A stroke is damage to part of the brain due to interruption to its blood supply. The most common cause is blockage of a cerebral artery by a blood clot, which may have formed in the artery (thrombosis) or may have been carried in the circulation from another part of the body (embolism). Stroke may also result from localized haemorrhage due to a ruptured blood vessel (including an aneurysm or bleeding disorder) in or near the brain.

The incidence of stroke rises with age and is higher in men. Certain factors increase the risk; the most important are hypertension and atherosclerosis, which damage artery walls. Smoking increases the risk of stroke by increasing the risk of all these disorders; and the risk of atherosclerosis is increased with diabetes. Atrial fibrillation (irregular, rapid heartbeats), a damaged heart valve, or a recent myocardial infarction (heart attack) can cause clots in the heart that may migrate to the brain.

Symptoms and signs 

Symptoms usually develop abruptly and depend on the site, cause, and extent of the brain damage. The movement, function, or sensation controlled by the damaged area of the brain is impaired.The possible complications of a major stroke include pneumonia and blood clots in the legs, which could lead to pulmonary embolism. Roughly one third of major strokes are fatal, a third result in some disability, and the final one third have no lasting ill effects.

Diagnosis and treatment 

A range of techniques may be used to investigate the cause and extent of brain damage. These include ECG, CT scanning,chest X-rays, blood tests (which may include clotting tests), angiography, MRI, and carotid doppler scanning (to assess narrowing of the carotid artery). In some cases of stroke, urgent treatment may improve the chances of recovery. If a scan shows that the stroke is due to thrombosis, thrombolytic drugs may be given to dissolve the clot. Anticoagulants may be given if there is an obvious source of an embolism, such as atrial fibrillation. Antiplatelet agents such as aspirin may be used in order to help prevent further clotting. Attention to hydration and pressure areas, and good nursing care, are important influences on outcome. Physiotherapy may restore lost movement or sensation; speech therapy may help language disturbances.

Stroke in detail - technical


Cerebrovascular diseases include many pathological conditions but the principal categories are (1) infarction—through occlusion of major arteries, small arteries or venous sinuses; and (2) haemorrhage—most often through rupture of small arteries, arterial aneurysms or capillaries.


Strokes are common, with annual incidence rates for subjects aged over 55 ranging from 420 to over 1000 per 100 000. They are the most important case of disability in developed Western nations and the second most frequent cause of death after coronary heart disease. About 80% of strokes are caused by cerebral infarcts, with the remainder due to haemorrhage, with 20% of these attributable to a bleeding cerebral aneurysm. The incidence of transient ischaemic attacks (TIAs) is about 50 per 100 000.

General considerations

Strokes typically present with sudden onset of focal neurological deficit. Urgent imaging with CT or MRI allows rapid differentiation between haemorrhagic and ischaemic causes—with critical benefit for the introduction of appropriate treatment—and serial imaging may reveal the development of important complications such as rebleeding, ischaemia, or hydrocephalus that mandate specific interventions.

The consequences of stroke are often devastating: sudden loss of a large amount of brain tissue affects much more than specific, localized functions such as movement, sensation, vision, and language. Mood, initiative, sense of humour, and speed of thought are among the essential aspects of mental life that can be severely affected, but are often sadly ignored. Management of patients with stroke requires complex needs to be comprehensively addressed. The introduction of specialized clinical units for the multidisciplinary care of patients with stroke has been shown to improve clinical diagnosis and recovery, reducing the risk of death or institutionalized care by 14%.

Pathophysiology of arterial occlusive disease

Atherothrombosis is the main cause of occlusion of major arteries in the brain, typically by embolism from a source in the carotid artery, aorta, or heart. Whether occlusion of an artery supplying brain tissue actually leads to ischaemia depends on collateral pathways, which may be by (1) circle of Willis, (2) connections between extracranial and intracranial vessels, (3) leptomeningeal anastomoses.

Arterial occlusive disease—transient ischaemic attacks

These, by definition, are due to ischaemia of a part of the brain, producing symptoms/signs that last for <24 h (usually for minutes). The main presentations are with transient (1) hemiparesis, (2) dysphasia, (3) monocular visual loss, or (4) hemianopia. Without treatment (see ‘Secondary prevention’), the risk of stroke after a TIA is up to 20% in the first year and 7% in subsequent years.

Arterial occlusive disease—cerebral infarction

Classification—this has been attempted on the basis of time course (‘progressive’, ‘completed’), anatomy (which artery?), cause (e.g. large artery atherosclerosis, cardio-embolism) or functional deficit (e.g. modified Rankin scale). The Oxfordshire Community Stroke Project system provides a useful and widely accepted compromise, defining four types of cerebral infarction: (1) total anterior circulation infarcts (TACI)—with both cortical and subcortical involvement; (2) partial anterior circulation infarcts (PACI)—more restricted and predominantly cortical infarcts; (3) posterior circulation infarcts (POCI)—clearly associated with the vertebrobasilar arterial territory; (4) lacunar infarcts (LACI), confined to the territory of the deep perforating arteries.

Specific treatments—several medical interventions aim at dissolving the occluding clot, or at least preventing it from growing. Those of widest application are: (1) Thrombolysis—treatment of highly selected patients within 4.5 h of the stroke event (and the sooner the better) with intravenous recombinant tissue plasminogen activator (rt-PA) will avoid death or dependence in 55 patients per 1000 treated. (2) Antiplatelet agents—treatment with aspirin, started within 48 h of onset, will avoid death or dependence in 13 patients per 1000 treated.

Secondary prevention—aside from control of lifestyle factors (cessation of smoking, reducing overweight, daily exercise), specific measures to reduce the risk of threatened stroke include (1) Carotid endarterectomy—substantially decreases the risk of disabling or fatal stroke in patients with severe, symptomatic carotid stenosis. (2) Antiplatelet drugs—aspirin produces relative risk reduction of 13%, with addition of dipyridamole providing some further benefit; clopidogrel should be prescribed in patients who are intolerant to aspirin. (3) Anticoagulants—with embolic sources in the heart, mostly from atrial fibrillation, coumadin anticoagulants (INR 2.5–4) are the first choice in the absence of contraindications. (4) Statins—reduce the risk of stroke by about 20% per mmol/litre reduction in low-density lipoprotein (LDL) cholesterol. (5) Antihypertensive drugs—the level of blood pressure is by far the most powerful risk factor for stroke; in individuals with a history of stroke or TIA, a blood-pressure-lowering regimen reduces the risk of stroke by 25 to 50% depending on the degree of blood pressure lowering, regardless of whether or not the blood pressure at baseline was considered too high.

Venous occlusive disease

Aetiology—cerebral venous thrombosis usually occurs in the context of a combination of predisposing factors, e.g. the oral contraceptive pill, pregnancy, or postpartum, in combination with some inherited disorder of coagulation such as the factor V Leiden mutation.

Clinical features, investigation and treatment—manifestations include headache, focal deficits, seizures, and impairment of consciousness. Imaging with CT/MRI reveals ‘venous’ infarcts that do not correspond to a known arterial territory and evidence of the underlying sinus thrombosis. Most physicians treat with anticoagulants, but evidence for their efficacy is not strong. Mortality is 5 to 30%.

Primary intracerebral haemorrhage

Aetiology—primary intracerebral haemorrhage usually occurs in the context of a combination of (1) anatomical factors—cerebral vascular lesions or malformations, e.g. arteriovenous or cavernous malformations, ruptured perforating arteries, amyloid angiopathy; (2) haemodynamic factors—most notably blood pressure; and (3) haemostatic factors—relating to platelet function or the coagulation system, e.g. oral anticoagulants.

Clinical features, investigation and treatment—in most cases history and examination provide few clues to the cause of an intracerebral haemorrhage, but specific enquiry should always be made about use of oral anticoagulants and the possibility of malignant disease (with bleed into a metastasis), and evidence of a generalized haemostatic disorder should be sought. Brain imaging (CT or MRI) is the most important single investigation in patients with suspected intracerebral bleeding. There is no specific treatment for most patients, but therapeutic anticoagulation should be reversed, and surgical evacuation of cerebellar haematomas should be considered.

Subarachnoid haemorrhage

Aetiology—ruptured aneurysms cause 85% of nontraumatic subarachnoid haemorrhages.

Clinical features and investigation—presentation is with sudden, severe, and unusual headache, with loss of consciousness at onset in 50%. Imaging with CT is the most important investigation: scanning within 3 days reveals extravasation of blood in the basal cisterns in 95% of cases. Lumbar puncture is indicated if the history is convincing but the CT scan negative: xanthochromia of the supernatant after centrifugation of the cerebrospinal fluid sample is the diagnostic finding. CT and MR angiography are the methods of choice for demonstrating or excluding an aneurysm as the source of haemorrhage.

Treatment—aside from general nursing and supportive measures, key issues are (1) Prevention of rebleeding—without intervention, the risk of this is 30% in the month after presentation, with immediate mortality of 50%. Endovascular treatment (‘coiling’) is the preferred method to occlude the aneurysm and prevent rebleeding, but not all aneurysms can be treated in this way and operative clipping is still necessary for these patients. (2) Delayed cerebral ischaemia—occurs in up to 25% of patients with a ruptured aneurysm, most commonly 5 to 14 days after the initial bleed. Nimodipine reduces the frequency of cerebral ischaemia and poor outcome by about one-third.



