Group Exercise in Later Life: Loneliness Effect Unproven in Review of 38 Trials
Published 2026-11-15
Trials have not yet shown that exercising together eases loneliness in later life, and they have not shown that it fails to: a 2018 review of 38 randomised trials of physical activity for older people found a small benefit for social functioning, strongest when the exercise was done in a group, but too few of the trials measured loneliness for their results to be pooled, and the review found no evidence of an effect on it[1]. Newer, smaller trials of group tai chi, dance and walking report lower loneliness scores, and they need testing at a larger scale.
Walking groups, tai chi, keep-fit and dance classes, bowls and seated exercise are among the most common things older people do together, and many are run by older people themselves. People come to move, keep well and enjoy themselves, and the company comes with it. That makes them hard to study: people who join may already be fitter, better off and better connected than those who do not. This study sets out what surveys, trials and long-running cohorts show about group activity and older people’s social lives. It reports population evidence only and gives no exercise advice to individuals.
Activity falls sharply at the oldest ages, though more older people in England are active than a decade ago
Worldwide, a 2024 analysis of 507 population surveys across 163 countries, funded in part by the World Health Organization, estimated that 31.3% of adults did not meet the guideline of 150 minutes of moderate activity a week (or its equivalent) in 2022, up from 23.4% in 2000[3]. These are modelled estimates for adults aged 18 and over; insufficient activity “was highest in the oldest age groups in all regions”[3].
England, which has a consistent yearly series, serves here as one worked example, not as a guide to other countries. Its Active Lives Adult Survey, which asks adults aged 16 and over about activity from walking for leisure to fitness classes, found in November 2024–25 that “activity levels generally decrease with age, with the sharpest decrease coming at age 75+”. Older adults also drove recent growth: 64.3% of people aged 55 to 74 were active for 150 minutes or more a week, against 57.0% in November 2015–16, and 44.3% of those aged 75 and over, against 33.4%[2]. The survey measured both years the same way, but it compares successive groups of older people, not the same people ageing. These figures do not show whether activity was done in company, and they say nothing about loneliness.
Pooled trials found a small gain in social functioning, strongest in groups; the evidence on loneliness was too thin to pool
The most direct test is a 2018 systematic review and meta-analysis by Shvedko and colleagues of randomised controlled trials comparing physical activity programmes for older adults living in the community with no activity or a non-activity comparison. Searching to February 2017, it found 38 trials with 5,288 participants (26 at low and 12 at high risk of bias) and pooled 23. Twenty-five of the 38 trials were in people with a long-term condition, and 23 of the programmes were run in groups. Its outcomes were social isolation, loneliness, social support, social networks and social functioning, reported as standardised mean differences, which put different questionnaires on one scale[1].
For social functioning (mostly a sub-scale of quality-of-life questionnaires: broadly, how far health allows people to keep up their usual social activities), physical activity had “a small significant positive effect”, a standardised mean difference of 0.30 (95% confidence interval 0.12 to 0.49). The effect was statistically significant for programmes run in groups (0.34) and not for those done individually (0.21), and it appeared in trials of people with a health condition (0.55), not in trials of healthy older adults (0.00). The authors say the result “must be treated with caution”: they found signs of publication bias, and the effect was no longer statistically significant once they adjusted for it[1].
On loneliness the review could say much less. Its summary reports that “no effect of PA was found for loneliness, social support, or social networks”, but the paper explains that “due to the lack of available data, meta-analysis for loneliness and social isolation outcomes was not performed”[1]. That is not proof that group activity cannot help. “Only one study explicitly targeted lonely or socially isolated people”, and it was a pilot; most trials measured social outcomes as a sideline to fitness or health[1]. A programme cannot reduce loneliness much in people who were not lonely to start with. We found no pooled estimate of the effect of exercise on older people’s loneliness that we could check at source.
Wider reviews point the same way. A 2024 review of 19 systematic reviews, covering 101 primary studies, found only 8% of the primary studies at minimal risk of bias, and that animal-assisted, psychological and skill-building interventions “were more successful than interventions focused on social facilitation or health promotion”[4]. A 2024 umbrella review of 29 reviews of randomised trials found social interventions most promising for social isolation and psychological ones for loneliness[5]. Our study of what works for loneliness covers this wider evidence.
