Status epilepticus (SE) is a major medical and neurologic emergency that is associated with a significant mortality (1–3). The incidence of SE is estimated based mainly on three population-based studies conducted in the United States (4,5) and Switzerland (6,7) in sociologically and ethnically distinct populations. So far, no population-based data for Germany have been reported. We prospectively examined the population-based, crude, and the corrected, age-adjusted, age- and sex-specific incidences of a first episode of SE in adults in the state of Hessen, Germany.
Summary: Purpose: To determine the incidence and case-fatality rate of status epilepticus (SE) in adults in Hessen, Germany, we performed a prospective, population-based study from July 1997 through June 1999.
Methods: All adult patients residing within the zip-code area 35 (area-35) with SE were included. Area-35 had 743.285 adult inhabitants, including 123.353 adult inhabitants of the primary service area of the University Hospital Marburg (PS-area). Patients were reported by 16 hospitals in the area and were prospectively identified and carefully reviewed within 5 days by one of the authors. Based on the crude annual incidence of SE and a rate of underascertainment of 10% determined for the PS-area, the corrected, age-adjusted incidence of SE in area 35, more representative of the population of Germany, was calculated.
Results: The crude annual incidence in the PS-area was 15.8/100,000 [95% confidence interval (CI), 11.2–21.6]. The calculated, corrected, age-adjusted incidence of SE in area 35 was 17.1/100,000. It was higher for men compared with women (26.1 vs. 13.7) and for those aged 60 years and older (54.5 vs. 4.2/100,000, p < 0.0001). The etiology was mainly remote symptomatic due to cerebrovascular disease. Epilepsy was previously diagnosed in only 50% of the patients. The case-fatality rate was 9.3%.
Conclusions: Based on our data, at least 14,000 patients would be affected by SE in Germany, associated with ∼1,300 deaths annually. The incidence of SE in Germany is similar to that found in the white United States population. Furthermore, this study confirms the higher incidence of SE in male patients and in the elderly population. This may be due to a higher incidence of cerebrovascular disease in these subpopulations.
PATIENTS AND METHODS
Between July 1, 1997, and June 30, 1999, we collected data on all episodes of SE reported within and around the primary service area of the University Hospital Marburg (PS-area) with 123,353 adult residents. The PS-area included the university city of Marburg with some 20,000 students. Therefore, the population structure was atypical for Germany, with a lower average age (23% of the population, 60 years or older; 51.7% women). The PS-area lay within the zip-code area 35 (area-35, 27% of the population 60 years or older; 51.4 % women), a part of middle Hessen with 743,285 adult residents and with a population structure representative for Germany (27.7% of the population, 60 years or older; 51.8% women). To allow comparison of results, we compared the age distribution of our study population with the U.S. population studied by Hesdorffer et al. (5) and found a similar age distribution as in the PS-area (21% in the PS-area as compared with 23% of the U.S. population 60 years or older).
Definition of SE
To allow comparison of results, SE was defined, as in previous epidemiologic studies, as a single clinical seizure lasting >30 min or repeated seizures over a period of >30 min without intervening recovery of consciousness (3–9). All patients alive >30 days after onset of SE were regarded as survivors.
Case ascertainment and data management
To ascertain all SE episodes in the PS-area as completely as possible, all 71 hospitals covering the investigation area and the neighboring hospitals within 20 km of the borders of area-35 were regularly informed about the study and asked to participate. We included only adults of 18 years or older, because a network of cooperation between departments of neurology and internal medicine taking care of adults with SE in the region already existed, and because legal reasons made it more difficult to obtain consent for the use of data regarding minors. Patients were reported by neurologists, intensive-care unit, and emergency department nurses and physicians by using prepared fax forms or by telephone. Cases were evaluated within 5 days by a member of the Status Epilepticus Study Group Hessen (SESGH) from Marburg, using a standardized data-entry form. Data regarding type, duration, clinical features, seizure semiology, therapy, and etiology of SE as well as medication, concomitant diseases, 30-day outcome, results of diagnostic procedures such as ictal and interictal EEG, magnetic resonance imaging (MRI), computed tomography (CT), blood tests, demographic data, and the medical history of the patient were collected. All reported cases (n = 150) were used to determine the etiology, recurrence rate, SE type, and the case-fatality rate. Only those patients living in the PS-area were included for calculating the corrected, age-adjusted population-based incidences for area 35, which was more representative of the German population. The rate of underascertainment was determined for the PS-area by computer-based retrospective review of discharge letters from the study period plus 2 months, searching for the terms “status,”“seizure series,”“seizure,” and “epilepsy,” and by careful review of all charts thus identified. For statistical analysis, only the first episode of SE was taken into account.
