Risk Factors for Stroke and Type of Stroke in Persons With

Isolated Systolic Hypertension

Barry R. Davis, MD, PhD; Thomas Vogt, MD; Philip H. Frost, MD; Alfredo Burlando, MD;
Jerome Cohen, MD; Alan Wilson, PhD; Lawrence M. Brass, MD; William Frishman, MD;
Thomas Price, MD; Jeremiah Stamler, MD;
for the Systolic Hypertension in the Elderly Program Cooperative Research Group
Background and PurposeWe sought to determine risk factors for stroke and stroke type in persons with isolated systolic
hypertension (ISH).
MethodsWe performed proportional hazards analyses of data from the Systolic Hypertension in the Elderly Program, a
double-blind, randomized, placebo-controlled trial of 4736 persons aged $60 years with ISH (systolic blood pressure,
160 to 219 mm Hg; diastolic blood pressure, ,90 mm Hg). One treatment group received chlorthalidone (12.5 to 25
mg/d) with step-up to atenolol (25.0 to 50.0 mg/d) or reserpine (0.05 to 0.10 mg/d), if needed. The other treatment group
received matching placebo. The main outcome measures were stroke, stroke or transient ischemic attack [TIA], and
stroke types: ischemic (including lacunar, atherosclerotic, and embolic) and hemorrhagic.
ResultsDuring an average follow-up of 4.5 years, 384 strokes or TIAs and 262 strokes (including 217 ischemic, 66
lacunar, 26 atherosclerotic, and 25 embolic strokes) were documented. In multivariate analyses, placebo treatment, older
age, smoking, history of diabetes, higher systolic blood pressure, lower HDL cholesterol, and ECG abnormality were
significantly associated (P,0.05) with increased incidence of stroke or TIA, stroke, or ischemic stroke. Greater lacunar
stroke risk was significantly related to placebo treatment, older age, history of diabetes (relative risk [RR]53.03; 95%
confidence interval [CI], 1.70 to 5.40), and smoking (RR53.04; 95% CI, 1.73 to 5.37). Greater atherosclerotic and
embolic stroke risk were significantly related to presence of carotid bruit (RR55.75; 95% CI, 2.50 to 13.24) and older
age (RR51.65 per 5 years; 95% CI, 1.25 to 2.18), respectively.
ConclusionsIn older persons with ISH, history of diabetes and smoking are important risk factors for lacunar stroke,
whereas carotid bruit and age are important risk factors for atherosclerotic and embolic stroke, respectively. (Stroke.
1998;29:1333-1340.)
Key Words: clinical trials n hypertension n lacunar infarction n stroke, ischemic

isk factors for stroke have been derived from analyses of

data collected during prospective and case-control studies of men and women.137 Reported predisposing factors are
high blood pressure, older age, male sex, black race, history
of coronary heart disease, stroke, TIA, diabetes mellitus,
sickle cell disease, presence of carotid bruit, atrial fibrillation,
cigarette smoking; alcohol consumption, elevated serum cholesterol, elevated hematocrit, LVH, higher BMI, and orthostatic hypotension.137 Few reports have assessed risk factors
for stroke in older persons3337 and in those with ISH, and few
have had the power to examine how risk factors differ for
varying types of strokes. This is one of very few opportunities
to assess risk factors prospectively in a major cohort.
The purpose of this article is to use data from the Systolic
Hypertension in the Elderly Program (SHEP) to assess risk

factors for the following cerebrovascular disease outcomes

stroke, stroke plus TIA, ischemic stroke, hemorrhagic stroke,
lacunar stroke, atherosclerotic stroke, and embolic stroke
among older persons with ISH.
SHEP was a double-blind, randomized, placebo-controlled
trial of treatment for ISH in persons aged 60 years and older.
Its primary objective was to determine whether antihypertensive drug treatment reduced risk of total stroke (nonfatal or
fatal) in a multiethnic cohort of men and women aged 60
years and older with ISH.38 40 Secondary goals were to
determine whether treatment of ISH would reduce coronary
and cardiovascular disease incidence as well as cause-specific
and all-cause mortality.
The trial showed a 36% reduction in incidence of stroke
(P,0.001), a 27% reduction in coronary heart disease

Received December 29, 1997; final revision received April 9, 1998; accepted April 9, 1998.
From the University of Texas School of Public Health, Houston (B.R.D.); Cancer Research Center of Hawaii, Honolulu (T.V.); University of California,
San Francisco (P.H.F.); University of California, Davis (A.B.); St Louis University School of Medicine (Mo) (J.C.); Robert Wood Johnson Medical
School, New Brunswick, NJ (A.W.); Yale University School of Medicine, New Haven, Conn (L.M.B.); Albert Einstein School of Medicine, Bronx, NY
(W.F.); University of Maryland School of Medicine, Baltimore (T.P.); and Northwestern University Medical School, Chicago, Ill (J.S.).
Complete listings of the Systolic Hypertension in the Elderly Program were published (JAMA. 1991;265:32553264).
Correspondence to Barry R. Davis, MD, PhD, University of Texas School of Public Health, 1200 Herman Pressler St, Houston, TX 77030. E-mail
davis@utsph.sph.uth.tmc.edu
1998 American Heart Association, Inc.

1333
Downloaded from http://stroke.ahajournals.org/
by guest on August 9, 2015

1334

Risk Factors for Stroke and Type of Stroke in SHEP

Selected Abbreviations and Acronyms

BMI 5 body mass index
CI 5 confidence interval
DBP 5 diastolic blood pressure
HDL-C 5 high-density lipoprotein cholesterol
ISH 5 isolated systolic hypertension
LDL-C 5 low-density lipoprotein cholesterol
LVH 5 left ventricular hypertrophy
RR 5 relative risk
SBP 5 systolic blood pressure
SHEP 5 Systolic Hypertension in the Elderly Program
TC 5 total cholesterol
TIA 5 transient ischemic attack

(P50.015), and a 33% reduction in combined incidence of

stroke or coronary heart disease (P,0.001). There was also a
25% reduction in the incidence of TIA (P50.089).40

Subjects and Methods

SHEP design and methods have been reported in detail.38 40 The
study recruited 4736 participants aged 60 years or older with ISH,
defined as mean SBP of 160 to 219 mm Hg and DBP ,90 mm Hg.
Persons were excluded if they had a history and/or signs of specified
major cardiovascular diseases, other major diseases, or medical
management problems. The study was approved by a review board at
each institution, and participants gave informed consent for screening and later study participation.
Participants were randomized in a double-blind manner. One
treatment group received chlorthalidone (12.5 to 25 mg/d) with
step-up to atenolol (25.0 to 50.0 mg/d) or reserpine (0.05 to 0.10
mg/d), if needed. The other treatment group received matching
placebo. Baseline SBP was used to establish a goal blood pressure
for each participant. For individuals with SBP $180 mm Hg, goal
was a reduction to ,160 mm Hg. For those at 160 to 179 mm Hg,
goal was a reduction of $20 mm Hg. Blood pressure above escape
criteria, despite maximal stepped-care therapy, was an indication for
prescribing open label active therapy.
The average follow-up of SHEP participants was 4.5 years. They
were seen monthly until their SBP reached goal level or until the
maximum level of stepped-care treatment was reached. All participants had quarterly visits from date of randomization. An ECG was
also done at entry and at the second and final annual visits.

