Residency · Residency · Gastroenterology

Upper GI Bleeding - Evaluation and Management

Definition and Epidemiology

Upper gastrointestinal bleeding is defined as hemorrhage originating proximal to the ligament of Treitz. The incidence ranges from 50 to 150 per 100,000 population per year, and overall mortality is 2 to 10%, with a notable disparity between inpatient-onset bleeding, which carries a mortality rate of 15 to 25%, and community-onset bleeding, which has a mortality of 5 to 8%. The most common etiologies include peptic ulcer disease, accounting for 30 to 50% of cases, followed by variceal bleeding at 10 to 20%, erosive disease at 10 to 15%, Mallory-Weiss tear at 5 to 10%, malignancy at 2 to 5%, Dieulafoy lesion at 1 to 2%, and angiodysplasia at approximately 5%.

Initial Assessment and Resuscitation

Hemodynamic Evaluation

The initial assessment of a patient presenting with upper GI bleeding centers on evaluating hemodynamic stability. Signs of shock include tachycardia with a heart rate exceeding 100 beats per minute, hypotension with systolic blood pressure below 90 mmHg, and orthostatic changes defined as a systolic blood pressure drop greater than 20 mmHg or heart rate rise greater than 20 beats per minute upon standing. Additional signs of hemodynamic compromise include altered mental status, cool and clammy extremities, delayed capillary refill, and oliguria. Two large-bore peripheral intravenous lines of at least 18-gauge and preferably 16-gauge should be established immediately, as flow rate is determined by catheter diameter according to Poiseuille's law, not by the size of the vein accessed.

Resuscitation

Initial resuscitation begins with isotonic crystalloid, either normal saline or lactated Ringer solution, though care should be taken to avoid over-resuscitation in patients with suspected variceal bleeding. Blood transfusion should follow a restrictive strategy, with a transfusion trigger of hemoglobin below 7 g/dL, as demonstrated by the landmark Villanueva trial, which showed that a restrictive approach was superior to a liberal threshold of hemoglobin below 9 g/dL, with reduced mortality (5% versus 9%), reduced rebleeding, and fewer complications in non-variceal upper GI bleeding. Exceptions to the restrictive threshold include patients with active exsanguination, acute coronary syndrome, or symptomatic anemia, in whom a liberal threshold of hemoglobin below 8 to 9 g/dL is appropriate. It is important to recognize that the hemoglobin at presentation may not reflect the true severity of blood loss, as equilibration with extravascular fluid takes 24 to 72 hours.

A massive transfusion protocol should be activated when the anticipated need exceeds 10 units of packed red blood cells in 24 hours or 4 units in 1 hour, with a target ratio of 1:1:1 for packed red blood cells, fresh frozen plasma, and platelets. Coagulopathy should be corrected in the setting of active bleeding: platelets should be transfused when the count falls below 50,000 per microliter, fresh frozen plasma is indicated for an INR exceeding 1.5 to 2.0, and cryoprecipitate should be administered when fibrinogen is below 150 mg/dL.

Anticoagulant/Antiplatelet Management

Management of anticoagulants and antiplatelet agents requires careful balancing of hemorrhagic and thrombotic risks. For patients on warfarin, endoscopy should not be delayed for supratherapeutic INR. In life-threatening bleeding, 4-factor prothrombin complex concentrate at 25 to 50 IU/kg is preferred over fresh frozen plasma for its faster onset of action, supplemented by intravenous vitamin K 10 mg administered as a slow infusion. For patients on direct oral anticoagulants, idarucizumab is the specific reversal agent for dabigatran, while andexanet alfa reverses rivaroxaban and apixaban, though it is expensive and carries prothrombotic risk. When specific reversal agents are unavailable, activated prothrombin complex concentrate at 50 IU/kg may be used.

