Social Prescribing and Loneliness: What the New Review Actually Shows

A systematic review published June 23 in Perspectives in Public Health looked at nine studies of social prescribing programmes aimed at loneliness. All nine reported positive effects at the individual level. Three reported reductions in downstream service use — fewer GP visits, fewer emergency contacts, fewer inpatient stays or referrals to social workers. On the face of it, this looks like exactly the kind of evidence a government would want before spending public money on an intervention: it works, and it might save money elsewhere in the system.

The question is whether nine studies with uniformly positive results is strong evidence or weak evidence dressed as strong evidence. The review does not resolve this, and the pattern of results across its included studies suggests the latter.

What social prescribing actually is

Social prescribing is a referral mechanism, not a treatment. A general practitioner or a “link worker” identifies a patient who is isolated or lonely and connects them to a non-clinical activity — a walking group, a gardening project, a befriending scheme, a community class. The UK’s 2018 loneliness strategy, the first national strategy of its kind, made social prescribing one of its central mechanisms and funded link workers as part of the National Health Service.

The theory of change is straightforward and plausible: connect an isolated person to a recurring activity, and their network grows, their loneliness falls, and their health outcomes improve as a downstream consequence. Eric Klinenberg’s account of social infrastructure — libraries, parks, and other shared physical spaces that structure contact between people — gives this theory a plausible mechanism. But a plausible mechanism is not the same as demonstrated effect, and the June review does not close that gap.

The problem is the studies, not the review

The review’s methodology is not in question. The problem is what it had to work with. Social prescribing research is dominated by small, uncontrolled, self-report studies — a person is referred, a person attends, a person later reports feeling less lonely. There is rarely a comparison group of similar people who were not referred. There is rarely a validated instrument; loneliness is often measured with a bespoke question rather than something like the UCLA Loneliness Scale, which makes results hard to compare across studies or against the wider literature. And follow-up periods are typically short, so nobody knows whether an effect measured at eight or twelve weeks persists at six months.

A related systematic review published in May, on social prescribing and wellbeing more broadly, found the same thing: real gains in self-esteem and self-confidence reported across studies, but “limited trial evidence and heterogeneity across programmes.” Two independent reviews, three weeks apart, arrive at the same qualified verdict: something appears to be happening, but the studies are not built to say what, how much, or for how long.

Nine studies reporting a positive result is also, on its own, a pattern that should raise a question rather than settle one. Social prescribing research is not conducted at random on programmes chosen at random; it tends to be conducted on programmes that already look promising, often by people invested in their success, and published disproportionately when the result is favorable. A literature with no negative or null findings at all is more consistent with that kind of selection than with an intervention that works reliably across contexts.

What this means for the policy built on it

None of this means social prescribing does not work. It means the evidence available in mid-2021 cannot tell policymakers how well it works, for whom, or relative to what. That distinction matters because social prescribing is no longer a pilot idea. It is embedded in the UK’s national loneliness strategy, and loneliness has become a formal item of government business elsewhere too — Japan appointed a loneliness minister in February, and the two countries held a joint ministerial meeting on the subject in June. Programmes at that scale are typically justified, at least in part, by exactly the kind of individual-level and service-use evidence the June review compiled.

The National Academies’ 2020 consensus report on isolation in older adults took a more cautious line, calling for routine assessment of isolation and loneliness in health care settings without endorsing a specific intervention as proven. That caution looks warranted. The AARP Foundation’s 2018 survey of adults 45 and older, which used the same UCLA scale the clinical literature relies on, found that the strongest predictors of loneliness were the size and diversity of a person’s social network and physical isolation — structural conditions that a single referral to a twelve-week walking group may or may not touch, depending on whether the person keeps attending after the study ends and whether the link workers making the referrals have any consistent way of tracking that.

What would settle the question

A study that would actually answer whether social prescribing reduces loneliness needs three things the current literature mostly lacks: a control group of comparable isolated or lonely people who were not referred, a validated instrument such as the UCLA scale applied consistently before and after, and follow-up at six and twelve months rather than at the end of the programme. Randomized referral — where eligible patients are randomly assigned to receive a social prescribing referral or usual care — would go further still, since it would separate the effect of the activity from the effect of being the kind of patient a clinician thought worth referring in the first place.

Until studies like that exist, the honest reading of the June review is not that social prescribing has been shown to work, but that nobody has yet tested it in a way capable of showing that it does not.