Blood Sampling, Laboratory Methods, and

ECG Readings
Baseline blood samples were obtained at the second baseline visit,
immediately before randomization. Sixty-four percent of all blood
samples were collected with the patients in a fasting state. Lipids
determined included TC, HDL-C, and triglycerides. NonHDL-C,
the difference between TC and HDL-C, was calculated for all
participants with these measures at baseline. LDL-C was estimated
in those individuals fasting at baseline and with triglycerides
,4.50 mmol/L.41 Methods of analysis and external laboratory
surveillance have been described.42
An ECG abnormality was defined as one or more of the following
Minnesota codes: 1.1 to 1.3 (Q/QS), 3.1 to 3.4 (high R waves), 4.1
to 4.4 (ST depression), 5.1 to 5.4 (T-wave changes), 6.1 to 6.8
(atrioventricular conduction defects), 7.1 to 7.8 (ventricular conduction defects), 8.1 to 8.6 (arrhythmias), and 9.1 to 9.3 and 9.5
(miscellaneous items).43,44 ECG abnormalities were classified further
as either ischemia (codes 1.1 to 1.3, 3.1 to 3.4, 4.1 to 4.4, and 5.1 to
5.4) or arrhythmia or conduction defect (codes 6.1 to 6.8, 7.1 to 7.8,
8.1 to 8.6, 9.1 to 9.3, and 9.5).

End Point Ascertainment

Total stroke was the primary end point. Stroke was defined as rapid
onset of a new neurological deficit attributed to obstruction or

rupture in the cerebral arterial system.38,39 The defined deficit had to

persist for at least 24 hours unless death supervened and had to
include specific localizing findings confirmed by neurological examination and by CT or MRI scan, if available, with lack of evidence
of an underlying nonvascular cause. Fatal stroke determination was
based on either autopsy or death certificate diagnosis plus data on
preterminal hospitalization with definite diagnosis of stroke.
TIA was defined as rapid onset of focal neurological deficit lasting
more than 30 seconds and less than 24 hours, presumed to be due to
cerebral ischemia and without evidence of underlying nonvascular
cause.
For suspected stroke or TIA, a standardized neurological evaluation was performed by a SHEP neurologist. This evaluation and
notes from the attending neurologist, scans, other studies of the
brain, and any additional information were forwarded to the coordinating center. Death certificates and autopsy reports were obtained
for decedents. For hospitalizations and nursing home admissions,
discharge or admission papers were obtained.
CT or MRI films were read independently according to specified
criteria by two neuroradiologists. Discrepancies between the two
readings were adjudicated by a SHEP neurologist. The adjudicated
reading was used by the three coding physicians, two of whom were
neurologists, in confirming possible neurological events. These
physicians were blinded to randomization allocation.
The method of determining stroke type (hemorrhagic, ischemic, or
unknown) was similar to that used in the Stroke Data Bank.45
Hemorrhagic stroke was diagnosed if intracranial bleeding was
found by CT scan, MRI, lumbar puncture, or at autopsy and there
was no evidence on the brain image of bleeding late into an ischemic
infarction. Ischemic stroke was diagnosed when a focal deficit was
present and no blood was observed in the brain image or was found
by lumbar puncture. Stroke type unknown was diagnosed when the
definition of stroke was satisfied but there was insufficient evidence
to determine whether the stroke was hemorrhagic or ischemic.
Hemorrhagic strokes were classified as due to subarachnoid hemorrhage if the blood was seen in the subarachnoid space and as
intraparenchymal if the blood was seen within the brain substance
itself.
Ischemic strokes were subdivided into lacunar, embolic, atherosclerotic, or other/unknown subtypes on the basis of clinical information and the brain image. The algorithm for deciding in which of
the categories the ischemic stroke would be placed is delineated in
Table 1.

Statistical Methods
Descriptive statistics were determined for baseline characteristics.
The outcomes were time to first event (eg, stroke, stroke or TIA,
ischemic stroke). For strokes and strokes plus TIA, an individual was
censored at the time of nonstroke death, loss to follow-up (10
individuals), or the completion of the study. For stroke subtypes, an
individual was censored at the time of nonstroke death, loss to
follow-up (10 individuals), the completion of the study for those who
did not have a stroke, or the time of stroke for those who had a
different subtype of stroke. For example, for the outcome of lacunar
stroke, individuals were (1) censored at the time of loss to follow-up
if they were lost to follow-up without having had a lacunar stroke,
(2) censored at the time of death if they died from a cause that was
not classified as lacunar stroke, (3) censored at the time of stroke if
the stroke was not classified as lacunar, (4) censored at the end of the
study if they completed the study without having had a lacunar
stroke, or (5) counted as an event at the time of a lacunar stroke if
they experienced one.
Cumulative event rates were calculated by life-table methods. RRs
and percent changes were calculated by proportional hazards regression46 based on the entire duration of follow-up. Univariate regression analyses were done to explore potential risk factors. Multivariate regression analyses were also done to examine risk factors
adjusted for potential confounding covariates. The number of factors
included in the multivariate models was limited to approximately
10% of the total number of events under consideration.47,48 These
factors included randomization group, age, sex, SBP, pulse, presence

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

Davis et al
TABLE 1.

Definitions for Subtypes of Ischemic Strokes

TABLE 2.

