Social Prescribing and the Limits of the Surgeon General's Advisory

The U.S. Surgeon General’s advisory of 2 May 2023 states that approximately half of American adults report experiencing loneliness, and puts the mortality risk of social disconnection on a par with smoking up to 15 cigarettes a day. It closes with a six-pillar National Strategy to Advance Social Connection. Vivek Murthy had made the underlying argument three years earlier in Together, which framed loneliness as a public health problem rather than a private misfortune; the advisory converts that argument into an instruction to institutions.

The question worth asking six days on is not whether the risk is real. It is whether the evidence base can support the specific things a health system would be asked to do — and here the advisory inherits a well-documented asymmetry. The observational literature on social connection and mortality is among the more consistent in population health. The trial literature on interventions is small, heterogeneous, and mostly uncontrolled.

The risk half of the case is the strong half

Holt-Lunstad’s 2010 meta-analysis in PLoS Medicine pooled 148 studies and 308,849 participants and found that stronger social relationships were associated with a 50% increased likelihood of survival over follow-up. Her 2015 review in Perspectives on Psychological Science separated the constructs and reported odds ratios for early mortality of 1.29 for social isolation, 1.26 for loneliness, and 1.32 for living alone, with effects persisting after adjustment for health status and, notably, running larger in samples averaging under 65.

The American Heart Association’s 2022 scientific statement, led by Crystal W. Cene, converted the same body of work into cardiovascular terms: roughly a 30% increased risk of heart attack, stroke, or death from either, with a 29% increase in heart attack or coronary death and a 32% increase in stroke, plus worse prognosis among people who have already had one. The AHA newsroom summary published alongside it noted that older adults and socially vulnerable groups carry elevated exposure.

That statement also did something the advisory does less prominently. It identified the absence of intervention evidence as the central research gap in the field. Not one gap among several — the gap.

Holt-Lunstad’s 2021 review in the American Journal of Lifestyle Medicine argues that social connection should sit alongside diet, exercise, and smoking in preventive frameworks, as a modifiable protective factor. The observational evidence establishes that it is a predictive factor. Whether it is modifiable at population scale, and by what, is a separate empirical claim requiring a different kind of study.

The instrument already on the shelf

If a U.S. health system is asked to act on loneliness, the mechanism most likely to be reached for is social prescribing: a clinician identifies social need and refers a patient to a non-clinical community activity, usually via a link worker. This is not a hypothetical. The UK’s A Connected Society strategy, published in October 2018 as the first national loneliness strategy by any government, funded social prescribing and embedded loneliness measurement into the Office for National Statistics. Japan created a loneliness ministerial post in February 2021, appointing Tetsushi Sakamoto, and the two governments held the first bilateral ministerial meeting on loneliness policy that June. The policy template exists and has been running for close to five years.

The National Academies’ 2020 consensus report on older adults gave the American version of the clinical ask, calling on the health care system to routinely assess social isolation and loneliness. It estimated that roughly a quarter of adults aged 65 and older are socially isolated. A clinician-facing commentary in the American Journal of Geriatric Psychiatry later that year worked through what routine assessment would actually require in practice — which is where the difficulty becomes concrete, because screening only makes sense if something effective sits on the other side of the referral.

Three syntheses published in 2021 and 2022 are the relevant evidence, and they should be read for their design as much as their conclusions.

A systematic review in Perspectives in Public Health in June 2021 examined social prescribing initiatives specifically against loneliness. All nine included studies reported positive individual impacts, and three reported reductions in GP, emergency, social worker, or inpatient service use. Nine studies, all positive, is a result that ought to prompt a question about what kinds of evaluations get designed and published in this area rather than confidence that the effect is robust.

A systematic review in the International Journal of Environmental Research and Public Health, published a month earlier, reported increases in self-esteem and self-confidence as the key outcomes across social prescribing programmes, while explicitly noting limited trial evidence and substantial heterogeneity between schemes. Self-esteem and self-confidence are plausible intermediate outcomes. They are not loneliness, and they are not isolation, and a programme can move them without changing either.

