An Overview of Statistical and Operational Considerations Related to the Shared Control in Platform Trials
摘要
Traditional, two arm randomized controlled trials are the gold standard for evaluating the efficacy of experimental interventions, but the need for flexible and efficient trial designs has increased. Trial designs using a master protocol, such as the platform trial, have been proposed as an option. Trials governed by a master protocol utilize an overarching study infrastructure to answer multiple research questions. Platform trials test multiple interventions against a common control arm and allow arms to enter and exit the study, increasing efficiency in evaluating the efficacy of experimental interventions. This increased efficiency, however, comes with unique considerations from both statistical and operational perspectives in the design, conduct, and analysis of platform trials. In this manuscript we discuss some key considerations for platform trials in general and specific challenges related to the shared control arm. Some of these challenges are related to fundamental trial features, including randomization and blinding. In a platform trial, participants are often randomized in a two-stage fashion to reduce operational complexities and more similarly mirror the process of randomization within a traditional two-arm randomized controlled trial. This approach to randomization, along with the chosen method for blinding within the trial, can have important implications when considering the ability to pool together control participants into a shared control arm for analyses. Overall, the platform trial requires unique considerations for its implementation, and users should be prepared to address these in order to realize the potential benefits and efficiencies of this trial design.