Abstract
Objective: To evaluate long-term mortality among people with status epilepticus (SE).
Methods: The authors performed a population-based retrospective cohort study to determine long-term mortality after SE. Between January 1, 1965, and December 31, 1984, all first episodes of SE receiving medical attention were ascertained through the Rochester Epidemiology Project Records-Linkage System. Cases surviving the first 30 days (n = 145) were followed until death or study termination (February 1996).
Results: At 10 years, cumulative mortality among 30-day survivors was 43%. The standardized mortality ratio (SMR) at 10 years was 2.8 (95% CI, 2.1–3.5). The mortality rate of those with idiopathic/cryptogenic SE was not increased (SMR = 1.1; 95% CI, 0.5–2.3). The following characteristics of SE increased long-term risk for mortality: SE ≥ 24 hours in duration vs SE < 2 hours (relative risk [RR] = 2.3; 95% CI, 1.1–5.1); acute symptomatic etiology vs idiopathic/cryptogenic etiology (RR = 2.2; 95% CI, 1.0–5.1) SE; myoclonic SE vs generalized convulsive SE (RR = 4.0; 95% CI, 1.3–13).
Conclusion: Forty percent of subjects who survived the first 30 days after an incident episode of SE die within the next 10 years. The long-term mortality rate was threefold that of the general population over the same time period. The long-term mortality rate at 10 years was worse for those with myoclonic SE, for those who presented with SE lasting more than 24 hours, and for those with acute symptomatic SE. The long-term mortality rate was not altered in those with idiopathic/cryptogenic SE. We conclude that SE alone does not modify long-term mortality.
No data in the literature exist on long-term mortality after an incident episode of status epilepticus. The majority of studies of SE have focused on short-term mortality with a follow-up period generally less than 1 year.1-11⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓
We examined the mortality rate of patients after a first episode of SE to evaluate the potential influence of SE on mortality after surviving the first 30 days. As in our study on short-term mortality,11 we evaluated possible clinical determinants of mortality such as seizure type, etiology, and duration of SE.
Materials and methods.
Study population.
The methods and the reference population of our study have been described previously.11 Briefly, medical records of residents of Rochester, MN, with a first seizure of any type between 1965 and 1984 were reviewed to identify those with an episode of SE. SE was defined as a seizure lasting ≥30 minutes or repeated seizures over a period ≥30 minutes without recovery between episodes. Consequently, subjects with the following were excluded: seizure lasting <30 minutes and stopped by antiseizure medication (ASM); electroencephalographic SE in the absence of clinical features; or a flurry of seizures, each <30 minutes in duration, with intervening periods of consciousness. We excluded 17 cases with incident febrile SE (all in children < 5 years of age) from the analysis. There were no deaths in this group. Febrile SE is a benign condition that has no effect on short- or long-term survival and it was not considered in this study.
Through the records linkage system, 184 cases of first afebrile SE were ascertained among residents of Rochester, MN, between January 1965 and December 1984. For analysis of long-term mortality, we considered only the 145 subjects who survived the first 30 days. This was done to eliminate the influence of the early events related to the underlying etiology of SE under the assumption that deaths occurring in the acute stage and those occurring later would be related to different factors.
Death or the end of the study period (February 1996) terminated follow-up. Vital status (living or dead) was identified for all cases through February 1996. All analyses are presented at 10 years after initial SE.
Measurement of study variables.
Status seizure type.
Seizure type was categorized by following criteria of the International League against Epilepsy (ILAE).12 SE was considered generalized when there was no clinical evidence of a focal or localized onset. SE was considered secondarily generalized when there was clinical evidence of a focal onset or focal features to a generalized seizure. Seizures were considered partial when the SE remain localized to one area without secondary generalization. Complex partial SE was categorized in this group. Absence SE was a seizure manifest by a (≥30 minutes) confusional state associated with a generalized spike and wave generalized EEG pattern. Myoclonic SE was characterized by random but subtle twitching of muscle groups frequently limited to face and eyelids. In this series, myoclonic SE was invariably associated with postanoxic encephalopathy.
Etiology.
