If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Epileptic seizures are well known sequelae of patients with stroke but only little is known about the different risk factors and about the influence of the different types of stroke including sinus thrombosis and bleedings on developing such seizures. Further, the association of post-stroke seizures and conventional vascular risk factors has not been evaluated to date.
We performed a cohort study on a sample of 593 consecutive patients with different types of cerebrovascular events. In 421 patients, sufficient data were obtained in a personal interview over a mean observation period of 30 months. Data regarding the clinical history were recorded from the patients’ charts.
The total prevalence of epileptic seizures was 11.6%, the total annual risk was 4.6%. We detected the following significant risk factors: younger age at stroke; higher NIH stroke scale score; any coagulopathy. TIA was found significantly less frequent as a cause of seizures as compared to infarction, bleeding, and sinus thrombosis. Patients with bleeding (14.3%) and with sinus thrombosis (16.3%) were significantly more frequent in the seizure group than in the non-seizure group (6.7% and 1.6%, respectively). The location of stroke, including cortical versus subcortical, did not influence the risk of seizures. The majority of patients developed secondary generalized seizures (57.1%). In adjusted analyses, the two major risk factors for post-stroke epilepsy were a higher NIH stroke scale and a sinus thrombosis as the initial cerebrovascular event. Common lifestyle, vascular, and metabolic risk factors of stroke and for dementia were not associated with the development of seizures.
In conclusion, our data show that epileptic seizures occur in particular after major strokes and in sinus thrombosis. Interestingly, conventional vascular risk factors were not associated with the occurrence of post-stroke seizures. Considering the risk for seizures after certain types of cerebrovascular events might help to early identify patients for anticonvulsive treatment. In the future, it should be investigated whether these patients might benefit from pre-emptive anticonvulsant treatment.
In an ageing society and with further improvement in stroke treatment, it becomes crucial to find methods for the prevention of seizures which have considerable social and psychological impact on the patients.
Some epidemiological studies have been published regarding seizures after stroke but they differ in number of patients, inclusion criteria, time of follow up, and diagnostic methods. Many studies have examined intracerebral haemorrhage (ICH), subarachnoid haemorrhage (SAH), and ischaemic stroke. There is, however, not much data concerning the development of seizures after other cerebrovascular events including transient ischaemic attack (TIA), different types of bleeding, and intracranial venous thrombosis (IVT).
In this hospital-based study, we aimed to find risk factors that contribute to the development of seizures and epilepsy after stroke in patients with different types of cerebrovascular events in univariate and multivariate analyses.
We evaluated all patients admitted to the Stroke Unit of the Department of Neurology, University of Münster between January 2003 and March 2010 for whom sufficient data were available in the database (n = 1611). Out of this number, we selected all patients with a first ever ischaemic stroke, TIA, ICH, SAH, or IVT. Only strokes with cerebral manifestation were included; strokes in cerebellar or brain stem location were excluded. Patients who had died while they were still in hospital were also excluded.
For the resulting 593 patients, a detailed questionnaire was completed. One part of the questionnaire (baseline demographic and clinical data) was filled in with data obtained from a patient chart review and a clinical interview with the patients who were in hospital between October 2009 and March 2010 with a minimum duration of 2 weeks after stroke. In the other patients, a follow-up telephone interview either with the patients themselves or with relatives on behalf of those patients who could not communicate was conducted at least 3 months after the initial event. Only cases with a successful interview were finally included in the study (n = 421). The mean time of follow-up was 30 months (maximum 78 months). In 172 cases, the follow-up interview could not be carried out for the following reasons: the patients had died meanwhile and no relative was available (n = 21); the contact data in the charts were not correct any more (n = 139); they did not want to participate for personal reasons (n = 12). At the time of interview, 41 patients (9.7%) had died but sufficient data was obtained by relatives. The flow chart of the different reasons for patients enrolled or excluded is presented in Fig. 1.
Seizures and epilepsies were classified according to the guidelines of the International League against Epilepsy (ILAE).
Early onset seizures were defined when they appeared within the first 7 days after stroke onset. Seizures were distinguished as being simple partial, complex partial, or generalized.
