Performance of a Diagnostic Test: A Decision Analysis Perspective

• Activity Number: 640
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Risk Analysis
• Author(s): Rama Lakshmi Vishnuvajjala*+
• Companies: FDA/CDRH
• Keywords: diagnostic test ; clinical consequences ; ROC AUC ; decision analysis

Abstract

A test for which is better than a random test is often called an informative test. But these tests may not necessarily have any clinical utility, depending on how much better than a random test they are and the consequences of false positives and false negatives. Several methods comparing tests implicitly assume false positives and false negatives have the same weight, which, in general is not true. For a screening test for cancer, a false negative would have much more serious consequences than a false positive. The degree of seriousness would depend on what the next step is for patients who are test positive. We will look at some examples of tests which may have the same sensitivity and specificity but can have entirely different consequences for the patients.