Social Prescribing Explained: What a Link Worker Can and Cannot Arrange

Social prescribing is a referral to a link worker, whose job is to connect somebody with non clinical support that already exists in their area. The link worker does not create provision, does not assess anybody clinically, and cannot conjure a service that is not there. Understanding those three limits explains most of what happens, and most of what fails to happen, when somebody is referred.

The conversation that shaped how I think about this was with a link worker who rang about places in a singing group I run. Once we had sorted out the practicalities I asked her, out of curiosity, roughly what proportion of her referrals were to anything creative. She thought about it and said maybe one in twelve. The rest was debt advice, housing, a lot of loneliness, exercise referrals, and helping people fill in forms. She was not being dismissive about the arts; she came to a session later and stayed for the whole thing. She was describing a job that the sector’s own literature does not much resemble, in which creative activity is a small and well liked corner of something much bigger and much more practical.

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 a link worker actually does

They have a conversation, work out what somebody needs that is not medical, and make a connection.

In England the role sits within primary care networks, so a link worker is generally attached to a group of practices rather than to a single surgery, and the National Academy for Social Prescribing is the body that describes what the work involves and what it is meant to achieve¹. The two things a link worker has that a clinical appointment does not are time and local knowledge. A first conversation is typically much longer than a GP appointment, and can happen more than once.

What they are not is a clinician. They do not diagnose, do not assess capacity or risk in a clinical sense, and do not treat anything. Nor are they therapists, and it is worth being explicit about that, because the conversation can be a personal one and people reasonably wonder. If somebody needs psychological therapy, the link worker’s job is to say so and point at the route, not to provide it.

What is on the other end of the referral

Mostly things that are not the arts, which is the part the creative health literature tends to skip.

A working list of what link workers commonly connect people with: benefits and debt advice, housing and homelessness services, food support, help with forms and appeals, exercise referral schemes, weight management, befriending and loneliness projects, walking groups, gardening, men’s sheds, carers’ support, volunteering, digital skills, and support groups for specific conditions. Creative activity is one item on that list.

That matters for a reader in a practical way. If you ask for social prescribing because you want a singing group, you will be offered a conversation that may quite reasonably start somewhere else, particularly if there is something more pressing in the way. That is the system working rather than failing. It also means the link worker is the right person to ask about several problems at once, which is not true of most people you will speak to in a health service.

Where arts on prescription fits

Arts on prescription is what happens when the connection made is to a creative course, and it has a recognisable shape.

The 2024 systematic review of the field found that most of its 25 included programmes ran for 8 to 10 weeks, in community settings, arts venues, GP surgeries, primary healthcare settings and in one case a school, across Australia, Denmark, Sweden, the United Kingdom and the United States². Painting, ceramics, singing, writing, photography, gardening and museum visits all appear under the heading. Sessions are usually weekly and a couple of hours long.

The fixed ending is a design feature of that model rather than an accident, and it is the thing participants most often find hard, because the ending tends to arrive at about the point where the group has become the reason to leave the house. What a ten week arts course involves goes through the shape of one week by week, and arts on prescription covers the route as a whole. Some schemes are organised around a venue type, such as museum prescriptions; some around a condition, such as singing for lung health.

What a link worker cannot do

Four limits, and knowing them saves a great deal of frustration.

They cannot create provision. If nothing exists in your area, referral does not produce it. This is the limit people meet most often, and it is much more common in rural areas, in places where a grant has recently ended, and for anything specialised.

They cannot arrange a registered arts therapy. 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 those four titles³. Access to them runs through a clinical referral or through private practice, not through social prescribing. See how to find a music therapist and art therapy.

They cannot prescribe anything. The word is a metaphor that stuck. Nothing is written on a prescription pad, nothing is dispensed, and there is no entitlement attached, no waiting time target and no national standard for what a scheme must offer.

They usually cannot override eligibility criteria. Where a place is free, the criteria belong to whoever is funding it: an age band, a postcode, a diagnosis, a referral source. Those rules come from the grant rather than from the activity, which is why they can seem arbitrary from the outside. Who pays for arts on prescription explains where they come from.

The ask that works

Specificity is the whole of it.

Asking a GP whether there is anything creative available produces goodwill and nothing else, because it is not a question they hold an answer to. Asking to be put in touch with the social prescribing link worker attached to the practice, by that name, produces a phone call. Many services also accept self referral, so ringing the practice and asking directly is legitimate and often faster than waiting to be sent.

Three further things worth asking once you have the conversation. What is running now, rather than what is listed. When the next block starts, since courses tend to run in fixed intakes rather than rolling admission and the wait is usually for the next start date. And how long the scheme is funded for, because that determines whether the group you join will still exist in a year. How to get referred to an arts scheme sets out the exact wording and what to do when the answer is that there is nothing.

What the evidence supports, and what it does not

For the arts half of social prescribing, the position is that the question has not been answered rather than that it has been answered badly.

The 2024 systematic review screened 7,805 records and included 25 studies: 10 qualitative, 7 mixed methods and 8 quantitative studies using uncontrolled before and after designs. In its own words, no randomised controlled trials were identified in the search. The quantitative studies were rated medium quality, most commonly for an absence of comparator groups, and exactly one included a treatment as usual comparison. Its pooled analysis found a statistically significant improvement in wellbeing, resting entirely on uncontrolled designs². That is Tier 3.

An uncontrolled before and after design cannot separate the activity from time passing, from regression to the mean, from who chose to take part, or from who stopped attending before the second questionnaire. People are referred at a low point, and low points are usually followed by better ones whatever happens next.

Two things need saying next to that, pulling in opposite directions. Tier 3 is not a verdict that it does not work: the reasons the trials do not exist are structural, since these programmes are hard to randomise, run on short grants, and are delivered by organisations whose money goes into delivery. And the qualitative half of the literature is not nothing. Social connection, psychological benefit and routes onwards were consistent themes across the included studies², and people are reliable witnesses to their own experience even where a design cannot prove causation.

The economic claims are a separate matter and they are weaker still. The figures in circulation trace to one uncontrolled evaluation of one Gloucestershire scheme, written in December 2011, which analysed 90 patients out of roughly 500 referred, had no control group, called itself “a simple observational study” and stated that it “does not imply causality”⁴. The full arithmetic is set out on does arts on prescription save money.

What it is reasonable to expect

A conversation, a connection, and no guarantee.

Reasonable: somebody with time who takes the whole of your situation rather than one symptom, knowledge of what exists locally, help getting through a door, and often a follow up call to see whether it worked. Those are worth having and they do not depend on any research finding.

Not reasonable: a treatment, an entitlement, a service that does not exist, or a promise about a clinical outcome. And not reasonable in the other direction either, which is worth saying because scepticism about the evidence sometimes turns into contempt for the provision: people who attend these courses generally value them, and a link worker who finds somebody a Tuesday afternoon they look forward to has done something real.

The sector’s own reports and case studies are gathered in one place⁵, and they are worth reading as descriptions of what is happening rather than as evidence that it works, a distinction worked through on creative health policy and reports. Nothing on this page is a reason to change, delay or decline any part of anybody’s treatment, and no community activity substitutes for one.