Leukoaraiosis is Associated

with Short- and Long-termMortality in Patients with Intracerebral Hemorrhage

Arnstein Tveiten, MD,* Unn Ljøstad, PhD,* �Ase Mygland, PhD,*†xand Halvor Naess, PhD‡

From the *Departmen

Hospital Kristiansand,

Bergen University Hosp

Medicine, University of B

Received December 20

accepted January 28, 201

Dr. Tveiten is supporte

Address corresponden

Neurology, Sørlandet H

Norway. E-mail: arnstein

1052-3057/$ - see front

� 2013 by National Str

http://dx.doi.org/10.1

Journal of Stroke and C

Background: There are few recent European studies of mortality after intracerebral

hemorrhage (ICH), particularly long-term follow-up studies. No previous European

studies have included information on leukoaraiosis.Methods:We studied all consec-

utive patients hospitalized with a first-ever intracerebral hemorrhage between 2005

and 2009 in awell-defined area and assessed the prognostic value of various baseline

clinical and radiologic factors. Leukoaraiosis was scored on the baseline computed

tomographic (CT) scan as described by van Swieten et al, with an overall score from

0 to 4. Results: One hundred thirty-four patients were followed up for a median of

4.7 years (interquartile range 2.5-6.6). Overall mortality was 23% at 2 days, 30% at

7 days, 37% at 30 days, 46% at 1 year, and 53% at 2 years. Factors independently

associated with increased 30-day mortality were warfarin use, leukoaraiosis

score, intraventricular hemorrhage, and Glasgow Coma Scale (GCS) score. Factors

independently associated with long-term mortality in the 85 patients who survived

the first 30 days were leukoaraiosis score, coronary heart disease, and initial GCS

score. Recurrent ICH occurred in 4.5% and was significantly more frequent after lo-

bar ICH than after ICH in other locations (11.1% v 0%; P5 .025). Conclusions: In un-

selected patients in Southern Norway with first-ever ICH, severe leukoaraiosis is

independently associated with both 30-day and long-term mortality in 30-day sur-

vivors. Warfarin is independently associated with 30-day mortality and coronary

heart disease with long-term mortality in 30-day survivors. Recurrent ICH is more

frequent after lobar ICH than after ICH in other locations. Key Words: Coronary

heart disease—diabetes mellitus—fatality—hemorrhagic stroke—intracerebral

hemorrhage—leukoaraiosis—mortality—prognosis—prognostic factors—stroke—

warfarin—white matter changes.

� 2013 by National Stroke Association

Ten percent to 15% of all strokes are caused by intracere-

bral hemorrhage (ICH).1 Thirty-day mortality in ICH is

higher than in ischemic stroke (40-50% in most studies).2

ts of Neurology; †Habilitation, Sørlandet

Kristiansand; ‡Department of Neurology,

ital, Bergen; and xDepartment of Clinical

ergen, Bergen, Norway.

, 2012; revision received January 25, 2013;

3.

d by the Norwegian Health Association.

ce to Arnstein Tveiten, MD, Department of

ospital Kristiansand, 4604 Kristiansand,

.tveiten@sshf.no.

matter

oke Association

016/j.jstrokecerebrovasdis.2013.01.017

erebrovascular Diseases, Vol. 22, No. 7 (Octob

In a recent review, the median 30-day mortality was 40.4%

(range 13.1-61.0%) and the median 1-year mortality was

54.7% (range 46.0-63.6%).3 Factors found to be associated

with 30-day mortality include age, Glasgow Coma Scale

(GCS) score, ICH volume, ICH location, the presence of

intraventricular hemorrhage (IVH), and oral anticoagula-

tion.4-11 Studies on factors associated with long-term

survival are fewer, particularly follow-up beyond 1 year.

Factors found to be associated with long-term mortality in-

clude age, sex, diabetes mellitus, anticoagulation, heart dis-

ease, and ICH location and volume.5,8,12,13 Demographics

and the distribution of risk factors vary between regions

and with time. The proportion of warfarin-associated

ICH has been increasing, and this has been linked to the in-

creaseduseofwarfarin in atrial fibrillation in the elderly.14-16

er), 2013: pp 919-925 919

A. TVEITEN ET AL.920

Leukoaraiosis is a common finding in neurologic im-

ages of stroke patients. It is a feature of cerebral small ves-

sel pathologies, including hypertensive arteriopathy,

amyloid angiopathy, and cerebral autosomal dominant

arteriopathy with subcortical infarcts and leukoencephal-

opathy. There is increasing evidence of an association

between severe leukoaraiosis and outcome in ischemic

stroke.17-19 For ICH, we have found only 2 studies

reporting an association between leukoaraiosis and

outcome, both from South Korea.20,21 There are no

European studies including leukoaraiosis in the

prediction of outcome in ICH.

