LAUSR

BACKGROUND AND PURPOSES Previous meta-analyses on neurocognitive test performances in depression have provided effect sizes for cognitive domains. Most domain effect sizes have medium to high variance heterogeneity. Restriction to each test performance could reduce variance and clarify differing test effect sizes. This systematic review and meta-analysis were done to 1. provide effect sizes for cognitive performances (test, subtest, or multiple measures within tests), and 2. investigate age as an effect modifier. METHODS Inclusion criteria were: 1. active major depression episode (MD), 2. a control group, 3 reported means and standard deviations, 4. non-computerized tests previously studied at least 3 times. Meta-analyses were performed using Cochrane Review Manager. Age under versus over 45 was investigated as an effect moderator. RESULTS Twenty-seven studies met criteria. MD patients performed significantly poorer on 16 of 16 neurocognitive measures (random effects d = -0.47 to -0.92 across tests). Variance was heterogeneous for 11 of 16 measures. Differences between cognitive measures were largely absent based on overlapping 95% confidence intervals. Effect sizes did not differ under versus over 45 years. LIMITATIONS Bias risk assessment showed limited control for subject selection, comparability of depressed and control groups, pre-morbid intelligence, drug treatment, and effort in testing. CONCLUSIONS The depression - cognition effect was in the moderate to large range regardless of test type. Variance heterogeneity was substantial despite exclusion of inactive depression and the absence of test pooling. The size of the depression - cognition effect was not a function of age.