Cost analysis of early supported hospital discharge for stroke

Transcription

1 Age and Ageing 1998; 27: Cost analysis of early supported hospital discharge for stroke PAUL MCNAMEE, JAKOB CHRISTENSEN, JENNIFER SOUTTER, HELEN RODGERS, NEIL CRAIG 1, PAULINE PEARSON 2, JOHN BOND Centre for Health Services Research, 21 Garemont Place, School of Health Sciences, University of Newcastle upon Tyne, Newcastle upon Tyne NE2 4AA, UK 'Ayrshire and Arran Health Board, Seafield House, Doonfoot Road, Ayr KA7 4DW, UK Department of Primary Health Care, The Medical School, School of Health Sciences, University of Newcastle upon Tyne, Newcastle upon Tyne NE2 4HH, UK Address correspondence to: R McNamee. Fax: (+44) Abstract Objective: to measure the net costs to the health and personal social services of an early supported discharge policy for stroke. Design and setting: cost analysis, using data from a pragmatic randomized controlled trial conducted in three hospitals in Newcastle upon Tyne, UK. Subjects: 92 people admitted with acute stroke within 72 h of onset from their own homes with no co-morbidity likely to affect rehabilitation. Main outcome measures: health and personal social service costs. Results: early supported discharge reduced median length of hospital by almost half (14 days vs 26 days, P = 0.02). The costs of the service were 7155 per patient compared with.7480 for conventional hospital care. Sensitivity analysis demonstrated that this result was robust to changes in values of bed days and exclusion of particular resource use items. Sub-group analysis suggested that costs were related to physical dependency. Conclusions: early supported discharge provided a cost-effective alternative in the management of stroke care. However, a larger study is required to assess the generalisability of the results and long-term cost effectiveness. Keywords: cost analysis, earty supported hospital discharge, stroke Introduction Stroke is one of the major causes of mortality and morbidity in older people and places a burden on scarce health care resources [1, 2]. As die incidence of stroke increases with age [3] and leaves many people with serious disability [4, 5], there is growing pressure to ensure diat stroke management is cost effective [6]. Recent studies indicate that the way hospital rehabilitation is organized can affect mortality and morbidity [7, 8]. However, there is uncertainty over how services should be best provided after discharge. While previous studies indicate no differences in health outcomes between different types of post-discharge rehabilitation [9,10], they provide conflicting evidence over whether costs are lower with community-based rehabilitation. However, recent research suggests that such care can generate greater improvement in physical functioning at modest additional cost [11]. None of the studies however have fully considered the net costs of the interventions. Furthermore, no research has directly considered whether shorter hospital stays combined with community rehabilitation is a feasible, cost-effective alternative to conventional management of stroke. This paper considers the costs of an early supported discharge service operated by Newcastle City Health Trust versus conventional hospital care for stroke using data from a pilot study using a randomized controlled trial design. Methods Description of the service The early supported discharge service was established in February 1995 as part of a pilot project to establish the feasibility of an earty supported discharge policy following acute stroke. An interdisciplinary team was created, consisting of full-time service coordinator, 345

