Home or Hospital for Stroke Rehabilitation? Results of a Randomized Controlled Trial
II: Cost Minimization Analysis at 6 Months
Background and Purpose—The goal of the present study was to examine the resource and economic implications of an early hospital discharge and home-based rehabilitation scheme for patients with acute stroke.
Methods—A cost minimization analysis in conjunction with a randomized controlled trial was carried out at 2 affiliated teaching hospitals in the southern metropolitan region of Adelaide, South Australia, between 1997 and 1998. Eighty-six hospitalized patients with acute stroke who required rehabilitation were randomized to receive both early hospital discharge and home-based rehabilitation, or conventional in-hospital rehabilitation and community care. Direct and indirect costs related to stroke rehabilitation were calculated, including hospital bed days, home-based intervention program, community services, and personal expenses during the 6 months after randomization.
Results—The mean cost per patient was lower for patients randomized to the early hospital discharge and home-based rehabilitation ($8040) compared with those who received conventional care ($10 054). This cost saving was not statistically significant (P=0.14). However, sensitivity analyses indicated that the cost of home-based rehabilitation was consistently lower than that of conventional care except when hospital costs were assumed to be 50% less than those used in the main analysis. Multiple regression analysis demonstrated that the cost of the home-based program was significantly related to a patient’s level of disability after adjustment for age, comorbidity, and the presence or absence of a caregiver.
Conclusions—The early hospital discharge and home-based rehabilitation scheme was less costly than conventional hospital care for patients with stroke. Limitation of the provision of such services to patients with mild disability is likely to be most cost effective.
In developed countries, a recent trend in health care has been the development of community-based alternatives to hospital care.1 2 Interest in the establishment of such “hospital at home” schemes stems from a new emphasis on primary and community care, on offering a greater choice to consumers, and in being an attractive solution to the rising costs of rehabilitation for patients in hospital and the pressure to maximize productivity.3 4 Although considerable heterogeneity exists within the framework of hospital-at-home schemes, broadly speaking, they can be divided into 2 groups: those that avoid admission and the inappropriate use of a hospital bed and those that achieve early hospital discharge from an acute or a planned admission and use home settings within an episode of care. The evidence to date indicates that these schemes can reduce the demand on hospital resources, but there remains uncertainty about the cost-effectiveness of hospital-at-home compared with in-hospital care.2 5 6 Most randomized trials have been small, follow-up has generally been short, and few have incorporated detailed economic evaluations.2
The provision of home-based nursing and rehabilitation services to facilitate early discharge from hospital is an attractive health care model for the management of patients with stroke.7 8 9 10 Not only is there an opportunity for a coordinated and seamless pathway of care for patients from hospital to home, but also there is the potential to reduce the risk of iatrogenic illness, distress associated with prolonged hospital stay, and travel to outpatient clinics for rehabilitation. In the United Kingdom and Sweden, such schemes have been shown to be cost effective, safe, and acceptable to patients and their families,7 8 9 10 11 but more data are required to determine the generalizability of these data within the context of other health care systems.
The present study was carried out in parallel with a pragmatic randomized controlled trial to evaluate the effectiveness and acceptability of early hospital discharge and home-based stroke rehabilitation. We concluded that the scheme does not have an adverse impact on clinical patient outcomes, although it may worsen the mental health of caregivers.12 Here, we report details of the economic consequences of a shift in the balance of care for patients with stroke.
Subjects and Methods
A comparison was made of continued care and rehabilitation in hospital with early discharge from hospital and rehabilitation at home for patients with acute stroke who had been admitted to hospital and required rehabilitation for residual disability. Because both forms of care were found to be equally effective for clinical patient outcomes,12 a cost-minimization study was the appropriate form of economic analysis. The evaluation was conducted from the perspective of both the health care system and the patients and their families. Inclusion of costs from the latter 2 viewpoints was considered important because a reduction in hospital costs could simply be offset by an increase in the costs to patients and families.
Full details of recruitment and randomization are outlined in the accompanying report.12 In brief, patients randomized to the hospital arm (n=44) received the usual care in hospital and in the community, whereas patients randomized to early discharge from hospital (n=42) received any necessary aids and adaptations in the home to facilitate early discharge and input from a community rehabilitation team. The team consisted of a full-time program coordinator, a consultant in rehabilitation, and physiotherapists, occupational therapists, social workers, speech therapists, and rehabilitation nurses, whose time was contracted by the service according to demand. Discharge from both groups occurred when the patient was deemed to have made a satisfactory recovery so he or she could manage within the existing support system and by using community services where necessary. All patients were followed up at 3 and 6 months after randomization by a research nurse, who was independent of the rehabilitation team and blind to treatment allocation. Information obtained for each patient included health-related outcomes and economic outcomes such as the use of health and community services. Economic data collected at 3 and 6 months were amalgamated, and the overall economic outcomes at 6 months are presented here.
