Risk factors for intracerebral hemorrhage differ according to

hemorrhage location
Sharyl R. Martini, Matthew L. Flaherty, W. Mark Brown, et al.
Neurology 2012;79;2275-2282 Published Online before print November 21, 2012
DOI 10.1212/WNL.0b013e318276896f

This information is current as of November 21, 2012

The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://www.neurology.org/content/79/23/2275.full.html

Neurology ® is the official journal of the American Academy of Neurology. Published continuously
since 1951, it is now a weekly with 48 issues per year. Copyright © 2012 American Academy of
Neurology. All rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.
 Risk factors for intracerebral hemorrhage
differ according to hemorrhage location

Sharyl R. Martini, MD, ABSTRACT

PhD Objectives: Risk factors have been described for spontaneous intracerebral hemorrhage (ICH); their rel-
Matthew L. Flaherty, MD ative contribution to lobar vs nonlobar hemorrhage location is less clear. Our purpose here was to inves-
W. Mark Brown, MA tigate risk factors by hemorrhage location.
Mary Haverbusch, RN,
Methods: This case-control study prospectively enrolled subjects with first-ever spontaneous ICH and
BSN
matched each with up to 3 controls by age, race, and gender. Conditional stepwise logistic regression
Mary E. Comeau, MA
modeling was used to determine significant independent risk factors for lobar and nonlobar ICH.
Laura R. Sauerbeck, RN,
MS Results: From December 1997 through December 2006, 597 cases and 1,548 controls qualified for
Brett M. Kissela, MD, MS the analysis. Hypertension, warfarin use, first-degree relative with ICH, personal history of ischemic
Ranjan Deka, PhD stroke, less than a high school education, and APOE e2 or e4 genotype were more common in ICH
Dawn O. Kleindorfer, cases. Hypercholesterolemia and moderate alcohol consumption (#2 drinks per day) were less common
MD in ICH cases. The associations of hypertension and hypercholesterolemia were specific for
Charles J. Moomaw, PhD nonlobar ICH. Conversely, the association of APOE e2 or e4 genotype was specific for lobar ICH.
Joseph P. Broderick, MD Conclusions: APOE e2 or e4 genotype was associated specifically with lobar ICH. Hypertension was
Carl D. Langefeld, PhD associated specifically with nonlobar ICH. A protective association was seen between hypercholesterol-
Daniel Woo, MD, MS emia and nonlobar ICH; no such association was identified for lobar ICH. Neurology
2012;79:2275–2282

GLOSSARY
Correspondence & reprint ARIC 5 Atherosclerosis Risk in Communities; CHS 5 Cardiovascular Health Studies; CI 5 confidence interval; GCNK 5
requests to Dr. Woo: Greater Cincinnati/Northern Kentucky; GCNKSS 5 Greater Cincinnati/Northern Kentucky Stroke Study; GERFHS 5 Genetic
[email protected] and Environmental Risk Factors in Hemorrhagic Stroke; ICD-9 5 International Classification of Diseases, Ninth Revision;
ICH 5 intracerebral hemorrhage; NHANES I 5 National Health and Nutrition Examination Survey; OR 5 odds ratio; SAH 5
subarachnoid hemorrhage.

Intracerebral hemorrhage (ICH) accounts for approximately 20% of strokes worldwide, with 30-day
mortality estimates of 32%–50%.1–4 Of patients who survive, only 28%–35% are independent at 3
months.5,6 Prior reports from our group and others support the hypothesis that risk factors for ICH
vary according to hemorrhage location.7–10 The Genetic and Environmental Risk Factors in Hem-
orrhagic Stroke (GERFHS) Study is designed to examine the genetic and environmental variables
associated with hemorrhagic stroke in the biracial population of the Greater Cincinnati/Northern
Kentucky (GCNK) region (population 1.3 million, 16% black). In 2002, our preplanned interim
analysis examined the hypothesis that risk factors for ICH varied according to hemorrhage
location. Hypertension had the highest attributable risk for nonlobar ICH (e.g., basal ganglia,
Supplemental data at thalamus, brainstem, cerebellum, or periventricular white matter), whereas APOE alleles e2 and
www.neurology.org
e4 had the highest attributable risk for lobar ICH7—findings that have been replicated in other
Supplemental Data
studies.8–10 That report had a limited sample size and therefore limited ability to differentiate risk
factors by subgroup. The current report, with over 3 times the sample size of our interim report,
re-examines the hypothesis that risk factors for ICH vary in prevalence and attributable risk for lobar
vs nonlobar ICH.

METHODS Study design. GERFHS is a case-control study of hemorrhagic stroke that used population-based case ascertainment.
CME

From the Departments of Neurology (S.R.M., M.L.F., M.H., L.R.S., B.M.K., D.O.K., C.J.M., J.P.B., D.W.) and Environmental Health (R.D.),
University of Cincinnati College of Medicine, Cincinnati, OH; and Department of Biostatistical Sciences (W.M.B., M.E.C., C.D.L.), Wake Forest
University School of Medicine, Winston-Salem, NC.
Study funding: Supported by NIH (National Institute of Neurological Disorders and Stroke: R-01-NS 36695 and T-32-NS 047996; NIEH: R-01- ES 06096).
Go to Neurology.org for full disclosures. Disclosures deemed relevant by the authors, if any, are provided at the end of this article.

