Risk Factors

Public Health Burden of Stroke

The distribution of the burden of stroke morbidity and mortality is heterogeneous in the US population and is changing dramatically with time. Stroke mortality remains the third leading cause of death in the United States, accounting for 1 in every 15 deaths during 1992.¹ Despite this burden, US stroke mortality rates are among the lowest in the world.² The estimated US stroke mortality rate for women was 36.7 per 100 000; for men it was 46.6 per 100 000. There has been a striking 60% decline in US stroke mortality between 1960 and 1990.¹ Despite this decline, nearly 150 000 Americans died of a stroke during 1995, which corresponds to 1 death every 3.5 minutes.¹

The burden of stroke is heterogeneous and is greater among the elderly, men, and African-Americans. In the southeastern United States, stroke risk is approximately 1.4 times that of other regions.¹²³⁴

Unlike stroke mortality estimates derived from vital statistics data, incidence estimates have been made indirectly or by extending estimates in small communities to the entire nation. Only a few communities in the United States have systematically collected incidence data.⁵⁶ In Olmsted County (Rochester, Minn), stroke incidence rates declined from 205 per 100 000 in the period 1955 to 1959 to 128 per 100 000 from 1975 to 1979.⁵ However, from 1980 to 1984, incidence increased to 153 per 100 000 and has remained relatively constant (145 per 100 000) from 1985 to 1989. That stroke incidence has not substantially declined since the mid 1980s is also supported by data from Framingham⁶ and Minneapolis.⁷ Importantly, the most reliable estimates of stroke incidence are provided in predominantly white communities with a high access to health care. As such, stroke incidence data on groups at high risk of stroke mortality (African-Americans, residents of the southeastern United States) are lacking.

While stroke incidence rates have been level since the mid 1980s, the decline in stroke mortality has continued at least through 1992.² This decline in stroke mortality in the face of a likely stable incidence rate suggests a declining case fatality among stroke victims. This may beso, as 1-year survival after stroke improved from 49% to 62% in five North Carolina counties between 1970 and 1973 and 1979 and 1980.⁸ This trend of improving case fatality was also noted between 1980 and 1990 in the Minneapolis area, where 2-year survival after stroke improved from approximately 62% to 73% in men and from approximately 57% to 73% in women.⁷ However, mortality after stroke remains substantial, with approximately 25% dying in the year following stroke.⁶⁷

Besides mortality, morbidity in the more than 3 000 000 surviving stroke victims (prevalent cases) is also substantial, making stroke the leading cause of serious disability in the United States.¹ Among long-term (>6 months) stroke survivors, 48% have hemiparesis, 22% cannot walk, 24% to 53% report complete or partial dependence on activity of daily living (ADL) scales, 12% to 18% are aphasic, and 32% are clinically depressed.⁹ The average healthcare costs (inpatient and outpatient) for cerebral infarction have been estimated to be between $8000 and $16 500; for subarachnoid hemorrhage, between $27 000 and $32 911; and for intracerebral hemorrhage, between $11 100 and $12 881.¹⁰ Although these numbers are impressive, they do not include the additional costs associated with the residual morbidity after stroke (lost work, additional nursing care, etc.).

While stroke incidence appears stable and stroke mortality is slowly declining, the absolute magnitude of stroke is likely to grow over the next 30 years. In 1995 12.8% of the US population was older than 65. By 2025 that percentage is expected to increase to 18.7%.² Similarly in 1995 12.6% of the US population was African-American; by 2025 the African-American population in the United States is expected to increase to 14.5%. With the aging of the population and an increased proportion of African-Americans, the absolute number of stroke victims (and demands on healthcare and other support systems) is likely to increase substantially in the future.

Risk Factors for Ischemic Stroke

Nonmodifiable Risk Factors or Risk Markers

Age, gender, race, ethnicity, and heredity have been identified as markers of risk for stroke. Although these factors cannot be modified, their presence helps identify those at greatest risk, enabling vigorous treatment of those risk factors that can be modifed.

Age is the single most important risk factor for stroke. For each successive 10 years after age 55, the stroke rate more than doubles in both men and women.⁵⁶ Stroke incidence rates are 1.25 times greater in men, but because women tend to live longer than men, more women than men die of stroke each year.

