Suicide death over the first year of lithium versus valproate treatment in cohorts with and without bipolar disorder

This study examined if lithium's association with suicide risk varies by diagnosis. We performed separate 1:1 high-dimensional propensity score (hdPS)-matching in US Veterans with and without bipolar disorder starting lithium or valproate. Among individuals with bipolar disorder, actively receiving lithium (compared to valproate) was not associated with suicide risk. However, in intent-to-treat analyses (following all individuals with bipolar disorder starting lithium or valproate for all 365 days, regardless of whether they stopped the medication), starting lithium was significantly associated with higher one-year risks of suicide (HR = 1.50, 95% CI: 1.05-2.15, p = 0.03). These intent-to-treat risks were attributable entirely to transiently elevated suicide risks observed among individuals no longer receiving lithium (significant at 180 days [HR = 6.10, CI: 1.37-27.3, p = 0.02] but not 365 days [HR = 2.05, CI: 0.88-4.79, p = 0.10]). Among individuals without bipolar disorder, depending on the analysis, actively receiving lithium was associated with nonsignificantly (HR = 0.43, CI: 0.15-1.20, p = 0.11) or significantly (HR = 0.28, CI: 0.08-0.98, p = 0.047) decreased one-year suicide risks. Study limitations included limited power, brief follow-up, and potential residual confounding. Residual confounding is suggested by the observation that more individuals diagnosed with suicidal ideation started lithium than valproate (with this difference being statistically significant for individuals with bipolar disorder, p = 0.0012). If it were possible to correct for this potential confounding, then the suicide-related risks associated with among individuals discontinuing lithium would be expected to be less, and the suicide-related benefits associated with actively receiving lithium (already statistically significant in some analyses among individuals without bipolar disorder) would be expected to increase. Further research is needed.