Predicting 30-Day Case Fatality of Primary Inoperable

Intracerebral Hemorrhage Based on Findings at the

Emergency Department
Panagiotis Zis, MD, MSc, PhD, Leivadeas Pavlos, MD, Dimitrios Michas, MD,
Dimitrios Kravaritis, MD, Panagiotis Angelidakis, MD,
and Antonios Tavernarakis, MD

Background: Early survival of patients with intracerebral hemorrhage (ICH) depends on several factors, including the location and size of the hematoma and the
level of consciousness on admission. The aim of our study was to estimate the
case fatality of primary inoperable ICH 30 days after the event in our hospital and
to identify clinical and laboratory characteristics, recordable at the Emergency
Department (ED), which could predict death at 30 days. Methods: Clinical and laboratory data on all patients with primary ICH admitted to our hospital were retrospectively collected. Results: Between January 2011 and June 2013 191 patients with
primary ICH were admitted to our hospital. The 30-day case fatality rate was estimated to be 31.9%, as 61 patients died within 30 days after the ICH. Five variables
were independently associated with 30-day case fatality: each decreased point at
the Glasgow Coma Scale (GCS) is associated with a 1.3-fold increase in the odds
of death at 30 days; infratentorial location and intraventricular extension are associated with a 5.5-fold and a 4.7-fold increase in the odds of death at 30 days, respectively; each centimeter of the maximum diameter of the hematoma and each point
increase of the international normalized ratio (INR) are associated with a 1.9-fold
and a 3.5-fold increase in the odds of death at 30 days, respectively. Conclusions:
GCS score on admission, infratentorial location of the hematoma, intraventricular
extension of the hematoma, INR on admission, and maximum diameter of the hematoma are the 5 variables that are independently associated with 30-day case fatality
of primary inoperable ICH. EDICH is introduced as a new grading scale, which includes laboratory and clinical findings at the ED and has predicting value of the
30-day case fatality. Key Words: Intracerebral hemorrhagepredicting factors
30-day case fatalitystrokeEDICH.
2014 by National Stroke Association

Introduction
From the Department of Neurology, Evangelismos General Hospital, Athens, Greece.
Received November 28, 2013; revision received January 28, 2014;
accepted February 2, 2014.
Source of finding: none.
Address correspondence to Panagiotis Zis, MD, MSc, PhD, Department of Neurology, Evangelismos General Hospital, 45-47 Ipsilantou
Street, 10676 Athens, Greece. E-mail: takiszis@gmail.com.
1052-3057/$ - see front matter
2014 by National Stroke Association
http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.02.006

Intracerebral hemorrhage (ICH) is the second most

common type of stroke after ischemic strokes. Spontaneous ICH represents 8%-25% of strokes.1,2 The overall
incidence of ICH is estimated to be 24.6 per 100.000
person years.3 Median case fatality at 30 days is 40.4%3;
however, it ranges from 13%4-61%,5 with most deaths
occurring within the first 48 hours.6
Prognostic factors of case fatality at 30 days include the
location and size of the hematoma, level of consciousness,

Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-6

P. ZIS ET AL.

patients age, comorbidities, and preceding anticoagulation and antiplatelet therapy.7-9 The ICH score is a
simple 6-point clinical grading scale that has been devised
to predict mortality after ICH.10 The FUNC score is an
11-point clinical grading scale that identifies patients
with ICH who will attain functional independence and,
thus, can provide guidance in clinical decision-making
and patient selection for clinical trials.11 Both these scales
incorporate severalclinical onlycomponents that
may be independent predictors of outcome.
The aim of our study was to estimate the case fatality of
primary inoperable ICH 30 days after the event in our
hospital and to identify clinical and laboratory characteristics, recordable at the Emergency Department (ED),
which could predict death at 30 days.

Subjects and Methods

Participants
This is a hospital-based retrospective study conducted
at Evangelismos General Hospital, which is the largest
Greek hospital located in Athens, Greece. Our cohort consisted of all patients admitted to the Department of
Neurology between January 2011 and June 2013 with a
primary diagnosis of ICH confirmed with computed tomography scanning of the brain on admission. All patients included in this study had received neurosurgical
evaluation on admission and for none of them intervention was deemed necessary at this point.
Approval was gained from the local Research Ethics
Committee to review the case notes of this group of patients.

