Some unanswered problems with probability analysis: Possible impact on exercise thallium test use
Probability analysis has been advanced as a means of selecting patient-test combinations that most efficiently lead to a diagnosis in patients with suspected coronary artery disease (CAD).1 Results of exercise cardiovascular nuclear medicine tests have yielded the greatest shifts in disease likelihood among patients with intermediate pretest likelihoods for disease. Furthermore, the differences in posttest probability produced by nuclear and electrocardiographic exercise tests are greatest among patients with intermediate pretest likelihoods for disease. However, when we examined the actual use of exercise thallium scintigraphy, most patients referred by both house staff and attending physicians had either very high or low pretest likelihoods for disease.2 One explanation for these ordering practices is a lack of familiarity with probability analysis.3 Alternatively, the clinical efficacy of a test may not be valued in proportion to its ability to alter disease likelihood. For example, exercise thallium scintigraphy has been advocated as a valuable means of excluding significant coronary disease, particularly when the pretest likelihood for coronary disease is relatively low.4 In this study we asked house staff to calculate the predictive value of an exercise thallium test result. We then examined the relation between likelihood estimates and stated preferences for testing in an attempt to discover motivations for test ordering in actual practice. © 1985.
Digital Object Identifier (DOI)
Additional Document Info
has global citation frequency