Clinical manifestations

Presentation and course

Headache in subarachnoid hemorrhage. The characteristic presenting complaint in subarachnoid hemorrhage is the sudden onset of a painful headache reaching its peak intensity instantaneously or within minutes, often described by patients as “the worst headache of my life”. This stereotypical presentation with a sudden, severe, “thunderclap”-type headache occurs in 85% to 95% of patients (19; 56). The headache is usually followed by pain radiating into the occipital or cervical region. Typically, the headache is bilateral, but occasionally the headache can be lateralized to one side. Vomiting preceding the onset of headache has been frequently reported. As many as 17% of patients present with a “warning headache” or “sentinel headache” that precedes the thunderclap headache. Such warning headaches are thought to result from small “sentinel” bleeding from an underlying aneurysm. Sentinel bleeds increase the odds of early rebleeding after the index subarachnoid hemorrhage by 10-fold (04).

Other presenting symptoms of subarachnoid hemorrhage. Meningismus frequently accompanies the headache of subarachnoid hemorrhage. Meningismus develops as blood flows into the spinal subarachnoid space. Some patients complain of pain radiating down the legs due to pooling of blood in the lumbar cistern and irritation of nerve roots. Impaired alertness, stupor, or coma can be manifestations in severe cases. Increased intracranial pressure with accompanying decrease in cerebral perfusion pressure and herniation syndromes during the ictus can lead to a sudden loss of consciousness as the presenting event, often accompanied by intermittent myoclonus or posturing, and this phenomenon may be difficult to distinguish from seizures, which can also occur during the initial presentation.

Clinical signs that accompany the presenting symptoms often include a mild temperature elevation and hypertension. Vitreous hemorrhage, known as Terson syndrome, may occur in some patients with subarachnoid hemorrhage and is associated with poor prognosis. Cranial nerve palsies and focal neurologic deficits may also be present--classically 3rd nerve compression from a posterior communicating artery aneurysm--though 3rd nerve compression can also occur with posterior cerebral or superior cerebellar artery aneurysms. Uncal herniation causing pupillary dilatation, third nerve palsy, and deteriorating mental status is an ominous sign. Unilateral or bilateral lateral rectus (6th nerve) paresis may signify increased intracranial pressure or early hydrocephalus and is considered a nonlocalizing sign. Focal deficits arise particularly in the case of concomitant intracerebral hemorrhage (more common with middle cerebral artery aneurysm rupture). Finally, patients may present with various cardiac and pulmonary abnormalities, including stunned myocardium, neurogenic pulmonary edema, hypotension, various arrhythmias, ST segment changes, and even cardiopulmonary arrest.

Prognosis and complications

The course of subarachnoid hemorrhage is fraught with complications, and mortality is high, at about one-third of patients (19; 46). On the other hand, outcomes can be excellent, with return to full independence, though often with persisting subtle cognitive effects. The prognosis is highly dependent on the initial severity. Patients with subarachnoid hemorrhage are classified according to standardized grading systems that correlate with mortality and outcome. The benefits of using a classification system for patients with subarachnoid hemorrhage include the potential for accurate description and communication of a patient's baseline neurologic status and a coarse measure of the severity of the hemorrhage. Although scoring on the classification systems has been associated with prognosis, the scoring systems should not be used in isolation to determine the individual patient’s chance of a good outcome. Furthermore, many patients may improve by several points on the classification scales after initial neurologic (including external ventricular drain placement for hydrocephalus) and systemic resuscitation (67). Finally, with the exception of obvious irreversible catastrophic neurologic injury at presentation, prediction of long-term neurologic recovery and accompanying decisions regarding goals of care should only be made with great caution and consideration, particularly within the first 72 hours of presentation.

The Hunt and Hess classification system is perhaps the best known and most widely used grading system in the United States (45). The Hunt and Hess grade at presentation is predictive of mortality and functional outcome, reflected in the Glasgow Outcome Score, in which 1 represents death, 2 represents persistent vegetative state, 3 severe disability, 4 moderate disability, and 5 a low level of disability (Table 1).

Table 1. Hunt and Hess Classification of Subarachnoid Hemorrhage

The World Federation of Neurological Surgeons (WFNS) grading system is also used and is associated with outcome. This system uses the Glasgow Coma Scale to evaluate level of consciousness and uses the presence or absence of major focal neurologic deficits to distinguish grade 2 from grade 3 (98). The WFNS scale has less interobserver disagreement compared to Hunt and Hess score (24).

Table 2. World Federation of Neurologic Surgeons Grading of Subarachnoid Hemorrhage

Further refinement of prognostication, from the Subarachnoid Hemorrhage International Trialists’ group (SAHIT), based on pooled datasets from multiple prospective studies comprising over 10,000 patients, has resulted in prediction models with very good ability to predict functional outcome based on detailed initial clinical features (76; 46).

Clinical vignette

A 46-year-old woman with a past medical history of tobacco use and hypertension presented to the emergency room with severe headache that reached maximum intensity within moments after onset, accompanied by nausea, vomiting, and somnolence. The headache initially localized in the back of her head and quickly spread to the top of the head and behind her eyes. Physical examination showed a systolic blood pressure of 190 mmHg, heart rate of 85 beats per minute, and a mild systolic murmur at the apex. Neurologic examination exhibited a somnolent patient. When aroused, she was alert and oriented. She had mild paresis on the right lower half of her face and right limbs. Her sensation to pinprick and her reflexes were slightly decreased on the paretic side.

Biological basis

Etiology and pathogenesis

Subarachnoid hemorrhage is defined as the presence of blood within the subarachnoid space located between the arachnoid membrane and the pia mater. A subarachnoid hemorrhage may be categorized as traumatic or spontaneous (typically due to aneurysmal rupture). Traumatic subarachnoid hemorrhage is much more common than spontaneous or nontraumatic subarachnoid hemorrhage, and it is not associated with the same complications. The current review focuses on spontaneous subarachnoid hemorrhage.

Rupture of a cerebral aneurysm, most commonly a saccular or “berry” aneurysm, is the most common source of spontaneous subarachnoid hemorrhage, accounting for 75% of cases. The precise pathogenesis of saccular aneurysms is not completely known. Degeneration of the internal elastic lamina occurs through mechanisms that may involve inflammation. Risk factors for aneurysm occurrence include hypertension, moderate to heavy alcohol consumption, cigarette smoking, and abuse of sympathomimetics (19; 32). A minority of patients suffering spontaneous subarachnoid hemorrhage are found to have other conditions that increase the risk of saccular aneurysms (Table 3). These include autosomal dominant polycystic kidney disease, glucocorticoid-remediable aldosteronism, fibromuscular dysplasia, Moyamoya disease, and Ehler-Danlos type IV. Suffering a previous subarachnoid hemorrhage is among the strongest predictors of subarachnoid hemorrhage. Those with a history of subarachnoid hemorrhage form new aneurysms at a rate of 1% to 2% per year (05). Ninety percent of aneurysms develop in the anterior circulation, most commonly at the bifurcation of the anterior communicating artery and the anterior cerebral artery (30%), the internal carotid artery and posterior communicating artery (25%), the middle cerebral artery bifurcation (20%), the internal carotid artery bifurcation (8%), and other locations (7%). Ten percent of aneurysms arise from the posterior circulation.

Ten to twenty percent of patients with a spontaneous subarachnoid hemorrhage have an angiogram that fails to reveal a source of the hemorrhage (05). Twenty-four percent of them have an occult aneurysm; this number increases to 50% when excluding patients with obvious causes of bleeding and those with a typical pattern of perimesencephalic bleeding (47). These data justify repeat angiographic imaging prior to or after discharge in patients with a high suspicion of an aneurysmal source or subarachnoid hemorrhage, in whom initial angiographic imaging was unrevealing.

