← EMA Topics
EMA Topics · Gastrointestinal · Resuscitation

Upper GI bleeding

Updated Oct 17, 2026Adults · ED management23 sources · 16 checked in PubMed2 figures

Bottom line

  • A Glasgow-Blatchford score of 0–1 marks very low risk. These patients may go home with outpatient follow-up.1,2 In 3,012 patients, a score of 1 or less missed only 1.4% of those who needed intervention or died.3
  • Transfuse restrictively. Restrictive transfusion at a hemoglobin below 7 g/dL improved 6-week survival over a threshold of 9 g/dL (95% vs 91%).4 Guidelines set the line at 7 or 8 g/dL.1,2
  • Do not give tranexamic acid. In 12,009 patients it did not reduce death from bleeding, and venous clots were more common.5
  • Endoscopy within 24 hours after resuscitation. In high-risk patients, endoscopy within 6 hours did not lower 30-day mortality.1,6,7
  • Suspected varices: start a vasoactive drug and antibiotics now. Aim for endoscopy within 12 hours.8

First steps

  1. Resuscitate if unstable. Start volume resuscitation to restore and maintain perfusion.2,8 For massive bleeding, use your massive transfusion protocol.9
  2. Protect the airway when needed. In variceal bleeding, intubate before endoscopy for altered consciousness or active vomiting of blood.8 Otherwise avoid intubation if you can. If you must, resuscitate first, preoxygenate, and have suction ready.10
  3. Ask about liver disease. If varices are possible, start a vasoactive drug and antibiotic prophylaxis at presentation.8,9
  4. Get hemoglobin and BUN, and score the Glasgow-Blatchford at first assessment.9,11
  5. Transfuse restrictively (next section).1,2
  6. Check anticoagulants and antiplatelets, but do not delay endoscopy for them.2
  7. Call GI. Plan endoscopy within 24 hours, or within 12 hours if varices are suspected.6,8

Risk stratification

Use the Glasgow-Blatchford score (GBS) before endoscopy.6,9 It uses only bedside and lab findings. Scores range from 0 to 23.7

ItemFindingPoints
BUN (mg/dL)
BUN< 18.20
18.2–22.32
22.4–283
28–704
> 706
Hemoglobin (g/dL)
Men> 13 / 12–13 / 10–12 / < 100 / 1 / 3 / 6
Women> 12 / 10–12 / < 100 / 1 / 6
Systolic BP (mm Hg)
Systolic BP≥ 110 / 100–109 / 90–99 / < 900 / 1 / 2 / 3
Other findings
Pulse ≥ 100/minPresent1
MelenaPresent1
Syncope at presentationPresent2
History of liver diseasePresent2
Cardiac failurePresent2

Items and points as listed by MDCalc, which converts the original urea bands to BUN in mg/dL.11

  • It finds who needs a procedure. In 3,012 patients at 6 hospitals, the GBS predicted intervention or death better than the Rockall and AIMS65 scores (AUROC 0.86). A score of 1 or less had a sensitivity of 98.6%.3
  • A score of 1 or less is common. It applied to 19% of patients. Two of these 564 patients (0.4%) died within 30 days.3
  • It holds up in ED patients. In 990 patients admitted through Paris emergency departments, a score of 1 or less had a sensitivity of 0.99 for intervention or death.12
  • It does not predict death well. For 30-day mortality, its AUROC was only 0.64.3
  • It changes practice. When a GBS-based protocol was introduced at 2 UK hospitals, 84 of 123 low-risk patients went home without adverse events. Admissions for this bleeding fell from 96% to 71%.13
  • After endoscopy, NICE adds the full Rockall score.9 The International Consensus Group suggests against AIMS65 to pick out very-low-risk patients.2
Sources disagree

Which score is low enough to go home? NICE says consider early discharge only at a GBS of 0.9 The ACG suggests 0–1, and the International Consensus Group and ESGE use 1 or less.1,2,6 The original validation used 0.13 The larger 2017 study found 1 or less was the best cutoff.3 Follow your local protocol.

