|
Activity Number:
|
276
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #302145 |
|
Title:
|
Estimation After Step-Down Tests in the Dose-Finding Studies
|
|
Author(s):
|
Yi-Hsuan Tu*+ and Ying Kuen Cheung
|
|
Companies:
|
National Cheng Kung University and Columbia University
|
|
Address:
|
Department of Statistics, No. 1 Ta-Hsueh Road, Tainan, International, 70101, Taiwan
|
|
Keywords:
|
Dose finding ; Familywise error ; Multiple comparisons
|
|
Abstract:
|
This article discusses the estimation of treatment effect at the identified minimum effective dose (MED) in a dose finding placebo-controlled study. The identification of MED is often formulated as a multiple comparison problem, and as such, demands a strict control of the familywise error rate (FWE). While the Bonferroni procedure and the Dunnett's procedure control the FWE, step-down tests may generally improve the probability of selecting the correct dose by avoiding multiplicity adjustment. In order to preserve the nominal coverage probability, confidence lower bound for the treatment effect at the identified MED is equal to the hypothesized treatment effect. We propose a two-stage procedure in which the step-down test is conducted at a slightly more conservative level than nominal, so that the remaining error rate may be spent in the estimation stage.
|
