Health and Community Cannot Just Refer to Each Other

A GP refers a patient to a walking group run by a community charity. Six months later, funding for the walking group ends, nobody at the practice notices, and the referral pathway keeps sending people to a group that no longer exists. This is not a hypothetical failure. It is the default outcome of treating community organisations as a delivery arm of the health system rather than as a partner in it.

Social prescribing has grown fast because it offers something clinical services cannot: a route into ordinary, non-medical group life for people whose isolation is a bigger problem than anything a prescription pad can fix. The systematic reviews on social prescribing are consistent on two points. People who go report real benefits — self-esteem, self-confidence, a sense of purpose — and the trial evidence underneath those reports is thin and heterogeneous. A 2021 review found all nine included studies reported positive effects for participants, with three showing reduced use of GP, emergency, or inpatient services. A parallel review that year reached the same broad conclusion. A qualitative synthesis published in 2022 dug into why, finding that participants describe benefit that goes beyond simple social contact toward restored meaningful participation — belonging to something, not just attending it.

That last finding matters more than it looks. It means the mechanism is not “get lonely people into a room.” It is “get lonely people into something with enough structure and purpose that they want to keep showing up.” That’s a much higher bar, and it’s a bar community organisations clear far better than health services do — if they’re treated as partners with expertise, rather than as capacity to be booked.

Why the referral model breaks

Most social prescribing infrastructure is built as a one-way pipe: link worker identifies need, link worker refers, community activity absorbs the person. The pipe assumes the activity on the other end is stable, funded, and has capacity. In practice, community organisations running the walking groups, men’s sheds, and choirs that social prescribing depends on are often small, grant-funded year to year, and invisible to the commissioning process that sends people their way.

Call this the subcontractor problem: health systems treat community groups as a service they can specify, refer into, and evaluate, without any obligation to sustain them, understand their constraints, or share what they learn. The relationship runs one direction. Referrals flow out; nothing flows back.

Two things follow from this, and both are avoidable.

First, capacity mismatches go undetected until they’ve already failed a patient. A link worker keeps referring to a group that closed, is oversubscribed, or never suited the people being sent — because no one asked the organisation running it what it can actually absorb.

Second, community organisations have no reason to trust the arrangement. They see people referred in with complex needs and no clinical handover, absorb the burden of managing that, and get no resourcing or recognition for it. Eric Klinenberg’s argument about social infrastructure is relevant here: libraries, community centres, and third places do the quiet work of holding communities together, but that work is chronically undercapitalised relative to its value. Referral pipelines that treat this infrastructure as free capacity make the undercapitalisation worse, not better.

What a genuine partnership requires

The UK’s 2018 loneliness strategy was the first national government strategy to name this explicitly, funding social prescribing while also acknowledging that the community and voluntary sector delivering it needed direct investment, not just referral volume. That two-sided commitment is the part most local implementations quietly drop.

A partnership model, as opposed to a referral pipe, has four features a referral pipe doesn’t:

1. Two-way information flow. Community organisations tell health partners what they can take on, when capacity is full, and what kinds of need they’re not equipped for. This has to be a standing conversation, not an annual contract renewal.
2. Shared risk on funding. If a health system’s referral targets depend on a community group’s survival, that group’s funding stability is the health system’s problem too. Treating it as someone else’s grant application to worry about is how the walking group above disappears without anyone upstream noticing.
3. Clinical handover that respects non-clinical settings. Community organisers are not clinicians and shouldn’t be expected to manage clinical risk. Referrers need to be explicit about what a group is and isn’t equipped to hold, particularly for people with complex mental health or medical needs.
4. Joint interest in what’s actually working. Not just referral counts, but honest conversation about which activities produce the sense of participation the 2022 synthesis identified as the real mechanism, and which are just attendance-taking.

What this means in practice: before you refer into a community activity, find out — directly, from the organisation running it — what its funding runway is, what its capacity ceiling is, and what kind of participant it’s not built to support. If you don’t know the answer to those three questions, you don’t have a partnership. You have a pipe, and pipes leak.

Where the evidence is solid and where it isn’t

Claim	Evidence status
Social isolation is a serious health risk, including cardiovascular risk	Strong — the American Heart Association’s 2022 scientific statement puts the increased risk of heart attack, stroke, or death at around 30%
Older adults are disproportionately affected and health systems should routinely assess for isolation	Strong — the National Academies’ 2020 consensus report and its clinical commentary both call for routine assessment
Social prescribing produces self-reported wellbeing benefits	Reasonably consistent across qualitative and mixed-method reviews
Social prescribing reduces downstream health service use	Suggestive, not established — only a minority of included studies in the 2021 reviews found this, and none used strong causal designs
Community-health partnership structures (as opposed to referral-only models) improve outcomes	Not directly tested — this is inference from the mechanism literature, not a measured comparison

That last row is the honest caveat. Nobody has run a trial comparing a referral pipe against a genuine two-way partnership and measured the difference in outcomes. The case for partnership rests on understanding why referral-only models fail operationally — funding instability, capacity blindness, absent handover — not on a body of comparative evidence. Say that plainly to a funder rather than implying the case is more settled than it is.

What this does not solve

This framework addresses the relationship between the systems that refer and the organisations that receive. It does nothing for the deeper reach problem in all social prescribing: it depends on someone — a GP, a link worker — first identifying a person as isolated, which means it systematically misses people who never make it into a clinical conversation in the first place. Partnership between health and community organisations makes the pipe less likely to leak. It does not widen the mouth of the pipe.