Economic Consequences of Early Inpatient Discharge to Community-Based Rehabilitation for Stroke in an Inner-London Teaching Hospital
Background and Purpose—In an inner-London teaching hospital, a randomized trial of “conventional” care versus early discharge to community-based therapy found no significant differences in clinical outcomes between patient groups. This report examines the economic consequences of the alternative strategies.
Methods—One hundred sixty-seven patients received the early discharge package, and 164 received conventional care. Patient utilization of health and social services was recorded over a 12-month period, and cost was determined using data from provider departments and other published sources.
Results—Inpatient stay after randomization was 12 days (intervention group) versus 18 days (controls) (P=0.0001). Average units of therapy per patient were as follows: physiotherapy, 22.4 (early discharge) versus 15.0 (conventional) (P=0.0006); occupational therapy, 29.0 versus 23.8 (P=0.002); speech therapy, 13.7 versus 5.8 (P=0.0001). The early discharge group had more annual hospital physician contacts (P=0.015) and general practitioner clinic visits (P=0.019) but fewer incidences of day hospital attendance (P=0.04). Other differences in utilization were nonsignificant. Average annual costs per patient were £6800 (early discharge) and £7432 (conventional). The early discharge group had lower inpatient costs per patient (£4862 [71% of total cost] versus £6343 [85%] for controls) but higher non-inpatient costs (£1938 [29%] versus £1089 [15%]). Further analysis demonstrated that early discharge is unlikely to lead to financial savings; its main benefit is to release capacity for an expansion in stroke caseload.
Conclusions—Overall results of this trial indicate that early discharge to community rehabilitation for stroke is cost-effective. It may provide a means of addressing the predicted increase in need for stroke care within existing hospital capacity.
Throughout Europe, stroke is a major cause of mortality and morbidity, and care for stroke consumes a substantial amount of healthcare resources and costs.1 2 3 4 The majority of these costs and resources are related to hospital inpatient services, with an estimated 80% of stroke patients being admitted to the hospital in inner-city districts.5 Once admitted, a study of 9 hospitals in 6 European states demonstrated that stroke patients have long lengths of stay relative to other acute conditions.6
It has been predicted that the effects of an aging population will cause real expenditures on strokes to increase. For England and Wales, this increase has been estimated as 30% over the period 1983–2023.7 Similar rises in spending have been projected for the Netherlands, with a 30% increase during 1992–2010,8 and for Austria, with a 25% increase during 1992–2010.9
The current and projected increases in the costs of stroke care mean that healthcare commissioners and providers will want to identify innovative methods of delivering care that reduce costs per case while maintaining health outcomes. It has been argued that more emphasis should be given to community-based care for stroke.10 This is in keeping with a broader healthcare strategy that is now seen in Europe of shifting the balance of health care away from the secondary and toward the primary care sector.11 12 13 However, although a shift in the balance of care is being advocated, there is a shortage of evidence about the effectiveness and, in particular, the cost-effectiveness of initiatives that bring about such a shift.14
This report details the economic consequences of a shift in the balance of care for stroke: early acute hospital discharge to a package of home-based physiotherapy, occupational therapy, and speech therapy. This strategy was compared with “conventional” inpatient and community-based care in a randomized controlled trial undertaken in an inner-London teaching hospital, St Thomas’s Hospital. The intervention, methods of case ascertainment and randomization, and clinical results of the trial have been fully reported elsewhere.15 They are briefly summarized here.
Patients were randomized when they were medically stable and able to transfer independently or with the aid of a caregiver. Patients in the early discharge cohort were then eligible for home-based therapy. This consisted of a planned program of care available up to a period of 3 months. Patient progress was reviewed at weekly meetings. The required frequency of patient visits was determined by the therapists, although patients could have no more than 1 visit per day from each type of therapist.
In this group of patients, prompt discharge from inpatient care was facilitated by the patients having rapid access to aids and adaptations to support home-based care. A supply of high chairs, commodes, and toilet frames was immediately available, as was a person who could undertake any necessary modifications to a patient’s home.
After randomization, patients in the control group continued with the conventional program of inpatient-based care and therapy. Conventional outpatient services included a hospital-based stroke clinic, a geriatric day hospital, and generic therapy services, provided either in a patient’s home or in hospital outpatient clinics.
