Yellow over Dark Blue, Ellsworth Kelly, 1964-65
The the largest real-world study of bipolar pharmacotherapy to date favors lithium, but with a major caveat
STUDY: Okumura Y et al, Br J Psychiatry 2026; doi: 10.1192/bjp.2026.10636
STUDY TYPE: Population-based cohort study (within-individual design)
FUNDING: Japan Agency for Medical Research and Development
Background
This study used Japan’s national insurance database to compare mood stabilizers and antipsychotics, alone and in combination, across a massive patient population over a decade.
The Study
Let’s start with the limitation. They only included patients who were diagnosed with bipolar and received at least two script for lithium. The idea was to refine the database search, and they hoped that requiring the lithium scripts would ensure the bipolar diagnosis was not an error. However, this may bias the sample toward lithium-responsive patients, not to mention that unipolar depression is also an indication for lithium.
- 315,046 adults with bipolar disorder
- Treated in Japanese psychiatric settings between 2013 and 2022, following median of 7 years
- Primary outcome = time to psychiatric hospitalization
- Each patient served as their own control (periods on a given medication compared against periods off it)
The Results
- Lithium monotherapy performed best, lowering hospitalization risk by 33% (adjusted hazard ratio [aHR] 0.67) compared to no mood stabilizer use
- Valproate, lamotrigine, and carbamazepine followed closely (aHRs 0.71, 0.72, and 0.74)
- Among antipsychotics, aripiprazole performed well in both oral (aHR 0.73) and long-acting injectable forms (aHR 0.62)
- Quetiapine showed only modest benefit overall, but this was likely due to inclusion of lower doses used for sleep. At doses of 300 mg/day or higher it performed well.
Combinations beat monotherapy in several cases:
- Adding carbamazepine to lithium reduced risk an additional 27% versus lithium alone (aHR 0.73)
- Adding aripiprazole to lithium, valproate, or lamotrigine consistently provided additional benefit
- Lithium plus sulpiride was the one combination associated with worse outcomes (aHR 1.16) (it is an antipsychotic similar amisulpride).
Other limitations: the database can’t capture symptom severity, bipolar I vs II, medication adherence, or the reason for hospitalization (mania versus depression).
Practice Implications
- Despite its major limitation, this study adds further support to the lithium-carbamazepine combination.
- In another study, patients who did not respond to lithium, and did not respond to carbamazepine, responded to them in combination. The two also reduce each other’s side effects, which is why we highlighted them in our Carlat article on rational polypharmacy.
- Aripiprazole stands out as the most consistent add-on across all three mood stabilizers tested. What’s your experience with that one?
— Chris Aiken, MD
Director, Psych Partners
Editor in Chief, Carlat Psychiatry Report
