Activity Number:
|
214
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biopharmaceutical Section*
|
Abstract - #301416 |
Title:
|
A Test for Proportional Hazards in Stratified Time-to-Event Models
|
Author(s):
|
Lynn Eberly*+ and Christopher Mullin and Patricia Grambsch
|
Affiliation(s):
|
University of Minnesota and University of Minnesota and University of Minnesota
|
Address:
|
420 Delaware Stree SE, MMC 303, Minneapolis, Minnesota, 55455, USA
|
Keywords:
|
Cox regression ; failure time ; survival analysis ; information matrix ; Schoenfeld residuals ; profile likelihood
|
Abstract:
|
Most tests of proportional hazards that have been proposed are special forms of the general test developed by Grambsch and Therneau (1994), which tests whether the regression coefficient value changes over time. The computed test statistic depends on an information matrix estimated by averaging across all the observed event times. This averaging, however, may be inappropriate for stratified models when covariate distributions vary across the stratum-specific risk sets. We examine here the performance of a test based on combining the information matrix across all observed event times within each stratum, leading to a test that is appropriate for stratified models. This new test is examined with simulations and a case study.
|