The research, published in BMC Geriatrics by Honghui Pan of National Chung Cheng University and LiFan Liu of National Cheng Kung University’s Institute of Gerontology, is notable for its methodological ambition. Rather than treating loneliness as a purely personal or psychological problem, the authors explicitly modeled it as a phenomenon shaped at two distinct levels: the community in which an older person lives and the individual circumstances of that person’s daily life. This multilevel framing reflects a growing consensus in gerontology that loneliness is not simply a character trait or a private misfortune, but the product of layered social, physical, and environmental conditions that can be measured and, in principle, changed.

The data came from the University Responsibility dataset collected by National Cheng Kung University in Tainan city, Taiwan, a secondary dataset that the researchers repurposed for this analysis. Loneliness itself was quantified using the six-item De Jong Gierveld Loneliness Scale, a widely validated instrument from 2006 that captures both emotional loneliness, the absence of intimate attachment, and social loneliness, the absence of a broader social network. Because the scale is short and psychometrically robust, it has become a standard tool in international loneliness research, allowing findings from Taiwan to be compared meaningfully with studies from Europe and North America.

To identify what drove loneliness in this population, the team employed hierarchical multiple linear regression, a statistical technique that enters groups of predictor variables into the model in successive blocks. This approach allowed the researchers to estimate the unique contribution of community-level factors after accounting for individual-level factors, and vice versa, while also controlling for a battery of demographic covariates including age, gender, marital status, ethnicity, educational level, health, and income. The community-level predictors examined were a non-age-friendly environment, lower social capital in the community, and lower levels of social support from family and friends. At the individual level, the study tested negative perceptions of the physical living environment and lower levels of social participation.

The results delivered a strikingly clear verdict. At the community level, the single significant predictor was social capital, the web of trust, reciprocity, and shared norms that binds neighbors together, with a standardized coefficient of β = -0.562 (p < 0.01). In practical terms, older adults living in communities rich in social capital reported substantially less loneliness, while those in communities where that fabric had frayed reported far more. Notably, neither the perceived age-friendliness of the environment nor the level of social support from family and friends emerged as statistically significant predictors once social capital and individual factors were accounted for, a finding that challenges the intuitive assumption that family support alone can shield older people from isolation.

At the individual level, the dominant predictor was housing satisfaction, with a standardized coefficient of β = -0.702 (p < 0.01), the strongest effect in the entire model. Older adults who felt positively about their physical living environment, their home, its comfort, and its suitability for their needs, were markedly less lonely, regardless of how much social capital surrounded them. Social participation, by contrast, did not reach significance in the final model. This pattern suggests that the subjective quality of the place where a person lives may do more to buffer loneliness during a period of enforced physical distancing than the frequency of social activities, which the pandemic had curtailed for nearly everyone.

Beyond the two focal levels of analysis, three demographic covariates also proved significant: self-rated health, marital status, and low-income status. These findings align with a long international literature showing that poorer subjective health, the loss of a spouse or the absence of a partner, and economic hardship each amplify the risk of loneliness in later life. Their persistence in a model that already included community and environmental variables underscores that loneliness among older adults is multiply determined, arising from the intersection of bodily vulnerability, relational status, material security, neighborhood cohesion, and the felt quality of one’s home.

The scale of the problem in Taiwan carries particular weight because the country’s demographic trajectory is among the steepest in the world. Taiwan has rapidly transitioned into a super-aged society, and the vast majority of its older citizens live in the community rather than in residential care. The pandemic’s restrictions on gatherings, visits, and everyday face-to-face contact therefore severed precisely the informal channels, temple groups, markets, neighborhood chats, and family dinners, through which community-dwelling elders traditionally sustain their social worlds. The finding that nearly 80 percent of the sample was lonely during this period suggests that those channels were not adequately replaced by digital or distanced substitutes, at least for this cohort.

From these results, the authors draw two practical recommendations for intervention. First, community-based initiatives aimed at alleviating loneliness should concentrate on building community social capital rather than merely delivering individual services. That implies investing in the collective infrastructure of neighborhood life, the shared activities, mutual-aid networks, and local organizations that generate trust and reciprocity among residents, so that older adults are embedded in a supportive social ecology even when formal support is disrupted. Second, programs should work to foster older adults’ housing satisfaction, since the subjective experience of one’s dwelling proved to be the strongest individual-level protector against loneliness. Improvements to housing conditions, adaptations that make homes safer and more comfortable, and attention to how older people feel about where they live may therefore yield loneliness-reduction benefits that conventional social programming alone cannot achieve.

The study’s implications extend well beyond Taiwan. Loneliness has climbed policy agendas across many Western countries, with governments appointing ministers and launching national strategies, but empirical evidence from East Asian ageing societies has remained comparatively scarce. By demonstrating that community social capital and housing satisfaction are the pivotal predictors in a Taiwanese context, the research suggests that anti-loneliness policy everywhere should be built on two pillars simultaneously: strengthening the social fabric of neighborhoods and improving the lived experience of the home. As pandemics, demographic ageing, and urbanization continue to reshape how older people connect with the world around them, the lesson from Tainan is that loneliness is neither inevitable nor purely personal. It is a measurable outcome of community design, housing quality, health, and social structure, and that means it is something societies can deliberately engineer downward, one neighborhood and one home at a time.

Subject of Research: Loneliness and its community-level and individual-level predictors among community-dwelling older adults in Taiwan during the early Covid-19 pandemic

Article Title: Loneliness among community-dwelling older adults during the early stage of the Covid-19 pandemic in Taiwan: exploring community-level and individual-level predictors

Article References: Pan, H., & Liu, L. (2026). Loneliness among community-dwelling older adults during the early stage of the Covid-19 pandemic in Taiwan: exploring community-level and individual-level predictors. BMC Geriatrics. https://doi.org/10.1186/s12877-026-08170-7

Image Credits: AI Generated

DOI: 10.1186/s12877-026-08170-7

Keywords: loneliness, older adults, Covid-19, Taiwan, social capital, housing satisfaction, community health, gerontology, De Jong Gierveld Loneliness Scale, social participation, age-friendly environment, public health