MODELLING SERVICE INNOVATION IN STROKE CARE FINAL REPORT Core research team, Imperial College: Dr Benita Cox; Prof. James Barlow; Dr Christina Petsoulas; Dr Steffen Bayer Advisory team: Dr Stephen Morris (Brunel University); Dr Alasdair Honeyman (independent consultant) Prepared for: Department of Health, Information and Communication Research Initiative 2 (ICRI II) October 2008 EXECUTIVE SUMMARY The UK government published the National Stroke Strategy in December 2007. This emphasises the urgent need to change all aspects of stroke care, from patient awareness and prevention to long-term rehabilitation after stroke. Our study contributes to a better understanding of the need to reconfigure healthcare services for stroke and the challenges of supporting aspects of stroke care with telecare or telemedicine solutions. It was funded as part of the Department of Health’s Information and Communication Research Initiative II. The research had four main objectives: to map out the care journey for stroke patients, to identify components of the care journey which could potentially be improved, to identify appropriate interventions (ICT or others) that might improve stroke care delivery in line with new policy guidelines and thereby to support local care community with planning stroke services. The reconfiguration of stroke services requires choices to be made between a variety of models, for example whether to provide regional specialist stroke centres for the acute phase of stroke care or to provide district general hospitals with specialist 24/7 facilities for treatment. Other potential configurations may be to connect district general hospitals via telemedicine to specialist advice in stroke centres. For each configuration there are associated advantages and disadvantages which are discussed in this report. A major difficulty facing stroke care provision is the lack of service integration between the many authorities, professionals and stakeholders involved in the process. In addition, there are also difficulties experienced during both the acute phase of care as well as during the period of rehabilitation. Some of the more salient of these may be summarised as follows: Acute phase capacity pressures availability of CT scanning inadequate out-of-hours cover, lack of thrombolysis facilities, bed pressures in stroke units, inappropriately long lengths of stay. 2 Community care lack of workforce availability, limited facilities for long-term care, paucity of appropriate facilities for Early Supported Discharge At the interface between acute and community care, delays exist in seeing patients after discharge. In this project we used simulation modelling to map the flow of care in the acute and community parts of the care pathway for stroke patients and built alternative models of care provision. In recent years emphasis has been placed on the importance of patient self-care and the need to support independent living. The use of telemedicine has the potential to enhance patient independence and contribute to better chronic disease management. Technologies for the remote monitoring of an individual’s health condition (telecare) could also, in time, become a mainstream part of community care delivery. In general, these technologies have the potential to be used in all aspects of healthcare from prevention, to acute care, to care in the community. While there is knowledge of the impact of technological innovations on individual patient outcomes, their wider systemic impact is little understood. Simulation modelling can help to develop such an understanding and ensure that future services are appropriately designed and targeted. Through modelling our study captures the current process of stroke care delivery (in its acute and community settings) and offers the potential for scenario building in an interactive way. The modelling process highlights areas where sufficient data is not available as well as neglected problem areas in the care system. We suggest that this type of modelling is suitable for application to highly complex processes, such as stroke care, and can be used successfully as a communication and decision making tool before committing real resources. However, the use of modelling is made challenging by the paucity of robust data. 3 CONTENTS 1 Introduction: project background................................................................................................ 7 2 The policy context ........................................................................................................................ 9 3 2.1 Financial implications of stroke ......................................................................................... 9 2.2 The need for more efficient use of resources ................................................................. 9 2.3 Need for service re-configuration and improvement ................................................... 10 Literature review: services for stroke patients ....................................................................... 12 3.1 3.1.1 Integrated services ....................................................................................................... 12 3.1.2 Stroke units.................................................................................................................... 13 3.1.3 Weekend cover ............................................................................................................. 14 3.1.4 Early supported discharge (ESD) .............................................................................. 14 3.1.5 Post-stroke rehabilitation ............................................................................................. 15 3.2 Diagnostics and treatments ............................................................................................. 16 3.2.1 CT scanning .................................................................................................................. 16 3.2.2 Thrombolysis ................................................................................................................. 17 3.2.3 TIA and secondary prevention.................................................................................... 17 3.3 4 Organisational structures and services ......................................................................... 12 The use of telecare, telemedicine and telerehabilitation ............................................ 18 3.3.1 Telestroke ...................................................................................................................... 18 3.3.2 Telerehabilitation .......................................................................................................... 19 3.3.3 The potential benefits of telestroke – and the evidence ......................................... 19 Case study: local stroke services ............................................................................................ 21 4.1 Background ........................................................................................................................ 21 4.2 Current stroke care process – hospital .......................................................................... 21 4.3 Current stroke care process – community .................................................................... 22 4.4 Identified problems and potential improvements ......................................................... 23 4 4.4.1 Stroke unit capacity ...................................................................................................... 23 4.4.2 Scanning and other tests............................................................................................. 23 4.4.3 Thrombolysis ................................................................................................................. 24 4.4.4 Transient ischaemic attack (TIA)................................................................................ 24 4.4.5 Length of stay................................................................................................................ 25 4.4.6 Early supported discharge (ESD) .............................................................................. 25 4.4.7 Rehabilitation................................................................................................................. 25 4.5 5 6 Data analysis of local stroke services ............................................................................ 26 Modelling .................................................................................................................................... 30 5.1 The planning challenge in healthcare ............................................................................ 30 5.2 Modelling in healthcare planning .................................................................................... 31 5.3 The prototype stroke care model.................................................................................... 33 5.3.1 Introduction .................................................................................................................... 33 5.3.2 Implementation ............................................................................................................. 34 5.3.3 Features of the prototype ............................................................................................ 35 5.3.4 Parameters and data ................................................................................................... 37 5.3.5 Experiments with the prototype .................................................................................. 38 Discussion ................................................................................................................................... 43 6.1 The need for service integration ..................................................................................... 44 6.2 Regional reorganisation of stroke services ................................................................... 45 7 Conclusions and recommendations ................................................................................... 48 8 References .................................................................................................................................. 50 Appendix 1: Methods ......................................................................................................................... 59 Appendix 2: Dissemination ............................................................................................................... 61 5 Figures Figure 1: Length of stay haemorrhage patients ............................................................................. 27 Figure 2: Length of stay infarction patients .................................................................................... 27 Figure 3: Hospital admissions by type ............................................................................................ 28 Figure 4: Waiting time between discharge and therapy start ...................................................... 28 Figure 5: Care costs for patients receiving community rehabilitation ......................................... 29 Figure 6: Model overview .................................................................................................................. 35 Figure 13: Regional reorganisation of stroke services ................................................................. 46 6 1 Introduction: project background The improved management of chronic disease is a central policy aim of the UK government. The vision is one in which services are situated close to the patient and patients are supported in self-care. Technological advances have the potential to facilitate the delivery of this vision. Exploring the potential of telemedicine and telecare in the management of chronic disease is the general theme of the Department of Health’s Information and Communication Research initiative II under which this research was funded. This project is concerned with stroke. Stroke is the third leading cause of death (after cancer and heart disease) and the single largest cause of long-term severe disability in the UK and other economically advanced countries. In recent years there has been a widespread recognition of the need for improvements in the provision of stroke services (Markus 2007; Schwamm et al. 2005; Saposnik, 2007; Rodgers, et al. 2003; Norrving and Adams 2006; Langhorne 2006; Darzi, 2007). Radical changes are needed throughout the stroke pathway, i.e. from prevention to acute care to care in the community with wide scale service reconfiguration. Planning wide scale service re-configuration is complex and requires numerous factors and alternatives to be balanced. One approach to this problem is to use modelling. Modelling can facilitate planning complex processes such as stroke care and can be used successfully as a communication and decision making tool before committing real resources. This project considers the use of modelling to assess the likely impact of changes to care practices and services on the outcomes of stroke care. The research had four main objectives: 1. To map out the care journey for stroke patients in order to gain a detailed understanding of the whole patient pathway as well as the organisations and professionals involved in stroke care. 2. To identify components in the care journey that could potentially be improved. The stroke patient pathway is a complex one which involves the close collaboration between many organisations across primary, acute and social care. Fragmented services can mean unnecessary long delays and bottlenecks with harmful effects both for patients and for the organisations involved. Identifying the reasons for delays is a key step towards improving stoke care services. 7 3. To use modelling to identify appropriate interventions that might improve stroke care delivery in line with new policy guidelines. 4. To support the local care community in planning stroke services. Planning and re- structuring complex, multi-agency services requires the close collaboration between different stakeholders. Modelling can facilitate discussion around issues such as preferred priorities and cost-effectiveness, and lead to better decision-making. We adopted a case study research design and used a combination of methods for the research: semi-structured interviews, observation, literature scanning, policy documentation analysis, modelling (Appendix 1). 8 2 The policy context The financial burden of stroke care has been widely recognised creating an awareness of the need for a more efficient use of resources. Inefficiencies in resource use as well as dissatisfaction with care outcomes and care processes have triggered calls for a reconfiguration and improvement of services. 2.1 Financial implications of stroke As a leading cause of the need for inpatient hospital stay, long-term rehabilitation and care in the community, stroke represents a huge financial burden on NHS resources. According to one estimate this is as high as 4-6% of the total NHS expenditure (Saka, et al., 2005). The national average length of stay in hospital for stroke patients is 28 days, so a large proportion of NHS spending on stroke is for hospital inpatient stay. According to a recent National Audit Office report, £2.8 billion represent direct care costs, £2.4 billion are informal care costs (e.g. home care by families) and £1.8 billion are wider economic costs. The same source reports that on average 20% of nursing home residents are there as a result of stroke, with total annual care costs of approximately £680 million (NAO, 2005). 