Emergency Care Institute –

Upper gastrointestinal bleeding

Published: May 2024. Printed on 25 Apr 2025.

On this page

The incidence of gastrointestinal haemorrhage ranges from 50 to 150 per 100, 000 people each year. The reported mortality rates are between 11% and 33% for patients admitted primarily due to gastrointestinal (GI) haemorrhage or who develop it as a complication of their hospital stay.

Rapid evaluation of the patient for evidence of haemodynamic compromise and identifying risk factors for serious haemorrhage is crucial. This enables early resuscitation and timely involvement of interventional services when indicated.

General resources

Assessment

Haemodynamic instability

  • Assess for clinical evidence of hypovolemic shock (↑heart rate, ↓pulse pressure, ↓blood pressure (beware compensation) ↑respiratory rate, ↑capillary refill time, cold clammy skin, ↓urine output, ↑agitation or confusion).
  • Mild, moderate or severe hypovolaemia is indicated by resting tachycardia, orthostatic hypotension or supine hypotension respectively, and equates to approximately 15%, 30% and over 40% loss of total circulating blood volume (5L in 70kg male).

Relevance of history

  • Alcohol abuse, previous GI bleed, liver disease or other causes of abnormal coagulopathy that may require correction or replacement.
  • Presence of significant or unstable coronary artery disease or renal disease which may place a patient at risk of volume overload during resuscitation attempts.
  • Melaena on patient history then likelihood ratio of an upper GI Bleed = 5.1-5.9. Examination of stool may give a clue for location of bleeding but is not a reliable indicator. If melaena present on PR, the likelihood ratio of an upper gastrointestinal (UGI) bleed is 25. Melaena may be seen with as little as 50mls of blood loss. Red stool or an altered blood in stool has 11% chance of being UGI in origin. When abdominal pain is present, consideration should be given to the possibility of perforation and the need for early surgical intervention.

Medications

  • Anticoagulants, antiplatelet, non-steroidal anti-inflammatory drugs, aspirin, steroids, iron supplements should all be considered during assessment and may require specific treatment or product replacement.

Laboratory studies

Acute bleeding may not be reflected in the full blood count due to ~24hrs delay in fluid equilibration. Microcytic changes will only be present if the bleeding or iron deficiency has been chronic. Remember, excessive crystalloid resuscitation may result in the false lowering of the haemoglobin. Blood urea nitrogen/creatinine ratio also positively correlated with UGI bleeding with ratios >30 having a likelihood ratio of 7.5.

Causes

Table 1: Causes of UGI bleeding

Cause % of patients

Peptic ulcer disease

35-50%

Gastroduodenal erosions

8-15%

Oesophagitis

5-15%

Varices

5-10%

Mallory-Weiss tear

15%

Vascular malformations

5%

Other or rare

6%

Although gastro-oesophageal varices are less common, managing the underlying liver disease and the severity of bleeding can demand significant resources.

Risk scoring systems

It is increasingly recognised that early risk assessment is an important part of management, which helps direct appropriate patient care and the timing of endoscopy. Several risk scores have been developed, most of which include endoscopic findings, although a minority do not.

The Glasgow Blatchford Score (GBS)

  • The most relevant score for the purposes of emergency management as it can be calculated prior to endoscopy.
  • It was developed in 2000 to predict the need for hospital -based intervention (transfusion, endoscopic therapy, or surgery) or death following UGI haemorrhage.
  • It predicts the need for inpatient care for the purposes of endoscopy and transfusion.
  • Six recent studies from the United Kingdom and Taiwan have shown GBS to be superior to the admission Rockall score in predicting need for clinical intervention or death. The GBS has also been shown to be superior to both the full and admission Rockall scores in predicting need for transfusion.

Patients with a GBS score of 0 may be suitable for discharge.

Higher risk groups require inpatient endoscopy for full evaluation and therapy.

Calculate the Glasgow Blatchford Score

Management

General management

  • Nasopharyngeal oxygen, monitoring as appropriate based on the patient’s comorbidities and risk of bleeding.
  • Two large bore intravenous cannulas. Arterial line consideration for high- risk patients.
  • Nil by mouth in view of potential for endoscopy or other interventional procedure.
  • Resus bed if indicated.
  • Notifying theatres and specialty teams in advance of the likely need for operative management for high -risk patients.