Stroke severity was measured using the National Institute of Health (NIH) stroke scale which was obtained on admission to the hospital. In all cases, a CT or MRI brain scan were performed to diagnose the type of stroke. TIA was defined as symptoms lasting less than 24 h and no correlation to the symptoms in MRI and CT. Ischaemic stroke was diagnosed via plain CT, CT-angiography, CT-perfusion studies, or MRI with diffusion-weighted-imaging (DWI); bleeding was diagnosed via CT and MRI with DWI. IVT was also diagnosed via CT angiography or MRI angiography. TOAST criteria classification of ischaemic stroke was confirmed by carotid Doppler and CT angiography for ‘macroangiopathic stroke’, by transoesophageal echocardiography for ‘cardioembolic strokes’, and by CT/MRI scan (stroke size less than 15 mm) and absence of significant findings in the other two diagnostic tools for ‘microangiopathic stroke’. Further, events were classified as ‘other defined aetiology’ (e.g., cervical artery dissection) and of ‘unknown origin’ if none of the diagnostic approaches leads to inclusion in one of the aforementioned categories. Strokes were subclassified as having occurred in the anterior circulation if they occurred in the middle or anterior cerebral artery territory and in the posterior circulation if they had occurred in the vertebrobasilar territory. If a stroke had occurred in both it was excluded from this analysis in order to obtain homogenous subgroups. Only strokes clearly located in one of the two hemispheres were included in the respective analysis.
Statistical analysis was performed using SPSS version 18. In the univariate analysis, categorical variables were compared using χ2-test and Fisher's exact test (if applicable). For parametric variables, we described arithmetic mean and standard deviation. The significance of differences between two groups was tested by Mann–Whitney U test. For multivariate analysis, a logistic regression model (backward Wald) was used. For this model, we included factors that showed significant influence on seizure development in univariate analysis and allowed to test a group which was large enough to perform a logistic regression model. Significance level was set at p = 0.05.
Out of the 421 patients finally enrolled, 53.2% (n = 224) were male and 46.8% (n = 197) were female. The mean age was 64.8 ± 15.1 years. In total, 9.7% (n = 41) had deceased at the time of the interview but sufficient data was available. The diagnoses were as follows: TIA, n = 64 (15.2%); ischaemic stroke, n = 311 (73.9%); ICH, n = 32 (7.6%); IVT, n = 14 (3.3%). Patients who developed secondary ICH were included with the original aetiology (ischaemic stroke/IVT). N = 49 patients (11.6%) developed at least one epileptic seizures. The patients’ characteristics can be seen in Table 1. The calculated annual risk to develop a seizure was 4.6% for the total sample and 4.1% for ischaemic stroke.
Table 1Characteristics of the total patients sample (n = 421).
The results of the univariate analysis for all variables regarded as potential risk factors for seizure development are presented in Table 2. Patients who developed seizures were significantly younger, had a significantly higher NIH stroke scale score, and received significantly more often an inpatient rehabilitation than patients without a seizure. Surprisingly, pre-existing cardiac conditions were significantly more frequent in patients without seizures than in patients with seizures. However, in the subgroup of pre-existing cardiac conditions the existence of coagulopathy was significantly more common among patients with seizures (8.2% compared to 1.3%; p = 0.013, OR 6.5, 95% CI 1.7–25.2) while patent foramen ovale (PFO) was less frequent in the seizure group (2.0% compared to 11.6%; p = 0.044, OR 0.2, 95% CI 0.0–0.6). All other risk factors were not significantly different in the two groups.
Table 2Univariate analysis of stroke risk factors for development of epileptic seizures.
Among the subjects with seizures, 4.1% had TIA compared to 16.7% in the non-seizure group, 65.3% had ischaemic stroke compared to 75.0% in the non-seizure group. Patients with bleeding (14.3%) and with IVT (16.3%) were significantly more frequent in the seizure group than in the non-seizure group (6.7% and 1.6%, respectively). TIA was significantly less frequent in patients with seizures compared to those with bleeding (p = 0.006; OR 8.7, 95% CI 1.7–22.7) and IVT (p < 0.001; OR 41.3, 95% CI 7.1–240.6). IVT was also more common in patients with seizures than ischaemic stroke (p < 0.001; OR 11.6, 95% CI 3.8–35.6) and bleeding (p = 0.038; OR 4.8, 95% CI 1.2–18.4).
