Bayesian Framework for Estimating the Incremental Value of a Diagnostic Test in the Absence of a Gold Standard

• Activity Number: 194
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Nandini Dendukuri*+ and Madhukar Pai and Daphne Ling and Ian Schiller
• Companies: McGill University and McGill University and Health Quality Ontario and Division of Clinical Epidemiology, McGill Univerity Health Centre
• Keywords: Incremental value ; Area under the curve ; Integrated Discrimination Improvement ; Bayesian ; diagnostic tests ; non-identifiable

Abstract

Our goal was to estimate incremental value of a new, imperfect test when the reference standard is also imperfect. Using Bayesian estimation of a latent class model, we determined an individual's latent disease status. We extended two commonly-used measures of incremental value based on predictive values [area under the ROC curve (AUC) and integrated discrimination improvement (IDI)] to the context where no gold standard exists. Methods were illustrated using simulated data and data on a new interferon-gamma release assay (IGRA) for latent tuberculosis screening compared to the tuberculin skin test (TST). We also show how to estimate incremental value of IGRAs when decisions are based on observed results. We found the incremental value is greatest when the new test has better sensitivity and specificity and in the absence of conditional dependence between the two tests. The incremental value of the IGRA depends on the sensitivity and specificity of the TST and may thus vary in different populations. Even in the absence of a gold-standard reference, incremental value statistics may be estimated and can aid decisions about the practical value of a new test.