|
Activity Number:
|
28
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #309167 |
|
Title:
|
Multiple Imputation of the Euroqol EQ-5D: Testing Different Approaches Using IVEWare
|
|
Author(s):
|
Robert Gerzoff*+ and Theodore J. Thompson
|
|
Companies:
|
Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
|
|
Address:
|
MS K10, Atlanta, GA, 30341,
|
|
Keywords:
|
Imputation ; Diabetes ; Managed Care
|
|
Abstract:
|
The EQ-5D is a standardized and widely used health index consisting of five items each with three levels. Different response profiles are converted to a single index value using established valuation weights. When individual items are missing the analyst must decide whether to impute the EQ-5D directly, or impute the missing item(s) and recalculate the EQ-5D. Decisions must also be made about the distributional assumptions underlying the missing data and the alternative methods available to model the missingness. Using data from the TRIAD Study, a multicenter study of diabetic 11,927 patients enrolled in ten U.S. managed care plans in seven states, we simulate missing data patterns and use IVEW are to evaluate the performance of various imputation strategies.
|
