Forty-eight RCTs (6,674 patients) were included. Mean follow-up was 19.1 months (range four to 48 months). Study quality was generally good but some trials did not report details of randomisation and allocation concealment. Twenty-nine trials had fewer than 100 participants, resulting in few suicide or self-harm events.
Compared with placebo, lithium significantly reduced suicide (Peto OR 0.13, 95% CI 0.03 to 0.66; four RCTs) and all-cause mortality (Peto OR 0.38, 95% CI 0.15 to 0.95; eight RCTs) but there was no significant difference for self-harm (Peto OR 0.60, 95% CI 0.27 to 1.32; three RCTs). Heterogeneity was not significant.
Differences between lithium and active comparators for suicide and all-cause mortality were not significant. Lithium reduced deliberate self-harm compared with carbamazepine (Peto OR 0.14, 95% CI 0.02 to 0.83) but not other drugs. Heterogeneity was generally low.
In sensitivity analyses against placebo, lithium reduced suicide and all-cause mortality but not self harm in patients with unipolar depression. The planned analysis focusing on young people was not possible because of small numbers of events.