© 2012 American Academy of Neurology 2275

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 Setting. We prospectively surveyed hospital admission logs in an hypertensive (yes, no, or unknown) or elevated cholesterol/blood
attempt to identify all patients presenting to one of the 16 adult lipids (yes, no, or unknown). We elected to use self-report because
hospitals serving the GCNK region with potential cerebral hemor- physiologic variables may be altered as a result of the hemorrhage.
rhage from December 1997 to August 2001, and July 2002 to Study nurses asked about alcohol consumption (amount, frequency,
December 2006. Methods have been previously described.7,11 and type), cigarette smoking (never, former, or current smoker;
GERFHS defined ICH as the nontraumatic abrupt onset of severe number of pack-years), and education (less than high school, high
headache, altered level of consciousness, or focal neurologic deficit school or greater). We also recorded medication use by cases (imme-
associated with a focal collection of blood within the brain paren- diately prior to ICH onset) and controls.
chyma as observed on CT, on other neuroimaging, or at autopsy For APOE genotyping, we obtained 4 buccal brush samples or
(adapted from Classification of Cerebrovascular Disease III whole blood from each case and control at the time of interview.
[1989]).7 We abstracted data from medical charts of all individuals APOE genotype was determined by a PCR-based method to deter-
with possible hemorrhagic stroke, and followed individuals with mine genotype at single nucleotide polymorphisms rs429358 and
hemorrhagic stroke until hospital discharge (up to 30 days). rs7412.12,13 Our controls showed no significant departure from
Hardy-Weinberg expectations (p 5 0.3357 for white controls, and
Standard protocol approvals and patient consents. The Insti- p 5 0.0345 for black controls; a p value that rejects Hardy-Weinberg
tutional Review Board for each hospital system approved the study. equilibrium is ,0.0001).
We obtained a Certificate of Confidentiality from the Department
of Health and Human Services. Study personnel obtained informed Statistical analysis. Categorical variables are reported as propor-
consent for all subjects who underwent interview and genetic analysis. tions. Variables with yes/no/unknown responses were analyzed dichot-
omously as “reported” (“yes” responses) and “not reported” (“no” or
Participants. Eligibility criteria include age $18 years, residence “unknown” responses, or missing data); tables e-1, e-2, and e-3 on the
within 50 miles of the University of Cincinnati, and enrollment Neurology® Web site at www.neurology.org contain responses for
within 90 days of the index event. Hemorrhages associated with each variable by hemorrhage location. APOE genotype was analyzed
trauma, brain tumor, encephalitis, endarterectomy, hemorrhagic cere- in an additive model.10 Continuous variables are reported as mean with
bral infarction, or thrombolytic treatment of ischemic stroke did not standard deviation. Univariate analyses that examine the association of
meet study criteria. The GERFHS study collected data for subarach- independent variables with ICH are reported as odds ratios (ORs) with
noid hemorrhage (SAH) as well as ICH. This analysis of first-ever 95% confidence intervals (CIs).
spontaneous ICH excludes primary SAH and ICH due to a structural We conducted multivariate analysis using stepwise elimination.
cause (arteriovenous malformation, aneurysm, cavernous angioma, APOE genotype was analyzed as a 3-level continuous variable. We
venous angioma, dural fistula). We also excluded patients with prior analyzed hemorrhage location subgroups (lobar vs nonlobar) sepa-
ICH and those not of white or black race. We included patients with rately; an exploratory analysis stratified subjects by race. An additional
anticoagulant-associated ICH, pure intraventricular hemorrhage, and exploratory analysis examined hemorrhage location subgroups by
prior ischemic stroke. Study neurologists reviewed clinical and neuro- hypercholesterolemia, statin use, and APOE genotype. We calculated
imaging information for each patient and made the final decision attributable risk using the multivariate OR and the prevalence rate
about case eligibility. from the matched controls.
To assess completeness of ICH event ascertainment, we com-
pared the ICH events ascertained by the GERFHS study in 2005 RESULTS Of the 7,133 potential cerebral hemorrhage
with those ascertained independently by a population-based study
events screened between 1997 and 2006, we identified
of all stroke subtypes in our region, the Greater Cincinnati/North-
ern Kentucky Stroke Study (GCNKSS). The GCNKSS used hospi-
2,850 cases of ICH, of which 597 subjects with first-ever
tal discharge ICD-9 codes and outpatient-based ascertainment. We spontaneous ICH underwent interview and genetic test-
found 324 cases ascertained by both studies, 7 ascertained only by ing (figure). To determine whether these 597 inter-
GERFHS, and 33 ascertained only by GCNKSS (including 8 coro- viewed cases were representative of all the verified ICH
ner’s cases and 1 identified in an outpatient clinic). cases ascertained, we compared the abstracted data
Controls were selected from the general population by ran-
from interviewed cases vs noninterviewed cases. Non-
dom digit dialing. We attempted to match 3 controls of the same
gender, race, and age (65 years) to each case.
interviewed cases were significantly older. After adjusting
for age, interviewed cases had significantly lower mortal-
Variables. Study neurologists classified hemorrhages as lobar (involv- ity and higher documented histories of smoking and
ing predominantly the cortex and underlying white matter of the cere-
hypercholesterolemia. There were no significant differ-
bral hemisphere), deep (involving predominantly the basal ganglia,
periventricular white matter, thalamus, or internal capsule), cerebellar,
ences for the other documented risk factors (table 1).
or brainstem. When location was unclear, a group of study neurolo- We matched controls 3:1 for 385 of the 597 cases,
gists adjudicated the films. For the current analysis, we categorized 2:1 for 181 cases, and 1:1 for 31 cases, for a total of
hemorrhage location as lobar or nonlobar (all other categories). 1,548 controls. Although age was a criterion for match-
Study nurses interviewed each consented case (or proxy) and ing, the cases were, on average, 3 years older than the
control face-to-face in a highly structured, identical manner.7 To
controls (65.2 6 15.5 vs 62.2 6 14.6), which re-
determine ability to be interviewed, every case passed a screening test
consisting of 7 questions regarding orientation, ability to follow
sulted in a small but significant age effect (OR 1.13
commands, and attention; 31% did not pass the test and required per year, 95% CI 1.08–1.17; p , 0.001); thus, all
proxy interview. The first choice for a proxy was the spouse/live-in multivariate analyses were adjusted for age.
companion or designated power of attorney, followed by adult child,
All ICH. ICH cases were more likely to have a history of
parent, sibling, or close friend (in this order).
Race and risk-factor variables were determined by self-report. hypertension, current warfarin use, a first-degree relative
Among the interview questions, participants were asked whether a with ICH, a personal history of ischemic stroke, less than
doctor had ever told them they had high blood pressure/were a high school education, and APOE e2 or e4 genotype.

