53454ournal ofNeurology, Neurosurgery, and Psychiatry 1994;57:534-543 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from NEUROLOGICAL MANAGEMENT

Stroke and transient ischaemic attacks

Peter Humphrey

The management of is expensive and ANATOMY accounts for about 5% of NHS hospital costs. Carotid v vertebrobasilar arterial territory Stroke is the commonest cause of severe Carotid-Classifying whether a stroke is in the physical disability. About 100 000 people territory of the carotid or vertebrobasilar suffer a first stroke each year in England and arteries is important, especially if the patient Wales. It is important to emphasise that makes a good recovery. Carotid endarterec- around 20-25% of all affects people tomy is of proven value in those with carotid under 65 years of age. The annual incidence symptoms. Carotid stroke usually produces of stroke is two per 1000.' About 10% of all hemiparesis, hemisensory loss, or dysphasia. patients suffer a recurrent stroke within one Apraxia and visuospatial problems may also year. The prevalence of stroke shows that occur. If there is a severe deficit, there may there are 250 000 in England and Wales. also be an homonymous hemianopia and gaze Each year approximately 60 000 people are palsy. Episodes of or central reported to die of stroke; this represents retinal artery occlusion are also carotid about 12% of all deaths. Only ischaemic events. heart disease and cancer account for more Homer's syndrome can occur because of deaths. This means that in an "average" dis- damage to the sympathetic fibres in the trict health authority of 250 000 people, there carotid sheath. This especially follows carotid will be 500 patients with first strokes, with a dissection. prevalence of about 1500. After an internal carotid occlusion, there Transient ischaemic attacks (TIAs) are may be exaggerated pulsation in the branches defined as acute, focal neurological symp- of the external carotid artery (especially the toms, resulting from vascular disease, which superficial temporal artery). Increased collat- eral bloodflow through this artery shunts resolve in less than 24 hours; most settle in http://jnnp.bmj.com/ less than 30 minutes. The incidence of TIAs blood via the orbital vessels into the ophthal- is a quarter that of stroke. mic artery and then into the circle of Willis in Over the past 20 years, mortality from an attempt to compensate for the internal stroke has fallen both in the United Kingdom carotid occlusion, patients thus performing and United States by about a quarter. There their own extracranial-intracranial anastomo- has also been a fall in the incidence of stroke.2 sis. This is an almost universal finding on This is probably real but may be partly ultrasonography in internal carotid artery accounted for by the reclassification of stroke. occlusion. Sometimes the collateral flow is so on September 24, 2021 by guest. Protected copyright. The more successful treatment of hyperten- marked that an orbital bruit is heard and the sion is also likely to be relevant, but is superficial temporal artery on the side of the unlikely to be a complete explanation as this occlusion becomes tender and painful. This improvement had also been seen during can mimic temporal arteritis. It is particularly 1950-60, before the treatment of hyper- important that the temporal artery is not tension was widely practised. biopsied or a major collateral source of blood Stroke is not a diagnosis. It is merely a supply will be obliterated. description of a symptom complex thought to Vertebrobasilar-The terminal branches of this have a vascular aetiology. It is important to system are formed by the posterior cerebral classify stroke according to the anatomy of artery. Ischaemia of its territory usually pro- the lesion, its timing, aetiology, and patho- duces unilateral field defects. Bilateral symp- genesis. This will help to decide the most toms are not uncommon, with complete appropriate management. blindness or bilateral visual hallucinations, such as an impression of frosted glass or water running across the whole field of vision. Classification of stroke Sometimes amnesic symptoms may be Walton Centre for seen. In most patients, however, transient Neurology and Many neurologists have described vascular Neurosurgery, Rice syndromes in erudite terms. Most of these are global amnesia is no longer thought to be a Lane, liverpool of little practical use. A broadbased anatomi- TIA.3 L9 1AE, UK cal knowledge is important, however, as this The posterior cerebral artery also supplies P Humphrey part of the thalamus: infarction here produces Correspondence to: has significance in pathogenesis and manage- Dr Humphrey ment. sensory impairment over the contralateral Stroke and transient ischaemic attacks 535

side of the body. This may be accompanied Sometimes multiple lacunar infarcts occur. J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from by a very unpleasant pain which may be In such patients, there is often, but by no spontaneous or induced by light touch (thala- means always, a history of preceding minor mic pain) and often only reaches its peak stroke. The resulting syndrome is of a pseudo- some months after the stroke. bulbar palsy with dementia, dysarthria, small The signs after vertebrobasilar stepping gait (marche a' petits pas) unsteadi- ischaemia depend on the level of the lesion.4 ness, and incontinence. ischaemia may result in pupillary Bamford and colleagues6 have used the changes with impaired vertical gaze or oculo- Oxford Community Stroke Data to classify motor nerve dysfunction. Damage to the strokes clinically into-(a) lacunar infarcts; produces horizontal gaze palsy with (b) total anterior circulation infarcts; (c) par- facial weakness or sensory loss. In either case tial anterior circulation infarcts; and (d) pos- a tetraparesis or hemiparesis may occur. terior circulation infarcts (vertebrobasilar)- A wide range of other syndromes is (a) and (b) are classified as carotid artery reported to follow ischaemia of localised areas territory stroke. of the brain. The basic pattern is one of ipsi- Total anterior circulation infarct (TACI) lateral cranial nerve palsies and cerebellar dis- presents with a combination of higher cere- turbance combined with contralateral paresis bral dysfunction deficit (dysphasia, dyscalcu- or sensory loss which may affect the face, arm lia, and visuospatial disorder), homonymous and leg, or arm and leg, depending on the visual field defect and motor or sensory level in the brainstem at which this occurs. deficit, or both, of at least two areas of the Homer's syndrome may be seen. face, arm, and leg. Partial anterior circulation In the "locked-in" syndrome, patients infarcts (PACI) present with only two of the appear to be unconscious but are actually three components of the TACI syndrome fully conscious. They can only move their with higher cerebral function alone or with a eyes vertically; sometimes they can move their motor/sensory deficit more restricted than eyelids. It is good practice to introduce one- those classified as lacunar infarcts-for self to patients who appear to be uncon- example, confined to one limb, or to face and scious, and immediately ask them to move hand but not to the whole arm. their eyes before accepting that they are truly Patients with posterior circulation infarct unconscious. (POCI) present with any of the symptoms One word of caution-carotid artery dis- described in the section on vertebrobasilar section often presents with ipsilateral disease. Horner's syndrome and contralateral hemi- Using these simple clinical criteria, it paresis. It has also been described with ipsi- proved possible to classify most strokes into lateral cranial nerve palsies (especially one of these four different categories. This affecting nerves IX-XII) because the may be important as the prognosis, aetiology, expanded carotid artery damages these nerves and risk of recurrent strokes varies in the dif- in the neck. Classical teaching would have ferent groups. The TACI group had a very mistakenly put this vascular syndrome in the poor prognosis, with high mortality but a low

