Who Is Responsible When Something Goes Wrong in a Walking Group

A link worker refers a 78-year-old man with mild cognitive decline to a weekly walking group run by three volunteers. Six weeks later he doesn’t show up. Nobody in the group has his GP’s number, his emergency contact, or a clear sense of whose job it is to notice he’s gone. This is not a hypothetical failure of imagination — it’s the structural gap that opens up whenever a health or care system refers someone into a setting it doesn’t run.

Social prescribing exists precisely because informal social contact — a choir, an allotment, a walking group — produces outcomes that clinical services alone don’t reach. The UK’s 2018 loneliness strategy built social prescribing into national policy on that basis, and the qualitative evidence backs the underlying logic: a 2022 meta-synthesis found that participants describe the benefit of social prescribing as extending well beyond contact itself, toward restored purpose and participation. But the same feature that makes informal groups valuable — they are not clinical settings, run by people without clinical training — is what makes them a safeguarding blind spot. Nobody has systematically solved this. What exists is a set of adaptations social prescribing schemes have made to manage a risk they can’t eliminate.

The link worker as the load-bearing wall

The core adaptation is structural, not procedural: put a trained intermediary between the referral and the informal group, and make that person responsible for judgment calls the volunteer running the group should never have to make alone.

This is the whole logic of the link worker role as it has developed in social prescribing. The link worker takes the referral, assesses what the person needs and what they can handle, and places them into a community activity — but stays the named point of contact rather than handing responsibility to whoever happens to be running the knitting circle that week. The National Academies’ 2020 consensus report on isolation in older adults, and the clinical commentary that followed it, both argue that routine assessment of isolation and loneliness needs to sit inside the health care system rather than be improvised by whoever a person is referred to next. That argument extends naturally to risk: a volunteer running a walking group should not be the first or only person assessing whether someone is safe to be there.

The problem is that this only works if the link worker relationship persists after the referral. Several of the systematic reviews of social prescribing note that programme design and evaluation practice are inconsistent across schemes — one review of nine studies found all reported positive outcomes but noted wide heterogeneity in how programmes were structured and followed up. Nothing in the published evidence establishes a standard for how long a link worker stays engaged after a referral, or what triggers re-contact. In practice, many schemes treat the referral as the endpoint of their responsibility, which recreates the exact gap the link worker model was meant to close.

Structure lowers risk exposure even when nobody frames it that way

A separate but related finding: the 2022 meta-synthesis on perceived benefits of social prescribing observed that structured, purposeful group activity appears more effective for participants than unstructured social contact alone. This wasn’t framed as a safeguarding finding — it was about outcomes, participation, and meaning — but it has a direct safeguarding implication. A group with a defined activity, a set start and end time, and a named facilitator generates far more natural opportunities to notice absence, distress, or decline than an open drop-in space where people come and go informally.

This gives programme designers a lever that has nothing to do with training volunteers in risk assessment: choosing structured activities over unstructured ones changes how much a group is capable of noticing, independent of anyone’s skill.

What this means in practice: don’t ask volunteers to become safeguarding officers. Ask instead whether the activity itself generates a register — a sign-in sheet, a fixed roster, a facilitator who greets people by name — because that structure does more risk-management work than a training session volunteers will half-remember by week six.

The evidence gap the sector hasn’t closed

The honest position is that almost none of the outcome literature on social prescribing was designed to measure safeguarding failure. Reviews report reductions in GP contact, improved self-esteem, and positive subjective experience. None report on adverse events, near-misses, or how often risk was identified but not acted on, because none were set up to look for that.

This mirrors a gap named explicitly in a different but adjacent literature. The American Heart Association’s 2022 scientific statement on social isolation and cardiovascular risk is unusually direct about this: having established the association between isolation and cardiovascular harm, it states plainly that the absence of intervention evidence is the central research gap. The same is true one level down, for the safeguarding infrastructure around interventions. We know these programmes probably need someone watching for risk. We do not have published evidence on which watching arrangements work.

Claim	Evidence status
Social prescribing produces positive subjective outcomes for participants	Reasonably well supported across multiple qualitative reviews, though heterogeneous in method
Structured activity outperforms unstructured contact on participant-reported benefit	Supported by one meta-synthesis; not yet replicated at scale
Link worker continuity reduces safeguarding incidents in community placements	Plausible by extension from assessment literature; not directly measured
A defined escalation pathway from volunteer to clinical contact improves outcomes	No published evaluation identified; treated as good practice by inference, not evidence
Volunteer training alone (without structural backup) reduces risk	No supporting evidence located

The failure mode: the warm handoff that goes cold

Call it the cold handoff. A referral is made with real care — a link worker assesses the person, matches them to an appropriate group, makes a warm introduction. The record shows a completed referral. But nothing in the design specifies what happens if the person stops attending, discloses something concerning to a volunteer, or shows signs of decline three months in. The handoff was warm at the point of transfer and has gone cold by the point anyone would need it to still be functioning. Programmes that treat referral as a single event, rather than a relationship with a defined re-contact trigger, are building this failure mode into their design whether they intend to or not.

The fix isn’t more paperwork at the point of referral. It’s a specified trigger for re-contact — a missed-session threshold, a scheduled check-in at a fixed interval, a named person on the community side who knows who to call — decided before the person is placed, not improvised after something has already gone wrong.

What this does not solve

None of this addresses the deeper structural fact that social prescribing, like almost all voluntary and community-sector infrastructure, is staffed overwhelmingly by volunteers who did not sign up to carry clinical risk and are not equipped to. Nor does it address reach: the people most likely to be safely caught by a well-designed escalation pathway are people who were referred by a system in the first place. Someone isolated enough to never reach a GP, a link worker, or a referral pathway at all sits entirely outside this framework — and the U.S. Surgeon General’s 2023 advisory on connection makes the same point about loneliness interventions generally: the people least connected are often the hardest to reach with any structured programme, safeguarded or not.