The evaluation of tests

Abstract

In order to correctly evaluate a test, at least four attributes should be measured: namely, sensitivity, specificity, accuracy and precision. Sensitivity is the proportion of diseased animals which are correctly identified, whereas specificity is the proportion of healthy animals which are correctly identified. These two attributes are important, not only because of the reasons implied by their definition but because they influence both the apparent prevalence of disease and proportion of test-positive animals which are actually diseased. The ability of a test to give a true measurement of the substance being measured, its accuracy, and its ability to give consistent results on the same sample, its precision, are good measures of quality control. Both these attributes influence the sensitivity and specificity of the test. Inaccuracies and inconsistencies arise from the test itself, the technician and the nature of the sample being tested.