A systematic review of the epidemiology of status epilepticus.

Abstract

Population-based data on the incidence, aetiology, and mortality associated with status epilepticus (SE) are required to develop preventative strategies for SE. Through a systematic review, we aimed to assess the methodological quality as well as similarities, and differences between available population based studies in order to arrive at conclusions on the epidemiology of SE. All population-based studies where primary outcome was incidence, aetiology or mortality of SE were identified through a systematic search and synthesized. Methodological quality of studies were independently rated by two examiners using a unique scoring system. Seven population-based projects on SE yielding nine published reports and five abstracts were reviewed. Quality scores were in the range of 19-34 with a possible maximum of 40 (kappa scores 0.67-1.0). The incidence of SE has a bimodal distribution with peaks in children aged less than a year and the elderly. Most SE were acute symptomatic. Short-term mortality was 7.6-22% and long-term mortality was 43%. Age and aetiology were the major determinants of mortality. There are few population-based studies on SE but most are of good quality. Most studies are primarily or exclusively based on adult populations. There is limited information on the association of ethnicity and socio-economic status and SE.

Full text

Page 1

CME ARTICLE
A systematic review of the epidemiology of status epilepticus
R. F. M. China,b, B. G. R. Nevilleaand R. C. Scotta,c
aNeurosciences Unit, Institute of Child Health, University College London, London, UK and Great Ormond Street Hospital for Children NHS
Trust, London, WCIN IEH, UK;bThe Centre for Paediatric Epidemiology and Biostatistics; andcRadiology and Physics Unit, Institute of
Child Health, University College London, London, UK
Keywords:
epidemiology, status epil-
epticus, systematic review
Received 8 March 2004
Accepted 13 May 2004
Population-based data on the incidence, aetiology, and mortality associated with
status epilepticus (SE) are required to develop preventative strategies for SE. Through
a systematic review, we aimed to assess the methodological quality as well as simi-
larities, and differences between available population based studies in order to arrive
at conclusions on the epidemiology of SE. All population-based studies where primary
outcome was incidence, aetiology or mortality of SE were identified through a sys-
tematic search and synthesized. Methodological quality of studies were independently
rated by two examiners using a unique scoring system. Seven population-based pro-
jects on SE yielding nine published reports and five abstracts were reviewed. Quality
scores were in the range of 19–34 with a possible maximum of 40 (kappa scores 0.67–
1.0). The incidence of SE has a bimodal distribution with peaks in children aged less
than a year and the elderly. Most SE were acute symptomatic. Short-term mortality
was 7.6–22% and long-term mortality was 43%. Age and aetiology were the major
determinants of mortality. There are few population-based studies on SE but most are
of good quality. Most studies are primarily or exclusively based on adult populations.
There is limited information on the association of ethnicity and socio-economic status
and SE.
Introduction
Data on the incidence, aetiology, and mortality associ-
ated with status epilepticus (SE), the most common
medical neurological emergency (Leppik, 1985; DeL-
orenzo et al., 1995, 1996), are required for decisions on
allocation of resources by institutions, and even gov-
ernment, that may be used to develop primary and sec-
ondary preventative strategies for SE. Most published
data are from hospital-based studies and may therefore
be biased and inappropriate. What is required is epide-
miological data from the general population, and such
datamaybeobtainedthroughpopulation-basedstudies.
Through a systematic review, we aimed to assess the
methodological quality as well as similarities, and dif-
ferences between available population-based studies in
ordertoarriveatconclusionsontheepidemiologyofSE.
The advantages of a systematic review rather than a
traditional review have been well described previously
(Mulrow, 1987). On the whole, systematic reviews have
been conducted for randomized
(RCTs) as they have long been viewed as the ?gold
standard? of evidence; observational studies have been
controlled trials
largely overlooked. This has resulted in the develop-
ment of methodology for systematic reviews, mainly for
RCTs. However, in some situations, only data from
observational studies, such as epidemiological studies,
are available, possible or appropriate (Berlin, 1995).
Observational studies have been used to evaluate edu-
cational programmes (Sipe and Curlette, 1997) and
assess risk factors for disease (Ioannidis and Lau, 1999).
Studies on the latter may not be randomized because
the risk factors may be related to intrinsic human
practices, and it is unethical to expose subjects to
harmful risk factors (Lipsett and Campleman, 1999).
Thus, it is increasingly being realized that observational
studies provide useful evidence, resulting in more
recent, active development of methodologies for sys-
tematic reviews of observational studies (Mulrow et al.,
1997; Sutton et al., 1998; Stroup et al., 2000).
The specific aim of this systematic review is to iden-
tify and synthesize all available population-based data
on SE to arrive at conclusions on the epidemiology of
SE including:
• The incidence of SE worldwide, and the effect of
gender, ethnicity and age on incidence.
• The range of aetiologies of SE.
• The association of SE with epilepsy.
• The recurrence of SE.
• The mortality associated with SE.
Correspondence: Richard F. M. Chin, The Wolfson Centre, Meck-
lenburgh Square, London WC1N 2AP, UK (fax: +44 2078339469;
e-mail: r.chin@ich.ucl.ac.uk).
800
? 2004 EFNS
European Journal of Neurology 2004, 11: 800–810

