Lithium and valproate come out level on suicide risk — so choose on everything else

Design

territory-wide retrospective cohort with inverse probability of treatment weighting, plus pairwise and network meta-analysis

Population

1,379 lithium and 5,556 valproate monotherapy starters with bipolar disorder, 2003–2023; 3 RCTs and 12 observational studies in the synthesis

Primary outcome

suicide risk

Effect

cohort weighted HR 0.76 (95% CI 0.54–1.08); pooled HR 0.87 (0.69–1.09); versus no treatment, lithium 0.46 (0.33–0.64) and valproate 0.63 (0.45–0.88)

Lithium's reputation as the anti-suicide mood stabiliser has driven prescribing for decades, but the direct comparison against valproate has been thin. This study did both halves of the job: a territory-wide cohort from Hong Kong Hospital Authority records of patients with bipolar disorder starting lithium or valproate monotherapy between 2003 and 2023, and a meta-analysis placing that cohort alongside the existing literature.

In the cohort — 1,379 lithium users and 5,556 valproate users, weighted to balance baseline characteristics — suicide risk did not differ significantly (weighted HR 0.76, 95% CI 0.54 to 1.08). Pairwise meta-analysis of eight observational studies agreed (pooled HR 0.87, 95% CI 0.69 to 1.09). The network meta-analysis found both drugs better than no treatment — lithium HR 0.46 (0.33 to 0.64), valproate HR 0.63 (0.45 to 0.88) — with no significant difference between them (HR 0.73, 95% CI 0.52 to 1.02).

The practical consequence is that the choice can be made where it should have been made all along: on renal and thyroid function, on pregnancy potential, on monitoring feasibility, on adherence and on tolerability. Valproate's teratogenicity keeps it off the table for a woman who could become pregnant, whatever this analysis shows about suicide. And the strongest single signal here is not lithium versus valproate at all — it is that both roughly halve suicide risk compared with being on neither, which makes the patient on no mood stabiliser the one to act on.

• Choose between the two on renal function, thyroid function, monitoring access and tolerability.
• Valproate remains contraindicated in women of childbearing potential without the full risk programme, regardless of this analysis.
• The untreated patient is the priority — both agents beat no treatment substantially.
• Do not switch a patient stable on valproate to lithium purely for suicide prevention on this evidence.
• Where lithium monitoring is not reliably available, that is a legitimate reason to choose otherwise.

Why it matters

It removes suicide prevention as the tie-breaker in a choice most psychiatrists make on that basis.

Don't overread it

Overlapping confidence intervals mean no difference was found, not that the two drugs are proven equivalent.

The statistics, in plain English

Every comparison between the two drugs has a confidence interval crossing 1.0 — 0.54 to 1.08 in the cohort, 0.69 to 1.09 pooled, 0.52 to 1.02 in the network. That means no difference was detected, which is not the same as showing the two are equivalent: the intervals still admit a meaningful lithium advantage of up to about a 45% relative reduction. The comparisons against no treatment are the ones that clear 1.0 decisively. The cohort is observational and weighted, so unmeasured confounding by illness severity remains possible — sicker patients are not randomly assigned to lithium.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.