Abstract
Background: Few population-based studies of status epilepticus have been performed in the United States.
Objective: To determine the incidence, case fatality, and demographics of generalized convulsive status epilepticus (GCSE) in the state of California.
Methods: Using a state-wide hospital discharge database, the authors identified all hospitalizations from 1991 through 1998 with a discharge diagnosis of convulsive status epilepticus. They identified the first admission for each individual to estimate the incidence of GCSE. In-hospital case fatality rates were calculated, and multivariate analysis was performed to determine predictors of death during hospitalization. Secondary diagnoses were analyzed by retrieving all discharge diagnoses accompanying the diagnosis of GCSE.
Results: The incidence rate of GCSE was 6.2/100,000 population and fell by 42% between the years 1991 and 1998 from 8.5 to 4.9/100,000. The rate of GCSE was highest among children under the age of 5 (7.5/100,000) and among the elderly (22.3/100,000). Blacks also demonstrated a relatively high incidence of GCSE (13.4/100,000). The case fatality for incident admissions was 10.7%, with increasing age being the only significant predictor in multivariate analysis. Case fatality was highest in patients who also carried a diagnosis of anoxia, CNS infection, or stroke.
Conclusions: The incidence of GCSE requiring hospitalization has fallen over the last decade and is lower than that reported in previous studies. The case fatality is also lower than that reported previously. Further studies are needed to determine the cause of this decline in incidence and mortality of GCSE.
Status epilepticus is defined as continuous seizure activity lasting at least 30 minutes or two or more sequential seizures without full recovery of consciousness between episodes.1 Although numerous authors have addressed the acute management and treatment of status epilepticus,1-8⇓⇓⇓⇓⇓⇓⇓ the epidemiology of status epilepticus has been less well studied. Population-based studies of status epilepticus include a prospective study in Richmond, VA,9 a retrospective study in Rochester, MN,10,11⇓ and two recent studies in Europe.12,13⇓ The incidence of status epilepticus in these reports varied considerably, from 10 to 61/100,000, with case fatalities ranging from 7.6 to 22%. The only study that evaluated time trends in the incidence of status epilepticus found an increase in incidence between the mid-1940s and mid-1980s.10,11⇓ However, no subsequent studies have been performed to determine whether the incidence of status epilepticus has continued to increase over time.
We conducted a retrospective evaluation of all inpatient hospitalizations for generalized convulsive status epilepticus (GCSE) in the state of California from 1991 through 1998. We report the incidence and case fatality of hospital admission for GCSE stratified by sex, age, ethnicity, and insurance type.
Methods.
The Office of Statewide Health Planning and Development (OSHPD) of California maintains a database of discharge abstracts for all admissions to nonfederal hospitals. Nonfederal hospitals constitute 571 of all 594 hospitals in the state. Hospitals began contributing cases in January 1990, with complete participation after June 1990. To identify admissions for GCSE, we searched the OSHPD hospital discharge database from January 1991 through December 1998 for patients with a diagnosis of grand mal status (International Classification of Diseases–9–Clinical Modification [ICD-9-CM] code 345.3), which is defined as “extended convulsive seizures with tension of limbs (tonic) and/or rhythmic contractions (clonic).”14 To estimate the incidence rate of first events, we used a unique patient identifier to determine the first date that each individual was admitted during the study years with a diagnosis of GCSE. Patients whose zip code did not indicate residence in the state of California were excluded. The remaining first admissions defined the study cohort. Annual incidence of GCSE hospitalizations was calculated as the total number of first admissions for the diagnosis of GCSE divided by the population of California for that year. California population figures were geometrically interpolated from published 1990 and 2000 US Census data.
To evaluate the possibility that variation in hospital discharge coding practices over time may have affected our incidence calculations, we analyzed a variety of other discharge diagnoses that are related to GCSE and that could be confused with this diagnosis during the coding process. These diagnoses included generalized convulsive epilepsy (ICD-9-CM 345.10), partial epilepsy with impairment of consciousness (ICD-9-CM 345.40 and 345.41), epilepsy unspecified (ICD-9-CM 345.90 and 345.91), and convulsion (ICD-9-CM 780.3). We calculated annual incidence rates for each of these diagnoses in the same manner as described above. We also determined annual incidences for GCSE when this diagnosis was listed as the primary discharge diagnosis.
