Isolation and Loneliness Are Not the Same Number

The National Academies report that roughly a quarter of U.S. adults aged 65 and older are socially isolated. Cigna reports that 61% of U.S. adults across all ages say they sometimes or always feel lonely. Harvard’s Making Caring Common project puts serious loneliness among American adults at 36%, but among young adults aged 18–25 at 61%. These numbers get cited together constantly, often in the same paragraph, as if they were three measurements of the same thing. They are not. Two are about loneliness. One is about isolation. The gap between them is not a rounding error; it is a difference in what is actually being counted.

Two different objects

Isolation is structural: the number of people in a person’s network, the frequency of contact, whether someone lives alone. It can, in principle, be observed by a third party — a researcher could count the visits, the phone calls, the household size. Loneliness is a subjective state: a felt gap between the connection a person wants and the connection a person has. Two people with identical contact logs can report opposite experiences. The 2020 National Academies consensus report treats this distinction as foundational, and its recommendation — that health systems routinely assess both isolation and loneliness — only makes sense if the two are measured separately, because a screening tool built for one will miss cases defined by the other.

Julianne Holt-Lunstad’s 2015 meta-analysis in Perspectives on Psychological Science is the clearest demonstration of why this matters empirically, not just conceptually. Analyzing mortality outcomes across many studies, it reports an odds ratio of 1.29 for social isolation, 1.26 for loneliness, and 1.32 for living alone. The three are close in magnitude but were entered as distinct variables, drawn from distinct instruments, and the paper is explicit that they do not simply substitute for one another. A person can score high on isolation and low on loneliness — content living alone, in infrequent contact, by choice — or the reverse: embedded in a large family network and still lonely within it. The mortality risk attaches to both, independently, which is a stronger empirical claim than saying “loneliness is bad for you.” It is a claim that two separable conditions are each, on their own, associated with worse survival odds.

What the surveys are actually asking

The instrument mismatch compounds the conceptual one. AARP’s 2018 survey of adults 45 and older used the 20-item UCLA Loneliness Scale, a validated academic instrument, and found one in three respondents scoring as lonely. That is a loneliness figure, arrived at through an indirect measure — the UCLA scale asks about feelings of being understood, feeling part of a group, feeling in tune with people around oneself — without directly asking “are you lonely.” Cigna’s 2020 workplace report, by contrast, asks adults directly whether they “sometimes or always” feel lonely, and gets 61%, a far higher figure. Neither number is wrong. They are answers to different questions, administered to different populations, and reporting them side by side without that caveat manufactures a discrepancy that looks like disagreement about reality when it is actually disagreement about method.

None of the widely cited recent surveys — AARP, Cigna, Harvard’s Making Caring Common, the AEI friendship survey — report a structural isolation measure comparable to the National Academies’ 25% figure for isolated older adults. The AEI survey on American friendship comes closest to a structural measure: it tracks the number of close friends people report, finding that 12% of Americans now report having no close friends at all, up from 3% in 1990, and that the share of men reporting six or more close friends fell from 55% to 27% over the same period. That is a network-size statistic, not a feelings-about-connection statistic, and it happens to show a much longer and steeper trend line than any single-year loneliness survey can, because it is describing an erosion of a structure, not a fluctuation in a mood.

Why the conflation persists

Part of the problem is convenience. A single well-worded question about loneliness is cheap to administer in an omnibus survey; a proper structural map of someone’s network — size, frequency of contact, diversity, reciprocity — is not. So loneliness gets measured far more often, and then gets used, loosely, to stand in for the whole phenomenon of social disconnection, isolation included.

Holt-Lunstad’s 2021 review in the American Journal of Lifestyle Medicine argues for treating social connection as a modifiable risk factor on par with diet and smoking — a framing that only works if clinicians know which deficit they are screening for. A screening question aimed at feelings will not catch the isolated person who denies loneliness, and a headcount of contacts will not catch the well-connected person who feels alone regardless. The two require different interventions: one plausibly addressed by expanding a network, the other not necessarily solved by adding people to it at all.

A stronger evidence base would report isolation and loneliness as two rows in the same table, from the same sample, using both a structural instrument and a validated subjective scale — rather than letting one stand in for the other because it happened to be the question that got asked.