Abstract
We performed the first population-based study that determined the magnitude of the risk and identified the factors predictive of developing seizure disorders after cerebral infarction. Five hundred thirty-five consecutive persons without prior unprovoked seizures were followed from their first cerebral infarctions until death or migration out of Rochester, Minnesota. Thirty-three patients (6%) developed early seizures (within 1 week), 78% of which occurred within the first 24 hours after infarction. Using multivariate analysis, the only factor predictive of early seizure occurrence was anterior hemisphere location of infarct (odds ratio 4.0, 95% CI 1.2 to 13.7). Twenty-seven patients developed an initial late seizure (past 1 week), whereas 18 developed epilepsy (recurrent late seizures). Compared with the population in the community, the risk during the first year was 23 times higher for initial late seizures and 17 times higher for epilepsy. The cumulative probability of developing initial late seizures was 3.0% by 1 year, 4.7% by 2 years, 7.4% by 5 years, and 8.9% by 10 years. Independent predictive factors on multivariate analysis for initial late seizures were early seizure occurrence (hazard ratio of 7.8 [95% CI 2.8 to 21.7]) and stroke recurrence (3.1 [1.2 to 8.3]). Both early seizure occurrence (16.4 [5.5 to 49.2]) and stroke recurrence (3.5 [1.2 to 10.5]) independently predicted the development of epilepsy as well. We also found that early seizure occurrence predisposed those with initial late seizures to develop epilepsy.
NEUROLOGY 1996; 46 350-355
There have been many reports about the development of seizures subsequent to ischemic stroke, [1-23] but the proportion of stroke patients who experienced seizures varied from 0.4% to 43%. This wide range was undoubtedly due to dissimilarities among studies in patient attributes, in diagnostic criteria, and in the duration of follow-up. For example, the lowest reported seizure rate of 0.4% was based on observations limited to thrombotic infarct patients seen by a selective group of five neurologists. [6] On the other hand, the highest rate of 43% was derived from a small group of 42 patients seen at a referral center for cerebral embolic disorders. [4] Although most studies presumably included all hospitalized stroke patients, some were restricted to patients under the care of stroke services, [3,12,18] or to patients requiring rehabilitation. [5,21] One study excluded patients who were critically ill or older than 85 years. [10]
Except for a few, [3,4,6,7] previous reports failed to specify diagnostic criteria for types of strokes and seizures. In most studies, there was no distinction between initial and recurrent strokes or seizures or between provoked and unprovoked seizures. Most past series also did not specifically exclude other possible etiologies before ascribing seizures to strokes.
Nearly all previous studies reported only the crude proportions of stroke patients developing seizures. The risk of developing seizures relative to the general population has not been evaluated. To address the above limitations and to obtain more clinically useful information, we conducted the first population-based study of the association between ischemic strokes and subsequent seizure disorders. The main objectives of our study were to determine the magnitude of the risk of developing epileptic seizure disorders after cerebral infarction and to identify the factors predictive of their development.
Methods.
Population characteristics.
Rochester is an urban community in southeastern Minnesota with a population that has steadily risen from approximate 25,000 in 1940 to approximate 71,000 in 1990. Approximately 95% of the residents obtain some or all of their medical care at the Mayo Clinic. Practically all residents with serious illnesses are admitted at one of the hospitals affiliated with the Clinic. Every year, more than half of the population is seen at the Clinic or at one of its affiliated facilities. All autopsies in the community are performed by pathologists from the Mayo Clinic. The autopsy rate has consistently been approximate 60% since 1935.
Source of data.
From the beginning of this century, our institution used a unified medical record for each patient. Each record documents all aspects of the patient's medical care, including outpatient visits, hospital admissions, surgical reports, laboratory and pathologic results, and medical correspondence. In 1907, the Clinic began a system of classifying and coding diagnoses. Each medical record is indexed by diagnostic codes to permit efficient identification and retrieval of records according to diagnoses. In 1966, this system of diagnostic indexing was adopted by other health care providers in the community. A central file was subsequently established to link the medical records from all health care providers in Rochester. This allows virtually complete case ascertainment of major and minor medical conditions that occur in the community.
For the past 40 years, our EEG laboratory used a uniform system of diagnostic classification of EEG findings for every record performed. The classification permits us to reliably retrieve and analyze EEG results along with other variables.
Cohort identification and follow-up.
Subjects for the study were identified from our stroke database, which has been maintained and updated in our ongoing epidemiologic studies of these conditions. Study criteria were first cerebral infarction occurring while residing in the community between 1960 and 1969, no prior history of hemorrhagic strokes, and no prior history of unprovoked seizures. Seizure occurrence was ascertained by chart review. Each index case was followed until death migration out of Rochester, or December 31, 1992.
Terminology.
