Clinical Trial Design and Power Analysis

Efficient design of randomized trials in neurology, psychiatry, and chronic disease

2026-06-07 18:16 PDT

Overview

Designing a clinical trial requires balancing statistical rigor against operational constraints: visit frequency, recruitment timelines, early stopping rules, and covariate structure all interact with power in ways that are rarely captured by simple two-sample formulas. This program develops and evaluates design methods for complex randomized trials, with particular emphasis on longitudinal endpoints, run-in periods, and multi-center settings common in Alzheimer’s disease and PTSD research.

Work spans analytic power derivations, Monte Carlo simulation, and applied design consulting on trials conducted through the Alzheimer’s Disease Cooperative Study (ADCS) and the VA Cooperative Studies Program (CSP).

Current work

  • Adaptive allocation with survival endpoints. Evaluating response-adaptive randomization when the primary endpoint is time-to-event, including comparison with fixed allocation under MMRM-based analyses.

  • Adaptive allocation with MMRM. Extending adaptive randomization methodology to repeated-measures continuous endpoints, the dominant endpoint structure in Alzheimer’s and PTSD trials.

  • Run-in period design. Characterizing the power and bias trade-offs of placebo run-in periods in trials with non-normal baseline distributions.

  • Optimal visit placement. Deriving visit schedules that maximize power for a fixed number of assessment occasions under realistic dropout and attrition models.

  • Multicenter RCT methods. Accounting for site-level clustering and heterogeneity in multicenter randomized designs.

  • Conditional power and interim analysis. Methods for re-estimation of sample size at interim looks, with application to rare-event and chronic-disease settings.

  • Longitudinal power. General power formulas for longitudinal repeated-measures designs accommodating arbitrary covariance structures.

Methods

Mixed-effects models for repeated measures (MMRM); Monte Carlo simulation under the ADEMP framework; analytic power derivations for linear and survival endpoints; adaptive randomization algorithms; covariance-structure estimation from historical ADCS trial data.

Software

  • zzpower – Interactive power analysis calculator supporting a range of clinical-trial designs.
  • nof1power – Power analysis and simulation for N-of-1 and parallel-group designs.
  • zzcollab – Reproducible research framework used for all simulation compendium projects.

Publications

Selected publications from this program are accessible through the full publications list, filtering by the keywords sample-size, clinical-trials, adaptive-designs, or longitudinal.

Key areas represented in the publication record include power analysis for MMRM endpoints, adaptive allocation theory, and design methods for multi-site Alzheimer’s disease trials conducted through ADCS.