Airway

The mental state should be noted carefully because a drowsy or comatose patient is at high risk of aspiration if he or she continues to vomit blood. The mental state may be impaired by a number of factors, including:

If the patient is unconscious or has an impaired gag reflex, the airway must be protected, especially if vomiting continues. The patient should be kept flat on his or her side. A cuffed endotracheal tube may need to be inserted. Such patients should be managed in an intensive care facility.

Hypotension and shock

Assessment of hypovolaemia involves simple bedside clinical observations. The patient may be clammy and sweaty with cold peripheries and a fast thready pulse. There may be associated confusion. The blood pressure will be low, often under 90 mmHg systolic. If any of these signs are seen, resuscitation should commence immediately. At least two large-bore intravenous cannulae should be inserted into large peripheral veins. A central venous line should be placed in high-risk cases (see below). Rapid infusion of isotonic saline followed by a plasma expander such as Haemaccel® should be commenced and blood samples should be drawn urgently for full blood count, coagulation screen, blood group and cross-matching of four to six units of packed cells, and urea, electrolytes and liver function tests.

As a rule of thumb, patients who have obvious signs of shock, with clammy peripheries and low blood pressure, may have lost up to 50% of their circulating blood volume. If these signs are not present, the patient may be sat up carefully and a check made for a postural drop in blood pressure. If this is present, it is likely that 10–20% of blood volume has been lost. Remember that with haemoconcentration immediately after a bleed, the haemoglobin may initially be near normal despite the loss of a considerable amount of blood.

Patients with a gastrointestinal bleed have lost ‘whole blood’ and there is, therefore, logic in transfusing them with whole blood (Box 10.2). In practice, however, donated blood is separated into packed cells and other products, such as platelets and plasma, which may be used in different clinical situations. Thus, in current clinical practice, patients with a significant bleed are given packed cells alternating with a plasma expander such as Haemaccel if they are hypovolaemic and packed cells alone if they are normovolaemic but anaemic. The aim of transfusion is to restore circulating blood volume so that the blood pressure is normal and to correct anaemia so that the oxygen carrying capacity of the blood is satisfactory. This generally means maintaining a haemoglobin level of approximately 100 g/L. One unit of packed cells will increase the haemoglobin level by 10 g/L (haematocrit by 3%).

In hypovolaemic patients, packed cells are transfused rapidly until the patient is haemodynamically stable. Rarely, group-specific uncross-matched blood or O rhesus-negative blood will be required. In haemodynamically stable patients, packed cells are transfused slowly, approximately one unit every 2 hours.

If the patient is coagulopathic or needs more than four units of packed cells, fresh frozen plasma (two units initially) should also be given to provide clotting factors.

Active bleeding

As resuscitation is proceeding, it is important to consider whether bleeding is still active. This assessment is straightforward if the patient continues to vomit bright red blood or if the blood pressure keeps falling. Passing melaena does not necessarily signify active bleeding; it may simply be old blood that has worked itself through the bowel. However, if the patient starts to pass more ‘fresh’ melaena, which is maroon-coloured or even bright red with visible clots, active bleeding is likely.

Source of the bleed

A targeted history should be obtained if possible from the patient or from the family. Particular attention should be given to:

Endoscopy

Even if the history points to a likely diagnosis (e.g. past history of duodenal ulcer), the cause of bleeding may be different on this occasion. Studies have shown that the clinical diagnosis as to the most likely cause of an upper gastrointestinal bleed is correct in only 60% of cases. Thus, investigation is necessary to establish a correct diagnosis.

Upper gastrointestinal endoscopy is the single most useful test. If performed within 24 hours of presentation, the cause of bleeding will be found in 90–95% of patients (Figs 10.1 to 10.3). Furthermore, it may permit the endoscopist to perform therapeutic interventions that in turn may arrest the bleeding or minimise the chance of further bleeding. It requires considerable expertise, especially in the situation of an actively bleeding lesion, and is not without some risk.

Figure 10.1 Arterial spurting from a gastric ulcer.

Figure 10.2 Gastric varices.

Figure 10.3 Gastric arteriovenous malformation.

