Evaluation of Diagnostic Tests with Binary/Dichotomous Output: a Decision Analytic Approach

• Activity Number: 410 - Diagnostic Tests: Regulatory Considerations of Intermediate Outputs and Oncology Screening, and Optimal Point of Heterogeneous ROC Curves
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Arianna Simonetti* and Bipasa Biswas
• Companies: U.S. Food and Drug Administration - CDRH and U.S. Food and Drug Administration - CDRH
• Keywords: diagnostic tests; sensitivity ; specificity; binary output; decision analysis

Abstract

Evaluating performance of a diagnostic test with binary/dichotomous output is not always straightforward and may present with several statistical challenges. Common metrics used for the performance of a single diagnostic test device include sensitivity, specificity, positive and negative predictive values and likelihood ratios. However, employing such metrics for the comparison between two tests, either separately or as summary measures, is complicated. A decision analytic approach using patient threshold for undergoing a diagnostic test may provide a better tool in solving issues related to inference-based comparison and clinical consequences of misclassification associated to the comparison of diagnostic tests. Examples of comparison between two dichotomous diagnostic tests will be presented.