More precise details of the early discharge and control strategies for delivering care for stroke are provided in Results (Tables 1 to 3⇓⇓⇓). These results compare the 2 patient cohorts in terms of the amounts of care they received from health and social services.
For each patient group, demographic data and data on severity of stroke were collected within 48 hours of randomization. Outcomes were assessed at 2, 4, 6, and 12 months after randomization in terms of impairment (Motricity Index, Mini-Mental State Examination, Frenchay aphasia screening test); disability (modified Barthel Index score, Rivermead Activities of Daily Living score, Hospital Anxiety and Depression score, 5-meter timed walk); general health status (Nottingham Health Profile); caregiver stress (Caregiver Strain Index); and patient and caregiver satisfaction (Pound questionnaires).
A total of 167 patients were randomized to the early discharge strategy and 164 to conventional care. These represented ≈45% of all stroke admissions during the period of the study and ≈60% of those who did not die in the hospital.
The early discharge cohort had a mean and median age of 70 years (controls, 72 and 73 years, respectively); 55% were males (controls, 57%); 74% were white, 16% black Caribbean, and 10% black African or other (controls, 74%, 17%, and 9%, respectively). Before their stroke, 31% had lived alone, 62% with another person, and 7% in an institution (controls, 38%, 53%, and 9%, respectively).
Stroke type for the early discharge group was 1% subarachnoid hemorrhage, 95% infarction, and 4% unknown (controls, 1%, 95%, and 4%, respectively). At the time of randomization, 42% of the early discharge group had a Barthel score of 0 to 14, 47% of 15 to 19, and 10% of 20 (controls, 40%, 48%, and 12%, respectively); 42% had dysphasia (controls, 35%); and 29% were incontinent (controls, 29%). There were no statistical differences (P<=0.05) between the 2 patient groups in terms of their characteristics at baseline and at the time of randomization.
Analysis of outcomes focused on differences at 12 months after randomization. Subjects in the early discharge group were shown to be more satisfied with their hospital care, with this difference being significant (P=0.032). All other outcome differences were nonsignificant, leading to the general conclusion that the early discharge package of care had a neutral effect on health outcomes.
The analysis and results presented in this report first compare the 2 strategies in terms of their impacts on patient utilization of care provided by health and social services from the point of hospital admission up to the year after randomization. Next, the cost implications of these patterns of utilization are assessed to determine the cost-effectiveness of the early discharge scheme.
Subjects and Methods
Patient Characteristics and Utilization of Services
Data on patient demographic characteristics and stroke severity, including presence of motor deficit or speech or swallowing problems, were extracted from case notes before randomization. Inpatient acute care data collected included patient stay by type of ward, both before and after randomization, and patient use of diagnostic tests (Table 1⇑).
Standardized forms, completed by physiotherapists, occupational therapists, and speech therapists, were used to prospectively collect data on patient use of inpatient and community-based rehabilitation. Therapists recorded the number of units of therapy received by patients. One unit of therapy was equal to 20 minutes of the time of 1 trained therapist. Hence, if a patient had 20 minutes of therapy with 2 therapists, the patient would have consumed 2 units of therapy.
For occupational and speech therapy services, the number of units of direct and indirect time was recorded. Direct time referred to actual contact time with patients, and indirect time referred to other noncontact activities linked to the care of individual patients, such as the completion of documentation about patients and, in the case of community-based care, time associated with traveling to patients’ homes. For physiotherapy only direct time was recorded. On the basis of the advice of the departmental head, indirect time was assumed to be 15% of direct time.
Data on patient contacts with hospital physicians, general practitioners (GPs), and social services were obtained with the use of questionnaires administered to patients and caregivers during follow-up at 2, 4, 6, and 12 months after randomization (for the items covered, see Table 3⇑). These time points reflected those used for the collection of data about patient outcomes in the clinical component of the trial. For hospital physician and GP services, individuals were asked about the number of contacts they had had during the period since their last follow-up interview. For other services, individuals were asked the number of times per week they currently received the service.
Costs of Services
The vast majority of stroke patients at the study hospital were treated on care of the elderly wards and by care of the elderly physicians. Nurse staffing levels on these wards were obtained from departmental heads, and the corresponding employment costs were obtained from the local finance department. A nurse staffing unit cost per day was then derived. Data on care of the elderly medical staff costs per day were obtained from the hospital finance department, as were unit costs for the diagnostic tests used by patients.