Data Recorded
All variables used for outcome model development
were extracted from data available at the time of initial
ICH evaluation in the ED. Demographics included age
and sex. Risk factors included history of hypertension,
diabetes mellitus, smoking, alcohol abuse, anticoagulant
use, and antiplatelet use. Clinical characteristics at the
ED included the Glasgow Coma Scale (GCS) score at presentation to the ED and parameters based on the brain
computed tomography on admission: location (basal
ganglia, lobar, cerebellar, thalamic, brainstem, and multiple) and consequent characterization of infratentorial or
supratentorial origin of the ICH, intraventricular extension, and maximum diameter of the hematoma. Laboratory findings included baseline blood tests: hematocrit,
hemoglobin, platelet count, serum glucose, serum urea,
serum creatinine, serum potassium, serum sodium, and
international normalized ratio (INR).

Statistical Analyses
A database was developed using the Statistical Package
for Social Science (version 16.0 for Mac; SPSS, Chicago, IL).
Frequencies and descriptive statistics were examined for

each variable. Comparisons between patients who died

and patients who survived were made using the Student
t tests for normally distributed continuous data, the
MannWhitney U test for non-normally distributed data,
and the chi-square test for categorical data.
When statistically significant differences were found,
correlations were carried out to identify variables that
could be entered into a logistic regression model to identify determinants of 30-day case fatality of ICH. When
variables were correlated, the most clinically relevant variable was selected for inclusion in the model. The variables were entered as independent variables and death
at 30 days was entered as the dependent variable.
An outcome risk stratification scale (the EDICH
grading scale) was developed based on the parameters
associated with 30-day case fatality in the logistic regression model, with weighting based on the strength of independent association of each variable. Receiver operating
characteristic (ROC) analysis was calculated to assess
the utility of the EDICH score to predict death within
30 days. The area under the curve and its 95% confidence
intervals for the ROC curve were calculated.
A value of P less than .05 was considered to be statistically significant.

Results
Study Population and 30-Day Case Fatality
Our study population included 191 patients with ICH,
of which 78 were female (40.8%) and 113 (59.2%) were
male. Mean age was 70.0 6 13.2 years. The most frequent
location was lobar (39.8%), followed by basal ganglia
(37.2%), when the least frequent location was cerebellar
(4.2%).
The 30-day case fatality rate was estimated to be 31.9%,
as 61 patients died within 30 days after the ICH. Both patients who underwent neurosurgical intervention did not
survive. For none of the patients of the total population
withdrawal of care was decided at any point of their
care and none of the patients, or next of kin when appropriate, had requested a do not resuscitate.

Univariate Analysis
Table 1 summarizes the demographic characteristics,
the risk factors, and the clinical characteristics, all recordable at the ED, for the cohort in total and the 2 groups (patients who survived/patients who died) separately.
There were no statistically significant differences
regarding the age and the gender between the 2 groups.
However, patients who died were using more frequently
anticoagulants (23.0% versus 11.5%, P 5.040) and antiplatelets (27.9% versus 10.8%, P 5 .003). Regarding the clinical characteristics, patients who died had a decreased
level of consciousness, as measured by the GCS score, at
admission (9.2 versus 14.0, P , .001). Moreover, patients

PREDICTING ICH 30-DAY MORTALITY BASED ON ED FINDINGS

Table 1. Demographics, risk factors, clinical and laboratory findings at the ED

Demographic characteristics
Female sex (%)
Age, y (SD)
Risk factors
Hypertension (%)
Diabetes mellitus (%)
Smoking history (%)
Alcohol consumption (%)
Anticoagulant use (%)
Antiplatelet use (%)
Clinical characteristics at ED
Glasgow Coma Scale score (SD)
Location
Basal ganglia (%)
Lobar (%)
Cerebellar (%)
Thalamic (%)
Brainstem (%)
Multiple (%)
Infratentorial location (SD)
Intraventricular extension (SD)
Maximum diameter, cm (SD)
Laboratory findings at ED
Hematocrit, % (SD)
Hemoglobin, g/dL (SD)
Platelets, 1000/mL (SD)
Serum glucose, mg/dL (SD)
Serum creatinine, mg/dL (SD)
Serum urea, mg/dL (SD)
Serum potassium, mmol/L (SD)
Serum sodium, mmol/L (SD)
INR (SD)

Total population
(n 5 191)

Patients who
died (n 5 61)

Patients who survived

(n 5 130)

P value

78 (40.8)
70.0 (13.2)

23 (37.7)
72.2 (12.4)

55 (42.3)
68.9 (13.5)

.546
.103

119 (62.3)
59 (30.9)
44 (23.0)
29 (15.2)
29 (15.2)
31 (16.2)

33 (54.1)
15 (24.6)
12 (19.7)
8 (13.1)
14 (23.0)
17 (27.9)

