When to Close a Peer Support Group, Not Just How to Start One

Somewhere in most social prescribing portfolios there is a group that has been running for three or four years, has a core of the same six or eight people, and hasn’t taken a new referral in a year. No one wants to close it. It clearly matters to the people in it. And that is precisely the problem: the case for keeping it open and the case for closing it rest on the same fact, and the guidance available to link workers and commissioners says almost nothing about how to tell them apart.

The literature on social prescribing is dense with advice on starting groups — referral pathways, engagement, what “good” looks like at week one. It is almost silent on ending them. That silence has a cost: budgets get committed indefinitely to activities that stopped generating new social contact long ago, while people on a waiting list for a first referral get told there’s no capacity.

What the evidence actually supports

Two systematic reviews of social prescribing for loneliness report real, if modest, effects. Nine studies reviewed in 2021 all reported positive individual impacts, and three found reductions in use of GP, emergency, or inpatient services. A separate 2021 review of social prescribing and wellbeing found gains in self-esteem and confidence, but flagged limited trial evidence and wide variation between programmes — the kind of caveat that should make anyone cautious about treating any single group as self-evidently working just because it has run a long time.

A 2022 qualitative synthesis adds a more specific and more useful finding: participants describe benefit that goes beyond social contact itself, toward restored purpose and meaningful participation. Structured, purposeful activity appears to do more than unstructured contact alone. That distinction matters for a closure decision, because a group that has drifted from structured activity into a standing social gathering has changed what it is, even if attendance looks stable.

None of this evidence tells you when to end a specific group. But it does tell you what a group is supposed to be doing, which is the only honest starting point for asking whether it still is.

The legacy group problem

Call it the legacy group problem: a group kept running past the point where it produces new social connection, because ending it feels like withdrawing care from the people still attending. The signal that a group has slipped into this state is not falling satisfaction — legacy groups often score well on satisfaction, because the remaining members are exactly the people the group works for. The signal is a stalled referral pipeline and a closed membership. If no new person has successfully joined in the past six to nine months, the group has functionally become a private friendship circle operating inside a commissioned service. That may be a fine thing for the people in it. It is not a fine thing to keep counting as reach.

Robert Putnam’s distinction between bonding and bridging social capital is useful here, even though it predates social prescribing as a field. Bonding capital — the tight, closed ties within an established group — deepens support among people who already have it. Bridging capital creates new ties, often across difference. A commissioned service that only produces bonding capital for a fixed membership, and stops producing bridging capital for anyone new, has quietly changed its function without anyone deciding it should.

Eric Klinenberg’s argument about social infrastructure — that libraries, parks, and shared physical spaces shape rates of contact by design — cuts the other way. Some social infrastructure earns its keep precisely by staying open regardless of who’s using it this month, because its value is in being reliably there. Ray Oldenburg’s account of the third place makes a similar point: durability and predictability are part of what makes a hangout function as one. The distinction that matters is between infrastructure — open access, low barrier, no referral needed — and a commissioned, resourced group with a waiting list behind it. The first can run indefinitely by design. The second cannot, without an ongoing justification tied to who it is currently serving.

Claim	Evidence status
Structured group activity outperforms unstructured contact for social prescribing outcomes	Supported — 2022 qualitative synthesis
Social prescribing groups reduce use of GP and emergency services	Suggestive — three of nine studies in one 2021 review
A stalled referral pipeline indicates a group has become bonding-only	Reasoned inference, not directly tested
Closing a legacy group and reallocating its slot improves aggregate reach	Not evaluated in the published literature

What this means in practice: Before renewing funding for any recurring group, ask one question a satisfaction survey won’t answer: how many people who were not already in this group joined it in the last two quarters? If the answer is zero, the group is not a social prescribing intervention anymore. It is a standing social club that happens to be paid for out of a health or council budget. That may still be worth funding — but say so explicitly, and stop counting it against referral targets it can no longer meet.

The AARP’s 2018 survey of adults 45 and older found that network size and diversity, not contact frequency alone, predicted lower loneliness — and that people who talk to neighbours are roughly half as likely to be lonely as those who never do. That is an argument for churn, not permanence: a portfolio of groups that keeps producing new, moderately diverse contacts will do more for population-level loneliness than the same budget locked into groups that have become closed and stable.

What this does not solve

None of this tells a link worker how to actually close a group without harming the people in it — that’s a duty-of-care question the published evidence doesn’t address, because almost no study has followed what happens to participants after a group ends. It also doesn’t help with the harder political problem: a group that has stopped serving new referrals is often the easiest one to defend to a funder, because it has years of satisfaction data behind it, while a new group with six months of uncertain outcomes does not. Evidence-based closure decisions will usually be less popular than evidence-free continuation. That asymmetry is not something a measurement framework fixes.