The early discharge team consisted of a full-time senior grade-1 physiotherapist and therapy aide, who jointly provided a unit of physiotherapy, a senior grade-1 occupational therapist, and a half-time grade-2 speech and language therapist. The average staffing establishment per unit of inpatient physiotherapy was 0.17 whole-time equivalent senior grade-1 therapists, 0.39 whole-time equivalent senior grade-2 therapists, and 0.44 basic therapists. An inpatient unit of occupational therapy was provided by a senior grade-1 therapist and that of speech therapy by a grade-2 therapist. Data on the costs of employing such staff were obtained from the local finance department, and the relevant costs per unit of therapist time were derived.
Data on the unit costs per outpatient physician contact were obtained from the local finance department. Unit costs per GP contact and per contact with other community-based health and social services were extracted from a publication by Netton and Dennett.16 This publication stems from a project that derives national statistics for England and Wales.
Staff costs were based on the midpoints of the appropriate pay scales, including any employee-related costs (eg, benefits packages) but excluding London weighting. All costs included an allowance for general overhead. For hospital staff costs, this allowance was taken as 69% of direct costs, the current percentage for the care of the elderly specialty in the study hospital. For the community-based therapy team, the allowance was taken as 23% of staff costs, in accordance with the rate quoted by Netton and Dennett.16 Costs were at 1997 prices.
For utilization data, statistical differences in categorical variables were examined with the χ2 test, and differences in continuous variables were examined with the Mann-Whitney test. Analysis of patient use of community-based nonclinical services (Table 3⇑) focused on individuals who were not receiving these services before their stroke. This allowed changes in an individual’s circumstances that might be linked to the stroke to be identified.
Overall annual average costs for each cohort of patients were derived by combining unit cost data with the data on annual patient utilization of services. When the multiplier was days of care, the overall stay of patients was included. One acute test or procedure was assumed for patients who had had an investigation. Cost totals were then divided by the number of patients in each arm of the trial to derive the average cost per patient.
The robustness of cost findings was explored with the use of sensitivity analysis. This focused on changes that were considered plausible and variables for which uncertainty might exist regarding their value if the early discharge strategy was introduced elsewhere.
Utilization of Services
A total of 167 patients were assigned to the early discharge arm and 164 to conventional care. After randomization there was a 6-day difference between groups in mean stay per patient (P=0.0001). The 2 patient groups were similar in their use of major diagnostic tests (Table 1⇑).
For each type of rehabilitation therapy, a similar percentage of patients with an identified motor deficit or speech or swallowing problem obtained access to inpatient rehabilitation, but a much larger percentage of such patients in the early discharge group received non-inpatient rehabilitation. For the total patient cohorts, the early discharge group received more units of rehabilitation: physiotherapy, 22.4 (early discharge) versus 15.0 (conventional) (P=0.0006); occupational therapy, 29.0 versus 23.8 (P=0.002); speech therapy, 13.7 versus 5.8 (P=0.0001) (Table 2⇑).
For outpatient and community-based clinical services, the early discharge group had more contacts during the year than the conventional group (Table 3⇑). The differences in annual physician contacts (P=0.015) and GP clinic visits (P=0.019) were statistically significant.
Average utilization per patient of Meals on Wheels, home help, and district nursing services was greater for the early discharge group. The conventional group utilized more lunch club and day hospital services (P=0.04). Patient contacts during the year with psychiatric nurse and respite care services were negligible and are not reported.
Annual Average Costs of Services
The results in Table 4⇓ combine data on patient utilization and on unit costs per item of service to derive total costs for the period from hospital admission up to 12 months after randomization. Annual costs of inpatient care for the early discharge cohort were £811 984 (£4862 per patient) and for the conventional cohort £1 040 276 (£6343 per patient). Costs of non-inpatient care were £323 625 (£1938 per patient) and £178 526 (£1089 per patient), respectively.
Across the complete continuum of care, the average annual costs for the early discharge cohort were £1 135 609 (£6800 per patient) and for the conventional cohort £1 218 802 (£7432 per patient). Average costs per patient were therefore ≈8% lower for the early discharge cohort.
