Bipolar depression is where most of the disability in bipolar disorder sits, treatment options are thin, and ketamine has been withheld from these patients largely on theoretical grounds. Ket-BD tested the theory.
This investigator-led, double-blind, midazolam-controlled randomised trial ran at three Ontario sites. It enrolled 68 outpatients aged 21 to 65 with DSM-5 bipolar I or II disorder, a current moderate-to-severe depressive episode with a MADRS of 21 or above, and at least two failed trials of evidence-based pharmacotherapy. Participants received four flexibly dosed 40-minute infusions over two weeks — ketamine 0.5 to 0.75 mg/kg or midazolam 0.02 to 0.03 mg/kg — adjunctive to a stable dose of at least one mood stabiliser or antipsychotic. Mean age was 44.4 and 55.8% were women.
At day 14, adjusting for sex, bipolar type and baseline MADRS, the ketamine group scored 7.3 points lower (95% CI -12.0 to -2.5, p=0.003), a Cohen's d of 0.7.
The safety finding is the one that changes anything. There were no cases of mania, hypomania, psychosis or suicide attempt in either arm, and one case of subthreshold mixed features in each. That is not proof of safety in a trial of 68 people, but it is the first randomised evidence in this population and it does not support the fear that has kept ketamine out of it.
Two constraints. Every participant was on a mood stabiliser or antipsychotic, so this tests adjunctive ketamine on mood-stabilised patients, not ketamine alone. And 47% correctly guessed their allocation after the first infusion, which is the perennial problem with a drug that produces dissociation — midazolam is the best available control and it is still imperfect.
- Consider adjunctive intravenous ketamine in treatment-resistant bipolar depression, on a stable mood stabiliser or antipsychotic
- Do not withhold it on the assumption it will precipitate mania — no manic or hypomanic switches occurred in this trial
- Maintain the mood stabiliser: every participant was on one, and ketamine monotherapy in bipolar disorder remains untested
- Monitor for mixed features rather than frank mania; one subthreshold case occurred in each arm
- The effect was measured at day 14 after four infusions — durability beyond that was not the endpoint
The statistics, in plain English
A MADRS difference of 7.3 points is clinically meaningful — around 2 points is usually considered the minimum detectable change, and 6 or more is generally regarded as a substantial response. Cohen's d of 0.7 is a moderate-to-large effect. The confidence interval, -12.0 to -2.5, is wide because only 63 participants entered the efficacy analysis, so the true effect could be anywhere from modest to very large. On safety, absence of events in 68 people is weak evidence: with zero manic switches observed, the upper bound of the true rate is still around 5%. The blinding failure — 47% guessing correctly after one infusion — inflates apparent effect on a subjective rating scale, though the observer-rated MADRS mitigates this somewhat.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free