Why Older Adults Score Lowest on Loneliness and Highest on Isolation

Gallup’s 2023 global survey found that adults 65 and older report the lowest rate of loneliness of any age group: 17%, against 27% among 19- to 29-year-olds, across 142 countries. The World Health Organization’s 2025 Commission on Social Connection report reached a similar conclusion at global scale, noting that loneliness rates are highest among young people and in low-income countries. Read quickly, this suggests older adults are the age group least in need of attention.

The National Academies of Sciences, Engineering, and Medicine’s 2020 consensus report says almost the opposite: roughly one quarter of U.S. adults aged 65 and older are socially isolated. That figure did not come from asking people whether they felt lonely. It came from counting who they see, how often, and how large and varied their networks are. Both figures are correct. They are not measuring the same thing, and the gap between them is the most consequential measurement problem in this literature as applied to older adults.

Two questions, two instruments

Loneliness surveys ask a subjective question, in some version of “how often do you feel lonely.” The UCLA Loneliness Scale, used in AARP’s 2018 and 2025 surveys of adults 45 and older, asks 20 items along these lines and produces a single score. Isolation measures ask a structural question: network size, contact frequency, living arrangement, participation in groups. NASEM’s 25% isolation estimate for older Americans comes from this second family of instruments.

A 2024 study in Scientific Reports, examining the interplay between isolation and loneliness during the pandemic, found that the relationship between the two varies by age rather than holding constant across the life course. A companion 2026 study on risk factors among community-dwelling socially isolated older adults goes further, identifying which isolated older adults go on to become lonely and which do not — treating isolation as the input and loneliness as one possible, but not automatic, outcome. Isolation without loneliness is common in older age: a person can have a small network, see few people, and not feel distressed by it. Loneliness without isolation also occurs, particularly in midlife caregiving contexts. But by NASEM’s own logic, the two diverge most sharply at the older end of the age range, which is exactly where global loneliness surveys report the lowest rates.

What AARP’s own instrument shows over time

AARP’s advantage in this literature is rare: it used the same 20-item UCLA Loneliness Scale in 2018 and again in its 2025 follow-up survey, “Disconnected.” That consistency means the two AARP surveys are directly comparable to each other in a way that most instrument comparisons across this field are not. The 2018 survey, based on 3,020 adults 45 and older, found one in three lonely, with network size and diversity and physical isolation as the strongest predictors — and a striking gradient by neighbourly contact: 33% of respondents who had spoken with neighbours were lonely, against 61% of those who never had. The 2025 follow-up reports loneliness rising further in this population. Within a single, held-constant instrument, older Americans look considerably lonelier than Gallup’s global cross-sectional snapshot of the 65-plus bracket would suggest.

Part of the discrepancy is geographic. Gallup’s figure is a 142-country average pulling in populations with very different baseline social structures; AARP’s is U.S.-specific. Part of it is also generational framing within the same age band — AARP surveys people from 45 upward, a range that includes a large working-age population, while Gallup’s 65-plus bracket excludes it. Neither survey is wrong. They answer different questions about different populations using different instruments, and stacking their headline numbers side by side without noting that produces a false comparison.

The United States’ own numbers do not agree with each other

Even within a single country, instrument choice moves the number substantially. The CDC’s Morbidity and Mortality Weekly Report, published in 2024 using 2022 survey data, produced federal surveillance estimates of loneliness and lack of social and emotional support among U.S. adults. A separate analysis of the 2022 Health Information National Trends Survey (HINTS-6), published in the American Journal of Preventive Medicine in November 2025, found roughly one in seven U.S. adults reporting severe loneliness and one in five reporting moderate loneliness, for a combined moderate-to-severe rate above 37%. Both draw on 2022 data. Both are federally sponsored. They still do not use identical loneliness categorisations, and readers comparing the CDC’s surveillance framing against HINTS-6’s severity tiers without checking the underlying instrument will draw conclusions the data cannot support.

Japan’s longitudinal advantage

Most of what drives cross-national confusion is that survey waves are cross-sectional snapshots taken once, in different years, with different wording. Japan’s JACSIS study, tracking the same population from 2020 to 2021, is one of the few sources in this area that can speak to change over time within a fixed cohort rather than comparing one country’s 2018 snapshot against another’s 2023 snapshot. That longitudinal design is a genuinely different kind of evidence from the repeated cross-sectional approach used by AARP, Gallup, and the CDC, and it deserves to be weighted differently when policy documents cite it — not folded into the same table of percentages as if all these figures were fungible.

Why this matters for where money goes

The WHO’s 2025 report puts the global toll of loneliness at an estimated 871,000 deaths annually and calls on member states to treat social connection as a public health priority. That call, and NASEM’s parallel call for the health care system to routinely screen for isolation, point in a specific direction: toward identifying isolated older adults, not merely lonely ones, because isolation is the more stable and more measurable structural condition, and the mortality literature associates it with elevated risk independent of self-reported distress. If national surveys keep reporting that older adults are the least lonely age group, and policy attention follows loneliness prevalence rather than isolation prevalence, older adults who are structurally cut off but not subjectively distressed will fall outside the net that screening programmes are designed to catch.

A better cross-national comparison would report both figures side by side for every age band, using a harmonised instrument for each construct rather than each country’s preferred survey — something closer to what AARP has managed within the United States by holding its own instrument constant across seven years, but extended internationally and paired with a structural isolation measure rather than a loneliness scale alone. Until that exists, any claim that older adults are “doing fine” on connection, or “in crisis,” should specify which of the two questions was asked, and whose instrument produced the answer.