Upper Gastro Intestinal Bleeding at St.Emlyn’s.


Earlier this week, Simon and Iain released a podcast on the management of acute upper gastrointestinal bleeding. Welcome to the blog post version, for all your textual needs.

Around 60,000 people are admitted into hospital every year with an upper GI bleed. With emergency department attendances in the UK totalling just over 22 million in the last year, this equates to roughly one person in every 300 seen. For those of you who work in bigger departments, on average you’ll refer one patient in every day to be investigated or treated. And this doesn’t take into account the patients you discharge.

It’s an important presentation to be aware of, and with an overall mortality of around 10% (which has remained static for 50 years despite advances in medical management), early recognition and treatment can really make a big difference 1.

NICE guidance for the management of acute upper gastrointestinal bleeding has recently been reviewed and updated 2, with only one difference in the new advice compared with that released in 2012 3. This relates to H2-antagonist and proton-pump inhibitor use for ulcer/GI bleeding prophylaxis in the critically ill. Whilst NICE recommends either can be used, they note that only ranitidine and cimetidine are licensed for this purpose. The use of PPI therapy would be off-licence, and whilst this will not affect the emergency physicians amongst us, it’s something for our critical care colleagues to be aware of.

Causes of Upper GI Bleeding

Peptic ulcer disease accounts for the majority of bleeds – around 40-50% of cases. This could be in the form of either a gastric ulcer or duodenal ulcer. In peptic ulcer disease, haemorrhage results from erosion into a blood vessel at the ulcer base . In patients with liver cirrhosis, increased blood flow through the portal venous system causes submucosal vessels in the oesophagus and stomach to become dilated and tortuous. These blood vessels, essentially the haemorrhoids of the upper GI tract, are called varices and have a strong tendency to bleed. Variceal bleeding is another important cause to be aware of as severe bleeds can be acutely life-threatening if not controlled quickly. Mortality for variceal bleeding in the short and long term is high and although it was as high as 50% in the past, modern techniques have reduced this to a 6-week mortality of 20% as compared to 50% in the 1980s 4. Other causes for bleeding include gastritis or duodenitis and erosions, oesophagitis, Mallory-Weiss (a mucosal tear in the oesophagus caused by repeated forceful vomiting or retching), malignancy, or more rare conditions such as a vascular malformation or aortoenteric fistula 5 (see table). Endoscopy is generally required in order to establish a diagnosis. In approximately 10-20% of cases, however, no cause is found.


Clinical Assessment

Patients can present in a variety of ways. Of course, the most obvious clue to an upper gastrointestinal bleed is if they are vomiting blood in front of you, however this is not always the case, and sometimes all you might get is a vague history of dark vomit – altered blood or coffee-ground vomit, or an episode of dark stools. Patients who are bleeding may just come to the ED with a history of abdominal pain, dizziness or fainting, and in rare circumstances you may just have an unwell patient who is tachycardic and hypotensive. Always keep gastrointestinal bleeding in mind when dealing with patients who appear to be shocked but with no obvious cause.

When taking a history, ask about previous bleeding or ulcer disease, liver disease, alcohol history (though just because there is an alcohol history, don’t automatically jump to variceal bleeding – peptic ulcer disease is still more common in these patients!), and recent profuse vomiting (as this may suggest a Mallory-Weiss tear. It is also important to ask about drug history, and any recent increased or new use of tablets which can promote gastric bleeding, such as ibuprofen, aspirin or warfarin. It’s also important at this point to exclude things that can look like an upper GI bleed, such as your patient with a nosebleed who has tipped their head back instead of forwards, swallowed a load of blood and then vomited it back out later, or patients who actually have haemoptysis. Particularly in the latter case, sometimes language barriers can make it difficult to determine whether the blood was vomited or coughed, but it’s an important distinction if it can be made.

On examining your patient, look for signs of decompensated liver disease (encephalopathy, ascites, jaundice), as this makes a variceal bleed more likely, and will impact on how the patient is treated. It’s also important to perform a digital rectal examination early to look for melaena.