Page 2

Methods
Search strategy
The following steps for ascertainment of the appropri-
ate articles for review were carried out:
Step 1: Computer-assisted literature searches of the
bibliographic databases: Medline, Cochrane Collabor-
ation database, Embase, Web of Science, and CINALH
for papers published between 1966 and November 2002
using the following search terms:
• status epilepticus and population*
• status epilepticus and epidemiology
• status epilepticus and incidence
• status epilepticus and aetiolog*
• status epilepticus and etiolog*
• status epilepticus and cause*
• status epilepticus and seizure type
• status epilepticus and mortality
• status epilepticus and fatality
All of the above searches were repeated with prolonged
febrile convulsion* or prolonged febrile seizure* instead
of status epilepticus.
All the searches were combined with (or stepwise limits
conducted in Medline): (i) review (ii) meta-analysis.
Step 2: The reference list of articles identified in step 1
were examined to identify additional relevant studies.
Step 3: An internet search up to 31 December 2003
using Google as the search engine using the search
strategy in step 1 was conducted.
Step 4: The conference proceedings and abstracts of the
British Paediatric Neurology Association (BPNA), the
American Epilepsy Society (AES), European Paediatric
Neurology Society (EPNS) year 1995 to year 2003 were
handsearched.
Selection criteria
All population-based studies on SE where the primary
outcome was either the incidence, aetiology, seizure
types, or mortality of SE were included. Studies of all
age groups were included. Studies that were based in a
single institution or did not involve a network of hos-
pitals serving a well-defined general population, and
non-primary studies were excluded.
Data extraction and assessment of study quality
Data to be extracted were agreed by two of the authors
and included study period, details of the study popu-
lation, selection criteria, definitions, classification of
seizure types, validation of the databases, and outcome
measures. One of the reviewers (R.F.M.C.) extracted
the data using a data extraction form and a second
reviewer (R.C.S.) confirmed them. To assess the meth-
odological quality, we developed a scoring system based
on a published proposal for reporting meta-analysis of
observational (Stroup et al., 2000) and the guidelines
for epidemiologic studies on epilepsy of the Interna-
tional League Against Epilepsy (ILAE) (International
League Against Epilepsy, Commission on Epidemiol-
ogy and Prognosis, 1993) (see Table 1). Each study was
scored on the following items: type of study design,
description of study population, adequacy of case def-
initions, selection criteria of study population, defini-
tion of seizure types and aetiologies according to the
ILAE guidelines, adequacy of definition of outcome
measures, type of rate/effect/impact, and whether there
were adjustments for degree of ascertainment. In
accordance with the ILAE guidelines for epidemiologic
studies on epilepsy, the more important determinants of
quality were considered to be type of study design,
definitions of cases and outcomes, classification of
seizure type and adjustment of rates. Thus, these items
were given greater scores. Using the scoring system,
each study was assessed independently by two reviewers
(R.F.M.C. and R.C.S.). Inter-rater agreement (kappa
Table 1 Scoring system for assessment of methodological quality of
population-based studies on SE
Study design12 ¼ prospective
5 ¼ retrospective
2 ¼ clearly defined
1 ¼ unclearly defined
0 ¼ not defined
4 ¼ all clear
2 ¼ most clear
1 ¼ most not clear
4 ¼ clear
2 ¼ not clear
1 ¼ not defined
4 ¼ clear
2 ¼ not clear
1 ¼ no
4 ¼ clear
2 ¼ not clear
1 ¼ no
2 ¼ all clear
1 ¼ most clear
0 ¼ most not clear
4 ¼ either age-specific or
age-adjusted as well as
sex-specific
2 ¼ age-specific or
age-adjusted or
sex-specific
1 ¼ crude only
4 ¼ clear
1 ¼ not clear/none
Sample size
Methodological definitions in
individual studies
Inclusion criteria for selection of
cases of SE from study population
Seizure type defined according to
the ILAE guidelines?
Risk factors/aetiology defined
according to ILAE guidelines?
Definitions of outcome parameters
Standardization of rates/ratios
Ascertainment adjustment
Maximum score ¼ 40, minimum score ¼ 11.
A systematic review of the epidemiology of status epilepticus
801
? 2004 EFNS European Journal of Neurology 11, 800–810

