[Principles for evaluating simple, diagnostic tests].

Abstract

The fundamental terms in the evaluation of diagnostic tests are described and the differences to corresponding test parameters in clinical chemistry are emphasized. The derivation of a two-by two contingency table is explained in detail and basic characteristics of diagnostic tests, i.e. sensitivity and specificity as well as their independence of the prevalence (in the sense of a-priori-probability) is pointed out. The importance of the so-called predictive values is shown graphically, also the problem of selecting a cut-off-level (discrimination value) for the evaluation of quantitative data. The various attempts to find a single value for the description of a diagnostic test are discussed and it is stressed that there is no principal necessity to use other terms than those mentioned above. Nevertheless it is possible to give a prevalence-independent measure which has some additional advantages in rating and comparing simple diagnostic tests.