Safeguarding in Informal Social Programmes: What Actually Needs a Policy

A volunteer befriender visits an isolated older adult weekly for four months. In month five, the older adult stops eating properly and starts giving the volunteer cash “for helping so much.” Nobody trained the volunteer to recognise either as a safeguarding concern, because the programme was designed around warmth, not risk. This is not a hypothetical edge case. It is the standard shape of harm in informal social programmes, and most programmes have no procedure for it because nobody designed one to feel this ordinary.

Social prescribing, befriending, peer support, and community-run groups all sit deliberately outside clinical structures. That is their strength: the systematic review evidence on social prescribing points to restored participation and purpose as the outcomes people actually value, not clinical contact. But “informal” is a description of tone, not of risk. The moment you put an unsupervised adult into a relationship with another adult who has reduced capacity, cognitive decline, or dependency needs — which is exactly the population much of this work targets, given that roughly a quarter of adults 65 and older are considered socially isolated — you have created a safeguarding context whether or not you call it one.

The three risk categories, and why treating them alike is the mistake

Most programmes either over-formalise everything (a 40-page policy nobody reads) or under-formalise everything (a values statement and a hope). Neither works because the risks are not the same kind of risk.

Risk type	Example	What it needs
Capacity and consent	Participant’s cognitive state means they can’t meaningfully agree to the activity or to disclosures made within it	A capacity check built into intake, and a clear escalation route — not volunteer judgement alone
Boundary drift	Volunteer or peer relationship shifts into financial, romantic, or dependency territory	Explicit written boundaries stated at induction, plus a named person volunteers must tell when a relationship starts to feel unusual
Disclosure of harm	Participant discloses abuse, neglect, self-harm risk, or domestic violence during an informal activity	A designated safeguarding lead, a reporting pathway, and volunteer training on what to do in the first ten minutes after a disclosure

Only the third has anything like established statutory infrastructure behind it in the UK, where the national loneliness strategy pushed social prescribing into mainstream commissioning without correspondingly clarifying who holds safeguarding accountability when the referral sits outside the health system. The other two are frequently left to individual judgement, which is where things go wrong.

The failure mode: “it’s just a walking group”

Programmes that involve unstructured time — walking groups, shared meals, drop-in social clubs — tend to assume that because the activity is casual, the relationships within it are low-stakes. Call this the walking-group blind spot: informality gets read as safety. It is the opposite. Structured, purposeful activity with clear roles gives everyone a script to fall back on when something feels wrong. Unstructured time is exactly where boundary drift and undisclosed vulnerability surface, because nobody is watching for it and nobody was told to.

The qualitative evidence on social prescribing found that participants valued structured, purposeful group activity more than unstructured contact alone — a finding usually cited for engagement reasons, but it has a safeguarding reading too. Structure is not just what makes activities effective. It is what makes them supervisable.

What this means in practice: if your programme includes any one-to-one contact (befriending, mentoring, peer visiting) or any unsupervised time between participants with unequal vulnerability, you need a named safeguarding lead who is not the volunteer or peer themselves, a written escalation pathway with a maximum response time, and a boundary-setting conversation at induction — not a boundary-setting clause buried in a handbook. If your programme is a supervised group activity with a paid or trained facilitator present throughout, a lighter policy — clear roles, a code of conduct, a way to flag concerns — is proportionate. Do not build the heavy version for the light case; it will not get followed.

What training has to cover that most inductions skip

Volunteer and peer-support training in this space tends to focus on empathy and communication skill, which matters but is not the gap. The gap is procedural knowledge under pressure: what to do in the ninety seconds after someone discloses something you weren’t trained to hear. A clinician-facing commentary on the National Academies’ work on isolation in older adults makes a related point in the health system context — that routine assessment of isolation requires clinicians to know not just how to ask, but what to do with the answer. The same logic applies to volunteers, who have far less institutional backup than clinicians do.

Training should cover, specifically:

1. What counts as a disclosure requiring escalation, with two or three concrete examples, not a definition.
2. Who to contact, by name, with a phone number, not “your coordinator” as an abstraction.
3. What not to do — do not promise confidentiality, do not investigate yourself, do not confront the other person.
4. How to end a shift or visit that has become distressing, without abandoning the person mid-crisis.

None of this needs to be long. It needs to be rehearsed, ideally through a short scenario walk-through rather than a document to read once and file.

Evidence status: be honest about what’s actually been tested

Claim	Evidence status
Structured social activity produces better wellbeing outcomes than unstructured contact	Reasonably supported — qualitative synthesis across social prescribing studies
Social prescribing reduces use of GP, emergency, or inpatient services	Weak — only three of nine studies in one systematic review found this, and none were controlled trials
Formal safeguarding training reduces harm in informal befriending/peer programmes	Not directly tested — this note is an inference from safeguarding practice generally, not a finding specific to this setting
Physical social infrastructure (libraries, parks) affects rates of informal contact	Supported by observational and historical analysis, notably around the 1995 Chicago heat wave

The honest position is that almost nobody has run a controlled study of safeguarding procedure in informal social programmes specifically. The evidence base for social prescribing’s benefits is itself heterogeneous and short on trials. That is not a reason to skip a safeguarding policy — the case for one rests on general safeguarding practice and plain risk logic, not on a loneliness-specific trial that doesn’t exist — but it is a reason not to claim more rigour for your policy than you actually have.

What this does not solve

A good safeguarding policy protects against harm within the relationships your programme creates. It does nothing for the person who never joins — the isolated adult with no referral pathway, no digital access, and no volunteer knocking on their door. Safeguarding design is a precondition for running a responsible programme, not a substitute for reach, and programmes that get very good at protecting the people already inside the room can mistake that competence for solving the wider problem.