2.2 The need for more efficient use of resources There is wide variation in the quality of stroke care provision between the UK and other economically advanced European countries, with the UK consistently achieving poor outcomes. Markus (2007) found that spending on stroke services in the UK is as high or higher than in other European countries which achieve better outcomes. According to the National Director for Heart Disease and Stroke, spending on the NHS has increased significantly, and currently the UK’s average spending on healthcare costs, as a percentage of GDP and per capita, including the percentage spent on stroke services, is double the European average. Despite this increase in spending, the stroke mortality rates in the UK have remained relatively high and the use of effective treatments, such as thrombolysis, is making slow progress (Boyle, 2007). Grieve et al. (2001) note that high cost stroke care in London is accompanied by high levels of post stroke mortality and dependence. Together, these findings suggest that the organisation of stroke care delivery, rather than levels of funding, is an important factor contributing to poor outcomes. The key question is thus how existing resources may be deployed in a more cost-effective way (Grieve et al. 2001). According to the estimates of the NAO report, better organisation of stroke services could result in £20m annual savings, 550 fewer deaths and 1700 fewer cases of disability. (NAO, 2005) 9 2.3 Need for service re-configuration and improvement The need for improving stroke services has been recognised in the National Service Framework for Older People (DH, 2001), which stated that every general hospital should introduce a specialist stroke service by 2004. However, a national survey conducted by the British Association of Stroke Physicians found that many hospitals lacked a number of factors required by the NSF (Rodgers et al. 2003). The The findings are summarised in the box. British Association of Stroke Physicians survey Only half of the acute stroke units admitted patients within 24 hours. (Rodgers et al. 2003). The number of beds available on acute stroke and stroke rehabilitation units did not reflect the total number of stroke inpatients, and a large number of these units (53% and 79% respectively) had admission criteria.1 The impact of limited bed availability in stroke units means that not all patients have access to dedicated stroke services throughout the entire duration of their stay in hospital. The authors also found that stroke unit therapy staffing levels in UK hospitals were often lower than the staffing levels present in randomised controlled trials. In order to provide the early and intensive rehabilitation required by the NSF, therapy staffing levels would have to be increased (Rodgers et al. 2003). The need to restructure stroke services in the UK is urgent. In other countries, for example, stroke care is an integral part of neurology. In the UK it tends to fall between neurology and general geriatric medicine (Markus 2007). Greater emphasis should be placed in the UK on improving the acute phase of stroke care. This would not only significantly improve patient outcomes but would also be costeffective by reducing the length of stay in hospitals and by resulting in less long-term disability (Markus 2007). The vast majority of the cost incurred in in-hospital stay in the UK is for nursing and overheads rather than investigations and medical care (Markus 2007). Patient awareness and recognition of stroke symptoms needs to be improved with immediate ambulance assessment and transfer to specialist stroke centres. (Markus 2007). While in many European countries CT or MRI scanning is performed on admission to A & E departments, many UK hospitals struggle to provide it within 24 hours (Markus 2007). There is inadequate access to stroke specialists for administering thrombolysis where suitable. According to Markus (2007), while 20-30% of eligible patients are given 1 According to the 2004 National Sentinel Stroke Audit, 67% of hospitals nationally applied selection criteria for their stroke units. 10 thrombolysis in Australia2, North America and Europe, only 1% of patients receive the therapy in the UK. A severe shortage of specialists trained in the management of acute stroke care exists. (Markus 2007). Despite the evidence for the benefits of organised stroke units, only 62% of patients in the UK were treated in such a unit at any time during their stay and only 54% spent more than half of their stay in such specialised units (Royal College of Physicians, Sentinel Audit, 2006) These issues suggest that new strategies and possibly a substantial reconfiguration of services are required in the UK. The recently published National Stroke Strategy (DH, 2007) makes the following major recommendations: improve awareness of the symptoms and urgency of stroke among NHS staff and the public, educate the public about stroke risk factors and provide advice on lifestyle and treatment options, improve the management of minor strokes or transient ischaemic attacks (TIAs) (e.g. through expert assessment, MRI within 24 hours, follow up care), radically change existing services of acute (e.g. 24/7 access to clinical assessment, rapid imaging and the ability to deliver intravenous thrombolysis, prompt access to high quality stroke unit, stroke networks, regional stroke centres) and post-acute stroke care provision (e.g. early supported discharge, rehabilitation as long as needed), provide adequate specialist therapy and support in the community, provide active end-of-life care, deliver patient care through stroke skilled workforce. Implementing these recommendations will require careful planning and close collaboration of those involved in the provision of stroke care e.g. decision makers, professionals as well as patients and carers. 2 An Australian government research paper, however, claims that thrombolysis is not widespread in either Australia or New Zealand and in the US, at least until 2005, the rate was only between 2% and 4% (Wojner-Alexandrov and Malkoff, 2006). 11 3 Literature review: services for stroke patients The purpose of the review was to identify published research and trials on stroke care and the main issues concerning the optimal organisation of stroke care services. The literature may be broadly classified into two domains: organisational structures and services, and diagnostic treatments. The literature on the role of telecare, telemedicine and telerehabilitation in stroke services is also reviewed. 3.1 Organisational structures and services 3.1.1 Integrated services Stroke is a complex condition which requires a systemic integration of services e.g. primary care, ambulance services, acute treatment and rehabilitation, post-acute rehabilitation and often long-term health and care support in the community. For this reason, it presents a significant challenge to existing generally fragmented health and social care services, not only in the UK but in other advanced countries (e.g. the US, Canada). There is a need for integrated planning and delivery of stroke services (Plochg and Klazinga 2002). A recent OECD report comparing the treatment, costs and outcomes of stroke services in seventeen European countries concludes that there are benefits from taking an holistic systemic approach to managing stroke, including prevention, acute care and rehabilitation and, second, that stroke units appear to have the potential to deliver real benefits, which are not currently fully realised (Moon et al. 2003). The recent National Stroke Strategy (DH, 2007b) and the American Stroke Association’s Task Force on the Development of Stroke Systems (Schwamm et al. 2005) both recommend the implementation of ‘stroke systems’ or ‘stroke networks’. As well as raising public awareness of the symptoms of stroke and providing information for appropriate action, such systems help to ensure that patients are brought to ‘hyper-acute’, fully-equipped hospitals for the acute phase of stroke. They also enable a smooth transition from hyper-acute to subacute care and then from inpatient to outpatient care and rehabilitation in the community. Stroke networks include all healthcare organisations involved in the provision of services. . In order to work effectively, stroke networks require appropriate communication links and transportation protocols which can be facilitated by advances in ICT (see section 3.3. on telecare and telemedicine). Despite long standing scientific proof of the effectiveness of organised stroke care, stroke systems have been slow to implement (Norrving 2005; 12 Norrving and Adams 2006). To move from the current fragmented approach to integrated stroke systems is a complex task. Radical service re-configuration needs time, resources and careful planning with input from all relevant stakeholders. Resource limitations or delays in reaching agreement between stakeholders may mean that adopting an incremental instead of comprehensive service reconfiguration approach is more realistic. It may be necessary, for example, to start by improving the acute aspect of stroke care before moving on to rehabilitation in the community. Simulation modelling can be useful during the planning stage of complex service re-configuration (see ‘Modelling’ section below). 3.1.2 Stroke units Dedicated specialist stroke units, typically including a co-ordinated multidisciplinary team operating within a discrete ward, can be either acute only, rehabilitation only or a combination of the two (comprehensive). A number of studies show that patients treated in stroke units rather than in general medical wards benefit in terms of survival, regained independence and return home (Stroke Unit Trialists’ Collaboration 2001; Evans, Perez et al. 2001). Studies that have extended the follow-up for five or ten years have found that the benefits are sustained for patients treated in stroke units (Jorgensen et al. 1999). There is wide variation in practice which makes it difficult to give a precise definition of ‘stroke units’. This means there is some uncertainty about which components of stroke units contribute to better outcomes. A review of trials (Langhorne and Pollock 2002) suggests that the characteristics that make stroke units effective appear to be: early comprehensive assessment of medical problems and complications (e.g. medical history and diagnostic examinations), early and active management of physiological abnormalities (including early mobilisation of patients), skilled nursing care, early setting of rehabilitation goals and early planning to discharge. Effective stroke units require a core multidisciplinary team of medical, nursing and therapy staff (physiotherapy, occupational therapy and speech and language therapy). Despite the problems in measuring the impact of stroke units, a number of studies have pointed to their effectiveness. A study in Norway, for example, shows that stroke units provide better treatment in terms of reducing death and disability, with beneficial long-term effects on survival, disability and quality of life. In Norway, where most acute hospitals have dedicated stroke units, the recommended model is a non-intensive combined unit able to focus simultaneously both on acute care and rehabilitation. This has been well evaluated in randomised trials and has achieved some of the best results regarding acute stroke care (Indredavik 2003). 13 Evans et al. (2001) argue that beneficial practices such as thrombolysis, physiological homoeostasis, early prescription of aspirin, anticoagulation and early mobilisation are likely to contribute significantly to reducing mortality and dependence, and that such interventions are more common in stroke units than in non-specialist settings. Diagnostic investigations, such as CT scans and carotid duplex studies, are also undertaken earlier in stroke units than in general wards. In addition, aggressively managed supportive care to assist neurological recovery and which is applicable to most stroke patients can be provided in stroke units. Another study (Kalra et al. 2000; 2005) compared stroke units, stroke teams and care in the patient’s home, and found that stroke units were more effective than a specialist stroke team or specialist domiciliary care in reducing mortality, institutionalisation and dependence after stroke. The authors found little support for either specialist domiciliary services for acute stroke or for stroke management in general medical wards, even with specialist team input. The stroke unit intervention was less costly per patient day alive and more effective than the stroke team intervention. The stroke unit was found to be a more cost-effective intervention than either stroke team or home care (Kalra et al. 2000; 2005). Another study concludes that, compared to stroke teams or domiciliary care, stroke units are more expensive, but they have better quality outcomes. (Patel et al. 2004). In general, the value of stroke units is now widely recognised. According to one expert, ‘although recommendations and guidelines on stroke care may differ in details, they all point towards stroke unit care and speed. Cost may be an argument against such units, but the benefits are substantial’ (Hacke 2000). 3.1.3 Weekend cover Studies show that reduced weekend and holiday cover has a negative impact on stroke outcomes. Hasegawa et al. (2005) and Saposnik (2007) investigated the effects on stroke outcomes of low-volume care resulting from weekends and holidays, when staffing levels are lower and without rehabilitative services. Weekday admission was found to be an independent negative predictor of case fatality and a positive predictor of favourable outcome at discharge from acute stroke units. In patients with rehabilitative therapy, a reduction in the weekday ratio was also associated with unfavourable outcome, probably due to a reduction in multidisciplinary care. 3.1.4 Early supported discharge (ESD) The national length of stay in NHS hospitals is in general a lot higher than in the US (Ham et al. 2003), and the biggest differences are for stroke and hip fracture. Although this study has been criticised (Mallet 2004), the authors argue lessons can be learnt about the scope for 14 reducing hospital lengths of stay for stroke by providing care in alternative, non-acute settings Plans for introducing alternative arrangements for the provision of therapy for stroke patients have recently been the focus of health policy in the UK, for example through ‘early supported discharge’ (ESD) for patients following a mild or moderately severe stroke. ESD is a system of care which aims to provide earlier discharge with rehabilitation and support in a home setting, as opposed to conventional hospital based rehabilitation. ESD is typically provided by a specialist multi-disciplinary team including medical, nursing and therapy staff as well as social work input. A number of studies have found that ESD can be beneficial both in terms of cost effectiveness and care outcomes (Larsen et al. 2006). A review by Anderson, et al. (2005) showed that the greatest economic benefits have been seen in trials involving a co-ordinated multi-disciplinary ESD team3 and in patients with mild-moderate disability. These benefits arise from reductions in length of hospital stay, long-term dependency and admission to institutional care (Anderson et al. 2005). This review also found that patients receiving ESD services were more likely to be independent and living at home six months after stroke than those who received conventional care. Thorsen et al. (2005) show that the good outcomes experienced by patients receiving ESD were still present five years later Fjaertoft et al. (2005) found that ESD is most cost-effective for patients with a moderate stroke and concludes that acute stroke unit care combined with an ESD programme may reduce the length of institutional stay without increasing the costs of outpatient rehabilitation compared with traditional stroke care Langhorne (2006) also found that ESD reduces hospital length of stay and long-term dependence, mostly for patients with less severe strokes. This review points out, however, that more research is required to define the essential characteristics of ESD services, as well as to determine cost-effectiveness for different patient and service groups (Langhorne 2006). 