6 Evaluating diagnostic tests

Abstract

Diagnosis-based decision-making is fundamental to clinical practice. In this chapter, we have seen how clinical symptoms can select a group of patients in whom diagnostic tests can then help the clinician to reach or exclude a diagnosis. Informed use of a diagnostic test requires identification of valid evaluations of its diagnostic abilities. Conventional measures of a test's effectiveness, such as sensitivity and specificity, are of limited use when applying diagnostic tests in clinical practice. Likelihood ratios provide an explicit tool for updating diagnostic probabilities according to test results, and can incorporate variations between individuals in their risk of disease. Their use may help both individual patient management and the appropriate allocation of diagnostic and therapeutic resources.