An increased incidence of stroke in families has long been noted. Potential reasons are a genetic tendency for stroke, a genetic determination of other stroke risk factors, and a common familial exposure to environmental or lifestyle risks. Earlier studies suggested an increased risk for men whose mothers died of stroke and women who had a family history of stroke.¹¹ In the Framingham Study an offspring analysis revealed that both paternal and maternal histories were associated with an increased risk of stroke.¹²

Stroke incidence and mortality rates vary widely between racial groups. Blacks are more than twice as likely to die of stroke as whites are.¹³ Between the ages of 45 and 55, mortality rates are four to five times greater for African-Americans than for whites; the difference decreases with increasing age.¹⁴ However, some race-related risk for stroke may be related to environmental factors or inherited risk factors other than race. In the National Health and Nutrition Examination Survey, the rate ratio of mortality for blacks versus whites decreased from 2.3 to 1.9 when adjusted for six well-established risk factors and decreased from 1.9 to 1.4 when further adjusted for family income.¹⁵ Thus, 38% of excess stroke mortality in blacks could be explained by the six risk factors and family income. Epidemiological studies of Hispanics in the United States are handicapped by the diversity of origin and heterogeneity of the groups; however, stroke death rates were similar in Hispanics and whites younger than 65 and lower in those older than 65. This may be changing. In New Mexico, between 1958 and 1987, Hispanics had lower cerebrovascular disease mortality rates than whites, but during the most recent 5-year period, rates were higher.¹³ In a hospital and community-based cohort study of all cases of first stroke in northern Manhattan, blacks and Hispanics had an overall age-adjusted 1-year stroke incidence rate 2.4 times and 1.6 times, respectively, that of whites.¹⁶

Stroke was a leading cause of death among Native Americans in 1990, but death rates were lower than in whites.¹⁷ From 1988 through 1990, stroke death rates were similar in Native Americans and whites younger than 65, and, like Hispanics, lower than in whites older than 65.

Asians, specifically Chinese and Japanese, have high stroke incidence rates.¹⁸ Stroke incidence and mortality rates in Japan were very high for most of this century and exceeded those for heart disease. As in the United States, stroke death rates in Japan have fallen dramatically since World War II. In recent years stroke incidence rates in Japanese men in Hawaii were similar to those of white Americans and between the rates of Japanese men in Japan and in California.¹⁹

Potentially Modifiable Risk Factors for Ischemic Stroke

Hypertension

Hypertension is the single most important modifiable risk factor for ischemic stroke. Most estimates for hypertension indicate a relative risk of stroke of approximately 4 when hypertension is defined as systolic blood pressure ≥160 mm Hg and/or diastolic blood pressure ≥95 mm Hg. A summary of seven studies assigning a relative risk of 1 for borderline or mild hypertension determined the relative risk to be about 0.5 at a blood pressure of 136/84 mm Hg and about 0.35 at a blood pressure of 123/76 mm Hg.²⁰ From the lowest to the highest level of blood pressure in this summary, risk is increased about 10-fold. Although clearly important even in the elderly, the impact of hypertension may decrease with age: the odds ratio is 4 at age 50, decreasing to 1 by age 90.²¹ From population surveys the prevalence of hypertension is about 20% at age 50, about 30% at age 60, 40% at age 70, 55% at age 80, and 60% at age 90.²² When the Joint National Committee V definition is used (≥140/90 mm Hg or on antihypertensive medication), prevalence increases to about 45% at age 50, >60% at age 60, and >70% at age 70.²² The prevalence of hypertension is greater in blacks than in whites.

The efficacy of antihypertensive treatment has been well established in clinical trials. In a summary of 17 treatment trials of hypertension throughout the world involving nearly 50 000 patients, there was a 38% reduction in all stroke and a 40% reduction in fatal stroke favoring systematic treatment of hypertension.²⁰ This effect was true in whites and blacks and at all ages. Treatment was also highly effective in preventing stroke in elderly persons with isolated systolic hypertension (Systolic Hypertension in the Elderly Program [SHEP]), the most prevalent form of hypertension in persons older than 65. Importantly, there was no less impact on stroke prevention above age 80, with incidence reduced by 40%.²³

Cardiac Disease

Various cardiac diseases have been shown to increase risk of stroke (Table 1). Atrial fibrillation (AF) is the most powerful and treatable cardiac precursor of stroke. The incidence and prevalence of AF increase with age. With each successive decade of life above age 55, incidence of AF doubles.²⁴ Using data from four population-based studies and the US census, it has been estimated that 2.2 million Americans have intermittent or sustained AF.²⁵ Prevalence above age 65 is estimated to be 5.9%. Data from the Framingham Study and hospital discharges suggest that the prevalence of AF in the US population is increasing.²⁶ The aging of the US population, the increasing incidence of AF with age, and the increasing prevalence of AF suggest that AF will result in increasing rates of morbidity and mortality in the population.

It is estimated that almost half of all cardioembolic strokes occur in the setting of AF. In the Framingham Study, nonvalvular AF was independently associated with a threefold to fivefold increased risk for stroke. The impact of hypertension, coronary heart disease, and cardiac failure on risk of stroke declined with advancing age, while the impact of AF persisted even into the ninth decade of life.²⁷ The attributable risk of AF for stroke rose from 1.5% in subjects aged 50 to 59 years to 23.5% in subjects aged 80 to 89 years; ie, nearly one stroke in four in persons older than 80 was a result of AF.