July 1998

1335

Baseline Characteristics

A. Lacunar stroke: 112, or 113, or 4

n

Mean
(SD) or %

1. Angiogram, if done, shows no evidence of adjacent major artery

occlusion or severe stenosis

Variable
Randomization, % active treatment

4736

50.0%

2. By CT or MRI, a deep area of decreased density ,2 cm in maximum

length in a location compatible with the clinical presentation

Age, y

4736

71.6 (6.7)

Race, % white and other nonblack

4736

86.1%

3. Clinical syndrome of pure motor hemiparesis, pure sensory stroke, ataxia

hemiparesis, dysarthria clumsy hand syndrome, and a normal CT or MRI

Sex, % men

4736

43.2%

SBP, mm Hg

4736

170.3 (9.4)

DBP, mm Hg

4736

76.6 (9.7)

Orthostatic hypotension, %

4732

12.1%

Pulse, bpm

4736

70.8 (10.6)

Prior antihypertensive medication use, %

4. Autopsy evidence of lacunar stroke due to small-vessel disease

B. Embolic stroke: 1 or 2 or 3 or 4
1. Cerebral hemisphere infarction with a recognized source for emboli or
systemic emboli and no lacune by CT compatible with the clinical
presentation. Sources for emboli include atrial fibrillation, endocarditis, mitral
valve disease, clot in the heart by echocardiogram or CT, recent cardiac
surgery or trauma, or myocardial infarction
2. Hemorrhagic infarction (mottled) by CT or MRI
3. CT shows small ,1/2 lobe cortical infarction compatible with clinical
findings with no prior TIAs in the same territory

4736

33.3%

BMI, kg/m

4678

27.5 (5.0)

Education $12 y, %

4723

30.8%

Current smoking, %

4730

12.7%

History of diabetes, %

4734

10.1%

4. Autopsy shows area of infarction thought to be due to embolus

Alcohol use of $1 drink per week, %

4736

29.8%

C. Atherosclerotic stroke: 1 or 2

History of MI, %

4736

4.9%

1. Focal infarct in the setting of evidence for large-vessel disease,

consisting of preceding TIAs in the same vascular territory or carotid artery
bruit over the proximate artery or internal carotid occlusion or severe
stenosis ($70%) by angiography (at a relevant artery) or carotid duplex
sonography at the carotid bifurcation if compatible, with no evidence of
lacunar, mottled infarction, or small cortical infarct by CT and no sources of
emboli

History of CHD (MI/CABG/PTCA), %

4736

5.4%

History of stroke, %

4736

1.4%

Presence of carotid bruits, %

4736

7.1%

History of CVD (CHD/stroke/TIA/endarterectomy), %

4736

7.1%

History of valvular heart disease, %

4734

2.3%

Intermittent claudication, %

4730

2.3%

2. Autopsy evidence of infarction caused by atherosclerosis

Rose Questionnaire angina, %

4710

5.5%

D. Other unknown stroke: 1 or 2 or 3

ECG abnormality, %

4680

61.0%

1. All cases not classified by the above rules for lacunar, embolic, or
atherosclerotic infarction

Presence of definite LVH, %

4672

7.4%

Ischemic ECG abnormality, %

4680

40.9%

2. All cases that could be classified in more than one of the above
categories

Conduction/arrhythmia ECG abnormality, %

4680

29.1%

Aspirin use in prior 2 weeks, %

4736

17.6%

Estrogen use (women only), %

2690

7.6%

Serum glucose, mmol/L

4571

6.02 (1.86)

3. All cases attributed to arteritis or dissection of the arterial wall

of carotid bruits, current smoking, history of diabetes, history of

stroke, intermittent claudication, alcohol use, education, HDL-C,
uric acid, hematocrit, ECG abnormalities, and BMI (and its square
term).

Serum uric acid, mmol/L

4476

315.2 (83.2)

Hematocrit, %

4173

42.2 (4.6)

TC, mmol/L

4428

6.11 (1.15)

HDL-C, mmol/L

4432

1.39 (0.38)

Baseline Findings

NonHDL-C, mmol/L

4393

4.72 (1.14)

Previous reports have presented detailed data on baseline

findings for SHEP participants overall and for subgroups.40
Table 2 presents data on the baseline variables used in this
report.
The mean age of participants was 72 years; 57% were
women; ethnicities were 82% white, 14% black, 2.5% Asian,
1% Hispanic, and 0.5% others. Of all participants, 1.4%
reported a history of stroke, and 4.9% reported a history of
myocardial infarction. Stroke with apparent residual effects
and myocardial infarction within 6 months of randomization
were exclusion criteria. On physical examination, 7% had
carotid bruits. Mean SBP was 170.3 mm Hg; mean DBP was
76.6 mm Hg. Approximately 60% had an ECG abnormality.

TC/HDL-C ratio

4393

4.69 (1.43)

NonHDL-C/HDL-C ratio

4393

3.69 (1.43)

2853

1.62 (1.00)

LDL-C, mmol/L

2740

3.98 (1.04)

LDL-C/HDL-C ratio

2740

3.02 (1.10)

Results

Stroke and Stroke Plus TIA Incidence, Active

Treatment, and Placebo Groups
By life-table analyses, 5-year cumulative stroke rates were
5.1/100 for the active treatment group and 7.9/100 for the

Fasting
Triglycerides, mmol/L
Fasting and triglycerides ,4.50 mmol/L

bpm indicates beats per minute; MI, myocardial infarction; CHD, coronary
heart disease; CABG, coronary artery bypass graft; PTCA, percutaneous
transluminal coronary angioplasty; and CVD, cardiovascular disease.

placebo group. On the basis of proportional hazards regression analysis, RR was 0.64 (95% CI, 0.50 to 0.82). Absolute
reduction in 5-year risk of stroke was 30/1000. Corresponding data for coronary heart disease were RR of 0.75 (95% CI,
0.60 to 0.94) and absolute reduction in 5-year risk of 15/1000;
for incidence of all major cardiovascular diseases, RR was

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

1336

Risk Factors for Stroke and Type of Stroke in SHEP

0.67 (95% CI, 0.55 to 0.82), and absolute reduction in 5-year

risk was 34/1000.