Perimesencephalic nonaneurysmal subarachnoid hemorrhage is typically distributed in the perimesencephalic cisterns anterior to the brainstem with potential extension to the ambient cistern and basal aspect of the Sylvian fissure. It constitutes 5% of subarachnoid hemorrhage, has a much more benign course, and has excellent prognosis when compared to aneurysmal subarachnoid hemorrhage. A typical presentation involves a sudden headache during a Valsalva maneuver. Long-term follow-up studies have shown virtually no risk of recurrence; only a few recurrences have been reported in the literature. However, vascular studies should be performed to rule out an underlying aneurysm, as about 10% of posterior circulation aneurysmal ruptures can present with perimesencephalic bleed pattern (63). The source of bleeding is not defined in the majority of cases but is thought to be related to the rupture of perforating arteries, bleeding from venous sources, or bleeding from basilar artery vasa vasorum. MRI of the cervical spine is often performed to rule out vascular malformation, however, one study showed low diagnostic yield if clinically not suspected (89).

Convexity subarachnoid hemorrhage, with blood products appearing focally in a pial distribution over a hemispheric surface, has been increasingly recognized with the advent of MR susceptibility brain imaging (20). These hemorrhages, located distant from the circle of Willis and the large artery branches where saccular aneurysms occur, are due to a different set of vascular pathologies. Besides trauma, causes include amyloid angiopathy, dural arteriovenous fistulas, small subpial vascular malformations, venous or sinus thrombosis, cerebral vasculitis, reversible vasoconstriction, or posterior encephalopathy syndromes (20; 11). Rarely, there may be an aneurysm in a pial vessel from an infective, traumatic, or neoplastic cause. An association of convexity subarachnoid hemorrhage with arterial stenosis from intracranial carotid atherosclerosis is also reported (38), and superficial ischemic infarctions can produce local subarachnoid bleeding. Commonly, these hemorrhages manifest in elderly patients with transient neurologic deficits or partial seizures, rather than severe sudden headache. Recurrence of bleeding is common (11). Repeated convexity subarachnoid hemorrhages are recognized by superficial siderosis on brain MR imaging, which along with parenchymal microhemorrhages, constitute a core radiological sign of amyloid angiopathy.

Although saccular aneurysms are responsible for the majority of spontaneous subarachnoid hemorrhages, a wide variety of etiologies can be responsible (Table 3). Awareness of and familiarity with these entities is key in their recognition and treatment.

Table 3. Etiologies of Subarachnoid Hemorrhage Beyond Saccular Aneurysms

Genetics. It has long been recognized that a positive family history is associated with a higher incidence of cerebral aneurysm formation and subarachnoid hemorrhage and a higher risk of rupture of known aneurysms (44; 33). A familial intracranial aneurysm trait should be considered when two or more first- to third-degree relatives have cerebral aneurysms. A number of chromosomal loci and specific genes have been implicated in aneurysm formation. Genetically determined conditions such as Polycystic kidney disease, Neurofibromatosis type 1, and Moyamoya disease are recognized risk factors for formation of intracranial aneurysms (Table 4). Intracranial aneurysm prevalence is increased 5-fold in autosomal dominant polycystic kidney disease, with reduced expression of the gene products polycystin-1 and polycystin-2 in vascular tissues (87). Mutations threatening the integrity of the vascular wall might be expected to lead to aneurysm formation and rupture, and, indeed, known connective tissue diseases including Ehlers-Danlos syndrome type IV and Marfan syndrome can lead to intracranial aneurysms. Studies have shown associations between polymorphisms in collagen genes and intracranial aneurysms (64). Several genes in gene families important to vascular development have also been associated with aneurysm development, including SMAD-3 (84), thrombospondin type 1 domain-containing protein 1 (92), and angiopoietin-like 6 (09). The rapid progression in genetic technology is accelerating the recognition of the wide range of genes influencing cerebrovascular development and structure that converge in determining the risk of intracranial aneurysm formation.

Table 4. Risk Factors for Aneurysmal Formation

Epidemiology

The annual incidence of spontaneous subarachnoid hemorrhage is 2 to 25 per 100,000 people, and approximately 30,000 spontaneous subarachnoid hemorrhages occur in the United States annually (05). The mean age of rupture is 55 years, with the highest incidence of aneurysmal subarachnoid hemorrhage between 40 to 60 years. The incidence of subarachnoid hemorrhage differs by race and ethnicity, being more common in African-American populations and women (05; 19; 32).

The risk burden for spontaneous subarachnoid hemorrhage largely parallels that of intracranial aneurysms. Other factors associated with increased risks of subarachnoid hemorrhage include black race, Hispanic ethnicity, hypertension, current smoking, alcohol abuse, use of sympathomimetic drugs, as well as larger aneurysms (56). Surprisingly, diabetes mellitus has been found to reduce the risk of subarachnoid hemorrhage in population-based studies (48).

The ISUIA study addressed treatment of unruptured aneurysms detected in patients with and without previous subarachnoid hemorrhage (102). Patients older than 50 years with large posterior circulation aneurysms are at the greatest risk for both rupture and repair. The 5-year rupture risk varies by aneurysm location and size for patients with no history of subarachnoid hemorrhage (Table 5):

Table 5. Aneurysm Size, Location, and 5-Year Rupture Risk

The decision whether to treat an unruptured aneurysm should involve a discussion of the risks and benefits of rupture and treatment. Small unruptured aneurysms are often followed with serial imaging to detect growth. However, there are no clear-cut data to guide how often to repeat imaging studies. The original aneurysm size has been shown to be a predictor of aneurysm growth, supporting the practice of imaging follow-up studies at regular intervals in these cases (10). A study of aneurysm growth found that initial aneurysm size, dome/neck ratio multilobularity, and patient smoking are risk factors for aneurysm growth (07). However, gender, age, hypertension, past medical history, or family history of subarachnoid hemorrhage were not associated with aneurysmal growth. Smoking deserves particular attention as a modifiable risk factor for growth of small aneurysms.

Prevention

Greater rates of discovery of unruptured intracranial aneurysms and prophylactic treatment of higher risk aneurysms with endovascular or open surgical procedures may be contributing to a discernable year-over-year decline in the number of aneurysmal subarachnoid hemorrhages relative to nonaneurysmal subarachnoid hemorrhage admissions, based on the U.S. National Inpatient Sample database (2004 – 2014) (91). Risk factors modification, such as smoking cessation, blood pressure control, and elimination of high-risk behaviors (cocaine and sympathomimetic use, heavy alcohol consumption) may also help prevent aneurysm rupture (17). Education of physicians and the general public regarding signs and symptoms of subarachnoid hemorrhage might result in fewer misdiagnoses and earlier neurosurgical referral. This may reduce the incidence of catastrophic subarachnoid hemorrhage.

Familial clustering of aneurysms has been described, with studies demonstrating that first-degree relatives of patients with aneurysmal subarachnoid hemorrhage are between two and five times more likely to develop subarachnoid hemorrhage. According to the American Heart Association / American Stroke Association guidelines in 2012, noninvasive screening of patients with at least one first-degree relative with aneurysmal subarachnoid hemorrhage or patients with previous aneurysmal subarachnoid hemorrhage may be reasonable; however, more studies are needed to weigh the risks and benefits (19). It is advisable to start screening high risk individuals at 10 years of age or earlier than the youngest family member diagnosed with cerebral aneurysm (44).

Differential diagnosis

Confusing conditions

Headache, the primary symptom of subarachnoid hemorrhage, is a common presenting complaint in office and emergency room settings, but only a small fraction of patients with headache have a subarachnoid hemorrhage, creating a diagnostic challenge. The headache of subarachnoid hemorrhage characteristically is severe, very rapid in onset, and holocephalic or posterior, but atypical presentations occur. Sudden onset severe headaches (“thunderclap headaches”) have a wide differential diagnosis that must be carefully considered with query for characteristic clinical features and performance of appropriate diagnostic testing (Table 6). The difficult challenge in the emergency setting is how to recognize which headache cases warrant further diagnostic testing, with brain imaging and lumbar puncture in particular, among all the patients presenting and seeking relief from a severe headache. In a multicenter study of 10 university-affiliated emergency departments, among alert and oriented patients with nontraumatic headache peaking in intensity within 1 hour of onset, after exclusion of patients with three or more similar previous recurrent headaches as well as those with other known causes for headache, there was a 6.2% rate of confirmed subarachnoid hemorrhage among remaining patients (81). A selection rule was applied to this population, calling for diagnostic investigation in patients with new severe headache and these exclusions, with the presence of any of the following high-risk variables:

This rule, termed the “Ottawa subarachnoid hemorrhage rule,” resulted in a very high sensitivity (100%) and low specificity (15.3%) for subarachnoid hemorrhage in this selected population of headache patients (81). Testing this rule in a population of 913 emergency department patients in China with recent onset acute headaches again found a 100% sensitivity with a specificity of 37% (105). The low specificity points to the common clinical features shared by several other conditions that can mimic subarachnoid hemorrhage (Table 6). Implementation of this rule has resulted in decreased rates of lumbar puncture and hospital admissions among patients with severe nontraumatic sudden headaches meeting the criteria for application of the rule (80).

