Transfusion Strategies for Acute Upper Gastrointestinal Bleeding

From Wiki Journal Club
Jump to: navigation, search
Villanueva C et al. "Transfusion strategies for acute upper gastrointestinal bleeding". The New England Journal of Medicine. 2013. 368(1):11-21.
PubMedFull textPDF

Clinical Question

In patients with an upper GI bleed, how does a restrictive transfusion goal compare with a liberal transfusion goal in regards to mortality?

Bottom Line

Among patients with acute upper GI bleeding, a restrictive transfusion threshold (Hgb ≥7) was associated with reduced mortality at 45 days compared to a liberal transfusion threshold (Hgb ≥10).

Major Points

TRICC (1999) changed transfusion practices among critically ill patients when it demonstrated improved survival with a restrictive transfusion strategy (Hgb ≥7 g/dL). However, actively bleeding patients were excluded from TRICC, and the Hgb threshold for transfusion in patients with acute UGIB was controversial.

This trial randomized 921 patients with acute UGIB in a single center in Spain to either a restrictive strategy (transfusion to maintain Hgb ≥7 g/dL) or a liberal strategy (Hgb ≥9 g/dL). Most patients presented with UGIB secondary to PUD (49%) followed by gastroesophageal varices (24%). About half of the patients in the restrictive group avoided transfusions, as opposed to 14% of those in the liberal group (1.5 vs. 3.7 mean transfusions per patient). As a result, restrictive strategy was associated with a 45% relative risk reduction in all-cause mortality at 45 days (5% vs. 9%), shorter LOS (9.6 vs. 11.5 days), and fewer adverse events. There were also fewer rebleeding events in the restrictive group, particularly among patients with suspected esophageal varices. However, among patients with severe cirrhosis (Child-Pugh class C), subgroup analysis did not demonstrate survival benefit nor decreased risk of further bleeding.

This trial argues for a restrictive transfusion strategy, even among patients with an acute upper GI bleed.


  • Single center, non-blinded, parallel group, randomized, controlled trial
  • N=921 patients with acute UGIB
    • Restrictive strategy (n=461)
    • Liberal strategy (n=460)
  • Setting: One center in Barcelona, Spain
  • Enrollment: 2003-2009
  • Follow-up: 45 days
  • Analysis: Intention-to-treat
  • Primary outcome: All-cause mortality at 45 days


Inclusion Criteria

  • Age ≥18 years
  • Hematemesis and/or melena

Exclusion Criteria

  • Declined blood transfusion
  • Massive exsanguinating bleeding requiring transfusions
  • Transfusion in prior 90 days
  • ACS
  • Symptomatic peripheral vasculopathy
  • Stroke or TIA
  • Recent trauma or surgery
  • Lower GI bleeding
  • Rockall score[1] of 0 with Hgb >12g/dL (patients at low risk of further bleeding)

Baseline Characteristics

  • Admission Hgb: 9.5 g/dL
  • Source of bleeding
    • Peptic ulcer: 49% (gastric 34%, duodenal 63%; visible vessel 56%, active bleeding 16%)
    • Gastroesophageal varices: 24%
    • Mallory-Weiss tears: 7%
    • Erosive gastritis or esophagitis: 8%
    • Malignancy: 4%
  • Cirrhosis: 31%
    • Child-Pugh A (27%), B (55%), C (19%)
    • Alcoholic: 45%
    • Source of bleeding:
      • Esophageal varices: 69%
      • Gastric varices: 8%
      • PUD: 14%
  • In-hospital bleeding: 6%
  • Rockall score: 5.4


  • Stratified according to presence or absence of cirrhosis, patients were randomized to either:
    • Restrictive strategy: Hgb transfusion threshold <7 g/dL targeting post-transfusion 7-9 g/dL
    • Liberal strategy: Hgb transfusion threshold <9g/dL targeting post-transfusion 9-11 g/dL
  • All patients underwent emergent EGD in first 6 hours with appropriate interventions:
    • Somatostatin and prophylactic norfloxacin or ceftriaxone for portal hypertension
    • Band ligation or sclerotherapy for bleeding esophageal varices. Cyanoacrylate injection for bleeding gastric varices. Portal pressures estimated with HPVG[2].
    • Omeprazole for PUD
    • Epinephrine plus electrocoagulation or clips for non-variceal bleeding lesion, non-bleeding visible vessel, or adherent clot
  • Hgb levels were measured every 8 hours for first 2 days then daily thereafter unless further bleeding
  • Transfusions were also allowed for symptomatic anemia, massive bleeding or surgery.


