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
1 Markus Leitinger and Eugen Trinka share the first authorship.
Markus Leitinger
Footnotes
1 Markus Leitinger and Eugen Trinka share the first authorship.
Affiliations
Department of Neurology, Christian Doppler University Hospital, Paracelsus Medical University Salzburg, AustriaCentre of Neuroscience, Christian Doppler University Hospital, Salzburg, Austria
1 Markus Leitinger and Eugen Trinka share the first authorship.
Eugen Trinka
Footnotes
1 Markus Leitinger and Eugen Trinka share the first authorship.
Affiliations
Department of Neurology, Christian Doppler University Hospital, Paracelsus Medical University Salzburg, AustriaCentre of Neuroscience, Christian Doppler University Hospital, Salzburg, AustriaDepartment of Public Health, Health Services Research and Health Technology Assessment, UMIT – University for Health Sciences, Medical Informatics nd Technology, Hall in Tirol, Austria
Department of Neurology, Christian Doppler University Hospital, Paracelsus Medical University Salzburg, AustriaDepartment of Mathematics, Paris-Lodron-University of Salzburg, Salzburg, Austria
Department of Neurology, Christian Doppler University Hospital, Paracelsus Medical University Salzburg, AustriaSheba Medical Center, Department of Neurology, Tel Hashomer, Israel
Department of Neurology, Christian Doppler University Hospital, Paracelsus Medical University Salzburg, AustriaCentre of Neuroscience, Christian Doppler University Hospital, Salzburg, Austria
Corresponding author at: Dept. of Public Health, Health Services Research and Health Technology Assessment, UMIT - University for Health Sciences, Medical Informatics and Technology, Eduard-Wallnoefer-Zentrum 1, A-6060 Hall i.T., Austria.
Department of Public Health, Health Services Research and Health Technology Assessment, UMIT – University for Health Sciences, Medical Informatics nd Technology, Hall in Tirol, AustriaCenter for Health Decision Science, Department of Health Policy and Management, Harvard T.H. Chan School of Public Health, Boston, MA, USAInstitute for Technology Assessment and Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
The annual incidence of status epilepticus (SE) in adults varies between 1.29 to 73.7/100,000 (81.1 SE episodes/100,000).
•
Clarification of terms is essential, e.g. SE patients or SE episodes, first ever SE or recurrent SE, and the term “per 100,000”.
•
Incidences in adults and children should be reported and adjusted separately to avoid bias.
•
Choice of reference population (RP) is crucial as study data will change as if the study was performed in that RP.
•
ILAE 2015 diagnostic time criteria T1 for SE are recommended as different types of semiology qualify at different times.
Abstract
Objective
Status epilepticus (SE) is a severe neurologic condition associated with high morbidity and mortality. Population-based studies in adults have found a wide range of incidences in various regions in the world. Although the incidence of SE increases almost exponentially in the elderly, data on census-based population statistics in these studies are scarce. This study provides a critical review with an emphasis on census-based population statistics and study characteristics in adults.
Methods
We performed a systematic search of population-based studies on SE in adults in PubMed using “status epilepticus” in combination with “epidemiology”, “population”, and “incidence” as search terms, and also screened references. For each identified study, we assessed and extracted the respective population pyramids of study and reference population, and study characteristics.
Results
We identified 22 population-based studies (eleven from Europe, six from North America, three from Asia, one from Africa, and one from Australasia). Incidence rates of patients with SE ranged from 1.29 to 73.7/100,000 adults (95% confidence interval (CI): 76.6–80.3) and of SE episodes up to 81.1/100,000 adults (95% CI: 75.8–87.0). The proportions of elderly and very old patients varied by a factor of 2.6 and 8.5, respectively, depending on study period and place. Further major reasons for heterogeneity were retrospective or prospective study design, definition of time to diagnose SE, variable detection of nonconvulsive SE (NCSE), different etiologies, inclusion of children, recurrent episodes, postanoxic patients, exclusion of patients with preexisting epilepsy or patients identified outside the emergency department, and choice of reference population for age- and gender adjustment. The most recent definition and classification of SE by the International League Against Epilepsy (ILAE) 2015 was used in two studies. Four studies (18.2%) reported incidences per ten-year age strata necessary for age adjustment to various reference populations.