vertebrobasilar territory. recurrence rate, presumably as most of the http://jnnp.bmj.com/ carotid territory had been destroyed by the Lacunar infarct. The PACI group had a good progno- These small, deep microinfarcts described by sis but a high early recurrence rate, as this Fisher are commonly seen in hypertensive type of stroke is frequently embolic, probably and diabetic patients.5 They rarely occur in from internal carotid artery atheroma. These patients with carotid artery stenosis. It is patients have much to lose if a second stroke important to recognise these lacunar syn- occurs. The lacunar group had an intermedi- dromes because of their good prognosis and ate prognosis but a low risk of recurrence. on September 24, 2021 by guest. Protected copyright. different pathogenesis. Lacunar infarcts com- This type of stroke is rarely due to embolic monly present as pure motor stroke, pure disease but follows microvascular thrombosis sensory stroke, sensorimotor stroke, or ataxic or haemorrhage, often as a result of hyper- hemiparesis (table 1). Acute focal movement tension with or without diabetes. disorders may also be lacunar. Patients must The POCI group had a good prognosis but have either complete face, arm and leg or high early recurrence rate. This Oxford major face and arm or leg involvement. Community study dispels the notion that Those with more restricted deficits-for brainstem strokes in general have a poor example, weak hands only-are not included. prognosis in the acute phase: it also empha- These are considered to be partial, anterior sises the significant risk of recurrence and the circulation infarcts in the cortex. need to give advice about risk factors, and offer early medical treatment to those with posterior circulation infarcts. Table 1 Common lacunar infarcts Subclavian steal syndrome Clinical type Site oflesion This syndrome is largely an irrelevance. Pure motor hemiplegia Internal capsule, pons, cerebral Subclavian stenosis is common in asympto- peduncle matic patients. The classic syndrome of Pure hemi-anaesthesia Thalamus Ataxic hemiparesis Pons, internal capsule brainstem ischaemia on exercising the arm is Dysarthria/clumsy hand rarely present. It has a very low risk of stroke syndrome Pons, internal capsule (under 2% annually).7 It can usually be 536 Humphrey J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from diagnosed by measuring the blood pressure in 10% to intracerebral haemorrhage (table 2). both arms. It is not necessary to consider any Thromboembolic stroke accounts for 85% form of surgical intervention unless there are of all stroke.'0 It is usually possible on clinical intractable vertebrobasilar TIAs. Angioplasty grounds to differentiate those with subarach- is a less invasive option. noid haemorrhage. The management of subarachnoid haemorrhage is discussed else- Border zone infarcts where." Sometimes infarction follows a generalised The clinical differentiation of thromboem- reduction in cerebral blood flow. This is most bolic disease and intracerebral haemorrhage commonly seen after a cardiac arrest or is difficult. Various authors have attempted to hypoxic damage during cardiac surgery. develop a clinical score.'2 Early loss of con- Ischaemia is then especially marked in the sciousness, early vomiting, bilateral extensor border zone between the territory of individ- plantars, and marked elevation of blood pres- ual arteries because here perfusion is least. sure all suggest haemorrhage. TIAs and the The parieto-occipital zone is the area most presence of peripheral vascular disease sug- often affected, where border zone infarcts gest thromboembolic disease. produce visual field defects (often partial and No clinical score is sufficiently accurate, easily missed on routine examination), read- however, to allow reliable differentiation. CT, ing difficulties, visual disorientation, and con- provided that it is performed within two structional apraxia.8 In the frontal border weeks of the first symptom, is the only reli- zone, slowing up, pathological grasp reflexes, able method. It should be used in patients gait disturbances, and incontinence may with stroke because the proper assessment occur. and treatment of thromboembolism and The clinical description is therefore helpful haemorrhage differ. in formulating an opinion about the anatomi- What percentage of strokes are thrombotic cal site, aetiology, prognosis, and risk of and embolic is even more difficult to ascer- recurrence. This may well have management tain. It is estimated that about half of all consequences. thromboembolic strokes are embolic (table 3) and half thrombotic. The percentage of all TIMING AND PATHOGENESIS strokes caused by other factors such as pri- The timing of events is important in our mary hypoperfusion (see above, border zone understanding of pathogenesis. Most TIAs infarcts), vasospasm, and arteritis is small. are embolic and usually arise from the inter- nal carotid artery or from the heart. A small percentage are haemodynamic-these usually occur when there is severe, widespread occlu- Table 2 Causes ofcerebral haemorrhage sive disease. * Hypertension Sometimes it is possible to identify patients * Berry aneurysm * Perimesencephalic haemorrhage with haemodynamic TIAs clinically. Embolic * Arteriovenous malformation TIAs usually occur for no apparent cause. * Anticoagulants * Bleeding into tumour Haemodynamic TIAs, however, may have a * Mycotic aneurysm http://jnnp.bmj.com/ trigger; for instance, standing, exercising, * Coagulation disorders * Arteritis lowering blood pressure, eating, and straining * Amyloid angiopathy have all been described as precipitating * Drug abuse-cocaine * Venous thrombosis events. Haemodynamic amaurosis fugax may be triggered by a bright light or sunlight. Unlike patients with embolic amaurosis fugax who usually describe a shutter or black on September 24, 2021 by guest. Protected copyright. shadow descending across the visual field, Table 3 Cardiac sources ofemboli those with haemodynamic attacks often ini- Left atrium: tially describe increased contrast between Thrombus (usually secondary to atrial fibrillation) Myxoma black and white and then whiteness of vision Paradoxical embolism before their vision goes. Haemodynamic Mitral valve: amaurosis fugax is often more gradual than Rheumatic endocarditis embolic amaurosis. Sometimes Infective endocarditis haemody- Marantic endocarditis namic TIAs may be preceded by symptoms of Prosthetic valve presyncope such as dizziness and faintness. Mitral valve prolapse Mitral annulus calcification Finally, they may occur many times a day over a considerable period of time: this is Left ventricle: unusual in embolic TIAs. Thrombus-myocardial infarction, cardiomyopathy TIAs are rare Aortic valve: Although haemodynamic Rheumatic endocarditis compared with embolic TIAs, it is important Infective endocarditis to as it is that Marantic endocarditis identify them, unlikely Bicuspid valve antiplatelet therapy will help these symptoms. Aortic sclerosis and calcification Reconstructive vascular surgery is more Prosthetic valve likely Syphilitic aortitis to be appropriate.9 Stroke is usually secondary to thrombo- Congenital cardiac disorders embolic disease. About 15% of all strokes are Cardiac surgery: Five per cent of these are sec- Air embolism haemorrhagic. Platelet/fibrin embolism ondary to subarachnoid haemorrhage and Stroke and transient ischaemic attacks 537

