Studies of seven projects (approximately 25,000 participants) were included. There were 2 RCTs (1,454 participants), 2 quasi-experimental studies with matched pairs (16,874 participants), 2 quasi-experimental non-randomised studies (approximately 7,000 participants) and 1 before-and-after study (115 participants).
Darlington (UK).
One government-funded project based in the UK (101 patients discharged from long-term hospital to the community) showed that a community care project compared with continuing institutionalised care reduced the rates of institutionalisation (50% at home after 12 months) and increased the number of days at home (137 days versus 12 days). The project also increased the use and appropriateness of community services, and significantly increased morale, patient satisfaction and depression; it showed no effect of the intervention on carer stress. The preferential selection of short-term patients who still had their own community accommodation was a potential source of bias. The project may have limited generalisability due to the requirement for extensive social support or only moderate dependency.
On Lok and PACE (USA).
Two similar Medicare, Medicaid and private premium-funded projects based in the USA sought to maintain participants in the community. One project (300 community-dwelling participants eligible for nursing home care; matched pairs design) showed that a community care project significantly improved some measures of functional independence and reduced the rates of hospitalisation and use of skilled nursing care. The other project (approximately 7,200 participant in a much greater catchment area; the control group was people who refused to enrol in the project) attempted to duplicate this project, but no results for the rates of institutionalisation were (as yet) available. There appears to have been problems in implementing the project due to the larger geographical area covered.
S/HMO (USA).
One Medicare, Medicaid, private premium and copayment-funded project based in a social health maintenance organisation in the USA (16,547 older participants; matched control from non-institutionalised Medicare clients) showed that a community project did not improve the health outcomes or service use in comparison with a fee for service system. The role of the case manager differed from other successful projects and the authors felt other organisational elements may also have had a negative impact.
Roverto and Vittorio Veneto (Italy).
One RCT (200 community-dwelling frail elderly in Italy) showed that, compared with standard home services, a government-funded comprehensive community-based project decreased the use of community services, institutionalisation and nursing home admission and significantly reduced acute hospital admission at one year. One before-and-after study (115 applicants for integrated home care service in Italy) showed that a government-funded comprehensive community-based project improved some functional outcomes and significantly reduced the number and duration of hospital admissions compared with the 6 months before the project.
SIPA (Canada).
The results were only available for the first 3 months of one RCT (1,254 community-dwelling frail elderly) of a primary care-based system. It showed that a government-funded project had a trend towards increased use of community services, reduced use of hospital emergency services, and an increase in preventative services such as vaccination.
Elements common to successful projects were case management, geriatric assessment, and multidisciplinary team. The one unsuccessful project did not use a multidisciplinary team and geriatric services were either weak or not present. Two successful projects (both non USA-based) used a single entry point. Financial levers were used by three projects (two successful and one with no results as yet).