4.3. Influence of analytical quality on the diagnostic power of a single S-CK B test in patients with suspected acute myocardial infarction

Abstract

We have compared two theoretical methods for assessing the effects of changing analytical quality of a clinical chemical test. The test considered was S-creatine kinase B subunit activity, used as the only diagnostic criterion for acute myocardial infarction. The two methods applied were based on (i) graphical analysis, and (ii) computer simulation. The results comprise the effects of changing analytical imprecision and bias on the weighted sum of misclassified cases on basis of the test results. The two methods yield comparable results at high analytical imprecision, but due to differences in assumptions about the error distribution of test results the difference increases with increased analytical imprecision. The graphical analysis is easily performed but is restricted in possible applications. The computer simulation is not a generally available methodology, but allows for mixing of different types of statistical distributions, which is not the case in conventional variance analysis.

Key Words:

• computer simulation
• graphical analysis
• loss function
• misclassification