Spontaneous subarachnoid haemorrhage

The incidence of spontaneous subarachnoid haemorrhage (SAH) increases with age and is about 5–10/100 000 population/annum, being somewhat more frequent in women than men, and is particularly common in Finland (Linn et al. 1996). A ruptured saccular aneurysm is by far the most common cause (Vermeulen et al. 1992). Some SAHs are due to bleeding from an intracranial vascular malformation, a few are due to rarities (Table 27.16), and—depending on the intensity of investigation—in about 15 per cent no cause can be identified in life (Rinkel et al. 1993).

In about two-thirds of the ‘no cause’ category there is a characteristic pattern of restricted bleeding around the midbrain, so-called perimesencephalic haemorrhage, where the case fatality and risk of re-bleeding are both so low that there is not even post-mortem information about the cause (van Gijn et al. 1985a; Rinkel et al. 1991). Chronic or repeated subarachnoid bleeding can produce the rare syndrome of superficial haemosiderosis of the central nervous system: sensorineural deafness, cerebellar ataxia, pyramidal signs, dementia, and bladder disturbance (Fearnley et al. 1995).

Primary intraventricular haemorrhage is very unusual, except in premature babies. In adults, a cause is not always found. Some may be due to a vascular malformation in the ventricular wall (Gates et al. 1986; Darby et al. 1988). The clinical features are so similar to SAH that it can only be differentiated on CT, or at post-mortem.

Subdural haemorrhage is more often traumatic, or due to ventricular decompression for hydrocephalus, than spontaneous (but remembering that trauma can so easily be ignored or forgotten). Rupture of a vascular malformation in the dura or of a very peripheral aneurysm (mycotic much more likely than saccular), a haemostatic defect (particularly therapeutic anticoagulation), or a peripheral cerebral tumour can be responsible. It is also a very rare complication of lumbar puncture. Often no convincing cause is found.

Subarachnoid haemorrhage (SAH) is commonly provoked by exertion and very rarely occurs during sleep (Ferro and Pinto 1994; Vermeer et al. 1997). Almost always there is sudden headache, usually severe and generalized, often described like an unexpected blow on the head, and not lasting less than an hour (Vermeulen et al. 1992; Warlow et al. 2000c; Schievink 1997). Loss of consciousness occurs in about half the patients but may only be brief. Nausea and vomiting are less common. Partial or generalized seizures occasionally occur at around the onset (Pinto et al. 1996).

Early focal symptoms and signs suggest an associated intracerebral haematoma, or local pressure from an aneurysm (e.g. posterior communicating artery aneurysm causing a third nerve palsy), later they are more likely due to delayed cerebral ischaemia. Meningism takes a few hours to develop and may not do so at all in deeply unconscious patients. In severe cases, however, the whole back becomes painful and stiff, and on occasion the pain radiates down the legs, mimicking sciatica. Preretinal and subhyaloid haemorrhages may occur, probably as a result of the acute rise in intracranial pressure being transmitted to the retinal veins (Keane 1979).

There may be a mild fever and reactively raised blood pressure. Patients are often irritable and photophobic and the headache lasts for days, if not for some weeks. SAH is one cause, albeit an unusual one, of sudden or relatively sudden death, and about 15 per cent of SAH patients die before receiving any medical attention (Bonita and Thomson 1985). The patients' state is usefully graded using the World Federation of Neurological Surgeons' Scale.

Some patients have experienced similar sudden headaches before, perhaps because of so-called ‘warning leaks’ from saccular aneurysms, or maybe simply because a previous diagnosable SAH had not been reported or was misdiagnosed as something else (Verweij et al. 1988). In the community, of patients presenting with sudden and severe headache lasting more than an hour, about a quarter turn out to have SAH, about an eighth have some other serious neurological disorder, and the rest are a mixture of acute neck pain, migraine, and other benign, if undiagnosable conditions (Linn et al. 1994).


Quite often the history of onset is unclear and one cannot be sure that the headache had started all that suddenly, and the differential diagnosis is then quite wide (Table 27.27). But, with a good history of a sudden and unexpected headache, the most likely possibilities other than subarachnoid haemorrhage (SAH) are ‘thunderclap headache’, cerebellar or intraventricular haemorrhage, benign orgasmic or exertional cephalalgia, and migraine which uncharacteristically can sometimes start suddenly rather than gradually.