3.1.5 Post-stroke rehabilitation The problems most frequently mentioned by stroke survivors and their carers are the social and emotional consequences of stroke and, second, service deficiencies. There is little clarity about the structure of a comprehensive primary care service providing long-term support to stroke patients and their carers (Murray et al. 2003). Co-ordination between 3 Comprising medical, nursing, physiotherapy, occupational therapy, and speech and language therapy staff. 15 health and social care service providers, in relation to the provision of community rehabilitation, can also be lacking (McKevitt et al. 2004). Stroke care units are seen as the appropriate setting for the acute phase of stroke. However, whether an acute hospital setting is the most appropriate environment for post-acute stroke rehabilitation has been questioned (Von Koch 1998). Much of the literature has concentrated on the appropriate length of time for rehabilitation to be provided and whether the optimal setting is within the hospital or elsewhere. According to the European Stroke Initiative, rehabilitation should begin as soon as patients are clinically stable and during the acute phase should not last longer than 6-12 weeks. After this time, it should be administered for as long as improvement in function is observed (Hacke 2003). It has been argued that rehabilitation should, whenever possible, take place in the patient’s home (Wade 2003). However other studies suggest that whilst rehabilitation in a stroke-unit care can be valuable for stroke patients in hospital, the effectiveness of outpatient care is less certain (Legg et al. 2004). Research suggests that therapy-based rehabilitation services for stroke patients living at home is beneficial in terms of maintaining independent living. 3.2 Diagnostics and treatments 3.2.1 CT scanning CT scanning is essential in diagnosing different types of stroke (e.g. infarction, haemorrhage and stroke mimics) and in determining the appropriate clinical management decisions. Limitations to immediate CT scanning may be due to lack of availability in scanners or a shortage of radiologists and radiographers. Wardlaw et al. (2004) argue that as CT scanning is an expensive and limited resource, determining its cost-effectiveness is important. The authors used modelling techniques to estimate the expected benefits and costs associated with different CT scanning policies. They recommend a policy of scanning all patients immediately. Costs are offset by savings in the length of inpatient stay. Because of the overall cost-effectiveness of immediate CT scanning, the cost of installing CT scanners (or training more radiologists) would quickly be recouped in reduced costs of stroke to health services. 16 3.2.2 Thrombolysis In some countries, the intravenous injection of clot-busting drugs such as t-PA (thrombolysis) in the treatment of stroke was given approval more than a decade ago (e.g. in the US in 1995). Although the availability of thrombolysis has revolutionised the treatment of stroke patients, its use is still limited. While in some countries thrombolysis is used widely (e.g. US, Germany), others (e.g. UK, Australia, New Zealand) have been slower in adopting the treatment (ANZHSN, 2006). This slow uptake may partly be due to the fact that it is only suitable for a relatively small proportion of patients within a short time window from the onset of stroke and because it requires careful co-ordination of services and 24/7 availability of medical expertise (including prompt CT scanning). Alternative models for offering thrombolysis (e.g. ‘hub and spoke’ systems in which hospitals form a network with one specialist centre offering support and advice via the use of telemedicine) may be more cost-effective. In the light of these challenges, the costeffectiveness of introducing thrombolysis to every single general hospital has not yet been proven. A 2003 review of thrombolysis trials suggests that the procedure can have promising outcomes but its widespread use in routine clinical practice is not supported (Wardlaw et al, 2003). Khaja and Grotta 2007argue that since only a small percentage of acute ischaemic stroke patients meet the criteria for thrombolysis, alternative acute treatment strategies should also be in place. 3.2.3 TIA and secondary prevention Effective management of mini-strokes or Transient Ischaemic Attacks (TIAs) is important for stroke care planning as these are warning signs for the onset of a major stroke in 15-20% of cases (Rothwell et al. 2006). These authors argue that the emphasis should be placed on identifying, including through brain imaging (CT and MRI), and treating the patients most likely at risk of a recurrence. Blight et al. (2000) suggest that compared to conventional practice of seeing the patient first, ordering of tests and then reviewing them, a single consultation dedicated Cerebrovascular Disease (CVD) clinic is cost effective in the outpatient management of minor stroke or TIA incidents. Educating the public to recognise the symptoms of TIA is also essential (Rothwell, Buchan et al. 2006). Urgent treatment of TIAs and minor strokes can result in effective secondary prevention and even if secondary prevention is ineffective, urgent admission to hospital can still have beneficial effects for high risk patients because of the potential for rapid thrombolysis in the event of an early recurrent stroke. Prompt surgery for suitable patients may also justify urgent hospital admission in high risk patients (Rothwell et al. 2006). 17 3.3 The use of telecare, telemedicine and telerehabilitation4 3.3.1 Telestroke There is a substantial literature on the use of ICT to support and enhance stroke care services 5. Applications broadly fall into the following categories: remote consultation between health professionals (called ‘telemedicine’ in this report), remote reading of scans (‘teleradiology’) and remote provision of rehabilitation support, for example through video links (‘telerehabilitation’). A ‘hub and spoke’ model is the most frequent type of telestroke system. In this model remote hospitals communicate with a centre of expertise. The data may be captured in a variety of formats (incl. for example video) and transmitted in real time to the hub (Cho et al. 2007). The patient’s clinical information and scans are then examined by a specialist. The decision as to whether or not to administer thrombolysis is jointly taken by the specialist and local physicians. Neurologists, neurosurgeons, nurses, programme administrators and technicians are part of the network. Existing telestroke networks are typically spread over distances of 160 to 400 kilometers (Deshpande et al, 2008). In the past ten years a number of pilot stroke networks supported by telemedicine have been established, notably the Telemedical Pilot Project for Integrative Stroke Care (TEMPiS) in Bavaria (Audebert, et al. 2006) and similar networks in the US (Wang, et al. 2003; Wang, 2003; Wang, et al. 2000; Hess et al. 2005). TEMPiS involves two specialised acute stroke centres which provide 24/7 consultation to twelve local hospitals via a two-way video conference system and CT/MRI image transfer with a high-speed data transmission. Each local hospital has specialised stroke wards with qualified stroke teams. The aim of the network is to administer thrombolysis to patients in rural areas at their local hospital within the 3 hour window. The Remote Evaluation in Acute Ischemic Stroke (REACH) programme in rural Georgia in the USA provides community rural hospitals with access to specialist neurologists who can 4 This section is partly based on research carried out by Erik Yves Landriault for his MSc research report on ‘Towards the Development of a Stroke System of Care. Implementation of a Regional Telestroke Programme in Malaysia.’ Imperial College Business School, September 2008. 5 Susman 1997; Levine and Gorman 1999; Shafqat et al. 1999; Gagliano et al. 2000; LaMonte et al. 2003; Gourlay et al. 2000; Dawson et al. 2001; Scheideman-Miller et al. 2001; Silverman 2001; Brennan et al. 2002; Piron et al. 2002; Richardson-Nassif et al. 2002; Tran et al. 2002; Johnston and Worrall 2003; Wang 2003; Wiborg and Widder 2003; Audebert et al. 2004; Buurke et al. 2004; Choi et al. 2004; Lai et al. 2004; Piron et al. 2004; Schwamm et al. 2004; Waite et al. 2004; Levine and McConnochie 2007; Schwab et al. 2007. 18 remotely assess patients, read CT scans and potentially guide non-stroke rural hospital physicians in administering thrombolysis. 3.3.2 Telerehabilitation There is growing interest in the use of virtual reality within stroke rehabilitation, driven partly by concerns that clinicians are not often able to provide adequate rehabilitative environments (O'Brien, 2008; Rizzo et al, 2004, Rizzo and Kim, 2005). Authors argue that such systems might provide performance feedback and support the learning process underlying rehabilitation. Virtual environments have been developed for a variety of rehabilitation purposes, including interventions for assisting with activities of daily living, way-finding in urban environments, wheelchair training and public transport navigation. Some feasibility trials have focused on post-stroke, home-based intervention. Lewis-Brooks (2004) describes a system whereby patients’ limb movements and body posture are tracked and converted into pleasant, abstract images and a melody to heighten activity, function and motivation. Should this approach prove beneficial, it is possible to envisage it also contributing to remote monitoring of at risk patients via a more conventional telecare system. The technology has also been used to train therapists in remote areas (Iwatsuki et al, 2004). Various reviews of the research literature have been published (e.g. Crosbie et al, 2007; Henderson et al, 2007). O’Brien (2008). 3.3.3 The potential benefits of telestroke – and the evidence Telestroke allows remote access to clinical advice. Many studies have recognized that telemedicine is very well suited to the evaluation of acute stroke, since most symptoms can be assessed through audio and visual information. Telestroke enables immediate remote access to an expert stroke diagnosis, the avoidance of costly air-transport of stroke victims, expert supervision of patients receiving thrombolysis, as well as specialist advice on secondary prevention and risk management. Telestroke improves the uptake of evidencebased medicine/ (Bahouth, 2003; Deshpande et al, 2008; Gorman and Levine, 1999; Audebert et al, 2006). The evidence base, however, remains inconclusive. The Canadian Agency for Drugs and Technologies in Health, which has undertaken the only systematic review of telestroke studies, found that the literature is contains many low quality observational studies. However, it also concluded that most studies suggested that telestroke is consistently associated with enhanced process of care, with improvements in health outcomes, including 19 reduced mortality and morbidity at six months post-stroke (Deshpande et al, 2008) (see box for examples). Improvements in stroke consultation and door-to-needle times, and a decrease in transfers of patients, imply cost-savings through reduced morbidity or avoidance of expensive care, the resource implications of telestroke remain unclear (Deshpande et al, 2008). This is due to a lack of data on how much it costs to implement or manage a telestroke system in mainstream use. Compared to telemedicine, there are few studies reporting on home telecare or telerehabilitation specifically for stroke patients. A systematic review of home telecare applications found 8666 studies (early 2007), of which 98 met the inclusion criteria (Barlow et al. 2007). None of these focused on stroke patients. Results from the TEMPiS project show that adverse outcome of thrombolysis after remote consultation are comparable to those reported in the National Institute of Neurological Disorders and Stroke trials, and that the service is cost-efficient. The network made thrombolysis far more widely available without diminishing the quality of care received by the patients. (Audebert et al, 2007). Numerous studies have analyzed the reliability of face-to-face stroke consultation in person versus a live video link (e.g. Chang et al, 1999; Gaul et al, 2003; Al-Khoury et al, 2005; Chacon et al, 2008). Remote examination of acute stroke patients is generally reliable according to these studies. For example, in a study of 153 telestroke patients in all but four cases both the local physician and remote neurologist were satisfied that the neurologist had contributed with relevant advice to enhance stroke management (Wiborg, 2003). These findings are consistent with research from the Massachusetts General Hospital where all physicians involved felt that telestroke improved care and 96% of physicians approved of the quality of telestroke (Hirschberg et al 2004). The REACH programme highlighted problems in administering thrombolysis via a telemedicine network, notably the lack of facilities and expertise in community hospitals to manage stroke patients appropriately throughout their stay and deal with complications. Many patients may therefore need to be transferred to a tertiary stoke centre after remote thrombolysis, raising the question of why they should not be transferred directly there (Crome and Bahr 2003). Since studies have shown benefits in rural settings of transporting stroke patients by helicopter to specialist centres, this may represent a better solution than telemedicine in certain circumstances. In the future, pre-hospital thrombolysis might become a further option. Recent innovations in mobile ultrasound imaging equipment may potentially support the administration of thrombolysis even before the patient arrives in hospital. 20 4 Case study: local stroke services 4.1 Background Our case study involved a district general hospital in London and the primary care trust and social services areas within which it was located. In order to explore the potential impact of the changes and innovations discussed in the previous section, we considered how local stroke care services were currently provided and where potential improvements could be made. We then explored the possible impact of selected improvements using simulation modelling (see section 5). 