Resuscitation and transfusion

  • Threshold: the ACG suggests red cells at a hemoglobin of 7 g/dL for hospitalized patients.1 The International Consensus Group suggests below 8 g/dL (80 g/L) without cardiovascular disease.2
  • Cardiovascular disease: the International Consensus Group suggests a higher threshold but gives no number.2 An emergency medicine review uses 8 g/dL with myocardial ischemia.10
  • Variceal bleeding: transfuse conservatively, with a target of 7–8 g/dL.8
  • Unstable with heavy bleeding: transfuse now, whatever the hemoglobin.10 Base decisions on the whole picture. Over-transfusion may harm as much as under-transfusion.9

The transfusion trial

In 921 patients with severe upper GI bleeding, a restrictive threshold (below 7 g/dL) was compared with a liberal one (below 9 g/dL). Half of the restrictive group (51%) never needed blood. Restrictive care gave better 6-week survival (95% vs 91%; HR 0.55), less further bleeding (10% vs 16%), and fewer adverse events (40% vs 48%).4 In cirrhosis, the benefit was seen in Child-Pugh A and B but not in Child-Pugh C. Portal pressure rose only in the liberal group.4

Sources disagree

7 or 8 g/dL? The ACG uses 7 g/dL.1 The International Consensus Group uses 8 g/dL, and higher with cardiovascular disease.2 Baveno VII aims for 7–8 g/dL in variceal bleeding.8 All of them advise against liberal transfusion.

Platelets, plasma and anticoagulants

  • Platelets: give them to patients who are actively bleeding with a count below 50,000/µL. Do not give them to stable patients who are not bleeding.9
  • Plasma: NICE gives fresh frozen plasma for active bleeding with a PT (INR) or aPTT over 1.5 times normal, then cryoprecipitate if fibrinogen stays below 150 mg/dL (1.5 g/L).9
  • Warfarin: offer prothrombin complex concentrate if the patient is actively bleeding.9
  • Apixaban or rivaroxaban: NICE lists andexanet alfa as an option for life-threatening or uncontrolled GI bleeding.9
  • Do not wait on reversal. Do not delay endoscopy in patients on anticoagulants.2 The group could not name an INR level that should prompt correction. Vitamin K given around endoscopy was linked to more thromboembolism and showed no link with rebleeding.2
  • Low-dose aspirin for secondary prevention: ESGE advises not stopping it. If it is stopped, restart it within 3–5 days.6
  • Restarting anticoagulants: resume as soon as bleeding is controlled, ideally within or soon after 7 days, based on clot risk.6
Sources disagree

Plasma in cirrhosis. NICE offers plasma for a high INR during active bleeding.9 Baveno VII says standard clotting tests do not reflect hemostasis in advanced liver disease. It advises against plasma in variceal bleeding, because plasma does not correct the coagulopathy and adds volume.8

Tranexamic acid: no

HALT-IT randomized 12,009 adults with significant upper or lower GI bleeding. They got tranexamic acid (1 g over 10 minutes, then 3 g over 24 hours) or placebo. Death from bleeding within 5 days was 4% in both groups (RR 0.99). Venous thromboembolism was higher with tranexamic acid (0.8% vs 0.4%; RR 1.85).5 Baveno VII does not recommend it for variceal bleeding.8

Drugs before endoscopy

DrugAdult doseWhenNotes
All upper GI bleeding
Erythromycin IV250 mg IV30–120 min before endoscopySuggested by the ACG.1 Dose from Baveno VII (variceal bleeding), which advises it only without QT prolongation.8 The label warns of QT prolongation and torsades.14
IV PPI (e.g., pantoprazole)Dose per institutional protocol and labelAt presentation, if usedMay be considered, but must not delay endoscopy.2,15 Bleeding is not a labeled use.16
IV PPI after endoscopic hemostasis80 mg IV bolus, then 8 mg/hFor 72 hoursFor high-risk stigmata treated at endoscopy.2,6
Suspected variceal bleeding
Octreotide (or another vasoactive drug)Dose per institutional protocol and labelAs soon as possibleContinue 2–5 days.8 Bleeding is not a labeled use, and the label gives no dose for it.17
Ceftriaxone IV1 g every 24 hFrom admissionFor advanced cirrhosis. Adjust to local resistance.8 The usual labeled adult dose is 1–2 g daily.18
Not recommended
Tranexamic acidDo not give—No benefit and more venous clots.5,8
Octreotide for ulcer bleedingNot routine—Not routinely recommended for peptic ulcer bleeding.2

Unreviewed doses. Verify against the current label and your institution's protocol before giving anything.

Erythromycin

It empties the stomach of blood so the endoscopist can see. In 6 trials (558 patients), it improved the view of the gastric mucosa (OR 3.43) and cut the need for a second endoscopy (OR 0.47). It also slightly reduced blood use and hospital stay.19 If erythromycin is not available, ESGE suggests IV metoclopramide in selected patients with severe or ongoing bleeding.15

PPI before endoscopy

In 6 trials (2,223 patients), a PPI before endoscopy did not clearly change mortality (OR 1.14), rebleeding or surgery. It likely reduced the need for endoscopic treatment at the first endoscopy (OR 0.68, moderate certainty).20 In variceal bleeding, stop a PPI started before endoscopy once the procedure is done, unless there is another reason for it.8

Sources disagree

PPI before endoscopy. NICE says not to offer acid suppression before endoscopy for suspected non-variceal bleeding.9 The International Consensus Group says it may be considered.2 ESGE 2026 suggests considering high-dose IV PPI.15 All agree it must not delay endoscopy.

