Skip to main content
Making Friends

Is Loneliness Bad for Your Health? The Honest Answer

Is loneliness bad for your health? Yes, with caveats the headlines drop. What 148 studies found, what UK Biobank undid, and how to read the numbers.

12 min read
Making Friends on KF.Social

This guide is by KF.Social.

On this page
  1. The short answer
  2. If this is hitting you hard right now
  3. What the survival research actually found
  4. The counterweight almost nobody cites
  5. Why the comparison to smoking does not hold
  6. Heart disease, stroke and dementia
  7. How common this is in England
  8. What the evidence supports doing
  9. Who this will not help
  10. Key terms

You have probably seen the headline. Loneliness is as dangerous as smoking, it says, and a number follows that sounds too precise to argue with.

It does two things at once. It makes you take the question seriously, and it makes you feel worse about an evening you were already finding hard.

The research underneath it is real and it is very large. The headline version is not what that research says, and the gap between the two is wide enough to matter to you.

This guide covers what the survival studies actually found, the two enormous cohorts that complicate them, why the smoking comparison does not hold, what can honestly be said about heart disease and dementia, how common this is in England, and what any of it justifies doing.

The short answer

Yes, with a narrower meaning than the headline gives it, and the distinction it turns on is the whole thing.

Researchers measure two separate things here. Social isolation is a count of how connected you actually are: how many kinds of relationship you have, how often you see people, whether you live with anyone. Loneliness is how connected you feel, which is the gap between the relationships you have and the ones you want.

They are related and they are not the same, and the health evidence treats them very differently. Isolation keeps showing up as an independent risk in the largest studies. Loneliness measured as a feeling often does not survive once income, existing illness, health behaviour and depression are taken into account.

That does not make loneliness harmless. It is a real and serious distress, and it tends to travel alongside other disadvantages that carry much of the risk themselves. What it does mean is that nobody has shown that joining things adds years to your life.

Every mortality and disease finding below is observational. It describes people who already differed from each other in many ways. Nobody was assigned to be lonely, so none of it can tell you what would happen if your situation changed.

If this is hitting you hard right now

Reading about health risk while you already feel cut off is its own kind of unpleasant, so this comes before the rest of it rather than at the bottom.

If you are struggling, Samaritans are on 116 123 in the UK and Ireland, free, at any hour. The Shout text line is 85258. For urgent mental health support in England, call NHS 111 and choose option 2. For anxiety or low mood that has lasted weeks, you can refer yourself to NHS Talking Therapies without going through your GP, and your GP is a good route too.

Nothing on this page is treatment and no social activity is a substitute for one. The studies below describe populations. They do not describe you, and they cannot tell you what will happen to you.

What the survival research actually found

One paper sits underneath almost every claim in this area, and it is worth reading at its own size rather than through a headline.

The headline figure comes from 148 studies

Holt-Lunstad, Smith and Layton's 2010 meta-analysis pooled 148 studies covering 308,849 people, with a mean follow-up of 7.5 years. People with stronger social relationships had a 50 per cent higher likelihood of surviving the follow-up period, an odds ratio of 1.50.

Two details are usually left out. The mean age at baseline was 63.9, so this is mostly a finding about older adults. And the studies pooled were observational with inconsistent sets of statistical controls, which the authors say plainly in their own limitations.

The strongest measure is being embedded in several kinds of relationship

Inside that paper the effect is not uniform. Complex measures of social integration, meaning the number of different kinds of relationship a person has, gave the largest figure at an odds ratio of 1.91. Simply living alone or not gave the smallest, 1.19, with a confidence interval that crosses 1.0 and so is not statistically significant.

That ordering is the most useful thing in the paper. Having a household is the weakest version of connection in this data. Having several different sorts of tie is the strongest.

The same team put loneliness and isolation at a lower number

Five years later, Holt-Lunstad and colleagues pooled the loneliness literature directly. After controlling for confounds they reported social isolation at an odds ratio of 1.29, loneliness at 1.26 and living alone at 1.32 for mortality.

The finding in that paper that gets dropped most often is this: the social deficits predicted death more strongly in samples with a mean age under 65, not less. This is not only a question for later life.

The counterweight almost nobody cites

If the story ended there it would be tidy. Two of the largest cohort studies ever run then asked what happens when you adjust properly, and the answer changed.

In UK Biobank the loneliness link disappeared on full adjustment

Hakulinen and colleagues followed 479,054 UK Biobank participants for 7.1 years. Adjusted only for demographics, loneliness carried a hazard ratio of 1.49 for heart attack and 1.36 for stroke. Fully adjusted, those hazards fell to 1.06 and 1.04, and neither was statistically significant.

The isolation figures attenuated by 83 to 84 per cent on the same adjustment. The authors' own conclusion is that most of the risk was explained by conventional risk factors.

