Upper Gastrointestinal Hemorrhage: Treatment ControversiesUpper gastrointestinal bleeding remains a common reason for emergency department visits and is a major cause of morbidity, mortality, and medical care costs. Often when these patients arrive, the classic IV-O2-Monitor is initiated and hemodynamic stability is assessed. Some of the next steps often performed include:

  1. Determination of the site and rate of bleeding (upper vs lower)
  2. Initiation of proton pump inhibitors (PPIs)
  3. Somatostatin analogs if variceal bleeding is suspected
  4. Prophylactic antibiotics
  5. Packed red blood cell (PRBC) transfusion for low hemoglobin and hematocrit levels

What is the evidence for these treatments, and do they affect morbidity and mortality?

Does nasogastric (NG) lavage help decrease mortality or clinically relevant endpoints? 
[1] [2]

I have written about this once before on ALiEM’s NG Lavage: Indicated or Outdated?, but just in case you missed it, here are the highlights:

  • Bloody aspirate associated with high-risk lesion (OR 4.82 when compared to clear/bile aspirate)
  • Bloody aspirate associated with high-risk lesion (OR 2.8 when compared to coffee-ground aspirate)
  • 14.7% of clear/bile aspirate cases had a high risk lesion
  • Take home point: A clear NG lavage still misses almost 15% of patients with a high risk lesion.
  • 632 patients with GI bleeding had NG lavage performed to evaluate 30-day mortality, mean hospital length of stay (LOS), and transfusion requirements
  • No statistical difference in 30 day mortality, mean LOS, or transfusion requirements
  • NGL was associated with earlier time to endoscopy
  • Take home point: NGL is associated with earlier performance of endoscopy, but NO difference in clinical outcomes

NoConclusion: NG lavage DOES NOT change 30 day mortality and transfusion requirements; a clear aspirate can still miss almost 15% of high risk lesions.

Does initiation of PPIs reduce overall mortality in upper gastrointestinal bleeds (UGIB)? [3][4]

number_one_1600_clr_9875What they did:

  • Meta-analysis of six RCTs comprising 2,223 participants with unselected UGIB, undergoing active treatment with a PPI vs control (placebo or H2 blocker)
  • Outcomes evaluated at 30 days:
    • Mortality
    • Rebleeding
    • Surgery
    • Stigmata of recent hemorrhage
    • Length of hospital stay
    • Blood transfusion requirements


  • No statistically significant differences in mortality, rebleeding, or surgery between PPI and control treatment
  • Not sufficient evidence to assess for amount of blood transfused or decrease in hospitalized days

Take home point: PPI treatment initiated before endoscopy for UGIB reduces stigmata of recent hemorrhage and requirement for endoscopic therapy, but DOES NOT affect clinically important outcomes namely: mortality, rebleeding, or need for surgery.

number_two_1600_clr_9869What they did:

  • Meta-analysis of  21 randomised controlled trials comprising 2,915 patients
  • Comparison of PPI vs placebo or H2 receptor antagonist in endoscopically proven bleeding ulcer
  • Outcomes measured:
    • Mortality
    • Rebleeding
    • Surgical intervention


  • PPI treatment significantly reduced rates of:
    • Surgical intervention: 8.4% on PPI vs 13.0% control
    • Rebleeding: 10.6% on PPI vs 18.7% with control
  • NO significant difference in mortality rates between PPI vs control

Take home point: PPI treatment in peptic ulcer bleeding reduces rebleeding and surgical intervention rates, but has no effect on mortality.

NoConclusion: PPI treatment in undifferentiated UGIB does NOT improve mortality,* but in the subcategory of PUD may improve clinically relevant outcomes such as rebleeding and need for surgical intervention.

The NNT* Interestingly, The NNT did a review of this meta-analysis and found that PPIs may have a mortality benefit in the Asian populations and more harmful (or unhelpful) in the European studies. 

Also checkout this great podcast from Ken Milne over at The Skeptics Guide to Emergency Medicine on PPIs.