Once you’ve done all this, and you’re satisfied that you’ve diagnosed an upper GI bleed, it’s time to risk stratify and to determine whether this patient needs admission, or if they can go home

Resuscitate, resuscitate, resuscitate!

Of course, if your patient is very unwell and you suspect an upper GI bleed – they’re trying to vomit their entire blood volume up in front of you – you’re going to want to resuscitate them well first. Good resuscitation is essential to stabilise your patient before they can progress to an endoscopy (and also to keep them alive)2,64. These overtly bleeding patients are a bit easier to manage from an ED perspective. They need airway management, oxygen, large-bore IV access, bloods taken, and fluid resuscitation with crystalloid or blood products, depending on their clinical picture. A venous blood gas can help to quickly establish the patient’s metabolic state, as well as give you a near-immediate haemoglobin level, which can guide use of blood products. Once resuscitated, your patient will need endoscopy, and to be admitted to a critical care environment.

It’s important to get senior help early on, and even if you are senior, these are patients who can deteriorate rapidly so grab a friend7. They need a multi-disciplinary approach, and you’ll need some extra hands to help you out. In particular, task someone to start phoning key people that you might need – an anaesthetist if the airway is in trouble, the on-call endoscopist, blood bank, haematology (if your patient has clotting disorders or is on medication that can affect this) and critical care. Having someone else do this keeps you free to manage the patient rather than being tied to the phone.7


Blood transfusion decisions should be based on the full clinical picture. It should be noted that over-transfusion has complications, and that studies have shown a reduction in mortality when transfusion is restricted to patients with a haemoglobin less than 7g/dL (9g/dL in unstable coronary artery disease).8

Current NICE guidelines state that platelet transfusion should only be offered to patients with a platelet count less than 50 x 109/litre who are actively bleeding. Fresh frozen plasma should be used in actively bleeding patients with a prothrombin time (PT), INR, or activated partial thromboplastin time (aPTT) greater than 1.5 times the normal upper limit. Prothrombin complex concentrate (PCC) can be used in patients taking warfarin who are actively bleeding, but if bleeding as stopped then follow local warfarin protocols.2

If your department has thromboelastography in the form of TEG or ROTEM, this can be used, if readily available, to guide the transfusion of clotting products. This is not yet evidence based as a recommendation though anecdotally we have found it useful in a variety of major bleeding patients 9.

Risk assessment

The trickier patients to manage are those who are more in the grey area of having some symptoms of a GI bleed, but on examination look very well. Do they need their endoscopy as an inpatient or outpatient?

The Glasgow-Blatchford score is the risk assessment tool of choice in the emergency department 1011. It is calculated using a few components – history of heart failure or liver disease, symptoms of syncope, presence of melaena, tachycardia, hypotension, and results of blood tests for haemoglobin and urea. It’s important to mention here that the melaena portion of the score is based on presence of melaena on examination, rather than patient reporting of altered stools, so once again – make sure a rectal examination is performed. Use an online calculator such as MDCalc to work out the score – it’s tricky to calculate and better left to a computer, just make sure you’ve got it set to the right units!

Patients can be divided broadly into three categories:

  • Low risk – patients with a Glasgow-Blatchford score of 0 who are well with no other reason for admission can be discharged with endoscopy as an outpatient.
  • Moderate risk – patients with a score of 1 or more should be admitted to an appropriate specialty for endoscopy – in some hospitals there might be a designated unit for these patients.
  • High risk – patients who are haemodynamically unstable, requiring active resuscitation, or who have a known variceal bleed should be considered for critical care admission.

Of course, even if you have a patient with a score of 0, if they are unwell and you are concerned about sending them home, go and speak to someone about them, and don’t just discharge them just because of a score. The scoring system is no substitute for the patient’s overall clinical picture.