Sensitivity Analysis of Costs
The assumed overhead rate of 69% reflects the cost structure of an inner-London teaching hospital. Research indicates that costs are 22% higher in London than elsewhere.17 An alternative overhead rate of 56% was therefore assumed. The revised annual costs of the early discharge strategy were £1 079 755 (£6466 per patient) and of the conventional strategy £1 143 733 (£6974 per patient), which is a 7% difference in average costs per patient. For average costs per patient to be the same, the allowance for hospital overhead would have to be only 5%.
The postrandomization average stay of patients in the early discharge group was 12 days despite patients being medically stable and able to transfer at randomization. First, it was assumed that this stay could be further reduced by 6 days (12 days fewer than conventional) and second, it was assumed that patients could be discharged at the point of randomization (18 days fewer than conventional).
The annual average costs for the early discharge cohort under the first scenario fell to £1 011 167 (£6055 per patient and 19% less than conventional care [£7432 per patient]) and under the second scenario to £886 724 (£5310 per patient, 29% less than conventional care). The cost areas that were assumed to be affected by these changes were inpatient nursing and clinical staff costs, the overhead of these costs, and ward consumables; rehabilitation costs were assumed to be unchanged.
The prerandomization stay of the conventional group was 2 days more than that of the early discharge group. This chance effect affects the calculated cost differences of the 2 strategies. It was estimated that costs per patient for the 2 groups would be the same when the difference in overall stay between groups was 3 days. Hence, the prerandomization difference did not affect the conclusion that the early discharge cohort had lower average costs per patient.
Incremental cost analysis considered short-term changes in financial expenditure. If an annual eligible caseload of 167 strokes and a bed occupancy of 90% (representative figures for the study hospital) are assumed, a 6-day reduction in postrandomization average stay would free 3 beds. If the bed reductions were implemented, nurse and clinical staffing levels are likely to remain unchanged, with the only financial savings being in ward consumable items (£10 190 per annum [6×167×£10.17]). The extent to which this relatively small financial cost saving would be offset by increased financial costs would depend on (1) whether the community rehabilitation team was introduced by reorganizing existing staffing establishments and (2) whether the increased demands on community health and social services could be absorbed within existing staffing structures.
“Opportunity costs” are created by the release of inpatient bed capacity. Given the predicted rise of 30% during 1983–2023 in the level of resources needed for stroke,7 it was assumed that the beds released by the early discharge policy would be needed for an expansion in inpatient caseload for stroke.
Again an annual eligible caseload of 167 strokes was assumed. It was also assumed that the average inpatient stay of these strokes would be 34 days, the average for the early discharge cohort of trial patients. Given these assumptions, the 3 beds released by the early discharge policy would create capacity for an extra 29 strokes per year, a 17% rise in activity levels. A release of 9 beds (a reduced stay of 18 days, leading to an average overall stay of 22 days) would create capacity for an extra 134 strokes per year, an 80% rise in activity levels. Given the existence of fixed costs referred to in the incremental cost analysis, many of the financial effects of this rise in caseload are likely to be absorbed within existing resources.
A current strategic emphasis of healthcare delivery is to shift the balance toward primary care and away from secondary care. However, a review by Godber et al14 found a shortage of evidence about the effectiveness and cost-effectiveness of initiatives that aim to achieve this. For 1 condition, the results presented here contribute to this debate by assessing the merits of an integrated policy of early discharge from inpatient care to home-based rehabilitation for stroke.
Previous studies of stroke care have compared the effectiveness of hospital- and home-based rehabilitation.18 19 20 A combined analysis of the results of these 2 trials suggested that there was “little to choose” between the 2 modes of delivery, findings in keeping with the clinical results of the trial presented here.21 The study by Young and Forster22 also compared the cost-effectiveness of rehabilitation at home versus that given in a day hospital. Results suggested that home treatment was the preferred option.
A search by the authors also found examples of early discharge strategies being used for other conditions.23 24 25 26 However, only 1 similar economic study concerning an integrated policy of early discharge and home-based rehabilitation for stroke was identified.27 That report, based on a Swedish population, gave results of a pilot study covering only 15 patients. The results suggested that early discharge to home-based rehabilitation may be less expensive.
