The WHO Commission Just Made Loneliness a Global Problem. Social Prescribing Was Built for a Local One

The World Health Organization launched a Commission on Social Connection on 15 November, co-chaired by U.S. Surgeon General Vivek Murthy and youth envoy Chido Mpemba. The framing is deliberately global: loneliness is not a high-income problem, and the commission’s job is to treat it as a worldwide public health priority rather than a UK or US concern that other countries eventually catch up to.

That framing creates an immediate operational problem for anyone running or funding a social prescribing programme. The evidence base for social prescribing was not built globally. It was built almost entirely in the UK’s National Health Service, on populations that are older, already in contact with a GP, and navigating a single, relatively centralized referral pathway. The UK’s 2018 loneliness strategy — the first national strategy of its kind — explicitly funded social prescribing as a delivery mechanism through that system. Most of the systematic reviews that now get cited as “the evidence for social prescribing” are reviews of that system, in that country, with that population.

Gallup’s 2023 survey of 142 countries found 24% of people worldwide report feeling lonely, with the highest rate — 27% — among adults aged 19 to 29, and the lowest, 17%, among people 65 and older. That is close to a mirror image of the population social prescribing was designed around. The UK model assumes a patient who is retired or nearing it, has an existing relationship with primary care, and can be referred into gardening groups, walking clubs, or befriending schemes that meet during the day. The population the WHO commission is now asking systems to reach skews younger, is less likely to have a settled relationship with any health service, and in much of the world has no GP gatekeeper to refer them at all.

Where the model breaks first

Three assumptions built into UK-style social prescribing do not travel automatically.

The referral pathway assumes a link worker sits inside a healthcare relationship. That works where primary care is the front door to social support, as it functions in NHS-funded programmes. In health systems without a salaried primary care workforce, or where care is paid for episodically, there is no equivalent point of contact to build a referral pathway from. The mechanism, not just the funding, has to be rebuilt.

The activities on offer assume a population with daytime availability and physical mobility. Reviews of social prescribing report real gains in self-esteem and confidence, and — where measured — reductions in GP, emergency, and inpatient service use. But the activities behind those gains (walking groups, gardening, craft sessions, weekly community drop-ins) were designed for people not working full-time. Gallup’s youngest cohort, the group with the highest loneliness rate worldwide, is also the group least served by an activity roster built for retirees.

The outcome measures assume proximity to a health system that can record them. A meta-synthesis of qualitative studies found that what people actually value in social prescribing is not contact for its own sake but restored participation and purpose — structured, purposeful activity does more than unstructured mixing. That is a genuinely useful design principle. It is not, however, an outcome you can capture through GP attendance records in a country where most primary care is delivered privately, informally, or not at all.

What this means in practice: before adapting a social-prescribing model to a new population or country, identify what stands in for the three things the UK model assumes — a referral gatekeeper, daytime-available participants, and a health-system data trail. If none of the three exists, you are not adapting the model. You are building a different one that borrows its name.

The transfer failure mode: assuming the mechanism is the evidence

Call this the borrowed-mechanism problem. A funder reads that social prescribing “works” and commissions a version of it for a population — younger adults, informal workers, a country with no GP referral system — where none of the underlying conditions hold. The activities get copied. The referral pathway does not get rebuilt, because nobody names it as a separate design decision. Six months later the programme has low uptake and no way to show impact, and the conclusion drawn is “social prescribing doesn’t work here,” when the actual failure was skipping the redesign step.

The National Academies’ 2020 report on isolation in older adults is instructive here precisely because it stayed disciplined about its population: it called for routine assessment within the health care system, for older adults, where that system already exists. It did not claim the finding generalizes to people outside that system. Programme designers borrowing from UK social prescribing should hold themselves to the same discipline.

Evidence status, by claim

Claim	Evidence status
Social prescribing improves self-reported confidence and self-esteem	Reasonably supported, within UK/NHS-style programmes
Social prescribing reduces downstream health service use	Suggestive — three of nine studies in one review, not a settled finding
Structured purposeful activity outperforms unstructured social contact	Supported by qualitative synthesis, not by trial evidence
The UK model transfers to non-GP-gatekept health systems	No evidence either way — untested
The UK model reaches younger, high-loneliness populations without redesign	Evidence points against it — activity design assumes different availability

What this does not solve

The WHO commission’s global framing is a genuine advance in getting loneliness taken seriously as policy, following the UK’s 2018 strategy and the AARP and Harvard surveys that established loneliness rates by age group in the US. But a commission naming a problem globally does not produce global evidence for a specific intervention. Programme designers working outside the UK’s health infrastructure are, for now, building on a hunch borrowed from a different country’s healthcare architecture — not on evidence that the same mechanism produces the same result elsewhere. And everywhere this model runs, it still only reaches the person who got referred. The people least likely to be lonely-and-visible to any system — the isolated, the undocumented, the never-in-contact — remain outside its reach regardless of which country runs it.