Ischaemic stroke was subsequently divided into different stroke aetiologies using the TOAST criteria. We found that 12.1% of the patients with seizures had stroke of cardiac origin compared to 31.7% in the non-seizure group. None of the patients with stroke of microangiopathic origin suffered from seizures while microangiopathic ischaemic stroke was seen in 5.3% of the cases without seizures. Macroangiopathic ischaemic stroke (thrombotic stroke) was seen in 36.4% of the patients who later developed seizures while this aetiology was responsible for only 19.2% of the strokes not followed by seizures thus being significantly more common than the other aetiologies (p = 0.006; OR 4.9, 95% CI 1.5–16.1). Strokes of other defined aetiology were found in 9.1% of the patients with seizures and in 5.7% of those without. The aetiology of the stroke was unknown in 42.4% of the patients who suffered from seizures and 38.1% in the group that did not have seizures.
Neither cortical involvement nor multiple strokes were significantly more common among patients who developed seizures. Carotid circulation stroke was more common than basilar circulation stroke both in patients without and with epileptic seizures showing no statistical significance between them. Strokes that were limited to one hemisphere (left or right) were equally common in both groups. Detailed results can be seen in Table 3.
Table 3Impact of stroke aetiology and stroke localization on the development of epileptic seizures.
Table 4 presents the characteristics of the seizures in the patient sample. Partial seizures were observed in 40.8%. Out of these, 55.0% were simple partial and 45.0% were complex partial. 57.1% patients had generalized seizures. In one case, classification of the seizure could not be evaluated based on the patient's history. Status epilepticus was found in 16.7% of the cases, 6.3% presented with Todd's paralysis. Electroencephalography (EEG) was performed by our department in 49% of the cases directly after a (suspicious) seizure. 8.3% had EEG patterns suggestive of epilepsy, 50.0% had other pathological EEG findings, while 41.7% had no abnormalities in EEG. Anticonvulsant drugs were used by 91.8%. Seizure freedom under medication was reached in 81.6%. Monotherapy was prescribed in 86.7% of the cases.
Table 4Different characteristics of epilepsy and epileptic seizures in stroke patients (presented in %).
In Table 5, the differences between patients with early onset und late onset seizures (for definition see methods) are presented. Early onset seizures appeared in 5.5% of all cases. Of the patients with seizures, 46.9% had early onset seizures while 53.1% had late seizures accounting for 6.2% of all cases. Patients with early seizures had a significantly lower NIH stroke score on admission (4.3 ± 4.9 compared to 9.8 ± 6.0 for those with late seizures; p = 0.002). All patients with IVT and seizures had early onset seizures (p = 0.001, OR 0.4, 95% CI 0.3–0.6). Late onset seizures were significantly more common among patients with TIA, ischaemic stroke, or bleeding with cortical involvement (38.5% compared to 6.7% in the early seizure group; p = 0.003, OR 0.1, 95% CI 0.0–1.0). Cortical involvement was not specifically recorded for patients with IVT. Adjusted for cortical involvement in IVT, a rate of 39.1% in early seizures was recorded.
Table 5Comparison of early versus late (i.e., 7 days or later after stroke) seizures in stroke patients.
The differences between patients with post-stroke epilepsy and only one seizure are presented in Table 6. Out of all patients with a post-stroke seizure, 57.1% suffered from recurrent seizures, i.e., epilepsy. 73.1% of them were patients with late seizures (p = 0.022, OR 0.02, 95% CI 0.01–0.8). Patients who developed epilepsy had a mean age of 59.3 ± 13.2 years compared to 68.5 ± 7.9 for those with a single seizure (p = 0.040). None of the other parameters such as stroke type, vascular territory or pre-existing conditions were statistically significantly different. Also, in terms of seizure presentation such as type of seizure, status epilepticus, Todd's paralysis, or EEG findings, no statistically significant differences could be observed.
Table 6Risk factors for developing epilepsy versus only one single seizure after stroke.