Etiology of SE was assigned into broad categories—acute symptomatic, progressive symptomatic, remote symptomatic, and idiopathic/cryptogenic—based on our previous work and ILAE recommendations.13,14⇓ SE was considered acute symptomatic when it occurred within a week of an acute CNS or systemic metabolic insult. These insults included but were not limited to brain trauma, CNS infection, cerebrovascular disease, and acute systemic insults including general metabolic insults or alcohol or drug toxicity or withdrawal. SE was considered unprovoked in the absence of an identified acute insult. Unprovoked SE was further categorized according to the ILAE recommendations for classification of seizures or epilepsy for epidemiologic purposes as progressive symptomatic, remote symptomatic, or idiopathic/cryptogenic.13,14⇓ SE was considered progressive symptomatic in the presence of nonstatic CNS conditions such as CNS tumors and degenerative neurologic diseases. SE was considered remote symptomatic in the presence of a history of a CNS insult presumed to result in a static encephalopathy associated with an increased risk for epilepsy; examples include stroke, head trauma, or meningitis. The time between SE and the neurologic insult had to be more than 1 week but was usually months or years. SE was considered idiopathic/cryptogenic in the absence of an acute precipitating factor or a history of a prior neurologic insult. The term idiopathic/cryptogenic is consistent with the current ILAE recommendations and encompasses people with epilepsy of unknown cause.15
Duration.
Duration of SE was estimated through review of the notes from ambulance technicians, emergency room staff, intensive care units, attending physicians, and nurses. These notes generally had times appended to allow calculation of duration. In each case duration was assigned into 6 broad categories by the study neurologists: 30 to 59 minutes, 1 to <2 hours, 2 to <6 hours, 6 to <12 hours, 12 to <24 hours, and ≥24 hours. For the current analysis, duration was categorized as 30 minutes to 2 hours, 2 to 24 hours, or ≥24 hours.11
Data analysis.
External comparison.
The mortality rate of subjects with SE was compared with that expected in the general Minnesota population matched for age, sex, and time period (years 1969 and 1979) using the standardized mortality ratio (SMR).16
Prognostic factors within the cohort.
We used Kaplan–Meier survival curves to estimate the cumulative mortality rate across etiologic groups.17 The contribution of potential prognostic factors to mortality through 10 years after SE was determined through a Cox proportional hazards model.17 We estimated the association between mortality and each potential prognostic factor in a univariate model; point estimates of relative risk (RR) and 95% CI were calculated. We then constructed a final model including all factors to measure the adjusted RR and 95% CI for each factor. The following variables were included in the model as possible determinants: age, sex, duration of SE (<2 hours, 2 to 24 hours, ≥24 hours), and seizure type (myoclonic, absence, partial only, partial with secondary generalization, and generalized without apparent localized onset). For analytic comparisons, seizure duration <2 hours was used as a baseline reference under the hypothesis that this group would have a better prognosis than those with longer duration of SE. Age was categorized in years as <1, 1 to 19, 20 to 64, and ≥65. The 1–19 age group was the reference under the hypothesis that this group would have the lowest mortality rate. For assessment of risk by seizure type, we used generalized tonic-clonic seizures as the referent.
Results.
There were 145 thirty-day survivors (72 males/73 females) of an original group of 184 individuals with incident afebrile SE.11 The largest proportion of patients was in the oldest age group (n = 57; ≥65, 39%). Only 13% (n = 19) were younger than 1 year of age (table 1). Sixty-two deaths occurred in the next 10 years—27 among the acute symptomatic group (n = 66), 12 among the progressive symptomatic group (n = 16), 15 among the remote symptomatic group (n = 35), and 8 among the idiopathic group (n = 28) (see table 1).
Distribution of deaths at 10 years among survivors of first 30 days after status epilepticus by age, sex, seizure type, and duration across etiologies
SMR (external comparison with the general population).
Subjects with SE who survived the first 30 days had higher mortality rates compared with that expected in the age, sex, time period standardized Minnesota population (SMR = 2.8; 95% CI, 2.1 to 3.5). The SMR was significantly elevated in all groups with symptomatic SE (acute, remote, progressive). Subjects with no clearly identified cause of SE (idiopathic/cryptogenic SE) demonstrated no difference in mortality rate when compared with that expected in the general population (SMR, 1.1; 95% CI, 0.5 to 2.3) (table 2).