Conclusions
This critical review reveals a marked heterogeneity among population-based studies on SE in adults. It provides comprehensive details on census-based population statistics in study and reference populations and various study designs and characteristics essential for direct comparisons between studies. Reporting on these essential key features should be improved in population-based studies on SE.
Status epilepticus (SE) is a potentially life threatening neurological condition in which seizures do not stop spontaneously and require immediate treatment [
]. Incidences reported in population-based studies in adults in different regions in the world vary by a factor of more than 50, which is poorly understood [
]. The incidence of SE increases almost logarithmically with age; therefore, higher percentages of elderly and very old people in a study population may contribute substantially to the overall incidence [
]. However, details on census-based population statistics were usually not reported in epidemiological studies, which makes comparisons of these studies difficult, or even impossible.
The goal of this review was to critically assess the census-based demographic composition of each study and reference population, study designs, inclusion and exclusion criteria, and various definitions of diagnostic time T1 to establish the diagnosis of SE. Additionally, we identified advantages and limitations of adjustments usually applied during data processing in epidemiological studies. These findings led us to recommendations for targeted investigations and improved reporting in epidemiological studies on SE.
2. Methods
2.1 Systematic literature search
We performed a systematic literature search to identify all population-based studies including adults that reported incidence rates of SE. We used the search terms “status epilepticus” combined with “incidence”, “population”, or “epidemiology” in the database PubMed from 1950 to August 12th 2019. Studies only reporting about SE in children, subgroups of adults, or subtypes like refractory or superrefractory SE were excluded. Congress reports in abstract form were used as a potential source of later or other publications but were themselves not included into this review. Two authors (CG, ML) assessed the search results regarding eligibility in our review. In case of disagreement, an additional author (ET) made the decision.
2.2 Census-based population statistics
We analyzed the study period of each population-based study and investigated the age characteristics of study and reference populations per decade of life beginning with the sixth decade as the marked increase in incidence started in this age group [
]. The data for the study and reference populations were retrieved from the respective national census bureaus. In studies that included both children and adults, i.e., “total populations”, we calculated crude incidences specifically for adults if data on age strata were provided (age threshold for adulthood as reported, or 20 years).
2.3 Extraction of core characteristics
We systematically retrieved study design, inclusion and exclusion criteria, different times to establish the diagnosis of SE, inclusion of children, postanoxic patients, and patients with recurrent SE. We assessed reporting on proportion of convulsive SE (CSE, i.e., bilateral tonic–clonic SE), etiology, ethnicity, and health insurance system in a prespecified standardized extraction form. Etiology was devided according to the International League Against Epilepsiy (ILAE) 2015 [
] into “known” (i.e., symptomatic; with the subcategories “acute”, “remote”, “progressive”, and “defined in electroclinical syndromes”) and “unknown” (i.e., cryptogenic). Data on febrile SE were collected if reported by studies including also children. With ILAE 2015, CSE had a diagnostic time T1 of 5 min, whereas focal SE and absence SE had to last for at least 10 min [
]. Case fatalities were reported as in hospital or at 30 days, respectively.
All results of our assessments were summarized and reported in systematic evidence tables.
3. Results
3.1 Search results
The search algorithm and number of identified papers are presented in Fig. 1. Our search strategy identified a total of 1954 different papers. Five studies were excluded after full text analysis as these studies did not report incidences. Finally, 22 population-based studies including adults reported incidences of SE and were, therefore, included in our critical review. Years of publication ranged from 1995 to 2019 (Table 1) [
]. Patients from the Swiss Canton of Geneva investigated by Jallon et al. were also included in a later study performed in French-speaking Switzerland [
Govoni et al. 2008 adjusted to “European population” referring to Waterhouse et al. 1982 [39], which authors could not retrieve; earliest online data for Ferrara were from 2002.
