Premium
Design characteristics that influence attrition in geriatric antidepressant trials: meta‐analysis
Author(s) -
Heo Moonseong,
Papademetriou Eros,
Meyers Barnett S.
Publication year - 2009
Publication title -
international journal of geriatric psychiatry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.28
H-Index - 129
eISSN - 1099-1166
pISSN - 0885-6230
DOI - 10.1002/gps.2211
Subject(s) - attrition , placebo , depression (economics) , antidepressant , sample size determination , medicine , psychology , statistics , alternative medicine , mathematics , dentistry , pathology , hippocampus , economics , macroeconomics
Objective To identify study‐level and treatment group‐level characteristics that are associated with attrition in antidepressant trials for the depressed elderly. Methods This meta‐analysis used 68 published antidepressant randomized trials for the elderly depressed. Study‐level and treatment group‐level variables were extracted. The former consisted of: use of placebo arm, number of arms, unbalanced treatment allocation, year conducted, study duration, study location, number of centers, and patient pool. The latter consisted of: assigned treatments, baseline mean Hamilton Depression Rating Scale (HDRS), baseline mean age, gender ratio, attrition size, and sample size. Univariate logistic regressions with attrition as the dependent variable were applied, followed by application of a forward stepwise selection method for identification of independent attrition correlates. Results The 68 studies had a total of 153 treatment groups and 8,385 subjects. Among them, 2,287 subjects were terminated early, resulting in an overall 27.3% attrition rate. The stepwise results showed that higher attrition was significantly associated with active antidepressant groups as opposed to placebo groups, higher baseline HDRS, smaller sample size, unbalanced allocation of treatments, longer duration, and studies conducted in USA. Conclusions The attrition in geriatric antidepressant trials can be affected by study and group‐level design characteristics that include severity of depression symptoms and active treatments among others. Copyright © 2009 John Wiley & Sons, Ltd.