Pooling data from five randomized controlled trials of antithrombotic therapy in AF identified increasing age, history of hypertension, previous transient ischemic attack or stroke, and diabetes as risk factors for stroke.²⁸ On the other hand, investigators noted that patients younger than 65 who possessed none of these factors had a low annual stroke rate of 1%. Data from three of the trials and epidemiological studies suggest that left atrial enlargement, mitral annular calcification, and perhaps decreased left ventricular systolic function were associated with an excess of stroke during follow-up. Spontaneous echocardiographic contrast and left atrial thrombus have also been identified as transesophageal echocardiographic predictors of stroke with AF.

Warfarin anticoagulation reduced the risk of stroke by 68% in a pooled analysis of AF trials. The annual rate of stroke was 4.5% in the control group and 1.4% in the warfarin group for an absolute annual reduction of 3.1% (P<.001). The annual rate of major bleeding was low: 1% for patients on placebo or aspirin and 1.3% for those on warfarin. The effectiveness of aspirin for stroke prevention in AF is uncertain and based largely on results of the Stroke Prevention in Atrial Fibrillation (SPAF) trials. In SPAF warfarin was significantly more effective than 325 mg aspirin daily. Aspirin seemed to reduce noncardioembolic stroke but did not prevent more severe strokes classified as cardioembolic.²⁹³⁰ Hence, the current recommendation to prevent stroke in AF is to give warfarin to patients who are candidates for anticoagulation and reserve aspirin for young subjects at low risk of stroke or with contraindications for warfarin.³¹

Despite the convincing evidence supporting the efficacy of warfarin, in 1992 only 26% of outpatients with AF were treated with warfarin, and warfarin use was lowest in the elderly patients in whom it might have the greatest value.³²³³ The Agency for Health Care Policy and Research states that warfarin is widely underused, with less than half of the eligible AF patients receiving warfarin.⁹

Cardiac valve abnormalities, in particular mitral stenosis, are important risk factors for stroke discussed in “Etiology of Stroke.” The risk of stroke in the setting of mitral valve prolapse may have been overstated, based on early retrospective case-control studies. The prevalence of mitral valve prolapse has been reported as 4% to 5%; however, the prevalence in community cohorts using modern diagnostic criteria is unknown. Prospective studies with more stringent diagnostic criteria for mitral valve prolapse suggest that the risk of stroke is low in subjects with prolapse uncomplicated by endocarditis or AF.

Another valvular risk factor for stroke is mitral annular calcification. Prevalence on M-mode echocardiography has ranged from 10% in men to 16% in women. In the Framingham Study mitral annular calcification was associated with a doubled rate of stroke in follow-up (RR 2.1, P=.006) after adjusting for traditional risk factors for stroke.³⁴ As with mitral stenosis, the presence of AF and mitral annular calcification resulted in an amplification of risk for stroke. With both AF and annular calcification, stroke risk was increased fivefold, compared with a doubling in stroke risk with either factor present alone.

The most recent finding associated with stroke is valvular strands. These fine, filamentous, threadlike mobile processes have been detected by transesophageal echocardiography (TEE), attached to the mitral and aortic valves. Two preliminary studies suggest that these valvular strands are a risk factor for ischemic stroke, but further prospective data are needed.

Left atrial enlargement was found to be a risk factor for stroke. In the Framingham Study, for every 10-mm increment in left atrial size, the age-adjusted risk of stroke was approximately doubled in both men and women; after multivariate adjustment, the excess risk of stroke persisted in men (relative risk 2.4).³⁵

Epidemiological evidence is accumulating that the cardiac structural abnormalities of patent foramen ovale (PFO) and atrial septal aneurysm (ASA) increase risk for embolic stroke. The PFO provides a right-to-left interatrial shunt leading to paradoxical embolism. PFO is now noninvasively detected by TEE or transthoracic echocardiography (TTE) with agitated saline contrast injections. Two case-control studies in which contrast TTE was used in young patients with ischemic stroke found a significant relation between PFO and stroke.³⁶³⁷ Several case-control and cross-sectional studies have demonstrated this association among older cases with stroke,³⁸ while others have not. ASA is a congenital malformation characterized by a bulging of the septum into either atrium.³⁹ An increased frequency of ASA was found among patients with unexplained stroke compared with control subjects.⁴⁰ A strong association between ASA and PFO was found, with evidence of a synergistic effect for cryptogenic stroke when both were present.

Myocardial disease has long been recognized as a risk factor for stroke. In the Framingham Study, when multivariate analysis was used, risk of stroke was increased twofold by coronary heart disease, threefold by electrocardiographic left ventricular hypertrophy, and threefold to fourfold by cardiac failure.²⁷ In a separate analysis at Framingham, left ventricular mass assessed by echocardiography was also predictive of stroke in follow-up.⁴¹

While it is apparent that prevention of coronary heart disease and left ventricular hypertrophy form a cornerstone of cardioembolic stroke prevention, an effective means of preventing stroke once myocardial disease is present remains less clear. (See “Etiology of Stroke.”)

