|
Activity Number:
|
162
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301834 |
|
Title:
|
Predicted and Conditional Marginals for Cox's Proportional Hazards Model Using SUDAAN
|
|
Author(s):
|
Elvessa Aragon-Logan*+ and G. Gordon Brown and Beth G. Barnwell and Babubhai V. Shah
|
|
Companies:
|
RTI International and RTI International and RTI International and SAFAL Institute Inc.
|
|
Address:
|
3040 Cornwallis Rd., RTP, NC, 27709,
|
|
Keywords:
|
survival ; SUDAAN ; adjusted treatment means ; predictedmMargins ; direct standardization
|
|
Abstract:
|
Obtaining adjusted treatment means for regression models is often of interest to the researcher. The primary goal of our work is to produce functional forms for two types of adjusted treatment means--predicted marginals and conditional marginals--for Cox's proportional hazards model for cluster-correlated data with tied event times. We consider two methods, Breslow's and Efron's, that attempt to adjust the estimation from Cox's model in situations where ties occur. The predicted and conditional marginals are presented for both of these methods. The point estimates for the two types of marginals are derived using pseudo-likelihood equations, and variance estimates are determined using Taylorized deviations. The variance estimates derived from the Taylorized deviations tend to be robust for cluster-correlated data. We will present analytic results comparing the variance estimates using the Taylorized deviations and the model-based (naïve) variance estimates. All calculations for the analytic examples are derived using new features in SUDAAN's SURVIVAL procedure.
|
