|
Activity Number:
|
541
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #307049 |
|
Title:
|
A Multivariate Truncated Model Combined with Multiple Imputation for Longitudinal Data with Nonignorable Missing
|
|
Author(s):
|
Rong Liu*+ and Viswanathan Ramakrishnan
|
|
Companies:
|
Merck & Co., Inc. and Virginia Commonwealth University
|
|
Address:
|
770 Sumneytown Pike, West Point, PA, 19468,
|
|
Keywords:
|
multivariate truncated normal ; multiple imputation ; treatment related dropout
|
|
Abstract:
|
In longitudinal clinical trials, the data sets often are incomplete. In some cases, patients drop out due to treatment-related reasons, which lead the distribution of the observed data to resemble a truncated normal distribution. Ramakrishnan and Wang (2005) proposed a method under a truncated multivariate normal distribution to analyze such data. Although majority of multiple imputation (MI) procedures involve the use of ignorable missing models, MI could also be used for nonignorable missing data. A MI procedure under the MDT method to accommodate the uncertainty in imputation is proposed. The combination of MDT method with Rubin's multiple imputation will be presented. A data set will be used to illustrate the application of the presented method.
|
