• bipolar disorder;
  • cluster analysis;
  • suicide


The development of more sophisticated models for understanding suicide among people with bipolar disorder (BD) requires diagnosis-specific data. The present study aimed to elucidate differences between people who die by suicide with and without BD, and to identify subgroups within those with BD.


Data on all suicide deaths in the city of Toronto from 1998 to 2010 were extracted from the Office of the Chief Coroner of Ontario, including demographics, clinical variables, recent stressors, and details of the suicide. Comparisons of person- and suicide-specific variables between suicide deaths among those with BD (n = 170) and those without (n = 2,716) were conducted, and a cluster analysis was performed among the BD suicide group only.


Those in the BD suicide group were more likely than those in the non-BD suicide group to be female [odds ratio (OR) = 1.75, 95% confidence interval (CI): 1.27–2.42; p = 0.001], to have made a past suicide attempt (OR = 2.01, 95% CI: 1.45–2.80; p < 0.0001), and to have had recent contact with psychiatric or emergency services (OR = 1.59, 95% CI: 1.00–2.52; p = 0.049). Five clusters were identified within the BD group, with differences between clusters in age; sex; marital status; living circumstances; past suicide attempts; substance abuse; interpersonal, employment/financial, and legal/police stressors; and rates of death by fall/jump or self-poisoning.


The present findings identified differences between BD and non-BD suicide groups, providing support to the utilization of an illness-specific approach to better understanding suicide in BD. Subgroups of BD suicide deaths, if replicated, should also be incorporated into the design and analysis of future studies of suicide in BD.