poststroke epilepsy
TRANSCRIPT
-
8/13/2019 Poststroke Epilepsy
1/5
Poststroke Epilepsy in the Copenhagen Stroke Study:Incidence and Predictors
Lars Peter Kammersgaard, MD, and Tom Skyhj Olsen, DMSci
Poststroke epilepsy (PSE) is a feared complication after stroke and is reported in 3%
to 5% of stroke survivors. In this study we sought to identify incidence and
predictors of PSE in an unselected stroke population with a follow-up period of 7
years. The study was community-based and comprises a cohort of 1197 consecu-
tively and prospectively admitted patients with stroke. Patients were followed up
for 7 years. We defined PSE as recurrent epileptic seizures with onset after stroke
and requiring antiepileptic prophylaxis. PSE was related to clinical factors (age, sex,
onset stroke severity, lesion size on computed tomography scans, stroke subtype,localization, stroke risk factor profile, and early seizures) in univariate analyses.
Independent predictors of PSE were identified through multiple logistic regression
analyses. Overall, 38 patients (3.2%) developed PSE. Univariately, PSE was asso-
ciated with younger age, intracerebral hemorrhage, and larger lesions. PSE was less
frequently associated with atrial fibrillation and ischemic heart disease. In the final
multiple regression model for the dependent variable PSE, independent predictors
were younger age (odds ratio [OR] 1.7/10 years; 95% confidence interval [CI]
1.3-2.1), onset stroke severity (OR 1.3-/10-point decrease; 95% CI 1.0-1.6), lesion
size (OR 1.2-/10-mm enlargement; 95% CI 1.0-1.3), intracerebral hemorrhage (OR
3.3; 95% CI 1.3-8.6), and early seizures (OR 4.5; 95% CI 1.3-16.0). We conclude that
PSE occurs in about 3% of all patients with stroke within 7 years after stroke. Age,
intracerebral hemorrhage, lesion size, increasing stroke severity, and early seizures
are independent predictors of PSE. Key Words: Cerebrovascular diseaselong-
termrisk factorsepilepsystroke severityagecerebral hemorrhagelogis-
tic regression.
2005 by National Stroke Association
Epilepsy is a feared complication to stroke among
patients and their relatives. Hence, identification of fac-
tors that predict epilepsy in patients with stroke is war-
ranted. Age, stroke severity, cortical stroke, and hemor-
rhagic stroke are often found to be predictors in most
studies,1-3 but not in all reports.4 Timing of first seizure
(early v late after stroke)3,5 and stroke recurrence6 are
among other suggested predictive factors for poststroke
epilepsy (PSE).
In this prospective study, which consists of an un-
selected community-based cohort of hospitalized pa-
tients, we followed up 1195 patients with stroke for 7years after onset. The aim of the study was to investigate
the overall risk for PSE in a long-term perspective and to
determine independent predictors for PSE.
Methods
The study included all patients with acute stroke ad-
mitted consecutively during a 25-month period to the
stroke department at Bispebjerg Hospital, Copenhagen,
Denmark. It was a prospective and community-based
From the Stroke Unit, Hvidovre University Hospital, Hvidovre,
Denmark.
Received May 1, 2005; accepted June 28, 2005.
Supported by grants from the Danish Health Foundation, the
Danish Heart Foundation, Ebba Celinders Foundation, the Indepen-
dent Order of Odd Fellows, Scandinavia Lodge 2, and HjerneSagen.
Address correspondence to Lars Peter Kammersgaard, MD, Stroke
Unit, Hvidovre University Hospital, DK-2650 Hvidovre, Denmark.
E-mail: [email protected].
1052-3057/$see front matter
2005 by National Stroke Association
doi:10.1016/j.jstrokecerebrovasdis.2005.07.001
Journal of Stroke and Cerebrovascular Diseases,Vol. 14, No. 5 (September-October), 2005: pp 210-214210
-
8/13/2019 Poststroke Epilepsy
2/5
-
8/13/2019 Poststroke Epilepsy
3/5
were tested backwards to fit the full model with all
explanatory variables. Unimportant variables were then
removed one by one, until all those remaining in the
model each contributed with aPless than .2. To select the
final model, backward logistic regression, including only
variables with aP less than .2, was followed by forward
logistic regression including the same variables. The ex-
planatory power of the equation was determined by the
value of correct classification. The required level of sig-
nificance for all tests was set at .05.