Intracerebral haemorrhage was first recorded by the Swiss physician Wepfer (1620–95) and in more detail by Morgagni (1682–1771) in Padua. Nonhaemorrhagic stroke, ‘serous apoplexy’, greatly puzzled the medical community until cerebral softening (‘ramollissement’) was recognized as a pathological entity in 1820 by Rostan (1790–1866), in Paris. Initially it was regarded as an inflammatory condition. The relationship of brain softening with arterial occlusion and atherosclerosis gradually dawned on the pathologists; it was firmly established by Rokitansky (1804–78) in Vienna and Virchow (1821–1902) in Berlin. The term ‘infarction’ was coined by Cohnheim, one of Virchow’s disciples. Subarachnoid haemorrhages and their usual source, intracranial aneurysms, were first recognized at the start of the 19th century; the diagnosis could (sometimes) be made during life from the start of the 20th century. In 1931 the Edinburgh neurosurgeon Norman Dott (1897–1973) carried out the first intracranial operation for a ruptured aneurysm, by wrapping it in muscle.

Understanding cerebrovascular disease received great impetus from the advent of CT in the 1970s. Before that time, observations depended on postmortem studies and on indirect neuroradiological studies such as angiography and pneumoencephalography. CT allowed rapid and reliable distinction between haemorrhagic and ischaemic stroke during life. Subsequently, CT and the newer technique of MRI identified several subtypes of stroke, each requiring specific therapeutic measures. The rapid increase in diagnostic accuracy coincided with the emergence of the randomized clinical trial, which replaced pathophysiological reasoning as a basis for medical interventions.

Epidemiology of stroke

Worldwide, stroke is the second most common cause of death after coronary heart disease, although stroke is the most important single cause of disability in the Western world. Stroke incidence is not technically difficult to measure but requires a great deal of time and resources. The few reliable studies, mostly from industrial countries, show that age- and sex-standardized annual incidence rates for individuals aged 55 or more are between 420 and over 1000 per 100 000, depending on geographical region. More than half of strokes occur in people aged over 75 years. In western Europe the overall incidence rates are falling, despite the ageing population. The pathological type varies, even between studies with a high rate of CT, but a general estimate is that out of every five strokes four are infarcts and one a haemorrhage, and that one out the five haemorrhages is from a ruptured aneurysm (subarachnoid haemorrhage). The incidence of transient ischaemic attacks (TIAs) is about 50 per 100 000.

In terms of an average general practice of 2400 people in western Europe (1000 aged >55 years), four patients will have a stroke per annum, vs one having a TIA. Intracerebral haemorrhage will occur about twice every 3 years, and subarachnoid haemorrhage once every 8 years.

Arterial occlusive disease

The cerebral circulation and its disorders

Brain tissue is critically dependent on a constant supply of oxygen and glucose. The cerebral blood flow (c.800 ml/min) accounts for 15 to 20% of the entire cardiac output, whereas the brain (c.1350 g) accounts for only 2% of the normal adult body weight. Neurons in the brain require a constant supply of ATP to maintain concentration gradients of ions across their membranes, necessary for the generation of action potentials. The resting brain consumes energy at the same rate as a 20-W light bulb.

Whether occlusion of an artery in the brain or in the neck actually leads to ischaemia depends on collateral pathways. If an end-artery is occluded and there is no collateral circulation at all, ischaemic symptoms will occur within seconds. Neurons will start dying within minutes and within hours the entire supply area of the artery will be irreversibly damaged. In contrast, permanent occlusion of a major artery such as the internal carotid artery may be asymptomatic in the presence of adequate collateral circulation. Broadly speaking, three levels of collateral circulation can be distinguished; these can be thought of as three lines of defence):

  1. 1 The circle of Willis—even if no blood at all is flowing to the brain from one or even both internal carotid arteries, collateral flow from the other internal carotid artery or the basilar artery, via an intact circle of Willis, may ensure an adequate blood supply in the territory of the occluded artery.
  2. 2 Connections between extracranial and intracranial vessels—if the internal carotid artery is occluded at its origin, collateral channels may develop via the external carotid artery. Branches supplying the outer orbit may connect with branches to the retina, resulting in a reversed flow in the ophthalmic artery. From there, blood reaches the distal part of the internal carotid artery. Similarly, branches of the occipital arteries (normally supplying the neck muscles) may fill the basilar artery if this is occluded at its origin.
  3. 3 Leptomeningeal anastomoses—if, for example, the main stem of the middle cerebral artery is occluded, its terminal branches at the surface of the brain may anastomose with similar branches of the anterior and posterior cerebral arteries; in this way the cerebral cortex in the territory of the occluded artery is spared, partly or wholly, although the deep territory will still be ischaemic.

Atherothrombosis is the major cause of occlusion of major arteries in the brain or the neck. Two important qualifications should be made. First, atherosclerosis of intracranial arteries is relatively uncommon, at least in white people (vs black or East Asian people). This means that, in the Western world, brain infarction is usually caused by embolism, in which thrombus has been dislodged from an upstream lesion. The source can be the carotid artery, aorta, or heart. Second, atherosclerosis is not a sufficient cause in itself: not every person with severe atherosclerotic disease has an ischaemic stroke. Other relevant factors are collateral circulation, irregularity of the plaque, blood turbulence, platelet aggregation, and the balance of clotting factors.

Diagnosis of TIAs

TIAs are important to diagnose because they are potential harbingers of stroke. They precede cerebral infarction probably in only 15 to 20% (the precise proportion is uncertain because almost all such information has been retrospectively collected from patients once they have had a stroke).

The term ‘transient ischaemic attack’ is rather imprecise, because it tacitly implies three restrictions. To begin with, it refers only to the brain and not to angina pectoris or intermittent claudication. Also excluded is transient ischaemia of the entire brain, such as occurs in syncope or ventricular fibrillation. In medical usage, only ischaemia of a part of the brain corresponds with the term TIA. Finally, how transient is transient? Traditionally the limit for the duration of symptoms has been set at 24 h. Obviously this threshold has more to do with astronomy than with biology or disease. In fact most TIAs last minutes, not hours. The longer an attack lasts, the greater the chance that CT or MRI afterwards will show a relevant ischaemic lesion. In terms of patient management, the essential question is not whether the attack has lasted 3 minutes, 3 days, or 3 weeks, but what its cause is and how recurrences can be prevented.

What actually happens in the brain during a given period of ischaemia can often only be guessed at. The usual assumption is that an embolus, most often consisting of platelets or loosely organized thrombus, temporarily blocks an intracerebral vessel and then dissolves into smaller fragments. There is scant evidence for this phenomenon, apart from chance observations during fundoscopy, angiography, or surgery. Other explanations, applicable only to a minority, include marginal flow, secondary to severe narrowing or occlusion of arteries.

The diagnosis of a TIA is problematic. That one has to rely on the history alone is a first difficulty (it requires time, skill, and patience), but not a unique one. A greater source of error is that the term TIA is an interpretation rather than a description.

Main varieties of transient ischaemic attacks

There are four kinds of symptoms that can safely be regarded as TIAs, given that the onset is sudden (within seconds), all symptoms appear at the same time, without ‘march’, and there is no better explanation.

Transient weakness of one half of the body

Apart from weakness there may also have been numbness. Isolated numbness or pins and needles on one side of the body are a rare manifestation of transient cerebral ischaemia; other causes such as overbreathing are more likely. Weakness and numbness are closely related perceptions, and one should not take these or other expressions (‘an arm gone dead’) for granted. It is important to make sure that the problem had to do with moving the limbs or the face on one side (facial weakness on one side often manifests itself through slurred speech or drooling), and not with what it felt like when those body parts were touched or with spontaneous sensations. It is also important to verify whether the problem was in two of three body areas, and that it was not just a leg or arm gone to sleep after a nap in older people.

Transient loss of the ability to find words or to understand them

The medical term for this type of TIA is dysphasia or aphasia; in this case patients and relatives may not recognize the episode as representing a problem of language but will often describe the attack as ‘confusion’. It is helpful to ask specific questions about the ability to put thoughts into words (motor dysphasia), and about having been able to understand what was said (sensory dysphasia). If a patient can write sentences but cannot speak, the cause is almost certainly psychological. A frequent problem is the distinction between dysphasia (disorder of language) and dysarthria (disorder of articulation). To ask whether pronunciation was difficult may not be very helpful. After all, in both cases the patient’s thoughts are clear and the distinction between the right words and the right sounds is rather academic. A more useful question is whether the words made sense and whether they were in the right order. Dysphasia implies a lesion of the left hemisphere in right-handed people, and in 30% of strong left-handers.

Transient loss of vision in one eye

The difficulty in this case is to distinguish transient monocular blindness from loss of vision on one side in both eyes (hemianopia). Either type of attack can be interpreted by the patient as a problem in one eye. The distinction has practical implications, as monocular attacks of blindness should lead to investigation of the ipsilateral internal carotid artery in the neck with a view to angiography and surgery, whereas isolated hemianopia mostly (in 80%) reflects a disorder in the vertebrobasilar circulation, in which case treatment will often be medical. The key question to ask is whether patients have alternately covered each eye during the attack. A surprisingly large proportion of patients have done so, but they will not always offer this information without prompting. On having covered the ‘good eye’ in case of hemianopia, the patient should still have been able to see with the ‘bad eye’, although only the nasal half of the visual field. With a monocular disorder the blindness should have been complete after covering the unaffected eye.

Transient loss of vision in one hemifield

Hemianopia reflects dysfunction of the occipital lobe. It is also a common aura in migraine attacks; these auras may occur without ensuing headache, especially in older people. It is therefore important for the physician to enquire about the mode of onset: flashing lights, bright colours, zigzag lines, and a gradually expanding deficit all argue in favour of a migrainous attack rather than ischaemia in its restricted sense of a stroke warning.