Relation of Baseline Variables to Incidence of

Stroke, Stroke Plus TIA, and Stroke by Type:
Univariate Analyses
Stroke
For stroke and all other outcomes, only factors that showed
statistical significance are displayed in Table 3. Assignment
to placebo treatment, older age, higher SBP, higher pulse,
current smoking, history of diabetes, lower HDL-C, history of
stroke, presence of ECG abnormality, higher serum glucose,
and carotid bruit were significantly (P,0.05) related to
increased stroke risk (Table 3). With classification of ECG
abnormality into ischemic and conduction/arrhythmic, only
ischemic abnormality was significantly related to stroke.
Race, sex, prior antihypertensive drug use, hematocrit, ECG
LVH, DBP, orthostatic hypotension, and alcohol consumption were not significantly related to stroke incidence.
Stroke Plus TIA
Results for the broader cerebrovascular disease end point of
stroke plus TIA were similar to the foregoing with a few
differences. Additional statistically significant factors included DBP, history of cardiovascular disease, intermittent
claudication, and ECG LVH (Table 3).
Major Stroke Types: Ischemic and Hemorrhagic
There were 217 ischemic strokes, 28 hemorrhagic strokes,
and 17 strokes that could not be classified into either of these
categories. Assignment to placebo treatment, older age, black
race, higher SBP, current smoking, history of diabetes,
history of myocardial infarction, carotid bruit, ECG abnormality, lower HDL-C, higher hematocrit, higher serum glucose, and higher TC to HDL-C and nonHDL-C to HDL-C
ratios were significantly (P,0.05) related to increased incidence of ischemic stroke (Table 3). With classification of
ECG abnormality into ischemic and conduction/arrhythmic,
only ischemic abnormality was significantly related to ischemic stroke.
Since there were fewer hemorrhagic strokes than ischemic
strokes, the power to delineate factors significantly associated
with this outcome is low. BMI had a significant quadratic
relationship with hemorrhagic stroke. Those with low and
high BMIs had an increased risk. History of stroke, lower
serum uric acid, and estrogen use in women were associated
with increased hemorrhagic stroke risk.
Ischemic Stroke Subtypes: Lacunar, Atherosclerotic,
and Embolic
There were 66 lacunar, 26 atherosclerotic, 25 embolic, and
100 unknown type ischemic strokes. Assignment to placebo
treatment, older age, higher SBP, current smoking, history of
diabetes, higher serum glucose, and higher TC to HDL-C and
nonHDL-C to HDL-C ratios were significantly (P,0.05)
related to increased incidence of lacunar stroke (Table 2).
Carotid bruit and higher LDL-C to HDL-C ratio were
significantly (P,0.05) related to increased incidence of
atherosclerotic stroke, although the latter factor could only be
evaluated in a subset of participants with fasting blood

specimens (Table 3). Older age, lower DBP, and ECG

abnormality were significantly (P,0.05) related to increased
incidence of embolic stroke (Table 3). With classification of
ECG abnormality into ischemic and conduction/arrhythmic,
only ischemic abnormality was significantly related to embolic stroke risk.

Relation of Baseline Variables to Incidence of

Stroke, Stroke Plus TIA, and Stroke by Type:
Multivariate Analyses
Stroke and Stroke Plus TIA
In multivariate analyses, several baseline traits remained
significantly related (P,0.05) to increased risk of stroke and
stroke plus TIA (Table 4), including placebo treatment
assignment, older age, current smoking, history of diabetes,
higher SBP, lower HDL-C, and ECG abnormalities. Higher
pulse was significantly related to increased stroke risk.
Similar results were obtained within each treatment group
(active and placebo).

Major Stroke Types: Ischemic and Hemorrhagic

Placebo treatment assignment, older age, current smoking,
history of diabetes, higher SBP, lower HDL-C, and ECG
abnormalities were significantly related to increased ischemic
stroke risk in the multivariate analysis (Table 4).
History of stroke and BMI were significantly related to
increased hemorrhagic stroke risk (Table 4). BMI had a
significant quadratic relationship with hemorrhagic stroke.
Those with low and high BMIs had an increased risk.
In further multivariate analyses, aspirin use at baseline was
not significantly associated with ischemic or hemorrhagic
stroke risk. However, in an analysis among women, estrogen
use at baseline was associated with a significant increase in
hemorrhagic stroke risk (RR54.87; 95% CI, 1.49 to 15.90).
Ischemic Stroke Subtypes: Lacunar, Atherosclerotic,
and Embolic
Placebo treatment assignment, older age, history of diabetes
(RR53.03; 95% CI, 1.70 to 5.40), and smoking (RR53.04,
95% CI, 1.73 to 5.37) were significantly related to increased
lacunar stroke risk in the multivariate analysis (Table 4).
Carotid bruit (RR55.75; 95% CI, 2.50 to 13.24) and older
age (RR51.65 per 5 years; 95% CI, 1.25 to 2.18) were
significantly related to increased atherosclerotic and embolic
stroke risk, respectively, in the multivariate analysis (Table
4). Increased alcohol intake (RR50.31 for $1 drink per
week) and ECG abnormality (RR52.07) were of borderline
significance (P#0.055) for decreased atherosclerotic and
increased embolic stroke risk, respectively,
The multivariate models were run separately for all of the
above outcomes within each treatment group. The results
were essentially similar.

Discussion
In persons aged 60 years and older with ISH, risk factors for
stroke or stroke and TIA as a combined end point, in
multivariate analyses, were no antihypertensive treatment,
older age, higher SBP and heart rate, lower HDL-C, cigarette
smoking, history of diabetes mellitus, history of stroke, and

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

Davis et al
TABLE 3.

July 1998

1337

Univariate Cox Regressions

RR (95% CI)

Variable
Randomization (active vs placebo)
Age (per 5 y higher)
Race (nonblack vs black)
Sex (men vs women)
DBP (per 10 mm Hg higher)
SBP (per 10 mm Hg higher)
Pulse (per 10 bpm higher)
BMI (per kg/m2 higher)
Education ($12 y vs ,12 y)
Smoking (current vs otherwise)
History of diabetes (yes vs no)
Alcohol use ($1 vs ,1 drink per week)
History of MI (yes vs no)
History of stroke (yes vs no)
Presence of carotid bruits (yes vs no)
History of CVD (CHD/stroke/TIA endarterectomy
(yes vs no)
Intermittent claudication (yes vs no)
ECG abnormalities (yes vs no)
Presence of LVH (yes vs no)
Estrogen use [women only] (yes vs no)
Serum glucose (per 1.67 mmol/L higher)
Serum uric acid (per 59.5 mmol/L higher)
Hematocrit (per 5% higher)
HDL-C (per 0.39 mmol/L higher)
TC/HDL-C ratio (per unit higher)
NonHDL-C/HDL-C ratio (per unit higher)
Fasting and triglycerides ,4.50 mmol/L
LDL-C/HDL-C ratio (per unit higher)

Stroke
(n5262)

Stroke or TIA
(n5384)

Ischemic
Stroke
(n5217)

Hemorrhagic
Stroke
(n528)

Lacunar
Stroke
(n566)

Atherosclerotic
Stroke
(n526)

Embolic
Stroke
(n525)

0.64
(0.50.82)
1.25
(1.151.37)
0.77
(0.551.06)
1.17
(0.911.49)
0.95
(0.851.07)
1.22
(1.091.37)
1.08
(0.971.21)
0.99
(0.971.02)
0.76
(0.571.00)
1.49
(1.082.06)
2.14
(1.562.93)
0.91
(0.641.28)
0.79
(0.421.49)
2.70
(1.395.25)
1.62
(1.102.40)
1.13
(0.721.78)
1.73
(0.923.26)
1.57
(1.202.05)
1.42
(0.952.14)
1.24
(0.702.19)
1.20
(1.111.29)
1.01
(0.921.11)
0.88
(0.781.00)
0.78
(0.690.91)
1.10
(1.021.20)
1.10
(1.021.20)
1.10
(0.951.28)

