Isolation and dementia travel together, but the arrow may point backward

Picture a man in his late seventies who quits his Thursday card game. The others talk too fast, he says, and the room is loud. About a year later he is diagnosed with dementia, and his family starts asking a painful question: did dropping out of the game help bring this on?

Researchers ask the same question about hundreds of thousands of people and hit the same wall. From the outside, pulling away from people and the first stage of dementia look alike, and either can come first.

That wall sits under nearly every headline about loneliness and the brain. The link has been found again and again. What remains unsettled is which way it runs, and whether adding conversation or company to someone’s week changes how their thinking holds up.

Here is how scientists try to sort that out, what the few randomized trials found, and what to do if someone you love is drifting away from people.

The link shows up almost everywhere researchers look

The numbers are consistent. A 2015 meta-analysis in Ageing Research Reviews, led by Jisca S. Kuiper at the University of Groningen in the Netherlands, pooled 19 long-term cohort studies. People with low social participation had about 1.4 times the risk of a later dementia diagnosis. People with infrequent contact, and people who reported more loneliness, each had close to 1.6 times the risk. Network size gave inconsistent results, and being dissatisfied with one’s network showed no significant link at all.

Larger pools of data agree. A 2024 meta-analysis in Nature Mental Health, led by Martina Luchetti at Florida State University, combined 21 samples covering more than 600,000 people. Loneliness was associated with about 31 percent higher risk of dementia from any cause. The association held after the team accounted for depression and social isolation, although results varied a great deal from one study to the next.

The 2024 Lancet Commission on dementia prevention, an international expert panel that periodically reviews the field, built this evidence into its estimates. It classes social isolation as a late-life risk factor, assigns it a relative risk of about 1.6 drawn from observational studies, and attributes roughly 5 percent of dementia worldwide to it. That 5 percent is a population model, not anyone’s personal odds, and it assumes the link is causal. The Commission’s full roster of 14 factors, and how to read its percentages, appears in our piece on the dementia risk factors people seldom guess.

In the United States, a 2020 consensus report from the National Academies of Sciences, Engineering, and Medicine estimated that about 24 percent of adults 65 and older living in the community are socially isolated. The report ties isolation to a dementia risk about half again as high, and it frames that figure as an association rather than a demonstrated effect.

One distinction matters below. Isolation is a count of contacts; loneliness is the feeling that your connections fall short. A person can have either without the other, as our article on loneliness as a health risk explains, and studies of the two sometimes disagree.

The direction problem

The brain changes behind dementia typically begin years before anyone puts a name to them, and one of their early effects can be social withdrawal.

Conversations get harder to follow. Names slip. A noisy restaurant becomes exhausting. Skipping the card game can be a sensible response to a change the person has not yet named. If a study measures social contact during those years and then records a diagnosis, the isolation will look like a cause even when it was an early symptom. Researchers call this reverse causation.

People living through it describe the same thing. In a 2024 qualitative study in The Gerontologist, geriatrics researchers at the University of California, San Francisco analyzed interviews with 24 people living with dementia and 48 current or bereaved care partners. Social networks fell away as the disease progressed. Friends grew uncomfortable, caregiving swallowed the time, and the illness itself made socializing harder.

The Lancet Commission names the problem in its own review of the evidence. In one of the systematic reviews it drew on, seven of the eight studies followed people for less than four years. Over a window that short, the Commission judged, the disease causing the isolation is a likely explanation.

Medical records show the timing too. In a 2025 study in Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, researchers at the University of Sheffield and the University of Oxford used language-processing software to read the records of patients with dementia. Patients whose notes mentioned social isolation lost about 0.21 points a year faster on the Montreal Cognitive Assessment, a 30-point screening test, but only in the six months before diagnosis. Before that, they tracked everyone else. The authors suggest isolation may contribute to progression. The timing fits an equally plain reading: isolation surfacing as part of an illness already underway.

