-IT
REBEL EM – The HALT-IT Trial: TXA in Acute GI Bleeds
St. Emlyn’s – Halt! It’s not time for TXA! Or is it? HALT-IT results at St Emlyn’s