7.1. Specialist stroke units
7.1.1. Clinical introduction
Patients with stroke admitted to organised stroke care (usually a stroke unit) are less likely to die and more likely to leave hospital independent than those who are cared for in general (usually medical and care of the elderly) wards. The evidence for this, documented in a systematic review initially in 1997, was the catalyst for a marked change in stroke service organisation across the NHS. The National Service Framework for the Elderly (Standard 5: stroke)50 recommended that all stroke patients should be admitted to organised stroke units. The National Audit Office Report6 in 2005 noted that there had been no increase in stroke beds between 2001 and 2004 in the National Sentinel Audits; in 2004, half of eligible patients were treated in a stroke unit at some point and only 41% spent most of their hospital stay there.2 However, by 2006, 91% of Trusts in the UK had a stroke unit, 62% of patients were treated in a stroke unit at some point and 54% spent most of their hospital stay on a stroke unit.3
The development of thrombolysis and other acute treatments has led to an increased emphasis on acute management of stroke in addition to rehabilitation. 52% of UK Trusts now have an acute stroke unit, characterised by access to brain imaging within 24 hours, specialist ward rounds at least 5 times a week, and acute stroke protocols and guidelines. A significant proportion also have access to CT scanning within 3 hours, continuous physiological monitoring and policies for direct admission from A&E. There is much less trial evidence available for the efficacy of acute stroke units than for rehabilitation units.
The clinical question to be addressed is whether patients who are rapidly admitted to a specialist stroke unit have better clinical outcomes than those admitted through a general ward.
7.1.2. Clinical methodological introduction
Specialist stroke unit
For the purposes of this question, specialist acute care was restricted to those units which focused on assessment, diagnostic tests including brain imaging and monitoring rather than rehabilitation. Features of stroke units included continuous monitoring of physiological functions, high staff-to-patient ratio, rapid access to diagnostic tests and treatment interventions.
One Cochrane systematic review was identified comparing organised inpatient (stroke unit) care for stroke with alternative care. Here we report the subgroup analysis that compared acute (semi-intensive) stroke units (continuous monitoring, high nurse staffing levels but no life support) with ‘comprehensive wards’ (a cerebrovascular ward and a stroke unit) or mixed rehabilitation wards.51
Level 2++
One RCT (N=304) was identified that looked at differences in management processes in stroke units compared with stroke team care.52
Level 1+
Five non-randomised controlled trials or cohort studies were identified.53–57 Six case series/observational studies were identified. 58–62,54
Level 3
The patient populations were broadly comparable with the exception of two studies. One study restricted the analysis to those patients who were living at home without community support prior to the stroke63 and the remaining study involved patients with intracerebral haemorrhage.54
Level 3
7.1.3. Health economic methodological introduction
Three economic evaluations were identified that had an acute stroke or TIA population.
Launois et al. (2004)64 reported on a French population. Not enough description was given of what care was received for the results to be applied to a UK setting. The results were for cost per trimester spent in minor disability, which cannot be compared with other evaluations and so would be difficult to use as evidence of the cost effectiveness of a stroke unit.
Moodie et al. (2006)65 reported on an Australian population. The units involved were all in different hospitals and so care varied even under each definition. The stroke care units and mobile services were at teaching hospitals, whereas the conventional care was in smaller suburban hospitals.
Evidence from Patel et al. (2004)66 was based on a randomised controlled trial (Kalra et al. 2005)67 carried out in the UK. Stroke units were compared to care by a mobile stroke team on a general ward, or domiciliary care. The mobile stroke team comprised a specialist registrar, nurse, physiotherapist and an occupational therapist. The team assessed every patient at admission and recommended a diagnostic and treatment plan based on stroke unit guidelines for implementation by the ward team. Patients were allocated to care within 72 hours of stroke. Cost effectiveness was evaluated both including the costs of informal care and excluding them. Informal care costs were calculated by two alternative methods:
time was valued using the minimum wage
time was valued at the average wage of a social services home help.
The evaluation had a one-year time horizon to reflect the one-year trial follow-up. Utility scores were collected at various points during the year of follow-up.
7.1.4. Clinical evidence statements
below summarises the outcome data on mortality, mortality or dependency combined and measures of dependency for patients admitted to acute stroke unit care compared with alternative care. The results are reported for all patients with acute ischaemic stroke and for specific subgroups where appropriate.
A summary of the outcome data on mortality, mortality or dependency combined and measures of dependency for patients admitted to acute stroke unit care compared with alternative care.
