The treatment of variceal haemorrhage remains controversial. Resurgence of old therapies (endoscopic sclerotherapy) and the introduction of new modalites (obliterative angiotherapy) has exacerbated the disputeL. Problems in assessment of therapeutic value are related to many variables2 including the severity of the liver disease and the effectiveness of the measures used to control the haemorrhage3. Non-operative procedures for thc. control of bleeding oesophageal varices have received increased attention in recent y e a r~~3~. This is related to the high mortality rates associated with emergency operations and the frequency of incapacitating encephalopathy.Why patients with portal hypertension bleed from oesophageal varices is not known6. Bleeding is related to variceal size, but size and extent of varices are unrelated to portal pressure7. Gastro-oesophageal reflux does not cause bleedings. Following a variceal bleed, the mortality without surgery, primarily due to bleeding, is 42 per cent at 4-6 weeks"9 varying from 22 to 84 per cent3910. Approximately one-third of patients rebleed within 6 weeks and one-third of the survivors subsequently rebleed if not treated ~u r g i c a l l y~~. Any substantial improvement in this survival requires safe and effective treatment in the early periodI2. In the last few years advances have been made in non-operative procedures for controlling variceal bleeding.
Medical treatment Vasopressin (Pitressin)Following Kehne's13 initial successful report on using vasopressin in bleeding oesophageal varices, it is the first line of therapy provided there are no contra-indications to its use. Vasopressin reduces portal pressure and contracts the oesophageal musculature"+Js. Vasopressin can be given intravenously as a bolus, by continuous infusion or selectively into the superior mesenteric arteryl6-I8. The conventional bolus of 20 units vasopressin intravenously over a 2-20 min period will not permanently stop bleedingl9. Intravenous infusion of a low dose of vasopressin (0.4 unitslmin) stopped bleeding in 86 per cent of episodes compared with only 12.5 per cent when boluses of 20 units were administered (P < 0.01) (Reference 20). This study contained only a small number of sequential patients. A controlled study comparing continuous intra-arterial with intravenous infusions of vasopressin in 22 cirrhotic patients showed that either route controlled bleeding in 50 per cent of the patients with equal side effects2'. Continuous intravenous vasopressin can be administered more easily and has distinct advantages over bolus and arterial infusions20-22. The optimum intravenous dose is unknown. Low doses appear to be as effective as a high dose infusion but tachyphylaxis develops21. Several controlled clinical studies have now confirmed the effectiveness of v a s o p r e~s i n~~-~~. Side effects include decreased pulmonary bloodflow and cardiac o u t p~t~~-~~ which can be reversed by intravenous nitroglycerin27. Ischaemic colitis28, emphysematous gastritis29 and local gangrene'O may also occur...