Young Adults Loneliest, Older Adults Most Contented in Huge English Volunteer Survey

A survey of more than 135,000 adults in England has found that loneliness falls steadily with age, with people aged 16 to 25 reporting by far the highest levels, while older adults reported the lowest. The analysis, published in BMC Global and Public Health, also examined social capital — neighbourhood trust, cohesion and reciprocity — and found it rose with age, marriage and retirement, but was lower among people with disabilities or long-term health conditions.

The study is the second major paper from the Measuring Loneliness in England (INTERACT) Study, described by its authors as one of the largest volunteer-based investigations of loneliness and social disconnection conducted in the United Kingdom. Responses were collected between 1 March and 31 July 2023 from 135,725 adults aged 16 and over, recruited through NHS primary care networks, voluntary sector organisations and the National Institute for Health and Care Research’s Be Part of Research Network. Because participation was voluntary, the researchers stress that the results describe the responding sample only and cannot be treated as estimates for the population of England as a whole.

Loneliness was measured with two validated instruments: the University of California, Los Angeles three-item Loneliness Scale (UCLA-3), an indirect multi-item measure scored from 3 to 9, and the Office for National Statistics Direct Measure of Loneliness (DMOL), a single-item measure of how often a person feels lonely. Overall, 30.5 percent of respondents were classified as experiencing severe loneliness on the UCLA-3, while 16.5 percent reported feeling lonely often or always on the DMOL. Severe UCLA-3 loneliness was most frequent among the youngest respondents, affecting 54.9 percent of those aged 16 to 25 and falling progressively to 16.4 percent among those aged over 65.

To quantify how these patterns varied simultaneously across characteristics, the team fitted multivariable ordinal logistic regression models for the two loneliness outcomes and a binary logistic regression model for social capital, which was measured with a seven-item scale dichotomised at the median. The adjusted models summarise mutually adjusted associations within the responding sample; the authors explicitly state they were not intended to estimate population parameters, test hypotheses or establish causal or predictive relationships. Uncertainty is reported as 95 percent model-based intervals, which reflect the fitted regression model only and do not capture uncertainty from volunteer recruitment or self-selection.

Age showed the strongest inverse association with loneliness. Compared with respondents aged 16 to 25, those aged 26 to 35 had 35 percent lower adjusted odds of being in a more severe loneliness category on the UCLA-3 (adjusted odds ratio 0.65), while those aged 65 and over had 86 percent lower adjusted odds (aOR 0.14). A similar pattern appeared on the DMOL, where respondents aged 65 and over had an adjusted odds ratio of 0.19 relative to the youngest group. Men reported lower loneliness than women on both measures, with adjusted odds ratios of 0.81 on the UCLA-3 and 0.74 on the DMOL.

Social contact emerged as one of the most powerful correlates. Relative to respondents reporting no friends, those reporting nine or more friends had substantially lower adjusted odds of severe loneliness — an aOR of 0.09 on the UCLA-3 and 0.16 on the DMOL. Nine or more relatives were likewise associated with lower loneliness on the UCLA-3 (aOR 0.28). Marital status mattered as well: married or civil-partnered respondents had markedly lower adjusted odds of severe loneliness than single respondents (aOR 0.44 on both scales), whereas widowed respondents had higher adjusted odds than single respondents (aOR 1.34 on the UCLA-3 and 1.49 on the DMOL).

Employment and health were also consistently linked to loneliness. Relative to respondents in full-time employment, unemployed respondents had higher adjusted odds of severe loneliness (aOR 1.82 on the UCLA-3, 1.68 on the DMOL), as did unpaid carers (aOR 2.24 and 1.68 respectively). Respondents with a disability had 77 percent higher adjusted odds of severe UCLA-3 loneliness (aOR 1.77) and higher odds on the DMOL (aOR 1.44), while a long-term condition raised the adjusted odds by 64 percent on the UCLA-3 (aOR 1.64) and similarly on the DMOL (aOR 1.53). Interestingly, education showed divergent patterns: university graduates had higher adjusted odds of severe loneliness on the UCLA-3 (aOR 1.22) but slightly lower odds on the DMOL (aOR 0.96), a divergence the authors attribute to the different constructs captured by indirect and direct loneliness measures, alongside possible contextual factors such as differing expectations of social connectedness.

Social capital, analysed as a separate contextual outcome, showed the opposite age gradient to loneliness. Respondents aged 46 to 55 had more than twice the adjusted odds of high social capital compared with the youngest group (aOR 2.38), and those aged 65 and over had more than threefold higher odds (aOR 3.21). Married respondents (aOR 1.60), widowed respondents (aOR 1.39) and retired respondents (aOR 1.49) also reported higher social capital, while unemployed respondents and unpaid carers reported lower levels. Social contact again dominated: respondents with nine or more friends had more than fivefold higher adjusted odds of high social capital relative to those with none (aOR 5.22). In contrast, disability (aOR 0.77) and long-term conditions (aOR 0.82) were associated with reduced odds of high social capital.

The authors highlight that the consistency of several associations across the two loneliness measures — particularly those involving age, marital status, social contact, employment and health — provides a rationale for testing these patterns in representative, longitudinal research. They also caution against reading the findings as protective effects or policy prescriptions. Because the sample was recruited through voluntary self-selection, individual inclusion probabilities were unknown, and the direction and magnitude of selection bias cannot be quantified. The large sample size reduces random error but does not correct systematic selection bias, so the narrow model-based intervals do not establish population representativeness. The cross-sectional design likewise precludes causal inference, with reverse causation and residual confounding remaining plausible explanations.

Further limitations include the reliance on constrained loneliness measures that do not capture the multidimensional nature of the experience, the exclusion of income and finer socioeconomic indicators, the absence of area-level variables such as rurality and neighbourhood deprivation from the current models, and the use of main-effects models that cannot capture intersecting vulnerabilities. Online sampling may also underrepresent people with cognitive or digital access barriers, particularly among older adults. Within these constraints, the researchers position the work as hypothesis-generating: the observed gradients in age, social ties and health status identify groups and contextual factors that warrant investigation in probability-based samples, longitudinal designs and future intervention studies testing approaches such as intergenerational programmes, community hubs and neighbourhood-level initiatives aimed at strengthening social connection.

Subject of Research: Cross-sectional analysis of demographic, social and health-related factors associated with loneliness and social capital in a large volunteer sample of adults in England

Article Title: Factors associated with loneliness and social capital in a large volunteer sample of adults in England

Article References: El-Osta, A., Al-Ammouri, M., Alaa, A., Altalib, S., Tristán-López, A., & Majeed, A. (2026). Factors associated with loneliness and social capital in a large volunteer sample of adults in England. BMC Global and Public Health, 4(1), Article 96. https://doi.org/10.1186/s44263-026-00335-1

Image Credits: AI Generated

DOI: 10.1186/s44263-026-00335-1

Keywords: loneliness, social capital, INTERACT study, England, UCLA-3 Loneliness Scale, ONS Direct Measure of Loneliness, cross-sectional survey, young adults, social contact, disability, long-term conditions, non-probability sampling