Decision-making model for a non-invasive diagnosis of compensated liver cirrhosis

Liver biopsy is used as a gold standard in the diagnosis of chronic liver disease. However, this procedure is not without risk to the patients. This study was aimed to evaluate whether clinical, ultrasonographic, and biochemical variables may discriminate between well-compensated liver cirrhosis and non-cirrhotic chronic liver disease. Logistic regression analysis was used to assess the independent predictive value of each variable. Moreover, the post-test probability of the diagnostic variables was converted into "weights" which positively correlated with the likelihood of diagnosis of liver cirrhosis. We applied a decisional rule based on the diagnostic "weight" of each variable to 412 patients, 278 with well-compensated liver cirrhosis and 134 with non-cirrhotic chronic liver disease, diagnosed by liver biopsy with/without laparoscopy. By adding the diagnostic "weights" of each variable, liver cirrhosis and non-cirrhotic chronic liver disease were correctly diagnosed in 67% and 75% of cases, respectively. This result was validated in a split sample of 100 patients (50 with compensated liver cirrhosis and 50 with non-cirrhotic chronic liver disease) randomly selected from the acquisition sample. Based on this data, diagnosis of well-compensated liver cirrhosis can be reached in over 67% of cases without using invasive procedures. However, a prospective study is needed to confirm the clinical reliability of this method.