Patients who did not meet the definition of SE with seizures lasting <30 min, patients with a nonprecise clinical description, or patients with uncertain or incomplete medical history, and those who did not reside within the area of study were excluded.
Classification of status epilepticus
SE was classified in accordance with the seizure classification of the International League Against Epilepsy as primary generalized (tonic, clonic, tonic–clonic, and myoclonic), secondarily generalized, complex partial, simple partial, and absence status (10). If no clearly ictal symptoms (i.e., an epileptic clonus) were present, SE was diagnosed only if the EEG showed electrographic seizure activity lasting ≥30 min (11). Two EEG interpreters, at least one of them board certified (ABCN) had to agree that the EEG pattern represented seizure activity. The classification was performed by an epileptologist (F.R., H.M.H.) based on clinical features and, if available, on the ictal EEG.
Etiology was defined as the preceding factor that was assumed to cause SE. If more than one factor was assumed to cause SE, up to four etiologies could be registered in each patient. Etiologies were classified according to Hauser as remote symptomatic, acute symptomatic, and unknown (9). They were classified as acute symptomatic if the etiologic factor was causing SE within 7 days (i.e., acute stroke). If SE occurred without a precipitating acute brain insult but in close relationship to an underlying condition, occurring >7 days ago, it was classified as remote symptomatic. If SE occurred in the absence of any identified precipitating factor, the etiology was classified as unknown.
Information from the national census bureau for 1998 was used as the denominator for the total adult population counts. The influence of sex and age of 60 years or older on the likelihood to develop SE was investigated with the χ2 test. The crude average annual incidence rate for the PS-area was obtained by including all first-time cases of SE in this area in the numerator and by dividing them by the number of all adults in the PS-area. In addition, the Clopper Pearson′s confidence interval for the 1-year incidence rate in the PS-area (rPS ) was calculated.
Based on these data and the rate of underascertainment in the PS-area (10%), the corrected, age-adjusted, age- and sex-specific incidence rates for area-35 were calculated. This was done to obtain incidence rates more representative for Germany. The following formula to calculate the corrected, age-adjusted incidence for area-35 was used:
a for age group; g for gender; n for total number; k for correction factor for the rate of underascertainment in the PS-area; r a,g,area35 for calculated, age-adjusted incidence rate in the area 35; SEC a,g,PS for first cases of SE in the 2-year study period in the PS-area; r a,g,PS for age- and gender-specific average annual incidence rate for the PS-area, considering the 10% rate of underascertainment.
As the incidence of SE is age dependent, age adjustment was necessary because the population of the PS-area was younger than that of area-35. Gender adjustment was not necessary, because the sex distribution was similar in both areas (51.7% vs. 51.4% women). This calculation process was proven to be stable by an internal validation within the PS-area. This was performed by dividing the PS-area into two different regions: A and B. By using the formula cited earlier, the incidence for region B was estimated based on the data of region A. The result of this calculation was very similar to the crude incidence of region B observed (14.2 vs. 14.7/100,000). The age- and gender-specific rates were obtained accordingly by performing the calculation for women and men and for patients younger than 60 and 60 years or older separately.
We were also interested in the case ascertainment outside the primary service area of the tertiary care medical facility conducting the study. Because the rate of underascertainment for the rest of area 35 outside the PS area (area 35 – the PS area) could not be determined, it was estimated by relating the crude incidence in this area to the calculated corrected age-adjusted incidence in area 35.
During the study period, 150 adult patients with an episode of SE were reported by 16 hospitals in and around the PS-area. About half of the study population were men (46%) with a mean age of 65 years (range, 19–94 years). Twenty-eight patients (19%) were younger than 60 years. The mean duration of SE was 18.6 h (range, 0.5–430 h). Twenty-three (13.3%) recurrences occurred in 20 of the 150 patients during the observation period. Nearly all of the patients having recurrent episodes of SE have had epilepsy before (93.8%). SE type was mainly partial (76%), with (19%) or without (57%) secondary generalization (Table 1). In the majority of cases (74%), the etiology was a remote or an acute brain insult (Table 2). Fifty percent of all 150 patients had a history of epilepsy at the time of the SE, 33% of the 39 patients residing in the PS-area, and 55.9% of those living outside the PS-area. As 14 of 150 patients died within 30 days of onset of SE, the case-fatality rate was 9.3%. The etiology was acute in seven of those patients (intracerebral hemorrhage, acute ischemic stroke, hypoxia after cardiac arrest, renal failure), remote in six patients (stroke, angioma, brain tumor), and unknown in one.
