Distinction between chronic hepatitis and liver cirrhosis in patients with hepatitis C virus infection. Practical discriminant function using common laboratory data

In order to distinguish patients with cirrhosis from those with chronic hepatitis, multivariate discriminant analysis was performed using common laboratory data. A total of 205 consecutive patients were diagnosed by peritoneoscopy and biopsy as having chronic liver disease caused by hepatitis C virus (HCV), 168 with chronic hepatitis and 37 with cirrhosis. Twenty variables and their natural logarithmic transformation were employed in the multivariate analysis. After stepwise variable selection, the following function was finally obtained to discriminate the disease severity, z = 0.120 x γ-globulin (%) + 0.423 x In (hyaluronate) (μg 1-1) - 0.059 x platelet ( x 104 counts per mm3) - 0.364 x sex (male, 1; female, 2) - 3.953. Since the function contained an expression of logarithm and was slightly troublesome to apply, we prepared another discriminant function composed of usual figures without logarithmic transformation, z = 0.124 x (γ-globulin (%)) + 0.001 x (hyaluronate) (μg 1-1) - 0.075 x (platelet ( x 104 counts per mm3)) - 0.413 x gender (male, 1; female, 2) - 2.005. When a positive result is calculated in the latter equation, the diagnosis of the liver disease indicates cirrhosis, and negative result chronic hepatitis. Accuracy of the two discriminant functions was 90.3 and 91.2%, respectively. A concise linear discriminant function could successfully differentiate liver cirrhosis from chronic hepatitis with an accuracy of 91.2%. (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.