Low-dose aspirin at 81 mg should be continued for secondary cardiovascular prevention whenever possible, as the Sung trial demonstrated a 7-fold increase in cardiac events when aspirin was held during upper GI bleeding. In patients on dual antiplatelet therapy, the P2Y12 inhibitor (clopidogrel or ticagrelor) should be held while aspirin is continued, with early cardiology consultation obtained. Anticoagulation should be restarted as soon as safely possible, typically 48 to 72 hours after hemostasis, as the thrombotic risk generally exceeds the rebleeding risk.

Risk Stratification

Pre-Endoscopy Scores

The Glasgow-Blatchford Score (GBS) is a pre-endoscopy risk stratification tool scored on a 0 to 23 scale that incorporates blood urea nitrogen, hemoglobin, systolic blood pressure, heart rate, presence of melena, syncope, hepatic disease, and heart failure. A GBS of 0 to 1 identifies very low-risk patients who can be safely managed as outpatients with elective endoscopy, with a 95% positive predictive value for no intervention being needed. A GBS of 7 or higher identifies high-risk patients who require inpatient management and urgent upper endoscopy. The AIMS65 score (Albumin below 3.0, INR above 1.5, altered Mental status, Systolic blood pressure below 90, and age above 65) predicts inpatient mortality, with 0 points corresponding to a 0.3% mortality risk and 4 or more points corresponding to 25% mortality.

Post-Endoscopy Scores

The Rockall score incorporates age, hemodynamic status, comorbidities, endoscopic diagnosis, and stigmata of recent hemorrhage to predict rebleeding and mortality. The Forrest classification provides a standardized description of peptic ulcer bleeding stigmata with corresponding rebleeding risks: Forrest Ia (spurting arterial hemorrhage) carries a 90% rebleeding risk, Forrest Ib (oozing hemorrhage) carries a 50% risk, Forrest IIa (non-bleeding visible vessel) carries a 43% risk, Forrest IIb (adherent clot) carries a 22% risk, Forrest IIc (flat pigmented spot) carries a 7% risk, and Forrest III (clean-base ulcer) carries a 3% rebleeding risk.

<image>A visual reference chart of the Forrest classification for peptic ulcer bleeding stigmata. Display 6 panels arranged in a 3x2 grid, each showing an endoscopic view of a peptic ulcer crater. Panel Ia: crater with a visible spurting arterial jet of bright red blood. Panel Ib: crater with slow oozing of dark blood from the ulcer base. Panel IIa: ulcer with a raised, dome-shaped pigmented protuberance (visible vessel) without active bleeding, shown in both head-on and side profile views. Panel IIb: ulcer with a dark reddish-brown clot adherent to the base that resists irrigation. Panel IIc: ulcer with a flat, dark spot (hematin) on the base. Panel III: clean, white-yellow fibrin base with no stigmata. Label each panel with Forrest class, description, rebleeding risk percentage, and recommended management (endoscopic therapy vs. medical). Use realistic mucosal tones and include red blood where appropriate.</image>

Pre-Endoscopy Medical Therapy

Proton Pump Inhibitors

Administration of an intravenous PPI bolus (esomeprazole or pantoprazole 80 mg) before endoscopy reduces the proportion of high-risk stigmata found at endoscopy and decreases the need for endoscopic therapy, as demonstrated in the Lau trial. However, pre-endoscopy PPI does not reduce mortality, rebleeding, or the need for surgery. The post-endoscopy role of PPI therapy is of far greater clinical importance.

Prokinetics

Intravenous erythromycin at 250 mg or 3 mg/kg administered over 20 to 30 minutes, given 30 to 90 minutes before upper endoscopy, significantly improves gastric visualization by promoting gastric emptying, reduces the need for repeat endoscopy, and decreases blood transfusion requirements. Intravenous metoclopramide at 10 mg serves as an alternative when erythromycin is unavailable, though the evidence supporting its use is weaker. Prokinetics are contraindicated in patients with QT prolongation, those taking concurrent QT-prolonging medications, and those with macrolide allergy.

