Abstract:
|
The multivariate longitudinal outcome data are commonly encountered in biomedical research. In clinical trials, often it is desired to show that the experimental therapy has an overall effect on multiple co-primary endpoints across time. For comparison of multivariate longitudinal outcomes between two groups, the robust rank-sum-test (O'Brien (1984)) and the improved one (Huang et al. (2005)) are widely used to detect the treatment efficacy. However, the aforementioned tests can only give a partial view of the treatment effect, since the test procedures utilize only the changes from baseline to the last visit of the endpoints and thus the experimental therapy is not evaluated throughout the whole treatment period. In this paper, we develop a novel rank-based test in longitudinal study to test whether the experimental therapy has positive effect overtime on multiple endpoints globally. Asymptotic properties of the proposed global test procedure are derived and thoroughly examined. Simulation studies under different scenarios are performed. The test statistics is motivated and applied to a recently-completed clinical trial of Parkinson's disease.
|