Bullet list 1 Attacks that should generally not be regarded as TIAs

  • Any attack with loss of consciousness
  • Any attack with involuntary jerking
  • Any attack with positive visual phenomena (bright lights, etc.)
  • Any attack with only:
    • • numbness
    • • dizziness (with or without spinning sensations)
    • • double vision
    • • slurred speech
    • • unsteady walking

Differential diagnosis of TIAs

Bullet list 1 (above) lists types of attacks that should in general not be regarded as TIAs, either because of positive phenomena such as rhythmic jerking that are incompatible with the definition of focal ischaemia, or because other causes are much more likely. In particular, the tendency to label any episode of ‘dizziness’ in older people as ‘vertebrobasilar ischaemia’ or, even worse, ‘vertebrobasilar insufficiency’ should be strongly resisted.

In addition, some specific disorders other than atherosclerosis may cause attacks that are more or less indistinguishable from true TIAs as defined above. They are listed in Bullet list 2. These rare but important causes are reason enough to order a CT or MRI scan of the brain in patients with cerebral TIAs (not necessarily in those with transient monocular blindness). A chronic subdural haematoma should always be suspected in older people, especially if they are on anticoagulants. Hypoglycaemia should come to mind in patients with diabetes. Focal weakness may follow an epileptic seizure (Todd’s paralysis) and may be misdiagnosed as a TIA if the initial jerking is missed or misinterpreted. Tumours may also cause temporary deficits without focal epilepsy. Transient global amnesia is a disorder of memory possibly caused by migrainous vasospasm or venous congestion; although technically ischaemic in nature, it is not associated with an increased risk of stroke or other vascular disease.

Bullet list 2 Disorders that may mimic genuine TIAs

  • Chronic subdural haematoma
  • Intracranial tumour (glioma, metastasis, meningioma)
  • Hypoglycaemia
  • Focal deficits following a partial epileptic seizure
  • Transient global amnesia
  • Myasthenia gravis

Prognostic implications of TIAs

Without treatment, the risk of stroke after a TIA can be estimated at up to 20% in the first year and 7% in subsequent years, and the average risk of death, stroke, or myocardial infarction in the first 5 years at 10% per annum. Heart disease and stroke each account for about a third of all deaths. It is important to recognize that the risk of stroke is highest soon after the first episode: 8% in the first week, 12% at 1 month, and 17% at 3 months. Patients at particularly high risk can be identified by means of the ABCD score (see ‘Further reading’): A for age (>60 years), B for blood pressure (>140/90 mmHg), C for clinical features (2 points for unilateral weakness, 1 point for speech disturbance without weakness), and D for duration (2 points for >60 min, 1 point for >10 min) and for diabetes (1 point). The risk of stroke within 2 days is approximately 8% in patients with a score of 6 or 7, 4% in those with a score of 4 or 5, and much less in the others.

Investigations in patients with cerebral ischaemia

There is no great difference between searching for the cause of a TIA and searching for the cause of an ischaemic stroke. Very early CT or MRI is mandatory, mainly to exclude intracerebral haemorrhage and the occasional structural lesion mimicking stroke, not so much to demonstrate infarcts. Bullet list 3 lists the major and contributory causes of TIA and ischaemic stroke, with corresponding investigations. In general, first-line investigations are full blood count, erythrocyte sedimentation rate (ESR), plasma glucose, creatinine and electrolytes, plasma lipids, ECG, duplex ultrasound scanning of the arteries in the neck, and unenhanced CT or MRI of the brain.

Diagnosis of cerebral infarction

Distinction from other types of stroke

From a practical point of view, the first step is to distinguish ischaemic stroke from intracerebral haemorrhage. In the past, when a certain distinction could be made only at operation or postmortem examination, a decreased level of consciousness and headache were considered typical of intracerebral haemorrhage. After CT became available in the 1970s, it soon became clear that smaller haemorrhages are not associated with headache and drowsiness. Given that 4 out of 20 strokes are haemorrhagic, and on the assumption that half of all haemorrhages miss distinctive clinical features, a diagnosis of cerebral infarction based on clinical features alone will be wrong in approximately every tenth case. Even complex clinical scoring methods can hardly improve on this error rate.

On CT, acute parenchymal haemorrhage is of higher density than normal brain tissue. The hyperdensity occurs immediately—it is caused by the iron molecules in haemoglobin. Signs of infarction are more difficult to detect at an early stage. In the first decade of CT this was not possible until after 3 days, when frank tissue necrosis caused a hypodense lesion on the scan. With improved CT technology, subtle early signs of cerebral infarction have been recognized, at least when the area of infarction is large. These features include loss of normal differentiation between grey and white matter (such as the normal outline of the insular ribbon and the lentiform nucleus) and effacement of cortical sulci.

Bullet list 3 Major and contributory ‘causes’ of transient ischaemic attack (TIA) or ischaemic stroke, with corresponding investigations

Investigations marked with an asterisk (*) have proven implications for management.

Arterial atheroma
  • Internal carotid artery in the neck—duplex ultrasound study*
  • Intracranial arteries—angiogram (with MR, CT, or catheter)

Small-vessel disease

  • Aorta—transoesophageal echocardiography
Other arterial disease
  • Congenital arterial anomalies—angiogram (with MR, CT, or catheter)
  • Moya-moya syndrome—angiogram (with MR, CT, or catheter)
  • Arterial dissection—MRI; angiogram (with MR, CT, or catheter)
  • Giant cell arteritisa—ESR*, temporal artery biopsy*
  • Systemic vasculitis—antinuclear antibodies*, tissue biopsy*
  • Embolization from arterial aneurysms—MRI; angiogram (with MR, CT, or catheter)
  • Cholesterol embolization syndrome—biopsy of skin, muscle, or kidney
  • Meningitis, encephalitis—cerebrospinal fluid*, brain biopsy*
  • Drugs of abuse—toxicological screening of urine
  • Genetic conditions (MELAS, analysis of mitochondrial or nuclear DNA, CADASIL, Fabry’s disease), α-galactosidase A*
  • Irradiation
  • Migraine
Embolism from the heart
  • Atrial fibrillation—ECG*; long-term monitoring of heart rhythm*
  • Recent myocardial infarction—ECG*
  • Rheumatic valvular disease—echocardiogram*
  • Infective endocarditis—echocardiogram*, blood cultures*
  • Open foramen ovale—venography*, echocardiogram
  • Atrial myxoma—echocardiogram*
Haemostatic factors
  • Polycythaemia—haematocrit*
  • Sickle cell disease—peripheral blood smear, sickling test
  • Thrombocytosis—platelet count*
  • Leukaemia—white cell count*, morphological analysis*
  • Disseminated intravascular coagulation—platelet count, prothrombin, and activated partial thromboplastin times, fibrinogen, fibrinogen degradation products, D-dimers
Contributing risk factors
  • Hypertension—serial measurement of blood pressure*
  • Diabetes—fasting glucose*, HbA1c?
  • Hypercholesterolaemia—plasma cholesterol*
  • Hyperhomocysteinaemia—plasma homocysteine level

ESR, erythrocyte sedimentation rate. a Only with involvement of optic nerve or occipital lobe.

Within the first few days, CT will show that the area of infarction changes into a slightly hypodense, ill-defined, and somewhat swollen lesion; towards the end of the first week it becomes more clearly demarcated and hypodense. Occasionally there may be massive swelling with brain herniation or haemorrhagic transformation. During the second week the infarct may again gradually increase in density, because the degradation products of necrotic brain tissue more readily absorb X-rays; in the third and fourth weeks the infarcted area may even become isodense, being temporarily almost indistinguishable from normal brain, the so-called ‘fogging effect’. Eventually a sharply demarcated, atrophic, hypodense (similar to cerebrospinal fluid) defect remains. It is not always possible to determine how old an infarct is, or to distinguish it from the scar of a haemorrhage that occurred weeks or years before. Intravenous injection of X-ray contrast may in the first weeks cause some enhancement of adjacent brain tissue.

The proportion of patients in whom CT shows an appropriate infarct depends not only on the time of scanning and the generation of the scanner, but also on the size of the infarct and its location; eventually more than 90% of infarcts will show up.

MRI is especially useful for demonstrating small infarcts and lesions in the posterior fossa; it is also more sensitive than CT in the early phases of brain ischaemia. Signal changes on T 2-weighted images occur after 6 to 8 h. Infarcts of any size are visible within minutes on diffusion-weighted imaging, but this technique is not yet broadly available. The distinction from intracerebral haemorrhage is less obvious than on CT, but after a few hours the paramagnetic effects of deoxyhaemoglobin can be identified.

Classification of cerebral infarction

Time course has often been the guiding principle in the classification of stroke in the era before brain imaging. From the point of view of management and prognosis, however, the distinction between ‘progressive stroke’ and ‘completed stroke’ is hardly useful, let alone that between ‘permanent stroke’ and ‘reversible ischaemic neurological deficit’ (RIND, a kind of ‘extended TIA’ with complete recovery within 3 or 6 weeks, depending on local convention). What counts is the eventual severity of the functional deficit and, conversely, the remaining functions that are at stake.

The anatomical classification distinguishes infarcts according to the territory of major cerebral arteries: in the cerebral hemispheres infarcts can be located in the supply areas of anterior cerebral artery, middle cerebral artery, or posterior cerebral artery, or in the border zones between these three main branches; the cerebellum and brainstem are supplied by branches of the vertebral arteries and the basilar artery. The problems are that there is little if any relationship with handicap, mostly no distinction is made between partial and complete infarcts in a given territory, and the boundaries between different territories vary substantially between individuals.