For the study cohort, we collected demographic information including sex, age, ethnicity, and insurance status. Insurance status was coded as three categories: 1) Medicare; 2) private/health maintenance organization (HMO), which also included preferred provider organization and CHAMPUS; and 3) indigent, which included Medicaid, no insurance, self-pay, charity, county indigent, no charge, and other nongovernmental entities such as Easter Seals. Incidence of GCSE was calculated for sex, age, and ethnic categories, with population denominators interpolated from US Census data for the state of California.
We identified in-hospital deaths to determine case fatality. Case fatality rates stratified by demographic characteristics were compared in univariate analyses by calculating relative risks and 95% CI using the Cornfield method.15 Multivariate relative risks and 95% CI were calculated by including patient age, sex, ethnicity, and insurance type in a logistic regression model. The change in mortality during the 8-year study period was assessed using an analysis of trend based on linear regression.16
Although a detailed analysis of underlying causes of status epilepticus was beyond the scope of this study, we evaluated all major medical conditions that appeared as accompanying discharge diagnoses in >0.5% of the patients, in an attempt to characterize the conditions that were associated with GCSE. Discharge diagnoses potentially related to etiology of GCSE were grouped into the following categories: anoxia (ICD-9-CM 348.1, 427.41, 427.5), brain tumor (ICD-9-CM 191.0 to 191.9, 198.3), stroke (ICD-9-CM 431, 433.1, 433.3, 434.0, 434.1, 434.9), late effects of stroke/brain injury (ICD-9-CM 438, 438.20, 438.9, 907.0), alcoholism (ICD-9-CM 291.0, 291.8, 303.0, 303.9, 305.0), hydrocephalus (ICD-9-CM 331.4), subdural hematoma (ICD-9-CM 432.1), sodium imbalance (ICD-9-CM 276.0, 276.1), noncompliance (ICD-9-CM V1581), MS (ICD-9-CM 340), CNS anomaly (ICD-9-CM 742.0 to 742.4, 742.9), cerebral palsy (ICD-9-CM 343.0, 343.1, 343.2, 343.4, 343.8, 343.9), developmental delay (ICD-9-CM 315.3, 315.4, 315.5, 315.8, 315.9, 317, 318.0, 318.1, 318.2, 319, 783.4), CNS infection (ICD-9-CM 320.0 to 320.3, 320.8, 320.9, 321.0 to 320.2, 321.8, 322.9, 323, 323.9, 324.0), and fever (ICD-9-CM 780.6). Case fatality rates for GCSE associated with each of these underlying conditions were calculated.
The readmission rate for patients with a primary diagnosis of GCSE was determined by calculating the number of people who were readmitted within 1 year for the same primary diagnosis. Readmission rates were not calculated for patients first admitted in 1998 because data from the subsequent year were not available. The total number of readmissions for each individual was also determined during the 8-year period.
All statistical analyses were performed with the Stata statistical package (version 6.0; Stata Corp., College Station, TX).
Results.
Incidence.
From 1991 through 1998, there were 19,491 admissions for GCSE in 481 hospitals, representing 15,882 individuals. Patients not residing in the state of California were excluded (n = 281), leaving a total of 15,601 individuals. The incidence of first GCSE hospitalization was 6.18/100,000 over the entire study period, and the case fatality during first hospitalizations for GCSE was 10.7%. The annual incidence of GCSE decreased by 42% between the years 1991 and 1998 (figure 1), from 8.5 to 4.9/100,000 (p < 0.001).
Figure 1. Incidence of generalized convulsive status epilepticus hospitalizations in California, 1991 to 1998.
