Can Cannabis Help Agitation in Late-Stage Dementia?

Key Takeaways

• A 120-person trial called LiBBY tested an oral THC and CBD combination in people with late-stage dementia who were eligible for hospice care, and found meaningful reductions in agitation within two weeks [1].
• Close to nine in ten people receiving the treatment showed overall improvement by week 12, compared with roughly one in four on placebo [1].
• The results were presented at a conference in July 2026 and have not been published in a peer-reviewed journal, so the full methods and safety tables are not yet open to outside review [1].
• Nearly thirty years of earlier trials produced mixed and often negative results, and a 2026 review found the overall benefit disappeared once weaker studies were removed [2].
• The medicine used in the trial is not the same as anything sold in a dispensary, and the researchers who ran it said so directly [3].

---

If you have gone looking for whether cannabis might help someone with advanced dementia, you have probably met two answers. One says a trial has now proven it works. The other says there is no real evidence and you should stay away. Neither is right, and the space between them is where the useful information sits.

In July 2026, researchers reported that in a trial of 120 people with late-stage dementia, an oral combination of THC and CBD reduced agitation, with most of the benefit visible within two weeks. In a condition where almost nothing has worked, those numbers are striking.

The medicine that produced them is not for sale. It is in no dispensary, no physician can prescribe it, and it exists only inside a clinical trial. But cannabinoid products are on shelves in most states, so the real question is not whether you can buy something. It is whether anything you can buy bears any relationship to what was studied.

This article covers what the trial found, how it sits against thirty years of research before it, what separates the tested product from what is on the shelf, and what a careful attempt would look like for someone who tries anyway.

What the Trial Found

The trial, called LiBBY, enrolled 120 people with Alzheimer's disease or another dementia who were hospice-eligible or already receiving hospice care and had significant agitation [1]. People this ill are almost always excluded from dementia research, which is part of why so little is known about helping them.

If you are caring for someone in this situation, you already know agitation here is not simple restlessness. It includes pacing, calling out, moaning, hitting, and throwing things, and when someone has lost most of their speech, these behaviors become the only way left to signal pain, fear, or an unmet need [1]. It affects roughly half of people with dementia near the end of life, and more than a third still have symptoms while taking sedatives, opioids, or antipsychotics.

Participants received a purified oral THC and CBD combination or a placebo, twice daily for twelve weeks, with neither families, staff, nor clinicians knowing which. Agitation was scored by the caregiver on a standard 29-item scale [3].

At two weeks, treated participants' agitation scores had fallen 6.27 points further than placebo. By week 12 the gap widened to 8.23 points [1]. On a separate clinician-rated measure of overall change, 83.9% of treated participants improved at two weeks against 30.5% on placebo, and by week 12 the figures were 87.2% and 23.6% [1].

Side effect rates were similar, at 46.7% and 42.4%. Serious adverse events were roughly twice as common in the treatment group, 23.3% against 11.9%, though investigators concluded none were caused by the study medication [1]. In people near the end of life such events are expected, and that judgment is reasonable. It is also one outside researchers cannot yet check.

That is the single most important limitation. These results were presented at a scientific meeting rather than published in a peer-reviewed journal, so they have not been scrutinized the same way and the detailed methods are not yet available [1].

What the Researchers Said About Their Own Results

Trial investigators rarely spend their announcement warning people away from an entire product category. This team did.

Brigid Reynolds, a nurse practitioner at Georgetown University and one of the two lead investigators, stated that people should not assume products sold at dispensaries or online are equivalent to what was studied. She described the trial medication as formulated, manufactured, and given under close medical supervision, and said commercially available THC and CBD products vary widely in composition, quality, and dosing, which makes them "potentially ineffective or even harmful" [3].

Coming from the people who designed and ran the study, that carries considerably more weight than a legal formality tacked onto a press release, and it deserves the same attention as the effectiveness numbers.

Two Different Things That Sound Like the Same Thing

There are two separate tracks here, and confusing them is the most common mistake with results like these.

The first is drug development. The LiBBY product, a purified formulation delivering 2mg of THC and 100mg of CBD per dose and later double that, is investigational [1]. It needs a larger Phase 3 trial, an FDA application, agency review, and a scheduling decision before any doctor could prescribe it. That takes years and does not always succeed.

The second is the retail market. In states with licensed medical programs, CBD-dominant products containing little THC are widely available today.

What separates them has little to do with which cannabinoids are present, and almost everything to do with four things you cannot see from the outside: whether the label reflects what is in the bottle, whether one batch matches the next, whether dosing is monitored by a clinician, and whether anyone has checked the product against the person's other medications.

Those gaps are measurable, and what testing exists is not reassuring. Laboratory analysis of 105 hemp-derived topical products found that only 24% of those listing a CBD amount were labeled accurately. THC turned up in 35% of them, and among those, 51% made no mention of THC anywhere on the label while 11% were labeled THC free [4]. Testing of CBD extracts sold online for oral use has found the same pattern of unreliable labeling [6].

The doses involved make that gap wider still, because at full strength LiBBY participants received 400mg of CBD daily. Reaching that with retail products means consuming a lot of product, and any undisclosed THC scales up alongside it.

Thirty Years of Trials Came First

LiBBY did not arrive in a vacuum, and what came before was not encouraging.

The first study appeared in 1997, and it was not about agitation at all. Researchers gave dronabinol, a synthetic THC, to 15 Alzheimer's patients refusing food, hoping to improve appetite. Weight went up, and they also noticed disturbed behavior seemed to decrease [6]. That side observation launched everything that followed.

The next two decades did not go well for the idea.

In 2015, a carefully conducted trial in Neurology gave 50 people with dementia low-dose THC or placebo for three weeks. The groups did not separate on any measure, including agitation [7].