2 P. McNamee et al. physiotherapist and occupational therapist, and parttime speech therapist and social worker. In addition, a home care bank was set up, with district nurse and occupational therapy technician input obtained when required. A budget for loan equipment was also available and medical cover was provided by the patient's general practitioner with consultant support if required. In addition to carrying out rehabilitation within patients' homes after discharge, members of the team acted as an 'in-reach' service, planning and organizing discharge and community rehabilitation arrangements during hospitalization. Randomization and sampling All patients admitted to either Freeman Hospital, Newcastle General Hospital or Royal Victoria Infirmary in Newcasde upon Tyne with acute stroke between 1 February 1995 and 31 January 1996 were assessed for eligibility to take part in the study against the criteria in Table 1. In total, 402 patients were admitted with stroke, of whom 119 "were eligible to participate. Consent was sought from patients and carers for data collection, after which patients were randomized. Permission to approach patients randomized to early supported discharge was then obtained from their hospital consultant and general practitioner before further consent was sought from patients and carers. Nine patients refused consent to data collection and a further 18 were unable to participate for other reasons (11 were participating in stroke drug trials and seven were excluded for other reasons: either previous serious neglect, self-discharge, delayed diagnosis, alcoholism or admission from hostel). Thus, the study population consisted of 92 patients, with 46 randomized to early supported discharge and 46 to conventional hospital care. Data collection Data on age, sex and neurological deficit were collected at admission. Barthel activities of daily living index (ADL) information was collected at 7 days post-stroke and at discharge [13]. Patients were assessed at 3 and 6 months post-stroke. Data relating to physical health outcomes were collected using the Nottingham Extended Activities of Daily Living Scale [14], depression was assessed using the Wakefield Depression Inventory [15] and overall health status measured by the Dartmouth COOP charts [16]. Carer stress was measured at 3 months by die General Health Questionnaire [17] and carer global health status by the Dartmouth COOP charts. As the focus here is purely on costs, 6-month health outcome data will be reported separately [18]. Resource use data were collected from hospital records and interview schedules on length of hospital stay following admission, length of hospital stay for readmissions and the frequency of use of a range of community services (day hospital, general practitioner and outpatient care). In addition, for those randomized to receive the early supported discharge service, the frequency and duration of physiotherapy, occupational therapy, speech therapy, district nursing, social work and home care visits were calculated from records kept by the early supported discharge team. For the conventional care group, the level of receipt of these services was estimated from staff records (hospitalbased input) and interview schedules (communitybased input). As travel time was only recorded for early supported discharge patients, travel times per visit were estimated for conventional care patients using the average for the early supported discharge group. Costs Table I. Eligibility criteria for study participation The costs incurred by the health and personal social services were calculated for stroke patients who received the early supported discharge service and for conventional care patients. For both groups of patients, these comprise hospital costs, rehabilitation costs and the costs of other services. Local weighted average specialty costs per bed day, based on 1995/6 contract prices, were used to value the resources used during inpatient stays. Weights were derived for each patient to reflect relative physical dependency, which was measured by the average of 7-day and discharge Barthel ADL scores. For example, a patient with a score twice that of the Resident in the study district Admitted from a private address Not previously severely handicapped (pre-stroke Oxford Handicap Scale 0-3 [12]) Admitted within 72 h of onset of stroke Medically stable 72 h after stroke Barthel score of between 5 and 19 at 72 h post-stroke No co-morbidity likely to affect rehabilitation 346

3 Cost analysis of early supported hospital discharge for stroke mean for all patients would have a weight of 0.5. This approach has been used in other studies to adjust per diem costs to reflect relative resource consumption [19]. Thus, total hospital costs per patient were calculated by multiplying the total number of days by the weighted cost. For both early supported discharge and conventional care patients, rehabilitation costs were calculated using local staff unit costs per hour. These were derived by firstly calculating an annual cost, which equalled the sum of actual salary levels and apportioned overhead costs (capital, supplies, office equipment). The annual cost was then divided by the product of the number of weeks worked per year and number of hours per week to produce an hourly unit cost figure. Thus, total costs of rehabilitation per patient were found by multiplying the total number of hours each staff type spent with each patient (including non-face-to-face patient-related tasks such as travel time, record keeping and planning) by the hourly unit cost. The same methods were applied for patients who received conventional care. In addition however, the cost of the service coordinator, who was employed for 8 months, was applied equally across all patients in the early supported discharge group, as was any additional time incurred by staff as a result of setting up the service or new methods of working. The costs of other services used after discharge from hospital were calculated by applying unit costs per visit, which have been derived and reported elsewhere [20]. Thus, total costs of these services were calculated by simply multiplying number of visits by the appropriate unit cost. Data on costs measured in the 6 months following stroke are reported in this paper. Statistical methods Differences in lengths of hospital stay and costs between both groups were compared using Mann- Whitney non-parametric tests of association. In addition, three-band sub-group analysis (grouped by Barthel ADL scores measured at 7 days post-stroke) on costs was carried out. All measurements were performed using intention-to-treat analysis. Results Group characteristics Table 2 shows that there were no significant differences in the demographic or clinical characteristics of either group as measured at admission. Length of hospital stay Mean number of hospital days, including re-admissions, was 27 in the early supported discharge group and 54 for conventional care patients. As a small number of patients had long lengths of stay, differences in the median were used to test for the statistical significance of changes in hospital days. Inclusive of readmissions, the difference of 14 (range 4-106) versus 26 days (range 4-183) was found to be significant (i> = 0.02). Use of rehabilitation services Table 3 shows the level of service receipt by total and median number of face-to-face contact hours, calculated across all patients within each group, together with the number of patients who received services. There were no significant differences between groups in the level of physiotherapy and speech therapy input. For all other services, however, early supported discharge patients received significantly greater levels of input, with the largest difference occurring in the provision of home care. Use of other services The amount of other services received in the community after discharge by total and median number of face-to-face visits, calculated across all patients within each group, is also shown in Table 3. There were no statistically significant differences between groups, although early supported discharge patients had more contact with general practitioners and conventional care patients received higher levels of day-hospital care. Table 2. Early supported discharge (ESD) and conventional hospital care (CHQ: demography and severity of stroke Median age, years (and range) Female gender (and %) No. (and %) living alone Severity of stroke No. (and %) continent of urine at 24 h Median 7-day ADL score (and range) ESD 73(47-93) 20 (43) 22 (48) 42 (91) 15(2-20) CHC 73 (44-91) 22(48) 21 (46) 41(89) 13(2-20) ADL, Barthel activities of daily living. 347