Recurrent ICH has been estimated to be 4% to 7% per

year. Data on the distribution of ischemic strokes com-

pared to recurrent ICHs are scarce and inconsistent. In

a review, only 69 classifiable recurrent strokes were avail-

able for assessment. The authors concluded that the re-

view revealed the limitations of the available data and

implied the need for more research.22

The aim of this study was to assess the factors associ-

ated with both 30-day and long-term mortality in unse-

lected patients hospitalized with first-ever ICH in

Southern Norway. We hypothesized that leukoaraiosis

was associated with increased mortality. An additional

aim of the study was to assess the rate of ICH recurrence.

Methods

Patients

The study includes all consecutive patients hospital-

ized with a first-ever ICH between 2005 and 2009 at

Sørlandet Hospital Kristiansand in Kristiansand, Norway.

The hospital serves a well-defined catchment area with

152,000 inhabitants. There is a low-threshold policy of

admitting all patients with suspected stroke to the neuro-

logic department regardless of age and stroke severity.

There is no neurosurgical department. All patients with

suspected stroke are examined in the emergency room

by the neurology staff. In select cases when neurosurgical

treatment is considered, the patient is transferred to

a neurosurgical department in Oslo. The cohort has

been presented in detail elsewhere.16 We excluded trau-

matic ICH, ICH related to intracranial malignant tumors,

ruptured aneurysms, or thrombolytic treatment, and

cases with isolated intraventricular hemorrhage without

visible affection of the cerebral parenchyma.

Clinical and Radiologic Data

Risk factors and clinical and radiologic data were

recorded in a stroke registry from September 2007 and

retrieved from patient files in cases before that time. We

recorded smoking (ever), hypertension (a known

diagnosis), diabetes mellitus (a known diagnosis, either

insulin-dependent or -independent), atrial fibrillation

(AF; documented before or during the hospital stay),

and coronary heart disease (CHD; defined as having

had an acute myocardial infarct, coronary bypass surgery,

or percutaneous coronary intervention), and the use of

platelet inhibitors and warfarin. We categorized patients

by GCS score into 3 groups (3-4, 5-12, and 14-15) as sug-

gested by Hemphill et al.4 All patients had a noncontrast

computed tomographic (CT) scan at baseline. One of the

authors (A.T.) assessed all CT images. Hematoma location

was classified as lobar (predominantly cortical or subcor-

tical white matter), deep cerebral, brainstem, or cerebel-

lum. Thalamic hemorrhages were allocated to the deep

cerebral category. The brainstem category included pon-

tine and mesencephalic hemorrhages. Hematoma volume

was calculated with the A3 B3C/2 formula.23 The pres-

ence of intraventricular blood was registered. Leukoar-

aiosis was scored on the baseline CT scan. We used the

leukoaraiosis grading described by van Swieten et al.24

The anterior and posterior regions were examined sepa-

rately. Leukoaraiosis was distinguished from infarction

by its poorly defined borders. The severity of hypoden-

sity, if present, was expressed in one of two degrees for

each of the 2 regions. In grade 1, the abnormality was re-

stricted to the region adjoining the ventricles. In grade 2,

the increased hypodensity involved the entire region

from lateral ventricle to the cortex. The scores of the 2 re-

gions were added giving an overall score from 0 to 4.24

Outcomes

Death was automatically updated monthly through

a link to the National Population Register.25 The follow-

up period was closed on December 31, 2011. Patients

not registered as dead by this date were considered alive.

Causes of death, updated to December 31, 2011 were

obtained from the cause of death registry/Statistics Nor-

way.26 In addition, patient files were reviewed for clinical

and radiologic information on cause of death. In all pa-

tients who were discharged alive, we checked for recur-

rence of ICH by combining information from the cause

of death registry and from the review of patient files for

information on readmission, neuroimaging, or outpatient

contact.

Statistical Analysis

Descriptive statistics were used to summarize the num-

ber of patients and baseline characteristics using the Pear-

son Chi-square or Fisher exact tests as appropriate for

comparisons of categorical data, the t test for continuous

parametric variables, and the Mann–Whitney U test for

nonparametric variables. Correlations were tested using

the Pearson correlation. We used logistic regression to

assess associations between various baseline factors and

30-day mortality. For long-term mortality, we used Cox

proportional hazard regression and included only

patients who had survived 30 days. Both short- and

long-term mortality factors were first entered one by

LEUKOARAIOSIS AND ICH 921

one in univariate analyses. We checked for collinearity.