4.2 Current stroke care process – hospital There is no formal stroke pathway in the acute trust, but our interviews suggested that the following processes are in place: Patients present to A&E and if appropriate are referred to the stroke team (normally the stroke coordinator). If patients need specialist treatment that is not offered at the hospital, they are sent to the nearest trust offering such treatment. (Like the majority of acute trusts in the country, the trust does not offer specialist stroke treatments e.g. thrombolysis.) The patient is admitted to the stroke unit providing there is bed availability. At weekends the unit relies on the hospital’s site managers to put stroke patients into the stroke unit if there is a bed available. If there is no stroke unit availability, the patient is admitted to the Acute Medical Unit (AMU). Patients can stay in the AMU for a maximum of three days. If after three days there is still no bed available in the stroke unit, patients are transferred to other hospital wards. When a bed becomes available in the stroke unit, the most acute patient is admitted. For most patients the wait is 2-3 days, although outlying stroke patients in other wards can wait up to seven days to be transferred. Any outliers (i.e. patients in other wards) are visited only by the stroke co-ordinator and the therapists. The stroke consultants do not visit patients in other wards. However, at the time of the fieldwork this practice was about to change and consultants intend to visit all stroke patients in the hospital in the future. When patients first come into the stroke unit, the stroke coordinator makes sure that they are seen as soon as possible by a physiotherapist, an occupational therapist, a speech 21 and language therapist, and the dietician if necessary. The stroke coordinator also makes sure that CT scans, blood tests and any other necessary tests are ordered, and that patients are given aspirin or anything else prescribed by the stroke clinician. A scan is ordered from the radiology department. 95% of patients have the scan within a 24h period. Therapy starts on the day of admission if patients are fit, or as early as possible. There is no weekend cover for therapy. While in hospital, patients receive 45 minutes of physiotherapy per day for five days per week. They are seen by an occupational therapist within the first three days of arrival and then daily according to their needs. Patients continue to receive rehabilitation6 until either they stop achieving the goals set for them or until they have achieved them and are no longer making any progress. If there is still potential for progress, when they are discharged from hospital the community therapy team takes over. 4.3 Current stroke care process – community At discharge from hospital, if there is a need for further rehabilitation, patients are referred to the community’s intensive rehabilitation team7. Community based care is provided by a combined social and healthcare team. Relations between the acute trust, the PCT and social services were described by interviewees as good. A senior therapist from the community stroke team is the link between the acute trust and the PCT and attends the weekly multidisciplinary team meetings in the acute hospital. In this way, the community team is kept informed about the progress of the hospital’s stroke patients, especially those about to be discharged home so that an appropriate care package can be put in place. 6 The hospital stroke rehabilitation team comprises four physiotherapists (One Senior grade 1 (parttime i.e. two days a week), two Senior grade 2 full-time, and one Junior full-time), who cover the stroke unit and any other neurological patients in the hospital. But this is likely to change. There are two occupational therapists (One Band 7 and one Band 6) and a Band 5 therapist who rotates between different teams, e.g. medical, surgical, orthopaedic, old people and stroke care. There are also a full-time dietician, Band 7, and a speech and language therapist. In the acute trust, there are two multi-disciplinary team meetings every week, each led by one of the two stroke consultants in the unit, in which therapists and the consultants discuss patient progress. 7 The community stroke rehabilitation team consists of three physiotherapists (one Senior grade 2 fulltime and two part-time Band 7). There are two occupational therapists (one full-time Band 7 and one half-time Band 6). There is also one speech and language therapist and three assistants (two full-time and one part-time). 22 At the time of the research, interviewees perceived the local wait for community rehabilitation to be on average two weeks, but were aware that the figures were volatile because of service restructuring. Our subsequent analysis of PCT data showed waits to be frequently substantially longer than two weeks. With the introduction of early supported discharge, the PCT planned to bring this down to two days. There is no separate neuro-rehabilitation unit in the community. Patients discharged from hospital either go to their own home (where they receive rehabilitation from the community therapy team) or to a care/residential home. Patients requiring neuro-rehabilitation are referred to one of three units outside the borough. Patients who go into care or nursing homes are seen by the community rehabilitation team depending on their needs and potential for improvement. Patients with psychological or cognitive needs are also referred outside the trust (to either NHS or private hospital). 4.4 Identified problems and potential improvements 4.4.1 Stroke unit capacity There was agreement amongst the interviewees that that appropriate acute stroke management for the majority of patients will reduce long term disability, leading to major cost savings. Limited bed capacity in the hospital’s stroke unit due to beds not being ring-fenced for stroke patients was perceived to be a serious problem. Stroke patients admitted to other wards maybe under the care of non-stroke consultants. Research shows that specialised stroke units have better results (see section 3.1.2). The lack of a psychologist in the stroke unit (a Royal College of Physicians guideline) was another issue mentioned at interviews. ‘Our consultants are stroke specialists. And one consultant in the medical ward has stroke experience. But patients may on admission be put under a medical consultant without stroke experience. So it’s a matter of luck’. (Acute trust interviewee). 4.4.2 Scanning and other tests There is emergency cover for CT scanning 24/7, with radiographers or radiologists on call on evenings and weekends. However, scanning for stroke patients is not regarded as a routine emergency and it is likely to be performed out-of-hours only in special circumstances. There is one CT scanner in the acute trust which is very old and very slow, although there are plans to replace it. Currently scanning and other tests can take 2-3 weeks. The aim is to reduce this to one week although the local stroke strategy recognises the need for CT scanning ideally within three hours or a maximum of 24 hours. 23 ‘I know last year we had lots of incidents of the scan being broken. And it would be broken for a whole week. I remember sending patients to another hospital for scanning’. (Acute trust interviewee). 4.4.3 Thrombolysis The trust does not offer thrombolysis to stroke patients, but the possibility of offering it was discussed in the local stroke strategy meetings. Interviewees were in favour of thrombolysis and discussions were underway about how to plan a cost-effective service and supporting infrastructure. This includes establishing a co-ordinated regional approach to identify a specialist site to be ‘on take’ for stroke, and significantly improved scanning availability. ‘We don’t have thrombolysis here yet, and that’s something we are going to look towards but because we are a smaller district hospital, I don’t think we are going to get that within the next 3 years possibly. If we have that, we are going to need more equipment, monitoring etc.’ (Acute trust interviewee). ‘Yes, the clinicians are very keen on having a thrombolysis clinic but all this needs money and investment and knowledge about whether they are viable or not’. (Community interviewee) ‘A lot of emphasis is placed on thrombolysis but even if all patients presented at the hospital immediately after the stroke, only about 1/3 or less of them would be suitable for thrombolysis’. (Acute trust interviewee) 4.4.4 Transient ischaemic attack (TIA) Improvements could be made to the care of patients who have suffered a minor stroke or TIA. These patients often fall through the net because either the wait to see a consultant is too long, or they are sent back home without treatment. One solution would be TIA clinics and more generally improved prevention of stroke via better information about the dangers of stroke, and better management of patients at risk or who have suffered a TIA. ‘If GPs decide to refer patients e.g. to a cardiologist, of course they would have to wait until they get an appointment, whatever the waiting time is, it could be three months. And this is why we think that maybe we should have a TIA clinic, specifically for TIA patients so that we can manage them as quickly as possible’. (Community interviewee) 24 4.4.5 Length of stay In 2005, the local average length of stay (ALOS) for cerebral infarction was well above the national average (32 days compared to 25). The ALOS for all stroke patients was 27.1 against the national average of 23.6 days. For patients discharged to long term care (10.7%), the ALOS was 70 days. Length of stay in hospital is partly related to the provision of appropriate facilities and rehabilitation for stroke patients in the community, for which there is currently an average wait of two weeks. Delays in providing equipment or making home alterations can also increase the length of stay in the acute trust. One interviewee cited an example when it took about 4-5 weeks to move a patient’s furniture downstairs so they could return home. Improvements and changes to the discharge and rehabilitation process could be made locally. 4.4.6 Early supported discharge (ESD) An initial ESD pilot was not successful because the selection criteria8 made it hard to identify suitable patients. However, the acute trust and the PCT are trying to develop an ESD service for suitable patients. The local stroke strategy aims to Enhance the hospital’s stroke multi-disciplinary team by developing a stroke therapy team managed by a lead clinician. Enhance the community therapy team in order to support ESD (aiming for a two day wait from hospital discharge) and enable continuation of treatment to be needs led. Explore need for a Community Matron service for patients with complex needs. Provide training and development for the intermediate care at home team. Appoint a Stroke Association Family Support Worker to focus on the transition into community. 4.4.7 Rehabilitation Evidence suggests that in the post-acute phase the rehabilitation should be provided by teams specialised in stroke care in a non-hospital environment, either the patient’s home or a community rehabilitation centre. Our interviewees agreed. ‘We have intermediate beds but maybe we need a mini hospital out there, not as a hospital but a community place where people can go even in the interim, maybe they need further 8 The criteria were that patients needed a Barthel score of 9 or 10, were living in the local area and had no major cognitive problems. 25 rehab but they don’t need to stay in the acute hospital. Something for when they are not quite ready to go home, or when you have these big delays’. (Acute trust interviewee). At the moment there is no such centre in the area, and the community rehabilitation intensive therapy team treat all neurological conditions. According to the local stroke strategy, the reason for not having specialist rehabilitation stroke team is the low number of patients and available resources. The local strategy suggests that the preferred long term aim would be to transfer the rehabilitation element of stroke recovery from the hospital into a separate unit in the community offering specialised rehabilitation for stroke patients. This would make it easier to provide ESD. Such a step would involve major reorganisation of services with implication for staff posts, and impact on finances. The strategy also identifies other opportunities for improving long term support and rehabilitation in the community such as: Appointing a Stroke Association Family Support Worker and development of stroke clubs/support groups, aphasia groups and services by working with existing voluntary sector organisations. Working with local agencies to create more specific exercise opportunities for stroke patients and developing transport options for attending existing opportunities. 4.5 Data analysis of local stroke services The local hospital we studied improved from a difficult situation over time. Hospital episode statistics data shows that lengths of stay reduced by about 20% from 2002/3 to 2005/6 (figures 1 and 2). It is notable that over the same period of time the share of admissions classified as suffering from a non-specified type of stroke halved from over 40% to less than 20% (figure 3). This indicates improvements in the diagnosis of patients. 26 Figure 1: Length of stay haemorrhage patients Figure 2: Length of stay infarction patients 27 Figure 3: Hospital admissions by type At the interface between the hospital and the community long waiting times for rehabilitation were observed (figure 4). This indicates that the implementation of early supported discharge as an intervention to shorten length of hospital stays will require adequate resourcing and careful management. Figure 4: Waiting time between discharge and therapy start 28 For patients who receive rehabilitation services in the community, we compared estimates of hospital costs with rehabilitation costs. Estimates for hospital beds are based on the number of bed days in hospital for each patient and the average national bed day cost for stroke patients derived from the National Schedule of Reference Costs - NHS Trusts (Department of Health, 2006). Rehabilitation costs are based on the number of rehabilitation hours received for each of the patients and an hourly cost estimate which was derived using the PSSRU Units Costs of Health and Social Care (PSSRU, 2006). Clearly, these costs are not representative for all stroke patients. Patients with minor stroke not requiring rehabilitation are likely to have shorter hospital stays. Patients who die in hospital or those severely disabled and requiring long-term care will also differ in average length of hospital stays. The comparison of costs for stroke patients in hospital and in the community (figure 5) shows that rehabilitation cost are relatively small if compared to the hospital costs. Figure 5: Care costs for patients receiving community rehabilitation 29 5 Modelling 5.1 The planning challenge in healthcare While service innovations have often been judged desirable or positively evaluated in trials, their translation into mainstream practice is often difficult. Innovations whose implementation affects many actors across the health and social care system and which require significant changes in several organisations have often proved particularly challenging. Organisational, cultural and funding barriers within the care system, as well as the lack of shared knowledge about the potential impact or benefits of these innovations, make whole systems thinking difficult. Evidence-based care models, consistent with well-established guidelines, are important for widespread acceptance and mainstream adoption of innovation in service delivery. Gathering evidence for the impact of innovations on care organisations or a local care economy is frequently challenging, though pilot projects may provide evidence of the impact of an innovation on the individual and clinical level. However, the wider effect of its mainstream use is often difficult to judge, especially when substantial organisational changes, such as alterations in working practices, are required. Planning new services – perhaps on the basis of new opportunities opened up by technological innovation – requires the differing needs of stakeholders to be understood and addressed. Delivering effective care services to patients frequently requires the coordination of the activities of several providers: GPs, other primary care services including rehabilitation, acute hospitals and potentially social care providers might all need to come together to meet the needs of an individual patient. Planning such services is difficult. Communication and the coordination of the flow of resources between many stakeholders across organisational boundaries needs to be managed. The modernisation of stroke care exemplifies the challenges of whole systems planning in an evolving care system, where the locus of care is shifting out of secondary care, new private sector providers are becoming available and organisational relationships are changing. In health planning, rational decision-making may be highly desirable but experience shows that fixed, rigid decision-making processes often do not work because the context changes quickly. New policies are introduced and new targets mean attention needs to be focused elsewhere. Local decision-makers change frequently and new stakeholders have to be brought on board. Local political needs, complex and sometimes perverse financial 30 incentives, differing values between professional groups from different parts of the system (such as social services and acute care) can lead to open or hidden conflicts and misunderstandings. In arriving at decisions, not only is the factual accuracy of the assumptions and the rationality of the decision important, but also the decision process. Processes which give some degree of structure and rationality, by highlighting uncertainties, encouraging stakeholder dialogue, supporting and documenting the decision, are therefore desirable. 