|
Activity Number:
|
332
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #300551 |
|
Title:
|
Covariate Adjustment for Adaptation in Trial Design
|
|
Author(s):
|
Sue-Jane Wang*+ and Hsien-Ming James Hung
|
|
Companies:
|
Food and Drug Administration and Food and Drug Administration
|
|
Address:
|
CDER, 5600 Fishers Lane, Rockville, MD, 20857-0001,
|
|
Keywords:
|
covariate adjustment ; flexible design ; adaptive trial design
|
|
Abstract:
|
Covariate adjustment is widely considered in analysis of randomized clinical trials, particularly for improving statistical precision in estimation of a treatment effect and statistical power for detecting the effect. To avoid data dredging in the final analysis stage, a common recommendation is pre-specification of the covariates and the forms (e.g., original forms or some transformations of the covariates) of the selected covariates for adjustment in the study protocol. Such pre-specification helps alleviate the problem of multiple testing and minimize selection bias. In ANCOVA, the best function of a selected covariate for adjustment is the conditional expectation of the response variable given the covariate. This function is usually unknown. In this talk, we shall explore the impact of using an estimated transformed covariate for adjustment on the treatment effect estimate. We will then explore an adaptive design strategy in which one employs accumulated data at an interim look with treatment codes kept blinded to search for a proper form of covariate to be selected as the transformed covariate in the final analysis.
|