Newer trials of group tai chi, dance and walking report lower loneliness scores, but they are small and short
Since the 2018 review, a handful of trials have tested group activities with loneliness as a named outcome. Each used a different loneliness scale, so their results cannot be added together or compared directly.
| Trial | Who took part | What was tested | Outcome as measured |
|---|---|---|---|
| Group tai chi, Wuhan, China[6] | 140 people aged 65 to 85, all living alone | Tai chi in a group twice a week for six months, against no intervention | Loneliness on the De Jong Gierveld Loneliness Scale fell more in the tai chi group (group-by-time effect −1.837); social self-efficacy rose |
| Dance movement therapy and walking, Turkey[7] | 48 people aged 65 and over registered at a family health centre, 16 per group | Six sessions, twice a week, of dance movement therapy or walking, against a control group | Both groups improved on a loneliness scale for older people (the Loneliness Scale for the Elderly, the scale’s own name) and on self-neglect compared with controls; dance improved more than walking on most measures |
| Physical Activity Intervention for Loneliness (PAIL), Birmingham, England[8] | 25 inactive people aged 60 and over at risk of loneliness | A weekly group walk and health workshop for 12 weeks, against a waiting list | A feasibility trial: it measured recruitment, retention and attendance, not effectiveness |
Both trials that measured effects found loneliness scores fell more among participants than among controls. Each was run in one place; the Turkish trial was very small and lasted six sessions, and neither abstract reports how long any benefit lasted. In Wuhan the comparison group received nothing, so the trial cannot separate tai chi itself from simply meeting a group twice a week.
The PAIL pilot shows how hard this work is. Recruiting inactive older people at risk of loneliness, it reached a recruitment rate of 25% (48 of 195) over four months; 25 were eligible and randomised, average attendance in the walking group was 58.3%, and the trial missed its targets for a full-scale study. The authors concluded that “a more extensive and robust strategy would be needed to support adequate recruitment of lonely older adults”[8].
Walking groups and group dance have good evidence for mood and health, but loneliness is rarely measured
Much research on the activities community groups run asks about health and mood, not loneliness. A 2015 meta-analysis of outdoor walking groups by Hanson and Jones found 42 studies with 1,843 participants, adults of all ages, comparing measurements at the start and end of each programme. Walking groups were linked with lower blood pressure (systolic down 3.72 mm Hg on average) and lower depression scores (standardised effect size −0.67); the evidence was “less clear” for some outcomes, including the SF-36 mental health scale, and “there were no notable adverse side effects reported”[9]. Its abstract reports no pooled result for loneliness or social isolation.
Dance shows a similar pattern. A 2025 network meta-analysis of 13 studies with 1,083 older adults compared nine kinds of dance programme for depression, anxiety, well-being and cognitive function. Chinese square dance did better than control groups for depressive symptoms (standardised mean difference 1.11), anxiety and well-being; for cognitive function, “no significant differences were found”[10]. Loneliness was not among its outcomes.
Better mood is worth having, and depression and loneliness are linked (see loneliness and depression in later life), but a group that lifts mood cannot be assumed to ease loneliness unless loneliness is measured. Music and arts groups have their own evidence: see arts, singing and older people.
In Japan, older people who are not in a sports group are less likely to be homebound where more of their neighbours are
The Japan Gerontological Evaluation Study (JAGES), a nationwide survey of functionally independent older people, treats taking part in a sports group (at least once a month) as social participation, not only exercise. In one cohort of 40,308 older people in 346 communities, followed for six years, sports group participation averaged 25.2% of older residents and ranged from 0.0% to 56.5% between communities[11].
Each 10-percentage-point higher community participation rate was associated with a lower risk of cognitive impairment (hazard ratio 0.92), after adjusting for each person’s own participation, age, sex, illness, social isolation, depression, income, education and other factors[11]. A second analysis, of 2016 survey data, worked out participation rates from 157,233 older people in 1,000 communities and then looked only at people not in a sports group: those living where participation was higher were less likely to be homebound, leaving the house one day a week or less (odds ratio 0.94 per 10 percentage points)[12]. Being homebound is a measure of social isolation, not loneliness.