5.2 Modelling in healthcare planning In order to understand the effect of implementing a service delivery innovation, modelling approaches are often appropriate. Modelling has the potential to support both local and national or regional planning of service delivery. Since the 1960s, many different approaches, including spreadsheets and simulation models, have been used in healthcare planning. Over recent decades the amount of work in the area has increased substantially (Brailsford, 2008; Pitt et al., 2008). More powerful computers, more accessible software and growing capabilities in the computer graphics and animation area, have helped to spread the use of simulation modelling. While many models have been developed for planning purposes in health care, successful and sustainable use of modelling remains a concern (Brailsford, 2005; Pitt et al. 2008). Modelling in health care planning often occurs in situations where understanding has to be created across organisational and professional boundaries. This makes it necessary to take into account a larger scale of system than that which individual stakeholders are located in. ‘Whole systems thinking’ often needs to consider the consequences of decisions for social, primary and acute care as well as different providers and funders. However, creating crossdisciplinary understanding in healthcare settings is difficult (Ferlie et al, 2005). Simulation modelling can be used as a tool to facilitate whole systems planning by helping to explore the consequences of decisions across the system (including unintended consequences, bottlenecks and delays, distribution of costs and benefits). In addition modelling can support and facilitate communication between diverse stakeholders by creating a shared representation of the whole system. Models can also be used to make predictions about outcomes in the real world and allow experimentation with different courses of action in a safe, quick and cheap way. However, as has been shown in the case of engineering (Dodgson et al., 2007), simulation modelling can also help to shape the conversation between stakeholders in problem solving and foster collaboration. In the modelling literature a tension between using modelling for representing 31 reality and for negotiating mutual understanfing has been identified (Zagonel, 2002). In health care applications these different roles of modelling are visible and vary with the purpose, the accessibility of the model and the modelling process as well as the type of modelling approach and the preferences of those involved in the modelling process. The role a simulation model can have in a decision process in health care is constrained by the degree to which the model and its results are accessible to stakeholders such as clinicians and managers. The visual representation of the model and its results are likely to be of a major influence on the accessibility of the model to a non-specialist audience. Simulation modelling can be used to experiment with different options and assess the consequences of different choices on the behaviour of the system. Options which can be assessed include the availability of alternative treatments, capacities (e.g. number of beds, available scanning slots or therapy hours), rules for allocating treatment and resources to patients. Aspects of the system behaviour which could be of interest include costs, waiting times or care outcomes (e.g. independence, disability or death). However, the insights at the system level will depend on the understanding and availability of data on the relationships in the system – without this, insights may be rather limited and possibly misleading. Simulation modelling supports sensitivity analyses which can help deal with the uncertainty in available data that characterises much of health and social care research. It allows investigation of ‘what if’ scenarios for mainstream implementation of a new service or intervention (e.g. what if more training was available or more resources provided during the rehabilitation phase; what if the demand for services increases?). Not only do models allow the creation of a range of projections to illustrate possible impacts of innovation implementation, from worst-case to best-case scenarios, but they also allow comparison of different implementation paths. There are also benefits from the model development process itself, where participants are able to use the framework to gain a better understanding of implementation issues in the context of the larger system. Simulation modelling therefore offers opportunities for contributing to the implementation of complex innovation in health and social care. This is especially useful when a static comparison of different approaches of care delivery – for example through traditional cost-benefit studies exploring the effect of implemented changes against the current system – is unable to fully capture the effects of a gradual change towards a new approach. Among the major approaches to simulation modelling in healthcare system dynamics and discrete event simulation are particularly prominent, other approaches include Monte Carlo simulation and agent-based modelling. 32 In system dynamics flows in the system as opposed to individual patients are modelled. This approach is therefore best suited to problems where the relevant behaviour of the system is less influenced by what happens to the individual patient and more by influences on an aggregated level. In particular, the effect of feedback and delays within the whole system can be particularly well represented and studied (Wolstenholme 1993). Indeed, system dynamics can be argued to be the origin of the emphasis on whole systems in health care planning (Wolstenholme, 2005). System dynamics models are often used to analyse strategic questions, often in a quick and intuitive way. System dynamics can be used to study the whole systems effect of an innovation as it is implemented and becomes effective over time. It is ideal in cases where a static comparison of different models of care delivery is not enough and the time dimension has to be taken into account (Bayer et al 2004) and also when several types of intervention are to be evaluated together (Homer et al 2003; Hirsch et al, 2004). Discrete event simulation has a more disaggregated focus than system dynamics and is more suited for detailed, operational models. Applications of this modelling approach typically emphasise the journey of individuals through the care system. In discrete event models emphasis is often placed on the effects of random variations (e.g. of arrivals of new patients, treatment durations etc.). Such stochastic effects are less often considered in system dynamics models. Often discrete event simulation models are used for decision making about care operations, as opposed to strategy and policy. This type of simulation has been used for many different healthcare applications including the organisation of specific clinics such as a vascular-surgery (Dodds, 2005) or emergency department activity (Connelly and Bair, 2004), care service innovation such as intermediate care (Kotiadis, 2004) or the evaluation of screening programmes (Davies et al., 2002; Brailsford et al. 2006). While system dynamics models favour a visualization of systemic relationships such as feedback loops, discrete event simulation lends itself to a visualisation of the journey of individuals. 5.3 The prototype stroke care model 5.3.1 Introduction In order to develop the prototype stroke caremodel9 we have chosen to use a discrete event simulation approach. This was chosen over system dynamics because though our engagement with the partners in our case study it became clear that at any point the care 9 The model can be inspected at http://www.imperial.ac.uk/business/dynamic/sbayer/stroke.html 33 provided for a small number of patients had the potential to significantly impact on the system as a whole. For example,a patient admitted to the hospital as an emergency typically might be the only patient that day, the further journey in the hospital of that patient would depend on the availability of a bed in the stroke unit and/or the ability to scan that patient within an appropriate period of time. It is therefore important whether a single bed is available at a specific point in time – averages of occupancy over a period of time are not enough for understanding the process. Discrete event simulation allows us to follow the journey of single patients, while system dynamics typically works on a more aggregated level. In this example, it is important for the strategic understanding of the problem to take that individual and their journey into account. For other health and social strategy discussions this might not be the case and a more aggregated treatment, including through system dynamics modelling, could be more appropriate for supporting strategic or policy decision making. 5.3.2 Implementation Our model was developed using the simulation package Anylogic10. This general purpose simulation the software package allows the building of models in different application domains and permits following different modelling paradigms including discrete event simulation. It should be emphasised that the model is a prototype which needs to be adjusted to the specific circumstances in a locality11. The assumptions made in the model are appropriate for demonstrating the model’s features. However, it would be misleading to use this prototype for planning services without adapting it to the local circumstances. The insights of local stakeholders in particular those of managers and professionals from the acute, primary, secondary and voluntary sector need to be drawn together (through workshops and in interviews) and their buy-in is required for successful implementation. A model developed in collaboration with local stakeholders could then be used by commissioners to plan and implement stroke services. While the prototype does not predict precise service utilization due to the randomness of service demand and of outcomes for individual patients, it will help to make more informed decisions about the structure of the service and capacities. An adaptation of the model to a specific locality could then contain the particular features of the local services, including the 10 The package has been developed by XJ Technologies (http://www.xjtek.com/) and is distributed in the UK by Saker Solutions (http://www.sakersolutions.com/). 11 We are currently investigating the possibility to take this work forward in a locality and develop a tailored model which is based on local needs and validated to local data. 34 relationship to the wider regional setting as appropriate. Additional detail could include the separation of a hyperacute stroke unit (for the first 72 hours of treatment), inclusion of several institutions of a specific type (e.g. acute hospitals, rehabilitation centres, care homes), the distinction of several types of rehabilitation services, detailed consideration of issues around the transport of patients to A&E, TIA services or a more explicit consideration of workforce capacity and financial issues. 