Fluid resuscitation

  • Adequate but judicious fluid resuscitation to treat symptomatic hypotension with isotonic crystalloid in 500ml aliquots while waiting for appropriately cross-matched blood.
  • Beware of overaggressive crystalloid administration and the potential for hyperchloraemic acidosis if too much “normal” saline is given.
  • Aim for trauma style avoidance of over resuscitation with crystalloids.

Blood products

The initial resuscitation section recommends haemostatic blood product resuscitation for unstable patients in line with massive transfusion practice in trauma. The recommended ratio of blood products - packed cells: fresh frozen plasma (FFP): platelets - is 1:1:1. However, where the patient is sufficiently stable to wait for pathology results, current guidelines and evidence suggest restrictive transfusion strategies, especially in patients with known or suspected variceal disease. Transfuse only for:

  • haemodynamic instability despite crystalloid resuscitation
  • haemoglobin.

Also give fresh frozen plasma for INR >1.5; give platelets for thrombocytopenia (aim platelets >50 x 109/L).

In the bleeding patient always remember that “cold doesn’t clot” and it is much easier to prevent hypothermia than reverse it. To maintain normothermia:

  • avoid exposure
  • use a blood warmer for all products
  • consider a Bair Hugger (forced air warming blanket) early.

Avoid acidosis and ensure adequate ionised calcium levels are maintained, that is, above 1.13mmol/L by giving calcium gluconate.

Other medical treatments

  • Consider proton pump inhibitor (PPI) as a loading dose and maintenance infusion. PPI reduces the need for intervention during endoscopy if given beforehand. It also reduces the incidence of re-bleeding but no evidence of change in mortality.
  • Consider thiamine if there is a question of alcohol abuse.

Variceal bleeding interventions

Treatment specific to suspected variceal bleeding

Each episode of active variceal haemorrhage is associated with 30% mortality . Variceal bleeding stops spontaneously in over 50% of patients, but the mortality rate rises to 70 –80% in those with continued bleeding.

Specific to variceal haemorrhage resuscitation is the need to avoid over-transfusion in the initial management of these patients as well as ensuring rapid consideration and correction of any coagulopathies.

Medical therapy aims to reduce the portal venous pressures and so reduce bleeding.

Early vasoactive therapy

Early vasoactive therapy should be commenced as promptly as possible on presentation of the patient who has known or suspected varices and should not be held pending confirmation of the diagnosis. Vasoactive medications have been shown to significantly decrease mortality and improve haemostasis in patients with acute variceal bleeding. There is also some evidence to indicate their efficacy in non-variceal bleeding, this includes:

Endoscopy

Endoscopy is first-line in variceal disease for diagnosis and intervention with potential banding or injection and decreases the risk of rebleeding to approximately 30%.

Transjugular intrahepatic portosystemic shunt (TIPS)

Surgical intervention has little role in the management of varices and patients who do not respond to endoscopic therapies are best treated by TIPS.

Specifically, if the location of the varices is known:

  • Oesophageal varices:
    • Acute bleeding is typically managed with endoscopic variceal ligation within 12 hours (occasionally endoscopic sclerotherapy is used).
    • If the bleeding cannot be controlled endoscopically, TIPS placement or surgical shunting should be considered.
  • Gastric varices:
    • Treatment is with tissue adhesives (cyanoacrylate injection) where available.
    • If cyanoacrylate injection is not an option, TIPS placement is typically used.
  • Bleeding ectopic varices may be managed with TIPS placement or surgery.

See Intubating the unstable GI bleeder

Balloon tamponade

After intubation, balloon tamponade is an option of last resort to temporarily stop bleeding from oesophageal or gastric varices while definitive treatment is being arranged. However, it is associated with serious complications including oesophageal rupture. So, always remember to confirm placement of the gastric balloon before you fully inflate it. Confirmation of placement is by listening over the stomach and lungs injecting air into the stomach port, and then inflating 50 mls of air or saline and doing X-ray to confirm in the stomach.

There are three types of tubes used for balloon tamponade. The Linton, Blakemore, & Minnesota Tubes Overview video by EMRAP provides an excellent summary.

