|
Activity Number:
|
409
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #308905 |
|
Title:
|
A Robust Bayesian Analysis of a Crossover Trial in Smoking Cessation
|
|
Author(s):
|
Pulak Ghosh*+ and Mary Putt
|
|
Companies:
|
Georgia State University and University of Pennsylvania
|
|
Address:
|
30 Pryor Street, Atlanta, GA, 30303,
|
|
Keywords:
|
bayesian ; crossover design ; Dirichlet Process prior ; Gibbs sampling ; smoking cessation
|
|
Abstract:
|
In a smoking cessation crossover trial, chronic smokers sequentially received placebo and naltrexone. Naltrexone is an opioid antagonist expected to reduce the reward value of nicotine, thus reducing motivation to smoke. In addition to an overall effect, differences in the responses of males and females to naltrexone were of interest. The distribution of two outcome variables reflecting positive and negative reinforcement from smoking displayed substantial evidence of departures from normality. Here we implement a Bayesian approach for estimating the effect of naltrexone on these outcomes. Our analysis is novel in the crossover literature in accounting for non-normality of the data through the use of a mixture of Dirchlet process priors on the model parameters. The approach is illustrated for each of the individual variables as well as in a multivariate framework.
|