Classification according to the cause of ischaemic stroke is of interest for studies aiming to describe or influence the pathophysiological background of strokes. The so-called TOAST classification, for example, distinguishes five subtypes of ischemic stroke: (1) large-artery atherosclerosis, (2) cardioembolism, (3) small-vessel occlusion, (4) stroke with other specific cause, and (5) stroke with undetermined cause. At present about 40% of patients would currently end up in the category ‘unknown’, even in specialized stroke services. Also the classification may change, according to the extent of ancillary investigations. Finally, and most important, the system is not suited for assessing the severity of stroke.

Rehabilitation specialists and geriatricians are more interested in the functional abilities of patients than in the niceties of neurological nosology. They mostly grade patients’ disability on a scale for activities of daily life (such as the Barthel scale, which ranks 10 in-house activities in hierarchical order, from bowel continence to taking a bath), or on a scale that includes some elements of social role fulfilment (‘handicap’), such as the Rankin scale (Table 1).

A system that strikes a useful compromise between the restrictions of lifestyle and the anatomical point of view is the classification of the Oxfordshire Community Stroke Project. Four categories are distinguished:

  1. 1 Total anterior circulation infarcts (TACIs), with both cortical and subcortical involvement, representing about one-sixth of all ischaemic strokes in the community
  2. 2 Partial anterior circulation infarcts (PACIs), with more restricted and predominantly cortical infarcts (one-third of all infarcts)
  3. 3 Posterior circulation infarcts (POCIs), clearly associated with the vertebrobasilar arterial territory (one-quarter)
  4. 4 Lacunar circulation infarcts (LACIs), confined to the territory of the deep perforating arteries (one-quarter).

Although the classes are anatomically defined, they contain important prognostic information: case fatality is highest by far in the TACI group.

Syndromes of cerebral infarction

Occlusion of the internal carotid artery may cause no symptoms at all or infarction, at its worst in the entire territory of the ipsilateral anterior and middle cerebral artery (and sometimes of the posterior cerebral artery or contralateral anterior cerebral artery as well), depending on the presence of a complete circle of Willis and other collaterals. If arterial dissection is the cause of carotid occlusion, subadventitial bulging of the artery may cause Horner’s syndrome and lower cranial nerve palsies, with or without infarction. Occlusion of the anterior, middle, and posterior cerebral arteries may lead to complete or partial infarction in their respective territories, depending on collaterals at the surface of the brain. Obviously, branch occlusions cause smaller infarcts. What follows is a description of syndromes associated with complete infarction in the average territory of the main cerebral arteries.

Table 1  Modified Rankin scale (or Oxford Handicap Scale) for measuring outcome after stroke (but suitable for other purposes as well)
Grade Description
0 No symptoms
1 Minor symptoms that do not interfere with lifestyle
2 Symptoms that lead to some restriction of lifestyle but do not interfere with the patient’s capacity to look after himself
3 Symptoms that restrict lifestyle and prevent a completely independent existence
4 Symptoms that clearly prevent independent existence though no constant attention is required
5 Totally dependent patient requiring constant attention, night and day

Infarcts in the area of the anterior cerebral artery cause contralateral hemiparesis, more marked in the leg than in the arm, with no or only mild sensory deficit. Other frontal lobe features include mutism, incontinence, and apathy or, conversely, disinhibition.

Middle cerebral artery infarcts, if complete, typically present with contralateral hemiplegia (most marked in the arm), sensory deficit, hemianopia, and cognitive defects such as aphasia (dominant hemisphere) or contralateral neglect (nondominant hemisphere). Massive infarction of the entire territory of the middle cerebral artery may lead to massive brain swelling followed by herniation, especially in young patients without cerebral atrophy.

Occlusion of a vertebral artery involving the origin of the posteroinferior cerebellar artery causes Wallenberg’s syndrome, with ipsilateral cerebellar ataxia through infarction of the inferior part of the cerebellum, and a—for students slightly bewildering—combination of deficits from infarction of the dorsolateral medulla: decreased skin sensation in the ipsilateral half of the face and the contralateral half of the body; ipsilateral Horner’s syndrome; ipsilateral weakness of the soft palate, larynx, and pharynx; and rotatory vertigo.

The full basilar artery syndrome, with infarction of most of the pons and midbrain, consists of coma, tetraparesis including facial movements, and loss of all eye movements and of pupillary and corneal reflexes. There are two characteristic partial syndromes of the basilar artery. One is the locked-in syndrome (infarction of the base of the pons), with tetraparesis including facial movements and loss of horizontal eye movements. Consciousness is preserved through sparing of the reticular formation, but patients can communicate only through vertical eye movements; these may not always be correctly interpreted or even noticed. The other is the top-of-the-basilar syndrome, with variable combinations of hemianopia or complete cortical blindness (occipital lobes), amnesia (inferior temporal lobes) and vertical gaze palsies, pupillary disturbances, and hallucinations (perforating branches to the midbrain).

The posterior cerebral artery syndrome may include hemianopia (occipital lobe), amnesia (lower temporal lobe), and oculomotor disorders or disturbances of language or visuospatial function, by involvement of perforating branches to the thalamus.

Occlusion of a single perforating artery, one of the many that originate at right angles from a large parent artery to supply a small area in the deep regions of the brain or brainstem, may be clinically silent, or cause a so-called ‘lacunar syndrome’. A necessary condition for the clinical diagnosis of a lacunar syndrome is the absence of ‘cortical’ deficits such as aphasia, neglect, hemianopia, or conjugate deviation of the eyes. The most common and archetypal form is pure motor stroke. In those cases the small deep infarct strategically involves corticospinal fibres (pyramidal tract) to the motor neurons of the limbs, anywhere in its course. Analogous fibres to the facial nucleus in the pons may be affected as well. The infarct can be located in the corona radiata, adjoining the wall of the body of the lateral ventricle, or slightly more caudally, in the posterior limb of the internal capsule, or, less commonly, in the pons or the medulla. Other ‘lacunar syndromes’ are sensorimotor stroke (corona radiata or internal capsule), pure sensory stroke (thalamus), and ataxic hemiparesis (usually the base of the pons). Lacunar infarcts in the brainstem may lead to an almost infinite range of syndromes, often with the name of a French 19th-century neurologist attached to it. Often such syndromes consist of an ipsilateral cranial nerve deficit and a contralateral hemiparesis.

Treatment of acute cerebral infarction

Several medical interventions aim at dissolving the occluding clot, or at least preventing it from growing: thrombolysis, antiplatelet agents, and anticoagulants. A different strategy, not yet well developed, is to protect ischaemic brain tissue. In addition, some underlying causes of stroke need urgent treatment, such as endocarditis. Before considering these specific measures, it is appropriate to consider the appropriate hospital setting in which stroke patients should be cared for.

Stroke units compared with treatment on general hospital wards

A specially organized stroke unit can be a ward or team that exclusively manages stroke patients (a dedicated stroke unit) or a ward or team that provides a generic disability service (a mixed assessment or rehabilitation unit). According to a meta-analysis of 23 randomized trials, stroke unit care reduces the risk of death or institutionalized care by 14%. The observed benefits are independent of patient age, sex, or stroke severity, and appeared to be better in stroke units based in a separate ward. No single element responsible for the benefits of organized stroke care has so far been identified, and probably there is none. The strength of stroke units lies in the integration of multidisciplinary efforts: stroke physician, nursing staff, physiotherapists, occupational therapists, speech and language therapists, rehabilitation physicians, and social workers.


Restoration of blood flow, to reperfuse the ischaemic brain as soon as possible after the cerebral artery has been occluded and, irrespective of its cause, should theoretically lead to reduction in the volume of brain damaged by ischaemia and to improvement in clinical outcome, analogous to myocardial infarction.

The main agents tested so far in stroke (18 trials in >5000 patients) are intravenous recombinant tissue plasminogen activator (rt-PA) and intravenous streptokinase, each in about one-half of the patients. Almost all were treated within 6 h of stroke onset. The evidence for efficacy is statistically significant for rt-PA, if administered within 3 h (or 4.5 h if observational studies are taken into account) and after exclusion of intracerebral haemorrhage by CT. Even within this period the adage is the sooner the better. For patients treated within 90 min the point estimates for survival with at most moderate disability improved from 54% in the placebo groups to 63% for patients in the rt-PA groups (absolute gain 9%), whereas for patients treated between 91 and 180 min after stroke onset the gain was 7% (from 57% to 64%). Taken together, a benefit of 8% means that some 12 patients have to be treated to save a single patient from death or the nursing home. These calculations have already taken into account the fact that there is a risk of secondary haemorrhage after treatment with rt-PA.

There are still many unanswered questions about thrombolysis in ischaemic stroke. The first of these is the time window. The earlier treatment is given the better, which is confirmed by subgroup analysis of patients treated within 3 h (or 4.5h if observational studies are taken into account) but inevitably this subgroup is biased towards more severe strokes (these get to hospital quickest). Second, is one agent better than another? For rt-PA alone, the balance of risks and benefits seems more favourable than for all agents together: per 1000 patients treated, an excess of 25 fatal intracranial haemorrhages, and a net result of 55 patients avoiding death or dependence, but the difference with streptokinase is not statistically significant. Third, we can roughly identify patients in whom the risk of haemorrhage in the infarcted tissue is great (those with the most severe deficits and those with early signs of extensive infarction), but the potential benefits are also greatest in this group. More controlled studies are needed. There are many contraindications in view of the risk of cerebral haemorrhage, and only a minority of patients admitted with cerebral infarction can be treated with thrombolysis.