Comparisons are restrictive vs. liberal strategies.

Primary Outcome

All-cause mortality at 45 days
5% vs. 9% (HR 0.55; 95% CI 0.33-0.92; P=0.02; NNT=25)

Secondary Outcomes

Further bleeding associated with hemodynamic instability or Hgb drop ≥2 within 6 hours
10% vs. 16% (HR 0.62; 95% CI 0.43-0.91; P=0.01)
RBC transfusions
49% vs. 86% (P<0.001)
1.5 vs. 3.7 transfusions per patient (P<0.001)
1.2 vs. 2.9 transfusions during initial bleeding (P<0.001)
Other transfusions
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)
Hemoglobin level
Lowest Hgb <7 g/dL: 45% vs. 18% (P<0.001)
At discharge: 9.2 g/dL vs. 10.1 g/dL (P<0.001)
At 45 days: 11.6 g/dL vs. 11.7 g/dL (P=0.67)
Mean LOS
9.6 vs. 11.5 days (P=0.01)

Subgroup Analysis

All-cause mortality at 45 days, according to subgroups
Child-Pugh class A or B: 4% vs. 12% (P=0.02)
Child-Pugh class C: 38% vs. 41% (P=0.91)
Variceal bleeding: 11% vs. 18% (P=0.18)
PUD: 3% vs. 5% (P=0.26)
Further bleeding, among patients with cirrhosis
Child-Pugh A or B: 11% vs. 21% (HR 0.53; P=0.04)
Child-Pugh C: 15% vs. 28% (HR 0.48; P=0.33)
Esophageal varices: 11% vs. 22% (HR 0.50; P=0.05)
Treatment with balloon tamponade: 2% vs. 8% (P=0.03)
Treatment with TIPS: 4% vs. 11% (P=0.04)
Further bleeding, among patients with PUD
10% vs. 16% (HR 0.63; P=0.09)
Treatment with second endoscopy: 9% vs. 12% (P=0.21)
Treatment with emergency surgery: 2% vs. 6% (P=0.04)

Adverse Events

Transfusion reactions
3% vs. 9% (HR 0.35; P=0.001)
Cardiac complications
11% vs. 16% (HR 0.64; P=0.04)
Pulmonary complications
11% vs. 12% (P=0.67)
18% vs. 22% (P=0.13)
Stroke or TIA
1% vs. 1% (P=0.33)
Bacterial infections
27% vs. 30% (P=0.41)


  • Patients with massive exsanguinating bleeding or low rebleeding risk were excluded.
  • All patients underwent emergent EGD within mean 5 hours of admission; therefore, it is unknown whether results would be similar in patients who do not receive endoscopic therapy as quickly as the patients in this trial.
  • Subgroup analysis demostrates that the benefit extends beyond those with portal hypertension [3]
  • Upper endoscopies within 6 hours is unrealistic outside of a research setting[4]
  • 8% of patients in the restrictive group received transfusions for symptoms/signs, massive bleeding, or surgery - irrespective of hemoglobin value, likely skewing results[4]
  • Transfusion of pRBC rather than pRBC with plasma and platelets may have caused a coagulopathy in the liberal-transfusion group[4]


  • Fundació Investigació Sant Pau
  • Dr. Guarner with a monetary conflict of interest

Further Reading

  1. Rockall TA et al. "Risk assessment after acute upper gastrointestinal haemorrhage." Gut 1996;38:316-21.
  2. HVPG definition at Journal of Online Hepatology
  3. Concurrent editorial
  4. 4.0 4.1 4.2 NEJM Letters to the Editor