Evaluation of diagnostic tests

Measures of test efficacy are its repeatability or reliability and its validity. Validity is computed by comparing it against an older ‘gold standard’ test, which is supposed to unequivocally give the diagnosis. Sensitivity and specificity are inherent properties of a test. Predictive values give the probability of disease or no disease when the test result is known and vary with the prevalence of disease in the setting it is used. Receiver operator characteristic curves are used to determine the best cut off and also to choose between tests with numerical values. Likelihood ratios give the odds of disease at a particular test result.