Meta-analysis of sleep disturbance and suicidal thoughts and behaviors

The potential association of various sleep disturbances to suicidal thoughts and behaviors is the subject of several reviews. The current meta-analysis was conducted to estimate the size of the association generally as well as between more specific relationships.

Electronic databases for years 1966-2011 were searched to identify candidate studies using PubMed search terms suicide and sleep or sleep initiation/maintenance disorders or dreams or nightmares or sleep disorders/psychology or sleep disorders/epidemiology as well as Ovid search terms suicide and sleep or insomnia or nightmares. The search was supplemented by cross-referencing from identified articles and reviews.

Original studies reporting both sleep disturbance and suicide outcomes were identified with 39 of 98 studies (40%) comprising 147,753 subjects selected for inclusion.

Data were extracted by multiple independent observers and verified by a study author. The meta-analysis was performed using random-effects models. The size of associations was calculated for all types of sleep disturbances and suicide outcomes combined and for more specific categories including nightmares, insomnia, and insomnia subtypes and suicidal ideation, suicide attempts, and suicide. Moderator effects were evaluated.

Overall, sleep disturbance was significantly associated with an increased relative risk for suicidal ideation, suicide attempt, and suicide ranging from 1.95 (95% CI, 1.41-2.69) to a relative risk of 2.95 (95% CI, 2.48-3.50) in unadjusted studies. Associations were smaller, but remained highly significant among adjusted studies. Depression did not moderate the association between sleep and suicide variables.

This meta-analysis supports an association between sleep disturbance and suicidal thoughts and behaviors. Sleep disturbances in general, as well as insomnia and nightmares individually, appear to represent a risk factor for suicidal thoughts and behavior. This proposition is further bolstered by the result that depression did not show risk moderation.