Can you look up how strong the evidence is for your bipolar medication?
Yes. Researchers pooled 77 systematic reviews covering 116 different bipolar disorder treatments, then put every result on a free public website so patients and doctors can look them up. The review also showed how thin the evidence is in places: of 2,510 separate analyses, only 236 earned the top certainty rating.
Bipolar disorder is not one problem with one fix. It moves through phases. There is bipolar depression, where mood drops. There is mania or a mixed episode, where mood and energy climb too high. And there is maintenance, the long stretch between episodes when the goal is to stop the next one from starting. A drug that works well in one phase may do nothing in another. That is the core message of this project, and it is why a single “best drug for bipolar” answer does not exist.
The team searched PubMed, PsycInfo, and the Cochrane library from the beginning of each database through 19 November 2024. They pulled in reviews that combined randomised trials, the strongest study design in medicine. Then they graded how trustworthy each answer was.
What the data show
The 77 included studies were made up of 21 network meta-analyses and 56 pairwise meta-analyses. Between them they covered 116 unique treatments: 74 medications, 18 brain stimulation approaches, 13 nutraceuticals such as supplements, 8 psychosocial therapies, and 3 circadian rhythm based treatments that work on sleep and body clock timing. Five control treatments were included for comparison.
Those studies looked at 133 different outcomes, split into 45 measures of whether a treatment worked and 88 measures of whether it was safe. Running all the combinations produced 2,510 meta-analyses. Each one was scored with GRADE, a standard system that rates how confident a reader can be in a result. The ratings came out as high for 236, moderate for 827, low for 986, and very low for 461. So roughly 42 percent of the evidence sits at moderate or better, and about 58 percent is low or very low.
Where the evidence is strongest
For bipolar depression in adults, the treatments that worked across outcomes were cariprazine, divalproex or valproate, fluoxetine, ketamine added to existing treatment, lamotrigine, lumateperone, lurasidone, olanzapine, olanzapine combined with fluoxetine, and quetiapine.
For manic episodes in adults, the list was different and longer: aripiprazole, asenapine, carbamazepine, cariprazine, divalproex or valproate, haloperidol, lithium, olanzapine, paliperidone, quetiapine, risperidone, tamoxifen, and ziprasidone.
For maintenance in adults, the effective options were aripiprazole including its long acting injection, asenapine, divalproex or valproate, lithium, olanzapine, group psychoeducation added to treatment, quetiapine, and risperidone as a long acting injection. Group psychoeducation is structured teaching about the illness delivered in a group setting, and it earned its place on the list alongside medications.
A smaller group held up across every phase in adults, including aripiprazole, asenapine, cariprazine, cognitive behavioural therapy as an add-on, divalproex or valproate, lamotrigine, lithium, olanzapine, paliperidone, quetiapine, and risperidone.
Children and teenagers have far fewer proven options
The gap by age is stark. In adults, ten treatments cleared the bar for bipolar depression. In children and adolescents, two did: lurasidone, and olanzapine combined with fluoxetine. For mania, adults had thirteen options with support behind them, while children and adolescents had five: aripiprazole, asenapine, olanzapine, quetiapine, and risperidone.
Dr. Kumar’s Take
The most useful thing here is not any single drug name. It is the structure. Families come to me having read a forum post saying lithium is the answer, or that lamotrigine is the answer, without anyone asking which phase of the illness they are treating. This review makes that question unavoidable, because the lists really do change from depression to mania to maintenance.
The certainty numbers deserve attention too. Fewer than one in ten analyses reached high certainty. That does not mean the treatments do not work. It means the trials behind many of them were small, short, or inconsistent, and a future study could shift the picture. A drug appearing on one of these lists is a reason to have a conversation with your psychiatrist, not a verdict.
The thin evidence for children and teenagers is the part clinicians should sit with. Young patients are being treated every day with drugs whose pediatric evidence base is a fraction of the adult one. That is a call for more trials in this age group, not a reason to withhold care.
One more thing worth noting: this is a living review. The authors say it may be updated over the next two years as new evidence comes in, and the website is designed to reflect that.
Practical Takeaways
- Before discussing medication options, get clear with your doctor on which phase you are treating right now, because the effective treatments differ for depression, mania, and maintenance.
- Look up your own treatment on the free EBI-BD platform at ebibd-database.org, which publishes the full results including a tool covering 12 interventions and 17 safety outcomes.
- Ask your psychiatrist how certain the evidence is for the specific drug you are on, since most of the analyses in this review landed at low or very low certainty.
- If you are a parent of a child or teenager with bipolar disorder, expect fewer well-supported options and ask directly what evidence supports the plan being proposed.
Related Studies and Research
- CBT vs all other depression treatments: massive meta-analysis of 409 trials
- TMS for PTSD and depression: how brain networks heal from trauma
- Coffee health benefits: umbrella review of 67 meta-analyses
- Red light therapy for brain disorders: a comprehensive review
FAQs
What does a GRADE rating of “low certainty” actually mean for my treatment?
GRADE is a rating of how much confidence to place in a finding, not a rating of how well a drug works. A low or very low rating usually reflects problems in the underlying trials, such as small numbers of patients, short follow-up, or results that disagree with each other. A treatment with low certainty evidence may still help you a great deal. It simply means that if better trials are run, the estimate could move. Since 986 of the 2,510 analyses here were rated low and 461 very low, this describes most of the current evidence on bipolar treatment.
Why does a treatment work for mania but not depression?
Mania and bipolar depression are opposite poles of the same illness, and they respond to different biology. Several drugs on the mania list, such as haloperidol and ziprasidone, did not appear on the adult bipolar depression list at all. Others, including divalproex or valproate, quetiapine, and olanzapine, showed effects across phases. This is why a treatment plan built around one episode can fail when the mood swings the other way, and why maintenance is treated as its own separate question.
What is a “living” review and why does it matter here?
A normal systematic review is a snapshot. It is accurate on the day it is published and starts aging immediately. A living review is set up to be updated as new trials appear, and this one carries a note saying it may be updated over the next two years if more evidence emerges. Pairing that with a public website matters, because a reader checking the EBI-BD platform is meant to see the current state of the evidence rather than a fixed picture from an old publication date.
Are supplements and sleep-based treatments included?
Yes. Alongside the 74 medications, the review covered 13 nutraceutical treatments, 8 psychosocial therapies, 18 brain stimulation approaches, and 3 circadian rhythm based treatments. Being included is not the same as being effective, and the non-drug approaches that made the effective lists in adults were group psychoeducation for maintenance and cognitive behavioural therapy as an add-on across phases. Both were used as additions to ongoing treatment rather than replacements for it.
Bottom Line
This project pulled together 77 systematic reviews covering 116 bipolar disorder treatments and 2,510 separate analyses, then made the whole thing searchable for free. The headline is that effectiveness depends on the phase of illness and on age, with adults having many supported options and children and adolescents having only a handful. The second headline gets less attention but matters just as much: only 236 of those 2,510 analyses reached high certainty, so most bipolar treatment decisions are still being made on evidence that could change.
