Loneliness Statistics: Why the Numbers Disagree

Four honest numbers that appear to contradict each other

There is no single loneliness rate, and anyone quoting one without saying how it was measured is quoting a coincidence.

The UK Government's Community Life Survey asks one direct question — "How often do you feel lonely?" — and counts the people who answer "often or always". In 2023/24 that was 7% of adults in England, about 3.1 million people. The Meta-Gallup Global State of Social Connections asked a single question about how lonely people feel and counted "very or fairly lonely": 24% of people aged 15 and over, across 142 countries. The University of Michigan's National Poll on Healthy Aging asks how often people feel a lack of companionship and how often they feel isolated, counting "some of the time" or "often": in March 2024, 33% of US adults aged 50-80 were lonely on that measure and 29% felt isolated. And Cigna's 2018 US Loneliness Index used the full 20-item UCLA Loneliness Scale and classified everyone scoring 43 or above as lonely — on that method the national average score was 44, which put roughly half the sample over the line.

Seven per cent and roughly fifty per cent are not a factual dispute. They are a direct question with a strict "often or always" cut, versus a 20-item psychometric scale with a threshold the researchers chose. Both are defensible. Neither is the loneliness rate.

The instruments, and what each one actually asks

Four instruments do most of the work in this literature, and knowing which one you are looking at explains most of the apparent disagreement.

The UCLA Loneliness Scale (Version 3), published by Daniel Russell in the Journal of Personality Assessment in 1996, is the reference instrument in loneliness research. It has 20 statements, each rated 1 (never) to 4 (often), giving a total between 20 and 80. Russell reported internal consistency between 0.89 and 0.94 and a one-year test-retest correlation of 0.73. It has no diagnostic cut-off — researchers pick a threshold for their study, which is exactly what Cigna did at 43.

The three-item short form (Hughes, Waite, Hawkley & Cacioppo, Research on Aging, 2004) was built for telephone and large population surveys. It asks how often you feel you lack companionship, how often you feel left out, and how often you feel isolated from others, each scored 1-3, giving a total of 3 to 9. Many surveys treat 6 or more as "lonely" — a convention, not a diagnosis.

The De Jong Gierveld scale (six-item version: De Jong Gierveld & Van Tilburg, Research on Aging, 2006) is the main European instrument and does something the UCLA scales do not: it separates emotional loneliness (missing one close, intimate relationship) from social loneliness (missing a wider network). Three items measure each. That distinction matters, because someone with a devoted spouse and no friends and someone with many friends and no confidant can produce the same overall score for opposite reasons.

The ONS/GSS harmonised measure, used across UK government surveys, is four questions: one direct ("How often do you feel lonely?") plus the three indirect UCLA-style items on companionship, feeling left out, and feeling isolated. Published UK headline figures — including the 7% above — come from the direct question at the "often or always" cut.

Loneliness, social isolation and living alone are three different things

The research literature keeps these three separate, and so should any number you rely on.

Living alone is a household arrangement. It is counted from census and household surveys, it is a fact about a dwelling, and on its own it is a choice millions of people make happily.

Social isolation is objective: how many contacts, how much interaction, measured by counting them. A person can be socially isolated and not feel lonely at all.

Loneliness is subjective: the distress of a gap between the connection you have and the connection you want. It is measured by asking, which is why the instrument matters so much. People report loneliness in full houses and report none living alone.

The three overlap but they are not interchangeable, and the useful response differs for each. Isolation responds to more contact. Loneliness responds to more meaningful contact — more contact alone can miss it entirely. Living alone raises one specific, narrow, practical question that has nothing to do with either feeling: if something went wrong today, how long before anybody knew?

That last question is the only one a check-in app has any business answering.

What the mortality research actually found

The most-repeated claim in this field is that loneliness is as deadly as smoking 15 cigarettes a day. It is worth being precise about where that comes from and what it does and does not say.

The underlying meta-analysis is Holt-Lunstad and colleagues, published in Perspectives on Psychological Science in 2015. It pooled 70 studies and reported odds ratios for mortality of 1.26 for loneliness, 1.29 for social isolation and 1.32 for living alone — that is, average increases in the likelihood of dying during follow-up of 26%, 29% and 32% respectively. An earlier meta-analysis by the same lead author (PLoS Medicine, 2010) pooled 148 studies covering more than 300,000 people and found a 50% higher likelihood of survival among those with stronger social relationships.