Terminology regarding epileptic seizure disorders was according to that established in earlier publications on the epidemiology of epilepsy in Rochester. [24,25] Unprovoked seizures refer to seizures that occurred without identifiable acute precipitants. Epilepsy was defined as a condition characterized by recurrent unprovoked seizures. Early seizures were seizures occurring within 1 week after cerebral infarction and were considered to be acutely provoked by stroke events. Unprovoked seizures developing beyond 1 week after infarction were termed late seizures. Acute symptomatic seizure due to recurrent strokes or other causes were excluded. Classification of seizure types was according to the International Classification of Epileptic Seizures. [26] For diagnostic classification of the types of cerebral infarctions, we applied methods used in our ongoing epidemiologic studies.
Infarcts were localized by previously validated clinical criteria. [27] Embolic type of infarct was determined by an algorithm [28] that established the presence of major potential cardiac sources of emboli. Recurrent stroke was determined by the occurrence of the next cerebral infarction or intracranial hemorrhage.
Only EEGs performed within 7 days after cerebral infarction were included for analysis. EEGs were considered abnormal when focal or lateralized slowing or epileptiform discharges were present.
The degree of functional disability at 1 and 6 months after cerebral infarction was graded by Rankin Disability Scale, [29] as follows: grade I, no significant disability; able to carry out all usual activities; grade II, slight disability; unable to carry out some previous activities but able to look after own affairs without assistance; grade III, moderate disability; requiring some help but able to walk without assistance; grade IV, moderately severe disability; unable to walk without assistance; and unable to attend to own bodily needs without assistance, and grade V, severe disability; bedridden, incontinent, and requiring constant nursing care and attention.
Statistical analysis.
Event rates of the different types of seizure disorders were calculated according to demographic and infarct characteristics. The probability of developing either initial late seizure or epilepsy was predicated on survival of longer than 7 days after cerebral infarction. Comparison of the incidence of initial late seizure and epilepsy between the cohort and the general population was by standardized morbidity ratios (SMR). For the incidence rates in the general population, we used age-and gender-specific rates for the years 1935 to 1984. [30] Kaplan-Meier life Table methodswere used to determine cumulative probabilities of developing initial late seizures and epilepsy. When calculating the SMRs and the cumulative probabilities associated with recurrent strokes, follow-up of the initial infarcts was censored at the time of stroke recurrence. Follow-up of the recurrent strokes began at the time of recurrence.
Chi-square tests were performed for univariate analysis of the following factors in the development of early seizures: gender, type of infarct (embolic versus thrombotic), location of infarct (anterior hemisphere versus others), focal EEG abnormality, and recurrent stroke; rank sum test was used to evaluate age at onset of infarct. Multivariate analysis of these factors was by stepwise logistic regression. For the outcomes of initial late seizure and epilepsy, the same factors plus early seizure occurrence were analyzed univariately by Cox proportional-hazards model. In the multivariate analysis by stepwise Cox regression, recurrent stroke and the two Ranking scale grades were added to these factors as time-dependent covariables. Factors with p less than 0.05 were considered significant in the multivariate models.
Results.
Cohort characteristics.
Five hundred thirty-five patients met the study criteria. Age at cerebral infarction was 71.6 plus minus 12.0 years (mean plus minus SD, range 8 to 99 years). There were 280 men (52%) and 255 women (48%). Four hundred twenty-one (79%) had nonembolic infarcts and 114 (21%) had embolic infarcts. Infarcts were located in the anterior hemisphere in 388 patients (73%), brainstem and/or cerebellum in 54 (10%), and posterior hemisphere in 33 (6%). Location was uncertain in 60 patients (11%). EEG was performed in 92 patients within 7 days after cerebral infarction. Fifty-four (58.7%) showed focal abnormalities.
Duration of follow-up was 5.5 plus minus 7.1 years (range 0 to 33.1 years). Ninety-nine patients died within 7 days of cerebral infarction. There were 2,930 person-years of follow-up in the remaining 436 patients. Strokes recurred in 148 patients.
Early seizures.
Thirty-three of the 535 patients (6%) developed early seizures. Most (78.0%) of the early seizures occurred within 24 hours after cerebral infarction. Univariate analysis of the potential factors in the occurrence of early seizures showed that infarcts in patients with early seizures were more likely to be embolic in type and anterior hemisphere in location (p less than 0.05; Table 1). On multivariate analysis, the only factor predictive of early seizure occurrence was anterior hemisphere location of infarct (odds ratio 4.0; 95% CI 1.2 to 13.7).
Table 1. Univariate analysis of risk factors in early seizures
SMRs for initial late seizures and epilepsy.