We investigated the possibility that incidence rates for GCSE hospitalizations declined because of changes in coding practices. It is possible that hospitalizations for GCSE have been increasingly miscoded under different but related seizure and epilepsy ICD-9-CM discharge codes. However, the incidence of hospitalizations for generalized convulsive epilepsy decreased from 6.3 to 2.0/100,000 between the years 1991 and 1998, and similar declines in incidences of hospitalizations for convulsion (23.1 to 20.5/100,000) and epilepsy unspecified (0.75 to 0.52/100,000) were recorded. Incidence of hospital admissions where GCSE was listed as the primary diagnosis also decreased steadily during the study years, from 5.1 to 3.2/100,000.
The incidence of GCSE varied by sex, age, and ethnicity (table 1). Men had a significantly higher incidence of GCSE than women, though the difference was small. The median age of individuals admitted with GCSE was 48 years and dropped between the years 1991 and 1998 from 52 to 45 years. A plot of incidence by age produced a J-shaped curve, revealing a high incidence of GCSE during the first few years of life as well as in the elderly (figure 2).
Incidence of convulsive status epilepticus requiring hospitalization in state of California during years 1991 to 1998
Figure 2. Age and incidence of generalized convulsive status epilepticus hospitalization in California, 1991 to 1998.
Compared with whites, blacks had a significantly higher incidence of GCSE, while Hispanics and Asians demonstrated a significantly lower risk of GCSE (see table 1). The decline in incidence of GCSE over the years was most notable in blacks (figure 3).
Figure 3. Trends in incidence of generalized convulsive status epilepticus hospitalization in California, 1991 to 1998, by ethnicity. Triangles = black; squares = white; diamonds = Hispanic; circles = Asian.
Case fatality.
The overall case fatality rate in our cohort was 10.7%. The case fatality rate for hospital admissions where GCSE was listed as the primary discharge diagnosis was much lower at 3.5%. Case fatality for all admissions associated with a diagnosis of GCSE remained unchanged over the years. However, case fatality among individuals with a primary discharge diagnosis of GCSE declined during the study years from 4.7 to 3.2% (p = 0.004), a decrease of 32%.
In multivariate analysis, age was the only significant predictor of in-hospital death following GCSE (table 2). Children under the age of 5 demonstrated the lowest case fatality (1.4%). When restricted to hospital admissions in which the primary diagnosis was GCSE, the case fatality rate in children under age 5 was even smaller (0.7%). Compared with children under the age of 5, adults in their twenties to fifties were almost six times as likely to die during hospitalization for GCSE, while individuals over the age of 75 had a 15-fold increased case fatality. Although Hispanics appeared to experience a lower case fatality than whites in univariate analysis, this difference may be explained by the fact that Hispanics tended to be younger in age than whites (Hispanic median age 31 vs white median age 53). When age and other factors were taken into account in the multivariate analysis, ethnicity no longer had a significant effect on mortality.
Case fatality associated with hospital discharge diagnosis of generalized convulsive status epilepticus
The median length of hospitalization for patients with GCSE who died during hospitalization was 7 days. Of the patients who died, 63% did so within 10 days of admission, and only 7% died after 30 days of hospital admission. During the study period, the mean length of hospital stay for GCSE decreased from 10.9 days in 1991 to 6.7 days in 1998 (p < 0.001).
Readmissions.
Of 9,779 incident admissions for a primary diagnosis of status epilepticus between the years 1991 and 1997, 1,075 were readmitted within 1 year with the same primary diagnosis, producing a 1-year readmission rate of 11.0%. This readmission rate remained constant over the study period. Mortality during readmissions was 1.4%. Over the entire study period, 5.9% of patients required multiple readmissions for a primary diagnosis of GCSE, ranging from 2 to 3 readmissions (4.6%) to 5 to 17 readmissions (0.7%).
Underlying conditions.
The most common major medical conditions that may have been related to the cause of GCSE were remote stroke or brain injury, developmental delay, sodium imbalance, alcoholism, anoxia, cerebral palsy, and noncompliance (table 3). Forty-eight percent of individuals identified as having a first-time admission for GCSE also carried 1 of the 16 diagnoses listed in table 3 that could have been related to the etiology of their prolonged seizures. This percentage did not change appreciably during the years of the study.
