Statistical analysis, trial design and duration in Alzheimer's disease clinical trials: a review

Paul Thompson, David Wright, Carl Edward Counsell, John P Zajicek

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

BACKGROUND: The social and economic burden of Alzheimer's disease (AD) and its increasing prevalence has led to much work on new treatment strategies and clinical trials. The search for surrogate markers of disease progression continues but traditional parallel group trial designs that use well-established, but often insensitive, clinical outcome measures predominate.

METHODS: We performed a systematic search across the Cochrane Library and PubMed abstracts published between January 2004 and August 2009. Information regarding the clinical trial methodology, outcome measures, intervention type and primary statistical analysis techniques was extracted and categorized, according to a standard protocol.

RESULTS: We identified 149 papers describing results from clinical trials in AD containing sufficient detail for our purposes. The largest proportion (38%) presented results of trials based on tests of cognition as the primary outcome measure. The primary analysis in most papers (85%) was a univariate significance test of a single primary outcome measure.

CONCLUSIONS: The majority of trials reported a comparison of baseline and end-point assessment over relatively short patient follow-up periods, using univariate statistical methods to compare differences between intervention and control groups in the primary analysis. There is considerable scope to introduce newer statistical methods and trial designs in treatment evaluations in AD.

Original languageEnglish
Pages (from-to)689-697
Number of pages9
JournalInternational Psychogeriatrics
Volume24
Issue number5
Early online date13 Sep 2011
DOIs
Publication statusPublished - May 2012

    Fingerprint

Keywords

  • Alzheimer's disease
  • clinical trials
  • statistical analysis
  • surrogate markers
  • outcome measures

Cite this