AIM: To estimate and test associations between substance use disorders (SUDs) and both completed suicides and suicide attempts in a population with severe mental illness.

DESIGN: Register-based cohort study with adjusted Cox regression of substance use disorders as time-varying covariates.

SETTING: Denmark.

PARTICIPANTS: People born in Denmark since 1955 with a diagnosis of schizophrenia (n = 35 625), bipolar disorder (n = 9279), depression (n = 72 530) or personality disorder (n = 63 958).

MEASUREMENTS: Treated SUDs of alcohol and illicit substances identified in treatment registers; suicide attempt identified in treatment registers; and completed suicides identified in the Cause of Death register. Covariates were sex and age at diagnosis.

FINDINGS: Having any SUD was associated with at least a threefold increased risk of completed suicide when compared with those having no SUD. Alcohol misuse was associated with an increased risk of completed suicide in all populations with hazard ratios (HR) between 1.99 [95% confidence interval (CI) = 1.44-2.74] and 2.70 (95% CI = 2.40-3.04). Other illicit substances were associated with a two- to threefold risk increase of completed suicide in all populations except bipolar disorder, and cannabis was associated with increased risk of attempted suicide only in people with bipolar disorder (HR = 1.86, 95% CI = 1.15-2.99). Alcohol and other illicit substances each displayed strong associations with attempted suicide, HR ranging from 3.11 (95% CI = 2.95-3.27) to 3.38 (95% CI = 3.24-3.53) and 2.13 (95% CI = 2.03-2.24) to 2.27 (95% CI = 2.12-2.43), respectively. Cannabis was associated with suicide attempts only in people with schizophrenia (HR = 1.11, 95% CI = 1.03-1.19).

CONCLUSION: Substance use disorders are associated strongly with risk of completed suicides and suicide attempts in people with severe mental illness.