Frequency of major medical conditions potentially related to etiology of generalized convulsive status epilepticus, n = 15,882
Case fatality varied widely according to the presence of specific underlying conditions. The highest case fatality was seen in patients who were given a diagnosis of anoxia (64%) or CNS infection (32%), while the lowest case fatality was seen in association with noncompliance (2%), CNS anomalies (1.8%), and otitis media (0.5%).
Compared with whites, a higher percentage of GCSE cases in blacks were associated with a diagnosis of noncompliance (6.8 vs 3.0%; p < 0.00001) or alcohol abuse (10.0 vs 7.7%; p < 0.00001). Hispanics demonstrated a significantly higher percentage of cases associated with a diagnosis of developmental delay, cerebral palsy, or CNS anomalies when compared with whites and fewer cases associated with anoxia or late effects of stroke/brain injury, consistent with the fact that Hispanics tended to be younger than their white counterparts. Other differences in the frequency of underlying conditions between the separate ethnic groups existed but were much smaller in magnitude.
The percentage of GCSE cases associated with specific underlying conditions was also explored over time. Comparing the years 1991 and 1998, there was a decline in the percentage of GCSE cases associated with stroke (2.5 vs 0.43%) and late effects of stroke/brain injury (11.8 vs 7.0%). A concomitant increase in percentage of cases associated with developmental delay (6.9 vs 13.1%) and cerebral palsy (3.1 vs 6.1%) was seen. These differences were all highly significant and are consistent with the decline in median age of our study cohort over time.
Discussion.
We found that the overall incidence of first GCSE requiring hospitalization in the state of California was 6.2/100,000 and that this incidence fell steadily and dramatically by 42% between the years 1991 and 1998. Children under the age of 5 as well as blacks had the highest incidence of GCSE, while increasing age and the nature of underlying medical conditions were the main predictors of in-hospital death.
The incidence of GCSE in our study is lower than that previously reported in the United States. The incidence rate of GCSE, which comprised 44 to 74% of all cases of status epilepticus in previous studies, was 8.3/100,000 in Rochester, MN, and 29/100,000 in Richmond, VA.9,10⇓ The lower incidence of GCSE found in our study compared with previous reports persists when stratified by race. Blacks demonstrated an approximate GCSE incidence rate of 39.9/100,000 in Richmond, where they made up the majority of the population,9 compared with an incidence rate of 13/100,000 in the current study. Whites constituted 96% of the population in Rochester, where they demonstrated a similar incidence of GCSE to whites in Richmond and a higher incidence than that found in our study.
Our finding that the incidence of GCSE has steadily decreased over the last decade supports the notion that GCSE may be less common now than in the past. Although the incidence of motor status epilepticus increased between the 1940s and mid-1980s in Rochester, MN,11 our study is the first to evaluate incidence trends of GCSE in the 1990s and suggests a more recent decline. Recent proposals to change the definition of status epilepticus to include any continuous seizures lasting over 5 minutes17,18⇓ would be expected to increase the overall incidence of GCSE, making the apparent decline in incidence of GCSE over the past decade all the more significant.
The incidence of GCSE may be decreasing for a number of reasons. First, the decline in hospitalizations for GCSE may reflect improved outpatient treatment of acute repetitive seizures in patients with known epilepsy.19,20⇓ However, epilepsy accounts for less than half of all patients presenting in status epilepticus,9,13,21⇓⇓ and other medical conditions were common causes of GCSE in our study. Second, improved prehospital treatment of status epilepticus may also be partly responsible for the apparent decline in hospitalized cases of GCSE.18,22⇓ Third, it is possible that errors in hospital discharge coding may have introduced bias into our annual incidence estimates. However, any systematic error in coding of GCSE is unlikely to exert a differential effect on the incidence rate across the years, and incidence trends for other hospital discharge codes pertaining to epilepsy also suggest a decline over time.
