Research reportRisk factors for suicide in bipolar I disorder in two prospectively studied cohorts

- Risk factors for suicide and bipolar I disorder can vary markedly depending on the sources of subjects and the methods of follow-up.
- The most consistently observed risk factor for suicide in bipolar I disorder is a history of suicide attempt.
- In neither study was a suicide attempt with high lethality or intent a stronger predictor of suicide than a history of any attempt.

BackgroundThese analyses were undertaken to determine whether similar risk factors for suicide emerged across two prospectively studied cohorts of individuals with bipolar I disorder.MethodsThe NIMH Collaborative Study of Depression (CDS) recruited 288 patients with bipolar I disorder from 1978-1981 as they sought treatment. Subjects were followed semiannually and then annually for up to 30 years. The Bipolar Genomics studies identified individuals through clinical referrals and advertisement. Clinical follow-up did not occur but personal identifiers of 1748 were matched with National Death Index (NDI) records. Kaplan-Meier survival analyses tested ten potential risk factors.ResultsThe CDS and Genomic follow-ups encompassed 12,667 and 4529 person-years, respectively. Suicides/100 person-years were 0.26 and 0.055. The demographic or clinical variables that predicted suicide differed considerably in the two cohorts. The odds ratio for suicide for those with any history of suicide attempt was 2.3 and 2.8, respectively, and was the third highest odds ratio of the tested risk factors in both studies.Limitations1.In one of the two cohorts, matching to the NDI was the only means available to ascertain death and cause of death.2.In one of the two cohorts, depressive symptom ratings were, in most cases, applied to those of a past major depressive episode.ConclusionsDifferences in the sources of participants in studies of suicide risk may result in marked differences across studies in both rates of suicide and in risk factors. A history of suicide attempt is a relatively robust risk factor across samples.