The study was approved by the Ethics Committee of
Frederiksberg and Copenhagen (approval numbers KF/
V.100.2263/91 and KF 01-287/98).
Results
The crude cohort consisted of 1197 patients, but two
patients were excluded because they had epilepsy before
stroke onset. Basic characteristics for the 1195 patients
included are shown inTable 1.PSE emerged in 16 (1.3%)
patients within 3 years, and in 38 patients within 7 years.
At stroke onset patients with subsequent PSE were
younger, more often had hemorrhagic strokes, and more
often had larger lesions. Fewer patients with subsequent
PSE had ischemic heart disease and atrial fibrillation. No
differences between patients with and without PSE and
PSE were found for sex, initial stroke severity, cortical
involvement, hypertension, diabetes, intermittent claudi-
cation, smoking, daily intake of alcohol, previous stroke,
or pre-existing disability. There was a trend toward a
higher frequency of early seizures among patients with
subsequent PSE.Fig 1gives the frequencies of PSE as a function of initial
stroke severity divided into mild (SSS score 45-58), mod-
erate (SSS score 30-44), severe (SSS score 15-29), and very
severe (SSS score 0-14), with frequencies being 2.3%,
3.6%, 4.5%, and 2.7%, respectively.
In the final multiple logistic regression model shown in
Table 2,independent predictors of PSE were hemorrhagic
stroke, early seizure, larger lesion diameter, increased
initial stroke severity, and younger age. Cortical involve-
ment was not found to be an independent predictor of
PSE.
Discussion
In this community-based study 3.2% of all patients
with stroke developed PSE during the follow-up period
of mean 7 years after stroke onset. A few smaller studies
reported frequencies of PSE in as many as 6% to 9%.14,15
In the Oxfordshire Community Stroke Project, PSE devel-
oped in 3.8% within 5 years after onset of first ever
stroke.2 In the Rochester, Minn, study that included 535
patients with ischemic strokes, 3.4% developed PSE
within a follow-up period of 5.5 years after onset of
stroke.6 In the prospective Seizures After Stroke Study
with a 9-month observation period, PSE emerged in 2.5%
of the 1897 patients included.3 In the current study we
found 1.3% to develop PSE within 3 years after onset.
Younger age appeared to predispose to PSE in our
cohort. After adjustment for other basic clinical charac-
teristics a 10-year decrease of age at stroke onset inde-
pendently increased the risk for PSE by 65%. Previous
studies of PSE have focused on other factors and not age
per se. We found only one previous study that had
analyzed age as a risk factor for developing PSE and in
Figure 1. Frequency of poststroke epilepsy as function of initial stroke
severity. Patients were stratified into 4 groups: mild (Scandinavian Stroke
Scale [SSS] 45-58 points), moderate (SSS 30-44 points), severe (SSS 15-29
points), and very severe (SSS 0-14 points) strokes.
Table 2. Independent predictors of poststroke epilepsy
Variable Covariate P OR 95% CI
PSE Cerebral hemorrhage .02 3.3 1.38.6
Lesion size, per 10-mm enlargement .02 1.2 1.01.3
Severity, per 10-point decrease .03 1.3 1.01.6
Age, per 10-year decrease .0001 1.7 1.32.1
Early seizure .02 4.5 1.316.0
Cortical involvement .67 1.2 0.52.7
CI, Confidence interval; OR, odds ratio; PSE, poststroke epilepsy.
Multiple regression analysis (final model): dependent variable PSE overall 2 0.0001, value of correct classification 95.4%.
