Premium
Commentary on “Meta‐analysis of six‐month memantine trials in Alzheimer's disease.” Memantine has negligible benefits in mild to moderate Alzheimer's disease
Author(s) -
Knopman David S.
Publication year - 2007
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2006.10.002
Subject(s) - memantine , disease , neurology , medicine , psychology , psychiatry , family medicine , gerontology , library science , dementia , computer science
Meta-analyses of clinical trials are intended to summarize the results of individual studies that might have used diverse populations and methods. Meta-analyses are thought of as a way of reducing bias originating in individual clinical trials. Implicit in the idea of meta-analyses is equipoise. The meta-analysis of clinical trials of memantine in Alzheimer’s disease (AD) by Doody et al breaks new ground in being substantially more biased than the individual trials on which it is based. Given the publication track record of the industry sponsor, it seems likely that this analysis would never have been submitted for publication if the result had been negative. Moreover, the academic authors acknowledged that they received compensation for preparing this article. Although I applaud the authors for disclosing this relationship, an analysis paid for by a drug’s sponsor does not pass the obvious test of equipoise. The article by Doody et al dealt with trials in both moderate to severe and mild to moderate AD. The more interesting story has to do with the data presented from the trials in mild to moderate AD patients. The attainment of a P value .05 in the combined analyses for what were the primary outcome measures in the individual trials in the mild to moderate patients can be accounted for by the very large number of subjects (more than 1300) when the three mild to moderate AD patient trials were combined.