|
Activity Number:
|
313
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #304908 |
|
Title:
|
Are Flexible Designs Sound?
|
|
Author(s):
|
Carl-Fredrik Burman*+ and Christian Sonesson*+
|
|
Companies:
|
AstraZeneca Pharmaceuticals and AstraZeneca Pharmaceuticals
|
|
Address:
|
R&D, Mölndal, SE-431 83, Sweden AstraZeneca R&D, Mölndal, SE-431 83, Sweden
|
|
Keywords:
|
adaptive design ; sample size reestimation ; sufficiency ; inference principles ; clinical trial
|
|
Abstract:
|
Some adaptive designs are controversial. Sample size re-estimation (SSRE) is one, in particular when based on interim efficacy estimates. The idea is attractive: why not modify the sample size in light of new information? With a weighted test, the type I error level can be protected. However, this test is not based on the sufficient statistic. The unblinded SSRE has been criticized as being less efficient than certain group-sequential designs. Burman & Sonesson argue that the violation of the sufficiency principle makes the inference invalid. They claim the weighted test may lead to paradoxical results: The null hypothesis of zero mean can be rejected, although the average response is zero. This session will contain a presentation and invited discussion of Burman & Sonesson's paper. The discussion will appear in Biometrics.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