Sources disagree

Erythromycin before endoscopy. The ACG suggests it.1 The International Consensus Group keeps its older advice not to use promotility agents routinely.2

Variceal bleeding

  • Vasoactive drug as soon as possible (terlipressin, somatostatin or octreotide), continued for 2–5 days.8 Across 30 trials (3,111 patients), vasoactive drugs lowered 7-day mortality (RR 0.74) and transfusion needs. No single agent was better than another.21
  • Antibiotics from admission. IV ceftriaxone 1 g every 24 hours in advanced cirrhosis.8 In 12 trials (1,241 patients), prophylaxis lowered mortality (RR 0.79), bacterial infections (RR 0.36) and rebleeding (RR 0.53). No antibiotic was better than another.22 In Child-Pugh A cirrhosis, the risk of infection and death is very low.8
  • Transfuse to 7–8 g/dL. Do not give plasma.8
  • Endoscopy within 12 hours of presentation once resuscitated. If the patient stays unstable, scope as soon as it is safe.8 Band ligation is the treatment for esophageal varices.9
  • Pass tubes gently. Be careful with airway manipulation and nasogastric tubes.8
  • Pre-emptive TIPS within 72 hours (ideally under 24) for Child-Pugh C under 14 points, or Child-Pugh B over 7 with active bleeding at endoscopy.8
  • Refractory bleeding: a balloon tube or an esophageal stent can bridge to definitive treatment. Baveno VII calls the stent just as effective and safer.8
  • Prevent hepatic encephalopathy: clear blood from the gut quickly with lactulose by mouth or enema.8
FIG 1Time targets in the first 24 hours. Drawn for this page from the guidelines' time targets.8,6,15 Endoscopy within 6 hours did not lower mortality in the trial, where time was counted from GI consultation.7 Pre-emptive TIPS can happen up to 72 hours. Vasoactive drugs and antibiotics continue past the chart.

Endoscopy timing

  • Within 24 hours of presentation, after resuscitation.1,2,6 NICE counts the 24 hours from admission.9
  • Not routinely within 6 or 12 hours. ESGE advises against it unless the patient stays unstable despite adequate resuscitation.15 NICE scopes unstable patients with severe bleeding right after resuscitation.9
  • The trial: 516 patients with a GBS of 12 or more got endoscopy within 6 hours or at 6–24 hours after GI consultation. Mortality at 30 days was 8.9% vs 6.6%, and further bleeding 10.9% vs 7.8%. Neither difference was significant.7
  • At endoscopy, treatment is recommended for ulcers that are spurting or oozing and for nonbleeding visible vessels.1
  • If bleeding recurs, repeat endoscopy. If endoscopic therapy fails, the next step is transcatheter embolization.1,6 NICE sends unstable patients who rebleed to interventional radiology, or to surgery if radiology is not quickly available.9
Endoscopic image of a gastric ulcer with a thin jet of bright red blood spurting from its base
FIG 2Spurting gastric ulcer. An endoscopic view of a gastric ulcer with an arterial jet of blood. This is a high-risk finding that needs endoscopic therapy.1 Cropped for this page to remove the on-screen text.Jeremias · Wikimedia Commons · CC BY-SA 3.0

Disposition

  • Very low risk (GBS 0–1, or 0 under NICE): consider discharge with outpatient follow-up and endoscopy.1,6,9
  • Everyone else: admit for endoscopy within 24 hours.1,6
  • Suspected varices: admit for endoscopy within 12 hours, with a vasoactive drug and antibiotics running.8
  • High-risk ulcer treated at endoscopy: most patients stay in hospital at least 72 hours.2
  • Low-risk findings at endoscopy: selected patients can go home soon after the procedure.2

Pitfalls

  • Giving tranexamic acid. It did not reduce death from bleeding and raised venous clot risk.5
  • Transfusing a stable patient at a threshold of 9 g/dL. That liberal strategy had more rebleeding and lower survival.4
  • Placing a nasogastric tube just to make the diagnosis. Lavage for diagnosis is not recommended.10
  • Using the GBS to predict death. It predicts the need for intervention, not mortality.3
  • Holding endoscopy to reverse an anticoagulant.2
  • Giving plasma to correct the INR in cirrhosis.8
  • Forgetting antibiotics in a patient with cirrhosis.8,22
  • Giving erythromycin to a patient with a long QT.8,14

How strong is the evidence?