We included age, NIH stroke score, type of stroke, and cardiac pre-existing conditions as factors in a multivariate analysis in order to evaluate independent risk factors for the development of seizures after stroke. The results are presented in Table 7. A higher NIH stroke score was a significant factor for the development of seizures (p = 0.001) with a slightly elevated risk with each additional point on the scale (OR 1.1, 95% CI 1.0–1.2). IVT was associated with an adjusted risk of 9.3 (95% CI 2.1–41.6, p = 0.004) to develop seizures when compared to all other types of stroke. In this model, no other aetiology was significantly and independently associated with the development of seizures. Also, the existence of vascular pre-existing conditions and age were not significant for seizure development after stroke. In another model, we looked for differences between ischaemic stroke of cardiac and macroangiopathic origin, age, NIH stroke score and pre-existing cardiac conditions. Here, the appearance of macroangiopathic stroke was a risk factor for seizure development (p = 0.002, OR 12.1, 95% CI 2.5–59.4). In this model, however, neither an elevated NIH stroke score nor age was significantly associated with seizure development.
Table 7Multivariate analysis of risk factors for seizure development after stroke.
project suggested a cumulative risk of 11.5% for the development of seizures within 5 years. This corresponds well with our findings although we included IVT. In our study, the rate of SAH was relatively small (n = 5) which could explain the similar findings. Bladin et al.
were more at risk of developing seizures. Aging brain undergoes degenerative and neuroplastic changes that lead to decreased excitability of the cortex. Given that the volume of cortical gray matter is higher in younger patients, it is easy to assume that these younger patients develop seizures more often. Nonetheless, age was not a significant risk factor in multivariate analysis of stroke types but when we looked at cardioembolic and macroangiopathic ischaemic stroke, age was a significant factor again. Stroke severity
but measurement of stroke severity was carried out with different instruments in different studies. We found that a higher NIH stroke scale score increases the risk for the development of seizures. This finding is supported by Bladin et al.,
found no significant differences in NIH stroke scores but only investigated ischaemic and haemorrhagic strokes. The higher rate of seizures in patients with inpatient rehabilitation treatment can be explained by the fact that patients with a more severe stroke are more likely to receive this procedure. In multivariate analysis of stroke subtypes, NIH stroke score was significantly associated with seizures, whereas in the setting of ischaemic strokes only, NIH stroke score was not significantly associated with seizures.
In our study, IVT was associated with seizures in multivariate analysis when compared to all other stroke types. In most studies
seizures following ICH and SAH are significantly more common than after ischaemic stroke but these studies did not include TIA and IVT. We found a significantly lower rate of seizures in patients with bleeding compared to those with IVT in univariate analysis but no significantly higher rates of bleeding when compared to ischaemic stroke. In contrast to other studies,
we did not find any significant difference in seizure occurrence with respect to left or right hemisphere or multiple infarction. However, seizures occurred clearly much more often after anterior circulation strokes.
used the NIH stroke score to measure stroke severity and found significantly higher NIH stroke scores in patients who had early seizures compared to those who did not develop seizures at all. The mean score in the patients with early seizures was notably higher than in our study. In our study, the NIH stroke scores are similar for patients with early onset seizures and patients without seizures but we used a 7-day interval to distinguish early seizures from late seizures and also included IVT. When adjusted for the 24-h interval, we see a mean score of 5.3 which is still lower than in the aforementioned study. In our study, all seizures that appeared after IVT were early onset seizures. Patients who had early onset seizures had significantly lower NIH stroke scores than those with late onset seizures.
Furthermore, having cardiac pre-existing conditions resulted in a lesser incidence of seizures. Since in our study seizures appeared in a relatively young population, it is likely that this younger population had less cardiac pre-existing conditions. In multivariate analysis, cardiac pre-existing conditions failed to show significance concerning seizure development. Even though coagulopathy was significantly more frequent among patients with seizures, this finding was not included in multivariate analysis due to the small number of cases. Since coagulopathies are associated with IVT,
stated that an elevated level of serum cholesterol protected against early seizures. Our results cannot confirm this finding. Familial disposition for stroke or epilepsy did not lead to a higher rate of seizures in the respective groups.
In the multivariate analysis, we found that macroangiopathic stroke was more likely to be followed by seizures than strokes of cardiac origin. In addition, patients with probable cardioembolic focus either suffered less from seizures (in patients with PFO) or showed no differences compared to those without. In the literature, there are either no differences
that support our finding. We hypothesize that subcortical strokes can cause cortical liability to develop seizures. Thus, stroke severity predicts the appearance of seizures more likely than the initial involvement of the cortex itself.