Accuracy of diagnosis Risk factors J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from The differential diagnosis of TIA and stroke Age is the most important risk factor for includes epilepsy, migraine, tumour, demye- stroke. Hypertension is the most important lination, syncope, subdural haematoma, treatable risk factor.'8 The risk of stroke after malignant hypertension, hyperventilation, a TIA is 30% in five years, the highest risk hypoglycaemia, and giant intracerebral being in the first year.'9 Other proven risk fac- aneurysm. 13 tors include cardiac disease, diabetes mellitus, Focal motor seizures may be mistaken for smoking,20 and hypercholesterolaemia.2' High TIAs, especially in patients with a very severe cholesterol levels are also a major risk factor carotid stenosis, in whom the jerking of the for heart disease which will be the cause of limbs occurs as part of the TIA. Focal sen- death in most patients with cerebrovascular sory seizures are even more difficult to distin- disease. guish, although the march of the sensory Alcohol, taken in excess, is probably a risk symptoms in a focal seizure may be helpful. factor for cerebrovascular disease, especially Migraine occasionally presents diagnostic haemorrhage. Raised homocysteine and fib- difficulties. The slow build up of a migrain- rinogen levels may be independent risk fac- ous aura, which often lasts 20 to 30 minutes, tors for vascular disease.2223 would be unusual in a TIA. Visual migraine It is not known if obesity, stress, or physi- often consists of positive visual symptoms, cal activity have any part to play in the aetiol- such as scintillating , unlike the ogy of stroke-if so, it is likely to be small. blackness of amaurosis fugax. The presence of a typical migrainous headache is unlikely in a TIA. Headache occurs in 16% of patients Investigations with TIAs.'4 Few basic investigations are necessary for The United Kingdom TIA Study Group most patients with TIAs. Measurements has recently presented their data on should include a full blood count, erythrocyte tumours mimicking TIAs. Patients who sedimentation rate, urea and electrolyte present with sensory or jerking TIAs, loss of levels, glucose and cholesterol levels. Many consciousness, or speech arrest should all be physicians request a chest radiograph and suspected of having a tumour until proven ECG, although it is debatable whether these otherwise. '5 are necessary if there are no symptoms of Demyelination is usually suspected cardiac disease. because of the age of the patient, past history Patients with carotid TIAs or stroke with of previous attacks, and a more gradual onset recovery should also be assessed with of hemiparesis compared with that seen in a Doppler/duplex ultrasonography to detect vascular hemiparesis. carotid stenosis-this is highly accurate but Subdural haematomas rarely present with very dependent upon the operator's skills2426; vascular-like symptoms. They do, however, our own experience has shown that most present a particular diagnostic difficulty. The radiology departments setting up Doppler/ diagnosis of a carotid TIA is usually reason- duplex ultrasound services are highly inaccu-

ably consistent,'6 17 but vertebrobasilar TIAs rate, and all such units should have their http://jnnp.bmj.com/ are more variable. It is important to be wary results substantiated either by angiography or of labelling the following as TIAs: loss of a proven ultrasound service. consciousness; dizziness; mental confusion; Table 4 lists other tests that should be con- incontinence of faeces or urine; or bilateral sidered. loss of vision with reduced level of conscious- ness. These are all often secondary to hypoperfusion, following primary cardiac Treatment disease. MEDICAL on September 24, 2021 by guest. Protected copyright. Single symptoms such as vertigo, , Vascular risk factors dysphagia, dysarthria, loss of balance, tinni- High blood pressure after acute stroke is tus, sensory symptoms confined to one part common, often settles spontaneously, and of one limb or face, amnesia, drop attacks, does not need to be treated in most people. and scintillating scotomas, should always be Treatment should be started only if hyper- interpreted cautiously when they occur in iso- tensive encephalopathy is considered to be lation. They may, however, be consistent likely (systolic more than 230 mmHg: dias- with TIAs, especially if they occur together or tolic over 130 mmHg) or the patient has had with other more definite symptoms of TIAs. a proven cerebral haemorrhage and the blood The reliability of the diagnosis can be pressure is markedly elevated. It is also improved if clear cut criteria in plain language important to check the blood pressure in all are used in the assessment of TIAs.'7 patients with stroke-one or two months after In the diagnosis of stroke, the false positive discharge from hospital, as a significant rate with no investigations is between 1 % and number will show a persistent rise in blood 5%, if a careful history is taken of the event.'0 pressure that is severe enough to require It is important to emphasise that CT is no treatment, even though their blood pressure more accurate than clinical opinion"3; this is was satisfactory when they were discharged. probably because some events that are clini- Hypertension is the most important risk cally strokes are mistakenly diagnosed on CT factor for stroke. The risk of stroke rises as tumours, a diagnosis that is not substanti- exponentially as diastolic blood pressure ated with time. increases in the range 70-100 mmHg. A 538 Humphrey

