|
Activity Number:
|
115
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
WNAR
|
| Abstract - #303990 |
|
Title:
|
Multinomial Group Sequential Design with Exact Method
|
|
Author(s):
|
Jiang Hu*+ and Aiyi Liu and Chengqing Wu
|
|
Companies:
|
FDA and National Institute of Child Health and Human Development and Yale University
|
|
Address:
|
1401 Rockville Pike, 400S, Rockville, MD, 20852-1448,
|
|
Keywords:
|
group sequential design ; exact method ; multinomial data ; parameter estimation
|
|
Abstract:
|
Group sequential testing has been widely applied in clinical trials to reduce the expected sample size. However, most group sequential designs were developed for normally distributed or simple binary data. In this paper we focus on analyzing and formulating group sequential designs and terminal point estimation for multinomial data. We first investigate group sequential design and parameter estimation for binomial data and then expand to multinomial data with exact distributions. The group sequential design involves determining the stopping regions and critical values. The MLE, a Whitehead bias-adjusted estimator and a Rao-Blackwell unbiased estimator are developed and studied. Numerical simulations are implemented to compare the performance of group sequential designs with exact method and that of group sequential designs with normal approximation method for multinomial data.
|
