Title: A method for testing multiple binary endpoints having a latent continuous distribution in clinical trials
Authors: Takuma Ishihara - Gifu university (Japan) [presenting]
Kouji Yamamoto - Yokohama City University (Japan)
Abstract: In confirmatory clinical trials, two or more primary endpoints are often used to assess the efficacy of a test treatment. For example, in a clinical trial for patients with rheumatoid arthritis, a percentage of patients achieving a response of 20\% improvement according to the American College of Rheumatology criteria (ACR20) in short term, and an achieving Disease Activity Score (DAS) below 3.2 in long term, are often used as primary endpoints for treatments. These endpoints can be observed as binary variables, but these variables have a latent continuous distribution. Furthermore, in general, it becomes more difficult to demonstrate that all of the endpoints are significant as the number of endpoints increases. Therefore, we propose a new test statistic for multiple binary endpoints which have a latent multivariate normal distribution in the framework. We confirm the efficacy of a test treatment when it is superior for at least one of the endpoint and not clinically inferior for the remaining endpoints. The performance of the proposed testing procedure is demonstrated through Monte Carlo simulations. We evaluate the empirical power and empirical type I error rate.