SP, standard population: data for European standard populations 1976 and 2013 are provided for comparison; DOM: Département d'Outre Mer — French overseas department; metrop: metropolitan (European part of France).
a Logroscino et al. 2001 adjusted to 1980 U.S. white population.
b Hesdorffer et al. 1998 adjusted to 1980 U.S. population.
c Dham et al. 2014 adjusted to 2000 US census.
d Betjemann et al. 2015 adjusted to 2000 US census.
e Coeytaux et al. 2000 adjusted to 1980 U.S. population.
f Knake et al. 2001 adjusted to “area 45” within Hessen 1998.
g Vignatelli et al. 2003 adjusted to Italy 1991, earliest online data for Bologna were from 2002.
h Vignatelli et al. 2005 adjusted to Italy 1991, earliest online data for population of Lugo di Romagna were Ravenna 2002.
i Govoni et al. 2008 adjusted to “European population” referring to Waterhouse et al. 1982 [
The place of study, study period (range: 1935 to 2016), census used as reference population, and percentages of elderly (patients 60 years of age or older) in the study and reference populations are presented in Table 1. The percentage of elderly ranged from 15.0% in Thailand in 2005 to 38.3% in the district North Savo, Finland in 2015, i.e., a variation of factor 2.55 [
]. There were no inhabitants over age 85 years in Rochester in 1940; otherwise, adults in this age group ranged from 0.6% in Thailand in 2005 to 5.1% in Florence in 2016, i.e., a factor of 8.5 [
Guidelines of the German Neuroloigcal Society 2008: bilateral tonic–clonic SE 5 min, other forms: 20 to 30 min, guidelines 2012: 5 min for all forms [41].
Incidences per 100,000; A: adults; T: total population, i.e., adults and children; pro: prospective; ret: retrospective; CSE: convulsive, i.e., bilateral tonic–clonic, status epilepticus; CSE T1: diagnostic time T1 for CSE n.a.: not available; Hypoxemic: patients after cardiac arrest; N: number of individuals.
a Nationwide or countrywide databases.
b “Primary service area” used for calculating the corrected age adjusted incidence.
c Neonates excluded.
d Incidence increased during study period from 1935 to 1984.
e Incidence increased during study period from 1965 to 1984, overall: 18.3.
f 61 refers to an estimate after correction for underascertainment (“validation correction”).
g Incidence decreased during study period from 1991 to 1998, overall: 6.18.
h Incidence increased during study period from 2000 to 2011, overall: 4.61.
i 5% of adults.
j Personal communication.
k SE was diagnosed after 30 min in the early parts of the study and after 5 min in the later parts.
l Guidelines of the German Neuroloigcal Society 2008: bilateral tonic–clonic SE 5 min, other forms: 20 to 30 min, guidelines 2012: 5 min for all forms [
Inclusion of patients with hypoxic etiology ranged from null to 16% [Logroscino]. Patients with a first episode of SE varied from less than 80% to 100% [
]. One study documented a 2.4-fold increase of nonconvulsive SE (NCSE) after introduction of diagnostic electroencephalography (EEG) criteria in 2013 [
In most cases, more than one factor Percentages of different etiologies were calculated from 150 patients with SE that included also patients outside the primary service area.
In most cases, more than one factor Percentages of different etiologies were calculated from 150 patients with SE that included also patients outside the primary service area.
a Proposal for definition and classification of SE by ILAE 2015.
b Primary service area.
c Calculated per status epilepticus episodes and not per patients.
d 1975–1984.
e In most cases, more than one factor Percentages of different etiologies were calculated from 150 patients with SE that included also patients outside the primary service area.
f Multifactorial: additional 14%.
g Multifactorial: additional 25.9%.
h Tumors.
i Only in children.
j Case fatality at 30 days, otherwise in hospital.