86 (66.2)
44 (33.8)
32 (24.6)
21 (16.2)
15 (11.5)
14 (10.8)

12.5 (3.7)

9.2 (4.6)

14.0 (1.9)

.109
.197
.449
.585
.040*
.003y
,.001z

71 (37.2)
76 (39.8)
8 (4.2)
13 (6.8)
13 (6.8)
10 (5.2)
31 (16.2)
80 (41.9)
4.5 (2.2)

14 (23.0)
24 (39.3)
2 (3.3)
2 (3.3)
13 (21.3)
6 (9.8)
21 (34.4)
44 (72.1)
5.9 (2.0)

57 (43.8)
52 (40.0)
6 (4.6)
11 (8.5)
0 (.0)
4 (3.1)
10 (7.7)
36 (27.7)
3.8 (1.9)

,.001z
,.001z
,.001z

40.8 (4.5)
13.8 (1.7)
223.5 (68.7)
141.1 (63.5)
1.1 (1.2)
45.8 (28.9)
4.1 (.5)
139.1 (3.8)
1.2 (.6)

41.3 (4.3)
14.0 (1.7)
223.6 (84.3)
160.5 (65.3)
1.2 (1.4)
49.7 (35.1)
4.0 (.5)
139.3 (3.4)
1.4 (.8)

40.6 (4.6)
13.7 (1.7)
223.5 (60.7)
131.9 (60.8)
1.1 (1.0)
43.9 (25.4)
4.2 (.6)
139.0 (4.0)
1.1 (.4)

.317
.315
.992
.004y
.552
.197
.091
.571
.001y

,.001z

Abbreviations: ED, Emergency Department; INR, international normalized ratio; SD, standard deviation.
Numbers in brackets correspond to SD for continuous and percentage (%) for categorical variables.
*p , .05
yp , .01
zp , .001

who died had a larger hematoma, as this was measured

by its maximum diameter (5.9 versus 3.8 cm, P , .001),
they presented more frequently with infratentorial hematomas (34.4% versus 7.7%, P ,.001), and their hematomas
extended in the ventricles (72.1% versus 27.7%, P , .001)
more frequently compared with the patients who survived.
Regarding the laboratory findings at the ED, patients who
died had higher serum glucose (160.5 versus 131.9 mg/dL,
P 5 .004) and higher INR (1.4 versus 1.1, P 5 .001).

Multivariate Analysis
The following independent variables were entered
into the multivariate logistic regression model: anticoagulant use; antiplatelet use; GCS score at admission; infratentorial location; maximum diameter; intraventricular

extension; serum glucose at admission; and INR at

admission. The results of the multivariate logistic regression are shown in Table 2. Adjusted odds ratios are presented. The full model significantly predicted death at
30 days (c2 5 100.64, df 5 8, P ,.001), with most variance
(48.9%-67.9%) being explained by 5 variables: GCS score
at admission; infratentorial location; maximum diameter
of the hematoma; intraventricular extension; and INR.
Each decreased point at GCS is associated with a 31.9%
increase in the odds of death at 30 days; infratentorial
location and intraventricular extension are associated
with a 5.5-fold and a 4.7-fold increase in the odds of
death at 30 days, respectively; each centimeter of the
maximum diameter of the hematoma and each point increase of the INR are associated with a 1.9-fold and a 3.5fold increase in the odds of death at 30 days, respectively.

P. ZIS ET AL.

Table 2. Patient characteristics investigated for their association with 30-day case fatality
Variable

OR (95% CI)

Wald

P value

Glasgow Coma Scale score, per decreased point

Infratentorial location
Maximum diameter, per cm
Intraventricular extension
Serum glucose, per mg/dL
Anticoagulant use
Antiplatelet use
INR, per point

1.319 (1.118-1.557)
6.122 (1.341-27.936)
1.908 (1.343-2.709)
4.690 (1.335-16.468)
.997 (.986-1.008)
2.659 (.347-20.357)
3.759 (.985-14.348)
3.494 (1.022-11.947)

10.716
5.472
13.027
5.816
.246
.887
3.753
3.977

.001*
.019y
,.001z
.016y
.620
.346
.053
.046y

Abbreviations: CI, confidence interval; INR, international normalized ratio; OR, odds ratio.
*p , .01
yp , .05
zp , .001

