Pitfalls in the management of acute upper gastrointestinal bleeding

Introduction

An audit undertaken by the National Blood Service and British Society of Gastroenterology recruited 6,500 patients with acute upper gastrointestinal bleeding over a two-month period in 2007. ¹ The mortality of these patients was approximately 10%, ranging from 7% for those admitted because of gastrointestinal bleeding to 25% in patients who were already in hospital for other reasons and subsequently developed acute gastrointestinal bleeding. The audit also showed that the commonest cause of bleeding was peptic ulcer (35% of cases) and that the incidence of varices due to alcoholic liver disease had more than doubled over ten years. Other causes of bleeding were self-limiting and had a good prognosis.

Acute upper gastrointestinal bleeding therefore remains a common medical emergency with a high mortality. Despite advances in medical and surgical treatments, patients continue to die from its complications and most deaths occur in elderly patients who have other medical co-morbidities.

In 2008 the National Patient Safety Agency became aware of anecdotal cases of unexpected deaths following hospital admission for acute gastrointestinal bleeding and as a consequence they and the British Society of Gastroenterology instigated a Working Party that recommended a series of minimum standards that every patient should receive. ² These comprise:

There should be a nominated clinical lead responsible for acute upper gastrointestinal bleeding;

Current management

When a patient presents with haematemesis and/or melaena they are first clinically assessed in order to gain insight into possible causes of bleeding and to define the severity of the bleed. The clinical team seeks evidence of liver disease since variceal bleeding tends to be severe and can precipitate other complications of liver failure. In all patients sudden blood loss causes tachycardia and hypotension that can lead to renal failure, particularly in patients with other medical co-morbidities. Several risk assessment scores have been developed both for non-variceal and for variceal bleeding, and good clinical practice dictates that each patient should be risk scored since this guides triage and is useful for audit purposes. The Rockall scoring system ⁴ is based upon age, pulse, systolic blood pressure, medical co-morbidities and endoscopic findings. The Rockall score accurately predicts mortality and, less reliably, the risk of re-bleeding. A major problem is that it is necessary to undertake an endoscopy to calculate a score. The Blatchford scoring system ⁵ is calculated at the time of presentation and is based upon the presence of shock, anaemia, medical co-morbidity, blood urea and reliably predicts the need to intervene, either by endoscopy or surgery.

Following clinical assessment and calculation of a risk score the clinical team should decide whether urgent intervention either by endoscopy or surgery is required or whether more elective investigation within 24 hours of the bleed is reasonable.

There has been considerable interest in the use of several drugs given before endoscopy. For patients with known liver disease and probable variceal bleeding, the vasopressin analogue Terlipressin lowers portal pressure, reduces the risk of continuing variceal haemorrhage and is administered before endoscopy. Terlipressin does not however by itself reduce mortality and is best regarded as complementary to endoscopic therapy. For all patients there is no evidence to support the routine use of proton pump inhibitor drugs given prior to endoscopy, since they do not influence mortality or re-bleeding, although they do improve the endoscopic findings. ²

The key investigation is endoscopy and the timing of this is dependent upon clinical factors. Unstable, actively bleeding patients with a high-risk score and the majority of patients with probable variceal bleeding should undergo urgent endoscopy. This is done once resuscitation has been optimized. Endoscopy is undertaken in the hospital endoscopy suite or, if this is unavailable, an operating theatre. The purpose of endoscopy is to establish a diagnosis, provide prognostic information and to apply a range of therapies that can stop active bleeding and prevent re-bleeding. For varices prognostic information includes the size of the varices, their site (oesophageal or gastric) and whether or not they are actively bleeding. For peptic ulcer bleeding the prognostic factors comprise active bleeding, a non-bleeding visible vessel (associated with a 50% of further bleeding) or a blood clot that is adherent to the ulcer bed. Endoscopic therapies comprise the application of rubber bands to oesophageal varices and the injection of tissue glues to gastric varices. The use of sclerosant substances (such as Ethanolamine or Pollidochanol) is discouraged since this is less effective than banding for oesophageal varices and completely ineffective in gastric varices. For peptic ulcers with major stigmata (active bleeding, visible non-bleeding vessel or adherent blood clot) therapy is based upon injection of Adrenaline or Thrombin around the bleeding point followed by either application of heat (using the heater probe or electrocoagulation) or an endo-clip.

Following endoscopic therapy a range of drugs is used. Terlipressin is continued for 24 hours as an adjunct to banding, for ulcers high-dose proton pump inhibitor therapy is prescribed for a 72-hour period.