The most useful of the three for design purposes is the qualitative meta-synthesis in BMC Health Services Research from October 2022. Participants described benefit that extended past social contact into restored meaningful participation and purpose, and the synthesis concluded that structured, purposeful group activity appears more effective than contact alone. That is a testable hypothesis about active ingredients, generated from participant accounts. It is not yet a finding from a controlled comparison, and the distinction matters if the policy question is what to commission.

It fits, at least, with the theoretical account. John Cacioppo and William Patrick’s 2008 book treated loneliness as an aversive signal evolved to motivate reconnection, comparable to hunger — which implies that the signal responds to relationships being restored, not to attendance being logged.

Screening depends entirely on the instrument chosen

The advisory’s headline figure of approximately half of adults is one of several numbers now circulating, and the spread between them is a measurement artefact before it is anything else.

The AARP Foundation’s 2018 survey of 3,020 adults aged 45 and older used the 20-item UCLA Loneliness Scale and found one in three lonely, making it directly comparable to the academic literature. Harvard’s Making Caring Common report of February 2021 found 36% of Americans reporting serious loneliness, including 61% of adults aged 18 to 25. Cigna’s report published in January 2020, describing data collected before the pandemic, put the figure at 61% of adults sometimes or always feeling lonely, with 73% of workers aged 18 to 22 — a much looser threshold. The Survey Center on American Life’s 2021 friendship survey measured something different again: 12% of Americans with no close friends, up from 3% in 1990, and 15% of men.

A screening programme’s caseload — and therefore its cost — is determined by which of these instruments and cut-points a health system adopts. A 20-item validated scale with a defined threshold produces a different population from a single item asking whether a person sometimes feels lonely. The 2020 National Academies isolation estimate of a quarter of over-65s is a network measure and picks out partly different people again. Since Holt-Lunstad’s 2015 analysis shows isolation and loneliness are independently predictive of mortality, screening for one does not identify those at risk from the other, and the interventions they imply diverge: a thin network is a structural problem, and subjective loneliness can persist in a dense one.

The infrastructure pillar has the same problem in a different form

The case for social infrastructure is intellectually strong and empirically observational. Ray Oldenburg’s 1989 account of third places, Robert Putnam’s 2000 documentation of declining civic participation and informal socialising, and Eric Klinenberg’s 2018 analysis of libraries and parks — including his reading of mortality patterns in the 1995 Chicago heat wave — collectively make it very plausible that shared physical space shapes rates of social contact. None of it constitutes evidence that a given capital investment reduces loneliness by a measurable amount over a defined period. That study has largely not been done.

The pandemic did not fix this. A December 2022 editorial in Frontiers in Public Health introduced fourteen papers on isolation, loneliness, and mental health collected during COVID-19 — a substantial addition to the observational stock, and almost entirely cross-sectional.

What would make the next advisory answerable

The most valuable design here is unglamorous: randomised or stepped-wedge evaluations of social prescribing with loneliness measured as a pre-registered primary outcome on a validated scale, follow-up long enough to detect relapse after the referral period ends, baseline stratification by isolation as well as loneliness, and a comparison arm that receives contact without structured purposeful activity, to test the BMC meta-synthesis hypothesis directly.

Longitudinal cohorts are also feasible. The JACSIS study published in March 2023 tracked isolation and loneliness prevalence in Japan across 2020 and 2021 and can therefore speak to change rather than a single snapshot — the design the U.S. currently lacks at national scale.

The UK has had a national strategy since 2018, statistical instruments embedded in the ONS since then, and funded social prescribing running through general practice. Five years of exposure data exist. That the field still cites nine small studies as its intervention base suggests the binding constraint on loneliness policy is not political will but evaluation design, and the United States is now in a position to build that in from the start rather than retrofit it.