Page 3

scores) for assessment of quality-of-study methodology
was determined for each study.
Results
Seven population-based projects on SE (Richmond
Group, Virginia, USA; Rochester Group, Minnesota,
USA; HessenGroup, Germany;
Cantons Group, Switzerland; California Group, USA;
St Louis Group, USA; and North London Group, UK)
yielding nine published reports (DeLorenzo et al., 1995,
1996; Logroscino et al., 1997, 2002; Hesdorffer et al.,
1998; Waterhouse et al., 1999; Coeytaux et al., 2000;
Knake et al., 2001; Wu et al., 2002) and five abstracts
(Morton et al., 2001; Towne et al., 2002; Trevathan
et al., 2002; Chin et al., 2003; Scott et al., 2003a) ful-
filled the inclusion criteria for entry into the review.
Four projects were conducted in the USA, three of
which were retrospective (Rochester, California, St.
Louis Groups) and one was prospective (Richmond
Group). The remaining three population-based projects
were European-based and prospective (Hessen, French-
speaking Cantons, North London Groups). All studies
were conducted after the ILAE’s guidelines for epi-
demiologic studies were published. There were neither
systematic reviews nor meta-analyses. The results are
presented for data collection up to 31 December 2003.
Studies were pooled only when their definitions of
relevant variables were similar or the results could be
processed and analysed as such. A glossary of defini-
tions used in the studies is listed in the Appendix.
French-speaking
Description of the studies
The description of the studies, methodological quality
scores and inter-rater agreement are shown in Table 2.
The quality scores ranged from 19 to 34 of a possible
maximum of 40. Inter-rater agreement was good to very
good with kappa scores between 0.67 and 1.0.
Seizure characteristics
The incidence of SE showed a bimodal distribution with
peaks in children less than a year of age and the elderly
(DeLorenzo et al., 1996; Hesdorffer et al., 1998; Wu
et al., 2002). Children less than a year of age had the
highest incidence (135–156/100 000/year) (DeLorenzo
et al., 1996; Hesdorffer et al., 1998). In general, males
were more likely to have SE, as were those of non-white
ethnicity. For further details, see Table 3.
The cause of SE varied according to age groups. In
children,prolonged febrile
(Hesdorffer et al., 1998; Chin et al., 2003) and low anti-
epileptic drug levels (21%) (DeLorenzo et al., 1996)
convulsions(33–35%)
were most common. In contrast, in adults, cerebro-
vascular disease/accidents, withdrawal or low levels
(34%) of maintenance antiepileptic drugs were most
likely. Acute symptomatic SE accounted for most
aetiologies across all age groups. Few episodes of SE
were idiopathic (see Table 4 for further details on
aetiology).
Status epilepticus occurred within the context of
epilepsy in <50% of cases overall. In children, 16–38%
had a history of epilepsy, whilst in adults, 42–50% had
a history of epilepsy (DeLorenzo et al., 1996; Knake
et al., 2001; Chin et al., 2003). The percentage of SE
patients with a history of epilepsy was higher in those
who resided in rural areas (49–56%) compared with
those who resided in urban areas (33%) (Coeytaux
et al., 2000; Knake et al., 2001).
Most SE lasted <24 h (DeLorenzo et al., 1996;
Hesdorffer et al., 1998; Knake et al., 2001). All pro-
longed febrile convulsions (PFC) were shorter than 2 h
(Hesdorffer et al., 1998). In the paediatric population,
the duration of continuous SE (median 60.5 min) was
shorter than that of intermittent SE (median 87 min,
P < 0.02). In contrast, adults had no difference in
duration of continuous SE (median 97.5 min) com-
pared with intermittent SE (median 85 min) (Water-
house et al., 1999).
Recurrence of SE occurred in 13.3–18.5% (DeL-
orenzo et al., 1996; Hesdorffer et al., 1998; Knake
et al., 2001). Recurrence was highest in the paediatric
population (DeLorenzo et al., 1996). In adults, almost
all (93.8%) patients with recurrences had a history of
epilepsy (Knake et al., 2001).
Mortality associated with SE across all age groups
ranged from 7.6 to 22% during hospitalization for SE
or within 30 days of SE (short-term mortality) and 43%
within 10 years following initial survival 30 days after
SE (long-term mortality). Age and aetiology of SE were
the major determinants of short- and long-term mor-
tality. Mortality is lowest in children (short-term mor-
tality from 3 to 9% and long-term mortality around
7%) and highest in the elderly (short-term mortality
from 22 to 38% and long-term mortality 82%).
Amongst children, those aged <1 year had the highest
short-term mortality (17–18%) (Morton et al., 2001).
The standardized mortality ratio (SMR), the relative
risk of mortality compared with the general population,
for subjects with long-term mortality associated with
SE was 2.8 (95% CI 2.1–3.5). The SMR in those older
than >65 years was 2.2 (95%CI 1.6–2.9), and in those
<65 years, SMR was 5.1 (95% CI 2.8–8.0).
Most deaths during hospitalization for SE or
within 30 days of SE occurred in those with an acute
symptomaticaetiology (DeLorenzo
Logroscino et al., 1997; Chin et al., 2003). No deaths
et al.,1996;
802
R. F. M. Chin, B. G. R. Neville and R. C. Scott
? 2004 EFNS European Journal of Neurology 11, 800–810

