Comparison of Sequential and Multistage Designs for Selecting Promising Therapeutic Modalities for Stroke in Pre-clinical Experiments

• Activity Number: 121
• Type: Contributed
• Date/Time: Monday, August 9, 2004 : 12:00 PM to 1:50 PM
• Sponsor: Biometrics Section
• Author(s): Paula K. Roberson*+ and William C. Culp
• Companies: University of Arkansas for Medical Sciences and University of Arkansas for Medical Sciences
• Keywords: sequential methods ; multistage designs ; bias ; pre-clinical experiments

Abstract

During preclinical development, it may be important to screen a number of competing alternative therapies simultaneously, identifying a small subset of sufficient promise for further intensive and costly development. In the case of binary outcomes, the goal is to maximize the likelihood of identifying procedures with high success probabilities while minimizing the chances of continuing investment in procedures with low success rates. For cost efficiencies, it is desirable to minimize the number of observations required to identify inferior therapies, while concurrently obtaining reasonable estimates of therapeutic success in study arms deemed adequate, to aid planning of future experiments. This poster will detail our development of such a plan as part of a developmental program of stroke therapy. Expanding on randomized phase II clinical trials designs, the experiments were designed as randomized concurrent single arm studies, rather than a single comparative study. Results of a simulation study to assess Type I/II errors, as well as estimation bias and variability, of sequential and multistage designs will be presented, including rationale for the design ultimately chosen.