A Hybrid Approach of Handling Missing Data That Combined Nonresponder Imputation with Multiple Imputation or Pattern Mixture Model

• Activity Number: 646
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Bidan Huang*+ and Yiran Bonnie Hu and Lei Shu and Qian Zhou
• Companies: AbbVie and AbbVie and AbbVie and AbbVie
• Keywords: Missing Data ; Randomized Clinical Trial ; Multiple Imputation ; Pattern Mixture Model

Abstract

In some randomized double-blind clinical trials (RCT), subjects who had disease flare or non-response to the study drugs are switched to open-label (OL) or put on rescue medication. Data from those subjects are regarded as "missing" at the end of the trial for endpoint evaluation, along with the subjects who prematurely discontinued from the study for various other reasons. A hybrid approach was proposed to handle these missing data differently. For those subjects who switched to OL or rescue medication per the study design, non-responder imputation (NRI) is appropriately applied. However, for the subjects who prematurely discontinued from the study, multiple imputation (MI) or pattern mixture model (PMM) is performed.