Naming the Gap Does Not Close It: Isolation Interventions for Older Adults

In February 2020, the National Academies of Sciences, Engineering, and Medicine published a consensus report concluding that roughly one quarter of adults aged 65 and older are socially isolated, and called on the health care system to screen for isolation and loneliness routinely. A clinician-facing commentary in the American Journal of Geriatric Psychiatry, published the same year, echoed the call for routine assessment and discussed what implementing it would actually require.

Neither document could point to a strong body of controlled trial evidence showing what to do once a clinician identified an isolated patient. That absence was not an oversight. It was, and largely remains, the honest state of the field.

The gap keeps getting named, not closed

Two years after the National Academies report, the American Heart Association published a scientific statement on social isolation and cardiovascular and brain health, led by Crystal W. Cene on behalf of several AHA councils. It reported that isolation and loneliness carry roughly a 30% increased risk of heart attack, stroke, or death from either. It also did something less commonly emphasized in press coverage: it explicitly identified the absence of intervention evidence as the central research gap in the field, distinguishing it from the abundant observational evidence on risk.

That distinction matters. The observational literature linking isolation and loneliness to mortality and cardiovascular outcomes is large, consistent, and by 2022 already well established. The literature testing whether a specific intervention reduces isolation or loneliness, in a design capable of ruling out the alternative explanation that people who improve would have improved anyway, is thin.

Systematic reviews of social prescribing — the practice of a health or social worker referring a patient to a non-clinical community activity, often used specifically to address loneliness in older adults — have said versions of the same thing for years. A 2021 review in the International Journal of Environmental Research and Public Health reported increases in self-esteem and self-confidence among participants but noted limited trial evidence and heterogeneity across programmes. A companion systematic review the same year, in Perspectives in Public Health, found that all nine included studies reported positive individual impacts and three reported reduced use of GP, emergency, social worker, or inpatient services — but the same review’s design constraints meant it was assembling uncontrolled or weakly controlled studies, not a body of randomized evidence. A 2022 qualitative meta-synthesis in BMC Health Services Research found that participants describe benefits extending beyond social contact to restored purpose and meaningful participation, which is a real finding, but it is a finding about perception, not about causal effect size.

By July 2025, a systematic review protocol posted to medRxiv was still making essentially the same observation the National Academies had made five years earlier: the effectiveness of social prescribing for older adults remains unclear despite growing adoption, and only one peer-reviewed randomized controlled trial exists in this specific area. Five years, several reviews, one trial. The gap did not close. It was simply re-described with each new publication.

What changed, narrowly, in 2024 and 2025

Two trials have since appeared that begin to fill the space the reviews kept pointing at, and it is worth being precise about what each one actually tested.

The HEAL-HOA dual randomized controlled trial, published in The Lancet Healthy Longevity in November 2024, tested prosocial engagement and volunteering against a control condition among lonely older adults in Hong Kong. It is one of the few RCTs of a loneliness intervention rather than an uncontrolled programme evaluation, and its existence is itself a data point about how rare this design still is in the field.

A December 2025 randomized controlled trial in Clinical Gerontologist tested befriending — trained volunteers providing regular social contact — against a control group among older people in residential aged care. Befriending reduced UCLA Loneliness Scale scores by 2.39 points at eight weeks and 2.71 points at sixteen weeks relative to control. That is a real, measured effect on a validated instrument, which is more than most of the social prescribing literature can offer.

Then, in a follow-up HEAL-HOA trial published in March 2026, telephone-delivered behavioral activation and mindfulness were tested against a befriending control among 1,151 older adults who were living in poverty, alone, and digitally excluded. The behavioral intervention — eight 30-minute telephone sessions delivered by trained laypeople who were themselves older adults experiencing loneliness — significantly reduced loneliness at 12 months compared with befriending. This result is worth sitting with, because befriending is the intervention most health systems and charities already deploy at scale. In a head-to-head randomized comparison, it lost to a more structured, psychologically grounded alternative.

Three trials, three different active conditions, three different comparators, three different follow-up windows — eight weeks, sixteen weeks, twelve months. None of them shares a control condition with another. This is not a body of evidence converging on an answer. It is three separate signals, each interesting on its own terms, that cannot yet be pooled into a claim about what works best for whom.

Why campaign-level evaluation does not fill the gap either

A separate strand of evidence tries to answer a related but distinct question: does population-level loneliness campaigning, rather than individual-level intervention, change outcomes? A 2023 difference-in-differences evaluation of the UK’s Campaign to End Loneliness, published in The American Journal of Geriatric Psychiatry, is one of the only quasi-experimental attempts to answer this at the campaign level rather than the programme level. It is valuable precisely because it tests the assumption that awareness campaigns move outcomes, which most policy documents simply take for granted. But a difference-in-differences design on a national campaign answers a different question than a randomized trial of a specific clinical or community intervention, and conflating the two would repeat the same error the field has been making with observational mortality data: treating association-level evidence as though it settles a causal, individual-level question.

What would actually close the gap

A genuinely comparable evidence base for isolation and loneliness interventions in older adults would need three things the current literature still lacks. First, a shared outcome measure across trials — the UCLA Loneliness Scale is available and validated, and its use in both the 2025 befriending trial and elsewhere shows it is feasible, but it is not yet the default. Second, active comparators rather than usual-care or waitlist controls, so that trials can say which intervention outperforms another rather than only that something outperforms nothing. Third, separate measurement of isolation as a structural property of a person’s network and loneliness as their subjective state, since the 2026 review of risk factors in socially isolated older adults and related work have shown the two do not move together in any simple way, and an intervention that reduces one may not touch the other.

Until reviews can point to a set of trials that share a comparator and an outcome measure, “the intervention evidence is thin” will keep being the correct sentence to write, five years from now, about a fifth review.