How to Choose Loneliness Outcome Measures a Commissioner Will Sign Off

The measure you pick determines the funding decision more than the programme does. A commissioner reading two similar applications will trust the one with a validated instrument and a defensible outcome chain over the one with a warmer narrative. This is a procedure for getting that instrument right the first time, not after a funder sends the application back.

Step 1: Separate what you are actually changing

Loneliness, social isolation, and low social support are three different things, and a growing body of research insists they be measured separately rather than treated as one construct. The National Academies’ 2020 consensus report on older adults distinguishes objective isolation — how many social contacts a person has and how often — from the subjective, felt experience of loneliness, and warns against collapsing the two into a single before-and-after number. A programme that increases contact frequency but does nothing for the felt experience of loneliness has still done something real. If your logic model doesn’t say which of these three you’re targeting, decide that before you touch an instrument.

This matters practically because the instruments are not interchangeable. A network-size measure will not move if your intervention is a one-off wellbeing group; a loneliness scale will not move if your intervention is a befriending call that increases contact but doesn’t touch how isolated someone feels day to day.

Step 2: Default to a validated scale, not a bespoke question

The single most common reason commissioners discount a self-reported outcome is that it was invented for the programme. A four-item “how connected do you feel” survey written in-house has no comparability to anything else the commissioner has seen, and no track record.

The UCLA Loneliness Scale is the instrument most consistently used across the academic literature, and its use is what makes a national survey like the AARP Foundation’s 2018 study of adults 45 and older comparable to decades of prior research, rather than a one-off snapshot. Where possible, use the UCLA scale or its short-form version rather than writing your own items. If a shortened version is unavoidable for a busy caseload, say so explicitly and cite which validated short form you used — do not silently invent one.

Step 3: Decide whether you need a trial-grade design or a plausible-story design

Most social connection programmes cannot run a randomised controlled trial, and commissioners generally do not expect one at programme scale. But it is worth knowing what the gold standard looks like, because it clarifies what your weaker design is missing. The HEAL-HOA trial, a dual randomised controlled trial of prosocial engagement and volunteering among lonely older adults in Hong Kong, is one of the few loneliness interventions tested against a genuine control group rather than assessed with a single-arm before-and-after survey. Most published social prescribing evidence is not built this way. A 2021 systematic review of social prescribing and loneliness found that all nine included studies reported positive impacts, but noted the studies themselves are heterogeneous and mostly uncontrolled — which is exactly the pattern that makes commissioners nervous, because a design with no comparison group cannot rule out that people who were already improving were the ones who signed up.

If you cannot run a controlled comparison, say so in the application rather than letting the commissioner discover it. Name the design as a pre-post cohort study, state what it cannot show, and explain why a fuller design isn’t feasible at your scale. This candour reads as competence, not weakness.

Step 4: Pair the subjective measure with one thing a commissioner can independently verify

Self-report on its own invites the objection that people say what they think evaluators want to hear. A 2021 systematic review found reductions in GP, emergency, social worker, or inpatient service use in three of nine studies of social prescribing’s effect on loneliness — and that kind of administrative data point is worth more to an NHS commissioner than another wellbeing scale, because it can be checked against records the commissioner already holds. If your setting allows it, pair your primary loneliness or isolation measure with one administrative or observable secondary measure: attendance records, GP contact data, or a referral count. You do not need this to prove causation. You need it so the commissioner has something outside self-report to anchor the story.

Step 5: Report the mechanism, not just the endpoint

A 2022 qualitative meta-synthesis of how people experience social prescribing found that participants describe the benefit as extending beyond social contact itself, toward restored meaningful participation and a sense of purpose — and that structured, purposeful activity appears to work better than unstructured contact alone. That is a measurement instruction as much as a design insight: if your programme is built around purposeful activity rather than pure contact, include a short purpose or participation measure alongside your loneliness scale. A commissioner assessing renewal funding wants to know why the programme worked, not only that a number moved, because “why” is what tells them whether it will transfer to a different population.

The named failure mode: the borrowed benchmark

The most common way an evaluation fails on review is what practitioners have started calling the borrowed benchmark: taking a national loneliness prevalence figure — such as the U.S. Surgeon General’s 2023 advisory finding that roughly half of adults report loneliness — and treating your local cohort’s baseline score as directly comparable, without accounting for the fact that your cohort was recruited precisely because they were already flagged as isolated. A programme that reports “we brought our cohort down to the national average” sounds strong until a commissioner asks what the baseline was for a group referred for isolation in the first place. State your own baseline plainly and let the change stand on its own.

Evidence-status table

Claim	Evidence status
Loneliness and isolation are distinct constructs requiring separate measures	Strong — consensus position, National Academies 2020
UCLA Loneliness Scale is the most comparable instrument across studies	Strong — dominant instrument in the literature
Social prescribing reduces self-reported loneliness	Moderate — consistent direction across small, mostly uncontrolled studies
Social prescribing reduces service utilisation (GP, A&E, inpatient)	Weak-to-moderate — reported in a minority of included studies
Purposeful structured activity outperforms unstructured contact	Emerging — qualitative synthesis, not yet tested experimentally
Loneliness measurement is standardised across the field	Weak — a 2023 review of the state of the field identifies inconsistent measurement as a persistent barrier to comparison

What this means in practice: pick a validated scale over a bespoke one, state your design’s limits before the commissioner finds them, and pair self-report with one thing that can be checked against a record. Commissioners fund applications that already anticipate the objection.

What this does not solve

None of this fixes the underlying reach problem. A well-measured programme still only tells you about the people who enrolled, attended, and completed a follow-up survey — typically the people already engaged enough to be reachable by measurement in the first place. The 2023 review of the state of loneliness research is explicit that inconsistent measurement across the field makes it hard to compare your results to anyone else’s, even when you have done everything on this list correctly. Good measurement makes your case defensible. It does not make your programme representative of who did not walk through the door.