Most International Consensus Group statements are conditional and rest on low or very low certainty evidence. That includes the GBS cutoff, the transfusion threshold and the 24-hour target.2 ESGE rates its 24-hour endoscopy advice as strong, with high-quality evidence.6 The ACG and ESGE full texts were not openly accessible, so their positions here come from their abstracts.1,6,15

Large trials anchor the key decisions: transfusion (921 patients), endoscopy timing (516) and tranexamic acid (12,009).4,7,5 HALT-IT pooled upper and lower GI bleeding and changed its primary outcome late in the trial. A critical-care review still judged it strong enough to stop using tranexamic acid.23 The pre-endoscopy PPI trials are older, and most of the evidence is low certainty.20

On-shift teaching

Questions to ask a learner at the bedside. Tap a question to see the answer. Every answer uses facts cited above.

Q1How safe is a Glasgow-Blatchford score of 1 or less?
In 3,012 patients, it had a sensitivity of 98.6% for intervention or death. It applied to 19% of patients, and 0.4% of them died within 30 days.3
Q2What did the restrictive transfusion trial show?
Transfusing below 7 g/dL instead of 9 g/dL gave better 6-week survival (95% vs 91%) and less further bleeding (10% vs 16%). Half of the restrictive group never needed blood.4
Q3Why not give tranexamic acid for a big GI bleed?
In HALT-IT (12,009 patients), death from bleeding was 4% in both groups. Venous thromboembolism was higher with tranexamic acid (0.8% vs 0.4%).5
Q4Should a high-risk patient (GBS 12 or more) be scoped within 6 hours?
Not routinely. Endoscopy within 6 hours of GI consultation did not lower 30-day mortality compared with 6–24 hours (8.9% vs 6.6%).7 Guidelines target 24 hours after resuscitation.6
Q5Why give antibiotics to a patient with cirrhosis and a GI bleed?
In 12 trials, prophylaxis lowered mortality (RR 0.79), bacterial infections (RR 0.36) and rebleeding (RR 0.53).22 Baveno VII suggests IV ceftriaxone 1 g every 24 hours in advanced cirrhosis.8
Q6What does a PPI before endoscopy actually do?
It likely reduces the need for endoscopic treatment at the first endoscopy (OR 0.68). It has not been shown to change mortality, rebleeding or surgery.20