0.67
(0.550.83)
1.19
(1.101.28)
0.97
(0.731.29)
1.21
(0.991.47)
0.90
(0.820.98)
1.16
(1.051.27)
1.06
(0.961.16)
0.98
(0.961.00)
0.87
(0.701.09)
1.43
(1.091.87)
2.07
(1.592.69)
1.01
(0.771.37)
1.05
(0.861.66)
2.97
(1.745.06)
1.84
(1.352.52)
1.44
(1.022.03)
2.07
(1.283.38)
1.36
(1.101.69)
1.56
(1.132.16)
1.18
(0.731.91)
1.16
(1.081.24)
1.04
(0.971.13)
0.87
(0.671.12)
0.81
(0.720.91)
1.11
(1.041.19)
1.11
(1.041.19)
1.16
(1.031.31)

0.63
(0.480.83)
1.27
(1.151.39)
0.71
(0.501.00)*
1.07
(0.811.38)
0.91
(0.811.02)
1.22
(1.071.38)
1.17
(0.991.38)
1.00
(0.971.02)
0.80
(0.591.08)
1.60
(1.122.23)
2.29
(1.633.21)
0.88
(0.651.19)
0.76
(0.380.54)
1.79
(0.744.35)
1.76
(2.162.68)
0.87
(0.501.53)
1.67
(0.823.39)
1.53
(1.142.04)
1.53
(0.992.35)
0.98
(0.501.93)
1.17
(1.081.28)
1.06
(0.961.17)
0.86
(0.750.99)
0.78
(0.670.91)
1.13
(1.041.24)
1.13
(1.041.24)
1.11
(0.651.19)

0.47
(0.211.04)
1.11
(0.851.46)
1.33
(0.404.41)
1.19
(0.562.49)
1.39
(0.852.21)
1.06
(0.721.56)
1.15
(0.821.62)
0.97
(0.901.05)
0.61
(0.251.51)
1.51
(0.594.10)
0.35
(0.052.59)
0.79
(0.331.84)
z z z\
zzz
5.77
(1.3724.30)
1.04
(0.204.39)
1.67
(0.495.41)
1.81
(0.2211.82)
1.46
(0.633.10)
0.50
(0.073.65)
4.50
(1.4314.14)
1.11
(0.851.41)
0.73
(0.540.99)
1.12
(0.731.70)
1.02
(0.691.44)
0.85
(0.631.14)
0.85
(0.631.14)
0.85
(0.531.36)

0.53
(0.320.88)
1.27
(1.071.51)
0.65
(0.351.19)
1.01
(0.621.64)
1.03
(0.801.34)
1.24
(0.991.55)
1.20
(0.971.50)
1.00
(0.951.04)
0.55
(0.301.01)
2.48
(1.424.31)
2.78
(1.564.94)
1.10
(0.651.84)
0.62
(0.152.53)
1.16
(0.168.36)
0.87
(0.322.39)
0.65
(0.202.06)
1.35
(0.335.53)
0.96
(0.591.56)
1.77
(0.853.71)
1.05
(0.323.42)
1.28
(1.141.45)
1.01
(0.841.21)
0.87
(0.671.12)
0.81
(0.621.07)
1.21
(1.051.40)
1.21
(1.051.40)
1.06
(0.791.44)

0.99
(0.462.13)
1.26
(0.961.66)
4.01
(0.5429.57)
1.17
(0.542.53)
0.80
(0.611.04)
1.31
(0.921.86)
1.05
(0.731.50)
0.95
(0.871.03)
0.41
(0.141.19)
1.30
(0.453.78)
1.72
(0.594.98)
0.31
(0.091.03)
0.80
(0.115.88)
3.01
(0.4122.21)
6.03
(2.6313.87)
1.14
(0.274.81)
3.63
(0.8615.36)
2.21
(0.895.51)
0.52
(0.073.81)
z z z\
zzz
0.99
(0.691.43)
0.90
(0.671.21)
1.06
(0.681.67)
0.90
(0.591.37)
1.07
(0.821.39)
1.07
(0.821.39)
1.54
(1.022.33)

0.55
(0.241.25)
1.72
(1.302.27)
0.64
(0.241.71)
0.79
(0.351.78)
0.77
(0.600.99)
1.30
(0.911.87)
1.24
(0.871.76)
1.03
(0.961.11)
1.07
(0.462.47)
0.30
(0.042.24)
1.83
(0.635.32)
0.59
(0.221.59)
1.74
(0.417.37)
3.19
(0.4323.60)
1.18
(0.285.07)
1.20
(0.285.07)
1.84
(0.2513.61)
3.53
(1.2110.28)
2.46
(0.847.16)
0.83
(0.116.25)
1.07
0.771.48)
1.05
(0.781.41)
0.89
(0.601.31)
0.76
(0.471.21)
1.12
(0.861.46)
1.12
(0.86.46)
1.04
(0.661.63)

P,0.05.
BMI plus BMI2 was not significant, P.0.05.
BMI plus BMI2 was significant, P,0.05.
P.0.05.
\Not calculable because of lack of convergence in Cox regression.

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

1338

Risk Factors for Stroke and Type of Stroke in SHEP

TABLE 4.

Multivariate Cox Regressions*

RR (95% CI)
Stroke or
TIA
(n5384)

Stroke
(n5262)

Ischemic
Stroke
(n5217)

Hemorrhagic
Stroke
(n528)

Lacunar
Stroke
(n566)

Atherosclerotic
Stroke
(n526)

Embolic
Stroke
(n525)

0.70

0.60

0.66

0.49

0.53

zzz

0.53

(0.570.86)

(0.450.79)

(0.500.87)

(0.221.08)

(0.320.88)

zzz

(0.241.21)

0.21

1.24

1.32

zzz

1.40

zzz

1.65

(1.121.31)

(1.121.38)

(1.191.47)

zzz

(1.171.66)

zzz

(1.252.18)

Sex (men vs women)

zzz

1.28

zzz

zzz

zzz

zzz

zzz

(0.951.73)

zzz

zzz

zzz

1.15

zzz
1.17

zzz

SBP (per 10 mm Hg higher)

zzz
1.12

zzz

zzz

zzz

zzz

(1.011.24)

(1.011.32)

(1.021.33)

zzz

zzz

1.15

1.14

zzz

zzz
1.19

zzz

1.08

zzz

zzz

zzz
5.75

zzz
zzz

Variable
Randomization (active vs placebo)
Age (per 5 y higher)

