Handling Missing Data in Rheumatoid Arthritis Trials

• Activity Number: 344
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Guoguang Ma*+ and Liyun Ni and Ling Chen
• Companies: Amgen Inc. and Amgen Inc. and Washington University School of Medicine
• Keywords: Longitudinal clinical trial ; Dropout ; Ignorability ; Index of sensitivity to nonignorability ; Partial imputation

Abstract

In longitudinal clinical trials, the comparison of treatment effect is often complicated by missing data caused by dropout. In this talk, an Index of Sensitivity to Nonignorability, or ISNI (Troxel et al. 2004) is applied as a first-order sensitivity analysis to measure the extent to which small departures from ignorability affect the maximum likelihood estimates. We also evaluate a partial imputation method proposed by Wei and Shih (2001) based on which some of the missing values are imputed using the last observation carried forward approach to balance the dropout distributions between the treatment groups. The method is shown to reduce the bias in the comparison of the treatment effect in the presence of nonignorable missing data. We extend their imputation method to be model-based. Simulation and real data are used to assess and illustrate the statistical properties of the method.