Who Doesn't Walk Through Your Door, and Why

A social prescribing service that fills its groups within two weeks of opening referrals has not necessarily succeeded. It may just mean the referral route only reaches people who were already going to find their way to something. The people the service was actually built for — the ones with the smallest networks, the least trust in institutions, the fewest existing reasons to walk into a building full of strangers — are statistically the ones least likely to respond to a leaflet in a GP waiting room.

This is not a minor sampling problem. It is the central design flaw in most social prescribing recruitment, and it is largely invisible unless you go looking for it.

The self-selection trap

Referral pathways for social prescribing typically run through a GP, a link worker, or a self-referral form. Each of those steps requires the person to have already done several things: noticed the problem in themselves, decided it’s worth mentioning, trusted the person they’re mentioning it to, and been willing to accept a suggestion involving unfamiliar people in an unfamiliar setting. A qualitative meta-synthesis of social prescribing found that participants who did engage described benefits going well beyond contact itself — restored purpose, a sense of meaningful participation, and structured activity mattering more than casual social contact. That is a genuinely useful finding. It is also a finding about people who made it through every one of those upstream filters. It tells you almost nothing about the people who didn’t.

The National Academies’ 2020 consensus report on isolation in older adults put the scale of the miss in plain terms: roughly one quarter of adults 65 and older are socially isolated, and the health care system is poorly positioned to identify most of them because isolation doesn’t present as a chief complaint. People experiencing the deepest isolation are, by definition, less visible to the systems designed to catch them. A companion clinical commentary on that report pushed further, arguing that routine assessment — not opportunistic conversation — is what it would take to find these patients at all, because they will rarely raise it themselves.

AARP’s tracking data adds a second layer to the problem. Its 2018 survey of adults 45 and older found that people who had spoken to their neighbours were lonely at 33%, against 61% among those who never had. Network size and diversity, not any single trait, were the strongest predictors of loneliness. The follow-up 2025 survey, using the same instrument, shows the pattern has not eased. The people recruitment needs to reach are precisely the ones with the fewest existing social ties to catch a referral, a poster, or a word-of-mouth mention in the first place.

Why the usual fixes don’t fix this

Programmes that notice a recruitment gap tend to reach for the same three moves: more posters, a wider referral network, a louder call to action. All three assume the barrier is awareness. It usually isn’t.

More posters assumes the target group passes through the same physical or digital spaces as everyone else. Research on third-place availability found that closures of coffee shops, libraries, recreation centres, and similar spaces between 2019 and 2021 were concentrated in census tracts with higher social vulnerability and in rural areas. If the community spaces that would carry your poster have already closed in the areas where isolation is worst, the poster strategy fails before it starts.

A wider referral network assumes GPs and link workers are seeing the isolated patients. Some are. Many aren’t, because isolated patients — especially those managing chronic illness, mobility limits, or distrust built from prior experiences of discrimination — attend health services less regularly, not more. A KFF survey found that experiences of discrimination correlate with smaller social support networks; the same populations navigating that discrimination are frequently the ones least likely to trust a referral coming from within a system that has not served them well.

A louder call to action assumes the barrier is motivation. For someone with a genuinely thin support network, the barrier is more often logistical and psychological at once: no one to go with, no transport, and an accurate prediction that walking into a room of strangers alone will feel worse before it feels better.

None of this means recruitment is hopeless. It means the standard playbook targets willingness, when the actual bottleneck is reach.

What changes recruitment design, not just recruitment volume

1. Route through existing relationships, not existing services. Isolated people are more likely to trust a person they already know than an institution they’ve been referred to. Eric Klinenberg’s account of social infrastructure argues that the physical and social spaces already embedded in a neighbourhood — libraries, churches, barbershops, community centres — carry more legitimacy than a clinical referral precisely because they are not clinical. Recruitment through those existing anchor points, rather than solely through health system contact, reaches people who have opted out of health system contact altogether.

2. Make the first step require nothing. Every additional requirement — a phone call, a form, a scheduled intake appointment — filters out someone with low trust or low capacity. A first contact that is a conversation at a place they already go, with no commitment attached, converts better than a referral slip.

3. Assess for isolation directly, don’t wait for disclosure. The clinical commentary on the National Academies report argued for structured assessment rather than relying on patients to raise isolation unprompted. The same logic applies outside clinical settings: community organisations serving older adults, carers, or new arrivals can build a routine isolation check into intake rather than treating it as something a person will volunteer.

4. Go to where the network has already collapsed, not where it’s thin. People who have spoken to no neighbours in the past year are a different recruitment target than people with a few loose ties. The former need an outreach worker who shows up more than once; the latter may respond to an invitation. Treating both groups the same wastes effort on the group that would have joined anyway.

5. Track who doesn’t come, not just who does. If a service can name the demographic and geographic pattern of its non-responders, it can redirect outreach. Most programmes never build this list because it requires deliberately going after the negative case.

What this means in practice: if your waitlist fills fast, that is evidence your recruitment channel works for people who were already reachable — not evidence the programme is reaching the people it was designed for. Before expanding capacity, map who is missing from the group, and redesign the first point of contact around their actual barriers, not a louder version of the same channel.

The evidence gap you should not paper over

Claim	Evidence status
Isolated older adults are underdetected by routine health care contact	Well supported (National Academies, 2020)
Structured, purposeful activity produces more benefit than casual contact	Supported by qualitative synthesis, not RCT evidence
Third-place closures are concentrated in already-vulnerable areas	Supported by a single 2025 geographic study
Reworking recruitment channels increases reach among high-isolation groups	Not directly tested; inferred from adjacent evidence
Social prescribing reliably reduces loneliness for the hardest-to-reach group specifically	Unclear — a 2025 systematic review protocol notes only one peer-reviewed randomised controlled trial exists for social prescribing in older adults at all

That last line matters. The intervention literature for social prescribing is thin even for people who do enrol. For people who never enrol because recruitment never reached them, there is effectively no outcome data — only the absence of a data point, which is a different kind of finding and an easy one to overlook.

What this does not solve

Redesigning recruitment gets more of the right people to the door. It does not make the thing behind the door effective, and it does not address the structural loss of the third places — libraries, senior centres, community halls — that made informal, low-barrier contact possible in the first place. A programme can fix its outreach perfectly and still be recruiting people into a group that meets in a room that will close in eighteen months. Reach and infrastructure are separate problems, and solving the first without attending to the second just produces better-targeted disappointment.