4 P. McNamee et al. Table 3. Early supported discharge (ESD) and conventional hospital care (CHC): level of service receipt ESD CHC No. of hours/visits No. of hours/visits Rehabilitation service (hours) Physiotherapy Occupational therapy Speech therapy District nursing Social work Home care Other services (visits) Day-hospital care Outpatient care General practitioner care, not significant at the 5% level. n Total Costs The costs of managing stroke over 6 months were found to total 7155 per patient in the early supported discharge group versus 7480 per patient in conventional care group. Table 4 indicates that the additional costs associated with implementing and running the scheme were more or less balanced by the value of bed days saved. Sensitivity analysis Two one-way sensitivity analyses were performed. First, as the estimates of bed day costs were based on specialty contract prices, it is uncertain whether these accurately reflected the level of resources used during acute inpatient care. Therefore, upper and lower weighted average specialty bed day costs were applied to hospital stays. In the upper band all general medicine and all geriatric medicine bed days were costed at 128 and 93 respectively, while in the lower band figures of 116 and SA6 were used. These values were chosen using the highest and lowest average bed Median (and range) 4(1-9) 8(4-14) 1 (0-3) 2 (0-3) 2 (0-4) 2(0-114) 0 (0-26) 0 (0-8) 2(0-13) n Total Median (and range) 4(2-15) 4 (0-8) 0 (0-2) 0(0-1) 0 (0-2) 0(0-41) 0(0-91) 1 (0-2) 1 (0-4) P value <0.001 < day cost estimates provided by Newcastle and North Tyneside Health Authority. In the lower band average costs per patient fell to 5583 in the early supported discharge group and 5005 in the conventional care group, while application of the upper band changed costs to 7442 and 8568 respectively. Second, some of the time used by the team did not relate to individual patients. Partly, this was due to factors relating to implementation of the new service (for example, the time taken to establish new networks with other agencies and the cost of the service coordinator) and carrying out extra documentation for research purposes. However, the bulk of this time was probably caused by activities associated with team working (goal-setting, case conferences and team meetings). As these latter activities are directly related to the new service, exclusion of service coordinator and implementation costs was undertaken to provide an estimate of recurrent costs. This reduced costs in the early supported discharge group by 800 to 6355 per patient. Table 4. Early supported discharge (ESD) and conventional hospital care (CHC): average costs per patient Cost( ) ESD CHC Difference Inpatient costs Rehabilitation/additional service costs Other service costs Total costs