The cigarette comparison is the US Surgeon General's, in the 2023 advisory Our Epidemic of Loneliness and Isolation. It is an effect-size analogy: the advisory is saying the size of the mortality association is in the same range as that of a well-known risk factor. It is not a biological claim, and nobody has shown that loneliness damages the body the way tobacco smoke does.

Two cautions that responsible coverage should carry and usually doesn't. First, these are associations. People who are lonely or isolated differ from people who are not in many ways — health, income, bereavement, disability — and reverse causation is entirely plausible, since poor health causes isolation as readily as isolation worsens health. Second, they are population averages across studies with different measures; they say nothing about the risk facing any individual person.

The World Health Organization's Commission on Social Connection put its own estimate on the scale of the problem in June 2025: loneliness and social isolation are associated with around 871,000 deaths a year worldwide, about 100 an hour. That is an attributable-burden estimate built on the same association literature, with the same caveats.

The loneliest age group is not the one people assume

Every instrument that reports an age breakdown finds the same shape, and it is the opposite of the stereotype: young adults report the most loneliness, and people over 65 often report the least.

On the UK Community Life Survey's direct question in 2023/24, 10% of 16-24-year-olds in England felt lonely often or always, against 4% of 65-74-year-olds and 6% of those 75 and over. Meta-Gallup found 27% of 19-29-year-olds worldwide very or fairly lonely, against 17% of people 65 and over. The WHO Commission reported the highest rate of all among 13-17-year-olds, at 21%.

This does not mean older people are fine. Two things are true at once. Self-reported loneliness peaks young; and the practical consequences of living without anyone nearby — falls, sudden illness, the discovery gap — fall hardest on the old. They are different problems and they need different answers. Confusing them produces the familiar mistake of selling companionship to people who wanted safety, and safety hardware to people who wanted company.

Figures we left out, and why

Several numbers that circulate widely in this subject are absent from this page on purpose.

We do not publish a single "X% of people are lonely" headline. There is no such quantity independent of an instrument, and manufacturing one is the specific error this page exists to correct.

We do not compare loneliness rates between countries in a ranked table. Cross-country loneliness league tables almost always mix instruments and thresholds; the ranking they produce is mostly a ranking of survey design.

We do not repeat percentage claims about how much any specific intervention reduces loneliness. Several such figures circulate attached to real organisations, and we could not trace them to a publication.

And we do not describe loneliness as a diagnosis. It is not in the diagnostic manuals, no score on any of these instruments is a clinical finding, and a questionnaire result is not a medical opinion.

What a daily check-in can and cannot do

A check-in app does not treat loneliness. It is not company, it is not care, and it is not therapy. If you are lonely, the things that help are the ones the literature has always pointed to — contact that means something, a reason to leave the house, professional help when the weight of it becomes too much. An app on a phone is not a substitute for any of those, and we will not pretend otherwise.

What a check-in does is answer one narrow question for people who live alone: if something goes wrong today, how long before anybody knows? You confirm you're okay once a day. If you don't, the people you chose are told, with your location — within hours, not days.

That is a safety mechanism, not companionship. It is worth having for the same reason a smoke alarm is worth having, and for no larger reason than that.

If you are in distress right now, please contact a crisis line rather than a questionnaire: in the US, call or text 988 (Suicide & Crisis Lifeline); in the UK and Ireland, call Samaritans free on 116 123; in Australia, Lifeline on 13 11 14. For other countries, findahelpline.com lists verified services.

I'm Alive is a daily check-in app. You confirm you're okay once a day; if you don't, the people you chose are told, with your location. The daily check-in is free (Try It, $0) and works on iPhone and Android anywhere in the world. Automatic alerts to your contacts start on Protect Me ($29.99/year), with location on the move on Protect Me On The Move ($39.99/year); there is a one-time Stay Connected lifetime upgrade at $4.99. Check-first phone fall detection is live on iPhone only (app v2.3.1) on Protect Me and above, stays off until you switch it on, and only runs while the app is open. It alerts your own trusted contacts — it never calls emergency services, and it is best-effort, not a medical device. The app interface is currently in English only.