[Predictive factors of the first digestive hemorrhage and death in cirrhotic patients with esophageal varices].

Predictive factors for the first digestive tract hemorrhage and for death in cirrhotic patients have been partially studied in prophylactic shunt trials and never prospectively according to multivariate analysis method. We prospectively followed 106 cirrhotic patients (88 p. 100 with alcohol abuse) with large esophageal varices without past history of bleeding during a mean period of 13.3 +/- 6.1 months. Of 23 clinical and biochemical variables recorded at entry as well as alcohol withdrawal evaluated during the course of the study, we looked for those which had independent prognostic value for the risks of bleeding and death according to a multiple regression analysis (Cox's model). At the end of the study, 30 patients had bled and 40 were dead. The cumulative one-year percentages of patients were 72 p. 100 free of bleeding and 65 p. 100 surviving (95 p. 100 confidence limits: 62 and 81, 54 and 73, respectively). The bleeding risk was positively correlated with variceal size, plasma creatinine and negatively correlated with age and mean arterial pressure. The bleeding risk varied according to variceal size: medium: 23 p. 100, large: 44 p. 100 (p less than 0.05). An increase in Child-Pugh score, plasma creatinine and mean corpuscular volume as well as hepatomegaly were independent predictive factors for death. The death rate was (according to Child-Pugh grade): A: 14 p. 100, B: 33 p. 100, C: 50 p. 100 (p less than 0.01). We conclude that bleeding risk is maximum in young patients with large varices and the death rate is dependent on severity of liver disease.(ABSTRACT TRUNCATED AT 250 WORDS)