The increasing complexity and prevalence of interventional cardiology treatments and procedures has resulted in cardiovascular complications, including stroke. The risk of stroke after cardiac catheterization and angioplasty is 0.2% to 0.3%. The perioperative cardiac surgery stroke rate is approximately 1% and is multifactorial in origin. Intracardiac devices may be complicated by thrombus or infection with resulting embolism. Electrophysiology procedures and devices, including radiofrequency ablation, pacing, and, more commonly, cardioversion have also been noted to lead to embolic complications.

Diabetes and Glucose Metabolism

Persons with diabetes have an increased susceptibility to atherosclerosis and an increased prevalence of atherogenic risk factors, notably hypertension, obesity, and abnormal blood lipids. Case-control studies of stroke patients and prospective epidemiological studies have confirmed an independent effect of diabetes with a relative risk of ischemic stroke in persons with diabetes from 1.8 to 3.0. Among Hawaiian Japanese men in the Honolulu Heart Program, those with diabetes had twice the risk of thromboembolic stroke of persons without diabetes that was independent of other risk factors.⁴² In a population-based cohort in Rancho Bernardo, persons with diabetes had a risk-factor adjusted relative risk of stroke of 1.8 in men and 2.2 in women. In Framingham, persons with glucose intolerance have double the risk of brain infarction of nondiabetic persons.

In addition to the role of glucose status (normal, impaired glucose tolerance, or diabetic), there are other aspects of glucose metabolism that may play a role as a risk factor for ischemic stroke—specifically hyperinsulinemia and increased insulin resistance (the relative inability of insulin to enhance glucose disposal). Both were shown to be risk factors for ischemic stroke among subjects with normal glucose status.⁴³ In non-Hispanic white and Hispanic subjects, increased insulin resistance is associated with increased atherosclerosis of the carotid arteries independent of glucose status, insulin levels, and other major cardiovascular risk factors.⁴⁴

Lipids

While hypercholesterolemia is an important modifiable risk factor for coronary heart disease, the link to ischemic stroke remains uncertain.⁴⁵⁴⁶ However, data clearly support the positive relation between total and LDL cholesterol and a protective influence of HDL cholesterol on extracranial carotid atherosclerosis.⁴⁷ In secondary analyses, the Scandinavian Simvastatin Survival Study (4S) found a reduction of fatal or nonfatal stroke with simvastatin versus placebo (RR=.70, 95% confidence interval .52, .96), and the Asymptomatic Carotid Artery Plaque Study (ACAPS) reported fewer strokes in the lovastatin versus placebo group (5 versus 0).⁴⁸ A pooled analysis of four pravastatin trials disclosed a 46% reduction in risk of stroke (P=.054).⁴⁹

Cigarette Smoking

Cigarette smoking increases risk (RR) of ischemic stroke nearly two times,⁵⁰ with a clear dose-response relation. In both the Framingham Study and the Nurses’ Health Study⁵¹⁵² cessation of smoking led to a prompt reduction in stroke risk—major risk was reduced within 2 to 4 years. This reduction in risk occurred throughout the age spans of these studies and in heavy as well as moderate smokers.

Alcohol

Moderate consumption of alcohol may reduce cardiovascular disease, including stroke. Recent epidemiological studies have shown a U-shaped curve for alcohol consumption and coronary heart disease mortality, with low to moderate alcohol consumption associated with lower overall mortality. In an overview analysis of stroke studies, a J-shaped association curve was suggested for the relation of moderate customary alcohol consumption and ischemic stroke.⁵³ This was most consistent for white populations; however, little if any association existed for Japanese and possibly black populations. Increasing alcohol consumption increases risk for brain hemorrhage.⁵⁴

Illicit Drug Use

Drug abuse is a major social problem, with cocaine the substance most commonly associated with stroke.⁵⁵ Other drugs linked to stroke include heroin, amphetamines, LSD, PCP, “T’s and Blues,” and marijuana. Case reports have also linked over-the-counter sympathomimetic decongestants, cold remedies, and diet aids (eg, phenylpropanolamine), ephedrine, and pseudoephedrine with hemorrhagic and, less often, ischemic stroke. The bulk of information about stroke and drug use and abuse is derived from case reports or case series, with many reports confounded by multiple drugs used. There are sparse epidemiological data relating drug use to stroke.

Lifestyle Factors (Obesity, Physical Activity, Diet, and Acute Triggers)

Various lifestyle factors have been associated with increased stroke risk. These include obesity, physical inactivity, diet, and acute triggers such as emotional stress. Obesity has been associated with higher levels of blood pressure, blood glucose, and atherogenic serum lipids, which are independent risk factors for stroke. In Framingham, obesity defined as a Metropolitan Life chart relative weight greater than 30% above average was a significant independent contributor to incidence of brain infarction in men aged 35 to 64 and women aged 65 to 94. In the Honolulu Heart Study, obesity was identified as an independent factor related to stroke incidence. The pattern of obesity may be important; central obesity manifested by abdominal deposition of fat, rather than obesity involving the hips and thighs, has been related to the occurrence of atherosclerotic disease.