2276 Neurology 79 December 4, 2012

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 Figure Study design

ICH 5 intracerebral hemorrhage; IVH 5 intraventricular hemorrhage; SAH 5 subarachnoid hemorrhage.

Conversely, ICH cases were less likely to have a history of We performed an exploratory analysis of nonlobar
hypercholesterolemia or moderate alcohol consumption ICH predictors, stratified by race. Black cases were 6
(#2 drinks per day; table 2). times more likely to have had a history of hypertension
than black controls (OR 6.15, 95% CI 2.78–13.57,
Lobar ICH. Independent risk factors associated with p , 0.0001); white cases were twice as likely to have
lobar ICH were warfarin use, a prior history of ische- had hypertension as white controls (OR 2.42, 95%
mic stroke, less than a high school education, and CI 1.74–3.37; p , 0.0001). Among cases, black cases
APOE e2 or e4 genotype. Less than a high school (n 5 93) were more likely to have had hypertension
education carried the highest attributable risk for than their white counterparts (n 5 287; p , 0.0001).
lobar ICH (table 3). Given the 59.8% prevalence of hypertension in black
controls and the 47.2% prevalence in white controls,
Nonlobar ICH. Nonlobar ICH cases had an increased the risk of ICH attributable to hypertension was
likelihood of hypertension, warfarin use, first-degree rel- 75.4% in black subjects and 40.1% in white subjects.
ative with ICH, prior history of ischemic stroke, and less
than a high school education. Hypercholesterole- Hypercholesterolemia. We performed an exploratory
mia was less frequent in nonlobar ICH cases (table 4). analysis of hypercholesterolemia with respect to statin