5.3.3 Features of the prototype The prototype (see figure 6) encompasses the main features of the journey of patients from the moment of the stroke occurrence. It includes treatment in the hospital as well as rehabilitation in the community and is a simplified representation of an acute hospital, the community rehabilitation service and care home provision. Depending on the policy option under discussion, other details of the system or additional elements could be included. A balance has to be found between the level of detail required to support decision making and implementation in a specific case and simplicity to facilitate understanding. An overly complex model is less likely to achieve buy-in and understanding by decision makers and other stakeholders, and might require too much effort to collect the underlying data. Figure 6: Model overview 35 The prototype distinguishes different types of strokes (infarction and haemorrhage). Incidents with either of the two diagnoses are distinguished in the animation using different colours (blue for infarction and red for haemorrhage). Stroke patients arrive at the hospital after a certain delay (this represents the time taken for the ambulance to be alerted and the patient to be transported to the hospital). Once admitted to hospital, stroke patients are put on a waiting list for scanning. Depending on scanner availability, they are scanned as soon as possible (differentiating scenarios with or without scanning outside standard working hours and accounting for the likelihood of scanners not being available is possible). For each patient the delay from stroke incident to hospital admission and scanning is recorded. The prototype allows us to experiment with the introduction of thrombolysis. In a scenario where thrombolysis is available, the following rule is applied: if a stroke patient with an infarction is scanned within the appropriate time window of three hours, thrombolysis is administered. In the prototype animation patients which have undergone thrombolysis are distinguished by a different colour (yellow). After the initial period in A&E, stroke patients are placed in the stroke unit if a bed is available. If a stroke unit bed is not available, the patient is placed on a different ward as an outlier and then transferred to the stroke unit as soon as a bed there becomes available. Beds in the stroke unit can be used for non-stroke patients if pressures dictate. The capacities of the stroke unit can be changed. The prototype also allows us to experiment with ring-fencing the stroke unit (to prevent the assignment of patients without stroke to those beds). Death rates, length of stays and outcomes (high vs. low level of disability on discharge) are treated probabilistically. They also vary for patients who have been thrombolysed or have spent more time in the stroke unit. Depending on the level of disability at the end of their hospital stay, patients are (with a certain probability) discharged to care homes or their own home with or without rehabilitation in the community. The waiting period for rehabilitation at home depends on the therapy capacity in the community. For different scenarios, it is possible to change the number of therapists who can deliver rehabilitation in the community. The prototype shows for each moment in time how many patients are in various parts of the hospital (A&E, stroke unit, outliers in other wards) and in various locations after discharge 36 (care home, living in the community, in the community receiving rehabilitation services). Outputs are produced showing performance indicators such as length of stays, percentage of patients having been scanned within 3 and 24 hours, percentage of patients having spent more than half their stay on the stroke unit. These indicators have been chosen to reflect the current concerns of policy and research in stroke care. 5.3.4 Parameters and data The data we used was extracted from existing records (community rehabilitation records) as well as local and national statistics (hospital episode statistics). We had local data for all those who received rehabilitation in the community provided by the PCT. We made the assumption that the gap between this number and those discharged from the hospital to their usual place of residence was made up of people not receiving rehabilitation. We assumed a Poisson process for the stroke incidences. Parameters (see table 1) describing the shortening of length of stays and the reduction of disability due to stroke unit treatment, early supported discharge and thrombolysis were based on the medical literature. Other parameter choices are based on data from the case study as well as some estimates. Better data both within and outside the acute setting would be desirable and would improve the reliability of the model and the granularity of our results. Table 1: Model parameters Parameter Value Sources Share of stroke patients having infarction 0.83 Rosamond et al, 1999 Reduction of LOS due to stroke unit (days) Standard hospital death probability 4 Stroke Unit Trialists’ Collaboration, 2007 0.32 Standard length of stay on general ward 27 Share of several disabled stroke patients after standard treatment Share of not-severely disabled going to care home Share of severely disabled stroke patients going to care home 0.24 Health Episode Statistics (2004/5) adjusted for stroke unit coverage and stroke unit effect Health Episode Statistics adjusted for stroke unit coverage and stroke unit effect Candelise et al., 2007 0.15 estimate 0.75 Probability of community rehabilitation 0.56 calculated from other parameters and discharge destinations found in Health Episode Statistics estimate based on case study data Effect of stroke unit (fraction of severe disability compared with general ward) Effect of stroke unit on death rate 0.79 Candelise et al., 2007 0.69 Candelise et al., 2007 37 Reduction of LOS due to ESD (days) 8 Early Supported Discharge Trialists, 2005 Death probability after thrombolysis 0.18 Wardlaw et al. 2003 Effect on share severely disabled after thrombolysis LOS thrombolysis patients 0.82 calculated based on Cochrane 3h thrombolysis Fagan et al, 1998 Weekly therapy slots per therapist 18 Duration of community rehabilitation 42 days 2 Average weekly sessions per client 11 calculated based on assumptions in Personal Social Services Research Unit (2006) taking into account annual leave and sickness etc. local data from case study local data from case study 5.3.5 Experiments with the prototype The prototype allows comparative scenarios for the implementation of different service modification both in hospital and the community. The policy choices which can be simulated include: Availability of thrombolysis Out of hours scanning and vailability of 24h coverage for the interpretation of brain scans (e.g. through teleradiology) Changed capacity of stroke unit Ring-fencing of stroke unit to prevent inappropriate admissions Telecare to prevent admissions to care homes Changed size of community rehabilitation team Telerehabilitation to increase community rehabilitation capacity In addition other parameters reflecting the consequences of some more indirect or complex measures such as shorter travelling times to hospital (e.g. due to changes in the ambulance service) or reduced waiting times for care home admissions can also be adjusted. During a run the model shows an animation of the flow of patients through the care system (figure 7). The prototype output includes statistics on activities which give rise to costs (e.g. bed days in stroke unit, general ward and the care home, scans performed and thrombolysis administered, community rehabilitation sessions provided, number of patients receiving telecare), as well as capacities whose provision incurs costs (stroke unit beds, therapists employed in the community). Based on averages of this activity and capacity data 38 over a large number of runs it is possible to arrive at estimates of the costs arising for the acute and the community sector under the different scenarios. Figure 7: Animation of patient flows 5.3.5.1 Base case In the base case thrombolysis is not available. There are capacity pressures on the (nonringfenced) stroke unit and on community care. (see table 2) Figure 8 shows the modelling output from the base case scenario. Table 2: Base case parameters Parameter Value Comments Average yearly stroke cases 365 Delay between stroke and presentation at hospital 2h Stroke unit capacity 22 ESD coverage 0 share of patients receiving early supported discharge support Average inappropriate admissions to stroke unit 0.25 dependent on bed management practices Length of stay of patients without stroke in stroke unit 4 dependent on bed management practices Start working hours - scanning 8 End working hours - scanning 17 Probability of scanner availability during working hours 0.8 dependent on competing demands, prioritisation and equipment breakdowns Percentage diverted by telecare 0.1 scenario depends on type of telecare package dependent on geography and ambulance service 39 Waiting for care home 7 days Number of therapists available 5 Effect of telerehabilitation on rehab capacity 0.1 dependent on discharge practices and local care home place availability dependent on technology used Figure 8: Base case scenario 5.3.5.2 Improved acute care scenario In the literature, treatment in a stroke unit has been identified as being preferable to treatment in a general ward both in terms of outcomes and the length of stay. The prototype takes these effects for the individual patient into account. In order to allow more patients to stay in the stroke unit for most of their time in hospital two measures can be undertaken: increasing the capacity of the stroke unit and preventing inappropriate admissions of patients without stroke to stroke beds. However, the prototype model of the stroke care system captures only a small part of the total care system. An intervention like ring-fencing the stroke unit such that patients without a stroke are not admitted would have consequences elsewhere in the hospital which are not covered in our model. Such consequences might include the impact on A&E waiting times if non-stroke patients can not admitted to the stroke ward if no other bed is available. 40 In this scenario we introduce thrombolysis and increase the size of the stroke unit by 20% compared to the base case and prevent inappropriate admissions to the stroke unit (see figure 9). The remaining parameters are chosen like in the base case. Figure 9: Improved acute care scenario 5.3.5.3 Telestroke scenario Three areas where telemedicine can be used in the system have been identified in: teleradiology to ensure permanent availability of expert guidance for the interpretation of brain scans even if no radiologist or other expert is available locally, telerehabilitation for some aspects of community rehabilitation work and telecare to support people after a stroke in the community. The model allows the effects of these innovations on the stroke care system to be studied. Clearly, there exists considerable uncertainty around these interventions. This constrains the possibility of predicting precise effects and makes an approach which involves stakeholders in exploring the possibilities through experimentation potentially more useful. This is particularly so for the case of telerehabilitation for which only very limited experience has been reported in the literature. 41 Teleradiology is captured in the prototype as a possible way of achieving a 24h scan interpretation capability; this could of course be also achieved by having staff available locally. Telecare is included in as far as it allows diverting some of the patients destined for a care home to the community. Telerehabilitation is conceptualized here as a way of improving the number of patients which can be seen by a given number of therapists (within a given number of time slots) since some face-to-face sessions with patients can be replaced by remote interaction. The telestroke scenario (see figure 10) illustrates the possibilities: we assume 24h scanning interpretation is covered by teleradiology, 20% of clients are diverted from a care home by telecare support in the community and that the rehabilitation capacity is increased by 10%. In this scenario we also assume that thrombolysis is available. The remaining parameters are chosen like in the base case Figure 10: Telehealth scenario 42 6 Discussion A large number of stakeholders have to come together to enable service integration. Some improvements can be made directly in the District General Hospitals (e.g. in relation to rapid diagnosis, the acute treatment available, the intensity of rehabilitation or the use of stroke units), while others require the collaboration of acute service providers with the health and social care services in the community. Early supported discharge requires that appropriate services are available in the community and that the interface between acute provision and community provision is carefully managed. Similarly, stroke prevention by identification of high risk groups (in particular those that have suffered a transient ischaemic attack) is likely to involve collaboration between stroke consultants in the acute trust and primary care providers (see figure 11). Figure 11: The need for service integration The provision of specialist acute or rehabilitation services (under some services models this will include thrombolysis) in regional specialist centres will require a wider reorganisation of stroke services. Close coordination with ambulance services is required to shorten the time from the incidence of stroke to provision of treatment in the appropriate acute facility. 43 6.1 The need for service integration Different innovations in stroke care will vary in the extent to which they require integration into the wider care system, with variation not only with regard to their integration requirements but also as to how narrow or specific the innovation is in relation to the conditions it is applicable to. The success of innovation implementation will also depend on availability of resources, the alignment (and attention) of stakeholders, the appropriate redesign of pathways as well as the supporting infrastructure (see figure 12). Figure 12: Types of stroke innovations Like other changes to stroke services, telecare and telemedicine interventions have to take these factors into account. Telemedicine (e.g. teleradiology to support remote thrombolysis) still needs appropriate systems around it in the district general hospital (e.g. urgent access to CT scanning, dedicated stroke care units, weekend cover) and cannot therefore just be “dropped in” as an easy solution in order to achieve better services. Care has to be taken to get the basics right before introducing telemedicine. The requirements for embedding ICT44 based technologies in the support of rehabilitation (e.g. speech and language therapy or the monitoring of exercises therapy via video link) into a wider service are likely to be comparatively less challenging. Telecare for frail elderly people as a more generic intervention will also be a valuable aid to support independent living for those who have suffered a stroke: stroke patients with mild or moderate disability can benefit from the same systems of telecare for the elderly. Clearly, patients with severe disability need more hands-on support and telecare can at most been an add-on to a more comprehensive package of care. 6.2 Regional reorganisation of stroke services When taking this work forward in an applied setting thought needs to be given to how embedded local services are in the regional setting. Our prototype model illustrates the use of this approach by outlining the case of a simple care system comprising one hospital and the community sector. Currently, there are several potential models for the regional organisation of stroke services under discussion (see figure 12). While the current model generally provides treatment and rehabilitation in the district general hospital nearest to where the stroke occurred, alternative service models coordinate and concentrate services on a regional basis. If the patient is taken to the nearest hospital travelling times are short resulting in faster treatment. However, the quality of care received will vary with services available at the particular hospital at that time which will depend on factors such as the specialist and equipment available, bed availability in the stroke unit, or the extent of out of hours service. 45 Figure 13: Regional reorganisation of stroke services An alternative service model would be to take patients to a specialist centre for the acute phase of stroke, after which sub-acute treatment and rehabilitation would be given in the local district general hospital. This model would help to improve equitable access to specialised care (including thrombolysis). However, because the time window for thrombolysis treatment is very short, the increased travelling times for acute admissions might become an issue especially if distances to the next specialised centre are large or traffic is congested. A second alternative service configuration would be to provide all acute services in a specialist neurological centre while rehabilitation would be provided at home or - depending on disability and need - in a specialist rehabilitation centre. Again travelling times for acute admissions might be a problem with this configuration. A third alternative would be to provide acute treatment in the nearest district general hospital (like in the currently predominant model), but to provide back-up to district general hospitals via a telemedicine link to the specialist neurological centre. In this way procedures such as thrombolysis can be delivered locally drawing on expert advice. Issues to consider include the 24h cover which will still be required in the district general hospital as well as the needs of patient requiring specialist care which can not be delivered in the district general hospital even with telemedicine back-up. 46 The model of the regional organisation of stroke care will need to be considered when applying the modelling approach to planning stroke services in a particular location. These considerations go, however, beyond what we considered to be the scope of this project. 