In 2019, a Canadian team tested nabilone, another synthetic THC relative, in 39 patients. Agitation did improve, but the surrounding picture was complicated. Overall clinical impression did not reach statistical significance, sedation occurred during 45% of treatment periods against 16% on placebo, and on one of the two cognitive tests patients did measurably worse on the drug [8]. The authors recommended close monitoring for sedation and cognitive decline.

A 2021 Cochrane review, which applies unusually strict standards for which trials it will accept, found only four that qualified, covering 126 people in total. It concluded that whether cannabinoids help or harm in dementia could not be determined, and that any benefit may be too small to matter clinically [9].

The largest trial before LiBBY was ambiguous too. Published in 2026, it gave dronabinol to 75 people with Alzheimer's agitation and set two co-primary measures of success. It met one, missed the other, and no secondary measure separated the groups [10].

Then in 2026, a review pooling ten trials and 328 participants found an apparent benefit for agitation. When the authors removed the studies at high risk of bias, the effect was no longer statistically significant [2]. That is the evidence LiBBY entered.

Why This Trial May Have Gone Differently

Almost every earlier study tested THC alone, or THC and CBD in equal amounts, while LiBBY used roughly fifty times more CBD than THC [1]. Laboratory work suggests CBD may act as a brake on THC rather than an amplifier, which could preserve the behavioral effect while reducing sedation, though that comes from cell studies rather than LiBBY and no trial has tested the ratio directly. The other difference is who was enrolled: people with severe, late-stage disease, precisely where the pooled evidence had already suggested cannabinoids work best [2].

What Is Actually Approved Right Now

Two medications carry FDA approval for agitation associated with dementia due to Alzheimer's disease.

Brexpiprazole, sold as Rexulti, was approved in May 2023 and was the first [11]. It is an atypical antipsychotic, a class carrying meaningful risks in older adults with dementia.

Dextromethorphan-bupropion, sold as Auvelity, was approved on April 30, 2026, and is the first approved treatment for this condition that is not an antipsychotic [12].

Neither was tested in a hospice-eligible population, so neither has been compared directly with what LiBBY studied. Comparing point scores across separate trials is not valid, because the patients, the disease severity, and the placebo responses all differ.

For the full evidence review behind this article, see the companion Hytiva Research article, Cannabinoids for Agitation in Advanced Dementia: A Positive Phase 2 Signal and Its Regulatory Limits.

What Separates a Careful Attempt From a Reckless One

Nothing above amounts to a recommendation. No trial has shown that a dispensary product reduces agitation in dementia, and the honest position is that the evidence does not support trying one.

People try anyway, usually out of exhaustion and usually when nothing else has helped. If that is where you are, the difference between a careful attempt and a reckless one is real and worth knowing.

Ask for a Certificate of Analysis, and treat the answer as the answer. A COA is batch-specific laboratory testing showing what is actually in the product. Given how often those labels turn out to be wrong [4, 5], it is the single most useful thing you can ask for. A shop that cannot produce one for the specific batch has told you what you need to know.

Tell the care team before, not after. Hospice and palliative teams manage complicated medication regimens, and a cannabinoid added quietly undermines that. Most will not be shocked, and many have seen it before.

Have the interactions checked. Cannabinoids inhibit a liver enzyme called CYP2C9 at realistic doses [13]. Warfarin and phenytoin both depend on it. A pharmacist can review this in a few minutes.

Watch for sedation specifically. It is the most consistently reproduced cannabinoid effect in this literature. In a frail, elderly person, sedation is not just drowsiness: it raises the risk of falls and aspiration, both serious in someone already near the end of life.

Frequently Asked Questions

Can I buy something at a dispensary that does the same thing? No product on the market has been shown to do what the trial product did, and the investigators said directly that commercial products may be ineffective or harmful [3]. Even an item with a similar ratio differs in verified contents, batch consistency, supervised dosing, and screening against other medications. If you are weighing one anyway, see the section above.

If it worked that well, why can't we get it? Because it is investigational, not approved. It needs a larger confirmatory trial, FDA review, and a scheduling decision before it could be prescribed. That takes years, and promising Phase 2 results do not guarantee a drug clears Phase 3.

Could it interact with my family member's other medications? Yes, and this applies to over-the-counter CBD too. Cannabinoids inhibit the liver enzyme CYP2C9 at realistic doses [13], and warfarin depends on it. Anyone on a blood thinner, an anti-seizure drug, or several psychiatric medications should have a pharmacist review it first.

Does this mean cannabis treats dementia? No. The trial measured agitation, not memory, thinking, or disease progression. Nothing here suggests cannabinoids slow dementia, and one earlier trial found cognition got worse on a cannabinoid [8]. Reducing distress is worthwhile on its own, and it is a different goal from treating the disease.

Would this help someone in the early stages? There is no evidence for that. Every participant had late-stage disease and was hospice-eligible, and the broader research suggests the effect concentrates in severe dementia rather than mild [2]. Applying this result to someone recently diagnosed goes well beyond what was studied.

The Bottom Line

A well-designed trial found that a specific, carefully manufactured THC and CBD combination reduced agitation in people with late-stage dementia, and the size of that effect is notable in a condition where almost nothing has worked.

Three things temper it, beginning with the fact that these results have not been peer reviewed. Thirty years of prior research was inconsistent enough that a 2026 analysis found the benefit vanished when weak studies were excluded. And the product tested is unavailable and unlike anything you can buy.

If someone you care for is struggling with agitation in advanced dementia, the useful next step is a conversation with their hospice team or prescriber about what is approved, what is appropriate, and what to watch for. That is worth doing regardless of where this medicine ends up.