The Cost Case for Social Connection Is Still Mostly Missing

The 2023 Surgeon General’s advisory put a number on the health risk of disconnection that has since been repeated in nearly every article on the subject: a mortality risk comparable to smoking up to 15 cigarettes a day. That figure has traveled well because it is vivid and because it rests on real meta-analytic work — Julianne Holt-Lunstad’s 2010 review of 148 studies and more than 308,000 participants, and her 2015 follow-up putting the odds ratio for social isolation at 1.29 and for loneliness at 1.26. Those are large, reasonably consistent numbers built from a genuinely large evidence base.

No comparable number exists for cost. Ask what it would save a health system, an employer, or a government to reduce isolation at population scale, and the literature has almost nothing to offer with the same confidence. This is worth stating plainly, because the health claims and the cost claims are frequently presented in the same breath, as if they rested on the same kind of evidence. They do not.

Where the chain breaks

The logic usually runs: isolation and loneliness raise the risk of cardiovascular disease, stroke, and early death; those conditions are expensive to treat; therefore reducing isolation should reduce cost. Each link is plausible. The first is well supported. The second is trivially true. The third is the one nobody has actually measured at scale.

The American Heart Association’s 2022 scientific statement is unusually candid about this. It reports that social isolation and loneliness carry roughly a 30% increased risk of heart attack, stroke, or death from either — 29% for heart attack or death from heart disease, 32% for stroke — and it explicitly names the absence of intervention evidence as the central gap in the field. That is a scientific body summarizing hundreds of association studies and then saying, in effect: we know the risk factor is real, and we do not know what happens, in dollars or in outcomes, when someone intervenes on it. A statement that confident about the gap deserves to be taken at face value rather than papered over by inference.

The National Academies’ 2020 consensus report on older adults makes a related point from the clinical side. It calls on the health care system to routinely assess isolation and loneliness — sensible, given how disproportionately older adults screen positive for isolation, at roughly one in four by that report’s estimate — but its supporting commentary, published later that year, is candid about what would be required to make that “opportunities for the health care system” framing accurate: sustained assessment, workflow integration, referral pathways, and a data infrastructure to track downstream effects, essentially none of which existed at the time the report was written. Recommending assessment is not the same as demonstrating that assessment, followed by intervention, reduces cost.

What the social prescribing evidence actually shows

Social prescribing — the practice of a clinician referring a patient to a community group, activity, or social program rather than, or alongside, a medical treatment — is the closest thing the field has to an economic test case, because it has been running long enough in the UK to generate multiple systematic reviews. The results are informative mainly for how modest they are.

A 2021 systematic review in the International Journal of Environmental Research and Public Health found the clearest outcome across included studies was increased self-esteem and self-confidence — a real finding, but not a cost figure, and the review notes limited trial evidence and substantial heterogeneity across programs, meaning the studies are not really measuring the same intervention. A second 2021 review, in Perspectives in Public Health, looked specifically for service-use effects and found that of nine included studies, three reported reductions in GP visits, emergency department use, social worker contact, or inpatient admissions. Three out of nine is not nothing, but it is also not the basis for a defensible savings estimate, particularly because none of the nine used a randomized design with a proper economic evaluation arm — the standard tool for turning a service-use reduction into a cost figure that a health system could act on.

A 2022 qualitative meta-synthesis in BMC Health Services Research adds a different kind of caution. Participants in social prescribing programs described benefit that went beyond social contact itself — restored purpose, meaningful participation — and the synthesis suggests structured activity with a purpose attached works better than unstructured contact. That is a useful design insight. It is also, again, not an economic finding. Qualitative meta-synthesis is the wrong tool for estimating averted spending, and none of the reviews claims otherwise.

The instrument problem compounds the cost problem

Part of why cost evidence lags so far behind mortality evidence is a measurement issue the field has been slow to resolve. A 2023 review in BMC Public Health mapping the state of loneliness and isolation research names inconsistent measurement as a structural barrier to comparing findings across studies. If isolation is measured with the UCLA scale in one study, a single-item self-report in another, and a network-size proxy in a third, then “did the intervention reduce isolation” cannot be answered consistently, let alone “did it reduce cost.” Economic evaluation depends on being able to attach a stable, comparable outcome to a stable, comparable exposure. The field has neither yet, for this particular question.

What would actually settle it

A defensible cost estimate would need a randomized trial — not an observational cohort — comparing a defined social connection intervention against a control, using a single validated loneliness or isolation instrument, tracked long enough to observe health service utilization and mortality, with a prespecified economic evaluation built in from the start rather than bolted on afterward. Nothing in the current literature does all of that. The mortality risk case for taking isolation seriously is arguably one of the stronger associations in public health, at Holt-Lunstad’s scale of evidence. The cost case for any particular intervention remains an inference from a much thinner and much more heterogeneous body of work, and the AHA’s own statement is right to say so.