Music Therapy in Palliative Care: What It Is For, and Why It Is Hard to Measure

Music therapy in palliative care is delivered by a registered therapist to people approaching the end of life and to the people around them, and the reasons hospices employ it are largely reasons that do not sit on an outcome scale. That is not a way of excusing thin evidence. It is a statement about what the work is for, and it is the beginning of a real methodological problem rather than a way around one.

I have a partial claim to first hand knowledge here and I want to set out how partial it is. A family in one of the care homes where I sing asked me to sing to their mother in her last weeks. I went four times. She had been in my Tuesday group for two years and by then she was in bed and mostly not speaking, and I sang the things she had liked in the group, quietly, for twenty minutes at a time, with a daughter in the chair by the window. On the third visit the daughter joined in, and afterwards told me she had not been able to think of anything to say to her mother for a fortnight. That is what I have. I am not part of any palliative care team, I have never watched a hospice music therapist work, and I have never sat in a meeting where this work was planned. Everything else here comes from the profession’s published description of its own work and from the reviewer who checks the page, and I would rather say so than write a scene I have not been in.

Where this page makes a claim about effect it carries a label: Tier 1 supported, Tier 2 promising but limited, Tier 3 practice and anecdote. Tier 3 covers absent evidence and null evidence, which are different things, and the page says which one it means. The reasoning behind all of it is on reading arts in health research.

Where the work sits

Palliative care music therapy happens in three settings that look quite different from each other: hospices, hospital palliative care teams, and people’s own homes in the community.

In a hospice the therapist is usually part of the establishment, employed or contracted, attending team meetings, reading the notes and taking referrals from nurses, doctors and social workers as well as from families. In a hospital, palliative care is typically delivered by a team that moves around the wards rather than occupying one, so a therapist attached to that team works in whoever’s bay the patient happens to be in, with the curtains as the only privacy available. In the community the therapist visits at home, which offers the most privacy and the least control: a living room with a television on, a dog, and whoever is in the house that afternoon.

Two consequences follow. The work has to be portable and adaptable in a way a clinic based practice does not, which is one reason live music matters more here than recordings. And provision is extremely uneven, because a hospice’s music therapy post depends on that hospice’s fundraising rather than on any national arrangement. With 6,103 arts therapists on the HCPC register across all four protected titles in July 2026¹, and music therapists forming part of that total, the number of hospices with a therapist on the staff is limited by the size of the profession before anybody reaches the question of budgets. How to find a music therapist sets out the registers.

What the work actually is

The word usually attached to this work is comfort, and it is too small a word. The published description of the profession’s own practice covers a set of tasks that are more various and more specific than that².

Legacy work. Recordings, songs written with a patient for particular people, sometimes a message spoken over music, sometimes a lullaby for a grandchild too young to remember. This is one of the few pieces of work in health care explicitly made for people who are not the patient and will receive it after the patient has died. It also has a deadline nobody can predict, which shapes how a therapist paces it.

Breath, pacing and physical settling. Live music can be matched to somebody’s breathing and then moved, slightly, and it is one of the few things available at a bedside that can be adjusted second by second for a person who cannot say what they want. That is a description of a technique rather than a claim about an outcome, and it is a large part of why the work is live rather than recorded.

Work with a family who cannot talk to each other. Rooms at the end of life are frequently silent for reasons that have nothing to do with peace. Music gives people something to do together that is not conversation, and a reason to be in the room that does not require them to have thought of anything to say. The daughter by the window in my four visits is the only version of this I have witnessed, and the profession describes it as routine.

Work with the family after a death. In many services the therapist’s involvement does not end when the patient does. Bereavement work with partners, and particularly with children and siblings, is a recognised part of the role, and one of the clearest signs that the patient is not the only client.

The sessions where nothing is played. A significant proportion of contacts involve no music: an assessment, a conversation, a visit where the person is asleep and the therapist sits with the relative instead, a session where the patient wants silence and the skill is in providing it without leaving. A service counting sessions delivered records those identically to the others, which is a small illustration of the measurement problem below.