Octreotide/Somatostatin (for suspected variceal bleeding)

When variceal bleeding is suspected based on clinical features such as known cirrhosis or stigmata of chronic liver disease, octreotide should be started empirically with a 50 mcg intravenous bolus followed by a continuous infusion at 50 mcg per hour for 3 to 5 days. Octreotide reduces splanchnic blood flow and portal pressure. Prophylactic antibiotics with ceftriaxone 1 g intravenously daily for 7 days should be administered concurrently in all patients with suspected variceal bleeding, as this reduces both bacterial infection and rebleeding rates.

Timing of Endoscopy

The timing of upper endoscopy is guided by clinical severity. Urgent endoscopy within 12 hours is indicated for patients with hemodynamic instability despite resuscitation, those with high-risk features such as a Glasgow-Blatchford Score of 12 or higher, and patients with inpatient-onset bleeding. Early endoscopy within 24 hours is the standard recommendation for all hospitalized patients with upper GI bleeding. Patients with a GBS of 0 to 1 who are hemodynamically stable may be managed with elective outpatient endoscopy within 2 to 7 days. Immediate endoscopy within 6 hours may be considered in suspected variceal bleeding with hemodynamic instability, although no mortality benefit of endoscopy before 6 hours compared to before 24 hours has been demonstrated in non-variceal upper GI bleeding.

Endoscopic Therapy

Peptic Ulcer Bleeding

Indications for Endoscopic Therapy

Endoscopic therapy is strongly indicated for Forrest Ia, Ib, and IIa lesions. For Forrest IIb lesions with an adherent clot, targeted irrigation should be attempted to dislodge the clot; if an underlying Forrest IIa lesion or active bleeding is revealed, endoscopic therapy should be applied. If a flat spot is found beneath the clot, no therapy is needed, though empiric treatment of adherent clots is a reasonable approach. Forrest IIc and III lesions do not require endoscopic therapy and are managed with medical therapy alone.

Modalities

Epinephrine injection using a 1:10,000 to 1:20,000 dilution is administered in 1 to 2 mL aliquots per quadrant around the visible vessel. Critically, epinephrine injection must always be combined with a second hemostatic modality, as injection monotherapy is associated with an unacceptably high rebleeding rate. Thermal coagulation using bipolar electrocoagulation (Gold probe) or heater probe employs firm tamponade combined with coagulation in a technique known as coaptive coagulation, applied in 7 to 10 second pulses at 15 to 20 watts. Through-the-scope hemoclips are deployed directly onto the visible vessel and are particularly useful for achieving definitive mechanical hemostasis, though they can be technically challenging in fibrotic ulcers or those located on the posterior duodenal bulb.

Over-the-scope clips (OTSC or Padlock clips) are large-jaw full-thickness clips that have demonstrated superiority to standard therapy for Forrest IIa and Ib lesions in the setting of refractory or recurrent bleeding, with the ability to capture the vessel and surrounding tissue. Hemostatic powders such as TC-325 (Hemospray) are applied as a spray and serve as a temporizing measure for diffuse oozing or technically challenging lesions; however, they carry a high rebleeding rate of 30 to 50% when used as monotherapy and are best utilized as bridge therapy. The Doppler ultrasound probe provides real-time assessment of submucosal arterial flow beneath the ulcer base and can confirm adequate hemostasis, as persistence of an arterial signal predicts rebleeding.

Combination Therapy

The standard of care for endoscopic hemostasis of peptic ulcer bleeding is combination therapy, consisting of epinephrine injection plus either thermal coagulation or mechanical clip placement. Dual therapy is superior to injection monotherapy, achieving a relative risk reduction for rebleeding of 35 to 45%.

