Multivariate dose response test

In neuroscience clinical research, patients are frequently treated with different doses of the same drug to determine a dose level that is efficacious with tolerable adverse effects. Efficacy is usually evaluated using multiple psychometric instruments. Most often the instruments are either binary or ordinal scaled. Conventional assessment of treatment efficacy is conducted per each instrument individually. However, it is important to combine these to provide an overall evaluation of the treatment efficacy. In this paper, we consider a statistical test of dose-response based on the simultaneous assessment of efficacy measured via the various instruments using a multivariate method. We derive the asymptotic distribution of the proposed test. Data from a neuroscience clinical trial is used to demonstrate the proposed method.