Table 4 Additional investigations in patients with TIAs Prescription costs for lipid lowering drugs are J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from Thyroid function tests (especially when cholesterol is raised or increasing by 20% per year. Drug trials have the patient is in atrial fibrillation) as yet shown no clear benefit from lipid CT if there is: lowering drugs, although the United doubt about the diagnosis especially if the stroke onset was Kingdom Healthy Heart Study is about to gradual or the history unclear a need to exclude cerebral haemorrhage (ideally all strokes) start, in which 20 000 subjects are expected cerebellar stroke with a deteriorating level of consciousness to be recruited. Cerebral angiography A small number of patients, who have a (Digital subtraction/magnetic resonance) may be indicated packed cell volume of over 50, need assess- if the patient is a candidate for carotid endarterectomy To look for evidence of arteritis ment for polycythaemia followed by appropri- In cerebral haemorrhage ate treatment. Echocardiography if: abnormal cardiac findings are present, suggestive ofvalvular Acute stroke heart disease or evidence of recent myocardial infarction, or left ventricular aneurysm either clinically or on chest In recent years there has been much debate radiograph or ECG. over the value of stroke units. A recent all young patients with stroke (<50 yrs) if no clear cut cause is found overview leaves little doubt that patients multiple territory stroke for no clear cause treated in stroke units do better than those antiphospholipid syndrome or systemic lupus erythematosus? Iibman-Sacks endocarditis treated in general medical wards.30 There is positive blood cultures no specific reason for this: it may just be an Blood cultures if the patient is febrile, to detect subacute organisational matter rather than due to any bacterial endocarditis one specific treatment. Temporal artery biopsy to detect arteritis More difficult is the question of whether patients do better at home or in hospital. The 24-Hour ECG in case of arrhythmia (very rarely necessary: grossly over used) recent papers from Gladman et al,31 and Young and Forster," have suggested that Work up for young patients with stroke (see table 6) home treatment may be better. More work needs to be done on this. If patients are kept at home, it is clearly important that they should still be investigated in the most appro- 7.5 mmHg rise in diastolic pressure within priate manner. Trials comparing home treat- the range 70-110 mmHg is associated with a ment with stroke units are needed. We should doubling in the risk of stroke. not assume that hospital is better. Motivation This underlines the importance of blood and "do it yourself' physiotherapy are proba- pressure control. There is a risk of precipitat- bly greatly enhanced by staying at home, pro- ing hypotension in a small number of vided that adequate support from social patients, especially the elderly; this is often services, paramedical teams, and the family is overstated as a reason for not being more available. aggressive in the treatment of high blood There is no proven medical treatment for pressure, especially isolated systolic hyper- acute stroke. Dextran has been shown not to tension. be beneficial. The trials of calcium antagonis-

In the population at large, a modest fall of tics, steroids, and glycerol are inconclusive," http://jnnp.bmj.com/ 5 mmHg in mean diastolic pressure, achiev- and this is an area of active research. able by reducing the mean daily salt intake by The International Stroke Trial (IST) is 50 mmol/l, might reduce overall stroke mor- comparing heparin, aspirin, and placebo. The tality by 22%.27 This would have a greater Multicentre Acute Stroke Trial (MAST) is effect on the total number of strokes than just assessing thrombolysis. These trials have treating high blood pressure in people with recently been reviewed.'4 Despite the lack of diastolic pressures of over 100 mmHg. any proven treatment, it is one of the most Treating all hypertensive patients would exciting frontiers in the field of acute neurol- on September 24, 2021 by guest. Protected copyright. reduce the mortality of stroke by 15%. This ogy. There is no doubt that successful treat- compares with aspirin, which reduces the ment will be found: the increase in, and overall incidence of stroke by 1-2%, and interest created by, large multicentre trials is carotid endarterectomy, which reduces the rightly unstoppable. Stroke units make these overall incidence by 0.5%.2s Recent data sug- trials much easier to perform. We have only gest that inadequate monitoring and treat- to see how the ISIS trials have transformed ment of high blood pressure is common and the care of myocardial infarction to know that is the most important, avoidable risk factor.29 we must encourage stroke units to investigate These figures emphasise that treating high treatments for this most debilitating disease. blood pressure will do more for stroke pre- vention than any other treatment, either sur- TIAs and stroke with recovery gical or medical. There is no doubt that aspirin reduces the Advice about tobacco smoking is clearly risk of stroke and death in patients with TIAs important. Good control of diabetic symp- by approximately 25%." The exact dose is toms is also to be encouraged, although there unclear. The evidence is best for doses of are no data proving that good control reduces around 300 mg. Some believe smaller doses the risk of stroke. (of 37-5 or 75 mg) may be adequate but it is There is no consensus about the value of possible that these trials may be too small, cholesterol lowering drugs. There is, how- and that a difference between 37*5 mg and ever, no doubt that the lower the cholesterol, larger doses may have been missed because of the lower the chance of a heart attack. a type II statistical error. Stroke and transient ischaemic attacks 539

Ticlopidine is also an effective antiplatelet In the ECST trial, the crossover was at J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from agent, perhaps more effective than aspirin.36 approximately five months and in NASCET It is available in the United States but is not three months. All fit patients with a tight on general release in the United Kingdom. symptomatic stenosis should therefore be Unfortunately it sometimes causes skin offered surgery. They should be told the risk rash, diarrhoea, and reversible neutropenia- of stroke-that is, 10% annually, the local patients, therefore, need more careful moni- operative risk, and advised that operation toring. A related drug, clopidogrel, is now reduces the risk of stroke to 2-3% annually. undergoing trials comparing it with aspirin. Patients can then make up their own minds. Warfarin is indicated for definite cardiac If the surgical risk is over 10%, then the emboli. Lone atrial fibrillation has now been benefit is lost: all units should aim for less added to this list. The European Atrial than 5%. Fibrillation Study shows that warfarin There are a possible 5000 candidates for reduces the risk of subsequent stroke by carotid endarterectomy in England and 60-70% compared with placebo in patients Wales: operation in this group would prevent who have had an episode of cerebral 500 strokes in the first year. It is not a cheap ischaemia. The annual risk of serious bleed- form of treatment but, for the individual with ing was only 3% with 0-2% intracranial a carotid TIA and a tight stenosis, surgery bleeds.37 reduces the risk of a stroke by 75%. I also use warfarin if a patient has had The average district general hospital in the several TIAs that were not controlled by United Kingdom serving a population of aspirin. I give warfarin for six to 12 months 250 000 would expect to have about 20 and then switch back to aspirin, provided no patients who are fit, symptomatic, and have a further ischaemic events occur. There are no carotid stenosis of more than 70%. The value good data to support this as yet but trials are of this operation is dependent on low opera- being performed in the United States, Italy, tive mortality and morbidity. I believe that and The Netherlands comparing aspirin with there should be a small number of designated warfarin. surgeons in each region who do this opera- A very common question is when to start tion. Each should do at least 50 operations warfarin after a definite stroke. The risk of per year and the results should be indepen- recurrent emboli is high after the first event dently audited. but there is the danger of secondary haemor- It is important to appreciate that this oper- rhage into an infarct if warfarin/heparin is ation is only for patients with recent carotid started too early. The Cardiac Embolism symptoms, such as amaurosis fugax, hemi- Study Group has shown that the risk of sec- paresis, hemisensory loss, and dysphasia. Just ondary haemorrhage is very low 11 days after how much this operation has been overused is the initial event. The risk is also very low in emphasised by the report in 1988 (before the the first 11 days if the infarct is small or the results of the ESCT and NASCET trials) deficit mild. I therefore start anticoagulants which showed that only 35% of patients had immediately if the deficit is mild but delay for this operation for appropriate reasons in a 11 days if the deficit is severe (for example, sample of 1302 patients in the United severe hemiparesis, sensory loss and dys- States.40 http://jnnp.bmj.com/ phasia with a large infarct visible on CT. Clearly on the first day, CT may be negative Ultrasonographylangiography and the decision then has to be made on the None of these surgical trials included the severity of the clinical deficit only).38 angiographic risk. Although the risk of stroke after carotid angiography is generally quoted SURGICAL TREATMENT as 1%,41 it is almost certainly higher in