Antiplatelet agents

More than 99% of the evidence from randomized trials is this area relates to aspirin. The pooled results of two trials with aspirin (160–300 mg), started within 48 h of onset, concluded that 13 fewer patients are dead or dependent for every 1000 patients treated. In some 800 patients who had been inadvertently randomized after a haemorrhagic stroke there was no evidence of net hazard. Only the combination with thrombolytic treatment should be avoided, because there are indications that aspirin enhances the danger of intracerebral haemorrhage.


Anticoagulants tested in clinical trials are standard unfractionated heparin, low-molecular-weight heparins, heparinoids, oral anticoagulants, and thrombin inhibitors. There is no evidence that anticoagulant therapy reduces the odds of being dead or dependent at the end of follow-up.

Neuroprotective agents

There are many steps in the destructive cascade between vessel occlusion and irreversible cell death where pharmacological intervention might be beneficial, at least theoretically. The pharmaceutical industry has developed several compounds for clinical development and testing. There is no doubt that in animal models many neuroprotective agents, given either before or after the onset of ischaemia, reduce the area of cerebral infarction. So far, none of these agents has been proven to reduce disability in patients, despite dozens of clinical trials. Many other trials are under way, but reduction of disability by neuroprotective drugs is likely to be modest at best.

Surgical decompression of space-occupying infarcts

To prevent brain herniation and death from supratentorial infarction, a large part of the skull vault can be removed. A pooled analysis (three trials) of 93 patients randomized within 48 h of stroke onset showed not only that the case fatality rate was much lower in operated patients (22%) than in patients who were treated conservatively (71%), but also that operated patients survived significantly more often (43%) with mild or moderate disability (modified Rankin grade 3 or less, see Table 1), against 21% in the conservative group. This has to be weighed against an increased proportion of survivors with severe disability (modified Rankin grade 4 or 5): 35% against 7%. Also the overall benefit is less clear in subgroups of patients with aphasia, those who underwent surgery on day 2, and those over 50 years of age.

With operations for space-occupying infarcts of the cerebellum there is no controlled evidence, but less uncertainty. Without surgery swelling of a cerebellar infarct can be fatal, whereas the deficits after surgical evacuation are surprisingly mild. In some patients it is sufficient to relieve obstructive hydrocephalus by external ventricular drainage.

Secondary prevention of ischaemic stroke

In the long-term management of patients with TIAs or moderately disabling ischaemic strokes, control of lifestyle factors is a primary concern: cessation of smoking, reducing weight if overweight, and daily exercise.

Specific measures to reduce the risk of threatened stroke are mostly pharmacological. Carotid endarterectomy is the only local treatment.

Carotid endarterectomy

This operation was increasingly performed from the 1960s onwards, but not until the 1980s were two randomized trials performed, one in Europe and one in North America. In patients with severe, symptomatic, carotid stenosis (70–99% lumen diameter reduction) the risk of disabling or fatal stroke substantially decreases after endarterectomy. On average, about six patients need to undergo surgery to prevent one ipsilateral ischaemic stroke within 5 years. This basic risk difference varies with age and sex, and it levels off after 3 or more years from randomization (i.e. 3.5 years after the qualifying event). It should be kept in mind that carotid endarterectomy is indicated in only a minority (<10%) of patients with TIAs or moderately disabling ischaemic strokes: the attacks have to be in the carotid territory, the patients should be fit and willing to undergo the operation, and the angiogram should show an accessible stenosis of over 70% at the carotid bifurcation.

A possible alternative for endarterectomy is carotid stenting, with or without devices to prevent emboli being carried downstream during the procedure. Despite the less invasive nature of this procedure, it is far from certain that the balance of risks and benefits is more favourable than for endarterectomy. So far the information from clinical trials is far from definitive.

For the demonstration of severe carotid stenosis, it is no longer necessary to perform catheter angiography, at least if the results of duplex ultrasound agree with those of CT angiography or MR angiography.

Antiplatelet drugs

The preventive effect of aspirin, in different doses, has been studied in placebo-controlled randomized trials in over 8000 patients after a TIA or moderately disabling stroke. There is virtually no difference between the risk reduction for daily doses between 30 mg and 1300 mg. The overall relative risk reduction is 13% (95% confidence interval 6–19%). Side effects, mainly indigestion, nausea, heartburn, and gastrointestinal bleeding, are more common as the dose is higher. Addition of dipyridamole 200 mg twice daily to aspirin provides a further risk reduction of approximately 18%, compared with aspirin alone. Headache is a common side effect of dipyridamole; it can be avoided by starting with smaller doses.

Clopidogrel, a thienopyridine derivative, is marginally more effective than aspirin, with a relative risk reduction of 8.7% (95% confidence interval 0.3–16.5), whereas the combination of clopidogrel and aspirin has no advantage over aspirin alone; clopidogrel should be prescribed only in patients who are intolerant to aspirin.


With sources in the heart, mostly from atrial fibrillation, coumadin anticoagulants (INR 2.5–4) are the first choice in the absence of contraindications; no evidence exists for a fixed age limit. In patients with a presumed arterial cause of cerebral ischaemia, anticoagulants may prevent some ischaemic events, but this benefit is offset or even outweighed by the risk of haemorrhages, especially in the brain, depending on the intensity of anticoagulation.


The protective effect on major cardiovascular events of statins, or, with the full name of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors, has been proved in clinical trials involving more than 90 000 patients with cardiovascular disease or risk factors, including cerebral ischaemia. The reduction in the 5-year incidence of major coronary events, coronary revascularization, and stroke is about 20% per mmol/litre reduction in low-density lipoprotein (LDL)-cholesterol, largely irrespective of the initial lipid profile or other presenting characteristics.

Antihypertensive drugs

Observational studies provide overwhelming evidence that the level of blood pressure is by far the most powerful risk factor for stroke. The obvious but at the same time precarious corollary that lowering the blood pressure prevents stroke has recently been confirmed by controlled evidence. In individuals with a history of stroke or TIA, a blood pressure-lowering regimen reduces the risk of stroke, regardless of whether or not the blood pressure at baseline was considered too high. The reduction of the stroke risk ranges from one-quarter to one-half, depending on the degree of blood-pressure lowering.

In conclusion, the optimal strategy for secondary prevention of stroke depends on the probable source of embolism. In patients with atrial fibrillation oral anticoagulants are the preferred treatment. Patients in sinus rhythm should be treated with a trio of drugs: an antiplatelet agent (preferably a combination of 30–300 mg aspirin once daily and dipyridamole 200 mg twice daily), a statin, and an antihypertensive agent. In most cases, baseline values for cholesterol or blood pressure should not influence the decision to administer statins and antihypertensive drugs, only the choice of the dose. In addition, patients with ischaemic events in the carotid territory should be investigated noninvasively for the presence of severe stenosis in the ipsilateral internal carotid artery, with a view to carotid endarterectomy.

Venous occlusive disease

The advent of noninvasive brain imaging methods in the last few decades has resulted in increased recognition of cerebral venous thrombosis. Before that time, physicians only rarely considered the diagnosis in patients with otherwise unexplained headache, focal deficits, seizures, impaired consciousness, or combinations of these features.

Causal factors

Unlike arterial occlusion, cerebral venous thrombosis is only rarely (in c.10%) associated with damage to the vessel wall—by infection, tumour growth, or trauma. Much more frequent causes are inherited disorders of coagulation. The most common form is factor V Leiden mutation, found in some 20% of patients without other causes. Stagnant flow (the third factor in Virchow’s triad of causes of thrombosis) contributes no more than a few per cent. In 20% of patients no causal factors can be identified.

Often there is not a single cause but a combination of contributing factors, e.g. the postpartum period and protein S deficiency; pregnancy and Behçet’s disease, or oral contraceptive drugs and the factor V Leiden mutation. The risk of cerebral venous thrombosis in the postpartum period increases with maternal age and with the performance of caesarean section.

In neonates, cerebral venous thrombosis is usually associated with acute systemic illness, such as shock or dehydration; in older children the most frequent underlying conditions are local infection (the leading cause before the antibiotic era), coagulopathy, and in Mediterranean countries Behçet’s disease.

Diagnosis of cerebral venous thrombosis

The clinical features of cerebral venous thrombosis consist essentially of headache, focal deficits, seizures, and impairment of consciousness, in various combinations and degrees of severity. The symptoms and signs depend on which sinus is affected, and for a large part on whether the thrombotic process is limited to the dural sinus or extends to the cortical veins.

In the case of the superior sagittal sinus, which is affected in 70 to 80% of all cases, cerebral venous thrombosis alone will lead to the syndrome of intracranial hypertension, i.e. headache and papilloedema. Up to 30% of patients with so-called ‘benign intracranial hypertension’ (BIH) may in fact have sinus thrombosis. Papilloedema can cause transient visual obscurations and sometimes irreversible constriction of visual fields, beginning in the inferonasal quadrants. The increased pressure of the cerebrospinal fluid may also give rise to cranial nerve VI palsies, and sometimes to other cranial nerve deficits. The onset of the headache is usually gradual, but in up to 15% of patients it is sudden and may initially suggest the diagnosis of a ruptured aneurysm.

Involvement of cortical veins causes one or more areas of venous infarction, with or without haemorrhagic transformation. If the affected veins drain into the superior sagittal sinus the venous infarcts are typically located near the midline in the rolandic and parieto-occipital regions, often on both sides. In the case of the lateral sinus the venous infarct is usually located in the posterior temporal area.