Any delay in the diagnosis of aneurysmal SAH can be overtaken by rebleeding, which is potentially disastrous and avoidable (Mayer et al. 1996; Neil-Dwyer and Lang 1997). Therefore, in patients with a suggestive clinical syndrome, an urgent unenhanced CT scan must be done which is the quickest, most informative, and cost-effective confirmatory investigation for blood in the subarachnoid space (Warlow et al. 2000c). On the first day it will show subarachnoid and/or intraventricular blood in over 90 per cent of cases, but less frequently thereafter; by a week or so the blood is gone, perhaps sooner with very mild SAH (Brouwers et al. 1992). Importantly, in addition, CT provides a baseline for the diagnosis of later rebleeding; shows any associated intracerebral, ventricular, or subdural haematoma which may require removal; shows any complicating hydrocephalus which may need treatment; and provides the best clue to which aneurysm has bled if more than one is found on later angiography (Adams et al. 1983; Vermeulen and van Gijn 1990).

The scan may also show calcification in the rim of an aneurysm or the characteristic pattern of benign perimesencephalic haemorrhage (Rinkel et al. 1991). CT may also show unsuspected evidence of traumatic head injury: brain contusions, soft-tissue swelling of the scalp, and skull fracture. The patient may either have fallen as a result of a spontaneous SAH, or any haemorrhage on the scan might be the result of a head injury (at which point the history must be reviewed to sort out which came first). Just occasionally, in almost brain dead patients, CT may seem to show SAH but this is probably blood in congested meningeal vessels (Opeskin and Silberstein 1998).

If CT is unavailable, or if it is available but no intracranial blood is visible, then a lumbar puncture must be done to confirm blood in the CSF and exclude bacterial meningitis (van der Wee et al. 1995). In fact, bacterial meningitis is unlikely if the headache onset really is sudden, but this may not necessarily be known if the patient is unconscious or confused. It is important to remember that blood, and xanthochromia, may well not appear in the lumbar CSF until about 12 hours after the onset (Vermeulen and van Gijn 1990). Therefore, if the history of sudden onset headache is definite, consciousness is normal, and there are no signs of infection (normal temperature, etc.), it is best to delay lumbar puncture for this period of time. Lumbar puncture can precipitate transtentorial herniation and cerebellar coning, particularly if there is an intracerebral or cerebellar haemorrhage, but this ought to be suspected from the presence of focal signs or coma (Duffy 1982).

As well as frank blood, the CSF must be inspected for xanthohromia, the result of breakdown of haemoglobin to pigments (oxyhaemoglobin, bilirubin) which have been released into the CSF some hours before (Vermeulen 1996; Beetham et al. 1998). These colour the supernatant of the spun-down CSF yellow. If the supernatant is clear to the naked eye, then spectrophotometry should probably be done as a more sensitive test for these pigments, but often it isn't. Failing to look for, and therefore missing, xanthochromia is a common cause for the incorrect assumption of a ‘traumatic tap’ when the CSF is bloody to the naked eye; this cannot be reliably confirmed by the well-worn method of observing clearing of blood across three consecutive tubes of CSF. A bloody CSF without xanthochromia (because the blood has entered the CSF at the time of the lumbar puncture) on spectrophotometry is the best indication of a traumatic tap. Red blood cells and xanthochromia clear from the CSF in about 2 weeks, or maybe less if the SAH is mild. In the acute stage the CSF glucose may be low, the protein slightly raised, and not uncommonly there is a slight excess of lymphocytes and polymorphs.

If the history is suggestive of SAH, but an early CT scan and CSF are both normal within days of onset, then SAH is most unlikely and the risk of a true SAH later is negligible (Wijdicks et al. 1988; Markus 1991). Therefore, cerebral angiography is not required under these circumstances. Similar headaches may recur (so-called thunderclap headaches) or they may become more obviously migrainous or tension-related. If, on the other hand, a patient presents more than about a week or two after a typical SAH story, then four-vessel (both carotid and vertebral arteries) catheter angiography is almost unavoidable, whatever the results of the CT scan and CSF. Despite a very small risk, it is still the definitive investigation to demonstrate saccular aneurysms , arteriovenous malformations and the very occasional intracranial arterial dissection or mycotic aneurysm, although CT angiography and MR angiography are both improving rapidly and may become preferred.

