Upper Gastrointestinal Bleeding (Upper GI haemorrhage)

Warning

Objectives

To provide guidance in the recognition and management of bleeding from the upper gastrointestinal tract.

Scope

This guidance covers the management of suspected upper gastrointestinal bleeding for a deployed medical facility including a brief overview of the required definitive management.

Audience

This guideline is intended for the use of registered healthcare professionals fulfilling a general role in a forward medical location or in an Emergency Department on Deployed Operations 

 

Background

Upper GI bleeding is a common and potentially life threatening medical presentation. This often presents with haematemesis, melaena or in massive upper GI bleeding can present with fresh blood PR. It can also present more insidiously with presentations such as abdominal pain, anaemia or syncope without overt haemorrhage.

There are a range of causes: peptic ulcer disease, oesophagitis/gastritis, Mallory Weiss tear, foreign body/chemical ingestion and variceal bleeding secondary to liver disease.

Each have similar initial assessment and management, however variceal bleeding is often high volume and treatments are aimed at reducing portal venous pressure as well as resuscitation. These patients will often have other sequalae of liver disease including coagulopathy and hepatic encephalopathy.

Severity of upper GI haemorrhage can be difficult to assess at initial presentation, these patients should be evacuated to deployed hospital care without delay (if possible) as the requirement for specialist intervention and assessment is high.

Initial Assessment & Management

Undertake a systematic approach.

History and examination - History taking should focus on the presence or absence of melaena, composition of vomitus for fresh blood versus coffee ground vomit. A PR exam may be required to confirm the presence of melaena. Assess for signs of liver disease to assess for possibility of variceal bleeding.

Investigations - A full set of observations should be taken and IV access obtained. If point of care testing is available then measure haemoglobin levels as a marker of severity, and to guide transfusion if blood products are available: Hb >70 g/dl is an appropriate target if transfusion is needed.

Medications - Start a proton pump inhibitor (e.g. omeprazole 20mg bd). Oral preparations are equally effective so there is no need to administer intravenously unless the patient is continuing to vomit; anti-emetics should be given if they are, but note that the available evidence does not presently support the use of TXA in gastrointestinal haemorrhage.

Fluid resuscitation - tolerate hypotension: resuscitate to a systolic blood pressure of 100 mmHg (or a palpable radial pulse) as part of a strategy of permissive hypotension. If fluid resuscitation is required then blood products should be used (if available) if the bleeding appears to be large volume or if the patient appears shocked, otherwise use crystalloid fluid.

Antibiotics - if variceal bleeding is suspected then start prophylactic antibiotics, as per current antimicrobial guidance.

 

Most patients who present with GI bleeding require evacuation for further assessment including bloods, noting that risk stratification using the Glasgow-Blatchford Bleeding Score is impossible without serum urea measurement.

The only exception is patients who present with clear features of a Mallory-Weiss tear, and who meet the criteria detailed below are suitable for monitoring in location and if they remain well for 24 hours, may be discharged.

Mallory-Weiss tears (MWT) are longitudinal, non-perforating mucosal lacerations at the gastroesophageal junction caused by a sudden, intense increase in intra-abdominal pressure. They are a common cause of upper gastrointestinal bleeding, often triggered by severe vomiting, retching, or coughing.  More than 80% heal spontaneously within 10 days.

Low risk patients:

  • Clear history of forceful vomiting, coughing or straining before blood observed in vomitus
  • No severe bleeding (Hb within normal range if measurable)
  • No ongoing bleeding
  • Normal vital signs including normal blood pressure and heart rate
  • No history of syncope
  • No history or evidence of melaena
  • No history or features of liver disease, renal failure or cardiac disease

Treat with PPI and monitor for 24 hours. If the patient remains well, discharge with follow up medical review after a further 24 hours. If bleeding recurs at any point then the patient requires assessment in a deployed hospital facility.

Advanced Assessment & Management

Assess and if necessary resuscitate the patient as described above.

Check bloods, in particular:

Full blood count - monitor Hb with serial measurement, but also check platelet count and look for microcytosis (low MCV) which may be indicative of iron deficiency anaemia due to longstanding GI blood loss.

Biochemistry should include urea, electrolytes and liver function tests. Elevated urea is an indicator of possible blood in the gut lumen, especially if creatinine is normal.

Check coagulation

Cross-match for transfusion even if the patient is stable, in case of unexpected deterioration.

Patients who do not present with severe bleeding and who are not experiencing ongoing bleeding may be considered for discharge if Glasgow-Blatchford score is 0 (see below). Otherwise further investigation including specialist review and endoscopy is required.  This will normally require evacuation from theatre are endoscopy capabilities are not normally available in the deployed environment. However, consider the possibility of accessing capabilities offered by host/partner nations if prompt evacuation is likely to be complex or difficult.  

If equipment and expertise are available, ultrasound imaging may aid the diagnosis of underlying liver pathology.

Patients with ongoing haemorrhage, haemodynamic instability or experiencing confusion/agitation from hepatic encephalopathy are likely to require critical care support both in theatre and for evacuation.

NICE guidelines recommend terlipressin 2 mg 4 hourly until bleeding is controlled for suspected variceal bleeding. This drug is not routinely carried in deployed modules but may be available from other sources such as host/partner nations. Alternatively noradrenaline can be used in place of terlipressin for suspected variceal bleeding in order to reduce portal pressure - consult separate guidance in the CGO on use of vasopressors.

Prolonged Casualty Care

If evacuation is delayed, continue proton pump inhibitor therapy twice daily.

Paediatric Considerations

Management of the paediatric patient mirrors that of the adult. Omeprazole dosing should be 2mg/kg/day, ideally in split (bd) doses.

Glasgow-Blatchford Score

In simple terms, patients with a Glasgow-Blatchford score of 0 may be considered for discharge.

This means that:

  • Serum urea <6.5 mmol/l
  • Hb >130 g/l if the patient is male or >120 g/l if the patient is female
  • Initial systolic blood pressure was >110 mmHg
  • Heart rate <100
  • No melaena
  • No recent syncope
  • No hepatic disease
  • No heart failure

If any of the list above do not apply then the patient will score at least 1 point and will therefore not be suitable for discharge without at least specialist discussion via reach back, and will likely require specialist review and endoscopy.

The full scoring system is below:

Urea (mmol/l) 0 points if < 6.5: 0 points
2 points if 6.5 – 7.9: 2 points
3 points if 8.0 – 9.9: 3 points
4 points if 10.0 – 25.0: 4 points
6 points if > 25.0: 6 points 
Haemoglobin (g/l)

For men:
0 points if >13
1 point if 12-13
3 points if 10-12
6 points if <10

For women:
0 points if >12
1 point if 10-12
6 points if <10

Initial systolic blood pressure (mmHg)

0 points if >110
1 point if 100-109
2 points if 90-99
3 points if <90

Pulse >100 beats per minute 1 point
Melena present 1 point
Presentation with syncope 2 points
Liver disease history 2 points
Cardiac failure present 2 points

 

Last reviewed: 13/10/2026

Next review date: 13/10/2027

References

Blatchford, Oliver et al. A risk score to predict need for treatment for upper gastrointestinal haemorrhage. (2000) The Lancet, Volume 356, Issue 9238, 1318 - 1321