Pulse (per 10 bpm higher)

(0.981.20)

(1.011.30)

(1.011.30)

zzz

(0.951.48)

Presence of carotid bruits (yes vs no)

zzz

zzz

zzz

zzz

zzz

zzz
1.56

zzz
1.63

zzz
1.81

zzz
zzz

zzz
3.04

(2.5013.24)

Smoking (current vs otherwise)

zzz

zzz

(1.172.08)

(1.142.34)

(1.252.62)

zzz

(1.735.37)

zzz

zzz

zzz

3.03

zzz

zzz

zzz

zzz

History of diabetes (yes vs no)

History of stroke (yes vs no)
Intermittent claudication (yes vs no)
Alcohol use ($1 drinks per week vs not)
Education ($12 y vs ,12 y)

2.03

1.80

2.29

(1.542.67)

(1.262.57)

(1.603.29)

2.41

2.13

zzz

zzz
5.86

(1.705.40)
zzz

zzz

zzz

(1.374.23)

(1.004.56)

zzz

(1.3924.80)

zzz

zzz

zzz

1.51

zzz

zzz

zzz

zzz

zzz

zzz

(0.902.52)

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz
0.31

zzz

zzz

zzz
0.78

zzz

zzz

zzz

(0.091.03)

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

HDL-C (per 0.39 mmol/L higher)

Serum uric acid (per 59.5 mmol/L higher)

zzz

zzz

zzz
0.82

(0.571.06)
0.82

zzz
0.81

(0.730.93)

(0.700.96)

(0.690.95)

zzz

zzz

zzz

zzz

zzz

zzz

1.08

zzz

zzz

zzz

zzz

zzz

zzz
0.88

(0.971.20)

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz

zzz
2.07

zzz
2.91

zzz

(0.835.16)

(0.998.56)

Hematocrit (per 5% higher)

zzz

(0.771.01)

ECG abnormalities (yes vs no)

zzz
1.27

1.47

zzz
1.39

(1.011.59)

(1.091.99)

(1.021.90)

zzz

zzz

zzz

zzz
0.60

zzz

zzz

zzz

zzz

zzz

zzz

(0.430.82)

zzz

zzz

zzz

zzz

zzz

zzz

1.01

zzz

zzz

zzz

zzz

(1.001.01)

zzz

zzz

zzz

BMI (per kg/m2 higher)

BMI2 (per [kg/m2]2 higher)

zzz
zzz
Ellipses indicate that variable was not included in the multivariate analysis.
P#0.05.

presence of ECG abnormality. With the exception of ECG

abnormality, which has only been previously reported as
significantly related to stroke incidence in Manolio et al,33 our
data confirm those reported in previous studies. There are
reports that ECG signs of ventricular hypertrophy are associated with increase in stroke risk.2,3,19 In SHEP, the major
ECG abnormality that contributes to stroke risk is the
ischemic component.
SHEP data contribute to understanding causes of and
potential for reduction of stroke. The primary risk factors

identified in SHEP are those often identified as contributors to atherosclerosis or markers of established atherosclerotic disease, and benefit resulted from antihypertensive treatment. SHEP active antihypertensive therapy was
associated with a 36% reduction in incidence of stroke.
The reduction in stroke rate at 5 years was estimated at
30/1000. This large effect occurred even though 35% of
those assigned to placebo took known antihypertensive
medication during the trial. This reduction was seen in
varying degrees across types of stroke.

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

Davis et al
Within the large category of ischemic stroke, lacunar
stroke risk appears to be associated particularly with smoking
and history of diabetes, whereas atherosclerotic and embolic
stroke risk appears to be related especially to signs of
established cardiovascular disease, eg, carotid bruit and ECG
abnormality (although of borderline significance). Carotid
bruit may indicate some degree of carotid stenosis, which is
known to be significantly associated with stroke incidence.32,33 Although atherosclerotic stroke includes presence
of a carotid bruit in its definition, other stroke subtypes
occurred in individuals with carotid bruits. In SHEP, of the 16
people with bruits who had a classifiable stroke, 8 (50%)
were atherosclerotic. In addition, embolic stroke includes the
presence of atrial fibrillation or recent myocardial infarction
in its definition, either of which might have been noted on the
baseline ECG. Of the 99 people with baseline ECG abnormalities who had a classifiable stroke, 21 (21%) had embolic
stroke. Among these 21 individuals, none had atrial fibrillation, and only three had some evidence of old infarction.
These three participants had no clinical history of a prior MI,
and they experienced their strokes between 20 and 27 months
after entry into the trial.
Risk factors for lacunar stroke found in past studies include
diabetes, smoking, hypertension, and physical inactivity.11,26
In SHEP, diabetes and smoking were significantly related to
lacunar stroke risk. All subjects in SHEP had systolic hypertension; those randomized to active treatment experienced
reduction in their risk of lacunar stroke. Physical activity was
not measured in SHEP, but those with higher baseline heart
rate did have higher risk.
According to the lacunar hypothesis, hypertensive smallvessel disease is the most important cause of lacunar stroke as
opposed to the causes of atherosclerosis and embolism.26 In
addition, diabetes mellitus may cause microatheroma in small
vessels, and such changes may be present in lacunar infarction.267 In SHEP, lacunar strokes accounted for 56% (66/117)
of the classifiable ischemic strokes and the largest reduction
in type of ischemic strokes between treatment groups.
Stroke and ischemic stroke risk factors found in other
studies were in general agreement with those found in SHEP.
A recent review article27 listed ranges of RRs for modifiable
ischemic stroke risk factors that were similar to those of
SHEP: RRs of 1.5 to 3.0 (2.3 in SHEP) for diabetes; 1.5 to 2.9
(1.8 in SHEP) for cigarette smoking; and 2.0 to 4.0 for
cardiac disease (1.4 in SHEP for ECG abnormality). A recent
case-control study showed that TIAs, diabetes, smoking, and
ischemic heart disease were risk factors for ischemic stroke.28
HDL-C was inversely associated with ischemic stroke mortality in one study by a magnitude similar to that found for
ischemic stroke risk in SHEP.31 In addition, the other factors
in that study associated with ischemic stroke mortalityage,
SBP, diabetes, and smokingwere significant risk factors for
ischemic stroke in SHEP. Among Japanese-American men in
the Honolulu Heart Program, older age, elevated SBP, increased glucose, smoking, LVH by ECG, and history of CHD
were significantly associated with increased thromboembolic
stroke risk.24 Elevated blood pressure, cigarette smoking,
diabetes, elevated serum cholesterol, and history of myocardial infarction were associated with increased nonhemor-