|Type of status epilepticus||Frequency|
|Simple partial||20 (13.3%)|
|Complex partial||65 (43.3%)|
|Secondarily generalized||29 (19.3%)|
|Primary generalized||21 (14.0%)|
|Absence status||9 (6.0%)|
|Etiologic factor||Frequency |
[n (% of 150)]
|Remote stroke||54 (36.0%)|
|Other remote symptomatic||40 (26.7%)|
|Binswanger's disease||25 (16.7%)|
|Acute stroke (ischemic and hemorrhage)||21 (14.0%)|
|Medication induced||16 (10.7%)|
|Metabolic disorders||13 (8.7%)|
|Low antiepileptic drug level||13 (8.7%)|
|Trauma (remote)||11 (7.3%)|
Of the 150 patients identified, 95 lived within the area-35. Thirty-nine of these 95 patients resided within the PS-area. The crude average annual incidence of SE for the PS-area was 15.8/100,000 adult residents (95% CI, 11.2–21.6/100,000). It was significantly higher in the population 60 years or older (p < 0.0001), whereas the gender difference was not significant (p = 0.75). The rate of underascertainment for the PS-area determined retrospectively was 10%. Based on these data from the PS-area, the calculated age-adjusted annual incidence of SE for adults in area-35 was 17.1/100,000. It was higher in the population of aged 60 years or older (54.5/100,000) than in younger adults (4.2/100,000) and in the male as compared with the female adult population (26.1 vs. 13.7/100,000) (Fig. 1).
To estimate the rate of underascertainment in area 35 outside the PS area (area 35 – the PS area), the crude average annual incidence of SE in adults was found to be 4.5/100,000 (i.e., 26.4 % of the calculated corrected age adjusted incidence for area 35). This suggests a rate of underascertainment in area 35 outside the PS-area of ∼73.6%, similar to the rate estimated by DeLorenzo et al. (4).
The corrected, age-adjusted incidence of first episodes of SE in adults in zip-code area 35, Hessen, Germany, was 17.1/100,000. As this area has an age and sex distribution representative of Germany, we would expect at least 14,000 new cases of SE in Germany, associated with ∼1,300 deaths annually. The incidence of SE was age and sex dependent, with higher incidences in adults older than 60 years (54.5 vs. 4.2/100,000) and in men (26.1 vs. 13.7/100,000). In a retrospective study in Rochester, Minnesota, the primary service area of the Mayo Clinic, Hesdorffer et al. (5) reported very similar results. They calculated an age-adjusted incidence of 18.3/100,000. The incidence was 19.8/100,000 in the age group younger than 20 years and 17.4/100,000 among adults 20 years or older. The annual incidence was higher in those of older than 60 years (62.5/100,000) than for younger adults (5.3/100,000). The population of Rochester is 96% white and 4% nonwhite, has a similar age distribution, and is, therefore, comparable to the German population, which may explain the similarity of the data.
In a prospective, population-based study in Richmond, Virginia, U.S.A., DeLorenzo et al. (4) obtained a total annual incidence of 41/100,000. However, the incidence for the white population was 20/100,000, including infants and children older than 1 month. As in this study, the underascertainment in the primary service area of the hospital initiating and conducting the study was determined to be 10%, whereas the rate of underascertainment in community hospitals outside the primary service area was found to be 67%. These data suggest that case ascertainment heavily depends on the presence of specialized medical personal with an interest and expertise in SE and appropriate diagnostic equipment such as the availability of EEG 24 h/day.
There are several other sources for underascertainment of SE. Nonconvulsive SE, especially in the setting of acute neurologic illnesses and when no EEG is available, is frequently underdiagnosed (11,12). Some cases of SE are treated by family doctors or paramedics and are not admitted to hospitals; others die before the diagnosis is made or before they reach the hospital. Furthermore, when a seizure duration of >30 min is not clearly documented, patients are excluded from the analysis. For these reasons, incidences reported in the literature should be regarded as minimal incidences, and efforts should be made to determine sources and rates of underascertainment.
Recently an annual incidence of 9.9/100,000 for the French-speaking Switzerland including the Geneva area was reported (6). The study was initiated and conducted by a tertiary care academic hospital in Geneva. The age-adjusted annual incidence for the Geneva area was 16.3/100,000 (6,7). In the Rochester study, patients with postanoxic myoclonic encephalopathies with an age-adjusted incidence of 1.9/100,000 were included (5). These patients were excluded in the Swiss study, which may explain the somewhat lower incidence found (7). Outside the Canton of Geneva, especially in the more rural areas, age-adjusted incidence rates were consistently lower, between 7.0 and 10.9/100,000. Rates of underascertainment were not reported. The high rates of underascertainment outside the primary service areas of the tertiary care academic hospitals conducting the epidemiologic study found by DeLorenzo and in the current study suggest that the lower incidences found in the rest of the French-speaking Switzerland may have been due to underascertainment (4–7). As the population of more rural areas tends to be older, a higher instead of a lower incidence of SE would be expected there.
