Minimally Clinically Important Differences

Anchor-based MCID estimation and its sample-size implications for neurodegenerative disease trials

2026-06-07 18:16 PDT

Overview

A statistically significant treatment effect is not automatically clinically meaningful. The minimally clinically important difference (MCID) – the smallest change in a patient-reported or clinician-rated outcome that a patient would recognize as important – provides the bridge between statistical inference and clinical relevance, and directly determines the sample sizes required for definitive trials.

This program develops anchor-based MCID estimates for rating scales used as primary endpoints in Alzheimer’s disease and related neurodegenerative conditions, derives the implied sample-size requirements, and examines the sensitivity of those requirements to MCID definition and covariance structure.

Current work

  • CDR-SB MCID and sample size for AD trials. Anchor-based estimation of the MCID for the Clinical Dementia Rating Sum of Boxes (CDR-SB), the primary endpoint for a growing number of confirmatory AD trials (including the lecanemab and donanemab programs). Analysis derives CDR-SB trajectories from registry data, estimates the MCID via multiple anchor methods, and produces power and sample-size curves for trial planning.

  • PSPRS subscale MCID (published). Anchor-based MCID estimation for the PSPRS-15 and PSPRS-10 subscales of the Progressive Supranuclear Palsy Rating Scale, extending previous PSPRS-28 MCID work. Published in Alzheimer’s & Dementia: Translational Research & Clinical Interventions (TRCI), 2026.

  • Empirical variance-covariance matrices for simulation. Systematic extraction of variance-covariance matrices from completed ADCS trials, providing realistic covariance inputs for sample-size simulation across the broader research program.

Methods

Anchor-based MCID estimation (distribution-based, anchor-based with global ratings of change); CDR trajectory modeling from multi-cohort registry data; power and sample-size curves via pwr and custom simulation; zzlongplot for CDR visualization; APA-format citation and reproducible Rmd rendering.

Selected publication

Golbe, L. I., & Thomas, R. G. (2026). Minimally clinically important differences for PSPRS subscales in progressive supranuclear palsy. Alzheimer’s & Dementia: Translational Research & Clinical Interventions. [Published]

Publications

Additional work on MCID methodology and related sample-size methods is available through the full publications list by filtering on sample-size or alzheimers-disease.