Our study confirms the previously described bimodal age distribution of status epilepticus, with young children and elderly patients experiencing the highest incidence of GCSE.9,10,13⇓⇓ Numerous studies have shown that children experience a particularly high rate of status epilepticus9,10,13,23,24⇓⇓⇓⇓ and that 74% of children with status epilepticus are under the age of 5.25,26⇓ However, the mortality of status epilepticus in children is low and has decreased from 11% in the 1960s to 3.6% in more recent reports.23,24⇓ The childhood case fatality associated with GCSE in our study is the lowest reported to date (1.4%). The excellent prognosis of GCSE in children has been attributed at least in part to the high percentage of cases due to febrile status epilepticus, which carries a particularly low morbidity and mortality.26,27⇓
Elderly patients, conversely, experience both a high incidence and a high case fatality rate for GCSE. The high incidence is not surprising, given that many common causes of status epilepticus are diagnosed primarily in the elderly, including stroke, anoxia, and brain tumor.9,12⇓ In our study, the major medical condition most commonly associated with a diagnosis of GCSE was late effect of stroke or brain injury, and the median age of this group of patients was 69 years.
High case fatality among elderly patients can also be explained by the nature of the underlying brain insults causing the status epilepticus. Previous reports suggest that death following status epilepticus is usually attributable to the underlying cause rather than to the prolonged seizure itself.21,28⇓ In a multivariate analysis, Towne et al.29 found that only increasing age and anoxia were significant predictors of mortality. We also found a high case fatality for GCSE associated with anoxia as well as with other entities such as CNS infection and stroke.
Our study is subject to a number of limitations. First, our incidence rates represent an underestimate of the true incidence of GCSE in the community, since our cases are limited to those leading to an inpatient hospital admission. Of the other existing population-based studies, only the retrospective analysis of Hesdorffer et al.10 of status epilepticus in Rochester, MN, during the 1960 to 1980s included information from outpatient and home visits as well as from inpatient hospitalizations. Our study also excludes cases of GCSE treated at federal hospitals such as Veterans Affairs and military base hospitals, which again contributes to a mild underestimate of the population incidence. Second, our study is limited by potential errors in hospital discharge coding. It is possible that systematic and nonsystematic errors in the coding of GCSE have led to bias in our incidence and case fatality estimates, which may very well have also led to an underascertainment of cases. However, decreased reporting of status epilepticus over time is an unlikely explanation, since insurance reimbursement issues continue to exert pressure on hospitals to list status epilepticus as a discharge diagnosis when it is noted in the medical charts and since prolonged convulsive seizures are easy to diagnose and unlikely to be missed by medical practitioners. Third, our study lacks detailed information on seizure type and underlying etiologies. We are unable to distinguish between patients who had generalized-onset seizures from those who had partial seizures at the onset of their generalized convulsions. We are also unable to specify the exact cause of the prolonged convulsions in our patients.
Despite these limitations, our study has a number of interesting findings. The decreasing incidence of GCSE over the past decade is less likely to be influenced by the limitations stated above and suggests that we are either treating epilepsy better or perhaps treating the coexisting conditions more effectively. Our case fatality of 10.7% is half of that reported in previous population-based studies,9,10⇓ with the exception of a Swiss study that excluded patients with anoxia,13 suggesting that the prognosis of GCSE may have improved. Why the incidence of GCSE varies by ethnicity is not clear. Blacks were almost twice as likely as whites to have GCSE, a finding consistent with DeLorenzo et al.’s report9 of status epilepticus in Richmond, VA. Conversely, Hispanics and Asians exhibited a significantly lower incidence of GCSE than whites. Whether such ethnic differences are due to differences in the distribution of underlying causes, as suggested by the higher percentage of cases associated with noncompliance and alcohol abuse found in blacks than whites, or due to other differences such as varying levels of susceptibility to GCSE between ethnic groups cannot be determined by this study.
Acknowledgments
Supported in part by the Neurologic Sciences Academic Development Award (grant 5 K12 NSO1692-05).
Acknowledgment
The authors thank Donna M. Ferriero, MD, and Brian K. Alldredge, PharmD, for their careful reviews of the manuscript.
- Received August 21, 2001.
- Accepted December 13, 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.
You May Also be Interested in
Cutaneous α-Synuclein Signatures in Patients With Multiple System Atrophy and Parkinson Disease
Dr. Rizwan S. Akhtar and Dr. Sarah Brooker