Random Effects Design for Multiregional Trials

• Activity Number: 427
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Fei Chen*+ and Gordon Lan and Jose Carlos Pinheiro
• Companies: Janssen Research & Development and Janssen Pharmaceutical Companies of Johnson & Johnson and Janssen R&D
• Keywords: discrete random effects model ; multi-regional clinical trial design ; meta-analysis

Abstract

Design of multi-regional trials has played an increasingly important role in clinical trials. Traditionally, a multi-regional trial would assume a fixed effects model, but its validity has been challenged on various fronts, with many interesting and insightful articles have been published recently. There are two different random effects models suggested for such studies: (i) Level 1 Discrete Random Effects Model (DREM), and (ii) Level 2 Normal random Effects Model (NREM). NREM was originally introduced by DerSimonian and Laird in 1986 for meta-analysis of clinical trials. We will use numerical examples to illustrate the difference between these two models, and make some practical suggestions for the design under DREM.