Page 4

Table 2 Description, methodology scores and interrater agreement of population-based studies on the epidemiology of status epilepticus
Studies
Country
Design
Sample size
Methods
definitions
Inclusion
criteria
ILAE
SE type
ILAE
aetiology Outcome
Standardization of rates/ratios
Ascertainment
adjustment
Methodology
score
j-value
Published papers
DeLorenzo et al.
(1995, 1996)
USA
Prospective
137 adults, 29 children
Most clear Unclear
Clear
Unclear
Clear
Age-specific
Clear
32
0.67
Logroscino et al., 1997
USA
Retrospective 132 adults, 52 children
Most clear Clear
Clear
Clear
Clear
Age- and sex-specific No
28
1.0
Hesdorffer et al., 1998
USA
Retrospective 130 adults, 69 children
Most clear Clear
Clear
Clear
Clear
Age- and sex-specific No
28
1.0
Waterhouse et al. (1999) USA
Prospective
433 adults, 212 children Most clear Clear
Clear
Unclear
Clear
Crude
Unclear
31
0.89
Coeytaux et al. (2000)
Switzerland Prospective
108 adults, 64 children
Most clear Unclear
Clear
Unclear
Clear
Age- and sex-specific No
31
1.0
Knake et al. (2001)
Germany
Prospective
150 adults
Most clear Unclear
Clear
Unclear
Clear
Age-adjusted and
sex-specific
Clear
34
0.67
Logroscino et al. (2002)
USA
Retrospective 110 adults, 35 children
Most clear Clear
Clear
Clear
Clear
Age- and sex-specific No
28
1.0
Wu et al. (2002)
USA
Retrospective 12716 adults,
2885 children
Most clear Clear
No
No
Clear
Crude
No
19
0.78
Abstracts
Morton et al. (2001)
USA
Prospective
598 children
Most clear Clear
Clear
Unclear
Clear
Crude
Unclear
31
0.78
Towne et al. (2002)
USA
Prospective
318 adults
Most clear Clear
Clear
Unclear
Clear
Crude
Unclear
31
0.78
Trevathan et al. (2002)
USA
Retrospective 39 649 subjects
Most clear Clear
No
No
Clear
Age-specific
No
22
0.78
Scott et al. (2003a)
UK
Prospective
55 children
Most clear Clear
Clear
Unclear
Clear
Crude
Clear
33
1.0
Chin et al. (2003)
UK
Prospective
79 children
Most clear Clear
Clear
Unclear
Clear
Crude
Clear
33
1.0
Methodology scores are based on the scoring system detailed in Table 1. Methods definitions refers to definitions outlined in the methodology of individual studies. Inclusion criteria refers to the criteria
for selection of cases of SE from study population. ILAE SE type and ILAE aetiology refer to whether the types of SE and aetiology of SE were classified according to ILAE guidelines. Outcome refers to
whether the outcome parameters were clearly defined or not. Standardization of rates/ratios refers to whether rates/ratios were standardized and the type. Ascertainment adjustment refers to whether an
adjustment was made for degree of ascertainment. Maximum methodology score ¼ 40, minimum score ¼ 11. Displayed scores are that of one of the authors (RCS). Note that the authors of this paper
are also the authors of the last two abstracts in this table.
A systematic review of the epidemiology of status epilepticus
803
? 2004 EFNS European Journal of Neurology 11, 800–810

