[Non-endoscopic predictors of large esophageal varices in patients with liver cirrhosis].

BACKGROUND/AIMS: The aim of this study was to identify non-endoscopic predictors for the presence of large esophageal varices in Korean patients with liver cirrhosis. METHODS: Among 736 patients with liver cirrhosis newly diagnosed between the year 2001 and 2005, 245 patients (171 men and 74 women, mean age of 51.9 years) fulfilled the inclusion criteria and underwent EGD as screening tests for esophageal varices. Fifteen variables were analysed to identify the presence of large esophageal varices. RESULTS: Esophageal varices were noted in 186 patients (75.9%) and large varices in 55 patients (22.4%), while 59 patients (24.1%) had no varices at the time of initial diagnosis of cirrhosis. The causes of liver cirrhosis were viral hepatitis (41.2%), chronic alcoholism (42.4%), viral hepatitis/alcoholism (9.8%), and others (6.6%). Fifty-one percent, 35.1% and 13.9% of the patients belonged to Child-Pugh class A, B, and C, respectively. Variables associated with the presence of large esophageal varices on univariate analysis were the presence of ascites, splenomegaly (long-axis >or= 12 cm by ultrasound measure), alcoholism, Child-Pugh class, platelet count, prothrombin time, and albumin. On multivariate analysis, alcohol, splenomegaly, and ascites were significantly associated with the presence of large esophageal varices. If the patients have two of them, sensitivity and negative predictive value were 80% and 91.7%, respectively. Patients without all three factors had no large esophageal varices. CONCLUSIONS: These results suggest that patients who have at least two among ascites, splenomegaly, and alcoholism would have an increased risk of having large esophageal varices.