Around all of it sits the same apparatus that defines the profession anywhere: assessment, goals, notes in the clinical record, supervision, and a place in the team’s meetings. What happens in a music therapy session describes that framework in general, and music therapy covers the profession as a whole.

Who the client is

In most of health care the person being treated is obvious. Here it frequently is not, and that ambiguity is a feature of the work rather than a looseness in it.

A single referral can produce work with the patient alone, with the patient and their partner together, with a child who is frightened of the room, with siblings who disagree about what is happening, and with a widow eight weeks later. The therapist has to decide, repeatedly, who the session is actually for today, and that judgement changes as the person’s condition does: work beginning with a patient who can sing often ends with a family who need something to hold.

That has an obvious consequence for research, which the next section takes up, and a less obvious one for how services are described: a hospice reporting the number of patients seen by its music therapist is undercounting the work by a wide margin and has no straightforward way to count it correctly. It also bears on consent, since a family member who becomes a recipient of the work has not usually been referred and may not realise they have become one. The broader shape of creative work with people who are dying, including the parts that involve no therapist at all, is on creative work at the end of life.

Why measurement is genuinely hard here

Palliative care is one of the hardest settings in medicine to run a trial in, and the difficulties are structural rather than a matter of effort. They are worth setting out one at a time, because “hard to measure” is otherwise indistinguishable from an excuse.

Attrition through death. In most trials, people who drop out are a nuisance. Here they are the outcome, and they are not lost at random: those who complete a follow up at six weeks are systematically less unwell than those who do not. Any analysis restricted to completers is therefore an analysis of a different population from the one recruited, and the statistical repairs for this are imperfect and contested.

The timing of measurement. A trial has to specify when the outcome is assessed, and in a population whose condition can change within a day, any fixed interval is arbitrary. Measuring anxiety at two weeks captures one person on a settled afternoon and another six hours after a deterioration. Measuring immediately after a session captures something real and short lived, and the field’s reliance on such measures is one reason its findings are hard to interpret.

The consent problem. Recruiting people who are exhausted, in pain, frightened or sedated raises real ethical difficulty, and ethics committees are properly cautious about it. So the people who end up in studies are disproportionately those well enough to be approached and to consent, who are not the people the service exists for.

Outcomes that do not sit on a scale. Much of what this work is for has no validated questionnaire behind it. A family who managed to say something to each other. A recording a nine year old will still have in twenty years. A death the people present remember as settled rather than frantic. Researchers measure anxiety, pain and quality of life instead, which are measurable and not quite the thing, and a trial finding no change in an anxiety score has not established that nothing of value occurred.

The impossibility of blinding. Nobody in the room is unaware that a musician is present: not the patient, not the family, not the nurse recording an observation. Expectation effects cannot be separated from the intervention, and the usual solution, an active comparator that looks similar, is difficult to design and hard to justify to a dying person’s family.

None of that means the work cannot be studied. It means the studies that exist are small, short, difficult to pool, and prone to exactly the weaknesses that reduce a certainty rating. Understanding why is the difference between reading a thin literature as a verdict and reading it as a description of how hard the question is.

The evidence position, stated exactly

There is no effect size for music therapy in palliative care on this page, and that is a decision rather than an omission.

For one to appear, this site would need a systematic review or an adequately powered randomised controlled trial of music therapy in a palliative care population, reporting the design, a named comparator, a stated primary outcome, the number of participants, the follow up interval, a certainty rating and the year. All seven. Nothing of that description is locked here, so no number appears, and none will be borrowed from an adjacent population and relabelled.

The honest label is therefore Tier 3, of the not adequately studied kind, not the null kind. Nobody has produced the evidence that would settle the question. That is a statement about the research literature and about how difficult the setting is, and it is emphatically not a finding that music therapy does nothing for dying people. Reading an empty file as a negative result does more damage here than almost anywhere else, because the people it lands on have no time to wait for better research and are being asked to choose now.

A good deal can still be said without a number. The profession’s own published account of the work is a legitimate description of practice and is used on this page as exactly that. Hospices employ therapists on the basis of what their teams observe, which is testimony rather than evidence of effect, and testimony from experienced clinicians is worth having as long as nobody dresses it up as something else.