We then performed multivariate analyses using the for-

ward stepwise method, including age, sex, and factors

with P values,.15 in univariate analyses. The leukoaraio-

sis score was used as a continuous variable. P , .05 was

considered statistically significant. Statistical analyses

were performed with SPSS software (version 18; SPSS

Inc., Chicago, IL).

Ethics Approval

The study was approved by the regional committee of

medical research ethics.

Results

In total, 134 patients with a first-ever ICH were fol-

lowed up for a median of 4.7 years (interquartile range

2.5-6.6 years), for a total of 630 patient-years. No patients

were lost to follow-up. Baseline characteristics are shown

in total and for each sex in Table 1. There were some sex

differences in baseline characteristics. Men were signifi-

cantly younger, were more often smokers, and had some-

what larger hematomas than women. White patients

accounted for 131 of 134 (97.8%) cases. Five patients

(3.8%) were treated surgically. Twenty-one (15.6%) pa-

tients had a reversal of anticoagulant treatment with co-

agulation factor concentrate (15 patients) or fresh frozen

plasma (6 patients). Thirty patients had intravenous

blood pressure–lowering treatment.

Overall mortality at 2 days was 23.1%, at 7 days 29.9%,

at 30 days 36.6%, at 1 year 46.3%, and at 2 years 53.0%. At

the end of follow-up, 84 of 134 (63%) patients were dead.

In Table 2, the 30-day mortality is presented listed by

ICH location and ICH volumes .30 mL or ,30 mL.

30-day Mortality

All baseline factors listed in Table 1 were tested in uni-

variate analysis for association with 30-day mortality. The

following factors met the criteria for additional analysis in

multivariate analysis (P values): age (.118), sex (.290),

diabetes mellitus (.025), coronary heart disease (.069),

warfarin (.001), atrial fibrillation (.013), GCS score (.001),

ICH volume (.001), brainstem location (.049), intraventric-

ular hemorrhage (.001), and leukoaraiosis score (.008).

The results of the multivariate analysis are shown in

Table 3. Significant independent factors were: warfarin,

GCS score, intraventricular hemorrhage, and leukoaraio-

sis score. For leukoaraiosis, the odds ratio for 30-day

mortality was 1.6 for each increasing point in the leukoar-

aiosis score.

The 2 independent variables that had the highest corre-

lation were ICH volume and GCS score (Pearson correla-

tion 0.569; P, .001). We also ran the multivariate analysis

without inclusion of the GCS score. ICH volume and

brainstem location were then significant factors in addi-

tion to the otherwise unaltered warfarin, intraventricular

hemorrhage, and leukoaraiosis score.

In 16 cases, all of which were fatal within 30 days, the

baseline CT scan was uninterpretable with respect to the

leukoaraiosis score. Therefore, these cases were excluded

in the multivariate analysis of 30-day mortality shown in

Table 3. We performed the same multivariate analysis

without inclusion of leukoaraiosis score as a variable,

including 133 of 134 patients. The remaining significant

factors (warfarin, GCS score, and intraventricular hemor-

rhage) were unaltered; the 16 patients with uninterpret-

able leukoaraiosis scores had significantly greater ICH

volumes than the 118 with interpretable scores (median

125 v 15 mL; P , .001). There was no difference in age

(75.9 v 75.1 years; P 5 .794).

For warfarin users, the median international ratio (INR)

was 2.6 (interquartile range 2.2-2.9). Only 2 patients

(5.6%) had an INR of .4. Therefore, most cases of

warfarin-associated ICH occurred with therapeutic INR

levels.

Long-term Mortality in 30-day Survivors

There were 85 30-day survivors, 35 of whom died

during the follow-up period. Twenty of 44 (45.5%) men

and 15 of 41 (36.6%) women died (P 5.509). All baseline

factors listed in Table 1 were tested in univariate analysis

for association with mortality. The following factors met

the criteria for additional multivariate analysis (P values):

age (.001), sex (.454), coronary heart disease (,.001),

GCS score (.074), ICH volume (.056), and leukoaraiosis

score (.004).

The results of multivariate analysis are shown in

Table 4. The following factors were significantly associ-

ated with mortality after 30 days in multivariate analysis:

CHD, GCS score, and leukoaraiosis score. For leukoaraio-

sis, the odds ratio for long-term mortality was 1.6 for each

increasing point in the leukoaraiosis score. Only 1 patient

with an initial GCS score of 3 to 4 survived 30 days. We

therefore merged the groups GCS 3 to 4 and GCS 5 to 12.