In 1954 the first carotid endarterectomy was patients with carotid stenosis, leaving approx- on September 24, 2021 by guest. Protected copyright. performed. The annual risk of stroke in imately 2% with a permanent disability.42 patients with a carotid stenosis who have had Doppler/duplex ultrasonography is undoubt- a TIA is about 10%. The surgical risk of edly the best screening test but it is very carotid endarterectomy varies from 1% to dependent on the skill of the operator. Some 25%. It is not surprising therefore that, for 37 units operate on the results of ultrasono- years, it was not known if this operation was graphy alone; unfortunately, obtaining infor- worthwhile. It was only with the publication mation on the intracranial circulation is of the European Carotid Surgery Trial difficult with ultrasound. Combined with (ECST) and the North American Trial magnetic resonance angiography (MRA), (NASCET) in 1991 that the true value of this ultrasound gives highly accurate information operation became known.39 about both the carotid artery in the neck, as The incidence of serious complications in well as an angiographic picture of the whole ECST was 3-7% and in NASCET 2-1%. intracerebral circulation. Patients were randomised to surgery and Our policy is to operate on the basis of an medical treatment or best medical treatment. entirely non-invasive work up with ultrasound In the group with 70-99% stenosis, there was and MRA if both tests agree (Young et a143). a highly significant benefit from surgery- We have found that, in patients with 70-99% there were 75% fewer strokes in those treated stenosis, the tests agree in 96%: we only pur- with carotid endarterectomy. Clearly the sue digital subtraction angiography in the lower the surgical complication rate, the other 4%. If MRA proves to be less depen- sooner the patient benefits from surgery. dent on the operator and becomes widely 540 Humphrey

available, it may replace the need for ultra- name but a few. In a population of a million J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from sonography; however, this would require an people, there are about 50 to 100 who are enormous expansion in MRA to all district candidates for carotid endarterectomy each general hospitals. year: this compares with around 15 000 with In expert hands, a bruit is the best clinical asymptomatic carotid stenosis. The scope for guide to detect an underlying internal carotid inappropriate surgery is substantial. It is un- stenosis.445 The bruit is lost, however, in very reasonable to expect a vascular surgeon to tight stenoses (false negative). differentiate these other conditions. A false positive bruit is not infrequent in There is little doubt that the risk of a the presence of either a contralateral occlu- stroke is highest in the first six months after sion, external carotid stenosis, or just internal the initial event and, if the time from first carotid atheroma. Furthermore, if the pres- symptom to assessment, investigation, and ence of a bruit is to be of value, it needs to be surgery takes several months, then we are fail- useful to those who are making the initial ing to meet the needs of many patients. In the assessment. The presence or absence of a United Kingdom we need to assess these bruit mentioned in the referral letter to a patients within a few days of their symptoms cerebrovascular clinic showed a specificity of and prepare them for surgery, if appropriate, 70% and sensitivity of 57% for patients with within two weeks after TIAs and six to eight 70-99% stenosis (Davies and Humphrey45a). weeks after recovery from stroke. This will On this basis, many patients with a carotid require changes in organisation, more neurol- stenosis would be denied surgery if only those ogists with an interest in vascular disease, and with a bruit were referred. perhaps more vascular surgeons. All patients with carotid TIA or who have recovered from stroke should have carotid ultrasonography in a department with a Hydrocephalus is a complication of cerebellar proven track record. I no longer listen for a strokes (both haemorrhage and infarction) bruit; if I wish to detect a carotid stenosis I which may be amenable to surgical treat- use ultrasonography. ment. My own personal work up for carotid endarterectomy is therefore a careful history, Asymptomatic bruits simple examination of the cardiovascular These are common in the elderly population system (occasionally I examine the neurologi- (approximately 7% over the age of 65). The cal system!), routine blood tests, chest radio- annual risk of ipsilateral stroke is approxi- graphy, and ECG. This, combined with mately 2%. Surgery is of no proven value, Doppler/duplex ultrasonography, is all done although trials are in progress. at the first clinic visit. If the patient is found to have a carotid stenosis and is prepared to OTHER ASPECTS take the risk of surgery, then an urgent MRA Emotional is booked in the outpatient department. Psychiatric factors are very important; few doctors have sufficient time to address these CTIMRI fully. Depression and anxiety are common: I do not routinely perform CT or MRI on with reassurance, especially about the risk of http://jnnp.bmj.com/ these patients. The value of CT was evalu- recurrence and advice about treatment to ated in a prospective study of 469 patients prevent further events, this will often being considered for carotid endarterec- improve. Counselling the spouse and close tomy-the cost was considerable and the family is also important.47 results did not alter management.46 In this Emotionalism is also common: it is present increasingly cost conscious health service, we in both bilateral and unilateral strokes. It is