These are observational associations, not causes: communities with lively sports groups may differ in parks, transport or income in ways adjustment cannot fully capture, and the homebound analysis was a one-off snapshot. They do suggest benefits may reach beyond members, which matters when groups are judged only by who signs up.
Who joins differs by sex and background, so groups may miss the people with most to gain
A Japanese study of 4,211 men and 4,944 women aged 65 and over without disabilities compared exercise groups, exercising informally with others and exercising alone, measuring mental health with the SF-8 survey. Among women, the link with better mental health “was limited to participation in exercise groups”; among men, only exercising alone showed a consistent relationship[13]. It is a cross-sectional association, but one model of group exercise will not suit everyone.
In the 2010 JAGES survey of 23,320 people aged 65 and over, those who grew up in poorer households were less likely to join sports groups (prevalence ratio 0.82 for men, 0.88 for women), a gap “greatly attenuated” once education was taken into account[14], which suggests that differences in opportunity across a lifetime shape who turns up. The groups easiest to fill are not always those that reach people who are housebound, unwell or on low incomes; see social infrastructure and places to meet.
What the evidence asks of us
The evidence does not show that exercise groups reduce loneliness, and it does not show that they fail to. It supports a few practical conclusions for funders, commissioners, researchers and the groups themselves.
- Value group activity for what it has been shown to do. Pooled trials found a small benefit for social functioning, strongest in group settings, though the reviewers urge caution about it[1], and walking groups and group dance have evidence for mood and health[9][10]. Funders can support these groups on that basis without promising loneliness outcomes the trials have not shown.
- Measure loneliness when loneliness is the aim. Too few trials name loneliness as an outcome, and those that do use different scales. Programmes funded to reduce loneliness should measure it with an established scale, include a comparison group and follow people up for long enough to see whether any change lasts. Our study of how loneliness is measured explains the main scales.
- Invest in reaching people, not only in running sessions. The PAIL pilot struggled to recruit and keep older people at risk of loneliness[8], and participation follows lifelong differences in education and income[14]. Outreach, transport, low or no cost and sessions suited to people with health conditions are part of the intervention, not extras.
- Treat groups as community infrastructure. The WHO Commission on Social Connection, reporting in 2025, said solutions “exist at multiple levels – national, community and individual”, including “strengthening social infrastructure (e.g., parks, libraries, cafés)”, and set out a roadmap built on “policy, research, interventions, improved measurement” and public engagement[15]. The JAGES findings that whole communities with more sports group participation do better, including non-members[12], fit that view, as associations still to be tested.
More than anything, the people who run walking groups, tai chi classes and dance sessions, many of them older people themselves, deserve research that asks the right questions of their work. The evidence on loneliness for older people as a whole is gathered in our Loneliness in Later Life library.
Language: this piece says "older people" and treats loneliness as a common human experience that anyone can have, not a personal failing or an illness. It keeps apart three things that are often mixed up: loneliness (how people feel about their relationships), social isolation (how few social contacts they have) and living alone (a household arrangement). Where a study uses another term, such as "seniors" or "older adults", we keep it when quoting that study.
Why we're publishing this
Community groups, u3a groups and older people’s charities run a great deal of the activity older people do together, often on small budgets and volunteer time, and they are increasingly asked to show that it reduces loneliness. They deserve an honest account of what the research has and has not tested, so that they can make their case on solid ground, and so that funders do not judge their work against a promise the trials never made.
How to read this data
Activity levels come from population surveys (a modelled global analysis of adults in 2022[3] and England’s Active Lives survey[2]) that do not measure loneliness or whether activity was done in a group. Intervention evidence comes from randomised trials and reviews. The trials measured loneliness on different scales (the De Jong Gierveld Loneliness Scale in Wuhan, a loneliness scale for older people in Turkey), and the 2018 review pooled questionnaires on social functioning, social support and social networks, but not loneliness, as standardised mean differences[1]; scores from different scales are not directly comparable. The walking-group and dance reviews measured mood and health, not loneliness, and start-and-end comparisons cannot rule out change that would have happened anyway. JAGES results are observational and measure social participation and being homebound, not loneliness. Most trials were small, short and single-site, and where we could read only an abstract we used only its figures. People who join exercise groups differ from those who do not in health, income and education, so trials with comparison groups carry more weight than member surveys.
