Cipriani et al. Lithium in the prevention of suicide in mood disorders: updated systematic review and meta-analysis. BMJ. 2013;346:f3646. PMID 23814104¶
One-paragraph summary¶
This systematic review and meta-analysis searched randomized long-term trials comparing lithium with placebo or active treatment in bipolar or unipolar mood disorders. Forty-eight trials with 6,674 participants and 15 comparisons were included. Against placebo, lithium was associated with fewer suicides, OR 0.13 (95% CI 0.03–0.66), and fewer deaths from any cause, OR 0.38 (0.15–0.95), while deliberate self-harm was not significantly reduced, OR 0.60 (0.27–1.32) (Cipriani 2013, PMID 23814104).
Key findings¶
- Evidence base: 48 randomized controlled trials, 6,674 participants.
- Suicide versus placebo: OR 0.13 (95% CI 0.03–0.66).
- All-cause death versus placebo: OR 0.38 (0.15–0.95).
- Deliberate self-harm versus placebo: OR 0.60 (0.27–1.32), not statistically significant.
- In unipolar depression, suicide OR was 0.36 (0.13–0.98) and all-cause death OR 0.13 (0.02–0.76).
- Against individual active comparators, the only statistically significant difference was versus carbamazepine for deliberate self-harm.
Limitations¶
- Suicide is rare in randomized maintenance trials, making estimates sensitive to zero-event handling and analytic method.
- The synthesis combined unipolar and bipolar mood disorders.
- Trials were designed primarily for mood recurrence, not suicide, and ascertainment quality varied.
- The abstract does not establish whether any effect is specific beyond relapse prevention.
- A later modern-trial reanalysis found suicide OR 0.41 (95% CI 0.03–2.49), not significant, so the magnitude remains disputed (Nabi 2022, PMID 36111461).
Why it matters¶
The paper supplied the most influential randomized quantitative support for lithium’s anti-suicide reputation and shaped risk-benefit discussions far beyond ordinary relapse prevention. Its rare-event vulnerability is equally important: newer analysis does not confirm a statistically significant effect, leaving a genuine evidence conflict rather than a settled protective magnitude. The paper is therefore landmark both for its clinical influence and for what it teaches about meta-analysis of rare outcomes.
Cited by wiki pages¶
- Lithium