Three ways researchers try to break the loop

Measure decades ahead

If isolation were only an early symptom, the link should fade when contact is measured long before diagnosis. The Whitehall II study put that to the test. Researchers at University College London followed 10,228 British civil servants who were 35 to 55 when the study began in the mid-1980s. Participants reported their social contact six times, and dementia was tracked through linked health records until 2017, a span of 28 years.

In their 2019 paper in PLOS Medicine, more frequent contact at age 60 went with lower dementia risk: about 12 percent lower for each step up the contact scale (one standard deviation). Estimates at ages 50 and 70 were similar in size but not statistically significant. Contact with friends drove the association. Contact with relatives did not.

A lead time that long makes pure reverse causation less likely. The authors still named the remaining loophole: keeping up a busy social life may itself be a sign of a brain with more reserve. The thinking difference they found in midlife was also small, 0.07 standard deviations between the most and least socially active thirds of the group.

Adjust for depression and health

Depression, anxiety, chronic illness and disability all make people withdraw, and all are tied to dementia. If they explain the link, accounting for them should shrink it. Two University of Memphis researchers tested this in 2,977 older Americans from the National Health and Aging Trends Study, followed for an average of 3.6 years. Writing in Innovation in Aging in 2024, the Memphis team reported: “This association between social isolation and incident dementia became nonsignificant after further adjustment for depression, anxiety, and health indicators.”

Loneliness has held up better in some data. In the Luchetti meta-analysis, the link persisted after adjusting for depression. A 2025 study in Psychological Medicine pooled harmonized cognitive testing from seven countries, including the United States, England, India and China, with more than 20,000 participants. Depression weakened the association between loneliness and test scores without erasing it. But the association was small: a correlation of 0.10, which statistically accounts for about 1 percent of the differences in scores between people.

These studies measured different exposures, objective isolation in one and the feeling of loneliness in the others, so they do not cancel out. Neither design can prove cause.

Track change over time

A single survey answer cannot separate a lifelong loner from someone whose spouse died last spring. A 2025 study in the International Journal of Geriatric Psychiatry, from Monash University in Australia, asked more than 12,000 adults 70 and older about loneliness twice, about three years apart, then followed them for a median of eight years.

Loneliness reported both times went with higher dementia risk in women, a hazard ratio of 2.14, meaning roughly double the rate over the follow-up. New loneliness in men carried a hazard ratio of 1.52. Loneliness that came and went showed no increased risk. Social isolation and poor support were linked to faster cognitive decline but not to more dementia diagnoses.

Duration seems to matter. The finding is still observational.

What the randomized trials found

The cleanest way to settle direction is to assign people at random to more social contact and watch their thinking. The Lancet Commission’s 2024 judgment was blunt: intervention studies so far are too small, and their follow-up too short, to tell whether they change how many people develop dementia. Three trials come up most often.

I-CONECT: conversation by video

The Internet-Based Conversational Engagement Clinical Trial, led by Hiroko H. Dodge of Massachusetts General Hospital and Oregon Health & Science University, recruited in Portland, Oregon, and Detroit. It enrolled 186 socially isolated adults aged 75 and older who had either normal cognition or mild cognitive impairment, meaning memory or thinking problems that show up on testing but do not yet disrupt daily life. People with dementia or moderate-to-severe depression were excluded.

The intervention group held 30-minute video conversations with trained staff four times a week for six months, then twice a week for six more. Partners rotated, and topics came from more than 150 prepared themes, from historical events to travel. Both groups received a weekly phone check-in of about 10 minutes.

Then the pandemic hit. Testing moved to the phone, and the research team and its safety monitoring board limited the main analysis to people who finished their six-month in-person testing before March 18, 2020. That left 56 people. Among the 31 with mild cognitive impairment, 13 in the conversation group and 18 controls, Montreal Cognitive Assessment scores were 1.75 points higher with conversation. Among the 25 with normal cognition, there was no difference.

