Case 11: Managing Haematemesis – Expert comment by Dr Graham

Thanks to everyone who contributed to the discussions about Mrs Tollings a 43 year old secretary who had vomited blood.

drip stand

You can see the highlights of the discussion on Twitter in the storify (add link).

This week’s question setter was Dr Beth Good, ST4 Acute Medicine & Clinical Teaching Fellow, UCL Medical School. Our expert was Dr David GrahamST7 Gastroenterology & PhD Research Fellow at UCL (Interventional Science Department)

Information available at the start of the week:

Mrs Tollings is a legal secretary who is normally in good health and on no prescribed medications. She lives alone, is a non-smoker and drinks approximately 5 bottles of white wine per week.

On admission Mrs Tollings has active haematemesis and is reviewed by the Emergency Department Registrar. ABCDE assessment is as follows:

A – clear, talking

B – RR 28, Sats 99% on air, clear chest

C – HR 124 (regular), BP 110/68. Cool, clammy hands and feet.

D – Alert. Capillary blood glucose 7.8.

E – Soft, mildly tender abdomen. Black tarry stool (melaena) noted on rectal examination.

Initial blood tests:
Hb 9.2 g/dL
WCC 12.0 x109/L
Plt 120 x109/L
Na 146 mmol/L
K haemolysed (3.4 mmol/L on a venous blood gas sample)
Urea 19 mmol/L
Creatininer 120mmol/L
LFTs haemolysed
Clotting sample bottle under filled.

Question 1:

  1. Is there anymore information you would like to know from Mrs Tolling’s history and examination that would help to narrow your diagnosis and formulate a management plan?

Further information released during the week:

Further history:

–       No vomiting prior to the haematemesis

–       Has some longer term symptoms of epigastric burning and thinks she may have had black stools a few times

GI Bleeding Risk factors

–       Takes Ibuprofen almost every day for tension headaches

–       Has drunk this amount of alcohol for one year but has no stigmata of chronic liver disease on examination

–       Has not had peptic ulcer disease in the past/never knowingly been infected with H.Pylori.

–       She has no known bleeding disorders and reports no blood loss from other sites

After initial medical management to ensure she was stable Mrs Tolling’s had an OGD (picture shown below).

Endoscopy finding

Further questions discussed on Twitter:

Question 2: Which clinical findings/results in this case help to clarify how severe the bleeding is & are there any risk assessment scores?

Question 3: What is your fluid management plan? Which ones and how much?

Question 4: What other management,specific to UGI bleeds,would you initiate prior to endoscopy?

Question 5: What endoscopic markers indicate risk of re-bleeding? What treatment should be prescribed post endoscopy?

Expert Comment

Key points:

1. The key initial management steps: including using the right fluids and monitoring response.

2. The importance of differentiating between suspected variceal and non-variceal bleeds and their different management.

3. Knowing about scoring systems to help decide the urgency of endoscopy

4. Appropriately managing these patients afterwards.


Assessment of patients with GI bleeding:


Dr David Graham , ST7 Gastroenterology & PhD Research Fellow at UCL (Interventional Science Department)says:

The advances in the management of upper gastrointestinal (UGI) bleeds, in particular in the field of endoscopy and the eradication of H. pylori, have certainly improved morbidity and mortality. However, despite this it remains a potentially life threatening condition with an associated mortality of 10%.

1 – Key initial management steps

In the interests of ease I will approach this case in stepwise manner, however, it is important to be aware that one’s approach to a patient with a suspected UGI bleed will differ depending on the severity. For example, it is obviously inappropriate to try to obtain a thorough history from a patient in severe hypovolaemic shock. In reality, in this instance, I would take a very targeted history whilst actively resuscitating the patient at the same time.


It is important to attempt to ascertain the possible cause for this event. In particular whether a patient is a suspected of having chronic liver disease as the management of these patients is very different. Questions in relation to chronic liver disease (alcohol abuse, history of hepatitis B and C, family history etc.) should be asked. However, the most common cause of an UGI bleed is peptic ulcer disease. Therefore questions relating to dyspepsia and importantly NSAID use must be asked. Finally, it is vital to be aware of anti-coagulant medication (aspirin, clopidogrel, warfarin etc.) that could be contributing to this event. Often these medications will have to be stopped and reversed (in the case of warfarin) to aid in the cessation of the bleed. This can have an impact on a patient’s co-existent medical conditions and therefore this decision is led by a gastroenterologist who will consult relevant other medical specialties if appropriate. More often than not one accepts that the risks of the UGI bleed outweigh the risks of stopping these medications in the short term.