L.P. KAMMERSGAARD AND T.S. OLSEN212
-
8/13/2019 Poststroke Epilepsy
4/5
that study age was not found to be an independent
predictor of PSE.1 However, in that study the follow-up
period was considerably shorter and the cohort investi-
gated smaller than the current one. We found no differ-
ences in the univariate analyses when stroke severity at
onset between patients with versus without PSE was
compared. As demonstrated inFig 1, the frequencies of
PSE increased with increasing stroke severity. Althoughthere was a decline in frequency for very severe strokes,
we think this was caused by a higher long-term mortality
among patients with the most severe strokes. In line with
this, the frequency of PSE was directly related to lesion
diameter. After adjustment for differences in basic char-
acteristics we found increasing stroke severity to inde-
pendently raise the relative risk for developing PSE by
27% per 10-point decrease of the SSS score. In the Sei-
zures After Stroke Study, patients who had a seizure after
stroke had more severe strokes as measured by the Ca-
nadian Neurological Score, but no relation between PSE
and stroke severity was detected.3 In the Oxfordshire
Community Stroke population, severe stroke was associ-
ated with both single and recurrent seizures, but stroke
severity was not measured by a stroke scale.2 A Norwe-
gian study found an initial stroke severity below 30
points on the SSS score to significantly increase the risk
for PSE within 12 months after stroke onset.1
Lesion size appeared to be a significant independent
predictor for developing PSE. An increase of the lesion
diameter by 10 mm on CT scans corresponded to a 16%
higher relative risk for PSE within 7 years after stroke
onset. In the Seizures After Stroke Study, large lesion
volumes were not found to be independent predictors for
PSE.3
However, in the Oxfordshire Community StrokeProject, only 1% of patients with lacunar strokes devel-
oped PSE, whereas 11% of patients with total anterior
circulation infarcts developed PSE.2 In addition, two
other studies have suggested that, in patients with very
small stroke lesions, development of seizures at all is
extremely rare.16,17
In the current study 21% of those who developed PSE
had cerebral hemorrhages, whereas that was only the
case in 7% of those without PSE. Furthermore, cerebral
hemorrhage was found to be an independent predictor
for PSE and increased the risk more than 3 times relative
to the risk for patients with ischemic strokes. The moresevere strokes that often accompany cerebral hemor-
rhages is not the only explanation for this finding because
we were able to adjust for differences in stroke severity.
Previous studies have reported seizures to be more fre-
quent among patients with cerebral hematomas3,17-19 and
the risk of seizures after cerebral hematomas has been
reported to be as high as 10 relative to ischemic strokes.2
A study that included a large number of patients with
only intracerebral hematomas found that PSE developed
in 2.5%. However, the prognostic significance of cerebral
hemorrhage per se for the development of PSE has not
been investigated previously using multiple regression
techniques.
It has been discussed that the risk for PSE differs
between patients with early- (occurring within 14 days
after stroke onset) versus late-onset seizures. Some au-
thors have suggested that early seizures predict PSE,6,20
others have found this relation for late seizures,21 and no
relation was reported in one study.
22
In our study, pa-tients who subsequently developed PSE had early sei-
zures twice as often as those without PSE. Although this
difference was only of borderline statistical significance,
early seizure was found to increase the relative risk for
PSE more than 4 times after adjustment for other clinical
characteristics.
In this study we found no correlation between cortical
localization of the stroke lesion and subsequent develop-
ment of PSE. Although stroke with cortical involvement
is well known to be associated with an increased risk for
early seizures,23,24 very few data are available for the
relation to the development of PSE. In a study of patients
with ischemic strokes, hemispheric stroke was found to
be associated with increased risk for recurrent seizures.25
In two studies, cortical stroke was not found to be an
independent predictor of PSE,3,18 and one study even
suggested that only patients with lacunar infarctions are
at greater risk for PSE.21
In conclusion, the current study shows that in an
unselected stroke population PSE develops in very few
(3.2%) patients during a period of 7 years after stroke.
The risk for PSE appears to be highest for younger
patients, patients with severe strokes on admission, those
with large lesions on CT scans, those with hemorrhagic
strokes, and when the patient has a seizure within 14days after onset of stroke.
References
1. Lossius MI, Ronning OM, Mowinckel P, et al. Incidenceand predictors for post-stroke epilepsy: A prospectivecontrolled trial; The Akershus stroke study. Eur J Neurol2002;9:365-368.