The results presented in this report build on the knowledge provided by these previous studies. For a period up to 1 year after randomization, a detailed comparison has been given of the health and social service costs of early inpatient discharge to community-based rehabilitation versus conventional inpatient and community-based care. The results indicated that the early discharge strategy had lower average costs per patient, particularly if further reductions in stay were feasible.
When set alongside the previously reported clinical implications of the intervention,15 these results indicate that the early discharge strategy is cost-effective, with the same outcomes achieved at lower average costs. This finding was shown to be robust to large changes in the assumed allowance for hospital overheads and in the actual impact of the intervention strategy on average patient stay.
Two factors affect the generalizability of the cost results: (1) the generalizability of the cost data used in the analysis and (2) the generalizability of the data relating to patient utilization of services.
The total average cost of inpatient care for the conventional group of patients was £31 040 276. This translates into a cost of £151 per inpatient day [£31 040 276/(164×42)]. Netton and Dennett16 quote a figure of £118 for the costs of an acute inpatient day within elderly care. These 2 cost figures are of a similar magnitude when allowance is made for the higher costs of London (≈22% more than elsewhere within the National Health Service17 ).
The total average costs of hospital-based non-inpatient care for the conventional cohort was £38 187 or ≈£160 per outpatient attendance (annual attendance of 239, extracted from Table 3⇑). The figure for an acute outpatient elderly care attendance quoted by Netton and Dennett16 is £91. This indicates that the costs per outpatient attendance used in this analysis are somewhat higher than the national average, even allowing for the higher costs of London. However, outpatient costs only accounted for ≈3% of the estimated annual costs of stroke care.
All other cost data used in the analysis were extracted from the publication by Netton and Dennett.16 These represent national averages in terms of cost.
Hence, after we allowed for the higher costs of London, the majority of the cost data used in the analysis appear to be representative of costs elsewhere in England and Wales. Beyond these countries, the cost data may not be representative. To help others to interpret the importance of these potential differences, the results exposed the ways in which data on patient utilization of services and the unit costs of these services had been combined to generate both total costs per cohort and average costs per patient. Others could substitute their local unit cost.
The second factor affecting the generalizability of the results relates to the impact of the early discharge strategy on the utilization of health and social services by patients. The precise magnitudes of these changes will vary from locality to locality and will be influenced by factors such as the existing supply of health- and community-based services, in other words, by what constitutes conventional care.
Within England and Wales, the types of services covered by the results are likely to exist, but overall levels of availability will vary. Hence, the services that are likely to be generalizable, in the opinion of the authors, are the specific services that would be affected by the intervention and control strategies and the relative contributions of these services to patient care. Beyond England and Wales, some of the services covered by the results may not exist.6
Potential adopters of the early discharge strategy may also want to consider additional factors that were not covered by the results presented here. One such issue relates to any differences between strategies in the costs of readmissions to hospital. However, the inclusion of these costs would not have affected the overall conclusions of this report, since patient readmission levels during the year were the same in both patient groups: 26% of the total cohort.15
An additional issue surrounds the relative impact of early discharge on any costs incurred by patients and informal caregivers, such as a patient’s relatives and friends. The authors recognize the importance of such costs28 and attempted to compare strategies in terms of their impacts on expenses incurred, earnings lost, and time away from normal activities by patients and caregivers. However, there were concerns about the quality of these data, and therefore they were not reported.
What, then, according to the results presented in this report, are the main benefits that would stem from implementation of the early discharge strategy? Although it has lower average costs per patient, incremental cost analysis indicated that it would lead to relatively small financial savings. Although crude, this analysis suggests that early discharge should not be perceived as a means of generating financial savings.
Probably its major benefit is to release beds to increase caseload. Demographic changes suggest that this capacity might be needed to cope with potential increases in demands for stroke care.7 The policy also results in increased access to rehabilitation for patients with an identified deficit, although this increase in access was not translated into a net improvement in the clinical outcomes covered by this project.
In conclusion, the overall results of this trial indicate that early discharge to community rehabilitation for stroke is both feasible and cost-effective. Although this release of resources is unlikely to result in financial savings, it does release capacity to treat more patients. Hence, the strategy may provide a means of addressing the predicted increase in need for stroke care within existing hospital capacity.
