|
Activity Number:
|
261
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #304920 |
|
Title:
|
A Cure Mixture Model for Multivariate Time-to-Event Data
|
|
Author(s):
|
E. Paul Wileyto*+
|
|
Companies:
|
University of Pennsylvania
|
|
Address:
|
School of Medicine, Psychiatry, Tobacco Use Research Center, Philadelphia, PA, 19104-3309,
|
|
Keywords:
|
cure-mixture models ; recurrent events ; multivariate survival ; behavioral medicine ; smoking cessation
|
|
Abstract:
|
The attempt to quit is, for most smokers, a series of alternating states, terminated by transitions that are either a return to smoking or recovery of abstinence. Ordinary recurrent failure time modeling leads to poor fit and biased estimates of treatment effects because long-term survivors may arise at each small step. We introduce a cure-mixture regression model for multivariate or recurrent failure time data based on Farewell's 1982 model, which used logistic regression to predict probability of membership in cured and noncured classes and Weibull regression to predict time to event in the noncured class. Parameter estimates are obtained using maximum likelihood, with standard errors adjusted to account for repeated measures using the cluster-correlated robust variance estimate. The model provides an excellent fit to data from a smoking cessation clinical trial.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
