Power and sample size determination when assessing the clinical relevance of trial results by 'responder analyses'

Stat Med. 2004 Nov 15;23(21):3287-305. doi: 10.1002/sim.1910.


A fundamental issue in regulatory decision making is the assessment of the benefit/risk profile of a compound. In order to do this, establishing the existence of a treatment effect by a significance test is not sufficient, but the clinical relevance of a potential benefit must also be taken into account. A number of regulatory guidelines propose that clinical relevance should be assessed by considering the rate of responders, i.e. the proportion of patients who are observed to achieve an apparently meaningful benefit. In this paper, we present methods for planning clinical trials that aim at demonstrating both statistical and clinical significance in superiority trials. Procedures based on analytical calculations are derived for normally distributed data and the case of a single endpoint as well as multiple primary outcomes. A bootstrap procedure is proposed that can be applied to non-normal data. Application is illustrated by a clinical trial in Alzheimer's disease.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Activities of Daily Living
  • Alzheimer Disease / drug therapy
  • Alzheimer Disease / pathology
  • Clinical Trials as Topic / methods*
  • Cognition / drug effects
  • Cysteine / therapeutic use
  • Drug Combinations
  • Humans
  • Pantothenic Acid / therapeutic use
  • Risk Assessment
  • Sample Size*
  • Treatment Outcome*


  • Drug Combinations
  • calcium D-pantothenate, L-cysteine drug combination
  • Pantothenic Acid
  • Cysteine