doi:10.1111/j.1468-1331.2012.03782.x
Department of Medicine, Institute of Post Graduate Medical Education and Research, Kolkata; bDepartment of Medicine, North Bengal
Medical College and Hospital, Darjeeling; and cDepartments of Medicine and Rheumatology and Clinical Immunology, Institute of Post
Graduate Medical Education and Research, Kolkata, India
Keywords:
alcoholism, early
mortality, early seizures,
haemorrhagic stroke,
India, stroke
Received 29 January 2012
Accepted 4 May 2012
Background and purpose: Stroke-associated early seizures (ES) often complicate the
initial course of acute stroke. This study intended to estimate the rate of and the
predictive factors for ES and the impact of ES on the clinical outcome in patients
with rst-ever acute stroke.
Materials and methods: Consecutive patients with rst-ever acute stroke admitted
in the Department of Medicine from June 2010 to December 2011 were prospectively included. ES were dened as seizures occurring within 7 days from acute
stroke. Patients with history of epilepsy, transient ischaemic attack, subarachnoid
haemorrhage and cerebral venous thrombosis were excluded. Clinical outcomes
were measured under the subheadings of mortality and disability at discharge,
according to modied Rankin score.
Results: Of the 441 (56.92% male patients, median age 55 years, 49.43% had
haemorrhagic stroke) patients, 79 (17.91%, 95% condence interval (CI): 14.61
21.78%) suered from ES. At discharge, 37.64% were disabled, and 19.5% were
dead. In multivariate analysis, alcoholism, NIHSS at admission, haemorrhagic
stroke and cortical location were signicant predictors of ES. Thirty-day mortality
was predicted by NIHSS at admission [hazard ratio (HR): 1.14, 95% CI: 1.111.18,
P < 0.001], history of hypertension (HR: 3.79, 95% CI: 2.16.85, P < 0.001), history
of alcoholism (HR: 2.43, 95% CI: 1.493.95, P < 0.001) and early seizure (HR:
2.58, 95% CI: 1.544.34, P = 0.001).
Conclusions: Early seizures occurred in about 18% acute stroke patients. Alcoholism, haemorrhagic stroke, cortical and severe strokes predict development of ES. ES
are an independent important risk factor for early mortality.
Introduction
After ischaemic heart disease, stroke is the commonest
cause of mortality worldwide [1], causing nearly 5.7
million deaths in 2005 [2]. More than 80% of these
occurred in low- and middle-income countries [3].
Further, there may be 62 million survivors of stroke
worldwide, with up to a third living with severe disability [4].
Stroke, both ischaemic and haemorrhagic, has been
considered an important cause of epilepsy, especially
in the elderly [5,6].
Correspondence: R. P. Goswami, Abhyudoy Housing, Flat 18/14,
ECTP, Ph IV, Type B, EM Bypass, Kolkata, West Bengal
700107, India (tel.: +919433534335; fax: +913324331574;
e-mail: rudra.goswami@gmail.com).
1361
1362
gorical baseline variables were analysed with chisquared test and continuous variables with Students
two-tailed t test. Statistical signicance was set at
P < 0.05. For multivariate analysis, risk factors were
entered in the model using a forward stepwise selection and setting at 0.05, the signicance level for
retention in the model that derived the maximum likelihood estimates of the odds ratio (OR) and the corresponding 95% condence interval for each variable.
For analysing the predictors of severe outcome in
univariate analysis, three groups were created (group
1: mRS = 0, 1, 2; group 2 = mRS 3, 4, 5; group 3 =
death). Continuous variables were analysed using ANOVA and post hoc analysis. Categorical variables were
analysed by the KruskalWallis test. In multivariate
survival analysis, Cox proportional hazard regression
was utilized. All analyses were performed with the
SPSS 16.0 (SPSS Inc., Chicago, IL, USA).
Results
A total of 441 [49.2% had haemorrhagic stroke
(n = 217), 48.8% had ischaemic stroke (n = 215), and
2% had haemorrhagic transformation of ischaemic
stroke (n = 9)] patients were enrolled in the study, of
which 79 (17.91%, 95% condence interval: 14.61
21.78%) patients suered from ES. Amongst the
patients, 56.92% were men. Median age of the population was 55 years (interquartile range 20 years).
Clinical and radiological predictors of ES were
analysed by univariate and multivariate statistics.
