Statistical approaches to effectiveness measurement and outcome-driven re-randomizations in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) studies

Abstract

The design of the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia and Alzheimer's disease studies pose several statistical challenges, including issues related to performing multiple comparisons, defining effectiveness outcomes, and collecting and analyzing data from a design with multiple outcome-driven re-randomizations. We discuss the CATIE strategy for addressing many hypotheses within the context of one clinical trial while controlling the overall type I error rate. We provide motivation for the use of two effectiveness outcomes: time to all-cause discontinuation and composite endpoints that combine outcomes from multiple domains, such as efficacy, safety, cost-effectiveness, and quality of life. Methods for statistical analysis of an outcome-driven re-randomization trial are compared and evaluated. We describe analysis within each phase, analysis based on the first randomization or treatment algorithms, and repeated measures modeling. Finally, strategies are described for designing an electronic data collection system for trials with repeated outcome-driven re-randomizations.