The Redundancy of Liver Tests in the Diagnosis of Cirrhosis Estimated by Multivariate Statistics

Abstract

Multivariate discriminant analysis was applied to 9 laboratory tests, per formed in 17 patients with cirrhosis and 19 control subjects. Bromsulfalein transport maximum was the most effective test in discriminating between the two groups. The combination of all 9 tests could not be shown to discriminate significantly better than the combination of bromsulfalein transport maximum and gamma-globulin. These 2 tests, however, could be replaced by the remaining 7 tests without a significant loss of discriminative information. The result of the discriminant analysis is greatly dependent on the size and selection of the materials. In larger materials more tests with significant contribution to the discrimination may be found.

Key Words:

• A priori probability
• discriminant analysis
• laboratory diagnosis
• liver cirrhosis
• liver function tests