Isolated Is Not Lonely: Untangling Two Statistics About Older Adults

Two figures circulate constantly in writing about older adults and social connection. Roughly one in four adults aged 65 and older is socially isolated, according to the 2020 National Academies of Sciences, Engineering, and Medicine consensus report. Roughly one in three adults aged 45 and older is lonely, according to the AARP Foundation’s 2018 national survey of 3,020 midlife and older adults, which used the 20-item UCLA Loneliness Scale.

These numbers get cited together often, and usually as if they measured the same thing at two different thresholds. They did not. One counted the structure of a person’s social network. The other counted a subjective state. The distinction is not academic hair-splitting. It changes what an intervention should do, and it changes who counts as at risk.

What each instrument actually asked

Social isolation, in the National Academies’ usage, is an objective property of a person’s relationships: how many social ties they have, how often they see people, whether they live alone, whether they participate in groups or organizations. It can be counted from the outside. A researcher, or a clinician using a checklist, can assess it without asking the person how they feel about it at all.

Loneliness is the subjective gap between the connection a person has and the connection they want. The UCLA Loneliness Scale, the instrument AARP used, asks about felt states — how often someone feels left out, how often they feel that no one really knows them — not about the number of people in their address book. Cacioppo and Patrick’s account of loneliness as an aversive signal, akin to hunger or thirst, is built entirely on this subjective register: the discomfort exists to motivate reconnection, and it can exist in someone surrounded by people.

The AARP survey’s own findings illustrate the split without needing outside cross-referencing. Its two strongest predictors of loneliness were the size and diversity of a person’s social network, and physical isolation — objective network properties predicting a subjective state. But the survey also found that 33% of people who talk to their neighbors are lonely. A structurally connected life does not close the gap for a third of the people living it.

Why treating them as one number is a mistake

Holt-Lunstad’s 2015 meta-analysis is the paper that actually forces the separation, because it reports isolation and loneliness as distinct predictors with distinct odds ratios: social isolation carried an odds ratio of 1.29 for early mortality, loneliness 1.26, living alone 1.32. These are close in magnitude but not identical, and the fact that all three survived adjustment for health status as separate terms means they are not simply redundant measurements of the same underlying deficit. If loneliness were just isolation experienced unpleasantly, one term would absorb the other in a regression model. It does not.

The American Heart Association’s 2022 scientific statement makes the same move at the level of cardiovascular outcomes, treating “objective and perceived social isolation” as titled components rather than synonyms, and reporting roughly a 30% increased risk of heart attack, stroke, or death from either — a figure derived from evidence on both structural and subjective disconnection, not one or the other. The statement is explicit that the mechanism connecting either pathway to cardiovascular disease remains unclear, which is itself a reason not to collapse the two into a single number: a policy aimed at expanding networks addresses one plausible pathway, and a policy aimed at reducing the subjective distress of disconnection addresses a different one, and nobody yet knows which matters more for heart disease specifically.

A 2023 review in BMC Public Health surveying the state of loneliness and isolation research names inconsistent measurement as a persistent barrier to comparing findings across studies. Part of that inconsistency is exactly this: some studies operationalize disconnection structurally, some subjectively, and many papers report a headline figure without saying which.

The population implication

Because the two constructs are imperfectly correlated, the people captured by “one in four isolated” and the people captured by “one in three lonely” are not the same people. Some structurally isolated older adults — living alone, with a thin network — report low loneliness; a stable routine and low expectation of contact can coexist with structural thinness. Some densely networked older adults report high loneliness, the AARP survey’s neighbor-talkers among them. A program that expands contact frequency, such as befriending schemes long promoted by organizations like the Campaign to End Loneliness, targets the first group. A program addressing the quality or meaning of existing contact targets the second. Aiming interventions using only one statistic risks missing whichever group that statistic does not describe.

The American Journal of Geriatric Psychiatry’s clinical commentary on the National Academies report pushes toward routine assessment in health care settings, but assessment of what, precisely, is the open question. A checklist counting contacts and a scale asking about felt connection will flag different patients, and a clinic that adopts only one will systematically miss people who fail the other.

What would resolve this

The clearest gap in the current literature is a dataset that measures both constructs in the same older-adult sample with validated instruments for each, reports the overlap directly, and tracks whether interventions aimed at one construct move the other. Some studies gesture at this; few report the cross-tabulation plainly enough to say what fraction of isolated older adults are also lonely, and what fraction of lonely older adults are not isolated at all. Until that overlap is reported as a number rather than assumed, citing “a quarter are isolated” and “a third are lonely” in the same sentence will keep implying a single crisis where the evidence actually describes two related but separable ones.