Background: Lithium may exert an antisuicidal effect in bipolar disorder patients, but this hypothesis requires further testing by direct comparison of patients with and without lithium treatment.

Method: Risk of life-threatening suicidal acts over time and associated factors were analyzed in 310 patients with DSM-IV bipolar I (N=186) or II (N=124) disorder evaluated for a mean of 8.3 years before, and prospectively during, a mean of 6.4 years of lithium maintenance in a mood disorder clinic; 185 were also followed for a mean of 3.7 years after clinically discontinuing lithium.

Results: In 5233 patient-years of observation, 58 patients made 90 suicide attempts (8 were fatal). Survival analyses with Weibull modeling with adjustments for covariates indicated a highly significant 6.4-fold adjusted hazard ratio during versus before and 7.5-fold ratio after versus during lithium maintenance. Suicidal acts were more common early in the course of illness before lithium and were associated with prior suicide attempts, greater proportion of time depressed, and younger age. After the discontinuation of lithium, suicidal acts were more frequent in the first year than at later times or before start of lithium treatment. Fatalities were 9 times more frequent after versus during treatment.

Conclusion: Lithium maintenance was associated with marked reduction of life-threatening suicidal acts, the number of which sharply increased after discontinuing lithium. Suicidal behavior was strongly associated with prior suicide attempts, more time depressed, and younger age or recent onset. Greater attention to suicidal risk in patients with bipolar depression and assessment of all proposed mood-stabilizing agents for antisuicidal effects are strongly encouraged.