So who should be endoscoped and when should it be done? This decision is more easily made if consideration is given to why the endoscopy is being done. First, the aim is to make an accurate diagnosis. Secondly, a prognosis for further bleeding may be given, based on the endoscopic findings. Finally, a bleeding lesion or one at high risk of re-bleeding may be able to be treated. However, if the patient is not in a high-risk category, it may not be necessary to do an emergency procedure. In general terms, endoscopy should always be done within 24 hours but, for patients considered being at ‘high risk’, particularly if there is a possibility of oesophageal varices, emergency endoscopy should be arranged once the patient has been adequately resuscitated.

Risk factors for greater morbidity and mortality from haematemesis are now well known and are listed in Box 10.3.

These patients should be targeted for the most aggressive management with emergency endoscopy. Endoscopy for acute haematemesis requires a high level of skill and experience. The main risk to the patient is of aspiration of blood, especially if sedation is used, and all staff must be aware of the need to protect the patient’s airway with a cuffed endotracheal tube, if necessary. There is some evidence to support the administration of intravenous erythromycin (approximately 250 mg) prior to endoscopy. The prokinetic effect of this drug clears blood from the stomach, thereby potentially improving visualisation of bleeding lesions and probably also reducing aspiration risk.

It should be noted that in about 80% of patients bleeding has stopped spontaneously upon presentation. In the remaining 20% of patients, bleeding persists or recurs during their period of hospitalisation. The mortality in this group increases as much as eightfold compared to those without further bleeding.

Oesophageal varices

Oesophageal varices are dilated submucosal veins forming a portal systemic circulation anastomosis in patients with portal hypertension. They look like large varicose veins and bulge into the oesophageal lumen. Arising from these veins are very thin-walled vascular channels, lined only by endothelium, extending into the squamous epithelium of the oesophagus. These have a high risk of rupture, which is considered to be mainly precipitated by sudden pressure rises.

About 50% of patients with cirrhosis of the liver have oesophageal varices, and 30% of these varices bleed within 2 years of their diagnosis. After bleeding, the risk of further bleeding is very high—about 70% over the ensuing 2 years. The mortality rate from bleeding oesophageal varices ranges from 40% to 70%. Predictors of haemorrhage include the presence of very large varices and varices with ‘cherry red spots’. These red spots represent the intraepithelial vascular channels arising from the varices. Ongoing alcohol ingestion, thrombocytopenia and poor liver synthetic function are also predictors of variceal bleeding.

If bleeding oesophageal varices are found during endoscopy, rubber band ligation is the current treatment of choice. Bleeding can be controlled in 80–90% of cases with a relatively low risk of complications. If passage of the rubber band ligating device mounted on the tip of the endoscope is not feasible or if equipment and expertise for rubber band ligation are not available, injection sclerotherapy may be performed (Fig 10.4). This involves direct injection of a sclerosant such as ethanolamine oleate or sodium tetradecyl sulfate into the varices. Injection sclerotherapy has a very high success rate in controlling the bleeding, although with a greater risk of complications including sepsis, oesophageal stricture and mediastinitis.

Figure 10.4 Injection sclerotherapy equipment for variceal bleeding.

Pharmacological treatment is frequently used in the control of variceal bleeding. The most widely used drug in this situation is intravenous octreotide, which is an analogue of somatostatin. It can be given acutely in the emergency room if there is a high index of suspicion that varices are the cause of the bleed, even before endoscopic confirmation. It is given by an initial bolus injection of approximately 25–50 mcg, followed by an infusion of 25–50 mcg octreotide per hour in 5% dextrose. An octreotide infusion is regarded as part of the resuscitation process. It works by decreasing portal venous blood flow and has been shown to control variceal bleeding in over 70% of cases. Octreotide is a safe and effective vasoactive agent. The benefit is more prominent if octreotide is prescribed early, even before endoscopy. Octreotide has also been shown to be effective when used as an adjuvant therapy in combination with endoscopic therapy. Recurrent bleeding episodes and hence requirement of transfusion are significantly reduced.

Sometimes, the above approaches fail and the patient continues to bleed. This is a very high-risk situation and a long-term management plan must be prepared. Important questions are: