|
Abstract:
|
The sequential parallel comparison design (SPCD), with sequence groups P:P, P:T, and T:T in which information from placebo responders in the second period are excluded, serves as an appealing design for studies with high placebo response, for example, psychiatric clinical trials (Fava et al. 2003). This presentation discusses methodology to study sources of comparison in the traditional SPCD design, i.e., the first period treatment difference in the overall population and the second period treatment difference in the placebo non-responders, as well as other sources of information that are available in this crossover design which could be of potential interest. We propose a nonparametric method for the point and confidence interval estimation of the correlated measurements. Randomization-based ANCOVA is introduced to adjust for baseline imbalance. Simulation studies are performed to study the statistical properties of the proposed methods, and their statistical properties are compared to those of the repeated measures model approach proposed by Doros et al. (2013). Potential adaptations to the proposed design to tailor to the needs of other therapeutic areas would also be addressed.
|
