|
Activity Number:
|
246
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #308994 |
|
Title:
|
Generalized Estimating Equations for Clustered Survival Data
|
|
Author(s):
|
Xiaohong Zhang*+ and Kenneth Koehler and Terry M. Therneau
|
|
Companies:
|
Amgen Inc. and Iowa State University and Mayo Clinic
|
|
Address:
|
One Amgen center drive, Thousand Oaks, CA, 91320,
|
|
Keywords:
|
clustered survival data ; generalized estimating equations ; Cox model ; efficiency
|
|
Abstract:
|
The analysis of clustered survival data is an important statistical question. A commonly used method is to obtain the estimates of the regression parameters under the assumption of independence. A robust variance estimate is then computed which accounts for correlation. The main deficiency of the method is the loss of efficiency when the within cluster correlation is strong. We propose generalized estimating equations (GEEs) to improve the efficiency in estimating regression coefficients in the Cox model without imposing an overwhelming computational burden. Simulation was used to assess bias, variance and relative efficiency of the proposed estimators. GEEs may provide substantial gains in efficiency if within cluster correlation is sufficiently strong and the censoring rate is low. Gains in efficiency also depend on the level of homogeneity of treatment assignments within clusters.
|
