Suicide risk with selective serotonin reuptake inhibitors and other new-generation antidepressants in adults: A systematic review and meta-analysis of observational studies

Background: There is ongoing controversy whether antidepressant use alters suicide risk in adults with depression and other treatment indications.

Methods: Systematic review of observational studies, searching MEDLINE, PsycINFO, Web of Science, PsycARTICLES and SCOPUS for case-control and cohort studies. We included studies on depression and various indications unspecified (including off-label use) reporting risk of suicide and/or suicide attempt for adult patients using selective serotonin reuptake inhibitors (SSRI) and other new-generation antidepressants relative to non-users. Effects were meta-analytically aggregated with random-effects models, reporting relative risk (RR) estimates with 95% CIs. Publication bias was assessed via funnel-plot asymmetry and trim-and-fill method. Financial conflict of interest (fCOI) was defined present when lead authors’ professorship was industry-sponsored, they received industry-payments, or when the study was industry-sponsored.

Results: We included 27 studies, 19 on depression and 8 on various indications unspecified (n=1.45 million subjects). SSRI were not definitely related to suicide risk (suicide and suicide attempt combined) in depression (RR=1.03, 0.70-1.51) and all indications (RR=1.19, 0.88-1.60). Any new-generation antidepressant was associated with higher suicide risk in depression (RR=1.29, 1.06-1.57) and all indications (RR=1.45, 1.23-1.70). Studies with fCOI reported significantly lower risk estimates than studies without fCOI. Funnel-plots were asymmetrical and imputation of missing studies with trim-and-fill method produced considerably higher risk estimates.

Conclusions: Exposure to new-generation antidepressants is associated with higher suicide risk in adult routine-care patients with depression and other treatment indications. Publication bias and fCOI likely contribute to systematic underestimation of risk in the published literature.