EDICH Grading Scale

An outcome risk stratification scale (the EDICH
grading scale) was developed from the logistic regression
model for all ICH patients. The 5 characteristics that remained statistically significant predictors of the 30-day
case fatality were each assigned points on the basis of
the strength of association with outcome. The total
EDICH score is the sum of the points of these variables.
Table 3 indicates the specific point assignments used in
calculating the EDICH score. In Figure 1, 4 cases and their
corresponding details are presented as examples.
ROC analysis showed an area under the curve of 92%
(95% confidence interval, 87%-96%; standard error, .023;
P , .001). All patients with a score of 6 or more died
within 30 days. At a cutoff score of 3 or more, the EDICH
showed a sensitivity of 84%, a specificity of 87%, in predicting 30-day case fatality.

tality in patients with supratentorial ICH,12 and it has

been suggested that intensive lowering of glucose level
should be further investigated in ICH patients.13 In our
study, although serum glucose differed significantly between patients who died and patients who survived at
30 days, the multivariate analysis did not confirm this difference.
Moreover, recent evidence highlights the contribution
of antiplatelet therapy to clinical severity and increased
mortality of ICH, but results are discrepant14; some
studies conclude that the previous use of antiplatelet
agents is an independent predictor of 30-day case fatality,12,14 whereas others argue that pretreatment with
antiplatelet agents is not.15 In our cohort, although the

Table 3. EDICH grading scale

Variable

Discussion
In our study population, the 30-day case fatality rate of
ICH was estimated to be 31.9%, which is lower than the
median case fatality reported in a recent meta-analysis.3
Moreover, we showed that the GCS score at admission, infratentorial location of the ICH, maximum diameter of the
hematoma, intraventricular extension of the hemorrhage,
and INR are recordable factors at the ED that are associated with 30-day case fatality.
Despite the fact that van Asch et al3 in their metaanalysis showed that case fatality of ICH increases with
age, in our cohort age did not appear to be predictive of
death, which has also remained after we grouped the patients in age groups of greater or equal to 80 years and
lower than 80 years. One possible explanation is that
our cohort included only Greek patients, whereas the
meta-analysis included patients of different ethnic origins.
In recent studies, it has been suggested that admission
glucose levels are independent predictors of 30-day mor-

EDICH score points

Glasgow Coma Scale score

3-7
8-11
12-15
Maximum diameter of the hematoma
.6 cm
3-6 cm
,3 cm
International normalized ratio
.3
2-3
,2
Intraventricular extension
Yes
No
Supratentorial location
Yes
No
Total EDICH score

2
1
0
2
1
0
2
1
0
1
0
1
0
0-8

Abbreviation: EDICH, Emergency Department intracerebral

hemorrhage.

PREDICTING ICH 30-DAY MORTALITY BASED ON ED FINDINGS

Figure 1. (Case A) 51-year-old male patient

with primary hemorrhage in the basal ganglia
without intraventricular extension. Maximum
diameter of the hematoma 2.4 cm, INR on admission 1.02, GCS score on admission 15. EDICH
score 0. Patient survived. (Case B) 58-year-old
male patient with primary lobar hemorrhage
with intraventricular extension. Maximum
diameter of the hematoma 6.8 cm, INR on admission 5.09, GCS score on admission 15. EDICH
score 5. Patient died on the 11th day. (Case C)
52-year-old male patient with primary hemorrhage in the brainstem without intraventricular
extension. Maximum diameter of the hematoma
2.0 cm, INR on admission .95, GCS score on
admission 10. EDICH score 2. Patient survived.
Abbreviations: EDICH, Emergency Department
intracerebral hemorrhage; GCS, Glasgow Coma
Scale; INR, international normalized ratio.

univariate analysis revealed statistical significant differences regarding the antiplatelet use between patients
who died and patients who survived at 30 days, the
multivariate analysis did not confirm this difference.
It was known that patients taking warfarin had a
doubling in the rate of ICH mortality in a dose-dependent
manner.12 Interestingly, in our study, INR was found to
have a predictive value of the 30-day case fatality, after having adjusted for anticoagulant use. This means that even in
patients who are not on anticoagulant therapy, high INR increases the chances of a poor outcome over 30 days.
The strengths of our study include that we managed to
review a significant number of primary ICHs over a
period of 30 days. Our cohort consisted of patients of
both sexes, of most age groups (range, 36-94 years), of
all possible ICH locations, of all levels of consciousness
on admission (GCS score range, 3-15), and of different
sizes of the hematoma (maximum dimension range, .515.0 cm). Moreover, we had no missing data regarding
the outcome.