need to look for value for money. usually sufficient to explain to the patient that on September 24, 2021 by guest. Protected copyright. Tumour "TIAs"'5 are rare and can often this is a physical symptom which will improve be suspected on clinical grounds-patients with time. Small doses of amitriptyline with speech arrest, pure sensory TIA, black- (10-25 mg) may be beneficial.48 outs, and jerking during their attacks should all raise the suspicion of alternative pathol- Epilepsy ogy. CT need only be performed in this Early epilepsy occurs in about 10% of all group. patients with stroke. It should be energetically treated with, for example, intravenous pheny- Assessmentfor surgery toin, as the cerebral metabolic rate doubles I remain convinced that a neurologist or during a fit. Late epilepsy after a stroke is a physician with a major interest in vascular common cause of epilepsy in the elderly pop- disease should perform the initial assessment. ulation. It is rare in the individual patient, These are not patients who should be referred however, and indicates that the validity of the primarily to the vascular surgeons. In our diagnosis of cerebrovascular disease should cerebrovascular clinics, of the 25 new patients be reassessed. we see each week, only two or three on average meet all the criteria for carotid Dysphagia endarterectomy. The differential diagnoses This is common, even after unilateral strokes. recently seen in our clinic include migraine, It usually improves but predisposes to aspira- epilepsy, hyperventilation, tumours, Parkin- tion, chest infection, dehydration, and death. son's disease, and motor neuron disease, to As it can be assessed by simply asking the Stroke and transient ischaemic attacks 541

patient to drink 50 ml of water, this test Stroke in the young J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from should be mandatory in all patients with Stroke in youth is often not caused by prema- stroke.49 ture atheroma (table 5). It should be investi- gated by a neurologist. The most common Thalamic pain causes are emboli from the heart, carotid or This is more common than is generally vertebral dissection, antiphospholipid syn- appreciated; it frequently starts weeks after drome, arteritis, cerebral venous thrombosis, the patient is discharged from hospital. It is a and premature atherosclerotic or hypertensive cause of great distress and may be helped by vascular disease. These require active assess- a variety of strategies.50 ment and treatment (table 6).52 It is likely that cerebral venous thrombosis should be treated Other common problems by anticoagulation.53 After acute stroke, deep venous thrombosis occurs in more than 50% of paretic legs, Care ofpatients with TIA and stroke although a relatively small number develop In England and Wales there are approxi- symptomatic pulmonary embolism. It can mately 100 000 patients with first stroke each usually be managed with elasticated stock- initial TIAs. ings. Pressure sores, septicaemia (often sec- year and 25 000 with ondary to urinary tract or chest infection), and hyperglycaemia may arise. Frozen shoul- TIA der is a common problem and can markedly Most TIAs will be managed by general prac- slow titioners. All reasonably fit patients with recovery. carotid events aged under 80 years should be referred to a neurologist or physician with an interest in vascular disease, to investigate the possibility of carotid stenosis, provided the Prognosis patient is prepared to take the risk of opera- The risk of a stroke after TIA is approxi- tion. All patients should be told the local mately 30% in five years. It is highest in the surgical risk. The results should be indepen- first year: in patients with carotid stenosis it is dently audited to ensure a low complication about 10-12%. The figures for stroke are rate. similar. It is therefore important to reassure I believe that patients under 50 years of age all patients that a second stroke is not immi- with TIAs should receive specialist opinion. nent. Even patients with bilateral internal carotid occlusions only have a recurrent stroke risk of 13% per year. Table 5 Causes ofstroke in the young Death in most patients with TIAs and Common: stroke is caused by cardiac arrest.5 Premature atherosclerosis After an acute stroke, 20-30% of patients Cardiac embolism Dissection-carotid or vertebral die. A poor prognosis is associated with Antiphospholipid syndrome including Sneddon's syndrome reduced consciousness, conjugate gaze palsy, Migraine

Arteritis (including postinfective e.g. ophthalmic zoster) http://jnnp.bmj.com/ signs of severe brainstem dysfunction, pupil- Venous thrombosis lary changes, and incontinence persisting Pregnancy beyond the first few days. Strokes resulting in Uncommon: cognitive impairment such as apraxia and Fibromuscular dysplasia Drug abuse especially cocaine, heroin, amphetamine neglect, and visuospatial dysfunction also Late effect of radiotherapy carry a poor prognosis for recovery. Moya moya syndrome Takayasu's syndrome, Behcet's syndrome One year after a stroke, 33% of patients Amyloid angiopathy

will be dead, 22% dependent, and 45% inde- Homocystinuria on September 24, 2021 by guest. Protected copyright. Fabry's disease pendent. Most recovery occurs in the first few Pseudoxanthoma elasticum, Marfan's syndrome, weeks; less occurs in months three to six and Ehlers-Danlos syndrome Haematological causes even less (but still useful recovery) occurs in Mitochondrial cytopathy months six to 12. It is known that some Syphilis AIDS symptoms, such as hemiplegic leg, can im- Neoplastic angioendotheliosis prove over a long period of time, but others often do not improve unless there is early recovery-for example, retinal infarction, homonymous hemianopia, and isolated Table 6 Additional tests in young patients with stroke spinothalamic sensory loss. In the hemiplegic * MRI/MRA* hand, if there is no active hand grip after * Echocardiography* * Serology for syphilis* three weeks, there is unlikely to be much * Lupus anticoagulant, antinuclear factor* improvement. It is crucial to take the natural * Anticardiolipin antibody* * Conventional angiography history of disability into account when plan- * Haemoglobin electrophoresis ning rehabilitation. * Haematological opinion including pro-antithrombin III * 24-Hour ECG Six months after a stroke, almost half the * Screening tests for homocysteinuria patients will be physically independent, 15% * Lumbar puncture * Brain biopsy/meningeal biopsy will have speech problems, 11% will be * White blood cell a-galactosidase incontinent of urine and 7% incontinent of * Muscle biopsy faeces, and 33% will still need assistance with * HIV screen feeding. *Should be performed in all young patients with stroke. 542 Humphrey