If an aneurysm is found, and angiography is not of all four vessels, then it is not impossible that an asymptomatic unruptured aneurysm has been demonstrated, which would lead to treatment of the wrong lesion. Therefore, unless it is very obvious from CT where an aneurysm has bled, and there is no intention of treating any coincidental unruptured aneurysm, angiography should aim to display all the cerebral vessels and be timed shortly before surgery (or coiling) is contemplated. An aneurysm may not be visualized if it is filled with thrombus, or if there is considerable vasospasm of the aneurysm-bearing artery, and so angiography may have to be repeated later (but not if CT has shown only perimesencephalic haemorrhage, see above).

Spinal subarachnoid haemorrhage

Spinal subarachnoid haemorrhage is very rare. It is due to a vascular malformation, haemostatic failure, coarctation of the aorta, inflammatory vascular disease, mycotic aneurysm, or a vascular tumour (e.g. ependymoma). There is sudden back or neck pain, radiating often to the head and down the legs, and meningism develops, so making for confusion with intracranial SAH. Complicating haematoma may compress the cord. Suspicion is aroused if the cerebral angiogram is negative and the patient develops spinal cord signs.


In the population at large, about half the patients with aneurysmal SAH die in the first month, of whom half of them die on the first day as a result of the initial bleed. Admitted patients, particularly to neurosurgical centres, do better because those who die very early and often those in coma are not included (Hop et al. 1997). Coma on admission, old age, and a large amount of blood on the initial CT scan are all associated with a worse prognosis (Kassel et al. 1990a,b). But, even after correcting for these adverse prognostic factors, any comparison between the management results of centres, or within centres at different times, is still almost certainly confounded by case mix and chance, and is a fruitless exercise, despite exactly these sorts of comparisons cluttering up the literature.

Early deterioration, sometimes leading to death in the first week or so, may be caused by aneurysmal rebleeding, hydrocephalus, delayed cerebral ischaemia, hyponatraemia, hypoxia, hypotension, seizures, and cardiac failure (Broderick et al. 1994; Warlow et al. 1996d) (Section 27.7.4 and Section 27.8.3). Therefore, in a deteriorating patient it is important not only to do an urgent CT scan (to reveal rebleeding, hydrocephalus, infarction) but also to check on the blood pressure, Pao2, and electrolytes.

Focal neurological deficits and, more commonly, cognitive deficits, behavioural disorders, seizures, anxiety, depression, and poor quality of life are frequent long-term sequelae (Saveland et al. 1986; Desantis et al. 1989; Hop et al. 1998). About 15 per cent of SAH survivors are dependent and 50 per cent disabled to some extent (Hijdra et al. 1987a; Hop et al. 1997). If no cause for SAH is found, early case fatality and risk of rebleeding are very low (less than 1 per cent per annum). This is probably because the diagnosis is sometimes wrong and the CSF obtained by traumatic tap, and because some patients have a perimesencephalic pattern of bleeding which has an exceptionally good prognosis (Rinkel et al. 1993).


In aneurysmal cases, without clipping or coiling, about 10 per cent rebleed within hours and another 30 per cent within a few weeks, during which there is no particularly high-risk time; no known factors predict further rebleeding, which is more likely to be fatal than the first bleed (Hijdra et al. 1987b; Hijdra et al. 1988). Subsequently, the rebleeding rate is about 2–3 per cent per annum (Winn et al. 1977). Deterioration is usually sudden, with reduced conscious level and, like the original bleed, there are few, if any, focal neurological features. But if the patient is being ventilated, these signs will not be seen and the clue is sudden fixed dilatation of the pupils (Hijdra et al. 1987b). Rebleeding can only be diagnosed by repeat CT scan to show fresh haemorrhage, and even then it can be difficult to be sure. Repeat lumbar puncture is not helpful and sometimes dangerous (i.e. if deterioration is actually due to intracerebral haematoma).

The breakdown of the fibrin clot sealing an aneurysmal leak can be inhibited, and so the risk of rebleeding reduced before the aneurysm is clipped or coiled, by prophylactic antifibrinolytic drugs, such as ε-aminocaproic acid and tranexamic acid. Unfortunately these drugs exacerbate delayed cerebral ischaemia and so there is no overall effect on outcome (Roos et al. 1999).

Ruptured arteriovenous malformations have a lower case fatality than aneurysmal SAH and are less likely to rebleed, certainly in the early period after the initial haemorrhage and if they have not already bled (Crawford et al. 1986; Ondra et al. 1990; Mast et al. 1997). Exactly which features of the vascular anatomy, or other factors, are associated with particularly high risks of bleeding or epilepsy is not clear (Duong et al. 1998).

