Understanding and using diagnostic tests

Abstract

Every diagnostic test causes additional cost if not additional risk. The cost and risk are justified if the test result is likely to change what is done for patients. This includes a diagnostic test that does not alter therapy but provides important prognostic information for the parents. Whether the test result changes therapy depends on: the action threshold (the probability of disease at which the clinician offers the therapy or intervention because it would do more good than harm). the patient's pretest probability of disease. the test performance (measured by LRs, sensitivity, and specificity). For all tests there is an inverse relationship or trade-offbetween sensitivity and specificity. A good diagnostic test is one in which this trade-off is relatively small within the range of sensitivities and specificities that are clinically relevant. When diagnostic tests are used in low-risk subjects (as in screening tests), the performance of the test is critical because missing a significant number of cases is considered unacceptable and subjecting healthy individuals to the repercussions of false-positive test results is also problematic. The decision to introduce a screening program also depends on the availability and cost of screening and diagnostic tests, the potential for testing to cause harm, the prevalence and prognosis of the condition, the availability of effective treatment, and the importance of presymptomatic treatment.