Recurrence

Four of 88 (4.5%) patients who were discharged alive

after the index ICH had an imaging-confirmed recurrent

ICH during follow-up. One was fatal. Recurrent ICH

was more frequent after lobar index ICH (4/36 [11.1%])

than after index ICH in other locations (0/52 [0%]; P 5

.025). Five patients (5.7%) had confirmed ischemic stroke

after the index ICH, none of which were fatal. The index

ICH was lobar in 2, deep cerebral in 1, and cerebellar in 1.

Cause of Death

Causes of death are shown in Table 5 and are listed by

in- or out-of-hospital deaths. In the 11 out-of-hospital

deaths with unspecified stroke as the reported cause,

Table 1. Baseline characteristics in 134 patients with first-ever intracerebral hemorrhage

Total Men Women P value*

N (%) 134 (100) 74 (55) 60 (45)

Risk factors

Age (y), mean (SD) 75.3 (12) 72.9 (12) 78.2 (12) .010

Hypertension, n (%) 72 (55) 41 (57) 31 (52) .532

Diabetes mellitus, n (%) 18 (13) 12 (16) 6 (10) .294

Coronary heart disease, n (%) 20 (15) 15 (20) 5 (8) .054

Atrial fibrillation, n (%) 35 (26) 20 (27) 15 (25) .791

Previous stroke/TIA, n (%) 28 (21) 16 (22) 12 (20) .818

Warfarin, n (%) 36 (27) 23 (31) 13 (22) .221

Platelet inhibitor, n (%) 48 (36) 28 (38) 20 (33) .598

Smoking ever, n (%) 51 (41) 38 (56) 13 (23) ,.001

Glasgow Coma Scale score, n (%) .425

13-15 82 (62) 44 (60) 38 (64)

5-12 29 (22) 15 (20) 14 (24)

3-4 22 (17) 15 (20) 7 (12)

ICH volume (mL), median (IQR) 17.5 (5-44) 21 (9-60) 15 (4-37) .046

Location, n (%) .170

Lobar 49 (37) 23 (31) 26 (43)

Deep 61 (46) 39 (53) 22 (37)

Brain stem 11 (8) 7 (10) 4 (7)

Cerebellum 13 (10) 5 (7) 8 (13)

Intraventricular hemorrhage, n (%) 50 (37) 31 (42) 19 (32) .224

Leukoaraiosis score, n (%)

0 15 (13) 9 (15) 6 (11)

1 25 (21) 15 (24) 10 (18)

2 22 (19) 9 (15) 13 (23)

3 7 (6) 4 (7) 3 (5)

4 49 (42) 25 (40) 24 (43)

Total score, median (IQR) 2 (1-4) 2 (1-4) 2 (1-4) .500

Abbreviations: IQR, interquartile range; SD, standard deviation; TIA, transient ischemic attack.

Missing data (no. of cases): Hypertension (2), smoking (10), GCS score (1), leukoaraiosis score (16).

*Men versus women.

A. TVEITEN ET AL.922

none had imaging confirmation, and it is unclear whether

the reported cause refers to a clinically diagnosed recur-

rent stroke or the index ICH. The 6 cases of malignancies

were: colon cancer (n 5 1), malignant melanoma (n 5 1),

bronchial cancer (n5 1), glottis cancer (n5 1), andmyelo-

matosis (n 5 1). The causes merged as ‘‘other’’ were:

Alzheimer disease (n 5 1) paralytic ileus (n 5 1), urinary

tract infection (n 5 1), depressive episode (n 5 1), hema-

temesis (n 5 1), chronic obstructive pulmonary disease

Table 2. Thirty-day mortality in relation to intra

Location

All volumes ,30

n/dead % n/dead

Lobar 49/15 31 25/3

Deep 61/24 39 40/7

Brainstem 11/7 66 8/4

Cerebellum 13/3 23 11/1

All locations 134/49 37 84/15

*Volumes ,30 mL versus .30 mL.

(n 5 2), ruptured aortic aneurysm (n 5 1), systemic lupus

erythematosus (n5 1), and type 2 diabetesmellitus (n5 1).