Complications of subarachnoid haemorrhage and their management

Hydrocephalus, due to blood obstructing CSF flow, occurs within days of onset in about 20 per cent of patients. It may be asymptomatic, but if there is no other reason for clinical worsening and the conscious level is deteriorating, and particularly if the ventricles are enlarging on repeat CT, then temporary external ventricular drainage may lead to dramatic improvement. Ventriculitis, and possibly rebleeding if the aneurysm has not been dealt with first, are complications (van Gijn et al. 1985b; Hasan et al. 1989; Heros 1989). Therefore, in patients who do not have blood obstructing the ventricles or any brain shift, repeated lumbar puncture may possibly be a safer alternative (Hasan et al. 1991). Months or years later, organized thrombus and fibrosis in the CSF pathways can lead to the syndrome of normal-pressure hydrocephalus.

Delayed cerebral ischaemia appears 4–14 days after onset in about 25 per cent of cases and has a bad prognosis. Poor initial clinical state, large quantities of subarachnoid or intraventricular blood on CT, hyponatraemia, and the use of antifibrinolytic drugs are all risk factors (Wijdicks et al. 1985; Hijdra et al. 1986; Adams et al. 1987; Hasan et al. 1990).

Cerebral ischaemia seems to be caused by vasospasm, often with structural changes in the vessel wall, of one or more cerebral arteries, not necessarily the one bearing the ruptured aneurysm. It is a much less frequent phenomenon in non-aneurysmal subarachnoid haemorrhage (Kassell et al. 1985). Clinical onset is usually gradual with deteriorating conscious level accompanied, or followed within hours, by evolving focal neurological signs in most cases. To make the diagnosis, there should be no other explanation for deterioration, and in particular a repeat CT should not show any rebleeding (Hijdra et al. 1986). Plasma volume expansion and induced hypertension have been suggested as treatments but are not always effective (Awad et al. 1987). Fortunately, the risk of delayed cerebral ischaemia can be reduced and the overall outcome improved by prophylactic calcium blockers, specifically nimodipine, 60 mg 4-hourly, administered orally or by nasogastric tube, for 21 days (Feigin et al. 1998). If this causes hypotension, then the dose should be reduced. There is no good evidence to support intravenous nimodipine, which is particularly likely to cause unacceptable hypotension.

Hyponatraemia occurs in about one-third of patients in the first week or two after subarachnoid haemorrhage and is related to the severity of the initial presentation. It is not usually due to inappropriate ADH secretion and dilutional hyponatraemia but to ‘salt wasting’, i.e. excess loss of both salt and water by the kidneys with a decrease in plasma volume. Below a plasma sodium of about 125 mmol/l, correction is necessary, not usually by water restriction, but by plasma volume expansion (using dextrose- saline or plasma) while monitoring central venous or pulmonary wedge pressure (Wijdicks et al. 1985; Berendes et al. 1997).

Intracerebral haematoma may cause a focal deficit initially, or later, and may be worth removing if there is coma, clinical deterioration, and brain shift, particularly if the ruptured aneurysm or vascular malformation can be dealt with at the same time (Heiskanen et al. 1988).

Unruptured saccular aneurysms

Unruptured aneurysms, either discovered while investigating SAH or in some other way, should normally be clipped or coiled if they are symptomatic; for example, eye pain from a posterior communicating artery aneurysm (Raps et al. 1993). Unruptured asymptomatic aneurysms are more problematic because their risk of rupture is, on average, perhaps about 4 per cent per annum if over 10 mm in diameter, and rather less than 1 per cent per annum for smaller aneurysms (Rinkel et al. 1998). But the rates were much lower in the recent large international series (International Study of Unruptured Aneurysm Investigators 1998). These risks have to be set against the risk of clipping or coiling, about 8 per cent dead or dependent (Raaymakers et al. 1998).

Also, account must be taken of how long the patient is likely to live with an untreated aneurysm, and how an individual patient feels about the competing risks. The problem of screening high risk of aneurysm populations (such as first-degree relatives of patients with ruptured aneurysms, families with multiple members affected by aneurysmal rupture, and polycystic kidney families), at what age, how often, and with what imaging technology, is even more fraught, and no firm conclusions are possible.