<image>An endoscopic hemostasis technique illustration showing four panels demonstrating the major modalities for peptic ulcer bleeding. Panel 1 "Epinephrine Injection": cross-sectional view of ulcer base with injection needle deploying 1:10,000 epinephrine in four quadrants around a visible vessel, with labeled arrows showing submucosal wheal formation and tamponade effect. Panel 2 "Bipolar Electrocoagulation": Gold probe tip applying coaptive coagulation to visible vessel with firm pressure, showing thermal injury zone in cross-section (coagulated vessel walls fused together). Panel 3 "Hemoclip Application": through-the-scope clip being deployed directly across visible vessel with arrows showing mechanical compression, and a deployed clip shown en face. Panel 4 "Over-the-Scope Clip (OTSC)": large bear-claw type clip grasping full-thickness tissue including ulcer base and surrounding mucosa, with cross-sectional inset showing depth of tissue capture. Use anatomical layers (mucosa, submucosa with vessel, muscularis propria) in cross-sections with appropriate coloring. Label all key structures and mechanisms of hemostasis.</image>

Post-Endoscopy Management

High-Dose PPI Therapy

Following successful endoscopic therapy for high-risk ulcers classified as Forrest Ia through IIb, high-dose intravenous PPI therapy is administered to maintain gastric pH above 6 and stabilize the fibrin clot. The Sung protocol uses a continuous infusion of esomeprazole or pantoprazole at 8 mg per hour for 72 hours. Recent meta-analyses suggest that intermittent high-dose intravenous PPI therapy (pantoprazole 40 mg every 8 or 12 hours) may be non-inferior to continuous infusion. After the initial 72-hour period, therapy is transitioned to oral PPI twice daily and subsequently reduced to daily dosing for 4 to 8 weeks. For low-risk stigmata (Forrest IIc and III), standard-dose oral PPI is sufficient, and early discharge, even on the same day, is clinically appropriate.

Diet Advancement

Clear liquids may be introduced within 24 hours of achieving hemostasis for most patients, with advancement to a regular diet as tolerated within 48 to 72 hours. There is no evidence supporting prolonged nothing-by-mouth status after successful endoscopic hemostasis.

Second-Look Endoscopy

Routine second-look endoscopy is not recommended, as meta-analyses have not demonstrated a mortality benefit. However, it may be considered in patients with suboptimal initial hemostasis, very high-risk lesions, or clinical concern for rebleeding.

Refractory and Recurrent Bleeding

Management of Rebleeding

When rebleeding occurs after initial endoscopic therapy, repeat endoscopy with hemostasis is successful in 70 to 75% of episodes. Over-the-scope clips should be considered for rebleeding after failure of initial standard hemostasis techniques. After two failed endoscopic attempts, management should escalate to interventional radiology or surgery.

Interventional Radiology

Transcatheter arterial embolization is indicated for hemodynamically unstable patients who have failed endoscopic therapy or who are at high surgical risk. Initial hemostasis is achieved in 70 to 90% of cases, though rebleeding occurs in 10 to 30%. Complications include bowel ischemia in 5 to 10% and access site complications. When the patient is sufficiently stable, CT angiography should be obtained before embolization to localize the bleeding source and guide the intervention.

Surgery

Surgical intervention is indicated for massive hemorrhage that is uncontrolled by endoscopy or interventional radiology, for perforation, or for failure of all other modalities. For duodenal ulcers, surgery typically involves oversewing of the bleeding vessel with ligation of the gastroduodenal artery, with or without pyloroplasty and vagotomy. For gastric ulcers, wedge resection or partial gastrectomy is performed, with the specimen sent for pathologic examination.

Special Scenarios

Dieulafoy Lesion

A Dieulafoy lesion is a submucosal caliber-persistent artery measuring 1 to 3 mm that erodes through the overlying mucosa without an associated ulcer. Approximately 75% of these lesions are located in the proximal stomach, typically on the lesser curvature within 6 cm of the gastroesophageal junction. The clinical hallmark is intermittent, massive, recurrent bleeding. Endoscopic therapy includes mechanical hemostasis with clips (preferred), epinephrine injection, thermal coagulation, or band ligation. Multiple endoscopic examinations may be required to identify the lesion, as bleeding is characteristically intermittent.