I spend much time reassuring patients and follow up every three to six months may help J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from refuting the diagnosis of vascular disease, to reverse or slow down the late decline in usually with enormous relief to the patient. mobility seen after stroke." An integrated Patients with TIAs not controlled by aspirin stroke service must be coordinated by a team or causing diagnostic difficulty should also be leader. In hospital, this will be the consultant seen for specialist opinion. in charge of the stroke unit. In the commu- nity, the integrated stroke service liaising with STROKE the general practitioner will organise the nec- This is more difficult-the first question is essary service. We do not want two com- whether the patient should be admitted to pletely separate services, one led by the hospital. This is partly dependent on the hospital and one by the general practitioner. severity of the deficit and age but, more often, Large, randomised trials with detailed cost- on social factors such as the presence of car- ings are needed to find out what is the most ers and support services. Whatever policy is effective method of delivering optimal care. followed, patients should be investigated In addition, all therapists must be aware of appropriately and, ideally, all should have their role as counsellors." CT. It is essential that we use the charities fully. I have no doubt that there will be treat- In the United Kingdom, the Stroke ment for acute stroke soon (as in myocardial Association is becoming more and more infarction), and that acute stroke units will be active, and a host of support clubs are-spring- needed to administer this. Acute investiga- ing up. Booklets about many aspects of stroke tions will all be performed at the same time. from TIA to wheelchairs, and from epilepsy After a short period in such a unit (perhaps after stroke to stroke in the young are avail- 24-72 hours), there are likely to be three able. options. For patients who have a mild deficit Delivering stroke care is expensive; it uses and will do well in any case, they can go a large percentage of the NHS budget and we home, needing only a small amount of domi- need to deliver care in the most efficient and ciliary services. For the severely disabled who cost effective manner. No view, however will not do well whatever is done, these steeped in tradition, should be exempt from patients may be managed at home or in nurs- proper clinical assessment with large, prop- ing homes or other long stay institutions. It is erly conducted trials. The progress made in a waste of time and resources putting these the past 10 years tells us this must be the way people through a long and arduous rehabilita- forward; it is exciting to answer questions that tion programme which will do nothing except everyone involved in the care of patients with lead to frustration and disappointment for stroke will face daily. The ultimate benefi- staff, patient, and family alike. Realistic goals ciary is the patient. must be set at all times and the patient and carers must understand what these are. 1 Oxfordshire Community Stroke Project. Incidence of Stroke in Oxfordshire. First year's experience of a The rehabilitation of those patients with Community Stroke Project. BMJ 1983;287:713-6. intermediate disability will be discussed in a 2 Whisnant JP. The decline of stroke. Stroke 1984;15: 160-8. separate article in this series. 3 Hodges JR, Warlow CP. The aetiology of transient global Clinical criteria to identify these groups are amnesia. Brain 1990;113:639-57. http://jnnp.bmj.com/ 4 Caplan LR. Vertebrobasilar disease. Stroke 1981;12: slowly being formulated. Where doubt exists, 111-4. the patient should be assumed to be able to 5 Fisher CM. Lacunar strokes and infarcts: a review. Neurology 1982;32:871-6. benefit from rehabilitation. New data are 6 Bamford J, Sandercock P, Dennis M, Burn J, Warlow C. beginning to identify patients at an early stage Classification and natural history of clinically identifi- able subtypes of cerebral infarction. Lancet 1991;337: who will not benefit from rehabilitation. 1521-6. A decision is needed whether rehabilitation 7 Hennerici M, Klemm C, Rautenberg W. The subclavian steal phenomenon: A common vascular disorder with is best delivered at home or in hospital. It is rare neurologic deficits. Neurology 1988;36:669-73. on September 24, 2021 by guest. Protected copyright. also necessary to ascertain what aspects of 8 Ross Russell RW, Bharucha N. The recognition and prevention of border zone cerebral ischaemia during disability respond to physiotherapy, speech cardiac surgery. QJf Med 1978;47:303-23. therapy, and occupational therapy. Could one 9 Ross Russell RW, Page NGR. Critical perfusion of brain and retina. Brain 1983;106:419-34. type of generalised therapist deliver most of 10 Sandercock P, Molyneux A, Warlow CP. Value of com- this care and advice, only calling on a more puted tomography in patients with stroke: Oxfordshire Community Stroke Project. BMJ 1985;290:193-7. specialised service if necessary? This would 11 Van Gijn J. Subarachnoid haemorrhage. Lancet 1992;339: certainly simplify a lot of domiciliary care. 653-5. 12 Allen CMC. Clinical diagnosis of acute stroke syndrome. Rehabilitation research needs to answer QJ7Med 1983;52:515-23. questions such as this to provide a useful, 13 Norris JW, Hachinski VC. Misdiagnosis of stroke. Lancet 1982;i:328-31. long term answer about the role of these ther- 14 Koudstaal PJ, Van Gijn J, Kappelle U. Headache in tran- apists. Multicentre trials with simple proto- sient or permanent cerebral ischaemia. Stroke 1991;22: 754-9. cols are just as relevant to rehabilitation as 15 Coleman RJ, Bamford JM, Warlow CP. For the UK TIA drug trials. Study Group. Intracranial tumours that mimic transient cerebral ischaemia: lessons from a large multicentre I suspect that rehabilitation is best man- trial. J Neurol Neurosurg Psychiatry 1993;56:563-6. aged at home: one hour daily (at best) of 16 Kraaijeveld CL, Van Gijn J, Schouten HJA, Staal A. Interobserver agreement for the diagnosis of transient inpatient physiotherapy will be trivial com- ischaemic attacks. Stroke 1984;15:723-5. pared with the amount of "physiotherapy" 17 Landi G, Candelise L, Cella E, Pinardi G. Interobserver reliability of the diagnosis of lacunar transient ischaemic motivated patients will do in their own home. attack. Cerebrovasc Dis 1992;2:297-300. The two most important factors affecting 18 Kannel WB, Wolf PA. Epidemiology of cerebrovascular disease. In: Ross Russell RW, ed. Vascular disease of the length of stay in hospital are stroke severity central nervous system, Edinburgh: Churchill Livingstone, and the absence of a carer at home.5 Routine 1983:1-24. Stroke and transient ischaemic attacks 543

