Residency · Residency · Emergency Medicine
Upper GI Bleeding: Risk Stratification and Resuscitation
Introduction
Upper gastrointestinal bleeding (UGIB), defined as hemorrhage originating proximal to the ligament of Treitz, accounts for approximately 300,000 hospitalizations annually in the United States. Mortality ranges from 2 to 10% and has remained stable despite therapeutic advances, largely due to the aging population and increasing comorbidity burden. The emergency physician's role centers on hemodynamic resuscitation, risk stratification, appropriate blood product utilization, and timely gastroenterology consultation.
Etiology
Common Causes
Peptic ulcer disease is the most common cause, accounting for 30 to 50% of all UGIB, and involves both gastric and duodenal ulcers associated with H. pylori and NSAID use. Esophageal and gastric varices account for 10 to 20% and result from portal hypertension due to cirrhosis; they carry the highest mortality at 15 to 25%. Mallory-Weiss tears represent 5 to 10% and consist of longitudinal mucosal lacerations at the gastroesophageal junction from forceful vomiting or retching. Erosive esophagitis and gastritis produce diffuse mucosal injury from acid, alcohol, or stress. Dieulafoy lesions are aberrant submucosal arterioles that erode through the mucosa and cause intermittent massive bleeding. Malignancies of the stomach or esophagus typically cause chronic oozing but can present with acute hemorrhage. Aortoenteric fistulae are rare but catastrophic and should be considered in patients with prior aortic graft surgery, as a "herald bleed" may precede massive hemorrhage.
Medications Associated with UGIB
NSAIDs, including aspirin, carry a synergistic risk with H. pylori. Anticoagulants such as warfarin and DOACs, along with antiplatelet agents, increase bleeding risk. Corticosteroids are particularly dangerous when combined with NSAIDs. SSRIs impair platelet aggregation and represent an underappreciated risk factor.
Clinical Presentation
Hematemesis presents as bright red blood or coffee-ground emesis from partially digested blood. Melena, characterized by black, tarry, foul-smelling stools, indicates at least 50 to 100 mL of blood in the GI tract. Hematochezia, or bright red blood per rectum, usually indicates lower GI bleeding but can occur with brisk UGIB involving more than one liter of blood loss. Hemodynamic compromise manifests as tachycardia, hypotension, orthostatic changes, and altered mental status. Syncope may be the presenting complaint in significant UGIB.
Initial Assessment and Resuscitation
Airway Considerations
Massive hematemesis poses a significant aspiration risk, and intubation should be considered for airway protection in actively vomiting patients with altered mental status. Left lateral decubitus positioning may reduce aspiration risk during active hematemesis if intubation is not immediately performed.
Hemodynamic Resuscitation
Two large-bore IVs of 18-gauge or larger should be placed, avoiding central lines as the sole access because their smaller lumens produce slower flow rates. Permissive hypotension targeting a MAP of 65 mmHg or systolic blood pressure of 90 mmHg is recommended for variceal bleeding to reduce portal pressure and re-bleeding risk. Crystalloid resuscitation provides initial volume expansion while blood products are being prepared. The massive transfusion protocol should be activated for hemodynamically unstable patients not responding to an initial crystalloid bolus.
Blood Product Transfusion
A restrictive transfusion strategy, transfusing packed red blood cells for hemoglobin below 7 g/dL, has demonstrated superior outcomes compared to liberal strategies using a threshold of 9 g/dL in the landmark Villanueva trial. Exceptions include active exsanguination, hemodynamic instability, and acute coronary syndrome, which warrant higher transfusion thresholds. The initial hemoglobin may be falsely normal due to hemoconcentration, making serial monitoring essential. Platelet transfusion is indicated if the platelet count falls below 50,000 with active bleeding. Fresh frozen plasma should be given for an INR greater than 1.5 with active bleeding, guided by coagulation studies.
<image>Resuscitation algorithm for upper GI bleeding showing initial assessment with two large-bore IVs, simultaneous lab draw (CBC, BMP, coagulation, type and crossmatch), decision tree for hemodynamic stability versus instability, transfusion thresholds, and indications for massive transfusion protocol activation</image>
Risk Stratification
Glasgow-Blatchford Score (GBS)
The Glasgow-Blatchford Score is a pre-endoscopic scoring system that identifies patients who can be safely discharged for outpatient endoscopy. Its components include BUN, hemoglobin, systolic blood pressure, heart rate, melena, syncope, hepatic disease, and heart failure. A GBS of 0 to 1 indicates very low risk, and these patients can be considered for outpatient management with close follow-up. A GBS of 7 or higher indicates high risk and requires admission with early endoscopy.
Rockall Score
The Rockall Score is a post-endoscopic scoring system that predicts re-bleeding and mortality. It incorporates age, shock status, comorbidities, endoscopic diagnosis, and stigmata of hemorrhage, and is useful for disposition planning after endoscopy.