Table 6. Differential Diagnosis of Severe Sudden Headache (“Thunderclap Headache”)

When brain CT imaging shows features suggesting subarachnoid hemorrhage, in addition to aneurysmal rupture and the various etiologies of spontaneous subarachnoid hemorrhage outlined in Table 3, other diagnoses must be considered:

Traumatic subarachnoid hemorrhage. Traumatic subarachnoid hemorrhage is usually recognized with a clear history of trauma. It typically affects the cerebral convexities and is often accompanied by other signs of injury such as orbital frontal contusions, skull fracture, or external scalp trauma.

Pseudo-subarachnoid hemorrhage. Pseudo-subarachnoid hemorrhage is defined as increased density of the basal cisterns and subarachnoid spaces on computed tomography, not due to blood products. It is due to radiographic mimics of subarachnoid hemorrhage, which could be due to pyogenic leptomeningitis, intrathecal administration of contrast material, high-dose intravenous contrast, and diffuse cerebral edema (39).

Diagnostic workup

Misdiagnosis of subarachnoid hemorrhage is thought to be as high as 12% (particularly in good-grade patients with mild symptoms). The most common diagnostic error is failure to obtain a noncontrast head CT. Failure to diagnose subarachnoid hemorrhage is associated with 4-fold increase in risk of death or severe disability. Additionally, in a prospective study that included 401 patients with subarachnoid hemorrhage, a delay in diagnosis in low-grade subarachnoid hemorrhages (Hunt and Hess groups 1 and 2) was independently associated with poor outcome, delayed cerebral ischemia, and procedure-related complications (75). Because treatment is urgent and the consequences of misdiagnosis are severe, a high index of suspicion for subarachnoid hemorrhage should be maintained (78).

Imaging studies.

Noncontrast CT scan. Noncontrast CT scan has a sensitivity of almost 100% when performed within 6 hours of onset, 85% sensitivity at 5 days, and 50% sensitivity at 1 week (30). False negatives can occur with low hematocrit levels. The thickness of the subarachnoid hemorrhage clot and the presence of intraventricular hemorrhage both predict the risk of vasospasm and delayed cerebral ischemia. The modified Fisher scale incorporates into its grading system the risk of symptomatic vasospasm or delayed cerebral ischemia due to both subarachnoid hemorrhage and intraventricular hemorrhage.

Table 7. Modified Fisher Grading Scale for Subarachnoid Hemorrhage

Cerebral angiogram. Digital subtraction cerebral angiography (DSA) is the gold standard for ruling out a ruptured cerebral aneurysm, for defining the relevant neuroanatomy, and possibly for providing immediate endovascular treatment. It has very low risks, with reported complication rate of 1% to 2.6% (renal failure, arterial occlusion, pseudo-aneurysm, and hematoma formation). The angiogram must visualize all intracranial vessels with multiple views, and the origins of both posterior inferior cerebellar arteries should be demonstrated. A 6-vessel angiogram is appropriate, including contrast to the external carotid arteries, to improve the diagnostic yield for the unusual dural arteriovenous fistulas as well as aneurysms. The goals of angiography are to determine the cause of subarachnoid hemorrhage and, if aneurysmal, to delineate the anatomy of the aneurysm (neck, nearby vessels, etc.), to determine if multiple aneurysms are present, to assess the degree of vasospasm, and to consider and potentially perform a therapeutic intervention. Ten percent to 20% of patients with subarachnoid hemorrhage have negative angiograms. Angiography should be repeated as false negative results can be due to vasospasm, thrombosis of the aneurysm, small aneurysm, or blood in the cisterns. However, when to repeat the cerebral angiogram is debatable. Some advocate repeating it within 6 to 14 days for early management, and others recommend waiting 4 to 6 weeks post ictus to allow for the vasospasm and the hematoma to subside (61). There is no need to repeat angiography for typical perimesencephalic hemorrhages because they are not usually associated with underlying aneurysm (52).

Though cerebral angiography remains the gold standard test for detection of aneurysms, noninvasive testing with CT angiography often is chosen as the initial vascular imaging study in patients identified with subarachnoid hemorrhage, with DSA performed in cases with negative CTA or need to better define aneurysm anatomy for treatment planning (19).

CT angiography. The sensitivity of CT angiography for detecting cerebral aneurysms depends on the size of the aneurysm. Larger aneurysms are reliably detected, whereas the smallest aneurysms may require DSA for detection. Post-image processing can provide detailed 3-D images with morphologic data. Sensitivity has improved as CT technology advanced, so that modern 16 slice or 64 slice CTA techniques have approached 100% sensitivity for aneurysms greater than 3 mm in diameter (57; 106). CT angiography is superior to digital subtraction angiography for defining aneurysmal wall calcification, intraluminal thrombus, and the relationship of the aneurysm to bony architecture. In addition, it has fewer complications, is a noninvasive, fast, diagnostic tool, and can triage patients to endovascular versus surgical treatment of the aneurysm. In previously treated aneurysms, metal artifact can limit the interpretation of CT angiography.

Magnetic resonance angiography. Magnetic resonance angiography is sensitive for aneurysms larger than 5 mm. A meta-analysis found evidence of improving sensitivity over time, showing a 95% sensitivity overall for detection of cerebral aneurysms in 960 pooled patients, with most of the missed aneurysms being smaller than 5 mm in diameter; a trend was found towards improved sensitivity with higher field strength (3 Tesla) magnets (90). MRI has higher sensitivity than CT for the detection of subacute hemorrhage and should be used in the patient presenting 3 to 4 days after onset, with normal head CT. In addition, contrast is not needed, and MR scanning is free from radiation. Obtaining MRI of the brain and cervical spine with and without gadolinium is a key step in the diagnosis of aneurysmal negative spontaneous subarachnoid hemorrhage. Wall enhancement of the aneurysmal wall on MRI is a promising marker of aneurysmal rupture. This could be useful in identifying the culprit in patients with multiple aneurysms (61).

Lumbar puncture. Lumbar puncture should generally be performed if the history is suspicious for subarachnoid hemorrhage, but the head CT is negative. Though there is a theoretical risk of aneurysmal re-rupture due to changing transmural pressure over an aneurysm, this risk is small with diagnostic spinal taps, in which a small amount of CSF is removed. Commonly, the opening pressure is elevated, and the CSF appears grossly bloody. With subarachnoid hemorrhage, the red blood cell count remains relatively constant from the first tube to the fourth; this contrasts with traumatic lumbar puncture in which there is a diminishing red blood cell count in subsequent CSF tubes. Xanthochromia, a yellowish discoloration of the cerebrospinal fluid from hemoglobin degradation products, develops 2 to 4 hours after the ictus and persists for about 2 weeks. It may be detected visually or with high sensitivity by spectrophotometry. There is a progressive rise in white blood cell counts in CSF samples taken after the ictus. Protein is usually slightly elevated, whereas glucose is ordinarily normal. With time after aneurysm rupture, chemical meningitis can develop. The CSF profile of chemical meningitis is indistinguishable from infectious meningitis and is characterized by elevated white blood cells, including elevated polymorphonuclear cells and low glucose. However, this must be differentiated from the infectious ventriculitis that is a risk in patients with prolonged CSF diversion.