In non-variceal bleeding, treatment is pretty much limited to that already described above.12 There is no role for proton pump inhibitor therapy in the ED13, and NICE guidelines do not recommend their use pre-endoscopy, as they have not been shown to have any benefit on mortality, re-bleeding rate or need for surgery.2 The only medication which may benefit patients is tranexamic acid. At present, there is no evidence for antifibrinolytic therapy in upper GI bleeding. The HALT-IT trial has been running since 2013 and is investigating whether tranexamic acid reduces mortality in acute upper GI bleeding.14 So far, over 5,000 patients have been randomised, and the study is planned to finish in 2017, so it shouldn’t be too long before we have a definite answer.

For patients with suspected variceal bleeds, vasopressins and antibiotics have good evidence for their use. Terlipressin has been shown to reduce variceal bleeding, and reduce mortality by a third, so give this early on.15 20% of cirrhotic patients with variceal bleeding will develop a bacterial infection within 48 hours so all patients with suspected bleeding varices should be given prophylactic antibiotic therapy, such as ciprofloxacin or ceftriaxone – both of which have good evidence behind them.16

Two holes, two tubes

Of course, no self-respecting blog post on GI bleeding would be complete without a section on using Sengstaken-Blakemore or Minnesota tubes in the management of patients with variceal bleeding. It’s kind of a last ditch attempt to use a balloon catheter to tamponade bleeding and should be undertaken in patients with a suspected or confirmed variceal haemorrhage that has continued to bleed despite medical therapy when endoscopy is not immediately available. Tamponade provides good control of bleeding in 90%, although most will re-bleed within 24 hours. The main role of these tubes is to buy time to endoscopy, and the patient will need a tube in the airway hole first in order to secure it, so call your friendly anaesthetist down to assist as this is not an easy anaesthetic, make sure you are well prepared for complications and a crashing patient on induction 76. They have some quite serious complications such as oesophageal necrosis and perforation, so be sure you have practised using it in simulation so that in an emergency you’re familiar with the device.

Scott Weingart over at EMCrit has a fantastic page with some videos demonstrating its use, so check those out here 17..

Final thoughts

So, there you go.

Resuscitate, risk stratify, and refer for endoscopy.

And don’t forget to listen to the podcast that this blog accompanies.




Before you go…



Barkun AN. International Consensus Recommendations on the Management of Patients With Nonvariceal Upper Gastrointestinal Bleeding. Annals of Internal Medicine. 2010;152(2):101. doi: 10.7326/0003-4819-152-2-201001190-00009 [Source]
National Institute for Clinical Excellence N. Acute upper gastrointestinal bleeding in over 16s: management. CG141. NICE. https://www.nice.org.uk/Guidance/cg141. Published August 2016.
Dworzynski K, Pollit V, Kelsey A, Higgins B, Palmer K. Management of acute upper gastrointestinal bleeding: summary of NICE guidance. BMJ. 2012;344(jun13 2):e3412-e3412. doi: 10.1136/bmj.e3412 [Source]
Tripathi D, Stanley AJ, Hayes PC, et al. UK guidelines on the management of variceal haemorrhage in cirrhotic patients. Gut. 2015;64(11):1680-1704. doi: 10.1136/gutjnl-2015-309262 [Source]
Upper GI Bleeding. lifeinthefastlane.com. http://lifeinthefastlane.com/ebm-upper-gi-haemorrhage/. Published September 2016. Accessed September 17, 2016.
Intubation in Gastrointestinal haemorrhage. lifeinthefastlane.com. http://lifeinthefastlane.com/ccc/intubation-in-upper-gastrointestinal-haemorrhage/. Published September 2016. Accessed September 17, 2016.
Intubating the critical GI bleeder. emcrit.org. http://emcrit.org/podcasts/intubating-gi-bleeds/. Published June 2009. Accessed September 17, 2016.
Villanueva C, Colomo A, Bosch A, et al. Transfusion Strategies for Acute Upper Gastrointestinal Bleeding. New England Journal of Medicine. 2013;368(1):11-21. doi: 10.1056/nejmoa1211801 [Source]
Thromboelastography aka The TEG. tamingthesru.com. http://www.tamingthesru.com/blog/grand-rounds/teg. Published August 2015. Accessed September 17, 2016.
Glasgow scores, not just for Coma. stemlynsblog.org. http://stemlynsblog.org/glasgow-scores-not-just-for-coma-any-more/. Published July 2012. Accessed September 17, 2016.
Blatchford O, Murray W, Blatchford M. A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet. 2000;356(9238):1318-1321. [PubMed]
Laine L, Jensen DM. Management of Patients With Ulcer Bleeding. Am J Gastroenterol. 2012;107(3):345-360. doi: 10.1038/ajg.2011.480 [Source]
Sreedharan A, Martin J, Leontiadis GI, et al. Proton pump inhibitor treatment initiated prior to endoscopic diagnosis in upper gastrointestinal bleeding. Sreedharan A, ed. Cochrane Database of Systematic Reviews. July 2010. doi: 10.1002/14651858.cd005415.pub3 [Source]
Roberts I, Coats T, Edwards P, et al. HALT-IT – tranexamic acid for the treatment of gastrointestinal bleeding: study protocol for a randomised controlled trial. Trials. 2014;15(1). doi: 10.1186/1745-6215-15-450 [Source]
Ioannou G, Doust J, Rockey D. Terlipressin for acute esophageal variceal hemorrhage. Cochrane Database Syst Rev. 2003;(1):CD002147. [PubMed]
Lee YY. Role of prophylactic antibiotics in cirrhotic patients with variceal bleeding. World Journal of Gastroenterology. 2014;20(7):1790. doi: 10.3748/wjg.v20.i7.1790 [Source]
Blakemore Tube Placement for Massive Upper GI Hemorrhage. emcrit.org. http://emcrit.org/procedures/blakemore-tube-placement/. Published October 2013. Accessed September 17, 2016.