Clinically the infarcts manifest themselves through epileptic seizures or focal deficits such as hemiparesis or dysphasia. If unilateral weakness develops (with thrombosis originating in the superior sagittal sinus), it tends to predominate in the leg, in keeping with the parasagittal location of most venous infarcts. Obstruction of cortical veins draining into the posterior part of the superior sagittal sinus or into the lateral sinus will commonly lead to hemianopia, dysphasia, or a confusional state. Impairment of consciousness may result from multiple lesions in the cerebral hemispheres, or from transtentorial herniation and compression of the brainstem. Either epilepsy or a focal deficit is a presenting feature in 10 to 15% of patients; in the course of the illness seizures occur in 10 to 60% of reported series, and focal deficits in 30 to 80%.

Involvement of the cortical veins alone, without sinus thrombosis and its associated signs of increased cerebrospinal fluid pressure, is an extremely rare occurrence. Thrombosis of the deep venous system, including the great vein of Galen (great cerebral vein), may lead to bilateral haemorrhagic infarction of the corpus striatum, thalamus, hypothalamus, ventral corpus callosum, medial occipital lobe, and upper part of the cerebellum. In those cases the clinical picture is often dominated by deep coma and disturbance of eye movements and pupillary reflexes.


CT will readily show ‘venous’ infarcts. These do not correspond to a known arterial territory, and often show haemorrhagic transformation; they are sometimes bilateral, in the parasagittal area, supra- as well as infratentorial, or in the deep regions of the brain. In addition, CT will often provide evidence of the underlying sinus thrombosis: the hyperdense sinus sign or, less reliable, the so-called ‘empty delta sign’ after injection of intravenous contrast material.

MRI has made catheter angiography redundant in the diagnosis of cerebral venous thrombosis. It is not sufficient to rely on nonvisualization of a cerebral sinus on MR venography, because this may represent hypoplasia. Demonstration of the thrombus itself is essential, but this depends very much on the interval from disease onset. Three stages can be distinguished. In the acute stage (days 1–5) the thrombus appears strongly hypointense in T 2-weighted images and isointense in T 1-weighted images. In the subacute stage (up to day 15) the thrombus signal is strongly hyperintense, initially on T 1-weighted images and subsequently also on T 2-weighted images. The third stage begins 3 or 4 weeks after symptom onset: the thrombus signal becomes isointense on T 1-weighted images but on T 2-weighted images it remains hyperintense, although often inhomogeneous. Recanalization may occur over months in up to one-third of patients, but persistent abnormalities are common and do not signify recurrent thrombosis.

Treatment and prognosis

Anticoagulant treatment is plausible, but the evidence from controlled clinical trials is sparse. In the acute phase heparin (either by intravenous route or, subcutaneously, as low-molecular-weight heparin) seems preferable to oral anticoagulants, because its intensity can be closely monitored. The evidence for heparin treatment rests on no more than 80 randomized patients; there is a nonsignificant trend towards better outcome in treated patients. At least heparin treatment seems safe, even in patients with haemorrhagic infarcts. Local thrombolysis via endovascular catheters has been performed only in uncontrolled studies.

Death rates in different series range between 5% and 30%, and probably depend more on case mix than on treatment. Residual deficits consist mostly of hemispherical deficits or visual impairment from optic atrophy.

The risk of recurrence has seldom been systematically addressed; it is probably of the order of 10%. It seems wise to advise other means of contraception than the combined oral contraceptive ‘pill’. In women with a peripartum episode of cerebral venous thrombosis, the available evidence does not warrant the advice to avoid a further pregnancy, although in patients with the factor V Leiden mutation the risk of a recurrent episode is probably higher than average.

Primary intracerebral haemorrhage

Causes of primary intracerebral haemorrhage

In most cases there is no single cause for primary intracerebral haemorrhage. Even in the classic example of a so-called hypertensive haemorrhage in the region of the basal ganglia, the question remains about which anatomical or other factors distinguished this patient from others, with similar degrees and duration of hypertension but without brain haemorrhage. Even a combination of recognized ‘causes’, such as that of hypertension and anticoagulants, does not invariably lead to intracerebral haemorrhage. In general, therefore, several causal factors combine. These can be broadly distinguished into three categories (Bullet list 4): anatomical factors (lesions or malformations of brain vasculature), haemodynamic factors (blood pressure), and haemostatic factors (to do with platelet function or the coagulation system). Abnormalities of the vascular system account for the vast majority of haemorrhages. The type of underlying abnormality varies with age: below the age of 40 arteriovenous or cavernous malformations are the most common single causes, whereas between 40 and 70 the most frequent sources are ruptured perforating arteries (deep haemorrhages); in older people one also finds haemorrhages in the white matter (‘lobar’ haemorrhages), commonly attributed to amyloid angiopathy.

‘Hypertensive’ intracerebral haemorrhage

‘Hypertensive’ intracerebral haemorrhage results from degenerative changes in small perforating vessels, in the deep regions of the brain (basal ganglia and thalamus), or in the cerebellum or brainstem. Risk factors other than hypertension are age, male sex and high alcohol intake. Microaneurysms occur on these vessels but are not necessarily the site of rupture. It is probable that rupture of a single small artery leads to a cascade of secondary haemorrhages from adjacent arterioles. This might explain rapid expansion of intracerebral haematomas, during a single scanning procedure or on serial scanning. A stable phase is usually reached in a matter of hours.

Deep brain haemorrhages are not always a one-time event. The recurrence rate in the first year is 7%, against 2% per year over the subsequent 6 years.

Amyloid angiopathy

This condition accounts for about 10% of intracerebral haemorrhages. Its frequency rises with age, but so does that of ‘hypertensive’ haemorrhage. The underlying abnormality consists of patchy deposits of amyloid in the muscle layer of small and medium-sized cortical arteries of the occipital, parietal, and frontal lobes. Amyloid can also be found in asymptomatic individuals, the proportion increasing with age. It is not found outside the brain and does not represent generalized amyloidosis. Haemorrhages associated with amyloid angiopathy typically occur at the border of the grey and white matter of the cerebral hemispheres. Recurrent haemorrhage associated with amyloid angiopathy is much more common than with ‘hypertensive’ small-vessel disease. Hereditary forms of amyloid angiopathy account for only a small minority of all cases.

Bullet list 4  Causes of primary intracerebral haemorrhage

Anatomical factors
  • Lipohyalinosis (complex small vessel disease) and microaneurysms
  • Cerebral amyloid angiopathy
  • Saccular aneurysms
  • Cerebral arteriovenous malformations
  • Cavernous angiomas
  • Venous angiomas
  • Telangiectasias
  • Dural arteriovenous fistulae
  • Haemorrhagic transformation of an arterial infarct
  • Intracranial venous thrombosis
  • Septic arteritis and mycotic aneurysms
  • Moya-moya syndrome
  • Arterial dissection
  • Caroticocavernous fistula
Haemodynamic factors
  • Arterial hypertension
  • Migraine
  • Haemostatic factors
  • Anticoagulants
  • Antiplatelet drugs
  • Thrombolytic treatment (for non-neurological indications)
  • Clotting factor deficiency
  • Leukaemia and thrombocytopenia
Other factors
  • Intracerebral tumours
  • Alcohol
  • Amphetamines
  • Cocaine and other drugs
  • Vasculitis
  • Trauma (‘Spät-Apoplexie’)

Possible manifestations of amyloid angiopathy other than haemorrhage are transient episodes of focal neurological deficits, and also intellectual deterioration, associated with diffuse demyelination of the subcortical white matter (leucoaraiosis).

Cerebral arteriovenous malformations

Arteriovenous malformations (AVMs) are tangles of dilated arteries and veins, without a capillary network between them. On angiography, they are recognizable by large feeding arteries and rapid shunting of blood to enlarged and tortuous veins, via a central nidus of dilated vessels. Haemorrhage is the initial clinical manifestation in 50 to 60% of symptomatic AVMs. Other clinical features include epileptic seizures, headaches, and progressive neurological deficits. Demonstrable AVMs are the most common single cause of intracerebral haemorrhage in patients under 45 years (c.30%).

In 10 to 20% AVMs are associated with thin-walled saccular aneurysms. These occur on peripheral feeding arteries, not at the classic sites at the circle of Willis, and are likely sources of bleeding. In AVMs in which one or more aneurysms have formed, the annual risk of rebleeding is as high as 7%, against 2 to 3% per annum for other AVMs. If there is no associated aneurysm, the site of rupture is mostly on the venous side of the malformation.

Cavernous malformations

Cavernous malformations consist of sharply demarcated areas with widely dilated and thin-walled vascular channels, without intervening brain tissue. They are often asymptomatic and are encountered in 0.5% of routine post-mortem examinations, in the white matter or cortex of a cerebral hemisphere in about one-half of all cases, in the posterior fossa in one-third, and in the basal ganglia or thalamus in one-sixth. If a cavernous malformation is at all symptomatic, epileptic seizures are at least as common a manifestation as haemorrhage. The annual risk of haemorrhage in patients in whom the lesion presents with seizures or focal deficits is rather low, between 0.25% and 0.6%. After a first rupture, rebleeding is more frequent, around 4.5% per annum. Haemorrhages from a cavernous malformation are rarely fatal.

Familial forms of the disorder occur in several countries around the world, and should be suspected if multiple lesions are found.

Diagnosis of primary intracerebral haemorrhage

The history sometimes suggests the cause of the haemorrhage. Previous epileptic seizures should raise suspicions about the presence of an AVM, cavernous malformation, or tumour. Amyloid angiopathy should come to mind in patients over 55 years of age with a history of TIAs, intellectual deterioration, or both. A record of long-standing hypertension indicates small vessel diseases as the most probable underlying condition in a patient with a haematoma in the basal ganglia or in the posterior fossa; on the other hand, hypertension is so common that it may coexist with other conditions. If the patient is known to have had cancer, haemorrhage into a brain metastasis is a strong possibility. The use of oral anticoagulants is a vital piece of information in patients with intracerebral haemorrhage, because their action should be neutralized as soon as possible. It is equally important to know about the use of recreational drugs, particularly cocaine and amphetamines. Finally, the circumstances preceding intracerebral haemorrhage may help to identify its cause, such as puerperium (intracranial venous thrombosis, choriocarcinoma) or neck trauma (dissection of the vertebral or carotid artery).