Because of the very high sensitivity of noncontrast head CT for subarachnoid blood in the early hours following onset of headache, there has been increasing interest in reevaluating whether lumbar puncture is a necessary test in cases of suspected subarachnoid hemorrhage (56). CT head within the first 6 hours of ictus is very sensitive in detecting subarachnoid hemorrhage. In a meta-analysis involving almost 9000 patients, noncontrast head CT had an overall sensitivity of 98.7% when performed within 6 hours of onset of headache (30). However, its sensitivity beyond this time window declines (12). Lumbar puncture and CSF analysis can definitively rule out subarachnoid hemorrhage but can produce false positive or misleading results. In addition, lumbar puncture is an invasive procedure that can be hard to perform, can have contraindications and risks, and can be uncomfortable for patients.

As a result, many physicians prefer omitting lumbar puncture and instead order a CT angiogram if the head CT is nonrevealing. A number of studies have examined whether noncontrast head CT alone, or followed by CT angiography, is sufficiently sensitive for aneurysmal subarachnoid hemorrhage in the emergency room setting so that lumbar puncture can be foregone when the CT-based imaging is negative. A metaanalysis showed a high reliability of noncontrast head CT scan when combined with clinical criteria for detecting subarachnoid hemorrhage in the emergency department setting (12). However, a structured literature review of 31 appropriate publications addressing this issue found insufficient evidence to conclude that noncontrast CT head alone is a safe approach but that noncontrast CT head followed by CT angiogram is a reasonable approach to the evaluation of select patients with possible subarachnoid hemorrhage in the emergency room setting (65). The practice of relying on CT head and CT angiogram alone is growing, supported by these findings.

However, it must be noted that CT angiography has low sensitivity for very small aneurysms and does not rule out nonaneurysmal causes of subarachnoid hemorrhage. Also, given that 1% to 2% of the population carry asymptomatic aneurysms, discovery of an aneurysm does not necessarily imply that a sentinel hemorrhage is the reason for the presenting headache; the aneurysm may be incidental or may be producing a headache without rupture. Thus, the gold standard for detecting subarachnoid hemorrhage for cases with high clinical suspicion and negative head CT scan, particularly for scans performed more than 6 hours after the headache onset, remains lumbar puncture and CSF analysis.

Laboratory studies. Routine testing should include coagulation studies and platelet count. Toxicology screening should be performed in at-risk populations.

Management

Early management. The initial management of subarachnoid hemorrhage is directed at diagnosing the etiology, responding to immediate threats to the patient’s survival, and preventing life threatening or disabling complications. When a ruptured aneurysm has been identified as cause of the hemorrhage, rebleeding is generally the greatest initial risk.

Rebleeding. Rebleeding following subarachnoid hemorrhage accounts for about 50% of the mortality of the condition. Early management of this syndrome is focused on avoiding this preventable complication. There is a 3% to 4% risk of rebleeding during the first 24 hours, a 2% risk the second day, and ongoing risk each subsequent day, amounting to a 15% to 20% rebleeding risk within the first 2 weeks and up to 50% risk during the first 6 months if the aneurysm is not repaired (05). Rebleeding is the most treatable cause of poor outcome after subarachnoid hemorrhage. Risk factors for rebleeding include aneurysm size, longer duration to securing the aneurysm, severity of the initial bleed, elevated blood pressure (which may be causative or secondary to the rebleeding), seizures, loss of consciousness at ictus, sentinel bleed, and the presence of intraventricular hemorrhage.

Although the only definitive method to prevent rebleeding is to secure the aneurysm by excluding it from the intracranial circulation by neurosurgical clipping or endovascular obliteration, initial medical management decisions can modulate this risk, as reviewed below.

Blood pressure management. Elevated blood pressure may contribute to aneurysm re-rupture. Blood pressure should be monitored and controlled to balance the theoretical risk of hypertension-related rebleeding and need to maintain cerebral perfusion pressure (05). In patients who are awake, and, thus, can be inferred to have a cerebral perfusion pressure within tolerable limits, it is reasonable to control the systolic blood pressure with short-acting titratable medications, to a goal systolic blood pressure of less than 160 mmHg (19; 29). Excess blood pressure lowering may be harmful, particularly if intracranial pressure is elevated (66). If intracranial pressure is monitored, it is safer to lower the systolic blood pressure below 140 mmHg, provided the cerebral perfusion pressure remains over 60 mmHg. Lowering of pressure is usually done with a titratable drip such as intravenous nicardipine, labetalol, or clevidipine. Nitroglycerine and sodium nitroprusside are vasodilators and should be avoided as they may increase cerebral blood flow and intracranial pressure. Once the aneurysm is secured, blood pressure parameters can be liberalized as patients enter the vasospasm period (typically 3 to 14 days after rupture). Further studies are needed to better elucidate acute management of blood pressure in aneurysmal subarachnoid hemorrhage.

Seizure prophylaxis. Because seizures can increase intracranial pressure and may lead to aneurysm re-rupture, it has been a common practice to provide seizure prophylaxis in subarachnoid hemorrhage patients, particularly prior to aneurysm repair. However, there is no high-quality evidence showing that seizure prophylaxis improves outcome in subarachnoid hemorrhage. There is evidence for high geographic variability in use of prophylactic anticonvulsants for subarachnoid hemorrhage, and that such use was associated with worse outcome, even after adjustment for age, neurologic grade, and admission systolic blood pressure (86). In this study, 65% of patients received prophylaxis, and phenytoin and phenobarbital were the most commonly prescribed antiepileptics. Patients treated with prophylactic anticonvulsants had higher odds ratios for developing cerebral vasospasm, cerebral infarction, and neurologic deterioration (86). Phenytoin and phenobarbital cause an increment in the enzymatic activity of CYP450, which can subsequently increase the metabolism of nimodipine, potentially explaining the higher incidence of delayed cerebral ischemia and worse outcomes (32). Other data have shown that patients exposed to phenytoin have worse cognitive outcomes at 3 months, although eventually these patients slowly recover to the level of those not exposed (72). One article reflecting current practice in seizure prophylaxis in centers with high volumes of patients with subarachnoid hemorrhage showed that most (two thirds) used prophylactic antiepileptic treatment for 7 to 14 days duration, and levetiracetam was the most used antiepileptic (25). A systematic review performed by Fang and colleagues demonstrated that the use of levetiracetam in patients with acute neurologic injury, including aneurysmal subarachnoid hemorrhage, did not show a reduction in seizure incidence, despite significant concern for bias and subtherapeutic doses of levetiracetam (34).

A consensus conference organized by the Neurocritical Care Society suggested that the risks and benefits of seizure prophylaxis should be considered, particularly in patients with unsecured ruptured aneurysms, given the devastating potential consequences of seizures; if prophylactic anticonvulsants are used, they should be given for a short course (3 to 7 days), and phenytoin should be avoided (27).

Until new guidelines are released, we recommend following the American Heart Association/American Stroke Association guidelines, which suggest that antiepileptics may be considered for prophylaxis in the acute phase of subarachnoid hemorrhage; however, they should not be routinely used as long-term treatment unless there are risk factors for the future development of seizures (19).

Antifibrinolytics. Antifibrinolytics (epsilon-aminocaproic acid 4 g intravenous load, then 24 g per day continuous infusion; or tranexamic acid 1 g intravenously immediately, then 1 g every 6 hours until occlusion) have been used to reduce the incidence of early rebleeding when early definitive aneurysm treatment is not available (32). In some studies, these medications have reduced rebleeding from 10% to 2% (43). Because there is some evidence that antifibrinolytics are associated with cerebral vasospasm and that prolonged use is associated with an increased risk of thromboembolism, their use is not uniform amongst institutions.

Post and colleagues prospectively studied the effect of ultra-early tranexamic acid in patients with subarachnoid hemorrhage (83). A total of 480 patients were randomized to receive 1 g of tranexamic acid as a bolus followed by 1 g every 8 hours for a maximum of 24 hours or until the aneurysm was secured, whichever occurred first. The study showed no difference in clinical outcome as measured by the modified ranking scale at 6 months. However, when definitive treatment of the aneurysm is unavoidably delayed and there are no other contraindications to treatment (presence of coagulopathy, recent myocardial infarction, ischemic stroke, pulmonary embolism, or deep vein thrombosis), short-term therapy with tranexamic acid or aminocaproic acid is reasonable, though evidence of long-term benefit is lacking (19; 29). Antifibrinolytics should be discontinued prior to catheter angiography because they can precipitate catheter-induced vasospasm.

Regionalized care. Management of subarachnoid hemorrhage requires a team of experienced physicians, including vascular neurosurgeons, endovascular neurointerventionalists, and neurointensivists. Repair of the aneurysm is only the first step in the care of a patient with subarachnoid hemorrhage. Appropriate management of delayed complications is essential to good outcome. Low volume hospitals should consider transferring their patients to a high-volume hospital where a multidisciplinary neurointensive care team can provide optimal care (19).

Secure the aneurysm. Untreated ruptured aneurysms portend a 1-year mortality rate of 65% with a median survival time of 20 days from symptom onset (51). Rapidly securing the aneurysm is imperative to lessening this high mortality. The optimal method of aneurysm protection for each patient should be individualized based on several factors, including (1) aneurysm morphology and location, (2) patient characteristics, and (3) the experience of the treating facility. For good-grade patients, early treatment is recommended to decrease the 67% mortality rate that is associated with rebleeding (05). It is not certain that early aneurysm control benefits the overall outcome of patients presenting with poor Hunt and Hess grades (IV and V). These patients may not be stable enough to undergo any sort of emergent intervention. When stable, aneurysm control with an endovascular approach is usually logistically easier to perform provided it is technically feasible. Studies have shown that coiling a poor grade aneurysm within 24 hours of subarachnoid hemorrhage was associated with better outcomes compared to after 24 hours (60; 33).

Endovascular techniques are usually the rule for aneurysms affecting the posterior circulation. These techniques include coil embolization, stent-assisted coil embolization, onyx embolization, and placement of flow-diverting stents. Complications related to endovascular treatment include complications related to catheter access, including local hematoma, aortic and cervicocephalic vessel dissection, and myocardial infarction and the specific complications related to the coiling, including perforation or rupture of aneurysm, migration of the coil, herniation of the coil, and local thromboembolism (causing cerebral ischemia) (05; 82). Rupture of aneurysm is reported in up to 5% of patients. Risk factors for rupture include small aneurysm size, middle cerebral artery location, hypertension, and ruptured aneurysm. Newer endovascular techniques include flow diverter stents and liquid embolization materials (82). The requirement for subsequent treatment with dual antiplatelet therapy to prevent stent thrombosis should be considered in choosing treatment, especially in patients with an intraparenchymal hematoma or CSF diversion.

Craniotomy with open microsurgical clipping is generally preferred for distal artery aneurysms and middle cerebral artery bifurcation aneurysms. This procedure may be favored in cases that require concomitant clot evacuation or decompressive surgery. Risks associated with surgical clipping include new or worsening neurologic deficits caused by brain retraction, temporary artery occlusion, occlusion of parent vessel, and intraoperative hemorrhage.

The International Subarachnoid Aneurysm Trial (ISAT) randomized 2143 patients with spontaneous subarachnoid hemorrhage to clipping versus coiling within 28 days of subarachnoid hemorrhage onset (69). The majority of patients were World Federation of Neurological Surgeons (WFNS) grade 1 to 2; 97% had anterior circulation aneurysms, and most aneurysms were smaller than 10 mm. At 1 year, 24% of endovascularly treated patients had disability or death (modified Rankin Scale 3 to 6) compared with 31% of surgically treated patients (p = 0.0019). At 7-year follow-up, the mortality and seizures were higher in the surgical group (p = 0.03). The early rebleeding risk (up to 30 days after the initial procedure) was higher with endovascular repair.

Due to the concern of losing the beneficial effect of endovascular treatment over a longer period, the UK cohort of ISAT was followed for 18 years and showed that although there is a small increased risk of rebleeding in the endovascular group, there was no significantly worse outcome compared to the surgical group and the probability of survival free from disability was significantly higher in the endovascular group compared to the surgical group at 10 years (68).

The Barrow Ruptured Aneurysm Trial (BRAT) is an ongoing randomized controlled study comparing aneurysm clipping to coiling. The 6- and 10-year results showed no difference in outcome between the two interventions for anterior circulation aneurysms. As for posterior circulation aneurysms, results were similar to the ISAT study, which favored coiling, however, with higher retreatment rate (97).

Management and prevention of subarachnoid hemorrhage complications.

Increased intracranial pressure and cerebral hypoperfusion. Sudden increase in intracranial pressure after subarachnoid hemorrhage can lead to a decrease in the cerebral perfusion pressure, leading to global cerebral ischemia and death before reaching the emergency room. The mechanisms leading to the increase in intracranial pressure are hydrocephalus, cerebral edema, intracranial hematoma, ischemia, and impaired autoregulation. Intracranial pressure peaks at day 3 after the aneurysmal rupture and declines after day 7. Mean intracranial pressure correlates with mortality and severity of early brain injury (107) and the total “dose” of intracranial pressure, meaning the magnitude times the duration predicts the 12-month outcome better than just the time exceeding a high threshold of 20 mmHg (13).

Elevated intracranial pressure can be treated initially with sedation (ideally with a rapidly reversible agent like propofol), mannitol 20% 1 gms/kg intravenous bolus, or 23.4% saline intravenous push (30 cc over 10 to 20 minutes via a central or peripheral line). Hyperventilation to a PCO2 goal of 30 to 35 mmHg should only be used transiently until the medical and surgical measures show effect, after which PCO2 should ideally remain between 35 to 40 mmHg. General measures to prevent increased intracranial pressure include placing the patient's head at a centered position, elevating the head of bed 30 degrees, and avoiding tight bandages or extrinsic structures from compressing the neck. Finally, certain measures can be attempted to control the cerebral metabolic rate, including avoiding hyperglycemia and fever. The use of paralytics, barbiturates, or hypothermia to reduce intracranial pressure is typically reserved for refractory patients who do not have evidence of extensive irreversible intracranial injury. Decompressive craniotomy can be considered in high-grade subarachnoid patients. It is very effective in reducing intracranial pressure and may reduce mortality, particularly in patients with concomitant hematomas and herniation, but evidence that it improves functional outcome in survivors is lacking (02).

Hydrocephalus. Acute hydrocephalus may be noncommunicating or obstructive in nature, due to blocked CSF outflow through the aqueduct of Sylvius, fourth ventricular outlet, basal cisterns, or the subarachnoid space, resulting from physical obstruction due to clot or increased CSF viscosity. Twenty percent of the patients will develop acute hydrocephalus. Hydrocephalus is associated with poorer clinical grade, with increased blood on CT, and with intraventricular hemorrhage. It occurs in 15% of patients radiographically; 40% of these patients are symptomatic. Temporary CSF diversion is achieved by external ventricular drain and can lead to a significant improvement in the clinical grade. Typical indications for external ventricular drain include ventriculomegaly, depressed level of consciousness, or suspicion of elevated intracranial pressure. Most external ventricular drains are set to drain at 15 to 20 mmHg initially, with a goal intracranial pressure of lower than 20 mmHg. This setting is often lowered after the aneurysm is secured, when a sudden change in transmural pressure is less likely to cause aneurysm re-rupture. Patients who undergo surgical clipping of their aneurysm may also be treated with fenestration of the lamina terminalis, which might reduce the rate of shunt dependent hydrocephalus (104). Chronic ventriculoperitoneal shunting is needed in 9% to 48% of patients suffering aneurysmal subarachnoid hemorrhage (19).

Seizures. In subarachnoid hemorrhage, the incidence of nonconvulsive seizures is 7% to 18%, and that of nonconvulsive status epilepticus is 3% to 13%. Nonconvulsive status epileptics is associated with higher rates of mortality and morbidity (50). Seizures may precipitate parenchymal shift and increase the intracranial pressure in noncompliant brains. Periodic epileptiform discharges, nonconvulsive status epilepticus, nonreactive background, and the absence of normal sleep architecture are independent predictors of poor outcome in patients with subarachnoid hemorrhage. About 10% of patients experience clinical seizures during the hospital course. Middle cerebral artery aneurysm, thickness of the subarachnoid hemorrhage, presence of a hematoma, infarct or rebleed, poor grade, and past medical history of hypertension have been proposed to be risk factors for early seizure development. Endovascular treatment seems to have lower incidence of seizures (19).

Hyponatremia. Hyponatremia occurs in almost half of patients sustaining spontaneous subarachnoid hemorrhage after the ictus and can be due to several causes. According to a prospective study, the syndrome of inappropriate antidiuretic hormone secretion (SIADH) is the most common cause of hyponatremia (42). Other implicated factors are cerebral salt wasting, acute cortisol deficiency, incorrect IV fluid administration, and hypovolemia. It may be difficult to differentiate cerebral salt wasting from SIADH given that serum osmolality is low and urine sodium is elevated in both conditions. However, urine output is typically elevated in cerebral salt wasting, which can result in volume depletion; in contrast, SIADH is associated with normovolemia/ hypervolemia with normal or slightly low urine output. Volume status estimation is difficult in critical care patients. Inferior vena cava compressibility, presence of lung B-lines, stroke volume variations, and pulse pressure variations are some of the more reliable measures of intravascular volume (71; 36). Cerebral salt wasting, in particular, often heralds cerebral vasospasm and should alert the clinician to the possibility of this complication.

Sodium should be monitored closely, and volume carefully maintained, in patients with hyponatremia. Patients with cerebral salt wasting often require aggressive volume repletion to avoid precipitating delayed cerebral ischemia. If volume depletion with hyponatremia is present, careful repletion with isotonic to hypertonic saline (0.9%, 1.5%, 2%, or 3% saline) can be considered; additionally, fludrocortisone acetate (0.5 to 2 mg by mouth or intravenously twice daily) can be utilized (19). Rapid overcorrection may result in pontine or extrapontine myelinolysis, although this is rare in patients with hyponatremia for less than 24 hours. Increases in serum sodium exceeding 8 mEq per 24 hours should be particularly avoided in patients who are chronically hyponatremic. In general, it is best to limit sodium increases to no more than 10 mEq per day in all patients. In cases with more profound or symptomatic hyponatremia with volume depletion, the concentration of saline and the rate of infusion may need to change with time. In cases of SIADH in subarachnoid patients, fluid restriction of patients with subarachnoid hemorrhage should be strictly avoided because this may exacerbate the development of infarcts related to vasospasm. Conivaptan, an inhibitor of vasopressin V1a and V2 receptors, can be used to treat SIADH. Studies showing efficacy of prophylactic measures have reported conflicted results (94).

Cardiovascular and pulmonary problems. Patients suffering aneurysmal subarachnoid hemorrhage are affected by a number of cardiovascular complications, many of these due to preexisting cardiac pathology, as subarachnoid hemorrhage and cardiopulmonary disease share risk factors such as hypertension and coronary artery disease. Furthermore, aggressive management of the complications of subarachnoid hemorrhage may include the prolonged use of sedatives and vasopressors, which can create strain on the cardiovascular system, particularly in patients with preexisting pathology. Conversely, even patients who have a healthy cardiovascular system at the time of the hemorrhage are at higher risk of developing disorders affecting heart, blood vessels, and lungs as a result of the significant influence that injured brains have on the cardiovascular system. Patients with subarachnoid hemorrhage can have electrocardiogram abnormalities, such as ST and T wave changes, elevated troponin levels that can be associated with higher risk of vasospasm causing delayed cerebral ischemia, and arrhythmia (23).

The apical ballooning syndrome or Takotsubo cardiomyopathy, first described in Japan in the 1990s, is characterized by signs and symptoms of acute coronary syndrome potentially followed by congestive heart failure due to reversible apical and midventricular wall motion abnormalities with normal coronary arteries. When adequately supported, this syndrome has a good prognosis as most patients have normalization of left ventricular function at 3 months from discharge. When associated with subarachnoid hemorrhage, this neurocardiogenic stunning or neurogenic stress cardiomyopathy has a particular risk for fatal ventricular arrhythmias and worse cerebral vasospasm (36). Three main proposed mechanisms that are associated with the tremendous catecholamine surge that takes place at the time of bleeding include epicardial coronary spasm, coronary microvascular dysfunction, and direct myocardial injury due to catecholamines. Damage to the right insular cortex has been particularly associated with systemic increases of catecholamines.

Pulmonary edema affects patients suffering from subarachnoid hemorrhage as a result of either cardiac dysfunction due to artificially increased intravascular volume, artificially elevated blood pressure, or neurocardiogenic stunning. Pulmonary edema that is not associated with obvious cardiac abnormalities is called neurogenic pulmonary edema. The proximate etiology is thought to be similar to stress-induced cardiomyopathy in that it is suspected to result from the sympathetic surge at ictus that maintains cerebral perfusion pressure in the face of spiking intracranial pressure with aneurysm rupture. It has an incidence of 8% and particularly affects patients with increased intracranial pressure, posterior circulation aneurysmal rupture, and higher clinical grade on presentation. The mainstay of treatment includes very careful diuresis and positive pressure ventilation. It usually resolves within 72 hours in half of the patients (36).

Deep venous thrombosis. Deep venous thrombosis is common in subarachnoid hemorrhage, with an incidence between 1.5% and 24%. Pulmonary embolism is a serious complication frequently encountered. The Neurocritical Care Society recommends implementing intermittent pneumatic compression devices for prophylaxis immediately on admitting the subarachnoid patient to hospital and starting prophylaxis with unfractionated heparin as soon as 24 hours after securing the aneurysm (74).

Vasospasm and delayed cerebral ischemia. Up to 70% to 90% of patients with aneurysmal subarachnoid hemorrhage suffer from angiographic vasospasm between days 3 to 14, and 50% of these patients go on to develop delayed cerebral ischemia. In contrast, vasospasm and delayed cerebral ischemia are extremely rare in nonaneurysmal subarachnoid hemorrhage, and prophylaxis is not commonly practiced for these patients, with rare exceptions. Risk factors for vasospasm/delayed cerebral ischemia include poor clinical grade, thick blood on CT (subarachnoid hemorrhage and intraventricular hemorrhage), fever, hypertension, sentinel bleed, ultra-early angiographic spasm, volume depletion, low cardiac output, and smoking.

The mechanism of vasospasm is thought to be related to an imbalance of endogenous vasoconstrictors and vasodilators, including the vasoconstrictor endothelin-1, leading to increased intracellular calcium, causing smooth muscle contraction (79). Clazosentan, an endothelin A antagonist, has been studied extensively, with the hope of reducing cerebral vasospasm and delayed cerebral ischemia, with conflicting results. A study in Japanese patients showed that clazosentan decreased the incidence of vasospasm-related morbidity and all-cause mortality at 12 weeks in patients with aneurysmal subarachnoid (31).

Delayed cerebral ischemia is defined as the presence of new focal neurologic deficit or a decrease in 2 points in the Glasgow Coma Scale, lasting at least an hour, and not due to other causes. It affects 30% of patients and, after rebleeding, it is the most feared complication affecting patients suffering from spontaneous subarachnoid hemorrhage as it can lead to cerebral infarction (100). Although delayed cerebral ischemia was previously considered to be solely the consequence of vasospasm, there is an emerging body of literature showing that delayed cerebral ischemia is likely to have a multifactorial etiology beyond pure cerebral arterial constriction (88). Other postulated mechanisms contributing to delayed cerebral ischemia are micro-thrombi, cell death, dysautoregulation of distal vasculature, inflammation, and cortical spreading depression. Cerebrovascular reactivity is reduced in patients with clinically significant delayed cerebral ischemia and shows promise in predicting the clinical deterioration (08).

Diagnosing vasospasm and delayed cerebral ischemia. Transcranial Doppler is a very useful tool for daily noninvasive vasospasm screening in asymptomatic patients. Routine vasospasm screening of subarachnoid patients’ CT or MR or angiography is generally not indicated, but it may demonstrate focal or multifocal luminal narrowing in both proximal and distal vessels (03). Perfusion imaging may demonstrate symmetric or asymmetric areas of decreased or absent perfusion, commonly in the watershed areas. Cerebral angiography can confirm and potentially allow treatment of delayed cerebral ischemia. Abnormal perfusion studies accompanied with normal angiogram point to spasm of the smaller vasculature.

Transcranial Doppler ultrasound may demonstrate elevated velocities that can precede clinical symptoms by 24 to 48 hours. False-negative rates are as high as 30%, and the inability to insonate intracranial vessels occurs in 5% to 18% of patients depending on age, gender, and the thickness of the temporal bone on CT. For angiographic spasm, the positive predictive value of middle cerebral artery mean flow velocity greater than 200 cm/s is 87%, and the negative predictive value for middle cerebral artery mean flow velocity less than 120 cm/s is 94%. However, the positive predictive value in other territories, such as the anterior cerebral artery, is poor, although the negative predictive value (for mean flow velocity less than 120 cm/s) for symptomatic spasm remains excellent. The Lindegaard ratio corrects the mean flow velocity for hyperemia (due to increased cardiac output, pressor use, or anemia) and is defined as the ratio of velocity of the middle cerebral artery to that of the internal carotid artery velocity.

Table 8. Lindegaard Ratio

Transcranial Doppler is typically performed on a daily basis, and the trend serves to indicate which patients may require closer observation for impending symptomatic vasospasm/delayed cerebral ischemia.

Continuous electroencephalography (cEEG) has great potential as a monitoring modality for detecting delayed cerebral ischemia, as it is continuous and noninvasive. In a systematic review, delayed cerebral ischemia was detected in 20% to 46% of patients on continuous EEG monitoring. It also showed that quantitative EEG analysis detecting decreased alpha/delta ratio, relative alpha variability, and total power may predict delayed cerebral ischemia (50). Quantitative EEG can detect vasospasm/delayed cerebral ischemia 2.3 days ahead of transcranial Doppler, neuroimaging, or clinical deterioration. In addition, up to 8.6% of patients with subarachnoid hemorrhage have subclinical seizures detected on cEEG monitoring (101).

Treatment of vasospasm is generally not based exclusively on transcranial Doppler or EEG readings, but findings of concern often prompt conservative preventative strategies aimed at avoiding the precipitation of delayed cerebral ischemia. These include close monitoring of the neurologic exam, bedrest, and strict monitoring of fluid status with aggressive maintenance of normovolemia. If symptoms of delayed cerebral ischemia develop, prompt and aggressive treatment (outlined below) is indicated.

Prophylaxis of vasospasm and delayed cerebral ischemia. Nimodipine is the only medication proven in large trials to improve outcome after subarachnoid hemorrhage. In a series of studies, nimodipine has been shown to reduce death or severe disability from delayed cerebral ischemia. Initially it aimed to target symptomatic vasospasm, but its mechanism seems to be independent of prevention of large vessel narrowing (19). Nimodipine is given for 21 days and typically dosed as 60 mg every 4 hours. Hypotension associated with nimodipine can be mitigated by doubling the dosing interval and halving the dose. Particularly when radiographic spasm is present, doses of nimodipine may be held altogether.

Intravenous nicardipine has been used for blood pressure control and has been shown to be associated with lower incidence of angiographic and clinical vasospasm, without affecting overall outcome (06).

“Triple-H” therapy, or medically induced hypertension, hypervolemia, and hemodilution, has historically been used both for prophylaxis against vasospasm and treatment of vasospasm. Systematic review of published studies of this approach has found the evidence insufficient to determine the efficacy of induced hypertension, hypervolemia, and hemodilution for either prevention or treatment of vasospasm following subarachnoid hemorrhage (59). Guidelines do not recommend prophylactic hypervolemia, noting evidence for harm from aggressive volume expansion, and suggesting instead that maintenance of euvolemia is preferred for prophylaxis of vasospasm and delayed cerebral ischemia (27; 19).

The evidence for use of high-dose statins to prevent vasospasm or delayed cerebral ischemia is conflicting. Two meta-analyses have been published. One analyzed 10 randomized controlled studies and showed a significant beneficial effect of statin therapy on decreasing vasospasm and delayed ischemic neurologic deficit, with no significant effect on mortality and functional outcome (01). The second study collected six randomized controlled trials and two prospective cohorts and showed that statins only decreased the incidence of vasospasm, without a significant effect on delayed deficits, infarction, or mortality (96). The neurocritical society recommends continuing statin medications in patients already taking them and weakly recommends adding them for the prevention of delayed cerebral ischemia in statin-naïve patients (Diringer et 2011), whereas guidelines from the Chinese Stroke Association recommend against routine use of statins in aneurysmal subarachnoid hemorrhage (29).

Hypomagnesemia has been associated with delayed cerebral ischemia and poor outcome. Studies have shown conflicting results; however, most studies did not show benefit in prophylactically administering magnesium to prevent delayed cerebral ischemia (06).

The selective phosphodiesterase 3 inhibitor cilostazol has shown promise for prevention of delayed cerebral ischemia in subarachnoid hemorrhage. A meta-analysis included three randomized controlled studies and four observational studies comparing cilostazol to placebo, showing significant reduction in the incidence of angiographic vasospasm, delayed cerebral ischemia, strokes, and poor outcome, with no difference in mortality (95). At a dose of 100 mg orally twice a day, there were no major side effects. However, larger studies are still needed.

After the aneurysm is secured, and especially in patients with high risk of asymptomatic vasospasm, permissive hypertension with avoidance of unnecessary treatment of hypertension should be the mainstay of therapy. Prophylactic hypervolemia should not be induced, as a positive fluid balance is independently associated with poor neurologic outcome (49). Management should target a euvolemic state.

Treatment of symptomatic vasospasm. The clinical management of patients with symptomatic vasospasm follows therapeutic paradigms building on those used to prevent vasospasm. Efforts to maximize cerebral perfusion should be employed immediately upon symptom recognition and should escalate until the deficit is resolved, or until the risks of induced hypertension outweigh the benefits. First, if no contraindications are present, the head of the bed should be laid flat and a fluid bolus should be administered. Second, if the patient has an external ventricular drain, a decrease in the setting to a lower level should be considered to further increase the cerebral perfusion pressure by decreasing the intracranial pressure. Finally, vasopressor agents such as phenylephrine or norepinephrine are subsequently administered to target the lowest blood pressure that produces resolution. However, the use of vasoactive medications can be limited by the development of end-organ damage (myocardial infarction, congestive heart failure, renal insufficiency, digital ischemia, etc.). Support for the use of induced hypertension is based on individual experience and case series. A randomized controlled study showed increased rates of serious adverse effects, with no added benefit, using induced hypertension for treatment of delayed cerebral ischemia in aneurysmal subarachnoid hemorrhage (37). However, this study was terminated early and was underpowered. A systematic literature review found insufficient evidence to determine the efficacy of medically induced hypertension for the treatment of vasospasm (59). Nevertheless, the practice is commonly followed. Impressive improvements in the neurologic exam can be witnessed while increasing the blood pressure. The Neurocritical Care Society guidelines and those of the American Heart Association both recommend inducing hypertension in appropriate patients with clinical suspicion of delayed cerebral ischemia (27; 19).

Failure for symptoms to resolve with medical management necessitates prompt referral for endovascular therapy to further improve cerebral blood flow by increasing the caliber of the vessel lumen. Endovascular treatment can reverse symptoms of delayed cerebral ischemia in 30% to 70% of patients. Options include intraarterial vasodilators such as papaverine, verapamil, or nicardipine, or balloon angioplasty. Vasodilators are short-lived, and angioplasty has a more durable effect but also carries a 1% risk of vessel rupture. Endovascular treatment is most successful if performed early, preferably within 2 hours of symptom onset.

Steroid use in subarachnoid hemorrhage is controversial. It has no effect on delayed cerebral ischemia and has side effects. However, a study showed that it might have a beneficial effect in the subgroup of patients that had their aneurysm clipped (21).

Blood transfusion improves cerebral oxygen delivery globally and to vulnerable areas at risk for delayed cerebral ischemia (22). Restricting transfusion to cases with hemoglobin levels below 7 g/dL, as in general medical patients, may not be appropriate in subarachnoid hemorrhage patients.