  1. Pingback: Global Intensive Care | Upper Gastro Intestinal Bleeding at St.Emlyn’s.

  2. jducanto

    Great summary of the topic! I plan to build the Minnesotta and/or Sengstaken-Blakemore into the SALAD Simulation. Having helped place one last year, I recall it was a very frustrating affair, even with McGrath MAC laryngoscopy and MaGill Forceps.

    1. Simon Carley

      Thanks James, that means lot from you. I’s love to see the SALAD simulator used for this as clearly a simulation is the best way to train.

      Thanks for the comment.


  3. Jo tillett

    Replacing blood loss with crystalloid always seems perverse. If someone is haemdynamically unstable from massive GI bleed I would reach for O neg blood not crystalloid. The Hb which looked OK on arrival is suddenly 50 after 2 litres of fluid. Am I backed by evidence though or just common sense? Also is there any role for permissive hypotension as in trauma or leaking AAA until definitive management?

    1. Chris Gray (Post author)

      Hi Jo,

      Thank you very much for your comments. If our patient is actively bleeding in front of us and unstable then it is sensible to reach for the O neg, and start the crossmatch process for further products as described in the NICE guidelines, according to our patient’s needs, as if the bleeding hasn’t stopped then we need to keep filling the bath until such time as we can put the plug back in. However, our patient could be haemodynamically unstable having had a large GI bleed, but with no evidence of ongoing bleeding (i.e. haematemesis), in which case a crystalloid bolus would be appropriate together with a venous gas for a quick haemoglobin assessment. The patient should be reassessed after every bolus given to see how they respond – clinically and with repeat gases if we have the ability. Those who don’t respond should be considered for blood products (or inotropic support!). The evidence is there that patients with a haemoglobin above 7 don’t benefit from blood, and may fall foul of the risks that come with giving it.

      As far as I’m aware there’s no specific evidence for permissive hypotension in this population, but like trauma, the first clot is the best clot and so we need to be careful not to over-resuscitate.

      Hopefully this helps! Decisions on what to resuscitate with are always based on the patient in front of you, but there is good evidence that blood isn’t always best!


      1. Jo

        Thanks Chris

  4. Pingback: Upper GI Bleed | The Resus Room

Thanks so much for following. Viva la #FOAMed

Translate »