In this case we do not have too much of a history. We are made aware she drinks 5 bottles of wine a week. The average bottle of wine is 9 units; therefore she is drinking approximately 45 units a week (over three times the recommended amount for women). However, remember that only a small percentage of people who abuse alcohol develop chronic liver disease and therefore this history alone is not sufficient for me to treat this patient as suspected of having chronic liver disease. It would raise my index of suspicion and therefore I would examine her closely for signs of liver disease and interrogate her liver function blood tests.

An alcohol history is often taken poorly in medical clerking’s. It is not uncommon to see terms such as “occasional” or “social” in the notes. This is not acceptable as it is open to our own personal interpretation. Also, one must be aware of the alcohol content of the drink the patient consumes. For example, if asked how much alcohol I drank, I may reply that I just drink 10 cans of beer a week. However, this is not enough. If I were drinking 10 cans of standard Fosters lager (4%) a week my weekly alcohol intake would be under the acceptable 21 units a week for a man. Change that to 10 cans of Tennents Super (9%) and my weekly alcohol intake becomes 45 units per week.


Putting aside advanced life support for this section (although it should always take priority) one should be aware of and look for signs of chronic liver disease. The gastroenterology team in any hospital at any given time will always have at least one inpatient with signs of chronic liver disease under their care. Ask your friendly gastroenterology registrar to take you to see these patients. They have good clinical signs that should not be missed. Picking up on these signs, in the context of a patient admitted with an acute UGI bleed will help to save their life.

Obviously it is important to examine all the systems of a patient but the only other aspect I will focus on in this section is the rectal examination. All patients presenting with a suspected UGI bleed must have a PR examination. Often patients present with “coffee ground” or dark vomit that commonly is not blood. Melaena is a sign that cannot be confused with anything else (although I accept iron supplementation causes dark stool). Its presence confirms an UGI bleed has occurred and shows that a significant amount of blood has been lost.

Advanced life support:

It is vital that you recognise a suspected UGI bleed early and initiate appropriate management. A patient cannot undergo an endoscopy until they have been made stable. Early escalation of care with the involvement of ITU / critical care outreach is often required. I will run through the ABC’s and interventions needed specific to this case:

A (airway) – She is talking and is giving a clear history. One presumes form this her GCS is 15. Be aware that this is not always the case. Airways can be compromised due to vomiting or due to a low GCS as a result of shock or encephalopathy (liver disease).

Intervention required in this case: None.

B (breathing) – Her respiratory rate is elevated. Many things could cause this but in this case I would suggest it is a sign that she is unwell and that this event is serious. Be aware that patient’s can sometimes develop an aspiration pneumonia (particularly if their GCS is low) that carries with it a significant morbidity and mortality.

Intervention required in this case: None.

C (circulation) – She is tachycardic, the first sign of shock. She is cold and clammy suggesting that she is peripherally shut down, again another sign of shock. Do not be lulled into a false sense of security with her normal blood pressure. Patients, particularly young patients, are often able to conceal significant blood loss by maintaining a normal blood pressure. A good way to demonstrate significant blood loss is to check for postural changes in blood pressure. It may not be appropriate to get this patient to stand up (given the other signs of shock) but one can always perform a lying / sitting blood pressure simply by raising and lowering the head of the bed. I would wager in this case we would see a significant postural drop.

Intervention required in this case:

Access – All patients with a suspected UGI bleed must have two large bore cannulae inserted. Anything else is not safe and therefore not acceptable. Gaining intravenous access can sometimes be tricky in these patients (they are peripherally shut down and may have bad veins due to previous IV drug use) if this is the case call for senior help immediately.

Fluid resuscitation – What to use and how much?

All patients with a suspected UGI bleed should be at least “Group and Saved” with many needing an urgent cross-match of blood and blood products. Again, it is not acceptable not to do this.

