Which Outcome Measures a Commissioner Will Actually Accept

A commissioner does not want to hear that your programme “improved wellbeing.” They want a number that maps to something they already have to report on, that was measured the same way before and after, and that will not fall apart if a colleague asks how it was collected. Most social prescribing evaluations fail this test not because the programme didn’t work, but because the measure chosen couldn’t carry the weight put on it.

This note is about choosing the measure, not proving the programme works. Get the measure wrong and no amount of subsequent analysis rescues it.

Start from what the commissioner already tracks

Before picking an instrument, find out what outcome the commissioning body is already accountable for. Integrated care boards, local authorities, and employers each report upward on different things, and your evaluation is far more persuasive if it speaks their existing language rather than introducing a new one.

1. Ask what outcomes are in the current commissioning contract or service specification, not what you think should be in it.
2. Check whether a validated scale is already in use elsewhere in the local system — GP practices running social prescribing link workers, for instance, may already be using a specific loneliness or wellbeing tool, and matching it lets your data sit alongside theirs.
3. Only introduce a new instrument if nothing suitable exists. A second, incompatible measure is worse than no measure.

The three things commissioners will ask about a scale

Whatever you pick, expect these three questions, and have answers ready.

1. Is it validated? A scale built for this evaluation, or lightly adapted “for our population,” has no track record and will be treated with suspicion. Use an instrument with published psychometric properties.
2. Has it been used elsewhere with populations like ours? A tool validated only on university undergraduates will not reassure a commissioner funding work with isolated older adults.
3. Can it be repeated at follow-up without contaminating the result? If completing the scale itself changes how people answer it next time — because it primes them to think about loneliness, say — that needs to be acknowledged, not hidden.

The UCLA Loneliness Scale is the closest thing to a safe default here: it is the instrument behind the AARP Foundation’s 2018 national survey of adults 45 and older, which is itself useful as a comparison benchmark rather than a novel commissioning of your own. Reporting your programme’s before-and-after scores next to a large national baseline measured the same way is far more persuasive than reporting your own numbers in isolation.

Evidence status of the outcomes commonly claimed

Programme managers routinely claim outcomes the underlying research cannot support at the strength implied. Here is where things actually stand.

Claim	Evidence status
Social prescribing improves self-esteem and confidence	Reasonably well supported; a 2021 systematic review lists these among the more consistently reported outcomes
Social prescribing reduces loneliness	Supported at individual level across all nine studies reviewed in 2021, but the review notes wide heterogeneity in how loneliness was measured, which weakens comparability
Social prescribing reduces GP or A&E attendance	Only three of nine studies in that same review reported this, and the effect is not consistent enough to promise a commissioner as a headline metric
Structured, purposeful activity produces better outcomes than social contact alone	Supported by qualitative synthesis; participants describe benefit tied to restored purpose and participation, not contact volume
Loneliness and social isolation are distinct constructs requiring separate measurement	Well established, and a frequent source of confused evaluation design
Volunteering-based interventions reduce loneliness in older adults	Supported by one randomised trial in Hong Kong — rare in this field, where most intervention evidence is uncontrolled

That last row matters more than its size suggests. The overwhelming majority of social prescribing intervention studies are uncontrolled before-and-after designs, a gap the wider loneliness research literature has flagged repeatedly. A 2023 review of the field’s state of research names inconsistent measurement as one of the central barriers to comparing results across studies at all. If a commissioner asks whether your evaluation design has a comparison group, and it doesn’t, say so plainly rather than letting the number speak for itself.

The named failure mode: the borrowed-scale problem

Programmes frequently borrow a scale from a national report — citing, for instance, the finding that social isolation carries a mortality risk comparable to established clinical risk factors, drawn from the 2015 meta-analysis behind that figure — and then apply it to a small local cohort assessed by a link worker rather than a researcher, with no control group, no baseline period, and a follow-up window of weeks rather than months. The scale is legitimate. Its use here is not. National mortality-risk statistics describe population-level relationships over years; they cannot validate a six-week local outcome measure by association.

This is the borrowed-scale problem: reaching for a credible instrument’s evidence rather than its conditions of use. Commissioners who have seen several of these evaluations start discounting the citations altogether, because the pattern is familiar. The fix is not to avoid citing strong national evidence — it is to be explicit that it establishes the importance of the outcome, not the validity of your particular measurement of it.

What this means in practice: pick one primary outcome measure with a track record, match it to what the commissioner already reports on, and state your evaluation design’s limitations in the same paragraph as your results — not buried in a methods appendix. A modest, honestly-caveated finding survives scrutiny. An impressive, undercaveated one gets picked apart in the first meeting.

Building in a clinical hook, if the setting allows it

Where a programme sits alongside primary care, the National Academies’ 2020 consensus report calling for routine assessment of isolation and loneliness in health care settings gives you a legitimate reason to align your outcome measure with something a GP practice might eventually be asked to collect anyway. A 2020 clinician-facing commentary on that report discusses what routine assessment would actually require in practice — a short, validated tool that a non-specialist can administer without training. That is a useful constraint to hold yourself to even outside a clinical setting: if your measure needs a research background to administer or interpret, it will not survive contact with an under-resourced service.

What this does not solve

Choosing the right measure does not create a control group where none exists, and it does not turn a six-week pilot into evidence of durable change. It also does nothing for the reach problem: outcome measures describe what happened to the people who engaged with the programme, not the ones who never heard about it or never got referred. A well-chosen measure makes an honest programme’s case credible. It cannot make a weak programme look strong, and it should not be asked to.