Neurology 79 December 4, 2012 2277

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 support to the concept that lobar and nonlobar
Table 1 Comparison of variables between interviewed and noninterviewed
casesa
ICH result from different pathophysiologic factors.
We found hypercholesterolemia less frequently in
Interviewed Noninterviewed nonlobar ICH cases compared to controls, but found
Variable (n 5 597) (n 5 2,022) p Value
no difference between lobar ICH cases and controls.
Age, y 65.2 6 15.5 70.3 6 15.1 ,0.0001
The protective association between hypercholesterolemia
Male 49 46 0.67 and all ICH is supported by several studies.8,14–16 Our
White 79 82 0.96 results suggest that this finding is driven specifically by
Hemorrhage location, lobar 36 35 0.64 the association of hypercholesterolemia with nonlobar
30-day mortality (all cause)b 4 55 ,0.0001
ICH. The Rotterdam Scan Study found that those
with the highest quartile of triglycerides had the lowest
History of hypertension 67 69 0.57
rate of deep or infratentorial microbleeds; no such
History of high cholesterolb 22 19 0.01
relationship was seen with lobar microbleeds.17 Hyper-
Frequent alcohol use 8 5 0.36 cholesterolemia may therefore play a more relevant path-
(>2 drinks/day)
ophysiologic role in maintenance of deep penetrating
Smoking (current or former) 37 29 0.03
arterioles than in cortical arterioles. The mechanism of
Cocaine use 2 2 0.13
hypercholesterolemia’s association with nonlobar ICH is
Diabetes 19 20 0.65 unclear, but low cholesterol has been hypothesized to
Anticoagulation c
16 22 0.07 contribute to vascular fragility.18 We found no associa-
First-degree relative with 0.7 0.5 0.69 tion between statin use and ICH, regardless of hemor-
ICH
rhage location or APOE genotype. This lack of
History of ischemic stroke 8 10 0.44 association would seem to be at odds with the results
History of heart disease d
30 37 0.21 of the Stroke Prevention by Aggressive Reduction in
Cholesterol Levels trial, which found an increased risk
Abbreviation: ICH 5 intracerebral hemorrhage.
a
Values are percentages or mean 6 SD. of ICH among subjects randomized to atorvastatin 80
b
p Value ,0.05, after adjusting for age. mg vs placebo.19 Further research on statin dosage and
c
Warfarin, heparin, or low-molecular-weight heparin. target low-density lipoprotein would help clarify the
d
History of myocardial infarction, atrial fibrillation, prosthetic valve, cardiac bypass, cardiac
angioplasty, or pacemaker.
relationship between statin usage and ICH.
We found self-reported hypertension more fre-
quently in ICH cases compared to controls. Nonlobar
use and APOE genotype, in addition to hemorrhage ICH cases appear to drive this association, for we
location. We found no association between statin use found no such association with lobar ICH. Other
and ICH, regardless of hemorrhage location or APOE studies also found hypertension more frequently in
genotype. Moreover, the protective association between nonlobar compared to lobar ICH, but did not collect
high cholesterol and nonlobar ICH was of similar mag- hypertension frequency in the population as a whole,
nitude in statin users and nonusers (table e-4). preventing calculation of attributable risk.9,11 In the
current analysis, we determined the attributable risk
DISCUSSION Our data add to the growing body of for hypertension in nonlobar ICH, and found the
literature assessing risk factors for ICH and lend attributable risk to be significantly higher for black than

Table 2 Distribution of APOE genotypesa

ICH location Cases or controls (n) e2/e2 e2/e3 e2/e4 e3/e3 e3/e4 e4/e4 Missing

All ICH

Cases (597) 1.7 16.3 4.7 49.1 22.3 3.9 2.2

Controls (1,548) 1.2 12.3 2.4 55.0 20.2 2.8 6.2

Nonlobar ICH

Cases (380) 2.1 13.7 2.9 54.0 22.6 2.6 2.1

Controls (919) 1.2 13.2 2.2 53.6 20.0 2.6 5.0

Lobar ICH

Cases (217) 0.9 20.7 7.8 40.6 21.7 6.0 2.3

Controls (556) 1.1 10.6 2.7 53.6 20.7 3.1 8.3

Abbreviation: ICH 5 intracerebral hemorrhage.

a
Values are percentages.

2278 Neurology 79 December 4, 2012

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 Table 3 Univariate and multivariate ORs and attributable risks for intracerebral hemorrhagea

Cases Controls Univariate OR (95% CI); Multivariate OR (95% CI); Attributable

Variable (n 5 597)b (n 5 1,548)c p value p value risk (%)

Age, y 65.2 6 15.5 62.2 6 14.6

Hypertension 67.3 51.3 1.97 (1.60–2.42); ,0.001 1.92 (1.52–2.43); ,0.001 31.8

Warfarin use 14.4 3.3 4.63 (3.17–6.76); ,0.001 3.37 (2.20–5.16); ,0.001 7.4

Smoking (current or former) 60.6 61.4 0.97 (0.80–1.18); 0.745

First-degree relative with ICH 6.0 2.1 3.00 (1.84–4.90); ,0.001 2.45 (1.39–4.33); 0.002 2.9

Prior history of ischemic stroke 9.7 2.1 5.24 (3.26–8.44); ,0.001 3.87 (2.26–6.61); ,0.001 5.7

Education (<12th grade) 63.2 45.7 2.12 (1.73–2.60); ,0.001 1.86 (1.48–2.34); ,0.001 28.4

APOE e2 (per allele)d

1.40 (1.12–1.74); 0.003 1.50 (1.17–1.93); 0.001 7.8

APOE e4 (per allele)d 1.23 (1.02–1.48); 0.029 1.31 (1.06–1.62); 0.012 7.7

History of high cholesterol 34.2 42.6 0.67 (0.55–0.83); ,0.001 0.57 (0.45–0.72); ,0.001 222.9

Alcohol (>2 drinks/day vs nondrinkers) 7.4 5.8 1.19 (0.80–1.76); 0.058 1.28 (0.83–1.99); 0.083

Alcohol (£2 drinks/day vs nondrinkers) 23.3 31.2 0.67 (0.53–0.84); ,0.001 0.77 (0.59–1.0); 0.014 27.8

Abbreviations: CI 5 confidence interval; ICH 5 intracerebral hemorrhage; OR 5 odds ratio.

a
Age is expressed in years as mean (6SD); other variables are expressed as percentages. Referent group for smokers was never smokers, referent group for
education level was completion of high school or beyond, referent group for alcohol consumption was never or former drinkers. ORs/attributable risks are not
provided for age (subjects were matched by age), or for variables not included in multivariate analysis. For APOE genotype: b n 5 580; c n 5 1,418; d see table 2 for
distribution of APOE genotypes.