47 7 Conclusions and recommendations The UK government recognises the need for attention to be given to stroke care. The National Stroke Strategy (December 2007) emphasises the urgent need to change many aspects of stroke care, from patient awareness and prevention to appropriate acute phase treatment and to long-term rehabilitation and care in the community. The Strategy also places emphasis on the need for integration of services and close collaboration of all those involved in the delivery of care. Major technological and scientific advances in the last 20 years offer the potential for lifesaving interventions in stroke patients. However, significant challenges still exist which prevent the practical implementation of such advances. It is necessary to first identify these challenges and secondly to re-organise stroke services in a way which takes full advantage of recent scientific developments. Our study has contributed to a better understanding of some of these barriers and the challenges facing stroke service restructuring or the introduction of telemedicine and telecare solutions. We used a case study research design, focusing on one particular locality in London. We examined the literature on technology and service delivery innovations for stroke care, conducted interviews with key local stakeholders and experts in the field and collected and analysed primary data from the community. We captured through simulation modelling the current stroke pathway (in its acute and community settings) and built scenarios of changes in the pathway and their potential impact on outcomes (clinical and costs). This type of modelling is suitable for application to highly complex processes, such as stroke care, and can be used successfully as a communication and decision making tool before committing real resources. We selected stroke care as an example of chronic disease management because, as a leading cause of adult long-term disability, stroke poses a huge demand on health and social care resources. The key insights from this study can be summarized as follows: Stroke is a complex disease which needs widespread service integration. Within the current fragmented services, achieving such integration poses a significant challenge for service planners. It involves examining various alternatives and assessing their likely impact before deciding to commit real resources. 48 The potential benefits of introducing telemedicine and telecare in assisting the delivery of stroke care have been a subject of widespread research and debate. During the course of this research we have come to the conclusion that if telecare is to be used to its maximum advantage in supporting patient care, it cannot be just ‘dropped in’ as an easy solution on existing highly fragmented services. The success of introducing innovations in healthcare depends on the availability of resources, stakeholder agreement, careful pathway design, and appropriate supporting infrastructures. Introducing technological innovations such as telemedicine/care in stroke care primarily requires the creation of a relatively stable service configuration. Modelling can help to determine what such a configuration might look like. However, the development of a model in itself raises a number of interesting and challenging issues. To develop a robust model necessitates the collection of appropriate data and information on which to ground the model. It was our experience that the collection of this information was problematic. Frequently the requisite information was unavailable or not available in an appropriate format. The exercise of model development in itself highlights problem areas and issues of paucity of data. Highlighting such areas has a value in its own right and is of relevance to health care professionals, policy makers and commissioners. Modelling helps to map out the care journey (thereby helping to develop a shared understanding of different stakeholders) and to compare alternatives before committing real resources. Modelling can in this way both support stakeholder communication and decision making. Acknowledgements We are grateful to the Department of Health for their support in funding this research. We also acknowledge the contribution of EPSRC for funding James Barlow and part of Steffen Bayer’s time via the Health and Care Infrastructure Research and Innovation Centre (HaCIRIC). We thank Erik Yves Landriault for use of material on telestroke from his MSc dissertation. Finally, we wish to thank our advisory board and all the participants at the case study site. Favourable ethical opinion obtained from the local research ethics committee (reference number 06/Q0707/70) on October 31st, 2006. 49 8 References Al-Khoury, L., J. Beer, Y. Cheng, R. Fellman, P.D. Lyden, B.C. Meyer, R. Raman, R. Rao, J.A Zivin. (2005). Prospective Reliability of the STRokE DOC Wireless/Site Independent Telemedicine System. Neurology. 64: 1058-1060. Anderson, C., E. Bautz-Holter, et al. (2005). Services for reducing duration of hospital care for acute stroke patients. Cochrane Database of Systematic Reviews (2). Audebert, H. J., C. Kukla, et al. (2005). Telemedicine for Safe and Extended Use of Thrombolysis in Stroke: The Telemedic Pilot Project for Integrative Stroke Care (TEMPiS) in Bavaria. 36: 287-291. Audebert, H. J., C. Kukla, et al. (2006). Comparison of Tissue Plasminogen Activator Administration Management Between Telestroke Network Hospitals and Academic Stroke Centers: The Telemedical Pilot Project for Integrative Stroke Care in Bavaria/Germany. 37: 1822-1827. Audebert, H. J., M. L. Wimmer, et al. (2004). Telemedicine stroke department network. Introduction of a telemedicine pilot project for integrated stroke management in South Bavaria and analysis of its efficacy [German]. Nervenarzt 75(2): 161-5. Audebert, H., Bogdahn, U. , Fürst, A., Horn, M., Kukla, C., Hauchwitz, M., Schwab, S., Vatankhah, B. (2007) Long-Term Outcome After Thrombolysis in Telemedical Stroke Care. Neurology. 69: 898-903. Audebert, H. (2006) Telestroke: Effective Networking. Lancet Neurology 5: 279-282. Bahouth, M., Crarey, P., Gunawardane, R., Hu, P., La Monte, M., Page, W., Pathan, M., Yarbrough, K. (2003) Telemedicine for Acute Stroke: Triumphs and Pitfalls. Stroke. 24: 725728. Barlow, J.; Singh, D.; Bayer, S.; Curry, R. (2007) A systematic review of the benefits of home telecare for frail elderly people and those with long-term conditions. Journal of Journal of Telemedicine and Telecare; 13: 172–179. Bayer S, Barlow J and Curry R (2007) Assessing the impact of a care innovation: telecare. System Dynamics Review, 23(1), pp. 61-80. Blight, A., A. Pereira, et al. (2000). A single consultation cerebrovascular disease clinic is cost effective in the management of transient ischaemic attack and minor stroke. Journal of the Royal College of Physicians of London 34(5): 452. Brailsford SC, V.A. Lattimer, P.Tarnaras and J.A. Turnbull (2004), Emergency and OnDemand Health Care: Modelling a Large Complex System, Journal of the Operational Research Society, 55, 34-42. Brailsford SC (2005). Overcoming barriers to the implementation of OR simulation models in healthcare. The Journal of Clinical Investigative Medicine, Vol 28 No 6, 312-315. Brailsford S, Rauner M, Gutjahr E and Zeppelzauer W (2006). Combined Discrete-Event Simulation and Ant Colony Optimisation Approach for Selecting Optimal Screening Policies for Diabetic Retinopathy. Computational Management Science, 4:59-83. 50 Brailsford, SC (2008). Advances and Challenges in Healthcare Simulation Modeling. Proceedings of the 2007 Winter Simulation Conference, Washington, DC, ed S. G. Henderson, B. Biller, M.-H. Hsieh, J. Shortle, J. D. Tew, and R. R. Barton, pp 1436-48. Brennan, D., A. Georgeadis, et al. (2002). Telerehabilitation tools for the provision of remote speech-language treatment. Topics in Stroke Rehabilitation 8(4): 71-8. Buurke, J. H., R. F. M. Kleissen, et al. (2004). A feasibility study of remote consultation to determine suitability for surgery in stroke rehabilitation. Journal of Telemedicine and Telecare 10(2): 108-12. Candelise, L., Gattinoni, M., Bersano, A., Micieli, G., Sterzi, R., and A Morabito, (2007) Stroke-unit care for acute stroke patients: an observational follow-up study, Lancet 369: 299–305. Chacon, M., M. Jensen, B. Meyer, R. Raman, J. Werner. (2008) Reliability of SiteIndependent Telemedicine when Assessed by Telemedicine-Naïve Stroke Practitioners. Journal of Stroke and Cerebrovascular Diseases. 17, 4: 181-186. Chang, Y., M. Guanci, J. Kvedar, L. Schwamm, S. Shafqat (1999) Role for Telemedicine in Acute Stroke: Feasibility and Reliability of Remote Administration of NIH Stroke Scale. Stroke. 30: 2141-2145. Cho, S., Hess, D., Khasanshina, E., Mathiassen, L., Stachura, M., Wang. S. (2007) An Analysis of Business Issues in a Telestroke Project. Journal of Telemedicine and Telecare. 13: 257-262. Choi, J. Y., A. W. Wojner, et al. (2004). Telemedicine physician providers: augmented acute stroke care delivery in rural Texas: an initial experience. Telemedicine Journal and e-Health 10(Suppl 2): S90-4. Connelly LG and Bair AE (2004). Discrete event simulation of emergency department activity: a platform for system-level operations research. Acad Emergency Med, 11: 1177– 1184. Crome, O. and M. Bahr (2003). Editorial Comment--Remote Evaluation of Acute Ischemic Stroke: A Reliable Tool to Extend Tissue Plasminogen Activator Use to Community and Rural Stroke Patients? 34: e191-192. Crosbie, J., Lennon, S., Basford, J., McDonough, S. (2007) Virtual Reality in Stroke Rehabilitation: Still More Virtual that Real. Disability and Rehabilitation, 29(13):1139-1146. Davies, R., P. Roderick, D Crabbe, J Raftery, P Patel and JR Goddard (2002) A simulation to evaluate screening for helicobacter pylori infection in the prevention of peptic ulcers and gastric cancers, Health Care Management Science, 249-258. Dawson, S., P. G. Clark, et al. (2001). Case studies of stroke rehabilitation in rural settings [abstract]. Telemedicine Journal and e-Health 7(2): 131. Department of Health (2001) National Service Framework for Older People. Department of Health (2005) The National Service Framework for Long-term Conditions. Department of Health (2006a). Mending Hearts and Brains. Clinical Care for Change. Report by Professor Roger Boyle, National Director for Heart Disease and Stroke. 51 Department of Health (2006b) ASSET 2 – Action on Stroke Services: an Evaluation Toolkit for Commissioners. Department of Health (2007a) A New Ambition for Stroke. A consultation on a national strategy. Department of Health ( 2007b). National Stroke Strategy. Department of Health (2006). National Schedule of Reference Costs 2005-06 for NHS Trusts. Deshphande, A., Jadad, A., Khoja, S., McKibbon, A., Rizo. C. (2008) Telehealth for Acute Stroke Management (Telestroke): Systematic Review of Analytic Studies and Environmental Scan of Relevant Initiatives. Technology Report 99. Ottawa: Canadian Agency for Drugs and Technologies in Health. Dodds S (2005) Designing improved healthcare processes using discrete event simulation. British Journal of Healthcare Computing and Information Management. Early Supported Discharge Trialists. Services for reducing duration of hospital care for acute stroke patients. Cochrane Database of Systematic Reviews 2005, Issue 2. Art. No.: CD000443. DOI: 10.1002/14651858.CD000443.pub2. E-Health News, (2008). http://www.ehealthnews.eu/content/view/982/27/ (accessed 10 March 2008). Evans, A., I. Perez, et al. (2001). Can differences in management processes explain different outcomes between stroke unit and stroke-team care? The Lancet 358(9293): 15861592. Fagan, S., Morgenstern, L., Petitta, A., Ward, R., Tilley, C., Marler, J., Levine, S., Broderick, j., Kwiatkowski, T., Frankel, M., Brott, T., Walker, M., and The NINDS rt-PA Stroke Study Group (1998) Cost-effectiveness of tissue plasminogen activator for acute ischemic stroke Neurology: 50:883-890. Ferlie, E., Fitzgerald L, Wood M, Hawkins C (2005). The nonspread of innovations: The mediating role of professionals. Academy of Management Journal 48(1): 117-134. Fjaertoft, H., B. Indredavik, et al. (2005). Early supported discharge for stroke patients improves clinical outcome. does it also reduce use of health services and costs? One-year follow-up of a randomized controlled trial. Cerebrovascular Diseases 19(6): 376-383. Gagliano, D., M. P. LaMonte, et al. (2000). Design and evaluation of a real-time mobile telemedicine system for ambulance transport. Journal of High Speed Networks 9(1): 47-56. Gaul, C., R. Handschu, J. Heckman, R. Littmann, B. Neundörfer, U. Reulback, M. Scibor (2003) Telemedicine in Emergency Evaluation of Acute Stroke: Inter-rater Agreement in Remote Video Examination With a Novel Multimedia System. Stroke. 34: 2842-2846. Gorman, M. and Levine, S. (1999) Telestroke: The Application of Telemedicine for Stroke. Stroke. 30: 464-469. Gourlay, D., K. C. Lun, et al. (2000). Virtual reality and telemedicine for home health care. Computers & Graphics 24(5): 695-9. 52 Grieve, R., J. Hutton, et al. (2001). A Comparison of the Costs and Survival of HospitalAdmitted Stroke Patients Across Europe. 32: 1684-1691. Hacke, W. (2000). A late step in the right direction of stroke care. The Lancet 356(9233): 869-870. Hacke, W. (2003). European Stroke Initiative Recommendations for Stroke Management Update 2003. Cerebrovascular Diseases 16(4): 311-337. Ham, C., N. York, et al. (2004). Hospital bed utilisation in the NHS and Kaiser Permanente: Authors' reply. 328: 584-a-. Ham, C., N. York, et al. (2003). Hospital bed utilisation in the NHS, Kaiser Permanente, and the US Medicare programme: analysis of routine data. 327: 1257-. Handschu, R., A. Garling, et al. (2001). Acute Stroke Management in the Local General Hospital. 32: 866-870. Hankey, G. J. and C. P. Warlow (1999). Treatment and secondary prevention of stroke: evidence, costs, and effects on individuals and populations*. The Lancet 354(9188): 14571463. Henderson, A., Korner-Bitensky, N. and Levin, M. (2007) Virtual Reality in Stroke Rehabilitation: A Systematic Review of its Effectiveness for Upper Limb Motor Recovery. Topics in Stroke Rehabilitation 14(2): 52-61. Hasegawa, Y., Y. Yoneda, et al. (2005). The effect of weekends and holidays on stroke outcome in acute stroke units. Cerebrovascular Diseases 20(5): 325-331. Hess, D. C., S. Wang, et al. (2005). REACH: Clinical Feasibility of a Rural Telestroke Network. 36: 2018-2020. Hirsch, G and Homer J (2004). Modelling the dynamics of health care services for improved chronic illness management. Proceedings, International System Dynamics Society Conference, Oxford, 25-29 July. Hirshberg, A., W. Koroshetz, J. Kvedar, S. Levine, E. Little, I. Petkovska, E. Rosenthal, P. Schaefer, L. Schwamm. (2004) Virtual TeleStroke Support for the Emergency Department Evaluation of Acute Stroke. Academic Emergency Medicine. 