The Screening Recommendation Built for Older Adults, Applied to a Population It Was Never Tested On

The 2023 Surgeon General’s advisory on social connection asked the health sector to treat loneliness as a clinical concern across the lifespan, not just among older patients. Pillar three of the accompanying National Strategy calls for training health care providers to recognize and address social disconnection, and the Department of Health and Human Services has published that strategy as applying broadly, without carving out an age threshold. On its face this is a reasonable extension: the Harvard Making Caring Common survey found 61% of young adults aged 18-25 reporting serious loneliness, higher than any other age group measured, and a 2026 eight-country study reported that nearly half of young adults report loneliness, with the loneliness-depression and loneliness-anxiety associations holding after adjustment for prior diagnoses. If any age group needs a clinical response, the numbers suggest young adults are it.

The question this brief asks is narrower: does the trial evidence support extending a screening and referral recommendation, specifically, to young adults, or does it only support recognizing that young adults are lonely.

Where the recommendation actually comes from

The clinical screening idea has a clear lineage, and it does not start with young adults. The National Academies of Sciences, Engineering, and Medicine’s 2020 consensus report, which is the most-cited foundation for routine assessment in health care settings, is titled Social Isolation and Loneliness in Older Adults and confines its evidence base to that population. Its recommendation that health systems routinely assess isolation and loneliness was built on studies of adults 65 and older, and a subsequent clinician-facing commentary in the American Journal of Geriatric Psychiatry discussed what routine assessment would require in geriatric and primary care settings specifically. Neither document makes a claim about adolescents or young adults, and neither was designed to.

The Surgeon General’s advisory broadened the framing to “connection across the lifespan,” but the advisory is a call to action, not a systematic review, and it does not present young-adult-specific intervention trials as its evidentiary basis. What it cites for prevalence and risk is largely survey data. What it cites for what to do clinically leans on the older-adult isolation literature because that is where the intervention research exists.

What the young-adult evidence base actually contains

The prevalence data on young adults is genuinely strong by the standards of this field. The Harvard MCC 2021 survey, the American Enterprise Institute’s 2021 survey on American friendship, Cigna’s 2025 national survey, and the 2025 HINTS-6 analysis all converge on the same basic pattern: loneliness is higher among younger adults than older ones, and it has been rising in cross-sectional comparisons since at least 1990, when the AEI survey’s baseline shows only 3% of Americans reporting no close friends, against 12% in 2021. That is prevalence and trend evidence, and it is reasonably consistent across instruments.

What is missing is the other half of a clinical recommendation: evidence that screening young adults in a health care setting, and then referring them to something, changes an outcome. The randomized controlled trials that exist test interventions in older populations. The HEAL-HOA trial in Hong Kong tested prosocial engagement and volunteering against a control among lonely older adults. A 2026 follow-up HEAL-HOA trial tested telephone-delivered behavioral activation and mindfulness against a befriending control in 1,151 older adults living in poverty and digital exclusion, finding the behavioral intervention outperformed befriending at 12 months. A separate randomized trial of befriending in residential aged care reduced UCLA Loneliness Scale scores by 2.39 points at eight weeks. All three trials recruited people over 60, in settings — aged care, poverty, digital exclusion — that describe almost no young adult.

The closest thing to young-adult intervention evidence is a 2025 study in Frontiers in Psychology examining which kinds of social connection predicted wellbeing among college students coming out of the pandemic. That is useful and directly relevant to the population in question, but it is an observational study of association, not a trial of a clinical screening pathway. It cannot say whether asking a 20-year-old about loneliness in a primary care visit, and referring them somewhere, changes anything measurable.

What that gap means for the recommendation

This is the pattern the field keeps producing: a well-supported prevalence claim borrowed to justify a clinical action whose supporting trials were run on a different population. It is not evidence that the recommendation is wrong for young adults. Loneliness in that age group tracks depression and anxiety closely enough, per the 2026 eight-country study, that a plausible mechanism exists. But plausibility is not the same as demonstrated benefit, and the specific mechanics that worked in aged-care trials — telephone-delivered behavioral activation, structured volunteering, befriending — were designed around older adults’ schedules, mobility, and social contexts. A college student’s isolation looks structurally different from a homebound 80-year-old’s, and there is no trial evidence indicating which, if any, of the existing interventions transfers.

Foundation for Social Connection and other groups translating the Surgeon General’s advisory into practice have generally been careful to note this is an emerging area rather than a settled one. What would close the gap is not another prevalence survey — that evidence is already adequate — but a randomized trial of a screening-and-referral pathway conducted specifically in primary care, college health, or urgent care settings that serve adults under 30, measuring loneliness at baseline and follow-up against a comparison arm. Until that trial exists, the honest description of the current recommendation is that it rests on strong prevalence data and borrowed intervention evidence from a different age group, not on evidence that the recommended clinical action works for the population it is now being extended to cover.