for white subjects. The higher attributable risk in black for stroke prevention in high-risk patients with
subjects is related in part to the higher frequency of atrial fibrillation.28,29 The risk of anticoagulant-associ-
hypertension in black controls, but may also be ated ICH in recent clinical trials has ranged from 0.38
related to blood pressure control, which could not be to 0.47% per patient year, and was lower (0.10%–
determined from our data. Given the relatively small 0.24% per patient year) for the newer anticoagulants
sample size of nonlobar ICH among black subjects, dabigatran and apixiban.30,31 Our results can serve as a
variables beyond hypertension may not have reached baseline for ongoing and future studies of ICH in
significance due to limited power. The absence of such the setting of anticoagulation. As the newer anti-
variables in a multivariate model may overestimate the coagulants are adopted, the per-person risk of antico-
influence of hypertension as a risk factor among agulant-associated ICH would be expected to
black subjects. decline if real-world experience with these agents
APOE alleles e2 and e4 appear to play a role in the mirrors that of the clinical trials.
pathogenesis of amyloid angiopathy.20,21 Most studies Alcohol intake has been variably reported to be
have found APOE alleles e2, e4, or both more fre- associated with ICH.15,16 Here we found a small pro-
quently in lobar ICH cases, although other studies have tective association of moderate alcohol consumption
found no association.10,12,22–25 The largest and most (#2 drinks/day) with ICH. This association was no
detailed analysis of APOE genotype and ICH was a longer significant when analyzed by ICH location,
meta-analysis that included our cases as well as those likely reflecting loss of power after stratification. We
from 6 other studies. That study found that alleles found no association of heavy alcohol use with ICH.
e2 and e4 were both associated with lobar ICH in Studies are difficult to compare because of the varying
white subjects, most strongly in subjects with a high definitions of alcohol use, but alcohol has not been
probability of cerebral amyloid angiopathy.10 Here consistently found to play a role in ICH.
we similarly found APOE alleles e2 and e4 to be Our finding that less than a high school education
associated with lobar ICH. is associated with approximately double the risk of
Our analysis confirms that use of warfarin is asso- ICH is consistent with the findings from the National
ciated with both lobar and nonlobar ICH.26 Warfarin Health and Nutrition Examination Survey (NHANES I)
use increased the odds of ICH in our present analysis study.32 Conversely, combined analysis of Atheroscle-
by 3- to 4-fold; however, the frequency of warfarin rosis Risk in Communities (ARIC) and the Cardiovas-
use in the population is relatively low so the attribut- cular Health Studies (CHS) found only a nonsignificant
able risk for anticoagulation was less than 10%. trend toward higher risk of ICH with lower education
The frequency of anticoagulation has increased over levels.15 Low education level is thought to reflect socio-
time,27 as studies have consistently demonstrated economic status and associated issues such as health
the superiority of warfarin over antiplatelet agents care access,33 thus the strength of its association with

Neurology 79 December 4, 2012 2279

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 2280
ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Neurology 79

Table 4 Univariate and multivariate ORs and attributable risks for intracerebral hemorrhage by locationa

Lobar Nonlobar
December 4, 2012

Casesb Controlsc Univariate OR Multivariate OR Attributable Casesd Controlse Univariate OR Multivariate OR Attributable
Variable (n 5 217) (n 5 556) (95% CI); p value (95% CI); p value risk (%) (n 5 380) (n 5 992) (95% CI); p value (95% CI); p value risk (%)

Age, y 66.4 6 63.3 6 64.4 6 61.5 6

16.0 15.5 15.2 14.0

Hypertension 54.8 53.1 1.04 (0.75–1.44); 74.5 50.3 2.95 (2.24–3.90); 2.87 (2.13–3.86); 48.4
0.834 ,0.001 ,0.001

Warfarin use 16.6 3.4 5.23 (2.88–9.50); 4.33 (2.22–8.45); 10.7 13.2 3.23 4.24 (2.60–6.94); 3.10 (1.78–5.40); 6.4
,0.001 ,0.001 ,0.001 ,0.001

Smoking (current or former) 58.5 60.3 0.92 (0.66–1.27); 61.8 62.0 1.0 (0.78–1.28);
0.595 0.991

First-degree relative with ICH 6.0 1.8 3.46 (1.51–7.91); 2.35 (0.93–5.96); 2.6 6.1 2.3 2.78 (1.51–5.10); 2.77 (1.38–5.58); 3.9
0.003 0.073 0.001 0.004

Prior history of ischemic 7.8 1.3 5.84 (2.40–14.19); 4.20 (1.46–12.05); 3.7 10.8 2.5 5.01 (2.85–8.81); 3.96 (2.11–7.40); 6.9
stroke ,0.001 0.008 ,0.001 ,0.001

Education (<12th grade) 60.8 44.1 2.01 (1.44–2.81); 1.86 (1.28–2.71); 27.9 64.5 46.7 2.19 (1.69–2.85); 2.12 (1.60–2.82); 34.3
,0.001 0.001 ,0.001 ,0.001

