Villanueva: Transfusion strategies for acute upper gastrointestinal bleeding

Villanueva, N Engl J Med 2013; 368(1):11-21

Clinical Question

  • In patients who present with an upper gastrointestinal (GI) bleed, does a restrictive transfusion goal compared with a liberal transfusion goal affect mortality?


  • Randomised controlled trial
  • Computer generated, block randomisation
  • Stratified according to presence of cirrhosis
  • Non-blinded
  • Follow up: 45 days
  • Analysis: Intention to treat


  • Single hospital, Spain
  • June 2003 – December 2009


  • Inclusion: adult patients with haematemesis or malaena
  • Exclusion: minor or massive bleeding; history of stroke/TIA/acute coronary syndrome/symptomatic peripheral vasculopathy within last 90 days; recent trauma/surgery; patients with Rockall score of 0 with Hb>12g/dL
  • 921 patients randomised out of 2372 patients admitted with gastrointestinal bleeding


  • Restrictive strategy (n=461): transfusion threshold of 7g/dl
    • target of 7-9g/dl


  • Liberal strategy (n=460): transfusion threshold of 9g/dl
    • target of 9-11g/dl
All patients underwent emergent gastroscopy within first 6 hours of presentation in addition to other appropriate interventions e.g somatostatin, prophylactic antibiotics and definitive management of bleeding point (band ligation, injection with adrenaline or cyanoacrylate). Hb levels were measured every 8 hours for the first 2 days and then daily thereafter unless further bleeding


  • Primary outcome: All-cause mortality at 45 days
    • Mortality was significantly reduced in the restrictive strategy group compared to the liberal strategy group (5% vs. 9%, P=0.02)
  • Secondary outcomes:
    • Rate of further bleeding associated with haemodynamic instability or Hb drop of 2 or more within 6 hours: Significantly lower in restrictive group (10%) vs. 16% in liberal group, P=0.01
    • Fluids & blood products
      • RBC transfusion: 49% vs. 86% (P=<0.0001)
        • mean (+/-SD) no of units transfused was 1.5 (+/-2.3) vs. 3.7(+/-3.8)
      • FFP: 6% vs. 9% (P=0.13)
      • Platelets: 3% vs. 4% (P=0.27)
      • Crystalloids in first 72 hours: 5.5L vs. 5.8L (P=0.19)
      • Colloids: 19% vs. 21% (P=0.62)
  • Subgroup analysis
    • All-cause mortality at 45 days according to subgroups
      • Child Pugh A or B 4% vs. 12% (P=0.02)
      • Child Pugh C 38% vs. 41% (P=0.91)
      • Variceal bleeding 11% vs. 18% (P=0.18)
      • Peptic ulcer disease (PUD) 3% vs. 5% (P=0.26)
    • Probability of survival
      • Slightly higher with the restrictive strategy than with the liberal strategy in the subgroup of patients who had bleeding associated with PUD (hazard ratio, 0.7; 95% CI 0.26-1.25)
      • Significantly higher in patients with cirrhosis and Child Pugh A or B (hazard ratio, 0.3; 95% CI, 0.11-0.85)
      • No different if cirrhotic and Child Pugh C (hazard ratio, 1.04; 95% CI 0.45-2.37)
    • Within first 5 days, the portal pressure gradient increased significantly in the liberal transfusion group compared with those assigned to the restrictive group (P=0.03)
    • Overall rate of complications: Significantly lower in restrictive group (n = 179; 40%) vs. liberal (n=214; 48%) P=0.02
      • Transfusion reactions and cardiac events (mainly pulmonary oedema) more frequent in the liberal-strategy group


  • Randomised, computer generated
  • Appropriate inclusion/exclusion criteria
  • Appropriate power calculation


  • Single centre, non-blinded
  • >700 patients admitted with GI bleeding were not screened for inclusion in trial
  • All patients had an endoscopy within 6 hours. These results should not be generalised to centres where this does not happen
  • Transfusion protocol violated in 9% of restricted group vs. 3% in liberal group, P<0.001
  • Transfusion of packed red cells rather than packed red cells in combination with other blood products (plasma and platelets +/- cryoprecipitate) may have caused a coagulopathy in the liberal transfusion group

The Bottom Line

  • Among patients with an upper GI bleed, a restrictive strategy for blood transfusion can be the standard treatment, providing early source control with an endoscopy is achievable. The study findings are as much about highlighting the importance of 24 hour emergency endoscopy facilities as the resuscitation management of patients with upper GI bleeding.


Full text pdf / abstract / doi: 10.1056/NEJMoa1211801

Editorial, Commentaries or Blogs

  • None


Summary author: @DavidSlessor
Summary date: 9 May 2014
Peer-review editor: @stevemathieu75

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.