|
Activity Number:
|
472
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #300275 |
|
Title:
|
An Enhanced Estimate of Treatment Effect Based on Joint Models
|
|
Author(s):
|
Weili He*+ and Weichung J. Shih
|
|
Companies:
|
Merck Research Laboratories and University of Medicine and Dentistry of New Jersey
|
|
Address:
|
PO Box 2000, Rahway, NJ, 07065-0900,
|
|
Keywords:
|
joint models ; clinical endpoint ; surrogate endpoint ; Markov chain Monte Carlo
|
|
Abstract:
|
Many clinical and epidemiologic studies often collect longitudinal measurements of a disease marker and time-to-event of a disease. Lately many approaches have been proposed to jointly model these two types of data. However, if the chief purpose of the joint model is to evaluate the treatment effect on the risk of developing a certain clinical disease and if the longitudinal measurement may in fact be a perfect surrogate endpoint for the clinical endpoint, then modeling the two types of data jointly may actually obscure the estimate of the treatment effect on the disease risk. We propose a joint model approach which enables the estimate of "an overall treatment effect" which combines the contributions from both the clinical endpoint and the surrogate endpoint. Markov chain Monte Carlo techniques are used to obtain the parameter estimates from the joint model. Extensive simulation studies indicate that the proposed approach results in an enhanced estimate of the overall treatment effect even in the case where the longitudinal measurement is only a partial surrogate endpoint. Data application to the Proscar Long-Term Efficacy and Safety Study also yields the similar conclusion.
|
