This entire article is available in PDF format to paid subscribers (certain restrictions apply).
If you have not already registered for Full Text Access to The Journal, then visit our registration page.

Study Design Factors and Patient Demographics and Their Effect on the Decline of Placebo-Treated Subjects in Randomized Clinical Trials in Alzheimer's Disease

Michael Gold, M.D.


Objective: To gain insight into factors affecting the rate of decline in placebo-treated subjects participating in Alzheimer's disease (AD) clinical trials.

Data Sources: Studies were identified through a MEDLINE search using a combination of 2 strategies. The first strategy searched for the term Alzheimer's disease and was restricted to those publications that reported a randomized controlled trial in human adults. A second strategy searched for the terms Alzheimer's disease and ADAS-Cog and restricted the results to human adults.

Study Selection: Studies were included if they reported results from a prospective, placebo-controlled, double-blind, parallel group study and used the Alzheimer's Disease Assessment Scale-Cognitive subscale (ADAS-Cog) as a primary outcome measure and included patients with mild to moderately severe probable AD. A total of 69 trials were identified. Three studies were removed from the analysis because of anomalous results. Two data sets were analyzed: an "intent-to-treat" set (69 trials) and a "plausible" set (66 trials).

Data Extraction: Lead or senior author, year of publication, total and placebo sample size, the use of acetylcholinesterase, maximum and minimum Mini-Mental State Examination (MMSE) score allowed, mean MMSE score and mean age at baseline, percentage of female subjects, number of investigational sites, investigational treatment, the number of treatment arms and study evaluations, study duration, percentage of patients with the genotype APOE ε4+, and the mean and SD of the change from baseline on the ADAS-Cog were recorded from each study and from the placebo-treated group specifically.

Data Synthesis: Linear regression models were used to explore the relationship between the date of publication and patient demographics. Multiple linear regression was used to explore the contributions of patient demographics and study design variables to the change from baseline score on the ADAS-Cog. Analyses suggest that patients enrolling in trials are progressively older and that trials are getting progressively longer. Other patient demographics and study design variables have not changed over time. The duration of a study was the most consistent predictor of decline in placebo-treated subjects. Increased number of investigational sites, increased number of evaluations, and milder dementia severity at baseline were associated with a loss of decline in placebo-treated subjects.

Conclusions: Study duration was the only variable that predicted decline consistently, whereas the number of evaluations, the number of sites, and baseline MMSE score are inversely proportional with decline and should be taken into account when planning future clinical trials.

(J Clin Psychiatry 2007;68:430-438)


Received Nov. 3, 2006; accepted Jan. 22, 2007. From GlaxoSmithKline, Research Triangle Park, N.C.

Dr. Gold is an employee of GlaxoSmithKline.

Corresponding author and reprints: Michael Gold, M.D., GlaxoSmithKline, P.O. Box 13398, Five Moore Dr., Research Triangle Park, NC 27709 (e-mail: michael.2.gold@gsk.com).