How-To Guides

Yes. In this 52-week multicenter retrospective cohort, past-year suicidal ideation was reported less often in the lithium group than in the non-lithium group (53.1% vs 65.4%, P = .003), and lithium use remained associated with lower odds of suicidal ideation after adjustment for clinical and treatment factors (adjusted odds ratio [aOR] = 0.57, 95% CI = 0.38–0.86, P = .008). The association was also seen for earlier forms of suicidal thinking, including wish to be dead (aOR = 0.59, 95% CI = 0.39–0.88, P = .01) and nonspecific active suicidal thoughts (aOR = 0.65, 95% CI = 0.44–0.96, P = .03).

Yes for suicidal behaviors overall, but not definitively for time to suicide attempt. Suicidal behaviors were less common in the lithium group than in the non-lithium group (23.8% vs 36.3%, P = .001), and adjusted analysis showed lower odds overall (aOR = 0.60, 95% CI = 0.39–0.90, P = .02).

However, when the authors examined time to first suicide attempt with a Cox model, the result was not statistically significant, although it favored lithium numerically (adjusted HR = 0.74, 95% CI = 0.53–1.04, P = .08). Kaplan-Meier analysis also showed a nonsignificant but lower cumulative incidence in the lithium group (log-rank P = .08).

No significant association was found between lithium use and nonsuicidal self-injury. NSSI was common in both groups, with 39.5% of the overall sample reporting it at least once in the past year, and the frequency did not differ significantly between lithium-treated and non-lithium participants (P = .167).

Yes, lithium users had lower hostility scores. On the Buss and Perry Aggression Questionnaire, the lithium group had a significantly lower hostility score than the non-lithium group (mean [SD], 21.42 [7.27] vs 23.03 [7.73], P = .01).

The article reports this as a group difference in hostility; it does not claim a causal mechanism, but notes that aggression-related traits may be relevant when considering suicidality in bipolar disorder.

In exploratory subgroup analyses, the protective association appeared more pronounced in females, adults, and higher-income participants, and in patients without comorbid anxiety or rapid cycling.

• Females: OR = 0.52, 95% CI = 0.32–0.86, P = .01
• Adults aged ≥18 years: OR = 0.46, 95% CI = 0.27–0.77, P = .003
• Higher income: OR = 0.41, 95% CI = 0.22–0.78, P = .006
• No comorbid anxiety: OR = 0.59, 95% CI = 0.36–0.96, P = .034
• No rapid cycling history: OR = 0.50, 95% CI = 0.30–0.81, P = .006

The study also found no significant association in adolescents younger than 18 years (OR = 0.82, 95% CI = 0.43–1.56, P = .54) or in those with predominant depressive episode type (OR = 0.82, 95% CI = 0.39–1.38, P = .45). The authors emphasize that these subgroup analyses were post hoc, exploratory, and hypothesis-generating.

Yes. The study reports that lithium remained associated with lower suicidal ideation regardless of psychotic symptoms. In subgroup analysis, the odds ratios were 0.14 and 0.64 in participants with and without psychotic symptoms, respectively, and both were statistically significant at P < .05.

This was a nationwide, 15-center retrospective cohort study in China with a 52-week observation period. It included 585 adolescents and adults aged 12–45 years with DSM-IV bipolar disorder confirmed by the MINI, with 290 in the lithium group and 295 in the non-lithium group.

The lithium group consisted of patients with at least 80% adherence to lithium during the first 6 months of the past year, and mean lithium exposure during the observation period was 338 days (SD = 79). The primary outcome was past-year suicidal ideation assessed with a structured interviewer-administered instrument adapted from the Columbia-Suicide Severity Rating Scale, and the main analysis used multivariable logistic regression with an IPTW sensitivity analysis.

This was a high-risk bipolar disorder sample. Across the full cohort, 59.3% reported suicidal ideation in the past year, 24.4% reported at least 1 suicide attempt in the past year, and about 70% reported suicidal ideation 1 year prior.

The sample also had a chronic illness course with a mean duration of 4 years, more than 30% had a history of rapid cycling, about 20% had experienced psychotic symptoms, and 61.5% had a predominant depressive episode type.

No. This study supports an association, not proof of causation or a definitive reduction in suicide deaths. The authors state that the observational retrospective design and nonrandom treatment allocation limit causal inference, and residual confounding by indication cannot be fully excluded even after adjustment and IPTW analysis.

The article also states that reduced suicidal ideation or suicidal behavior does not necessarily guarantee reduced completed suicide. Reverse causality and recall bias are additional limitations because outcomes were assessed retrospectively over a 12-month period.

The main limitations are observational design, retrospective assessment, possible reverse causality, and limited generalizability.

• Because treatment was not randomized, the study cannot establish causation, and unmeasured confounding by indication may remain.
• Suicidal ideation and behaviors were assessed retrospectively over the previous year, which introduces recall bias.
• The study could not definitively determine the order of events within the 1-year window, including whether suicidal ideation emerged after lithium discontinuation or lithium was prescribed in response to recent ideation.
• Most participants were from urban tertiary hospitals, so the findings may not generalize fully to rural or community settings.