Moderate and heavy levels of physical activity have been associated with reduced CHD incidence. In recent years evidence supports a protective effect of moderate physical activity on stroke incidence in men and women.⁵⁶⁵⁷ In Framingham, physical activity was protective in men; adjusted relative risk was 0.41. However, there was no evidence of a protective effect of physical activity on risk of stroke in women. In addition, as has been found in coronary heart disease, there was no evidence that heavy physical activity conferred greater benefit than moderate levels. Physical activity exerts a beneficial influence on risk factors for atherosclerotic disease by reducing blood pressure, weight, and pulse rate; raising HDL cholesterol and lowering LDL cholesterol; decreasing platelet aggregability; increasing insulin sensitivity and improving glucose tolerance; and promoting a lifestyle conducive to changing diet and promoting cessation of cigarette smoking. Studies regarding the association of stroke and diet have been inconclusive. Increased consumption of fish, green tea, and milk were protective of stroke, while diets high in fat and cholesterol could be deleterious.⁵⁸

Oral Contraceptives

Oral contraceptives with an estrogen content >50 μg, the preparations used in the 1960s and 1970s, were strongly associated with risk for stroke. Recently a study of low-dose oral contraceptives (<50 μg estrogen) disclosed no increased risk of stroke in more than 3.6 million woman-years of observation.⁵⁹

Migraine

While migraine has been identified as an independent risk factor for ischemic stroke in men older than 40 in the Physicians’ Health Study, no association was found in other studies after adjusting for other stroke risk factors.⁶⁰ Although there may be an association between migraine and stroke, this association must be put in the context of the absolute risk of stroke. It has been estimated that the presence of migraine increased the incidence of stroke in young women from 10 in 100 000 woman-years to 19 in 100 000 woman-years. Therefore, the absolute risk of stroke associated with migraine is very small.

Hemostatic and Inflammatory Factors

Hemostatic factors have been related to incidence of cardiovascular disease generally, and in two prospective studies fibrinogen has been linked to increased stroke risk. In Göteborg there was an independent graded relation between fibrinogen levels and incidence of stroke in 54-year-old men.⁶¹ The Framingham Study confirmed these observations in men, but among women the relation did not reach statistical significance.⁶² Fibrinogen has also been prospectively linked to both progression of carotid artery stenosis and risk of recurrent stroke. The mechanisms by which fibrinogen may be related to stroke risk include effects on viscosity, platelets, and atherogenesis, as well as its direct role in clot formation as the substrate for thrombin.⁶³

The endogenous tissue-type plasminogen activator (TPA) system, the primary mediator of intravascular fibrinolysis, has been independently associated with risk of myocardial infarction and stroke.⁶⁴ In a nested case-control study within the Physicians’ Health Study, a graded prospective relation was found between TPA antigen and risk of first ischemic stroke in men aged 40 to 84 years. The apparent paradox of an association between ischemic stroke and plasma levels of a factor associated with fibrinolysis may be explained by the fact that only a small portion of TPA exists in the free active state, and most circulates as an inactive complex bound to plasminogen activator inhibitor-1 (PAI-1).⁶⁵ Thus, elevated TPA antigen reflects impaired fibrinolysis, predominantly due to elevations in PAI-1.

Homocysteine

Blood levels of homocysteine, produced from the essential amino acid methionine, can be determined by genetic factors and by intake of vitamins B6, B12, and folic acid. Numerous case-control studies have shown a strong relation between stroke and both basal and postmethionine load moderate hyperhomocysteinemia. There was evidence of a prospective relation to ischemic heart disease and extracranial carotid artery stenosis.⁶⁶ Recently the British Regional Heart Study showed a strong, independent, and graded relation of homocysteine level to stroke risk among middle-aged men.⁶⁷ Compared with the first quartile of homocysteine, the fourth quartile had an adjusted relative risk for stroke of 4.7 (1.1 to 20.0). Levels of homocysteine have been inversely related to oral intake and blood levels of folic acid and pyridoxine.⁶⁸ Because high levels of homocysteine are both atherogenic and prothrombotic, the relation with stroke is biologically plausible and has been demonstrated in an animal model. As many as 40% of persons with normal fasting levels of homocysteine developed hyperhomocysteinemia in response to a methionine load. Whether these persons are also at increased risk of stroke is unclear. Furthermore, although supplemental vitamins B6, B12, and folic acid may reduce blood levels of homocysteine, it has not been shown that this intervention will reduce incidence of stroke (or myocardial infarction).

