Comparing lithium and valproate in reducing suicide risk: A territory-wide retrospective cohort study with pairwise and network meta-analysis.

Direct clinical comparisons between lithium and valproate for suicide prevention remain limited and inconclusive. We aimed to evaluate their effectiveness of suicide prevention using a retrospective cohort study and meta-analysis. We conducted a retrospective cohort study using electronic health records from the Hong Kong Hospital Authority, including patients with bipolar disorder who initiated lithium or valproate monotherapy between 2003 and 2023. Inverse probability of treatment weighting was applied to balance baseline characteristics, and weighted Cox proportional hazards regression estimated hazard ratios (HRs) for suicide risk. We also searched electronic databases for randomized controlled trials (RCTs) and observational studies comparing lithium, valproate, and nontreatment controls. Pairwise meta-analyses and network meta-analyses were performed to synthesize direct and indirect evidence. The cohort study included 1,379 lithium users and 5,556 valproate users, with no significant difference in suicide risk between groups (weighted HR = 0.76, 95% confidence interval [CI]: 0.54-1.08). The meta-analysis identified 3 RCTs and 12 observational studies. Meta-analysis of RCTs was precluded due to methodological incompatibilities. Pairwise meta-analysis of eight observational studies showed no significant difference between lithium and valproate (pooled HR = 0.87, 95% CI: 0.69-1.09). Network meta-analysis showed that both lithium (network HR = 0.46, 95% CI: 0.33-0.64) and valproate (network HR = 0.63, 95% CI: 0.45-0.88) were associated with lower suicide risk than no treatment, with no significant difference between the two agents (network HR = 0.73, 95% CI: 0.52-1.02). Lithium and valproate showed comparable effectiveness in reducing suicide risk, while both were associated with lower risk than no treatment. The choice between these two mood stabilizers should be guided by individual patient characteristics and safety considerations.