WITHDRAWN: Comparison of Imputation Methods for the Delayed Start Design in Studies of Alzheimer's Disease

• Author(s): Kathryn Dawson
• Keywords: Delayed start design ; Alzheimer's ; Imputation ; missing data

Abstract

In studies of Alzheimer's disease the Delayed Start Study design is a preferred method (D'Agostino, NEJM, 2010). This method is used to detect differences in cognitive outcomes over time that can be attributed to either symptom improvement or disease modifying effects. This type of study may run over a long period of time where outcomes are measured periodically. It is not uncommon in these studies to have drop out rates as high as 25%. In order to adequately test for treatment effects the causes for the missing data should be comsidered and imputation methods appropriatedly applied. Several imputation methods including multiple imputation and conditional mean imputation can be applied. In this abstract simulation studies will compare the performance of possible imputation methods under various assumptions.