Discussion

This study shows that leukoaraiosis is associated with

both 30-day and long-term mortality in patients with

first-ever ICH. It also shows that CHD is associated

with increased long-term mortality.

cerebral hemorrhage location and volume

mL .30 mL

P value*(%) n/dead (%)

12 24/12 50 .005

18 21/17 81 ,.001

50 3/3 100 .236

9 2/2 100 .038

18 50/34 68 ,.001

Table 3. Multivariate analysis of factors associated with

30-day mortality*

Factors OR (95% CI) P value

Warfarin 4.4 (1.2-15.5) .022

Leukoaraiosis score 1.6 (1.06-2.5) .026

GCS score

13-15 1 —

5-12 9.6 (2.8-33.3) ,.001

3-4 102.6 (8.9-1183.2) ,.001

Intraventricular hemorrhage 5.7 (1.5-20.4) .008

Abbreviations: CI, confidence interval; GCS, Glasgow Coma

Scale; OR, odds ratio.

No. of cases included: 117 of 134 (87.3%).

*Using forward stepwise logistic regression.

Table 5. Causes of death for in-hospital and out-of-hospital

deaths

Cause, n (%)

In-hospital

(N 5 59)

Out-of-hospital

(N 5 25)

Index ICH 37 (63) 3 (12)

Recurrent ICH 1 (2) 0 (0)

Cerebral infarct 0 (0) 0 (0)

Unspecified stroke 0 (0) 12 (48)

Cardiac disease 9 (15) 3 (12)

Pneumonia 2 (3) 1 (4)

Malignancy 5 (9) 1 (4)

Other causes 5 (9) 5 (20)

Abbreviation: ICH, intracerebral hemorrhage.

LEUKOARAIOSIS AND ICH 923

The overall mortality at standardized time points is in

general agreement with previous reports.5,8,13,27-29 With

volumes exceeding 30 mL, all hematoma locations other

than lobar carried a high 30-day mortality rate (Table 2).

Smaller cerebellar hematomas had the lowest mortality

rates. This is in line with other studies.5,8,9 The mean

age of 75 years is relatively high, and the proportion of

patients using warfarin (26.9%) is higher than in most

reports.

30-day Mortality

To our knowledge, our study is the first from Europe to

show an association between leukoaraiosis and mortality

after ICH, and the first based on an unselected cohort re-

garding age and severity. We have only found 2 studies

before ours that explored this association, both from

South Korea. In one single-center study, there was an in-

creased 90-day poor outcome (dependency or death, mor-

tality not specified). In a nationwide multicenter study of

1321 patients, leukoaraiosis was associated with in-

creased early and long-term mortality. Both study popu-

lations differed substantially from ours, with a low

mean age of 60 years in both studies, compared to 75

Table 4. Multivariate analysis of factors associated with

long-term mortality in 30-day survivors*

Factors HR (95% CI) P value

Coronary heart disease 5.7 (2.5-13.2) ,.001

GCS score

13-15 — —

3-12y 3.5 (1.4-8.8) .008

Leukoaraiosis score 1.6 (1.2-2.1) .001

Abbreviations: CI, confidence interval; HR, hazard ratio.

No. of cases included: 85 of 85 (100%).

*Cox regression, forward stepwise.

yGCS scores 3 to 4 and 5-12 were merged because there was only

1 case in the GCS 3 to 4 group.

years in our study, and a large proportion were treated

surgically (32% and 35%, respectively) compared to

3.8% in our study. In the first study, patients with antico-

agulant treatment were excluded; in the other, there was

no information on anticoagulation.20,21 In previous

studies, leukoaraiosis has been shown to be an

independent risk factor for warfarin-associated ICH and

for ICH after thrombolytic treatment for acute ischemic

stroke.30,31 A recent study of 79 patients in the United

States revealed an association between leukoaraiosis

and greater ICH volumes and a trend toward more

hematoma growth.32

Our findings regarding warfarin and 30-day mortality

are in agreement with previous studies. Warfarin was

a significant independent factor, as seen in several stud-

ies,15,33-35 and most cases of warfarin-associated ICH oc-

curred with therapeutic INR levels, which has also been

shown by others.14

The increasing proportion of warfarin-associated ICH

linked to increasing use of warfarin in AF14-16 and the

poor outcome are highly relevant issues in the current

ongoing shift toward the increased use of new oral

anticoagulants. Three new oral anticoagulants have

shown a reduced risk of ICH in head to head

comparison with warfarin in patients with nonvalvular

AF.36-38

GCS score, ICH volume, and the presence of intraven-

tricular hemorrhage have been identified as predictors

of a high early mortality in several studies.11,39-41 In our

study, ICH volume was not a significant independent

factor in multivariate analysis when the GCS score was

in the model, but was highly significant when the GCS

score was removed from the model. This probably

reflects an underlying impact of ICH volume on the

GCS score. We view ICH volume as the more causative

of the two. In our study, increasing ICH volume

correlated significantly with a lower GCS score. In

a community-based study in Texas, increased ICH

volume was identified as an independent predictor of

a lower GCS score.42

A. TVEITEN ET AL.924

Long-term Mortality in 30-day Survivors

Leukoaraiosis was independently associated with in-

creased long-term mortality in multivariate analysis.

