|
Activity Number:
|
541
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #307130 |
|
Title:
|
Jointly Modeling the Dropout Process and Repeated Categorical Outcome
|
|
Author(s):
|
Yuting Zhang*+ and Brent J. Shelton
|
|
Companies:
|
MedFocus and University of Kentucky
|
|
Address:
|
923 Shell Blvd., Foster City, CA, 94404,
|
|
Keywords:
|
missing data ; EM algorithm ; repeated categorical outcome
|
|
Abstract:
|
For clinical trials with multiple endpoints, analyzing categorical response data measured over time is a big challenge due to possible survival event or lost to follow-up. We proposed a pattern-mixture approach to jointly model the dropout process and the repeated categorical response. Given the distribution of dropout process, the missing data can be treated as missing at random for the categorical outcomes. The estimates are obtained by the method of the weighted Expectation-Maximization (EM) algorithm. The proposed approach is illustrated by an example. The robustness and validity of this approach is investigated by sensitivity analyses
|
