Which Outcome Measures Will Actually Survive a Commissioner's Review

A commissioner does not fund loneliness reduction. They fund a line item that reports up to something they are accountable for — reduced GP attendance, fewer safeguarding referrals, better delayed-discharge figures, a wellbeing indicator in a strategy document. If your outcome measure does not map onto that chain, it does not matter how well the programme ran. This note is about picking the measure before you design the evaluation, not after.

The satisfaction trap

Most community programmes default to an end-of-session feedback form: did you enjoy it, would you come back, how did it make you feel. These are cheap to collect and almost useless to a commissioner, because they measure delivery quality, not the outcome anyone is paying for. A programme can score 95% satisfaction and still fail to move loneliness at all — satisfaction and change are different questions, and a commissioner who has seen enough of these forms knows it.

The satisfaction trap is not that these forms are wrong to collect. It is that they get submitted as if they were the outcome evidence, and the funding conversation stalls when someone asks the follow-up question: so did anyone’s isolation actually change?

Match the measure to who is paying

Before choosing an instrument, work out which of three commissioner types you are talking to, because they read outcomes differently.

1. Health commissioners (ICBs, primary care networks) want something that plausibly connects to service use or clinical risk. The National Academies’ 2020 consensus report calls for routine assessment of isolation within health care settings, which means health commissioners increasingly expect programmes to speak that language — screening at referral, re-screening at exit.
2. Local authority and public health commissioners want population-level wellbeing indicators, often tied to a strategy document rather than a clinical pathway. They are more tolerant of self-report scales and softer outcomes like confidence or self-esteem.
3. Funders and trusts often want a story that combines a number with a mechanism — not just “loneliness fell” but a plausible account of why.

The same programme, reporting to all three, needs at least two measures, not one.

Four steps to a measure that holds up

1. Use a validated instrument, not a bespoke survey. A short, non-validated in-house questionnaire is the fastest way to lose credibility with a commissioner who has seen a validated scale before. The AARP Foundation’s 2018 survey of adults 45 and older used the 20-item UCLA Loneliness Scale rather than a bespoke tool, which is exactly why its finding — that a third of adults who report speaking to neighbours are lonely, against 61% of those who never do — is citable elsewhere. Borrow that discipline. A shortened UCLA scale (there are three- and eight-item versions used across the literature) gives you a number a commissioner can compare against published baselines, which a bespoke tool cannot do.

2. Set a baseline before the programme starts, not a single post-programme snapshot. A single measurement taken at week six tells you where people ended up, not whether they moved. Commissioners increasingly ask for pre/post, and some ask for a delayed follow-up at three or six months to check the change held.

3. Pick a second measure tied to the commissioner’s own KPI, where you can defensibly claim one. If you are reporting to a health commissioner, a systematic review of social prescribing found that three of nine included studies reported reductions in GP, emergency, social worker, or inpatient service use. That is a real but narrow evidence base — six of nine studies did not find this, and the review itself flags limited trial evidence. Use it to justify collecting the measure, not to promise the result.

4. Say what you are not claiming. If your evaluation has no comparison group, say so in the same paragraph as your headline number. Commissioners trust an evaluation more, not less, when it names its own limits — it signals you understand what the number can and can’t carry.

What the evidence actually supports

Claim	Evidence status
Social prescribing improves self-reported wellbeing, confidence, self-esteem	Reported across multiple qualitative studies, but the underlying systematic review notes limited trial evidence and high heterogeneity across programmes
Social prescribing reduces measured loneliness	Positive individual-level impact reported in all nine studies in one systematic review, but designs were mostly small and non-experimental
Social prescribing reduces NHS or social care service use	Reported in a minority of included studies (three of nine); not consistent enough to promise to a commissioner as a default outcome
Structured, purposeful group activity produces more benefit than contact alone	Supported by qualitative meta-synthesis; participants describe benefit from restored participation and purpose, not just from being around people
A short validated loneliness scale gives comparable, credible numbers	Well established methodologically; the main practical risk is choosing a bespoke tool instead
Social connection belongs in the same risk category as smoking or heart disease	Argued in the 2023 U.S. Surgeon General’s advisory and supported by Julianne Holt-Lunstad’s mortality meta-analyses, but this is a population-health framing, not evidence that any given local programme moves mortality risk

What this means in practice: if you can only fund one measurement instrument, make it a validated loneliness scale with a genuine pre/post design, not a satisfaction survey and not a bespoke in-house questionnaire. If you can fund a second, make it whichever service-use or referral metric your specific commissioner already tracks — but report it as suggestive, not as your proven outcome, because the review evidence behind it is thin.

The comparison-group problem you cannot solve on this budget

Almost every commissioner conversation eventually reaches the question a community programme cannot answer cheaply: compared to what? Holt-Lunstad’s 2015 meta-analysis and the broader mortality literature it belongs to draw their credibility from large samples and control for confounds that a local evaluation, run on a small grant, cannot replicate. A qualitative meta-synthesis of social prescribing found that participants describe real, felt benefit — restored participation, a sense of purpose — but that is evidence of subjective change, not evidence that the programme caused a measurable shift relative to people who received nothing.

Be honest about this rather than dressing a pre/post design up as causal. A commissioner who has read one evaluation with an inflated causal claim will discount every subsequent one you send them.

What this does not solve

None of this fixes the reach problem underneath all social connection programming: the people who show up to be measured are, almost by definition, people who already engaged enough to attend, be screened, and complete an exit survey. The AARP data on neighbour contact and the National Academies’ figure that roughly a quarter of adults 65 and older are socially isolated both describe populations that most referral-based programmes struggle to reach at all — people who never speak to a neighbour, never accept a referral, never fill in a baseline form. Choosing the right outcome measure makes your evidence more credible to the people already inside the programme. It says nothing about the isolation happening outside it, and no measurement instrument changes that.