Comparing the Predictive Values of Diagnostic Tests: Sample Size and Analysis for Paired Study Designs

Chaya S. Moskowitz, Memorial Sloan-Kettering Cancer Center
Margaret S. Pepe, Fred Hutchinson Cancer Research Center and University of Washington

Abstract

In this paper we consider the design and analysis of studies comparing the positive and negative predictive values of two diagnostic tests that are measured on all subjects. Although statistical methodology is well developed for comparing diagnostic tests in terms of their sensitivities and specificities, comparative inference about predictive values is not. We derive analytic variance expressions for the relative predictive values. Sample size formulas for study design ensue. In addition, two new methods for analyzing the resulting data are presented and compared with an existing marginal regression methodology.