Randomised controlled trial to evaluate early discharge scheme for patients with stroke

Case ascertainment and entry criteria

A hospital based stroke register was maintained at St Thomas's and King's College Hospitals, London, between January 1993 and July 1995. The World Health Organisation's definition of stroke was used.10 Twice weekly checks of the wards were undertaken by two dedicated research associates with nursing training. Patients were managed in the normal way until the criteria were fulfilled for entry into the trial. If patients lived alone they needed to be able to perform functional independent transfer, and if they lived with a willing carer they needed to be able to perform transfer with assistance. The point at which these criteria were met was decided after consultation with the hospital physiotherapist. All patients were assessed within one working day of notification by a consultant physician or at King's College Hospital by a medical registrar, who obtained informed consent from the patient or next of kin when the patient had cognitive impairment or dysphasia. Baseline data were collected after randomisation by a researcher blinded to which arm of the trial the patient was in. Ability to transfer did not necessarily imply that discharge could take place immediately. The rehabilitation team responsible for the patient was notified that the community team was available to provide care after discharge and it was left to the team to plan discharge whenever feasible. The study obtained approval from the local ethics committees. Patients were excluded if they lived too far away for the team to visit.

Randomisation

After consent randomisation was restricted in permuted blocks of 10 with random number tables provided in blank sealed opaque envelopes. Patients were allocated to continue conventional treatment or to receive community treatment. No stratification was performed.

Interventions

Patients allocated to conventional care continued with their treatment, discharge planning, and outpatient care in the normal way. Stroke care at both hospitals is similar and well coordinated. About half of the patients who are admitted receive treatment in a stroke unit, with the remainder being treated in general medical or elderly care wards. Outpatient resources available to them included a hospital based stroke clinic, geriatric day hospital, generic domiciliary physiotherapy and speech and language therapy, hospital outpatient physiotherapy, and the usual community resources. The maximum level of home care available in the study area to all patients was three one hour visits daily by a home help for personal care, meals on wheels, and community nurse visits for specific tasks.

Patients randomised to the community therapy team remained in hospital until the required package of social services care could be organised and any home adaptations undertaken. A store of commodes, high chairs, and toilet frames was kept by the team to expedite discharge. The patients were assessed for rehabilitation needs before discharge in conjunction with the hospital based therapists to set initial objectives and to ensure continuity of care. After discharge, patients were given a planned course of domiciliary physiotherapy, occupational therapy, and speech therapy, with visits as frequently as considered appropriate (maximum one daily visit from each therapist). Each patient had an individual care plan, which was reviewed at a weekly team meeting. Patients received care from the team for a maximum of three months. On discharge patients were referred to conventional services when appropriate. All other services apart from therapy were as described for the conventional group. There was no augmentation of social services resources.

The community therapy team comprised a senior physiotherapist grade 1 with neurological training, a senior occupational therapist grade 1, a half time speech and language therapist with adult neurological training, and a full time therapy aide. A consultant physician (AGR) coordinated the team and chaired the weekly clinical meeting.

Assessment of outcome

Baseline data on demography and severity of stroke were collected within 48 hours of randomisation. Two research associates conducted all the assessments at discharge and at 2, 4, and 6 months. A third research associate conducted the assessment at 12 months by interview, usually in the patient's home, and she was blind to the treatment group, although occasionally the patient did make clear which group they had been in. No evaluation of the efficacy of the blinding procedure was performed.

A range of standardised outcome assessments was used to measure aspects of impairment, disability, and handicap and could be completed in most cases in 45 minutes.11

Impairments were assessed with the following tests.

Motoricity index, which is a measure of limb function with a maximum score of 100 for normal subjects. Severe paralysis is defined as a score of 0-32, moderate as 33-64, and mild as 65-99.

Minimental state examination—Severe cognitive impairment is defined as a score of 23 or less out of a total of 30.

Frenchay aphasia screening test—is a screening instrument to detect aphasia in patients with stroke. A score of 13 or less out of 20 indicates aphasia.

Disability was measured with the following tests.

Modified Barthel score—To try to overcome the recognised differences between the intervals at the lower, middle, and upper ends of the scale, analysis was also performed by dividing the scale into 0-14 (severe disability), 15-19 (moderate disability), and 20.

Rivermead activities of daily living score has been validated for use in elderly patients with stroke. Scores range from 15-45, with 15 indicating dependence in all measured activities.

Hospital anxiety and depression scale is a 21 point scale, with a cut off point of greater than 10 identifying those with a high probability of the disorder and 8-10 being borderline. It is designed for self assessment, creating problems for those with language, cognitive, and concentration problems.