The physical examination will provide rather few clues to the cause of an intracerebral haemorrhage, except petechiae or bruising, which indicate a generalized haemostatic disorder, signs of malignant disease such as cutaneous melanoma, a collapsed lung or enlargement of the liver or spleen, or telangiectasias in the skin and mucous membranes. Finding a high blood pressure on admission is the rule, but only in about 50% is there evidence of long-standing hypertension. Retinal haemorrhages indicate intracranial bleeding in general, most often subarachnoid haemorrhage (SAH). Heart murmurs may be coincidental but should at least raise the possibility of infective endocarditis, as should the finding of needle marks in possible drug addicts. The neurological examination will show focal deficits corresponding to the site of the lesion, with or without a decreased level of consciousness.

Investigations should start with the usual tests of blood and serum. These will sometimes uncover a cause of intracerebral haemorrhage, such as a low platelet count or massive liver damage. Brain imaging (CT or MRI) is the single most important investigation in patients with suspected intracerebral haematomas. The location of the haematoma may to some extent indicate the underlying cause. Intraventricular extension of the haemorrhage occurs relatively often with deep, ‘hypertensive’ haemorrhages. A grossly irregular margin of a lobar haematoma in older patients suggests amyloid angiopathy, as do multiple or recurrent haemorrhages in the white matter. Intracranial venous thrombosis should be suspected with irregularly shaped haemorrhages in the parasagittal region. Repeat brain CT after injection of contrast may pick up underlying lesions. Sometimes these can be identified only weeks later, when the lesion is no longer obscured by mass effects.

Treatment of primary intracerebral haemorrhage

Factors predicting the chances of survival of patients with primary intracerebral haemorrhage are: level of consciousness (Glasgow Coma Scale); age; volume of haematoma (poor prognosis if supratentorial haematoma more than 50 ml), subsequent expansion of the haematoma, and intraventricular extension of haemorrhage (poor prognosis if volume >20 ml). The possible interventions outlined below of course apply only to patients who have a chance of survival.

In patients on oral anticoagulants the first step is intravenous injection of 10 to 20 mg of vitamin K, at not more than 5 mg/min, followed by infusion of a concentrate of the coagulation factors II, VII, IX, and X, or of fresh frozen plasma.

Intracranial pressure is often raised. Factors other than the local effects of the haematoma may contribute, such as fever, hypoxia, hypertension, seizures, and elevations of intrathoracic pressure. An unsolved question is the use, in comatose patients, of monitoring and, if judged appropriate, lowering intracranial pressure. There are many believers in this area but few controlled studies. Insertion of a ventricular catheter may be a definitive measure in patients with cerebellar haemorrhage and no signs of direct compression of the brainstem.

For surgical treatment of supratentorial haematoma, randomized trials have failed to show benefit, including those employing endoscopic evacuation. In patients with cerebellar haematomas there is no doubt that surgical evacuation can be life saving, often with surprisingly few neurological sequelae. Sound indications for evacuation of a cerebellar haematoma are the combination of a depressed level of consciousness with signs of progressive brainstem compression (unless all brainstem reflexes have been lost for more than a few hours, in which case a fatal outcome is unavoidable), or haematoma greater than 3 to 4 cm. If the patient has a depressed level of consciousness and hydrocephalus, without signs of brainstem compression and with a haematoma less than 3 cm, ventriculostomy can be carried out as an initial (and sometimes only) procedure.

Subarachnoid haemorrhage

Causes of subarachnoid haemorrhage

Ruptured aneurysms are by far the most common source of nontraumatic SAH, in about 85%. Around 10% are nonaneurysmal perimesencephalic haemorrhages, the remaining 5% is made up by rarities (Bullet list 5).

Cerebral aneurysms are not congenital; they develop during the course of life. Therefore aneurysmal haemorrhage in a child is extremely rare. The aneurysms are saccular in shape and mostly arise at sites of arterial branching at the base of the brain, at or near the circle of Willis. It is largely unknown why some adults develop aneurysms. There are families with two or more affected first-degree relatives, but these account for less than 5% of all SAHs. Many classic risk factors for stroke in general also apply to SAH: smoking, hypertension, heavy drinking, and oral contraceptives. Not all aneurysms rupture. Their prevalence can be estimated, from angiographic studies (for other purposes) and postmortem studies at approximately 2 to 3% in middle age, up to 5% at the end of life. On the assumption that this proportion is 1% for a standardized population across all age groups, and given that the incidence of SAH is approximately 6 per 100 000 (of the entire population), the annual risk of rupture of an aneurysm is 0.6%.

Bullet list 5 Causes of subarachnoid haemorrhage

  • Ruptured aneurysm (85%)
  • Nonaneurysmal perimesencephalic haemorrhage (of venous origin?) (10%)
  • Rarities (5%)
    • • Arterial dissection (transmural)
    • • Cerebral arteriovenous malformation
    • • Dural arteriovenous fistula
    • • Pituitary apoplexy
    • • Mycotic aneurysm
    • • Cardiac myxoma
    • • Sickle cell disease
    • • Tumours
    • • Spinal arteriovenous malformation or aneurysm
    • • Trauma (without contusion)
    • • Cocaine abuse

Nonaneurysmal perimesencephalic haemorrhage is a distinct and benign variety of SAH, in which the distribution of extravasated blood on the brain CT scan is different from that with aneurysms, in the cisterns around the midbrain or ventral to the pons. The angiogram is completely normal, and the long-term outcome is invariably excellent. This subtype constitutes 10% of all SAHs and two-thirds of SAHs with a normal angiogram.

Diagnosis of SAH

The key feature in the history is that of a sudden, severe, and unusual headache. In 50% there is loss of consciousness at onset; in them the headache may emerge only later. The diagnosis is most difficult in patients with headache as the only feature. In general practice, exceptionally sudden forms of common headaches outnumber ruptured aneurysms. The incidence of aneurysmal haemorrhage being about 6 per 100 000 population per year, the average general practitioner will, on average, see one such patient every 8 years. There are no single or combined features of the headache that distinguish reliably and at an early stage between SAH and innocuous types of sudden headache. The discomfort and cost of referring most patients for only a brief consultation in hospital are a reasonable price to pay for avoiding misdiagnosis of a ruptured aneurysm.

The physical examination is unhelpful in patients with headache alone, without loss of consciousness or focal deficits. Neck stiffness takes about 6 h to develop, so its absence soon after the onset does not exclude the diagnosis of SAH at all.

Of the investigations CT is the most important. This will show extravasation of blood in the basal cisterns of the brain in at least 95% of patients with a ruptured aneurysm, if the scan is performed within 3 days. After that interval the sensitivity of CT quickly decreases. In patients with a negative CT scan but a convincing history, lumbar puncture is indicated. If the cerebrospinal fluid is blood stained, it is essential to distinguish SAH reliably from a traumatic tap. For that purpose at least 6 and preferably 12 h should have elapsed from symptom onset. In case of SAH sufficient lysis of red cells will have occurred in the meantime for bilirubin and oxyhaemoglobin to have formed. These pigments give the cerebrospinal fluid a yellow tinge after centrifugation (xanthochromia); they are invariably detectable until at least 2 weeks later. The ‘three tube test’ (a decrease in red cells in consecutive tubes in the case of a traumatic puncture) is notoriously unreliable. If the supernatant seems crystal clear, the specimen should be stored in darkness until the absence of blood pigments is confirmed by spectrophotometry. Bilirubin can be formed only in vivo; its demonstration by spectrophotometry therefore proves that red blood cells cannot have been introduced during the lumbar puncture, whereas oxyhaemoglobin can be formed if a cerebrospinal fluid specimen with red blood cells is left standing before the sample is spun down. Catheter angiography is rapidly being replaced by CT and MR angiography as a method for demonstrating or excluding an aneurysm as the source of haemorrhage.

Treatment of aneurysmal SAH

Several complications may occur after a first episode of aneurysmal SAH, of which rebleeding and cerebral ischaemia are the most dreaded. Despite advances in surgical and medical management, the population-based case fatality rate is still around 50%, with half of survivors remaining more or less disabled.

As general nursing measures, continuous observation and an intravenous access are essential. A bladder catheter is necessary for monitoring fluid balance. Headache should be relieved in a stepwise approach, with paracetamol and codeine as first steps. Distressing anxiety can be alleviated with short-acting benzodiazepines. Stools should be kept soft with oral laxatives and also by an adequate intake of fluids.

Prevention of rebleeding is challenging, if only because any effective measure tends to be offset by an increased risk of ischaemia. Moreover, at least 10% of all patients with SAH suffer a further bleed within hours of the initial haemorrhage. Over the next 4 weeks the rate of rebleeding without intervention is at least 30%. The immediate case fatality of rebleeding is 50%. Endovascular treatment (‘coiling’) is the preferred method to occlude the aneurysm and prevent rebleeding, but not all aneurysms can be treated in this way and surgical treatment by clipping is still necessary for these patients. Antifibrinolytic drugs decrease the rate of rebleeding but do not improve overall outcome.

