A1450
Title: Comparison of the predictive values of two binary tests in the presence of categorical covariates
Authors: Jose Antonio Roldan Nofuentes - University of Granada (Spain) [presenting]
Abstract: The comparison of the predictive values of two binary diagnostic tests is a topic of interest in the study of statistical methods for medical diagnosis and has been the subject of different studies in the statistics literature. In clinical practice, it is frequently found that when comparing diagnostic parameters, covariates are observed in all of the individuals in the sample, and it is necessary to adjust for these covariates. In this framework, a global hypothesis test is proposed to simultaneously compare the predictive values of two binary diagnostic tests when all of the individual's categorical covariates are observed. The global hypothesis test is solved through logistic regression and multinomial logit models. Simulation experiments were carried out to study the asymptotic behavior of the method proposed when a binary covariate is observed, and this was compared to the asymptotic behavior of the global test when the covariate is not used. In general terms, the method based on the regression models shows better asymptotic behavior than the other method. The results were applied to a real example.