Information about seizure characteristics is inconsistent in the literature because of different methods to differentiate seizure types. We found seizures to be more often generalized than partial. This is supported by Berges et al.
found an incidence of only 9% but did not include SAH and IVT. While 50% of the patients with bleeding and seizure in our study developed status epilepticus, none of the patients with IVT had status epilepticus.
It is very difficult to determine the characteristics of seizures via phone interview since classification often depends on patients’ observations. For example, it is very difficult to evaluate the percentage of true status epilepticus when use of benzodiazepines to terminate a seizure is included in this definition rather than just based on direct clinical observation as in the study of Velioglu et al.
postulated that Todd's paralysis is partly due to cortical ischaemia and arteriovenous shunts which develop through vasomotor or metabolic changes. Following that theory it would be likely that cerebrovascular diseases could cause a higher rate of Todd's paralysis. We found Todd's paralysis in only 6.3% of the patients. In a study by Hornig et al.,
This is due to the fact that most early seizures appear within the first 24 to 48 h after onset of stroke. In our study, 17 seizures (73.9% of all early seizures) developed within the first 24 h which is in accordance with recent data.
stated that more female patients developed epilepsy, we found no significant difference.
Although we subdivided the patients with seizures in patients with single seizures and epilepsy, these findings might not be representative because the minimum follow up period was 3 months. It is possible that we missed recurrent seizures among some patients. In a phone interview, it is also likely that some patients denied having had seizures due to social expectancy or improper comprehension. In addition, Blum et al.
noted that only 26% of the patients are always aware of their seizures. This, however, is a problem that is always faced when examining longtime outcome concerning seizure development.
Our study has some limitations. First, most patients were enrolled retrospectively. This might cause a selection bias since patients with severe seizures could have passed away or were referred to nursing homes where they could not be contacted. Second, we lost 139 patients in the follow-up. The reasons for this loss remain unclear but it is likely that patients with severe impairment had to move to nursing homes or other places and could therefore not be contacted. This would mean a negative selection bias since patients with severe seizures were not detected. Third, we evaluated seizures and their semiology in the majority of patients by a telephone interview. This is of course not an appropriate method to detect all types of seizures. However, it is difficult to approach stroke patients with severe disability anyway, even if they are examined face to face. Furthermore, the telephone interview was performed by an experienced investigator and included also contact to the relatives of the patients in most cases. Therefore, we believe that our data are as correct as it is possible in this specific clinical situation.
Based on our data, we conclude:
Patients with IVT are significantly more at risk to develop a single seizure than patients with bleeding or ischaemic stroke but do not have a higher risk of developing epilepsy or status epilepticus.
Bleeding leads to the development of epilepsy in 50% of the cases. This finding was not statistically significant but shows a tendency for a more severe course compared to the other aetiologies.
In ischaemic stroke, macroangiopathy leads more often to seizures than cardioembolic strokes.
A higher score of the NIH stroke scale predicts seizure development. This is more likely for late seizures than for early seizures, but it is not helpful in predicting the development of epilepsy.
Younger patients are more at risk to develop single seizures and epilepsy than older patients but this finding failed to show statistical significance in multivariate analysis.
Neither cortical involvement in ischaemic stroke nor the presence of multiple infarcts or their location regarding vascular territories and hemisphere are useful in predicting the development of seizures.
We tested for common lifestyle, vascular, and metabolic risk factors of stroke and for dementia but could not observe any correlation between those and the development of seizures.
The present analysis shows that only the type of stroke and stroke severity have a significant potential in predicting seizure development after a first ever stroke on a multivariate level. All other factors do not exceed the univariate level. In the setting of ischaemic strokes only the presence of macroangiopathic stroke has predictive value.
Concerning these findings further research should clarify whether these factors are persistent if tested in different age groups. New diagnostic approaches such as diffusion and perfusion imaging for risk stratification of seizure development after stroke are needed as the pathomechanisms of stroke seem to be more important for the development of seizures than underlying conditions. Furthermore, clinical trials are warranted whether patients with a higher risk of developing epilepsy after stroke do benefit from early or pre-emptive prophylactic treatment.
Epileptic seizures in the elderly. I. Aetiology and type of seizure.