A companion analysis, Elovainio and colleagues on 466,901 participants, found the same shape for death from any cause. Loneliness went from a hazard ratio of 1.38 to 0.99 once socioeconomic conditions, behaviour, depression and biology were included. Isolation went from 1.73 to 1.26 and survived, weakened.

In the English Longitudinal Study of Ageing loneliness did not predict death

Steptoe and colleagues followed 6,500 English adults aged 52 and over for an average of 7.25 years. Social isolation predicted mortality with a hazard ratio of 1.26. Loneliness gave 0.92, which is not significant, and adding it to the model left the isolation estimate unchanged.

Read those three results together and the defensible sentence is short. Isolation looks like a real and partly independent health risk. Loneliness is a real and serious distress that usually travels with other disadvantages, and those disadvantages carry most of the measured risk.

Why the comparison to smoking does not hold

The single most repeated line in this field compares loneliness to a daily cigarette count. It is worth taking apart, because understanding why it is wrong tells you how to read everything else.

The comparison is not in the 2010 meta-analysis. That paper compares the odds of decreased mortality across several conditions, drawing its comparators from other meta-analyses of smoking cessation, cardiac rehabilitation, flu vaccination, physical activity, obesity and hypertension treatment. Its own conclusion says the effect is comparable with quitting smoking and exceeds several well known risk factors. A cigarette count is a later restatement by other people.

Three things go wrong in the restatement. It converts a mortality odds ratio into a general health equivalence, which it never was. It attaches a figure about social disconnection, which is a structural measure, to the feeling of loneliness, which is a different variable. And the version it came from carried a qualifier that almost always gets dropped on the way to a headline.

A number that has been stretched three times is not evidence. If you want the honest size of the thing, use the odds ratios: 1.26 to 1.50 depending on which measure you take, from observational studies, with the largest cohorts suggesting the loneliness end of that range is mostly explained by other factors.

Heart disease, stroke and dementia

These are the three outcomes people search for, and each one needs its own caveat rather than a shared one.

The heart figures attenuate by more than four fifths

Valtorta and colleagues pooled 16 longitudinal datasets covering 4,628 coronary heart disease events and 3,002 strokes. Poor social relationships were associated with a 29 per cent higher risk of incident heart disease and a 32 per cent higher risk of stroke. Both confidence intervals come close to 1.0.

Those are the numbers that circulate in government advisories. They are also the ones UK Biobank then attenuated by 83 to 84 per cent on full adjustment. Both facts belong in the same paragraph, every time.

The dementia evidence disagrees with itself

A 2024 meta-analysis of 608,561 people put loneliness and all-cause dementia at a hazard ratio of 1.306, holding after controls for depression and social isolation. A 2018 meta-analysis of 2,370,452 people found loneliness non-significant at a relative risk of 1.38, while a poor social network came in at 1.59 and good social engagement looked protective in longer studies.

Use both or neither. The line the two agree on is that social disengagement, which is something you can observe from outside, is more consistently associated with dementia risk than the feeling of loneliness is. Reverse causation is also live here, because early dementia causes withdrawal.

One more caution. The mechanism studies people quote, the ones about inflammation and gene expression, are small. The best known compared gene expression in 14 people. That is a reason to keep looking, not a population finding.

How common this is in England

The prevalence question has one answer worth quoting and several that are not worth quoting, so here is the one.

The Community Life Survey 2024/25, with more than 175,000 responses in England, found 6.6 per cent of adults felt lonely often or always. That is down slightly from 7.1 per cent the year before and above the 4.9 to 5.6 per cent recorded between 2013/14 and 2017/18.

The age pattern is the reverse of the assumption almost everybody carries. It runs 9 per cent among 16 to 24 year olds, 8 per cent at 25 to 34, 7 per cent at 35 to 49, 5 per cent at 50 to 64, and 4 per cent at 65 to 74. The youngest group reports the most, not the least.

Two other splits are larger than the age one. Disabled adults reported 14 per cent against 3 per cent for non-disabled adults, and the most deprived tenth of areas reported 11 per cent against 4 per cent in the least deprived. That shape is the same one the cohort studies keep running into: this clusters with other disadvantage, which is exactly why the adjusted hazard ratios fall.

What the evidence supports doing

Here is where most articles on this subject promise that getting out more will fix your health. The evidence does not support that promise, and the honest version is more useful anyway.

Turning up is not the active ingredient

In Masi and colleagues' meta-analysis of loneliness interventions, the randomised trials are the ones that count. Within them, approaches that simply increased opportunities to socialise produced no significant effect. Approaches that changed how people interpret social situations produced the largest one.