2. Burn J, Dennis M, Bamford J, et al. Epileptic seizuresafter a first stroke: The Oxfordshire community strokeproject. BMJ 1997;315:1582-1587.
3. Bladin CF, Alexandrov AV, Bellavance A, et al. Seizuresafter stroke: A prospective multicenter study. Arch Neu-rol 2000;57:1617-1622.
4. Kilpatrick CJ, Davis SM, Hopper JL, et al. Early seizuresafter acute stroke: Risk of late seizures. Arch Neurol1992;49:509-511.
5. Sung CY, Chu NS. Epileptic seizures in thromboticstroke. J Neurol 1990;237:166-170.
6. So EL, Annegers JF, Hauser WA, et al. Population-basedstudy of seizure disorders after cerebral infarction. Neu-rology 1996;46:350-355.
7. Jrgensen HS, Nakayama H, Raaschou HO, et al. Effectof blood pressure and diabetes on stroke in progression.Lancet 1994;344:156-159.
POSTSTROKE EPILEPSY 213
-
8/13/2019 Poststroke Epilepsy
5/5
8. Jrgensen HS, Nakayama H, Raaschou HO, et al. Recoveryof walking function in stroke patients: The Copenhagenstroke study. Arch Phys Med Rehabil 1995;76:27-32.
9. Stroke1989. Recommendations on stroke prevention,diagnosis, and therapy: Report of the WHO task force onstroke and other cerebrovascular disorders. Stroke 1989;20:1407-1431.
10. Commission on Classification and Terminology of theInternational League Against Epilepsy. Proposal for clas-
sification of epilepsies and epileptic syndromes. Epilep-sia 1985;26:268-278.
11. Scandinavian Stroke Study Group. Multicenter trial ofhemodilution in ischemic strokebackground and studyprotocol. Stroke 1985;16:885-890.
12. Reith J, Jrgensen HS, Pedersen PM, et al. Body tempera-ture in acute stroke: Relation to stroke severity, infarct size,mortality, and outcome. Lancet 1996;347:422-425.
13. Norusis MJ. Logistic regression analysis: SPSS for Win-dows, advanced statistics. Chicago, IL: SPSS; 1993:130.
14. Louis S, McDowell F. Epileptic seizures in nonemboliccerebral infarction. Arch Neurol 1967;17:414-418.
15. Olsen TS, Hgenhaven H, Thage O. Epilepsy afterstroke. Neurology 1987;37:1209-1211.
16. Arboix A, Garcia-Eroles L, Massons JB, et al. Predictivefactors of early seizures after acute cerebrovascular dis-ease. Stroke 1997;28:1590-1594.
17. Kilpatrick CJ, Davis SM, Tress BM, et al. Epileptic sei-zures in acute stroke. Arch Neurol 1990;47:157-160.
18. Berges S, Moulin T, Berger E, et al. Seizures and epilepsyfollowing strokes: Recurrence factors. Eur Neurol 2000;43:3-8.
19. Giroud M, Gras P, Fayolle H, et al. Early seizures afteracute stroke: A study of 1,640 cases. Epilepsia 1994;35:959-964.
20. Kilpatrick CJ, Davis SM, Hopper JL, et al. Early seizuresafter acute stroke: Risk of late seizures. Arch Neurol1992;49:509-511.
21. Heuts-Van Raak EP, Boellaard A, De Krom MC, et al.Supratentorial brain infarcts in adult-onset seizures; TheMaastricht epilepsy case register. Seizure 1993;2:221-227.
22. Gupta SR, Naheedy MH, Elias D, et al. Postinfarctionseizures: A clinical study. Stroke 1988;19:1477-1481.
23. Lo YK, Yiu CH, Hu HH, et al. Frequency and character-istics of early seizures in Chinese acute stroke. ActaNeurol Scand 1994;90:83-85.
24. Cheung CM, Tsoi TH, Au-Yeung M, et al. Epilepticseizure after stroke in Chinese patients. J Neurol 2003;250:839-843.
25. Lamy C, Domigo V, Semah F, et al. Early and lateseizures after cryptogenic ischemic stroke in youngadults. Neurology 2003;60:400-404.
L.P. KAMMERSGAARD AND T.S. OLSEN214