Who Owns the Referral After You Make It

A GP or link worker identifies loneliness, makes a referral to a community organisation, and closes the case. That is the moment most partnerships between health systems and community organisations quietly fail. Not at referral. After it.

The National Academies’ 2020 consensus report called on health systems to routinely assess isolation and loneliness. Clinician-facing commentary on that report went further: it argued that assessment without a functioning place to send the result is close to useless, and that most systems have built the first half of that pipeline and not the second. Assessment is now common. What comes after it is not standardised, not resourced, and rarely owned by anyone in particular.

The gap has a shape

A referral from a clinical setting to a community organisation crosses an accountability boundary. The clinician’s job, in most systems, ends at the referral. The community organisation’s job — a walking group, a men’s shed, a lunch club — begins when someone shows up. Nobody owns the space between those two events: whether the person called, whether they were offered a next step if the first group did not suit them, whether they turned up once and never came back.

Systematic reviews of social prescribing report real individual-level benefits — increases in self-esteem and confidence, and in some studies reductions in GP or emergency service use. But the same reviews are consistent about a second finding that gets less attention: the evidence is heterogeneous, thin on trial data, and says almost nothing about what happens to the substantial share of referred people who never make it to a session. A qualitative synthesis of how people experience social prescribing found that benefit came from restored participation and purpose, not from contact alone — which means a referral that does not result in sustained participation has not delivered the thing the evidence says works. It has delivered paperwork. A 2025 systematic review protocol focused specifically on older adults notes that despite growing adoption of social prescribing, its effectiveness remains unclear, and that only one peer-reviewed randomised controlled trial exists in this area at all. Adoption has outpaced evidence, and the accountability gap is a large part of why the evidence is so hard to generate: nobody is tracking the middle of the pathway consistently enough to study it.

Naming the failure: the handoff cliff

Call it the handoff cliff. It is the point at which a referral is recorded as “made” in the clinical system and the community organisation has no obligation, capacity, or in many cases even the contact information to confirm the person ever arrived. Both sides can report success. The clinician’s metric is referrals issued. The organisation’s metric is sessions run. Neither metric captures the person who fell between them.

This is not a training problem or a motivation problem on either side. It is a design problem: two organisations with different funding cycles, different data systems, and different definitions of “done” are handed a single continuous task and no shared handoff protocol.

What a functioning partnership needs to specify

1. A named receiving contact, not a generic organisational inbox, on the community side — someone who commits to acknowledging a referral within a fixed window, typically 48 to 72 hours.
2. A closed-loop confirmation back to the referrer: attended, did not attend, attended once and stopped, declined. This should be a data field, not a narrative note nobody reads.
3. A second-offer rule. If the first activity does not suit the person, the receiving organisation offers at least one alternative before the case is considered closed. Reviews of social prescribing repeatedly find that structured, purposeful activity works better than generic contact — which means the first match matters, and one bad match should not end the pathway.
4. A shared minimum dataset, agreed before referrals start, covering who was referred, what for, what happened, and at what cost to each side. Without this, no one can later show a funder or commissioner what the partnership achieved, and the 2025 protocol’s point about thin trial evidence will remain true indefinitely, because nobody is generating the data that would fix it.
5. A capacity ceiling, stated in advance. Community organisations are frequently referred more people than they can absorb. If there is no agreed ceiling, the organisation either quietly stops accepting referrals or dilutes what it offers, and the referrer keeps sending people into a pathway that has already collapsed without being told.

What this means in practice: before a referral pathway goes live, put the receiving contact’s name, the confirmation window, and the second-offer rule in writing, signed by both sides. If a commissioning conversation about a social prescribing partnership does not produce these three things, the partnership is not ready to take referrals yet, whatever the launch date says.

Why physical space belongs in this conversation

Part of what community organisations are being asked to receive people into is disappearing. Research tracking third places — the term Ray Oldenburg used for informal gathering spaces such as libraries, cafes, and recreation centres — found closures across every category studied between 2019 and 2021, concentrated in places with higher social vulnerability and in rural areas. Eric Klinenberg’s account of social infrastructure argues these losses are not incidental; they shape the rate of social contact directly. A health system and a community organisation can build a perfect handoff protocol and still have nowhere reliable to send someone in a town that has lost its senior centre.

What this does not solve

None of this addresses reach. A handoff protocol makes the pathway work better for people who were already identified and referred. It does nothing for people who never see a clinician, never get asked about loneliness, or live somewhere the receiving organisation no longer exists. Fixing the handoff cliff is necessary and cheap relative to most health interventions. It is not the same as solving the underlying shortage of places to send people, and no amount of process design will substitute for that shortage being addressed directly.