APOE e2 (per allele)f 2.03 (1.40–2.95); 2.28 (1.50–3.46); 16.8 1.14 (0.86–1.50);
,0.001 ,0.001 0.363

APOE e4 (per allele)f 1.40 (1.05–1.86); 1.43 (1.03–1.99); 10.9 1.12 (0.88–1.43);
0.023 0.033 0.339

History of high cholesterol 40.1 43.0 0.85 (0.61–1.19); 30.8 42.4 0.58 (0.45–0.76); 0.46 (0.34–0.62); 229.9
0.342 ,0.001 ,0.001

Alcohol (>2 drinks/day) 6.5 5.9 0.91 (0.47–1.77); 7.9 5.8 1.39 (0.85–2.27);
0.602 0.048

Alcohol (£2 drinks/day) 22.1 31.8 0.59 (0.4–0.87); 24.0 30.9 0.72 (0.54–0.96);
0.037 0.005

Abbreviations: CI 5 confidence interval; ICH 5 intracerebral hemorrhage; OR 5 odds ratio.

a
Age is expressed in years as mean (6SD); other variables are expressed as percentages. Referent group for smokers was never smokers, referent group for education level was completion of high school or beyond,
referent group for alcohol consumption was never or former drinkers. ORs/attributable risks are not provided for age (subjects were matched by age), or for variables not included in multivariate analysis. For APOE
genotype: b n 5 210, c n 5 499, d n 5 370, e n 5 919; f see table 2 for APOE genotype distribution.
 disease is likely to vary among populations. Notably the nonlobar hemorrhages result from different pathophys-
NHANES I study was designed to oversample in- iologic processes. Furthermore, ours is the first study to
dividuals of lower socioeconomic status, whereas suggest that the protective association of hypercholes-
the ARIC and CHS studies were not. terolemia varies by hemorrhage location. Ongoing stud-
We found no association between smoking and ies will further advance our knowledge of risk factors for
ICH, and this lack of association persisted regardless ICH, and will allow assessment of risk factors by both
of hemorrhage location. The ARIC and CHS studies race and hemorrhage location.
similarly found no association between ICH and smok-
ing, even when stratified by current smoking status and
AUTHOR CONTRIBUTIONS
pack-years.15 Conversely, the INTERSTROKE study
Dr. Martini: drafting and revising the manuscript for content, data acqui-
found a small association between current smoking sition, analysis and interpretation of the data. Dr. Flaherty: revising the
and ICH.16 Thus, if smoking is a risk factor for ICH, manuscript for content, study concept/design, analysis and interpretation
its effect is weak. of data, acquisition of data. Mr. Brown: statistical analysis, analysis and inter-
pretation of the data, revising the manuscript for content. Ms. Haverbusch:
Our study is not without limitations. Cohort study acquisition of data, analysis and interpretation of the data, study coordination.
design offers advantages over case-control design; how- Mrs. Comeau: statistical analysis, revising the manuscript for content. Ms.
ever, a cohort study with 597 cases at an incidence rate Sauerbeck: acquisition of data, study coordination. Dr. Kissela: revising the
manuscript for content, data acquisition. Dr. Deka: acquisition of data, anal-
of 20 per 100,000 per year4 would require 3 million
ysis and interpretation of data. Dr. Kleindorfer: revising the manuscript for
patient years of follow-up. Potential limitations of our content, data acquisition. Dr. Moomaw: revising the manuscript for content,
case-control study include inadequate capture of cases, analysis and interpretation of data, data management. Dr. Broderick: study
survival bias, and recall bias. Comparison with the concept/design, analysis and interpretation of data, revising the manuscript
for content, study supervision, obtaining funding. Dr. Langefeld: analysis
GCNKSS34,35 suggests that our method captures
and interpretation of data, statistical analysis, revising the manuscript for con-
.90% of primary ICH cases that occur in our region. tent. Dr. Woo: study concept/design, acquisition of data, analysis and inter-
Despite our aggressive prospective enrollment, the dif- pretation of data, revising the manuscript for content, study supervision,
ference in mortality between interviewed and non- obtaining funding.

interviewed subjects reveals a significant survival bias.

