Noninvasive predictors of large varices in patients hospitalized with gastroesophageal variceal hemorrhage

Aim: To identify noninvasive factors predicting the presence of large varices (LV) in patients hospitalized with gastroesophageal variceal hemorrhage (GEVH). Methods: Case records of patients admitted with GEVH between January 1998 and June 2005 were retrospectively analyzed. Relevant clinical parameters assessed included Child-Pugh class, ascites (clinical and/or on ultrasound), portosystemic encephalopathy (PSE), splenomegaly (clinical and/or on ultrasound), and hemodynamic instability. The laboratory parameters assessed were hemoglobin level, platelet count, prothrombin time, serum bilirubin, and albumin. The ultrasonographic characteristics noted were splenic size, presence of splenic varices, and portal vein diameter. Results: A total of 420 patients (264 men) presented with GEVH during the study period. The mean age, gender distribution, and presence of cirrhosis were similar in the two groups. Liver cirrhosis with hepatocellular carcinoma (HCC), Child-Pugh class C, presence of clinically detectable ascites, grade 3-4 PSE, detectable splenomegaly, previous history of GEVH, hemodynamic instability and platelet count <91,000 were more common in the LV group. The frequency of radiologically detected ascites, splenomegaly, and portal vein diameter were similar in both groups. On multivariate analysis, the independent predictors for the presence of LV were cirrhosis with HCC, clinically detectable splenomegaly, hemodynamic instability, a previous history of GEVH, platelet count <91,000, and splenic size ≥158 mm. Conclusion: Cirrhosis with HCC, clinical splenomegaly, hemodynamic instability, a previous history of GEVH, thrombocytopenia (i.e., platelet count <91,000), and splenic size ≥158 mm are independent noninvasive predictors of large varices in patients hospitalized with gastroesophageal variceal hemorrhage.