This is in agreement with the only existing previous

study, which was conducted with nonwhite patients.21

CHDwas not significantly associated with 30-day mortal-

ity, but it was independently associated with increased

long-term mortality. Two Swedish studies found that

CHD was associated with increased 30-day and long-

term mortality (1 year in 1 study and 3 years in the other).

However, for long-term mortality, CHD was not an inde-

pendent factor in multivariate analysis in either of the

studies.8,13 In a study from Finland, CHD was an

independent predictor of 3-month mortality.43 Our study

adds to the existing knowledge by showing an indepen-

dent association between CHD and long-term mortality

in 30-day survivors. Some previous studies of predictors

of long-term mortality included all patients,5,8 while

others included only 30-day survivors.12

Warfarin was not associated with long-term mortality.

A similar finding was reported in a 1-year follow-up

study in Finland, where Kaplan–Meier survival curves

for warfarin users and nonusers diverged over the first

days but appeared to parallel each other thereafter.15 In

a large study from the United States, anticoagulant ther-

apy was an independent predictor of long-term mortality,

but in this analysis, all cases were included, not just 30-

day survivors, and therefore excess early mortality in

warfarin users may have influenced the findings.5

In our study, as in several others, sex was not associated

with long-term mortality.5,6,10 Some studies have shown

higher long-term mortality in men versus women.8,12 In

Sweden, one study reached the opposite conclusion.13

Recurrence

Our finding of recurrent ICH in 4.5% of the population

during the follow-up period may be low compared with

previous estimates of 4% to 7% per year,22,44,45 but

overall numbers are small. Recurrent ICH was more

frequent after lobar ICH than ICH in other locations.

This is in agreement with findings in a review and in

a cohort study in the United States, where both reported

a higher risk of recurrent ICH after lobar ICH.22,45 In

Southern Sweden, no difference in recurrence risk was

found between lobar and deep cerebral hemorrhage.8

We find it reasonable to speculate that our findings may

be linked to underlying amyloid angiopathy. Published

data on the proportional distribution of ischemic and

hemorrhagic recurrent events are scarce and inconsistent.

In a review, there was an overweight of recurrent ICH in 7

hospital-based studies but not clear in 3 population-based

studies.22 In Southern Sweden, similar numbers of hem-

orrhagic and ischemic recurrences were found.8 In our

study, numbers were very small, but we found a similar

distribution.

Strengths and Limitations

In our opinion, the major strength of this study is the un-

selected cohort. By using all identified cases of first-ever

ICHs within a defined area, we believe the cohort is repre-

sentative of current demographic and risk factor distribu-

tion in Southern Norway. The study also has some

limitations. Overall, the numbers are relatively small, espe-

cially in somesubgroups.Thismustbekept inmindand the

results interpreted with caution. For leukoaraiosis, CT im-

ages were uninterpretable in some cases. However, we be-

lieve that other researchers assessing leukoaraiosis in acute

ICH will be confronted with the same limitation in some

cases—particularly in patients with large hematomas. A

possible limitation is the use of the forward stepwise

method,whichcreatesmodels by statistical selectionofvar-

iables. It is, however, among themostwidely usedmethods

inmultivariate analysis. In the current study, the candidate

factorswere all chosen as clinically relevant and commonly

used in similar studies. We got the same results with the

backwardstepwisemethod,andwhenweremoved the leu-

koaraiosis score from the model, the remaining significant

factorswere the same.Although the national death register

wasused,uncertainty remains about the cause of someout-

of-hospital deaths. It has been argued in bothDenmark and

Norway that the quality of the cause of death registers is

poor for research purposes.46,47

In conclusion, this study shows that in unselected pa-

tients in Southern Norway with first-ever ICH, leukoar-

aiosis is independently associated with both 30-day and

long-term mortality in those who survive 30 days. CHD

is independently associated with long-term mortality.

This study shows that long-term mortality strongly re-

lates to the burden of vascular disease. Recurrent ICH is

more frequent after lobar than deep cerebral hemorrhage.

Acknowledgment: We thank Siv Pettersen and Karen

Johanne Olsen for their assistance with data collection and

statistician Hugo Pripp, PhD, for assistance with statistical

analysis.

