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 HALT-IT Trial Collaborators

Lancet 2020, 395: 1927–36


To assess the effects of tranexamic acid in patients with gastrointestinal bleeding


Randomised, double blinded, placebo-controlled.

Multi-centred – 164 hospitals across Multi-nations (15 countries)

World Bank (income)

  • High:                   UK, Australia, Spain, Ireland, Saudi Arabia, Romania
  •  Mid-high:           Malaysia, Albania, Georgia
  •  Mid-low:             Pakistan, Egypt, Nigeria, Nepal, Papua New Guinea
  • Low:                    Sudan


  • UK NRES Committee East of England (reference number 12/EE/0038); or,
  • national / local research ethics committees non-UK participant countries

Data collection – Jul 2013 to Jul 2019

Funding – NIHR



  • Numbers generated by Sealed Envelope (London, UK), independent statistician.

Treatment packs created/masked by:

  • Sharp Clinical Services UK (Crickhowell, UK), a Good Manufacturing Practice certified service
  • Removing manufacturer’s label and replacing it with the clinical trial label and randomisation number
  • Coding checked by testing each batch of ampoules with high-performance liquid chromatography



Adult (age ≥ 16-18)

Significant bleeding as defined by:

  • ↓BP, ↑HR, shocked
  • transfusion likely
  • urgent endoscopy or surgery)

Clinician unsure about TXA

 Exclusion:  none


  • If capacity was impaired and a personal or professional representative was available, consent was sought from the representative.
  • If neither were able to provide consent, it was waived, and the patient was informed about the trial and consented for ongoing data collection as soon as possible afterwards.

Eligible patients:

  • randomly assigned as possible
  • lowest numbered treatment pack was taken from a box of eight packs.
  • treatment was started immediately.


  • Loading: 1g TXA in 100mL 0.9% NaCl infused over 10 minutes.
  • Maintenance: 3g TXA in 1000ml of any isotonic IV solution 125mg/h for 24h

Control –

  • placebo (sodium chloride 0·9%)

Outcome data collected at death, discharge from the randomising hospital, or 28 days after randomisation, whichever occurred first.


Death due to bleeding within 5 days



Arterial thromboembolism (MI/CVA)

Venous thromboembolism (PE/DVT)

Surgical, endoscopic, or radiological intervention

Transfusion of blood products

Other complications (pneumonia, liver/renal failure, sepsis, seizures)



  • block non stratified
  • 12,009 patients. 5994 Intervention (49·9%) and 6015 Placebo (50.1%)

Original sample size

-based on all-cause mortality 1° outcome, assuming baseline 10% mortality.

-powered (90%) to detect ↓25% mortality 10% to 7.5%; estimated sample size 8000

Amended sample size:

-1° outcome changed to death within 5days of randomisation, 9 months before end of trial

-based on TXA T½=2–3h (99% eliminated by 2d of randomisation);

-to avoid dilution of sample from all-cause mortality

-powered (85%) ↓25% death due to bleeding from 4 to 3%; estimated sample size 12000

Modified intention-to-treat used. Excluded those who received neither dose or where outcome data on death were unavailable:

  • received neither: dose 52 (TXA 29, Placebo 23) /12,009 [0.43%]
  • consent withdrawn and outcome unavailable:
    • 17 (TXA 6, Placebo 11) /12,009 [0.14%]
  • Consent withdrawn but outcome available:
    • 12 (TXA 5, Placebo 7) /12,009 [0.1%]

Statistical analysis plan published by Brenner et al; Trials; (20:467; 2019);

  • DOI:10.1186/s13063-019-3561-7
  • Exact statistical tests applied – not published.


Death: No difference

TXA: 222/5956 [4%]       vs           placebo 226/5981 [4%]

(RR 0·99, 95% CI 0·82–1·18)

ATE: No difference

TXA 42/5952 [0.7%]       vs          placebo 46/5977 [0.8%]

(RR 0·92, 95% CI 0·60 to 1·39)

VTE: higher in TXA group

TXA 48/5952 [0.8%]       vs           26/5977 [0.4%]

RR 1·85; 95% CI 1·15 to 2·98).


TXA did not reduce death from gastrointestinal bleeding.

  • Ruled out large mortality reduction suggested by the Cochrane systematic review (2012)

TXA should not be used outside the context of a randomised trial.


  • use of the pre-endoscopy Rockall score might have misclassified baseline risk.
  • strict criteria to diagnose thromboembolic event (i.e. positive ultrasound imaging or post mortem) might have led to some under reporting.
  • in gastrointestinal bleeding it is difficult to determine the time of bleeding onset, presentation is often delayed. This is unlike in trauma / obstetrics where TXA can be given ≤3h.
  • cannot rule out a modest increase or decrease in death due to bleeding with tranexamic acid.



  • Good overall design and well powered for 1° outcome.
  •  Multi-centred, multinational.
  •  Baseline characteristics well matched
  •  Minimal loss to follow up (<1%) & small number of protocol violations


  • Enrolment by principle of uncertainty. Unclear if this over or under reports the benefits/risks of TXA where if patient outcome is more clearly indicated/contraindicated.
  • VTE incidence very small for large sample, data lacking clarity on timing and severity.
  • Uncertainty for patients with liver disease with their complex coagulation picture
  • No recommendation for or against TXA usage identified by various gastroenterology colleges of UK / USA / Canada / Europe

Does this change practice?

  • Raised awareness for caution in GI bleeding cases when uncertain of TXA benefit/risk.
  • VTE risk needs to be factored into decision making process, but small incidence numbers suggest risk not as big as previously thought.
  • To look into alternative/additional data (ie TEG) to assess VTE risk

Summary by Dr J He. Journal Club Meeting 2 February 2023.

Comments are closed.

Blog at WordPress.com.

Up ↑

%d bloggers like this: