Picking Loneliness Metrics a Commissioner Will Actually Fund

A commissioner reading your quarterly report does not want to know that staff “felt more connected.” They want to know whether the number moved, whether the number is credible, and whether it will still mean the same thing next year. Get the measure wrong and the whole programme becomes hard to defend at renewal — not because it failed, but because nobody can tell.

This is the decision in front of you before you launch anything: which outcome measure to commit to reporting. Get it wrong in month one and you are stuck with it, or you have to explain to a funder why you switched.

The three failure modes

The bespoke-survey problem. Programme teams often write their own three-question pulse survey because it feels quick and relevant. It is quick. It is also worthless to a commissioner, because there is no benchmark to compare it against and no way to know if your 15-point improvement means anything. The AARP Foundation’s 2018 national survey of adults 45 and older used the 20-item UCLA Loneliness Scale precisely so its findings — one in three older US adults report loneliness — could sit alongside decades of academic research. Your homemade scale can’t do that.

The vanity-metric problem. Attendance, sign-ups, and satisfaction scores measure whether people showed up and liked it. They do not measure whether anyone is less isolated. A commissioner who has seen a few of these decks will ask the obvious follow-up question, and “we didn’t track that” ends the conversation badly.

The over-claiming problem. Reporting isolation reductions as if they translate directly into reduced sickness absence or healthcare cost, without the mediating evidence to back it, is the fastest way to lose credibility with a commissioner who checks sources. The evidence connecting loneliness to costly downstream harm is real — but it is population-level epidemiology, not a guarantee that your six-week programme will show up in next year’s absence figures.

What actually holds up

Three categories of measure survive scrutiny, in descending order of rigor and ascending order of ease.

Validated psychometric scales. The UCLA Loneliness Scale and the shorter three-item loneliness scale derived from it are the closest thing to a common currency in this field. They are used across the survey and cohort literature that commissioners’ own analysts are likely to have read — including the Perspectives on Psychological Science meta-analysis by Julianne Holt-Lunstad and colleagues, which pooled results across dozens of studies using comparable instruments. Using one means your before-and-after numbers can be placed against a published baseline instead of floating alone.

Service-use proxies. A systematic review in Perspectives in Public Health found that of nine studies evaluating social prescribing’s effect on loneliness, three reported reductions in GP visits, emergency contacts, or inpatient use. These are commissioner-native metrics — they speak directly to cost, which is usually the actual question behind the loneliness question. If your programme sits inside a health or occupational health pathway, tracking absence days, EAP contacts, or occupational health referrals alongside a loneliness scale gives you a bridge between the wellbeing story and the budget story.

Structured qualitative synthesis. A 2022 meta-synthesis in BMC Health Services Research found that participants in social prescribing describe benefit less in terms of “feeling less lonely” and more in terms of restored purpose and meaningful participation — outcomes a numeric scale won’t capture well. That’s a real limitation of the scales, not a reason to abandon measurement altogether. Structured qualitative interviews, coded against a small fixed framework rather than left as open testimonial, can sit as a secondary measure that explains the “why” behind a scale movement. Commissioners generally accept this as supporting evidence, not as the primary outcome.

The evidence-status table

Claim	Evidence status
Loneliness is a modifiable, measurable state comparable across populations using validated scales	Strong — replicated across dozens of studies
Social prescribing-type interventions reduce loneliness for participants	Moderate — consistent direction across reviews, but small samples and heterogeneous designs
Reduced loneliness translates into reduced healthcare or absence costs at programme scale	Weak — plausible from population epidemiology, not demonstrated at intervention level
A bespoke, unvalidated in-house survey shows programme impact	Not evidence a commissioner should accept

What this means in practice: commit to one validated scale (the UCLA scale or its three-item short form) as your primary outcome before launch, pair it with one service-use or absence proxy if your setting allows it, and treat qualitative interviews as explanatory colour, not headline data. Report the scale movement honestly even when it’s small — a modest, well-measured effect is more fundable than an impressive, unverifiable one.

Why the sample size problem doesn’t go away

Workplace pilots are often 40 to 80 people. Loneliness research at the scale commissioners cite — the National Academies’ consensus finding that roughly a quarter of adults 65 and older are socially isolated, or the Surgeon General’s 2023 advisory putting loneliness at roughly half of US adults — comes from samples in the thousands or hundreds of thousands. A 60-person pilot cannot produce a statistically stable effect size, and no measurement choice fixes that. What a good measure gets you is comparability and credibility, not statistical power. Say this to the commissioner up front, in the proposal, rather than let them discover it in the evaluation.

What this does not solve

None of this addresses reach. The people who complete a workplace wellbeing survey and show up to a connection programme are disproportionately those already engaged enough to opt in. A validated scale measured well among volunteers tells a commissioner nothing about the disengaged third of the workforce who never entered the programme at all — and current review evidence on social prescribing, drawn largely from self-selected participants, has the same blind spot. Measurement rigor and reach are separate problems, and solving the first does not touch the second.