References

1. Thrift AG, Donnan GA, McNeil JJ. Epidemiology of intra-cerebral hemorrhage. Epidemiol Rev 1995;17:361-381.

2. Mathew L, Flahery DW, Broderick JP. The epidemiologyof intracerebral hemorrhage. In: Carhuapoma JR,Mayer SA, Hanley DF, eds. Intracerebral hemorrhage.New York: Cambridge University Press, 2009:1-16.

3. van Asch CJ, Luitse MJ, Rinkel GJ, et al. Incidence, casefatality, and functional outcome of intracerebral haemor-rhage over time, according to age, sex, and ethnic origin:A systematic review and meta-analysis. Lancet Neurol2010;9:167-176.

4. Hemphill JC 3rd, Bonovich DC, Besmertis L, et al. TheICH score: A simple, reliable grading scale for intracere-bral hemorrhage. Stroke 2001;32:891-897.

5. Flaherty ML, Haverbusch M, Sekar P, et al. Long-termmortality after intracerebral hemorrhage. Neurology2006;66:1182-1186.

LEUKOARAIOSIS AND ICH 925

6. Sacco S, Marini C, Toni D, et al. Incidence and 10-year sur-vival of intracerebral hemorrhage in a population-basedregistry. Stroke 2009;40:394-399.

7. Nilsson OG, Lindgren A, Stahl N, et al. Incidence of intra-cerebral and subarachnoid haemorrhage in southernSweden. J Neurol Neurosurg Psychiatry 2000;69:601-607.

8. Zia E, EngstromG, Svensson PJ, et al. Three-year survivaland stroke recurrence rates in patients with primary in-tracerebral hemorrhage. Stroke 2009;40:3567-3573.

9. Inagawa T, Ohbayashi N, Takechi A, et al. Primary intra-cerebral hemorrhage in Izumo City, Japan: Incidencerates and outcome in relation to the site of hemorrhage.Neurosurgery 2003;53:1283-1297.

10. Hosomi N, Naya T, Ohkita H, et al. Predictors of intrace-rebral hemorrhage severity and its outcome in Japanesestroke patients. Cerebrovasc Dis 2009;27:67-74.

11. Broderick JP, Brott TG, Duldner JE, et al. Volume of intra-cerebral hemorrhage. A powerful and easy-to-use predic-tor of 30-day mortality. Stroke 1993;24:987-993.

12. Fogelholm R, Murros K, Rissanen A, et al. Long termsurvival after primary intracerebral haemorrhage: A ret-rospective population based study. J Neurol NeurosurgPsychiatry 2005;76:1534-1538.

13. Nilsson OG, Lindgren A, Brandt L, et al. Prediction ofdeath in patients with primary intracerebral hemorrhage:A prospective study of a defined population. J Neurosurg2002;97:531-536.

14. Flaherty ML, Kissela B, Woo D, et al. The increasing inci-dence of anticoagulant-associated intracerebral hemor-rhage. Neurology 2007;68:116-121.

15. Huhtakangas J,Tetri S, JuvelaS, et al.Effectof increasedwar-farin use onwarfarin-related cerebral hemorrhage: A longi-tudinal population-based study. Stroke 2011;42:2431-2435.

16. Tveiten A, Ljostad U, Mygland A, et al. Intracerebralhemorrhage in Southern Norway—A hospital-based inci-dence study. Eur Neurol 2012;67:240-245.

17. Smith EE. Leukoaraiosis and stroke. Stroke 2010;41(10Suppl):S139-S143.

18. Arsava EM, Rahman R, Rosand J, et al. Severity of leu-koaraiosis correlates with clinical outcome after ischemicstroke. Neurology 2009;72:1403-1410.

19. Oksala NK, Oksala A, Pohjasvaara T, et al. Age relatedwhitematter changespredict strokedeath in long term fol-low-up. J Neurol Neurosurg Psychiatry 2009;80:762-766.

20. Won YS, Chung PW, Kim YB, et al. Leukoaraiosis predictspoor outcome after spontaneous supratentorial intracere-bral hemorrhage. Eur Neurol 2010;64:253-257.

21. Lee SH, Kim BJ, Ryu WS, et al. White matter lesions andpoor outcome after intracerebral hemorrhage: A nation-wide cohort study. Neurology 2010;74:1502-1510.

22. Bailey RD, Hart RG, Benavente O, et al. Recurrent brainhemorrhage is more frequent than ischemic stroke afterintracranial hemorrhage. Neurology 2001;56:773-777.

23. Kothari RU, Brott T, Broderick JP, et al. The ABCs of mea-suring intracerebral hemorrhage volumes. Stroke 1996;27:1304-1305.

