An Evaluation of Factors Influencing the Discriminative Effectiveness of a Group of Liver Function Tests

Abstract

Summary An evaluation has been made of the relative effectiveness of nine representative liver function tests in differentiating between a group of patients with cirrhosis and a group of normal subjects. The Bromsulfalein test was the most effective single test. The intravenous galactose tolerance and hippuric acid tests were approximately three-fourths as effective as the Bromsulfalein test; the zinc sulfate turbidity, 24-hour urine coproporphyrin, and per cent cholesterol ester tests were approximately three-fifths as effective; and the 24-hour urine urobilinogen, thymol turbidity, and total serum bilirubin tests were approximately two-fifths as effective. After allowing for interdependence among the tests, only four tests—the Bromsulfalein, zinc turbidity, hippuric acid, and urine coproporphyrin—were found to contribute independently to the discrimination between the normal and cirrhotic groups. The other five tests made little contribution beyond that already represented in the four independently significant tests. A weighted linear combination of these four tests, called the cirrhosis abnormality score, was significantly, though not greatly, more effective than the Bromsulfalein test alone for purposes of group distinction. It also misclassified fewer abnormal individuals. The independent contribution of the Bromsulfalein test to the optimal combination was one and one-half times that of the coproporphyrin test and two and one-half times that of the zinc turbidity and hippuric acid tests.