A Referral Is Not a Membership

A link worker refers someone to a walking group, a craft session, a gardening project. The referral is logged, the person turns up once, maybe twice, and then stops. The outcome data records this as a failed engagement. Usually the group wasn’t the problem. The person was referred into a service, when what actually reduces loneliness is becoming part of a community — and those are not the same intervention, even when they look identical on a leaflet.

This distinction rarely gets named in social prescribing design, and that is a design failure worth fixing, not a minor semantic point.

What separates a service from a community

A service has a provider and a recipient. Someone runs it, someone attends it, and the relationship between them is the reason the person shows up. A community has members, and the relationship that matters is between the members, not between any one member and a coordinator. You can tell which one you’ve built by asking a blunt question: if the paid facilitator didn’t show up one week, would the group still meet?

If the answer is no, you have built a service with a social veneer. That is not worthless — services can be efficient, consistent, and easy to fund. But the evidence on loneliness points toward peer relationship, not facilitated contact, as the thing doing the work. Robert Putnam’s distinction between bonding and bridging social capital is relevant here: a service can produce bridging contact (a session with strangers) without ever producing the bonding tissue — mutual obligation, recognition, the sense of being missed — that keeps someone coming back.

Ray Oldenburg’s account of the third place makes the same point from a different angle. The pub, the barbershop, the café that works as social infrastructure isn’t organised around a single activity delivered to attendees. It’s organised around regulars who recognise each other, and the activity is often incidental to that. A council-run twelve-week befriending course rarely becomes a third place, because twelve weeks is a programme, not a habit, and it ends on a date someone else chose.

Why this matters more than the activity choice

Programme managers spend a lot of time on activity selection — walking versus singing versus allotments — and comparatively little on whether the activity is structured to produce membership. That allocation of attention is backwards. A 2022 qualitative meta-synthesis of social prescribing found that people describe the benefit not as social contact in isolation, but as restored participation and purpose — the sense of having a role, not just an appointment. Structured, purposeful group activity tracked as more valuable to participants than contact alone. That’s a vote for community-like structures over service delivery, but it is a preference finding from qualitative synthesis, not a controlled comparison, and it should be treated with that caveat.

A separate systematic review of social prescribing’s impact on loneliness found that all nine included studies reported positive individual impacts, with three showing reduced use of GP, emergency, or inpatient services. That’s real, and it’s the kind of number that gets a programme funded. But look at what the review is measuring: individual outcomes at a point in time, not durability. None of the underlying evidence in this area is strong on retention past the initial engagement window, and a 2025 systematic review protocol is explicit that the effectiveness of social prescribing for older adults remains unclear, noting that only one peer-reviewed randomised controlled trial exists in the field. The research infrastructure to answer “did this become a community that held” barely exists yet.

The week-six cliff

Programmes referring into short, facilitator-led courses tend to see a drop-off around week five or six — after the novelty has worn off but before any peer relationship has had time to form independent of the session itself. Call it the week-six cliff. It shows up as an engagement chart that looks fine for the first month and then falls off a shelf, and it gets read internally as “the activity wasn’t the right fit,” when the more accurate reading is usually “the group never became a group.”

The fix is not more weeks of the same structure. It’s redesigning for member-to-member contact earlier: rotating who brings something, letting participants set the next session’s plan, building in contact channels that don’t require the facilitator to relay them (a shared chat, a swapped phone number, a standing invitation to arrive early). None of this is exotic. It’s how the third places Oldenburg and Eric Klinenberg describe actually function — libraries and parks work as social infrastructure not because staff run brilliant programming, but because the physical space lets loose association happen on repeat, unsupervised, until it becomes a habit that doesn’t depend on anyone organising it.

What this means in practice: before referring someone into an activity, ask who is responsible for making them a member of it rather than an attendee at it. If the answer is “the facilitator will be friendly,” that is a service. Look instead for activities with rotating leadership, participant-set agendas, or a track record of continuing after paid coordination ends — and if none exist locally, treat building one as a commissioning priority, not an activity gap.

An evidence-status table for the claims in this piece

Claim	Evidence status
Social prescribing produces positive self-reported outcomes	Reasonably well supported across multiple systematic reviews, though studies are heterogeneous and mostly uncontrolled
Social prescribing reduces downstream service use (GP, A&E)	Supported in a minority of included studies (3 of 9 in one review); promising but thin
Community-structured groups outperform facilitator-led services on retention	Plausible and consistent with qualitative accounts of purpose and participation; no controlled trial evidence isolating this specific comparison
Social prescribing works durably for older adults specifically	Explicitly unresolved — a 2025 review protocol found only one RCT in the entire field
Third places (libraries, cafés, parks) function as social infrastructure independent of programming	Well established in the sociological literature (Oldenburg, Klinenberg), though this is theory and case description, not trial evidence

The honest summary: the loneliness-reduction claim for social prescribing generally is stronger than the claim that any particular activity structure sustains it. Commissioners buying “twelve sessions of X” are buying the well-evidenced part and hoping the weakly-evidenced part follows automatically. It doesn’t.

What to check before referring, and what to build if nothing qualifies

A short checklist for a link worker or programme manager assessing whether a local activity is a service or something closer to a community:

1. Does the group continue, in some form, if the paid coordinator is absent for two weeks?
2. Do members have each other’s contact details, or only the coordinator’s?
3. Is there a role for members beyond attending — hosting, teaching a skill, organising the next outing?
4. Has anyone who joined a year ago still be attending, and do they know the newer members’ names?
5. Is there a cost or barrier to leaving (a missed friend, an unfinished project) beyond simply not showing up?

If most answers are no, the group is a well-run service. That’s still a legitimate referral for someone who needs structured contact right now, particularly someone in acute isolation who isn’t ready for the informality of a third place. But log it accordingly, and don’t expect the twelve-week evaluation to show durable connection, because you didn’t build for durability. AARP’s 2018 survey of adults 45 and older found that the strongest predictors of loneliness were the size and diversity of a person’s network and physical isolation — not the absence of any single programme. Its 2025 follow-up shows loneliness rising further in this age group over the intervening years. A service can plug a gap for a defined period. Only something that behaves like a community changes the shape of someone’s network permanently.

What this does not solve

This distinction helps you evaluate and redesign existing referral options. It does not solve the fact that community-like structures are harder to build than services, take longer to mature, and resist the kind of standardised commissioning that funders prefer to buy. It also does nothing for reach: both services and communities depend on someone already engaged enough to accept a referral in the first place, and neither reaches the person who never gets as far as the link worker’s desk.