Mortality
When comparing all patients with acute ischaemic stroke, three studies reported significantly lower mortality rates at discharge and at follow-up associated with acute stroke units compared with alternative care.58, 59, 61 In the Cochrane review there was significant heterogeneity in the results and a random effects model was therefore used. There were no significant differences (NS).51
Level 3
Four studies reported significant reductions in mortality associated with stroke unit care but only for specific subgroups.61,63,60,55 However, the different types of care received and outcome measures used across the studies preclude any conclusion regarding any differential effects of stroke units on specific patient populations. Level 3
The study on patients with intracerebral haemorrhage reported a significantly lower mortality rate associated with a stroke unit compared with a general medical ward.54
Level 2++
Death or dependency
When comparing all patients with acute ischaemic stroke, three studies reported significant differences in favour of stroke units for the outcome of combined death or dependency or measures of disability or dependency.58,59,53
A further three studies reported significant differences in favour of stroke units for these outcomes but only for specific groups.60,61,63
Level 3
The study on intracerebral haemorrhage reported no significant differences for a stroke unit compared with a general medical ward with respect to number of patients discharged home or to institutionalised care (NS).54
Level 2++
A further study reported there was no significant difference at 6 months on quality-of-life measures when comparing patients admitted to a stroke unit with those admitted to a general medical ward (NS).56
Level 2+
Length of stay
The Cochrane review reported a significantly shorter length of hospital stay in patients admitted to stroke units compared with alternative care.51
Level 1++
One study reported a significantly shorter duration of hospital stay associated with stroke unit care59 and in one study patients admitted with unimpaired consciousness on admission had a significantly longer stay if they were admitted to a stroke unit compared with a general ward.63
Level 3
Three other studies found no significant differences in the duration of hospital stay for patients on stroke units compared with general wards.53,58,54
Level 3
Diagnostic procedures and treatment
One study found no processes of care (including setting, staffing, protocols, mobilisation and diagnostic exams available) were associated with a good outcome.58 Another study reported that patients admitted to a stroke unit were monitored significantly more frequently and were significantly more likely to receive measures to reduce aspiration and to receive early nutrition compared to patients on general wards.52 There was limited evidence to suggest that patients admitted to stroke units underwent diagnostic tests more frequently53 or more quickly54 than those admitted to a medical ward. Pharmacological interventions were more likely to be either inappropriately stopped or delayed if patients were admitted to a general ward compared with being admitted directly to a stroke unit.62,53,54
Level 3
7.1.5. Health economic evidence statement
Randomised evidence
Patel et al. (2004)66 found that stroke units provided the most expensive care (at one year), followed by the mobile stroke team, and then domiciliary care. This did not change regardless of whether informal care was included or not.
The mobile stroke team was less effective than domiciliary care, 69% of patients avoided death or institutionalisation compared to 78% in domiciliary care. The stroke unit was most effective with 87% of patients avoiding death or institutionalisation after one year.
When the stroke unit was compared to domiciliary care, the incremental cost per additional death/institutionalisation avoided was £49,600 (excluding costs of informal care), which related to an incremental cost per QALY of £64,097 with a one-year time horizon.
Observational evidence
Moodie et al. (2006)65 found that stroke units were more expensive than conventional care in the first 28 weeks (AU$15,000 vs $12,000, p=0.08). However, severe complications were significantly reduced (5.9% vs 25%, p<0.001).
7.1.6. From evidence to recommendations
The relatively low overall mortality rate in the systematic review compared to most unselected hospital-based cohorts may be due to selective entry of patients into trials. It was agreed that observational studies may be more representative of the stroke population as a whole. Three studies demonstrated that patients admitted to a stroke unit received therapeutic interventions and investigations more appropriately and quickly compared to those in a general medical ward. While better processes of care are linked to better outcomes there is currently no definitive trial support that these result in a reduction in mortality and morbidity. The lack of high-quality evidence was noted.
There is a need for a randomised trial comparing direct admission to an acute stroke unit vs admission to a medical ward at least while the latter remains standard clinical practice.
In the absence of evidence on whether rapid admission to an acute unit reduces mortality, morbidity and length of hospital stay, expert consensus led to the agreement that patients should be admitted where possible directly to an acute stroke unit. Trials outside the acute setting which demonstrate that direct admission improved the processes of care were noted. In the absence of any evidence identified in acute management, the group felt that there needed to be a very good reason not to generalise overall stroke unit results to those in the acute setting.
A cost-effectiveness analysis compared stroke units to care by a mobile stroke team on a general ward, or domiciliary care. Although the cost-effectiveness ratio of over £60,000 per QALY gained compared with domiciliary care would seem to imply that stroke units are not cost effective, this result must be treated with extreme caution since the one-year time horizon is likely to have dramatically under-estimated both the QALYs gained from averting deaths and the cost savings due to averting dependence. The consensus view of the GDG is that all patients should be directly admitted to a stroke unit.
7.1.7. RECOMMENDATION
- R17.
