brand logo

Am Fam Physician. 2004;69(2):393-394

Clinical Question: Are cholinesterase inhibitors safe and effective in patients with Alzheimer's dementia?

Setting: Various (meta-analysis)

Study Design: Meta-analysis (randomized controlled trials)

Synopsis: A previous meta-analysis of functional and behavior outcomes found little benefit with use of cholinesterase inhibitors in patients with Alzheimer's dementia (JAMA 2003;289:210–6). In the current study, the primary outcome was “global response,” which was defined as minimal improvement or better as evaluated by clinicians (Clinical Global Impression of Change scale) or clinicians and caregivers (Clinician's Interview-Based Impression of Change Plus Caregiver Input scale). A secondary outcome was “cognitive response,” defined as an improvement of 4 points or more on the Alzheimer's Disease Assessment Scale–Cognitive subscale.

The authors identified 40 studies but excluded 24 because of methodologic problems, leaving 16 studies with more than 7,800 patients for the final analysis. The JAMA meta-analysis used 29 studies, but the current authors were a bit more selective. All of the studies had a Jadad score for quality of 3 or more on a 5-point scale. Overall, they estimated a number needed to treat of 12 for global response and 10 for cognitive response. However, the number needed to harm was 12 for any adverse event and 16 for any adverse event severe enough to cause a dropout.

All drugs were not equal—donepezil appeared to be somewhat more effective than galantamine and was better tolerated. Higher dosages did not produce much benefit over lower dosages. A flaw in the study was that results from studies were combined even when there was significant heterogeneity (variability between studies). However, examining the individual study data for cognitive responders shows that much of this variability may have come from combining studies of donepezil and rivastigmine (which clustered together) with those of galantamine. When the drugs were examined separately, only rivastigmine exhibited significant heterogeneity.

If the definition of success was changed to greater than minimal improvement, the number needed to treat was 42 (95 percent confidence interval, 26 to 114), a much less impressive benefit. With global responders, variability came from a study done exclusively with Japanese patients, which detected a large treatment effect.

Bottom Line: Using a fairly loose criterion for success of “any benefit,” cholinesterase inhibitors helped one patient for every 12 treated but caused side effects significant enough to force discontinuation of the drug in one patient for every 16 treated. Donepezil appears to be more effective than galantamine, although head-to-head trials are lacking. Low dosages are similar to high dosages in efficacy. (Level of Evidence: 1a–)

POEMs (patient-oriented evidence that matters) are provided by Essential Evidence Plus, a point-of-care clinical decision support system published by Wiley-Blackwell. For more information, see http://www.essentialevidenceplus.com. Copyright Wiley-Blackwell. Used with permission.

For definitions of levels of evidence used in POEMs, see http://www.essentialevidenceplus.com/product/ebm_loe.cfm?show=oxford.

To subscribe to a free podcast of these and other POEMs that appear in AFP, search in iTunes for “POEM of the Week” or go to http://goo.gl/3niWXb.

This series is coordinated by Sumi Sexton, MD, editor-in-chief.

A collection of POEMs published in AFP is available at https://www.aafp.org/afp/poems.

Continue Reading


More in AFP

Copyright © 2004 by the American Academy of Family Physicians.

This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP.  See permissions for copyright questions and/or permission requests.