  • Sengstaken-Blakemore tube three ports (two for filling the balloons with air and one for gastric suction)
    • 250cc gastric balloon, an oesophageal balloon, and a single gastric suction port
  • Minnesota tube four ports (just a modified Sengstaken-Blakemore tube)
    • Also has an oesophageal suction port above the oesophageal balloon
  • Linton-Nachlas tube one port
    • Which has a single, but larger 600cc, gastric balloon

Remember, balloon tamponade should only be considered as an interim and dire measure pending endoscopy or where endoscopy has failed, and the patient is awaiting TIPS. More information on how to attempt balloon tamponade can be found on the The Blakemore Tube page of EM Curious

Beta blocker

Long term, the administration of a nonselective beta blocker such as propranolol can also decrease the risk of re-bleeding in variceal bleeders.

Non-variceal bleeding interventions

Endoscopy is considered the gold standard for diagnosis and intervention. Endoscopy is recommended within 24 hours of presentation for the diagnosis and treatment of active UGI bleeding and for the prevention of recurrent bleeding rather than waiting more than 24 hours.

Where endoscopy is not available, some evidence suggests that the use of somatostatin and its long-acting analogue octreotide may be of benefit even in non-variceal UGI bleeding.

For patients whose endoscopy has failed or is contraindicated, the options include the following.

Interventional angiography

  • Should be considered before surgery, local resources and expertise permitting.

Surgery

  • Emergency surgery is undertaken for uncontrolled bleeding or re-bleeding that cannot be controlled by further endoscopic therapy and/or where angiography expertise is not available.
  • Surgery is also considered in cases of perforation, haemodynamic shock associated with recurrent haemorrhage or ongoing hemodynamic instability despite vigorous resuscitation (more than a three unit transfusion).
  • Emergency surgery has a high (36%) mortality rate while early elective surgery has a much lower mortality rate of 0-7%.

With appropriate treatment, high-risk lesions have recurrent bleeding rates of 5- 20%, depending on the endoscopic appearance of the ulcer base. The majority of patients with upper gastrointestinal bleeding due to peptic ulcer disease will stop bleeding spontaneously. Most will not re-bleed during hospitalisation.

Remember to consider whether there are any modifiable predisposing risk factors for ulcer formation, e.g. stopping non-steroidal anti-inflammatory drugs, smoking, helicobacter pylori infection and treated as appropriate.

Intubating the unstable GI bleeder

Massive GI haemorrhage presents challenges during and after intubation

  • Vocal cords are likely to be obscured by blood and/or vomit.
  • Haemodynamic instability from haemorrhagic shock.
  • Aspiration risk high.
  • Exposure risk to staff from contact with body fluids.

Steps to maximise success

  1. Use your airway checklist, consider and PLAN for the worst and prep your team.
  2. All staff involved to wear personal protective equipment, especially goggles, gloves and mask.
  3. Empty the stomach (if tolerated insert a nasogastric tube and put it on low wall suction – varices are not a contraindication).
  4. Administer prokinetic, e.g. Metoclopramide or erythromycin.
  5. Intubate with patient with head of bed at 45°. But if the patient vomits reposition to Trendelenberg to get the vomit out of the lungs.
  6. Have an assistant allocated to the role of suctioning on the left. Have a spare Yanker sucker setup in case the first gets blocked.
  7. Preoxygenate well with good seal (if possible) but try to avoid non-invasive ventilation if vomiting.
  8. Apnoeic oxygenation (using continuous nasal prong oxygenation) is a must in these patients.
  9. BVM gently and slowly to re-oxygenate, if needed, (6-10 breaths per minute).
  10. Use video laryngoscopy if available but have a standard laryngoscope blade ready in case of blood and/or fluids obscuring the camera.
  11. Must paralyse to optimise your first pass success. Where suxamethonium contraindicated, paralyse with rocuronium to minimise time to onset of action. Paralysis does not reduce the lower oesophageal tone.
  12. Consider using a hemodynamically stable sedative like ketamine.
  13. No role initially for antibiotics in “chemical pneumonitis” from aspiration, but may require antibiotics (ceftriaxone) if variceal bleed. Systemic inflammatory response syndrome may require further fluid resuscitation +/- vasopressors.

Recent evidence suggests that all variceal bleeds should get antibiotics.

Further resources

Accessed from the Emergency Care Institute website at https://aci.health.nsw.gov.au/networks/eci/clinical/tools/upper-gastrointestinal-bleeding

Back to top