The percentage of patients with previously diagnosed epilepsy was lower in the PS area (33%) with a low rate of underascertainment as compared with that outside the PS-area (55.9%), with a high rate of underascertainment. Similar results were reported by the Swiss group, who reported 32.8% of patients with a history of epilepsy in the Canton of Geneva (7) and 48.6% outside the Canton of Geneva (6,7). As a SE is more likely to be diagnosed if a history of epilepsy is present, these differences likely reflect a selection bias, leading to higher proportions of patients with a history of epilepsy in areas with higher rates of underascertainment.
In all four studies, the incidence of SE was age dependent and higher in the older population. Taking the continued aging of the populations into account, the incidence of SE in these countries is likely to increase in the future. This study also confirms the higher incidence of SE in men as compared with women, previously determined by two studies (5,6). Remote cerebrovascular disease was the most frequent etiology of SE in this study. Therefore, the higher incidence in men and the older population likely relates to the overproportionate frequency of cerebrovascular disease in these subpopulations (5,13). Similar to other studies, the main portion of SE was partial or partial with secondary generalization (4–7). In comparison with these studies, the percentage of partial SE or partial SE with secondary generalization is the highest in this study (76%). This can be explained by the fact that we included adults only. In this age group, most of the SE are symptomatic because of focal acute and remote cerebral lesions causing partial SE (4–7,13). DeLorenzo et al. (4) reported that >70% of SE adults but only 50% of SE children had partial SE as the initial seizure type leading to SE, and Hesdorffer et al. (5) also reported a higher percentage of partial SE with (15%) or without (46%) secondary generalization in adults as compared with patients younger than 20 years. Therefore, age correlates not only with the frequency of SE but also with status type.
We conclude that the minimal age-adjusted incidence of SE in the white adult population in Germany and the United States lies similarly between 17 and 20/100,000. Lower rates found outside the primary service areas of the academic hospitals conducting the studies are usually due to underascertainment. The status type is dependent on etiology and age. This study confirms, that SE is more frequent in men and the elderly and is associated with a significant case fatality. Because of the aging of the population, the incidence of SE will increase in the future.
Institutions participating in the Status Epilepticus Study Group Hessen (SESGH) and their representatives are listed in alphabetical order.
M. Adelmann, M.D., Neurologische Klinik, Klinikum Weilmünster, Weilmünster; M. Auf dem Brinke, M.D., Neurologische Klinik Westend, Bad Wildungen; H. Bauerle, M.D., Klinik für Innere Medizin, Burgfeld-Krankenhaus Kassel, Kassel; H. Begenat, M.D., Abteilung für Innere Medizin, Kreiskrankenhaus Ziegenhain, Krankenhausbetrieb des Schwalm-Eder-Kreises, Ziegenhain; K. Boehm, M.D., Neurologische Klinik Braunfels, Braunfels; A. Ferbert, M.D., Neurologische Klinik, Klinikum Kassel, Kassel; L. Gerlach, M.D., Neurologische Akutklinik Bad Zwesten, Bad Zwesten; T. Kaltenbach, M.D., Klinik für Innere Medizin, Kreiskrankenhaus Alsfeld, Alsfeld; H-D. Langohr, M.D., Klinik für Neurologie, Städtisches Klinikum Fulda, Fulda; J. Meier, M.D., Abteilung Innere Medizin, Kreiskrankenhaus Bürgerhospital Friedberg, Friedberg; G. Mayer, M.D., Hephata-Klinik, Schwalmstadt-Treysa; M. von Reutern, M.D., Asklepios-Neurologische Klinik, Nidda-Bad Salzhausen; F. Rosenow, M.D., Klinik für Neurologie, Klinikum der Philipps-Universität Marburg, Marburg; P. Schneider, M.D., Abteilung für Neurologie, St.-Katharinen-Krankenhaus, Frankfurt; A. Simonow, M.D., Arzt für Neurologie, Herborn; A. Wirbatz, M.D., Klinik für Neurologie, Klinikum der Justus-Liebig-Universität Giessen, Giessen (all in Hessen, Germany).
Acknowledgment: This study was supported by grants by GlaxoWellcome, Desitin, and the Verein zur Erforschung der Epidemiologie der Epilepsien e. V. One of the authors (F.R.) is supported by the Ulran-Foundation-Professorship for Neurology/Epileptology, Germany. We are indebted to the Statistische Bundesamt, Berlin, for supplying demographic data regarding the population of Germany.