This patient requires fluid resuscitation. We now understand that a restrictive blood transfusion policy has been shown to improve outcomes in regards to UGI bleeds. Over transfusing patients can cause problems such as prolonging clotting which worsens mortality. A target haemoglobin of above 8g/dl is acceptable and if we do transfuse we should aim not to exceed 10g/dl. However, one must also use common sense. If we have a blood test result with a haemoglobin above 8g/dl but since the blood test was taken the patient has continued to have haematemesis and is haemodynamically unstable then a transfusion is likely to be necessary.

In this case, the scenario suggests she continues to have active bleeding. Given that she is haemodynamically unstable and having active bleeding I would certainly ask for blood to be cross-matched (4-6 units) but whilst I was waiting for this I would give her 500mls to 1000mls of colloid (stat) and assess her response.

If she did stop actively bleeding, whilst you were waiting for the blood, it may be that her haemodynamic status would stabilise with the colloid alone. In this case I may consider re checking her haemoglobin of 9.2g/dl before I transfused (to avoid over transfusion). However, if she continued to vomit blood and remained haemodynamically unstable I would transfuse her as that haemoglobin of 9.2g/dl is likely to be old news.

The fluid we prescribe to a patient (for any condition) is not a “one size fits all” scenario. It is individual dependent on how they respond and the clinical picture. Above is how I would start, in this case, and I would adapt according to how she responds. Being too generous with fluid resuscitation can cause as many problems as not being generous enough. If you are not sure also involve a senior Doctor who has had more experience.


It is important we order the right tests and know why we’re ordering them. A FBC, U&E’s, LFT’s, clotting profile (including INR) and a cross match / group and save should be taken for all patients. Specific to this case:

FBC – The haemoglobin has been covered above. Of note her platelets are low. This can be caused by a variety of factors (including her alcohol abuse) but does not require a platelet transfusion unless the count falls below 50.

U&E’s – Of note here her urea has risen out of proportion to her creatinine. This is caused by a combination of pre-renal failure and blood being broken down by the gut’s digestive enzymes to amino acids releasing urea. A raised urea suggests an UGI bleed has occurred.

LFT’s – Whilst these are helpful to determine whether this patient has a liver injury do not forget that it is the liver function tests of INR and albumin that tell us whether she may have chronic liver disease (synthetic liver function).

Clotting screen – The sample is under filled. You could’ve sworn it wasn’t. The lab has it in for you……….It happens. It’s annoying. Get over it and immediately send another. Take it to the lab yourself. Call ahead, be polite but emphasise the importance of this test. Not only will the INR suggest whether this patient has chronic liver disease but also any clotting abnormality must be corrected.

2 – Differentiating between suspected variceal and non-variceal bleeds

Any patient with suspected chronic liver disease should be treated as a suspected variceal bleed. As a general rule of thumb this includes: patients who say they have chronic liver disease, patients who say they have varices, patients with clinical signs of chronic liver disease and patients with blood tests suggestive of chronic liver disease (prolonged INR / PT and / or a low albumin). The reason this differentiation is important is because these patients are managed differently and will require an endoscopy as soon as possible. If you are not sure whether you should treat your patient as a suspected variceal bleed or not speak to a gastroenterologist or a senior immediately.

In this case although the patient has a high alcohol intake there are no other signs of chronic liver disease (I accept we do not have her albumin or INR / PT back yet). I would treat her as a suspected non-variceal bleed.

Non-variceal bleed:

The decision when to contact a gastroenterologist and when to perform an endoscopy is discussed below in regards to the scoring systems. For these patients, until the endoscopy is performed, the mainstay of treatment is resuscitation outlined above. The latest NICE guidelines do not advise prescribing a PPI (oral or infusion) prior to endoscopy. After NICE had analysed all the studies they found that whilst there is evidence for PPI’s given before endoscopy reducing the need for endoscopic intervention and improving early healing there was not evidence that they improved morbidity and mortality. Therefore, they felt that one can not justify the cost of PPI’s pre-endoscopy if we are not improving these outcomes. This is a subject still debated by gastroenterologists but I won’t bore you with that here.

At endoscopy we use various methods to stop bleeding including clips, injection of adrenaline and a heater probe (to cauterise vessels). You should try to watch a UGI bleed being treated in endoscopy to fully understand what happens.

