Selection and Evaluation of Weighted Composite Outcome Measures for Clinical Trials in Alzheimer's Disease

• Activity Number: 3
• Type: Invited
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): M. Colin Ard* and Steven D. Edland
• Companies: UC San Diego and UC San Diego
• Keywords: Alzheimer's ; composite scale ; clinical trial ; sample size ; power ; mild cognitive impairment

Abstract

The development of efficient composite outcome measures for use in longitudinal clinical trials is an active area of research. For example, novel composites constructed as weighted sums of individual item, subtest, and/or test scores have been proposed as primary endpoints for Alzheimer clinical trials (Donohue et al., 2014; Langbaum et al., 2014). The statistical power of composite endpoints depends not only on the chosen set of component tests and items, but also on the weights that determine the relative contribution each component makes to the final composite. A poor choice of weights can produce a composite that is actually less efficient than one or more of the component scales from which it is calculated. We outline important considerations in the selection of weighting algorithms for determination of efficient composites, identify some of the challenges involved in determining weights from pilot data sets, and propose methodological strategies for evaluating and optimizing the prospective efficiency of composites when the true parameters of the longitudinal data distribution are unknown.