Assessment of Costs
A summary of the types of costs and sources of costing information is provided in Table 1⇓. Data on the direct costs of health care and community services were calculated for each patient in the 6 months after randomization. Where possible, information on actual individual use of resources was used, such as the number of sessions of community services. The staff time (direct and indirect) and travel costs of the community rehabilitation team were estimated directly as they were routinely collected. However, detailed information for all patients in all aspects of care could not be recorded. An average per-patient cost was therefore used for certain health and community services, including any payments made by patients. The indirect costs incurred by caregivers were estimated on a self-report questionnaire developed for the study. Pharmaceutical costs were excluded for both groups of patients because of the difficulties in the accurate assessment. All of these costing assumptions are crucial to the final results and therefore were checked in a sensitivity analysis.
Hospital and Residential Care Costs
An average per-day cost for a rehabilitation bed was obtained from the hospital finance department based on actual expenditure for the fiscal year 1997 to 1998. This cost also included overhead costs such as management, heating, and laundry. The costs of medical care vary according to the intensity of nursing care, investigations, and treatment and are usually the most costly in the first few days in the hospital. In the present study, patients were entered into the trial only when they were considered suitable for rehabilitation (and “safe” for early discharge), and therefore the average cost per bed day in rehabilitation is also likely to be a valid estimate of actual marginal (ie, end-of-stay) acute bed day costs. Average per-episode (or per-visit) costs were used to calculate for each patient the costs associated with a readmission to an acute hospital bed and visits to outpatient medical (or rehabilitation) clinics, with data provided from national costs data for the fiscal year 1997 to 1998. These data were augmented with the assignment of the cost of “urgent” and “nonurgent” transport by ambulance over 10 km for each readmission and outpatient visit, respectively.
Early Discharge and Home-Based Rehabilitation Program
Individual patient records were used to determine the costs incurred by the home-based rehabilitation team while treating patients randomized to this scheme. This included time spent with the patient plus any additional time (travel, administration, and so on) attributable to that particular case. An average cost was assigned to each patient for those costs not directly attributable to individual patients, such as time spent by staff in multidisciplinary meetings, plus nonstaff administration and overhead costs (stationery, telephone, vehicle maintenance, and so on). The standard mileage rate for health services at the time of the study ($Aust 0.51/ km) was used to place a value on all travel for the team. Market prices were used to place a value on all aids and adaptations. For the coordinator costs, research time (eg, randomization and consent) was excluded. A proportion of the coordinator’s time and associated costs (20%) was allocated to the conventional care group based on the assumption that in any such service, time would be spent by the coordinator in screening patients for early discharge and home-based rehabilitation.
Community services that were examined included visits to the general practitioner, community therapy services that were outside of the experimental program (eg, physiotherapy, occupational therapy, dietitian), alternative therapy services (eg, chiropractor, naturopathologist), district nursing, attendance at community day centers, Meals on Wheels, and admission to hostel or nursing home facilities for either short-term respite or permanent care. The use of these services was recorded for each patient on a per-visit or per-bed-day basis. Assumed costs were applied to items for which it was not possible to obtain accurate information.
Other Health Patient and Family Costs
Patients were asked whether they had made any changes to their mode of transport, if any additional modifications had been made to their home, or if they had required any special equipment after the stroke, but these costs were not estimated. Similarly, caregivers were asked whether they had taken time off work to provide care, but no attempt was made to measure or to value these costs because it was assumed that these would be absorbed within employment contracts. However, caregivers were also asked to estimate what proportion of their time during the preceding 24 hours was spent providing help or supervision in everyday activities for the patient. With the assumption that caregivers were crucial in the prevention of the admission of patients to residential care, an estimate was made for each patient of the cost of informal care based on the cost of hostel-level residential care.
The inclusion of an average per-day cost in the valuation of hospital care may overstate the actual hospital costs. To approximate the marginal (ie, end-of-stay) costs and therefore potential savings from the release of hospital beds, sensitivity analysis assumed that resources released would be either 75% or 50% of the average per-day cost. Because the home-based rehabilitation program was a service with input from predominantly part-time staff who were contracted from existing services according to demand, the costs of staff, travel, and overhead might prove to be higher for a stand-alone specialty service. Therefore, for sensitivity analysis, alternative increased costs of 25% and 50% for the home-based rehabilitation team were assumed as well as a reduced cost (75% of baseline) based on the assumption that a new service would become more efficient and cost effective as it developed over time. Finally, the effect on costs according to patient level of disability was also determined.