Standardized mortality ratio (SMR) at 10 years by etiology after incident status epilepticus
Most deaths occurred in those ≥65 years old (76%). We dichotomized the analysis into those age ≥65 vs those <65 years old. Compared with the expected mortality rate in the elderly general population of Minnesota ≥65 years old, those with SE were 2.2 times more likely to die (95% CI, 1.6 to 2.9). Among the subjects <65 years old, the SMR was higher (SMR, 5.1; 95% CI, 2.8 to 8.0).
Prognostic factors (internal comparison within the cohort).
Within the cohort of patients with SE, we assessed the contribution of various prognostic factors to long-term mortality. Kaplan–Meier curves demonstrated a difference in predicted long-term mortality across etiologic subgroups (log rank test, 9.66; p = 0.02) (figure). The group of progressive SE shows the worst survivorship: 76% mortality rate by 10 years. When tested independently against each of the other groups, predicted survivorship in the progressive symptomatic group was significantly less that that of the idiopathic, the remote, and acute symptomatic patient group. Survivorship was similar in the acute symptomatic (59%) and the remote symptomatic groups (57%) at 10 years. Subjects with idiopathic/cryptogenic SE have the best survivorship at 10 years (71%) (see table 1). These three groups were not significantly different from one another when dichotomous comparisons were made.
Figure. Ten-year Kaplan–Meier curves for 30-day survivors after incident status epilepticus by etiology. × = idiopathic/cryptogenic; filled circle = remote; solid line = progressive; broken line = acute.
The effect of etiology, sex, duration, and SE seizure type on long-term mortality was evaluated in a univariate and multivariate analysis using the Cox proportional hazards model (table 3).
Multivariate analysis of determinants of long-term (10 years) mortality within a cohort of subjects after an incident episode of status epilepticus (internal comparison)
Age, duration, seizure type, and etiology each significantly contributed to mortality in the multivariate analysis. When compared with the 1–19 year age group, mortality was increased among those between 20 and 64 years old (RR = 13.3; 95% CI: 1.7 to 103) and those ≥65 years old (RR = 67; 95% CI, 8.9 to 503). Using generalized SE as the reference group in the model, only those with myoclonic SE showed a significant increase in mortality (RR = 4.0; 95% CI, 1.3 to 13). Using those with idiopathic/cryptogenic SE as a referent, only those with acute symptomatic SE demonstrated a significant increase in mortality (RR = 2.2; 95% CI, 1.0 to 5.1).
When compared with subjects with SE lasting <2 hours, SE >24 hours in duration was associated with increased mortality (RR = 2.3; 95% CI, 1.1 to 5.1). To detect a possible interaction between duration and etiology, we stratified the analysis by etiology, separating acute symptomatic from unprovoked (idiopathic, remote symptomatic, and idiopathic/cryptogenic). The mortality rate increased fivefold for those in the acute symptomatic group with SE lasting >24 hours (RR = 5.1; 95% CI, 1.1 to 23.4). The mortality rate was not increased among those with unprovoked SE >24 hours in duration (RR = 1.0; 95% CI: 0.7 to 2.3).
Discussion.
Long-term mortality after SE has not been evaluated previously; rather, mortality after SE has been restricted to clinical studies terminating follow-up evaluation at hospital discharge1-3,5-10⇓⇓⇓⇓⇓⇓⇓⇓ and two population-based studies terminating follow-up evaluation at 30 days.4,11⇓ We demonstrate an increased long-term mortality after a first episode of SE.
More than 40% of subjects who survived the first 30 days after SE died in the next 10 years, a death rate almost 3-fold greater than that expected in the general population matched for age, sex, and time period. Within etiologic subgroups, the highest SMR was present in the acute symptomatic group, an almost fourfold increase over the expected. Because there are no published studies on long-term mortality among people with acute symptomatic seizures, a comparison cannot be made. The mortality rate in the etiologic subgroups of unprovoked SE (progressive, remote symptomatic, idiopathic/cryptogenic) was similar to the mortality rate reported for similar etiologic subgroups within incident cohorts of epilepsy.18-21⇓⇓⇓ In the group of SE without a known cause (idiopathic/cryptogenic), the mortality rate was similar to that of the general population (SMR = 1.1). This suggests that SE by itself, without any identifiable underlying cause, does not alter the long-term mortality.