The 5 metre timed walk—In a trial of late physiotherapy after stroke Wade et al used a 10 metre walk,12 suggesting gait speed offers a simple and sensitive measure of outcome. For many housebound patients 10 metres was impracticable, and 5 metres was therefore substituted.

Nottingham health profile was used to assess subjective health status across six domains: energy, pain, emotion, sleep, social, and physical mobility. The maximum total score is 45, with a high score indicating poor health status.

Caregiver strain index measured carers' stress, the index having 13 items. Overall stress score was obtained by summing the subjects' ratings across all the items.

Patient and carer satisfaction was assessed by using stroke specific questionnaires. Analysis was performed separately for questions relating to hospital care, rehabilitation, and home care.

To evaluate the differences in utilisation of services, length of stay, number of therapy sessions with one unit being defined as a 20 minute session per therapist, hospital readmission, and living conditions on discharge and at one year were documented.

Statistical analyses

The associations between the normally distributed continuous variables and treatment group were examined by using the unpaired t test. When the continuous variables showed evidence of a non-normal distribution, the Mann-Whitney test was used and the confidence interval calculated for the difference between medians. The χ² test was used to investigate the associations between the categorical variables and the treatment group. For ordered categorical variables the χ² test for trend was used. The sample size of 130 per group at one year has a power of 80% at the 5% level of detection for a 3.5 point difference in Barthel score based on a standard deviation of 10.

Methodological considerations

One year outcome measures support the notion that early discharge to a rehabilitation team in the community has no adverse effect on outcome. The Barthel score is the most widely used measure in stroke rehabilitation trials and was used in this study because of the importance of longer term disability. It could be argued that the intervention was more likely to affect the Rivermead at one year. A post hoc power calculation for the Rivermead activities of daily living score, with about a quarter of patients scoring the minimum score (15) at 1 year, showed a power of 80% to detect more than 41% or less than 12% of the early discharge group scoring 15. Satisfaction with care was not significantly different between the two groups from the perspectives of carers or patients, although patients in the community therapy group were more satisfied with the care provided in the hospital. The longer patients stay in hospital the more chance they have of being exposed to aspects of care that dissatisfy them. Alternatively the higher dissatisfaction among the patients receiving conventional care reflected the fact that they were exposed to a rehabilitation philosophy within the context of an acute hospital. Patients may expect to be cared for by nurses and not to be encouraged to independence, while at home the need for independence may be more evident.

The overall survival rates were similar in both groups, and early discharge did not result in greater rates of institutionalisation or hospital readmission. Although the total therapy for each discipline provided was not significantly different between the two groups for those having therapy, provision of therapy for patients with an impairment was better in the patients treated in the community, possibly reflecting better assessment and team working.

Nine patients (2.7%) were lost to follow up at 1 year, mainly because they had left the country. Completion rates for the outcome assessments were low because of a combination of cognitive and language problems in the study population but were similar to those achieved in a comparable group of patients with stroke.16 A considerable proportion of the patients (26%) were living alone at 1 year and were without a principal carer able to give information. Completion rates for the scales and follow up rates were not significantly different between the two groups and are unlikely to have affected the final results.

Comparison with other studies

Early discharge schemes have been developed for surgical patients after caesarean section,13 varicose vein and hernia operations,14 15 and myocardial infarction.16 The effects on outcome in these studies were small. Augmented support hospital discharge teams17 18 reduce readmissions to hospital of elderly patients, thereby saving hospital resources. The overall conclusion of Clarke's literature review was that with the very limited evidence reductions in length of stay have not had a major effect on health outcome,19 although the outcomes measured may not have been relevant to the intervention.

It has been argued that admission to hospital at all for patients with stroke is unnecessary except for a small proportion of patients.20 This is not the view taken by a WHO consensus meeting21 and is clearly not believed by most of the general practitioners who admitted up to 70% of their patients in London.2 The only non-randomised trial to provide intensive domiciliary rehabilitation as an alternative to hospital admission for stroke failed to reduce admission rates.22 Pound et al found that patients with stroke have important psychosocial needs during the acute stage of stroke that are often met by hospital admission.23

Conclusions

Provision of a specialist team of therapists led by a consultant provides a model of care for patients with stroke which is as effective as conventional hospital based care, with a significant reduction in hospital stay. As the numbers of patients with stroke increase with the aging population the only alternative to building more acute hospitals may be to develop alternative community based specialist care.