When to Stop Running a Social Prescribing Group

Most social prescribing guidance tells you how to start something. Almost none tells you how to end it. That gap costs programmes money and, more quietly, costs participants the trust they extended when they showed up.

Groups decline. Attendance that started at fourteen settles at five, then three. A walking group loses its organiser and limps along with whoever remembers to show up. The honest question is rarely asked out loud: is this worth continuing, or are we running it because closing it feels like admitting failure?

The three reasons groups keep running past their usefulness

Sunk referral cost. If link workers have been sending people to a group for eighteen months, closing it means telling referrers to stop, which feels like undoing work.

No decision trigger. Most programmes have a start date and a funding cycle, but no defined point at which someone is supposed to ask “should this continue?” Without that point, continuation is the default, not a choice.

Confusing attendance with harm. A shrinking group is often assumed to be failing participants. Sometimes it is. Sometimes three people who found each other in month one are getting exactly what they need, and low numbers reflect success at forming bonds rather than failure to attract a crowd. The qualitative literature on social prescribing consistently finds that participants value depth of connection and restored purpose over volume of contact — a 2022 meta-synthesis in BMC Health Services Research found benefit was described in terms of meaningful participation, not attendance counts. Shrinkage on its own tells you nothing until you know why.

A decision framework: fix, restructure, or close

Run this check at a fixed interval — quarterly is reasonable for most community groups — rather than waiting for a crisis.

Question	If yes	If no
Is attendance stable or growing among a core group, even if small?	Continue	Investigate further
Is decline traceable to a single fixable cause (venue, time slot, one difficult member, loss of a facilitator)?	Fix that cause	Move to restructure/close decision
Are the people still attending reporting benefit, not just habit?	Continue at reduced scale	Consider closing
Would closing free capacity for a referral pathway with higher demand?	Weigh against benefit to current members	Lower urgency to close
Has the group become a holding pattern — people attend but report no change in isolation, mood, or confidence over two consecutive reviews?	Close or substantially redesign	Continue monitoring

The last row is the one programmes skip. Attendance is easy to measure and almost meaningless on its own; the systematic review of social prescribing and loneliness found that most included studies reported positive impact, but the outcome measures used were inconsistent enough that “still coming” and “actually better off” are frequently conflated. If you are not separately tracking self-reported loneliness or confidence alongside headcount, you cannot use this framework — you’re guessing.

What this means in practice: before you can decide whether to stop something, you need a baseline measure taken when someone joins and a follow-up measure taken periodically after — even something as simple as a single loneliness item, or a short version of the UCLA scale used in the AARP Foundation’s 2018 survey of adults 45 and older. Without that, “the group feels quieter” is the only signal you have, and it points you toward closing groups that are working and keeping ones that aren’t.

The named failure mode: the hospice group

Call it the hospice group problem. A group has stopped generating new referrals, has stopped growing, and exists mainly to preserve continuity for its remaining members — who are, in fact, benefiting from it. The programme keeps it alive out of loyalty rather than evidence, and it slowly consumes facilitator time and room budget that a genuinely oversubscribed group elsewhere could use.

The hospice group is not automatically wrong to keep running. Robert Putnam’s account of declining civic associational life is a reminder that small, durable groups of the kind that Ray Oldenburg called third places have a value that isn’t captured by growth metrics — a group of four people who talk weekly may be doing more for those four than a group of twenty doing less for each. The error is not keeping it. The error is keeping it without ever explicitly deciding to, and without recording that decision so it gets revisited rather than defaulting on indefinitely.

How to close a group without harming the people in it

If closure is the decision, sequence it deliberately:

1. Tell current members first, separately from any public announcement. Give at least one full cycle of notice — for a weekly group, four weeks minimum.
2. Offer a specific alternative, not a general referral back to the link worker. “This group is ending; here is another group meeting the same need” beats “let us know if you want another referral.”
3. Hold a final session that is explicitly a final session, not a group that quietly stops appearing on the calendar. Unannounced disappearance is what erodes trust in the wider referral pathway — people who feel dropped are less likely to accept a future social prescription.
4. Record why it closed, separate from attendance numbers, so the next programme manager doesn’t restart an identical group for the same reasons it failed the first time.

What this does not solve

This framework assumes you are already collecting some outcome measure beyond attendance, which many programmes are not — and building that measurement capability is a separate piece of work, not a byproduct of deciding to close a group. It also does not help with the harder political version of this decision: closing a group that a funder specifically named in a grant, where the funding logic and the participant-benefit logic point in different directions. That is a negotiation, not a framework. And none of this addresses the people who never joined the group in the first place; deciding when to stop something says nothing about whether the thing was reaching who it needed to reach while it ran.