How Lonely Is America? Two Numbers, Two Instruments

Two figures, both describing loneliness in the United States, both released in 2025, sit roughly twenty percentage points apart. The Cigna Group’s Loneliness in America 2025 reports that 57% of American adults report loneliness. A study published in the American Journal of Preventive Medicine using the 2022 Health Information National Trends Survey (HINTS-6) puts moderate-to-severe loneliness at just over 37%, with roughly one in seven adults classified as severely lonely and one in five as moderately lonely.

Neither number is wrong. They are answers to different questions, asked with different tools, of different populations, at different points in time. Treating them as competing estimates of a single true prevalence rate misses the more useful lesson: in this literature, the instrument determines the answer almost as much as the underlying reality does.

What each study actually measured

Cigna’s fieldwork ran from 29 May to 14 June 2024, drawing on more than 7,500 U.S. adults, though the report itself was published in 2025. The HINTS-6 analysis is built on data collected in 2022 and published in November 2025 — a three-year gap between data collection and publication that is easy to miss if a reader only registers the byline date.

That gap matters less here than the choice of instrument. Cigna’s loneliness figure derives from its own proprietary index, developed for a corporate research and communications programme and not identical across successive years of the report, which makes trend comparison within Cigna’s own series harder than it looks. The HINTS-6 study, by contrast, uses loneliness items embedded in a federally administered survey designed primarily for cancer communication and health information research, with defined severity thresholds — none, moderate, severe — that allow the authors to report a distribution rather than a single yes/no rate.

A single combined “moderate-to-severe” figure of 37% is the number closest in spirit to Cigna’s headline 57%, since both are attempts to describe how much of the adult population experiences some meaningful degree of loneliness. Even so, the gap is large. Part of the explanation is simply that “some loneliness” is a generous threshold in a proprietary index built to produce a communicable topline, while HINTS-6’s severity bands were constructed to be more conservative and clinically legible.

The instrument problem is not new

This is not the first time two credible sources have produced wildly different loneliness prevalence figures. AARP’s 2018 survey of adults 45 and older found that one in three reported being lonely, using the 20-item UCLA Loneliness Scale — the closest thing this field has to a shared academic standard, precisely because it is used across dozens of studies and permits comparison over time. AARP’s 2025 follow-up, using the same instrument on the same age group, is directly comparable to 2018 in a way that Cigna-to-Cigna or Cigna-to-HINTS comparisons are not, and it reports rising loneliness in that population. That comparability is rare in this literature and worth naming as a model rather than a footnote.

The Surgeon General’s 2023 advisory cited “approximately half” of U.S. adults reporting loneliness, a figure that sits almost exactly between Cigna’s 57% and the HINTS-6 37%. That the advisory’s rounded, hedged framing lands in the middle is not a coincidence; it is what happens when a policy document synthesizes across studies that do not use the same yardstick. A 2023 review in BMC Public Health made the underlying point explicit: inconsistent measurement across the loneliness and social isolation literature is a structural barrier to comparing findings, not a peripheral inconvenience.

Why this is not just noise

It would be a mistake to conclude that because the numbers disagree, the underlying phenomenon is unmeasurable or the studies are interchangeable. Both surveys likely capture something real. But they are not measuring the same construct at the same threshold.

A proprietary index built by a health services company for a public-facing annual report is optimized for a clean, communicable statistic — one number, one year, one headline. A federally fielded survey instrument embedded in a broader health-information questionnaire is optimized for internal consistency across severity bands and for downstream epidemiological use, at the cost of being less quotable. Neither goal is illegitimate. But a reader who takes “57% of Americans are lonely” and “37% of Americans report moderate-to-severe loneliness” as two attempts to answer the identical question will end up either confused or falsely confident that one source got it wrong.

The more defensible reading is that the true share of U.S. adults experiencing some meaningful loneliness sits somewhere in a fairly wide band — roughly a third to a bit over half, depending on where the threshold for “lonely” is drawn — and that this band itself is the finding. Any single point estimate presented without its instrument attached is doing more rhetorical work than the data supports.

What would resolve it

The clean way to settle this is not more surveys with more respondents; both existing studies already have adequate sample sizes. It is convergence on instrument. If Cigna’s future editions adopted the UCLA scale, or reported findings at the same severity thresholds HINTS-6 uses, the two data streams would become genuinely comparable, the way AARP’s 2018 and 2025 surveys already are within their own series. Absent that, the responsible move for anyone citing loneliness prevalence is to name the instrument in the same sentence as the number — not “57% of Americans are lonely” but “57% by Cigna’s index” — and to resist collapsing distinct measurement traditions into a single figure that implies more precision than the underlying data supports.