Mallory-Weiss Tear

A Mallory-Weiss tear is a mucosal laceration at the gastroesophageal junction resulting from forceful retching or vomiting. Ninety percent of these tears stop bleeding spontaneously. When active bleeding is present, endoscopic therapy with clips, injection, or band ligation is effective. Angiographic or surgical intervention is rarely required.

Portal Hypertensive Gastropathy

Portal hypertensive gastropathy, covered in detail in the cirrhosis and portal hypertension lecture, produces diffuse oozing that is not responsive to PPI therapy. Treatment centers on non-selective beta-blockers, with transjugular intrahepatic portosystemic shunt (TIPS) reserved for refractory cases.

Cameron Lesions

Cameron lesions are linear erosions on the gastric folds at the diaphragmatic hiatus in patients with large hiatal hernias. They typically cause chronic occult bleeding and iron deficiency anemia rather than acute hemorrhage. Management includes PPI therapy and iron supplementation, with surgical hernia repair considered for refractory cases.

<image>A comprehensive illustration showing the anatomical locations of major UGIB sources in the upper GI tract. Display an opened view of the esophagus, stomach, and duodenum. Label and illustrate: (1) Esophageal varices at the distal esophagus with blue-purple variceal columns, (2) Mallory-Weiss tear at the GEJ as a longitudinal mucosal laceration, (3) Cameron lesion as linear erosions on gastric folds at the diaphragmatic hiatus, (4) Gastric ulcer on the lesser curvature with visible vessel, (5) Dieulafoy lesion on the proximal lesser curvature shown as a small arterial protrusion without surrounding ulcer (include cross-sectional inset showing the caliber-persistent artery), (6) Portal hypertensive gastropathy in the fundus/body shown as a snakeskin mosaic pattern, (7) Duodenal ulcer on the posterior wall with the gastroduodenal artery labeled behind it, (8) Angiodysplasia as red vascular ectasia in the gastric antrum (GAVE/watermelon stomach pattern). Include percentage contribution to UGIB next to each lesion. Use realistic anatomical coloring with clear leader lines and labels.</image>

Key Clinical Pearls

• Restrictive transfusion (Hb <7 g/dL) is superior to liberal (Hb <9) in non-variceal UGIB — Villanueva trial showed 45% mortality reduction
• Glasgow-Blatchford Score 0-1 identifies very low-risk patients safe for outpatient management
• IV erythromycin 250 mg given 30-90 minutes pre-EGD significantly improves visualization and reduces need for repeat endoscopy
• Epinephrine injection should NEVER be used as monotherapy — always combine with thermal or mechanical hemostasis
• Post-endoscopic hemostasis: high-dose IV PPI x 72 hours for Forrest Ia-IIb; intermittent dosing likely non-inferior to continuous infusion
• Continue low-dose aspirin for secondary cardiovascular prevention during UGIB management — Sung trial showed 10-fold higher cardiovascular mortality if held
• Over-the-scope clips are superior to standard hemostasis for recurrent ulcer bleeding — consider early in difficult anatomy
• Forrest IIc and III ulcers do NOT require endoscopic therapy — medical management with PPI alone

References

1. Laine L, et al. ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding. Am J Gastroenterol. 2021;116(5):899-917.
2. Villanueva C, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med. 2013;368(1):11-21.
3. Sung JJY, et al. Continuation of low-dose aspirin therapy in peptic ulcer bleeding. Ann Intern Med. 2010;152(1):1-9.
4. Barkun AN, et al. International Consensus Upper Gastrointestinal Bleeding Conference Group recommendations 2019 update. Ann Intern Med. 2019;171(11):805-822.
5. Jensen DM, et al. Doppler endoscopic probe as a guide to risk stratification and definitive hemostasis of peptic ulcer bleeding. Gastrointest Endosc. 2016;83(1):129-136.

Read this lecture as Markdown