Depression and memory in the elderly: A meta-analysis

Abstract

A meta-analysis (N = 40) of the effects of depression on memory in the elderly (sample mean age ≥ 55 years) examined variables potentially accounting for divergent findings in the literature. The distribution of effects was bimodal and the effect sizes were heterogeneous. Compared to controls, groups containing unipolar subjects only were significantly less impaired than were mixed unipolar-bipolar; five of six studies mixing depression subtype were associated with the more negative mode. Samples containing younger depressed subjects (< 45 years) were significantly more impaired and were associated with the more negative mode. Significant group differences were found between studies matching their comparison groups reasonably well on years of education and those that did not. Thoroughness of dementia screening yielded no group differences. Although correlated observations precluded significance tests, larger effects were found for (1) figural (vs. verbal) memory; (2) delayed (vs. immediate) memory; and (3) recognition (vs. free recall and incidental or cued recall; incidental and cued recall effects were nearly identical). Similar effects were found for composite memory scores versus constituent and for various presentation paradigms (e.g., single presentation, selective reminding). Effect sizes for these categories were in the moderate range. Difficulties synthesizing this literature are discussed as are suggested remedies and directions for future research.