24. van Swieten JC, Hijdra A, Koudstaal PJ, et al. Gradingwhite matter lesions on CT and MRI: A simple scale.J Neurol Neurosurg Psychiatry 1990;53:1080-1083.

25. Norway National Population Register web site. Availablefrom: http://www.norway.no/temaside/tema.asp?stikkord594303. Accessed February 9, 2013.

26. Statistics Norway web site. Available from: http://www.ssb.no/en/. Accessed February 9, 2013.

27. Dennis MS, Burn JP, Sandercock PA, et al. Long-term sur-vival after first-ever stroke: The Oxfordshire CommunityStroke Project. Stroke 1993;24:796-800.

28. Heuschmann PU, Di Carlo A, Bejot Y, et al. Incidence ofstroke in Europe at the beginning of the 21st century.Stroke 2009;40:1557-1563.

29. Hardemark HG, Wesslen N, Persson L. Influence of clin-ical factors, CT findings and early management on out-come in supratentorial intracerebral hemorrhage.Cerebrovasc Dis 1999;9:10-21.

30. Smith EE, Rosand J, Knudsen KA, et al. Leukoaraiosis isassociated with warfarin-related hemorrhage followingischemic stroke. Neurology 2002;59:193-197.

31. Neumann-Haefelin T, Hoelig S, Berkefeld J, et al. Leu-koaraiosis is a risk factor for symptomatic intracerebralhemorrhage after thrombolysis for acute stroke. Stroke2006;37:2463-2466.

32. Lou M, Al-Hazzani A, Goddeau RP Jr, et al. Relationshipbetween white-matter hyperintensities and hematomavolume and growth in patients with intracerebral hemor-rhage. Stroke 2010;41:34-40.

33. Hart RG, Boop BS, Anderson DC. Oral anticoagulantsand intracranial hemorrhage. Facts and hypotheses.Stroke 1995;26:1471-1477.

34. Gage BF, Birman-Deych E, Kerzner R, et al. Incidence ofintracranial hemorrhage in patients with atrial fibrillationwho are prone to fall. Am J Med 2005;118:612-617.

35. Rosand J, Eckman MH, Knudsen KA, et al. The effect ofwarfarin and intensity of anticoagulation on outcome ofintracerebral hemorrhage. Arch Intern Med 2004;164:880-884.

36. Granger CB, Alexander JH, McMurray JJ, et al. Apixabanversus warfarin in patients with atrial fibrillation. N EnglJ Med 2011;365:981-992.

37. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatranversus warfarin in patients with atrial fibrillation. NEngl J Med 2009;361:1139-1151.

38. Patel MR,Mahaffey KW, Garg J, et al. Rivaroxaban versuswarfarin in nonvalvular atrial fibrillation. N Engl J Med2011;365:883-891.

39. Qureshi AI, Safdar K, Weil J, et al. Predictors of earlydeterioration and mortality in black Americans withspontaneous intracerebral hemorrhage. Stroke 1995;26:1764-1767.

40. Lisk DR, Pasteur W, Rhoades H, et al. Early presentationof hemispheric intracerebral hemorrhage: Prediction ofoutcome and guidelines for treatment allocation. Neurol-ogy 1994;44:133-139.

41. Tuhrim S, Horowitz DR, Sacher M, et al. Validation andcomparison of models predicting survival following in-tracerebral hemorrhage. Crit Care Med 1995;23:950-954.

42. Zahuranec DB, Gonzales NR, Brown DL, et al. Presenta-tion of intracerebral haemorrhage in a community. J Neu-rol Neurosurg Psychiatry 2006;77:340-344.

43. Tetri S, Mantymaki L, Juvela S, et al. Impact of ischemicheart disease and atrial fibrillation on survival after spon-taneous intracerebral hemorrhage. J Neurosurg 2008;108:1172-1177.

44. Dennis MS. Outcome after brain haemorrhage. Cerebro-vasc Dis 2003;16(Suppl 1):9-13.

45. Viswanathan A, Rakich SM, Engel C, et al. Antiplateletuse after intracerebral hemorrhage. Neurology 2006;66:206-209.

46. Reseland S. Bad quality of the Cause of Death Registry [inNorwegian]. Tidsskr Nor Laegeforen 2009;129:894.

47. Gjersøe P, Andersen SE, Mølbak AG, et al. Reliability ofdeath certificates. The reproducibility of the recordedcauses of death in patients admitted to departments ofinternal medicine [in Danish]. Ugeskr Laeger 1998;160:5030-5034.