Subclinical Disease

Subclinical disease, or disease detected noninvasively and without clinical signs or symptoms, is known to be related to both prevalent and incident stroke. Commonly performed subclinical disease measurements include (1) carotid ultrasonography for measurement of intimal-medial thickness, assessment of plaque characteristics, and quantification of flow-reducing lesions; (2) ankle-brachial blood pressure ratio or ankle-arm index for assessment of lower extremity arterial disease; and (3) cerebral magnetic resonance imaging (MRI) and computed tomography (CT) for detection of infarctlike lesions, white matter disease, and cerebral atrophy. Other subclinical disease measures, such as Doppler-defined carotid microemboli, positron emission tomographic abnormalities, and magnetic resonance angiography, are still being developed as research tools. Aortic arch atheromas detected by TEE have also been added to the growing list of stroke risk factors and are discussed in “Etiology of Stroke.”

Ultrasound measures of intimal medial thickness have been related to pathologically defined atherosclerosis and are highly reproducible. Duplex carotid ultrasonography can detect focal wall thickening, increased vessel diameter, luminal narrowing, flow limitations, and turbulence. Intimal medial thickness has also been strongly associated with prevalent stroke as well as with other stroke risk factors and overt cardiovascular disease, but its relation to incident stroke risk has yet to be defined prospectively in large-scale cohort studies.⁶⁹ Carotid ultrasound has identified plaque characteristics such as heterogeneity and irregular surfaces, which have been associated in clinical series with increased risk of subsequent stroke.

Reduced ankle-arm index has been strongly related to prevalent cardiovascular disease and its risk factors as well as carotid ultrasound abnormalities.⁷⁰ It is a strong and independent predictor of cardiovascular and total mortality but has not been as strongly related to incident stroke, perhaps because of small sample sizes.

Infarcts and infarctlike lesions on CT scanning were first associated with stroke in a population-based series of stroke patients in the Framingham Heart Study. MRI-defined infarcts were similarly associated with prevalent stroke in a population-based case-control study from the Cardiovascular Health Study.⁷¹ White matter hyperintensities have also been related to prevalent stroke, and this association has been reported to be independent of other risk factors.⁷² Cerebral atrophy and sulcal widening detected on cerebral imaging have similarly been shown to be associated with prevalent stroke.⁷² The predictive value for stroke of cerebral MRI findings, although highly plausible biologically, also remains to be demonstrated.

Prospective, population-based data on risk of stroke associated with these subclinical disease measures and the strength and independence of this risk from known cerebrovascular disease risk factors are expected shortly from continued follow-up of several large established cohorts. Interventional studies may be appropriate for determining whether reduction (or slowed progression) of intimal medial thickness, white matter disease, or cerebral atrophy reduce risk of stroke. Interventions known to reduce stroke risk, such as antihypertensive therapy, might be examined for their effects on subclinical disease in an attempt to understand the pathogenic role of subclinical disease in stroke.

Asymptomatic Carotid Stenosis

Despite data from both natural history studies and randomized controlled trials of carotid endarterectomy, the optimal management of individual patients with asymptomatic stenosis of the extracranial carotid artery remains controversial. Asymptomatic carotid artery stenosis may be suspected by the presence of a cervical bruit on a routine physical examination or if a stenosis is found on screening.

The presence of a cervical bruit raises concern that the patient has an underlying stenosis of the carotid artery and is at increased risk of stroke.⁷³ Although the detection of a condition with low prevalence such as a cervical bruit can be unreliable, population-based studies indicate that cervical or carotid bifurcation bruits are present in about 4% to 5% of persons older than 45.⁷⁴ The prevalence of cervical bruits in the general population increases with age from approximately 1% to 3% for those between the ages of 45 and 54 to 6% to 8% in persons older than 75. The rate of stroke in persons diagnosed with an asymptomatic cervical bruit is approximately 1% to 2% annually, and the risk of stroke is more than double for a person with a neck bruit.⁷⁵

Observational studies suggest the rate of unheralded stroke, ie, stroke without an antecedent transient ischemic attack (TIA) ipsilateral to a hemodynamically significant extracranial carotid artery stenosis is about 1% to 2% annually. Risk of stroke may be higher in patients with progressing and more severe stenosis. As with bruits, asymptomatic carotid stenosis is an important indicator of coexisting atherosclerosis and ischemic cardiac disease.⁷⁵

Controversy surrounds the role of prophylactic endarterectomy in persons with asymptomatic extracranial carotid artery stenosis. Despite four published randomized controlled trials, treatment of these patients remains unclear. The largest study, the Asymptomatic Carotid Atherosclerosis Study (ACAS),⁷⁶ enrolled 1662 patients with asymptomatic carotid artery stenosis >60%. The 1.2% risk of stroke from angiography contributed to a 2.3% aggregate perioperative stroke risk. After 2.7 years the study was terminated, and the 5-year projected aggregate rate of ipsilateral perioperative stroke or death in surgically treated patients was estimated at 5%. The corresponding rate in the medical group was 11%. This 55% relative risk reduction from 2% per year to 1% per year (P=.004) was not seen in women, for whom there was a nonsignificant reduction of 16%. Surprisingly, there was no relation between benefit and the degree of carotid artery stenosis.⁷⁶⁷⁷