Of the 436 patients who survived longer than 7 days after cerebral infarction, 27 developed initial late seizures. The expected number of initial late seizure cases calculated from the general population was 4.2. Hence, the relative risk (SMR) of developing initial late seizures was approximately six times that of the general population Table 2. The risk was highest during the first year--almost 23 times that expected. It remained increased over the next 3 years.
Table 2. SMR of developing initial late seizure and epilepsy in the whole cohort by time period after cerebral infarction
Of the 27 patients with initial late seizures, 18 subsequently developed epilepsy. The cumulative probability of developing epilepsy among those available for follow-up approached 90% by 4.5 years after initial late seizure occurrence Figure 1. Similar to initial late seizures, SMR of developing epilepsy was highest during the first year Table 2. However, unlike initial late seizures, the elevated SMR persisted beyond 4 years after cerebral infarction.
Figure 1. Survival curves of the probabilities of developing epilepsy in initial late seizure patients with and without early seizures
Stratification of SMR according to age at cerebral infarction showed that the risk was highest in those younger than 55 years for either initial late seizures or epilepsy Figure 2. SMRs in those age 55 to 64 years were also much higher than in persons who were older at the time of infarct.
Figure 2. SMR (confidence interval) of initial late seizure and epilepsy by age groups
Six of the patients with recurrent strokes subsequently developed initial late seizures, five of whom eventually developed epilepsy Table 3. The expected numbers of cases were 0.7 for initial late seizures and 0.5 for epilepsy. Thus, the SMRs of 8.3 for initial late seizures and 9.4 for epilepsy after recurrent strokes were not significantly higher than those after initial infarcts.
Table 3. SMR after first infarct or after recurrent stroke and in the whole cohort
Cumulative probabilities of developing initial late seizures and epilepsy.
Table 4 shows the cumulative probabilities of developing initial late seizures and epilepsy after initial cerebral infarets and recurrent strokes. After initial infarcts, cumulative probability of developing initial late seizures gradually increased to almost 7% in the 4th year but remained stable through the 12th year. It then increased slightly to its peak of approximate 8% in the subsequent year. In contrast, cumulative probability of developing initial seizures after stroke recurrence peaked at 14% in the ninth year after recurrence.
Table 4. Cumulative risks in percent of developing initial late seizures or epilepsy after first infarct or after recurrent stroke*
Factors predictive of initial late seizures and epilepsy.
Univariate analyses showed that early seizure occurrence was the only factor with a significantly higher risk of subsequently developing initial late seizure and epilepsy Table 5. Compared with patients without early seizures, patients with early seizures were 9 times more likely to develop initial late seizures and 19 times more likely to develop epilepsy.
Table 5. Hazard ratios* by univariate analyses and by multivariate analysis of factors in initial late seizure and in epilepsy
Multivariate analysis showed that early seizures and recurrent strokes were the only factors predictive of the development of initial late seizures and epilepsy Table 5. Compared with patients without early seizures, patients with early seizures were nearly 8 times more likely to develop initial late seizures and approximate 16 times more likely to develop epilepsy. Stroke recurrence increased the risk of developing either initial late seizure or epilepsy by approximately three times.
Discussion.
Previous studies differed in how they distinguished early or acute seizures from late seizures. Whereas most defined early or acute seizures as those arising within 2 weeks after stroke, Walczak et al. [20] included all seizures occurring within 30 days after stroke. On the other hand, Shinton et al. [13] limited their report to patients who experienced seizures acutely within 24 hours of stroke. Hauser et al. [9] defined early seizures as those that developed within 1 week after cerebral infarction. We considered early seizures as having been acutely provoked by stroke events. When analyzing studies in the literature, it should likewise be noted that some [5,10,12,21] reported the rate of epilepsy and not the rate of initial seizure occurrence. This distinction is important because not all patients who experience a seizure subsequently develop epilepsy.
We found that most early seizures happen on the first day after infarction. Ninety percent of early seizures in other series [7,31] also occurred within a day or two of infarction. Other investigators [11,31] made similar observations in patients with intracranial hemorrhage. As for initial late seizures and epilepsy, the risks are highest during the first year after cerebral infarction. The risks decline in subsequent years.
Our patients who were less than 55 years old had the highest SMR for initial late seizures and for epilepsy. Their SMR for either of these seizure disorders was approximate 30, as compared with an SMR of 4.0 in those 75 years or older. The higher SMR was mostly due to the low expected incidence rates of these seizure disorders in the corresponding age group of the general population Below 55 years of age, the expected incidences were 53.7 in 100,000 for initial late seizures and 35.8 in 100,000 for epilepsy. In contrast, the expected rates in those 75 years and older were 212.7 in 100,000 for initial late seizures and 154.3 in 100,000 for epilepsy. Moreover, contamination of the expected rates in this age group by the higher prevalence of cerebral infarction in the standard population attenuated the SMR for late seizure disorders. Otherwise, the absolute risks were comparable between those younger than 55 years and those 75 years and older.
