Creative Activity in Dementia Care: What the 2025 Review Changed, and What to Expect

The 2025 Cochrane review of music based therapeutic interventions in dementia reports moderate certainty evidence of no effect on agitation and aggression, which is the outcome creative activity in dementia care is most often claimed to address. The figure is SMD -0.05, 95% CI -0.27 to 0.17, across 30 studies and 1,720 participants randomised¹. That is a positive finding of absence rather than a gap in the research, and it should change how anybody running or buying a programme describes it.

I need to be careful here, because I run these sessions and I have watched the thing the review says does not happen. A room that was restless settles. Somebody who has been calling out stops calling out. I have seen it often enough that my instinct disagrees with the number, and I want to say plainly why I do not trust my instinct. I am in the room for an hour a week. I do not see the two hours afterwards. I do not see the mornings when the same person was calm without me. I chose to be there, I want it to work, and the staff who tell me afterwards that it was a good afternoon know that too. Everything about my position is designed to produce the impression I have. That is exactly what a controlled trial exists to correct, and this is the page where I take the correction rather than argue with it.

Where a claim about effect appears below it carries a tier (Tier 1 supported, Tier 2 promising but limited, Tier 3 practice and anecdote), explained on reading arts in health research.

What the review reports

Four findings, and each needs its own sentence.

Agitation and aggression: SMD -0.05 (95% CI -0.27 to 0.17), moderate certainty. Tier 1, null result. The interval is narrow, it is centred essentially on zero, and the certainty rating is moderate rather than low. Read together, that is evidence that music based interventions do not meaningfully change agitation, not a statement that the question is open.

Depressive symptoms: SMD -0.23 (95% CI -0.42 to -0.04), moderate certainty. Tier 1. A real effect, and a small one: roughly a fifth of a standard deviation. Worth having. Not transformation.

Anything persisting four weeks after treatment ends: no evidence of one. Tier 3.

Quality of life: no separate estimate exists in the review¹. Any quality of life figure attributed to it has come from somewhere else.

The review covers 30 studies and 1,720 participants randomised and was published on 7 March 2025. Anything citing 22 studies and 1,097 participants is quoting the superseded 2018 version, which reached different conclusions on agitation, and that includes sources published well after the update.

Why the null is the important number

Because it is the claim the sector makes most, and because a moderate certainty null is a strong statement rather than a weak one.

Agitation, restlessness and distressed behaviour are what care settings most want help with, and they are what activity programmes are most often sold as addressing. A brochure that says music calms residents is making a claim about agitation. A commissioning case built on reducing incidents is making a claim about agitation. A funding bid citing behaviour management is making a claim about agitation.

The distinction that matters is between a wide interval at low certainty and a narrow interval at moderate certainty. A wide interval at low certainty means nobody knows. SMD -0.05 with an interval of -0.27 to 0.17, at moderate certainty, means the reviewers looked at a substantial body of randomised evidence they broadly trusted and found the effect to be about nothing. Describing that as more research is needed misdescribes it.

For contrast, this is what genuinely uncertain looks like: the 2021 Cochrane review of music interventions in cancer care pooled 81 trials and 5,576 participants and rated its findings very low certainty², which is Tier 2. Lots of studies, little confidence. The dementia agitation result is the opposite shape: reasonable confidence, in a finding of nothing.

What the evidence does support

A small effect on low mood, and a strong implication about how often sessions should run.

The depressive symptoms result is real. SMD -0.23 is small by convention and it is not nothing, and it comes with an interval that stays on the beneficial side of zero and a moderate certainty rating. That is a properly supported claim and it is the one worth making.

The finding about persistence matters more for planning than for advocacy. No evidence of any effect persisting four weeks after treatment ends is an argument against the block of six sessions delivered by a visiting project and then withdrawn, and an argument for regular continuing provision. It also means a home cannot bank a benefit: whatever the sessions do, they appear to do it while they are happening.

That has a practical shape. Weekly, ongoing, small, and person specific beats monthly, occasional, large and generic. It also means one to one work at a bedside, which never appears in a programme and never photographs well, counts as provision.

The citation that has to stop

NICE does not recommend music therapy for dementia, and a full text search of NG97 returns zero occurrences of music, art, dance, drama or creative activity³.