AIMS65 Score
The AIMS65 score predicts inpatient mortality using five variables: albumin less than 3.0, INR greater than 1.5, altered mental status, systolic blood pressure less than 90, and age greater than 65. A score of 2 or higher identifies high-risk patients, and mortality increases progressively with each additional point.
| Score | Purpose | Key Variables | Low Risk | High Risk |
|---|---|---|---|---|
| Glasgow-Blatchford (GBS) | Pre-endoscopic; identifies safe discharge | BUN, Hgb, SBP, HR, melena, syncope, liver/heart disease | GBS 0–1: outpatient | GBS ≥ 7: admit, early endoscopy |
| Rockall | Post-endoscopic; predicts re-bleeding/mortality | Age, shock, comorbidities, diagnosis, stigmata | Low score: safe discharge | High score: ICU, intervention |
| AIMS65 | Inpatient mortality prediction | Albumin < 3, INR > 1.5, AMS, SBP < 90, Age > 65 | Score 0–1 | Score ≥ 2: high mortality |
Pharmacologic Management
Proton Pump Inhibitors (PPI)
IV PPI consisting of an 80 mg pantoprazole bolus followed by an 8 mg per hour infusion was previously standard, but recent evidence suggests that intermittent IV or oral PPI dosing such as pantoprazole 40 mg IV every 12 hours is noninferior for outcomes. PPIs raise gastric pH to stabilize clots over ulcers, with the greatest benefit occurring after endoscopic therapy. Pre-endoscopic PPI may downstage lesions by reducing active bleeding to non-bleeding stigmata but has not consistently reduced mortality, re-bleeding, or the need for surgery.
Variceal-Specific Medications
Octreotide is given as a 50 mcg IV bolus followed by a 50 mcg per hour infusion for up to 5 days, reducing splanchnic blood flow and portal pressure. It should be started empirically whenever variceal bleeding is suspected based on cirrhosis or known portal hypertension, without waiting for endoscopic confirmation. Antibiotics are mandatory: ceftriaxone 1 g IV daily for 7 days or norfloxacin reduces infection and mortality in cirrhotic patients with UGIB.
Prokinetics
Erythromycin 250 mg IV given 30 to 90 minutes before endoscopy improves gastric visualization by promoting gastric emptying and reduces the need for second-look endoscopy. Metoclopramide is an alternative but is less well-studied for this indication.
<image>Comparison chart of risk stratification tools for UGIB showing Glasgow-Blatchford Score (pre-endoscopic, identifies low-risk patients), Rockall Score (post-endoscopic, predicts re-bleeding and mortality), and AIMS65 (predicts inpatient mortality) with their respective components and clinical utility</image>
Endoscopy
Early endoscopy within 24 hours is recommended for most patients with UGIB. Urgent endoscopy within 12 hours is indicated for patients who are hemodynamically unstable despite resuscitation, those with suspected variceal bleeding, and those with high-risk clinical features. Endoscopic therapies include epinephrine injection (always combined with a second modality), thermal coagulation, hemoclips, and band ligation for varices. The Forrest classification guides endoscopic management of peptic ulcers based on stigmata of hemorrhage.
Special Considerations
Anticoagulant Management
For warfarin, reversal with vitamin K 10 mg IV and 4-factor PCC is indicated for life-threatening bleeding, and endoscopy should not be delayed for full INR correction. For DOACs, reversal agents such as idarucizumab for dabigatran and andexanet alfa for factor Xa inhibitors should be considered in severe hemorrhage. Antiplatelet agents are generally not reversed, as platelet transfusion has not shown benefit and may worsen outcomes in some studies.
Balloon Tamponade
The Sengstaken-Blakemore or Minnesota tube serves as a temporizing measure for uncontrolled variceal hemorrhage when endoscopy fails or is unavailable. Complications include esophageal rupture, aspiration, and mucosal necrosis, and intubation is mandatory before placement. Maximum inflation time is 24 hours, and the device serves as a bridge to TIPS or repeat endoscopy.
<image>Anatomical illustration showing common causes of upper GI bleeding with labeled sites: esophageal varices, Mallory-Weiss tear at GEJ, gastric ulcer, duodenal ulcer, erosive gastritis, and Dieulafoy lesion with brief descriptions of each</image>
Clinical Pearls
A normal initial hemoglobin does not exclude significant hemorrhage because it takes hours for hemodilution to reflect true blood loss, and resuscitation should be guided by the clinical picture. The Glasgow-Blatchford Score of 0 to 1 reliably identifies very low-risk patients who may be safely managed as outpatients. In suspected variceal bleeding, both octreotide and ceftriaxone should be started empirically before endoscopy, as antibiotics reduce mortality in cirrhotic UGIB. Restrictive transfusion using a hemoglobin threshold of less than 7 g/dL improves survival compared to liberal transfusion, except in active exsanguination or acute coronary syndrome. Aspirin should be avoided in the bleeding patient, but anticoagulation should not be reflexively reversed in stable patients without discussing the bleeding versus thrombotic risk with gastroenterology.
References
- Villanueva C, et al. "Transfusion Strategies for Acute Upper Gastrointestinal Bleeding." New England Journal of Medicine. 2013;368(1):11-21.
- Laine L, et al. "ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding." American Journal of Gastroenterology. 2021;116(5):899-917.
- Garcia-Tsao G, et al. "Portal Hypertensive Bleeding in Cirrhosis: Risk Stratification, Diagnosis, and Management." Hepatology. 2017;65(1):310-335.
- Stanley AJ, et al. "Comparison of Risk Scoring Systems for Patients Presenting with Upper Gastrointestinal Bleeding." BMJ. 2017;356:i6432.