Details of univariate analyses are available at Table 1.
Of note, 23.5% [95% condence interval (CI): 18.33%
29.59%] of haemorrhagic stroke patients, 12.09%
(95% CI: 8.39%17.13%) of ischaemic stroke patients
and 22.22% (95% CI: 6.32%54.74%) of ischaemic
stroke patients with haemorrhagic transformation
developed ES (P = 0.008).
In multivariate logistic regression analysis, alcoholism [odds ratio (OR): 4.48, 95% condence interval
(CI): 2.099.63, P < 0.001], NIHSS at admission (OR:
1.06, 95% CI: 1.031.09, P = 0.001), haemorrhagic
stroke (OR: 2.58, 95% CI: 1.424.7, P = 0.002) and
cortical location (OR: 6.98, 95% CI: 3.8712.57,
P < 0.001) were signicant predictors of ES.
Amongst the 441 patients analysed, 189 patients
(42.86%) had mRS < 3, 166 patients (37.64%) had
mRS > 3, and 86 patients (19.5%) were dead. Regarding early outcome of stroke, ANOVA tables were created
between the three outcome groups (group 1: mRS < 3,
group 2: mRS > 3, group 3: death) for continuous variables (Table 2). Age, SBP and NIHSS at admission
values were signicantly dierent amongst the groups.
Post hoc analysis showed signicant age disparity
Variable
Mean age in years
(SD)
Male (n)
Diabetes mellitus (n)
Hypertension (n)
Dyslipidaemia (n)
Ischaemic heart
disease (n)
Alcoholism (n)
Smoking (n)
Mean systolic blood
pressure in mmHg
(SD)
Mean diastolic blood
pressure in mmHg
(SD)
Haemorrhagic
stroke (n)
Ischaemic stroke (n)
Ischaemic stroke with
haemorrhagic
transformation (n)
Cortical lesion (n)
Mean NIHSS score
at admission (SD)
Mean duration of
hospital stay in
days (SD)
Without
convulsion
(n = 362)
54.38 (14.32)
56.91%
33.42%
61.33%
23.48%
22.65%
(206)
(121)
(222)
(85)
(82)
With
convulsion
(n = 79)
55.23 (14.66)
56.06%
43.04%
60.76%
18.99%
24.05%
(45)
(34)
(48)
(15)
(19)
P value
0.6
0.99
0.1
0.9
0.38
0.78
14.92% (54)
25.41% (92)
160.88 (29.54)
34.17% (27)
20.25% (16)
169.34 (36.09)
<0.001
0.3
0.02
101.15 (59.38)
107.01 (65.64)
0.43
45.85% (166)
64.56% (51)
0.002
52.2% (189)
1.9% (7)
32.9% (26)
2.2% (2)
0.0013
0.86
15.19% (55)
13.53 (8.18)
54.43% (43)
18 (9.85)
10.00 (5.49)
11.01 (7.85)
<0.001
<0.001
0.17
Tests of signicance: categorical variables chi-squared test, continuous variables Students two-tailed t test.
amongst groups 1 and 3 (P = 0.03). Mean SBP dierences amongst all the groups were signicant (group 1
versus group 2: P = 0.009, group 2 versus group 3:
P = 0.03, group 3 versus group 1: P < 0.001). Similarly,
NIHSS at admission was also signicantly dierent
amongst all the groups at signicance levels < 0.001.
Whilst analysing categorical variables, Kruskal
Wallis test was used amongst the three outcome
groups (Table 3). Presence of hypertension, alcoholism, ES and haemorrhagic stroke predicted early
disability or death.
1363
Discussion
In the present study, ES occurred in 17.91% of
patients with acute stroke. The principle conclusions
regarding the predictors of ES following stroke were
history of alcoholism, higher NIHSS at admission,
haemorrhagic stroke and cortical involvement. Interestingly, early mortality was strongly predicted by the
development of ES.
Symptomatic seizures have been reported to occur
in 233% of patients with acute stroke [1720]. The
reported wide range of percentages is probably
explained by the analysis of retrospective studies and
dierent window for dening ES (range 130 days)
[12]. Using a ceiling of 2 weeks to dene ES, incidence
has been reported to vary from 2.4% to 14% after
acute stroke in studies from western countries
[9,10,12]. Reported prevalence from Asian countries
varies as follows: 0.5% (Pakistan) [21], 2.5% (China)
[22], 9.41% (Thailand) [23] and 10.03% (India) [13].
