Why the Older-Adult Isolation Checklist Doesn't Travel

Most of the measurement infrastructure now used in social prescribing was built for one population: older adults, often housebound, often widowed, often assessed after a fall or a hospital admission. The National Academies’ 2020 consensus report on isolation and loneliness in older adults, and the clinical commentary that followed it, both argue for routine screening in health care settings — and the instruments they describe assume a specific kind of loneliness: shrinking networks, reduced mobility, physical distance from family. That assumption made sense for the population it was written for. Roughly a quarter of adults 65 and older meet criteria for social isolation, and the tools reflect that reality.

The UK’s 2018 loneliness strategy funded social prescribing at scale, and the model has since been pointed at very different groups — new parents, working-age adults referred for anxiety, students, employees flagged through workplace wellbeing programmes. What has not kept pace is the measurement. Programmes serving these newer populations are, in large numbers, using the same screening questions, the same network-size proxies, the same “how often do you see friends or family” items that were validated on people whose loneliness looked nothing like theirs.

What the older-adult model actually measures

The instruments built for older isolated adults are mostly counting contact: network size, frequency of visits, whether someone lives alone. The AARP Foundation’s 2018 national survey of adults 45 and older used the 20-item UCLA Loneliness Scale and found network size and diversity, along with physical isolation, were the strongest predictors of loneliness in that group — a third of people who had spoken to a neighbour were lonely, against 61% of those who never had. For a population with shrinking social circles, counting contact is a reasonable proxy for the underlying problem.

It is a much worse proxy for a 22-year-old. Harvard’s Making Caring Common project found 61% of adults aged 18 to 25 reported serious loneliness, and about half of lonely young adults said no one had spent more than a few minutes recently asking how they were doing in a way that felt genuine. That is not a network-size problem. Many of these young adults have contact — housemates, classmates, group chats — and are still lonely, because the loneliness is about the quality and depth of connection, not its frequency. The American Enterprise Institute’s 2021 survey found something structurally similar for a different reason: the share of men reporting no close friends rose from 3% in 1990 to 12% by 2021, and the share of men with six or more close friends collapsed from 55% to 27% over the same period. A screening tool that asks “how often do you see people” will miss a man who sees people constantly at work and has no one he would call a close friend.

How programmes have handled the mismatch

Published accounts of social prescribing suggest three approaches, none of them fully satisfactory.

Some schemes keep the older-adult instrument unchanged and apply it to whoever is referred, on the logic that any validated tool beats no tool. Systematic reviews of social prescribing’s effect on loneliness report broadly positive individual outcomes across the studies included, but the reviews themselves flag that outcome measures vary widely and that trial evidence is thin — a 2021 review in Perspectives in Public Health found all nine included studies reported positive impacts, but with enough heterogeneity in what “impact” meant that comparing programmes is difficult. A 2021 review in the International Journal of Environmental Research and Public Health reports self-esteem and confidence gains as common outcomes, again without a consistent instrument behind them.

Others have swapped in workplace-specific instruments where the population is clearly occupational. Cigna’s 2020 workplace report used its own loneliness index rather than a geriatric screening tool, and found 61% of US adults reported loneliness sometimes or always, rising to over 80% among employed Gen Z workers, with lonely workers missing work roughly twice as often for illness and five times as often for stress-related absence. That is a genuinely different measurement frame — outcomes tied to absenteeism and presenteeism rather than network size — and it produces numbers that mean something to an employer deciding whether to fund an intervention.

A third group does neither: they borrow the older-adult framing’s language (“isolation,” “reduced contact”) without borrowing its instrument, producing bespoke, unvalidated surveys that cannot be compared to anything.

Claim	Evidence status
Older adults’ loneliness is well captured by network-size and contact-frequency measures	Reasonably supported — AARP’s 2018 survey and the National Academies’ 2020 report converge on this
The same measures capture young adults’ or employees’ loneliness equally well	Not supported — Harvard’s 2021 survey and the AEI’s 2021 survey both point to quality-of-connection and friendship-depth as the driver, not contact frequency
Social prescribing reduces loneliness across populations	Weakly supported — reviews report positive findings but flag thin, heterogeneous trial evidence
Loneliness is a modifiable, population-relevant risk factor worth measuring at all	Well supported — Holt-Lunstad’s 2015 meta-analysis and her 2021 review both treat it as comparable to other preventable health risks

What this means in practice: if you are adapting a programme originally built for isolated older adults to a younger, working, or newly-parenting population, do not reuse its outcome measure by default. Ask what the mechanism of loneliness actually is for the new group — contact scarcity, or connection shallowness — and choose or build an instrument that measures that, even if it means losing comparability with the older cohort’s data.

The frailty-frame problem

Call it the frailty frame: treating loneliness as a single condition with a single validated checklist, when the underlying biology described by Cacioppo — loneliness as an aversive signal driving reconnection — can be triggered by opposite social conditions. An isolated widower and an overcommitted, contact-saturated 24-year-old can register the same score on a scale that was never built to distinguish between them, and a programme evaluated against the wrong mechanism will look like it is failing when it is actually solving a problem the instrument cannot see.

What this does not solve

None of this addresses reach. Every measurement fix described here still depends on someone showing up to be measured in the first place, and social prescribing schemes reach people already inside a referral pathway — a GP visit, a workplace programme, a housing service. People outside those systems, including a large share of the loneliest young adults identified by the Harvard survey, are not being measured by anything, adapted or not.