A Referral Gets Someone There Once. It Does Not Make Them Stay

A link worker refers someone to a walking group, a men’s shed, or a singing session. The person attends once. That is the outcome most social prescribing dashboards will record: referral made, activity attended. It is also, on its own, close to worthless as a measure of anything that matters.

The distinction that gets lost is between a service — something delivered to a person, on a schedule, by staff, that ends when the funding or the referral period ends — and a community — a set of relationships among the people attending, which persists whether or not anyone is watching. Social prescribing is built to produce the first. It hopes, often silently, for the second. Those are different design problems, and treating them as the same one is why so many referral pathways produce attendance without connection.

Why this distinction matters more than the activity itself

The evidence on social prescribing is consistent on one point: participants report benefit that goes beyond the activity itself. A 2022 qualitative meta-synthesis found that people describe gains in meaningful participation and restored purpose, not just social contact for its own sake. A separate systematic review of loneliness-focused social prescribing found all nine included studies reported positive individual impacts, with three showing reduced use of GP, emergency, or inpatient services. That is a real signal. But both reviews also flag heterogeneity across programmes and thin trial evidence — this is not a body of work precise enough to tell you which design features drove the benefit.

What it does support is a working distinction long established outside health policy. Ray Oldenburg’s concept of the third place — a setting distinct from home and work where informal, recurring contact happens — describes what a community activity needs to become in order to outlast the referral. Robert Putnam’s account of declining civic membership in the United States is, in effect, a description of third places closing or thinning out one by one. Eric Klinenberg’s argument about social infrastructure makes the mechanism concrete: it is the physical and social scaffolding — a library table people return to, a regular slot at the same time each week, staff who remember names — that converts a one-off attendance into a standing habit. A social prescription can fund the activity. It cannot, by itself, build the infrastructure that makes the activity recurring.

The five differences that actually separate a service from a community

Feature	Service design	Community design
Who initiates future contact	Staff, via reminder or follow-up call	Members themselves, without prompting
What happens if staff are absent	Session cancelled	Group meets anyway
Membership boundary	Referral-in, discharge-out	Open-ended, self-perpetuating
Social ties formed	Participant-to-facilitator	Participant-to-participant
What ends when funding ends	Everything	The room booking, not the relationships

Most social prescribing activity sits firmly in the left column. That is not a failure of the link worker — referral pathways are, structurally, services. The question is whether the receiving activity has been designed to migrate participants into the right column, or whether it just keeps running the left column indefinitely and calling it community.

The naming problem

Call something a “12-week programme” and you have told participants, correctly, that it ends in twelve weeks. Anyone who has formed an attachment worth keeping now has an incentive to withhold that attachment, because the container is disposable. Programmes that survive past their funded cohort tend to have dropped the cohort language early — not the funding cycle, which is unavoidable, but the framing that told participants they were passing through rather than joining.

What this means in practice: if a referral pathway’s success metric stops at “attended,” it cannot tell you whether it built a service or a community, because both produce attendance. Track whether participants show up without a reminder by week eight, and whether any contact happens between participants outside the scheduled session. Those two things distinguish the categories; attendance alone does not.

The failure mode: the graduation cliff

Call it the graduation cliff. A programme runs for a fixed term, participants build real rapport, and then the term ends, the room booking lapses, and the facilitator moves on to the next cohort. Everyone involved experienced something real. None of it persists, because nothing was ever handed over to the participants themselves — no ongoing meeting slot, no member who took on informal convening, no mechanism for the group to continue without the organisation that started it.

This is a predictable consequence of funding cycles, not a mystery. Commissioners fund defined-length interventions because that is what is measurable and procurable. But a programme designer who accepts the funded term as the natural lifespan of the group has already decided the outcome will be a service, however warmly delivered. Designing against the graduation cliff means asking, from week one, who besides paid staff will be capable of keeping this going, and building in a deliberate handover rather than hoping one emerges.

What to check before referring into an activity

1. Does the activity have a life independent of its current funding cycle, or would it stop entirely if this grant ended?
2. Is there a physical or time anchor — same room, same day, same hour — that exists regardless of who is facilitating?
3. Can you name one thing participants do together that staff did not organise?
4. Has anyone who started as a participant taken on a convening role?
5. Does the activity’s own language describe an ending, or does it describe an ongoing thing people join?

An activity that fails most of these is not necessarily a bad referral — a well-run twelve-week course can still deliver real benefit and is easier to commission and evaluate than an open-ended community. But it should not be marketed to referrers as a cure for isolation, because the evidence base for that specific claim is thinner than the literature’s general tone suggests. Both systematic reviews cited above are explicit about heterogeneity and limited trial evidence; “improves wellbeing during participation” is supportable, “resolves loneliness” is not.

Where the AARP finding cuts against convenient assumptions

A 2018 AARP Foundation survey of adults 45 and older is worth citing precisely because it complicates the referral-to-activity model. It found the strongest predictors of loneliness were the size and diversity of a person’s social network and physical isolation — not simply whether they had recently attended something. Among respondents who had spoken to their neighbours, 33% were lonely; among those who never had, 61% were. That is a proximity and repetition effect, not an event-attendance effect. It supports designing for durable, local, low-friction contact over designing for a single well-produced session, however good that session is.

What this does not solve

None of this addresses reach. The people most likely to benefit from durable community — those most isolated, least confident walking into a room of strangers, least likely to have a network that reinforces attendance — are also the least likely to make it past a first referral, let alone stay long enough for a community to form around them. A programme redesigned along every principle above will still mostly retain people who were already inclined to show up twice. Distinguishing service from community improves what happens to people who arrive. It does nothing to change who arrives in the first place, and that remains the harder and largely unsolved problem in social prescribing.