July 1998

1339

rhagic ischemic stroke death risk in long-term follow-up of

those screened in the Multiple Risk Factor Intervention
Trial.23 In a case-control study to evaluate the effect of
alcohol, Palomaki and Kaste21 showed that light to moderate
alcohol intake has an inverse association with the risk of
ischemic stroke. For atherosclerotic strokes in SHEP, there
appeared to be an association (69% lowered risk) with
moderate alcohol intake (P50.055). In a prospective study of
a large cohort of men in England, elevated hematocrit was an
independent risk factor for all stroke, and this was most
apparent in those with hypertension.25 In SHEP, there was a
12% lowered risk associated with a 5% higher hematocrit
(P50.06).
In a case-control study from the Melbourne Risk Factor
Study, hypertension was the most important risk factor for
intracerebral hemorrhagic risk.29 Low serum cholesterol and
high BMI were also associated with an increased risk.
Aspirin-like drugs were not associated with an increased risk.
The results of studies examining the association between
estrogen replacement therapy and stroke have been inconsistent.49 The greater than 3.5-fold increase in hemorrhagic
stroke risk seen in women in SHEP needs further clarification
from other studies of stroke in women, especially hypertensive women.
Combining SHEP data with confirmed stroke risk factors
from other studies indicates that patients with ISH plus one or
more of the following traits are at considerably greater risk of
stroke: high SBP and/or DBP without adequate therapeutic
control, diabetes mellitus, low HDL-C, cigarette smoking,
and history of stroke. These patients are also at greater risk of
coronary heart disease.50 Patients with hypertension who also
have one or more of the other risk factors should be followed
and treated comprehensively with particular care.

Acknowledgments
The SHEP trial was supported by contracts from the National, Heart,
Lung, and Blood Institute and the National Institute on Aging. Drugs
were supplied by the Lemmon Company, Sellersville, Pa; Wyeth
Laboratories and AH Robins Company, Richmond, Va; and Stuart
Pharmaceutical, Wilmington, Del. It is a pleasure to acknowledge the
contribution of the investigators and staff at the 16 clinical centers
and coordination and service centers of the SHEP Cooperative
Research Group.

References
1. Paffenbarger RS Jr, Williams JL. Chronic disease in former college
students, V: early precursors of fatal stroke. Am J Public Health. 1967;
57:1290 1299.
2. Kagan A, Popper JS, Rhoads GG. Factors related to stroke incidence in
Hawaii Japanese men: the Honolulu Heart Study. Stroke. 1980;11:14 21.
3. Tanaka H, Ueda Y, Hayashi M, Date C, Baba T, Yamashita H, Shoji H,
Tanaka Y, Owada K, Detels R. Risk factors for cerebral hemorrhage and
cerebral infarction in a Japanese rural community. Stroke. 1982;13:6273.
4. Wolf PA, Kannel WB, Verter J. Current status of risk factors for stroke.
Neurol Clin. 1983;1:317343.
5. Gill JS, Zezulka AV, Shipley MJ, Gill SK, Beevers DG. Stroke and
alcohol consumption. N Engl J Med. 1986;315:10411046.
6. Donahue RP, Abbott RD, Reed DM, Yano K. Alcohol and hemorrhagic
stroke: the Honolulu Heart Program. JAMA. 1986;255:23112314.
7. Hillbom ME. What supports the role of alcohol as a risk factor for stroke?
Acta Med Scand. 1987;717(suppl):93106.
8. Colditz GA, Bonita R, Stampfer MJ, Willett WC, Rosner B, Speizer FE,
Hennekens CH. Cigarette smoking and risk of stroke in middle-aged
women. N Engl J Med. 1988;318:937941.

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

1340

Risk Factors for Stroke and Type of Stroke in SHEP

9. Wolf PA, DAgostino RB, Kannel WB, Kannel WB, Bonita R, Belanger
AJ. Cigarette smoking as a risk factor for stroke: the Framingham Study.
JAMA. 1988;259:10251029.
10. Gandolfo C, Caponnetto C, Del Sette M, Santoloci D, Loeb C. Risk
factors in lacunar syndromes: a case-control study. Acta Neurol Scand.
1988;77:2226.
11. Stampfer MJ, Colditz GA, Willett WC, Speizer FE, Hennekens CH. A
prospective study of moderate alcohol consumption and the risk of coronary disease and stroke in women. N Engl J Med. 1988;319:267273.
12. Gill JS, Shipley MJ, Tsementzis SA, Hornby R, Gill SK, Hitchcock ER,
Beevers DG. Cigarette smoking: a risk factor for hemorrhagic and nonhemorrhagic stroke. Arch Intern Med. 1989;149:20532057.
13. Dobkin BH. Orthostatic hypotension as a risk factor for symptomatic
occlusive vascular disease. Neurology. 1989;39:30 34.
14. Gorelick PB. The status of alcohol as a risk factor for stroke. Stroke.
1989;20:16071610.
15. Camargo CA Jr. Moderate alcohol consumption and stroke: the epidemiological evidence. Stroke. 1989;20:16111626.
16. Iso H, Jacobs DR, Wentworth D, Neaton JD, Cohen JD. Serum cholesterol levels and six-year mortality from stroke in 350,977 men screened
for the Multiple Risk Factor Intervention Trial. N Engl J Med. 1989;320:
904 910.
17. Kittner SJ, White LR, Losonczy KG, Wolf PA, Hebel JR. Black-white
differences in stroke incidence in a national sample: the contribution of
hypertension and diabetes mellitus. JAMA. 1990;264:12671270.
18. Kannel WB, Wolf P, Verter J. Manifestations of coronary heart disease
predisposing to stroke: the Framingham Study. JAMA. 1991;250:
29422946.
19. Wolf PA, Belanger AJ, DAgostino RB. Management of risk factors.
Neurol Clin. 1992;10:177191.
20. Gorelick PB, Kelly MA. Alcohol as a risk factor for stroke. Heart Dis
Stroke. 1992;1:255258.
21. Palomaki H, Kaste M. Regular light-to-moderate intake of alcohol and the
risk of ischemic stroke: is there a beneficial effect? Stroke. 1993;24:
1828 1832.
22. Haheim LL, Holme I, Hjermann I, Leren P. Risk factors of stroke
incidence and mortality: a 12-year follow-up of the Oslo Study. Stroke.
1993;24:1484 1489.
23. Neaton JD, Wentworth DN, Cutler J, Stamler J, Kuller L. Risk factors for
death from different types of stroke: Multiple Risk Factor Intervention
Trial Research Group. Ann Epidemiol. 1993;3:493 499.
24. Yano K, Popper JS, Kagan A, Chyou PH, Grove JS. Epidemiology of
stroke among Japanese men in Hawaii during 24 years of follow-up: the
Honolulu Heart Program. Health Rep. 1994;6:28 38.
25. Wannamethee G, Perry IJ, Shaper AG. Hematocrit, hypertension, and risk
of stroke. J Intern Med. 1994;235:163168.
26. You R, McNeil JJ, OMalley HM, Davis SM, Donnan GA. Risk factors
for lacunar infarction syndromes. Neurology. 1995;45:14831487.
27. Sacco RL. Risk factors and outcomes for ischemic stroke. Neurology.
1995;45(2 suppl 1):S10 S14.
28. Whisnant JP, Wiebers DO, OFallon WM, Sicks JD, Frye RL. A
population-based model of risk factors for ischemic stroke: Rochester,
Minnesota. Neurology. 1996;47:1420 1428.
29. Thrift AG, McNeil JJ, Forbes A, Donnan GA. Risk factors for cerebral
hemorrhage in the era of well-controlled hypertension: Melbourne Risk
Factor Study (MERFS) Group. Stroke. 1996;27:2020 2025.
30. Bogousslavsky J, Castillo V, Kumral E, Henriques I, Melle GV. Stroke
subtypes and hypertension: primary hemorrhage vs infarction, large- vs
small-artery disease. Arch Neurol. 1996;53:265269.
31. Tanne D, Yaari S, Goldbourt U. High-density lipoprotein cholesterol and
risk of ischemic stroke mortality: a 21-year follow-up of 8586 men from
the Israeli Ischemic Heart Disease Study. Stroke. 1997;28:83 87.
32. Sacco RL Benjamin EJ, Broderick JP, Dyken M, Easton JD, Feinberg
WM, Goldstein LB, Gorelick PB, Howard G, Kittner SJ, Manolio TA,
Whisnant JP, Wolf PA. Risk factors. Stroke. 1997;28:15071517.