Nevertheless, the minimal differences in risk factors DISCLOSURE
between the interviewed and noninterviewed cases sup- S. Martini, M. Flaherty, and M. Brown report no disclosures. M. Haverbusch
port the external validity of our results. In fact, the lower receives salary support from grant NS 36695. M. Comeau reports no disclo-
frequency of hypercholesterolemia in noninterviewed sures. L. Sauerbeck received salary support from grant NS 36695. B. Kissela,
R. Deka, D. Kleindorfer, C. Moomaw, J. Broderick, and C. Langefeld report
cases suggests that the data presented here may be an no disclosures. D. Woo serves as the PI of the grant (NS36695) from which
underestimate of the true effect size of hypercholester- the data are derived. Go to Neurology.org for full disclosures.
olemia. Similarly, the trend toward more frequent anti-
coagulation in noninterviewed cases suggests that Received March 6, 2012. Accepted in final form August 9, 2012.
our results may underestimate the effect size of
REFERENCES
anticoagulation. Our structured interview assesses var-
1. Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster G.
iables by self-report, which raises the possibility of recall Volume of intracerebral hemorrhage: a powerful and easy-
bias, creating the potential for misclassification of expo- to-use predictor of 30-day mortality. Stroke 1993;24:987–
sure. Furthermore, self-report does not account for 993.
the severity of the hypertension or hypercholesterole- 2. Hemphill JC III, Bonovich DC, Besmertis L, Manley GT,
mia. Sensitivities for self-report of hypertension, Johnston SC. The ICH score: a simple, reliable grading scale
hypercholesterolemia, tobacco use, and alcohol con- for intracerebral hemorrhage. Stroke 2001;32:891–897.
sumption in prior studies ranged from 51%–86%, 3. Rosand J, Eckman MH, Knudsen KA, Singer DE,
Greenberg SM. The effect of warfarin and intensity of
and specificities ranged from 57%–95%.36–38 Cases
anticoagulation on outcome of intracerebral hemorrhage.
and controls had risk factor assessment by the same
Arch Intern Med 2004;164:880–884.
methodology, however, and common atherosclerotic 4. Sacco S, Marini C, Toni D, Olivieri L, Carolei A. Inci-
risk factors such as diabetes and cigarette smoking did dence and 10-year survival of intracerebral hemorrhage in
not show significant differences between cases and con- a population-based registry. Stroke 2009;40:394–399.
trols, which suggests that recall bias was not a significant 5. Rost NS, Smith EE, Chang Y, et al. Prediction of func-
factor in our analysis. tional outcome in patients with primary intracerebral hem-
This is the largest study with population-based orrhage: the FUNC score. Stroke 2008;39:2304–2309.
6. Schwarz S, Hafner K, Aschoff A, Schwab S. Incidence and
ascertainment of ICH cases that analyzes genetic and
prognostic significance of fever following intracerebral
environmental risk factors by hemorrhage location and
hemorrhage. Neurology 2000;54:354–361.
provides attributable risk estimates. We confirmed that 7. Woo D, Sauerbeck LR, Kissela BM, et al. Genetic and
APOE e2 or e4 genotype is associated with lobar ICH, environmental risk factors for intracerebral hemorrhage: pre-
whereas hypertension and high cholesterol are associ- liminary results of a population-based study. Stroke 2002;
ated with nonlobar ICH, suggesting that lobar and 33:1190–1195.

