I appreciate the comments from Dr Swets. In fact, his 1966 book1 played a major role in my education on signal detection theory. He is, of course, quite right in stating that variability in decision criteria can ceteris paribus lead to considerable differences in the sensitivities and specificities reported for a diagnostic test.In earlier drafts of my article, I displayed the curves that Dr Swets has presented, but came to slightly different conclusions. Clearly, for certain diagnostic tests, it is possible that different observers might use different criteria for defining a positive test result, and it is possible that the resultant sensitivities and specificities might fall on a typical ROC curve. Likely examination results for which this effect might be observed are ultrasound, lymphangiography, and mammography—tests where there is wide latitude for observer interpretation. My graphs of the data for these tests were similar to Dr
Harris JM. Sensitivities and Specificities of Diagnostic Tests-Reply. JAMA. 1982;248(5):549–550. doi:10.1001/jama.1982.03330050032021
Customize your JAMA Network experience by selecting one or more topics from the list below.