Endoscopic therapies have transformed the management of acute upper gastrointestinal bleeding. This was reflected in the audit that showed that the need for surgical intervention has diminished from approximately 20% of patients fifteen years ago to 1% of cases in 2007. Nevertheless, some patients continue to bleed or re-bleed despite endoscopic therapy and these patients have a high risk of death. Rescue therapies for ulcer bleeding comprise an emergency surgical operation (usually under running or over running the ulcer) or arterial embolization. Transjugular intrahepatic portosystemic stenting (TIPS) is used when patients continue to bleed from varices despite endoscopic therapies.

Critical areas

These particular points within the diagnostic and treatment pathways were errors most likely to occur. These apply to clinical teams and to management structures.

Clinical teams

(i) Resuscitation. It is crucial that patients are properly resuscitated when they first develop acute gastrointestinal bleeding. Failure to do this may lead to renal failure and other medical disasters. It is important that medical co-morbidity is both recognized and supported and it is sometimes wise to obtain the advice of an appropriate specialist. A good example is patients presenting with acute upper gastrointestinal bleeding whilst taking anti-platelet agents such as Clopidogrel and Aspirin following coronary stent insertion. These drugs exacerbate bleeding and gastroenterologists often discontinue their use as part of their approach to controlling bleeding. The reality is that the biological half life of these drugs is approximately ten days; the effectiveness of stopping them is therefore limited and discontinuation in coronary stented patients increases the risk of cardiac events and death due to stent thrombosis. Thus, a discussion between gastroenterologists and cardiologists (or stroke specialists) is essential in any one patient to establish the risk benefit ratio of stopping Aspirin and Clopidogrel in bleeding patients.

(ii) Blood transfusion is clearly life-saving in patients with massive gastrointestinal bleeding but the benefits of blood transfusion in patients with less severe bleeding is uncertain. The bleeding audit actually showed that transfused patients tended to re-bleed more often than comparable groups of case matched patients who did not receive a blood transfusion and it is wise to be judicious about blood transfusion. Whilst guidelines have suggested a threshold haemoglobin of 10g/L for the administration of blood products, this may be too high and it is possible that a more restricted use of blood transfusion might lead to better outcome.

(iii) Drugs. As already stated, the value of proton pump inhibitor drugs is well established in patients with major endoscopic stigmata but there is no evidence for their routine use in all patients prior to endoscopy. It is occasionally said that proton pump inhibitors are better than nothing and that if endoscopy is delayed perhaps they have some value. There is no evidence to support this statement and clinicians are unwise if they believe that these drugs can safely be used instead of endoscopy or allow endoscopy to be safely delayed. Terlipressin reduces portal pressure, decreases the risk of continuing bleeding in patients with varices and supports renal function in patients with advanced liver disease. On the other hand, vasopressin analogues are well known to have complications including spasm of other arteries. Patients with significant coronary artery disease may develop cardiac events and those with peripheral vascular disease can develop limb ischaemia as a consequence of injudicious use of Terlipressin.

(iv) Endoscopy. Eighty percent of patients stop bleeding spontaneously and have a good outcome. The remaining 20% comprise the high-risk group of patients with non-variceal and variceal bleeding and need urgent endoscopic therapy. The degree of urgency is informed by the risk assessment score and clinical common sense. Thus, a patient with active haematemesis or melaena who is hypotensive despite intravenous fluids clearly is at considerable risk of exsanguination and complications of hypotension, and should be treated urgently. On the other end of the severity spectrum is a group of younger patients who lack co-morbidity, are haemodynamically stable and have stopped bleeding. These can be endoscoped electively; some indeed can be managed as outpatients. Most patients lie between these extremes and are endoscoped on the next elective list within 24 hours. Patients who require very urgent endoscopy because they are clinically unstable should first be resuscitated and when this is optimized will require endoscopy in a safe environment. In some hospitals facilities are not available for out-of-hours endoscopy and patients will have to be transferred to a referral centre where endoscopy can be undertaken. Policies for safe transfer must therefore be developed and applied on a case by case basis. Occasionally the patient is so ill that they cannot be transferred; a good example is a patient who develops acute gastrointestinal bleeding whilst in a High Dependency or Intensive Care Unit that is not served by 24/7 endoscopy. In these situations it may be better for the endoscopy team to travel to that HDU or ITU from the referral centre.

Therapeutic endoscopy is both difficult and potentially risky. Only endoscopists experienced in the management of acute gastrointestinal bleeding and able to administer endoscopic therapy for varices and ulcer bleeding should undertake the procedure. The major risk is aspiration pneumonia and for this reason it is sensible to involve anaesthetic support including endotracheal intubation for the most severely bleeding patients. This is particularly relevant to patients with variceal bleeding in whom bleeding is frequently severe and the conscious state is depressed from hepatic encephalopathy. Many patients with severe bleeding are elderly and have other medical co-morbidities. They can be very sensitive to the effects of sedative drugs and for all these reasons anaesthetic assessment and support is wise.