The abstract of the 2024 topline paper in The Gerontologist presents the 1.75-point gain as a result for the experimental group, without naming the subgroup. The 31-person detail sits in the results section. Two secondary findings, a verbal fluency gain in people with normal cognition and a story-recall gain in those with mild impairment, cleared the usual statistical bar. The I-CONECT authors were candid about that: “For the secondary outcomes, we used a conventional Type I error rate without adjusting for multiple comparisons.” Against the stricter threshold printed in their own table, neither result counts.

One detail is worth keeping. Social satisfaction rose about equally in the phone-call group and the video group, yet test scores moved only in the video group, and only in the small mild-impairment subgroup. The authors read this as a hint that mentally demanding conversation, rather than emotional closeness, may be the active ingredient. That is the authors’ reading, and the cognitive result behind it comes from 31 people. The trial also reached only 23 percent of its recruitment goal for Black participants, which narrows whom the results describe. The National Institute on Aging, in a March 2024 news story, called the results very preliminary and said they need far more testing in other groups and settings.

Experience Corps: volunteering in schools

The Baltimore Experience Corps Trial placed older adults in elementary schools to help children with reading, library work and classroom behavior. In a 2008 pilot with 149 people serving about 15 hours a week for a school year, overall gains in memory and executive function, the mental skills behind planning and self-control, fell short of significance, though people who started with weak executive function improved.

The full trial produced a striking brain-scan headline. In an MRI substudy of 111 participants, published in 2015 in Alzheimer’s & Dementia, volunteers’ cortical and hippocampal volumes held steady or grew over two years while controls’ shrank. The difference reached statistical significance in men only. Women in the program showed small gains against declines of about 1 percent in control women, a gap that fell short of significance. Brain volume is also a measurement on a scan, not something a person notices in daily life.

On thinking itself, a 2021 analysis of 336 trial participants in the Journals of Gerontology reported that the standard intention-to-treat comparison, which analyzes everyone in the group they were assigned to, showed no significant cognitive improvement. A benefit appeared only on an unconventional measure, the consistency of reaction times, among the most compliant volunteers.

ACHIEVE: an indirect route

In ACHIEVE, a hearing trial, older adults whose hearing loss was treated kept roughly one additional person in their social network over three years, compared with a health-education group, a difference the investigators said may or may not matter clinically. Its cognitive results get a full accounting in our companion article on hearing and the aging brain.

None of these trials showed fewer dementia diagnoses.

Where each claim stands

Graded claim by claim, using the library’s evidence verdicts:

• Isolation and loneliness travel with later dementia. Positive evidence, contested interpretation. The association is consistent across countries and designs. Whether it is cause, symptom or a shared third factor is unresolved.
• Adding conversation or volunteering improves thinking. Insufficient: a few small trials with mixed results, the positive findings concentrated in subgroups and secondary measures.
• Staying social prevents dementia. Insufficient: untested at the size and length a dementia outcome needs.
• A guideline recommends socializing to prevent cognitive decline. We found no US clinical guideline that names a social intervention for that purpose. The National Academies recommendation, described below, is about finding and addressing isolation itself.

None of this means connection is beside the point. The same National Academies report judged more than four decades of research strong in linking a lack of social connection to earlier death from all causes. That is also an association, but a far better established one than the brain-specific case.

Five questions for the next loneliness-and-brain headline

These five questions sort most stories in a minute.

1. How long before diagnosis was social life measured? If the gap is under about four years, the withdrawal could be the disease announcing itself.
2. Was dementia diagnosed, or estimated? A 2020 Health and Retirement Study analysis of 12,030 people, by Angelina R. Sutin and colleagues, found loneliness linked to 40 percent higher dementia risk. Dementia there was defined by a low score on a telephone cognitive screen, not a clinical diagnosis.
3. Was depression accounted for? In at least one US cohort, the isolation link lost statistical significance once it was.
4. Is it isolation or loneliness? They are different exposures and can give different answers.
5. If it is a trial, what happened to everyone enrolled? Look past subgroup results and secondary outcomes to the main comparison, and check whether the control group also got contact, as I-CONECT’s did.

When withdrawal deserves a clinician’s look

A new retreat from people in an older adult is information. It can reflect depression, a hearing problem, a mobility limit or early cognitive change, and these overlap. The National Academies report recommends that health care providers periodically check older patients for isolation and loneliness with validated questionnaires, then look for underlying causes, naming hearing loss and mobility limits as examples.

Signs that make the conversation worth having soon:

• Someone drops activities they kept up for years and gives reasons that do not quite add up.
• Group conversation has become hard to follow, or the television keeps getting louder.
• Low mood, poor sleep or loss of interest has lasted for weeks.
• Familiar tasks, such as bills, recipes or a well-known route, have started going wrong.

Whether a withdrawal reflects depression, hearing loss or a change in memory is a question for a primary care clinician, who can look at all three in one visit. Dated examples help more than impressions: she stopped going to choir in March tells a clinician more than she seems off lately. How clinicians judge memory change against normal aging is a separate subject in this collection.

One caution for families: the evidence does not support telling someone with early dementia that withdrawing caused their illness. The arrow may well run the other way.

If isolation comes with hopelessness or thoughts of suicide, call or text 988, the 988 Suicide & Crisis Lifeline. It is free and confidential, answers 24 hours a day, and is for emotional distress of any kind. In an emergency, call 911.

The most useful takeaway

Treat a new pulling-away as a symptom to investigate, not a moral failing and not a proven cause of dementia. The research says isolation and dementia keep company, that early illness drives some of that, and that the trials of adding social contact are too small to promise a brain benefit. Connection remains worth having for reasons that do not depend on that promise. When someone’s world starts to shrink, the first useful question is what changed: their hearing, their mood, their mobility or their memory.

Sources

1. University College London (UCL Discovery), accepted manuscript of The Lancet, 2024;404(10452):572-628. DOI: 10.1016/S0140-6736(24)01296-0. “Livingston G, Huntley J, Liu KY, et al. "Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission" (accepted version, text).” discovery.ucl.ac.uk
2. University College London (UCL Discovery), accepted manuscript table file. “Livingston G, Huntley J, Liu KY, et al. "Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission," Table 1 (relative risks and population attributable fractions).” discovery.ucl.ac.uk
3. National Academies of Sciences, Engineering, and Medicine. “Social Isolation and Loneliness in Older Adults: Opportunities for the Health Care System (2020), Summary.” nationalacademies.org
4. Ageing Research Reviews. “Kuiper JS, Zuidersma M, Oude Voshaar RC, et al. "Social relationships and risk of dementia: A systematic review and meta-analysis of longitudinal cohort studies." Ageing Res Rev, 2015. PubMed. DOI: 10.1016/j.arr.2015.04.006.” pubmed.ncbi.nlm.nih.gov
5. Nature Mental Health. “Luchetti M, Aschwanden D, Sesker AA, et al. "A Meta-analysis of Loneliness and Risk of Dementia using Longitudinal Data from >600,000 Individuals." Nat Ment Health, 2024. PubMed. DOI: 10.1038/s44220-024-00328-9.” pubmed.ncbi.nlm.nih.gov
6. The Journals of Gerontology, Series B. “Sutin AR, Stephan Y, Luchetti M, Terracciano A. "Loneliness and Risk of Dementia." J Gerontol B Psychol Sci Soc Sci, 2020. PubMed. DOI: 10.1093/geronb/gby112.” pubmed.ncbi.nlm.nih.gov
7. Psychological Medicine. “Lee JH, Sutin AR, Hajek A, et al. "Loneliness and cognition in older adults: A meta-analysis of harmonized studies from the United States, England, India, China, South Africa, Mexico, and Chile." Psychol Med, 2025. PubMed. DOI: 10.1017/S003329172500011X.” pubmed.ncbi.nlm.nih.gov
8. PLOS Medicine. “Sommerlad A, Sabia S, Singh-Manoux A, Lewis G, Livingston G. "Association of social contact with dementia and cognition: 28-year follow-up of the Whitehall II cohort study." PLoS Med, 2019. PubMed. DOI: 10.1371/journal.pmed.1002862.” pubmed.ncbi.nlm.nih.gov
9. Innovation in Aging. “Yang Y, Jiang Y. "The Association Between Social Isolation and Incident Dementia Among Older Adults: Evidence From National Health and Aging Trend Study." Innov Aging, 2024. PubMed. DOI: 10.1093/geroni/igae093.” pubmed.ncbi.nlm.nih.gov
10. International Journal of Geriatric Psychiatry. “Htun HL, Teshale AB, Sun H, et al. "Changes in Loneliness, Social Isolation, and Social Support: A Gender-Disaggregated Analysis of Their Associations With Dementia and Cognitive Decline in Older Adults." Int J Geriatr Psychiatry, 2025. PubMed. DOI: 10.1002/gps.70065.” pubmed.ncbi.nlm.nih.gov
11. Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring. “Myers JAC, Stafford T, Koychev I, et al. "Loneliness, social isolation, and effects on cognitive decline in patients with dementia: A retrospective cohort study using natural language processing." Alzheimers Dement (Amst), 2025. PubMed. DOI: 10.1002/dad2.70149.” pubmed.ncbi.nlm.nih.gov
12. The Gerontologist. “Kotwal AA, Allison TA, Halim M, et al. "“Relationships, Very Quickly, Turn to Nothing”: Loneliness, Social Isolation, and Adaptation to Changing Social Lives Among Persons Living With Dementia and Care Partners." Gerontologist, 2024. PubMed. DOI: 10.1093/geront/gnae014.” pubmed.ncbi.nlm.nih.gov
13. The Gerontologist. “Dodge HH, Yu K, Wu CY, et al. "Internet-Based Conversational Engagement Randomized Controlled Clinical Trial (I-CONECT) Among Socially Isolated Adults 75+ Years Old With Normal Cognition or Mild Cognitive Impairment: Topline Results." Gerontologist, 2024. PubMed. DOI: 10.1093/geront/gnad147.” pubmed.ncbi.nlm.nih.gov
14. National Institute on Aging, National Institutes of Health. “Online conversations show potential cognitive benefit for socially isolated older adults (March 7, 2024).” nia.nih.gov
15. The Gerontologist. “Carlson MC, Saczynski JS, Rebok GW, et al. "Exploring the effects of an “everyday” activity program on executive function and memory in older adults: Experience Corps." Gerontologist, 2008. PubMed. DOI: 10.1093/geront/48.6.793.” pubmed.ncbi.nlm.nih.gov
16. Alzheimer's & Dementia. “Carlson MC, Kuo JH, Chuang YF, et al. "Impact of the Baltimore Experience Corps Trial on cortical and hippocampal volumes." Alzheimers Dement, 2015. PubMed. DOI: 10.1016/j.jalz.2014.12.005.” pubmed.ncbi.nlm.nih.gov
17. The Journals of Gerontology, Series B. “Brydges CR, Carlson MC, Andrews RM, Rebok GW, Bielak AAM. "Using Cognitive Intraindividual Variability to Measure Intervention Effectiveness: Results from the Baltimore Experience Corps Trial." J Gerontol B Psychol Sci Soc Sci, 2021. PubMed. DOI: 10.1093/geronb/gbaa009.” pubmed.ncbi.nlm.nih.gov
18. JAMA Internal Medicine. “Reed NS, Chen J, Huang AR, et al. "Hearing Intervention, Social Isolation, and Loneliness: A Secondary Analysis of the ACHIEVE Randomized Clinical Trial." JAMA Intern Med, 2025. PubMed. DOI: 10.1001/jamainternmed.2025.1140.” pubmed.ncbi.nlm.nih.gov
19. 988 Suicide & Crisis Lifeline. “988 Suicide & Crisis Lifeline.” 988lifeline.org