Inside China’s Integrated Care Experiment: What Chengdu’s Older Adults Really Want

China is ageing faster than almost any society in history, and the pressure is landing hardest on the communities where older adults actually live. In Chengdu, a sprawling metropolis of more than 20 million people in the country’s southwest, researchers have now taken one of the most detailed looks yet at how the nation’s flagship policy for merging medical treatment with elderly care is performing on the ground. The results, published in Frontiers in Health Services, reveal a system that is popular in principle, uneven in practice, and riddled with gaps between what institutions can deliver and what older people say they need.

The policy at the centre of the study is known as integrated medical and elderly care, or IMEC. Since the mid-2010s, Beijing has pushed hard to dissolve the historical wall between the health sector, which treats disease, and the civil affairs sector, which looks after frail elders. The result has been a wave of hybrid institutions: hospitals that opened long-term care wings, and nursing homes that hired doctors and nurses. On paper, these facilities promise seamless care for people whose needs span both worlds. Whether they actually deliver it is a far more complicated question.

To find out, a team led by researchers at Sichuan Provincial Women’s and Children’s Hospital and Chengdu Medical College ran two parallel surveys between October 2023 and May 2024. On the supply side, they approached all 132 registered IMEC institutions in Chengdu and obtained usable data from 127 of them, split between 77 that grew out of medical providers and 50 that evolved from elderly care facilities. On the demand side, they interviewed 2,525 adults aged 65 and older recruited from communities across the city’s inner, middle and outer rings. The two datasets were never linked at the facility or neighbourhood level, so the comparison is deliberately descriptive rather than causal, but the juxtaposition is revealing.

The institutional survey exposed deep structural heterogeneity hiding beneath a single policy label. Medical-run care institutions, the hospitals-turned-care-providers, had a median of 62.5 per cent of their beds designated as medical beds, compared with just 12.9 per cent in care-run medical institutions. Staffing told the same story: per 100 beds, the medical-run facilities employed roughly twelve times more doctors, eight times more nurses and nearly ten times more technicians. The care-run institutions, by contrast, devoted 87 per cent of their beds to long-term care and carried far more elderly-care attendants relative to bed capacity. Yet when the researchers standardised by actual resident numbers, attendant density was statistically indistinguishable between the two models, at roughly 19 attendants per 100 residents each. The real difference, it turns out, lies in bed allocation and medical staffing, not in the intensity of hands-on care.

Perhaps the most striking finding is what did not differ: overall service breadth. Both organisational models reported considering a median of around 11 of 12 core service domains as needed by community-dwelling older adults, and both claimed the ability to provide only about four to five of them. In other words, whichever door an older person walks through, the menu of services looks broadly similar, and it is strikingly incomplete relative to what providers themselves believe is needed.

That gap between perceived need and actual capacity was the study’s second headline. For palliative care, 63.8 per cent of institutions said community elders needed it, but only 13.4 per cent could provide it. For family beds and home-based medical services, the figures were 58.3 versus 9.4 per cent. Medical services, rehabilitation guidance and psychological or spiritual support all showed shortfalls of roughly 40 percentage points. Even the humblest information function, establishing a health record, was deemed necessary by 72.4 per cent of institutions while only 44.1 per cent could actually deliver it. The authors are careful to note that this is a within-institution comparison, not proof of city-wide supply-demand mismatch, but the pattern is hard to ignore.

The resident survey echoed those gaps from the other direction. Among the 2,525 older adults, 68.8 per cent said they would be willing to choose a community-based integrated care model, a remarkably high figure that suggests the policy has genuine latent appeal. But willingness did not track health status. Sex, chronic disease burden, disability level and out-of-pocket medical spending showed no significant association with willingness in the multivariable logistic regression. What mattered instead were two things: perceived need for convenient home and community medical services, which nearly tripled the odds of willingness with an adjusted odds ratio of 2.76, and simple awareness that community-based integrated care existed, which raised the odds by 55 per cent. Awareness, in a system supposedly built around these services, was reported by only 65 per cent of respondents.

The importance-receipt differences among residents were equally telling. Medical services and health examinations were both widely received and widely valued, with 53.4 and 65.5 per cent of respondents respectively having used them. But the services that would actually keep people living independently at home lagged badly. Family beds and home-based medical services were considered important by 65.7 per cent of older adults yet had been received by only 36.2 per cent. Rehabilitation guidance showed a 58.3 versus 29.1 per cent split, and psychological or spiritual support a 38.5 versus 28.6 per cent gap. These are not fringe services; they are the backbone of ageing in place, the policy goal that Chinese planners, like their counterparts worldwide, have embraced as both humane and fiscally necessary.

The authors connect these findings to a broader conversation about data-supported continuity of care, including the emerging concept of personal health data spaces, or PHDS. The logic is straightforward: integrated care fails if a patient’s information does not travel with them between home visits, community clinics and hospitals. The Chengdu data suggest the organisational foundation for such data flows is shaky, since nearly a third of institutions cannot even establish basic health records. The researchers are careful to stress that their survey did not measure electronic interoperability, consent architecture or cross-provider data exchange, so the PHDS connection remains conceptual. Still, the finding identifies a concrete, measurable weakness in the care-continuity chain that any future data infrastructure would have to fix.

For policymakers, the study’s implications are pointed. Expanding the number of IMEC institutions, the authors argue, matters less than reshaping what those institutions actually offer. Home-based medical care, rehabilitation, chronic disease management and psychosocial support should be treated as core functions rather than optional extras. Service visibility deserves equal attention, since the strong awareness-willingness link suggests that many older adults may simply not know what is available, although the cross-sectional design means this remains a hypothesis rather than a demonstrated causal pathway. The authors also call for differentiated staffing standards and purchasing arrangements for the two organisational models, closer alignment with health insurance and long-term care financing, and spatially sensitive planning, given that the share of residents aged 65 and older in Chengdu ranges from under 10 per cent in some outer districts to over 21 per cent in others.

The study has honest limitations, which the authors catalogue in detail. The household sample was a geographically distributed convenience sample, not a probability survey, and counts of refusals and non-response were not retained. Community identifiers were lost, so within-neighbourhood clustering could not be modelled, potentially making the confidence intervals overly precise. Age, education and income were never collected, leaving the regression open to residual confounding. The perceived-need variable sits conceptually close to the willingness outcome, and the authors’ own sensitivity analysis, which dropped it, produced a model with weak discrimination. Willingness, moreover, is not the same as use. None of this undermines the central descriptive picture, however: a policy that older adults broadly want, delivered by institutions that differ enormously depending on their origin, and falling short precisely in the home-based, rehabilitative and psychosocial services that would make ageing in place real.

Subject of Research: Organisational capacity and older adults' willingness to use community-based integrated medical and elderly care in Chengdu, China

Article Title: Community-based integrated medical and elderly care in Chengdu, China: organisational heterogeneity, service-capacity and service-receipt patterns, and implications for care continuity

Article References: Wang, X., Peng, Y., Liu, J., Bai, M., Peng, J., Heng, C., Yang, Y., Lu, H., Wang, X., Li, C., & Luo, T. (2026). Community-based integrated medical and elderly care in Chengdu, China: organisational heterogeneity, service-capacity and service-receipt patterns, and implications for care continuity. Frontiers in Health Services, 6, Article 1875860. https://doi.org/10.3389/frhs.2026.1875860

Image Credits: AI Generated

DOI: 10.3389/frhs.2026.1875860

Keywords: integrated care, elderly care, China, Chengdu, ageing in place, long-term care, health policy, care continuity, community health services, cross-sectional study, health records, older adults