What the Loneliness Literature Actually Says About Cost

Cigna’s 2020 workplace survey reported that lonely workers miss work twice as often due to illness and five times as often due to stress. That is a striking pair of numbers, and it gets cited constantly in arguments for workplace connection programs, wellness apps, and remote-work policy. It is also, on inspection, not a cost estimate. It is a self-reported absenteeism ratio from a single cross-sectional survey, with no wage data, no productivity multiplier, and no attempt to translate missed days into dollars. Anyone using it to argue that loneliness costs employers a specific figure is doing arithmetic the source does not support.

This is the pattern across the literature on the economics of loneliness and social isolation: strong evidence on health association, thin evidence on health causation for any specific intervention, and almost no evidence at all on cost. The three tiers do not get distinguished nearly as often as they should.

The health evidence is comparatively solid

Holt-Lunstad’s 2015 meta-analysis, pooling many prior studies, found social isolation associated with an odds ratio of 1.29 for early mortality, loneliness at 1.26, and living alone at 1.32, with the effects persisting after adjustment for baseline health. The American Heart Association’s 2022 scientific statement, reviewing cardiovascular evidence specifically, put the isolation-and-loneliness risk at roughly 30% for heart attack, stroke, or death from either. These are meta-analyses and systematic reviews with defined statistical methods, large pooled samples, and explicit handling of confounding. Whatever else is uncertain in this field, the association between social disconnection and mortality risk is not.

The same AHA statement, though, is unusually candid about what comes next. It names the absence of intervention evidence as the central research gap in the field: knowing that isolation predicts worse cardiovascular outcomes is not the same as knowing that any specific program reverses that risk, and the statement says so plainly rather than assuming the two are equivalent. That gap between association and actionable intervention evidence is where cost estimates would need to sit, and it is exactly where they are missing.

What passes for cost evidence in social prescribing

Social prescribing is the closest thing this field has to a health-system-level intervention with a paper trail, and its evidence base is instructive for how far short of true economic evaluation most of this research falls. A 2021 systematic review in Perspectives in Public Health found that all nine included studies reported positive individual-level impacts, and three of those nine reported reductions in GP visits, emergency department use, social worker contact, or inpatient admissions. That is a real signal, and it points toward downstream cost savings. But a reduction in utilization reported by three uncontrolled or weakly controlled studies is not a cost-effectiveness analysis. None of the nine studies monetized the savings, adjusted for what those patients would have cost the system absent the program, or compared the intervention’s cost to its benefit in a common unit.

A companion 2021 review in the International Journal of Environmental Research and Public Health, focused on individual and community well-being outcomes, reports gains in self-esteem and confidence as the main documented effects, and is explicit about the limited trial evidence and heterogeneity across programs. A 2022 qualitative meta-synthesis in BMC Health Services Research adds that participants describe benefits extending beyond social contact to restored purpose and participation, which is useful for understanding mechanism but says nothing about cost. Most tellingly, a 2025 systematic review protocol on social prescribing for older adults notes that despite growing adoption across health systems, its effectiveness remains unclear, and that only one peer-reviewed randomized controlled trial exists in this specific area. A field cannot have settled cost-effectiveness findings when it does not yet have settled effectiveness findings, and this protocol says as much directly.

Technology-mediated work: two studies, no dollar figures

The technology angle sharpens rather than resolves the problem. Gallup’s 2024 State of the Global Workplace survey found that fully remote employees report loneliness at 25%, hybrid employees at 21%, and fully on-site employees at 16%, with loneliness generally higher among employees under 35. A 2025 nationally representative study in the Journal of Affective Disorders examined the association between remote work arrangements and loneliness in employed U.S. adults directly, and a 2024 cross-sectional study of healthcare workers usefully separated workplace isolation from loneliness as distinct constructs with different correlates, finding that perceived social support moderates the relationship between remote arrangements and well-being.

These are genuinely useful studies for understanding the shape of the problem. None of them contains a cost figure. None estimates turnover cost, lost productivity in dollar terms, or healthcare spending attributable to remote-work-associated loneliness. The Gallup figures are prevalence rates; the JAD and healthcare-worker studies are association studies. Any claim that remote work “costs” employers a specific sum through loneliness-driven attrition is importing a number from outside this evidence base, because the evidence base does not contain one.

The same caution applies to AI companionship tools, an increasingly common technology response to loneliness. George Mason University’s 2025 commentary frames AI companionship as a substitution risk rather than a straightforward remedy, a public-health framing rather than an economic one. There is no cost or cost-offset study in the current literature addressing AI companionship tools at all, which means claims about their economic value to health systems or employers rest on nothing published.

The trials that could have measured cost, and didn’t

Two recent randomized controlled trials are worth flagging precisely because they are methodologically strong enough that a cost-effectiveness arm would have been possible and valuable. The HEAL-HOA trial, published in The Lancet Healthy Longevity in 2024, tested prosocial engagement and volunteering against a control among lonely older adults in Hong Kong, and stands out simply for being a randomized trial in a literature otherwise dominated by uncontrolled program evaluations. A 2025 randomized trial in Clinical Gerontologist found that befriending in residential aged care reduced UCLA Loneliness Scale scores by 2.39 points at eight weeks and 2.71 points at sixteen weeks relative to control, a real and specific effect on a validated instrument.

Neither trial reports a cost-per-unit-of-loneliness-reduction, a cost-per-quality-adjusted-life-year, or any other health-economic outcome. Both were designed to answer “does this work,” not “is this worth what it costs,” and they answer the first question credibly. The second question remains open, and given how few randomized trials of any loneliness intervention exist, it is likely to remain open for some time.

What a real cost estimate would need

A defensible cost-effectiveness study of a loneliness intervention would need a randomized or well-matched comparison group, a validated loneliness or isolation instrument measured at baseline and follow-up, a monetized cost of delivering the intervention, and a monetized estimate of downstream savings, whether in health care utilization, absenteeism, or another quantified channel, ideally expressed as a standard health-economic ratio so it can be compared across interventions. Nothing in the current literature does all four. The social prescribing reviews get partway there with utilization data but no monetization. The RCTs get partway there with rigorous effect sizes but no cost data. The workplace surveys have absenteeism ratios but no experimental design and no monetization either. Until a study combines rigor on effect with rigor on cost, the widely repeated economic claims about loneliness, however intuitively plausible, remain estimates borrowed from adjacent fields and applied without the qualifications they were built with.