Variceal bleeding and its ensuing complications correlate positively with the severity of liver disease. The average risk of bleeding in patients with cirrhosis who have not previously bled is 30%, with a 50% mortality rate within 6 weeks. This mortality rate is the rationale for prophylaxis. However, although fatal bleeding causes 35% of all deaths, patients who die after the first episode of bleeding represent only 15% of patients with cirrhosis and varices. Portal and intravariceal pressure, the appearance of oesophageal varices on endoscopic examination, severity of liver disease and alcohol abuse are independent risk factors for the occurrence of the first bleeding episode. In sinusoidal portal hypertension, the presence of varices indicates a hepatic venous pressure gradient greater-than-or-equal-to 12 mmHg. Although hepatic venous pressure gradient tends to be higher in patients who bleed or have large varices, bleeding risk is not related linearly to pressure above this threshold. Tension on the variceal wall relative to varix radius may be critical and increasing variceal size, in conjunction with wall thinness, may favour rupture at lower intraluminal pressures. The North Italian Endoscopic Club's simplified index for the risk of a first bleeding episode is based on Child class, variceal size and presence of red wale markings, although there may be other independent risk factors. Abstention from alcohol can decrease variceal size and the number of cherry-red spots. Because large varices are unlikely to develop de novo within 2 years, biennial endoscopic screening is sufficient for patients without varices; annual endoscopy is recommended for those with small varices. The risk of bleeding is highest in the first year after entry into a clinical trial but nonlinear thereafter - an important consideration for prophylactic therapy. Presumably, patients with a history of varices that have not bled have a different risk profile from those whose varices are discovered on initial diagnosis of liver disease.