Teams need to be aware that endoscopic therapy for peptic ulcers can be complicated by perforation and exacerbation of bleeding. Whilst these are relatively unusual more aggressive multiple therapies do increase the risk of these complications. In some units the policy to re-endoscope and endoscopically re-treat patients within the first 24 hours of the first endoscopy for peptic ulcer bleeding may increase these risks.

(v) Rescue Therapies. Surgical and radiological procedures are used when patients continue to bleed despite endoscopic treatments. Urgent surgery undertaken for peptic ulcer disease that does not respond to endoscopic therapy has a very high mortality: approximately 30% in the UK Bleeding Audit. ¹ Patients undergoing surgery are almost always frail and have multiple medical co-morbidities. Deaths usually follow postoperative complications including pneumonia, pulmonary embolus and stroke. It is possible that delays in surgery from repeated endoscopies contribute to postoperative complications and it is therefore wise to consult with surgical teams so that the optimum timing of any surgical intervention can be established. It is sensible to re-endoscope patients who re-bleed following primary endoscopic therapy and in selected cases to repeat endoscopic treatment. In some patients this is doomed to failure because of massive haemorrhage and then urgent surgery is required. A team approach is thus necessary in this severely ill patient group in order to obtain the best outcome.

There is increasing evidence that interventional radiology comprising embolization of the bleeding artery is effective and safe, yet in the United Kingdom it is rarely undertaken because of a lack of resources and out-of-hours unavailability of interventional radiologists. For varices rescue therapy essentially comprises TIPS. This procedure requires expert radiologists and their team. TIPS is only available in referral centres; safe transfer of patients with variceal bleeding from peripheral hospitals is an important issue and protocols for this need to be established in all units. Patients and their carers must be aware that the TIPS procedure is both a major procedure and one that is associated with complications, particularly hepatic encephalopathy.

Management issues

(i) Management teams should ensure that the basic standards of care, defined in the Introduction, are applied to all patients presenting as new hospital admissions or occurring in established inpatients.

(ii) Availability of urgent endoscopy. Sixty percent of patients presenting to hospitals in the UK because of acute gastrointestinal bleeding are admitted out-of-hours yet the audit showed that 45% of units do not have out-of-hours endoscopy rotas. The issue of whether or not out-of-hours endoscopy is cost-effective is rather controversial; there is no doubt that urgent therapeutic endoscopy is effective and saves lives in patients with severe active upper gastrointestinal haemorrhage. In addition there are data that suggest that units managing a significant number (more than 300 per year) of bleeding patients should provide out-of-hours endoscopy since this is cost-effective for patients with mild to moderate bleeding, as these can be managed more efficiently and leads to a reduced duration of hospital admission.

Management issues therefore relate to the volume of admissions since a small remote and rural hospital clearly cannot justify 24/7 endoscopy, whilst a large urban teaching hospital dealing with significant numbers of variceal and non-variceal bleeding should provide such a service.

(iii) Issues concerning safe transfer of patients have already been referred to in this article. A particular dilemma is that the most critically ill patients are at greatest need of transfer to a referral centre for therapeutic endoscopy, yet this group is at greatest risk during transfer. Arrangements must therefore involve accompanying medical and medical staff, an issue that may stress smaller units where staff numbers tend to be low.

(iv) Inequalities of services. Provision of endoscopy, interventional radiology and emergency surgery was found to vary across the UK in the audit. ¹ Such inequalities were sometimes a consequence of small units (where, for example, out-of-hours endoscopy provision is not cost-effective or where interventional radiology is not viable), but this was not always the case (some large urban hospitals not offering emergency 24/7 endoscopy or angiography). Where necessary services are lacking, units should develop referral streams that can deliver life-saving therapies to patients; whilst accepting that the development of these networks can be particularly challenging for management teams, this is essential if current service inequalities are to be overcome.

(v) Quality assurance. A particular issue is the expertise of endoscopists. Undoubtedly therapeutic endoscopy in bleeding patients is technically difficult and the audit showed that approximately a third of individuals on endoscopy rotas felt unable to manage bleeding oesophageal varices. It is therefore encumbent upon management teams to ensure that endoscopists have the appropriate skills to manage acutely bleeding patients. Furthermore, it is clear that as endoscopic advances have occurred, the best outcome depends upon a team approach and that nursing and other support for endoscopy should be available.