Sample Size Determination in Clinical Trials with Multiple Co-primary Endpoints
- 2015-12-08 (Tue.), 10:30 AM
- Recreation Hall, 2F, Institute of Statistical Science
- Prof. Takashi Sozu
- Tokyo University of Science, Faculty of Engineering, Tokyo, Japan
Abstract
The determination of sample size and the evaluation of power are fundamental and critical elements in the design of a clinical trial. Most commonly, a single endpoint is selected and then used as the basis for the trial design including sample size determination, interim data monitoring, and final analyses. However, many recent clinical trials have utilized more than one primary endpoint. The rationale for this is that use of a single endpoint may not provide a comprehensive picture of the intervention’s multidimensional effects. For these reasons, use of more than one primary endpoint, i.e., multiple endpoints, has become a common design feature in clinical trials. For example, multiple endpoints are utilized as “co-primary” or “multiple primary” to evaluate the effects of the new interventions for the treatment of Alzheimer disease, irritable bowel syndrome, acute heart failure, and diabetes mellitus. “Co-primary” in this setting means that the trial is designed to evaluate if the intervention is superior to the control on all of the endpoints. In contrast, a trial with “multiple primary” endpoints is designed to evaluate if the intervention is superior to the control on at least one of the endpoints. In such clinical trials, the correlation among the multiple endpoints should be considered in order to obtain an appropriate sample size. ??? ?? In this presentation, we provide an overview of the concepts and the technical fundamentals regarding recent methodological developments for power and sample size calculations for comparative clinical trials with multiple co-primary endpoints. We also provide a convenient sample size formula with associated numerical tables. Finally, we briefly discuss developments for designing clinical trials with other design characteristics including: (i) other inferential goals, (ii) more than two intervention groups, (iii) group sequential designs, and (iv) endpoints with other measurement scales.