11: 1193-1197. Homer, J., Hirsch G, Minniti M, Pierson M (2004). Models for Collaboration: How System Dynamics Helped a Community Organize Cost-Effective Care for Chronic Illness. System Dynamics Review 20(3): 199-222. Indredavik, B. (2003). Stroke units - The Norwegian experience. Cerebrovascular Diseases 15: 19-20. Indredavik, B., F. Bakke, et al. (1999). Stroke Unit Treatment : 10-Year Follow-Up. Stroke, 30: 1524-1527. Iwatsuki, H., Fujita, C., Maeno, R. and Matsuya, A. (2004) Development of a Telerehabilitation System for Training Physiotherapists in Rural Areas. Journal of Telemedicine and Telecare, 10(1): 51-52. 53 Johnston, K. C. and B. B. Worrall (2003). Teleradiology assessment of computerized tomographs online reliability study (TRACTORS) for acute stroke evaluation. Telemedicine Journal and e-Health 9(3): 227-33. Jorgensen, H. S., L. P. Kammersgaard, et al. (1999). Treatment and Rehabilitation on a Stroke Unit Improves 5-Year Survival : A Community-Based Study. 30: 930-933. Kalra, L., A. Evans, et al. (2000). Alternative strategies for stroke care: a prospective randomised controlled trial. The Lancet 356(9233): 894-899. Kalra, L., A. Evans, et al. (2005). A randomised controlled comparison of alternative strategies in stroke care. Health Technology Assessment 9(18): 1-+. Khaja, A. M. and J. C. Grotta (2007) Established treatments for acute ischaemic stroke. The Lancet 369(9558): 319-330. Koch, S. (2006). Home telehealth--Current state and future trends. International Journal of Medical Informatics 75(8): 565-576. Kotiadis K and Mackenzie M (2004). Simulation Modelling for Intermediate Care, British Journal of Health Care Management, 10 (8) , 240-246. Lai, J. C. K., J. Woo, et al. (2004). Telerehabilitation: a new model for community-based stroke rehabilitation. Journal of Telemedicine and Telecare 10(4): 199-205. LaMonte, M. P., M. N. Bahouth, et al. (2003). Telemedicine for acute stroke: triumphs and pitfalls. Stroke 34(3): 725-8. Langhorne, P. (2006). Services for reducing the duration of hospital care for acute stroke patients. Stroke 37(1): 276-277. Langhorne, P. and A. Pollock (2002). What are the components of effective stroke unit care? 31: 365-371. Larsen, T., T. S. Olsen, et al. (2006). Early home-supported discharge of stroke patients: A health technology assessment. International Journal of Technology Assessment in Health Care 22(3): 313-320. Legg, L., P. Langhorne, et al. (2004). Rehabilitation therapy services for stroke patients living at home: systematic review of randomised trials. Lancet 363(9406): 352-356. Lewis-Brooks, A. (2004) HUMANICS 1 – A Feasibility Study to Create a Home Internet Based Telehealth Product to Supplement Acquired Brain Injury Therapy. In: Proceedings of the 5th International Conference on Disability, Virtual Reality and Associated Technologies (ICDVRAT 2004), pp. 11-18. Levine, S. R. and M. Gorman (1999). Telestroke: The application of telemedicine for stroke. Stroke 30(2): 464-9. Levine, S. R. and K. M. McConnochie (2007). Telemedicine for acute stroke: When virtual is as good as reality. 69: 819-820. Linard K (1996). System Dynamics Modelling and Aged Care, Modelling the Future – Techniques and Directions. Australian Institute of Health and Welfare Workshop. 54 Mallet, M. L. (2004). Hospital bed utilisation in the NHS and Kaiser Permanente: Authors did not compare like with like. 328: 584-. Markus, H. (2007). Improving the outcome of stroke. 335: 359-360. McKevitt, C., J. Redfern, et al. (2004). Qualitative studies of stroke - A systematic review. Stroke 35(6): 1499-1505. Moon, L., P. Moise, S. Jacobzone (2003). Stroke Care in OECD countries: A Comparison of Treatment, Costs and Outcomes in 17 countries, OECD Health Working Papers, 5. Murray, J., R. Ashworth, et al. (2003). Developing a primary care-based stroke service: a review of the qualitative literature. British Journal of General Practice 53(487): 137-142. National Audit Office (2005). Reducing Brain Damage. Faster Access to Better Stroke Care. Norrving, B. (2005). Editorial Comment--Organized Stroke Care: The Core of Effective Stroke Care Provision. 36: 1616-1618. Norrving, B. and R. J. Adams (2006). Organized Stroke Care. 37: 326-328. O’Brien, J (2008) Examining a Novel Technology Against End-user, Clinical and Management Demands with Reference to UK Care Provision. Doctor of Engineering thesis, Bartlett School of Graduate Studies, University College London. Panzarasa, S., S. Madde, et al. (2002). Evidence-based careflow management systems: the case of post-stroke rehabilitation. Journal of Biomedical Informatics 35(2): 123-139. Patel, A., M. Knapp, et al. (2004). Alternative strategies for stroke care - Cost-effectiveness and cost-utility analyses from a prospective randomized controlled trial. Stroke 35(1): 196203. Personal Social Services Research Unit (2006). Unit Costs of Health and Social Care 2006 Personal Social Services Research Unit (2007). Unit Costs of Health and Social Care 2007 Piron, L., P. Tonin, et al. (2002). Virtual environment system for motor tele-rehabilitation, Amsterdam, Netherlands, IOS Press. Piron, L., P. Tonin, et al. (2004). Motor tele-rehabilitation in post-stroke patients. Medical Informatics and the Internet in Medicine 29(2): 119-25. Pitt M, Bensley D, Brailsford S, Burnell S, Chaussalet T, Davies R, Dodds S, Pollard A, Wherry B, Worthington D. 2008. Simulation for strategic planning in health care – The state of the art. Briefing report for the NHS Institute, Mashnet. Plochg, T. and N. S. Klazinga (2002). Community-based integrated care: myth or must? International Journal for Quality in Health Care 14(2): 91-101. Richardson-Nassif, K., R. Swartz, et al. (2002). Implementing a community education program on stroke for health care providers and consumers. Education for Health 15(1): 5964. Rizzo, A., Schulteis, M,. Kerns, K., Mateer, C. (2004) Analysis of Assets for Virtual Reality Applications in Neuropsychology. Neuropsychological Rehabilitation. 14(1/2): 207-239. 55 Rizzo, A. and Kim, G (2005) A SWOT Analysis of the Field of Virtual Reality Rehabilitation and Therapy. Presence: Teleoperators & Virtual Environments, 14(2): 119-146. Rodgers, H., M. Dennis, et al. (2003). British Association of Stroke Physicians: benchmarking survey of stroke services. Age and Ageing 32(2): 211-217. Rosamond, W.; Folsom, A., Chambless, L., Wang, C., McGovern, P., Howard, G., Copper, L., Shahar, E., Rothwell, P. (1999) Stroke Incidence and Survival Among Middle-Aged Adults: 9-Year Follow-Up of the Atherosclerosis Risk in Communities (ARIC) Cohort. Stroke. 30(4):736-743. Buchan, et al. (2006). Recent advances in management of transient ischaemic attacks and minor ischaemic strokes. The Lancet Neurology 5(4): 323-331. Royal College of Physicians (2004). Intercollegiate Stroke Working Party, National Clinical Guidelines for Stroke (2nd ed.) Royal College of Physicians (2005). National Sentinel Stroke Audit 2004. Royal College of Physicians (2007). National Sentinel Stroke Audit 2006. Saka, R.O., A. McGuire, C.D.A. Wolfe (2005). Economic burden of Stroke in England, King’s College London. Saposnik, G. (2007). Weekends: A dangerous time for having a stroke? Stroke 38(4): 12111215. Scheideman-Miller, C., P. G. Clark, et al. (2001). Rural post acute stroke care using multidisciplinary telerehabilitation, Los Alamitos, CA, IEEE Computer Society Press. Schwab, S., B. Vatankhah, et al. (2007). Long-term outcome after thrombolysis in telemedical stroke care. 69: 898-903. Schwamm, L. H., A. Pancioli, et al. (2005). Recommendations for the Establishment of Stroke Systems of Care: Recommendations From the American Stroke Association's Task Force on the Development of Stroke Systems. 111: 1078-1091. Stroke Unit Trialists’ Collaboration. Organised inpatient (stroke unit) care for stroke. Cochrane Database of Systematic Reviews 2007,Issue 4. Art. No.: CD000197. DOI: 10.1002/14651858.CD000197.pub2. Schwamm, L. H., E. S. Rosenthal, et al. (2004). Virtual telestroke support for the emergency department evaluation of acute stroke. Academic Emergency Medicine 11(11): 1193-9. Shafqat, S., J. C. Kvedar, et al. (1999). Role for telemedicine in acute stroke: Feasibility and reliability of remote administration of the NIH stroke scale. Stroke 30(10): 2141-5. Silverman, G. (2001). Artificial intelligence in rehabilitation medicine: An emerging technology, Anaheim, CA, ACTA Press. Sundberg, G., A. Bagust, et al. (2003). A model for costs of stroke services. Health Policy 63(1): 81-94. 56 Susman, E. (1997). Telemedicine to give rural stroke victims fair chance of recovery with new treatment. Telemedicine and Virtual Reality 2(1): 1-2. Stroke Unit Trialists' Collaboration. (2007) Organised inpatient (stroke unit) care for stroke. Cochrane Database of Systematic Reviews, Issue 4. CD000197. DOI: 10.1002/14651858.CD000197.pub2. Thorsen, A. M., L. W. Holmqvist, et al. (2005). A randomized controlled trial of early supported discharge and continued rehabilitation at home after stroke - Five-year follow-up of patient outcome. Stroke 36(2): 297-302. Tran, B. Q., K. M. Buckley, et al. (2002). Selection and use of telehealth technology in support of homebound caregivers of stroke patients. Caring 21(3): 16-21. van Exel, N. J. A., M. A. Koopmanschap, et al. (2005). Cost-effectiveness of integrated stroke services. Qjm-an International Journal of Medicine 98(6): 415-425. Venketasubramanian, N., B. P. L. Chan, et al. (2002). Stroke disease management - A framework for comprehensive stroke care. Annals Academy of Medicine Singapore 31(4): 452-460. Von Koch, L. (1998). Rehabilitation in the home versus the hospital: the importance of context. Disability and rehabilitation 20(10): 367-372. Wade, D. T. (2003). Community rehabilitation, or rehabilitation in the community? Disability and Rehabilitation 25(15): 875-881. Waite, K., C. Jaigobin, et al. (2004). Telestroke: a strategy for acute stroke management in underserved communities [abstract]. Telemedicine Journal and e-Health 10(Suppl 1): S57. Walker B (2000). The Dynamics of Local Rules in Hospital Admission Processes. Caufield East: Faculty of Business & Economics, Monash University. Wang, D. Z. (2003). Telemedicine: the solution to provide rural stroke coverage and the answer to the shortage of stroke neurologists and radiologists. Stroke 34(12): 2957. Wang, D. Z., J. A. Rose, et al. (2000). Treating Acute Stroke Patients With Intravenous tPA : The OSF Stroke Network Experience. 31: 77-81. Wang, S., S. B. Lee, et al. (2003). Remote Evaluation of Acute Ischemic Stroke: Reliability of National Institutes of Health Stroke Scale via Telestroke. 34: e188-191. Ward, A., K. A. Payne, et al. (2005). Care needs and economic consequences after acute ischemic stroke: the Erlangen Stroke Project. European Journal of Neurology 12(4): 264267. Wardlaw J.M. , del Zoppo G, Yamaguchi T, Berge E. (2003) Thrombolysis for acute ischaemic stroke. Cochrane Database of Systematic Reviews, Issue 3., CD000213. DOI: 10.1002/14651858.CD000213. Wardlaw, J. M., J. Seymour, et al. (2004). Immediate Computed Tomography Scanning of Acute Stroke Is Cost-Effective and Improves Quality of Life. 35: 2477-2483. Wiborg, A. and B. Widder (2003). Teleneurology to improve stroke care in rural areas: the telemedicine in stroke in Swabia (TESS) project. Stroke 34(12): 2951-6. 57 Wiborg, A. and B. Widder (2003) Teleneurology to Improve Stroke Care in Rural Areas: The Telemedicine in Stroke in Swabia (TESS) Project. Stroke. 34: 2951-2957. Wolstenholme EF (2005). The potential of system dynamics, Leading edge briefing 10, NHS Confederation. Wolstenholme EF (1993). A case study in community care using systems thinking. Journal of the Operational Research Society 44(9): 925–934. Wu, O. and P. Langhorne (2006). The challenge of acute-stroke management: Does telemedicine offer a solution? International Journal of Stroke 1(4): 201-207. Zagonel A. 2002. Model conceptualization in group model building: A review of the literature exploring the tension between representing reality and negotiating a social order. Proceedings International Conference of the System Dynamics Society, Palermo. 58 Appendix 1: Methods Interviews We conducted 12 in-depth interviews with relevant local stakeholders gathering detailed information about the way stroke care is organised in the locality, any existing areas causing delays and bottlenecks and ideas about possible solutions. In addition, through a number of face-to-face interviews and attendance in conferences, we gathered expert information (e.g. stroke consultants, health economists, policy representatives) about the current organisation of stroke services in the country and ideas about possible alternative models. Observation We attended a weekly stroke multidisciplinary team meeting at the hospital site in which progress of patients was discussed. Informal conversations with the stroke co-ordinator about various aspects of current stroke care provision in the trust and relations between the trust and other relevant organisations e.g. PCT and Social Services. Documentary analysis In order to make the research and possible recommendations relevant to policy decision making, we became familiar with existing policy documents available in the public domain: 1. National Service Framework for Older People (DH, 2001). 2. National Clinical Guidelines for Stroke (2nd ed., 2004). Intercollegiate Stroke Working Party, Royal College of Physicians. 3. The National Service Framework for Long-term Conditions (DH, 2005) 4. Reducing Brain Damage. Faster Access to Better Stroke Care. (National Audit Office, November 2005). 5. National Sentinel Stroke Audit 2004. (Royal College of Physicians, 2005). 6. Economic burden of Stroke in England, (King’s College London, 2005). 7. Mending Hearts and Brains. Clinical Care for Change. Report by Professor Roger Boyle, National Director for Heart Disease and Stroke. (DH, 2006a). 59 8. ASSET 2 – Action on Stroke Services: an Evaluation Toolkit for Commissioners (DH, 2006b). 9. A New Ambition for Stroke. A consultation on a national strategy (DH, 2007a) 10. National Sentinel Stroke Audit 2006. (Royal College of Physicians, 2007). 11. National Stroke Strategy, (DH, 2007b). 60 Appendix 2: Dissemination Following the publication of the National Stroke Strategy we have described some of our findings in a press release. We also presented to or discussed our research with Roger Boyle (National Director for Heart Disease &Stroke), Mike Burrows (Chief Executive Salford PCT), Patricia Leahy (Director of Public Private Partnerships, NAO), Tom Allen (MIT), Stan Finkelstein (MIT and Harvard Medical School), National Advisory Council for Stroke (Scotland), NW London Cardiac and Stroke Network, Kensington and Chelsea PCT, NHS Tayside, and Pfizer.We presented our research in the clinical governance meeting for stroke at the acute hospital in our case study locality. We have ongoing discussion with Kensington and Chelsea PCT with a view of using modelling for supporting their implementation of the new stroke strategy for London. Engagement with them would allow us access to the data required to take our modelling work beyond the prototype stage. A further presentation at Brighton & Sussex University Hospitals NHS Trust is also planned. We are also exploring taking the work further in the Boston area with colleagues at MIT, Harvard Business School, and Harvard Partners. We are in the process of preparing a paper (“Towards the facilitation of stroke care planning using simulation modelling “) to be submitted to an academic journal. 61