Measuring the Handoff: Where Referral Pathways Actually Fail

Most social prescribing evaluations report a single number: did loneliness or wellbeing improve at three or six months, for the people who completed the programme. That number tells you almost nothing about whether the pathway works, because it only counts people who made it all the way through. The pathway has several handoff points before that — GP or link worker identifies need, makes a referral, the person attends an initial conversation, the person actually turns up to the community activity, the person keeps turning up. Each one loses people. If you only measure the far end, you cannot tell whether your problem is the activity itself or the five steps that came before it.

This matters because social prescribing referral is now official policy infrastructure in the UK, embedded through the 2018 national loneliness strategy, and is being examined elsewhere as a health-system response to isolation. The evidence base behind it is real but limited: a systematic review of social prescribing and wellbeing found consistent gains in self-esteem and confidence but noted the trial evidence is thin and the programmes it drew on were highly heterogeneous. A separate review focused specifically on loneliness found all nine included studies reported positive individual outcomes, and three found reduced use of GP, emergency, or social care services — but nine studies is not a lot to build a national infrastructure on, and none of them were designed to isolate where in the pathway the benefit actually occurred.

The dropped baton

Call it the dropped baton problem: a referral is made, logged as a successful outcome by the referring service, and then nothing happens. The person never attends. In many programme evaluations this looks like success — a referral was completed — right up until someone checks attendance data and finds a large gap between referrals made and first sessions attended. The National Academies’ 2020 consensus report on isolation in older adults pushed health systems toward more routine screening and referral, but the clinical commentary that followed it was blunt that assessment and referral infrastructure means little if the systems downstream aren’t equipped to receive and follow up on it. Screening more people without fixing the handoff just moves the failure point downstream and makes it less visible.

The dropped baton tends to cluster at two specific points:

1. Referral to first contact. The person is referred by a GP or clinician but the link worker or community organisation never reaches them, or reaches them once and gets no response.
2. First contact to sustained attendance. The person has one conversation or attends once, then does not return. A qualitative synthesis of how people experience social prescribing found that benefit tends to come from restored purpose and meaningful participation, not from contact alone — which means a single attendance, recorded as a successful referral, may deliver almost nothing if it doesn’t lead to repeat engagement.

Neither of these shows up if your evaluation only asks “did outcomes improve for people who completed the programme.”

What to actually track at each stage

Stage	Metric	Why it matters
Referral made	Volume and source	Tells you demand and where it’s coming from, not whether it works
Referral to first contact	Conversion rate, time elapsed	The most common drop-off point; delay itself predicts non-attendance
First contact to first attendance	Conversion rate	Distinguishes “agreed to try it” from “actually engaged”
First attendance to week 4-6	Retention rate	Captures whether the activity itself sustains interest once novelty wears off
Sustained engagement to outcome	Change in validated measure (e.g., UCLA Loneliness Scale)	Only meaningful once you know who’s still in the pathway

Reporting a single outcome score without the funnel above it is like reporting a conversion rate without the traffic numbers. A funder reading “wellbeing improved by X” should ask what proportion of original referrals that X applies to. If it’s 20%, the headline number is describing a self-selected group, not the pathway.

What this means in practice: before you report any outcome measure, report the funnel it sits inside. If half your referrals never reach a first session, fix the handoff before you commission another activity. No amount of activity quality compensates for a broken pathway.

Evidence status

Claim	Status
Social prescribing improves self-esteem and confidence for people who engage	Reasonably supported, small heterogeneous trial base
Social prescribing reduces use of GP, emergency, or social care services	Supported in a minority of studies; not consistent enough to promise a funder
Structured, purposeful activity outperforms unstructured social contact	Supported by qualitative synthesis, not yet by controlled comparison
Screening more people for isolation improves outcomes on its own	Not supported — clinical commentary explicitly flags this as dependent on downstream capacity
Network size and diversity predict loneliness better than frequency of any one contact	Supported by large survey data, relevant to what “success” should look like beyond attendance counts

Network size matters because it reframes what the pathway is for. A 2018 national survey of adults 45 and older found the strongest predictors of loneliness were the size and diversity of a person’s social network and physical isolation, not any single measure of contact. A referral pathway that gets someone to one group once has not moved that number. Retention data at week 4-6, not attendance at week one, is the closer proxy.

What this does not solve

Tracking the funnel tells you where your pathway breaks. It does not tell you who never entered it. Referral pathways, by design, only capture people already inside a system that routes them somewhere — a GP practice, a link worker’s caseload, a community organisation’s front door. People who never see a GP, or who are unregistered, or who avoid formal services entirely, do not appear in any of these metrics, however good your retention rate looks. Fixing the handoff makes the pathway work better for the people already on it. It does nothing for the people who were never referred at all, and no evaluation of the pathway will tell you how large that group is.