SEARCH

SEARCH BY CITATION

Keywords:

  • disease progression modelling;
  • Parkinson's disease;
  • statistical power;
  • delayed start design;
  • washout design;
  • Alzheimer's disease

Abstract

Treatment of a slowly progressive disease, such as Parkinson's and Alzheimer's disease should slow down or even reverse progression of the disease. For estimating the actual treatment effect a clinical trial should allow the time course of treatment effects to be distinguished from those due to natural disease progression. A symptomatic treatment effect has a rapid onset, which will disappear after treatment cessation, whereas disease modifying effects are slow and persistent. Our objective was to compare the power of washout and delayed start designs for distinguishing symptomatic and disease modifying effects using disease progression modelling. The effect of dropout due to worsening disease status on the power was also considered.

The change in the Unified Parkinson's Disease Rating Scale score over time was simulated, assuming that a treatment has a combined symptomatic and disease modifying effect. The simulated data were then fitted with the combined effect model and also with alternative models with either symptomatic or disease modifying effects alone. The power was derived from the number of times that the true model was correctly distinguished from the alternative models.

Including a washout period in which the subjects are followed up after cessation of treatment substantially increases the power to distinguish different treatment effect types (washout>80%; delayed start<60%). Dropout of 40%–50% during the trial has a small effect on power (7% decrease). Under the assumptions of the simulation we found a clear advantage from using the washout design compared with a delayed start design. Copyright © 2008 John Wiley & Sons, Ltd.