In patients with cirrhosis and portal hypertension, gastrointestinal haemorrhage is a major complication and cause of death. Oesophageal varices are present in approximately 50% of patients with cirrhosis; the prevalence is higher in Child-Pugh C patients than in those with Child-Pugh A-B. Among patients with varices, the risk of gastrointestinal haemorrhage ranges from 30% to 50% and half of these patients die within six weeks after bleeding. Thus 10–15% of unselected patients with cirrhosis die from gastrointestinal bleeding. The number of deaths has decreased in the last few decades as a result of modern techniques (see below). The exact mechanisms of variceal rupture have not yet been determined and thus no ideal treatment has been found. However, haemorrhage has been shown to cause different complications such as sepsis or renal failure which may be responsible for death. In patients admitted for variceal haemorrhage, certain prognostic factors for death have been determined such as age, Child-Pugh C, hepatocellular carcinoma, early rebleeding, hepatic encephalopathy, and renal failure.1

Ideally, patients admitted for acute variceal haemorrhage must be taken to a medical centre with an emergency facility and an intensive care unit with experience in variceal bleeders. In the last 20 years, various medical, surgical, radiological, pharmacological, and endoscopic treatments have been used and most are effective in controlling variceal haemorrhage compared with no treatment or placebo. Most of these treatments, except endoscopic therapy, act by decreasing portal pressure. There is a negative relationship between treatment efficacy and invasiveness of the method. For example, surgical shunts or transjugular intrahepatic portosystemic stent shunts are more effective than pharmacological and endoscopic treatments but are more invasive and thus associated with more side effects. Although all of these treatments stop variceal haemorrhage, a significant difference in mortality has only been found with endoscopic sclerotherapy or terlipressin administration compared with placebo or no treatment. The discrepancy between the efficacy of treatment on variceal bleeding and survival rate suggests that other factors play a role in survival. Emergency endoscopic sclerotherapy is now the gold standard in the management of acute variceal haemorrhage as it increases hospital survival.2 A recent meta-analysis showed however that endoscopic sclerotherapy and variceal ligation are equally effective, suggesting that variceal ligation may also improve survival rate with fewer complications.3 Two controlled studies have shown that terlipressin administration is effective both in controlling bleeding and improving survival rate. In one trial, terlipressin stopped bleeding in 90% of patients with a significant difference in survival rate before discharge.4 In a second study, terlipressin was administered with glyceryl trinitrate by an intensive care team, within one hour after an emergency call before admission.5 In this trial, mortality due to bleeding episodes was significantly lower in the terlipressin group than in the placebo group at day 15 and day 42. These findings suggest that this drug should be administered as soon as possible before endoscopic investigation. Moreover, it is interesting to note that these two treatments (sclerotherapy and terlipressin administration) are equally effective in the initial control of variceal bleeding and in the prevention of early rebleeding.6

The second type of therapy is for the prevention of the complications of haemorrhage. Patients with gastrointestinal bleeding are at a high risk of bacterial infection and thus antibiotics must be prescribed. A meta-analysis confirmed that the use of prophylactic antibiotics prevents bacterial infection but also the risk of early rebleeding and more important, significantly improves survival rate.7Prevention of liver failure, including encephalopathy and renal failure, may also play a major role in the improvement in survival rate. Administration of blood products must be limited as it has been suggested that excess blood products may induce early rebleeding. Similarly, limiting certain more or less invasive investigations may reduce severe complications. Accordingly, improvements in survival rate may depend in part on the type of treatment. This is relatively clear for terlipressin administration8 but no explanation exists for endoscopic treatment.

Finally, mortality due to acute variceal haemorrhage has significantly decreased in the last few decades. Preliminary results of one study showed that inhospital mortality has decreased by 50% over the past 15 years.9 Moreover, the authors showed that although there were no deaths in Child-Pugh A patients following gastrointestinal bleeding, 25% of Child-Pugh C patients still died from haemorrhage. A second recent study followed 5000 patients in two cohorts separated by an interval of 11 years, who were followed for more than six years.10 The authors demonstrated that mortality in the late cohort significantly declined at 30 days (approximately 30%) and at six years (approximately 6%). These results were despite the fact that patients in the late cohort had more severe liver disease. Thus improvement in survival rate is even more significant with multivariate analysis. This study also showed that patients treated with sclerotherapy during the first hospitalisation had a lower mortality rate than the rest of the late cohort. In this issue ofGut, McCormick and O'Keefe have calculated the survival rate in patients with cirrhosis admitted for a first episode of variceal haemorrhage (see page 682).11 They compared the survival rate from “control” groups in approximately 1500 patients from 1960 to 2000. The results show a significant reduction (approximately 40%) in bleeding related mortality.

Although this improved survival rate after variceal haemorrhage might be expected, variceal haemorrhage was controlled for more than 20 years without any significant effects on survival rate. The only treatments known to improve survival rate are pharmacological or endoscopic treatments and antibiotic administration (see above). Nevertheless, a certain number of patients still die of variceal haemorrhage, suggesting that other factors must be studied. Thus new ideas, hypotheses, and approaches are needed to further our understanding of the treatment and mechanisms of variceal haemorrhage.