No age or gender dierence was observed between
patients with or without ES. Although male sex and
age younger than 65 years have come up in certain
reports as predictive of ES [24,25], most of the earlier
reports do not assert this observation [9]. Consistent
with most reports, conventional vascular risk factors
(hypertension, diabetes mellitus, smoking, hyperlipidaemia or ischaemic heart disease) were not associated
with the development of ES [9,10].
However, alcohol consumption was signicantly
associated with ES. This eect was echoed earlier in
several reports by Leone et al. [26,27] who found that
alcohol consumption was a risk factor for acute symptomatic seizure after traumatic brain injury and
stroke.
Table 2 Demographical and clinical features of patients according to early functional outcome (ANOVA)
Variable
Rankin < 3
(n = 189)
Rankin > 3
(n = 166)
Death
(n = 86)
P value
(combined)
Age in years
Mean systolic blood pressure in mmHg
Mean diastolic blood pressure in mmHg
Mean NIHSS score at admission
Mean duration of hospital stay in days
53.07
154.96
94.56
8.56
8.61
54.51
164.52
106.78
16.7
13.92
57.79
174.63
110.16
22.88
6.42
0.04
<0.001
0.06
<0.001
<0.001
(14.20)
(25.58)
(14.76)
(5.76)
(4.47)
(14.11)
(32.53)
(85.34)
(6.91)
(6.61)
(14.29)
(34.21)
(64.13)
(8.17)
(3.05)
1364
Table 3 Demographical, clinical, biochemical and radiological features of patients according to early functional outcome (KruskalWallis test)
Male
Diabetes mellitus
Hypertension
Dyslipidaemia
Ischaemic heart disease
Alcoholism
Smoking
Percentage of patients with convulsions
Haemorrhagic stroke
Cortical lesions
Modied Rankin
score < 3 (n = 189)
Modied Rankin
score < 3 (n = 166)
Death (n = 86)
v2
P value
57.14% (108)
31.22% (59)
51.85% (98)
22.75% (43)
17.99% (34)
13.23% (25)
23.28% (44)
4.23%
44.44% (84)
16.93% (32)
57.22% (95)
40.36% (67)
60.84% (101)
26.51% (44)
28.92% (48)
17.47% (29)
27.11% (45)
22.89%
46.38% (77)
26.51% (44)
55.81%
33.72%
82.56%
15.12%
22.09%
31.39%
22.09%
38.37%
65.12%
25.58%
0.053
3.33
23.44
4.19
4.96
13.15
1.03
51.33
10.95
5.38
0.97
0.19
0.001
0.12
0.05
0.001
0.59
<0.0001
0.004
0.06
(48)
(29)
(71)
(13)
(19)
(27)
(19)
(56)
(22)
associated with cardioembolic infarction, such as seizures at onset are not specic for cardioembolic stroke
[32]. Also, ES were not a predictive clinical factor of
in-hospital death in a recent study carried out in
patients with cardioembolic infarction [33].
The mechanism of seizure initiation by haemorrhage is not well established. Products of blood
metabolism, like haemosiderin, may cause focal cerebral irritation leading to seizures. During acute ischaemic injury, accumulation of intracellular calcium and
sodium may result in the depolarization of the transmembrane potential and lower the seizure threshold.
Local cerebral glucose utilization (LCGU) studies
demonstrated increased LCGU within hours around
an intracerebral haemorrhage [34]. There is also evidence regarding eect of intracerebral haemorrhage
on glutamate uptake or glutamate receptors. A recent
paper suggests that intracerebral haemorrhage, possibly via thrombin activation of protease-activated
receptors, activates Src gene that phosphorylates
NMDA receptors and other proteins that mediate
injury after ICH [34]. These associated with the eect
of haemosiderin, and the factors like dissection of
blood from subcortical to cortical planes and rebleeding may explain the increased rate of ES in intracerebral haemorrhage. The importance of cortical damage
in seizure initiation, despite earlier negative results
[17], has been emphasized in most current literatures
[9,10,19,28].
Our study showed that ES were signicantly more
frequent in stroke patients with cortical lesions than
in those patients without cortical involvement even
with multivariate analysis.
