Social Prescribing Builds Referrals, Not Necessarily Communities

A link worker refers someone to a walking group. The person attends twice, likes it, then stops. The referral is logged as a success — activity completed, outcome recorded — and the case is closed. Six months later that person is exactly as isolated as before, but the data says the intervention worked.

This is the gap between a service and a community, and it is the single most consequential design decision in social prescribing that nobody makes on purpose.

Two different things wearing the same name

A service delivers a unit of contact: an assessment, a class, a call, a session. It has a start and an end, a named recipient, and a professional or volunteer on the other end whose job is to be there. A community has none of that structure. It persists whether or not any particular person shows up, has no one accountable for any individual’s presence, and generates belonging as a side effect of people simply being in the same place repeatedly.

Ray Oldenburg’s account of the third place captures what a real community setting does that a service cannot: it works because attendance is voluntary, low-stakes, and habitual, not because someone is checking a name off a list. Robert Putnam’s distinction between bonding and bridging social capital points at the same gap from another angle — a service can bond a small group briefly around a shared referral, but it rarely bridges anyone into a wider, self-sustaining network. Eric Klinenberg’s argument about social infrastructure makes the physical version of the point: a library or a park generates contact continuously, without a caseworker scheduling it.

Social prescribing pathways are, almost by construction, services. Someone is referred, someone delivers, an outcome is measured, the case closes. That is not a criticism of the model — it is a description of what a healthcare-adjacent pathway is built to do. The problem is when a service is marketed and evaluated as though it produced a community, because the mechanisms of benefit and the mechanisms of failure are different for each.

Why this distinction changes what you build

The evidence base for social prescribing is consistently positive on subjective outcomes and consistently thin on mechanism. A systematic review found improvements in self-esteem and confidence but flagged limited trial evidence and heterogeneity across programmes. A separate review of loneliness-specific outcomes found all nine included studies reported positive individual impacts, with three showing reduced service use — but a 2025 review protocol notes that only one peer-reviewed randomised controlled trial exists for older adults specifically, despite growing adoption of the model. That is a strong signal that people feel helped and a weak signal about what, structurally, helped them.

A qualitative meta-synthesis gives a clue: participants describe benefit extending beyond social contact itself to restored meaningful participation and purpose, and structured, purposeful activity appears to work better than contact alone. That finding cuts against the community framing, not toward it. Purpose and structure are service properties. A community rarely organizes itself around anyone’s sense of purpose; it just recurs.

The randomised evidence, thin as it is, tells a consistent story in the same direction. The HEAL-HOA trial found volunteering and prosocial engagement reduced loneliness among older adults in a controlled comparison — a structured, assigned activity, not an open-ended gathering. A 2025 randomised trial of befriending in aged care found measurable reductions in UCLA Loneliness Scale scores at 8 and 16 weeks, but noted that befriending was outperformed by more structured psychological approaches in head-to-head comparison. Read together, these two trials say something specific: where controlled evidence exists at all, it favours structured, service-like design over loose communal contact. The intervention literature is not proving that communities work. It is proving that well-designed services work, at least in the short term studied.

That does not mean community-building is worthless — Oldenburg and Klinenberg’s arguments about third places are not refuted by a lack of RCTs; the RCT literature simply hasn’t been built to test that kind of slow, ambient infrastructure. It means you should not expect a twelve-week referral pathway to produce third-place effects, and you should not evaluate it as though it were meant to.

What this means in practice

Decide, in writing, whether you are running a service or seeding a community, before you design the outcome measure. If it is a service — bounded, staffed, purposeful — measure the outcome the structure was built to produce (confidence, activity completion, reduced service use) and stop there. If you intend the referral to seed something that outlasts the programme, you need a second, later measurement point, six to twelve months out, asking whether the person is still attending anything at all, referred or not. Most programmes measure only the first and imply the second.

The AARP survey of adults 45 and older gives a rough proxy worth borrowing for that later check: it found people who had spoken with their neighbours were lonely at roughly half the rate of those who never had. That is not a social prescribing outcome measure at all — it’s a neighbourhood-contact measure — but it is closer to what “did this become a community” would actually mean than attendance counts at week six.

The handoff problem

Call it the handoff problem: a service that never plans its own obsolescence keeps producing dependent referrals instead of independent participants. A walking group run entirely by a paid facilitator, with no route for a regular attender to become the person who unlocks the gate or organises the route, will dissolve the moment funding for the facilitator ends — because it was never a community, it was a service with an audience. The groups that survive their own funding cycle tend to be the ones that quietly transferred ownership to participants well before the money ran out.

Evidence status

Claim	Status
Social prescribing improves self-reported confidence and wellbeing	Reasonably well supported across multiple reviews
Social prescribing reduces loneliness specifically	Supported by observational and qualitative work; only one RCT exists for older adults
Structured activity outperforms unstructured contact	Suggested by two small RCTs and a qualitative synthesis; not yet a robust evidence base
Social prescribing seeds lasting community, independent of the referral	Largely unmeasured; almost no programme evaluation looks past programme end

What this does not solve

This framework tells you how to stop conflating two different things. It does not tell you how to build a third place from scratch, and the evidence on doing that deliberately is thin to nonexistent — Oldenburg and Klinenberg describe why such places matter, not how a commissioner manufactures one on a budget cycle. It also does nothing for reach: everything above describes what happens once someone is already inside a pathway, referred by a GP or a link worker who noticed them. The people never referred, never registered, never inside any service at all remain outside the scope of this or any social prescribing design question.