Variceal bleed:

A gastroenterologist should be contacted early on in the resuscitation process for all patients with a suspected variceal bleed regardless of how severe you think it is. These patients can deteriorate rapidly. Often patients will firstly need to be stabilised with consideration of correcting a coagulopathy and blood transfusion. Specific management is outlined below:

Terlipressin, 2mg qds, should be given prior to endoscopy after initiating fluid resuscitation. This can be effective in controlling variceal bleeding by causing vasoconstriction, however it should not be used to delay specific endoscopic therapy. Be aware that it does not cause vasoconstriction specific to the portal system and it is therefore contraindicated in patients with ischaemic heart disease or peripheral vascular disease. The decision to prescribe or omit terlipressin in these cases should be discussed with a gastroenterologist.

All patients should receive prophylactic broad-spectrum antibiotic cover. This has been shown to reduce mortality in variceal bleeds and the rate of early re-bleeding. Which antibiotic to prescribe will be specific to the policy of the trust the hospital is based in. Normally Tazocin 4.5g tds is used (provided the patient is not penicillin allergic).

Vitamin K (10mg) and pabrinex should also be prescribed, as these patients are likely to be deficient with consideration and treatment for potential alcohol withdrawal if present.

At endoscopy, variceal band ligation is the treatment of choice for oesophageal varices with glue injection used for gastric varices. A Sengstaken Blakemore tube should always be at hand should it be required in the event of uncontrollable bleeding. Nowadays this seems to be only put in by gastroenterologists but it is potentially life saving and something everyone should know how to use. Some patients require Transjugular intrahepatic portosystemic shunts (TIPS) to control bleeding in the event that it cannot be stopped endoscopically.

3 – Scoring systems and how to decide the urgency of an endoscopy

There are two main scoring systems used in UGI bleeds: The Rockall Score and The Blatchford Score. Like all scoring systems in medicine they are a guide and not an absolute. It is possible for a patient with a severe, life threatening bleed to have a low score and the opposite can also be true. Common sense and clinical acumen is invaluable.

The Blatchford Score:

This is generally accepted as the score to use for patients prior to endoscopy as it can be used to estimate the likelihood that a patient with an UGI bleed will require intervention. As a general rule of thumb one should interpret the scores as follows:

Score 0 – Patients can be discharged from hospital. If it is felt appropriate an outpatient gastroscopy or clinic appointment can be offered.

Score <6Patients are unlikely to require endoscopic intervention. Cases should be discussed with a gastroenterologist. It is unlikely these patients will require an endoscopy out of hours and therefore discussion can wait until normal working hours where appropriate.

Score >6These patients may require endoscopic intervention.  Severe bleeds (higher Blatchford scores) or patients with on-going bleeding may require an endoscopy immediately after resuscitation. These cases should be discussed with a gastroenterologist.

In our scenario the patient’s pre endoscopy Blatchford score is 12. This is a serious UGI bleed and she is likely to require endoscopic intervention. Contact the gastroenterologist early on in your management.

The Rockall Score

This tends to be used following the endoscopy as it is a score to estimate the outcome following a UGI bleed. One can use the score to calculate your patient’s risk of mortality from the first bleed and their risk of mortality from a re-bleed. For arguments sake, lets say our patient had a duodenal ulcer that was bleeding and was successfully injected and clipped. Her Rockall score would be 4. That gives her a 5-10% mortality for the first bleed and 15-25% mortality should she re-bleed.

4 – Appropriately managing these patients afterwards

The management of these patients does not stop after the endoscopy and I hope the above scores highlight the potential severity of these events. These patients should always be under the care of a gastroenterologist and be managed in an appropriate area of the hospital (often HDU or ITU).  Their resuscitation should be on going afterwards and they should be monitored closely for any signs of re-bleeding.

In the case of a peptic ulcer bleed it has been proven that a 72 hour PPI infusion reduces morbidity and mortality and this should be initiated. The intricacies of managing a patient following a variceal bleed are not for this forum and are always led by a gastroenterologist. To learn more about this speak to a gastroenterologist on the ward.

One response to “Case 11: Managing Haematemesis – Expert comment by Dr Graham

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s