This study was supported by the Stroke Association; Lambeth, Southwark, and Lewisham Health Authority; the Special Trustees of St Thomas’s Hospital; the Nuffield Provincial Hospitals Trust; and Wandsworth Health Gain Fund. The members of the St Thomas’s Stroke Trial Research Group are G. Browning, F. Bunn, H. Campbell, T. Colella, V. Churchouse, V. James, S. Mackay, A. Moberly, P. Pound, E. Richardson, P. Trail, Y. Toun, C. Varden, F. Warburton, and C. Watson. We acknowledge the help and support of our colleagues at St Thomas’s Hospital and King’s College Hospital and Dr M. Andlaw, general practitioner.
- Received October 5, 1998.
- Revision received January 26, 1999.
- Accepted January 26, 1999.
- Copyright © 1999 by American Heart Association
Annual World Health Statistics. Geneva, Switzerland: World Health Organization; 1990, 1991, 1992.
Office of Health Economics. Stroke. London, England: Office of Health Economics; 1988.
Isard PA, Forbes JF. The cost of stroke to the National Health Service in Scotland. Cerebrovasc Dis. 1992;2:47–50.
Wolfe CDA, Taub N, Woodrow E, Richardson E, Warburton FG, Burney PGJ. Patterns of acute stroke care in three districts of southern England. J Epidemiol Community Health. 1993;47:144–148.
Beech R, Ratcliffe M, Tilling K, Wolfe CDA. Hospital services for stroke care: a European perspective. Stroke. 1996;27:1958–1964.
Malmgren R, Bamford J, Warlow C, Sandercock P, Slattery J. Projecting the number of first-ever strokes and patients newly handicapped by stroke in England and Wales. BMJ. 1989;289:656–660.
Bergman L, van der Meulen JHP, Limburg M, Halberma TDF. Costs of medical care after first-ever stroke in the Netherlands. Stroke. 1995;26:1830–1836.
Haidinger G, Waldhoer T, Tuomilehto J, Vutuc C. Assessment of costs related to hospitalisation of stroke patients in Austria for 1992 and prospective costs for the year 2010. Cerebrovasc Dis. 1997;7:163–167.
Young J. Is stroke better managed in the community? BMJ. 1994;309:1356–1358.
NHS Executive. Developing NHS Purchasing and GP Fund-Holding: Towards a Primary Care-Led NHS. Leeds, England: National Health Service Executive; 1996. EL (94) 79.
Ham C, Robinson R, Benzeval M. Health Check: Health Care Reforms in an International Context. London, England: King’s Fund Institute; 1990.
Department of Health. The New NHS: Modern, Dependable. London, England: The Stationery Office; 1997. Command paper CM3807.
Rudd AG, Wolfe CDA, Tilling K, Beech R. Randomised controlled trial to evaluate early discharge scheme for patients with stroke. BMJ. 1997;315:1039–1044.
Netton A, Dennett J. Unit Costs of Health and Social Care. Canterbury, England: Personal Social Services Research Unit, University of Kent at Canterbury; 1997.
Netton A. Unit Costs of Community Care. Canterbury, England: Personal Social Services Research Unit, University of Kent at Canterbury; 1994.
Gladman JRF, Lincoln NB, Barer DH. A randomised controlled trial of domiciliary and hospital-based rehabilitation for stroke patients after discharge from hospital. J Neourol Neurosurg Psychiatry. 1993;56:960–966.
Gladman JRF, Lincoln NB. Follow-up of a controlled trial of domiciliary stroke rehabilitation (DOMINO study). Age Ageing. 1994;23:9–13.
Young JB, Forster A. The Bradford Community Stroke Trial: results at six months. BMJ. 1992;304:1085–1089.
Gladman J, Forster A, Young J. Hospital- and home-based rehabilitation after discharge from hospital for stroke patients: analysis of two trials. Age Ageing. 1995;24:49–53.
Adler MW, Waller JJ, Creese A, Thorne SC. Randomised controlled trial of early discharge for inguinal hernia and varicose veins. J Epidemiol Community Health. 1978;32:136–142.
Beech R. Economic evaluation of stroke care. In: Wolfe CDA, Rudd A, Beech R, eds. Stroke Services and Research: An Overview With Recommendations for Future Research. London, England: The Stroke Association; 1996:281–299.