In symptomatic carotid artery stenosis, six endarterectomies can be expected to prevent one stroke in 2 years, whereas for asymptomatic patients 67 operations would be needed.⁶⁷ The single stroke prevented must be weighed against the risk of stroke, myocardial infarction, and death. Benefit of carotid endarterectomy in comparison with medical therapy alone is highly dependent on surgical risk. In asymptomatic stenoses, perioperative complication rates much higher than the 2.3% for stroke or death reported in the ACAS trial would eliminate the benefit of surgery. However, a review of carotid endarterectomies performed at 12 academic medical centers found that of those patients without ipsilateral symptoms, 4.5% had perioperative stroke, myocardial infarction, or died.⁷⁸ This rate is similar to that found in the Veterans Administration Cooperative study but nearly twice that reported by the ACAS investigators.

Unfortunately it has not been possible to identify a subgroup of patients at high risk of ipsilateral unheralded stroke. Strategies of performing prophylactic carotid endarterectomy in persons at low risk of perioperative complications or of deferring surgery unless the patient experiences ipsilateral symptoms have therefore been advocated. Although evidence now shows a potential benefit of prophylactic carotid endarterectomy in selected patients, the controversy over how these data are to be translated into clinical practice and healthcare policy continues. Results of the ongoing European Asymptomatic Carotid Endarterectomy trial and effectiveness studies may further help clarify the issue. In the interim, both physician and patient need to fully understand the risks and benefits of pursuing medical treatment or prophylactic endarterectomy before a management plan is outlined.

Transient Ischemic Attacks

The average risk of stroke in patients with TIAs is about 4%. After adjustment for major cardiovascular risk factors predisposing a patient to stroke, a TIA remains a significant independent risk factor for both stroke and myocardial infarction.⁷⁹ TIA referrable to a high-grade carotid artery stenosis carries a higher risk for stroke than those beyond a mild stenosis, and the risk with hemispheric ischemic symptoms is greater than for retinal ischemia.⁸⁰ Recent-onset TIA has a higher risk for ischemic stroke than remote TIA, and the same may be true for “crescendo” TIA. Various other clinical features have a major effect on the absolute risk for individual patients (Table 2). It is clear that antiplatelet therapy substantially reduces risk for stroke (and other atherothrombotic events such as myocardial infarction and vascular death) in all high-risk patients, including those with TIAs.⁸¹ Data from these and other studies of the management of TIAs are addressed in “Etiology of Stroke.”

Multiple Risk Factors

Risk factors independently increase the probability of stroke and may also interact to increase the probability of stroke. Moreover, many persons have multiple borderline elevations of risk factor levels. To identify persons at greatest risk of TIA and stroke, a risk profile was developed using 36 years of follow-up data from Framingham.⁸² Gender-specific tables allow stroke probability to be determined by a point system based on age, systolic blood pressure, use of antihypertensive therapy, presence of diabetes, cigarette smoking, history of cardiovascular disease (coronary heart disease or congestive heart failure), and electrocardiographic abnormalities (left ventricular hypertrophy or atrial fibrillation). This risk profile provides a quantitative determination of probability of stroke, relative to what is average for a person of this age. For example, a 70-year-old man with a systolic blood pressure of 120 mm Hg may have several times the risk of stroke of someone with a systolic pressure of 180 mm Hg who is free of other risk factors. Probability of stroke increases with the presence of other abnormalities in the risk profile. The realization that the probability of stroke is increased severalfold by the presence of multiple risk factors may help both patient and physician to more fully appreciate the need for serious risk factor management.

Risk Factors for Recurrent Ischemic Stroke

As mortality from stroke declines and life expectancy of the US population increases, the number of persons with recurrent stroke and the cost of their care will become greater public health concerns. Recurrence is frequent and is a major contributor to stroke morbidity and mortality. The immediate period after a stroke carries the greatest risk for recurrence. In the Stroke Data Bank, of 1273 patients with infarcts, 3.3% had an early recurrence within 30 days.⁸³ Nearly one third of the recurrent strokes in 2 years of follow-up occurred within the first 30 days. Early stroke recurrence increased motor weakness scores, early mortality, and duration of hospital stay.

Long-term stroke recurrence rates range from 4% to 14% per year, with aggregate annual estimates of 6.1% for minor stroke and 9.0% for major stroke.⁸⁴⁸⁵ In the Stroke Data Bank the cumulative 2-year stroke recurrence rate was 14.1%. In the Northern Manhattan Stroke Study⁸⁵ stroke recurrence was frequent, with 25% suffering a recurrent stroke by 5 years. Moreover, mortality after a recurrent stroke was greater than after the index stroke.

