Diagnostic Testing: Assessing the Value of Studies of Diagnostic Tests
Diagnostic testing is a core domain in any evidence-based medicine curriculum and is also a place where the concepts typically tested in early health professions education tend to differ from the concepts needed to apply data clinically. We believe it is critical to move beyond sensitivity, specificity, and predictive values into likelihood ratios. Only likelihood ratios allow us to apply a test result to a patient, moving from our pre-test probability to post-test probability for that patient. The diagnostic testing arena is a good place to touch on the math behind likelihood ratios so that learners understand how it is derived, but to quickly pivot to the application of the numbers. The process of arriving at a pre-test probability and deciding what post-test probability can allow us to stop testing and move to action is rich clinical material that learners should spend time grappling with. In this chapter we review sources of bias in studies of diagnostic tests, deciding on pre-test probability, calculating and applying likelihood ratios for multiple levels of a test and application to patient care.
