Five Years of Reports, One Missing Trial

In February 2020, the National Academies of Sciences, Engineering, and Medicine published a consensus report estimating that roughly one quarter of adults aged 65 and older in the United States are socially isolated. The report’s central recommendation was that the health care system routinely screen for isolation and loneliness, the way it screens for depression or fall risk. Its central caveat, buried a few pages later, was that almost no rigorous evidence exists on what to do once a clinician identifies an isolated patient.

Five years on, that caveat has not been resolved. It has been repeated, with minor variation, in nearly every major report since.

The same sentence, restated

The American Heart Association’s 2022 scientific statement on social isolation and cardiovascular and brain health reviewed evidence linking isolation and loneliness to roughly a 30% increased risk of heart attack, stroke, or death from either — a 29% increase in heart attack or cardiac death, a 32% increase in stroke. That is a substantial, well-replicated association. But the statement’s authors, writing on behalf of multiple AHA councils, went out of their way to identify the absence of intervention evidence as the central research gap. They had strong evidence that isolation predicts disease and almost no evidence about what reverses it.

The 2023 Surgeon General’s advisory on the epidemic of loneliness and isolation made a similar move at a different scale, comparing the mortality risk of social disconnection to smoking up to 15 cigarettes a day and proposing a six-pillar national strategy. The strategy is programmatic — strengthen social infrastructure, enact pro-connection public policy, mobilize the health sector — rather than evidentiary. It describes what should be tried, not what has been shown to work at the level of a controlled trial.

A 2023 review in BMC Public Health, surveying the state of loneliness and social isolation research more broadly, reached the same conclusion from the methods side: inconsistent measurement across studies is itself a barrier to knowing whether anything works, because studies claiming to reduce “isolation” often measure loneliness instead, or vice versa, using instruments that were not designed for comparison.

This is not four reports independently discovering the same problem. It is one problem, restated by four different bodies with different mandates, none of which has been positioned or funded to fix it.

Why the systematic reviews cannot close the gap

Social prescribing — the practice of a health or social worker referring an isolated or lonely patient to a community group, class, or activity rather than a clinical treatment — is the most widely deployed intervention in this space, and it has been reviewed repeatedly. A 2021 systematic review in the International Journal of Environmental Research and Public Health found improvements in self-esteem and self-confidence associated with social prescribing programs. A separate 2021 systematic review in Perspectives in Public Health found that all nine included studies reported positive individual impacts, and three reported reductions in use of GP, emergency, social worker, or inpatient services. A 2022 qualitative meta-synthesis in BMC Health Services Research added that participants describe benefits extending beyond social contact itself to a restored sense of purpose, suggesting that structured, purposeful activity outperforms unstructured contact.

These are genuinely encouraging findings. They are also, by design, not able to establish causation. The overwhelming majority of the underlying studies are uncontrolled program evaluations: people were referred, some outcome improved, and the study reports the improvement. Without a control group, there is no way to separate the effect of the program from the effect of being the kind of person who accepts a referral, or from regression to the mean, or from concurrent changes in health status. Systematic review is a tool for aggregating what exists; it cannot manufacture a randomized comparison that was never run.

The one trial that counts, and what it does not tell us

Against that backdrop, the November 2024 HEAL-HOA trial published in The Lancet Healthy Longevity stands out mainly for its design. It is a randomized controlled trial — one of a small number in this literature — testing volunteering and prosocial engagement against a control condition among lonely older adults in Hong Kong. Its importance lies less in the specific result than in the fact of the method: it is the kind of study the National Academies, the AHA, and the Surgeon General have each said is missing.

One trial from one city cannot answer a question this broad. It does not tell health systems in the United States, the United Kingdom, or Japan — all of which have made loneliness a formal policy concern — whether volunteering programs would produce comparable effects in different cultural and health care contexts, at different scales, or over longer follow-up periods. It is a proof of concept for the method, not yet a evidence base.

What the persistence of the gap implies

There is a structural reason this keeps happening. Isolation and loneliness interventions are cheap to run as community programs and expensive to test as trials — recruitment, randomization, and long follow-up for an outcome like cardiovascular events or mortality require resources that most social prescribing pilots never had. The consensus reports have consistently called for isolation to be treated as a modifiable risk factor on par with smoking or inactivity, a framing echoed in Julianne Holt-Lunstad’s meta-analytic work establishing that social isolation carries a mortality odds ratio around 1.29 independent of baseline health. But the prevention infrastructure built around smoking cessation — decades of randomized pharmacological and behavioral trials — has no counterpart here.

What would actually close this gap is not another synthesis of observational studies; the observational base is already large and consistent enough to support the underlying claim that connection matters for health. What is missing is a portfolio of adequately powered randomized trials, run across more than one health system, that test specific interventions — volunteering, group referral, structured versus unstructured contact — against genuine controls, with follow-up long enough to capture the cardiovascular and cognitive outcomes the observational literature has already flagged as at stake. Until that portfolio exists, health systems adopting screening and referral programs on the strength of the current evidence are, in effect, running the trial themselves, one uncontrolled rollout at a time.