5 Cost analysis of early supported hospital discharge for stroke Sub-group analysis To gain further insight into whether the level of physical dependency influenced costs, patients were divided into three sub-groups based on 7-day Barthel ADL scores. These cut-off scores were chosen following Granger et al. [21], who reported that scores of 12 or below indicate dependence and scores of eight or below were markers of severe dependence. Table 5 shows the level of cost variation and suggests that there is a clear inverse relationship between Barthel ADL score and costs, with the most dependent patients (Barthel 0-8) being the most expensive. In addition, there appears a slight cost advantage of early supported discharge care for more dependent patients (that is, Barthel scores of 12 or below). However, these differences did not reach statistical significance. In contrast, for the most independent patients, costs were significantly lower for those who received conventional hospital care compared with early supported discharge. Sensitivity analysis was also conducted on these sub-groups and there were no changes in the magnitude or direction of cost differences. Discussion Reduced hospital lengths of stay of up to 50% were achieved by the early supported hospital discharge service, with the additional costs at least balanced by the value of the released resources. Sensitivity analysis did not markedly alter this finding. Although the service was found to be more expensive when lower values for bed day costs were used, the difference was statistically non-significant. Indeed, as the baseline values themselves are lower than those reported elsewhere [20], the magnitude of cost differences between the two groups is likely to be nearer to those calculated using the higher estimate of bed day costs. Moreover, other changes, such as subtraction of implementation costs, also made the Table 5. Three-band analysis of costs Band 7-day Barthel score of 0-8 ESD 11 CHC 10 7-day Barthel score of 9-12 ESD 9 CHC 13 7-day Barthel score of ESD 26 CHC 23 n Median Barthel score service less expensive than conventional hospital care. If it can be shown that the service does not lead to inferior health outcomes [18], early supported discharge therefore appears to be a cost-effective way of managing patients with stroke. Inevitably, some caution should be exercised before accepting this conclusion. First, the power of this pilot study was not sufficient to detect clinically important differences in health outcomes. Second, only under one-quarter of those admitted with stroke were eligible to participate in the study. Third, the sub-group analysis suggests that the costs of the scheme are associated with levels of initial physical disability. Therefore, the results may only be generalizable to a select group of patients with similar Barthel ADL scores. Finally, only costs to the health and personal social services over 6 months are included. If early supported discharge leads to differences in admission to residential or nursing homes, or greater problems for carers in terms of psychological distress or lost work opportunities in the longer term, then costs are likely to have been underestimated. Despite these caveats, the finding that the service is no more expensive than conventional hospital care is particularly robust by virtue of the assumptions used in the baseline analysis. In particular, one important factor led to the results being biased against the early supported discharge group. Due to the way rehabilitation data were recorded in the control group, the level of service receipt amongst conventional care patients is likely to have been underestimated. Staff were asked to estimate from retrospective records the number of hours of hospital-based contact, while patients were asked about the level of post-discharge contact. In the early supported discharge group, however, the team kept detailed records of all contacts, although patients were also asked to provide estimates of the frequency of post-discharge services. Analysis of the difference between data collected by members of the team and reported by patients revealed that Costs ( ) P value ESD, early supported discharge; CHC, conventional hospital care. 349

