The overall objective of this proposal is to develop powerful global statistical tests for comparing Parkinson disease treatments. Since no known biological marker of disease progression is identified, treatment efficacy is evaluated through multiple correlated outcomes. Conventional clinical trial designs are based on single primary outcomes. Such designs often either make clinical findings from other outcomes unclear (i.e., result in a lack of power to detect), or make sample size unnecessarily large in order to assure an appropriate small overall Type I error. In recent years, there is a growing demand for cost-effective clinical trial designs and efficient data analyses. A global statistical test (GST) is a single test which can """"""""squeeze"""""""" significance out of many single non-significant tests to enhance the power of the overall test. Several GSTs have recently been introduced into the literature, but there are limitations to the applicability of these GSTs to Parkinson disease treatment comparisons. We propose to develop new global statistical tests, which are particularly suitable for Parkinson disease treatment comparisons and are more powerful in detecting specified clinically meaningful treatment differences than tests based on single outcomes. We will 1. Develop nonparametric global statistical tests for comparing treatments in Parkinson disease clinical trials with multiple correlated outcomes; 2. Develop large sample properties of the global statistical tests and evaluate their small sample behaviors; 3. Develop sequential Parkinson disease clinical trial designs using global statistical tests; 4. Apply global statistical tests in Parkinson clinical trials; 5. Develop statistical software to facilitate the application of global statistical tests in Parkinson disease clinical trials.