Novel techniques currently under investigation to decrease the risk of delayed cerebral ischemia include intrathecal thrombolysis CSF drainage through a lumbar drain, the use of phosphodiesterase inhibitors, and intraarterial administration of fasudil hydrochloride or nicardipine. Intrathecal administration of nicardipine or nimodipine has also been examined. A phase 1/2a study showed that intraventricular sustained release nimodipine is safe and may improve outcome; however, it was associated with higher rates of culture-positive ventriculitis (41). Systematic literature review of descriptions of intraventricular nicardipine found two nonrandomized prospective studies as well as a number of retrospective case series (40). The most widely used dosage was 4 mg of nicardipine twice a day, for up to 11 days. It reduced the incidence of angiographic vasospasm; however, it remains unknown whether it improved functional outcome, and it carried a 6% risk of ventriculostomy-associated infection. A meta-analysis of all studies evaluating the efficacy of intrathecal, intraventricular, and intracisternal nicardipine in subarachnoid hemorrhage showed that it can improve vasospasm, mainly in the proximity of the drug release, especially if vasospasm is refractory to conventional therapy; however, larger randomized controlled studies are needed to better evaluate its effectiveness (28).

The phosphodiesterase-3 inhibitor milrinone, given intravenously, has shown promise in treatment of delayed cerebral ischemia. Milrinone produces inotropic effects in the heart and vasodilatory effects peripherally and in cerebral arteries. A systematic review of 10 published trials found a signal for efficacy in improved resolution of delayed cerebral ischemia, with side effects of hypotension and hypokalemia, in the limited evidence available (14). Lakhal and colleagues showed that despite its poor tolerance, treatment of vasospasm with intravenous milrinone was associated with better neurologic and radiological outcomes compared to the standard treatment of induced hypertension (55).

Therapeutic hypothermia, although not typically used in subarachnoid hemorrhage patients, was associated with decreased risk of vasospasm and delayed cerebral ischemia in an observational study (54).

Multimodal monitoring. Although the clinical examination remains the gold standard for the assessment of patients with any neurologic illness, it is time-consuming and requires a high level of training and expertise. Additionally, examination may be confounded in poor grade patients with a need for ventilation, sedation, or paralysis, or with concomitant encephalopathy, masking important changes. Multiple techniques for monitoring of patients with subarachnoid hemorrhage are gaining popularity by allowing physicians and staff to collect objective data that are easy to compare and can potentially quantify effect of interventions and assess damage. Techniques being currently utilized to monitor such patients include brain tissue oxygen tension monitors, continuous and quantitative EEG, cerebral microdialysis, regional cerebral blood flow monitors, and jugular bulb oximetry. These devices have significant pitfalls, including the fact that they are invasive and provide data that are limited to a very small portion of parenchyma. Only standardized collection of data will allow the neurocritical care community to prove a positive impact in outcome from patients subjected to monitoring.

Outcomes

The devastating natural history of this disease is reflected in the fact that 12% to 15% of patients with subarachnoid hemorrhage die before reaching a hospital. The most important predictive factors for prognosis after subarachnoid hemorrhage include level of consciousness, neurologic grade on admission, patient age, and amount of blood on initial head computed tomography scan (19). The presence of rebleeding, hypoxemia, metabolic acidosis, MAP less than 70 and more than 130, renal failure, fever, and anemia are also major variables that affect the prognosis of the patient (18). In the United States, the median mortality rate of subarachnoid hemorrhage is 32% (excluding prehospital death). Data reflect a trend of decreasing mortality: among all patients hospitalized in Canada from 2004 to 2015, in-hospital mortality was 21.5% (15). In addition to death, persistent functional deficits are common following subarachnoid hemorrhage. In the International Subarachnoid Aneurysm Trial (ISAT), 12% of patients had significant disability and 6.5% were functionally dependent 1 year post aneurysmal subarachnoid hemorrhage. Many patients are left with cognitive deficits after subarachnoid hemorrhage (19).

There are several causes of poor outcome following subarachnoid hemorrhage. These include, in order of decreasing importance: (1) deleterious effects on the brain of the initial bleed (typically measured using the Hunt-Hess grade or World Federation of Neurological Surgeons grade); (2) aneurysm re-rupture; (3) cerebral vasospasm and delayed cerebral ischemia; (4) hydrocephalus; (5) hyponatremia; and (6) seizures. Recurrent hemorrhage carries a case fatality rate approaching 70% and is currently the most treatable cause of poor outcome (05).

Despite the very aggressive nature of this condition, a 2009 meta-analysis found a decrease in case fatality from 1973 to 2002, notwithstanding an increase in the mean age of patient presentation, and coinciding with the introduction of improved management strategies (73). A survey reported that more than 75% of survivors of subarachnoid hemorrhage have a favorable outcome, and nearly two-thirds return to work (93).

Following the initial event, the risk of subarachnoid hemorrhage recurrence has been estimated to be 15 to 22 times higher than the expected rate of a first subarachnoid hemorrhage in a healthy age- and sex-matched cohort (85). Reported cumulative 8- to 10-year incidences of late rebleeding (more than 1 year after initial subarachnoid hemorrhage) vary from 0.1% to 3.2%. Endovascularly treated patients appear to be at somewhat higher risk of rebleeding from the original aneurysm than surgically treated patients, but the overall risk is low (68); the aneurysms recur in 9% to 34% of endovascularly-treated cases (16).

Compared to those with aneurysmal subarachnoid hemorrhage, patients with nonaneurysmal spontaneous subarachnoid hemorrhage have much better prognosis and low mortality. Most patients with perimesencephalic subarachnoid hemorrhage return to their baseline level of functioning. Perimesencephalic subarachnoid hemorrhage has excellent outcome with extremely low risk of recurrence (63).

Special considerations

Pregnancy

Aneurysmal subarachnoid hemorrhage occurs in 1 in 5000 to 10,000 pregnancies, and up to 20% of all aneurysmal ruptures occur in pregnancy or in the early postpartum period (103). The clinical presentation of aneurysmal subarachnoid hemorrhage is no different in the pregnant patient. The risk of aneurysmal rupture increases with increasing maternal age and trimester. Diagnosis is made using the same radiological tests, with appropriate shielding of the uterus. MRA, because it does not entail radiation, may be a reasonable diagnostic alternative for pregnant patients (05; 62). As the natural history of a ruptured aneurysm in a pregnant patient is similar to that of a nonpregnant patient, treatment of the pregnant patient with an aneurysmal subarachnoid hemorrhage should not differ from the nonpregnant patient. If the aneurysm ruptures at or near term, an emergency Cesarean section can be performed. If the fetus is not yet capable of surviving without a respirator, the aneurysm is surgically treated, and the pregnancy is then allowed to take its course. Pharmacological adjuncts to the surgical treatment of aneurysmal subarachnoid hemorrhage should be used with caution (103).

Anesthesia

Preoperative anesthetic evaluation should include a brief survey of concurrent medical problems, an assessment of the severity of the subarachnoid hemorrhage, and identification of any associated cardiac abnormalities (particularly the possibility of myocardial injury). Awareness of the presence and degree of cerebral vasospasm, hydrocephalus, and intracranial mass effect is also necessary. The neurointensivist, neuroanesthesiologist, and neurosurgeon should review the appropriate radiographic studies and discuss the need for neurophysiologic monitoring, cerebral protective therapy, or rapid infusion devices.

Coexisting medical conditions also need to be evaluated. This includes an assessment of the patient's respiratory status (aspiration pneumonia, neurogenic pulmonary edema), electrolyte and metabolic abnormalities (hyponatremia, hypoglycemia, hypoproteinemia), and most importantly, the wide variety of cardiac abnormalities that are frequently encountered.

Intraoperative management should include prevention of arterial hypertension (especially during induction of anesthesia), prevention of intracranial pressure spikes during induction (if elevated intracranial pressure is of concern). Induction is usually performed with propofol, thiopentone sodium, or etomidate with short acting opiates. Muscle relaxants are used; however, succinylcholine can increase intracranial pressure. Nondepolarizing agents are preferred for intubation (53). The American Heart Association recommends minimizing the degree and duration of intraoperative hypotension (05; 19). Intraoperative-induced moderate hypothermia has not been shown to improve outcomes in patients with good grade subarachnoid hemorrhage (99), but it may be considered in selected cases (19). Intraoperative hyperglycemia should be avoided as it is associated with long-term cognitive decline and gross neurologic dysfunction (77).