Most studies characterize risk factors for initial stroke, whereas longitudinal studies often emphasize survival rather than stroke recurrence. Some studies have found no effect of hypertension and cardiac disease, while others suggest that these factors increased recurrence after stroke.

Risk Factors for Intracerebral and Subarachnoid Hemorrhage

While much of the public health stroke burden is attributed to ischemic stroke, accounting for 75% to 85% of the total, subarachnoid (SAH) and intracerebral hemorrhage (ICH), however, result in significant morbidity, mortality, and cost. Aneurysms and arteriovenous malformation are the most commonly identified causes of SAH, while hypertensive arteriolar disease, amyloid deposition, intra-infarct hemorrhage, and arteriovenous malformations are the bases of most ICHs. Both types of hemorrhage involve rupture of a damaged or abnormal vessel wall. Thus, differences as well as similarities in risk factors for ICH and SAH are expected (Table 3).

ICH is twice as common as SAH in population-based studies with CT documentation. ICH caused by gross structural vascular abnormalities such as vascular malformations or aneurysms make up only 4% to 5% of the total. Therefore, this discussion focuses on the remaining cases of spontaneous ICH where a defined pathological cause is often not found.

Age is the most important risk factor for spontaneous ICH; incidence increases exponentially with advancing age.⁸⁶ ICH is more common among men than women and among African-Americans than among whites.⁸⁷ The increase in incidence rates among African-Americans as compared with whites is due to increased rates in young and middle-aged adults. Asian populations have much higher reported incidence rates of ICH compared with population studies in Western countries, but most of the Asian studies were done before CT or without CT verification.

Hypertension is the most powerful modifiable risk factor for ICH. In the Hiroshima and Nagasaki cohort study, the risk of ICH increased with systolic blood pressure level.⁸⁸ Left ventricular hypertrophy has also been associated with a twofold to sevenfold increase in risk. Treatment trials have repeatedly shown treatment of hypertension to substantially decrease risk of ICH. The SHEP study was the first to show that treatment of isolated systolic blood pressure specifically decreased risk of ICH. Thus far, treatment of hypertension is the only proved preventive therapy for ICH.

Other modifiable risk factors significantly related to ICH include prior stroke, heavy use of alcohol, cocaine, anticoagulation, and thrombolytic therapy.⁸⁶ In the elderly, one of the most important identified risk factors for spontaneous lobar hemorrhage is amyloid angiopathy. Unfortunately, except for pathological documentation of the amyloid protein in brain blood vessels, there is no marker for this vasculopathy, even after hemorrhage has occurred. Advances in MRI and genetic testing for Alzheimer’s disease may help identify ICH due to amyloid angiopathy. Other risk factors such as very low serum cholesterol, ie, total serum cholesterol ≤160 mg/dL, and cigarette smoking require further study.

Most SAHs are due to aneurysmal rupture. In contrast to ICH, incidence of SAH increases only moderately with age. Although SAH is often considered “the young person’s hemorrhage,” the number of ICHs in persons under 65 is almost identical to the number of SAHs. For unknown reasons, women have a higher age-adjusted risk of SAH, and African-Americans have twice the age- and gender-adjusted risk of SAH of whites. The known excess hypertension among young and middle-aged blacks as compared with whites is one possible explanation for this difference. Approximately 5% to 6% of patients with a SAH due to aneurysms have a familial history of SAH. The usefulness of screening for unruptured aneurysms among subjects with a strong family history needs further study. Genetic defects are being investigated, and putative candidates involve abnormalities in structural proteins such as collagen. Other diseases associated with an increased risk of ruptured aneurysms include polycystic kidney disease, coarctation of the aorta, Marfan syndrome, Ehler Danlos syndrome, fibromuscular dysplasia, and other rare collagen or elastin disorders.

Cigarette smoking is the most important modifiable risk factor for SAH,⁸⁹ and giving up smoking decreases but does not eliminate the excess risk.⁵² The pathophysiology linking cigarette smoking and formation of aneurysms is not known. Speculative hypotheses include an elevated proteolytic enzyme activity released by macrophages in the lungs, early accelerated atherosclerosis, and transient elevations in blood pressure associated with use of nicotine.

Hypertension has also been shown to be an independent risk factor for SAH but not as strong a risk factor as smoking.⁸⁹ At present treatment of hypertension and smoking cessation are the two most effective preventive therapies for ICH and SAH. The importance of other potential risk factors for SAH, such as heavy use of alcohol and use of estrogen, are less clear.

Recommendations

Public Policy

Education

The Public

Healthcare Providers

Research Priorities

Hypertension

Cardiac Diseases

Diabetes and Glucose Metabolism

Use of Cigarettes, Alcohol, and Illicit Drugs

Hypercoagulable and Inflammatory Markers

Homocysteine

Lifestyle Factors

Subclinical Diseases

Intracerebral Hemorrhage

For reprint information, see page 1498.