The sample size was determined specifically for clinical rather than economic evaluations.12 Because the resource use distributions were positively skewed, median and interquartile range values are reported. The cost data distributions were also positively skewed. However, the provision of information about median costs is not helpful for service providers who require estimates of costs per patient. Therefore, mean costs were calculated and compared with the use of independent sample t tests, and mean differences were expressed with 95% CIs. For such skewed data, however, care must be exercised in the interpretation of SDs. Categorical variables were compared with use of the χ2 test. Multiple regression analysis was undertaken to explore the relationship between the dependent variable cost and baseline demographic and other independent variables (age, level of disability, living arrangements, and comorbidity) for the early hospital discharge and home-based rehabilitation program. All analyses were performed with SPSS for Windows version 6.0 and Arcus Quickstat (Longman Software Publishing)..
All 86 patients who participated in the randomized controlled trial were included in the economic evaluation. Table 2⇓ shows the time and costs of the different services provided by the home-based rehabilitation team and costs related to travel, equipment, and modifications to the homes of patients. The total cost of the home-based intervention program was $128 853, and the mean cost per patient was estimated to be $3068.
Table 3⇓ presents estimates of the median use of resources by each group. Apart from a reduced length of initial hospital stay and input of the home-based rehabilitation team for the experimental group, there were no significant differences in community services and other resource use between the 2 groups.
Table 4⇓ shows the results of combining resource use with information on valuation in terms of mean cost per patient across the major health service domains. The cost of hospital care was greatly reduced in the experimental group, but this was counterbalanced by additional direct and indirect costs associated with home-based rehabilitation. The mean overall cost difference between intervention groups was $2013, or approximately one fifth of the cost of conventional rehabilitation for patients with stroke. This difference, however, was not statistically significant.
Multiple regression analysis demonstrated that functional status, as defined with the modified Barthel Index (BI),13 was an independent predictor of cost in the home-based intervention program. When patients were grouped according to “mild” (scores 91 to 100) or “moderate” (scores 50 to 90) grades of disability, those in the former category had lower costs after adjustment for age, comorbidity (history of hypertension, angina, heart failure, diabetes, or arthritis), and presence or absence of a caregiver (regression coefficient −1446 [95% CI −2417 to −476], SE 479, P=0.005).
Table 5⇓ shows the results of the sensitivity analysis. The results obtained are robust to changes in the costing data and assumptions. Only when the costs of care in hospital were assumed to be 50% of baseline did the conventional care scheme become cheaper. Limitation of the between-group analysis to patients with mild disability (BI scores 91 to 100) showed the home-based program to be more cost effective than conventional care.
This economic analysis of early hospital discharge and home-based rehabilitation for patients with stroke found the cost of rehabilitation in the home to be less than the cost of conventional in-hospital rehabilitation and care in the community during the 6 months after randomization. The mean reduction in overall cost for patients allocated to the home-based rehabilitation scheme was ≈Aust$2000, or one fifth of the cost of conventional care and rehabilitation. Although the overall cost savings were not statistically significant, this clinically meaningful effect was mainly a result of a significant reduction in the initial length of stay in the hospital and the consequent reduction in hospital costs.
Hospital-at-home schemes involve a transfer of care from the secondary to primary sector and thus have the great potential to increase the use of community services and shift the costs of care onto family members and other informal caregivers. This is particularly relevant to rehabilitation services aimed at older patients with disabling conditions such as stroke. A strength of the present study compared with other studies2 8 9 is the inclusion of detailed analyses of the cost of community care. No significant difference was found in the use of routine community services between the 2 groups, and the associated costs for these services were less in the home-based rehabilitation group. This may reflect a better adjustment to residual disability on the part of patients who received rehabilitation in the home and, therefore, have less need for support from community services at the completion of the program.
Because more than half of patients indicated that they had a caregiver at the time of the stroke, it was important to measure the cost of informal care for patients after their discharge from hospital. On the assumption that the cost of caregiver time for patients in the home was equivalent to the cost of residential care at the hostel level for persons with a mild to moderate level of physical disability, the cost of informal care was higher for patients in the home-based rehabilitation scheme. However, this may simply reflect an early discharge from hospital and a longer period in the community for this group rather than an actual increase in costs for caregivers. Even so, taken together with the finding of worse mental health on the part of caregivers as reported in the accompanying report,12 the potential for an adverse impact of home-based schemes is clearly important and must be considered in the development of these services.