The nearest locked evidence, and what it is not

The closest thing this site holds to relevant controlled evidence is the 2021 Cochrane review of music interventions in cancer care, and it has to be labelled as cancer care every time it is mentioned here.

That review covers 81 trials and 5,576 participants. It reports anxiety 7.73 STAI-S units lower and pain at SMD -0.67, and it rates both findings very low certainty³. Both are Tier 2 on this site: controlled evidence exists and it is not strong enough to lean on.

Three qualifications matter before anybody carries those figures across. Cancer care and palliative care are overlapping populations, not the same one: many people receiving palliative care do not have cancer, many people with cancer are not receiving palliative care, and the trials pooled were largely conducted around diagnosis and treatment rather than at the end of life. The question is different too, since anxiety around a procedure is not what a hospice referral is asking about. And the interventions pooled include a great deal of recorded music delivered around treatment, which is not a registered therapist’s work at a bedside.

There is also a lesson inside those numbers that has nothing to do with palliative care, and it is the most useful thing on this page for reading anything else. Eighty one trials and 5,576 participants still came out at very low certainty. Volume is not strength. Certainty ratings ask how far a pooled estimate can be trusted given how the individual trials were run, and eighty one small, unblinded, inconsistent trials produce a precise looking average built out of weak material. The parallel case in the other modality, where the certainty problem has the same shape, is on art therapy during cancer treatment.

Consent and capacity at the end of life

Consent here is the hardest version of a problem that runs through all of arts in health, and it is treated as a check made in every session rather than as a decision taken once.

Somebody who is drowsy, in pain, frightened or on medication affecting their thinking may be able to agree on Monday and not on Wednesday, and agreement to a course of work does not license any particular visit. So the therapist reads the response in the room and stops on any indication of refusal, which at this stage of illness is far more often physical than verbal: a turned head, a hand raised, a change in breathing, eyes closing in a way that is not sleep. Treating those as things to work through rather than as answers is the failure mode, and its motive is usually benign.

Where capacity is absent, the work proceeds on a best interests basis with the people close to the patient involved, and it is revisited rather than banked. Two further things need care. Recording anybody, which is central to legacy work, is a separate decision from consenting to a session, and it involves a person who will not be there to change their mind later. And a family member who has become a recipient of the work has not been referred and should be told gently that this is what is happening, rather than discovering it afterwards. The general framework is on consent and safeguarding in arts in health.

How to ask, and what a hospice can realistically offer

Ask the hospice or the palliative care team directly, ask early, and be specific about which of the several possible pieces of work you actually want.

Early matters more than people expect. A therapist can do more with three weeks than with three days, particularly for anything involving recording or songwriting, and referrals here are frequently made at the last possible moment because nobody thought of it sooner. Specific matters too: a recording for grandchildren, help for somebody frightened at three in the morning, and support for a family who cannot talk to each other are three different jobs, and naming the one you have in mind gets a better answer than asking for music therapy in general.

Be prepared for the answer to be that there is no therapist. Many hospices do not have one, some buy in a few sessions a week, and some have a post funded by a legacy that will run out. Where the service exists it is normally free to the patient and the family, and the constraint is availability rather than money. If there is no music therapist, ask what else the hospice has, since chaplaincy, complementary therapies and volunteer musicians exist in some services and are different things offering different amounts. The equivalent question inside a care home, where a resident may be dying without any hospice involvement, is on arts in care homes.

One last thing, and it is what I would want said to my own family. Nothing here treats an illness or relieves a symptom in place of the drugs that relieve it, and asking for a musician is not an alternative to asking for a doctor; no good hospice would present it as one. What the work appears to be for, on the profession’s own account, is the part of dying that is not a medical problem: the room, the people in it, and what gets said or sung or left behind. I cannot give you a number for that, and I have seen enough of the edge of it to be unwilling to pretend the missing number means it is not there. The same discipline applied to a far better studied question is on music therapy for dementia, which is what a locked answer looks like when one exists.