Use this data
Free to cite with attribution to Understanding Aging. u3a groups, men's sheds, senior centres, lunch clubs, befriending services, charities, campaigners and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.
Important: This article summarises published research and statistics for general information. It is not medical or care advice and cannot be used to diagnose, treat or predict anything about an individual. If loneliness, low mood or isolation is affecting your health or someone else’s, a doctor or another health professional can help, and local services for older people can often put you in touch with groups and support nearby.
Sources
- Shvedko A, Whittaker AC, Thompson JL, Greig CA. Physical activity interventions for treatment of social isolation, loneliness or low social support in older adults: a systematic review and meta-analysis of randomised controlled trials. Psychology of Sport and Exercise, 2018;34:128–137. storre.stir.ac.uk/handle/1893/29872
- Sport England. Active Lives Adult Survey November 2024-25 Report. Sport England, 2026. sportengland.org/research-and-data/data/active-lives
- Strain T, Flaxman S, Guthold R, et al. National, regional, and global trends in insufficient physical activity among adults from 2000 to 2022: a pooled analysis of 507 population-based surveys with 5.7 million participants. The Lancet Global Health, 2024. europepmc.org/article/PMC/PMC11254784
- Patil U, Braun KL. Interventions for loneliness in older adults: a systematic review of reviews. Frontiers in Public Health, 2024. europepmc.org/article/MED/39091527
- Hansen T, Nes RB, Hynek K, et al. Tackling social disconnection: an umbrella review of RCT-based interventions targeting social isolation and loneliness. BMC Public Health, 2024. europepmc.org/article/MED/39020331
- Zhang G, Guo Z, Guo D. The influence of group Tai Chi practice on loneliness among older adults living alone in Wuhan: the mediating role of social self-efficacy. BMC Psychology, 2026. europepmc.org/article/MED/42393768
- Das Gecim GY, Ustun G. The effect of dance movement therapy and walking interventions on loneliness and self-neglect in older people aged 65 and over: a randomized controlled trial. BMC Geriatrics, 2026. europepmc.org/article/MED/41580654
- Shvedko AV, Thompson JL, Greig CA, Whittaker AC. Physical Activity Intervention for Loneliness (PAIL) in community-dwelling older adults: a randomised feasibility study. Pilot and Feasibility Studies, 2020. europepmc.org/article/MED/32489675
- Hanson S, Jones A. Is there evidence that walking groups have health benefits? A systematic review and meta-analysis. British Journal of Sports Medicine, 2015. europepmc.org/article/MED/25601182
- Zhang Y, Ma H. The impact of dance on the mental health of older adults: a network meta-analysis of anxiety, depression, and well-being. Frontiers in Psychology, 2025. europepmc.org/article/MED/41394050
- Tsuji T, Kanamori S, Miyaguni Y, Hanazato M, Kondo K. Community-level sports group participation and the risk of cognitive impairment. Medicine and Science in Sports and Exercise, 2019. europepmc.org/article/MED/31205226
- Tsuji T, Kanamori S, Miyaguni Y, Kondo K. Community-level sports group participation and health behaviors among older non-participants in a sports group: a multilevel cross-sectional study. International Journal of Environmental Research and Public Health, 2021. europepmc.org/article/MED/33435252
- Tomioka K, Shima M, Saeki K. Exercising alone in men and group exercise in women are cross-sectionally associated with positive mental health among older Japanese. Environmental Health and Preventive Medicine, 2026. europepmc.org/article/MED/41813155
- Yamakita M, Kanamori S, Kondo N, et al. Association between childhood socioeconomic position and sports group participation among Japanese older adults: a cross-sectional study from the JAGES 2010 survey. Preventive Medicine Reports, 2020. europepmc.org/article/MED/32123650
- World Health Organization. Social connection linked to improved health and reduced risk of early death (news release on the report of the WHO Commission on Social Connection). WHO, 30 June 2025. who.int/news/item/30-06-2025-social-connection-linked-to-improved-heath-and-reduced-risk-of-early-death