Excess mortality was present both in the adults and in the elderly. The case fatality was much greater among the elderly (82% of subjects ≥65 years old). This represented a twofold increase in the risk for mortality compared with similarly aged individuals in the general population who also have a high mortality rate. In contrast, the case fatality of subjects <65 years old was lower (17%), but the mortality rate of this group was 5-fold greater than expected in the general population. This paradox is related to the low mortality rate of the younger age group.
The etiology of SE has been considered an important risk factor for survival. The Kaplan–Meier curve is a useful representation of cumulative mortality of each etiologic group but can be misleading in terms of definition of risk factors because it does not take into account potential confounding factors like age. In the group of progressive SE, 75% died within 10 years and the survival rate is significantly different from the idiopathic/cryptogenic SE in which only 29% died. When we use the multivariate approach with Cox modeling, only those with acute symptomatic SE show a significant difference from the idiopathic/cryptogenic group. There was no significant increase in the mortality rate of those with progressive SE or remote symptomatic SE when compared with those with idiopathic/cryptogenic SE. This lack of difference may be a reflection of the power of the study.
The cumulative incidence of death predicted by the Kaplan–Meier curves reflects the impact of etiologic subgroups and the age structure of the groups in question. The increased mortality in the progressive and remote symptomatic groups is related to the underlying conditions such as stroke or neoplasm that occur most frequently in the oldest age group and are associated with markedly reduced life span after onset. In multivariate models, the symptomatic groups (progressive, remote symptomatic, and acute) demonstrate a twofold increase in risk for death when compared with the idiopathic/cryptogenic group.
We have separately categorized those with generalized convulsive SE into those with secondary generalized SE and those with generalized SE without evidence of focality. This distinction was not important for the long-term mortality of SE. We confirm the malignancy of myoclonic SE observed in previous studies.20 Myoclonic SE was associated with a more than fourfold increased risk for long-term mortality when compared with generalized convulsive SE. This poor long-term prognosis is largely attributable to the underlying cause of myoclonic SE, in all cases due to anoxic encephalopathy after cardiac arrest.22 Future studies should categorize the myoclonic group separately for the evaluation of prognostic factors.
Studies of the effect of duration of SE on short-term mortality are inconsistent. Duration was a predictor of short-term mortality in a population-based study in Richmond but not in our study in Rochester after adjusting for other risk factors.4,5,11⇓⇓ In the current analysis, long-term mortality is increased in those with long duration of SE (>24 hours) after adjusting for other risk factors. After stratification by etiologic subgroups, the increased risk in mortality associated with longer duration was restricted to those with acute symptomatic SE. In the unprovoked SE, prolonged SE was not associated with increased mortality. A study limited to children also found long duration of SE to be associated with increased mortality in acute symptomatic SE and that duration did not influence the mortality rate in those with unprovoked SE.23 Duration of SE, in the context of an acute insult, may be a surrogate for the severity of the underlying cause of SE.
There were other differences in predictors of short-term and long-term mortality in the current cohort study. After adjusting for other factors, males had a higher 30-day mortality than females, but there are no differences related to sex in the long-term mortality rate. Short-term mortality occurred predominantly among those with acute symptomatic SE. Long-term mortality was greatest among those with progressive symptomatic SE, although the SMR was greatest for those with acute symptomatic SE. The mortality was greatest in the elderly and those with myoclonic SE for both short-term and long-term mortality.
This is the first study of long-term mortality after SE. Age, long duration, and the presence of myoclonic SE should be considered as indicators of poor outcome for long-term survival after an incident episode of SE. The absolute mortality rate of subjects by 10 years after a first episode of SE is high, but there is no increase in mortality in the subgroup without an underlying cause of SE (idiopathic/cryptogenic) compared with the mortality rate of the general population. Based on this observation, we conclude that SE alone does not modify the long-term mortality.
Acknowledgments
Supported by National Institute of Neurological Disorders and Stroke Grant No. NS 16308.
Footnotes
-
See also page 515
- Received May 15, 2000.
- Accepted November 16, 2001.
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.