Our results should be interpreted with some caution

given the limitations of our design. Because retrospective
studies are susceptible to recall bias, we attempted to
minimize this through the use of case and not data. Our
thorough approach to review all available recorded data
by 2 independent investigators for each case minimized
the amount of missing data. Moreover, as we aimed to
include only recordable factors at the ED, we could not
control other blood tests, such as cholesterol and triglyceride levels, which take in place only in ward after admission, or parameters such as blood pressure variability and
time rate of blood pressure variation over the first
24 hours, which affect the outcome.16 Finally, our cohort
comprised patients of one hospital, and results may not
be generalizable to other settings.
Clinical grading scales play an important role in the
evaluation and management of patients with acute neurologic disorders.10 The ICH score is a well-established
grading scale that predicts accurately the 30-day mortality. The FUNC score is a valid clinical assessment tool

P. ZIS ET AL.

that identifies patients with ICH who will attain functional independence.11
EDICH could be a new grading scale, which incorporates laboratory findings at the ED and also predicts value
of the 30-day mortality. However, our results were only
based on the population of a single center, a multicenter
replication prospective validation study is advised to
confirm our findings and confirm the psychometric properties of EDICH.
Acknowledgment:

The authors have no conflict of interest.

References
1. Broderick J, Connolly S, Feldmann E, et al, American
Heart Association; American Stroke Association Stroke
Council; High Blood Pressure Research Council; Quality
of Care and Outcomes in Research Interdisciplinary
Working Group. Guidelines for the management of spontaneous intracerebral hemorrhage in adults: 2007 update:
a guideline from the American Heart Association/American Stroke Association Stroke Council, High Blood Pressure Research Council, and the Quality of Care and
Outcomes in Research Interdisciplinary Working Group.
Stroke 2007;38:2001-2023.
2. Kimura Y, Takishita S, Muratani H, et al. Demographic
study of first-ever stroke and acute myocardial infarction
in Okinawa, Japan. Intern Med 1998;37:736-745.
3. van Asch CJ, Luitse MJ, Rinkel GJ, et al. Incidence, case
fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin:
a systematic review and meta-analysis. Lancet Neurol
2010;9:167-176.
4. Inagawa T, Ohbayashi N, Takechi A, et al. Primary intracerebral hemorrhage in Izumo City, Japan: incidence
rates and outcome in relation to the site of hemorrhage.
Neurosurgery 2003;53:1283-1297.
5. Corbin DO, Poddar V, Hennis A, et al. Incidence and case
fatality rates of first-ever stroke in a black Caribbean population: the Barbados register of strokes. Stroke 2004;
35:1254-1258.

6. Broderick JP, Brott TG, Duldner JE, et al. Volume of intracerebral hemorrhage. A powerful and easy-to-use predictor of 30-day mortality. Stroke 1993;24:987-993.
7. Franke CL, van Swieten JC, Algra A, et al. Prognostic factors in patients with intracerebral haematoma. J Neurol
Neurosurg Psychiatry 1992;55:653-657.
8. Fogelholm R, Murros K, Rissanen A, et al. Long term survival after primary intracerebral haemorrhage: a retrospective population based study. J Neurol Neurosurg
Psychiatry 2005;76:1534-1538.
9. Thompson BB, Bejot Y, Caso V, et al. Prior antiplatelet
therapy and outcome following intracerebral hemorrhage: a systematic review. Neurology 2010;75:
1333-1342.
10. Hemphill JC 3rd, Bonovich DC, Besmertis L, et al. The
ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-897.
11. Rost NS, Smith EE, Chang Y, et al. Prediction of
functional outcome in patients with primary intracerebral hemorrhage: the FUNC score. Stroke 2008;
39:2304-2309.
12. Roquer J, Rodrguez Campello A, Gomis M, et al.
Previous antiplatelet therapy is an independent predictor of 30-day mortality after spontaneous supratentorial intracerebral hemorrhage. J Neurol 2005;
252:412-416.
13. Lee SH, Kim BJ, Bae HJ, et al. Effects of glucose level on
early and long-term mortality after intracerebral haemorrhage: the Acute Brain Bleeding Analysis Study. Diabetologia 2010;53:429-434.
14. Lacut K, Le Gal G, Seizeur R, et al. Antiplatelet drug use
preceding the onset of intracerebral hemorrhage is associated with increased mortality. Fundam Clin Pharmacol
2007;21:327-333.
15. Foerch C, Sitzer M, Steinmetz H, et al. Pretreatment with
antiplatelet agents is not independently associated with
unfavorable outcome in intracerebral hemorrhage. Stroke
2006;37:2165-2167.
16. Zis P, Vemmos K, Spengos K, et al. Ambulatory blood
pressure monitoring in acute stroke: pathophysiology of
the time rate of blood pressure variation and association
with the 1-year outcome. Blood Press Monit 2013;
18:94-100.