]. Logroscino et al. reported a drop of the annual adjusted incidence of SE in Rochester, MN by relative 27.6% from 15.2 in the period from 1955 to 1964 to 11.0/100,000 total population from 1965 to 1974 (Fig. 2) [
]. In this time, the percentage of patients 65 years or older was slightly decreasing by relative 2.6% from 39% of total population from 1955 to 1964 to 38% from 1965 to 1974 [
]. In the years 1950, 1960, 1970, and 1980, children aged up to 4 years in Rochester were 3027, 4991, 5071, and 4260, and 34.5%, 32.8%, 24.5%, and 24.1% of persons aged under 20 years, respectively, i.e., relative 25.3% lower in 1970 compared with the previous decade [
Dham et al. found a substantial variation of incidences of SE in USA over time with an almost threefold increase during the 1980s, which then reduced to almost a half in the 1990s, followed by another marked increase in the early 2000 years (Fig. 3) [
]. In the USA, the absolute number of children up to 4 years of age were 16.451,184, 18.856,447, and 19.178,293, in the years 1980, 1990, and 2000, representing 22.7%, 26.2% (i.e., an increase of 15.4% compared with previous decade), and 23.8% (i.e., a decrease of 9.2% compared with previous decade) of all children and adolescents under 20 years of age, respectively [
]. Similar data were also reported for California by Wu et al. who reported a reduction of incidence from 8.5 to 4.9 CSE/100,000 (drop of 42.4%) from 1991 to 1998, while percentages of people aged 60 years or older remained constant at 20.1%, and percentages of the people 75 years of age and older even increased from 1.4% to 1.8% (Table 1) [
(Reproduced with permission, courtesy of Neurocritical Care/SPRINGER 2014) a no data available for 1985. 1986; b no data available for 2006-2010; c no data from 2002 - 2004, as question differed.
The incidence of 18.3/100,000 total population in Rochester increased to 23.7/100,000 adults after age- and gender-adjusted to the population of Austria 2016, which was used in the recent Austrian study using ILAE 2015 criteria for SE [
In this critical review of 22 population-based studies, we identified several profound influence factors responsible for the wide range of annual incidences of SE of 1.29 to 73.7/100,000 adults and 81.1 SE episodes/100,000 adults. In part, these factors are amenable to mathematical correction, which increases comparability. To our knowledge, we investigated for the first time the census-based demographic statistics underlying the population-based studies on SE in adults. We implemented our findings into recommendations on reporting of incidences in population-based studies on SE (Table 4).
Table 4Recommendations for reporting in population-based studies (e.g., on status epilepticus).
The following items should be considered in any report on incidence rates (if applicable):
Item 1: Issues related to case ascertainment
Item 1a: Estimate underascertainment in rural areas and in the city separately.
Item 1b: If study is based on discharge diagnosis, estimate the number of patients given another discharge diagnosis. If patients are treated on an outpatient basis, make sure these patients receive a discharge diagnosis and are included into the study, or are ascertained in alternative ways.
Item 1c: Provide the number of patients with health insurance during study time. If risk of SE is not uniform within different ethnical or age groups, insurance status should be provided separately for each subgroup.
Item 1d: Is there evidence of gender-specific access to healthcare?
Item 2: Issues related to the calculation of incidence rates
Item 2a: Only first episode (e.g., status epilepticus) should be considered. However, also report the number of patients with recurrent episodes during study period.
Item 2b: Any details or features of study populations should be reported per patients and not per episodes.
Item 2c: Estimate incidences for children and adults separately. Provide data for ages of 15 and 20 years to define adulthood, respectively.
Item 2d: Provide number of patients who denied informed consent in prospective studies.
Item 3: Issues regarding the adjustment of incidence rates
Item 3a: It is of fundamental importance to unambiguously report place and time of the reference population, and a literature reference, which directly provides data on population statistics.
Item 3b: For public health issues, the reference population used for the adjustment should be representative of the geographical region and time period of interest.
Item 3c: The study and reference populations have to be similar with respect to factors other than age and gender, for example, various ethnical subgroups have a different risk of SE. If homogeneity cannot be achieved, age- and gender-adjusted incidence rates must be calculated separately for the subgroups.
Item 3d: Perform age and gender adjustment separately for children and adults to avoid distortion of results if proportions of adults and children are different in study and reference populations.
Item 3e: Number of patients and inhabitants per decade of life should be reported in study and reference population, separately for men and women. This would enable the interested reader to adjust the results to other reference populations.
Item 4: Characteristics of status epilepticus
Item 4a: Provide the proportion of patients after cardiac arrest, of patients with preexisting epilepsy, and of patients with known (acute, remote, progressive, or in defined electroclinical syndromes) or unknown etiology.
Item 4b: Definition: Provide the number of patients with SE diagnosed after 5 or 30 min for convulsive SE, that is, bilateral tonic–clonic SE, and 5, 10, or 30 min for focal SE or absences.
Item 4c: Classification: According to ILAE 2015, provide data for occurance of prominent motor phenomena (essential for NCSE) and level of consciousness before treatment.
