Comprehensive meta-analysis of excess mortality in depression in the general community versus patients with specific illnesses.

Objective: Several hundred studies have shown that depression is associated with an elevated risk of dying at follow-up. It is not clear, however, whether the mechanisms for this association are disease specific, leading to higher mortality in specific patient groups, or generic, resulting in comparable mortality rates in all patient groups as well as in community samples. The authors conducted a comprehensive meta-analysis of prospective studies of community as well as patient samples associating depression at baseline with excess mortality at follow-up.

Method: The authors conducted systematic searches of PubMed, PsycINFO, and Embase. Studies were included if depression was measured with a standardized instrument and mortality was reported for both depressed and nondepressed participants at follow-up.

Results: A total of 293 studies including 1,813,733 participants (135,007 depressed and 1,678,726 nondepressed) from 35 countries were included. The overall unadjusted relative risk of mortality in depressed relative to nondepressed participants was 1.64 (95% CI=1.56-1.76), with high heterogeneity (I2=83, 95% CI=80-84). After adjustment for publication bias, the overall relative risk was reduced to 1.52 (95% CI=1.45-1.59). No strong indications were found that the pooled relative risk was different across the relatively healthy community samples and specific patient samples with heart disease, cancer, kidney disease, or other disease, except for a significantly higher risk in patients with chronic obstructive pulmonary disease (p<0.05). Also, the relative risk was lower when the follow-up period was longer and when the quality of the study was higher.

Conclusions: The authors could confirm the presence of a significant association between depression and excess mortality, although this association may have been overestimated because of publication bias and low study quality. Few indications were found that this association is stronger in community or specific patient samples.