The community hospital is located in the market town of Leek, North Staffordshire, UK. The day-to-day medical cover (9a.m. to 5p.m., Monday to Friday) is provided by General Practitioners with Special Interest in Geriatric Medicine. Cover outside of these hours is provided by a local out-of-hours medical cooperative. In-patient care at Leek Hospital is multi-disciplinary and the team comprises medical and nursing staff, Physiotherapists, Occupational therapists, Rehabilitation assistants, Social Worker and a Discharge Facilitator. A Consultant Physician specialised in Geriatric Medicine conducts a detailed ward round once weekly.

Most patients are transferred to the unit from the local Acute University Hospital, Stoke-On-Trent, North Staffordshire, once they have recovered from acute illness but have not regained their pre-admission functional state and are unable to return to the community. A minority of patients are admitted directly from their home without first going to the Acute Hospital. All patients must be cooperative and have the physical capacity to undergo a rehabilitation programme. Patients with Dementia are accepted if their Dementia is mild or moderate, they have good insight, and are able to participate in the rehabilitation process. Each patient has a Collaborative Care Document file which includes initial assessment notes, medical, nursing, physiotherapy, occupational therapy and social work entries, multi-disciplinary goals-setting, and discharge documentation. All staff have access to these files allowing for a holistic approach to patient care, which is important in this group of elderly patients with complex needs. Multidisciplinary meetings are held once a week to discuss patients’ progress and any issues that need addressing.

The service began in 1997 as a small unit that gradually expanded to the current status of two similar wards with 21 beds in each. Our analysis focuses on patients in the 31 beds under the care of the Consultant Physician and his team. Patients in the remaining 11 beds are managed by the patients’ own General Practitioners under a different service contract and were excluded from this evaluation.

The primary aim was to assess the effectiveness of Intermediate Care and Rehabilitation in terms of individual patient outcomes in functional ability as measured by the Barthel index. Secondly, other variables (age, sex, length-of-stay, number of prescription items on discharge documents, admission Barthel score and where the patients have been admitted from) were analysed to identify any factors which may have a relationship to the primary outcome.

After review of the plan for the project by the Student Project Committee of the Undergraduate School of Medicine, the project was defined as a service evaluation and further scrutiny from the Local Research Ethics Committee was not required.

All Patients admitted between 1 ^st November 2009 and 30 ^th April 2010 were included.

As achieving a good functional status and independence are fundamental goals of Intermediate Care and Rehabilitation, measuring this outcome is useful in assessing effectiveness of the service. An appropriate method of measuring patients’ functional ability is by focusing on basic activities of daily living. The Barthel index is a tool widely used in routine clinical practice to measure how independent somebody is when performing activities such as feeding, dressing and washing themselves (7). The reliability and validity of the Barthel index have been studied and found to be high (8,9,10). The Barthel Index has been used before as a measure of outcome in older patients undergoing rehabilitation and was found to be effective (11).

The relevant information was obtained anonymously from the medical case-notes, collaborative care documents, and the ward admission book where details of all patients are recorded. The data for each patient was entered onto a separate data collection sheet and then transferred to the computerised database OpenOffice.org 3.2 Calc. The items collected comprised: age (years at admission); where patients were admitted from; original type of residence; length of inpatient stay (in days); Barthel scores on admission and at discharge for assessing change in functional status (these are recorded routinely in all patients); the number of prescription items at discharge as a proxy of morbidity burden (see more information below); and discharge destination. Patients who were transferred to the acute hospital or died did not go through the standard discharge process and so did not have the full set of discharge documentation. Patients were excluded from full analysis if they had incomplete data.

The number of prescription items on discharge was used as a measure of each patient’s morbidity burden as it has been shown to be a reliable a measure of chronic disease in a large community study on older patients (12). It is also easier to collect and verify.

Descriptive and statistical analyses of the data were performed using the StatsDirect software programme. Simple descriptive analysis was carried out on the whole dataset. For patients who were acutely transferred elsewhere or had died, only the descriptive analysis could be undertaken, whereas for patients who had complete data including admission and discharge Barthel scores (patient who had been discharged either home or to institutional care) additional statistical analyses were performed. The non-parametric Wilcoxon signed rank test for paired data was used to compare the admission and discharge Barthel scores to assess if there was a significant change. The Spearman’s rank correlation coefficient tests were used to test the association between the change in Barthel score and: age; length of inpatient stay; number of prescription items; and the admission Barthel score as a measure of initial level of disability. If these univariate analyses revealed significant associations, the intention was to perform multiple regression analysis to assess the relative independence of these factors as predictors of functional improvement. Statistical significance was accepted at p-values of <0.05 and all tests were two-tailed.

The total number of patients admitted to the Unit in the six month period of study was 180. Of these, 35 were excluded for the following reasons: 19 had incomplete records, in 12 patients the case-notes could not be located, and 4 patients were still on the ward. The remaining 145 (81%) had mean age of 82.1 years (SD 9.7), median length of inpatient stay 25 days (interquartile range 29), and 58% were female. 88% were admitted from the local Acute hospital and 12% were admitted directly from their place of residence.

Of the 145 patients who were analysed, 21 were transferred back to the Acute hospital and 16 died. These 37 patients did not return to the community as planned and therefore were not investigated further. Thus, 108 patients had been discharged back into the community and had complete clinical data for full analysis.

The group of 108 and the group of 37 patients were similar in age i.e. 81.8(SD10) and 82.8(SD8.8) years respectively (p>0.05), male/female distribution i.e. 59% and 54% respectively were female (p>0.05), and origin of admission i.e. 89% and 87% respectively were admitted from the Acute hospital (p>0.05) and 11% and 13% directly from their residence (p>0.05). However, the length of stay in Leek Hospital was significantly shorter for the 37 patients who died or were transferred back to the Acute hospital i.e. median 11 (interquartile range 23) days versus 32 (interquartile range 22), p=0.049 (Mann-Whitney U test), reflecting more severe illness and/or acute complications.

Of the 108 patients whose therapy was completed and were discharged back to the community, 77% went back to their own home and only 23% to a Residential or Nursing Home. The median number of drug prescriptions at discharge was 6 (interquartile range 4). The median admission and discharge Barthel scores were 10 (interquartile range 8) and 14 (interquartile range 8) respectively (p=0.0001), reflecting a significant improvement in functional status.

No significant association was found between the change in Barthel score and the following: patients' age (r =0.12, 95% CI ?0.06-0.31, p=0.12); length of inpatient stay (r = ?0.02, 95% CI ?0.21-0.17, p=0.81); number of prescription items at discharge (r = ?0.05, 95% CI ?0.23-0.14, p=0.62); and the admission Barthel score (r = ?0.14, 95% CI ?0.32-0.05, p=0.16).