33. Manolio TA, Kronmal RA, Burke GL, OLeary DH, Price TR, for the
CHS Collaborative Research Group. Short-term predictors of incident
stroke in older adults: the Cardiovascular Health Study. Stroke. 1996;27:
1479 1486.
34. Zeiler K, Siostrzonek P, Lang W, Gossinger H, Oder W, Ciciyasvilli H,
Kollegger H, Mosslacher H, Deecke L. Different risk factor profiles in
young and elderly stroke patients with special reference to cardiac disorders. J Clin Epidemiol. 1992;45:13831389.
35. Woo J, Lau EM. Risk factors predisposing to stroke in an elderly Chinese
population: a longitudinal study. Neuroepidemiology. 1990;9:131134.
36. Welin L, Svardsudd K, Wilhelmsen L, Larsson B, Tibblin G. Analysis of
risk factors for stroke in a cohort of men born in 1913. N Engl J Med.
1987;317:521526.
37. Wolf PA, DAgostino RB, Belanger AJ, Knell WB, Probability of stroke:
a risk profile from the Framingham Study. Stroke. 1991;22:312318.
38. The Systolic Hypertension in the Elderly Program Cooperative Research
Group: Rationale and design of a randomized clinical trial on prevention
of stroke in isolated systolic hypertension. J Clin Epidemiol. 1988;41:
11971208.
39. Borhani NO, Applegate WB, Cutler JA, Davis BR, Furberg CD, Lakatos
E, Page L, Perry HM, Smith WF, Probstfield J. The Systolic Hypertension
in the Elderly Program (SHEP), part 1: rationale and design. Hypertension. 1991;17(suppl II):II-2-II-15.
40. The Systolic Hypertension in the Elderly Cooperative Research Group.
Prevention of stroke by antihypertensive drug treatment in older persons
with isolated systolic hypertension: final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;265:32553264.
41. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein in plasma without the use of the
preparative ultracentrifuge. Clin Chem. 1972;18:499 502.
42. Hall WD, Davis BR, Frost P, Hoffmeier M, OBrien JE, Pace S, Page L,
Schneider KA, Stamler J. The Systolic Hypertension in the Elderly
Program (SHEP), part 7: baseline laboratory characteristics. Hypertension. 1991;17(suppl II):II-102-II-122.
43. Kostis JB, Prineas R, Curb JD, Lee M, Berkson D, Raines J, Frishman W,
Francis CK, Sheffield T. The Systolic Hypertension in the Elderly
Program (SHEP), part 8: electrocardiographic characteristics. Hypertension. 1991;17(suppl III):III-23-III-51.
44. Prineas RJ, Crow RS, Blackburn H. The Minnesota Code Manual of
Electrocardiographic Findings: Standards and Procedures for Measurement and Classification. Littleton, Mass: John Wright-PSG Inc;
1982.
45. Foulkes MA, Wolf PA, Price TR, Mohr JP, Hier DB. The Stroke Data
Bank: design, methods and baseline characteristics. Stroke. 1988;19:
547554.
46. Cox DR. Regression models and life-tables. J R Stat Soc. 1972;34:
187220.
47. Concato J, Peduzzi P, Holford TR, Feinstein AR. The importance of
events per independent variable (EPV) in proportional hazards analysis, I:
background, goals, and general strategy. J Clin Epidemiol. 1995;48:
14951501.
48. Peduzzi P, Concato J, Feinstein AR, Holford TR. The importance of
events per independent variable (EPV) in proportional hazards analysis,
II: accuracy and precision of regression estimates. J Clin Epidemiol.
1995;48:15031510.
49. Pagnani-Hill A. Estrogen replacement therapy and stroke. Prog Cardiovasc Dis. 1995;38:223242.
50. Frost PH, Davis BR, Burlando AJ, Curb JD, Guthrie GP, Isaacsohn JL,
Wassertheil-Smoller S, Wilson AC, Stamler J. Coronary heart disease
risk factors in men and women aged 60 and older: findings from the
Systolic Hypertension in the Elderly Program. Circulation. 1996;94:
26 34.

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015

Risk Factors for Stroke and Type of Stroke in Persons With Isolated Systolic Hypertension
Barry R. Davis, Thomas Vogt, Philip H. Frost, Alfredo Burlando, Jerome Cohen, Alan Wilson,
Lawrence M. Brass, William Frishman, Thomas Price and Jeremiah Stamler
for the Systolic Hypertension in the Elderly Program Cooperative Research Group
Stroke. 1998;29:1333-1340
doi: 10.1161/01.STR.29.7.1333
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 1998 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/29/7/1333

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/

Downloaded from http://stroke.ahajournals.org/ by guest on August 9, 2015