Does initiation of Somatostatin Analogues reduce overall mortality in UGIB from esophageal varices? [5][6]

number_one_1600_clr_9875What they did:

  • Meta-analysis of included 21 trials with a total of 2,588 patients comparing somatostatin or analogues with placebo or no treatment
  • Outcomes measured:
    • Mortality
    • Blood transfusions
    • Balloon tamponade
    • Rebleeding


  • No significant reduction in mortality
  • Reduction of 0.7 units of blood required
  • Not enough reports to comment on balloon tamponade

Take home points: Somatostatin analogues do not reduce mortality in UGIB from esophageal varices, but may reduce blood transfusion requirements.

number_two_1600_clr_9869What they did:

  • Meta-analysis of  eight trials involving 939 patients comparing EGD alone vs EGD plus somatostatin analogue therapy
  • Outcomes measured:
    • Initial hemostasis
    • 5-day hemostasis
    • 5-day mortality
    • Adverse events


  • Combined treatment (EGD + somatostatin analogue) improved: initial hemostasis (NNT = 8) and 5-day hemostasis (NNT = 5)
  • No increase in adverse events in either group
  • No difference in 5-day mortality

Take home points: Somatostatin analogues improve the efficacy of endoscopic therapy to achieve initial control of bleeding and 5-day hemostasis, yet fail to affect mortality.

NoConclusion: Somatostatin analogues DO NOT reduce mortality in UGIB from esophageal varices, but may reduce bleeding when combined with endoscopy.

Do prophylactic antibiotics reduce overall mortality in UGIB in cirrhotic patients? [7]

What they did:

  • Meta-analysis of 12 trials including 1,241 patients evaluated with antibiotic prophylaxis vs placebo or no antibiotic prophylaxis with cirrhotics having UGIB


  • Antibiotic prophylaxis versus no intervention or placebo was associated with beneficial effects on:
    • All-cause mortality
    • Mortality from bacterial infections
    • Rebleeding
    • Days of hospitalization
    • Bacteremia
    • Spontaneous bacterial peritonitis
  • No choice of antibiotic regimen for gram-negative infections was more superior to others

check_mark_green_1600_clr_3296Conclusion: Prophylactic antibiotic use in patients with cirrhosis and UGIB significantly reduces bacterial infections, all-cause mortality, bacterial infection mortality, rebleeding events, and hospitalization length.

Does transfusion of PRBCs reduce overall mortality in UGIB? [8]

What they did:

  • 921 patients with acute UGIB were randomly assigned to a restrictive strategy (transfusion only when the hemoglobin level fell below 7 g/dL) vs a liberal strategy (transfusion when the hemoglobin fell below 9 g/dL)


  • Survival at 6 weeks: 95% (restrictive) vs 91% (liberal)
  • Further bleeding: 10% (restrictive) vs 16% (liberal)

check_mark_green_1600_clr_3296Conclusion: A restrictive transfusion strategy significantly improves mortality in patients with acute UGIB.


Also checkout this great Podcast from Ken Milne over at The Skeptics Guide to Emergency Medicine on Restrictive Transfusion Strategy.


Summary of Treatment Modalities Effect on Mortality in UGIB

check_mark_green_1600_clr_3296Treatments that DO improve mortality in UGIB:

  • Antibiotic prophylaxis in cirrhotic patients
  • Restrictive transfusion strategy (transfuse if HgB < 7 g/dL)

x_symbol_1600_clr_5130Treatments that DO NOT improve mortality UGIB:

  • Somatostatin analogues
  • NG lavage
  • PPIs (although may have some mortality benefit in Asian studies)
  1. [ref id=”21737077″ link=true]
  2. [ref id=”14745388″ link=true]
  3. [ref id=”20614440″ link=true]
  4. [ref id=”15266462″ link=true]
  5. [ref id=”18677774″ link=true]
  6. [ref id=”11870374″ link=true]
  7. [ref id=”21707680″ link=true]
  8. [ref id=”23281973″ link=true]




Salim Rezaie, MD

Salim Rezaie, MD

ALiEM Associate Editor
Clinical Assistant Professor of EM and IM
University of Texas Health Science Center at San Antonio
Founder, Editor, Author of R.E.B.E.L. EM and REBEL Reviews