Item 4d: Provide information about the semiological sequence, for example, bilateral tonic–clonic evolving to nonconvulsive, or nonconvulsive evolving to focal motor SE.
Item 4e: Provide the number of patients with spontaneous cessation of SE with characterization of durations, and how the end of SE was measured.
Item 4f: Discuss potential sources of bias, such as incentives in the billing or coding system, or regulations, which cause a shift from inpatient to outpatient management.
]. The annual incidence in adults under age of 60 years of 18.1/100,000 (95% confidence interval (CI): 10.1–30.1) may increase more than fourfold to 79.9 [95% CI: 53.4–114.8] in the elderly (60 years or older), subgroups of 80 years and older reach more than 150, and above 90 years almost 200/100,000 with huge CIs (Supplementary Fig. 1) [
]. For comparison, in the very first population-based study in Virginia, the incidence in young adults was 27, and in the elderly, more than threefold higher with 86/100,000 [
]. Consequently, the higher the ratio between elderly and young adults, the more the elderly determine the incidence in adults. Population-based studies usually provide information about the study period and study region. However, the proportion of elderly is different for each time point in each country as illustrated for USA and Germany (Supplementary Fig. 2, Supplementary Fig. 3) [
]. For practical reasons, we can group the studies according to their percentages of elderly in their adult population, which helps to quickly estimate the amount of people affected by SE in this high risk group: 10–20% of adults [
]. The ethnical background may also influence the risk of SE. The first prospective population-based study performed in Richmond (Virginia, VA) found the incidence of SE for Americans with African descent almost threefold higher than in Americans with European descent (57 compared with 20/100,000 adults and children) [
]. In Virginia, the estimated incidence after correction for underascertainment of 61/100,000 adults and children was more than three times higher than the incidence of 18.3/100,000 adults and children in Rochester [
]. In USA, the annual incidence in Americans with African descent was almost double the incidence in Americans with European descent with 13.7 compared with 6.9/100,000 adults and children, respectively [
]. In California, the relative risks of Americans with African descent, Hispanic descent, or Asians descent were 1.92, 0.50, and 0.38 compared with Americans with European descent [
]. However, in Auckland, New Zealand, incidences found in people with Asian descent, with European descent, Pacific Islanders, and Maori were 17.8, 19.1, 26.6, and 29.3 patients/100,000 adults and children, respectively [
]. This variation of risks between European and Asian descent of 2.6 in California and of 1.1 in Auckland warrants the search for other factors. In particular, groups with different descents may have considerably different percentages of elderly people that could bias incidences. In Auckland in 2015, elderly people were 21.6% of adults (age limit: 15 years) in European descent compared with 8.34% in Asian descent [
]. This could explain a rate more than double in European descent compared with Asian descent, but the incidences were almost similar. Data of the second high incidence age group of very young children separate for each descent were not available [
]. However, the relative percentages of people with different descents should be considered when data are interpreted. Of note, Americans with African descent made 80% of the study population in Virginia [
], although exact percentages were only rarely reported. The population on the island of LaReunion included 23% of inhabitants with European descent, 53% “ethnically mixed”, and people with Indian, Chinese, and African descent [
Age and gender adjustment is a process that transfers the findings from a certain study with a specific study population into the respective (adjusted) findings of any reference population. In other words, the adjusted results indicate what the results would have been if the study had been performed in the reference population. For performing adjustment, one needs the incidences per age and gender stratum in the study population, and the numbers of persons within these strata both in the study and the reference population. An important application of age and gender adjustment is the comparison of different studies for which the results obtained for the respective study cohorts are adjusted to the same reference population.
For example, the Austrian study adjusted the findings from the Rochester study by Hesdorffer et al. between 1965 and 1984 to Austria 2016 and revealed an increase of incidence from reported 18.3/100,000 total population to adjusted 23.7 per 100,000 adults per year [
], which mainly addressed the different age structures of Rochester and Austria (Table 1). In addition to the adjustment algorithm, comparison of incidence rates across studies must consider the precision of incidence estimates in each individual study expressed by CIs, which were reported in 15 studies [
31036-4&id=gr1.jpg){kind=link}
31036-4&id=gr2.jpg){kind=link}
31036-4&id=gr3.jpg){kind=link}