|
Activity Number:
|
405
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #301273 |
|
Title:
|
An Application of Multiple Partial Imputation to Analysis of Longitudinal Quality-of-Life Data
|
|
Author(s):
|
Paul Kolm*+ and Wei Zhang and John Spertus and William Weintraub
|
|
Companies:
|
Christiana Care Health System and Christiana Care Health System and Mid American Heart Institute and Christiana Care Health System
|
|
Address:
|
131 Continental Drive, Newark, DE, 19713,
|
|
Keywords:
|
Missing data ; Longitudinal data analysis ; Multiple partial imputation
|
|
Abstract:
|
Missing data present a challenge for analysis, particularly when the data are missing not at random (MNAR). In clinical trials where longitudinal quality of life and health status data are obtained from patient self-report, missing observations may be MNAR, in that patients in poorer health may be reluctant to complete a questionnaire regarding their health. A multiple partial imputation (MPI) strategy has been suggested that assumes intermittent missing data are missing at random (MAR), but missing data due to dropout are MNAR (nonignorable). The imputed data sets are then analyzed by likelihood or quasi-likelihood methods that account for nonignorable dropout. In this study, MPI is applied to quality of life data obtained in a large clinical trial spanning a 7 year follow-up period. Issues relating to the assumptions and application of MPI are discussed.
|
