Case-Control Analyses of the Impact of Pharmacotherapy on Prospectively Observed Suicide Attempts and Completed Suicides in Bipolar Disorder: Findings From STEP-BD

Lauren B. Marangell, MD; Ellen B. Dennehy, PhD; Stephen R. Wisniewski, PhD; Mark S. Bauer, MD; Sachiko Miyahara, MS; Michael H. Allen, MD; Melissa Martinez, MD; Rayan K. Al Jurdi, MD; Michael E. Thase, MD

J Clin Psychiatry 2008;69(6):916-922

Article Abstract

Objective: Given high rates of suicide and suicide attempts in bipolar disorder and the data suggesting a suicide-protective effect of lithium, we evaluated the impact of pharmacotherapy on prospectively observed suicides and suicide attempts in subjects in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD).

Method: The STEP-BD study enrolled 4360 participants with DSM-IV bipolar disorder diagnoses from September 1998 through November 2004. There were 270 suicide events in STEP-BD (8 completed suicides, 262 attempts). These occurred in 182 of STEP-BD participants (cases). Inclusion criteria required cases to be white or Caucasian, have at least 1 postbaseline visit, and have prescription information within 30 days of the suicide event. This reduced the available cohort to 106 cases. Matching included age, gender, history of previous suicide attempt, and a propensity score that considered bipolar subtype, marital status, age at onset, and history of psychosis, resulting in 93 matched pairs. A secondary analysis added mood state status within 30 days of the suicide event to the propensity score (N = 54 pairs). The association of drug prescriptions with suicide attempts/completions was assessed using a conditional logistic regression model.

Results: The results do not indicate a relationship between lithium use and suicide attempts or completions (p = .41). Similar findings were found for exposure to valproate, carbamazepine, lamotrigine, and the atypical antipsychotic medications. An association between selective serotonin reuptake inhibitor (SSRI) prescription and suicide events was observed (p < .0001). Findings were similar in a secondary analysis that controlled for mood state.

Conclusion: Our data are not consistent with a suicide-protective effect of lithium. The association between suicide events and SSRI prescriptions requires cautious interpretation due to complex relationships between treatment, severity, and suicidality.

Trial Registration: clinicaltrials.gov Identifier: NCT00012558