19 Dennis M, Bamford J, Sandercock P, Warlow C. lation of embolic stroke: brain hemorrhage and manage- J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.5.534 on 1 May 1994. Downloaded from Prognosis of transient ischaemic attacks in the ment options. Stroke 1984;15:779-89. Oxfordshire Community Stroke Project. Stroke 1990;21: 39 Brown MM, Humphrey PRD on behalf of the Association 848-53. of British neurologists. Carotid endarterectomy: recom- 20 Donnan GA, McNeil JJ, Adena MA, Doyle AE, O'Malley mendations for management of transient ischaemic HM, Neill GC. Smoking as a risk factory for cerebral attack and ischaemic stroke. BMJ 1992;305:1071-4. ischaemia. Lancet 1989;ii:643-7. 40 Winslow CM, Solomon DH, Chassin MR, Kosecoff J, 21 Qizilbash N, Duffy SW, Warlow CP, Mann J. Lipids are Merrick NJ, Brook RH. The appropriateness of carotid risk factors for ischaemic stroke, overview and review. endarterectomy. N EnglJ Med 1988;318:721-7. Cerebrovasc Dis 1992;2:127-36. 41 Hankey GJ, Warlow CP, Sellar RJ. Cerebral angiographic 22 Clarke R, Daly L, Robinson K, et al. Hyperhomo- risk in mild cerebrovascular disease. Stroke 1990;21: cysteinaemia: an independent risk factor for vascular 209-22. disease. NEnglJ Med 1991;324:1 149-55. 42 Davies KN, Humphrey PRD. Complications of cerebral 23 Qizilbash N, Jones L, Warlow CP, Mann J. Fibrinogen angiography in patients with symptomatic carotid terri- and lipid concentrations as risk factors for transient tory ischaemia screened by carotid ultrasound. J Neurol ischaemic attacks and minor ischaemic strokes. BMJ Neurosurg Psychiatry 1993;56:967-72. 199 1;303:605-9. 43 Young G, Humphrey PRD, Shaw MDM, Nixon TE, 24 Humphrey P, Sandercock P, Slattery J. A simple method Smith ETS. A comparison of magnetic resonance to improve the accuracy of non-invasive ultrasound in angiography, Duplex ultrasound and digital subtraction selecting TIA patients for cerebral angiography. Jf Neurol angiography in the assessment of extracranial internal Neurosurg Psychiatry 1990;53:966-7 1. carotid artery stenosis. Jf Neurol Neurosurg Psychiatry 25 Hankey GJ, Warlow CP. Symptomatic carotid ischaemic 1994;57 (in press). events: safest and most cost effective way of selecting 44 Harrison MJG, Marshall J. Indications for angiography patients for angiography, before carotid endarterectomy. and surgery in carotid artery disease. BMJ 1975;1: BMJ 1990;300: 1485-91. 616-8. 26 Howard G, Cambless LE, Baker WH, et al. A multicentre 45 Wilson LA, Ross Russell RW. Amaurosis fugax and validation study of Doppler ultrasound versus angiogra- carotid artery disease: indications for angiography. BMJ phy. J Stroke Cerebrovasc Dis 1991;1: 166-73. 1977;ii:435-7. 27 Law M, Frost C, Wald N. By how much does dietary salt 45a Davies KN, Humphrey PRD. Do carotid bruits predict reduction lower blood pressure? II. Analysis of data disease of the internal carotid artery? Postgrad Med Jf from trials of salt reduction. BMJ 1991;302:819-24. 1994 (in press). 28 Dennis M, Warlow C. Strategy for stroke. BMJ 1991;303: 46 Martin JD, Valentime RJ, Myers SI, Rossi, MB, Patterson 636-8. CB, Clagett GP. Is routine CT scanning necessary in 29 Payne JN, Milner PC, Saul C, Bowns IR, Hannay DR, the preoperative evaluation of patients undergoing Ramsay LE. Local confidential inquiry into avoidable carotid endarterectomy? J Vasc Surg 1991;14:267-70. factors in deaths from stroke and hypertensive disease. 47 House A. Depression after stroke. BMJ 1987;294:76-8. BMJ 1993;307: 1027-30. 48 House A, Dennis M, Molyneux A, Warlow CP, Hawton 30 Langhorne P, Williams B, Gilchrist W, Howie K. Do K. Emotionalism after stroke. BMJ 1989;298:991-4. stroke units save lives? Lancet 1993;342:395-8. 49 Gordon C, Langton Hewer R, Wade DT. Dysphagia in 31 Gladman JRF, Lincoln NB, Barer DH. A randomised acute stroke. BMJ 1987;295:411-4. controlled trial of domiciliary and hospital based reha- 50 Bowsher D. Pain syndromes and their treatment. Curr bilitation for stroke patients after discharge from hospi- COpin Neurol Neurosurg 1993;6:257-63. tal. J7 Neurol Neurosurg Psychiatry 1993;56:960-6. 51 Sacco RL, Wolf PA, Kannel WB, McNamara P. Survival 32 Young JB, Forster A. The Bradford Community Stroke and recurrence following stroke: the Framingham Trial, results at six months. BMJ 1992;304:1085-9. Study. Stroke 1982;13:290-5. 33 Sandercock P, Willems H. Medical treatment of acute 52 Caplan LR. Stroke in children and young adults. In: ischaemic stroke. Lancet 1992;339:537-9. Stroke: a clinical approach. Oxford: Butterworth 34 Sandercock PAG, Van der Belt AG, Lindley RJ, Slattery J. Heinemann, 1993;469-85. Antithrombotic therapy in acute stroke: an overview of 53 Einhaupl KM, Villringer A, Meister W, et al. Heparin the completed randomized trials. Jf Neurol Neurosurg treatment in sinus venous thrombosis. Lancet 1991;338: Psychiatry 1993;56: 17-25. 597-600. 35 Antiplatelet Trialists' collaboration. Secondary prevention 54 Wade DT, Langton Hewer R. Hospital admission for of vascular disease by prolonged antiplatelet treatment. acute stroke. Who, for how long and to what effect? BMJ 1988;296:320-31. J Epidemiol Commun Hlth 1985;39:347-52. 36 Editorial. Ticlopidine. Lancet 1991;459-60. 55 Wade DT, Collen FM, Robb GF, Warlow CP. Physio- 37 European Atrial Fibrillation Study Group (EAFT). therapy intervention late after stroke and mobility. BMJ Secondary prevention in non rheumatic atrial fibrillation 1992;304:609-13. after transient ischaemic attack or minor stroke. Lancet 56 Forster A, Young J. Stroke rehabilitation: Can we do 1993;342: 1255-62. better? Emphasising physical recovery may be counter- 38 Cerebral Embolism Study Group. Immediate anticoagu- productive. BM3 1992;305: 1446-7. http://jnnp.bmj.com/ on September 24, 2021 by guest. Protected copyright.