All people with suspected stroke should be admitted directly to a specialist acute stroke unit following initial assessment either from the community or accident & emergency (A&E) department.
Definition of a stroke unit
a discrete area in the hospital
staffed by a specialist stroke multidisciplinary team
access to equipment for monitoring and rehabilitating patients
regular multidisciplinary meetings occur for goal setting.
7.2. Brain imaging for the early assessment of people with acute stroke
7.2.1. Clinical introduction
Brain imaging is essential in stroke to exclude haemorrhage and stroke mimics. The ‘National clinical guidelines for stroke’ (2004)29 recommended scanning within 24 hours of onset of symptoms to confirm diagnosis. Only 42% of patients in the 2006 Sentinel Audit3 achieved this standard. This is unacceptably low. It is recommended that by the time of the 2008 audit, 100% of patients should be scanned within a maximum of 24 hours after admission. Access to brain scanning has been difficult in the past because of a perceived lack of urgency for scanning, problems with access to scanning, or a lack of radiology or radiography support. Even though scanner availability has increased in recent years, systems are clearly not routinely in place to allow immediate or rapid access to scanning throughout the UK. Changes in clinical practice (increased availability, changes in scan request and reporting procedures) will be required to implement the new recommendation.
The clinical question to be addressed is how quickly brain imaging should be performed following an acute stroke.
7.2.2. Clinical methodological introduction
No relevant papers were identified.
7.2.3. Health economic methodological introduction
Two economic evaluations were identified that address early brain imaging following an acute stroke.
An evaluation in the US of the health economics of early scanning assessed usual US practice with practice based on National Institute of Neurological Disorders and Stroke (NINDS) recommendations on time from arrival to hospital to scanning.68
A UK study69 analysed the HE issues associated with the selection and timing of CT scanning after first ever stroke, including ischaemic and haemorrhagic stroke and stroke mimics, but excluding subarachnoid haemorrhage.
7.2.4. Health economic evidence statements
Both strategies in the Stahl et al.68 analysis involved taking stroke care through the following steps:
symptom onset
arrival at emergency department
thorough evaluation by an emergency medicine physician
CT scanning and interpretation of CT findings
administration of tPA to eligible patients.
The current practice described was an average time of 25 minutes to emergency medicine physician evaluation and approximately 1.6 hours from onset to administration of tPA.
The NINDS strategy recommended shorter times: 10 minutes to emergency medicine physician evaluation, neurologist assessment within 10 minutes, and 25 minutes to CT scan, allowing tPA administration within an hour.
The NINDS strategy was cost-saving. The results showed an increase of 0.01 QALYs and a saving of $434 per patient, although no time horizon was stated.
Wardlaw et al. (2004)69 compared thirteen different scanning strategies ranging from scanning immediately to scanning within 14 days; and scanning all patients to scanning no patients. Outcomes were quantified using the modified Rankin scale (mRS) as alive and independent, dependent, or dead at 6, 12, and 24 months after stroke. Life-years were estimated up to 5 years after first-ever stroke. Scanning all patients immediately was found to be the dominant strategy (less costly and more effective).
7.2.5. From evidence to recommendations
No clinical trial was identified to answer this question. However, it is clear that there are some patients in whom urgent scanning will result in immediate changes in clinical management. In the absence of reviewing the evidence on which patients should receive urgent scanning, a consensus was reached by the group. It was agreed that patients who are on anticoagulant therapy, have a known bleeding tendency, a depressed level of consciousness, unexplained progressive or fluctuating symptoms, papilloedema, neck stiffness or fever, severe headache at onset and/or indications for thrombolysis or early anticoagulation should receive immediate (next available slot or within 1 hour; within 1 hour out of hours) brain imaging. This consensus was based on both clinical experience and a recommendation made in the Intercollegiate Stroke Working Party guideline (2004 edition).29 The GDG felt that immediate imaging of this patient population would result in changes in clinical management. For the remaining acute stroke patients, the clinical consensus of the group was that scanning should be performed as soon as possible (certainly within 24 hours).The health economic evidence supports the cost effectiveness of immediate scanning, although there may be limitations to the UK study because of changes in radiology staff costings. Immediate scanning, whilst cost effective, maybe difficult to implement because of scanning availability.
7.2.6. RECOMMENDATIONS
- R18.
Brain imaging should be performed immediately* for people with acute stroke if any of the following apply:
indications for thrombolysis or early anticoagulation treatment (see
sections 8.1 and
8.2)
on anticoagulant treatment
a known bleeding tendency
a depressed level of consciousness (Glasgow Coma Score (GCS) below 13)
unexplained progressive or fluctuating symptoms
papilloedema, neck stiffness or fever
severe headache at onset of stroke symptoms.
- R19.
For all people with acute stroke without indications for immediate brain imaging, scanning should be performed as soon as possible.**