That opportunities cell rests on only two studies, so it is weak evidence rather than proof of nothing. It is still enough to rule out the confident claim that attending events lifts loneliness.

Building a group membership is a different thing from attending

The strongest trial in the area is GROUPS 4 HEALTH, a randomised controlled trial of 120 adults in psychological distress. Loneliness fell further in the programme arm than under treatment as usual at two months. What the programme teaches is how to build and keep group memberships, which is not the same as a diary of events. One trial, one research group, two months of follow-up.

What this justifies, stated plainly

It justifies treating isolation as worth doing something about, for reasons of ordinary quality of life that need no mortality statistic at all. It does not justify anyone telling you that a weekly activity will extend your life, and it does not justify the fear the headline version is designed to produce.

If you want the better supported mechanism rather than the health claim, it is about repetition. The same people, on a rhythm, in time that is leisure rather than task. That is what the friendship-formation research keeps pointing at, and it carries no promise about your arteries.

Who this will not help

If you came here worried about a specific symptom, this page is the wrong tool. Population associations cannot be applied to one person, and a doctor who can see you is worth more than any meta-analysis.

If what you are carrying is persistent low mood or anxiety rather than a quiet diary, the same applies with more force. Your GP is the route, and in England you can refer yourself to NHS Talking Therapies without one. Samaritans are on 116 123 at any hour and Shout is on 85258. Nothing here is treatment and no social activity is a substitute for one.

And if reading risk figures makes you feel worse rather than mobilised, you are allowed to stop reading them. The evidence does not reward worry, and worry is not a plan.

KF.Social is a make-friends app for adults: see people near you, join clubs around what you like, and pick the Plans in your area you would actually turn up to. Join something that repeats.

Key terms

Social isolation: a measure of how connected you actually are, counting kinds of relationship, frequency of contact and household composition. Observable from outside.

Loneliness: a measure of how connected you feel, meaning the gap between the relationships you have and the ones you want. You can feel it in a full room.

Odds ratio and hazard ratio: ways of expressing how much more likely an outcome was in one group than another. Neither says anything about cause on its own.

Observational study: research that measures people as they are rather than assigning them to conditions. It can show that two things travel together and cannot show that one produces the other.

Adjustment: the statistical step of accounting for other differences between groups, such as income, illness or smoking. When a result shrinks on adjustment, those other differences were carrying it.

Plan: on KF.Social, a specific activity in your area at a specific time that you can join. Plans come in several types and the app shows the details before you decide.

Club: on KF.Social, a group organised around an interest, with its own posts and group chat. A standing membership rather than a single occasion.

Area: the part of your city a Plan or a club is local to. What you are shown is local to where you are.

Related Questions

Is loneliness bad for your health?
It is associated with worse health outcomes, though the strongest and most consistent evidence is for social isolation rather than the feeling itself. All of it is observational, and in the largest cohorts the loneliness association shrinks sharply once income, illness, behaviour and depression are accounted for.
Does being lonely shorten your life?
No study has shown that. Pooled observational research puts stronger social relationships at a 50 per cent higher likelihood of surviving a follow-up period, but two very large cohorts found the loneliness link to death disappeared on full adjustment. Association is not the same as years gained or lost.
Is loneliness worse for you than smoking?
No, and the comparison people quote is a restatement rather than a measurement. The 2010 meta-analysis compared mortality odds across several conditions and said the effect was comparable with quitting smoking. Turning that into a daily cigarette count stretches a mortality figure into a general health claim.
What is the difference between loneliness and social isolation?
Isolation is how connected you actually are, counted from outside. Loneliness is how connected you feel. In the English Longitudinal Study of Ageing, isolation predicted mortality and loneliness did not, and adding loneliness to the model changed nothing about the isolation estimate.
Does loneliness cause heart disease or stroke?
No one has shown that. Pooled longitudinal data put poor social relationships at 29 per cent higher heart disease risk and 32 per cent higher stroke risk. In UK Biobank, covering 479,054 people, those hazards attenuated by 83 to 84 per cent once conventional risk factors were included.
How many people in the UK are lonely?
In England, 6.6 per cent of adults said they felt lonely often or always in the Community Life Survey 2024/25, from more than 175,000 responses. The rate is highest among 16 to 24 year olds at 9 per cent and lowest among 65 to 74 year olds at 4 per cent.
Will joining a club improve my health?
The evidence does not support that promise. In randomised loneliness trials, simply increasing opportunities to socialise showed no significant effect, while a programme teaching people to build and keep group memberships did reduce loneliness. Neither result says anything about your physical health.
Send this to someone who needs it

Make friends and make plans with KF.Social

Small groups near you doing something you enjoy, and your first hello is written for you. Make friends as an adult, without it being awkward.

🤝 More in Making Friends →
KF.Social
Make friends and make plans.
Get the app