Frequent Cognitive Tests Increase Power for Alzheimer’s Disease Clinical Trials

Abstract

BackgroundMany clinical trials use performance‐based cognitive or functional test results as primary outcome measures. As few as two measurements, typically pre‐ and post‐intervention, can be used to define a cognitive trajectory. Increasing the number and frequency of assessments can increase power to test the primary hypothesis. However, increased assessment frequency may be expensive and burdensome to participants and staff. Frequent neuropsychological testing may also introduce a practice effect or learning effect. Trials should be designed with attention to the costs and benefits of more frequent assessments.MethodWe conducted two clinical trials and performed Monte Carlo simulations. The Montreal Cognitive Assessment (MoCA) and the MCI Screen (MCIS) were used to assess cognition; the Functional Assessment Staging Tool (FAST) was used to assess function. In addition to simulated data, we employed data from the Coaching for Cognition in Alzheimer’s (COCOA) and Precision Recommendations for Environmental Variables, Exercise, Nutrition and Training Interventions to Optimize Neurocognition (PREVENTION) Trials.ResultEffect sizes and measurement variability were similar between the two cognitive screens. More frequent assessments increased power to reject a null hypothesis The major factor driving this improved power was the influence of data from individuals withdrawing from or dropping out of the trial prior to trial end. More frequent testing captured more of the effect of the intervention over time in those individuals, and therefore, increased power for the entire study. The design of the MCIS with rotating word lists facilitated frequent testing without a learning effect. Hypothetically, a more frequently administered assessment might have less precision than a different assessment administered less frequently, and the unit costs of these assessments might differ. Under a wide range of parameters modeled in simulations, the advantages of more frequent testing outweighed decreases in the precision of individual assessments.ConclusionMore frequent testing should be preferred in clinical trials. Use of shorter and/or remote tests may facilitate such frequency, reduce costs, and increase safety during pandemics. In addition to gains in statistical power, dense measurements have other benefits. They may also reveal non‐linear trajectories and better enable longitudinal analyses such as time dependencies and dose‐response relationships.