Lesion size [17] and stroke severity [34] have been
variably related to ES. The prospective Copenhagen
Stroke Study, which dealts with ischaemic stroke only,
found that stroke severity at admission (measured
with the Scandinavian Stroke Scale) was the only factor related to ES [35]. In 2006, Carrera et al. [36]
1365
inuence on both the development of ES and the functional outcome. So our results should be interpreted
with caution and needs to be conrmed in further
studies. We also could not employ EEG; some seizures
could have been missed or some paroxysmal non-epileptic events could have been called seizures. Also single-centre nature of our study might make our results
dicult to interpret in the general population.
Despite this, the relatively large sample size in our
study and the prospective study design are the major
advantages. Also being the rst study from India that
focuses on post-stroke ES and related morbidity and
mortality should raise viable questions for the future
works regarding the need for proper outlook, prevention and management of ES in this setting, thereby
inuencing public health policy in developing nations.
In conclusion, we have demonstrated relatively
higher frequency (17.91%) of post-stroke ES. In
Indian patients, conventional risk factors for ES apply
in similar fashion with the addition of alcoholism. In
this preliminary study on Indian patients, ES have
emerged as an independent important risk factor for
both post-stroke disability and 30-day mortality.
Acknowledgements
The authors sincerely acknowledge the encouragement
and support provided by the Director of Institute of
Post Graduate Medical Education, Kolkata, to continue the present work.
References
1366
7. Davenport RJ, Dennis MS, Wellwood I, et al. Complications after acute stroke. Stroke 1996; 27: 415420.
8. Bladin CF, Alexandrov AV, Bellavance A, et al. Seizures after stroke: a prospective multicenter study. Arch
Neurol 2000; 57: 16171622.
9. Labovitz DL, Hauser WA, Sacco RL. Prevalence and
predictors of early seizures and status epilecticus after
rst stroke. Neurology 2001; 57: 200206.
10. Lamy C, Domigo V, Semah F, et al. Early and late seizures after cryptogenic stroke in young adults. Neurology 2003; 60: 400404.
11. Ferro JM, Pinto F. Post-stroke epilepsy: epidemiology,
pathophysiology and management. Drugs Aging 2004;
21: 639653.
12. Alberti A, Paciaroni M, Caso V, Venti M, Palmerini F,
Agnelli G. Early seizures in patients with acute stroke:
frequency, predictive factors, and eect on clinical outcome. Vasc Health Risk Manag 2008; 4: 715720.
13. Dhanuka AK, Misra UK, Kalita J. Seizures after stroke:
a prospective clinical study. Neurol India 2001; 49: 3336.
14. World Health Organization. Recommendations on
stroke prevention, diagnosis and therapy. Report of the
WHO Task Force on Stroke and other Cerebrovascular
Disorders. Stroke 1989; 20: 14071431.
15. De Hann R, Limburg M, Bossuyt P, et al. The clinical
meaning of Rankin handicap grades after stroke. Stroke
1995; 26: 20272030.
16. International League Against Epilepsy. Commission on
epidemiology and prognosis, guidelines for epidemiologic studies on epilepsy. Epilespia 1993; 34: 592596.
17. Gupta SR, Naheedy MH, Elias D, et al. Post infarction
seizures. A clinical study. Stroke 1988; 19: 14771481.
18. Giroud M, Gras P, Fayolle H, et al. Early seizures after
acute stroke: a study of 1640 cases. Epilepsia 1994; 35:
89598964.
19. Kilpatrick CJ, Davis SM, Hopper JL, et al. Early seizure after acute stroke. Risk of late seizures. Arch Neurol 1992; 49: 509511.
20. Ryvlin P, Montavont A, Nighghossian N. Optimizing
therapy of seizures in stroke patients. Neurology 2006;
67(Suppl. 4): S3S9.
21. Khealani BA, Ali S, Baig SM. Post stroke seizures:
descriptive study from a tertiary care centre in Pakistan.
J Pak Med Assoc 2008; 58: 365368.
22. Lo YK, Yiu CH, Hu HH, Su MS, Laeuchli SC. Frequency and characteristics of early seizures in Chinese
acute stroke. Acta Neurol Scand 1994; 90: 8385.
23. Panitchote A, Tiamkao S. Prevalence of post-stroke seizures in Srinagarind Hospital. J Med Assoc Thai 2010;
93: 10371042.
Copyright of European Journal of Neurology is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.