Delayed cerebral ischaemia occurs in up to 25% of patients with a ruptured aneurysm, mainly between day 5 and day 14 after the initial bleed. Understanding of its pathogenesis has been impeded by simplistic notions about ‘vasospasm’ or ‘clots around vessels’. Narrowing of the arteries at the base of the brain is a factor but not a sufficient one. The total amount of subarachnoid blood is a potent risk factor, but only after rupture of an artery, and the distribution of blood in the subarachnoid space does not predict the site of ischaemia. The calcium antagonist nimodipine, in a dose of 60 mg every 4 h by mouth or nasogastric tube, reduces the frequency of cerebral ischaemia and poor outcome by about a third; its mode of action is incompletely understood. Hypertension should as a rule be left untreated; it is a compensatory reaction to maintain cerebral perfusion. The plasma volume should not be allowed to fall; hyponatraemia is caused by renal sodium depletion and not, as still often believed, by dilution as a result of inappropriate secretion of antidiuretic hormone. Fluids should therefore be replaced and not restricted. The basic intake should be at least 3 litres/day, with intravenous fluids supplementing oral intake; compensation should be made for fever or a negative fluid balance.

Further reading  

Algra A, van Gijn J (1999). Cumulative meta-analysis of aspirin efficacy after cerebral ischaemia of arterial origin. J Neurol Neurosurg Psychiatry, 66, 255. [Systematic review of aspirin in the secondary prevention of stroke.]
Ariesen MJ, et al. (2003). Risk factors for intracerebral hemorrhage in the general population: a systematic review. Stroke, 34, 2060–5. [Systematic review of risk factors for intracerebral haemorrhage.
Baigent C, et al. (2005). Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet, 366, 1267–78. [Meta-analysis of 14 randomized trials with statins in patients with cardiovascular disease or risk factors.]
Baigent C, et al. (2002). Collaborative meta-analysis of randomised trials of antiplatelet therapy for prevention of death, myocardial infarction, and stroke in high risk patients. BMJ, 324, 71–86. 
Bhatt DL, et al. (2006). Clopidogrel and aspirin versus aspirin alone for the prevention of atherothrombotic events. N Engl J Med, 354, 1706–17. [Shows that the combination of aspirin and clopidogrel has no advantage over aspirin alone in patients at high risk of atherothrombotic events.]
Coull AJ, Lovett JK, Rothwell PM (2004). Population based study of early risk of stroke after transient ischaemic attack or minor stroke: implications for public education and organisation of services. BMJ, 328, 326–8. [Shows that the risk of stroke after a transient ischaemic attack is greatest in the first week and month.][Abstract/Full Text]
Davis SM, et al. (2006). Hematoma growth is a determinant of mortality and poor outcome after intracerebral hemorrhage. Neurology, 66, 1175–81. [Shows that for intracerebral haematomas not only age and initial size but also subsequent expansion are important prognostic factors.][Abstract/Full Text]
EAFT (European Atrial Fibrillation Trial) Study Group (1993). Secondary prevention in non-rheumatic atrial fibrillation after transient ischaemic attack or minor stroke. Lancet, 342, 1255–62. [Proves effectiveness of oral anticoagulants.] [Web of Science] [Medline] 
Feigin VL, et al. (2003). Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol, 2, 43–53. [Review of population-based studies of stroke incidence.][CrossRef] [Web of Science] [Medline] 
Gent M, et al. (1996). A randomised, blinded, trial of clopidogrel versus aspirin in patients at risk of ischaemic events (CAPRIE). Lancet, 348, 1329–39. [Shows marginal advantage of clopidogrel over aspirin.][CrossRef] [Web of Science] [Medline] 
Gubitz G, Sandercock P, Counsell C (2004). Anticoagulants for acute ischaemic stroke. Cochrane Database Syst Rev, CD000024. [Systematic review.]
Hacke W, et al. (2004). Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials. Lancet, 363, 768–74. [Combined analysis of trials with rt-PA in ischaemic stroke.][CrossRef] [Web of Science] [Medline] 
Halkes PHA, et al. (2006). Aspirin plus dipyridamole versus aspirin alone after cerebral ischaemia of arterial origin (ESPRIT): randomised controlled trial. Lancet, 367, 1665–73. [Trial providing definitive evidence that the combination of dipyridamole and aspirin is superior to aspirin alone in the secondary prevention of stroke.][CrossRef] [Web of Science] [Medline] 
Halkes PHA, et al. (2007). Medium intensity oral anticoagulants versus aspirin after cerebral ischaemia of arterial origin (ESPRIT): a randomised controlled trial. Lancet Neurol, 6, 115–24. [Shows that medium intensity anticoagulation (INR 2 to 3) is ineffective in preventing major vascular events.][CrossRef] [Web of Science] [Medline] 
Johnston SC, et al. (2007). Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet, 369, 283–92. [Develops a simple risk score for patients with transient ischaemic attacks.][CrossRef] [Web of Science] [Medline] 
Koudstaal PJ, et al. (1992). TIA, RIND, minor stroke: a continuum, or different subgroups? Dutch TIA Study Group. J Neurol Neurosurg Psychiatry, 55, 95–7. [Shows how irrelevant it is to strictly distinguish ischaemic episodes of the brain according to their duration.]
MacMahon S, et al. (2001). Randomised trial of a perindopril-based blood-pressure-lowering regimen among 6105 individuals with previous stroke or transient ischaemic attack. Lancet, 358, 1033–41. [Randomized trial of antihypertensive drugs (perindopril with or without indapamide) in more than 6000 patients with a history of transient ischaemic attack or ischaemic stroke, showing a risk reduction in normotensive as well as in hypertensive patients.][CrossRef] [Web of Science] [Medline] 
Madden KP, et al. (1995). Accuracy of initial stroke subtype diagnosis in the TOAST study. Neurology, 45, 1975–9. [Classification of ischaemic strokes according to cause: the initial classification is confirmed after 3 months in only two-thirds.]
Mendelow AD, et al. (2005). Early surgery versus initial conservative treatment in patients with spontaneous supratentorial intracerebral haematomas in the International Surgical Trial in Intracerebral Haemorrhage (STICH): a randomised trial. Lancet, 365, 387–97. [Clinical trial of surgical treatment for supratentorial intracerebral haematomas, showing no overall benefit.] [Web of Science] [Medline] 
Molyneux AJ, et al. (2005). International subarachnoid aneurysm trial (ISAT) of neurosurgical clipping versus endovascular coiling in 2143 patients with ruptured intracranial aneurysms: a randomised comparison of effects on survival, dependency, seizures, rebleeding, subgroups, and aneurysm occlusion. Lancet, 366, 809–17. [Shows an advantage in survival with occlusion of ruptured aneurysms by endovascular coiling rather than clipping, maintained over several years.][CrossRef] [Web of Science] [Medline] 
O’ Collins VE, et al. (2006). 1,026 experimental treatments in acute stroke. Ann Neurol, 59, 467–77. [Explores the question why neuroprotective drugs after cerebral ischaemia that were promising in animal studies failed to work in humans.][CrossRef] [Web of Science] [Medline] 
Roos YBWEM, et al. (2003). Antifibrinolytic treatment in aneurysmal subarachnoid haemorrhage. Cochrane Database Syst Rev, CD001245. [Systematic review.]
Rothwell PM, et al. (2003). Analysis of pooled data from the randomised controlled trials of endarterectomy for symptomatic carotid stenosis. Lancet, 361, 107–16. [Systematic review based on individual patient data.][CrossRef] [Web of Science] [Medline] 
Rothwell PM, et al. (2004). Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet, 363, 1925–33. [Documents a substantial decrease in the age-specific incidence of major stroke in Oxfordshire.][CrossRef] [Web of Science] [Medline] 
Sandercock P, et al. (2003). Antiplatelet therapy for acute ischaemic stroke. Cochrane Database Syst Rev, CD000029. [Systematic review.]
Stam J (2005). Thrombosis of the cerebral veins and sinuses. N Engl J Med, 352, 1791–8. [Comprehensive review article.][CrossRef] [Web of Science] [Medline] 
Stroke Unit Trialists’ Collaboration (2002). Organised inpatient (stroke unit) care for stroke. Cochrane Database Sys Rev, CD000197. [Systematic review showing the benefits of organized inpatient stroke care.]
Vahedi K, et al. (2007). Early decompressive surgery in malignant infarction of the middle cerebral artery: a pooled analysis of three randomised controlled trials. Lancet Neurol, 6, 215–22. [Pooled analysis of three randomized trials.][CrossRef] [Web of Science] [Medline] 
Van der Zwan A, et al. (1992). Variability of the territories of the major cerebral arteries. J Neurosurg, 77, 927–40. [Shows the variability between individuals of boundaries between the territory of major cerebral arteries.] [Web of Science] [Medline] 
Van Gijn J, Kerr RS, Rinkel GJE (2007). Subarachnoid haemorrhage. Lancet, 369, 306–18. [Review.][CrossRef] [Web of Science] [Medline] 
Wahlgren L, et al. (2008). Thrombolysis with alteplase 3–4.5 h after acute ischaemic stroke (SITS-ISTR): an observational study. Lancet, 372, 1303–9. [Observational study suggesting substantial benefit of thrombolysis 3-4.5 h after stroke onset.][CrossRef] [Web of Science] [Medline] 
Warlow CP, et al. (2007). Stroke—practical management, 3rd edition. Blackwells, Oxford. [Comprehensive monograph about cerebrovascular disease, including a chapter about the historical background.]
Zhang-Nunes SX, et al. (2006). The cerebral beta-amyloid angiopathies: hereditary and sporadic. Brain Pathol, 16, 30–9. [Review.][CrossRef] [Web of Science] [Medline]