Nearly all previous series estimated only crude seizure rates that were difficult to compare because the duration of follow-up varied widely. Only two determined cumulative rates of developing late unprovoked seizures after cerebral infarction. Despite the inclusion of hemorrhagic strokes and TIA, the cumulative rates in one study [12] of 3% by 1 year and 5% by 5 years approximate the respective rates of 3.0% and 7.4% in our patients. The other study [9] observed a much higher rate of 19% by 5 years in 206 patients hospitalized for occlusive cerebrovascular disease. In that retrospective study, which was not population-based, the higher rate may be due to a greater likelihood of patients who developed seizures to return for medical attention and follow-up.
Location of infarcts with early seizures was more likely to be anterior hemisphere than in those without early seizures. Involvement of cerebral cortex may explain the association between anterior hemisphere infarction and early seizures. One autopsy study [2] and several clinical series using CT or MRI [10,14,18,23,32] found that cortical strokes were more likely to be followed by seizure disorders than were subcortical strokes. In the autopsy series of Richardson and Dodge, [2] none of the 47 patients with subcortical lesions developed seizures. In contrast, 13 of their 57 patients (22.8%) with cortical strokes experienced seizures. When epilepsy was the specific outcome measure, Olsen et al [10] obtained very similar findings. Epilepsy developed in 6 of their 23 patients (26%) whose lesions affected the cerebral cortex, whereas only 1 of the 54 patients (2%) without cortical involvement subsequently had epilepsy.
Our study revealed that embolic infarction is associated with early but not with late seizure occurrence. Embolic infarction has previously been observed to bear varying degrees of seizure risk. There was no general agreement among the seven studies [1,2,5-7,18,33] that analyzed seizure risk according to the mechanisms of infarction. Two [1,2] were autopsy studies that also yielded conflicting results. Richardson and Dodge [2] reported a sixfold higher risk after embolic infarcts compared with thrombotic infarcts, whereas Merritt and Aring [1] detected no difference. The reason for the variance is not obvious, but dissimilarities in the autopsy subjects may in part explain the conflicting results.
Large stroke lesions are generally expected to carry a higher risk of developing early seizures than small lesions. This was the observation of Kilpatrick et al. [18] when they initially assessed seizure risk without regards to stroke type. Yet, there was no difference when their analysis was stratified by stroke types. Lacunar infarcts, which are typically small and not associated with seizures, [6,18,20] accounted for the finding of lower risk with smaller lesions when all types of strokes were analyzed together. In another study of intracerebral hemorrhage, [11] there was also no association between lesion size and early seizures. We could not investigate the relationship between infarct size and early seizures because our study involved patients whose cerebral infarction developed before the advent of CT and MRI.
We found early seizure occurrence and stroke recurrence to be predictive of developing either initial late seizures or epilepsy. Hauser et al. [9] also found that late seizures were more likely to develop in patients with early seizures than in those without. Kilpatrick et al. [34] recently made the same observation despite a major flaw in their study design. They defined late seizures as ``seizures occurring following discharge.'' Hence, the risk of late seizure occurrence in their patients became dependent on the length of hospital stay.
After early seizure occurrence, approximately half of our patients (17/33) were started on antiepileptic medications. When initial late seizures occurred, 3 of 27 patients were taking antiepileptic drugs. Ten patients were treated after initial late seizure occurrence. In this retrospective study, we could not reliably verify whether antiepileptic drug therapy was optimized or not. Determination of serum antiepileptic drug concentrations was introduced only in the latter part of the study period.
Although early seizure occurrence increases risk of subsequent epilepsy, chronic antiepileptic drug treatment is not warranted in most early seizure patients. Only 5 of our 23 early seizure patients (21%) who survived 1 week after cerebral infarction developed epilepsy. However, the option of antiepileptic drug treatment may be considered when an early seizure is followed by an initial late seizure. All five of our patients with early and initial late seizures developed epilepsy by 1 year, whereas only 13 of the 22 initial late seizure patients who had no early seizures developed epilepsy (p less than 0.001; Figure 1). This observation agrees with the finding of Hauser et al. [35] that, in the general population, the risk of developing epilepsy after a first unprovoked seizure is increased by a history of acute symptomatic seizures. As in that group of patients, the benefits of treatment after an initial late seizure in patients with prior cerebral infarction and early seizure should be weighted against the risk of adverse drug effects. [36]
Acknowledgments
We are grateful to Ruth H. Cha, MS, and Peg E. Farrell, RN, for their invaluable assistance.
- Copyright 1996 by the Advanstar Communication Inc.
REFERENCES
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hastening the Diagnosis of Amyotrophic Lateral Sclerosis
Dr. Brian Callaghan and Dr. Kellen Quigg