The Evaluation of the Sequential Parallel Design: A Design Strategy to Deal with High Placebo Response in Psychiatric Clinical Trials

• Activity Number: 37
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yang Yang*+ and Yeh-Fong Chen and H.M. James Hung and Sue-Jane Wang
• Companies: U.S. Food and Drug Administration and U.S. Food and Drug Administration and U.S. Food and Drug Administration and U.S. Food and Drug Administration/CDER
• Keywords: Placebo Response ; Sequential Parallel Design ; Psychiatric Clinical Trials ; Missing Data and Imputation Methods

Abstract

Fava et al. (2003) proposed the sequential parallel design (SPD) to deal with high placebo response in psychiatric clinical trials. Since the applicability of such a novel design to confirmatory trials is uncertain, it is important to study and understand the design as well as the relevant statistical analyses. In this work, we evaluate the applicability of SPD with continuous primary endpoint through its performance characteristics. We conduct simulation studies to demonstrate the need of re-randomization at the second stage of SPD. We analytically show that, when data are missing completely at random, the weighted test statistic based on simple ordinary least square estimates is a viable alternative to the weighted test statistic based on estimates from seemingly unrelated regression proposed by Tamura and Huang (2007). We also compare three imputation methods (LOCF, multiple imputation, and MMRM) by simulation and find that, when data are missing at random and dropout rate is moderate, the weighted test statistic based on MMRM estimates appears to be the most robust test statistic under SPD in terms of type I error control, power performance, and estimation accuracy.