6 P. McNamee et al. patients reported significantly fewer visits. However, due to uncertainty of the size of the discrepancy in the control group, these data were left unadjusted. This means that the costs of conventional hospital care are likely to be greater than 7480 per patient. Viewed from the perspective of the health service, the attractiveness of an early supported discharge policy crucially depends on the assumptions made about the value of the freed bed days. The savings brought about by the early supported discharge service are not real savings but are estimates of the value of shorter lengths of stay, based on the assumption that released beds will be used by patients of similar need. If this assumption is violated, then the cost effectiveness of the scheme is correspondingly reduced. Similarly, should lengths of hospital stay fall for conventional hospital care of stroke in the future, the cost effectiveness of the service -will also diminish. The finding in sub-group analysis that costs vary by Barthel ADL score merits further investigation. The power of the study was not sufficient to determine whether differences in costs were matched by differences in health outcomes. For example, amongst less physically dependent patients, costs were significantly higher for patients who received early supported discharge with no differences in outcomes, even when unweighted costs were applied. The policy implication may be that early supported discharge is better targeted towards other groups, such as more physically disabled patients, if it can be shown that health outcomes are not inferior for these patients. However, previous research of a similar high-support hospital discharge scheme for older people demonstrated that benefits were greater in those who were less disabled [22], while others showed that Barthel scores of <9 were associated with problems of managing at home [21]. Further research with larger sample sizes is required to determine the statistical significance of the results in order to guide policy making. In conclusion, the economic evaluation of the early supported discharge service suggests that it is a costeffective alternative in the management of stroke. However, further investigation is required to determine the effects on costs in the long run, to assess whether interdisciplinary teams in other hospitals produce different costs and outcomes, and to investigate whether health outcomes are superior in particular groups. Acknowledgements We thank members of the early supported discharge service and all medical, nursing and therapy staff and patients who participated in the study. We also thank David Lewis of Newcastle and North Tyneside Authority and Jenni Bremner of Newcastle City Health Trust for help in supplying unit costs. We offer grateful acknowledgement to Wendy Kaiser, Ruth Dobson for data management and Joseph Hoben for secretarial support. The research was funded by the National CVD and Stroke Research and Development Programme; the service development by Newcastle Health Authority Primary Care Development Fund. Key points Early supported discharge for acute stroke is feasible and significantly reduced lengths of hospital stay. The additional costs associated with early supported discharge were balanced by the value of released bed days. Health outcomes were found to be similar compared with conventional hospital care. Costs were higher amongst the more physically dependant patients irrespective of the type of care offered. Early supported discharge was shown to be a costeffective alternative to conventional hospital care; larger studies are required to test the generalisability of the results. References 1. Bond J. The importance of social care for stroke patients and their families. In: DallJLC et al. eds. Prospects in Ageing. London: Academic Press, Forbes J. Cost of stroke. Scot MedJ 1993; 38 (suppl. 3): Malmgren R, Bamford J, Wariow C, Sandercock P, SlatteryJ- Projecting the number of patients with first ever strokes and patients newly handicapped by stroke in England and Wales. BrMedJ 1989; 298: Bamford J, Sandercock P, Dennis M et al A prospective study of acute cerebrovascular disease in the community: the Oxfordshire Community Stroke Project. J Neurol Neurosurg Psychiatry 1990; 53: BonitaR. Epidemiology of stroke. Lancet 1992; 339: Secretary of State for Health. The Health of the Nation: a strategy for health in England. London: HMSO, Indredavik B, Bakke F, Solberg R et al Benefit of a stroke unit: a randomized controlled trial. Stroke 1991; 22: Langhorne P, Williams BO, Gilchrist W, Howie K. Do stroke units save lives? Lancet 1993; 342: Young J, Forster A. Day hospital and home physiotherapy for stroke patients: a comparative cost-effectiveness study. J Roy CoU Physicians Lond 1993; 27: Gladman J, Whynes D, Lincoln N. Cost comparison of domiciliary and hospital-based stroke rehabilitation. Age Ageing 1994; 23: Byford S, Geddes JML, Bonsall M. Stroke rehabilitation: a 350

7 Cost analysis of early supported hospital discharge for stroke cost-effectiveness analysis of a placement scheme. Centre for Health Economics, Discussion Paper 140. University of York, van Swieten JC, Koudstaal PJ, Visser MC et at. Interobserver agreement for the assessment of handicap in stroke patients. Stroke 1988; 19: Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. Md State Med J 1965; 14: Nouri FM, Lincoln NB. An extended activities of daily living scale for stroke patients. Clin Rehab 1987; 1: Snaith R, Ahmed S, Mehta S, Hamilton M. Assessment of the severity of primary depressive illness. Psychol Med 1971; 1: Nelson E, Wasson J, Kirk J et al. Assessment of function in routine clinical practice: description of the COOP chart method and preliminary findings. J Chron Dis 1987; 40 (suppl. 1): 53-63S. 17. Goldberg D, Williams P. A User's Guide to the General Health Questionnaire. Windsor NFER-Nelson, Rodgers H, Soutter J, Dobson R et al. Early supported hospital discharge following acute stroke: length of stay and three month outcomes (Abstract). Cerebrovasc Dis 1996; 6 (suppl. 2): Farnworth MG, Kenny P, Shiell A. The costs and effects of early discharge in the management of fractured hip. Age Ageing 1994; 23: Netten A, Dennet J. Unit Costs of Community Care. Personal Social Services Research Unit, University of Kent, Canterbury, Granger C, Dewis L, Peters N et al. Stroke rehabilitation analysis of repeated Barthel index measures. Arch Phys Med Rehabil 1979; 60: Martin F, Oyewole A, Moloney A. A randomized controlled trial of a high support hospital discharge team for elderly people. Age Ageing 1994; 23: Received 13 February

8 Anna Coorey, who is 88 years old, goes swimming every day. Ulrike Preuss/Format.