The early hospital discharge and home-based rehabilitation scheme was a new service that underwent development during the study. In particular, much of the coordinator’s time was spent in publicity and in screening and recruiting patients. The cost per patient seen in this study would, therefore, be expected to reduce over time and is likely to be less for an established scheme. The crucial sensitivity analyses indicate that the finding of a cost advantage for home-based rehabilitation over conventional in-hospital rehabilitation is robust. The cost of the new scheme was also strongly related to the patient’s level of physical disability, which is at variance with the findings of McNamee et al.8 However, this may relate to differences between the studies in the set-up costs and cost estimates for the services. In the present study, therapists were contracted to provide care based on need and data were collected prospectively, whereas the study of McNamee et al included full-time staff as part of a stand-alone service, and the costs for hospital and therapy were estimated retrospectively. No such analyses were undertaken in the study of Beech et al.10
In conclusion, early hospital discharge and home-based rehabilitation in our health care system proved to be a less costly alternative to conventional care and rehabilitation for patients with stroke. Although the reduction in cost was not statistically significant, the economic results indicate that an established scheme could play an important role in the release of hospital beds and be a cost-effective approach to the rehabilitation of patients with stroke.
We are grateful to Sue McKechnie and the Ambulatory Care Unit of the South Australian Health Commission, who supported this project through a grant from the federal government. We are indebted to the following members of the community rehabilitation team for their dedication: Robyn Lister and Nicki Hayball (consecutive project coordinators), Heather Jesshope, Jan Van Emden, Penny Jacomos, Naomi Brill, Marion Reece, John Stanfield, Robyn Burnett, Belinda Magor, Edwina Reid, Martine Ledger, Colleen Gleeson, Helen Reid, Marline Reece, Margi Smart, Donna Lawrence, Claire Morris, and Beth Stronach. We thank Sara Laubscher and Kerry Clifford for their support in the study and Derrick Bennett and Paul Brown of the University of Auckland for their advice on statistical and economical analyses. We acknowledge the support of M. Bennett, A. Crockett, A. Darzins, B. Dodd, P. Finucane, J. Harvey, E. Hobbin, P. Lavelle, D. Law, M. McTab, M. Russell, H. Trenorden, L. White, and B. Wilson and our colleagues and staff at the Flinders Medical Center, Repatriation General Hospital, and Southern Domiciliary Care and Rehabilitation Service, South Australia.
- Received November 9, 1999.
- Revision received February 3, 2000.
- Accepted February 24, 2000.
- Copyright © 2000 by American Heart Association
Iliffe S. Hospital at home: an uncertain future. BMJ. 1996;312:923–924.
Iliffe S. Hospital at home: from red to amber? BMJ. 1998;316:1761–1762.
Marks L. Home and Hospital Care: Redrawing the Boundaries. London, UK: King’s Fund; 1990.
Lafferty G. Community-based alternatives to hospital rehabilitation services: a review of the evidence and suggestions for approaching future evaluations. Rev Clin Gerontol. 1996;6:183–194.
Shepperd S, Iliffe S. Hospital at home compared with in-patient care. In: Bero L, Grilli R, Grimshaw J, Oxman A, eds. The Cochrane library. Oxford, UK: Update Software; The Cochrane Collaboration, issue 1, 1998.
Coast J, Richards SH, Peters TJ, Gunnell DJ, Darlow M, Pounsford J. Hospital at home or acute hospital care? A cost minimisation analysis. BMJ. 1998;316:1802–1806.
Rodgers J, Soutter J, Kaiser W, Pearson P, Dobson R, Skilbeck C, Bond J. Early supported hospital discharge following acute stroke: pilot study results. Clin Rehabil. 1997;11:280–287.
McNamee P, Christensen J, Soutter J, Rodgers H, Craig N, Pearson P, Bond J. Cost analysis of early supported hospital discharge for stroke. Age Ageing. 1988;27:345–351.
Rudd AG, Wolfe CD, Tilling K, Beech R. Randomised controlled trial to evaluate early discharge scheme for patients with stroke. BMJ. 1997;315:1039–1044.
Beech R, Rudd AG, Tilling K, Wolfe CDA. Economic consequences of early inpatient discharge to community-based rehabilitation for stroke in an inner-London teaching hospital. Stroke. 1999;30:729–735.
Widén Holmqvist L, von Koch L, Kostulas V. Holm M, Widsell G, Tegler H, Johansson K, Almazan J, de Pedro-Cuesta J. A randomized controlled trial of rehabilitation at home after stroke in southwest Stockholm. Stroke. 1998;29:591–597.
Anderson C, Rubenach S, Ni Mhurchu C, Clark M, Spencer C, Winsor A. Home or hospital for stroke rehabilitation? Results of a randomized controlled trial, I: health outcomes at 6 months. Stroke. 2000:31:1024–1031.
Shah S, Vanclay F, Cooper B. Predicting discharge status at commencement of stroke rehabilitation. Stroke. 1989;20:766–769.