Choice of treatment in the management of bleeding oesophageal varices in patients with cirrhosis.

URGENT control of haemorrhage is a basic surgical principle, but it is not always achieved most effectively by surgery. Emergency portacaval shunt can undoubtedly control bleeding from oesophageal varices, but the huge mortality in non-selected patients, as reported by Orloff, is prohibitive (Orloff et al, 1975). The idea of selective management based on clinical and laboratory findings was first suggested by Child as far back as 1964 when he introduced his now widely used classification of patients with cirrhosis (Child, 1964). However, even if one follows his concept of reserving portal decompression for A cases, the results of shunt surgery are often disappointing. Most controlled trials have failed to show any clear-cut advantage following either prophylactic or therapeutic shunt, and thus there has followed a reappraisal of the value of shunt. Realizing that the vast majority of patients with bleeding varices will never be fit for shunt, how should these shunt rejects be managed? First it is necessary to establish that bleeding is coming from the oesophageal varices, since it is accepted that patients with cirrhosis may be bleeding from other lesions. However, rigid adherence to the now popular policy of routine emergency endoscopy in all patients should be resisted. The pendulum has swung too far since there is no evidence that emergency endoscopy in all patients with haematemisis reduces mortality (Sandlow et al, 1974; Morris et al, 1975; Lee et al, 1977). It must also be acknowledged that it does not always provide the correct diagnosis since observer error may be as high as 30 per cent (Conn et al, 1967). The much higher incidence of acute mucosal lesions in the Ameri,gn series, when compared with the figures for the British Isles, may represent a higher incidence of alcoholic patients, or perhaps reflect over-vigorous lavage, which can produce artefactual bleeding. Certainly, the practice of inducing vomiting in order to displace residual clot while the endoscope is still in place in the stomach, may explain Sagawa's 15 per cent incidence of the Mallory-Weiss syndrome (Sagawa et al, 1973). The fact that varices are not actually bleeding at the time of examination does not exclude them as the source of bleeding. Indeed, it is unusual to demonstrate active bleeding into the lumen of the oesophagus, even by direct injection of contrast medium into the left gastric vein (Lunderquist, 1977). It is probably acceptable to assume the diagnosis on the basis of exclusion of other sites of bleeding, although by using a wide bore rigid oesophagoscope to distend the varices, one may demonstrate the site of bleeding when the thinner, flexible endoscope has failed to do so. All patients should have endoscopy at some stage to confirm the diagnosis, but emergency endoscopy should be selective.