It is the most commonly repeated false citation in this field. It appears in home brochures, funding bids, service descriptions, training slides and press coverage, and the people repeating it are not being dishonest. They have received it from a source that looked reliable.

Two balancing points, because a bare correction gets misused. Absence from a guideline is not a finding that something is ineffective; it is a statement about what the guideline covers. And there is real evidence about music in dementia, set out above, which comes from a Cochrane review rather than from NICE. The right move is to cite the review and stop citing the guideline. How false citations propagate, and the six checks that catch them, is on creative health policy and reports.

While on figures that drift: the WHO dementia fact sheet, updated 3 July 2026, gives 57 million people worldwide living with dementia in 2021 and carries no 2050 projection⁴. Projections exist and belong to specific documents, and attributing one to the current fact sheet is wrong.

What this does not license

Three inferences that would be as wrong as the overclaim, and the third is the one this page is most at risk of provoking.

It does not mean creative activity in dementia care is worthless. The review tested music based therapeutic interventions on specific outcomes. It did not measure whether an afternoon with something in it is better than an afternoon without, whether relatives find visiting easier when there is something to do together, or whether staff have a better shift. Those are real and they are not in the trial.

It does not transfer to every activity. The trials covered structured music based interventions, many delivered by trained practitioners in research conditions. A generic conclusion about art, dance, poetry or crafts cannot be read off them, in either direction. The commonest error in this field is transferring a finding about one thing to a different thing with a similar name: see music therapy against community music.

It does not mean stop the singing group. A small real effect on low mood, plus the ordinary fact that people enjoy it, is enough reason to run one. What it means is describing it accurately: not as behaviour management, not as a treatment, and not as something that will still be working next month if it stops next week.

What to say instead, if you run or commission one

Language that is both honest and defensible, which is a narrower target than it sounds.

Say: people enjoy it; a day with something in it is better than a day without; the current evidence supports a small effect on low mood; there is no evidence of benefit persisting once sessions stop, which is why it runs continuously; it is an activity rather than a treatment.

Do not say: it reduces agitation; it improves quality of life, on the strength of that review; NICE recommends it; it reduces the need for medication; a block of sessions has achieved a lasting change.

If a family asks whether it will help, the honest answer has three parts: this is what the research supports, this is what it does not, and this is what I see in the room, which is not evidence. People take that answer better than the sales version, in my experience, and it does not have to be withdrawn later.

Who delivers it, and the distinction that carries the evidence

The trial evidence describes structured interventions, and most of what happens in care homes is something else.

Music therapy, art therapy, art psychotherapy and dramatherapy are clinical interventions delivered by practitioners holding titles protected in UK law, and in July 2026 the Health and Care Professions Council register listed 6,103 arts therapists across all four⁵. That is a very small workforce and most homes have never had one.

What most homes have is an activities coordinator, visiting artists and volunteers, and in England that provision sits within the home’s wider regulated obligations rather than being inspected as clinical care⁶. Both halves can help somebody. Only one is regulated, and a finding established about one is not evidence about the other. Arts in hospitals covers the same split inside acute settings, arts in care homes covers who provides what and the questions a family should ask, and music therapy for dementia sets out the clinical side of the same evidence.

Two related pages cover activities that come up constantly in this setting. Personalised playlists for dementia covers the headphones intervention and why it is not music therapy. Dementia choirs and singing groups covers community singing and what has actually been measured about it.

Consent, which is not a formality here

The people this page is about are frequently unable to consent to being described, photographed or recorded, and that constraint is live rather than procedural.

Capacity is decision specific and can fluctuate. Somebody may be able to decide whether to join in and unable to decide whether to appear on a website indefinitely, and enjoying a session is not consent to being filmed. A relative’s agreement does not reliably extend to publication. The clip everybody wants to share is very often the one thing that should not be shared, and where there is doubt the answer is not to publish. That binds this site’s own imagery as much as anybody else’s, and the full account is on consent and safeguarding in arts in health.

Nothing on this page is a reason to change, delay or decline any part of anybody’s care or treatment, and no creative activity substitutes for either. If somebody’s distress is increasing, that is a clinical conversation with the GP or the dementia team, not a reason to book more sessions.