References

  1. 1
    Laine L, Barkun AN, Saltzman JR, Martel M, Leontiadis GI. ACG Clinical Guideline: upper gastrointestinal and ulcer bleeding. Am J Gastroenterol. 2021;116(5):899–917.GuidelinePMID 33929377DOIPubMed record; full text not openly accessible
  2. 2
    Barkun AN, Almadi M, Kuipers EJ, et al. Management of nonvariceal upper gastrointestinal bleeding: guideline recommendations from the International Consensus Group. Ann Intern Med. 2019;171(11):805–822.GuidelinePMID 31634917PMCFull text read
  3. 3
    Stanley AJ, Laine L, Dalton HR, et al. Comparison of risk scoring systems for patients presenting with upper gastrointestinal bleeding: international multicentre prospective study. BMJ. 2017;356:i6432.CohortPMID 28053181PMCFull text read
  4. 4
    Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med. 2013;368(1):11–21.RCTPMID 23281973PubMed
  5. 5
    HALT-IT Trial Collaborators. 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. Lancet. 2020;395(10241):1927–1936.RCTPMID 32563378PMCPubMed
  6. 6
    Gralnek IM, Stanley AJ, Morris AJ, et al. Endoscopic diagnosis and management of nonvariceal upper gastrointestinal hemorrhage (NVUGIH): European Society of Gastrointestinal Endoscopy (ESGE) Guideline – Update 2021. Endoscopy. 2021;53(3):300–332.GuidelinePMID 33567467PubMed record; full text not openly accessible
  7. 7
    Lau JYW, Yu Y, Tang RSY, et al. Timing of endoscopy for acute upper gastrointestinal bleeding. N Engl J Med. 2020;382(14):1299–1308.RCTPMID 32242355PubMed
  8. 8
    de Franchis R, Bosch J, Garcia-Tsao G, et al. Baveno VII – Renewing consensus in portal hypertension. J Hepatol. 2022;76(4):959–974.ConsensusPMID 35120736PMCFull text read
  9. 9
    National Institute for Health and Care Excellence. Acute upper gastrointestinal bleeding in over 16s: management (CG141). Published 2012, last updated 2016, with later edits to the recommendations page.Guidelinenice.org.ukPage read
  10. 10
    Long B, Gottlieb M. Emergency medicine updates: upper gastrointestinal bleeding. Am J Emerg Med. 2024;81:116–123.ReviewPMID 38723362PubMed
  11. 11
    MDCalc. Glasgow-Blatchford Bleeding Score (GBS). Score items only.FOAMmdcalc.comPage read
  12. 12
    Thiebaud PC, Wassermann E, de Caluwe M, et al. Assessment of prognostic scores for emergency department patients with upper gastrointestinal bleeding. Ann Emerg Med. 2025;85(1):31–42.CohortPMID 39093247PubMed
  13. 13
    Stanley AJ, Ashley D, Dalton HR, et al. Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation. Lancet. 2009;373(9657):42–47.CohortPMID 19091393PubMed
  14. 14
    Erythromycin lactobionate for injection, USP. Prescribing information (Piramal Critical Care). DailyMed.Drug labelDailyMedLabel text read
  15. 15
    Gralnek IM, Morris J, Laursen SB, et al. Endoscopic diagnosis and management of peptic ulcer bleeding: European Society of Gastrointestinal Endoscopy (ESGE) Guideline – Update 2026. Endoscopy. 2026;58(8):899–924.GuidelinePMID 42127996PubMed record; full text not openly accessible
  16. 16
    Pantoprazole sodium for injection. Prescribing information (Avenacy). DailyMed.Drug labelDailyMedLabel text read
  17. 17
    Octreotide acetate injection. Prescribing information (Hikma). DailyMed.Drug labelDailyMedLabel text read
  18. 18
    Ceftriaxone for injection. Prescribing information (Hikma). DailyMed.Drug labelDailyMedLabel text read
  19. 19
    Theivanayagam S, Lim RG, Cobell WJ, et al. Administration of erythromycin before endoscopy in upper gastrointestinal bleeding: a meta-analysis of randomized controlled trials. Saudi J Gastroenterol. 2013;19(5):205–210.Meta-analysisPMID 24045593PMCPubMed
  20. 20
    Kanno T, Yuan Y, Tse F, Howden CW, Moayyedi P, Leontiadis GI. Proton pump inhibitor treatment initiated prior to endoscopic diagnosis in upper gastrointestinal bleeding. Cochrane Database Syst Rev. 2022;1(1):CD005415.Meta-analysisPMID 34995368PMCPubMed
  21. 21
    Wells M, Chande N, Adams P, et al. Meta-analysis: vasoactive medications for the management of acute variceal bleeds. Aliment Pharmacol Ther. 2012;35(11):1267–1278.Meta-analysisPMID 22486630PubMed
  22. 22
    Chavez-Tapia NC, Barrientos-Gutierrez T, Tellez-Avila FI, et al. Antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding. Cochrane Database Syst Rev. 2010;(9):CD002907.Meta-analysisPMID 20824832PMCPubMed
  23. 23
    Walker G. HALT-IT: tranexamic acid in acute gastrointestinal bleeding. The Bottom Line; June 2020.FOAMthebottomline.org.ukPage read
How this page was made
  • Written from scratch. No source text was copied; facts are restated and cited.
  • Anchor guideline: the International Consensus Group recommendations (2019), read in full on PubMed Central. Variceal care comes from the Baveno VII consensus, also read in full. NICE CG141 recommendations were read on the NICE website.
  • The ACG 2021 guideline, the ESGE 2021 and 2026 guidelines, and the AASLD 2024 portal hypertension guidance were not openly accessible. ACG and ESGE positions come from their PubMed abstracts only. AASLD guidance is not cited.
  • Each PubMed citation was resolved through NCBI E-utilities, and each number was checked against the abstract or open full text. Doses come only from guideline text read (Baveno VII, International Consensus Group) or DailyMed labels. No source gave an octreotide dose, so none is stated.
  • Glasgow-Blatchford items and points come from MDCalc, used for score items only.
  • Adults only. Children were not covered.
  • Figure 1 was drawn for this page from the guidelines' time targets. Figure 2 is an openly licensed Wikimedia Commons image, cropped to remove on-screen text; it keeps its own license.
  • Not for clinical use until physician review.
Not for clinical use

AI-generated draft. Not reviewed by a physician. This page is for education only and may contain errors, including in doses. Do not use it to diagnose, treat or make decisions about any patient. Rely on your own clinical judgment, your institution's protocols and current drug labeling. Reading this page does not create a physician-patient relationship. Full terms