|
Activity Number:
|
202
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #310056 |
|
Title:
|
Bayesian Subgroup Analysis in Clinical Trials
|
|
Author(s):
|
Claudia Pedroza*+
|
|
Companies:
|
The University of Texas Health Science Center at Houston
|
|
Address:
|
1200 Herman Pressler Dr, Houston, TX, 77035,
|
|
Keywords:
|
bayesian analysis ; subgroup analysis ; clinical trials ; center effects ; multiple comparisons
|
|
Abstract:
|
Clinical trials test for an overall treatment effect in the target population. However, clinical investigators are often interested in estimates of treatment effect for various subgroups. Typically these trials are not designed nor powered to conduct subgroup comparisons and this type of analysis raises the concern of multiplicity effects. Here, we review the common practices in dealing with subgroup analysis in clinical trials. We then assess the performance of a Bayesian approach to subgroup analysis which uses hierarchical models with random effects. We use simulations to calculate the rate of false-positive/negative results under various scenarios. In particular, we investigate how well the Bayesian method performs in data sets with small total sample size as well as with large number of subgroups with small sample sizes.
|