Neurology 79 December 4, 2012 2281

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 8. Zia E, Pessah-Rasmussen H, Khan FA, et al. Risk factors 24. Tzourio C, Arima H, Harrap S, et al. APOE genotype,
for primary intracerebral hemorrhage: a population-based ethnicity, and the risk of cerebral hemorrhage. Neurology
nested case-control study. Cerebrovasc Dis 2006;21:18–25. 2008;70:1322–1328.
9. Labovitz DL, Halim A, Boden-Albala B, Hauser WA, 25. Seifert T, Lechner A, Flooh E, Schmidt H, Schmidt R,
Sacco RL. The incidence of deep and lobar intracerebral Fazekas F. Lack of association of lobar intracerebral hemor-
hemorrhage in whites, blacks, and Hispanics. Neurology rhage with apolipoprotein e genotype in an unselected pop-
2005;65:518–522.
ulation. Cerebrovasc Dis 2006;21:266–270.
10. Biffi A, Sonni A, Anderson CD, et al. Variants at APOE
26. Flaherty ML, Haverbusch M, Sekar P, et al. Location and
influence risk of deep and lobar intracerebral hemorrhage.
outcome of anticoagulant-associated intracerebral hemor-
Ann Neurol 2010;68:934–943.
rhage. Neurocrit Care 2006;5:197–201.
11. Flaherty ML, Woo D, Haverbusch M, et al. Racial varia-
27. Flaherty ML. Anticoagulant-associated intracerebral hem-
tions in location and risk of intracerebral hemorrhage. Stroke
2005;36:934–937. orrhage. Semin Neurol 2010;30:565–572.
12. Woo D, Kaushal R, Chakraborty R, et al. Association of 28. Warfarin versus aspirin for prevention of thromboembo-
apolipoprotein e4 and haplotypes of the apolipoprotein E lism in atrial fibrillation: Stroke Prevention in Atrial Fibril-
gene with lobar intracerebral hemorrhage. Stroke 2005;36: lation II Study. Lancet 1994;343:687–691.
1874–1879. 29. Petersen P, Boysen G, Godtfredsen J, Andersen ED,
13. Rebeck GW, Reiter JS, Strickland DK, Hyman BT. Apolip- Andersen B. Placebo-controlled, randomised trial of warfarin
oprotein E in sporadic Alzheimer’s disease: allelic variation and and aspirin for prevention of thromboembolic complications
receptor interactions. Neuron 1993;11:575–580. in chronic atrial fibrillation: The Copenhagen AFASAK
14. Zhao CX, Cui YH, Fan Q, et al. Small dense low-density study. Lancet 1989;333:175–179.
lipoproteins and associated risk factors in patients with stroke. 30. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran
Cerebrovasc Dis 2009;27:99–104. versus warfarin in patients with atrial fibrillation. N Engl J
15. Sturgeon JD, Folsom AR, Longstreth WT Jr, Shahar E,
Med 2009;361:1139–1151.
Rosamond WD, Cushman M. Risk factors for intracerebral
31. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban
hemorrhage in a pooled prospective study. Stroke 2007;38:
versus warfarin in patients with atrial fibrillation. N Engl J
2718–2725.
Med 2011;365:981–992.
16. O’Donnell MJ, Xavier D, Liu L, et al. Risk factors for ischae-
32. Qureshi AI, Suri MF, Saad M, Hopkins LN. Educational
mic and intracerebral haemorrhagic stroke in 22 countries
attainment and risk of stroke and myocardial infarction.
(the INTERSTROKE study): a case-control study. Lancet
2010;376:112–123. Med Sci Monit 2003;9:CR466–473.
17. Wieberdink RG, Poels MM, Vernooij MW, et al. Serum 33. Kleindorfer DO, Lindsell C, Broderick J, et al. Impact of
lipid levels and the risk of intracerebral hemorrhage: the socioeconomic status on stroke incidence: a population-
Rotterdam study. Arterioscler Thromb Vasc Biol 2011;31: based study. Ann Neurol 2006;60:480–484.
2982–2989. 34. Broderick JP, Bonomo JB, Kissela BM, et al. Withdrawal
18. Konishi M, Iso H, Komachi Y, et al. Associations of serum of antithrombotic agents and its impact on ischemic stroke
total cholesterol, different types of stroke, and stenosis dis- occurrence. Stroke 2011;42:2509–2514.
tribution of cerebral arteries: The Akita pathology study. 35. Kleindorfer DO, Khoury J, Moomaw CJ, et al. Stroke inci-
Stroke 1993;24:954–964. dence is decreasing in whites but not in blacks: a popula-
19. Amarenco P, Bogousslavsky J, Callahan A III, et al. High- tion-based estimate of temporal trends in stroke incidence
dose atorvastatin after stroke or transient ischemic attack. from the Greater Cincinnati/Northern Kentucky Stroke Study.
N Engl J Med 2006;355:549–559.
Stroke 2010;41:1326–1331.
20. Garcia C, Pinho e Melo T, Rocha L, Lechner MC. Cere-
36. Natarajan S, Lipsitz SR, Nietert PJ. Self-report of high cho-
bral hemorrhage and APOE. J Neurol 1999;246:830–834.
lesterol: determinants of validity in US adults. Am J Prev
21. Viswanathan A, Greenberg S. Cerebral amyloid angiopa-
Med 2002;23:13–21.
thy in the elderly. Ann Neurol 2011;70:871–880.
37. Vargas CM, Burt VL, Gillum RF, Pamuk ER. Validity of self-
22. Peck G, Smeeth L, Whittaker J, Casas JP, Hingorani A,
Sharma P. The genetics of primary haemorrhagic stroke, reported hypertension in the National Health and Nutrition
subarachnoid haemorrhage and ruptured intracranial aneur- Examination Survey III, 1988-1991. Prev Med 1997;26:678–
ysms in adults. PLoS One 2008;3:e3691. 685.
23. O’Donnell HC, Rosand J, Knudsen KA, et al. Apolipopro- 38. Bonevski B, Campbell E, Sanson-Fisher RW. The validity and
tein E genotype and the risk of recurrent lobar intracerebral reliability of an interactive computer tobacco and alcohol use
hemorrhage. N Engl J Med 2000;342:240–245. survey in general practice. Addict Behav 2010;35:492–498.

2282 Neurology 79 December 4, 2012

ª 2012 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

Risk factors for intracerebral hemorrhage differ according to hemorrhage location
Sharyl R. Martini, Matthew L. Flaherty, W. Mark Brown, et al.
Neurology 2012;79;2275-2282 Published Online before print November 21, 2012
DOI 10.1212/WNL.0b013e318276896f

This information is current as of November 21, 2012

Updated Information & including high resolution figures, can be found at:
Services http://www.neurology.org/content/79/23/2275.full.html

Supplementary Material Supplementary material can be found at:

http://www.neurology.org/content/suppl/2012/11/21/WNL.0b013
e318276896f.DC1.html
References This article cites 38 articles, 15 of which you can access for free
at:
http://www.neurology.org/content/79/23/2275.full.html##ref-list-
1
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
All Cerebrovascular disease/Stroke
http://www.neurology.org//cgi/collection/all_cerebrovascular_dis
ease_stroke
Case control studies
http://www.neurology.org//cgi/collection/case_control_studies
Clinical trials Observational study (Cohort, Case control)
http://www.neurology.org//cgi/collection/clinical_trials_observat
ional_study_cohort_case_control
Intracerebral hemorrhage
http://www.neurology.org//cgi/collection/intracerebral_hemorrha
ge
Risk factors in epidemiology
http://www.neurology.org//cgi/collection/risk_factors_in_epidem
iology
Permissions & Licensing Information about reproducing this article in parts (figures,tables)
or in its entirety can be found online at:
http://www.neurology.org/misc/about.xhtml#permissions
Reprints Information about ordering reprints can be found online:
http://www.neurology.org/misc/addir.xhtml#reprintsus