Page 5

occurred in those with PFC. In subjects with long-
term mortality, the SMR was significantly elevated in
the acute and remote symptomatic SE groups. Acute
symptomatic SE was associated with a higher mor-
tality rate than idiopathic/cryptogenic SE. Amongst
seizure types, myoclonic SE in particular was associ-
atedwithhigh short-and
Subjects with myoclonic SE had a significant relative
risk (RR) of long-term mortality in comparison with
those with generalized SE (RR 4.0, 95% CI 1.3–13).
There was no difference in short-term case fatality
between continuous and intermittent SE in children,
but in adults, short-term case fatality was 31.4% in
continuous SE compared with 19.6% in intermittent
SE (P < 0.01).
The effect of duration of SE and gender on short-
term mortality is uncertain (Logroscino et al., 1997;
Waterhouse et al., 1999; Wu et al., 2002) but there was
increased long-term mortality in subjects with SE
>24 h when compared with those with SE <2 h (RR
2.3, 95% CI 1.1–5.1) (Logroscino et al., 2002). Ethni-
city had no effect on short-term mortality.
long-term mortality.
Discussion
The methodology for systematic reviews of non-inter-
ventional research is still in its infancy.
In this systematic review of observational studies, the
methodology has been based on recently published
proposals for reporting meta-analyses of observational
studies, and guidelines for systematic reviews of qual-
itative studies (Abou-Khalil et al., 1993; Stroup et al.,
2000). A similar methodology has been used in a re-
cently published systematic review of incidence studies
of epilepsy (Kotsopoulos et al., 2002).
Differences in definitions, classification of seizures,
and aetiologic criteria in epidemiological studies on
epilepsy have often yielded discordant results between
studies, and have made comparison of results difficult
(International League Against Epilepsy, Commission
on Epidemiology and Prognosis, 1993). In an attempt
to address this issue, in 1993, the ILAE published
guidelines for epidemiologic studies on epilepsy (Inter-
national League Against Epilepsy, Commission on
Epidemiology and Prognosis, 1993). We identified se-
ven projects with 13 eligible studies that were available
for this systematic review. Although they were con-
ducted after the ILAE’s published guidelines for epi-
demiologicstudies on
heterogeneity in the basic epidemiological parameters
listed above. Thus, a meta-analysis of the data from
these studies was not possible but through the devel-
opment of a unique scoring system based on a pub-
lished proposalfor reporting
observational studies (Stroup et al., 2000) and the
ILAE’s guidelines (International League Against Epi-
lepsy, Commission on Epidemiology and Prognosis,
1993), we were able to assess the methodological quality
of studies and to compare studies. Most population-
based studies on SE have high methodology scores. It is
interesting that studies that had the largest study sam-
ple size (Trevathan et al., 2002; Wu et al., 2002) had the
lowest methodology scores, suggesting that sample size
alone is not an adequate guide to the quality of any
individual study.
The range of reported crude incidence rates is wide
(6.8–41/100 000/year). The lowest incidence rate is
from a study with a low methodology score (Wu et al.,
2002). This suggests that the true incidence rate of SE is
higher than the 6.8/100 000/year in California. The
remaining five studies that reported incidence rates had
similar quality scores which were high (DeLorenzo
et al., 1996; Hesdorffer et al., 1998; Coeytaux et al.,
2000; Knake et al., 2001) but despite this there was still
a wide range of incidence rates (9.9–41/100 000/year).
It may be hypothesized that problems in case ascer-
tainment between studies may partly explain the wide
range of reported incidence rates. Quantification of
epilepsy,there isstill
meta-analysisof
Table 3 Incidence of SE according to research groups
Incidence rate
(per 100 000 persons/year)
Richmond
(Virginia, USA)
Rochester
(Minnesota, USA)
Hessen
(Germany)
French-speaking
Cantons (Switzerland)
California
(USA)
North London
(UK)
Crude
Ascertainment adjusted
Age-adjusted
In Children
In adults
In the elderly
Male : Female
White : Non-white
Peak age (years of age)
41
61
–
–
15.8
17.1
17.1
–
4
54
9.9
–
10.3
21
5
15
6.8*
–
–
3.86
4.58
14.6
1 : 1
2 : 1
<5, >60
–
–18.3
24
6
62
38
27
86
18
–
–
1.3 : 1
2.6 : 1
<5
– 2 : 12 : 1 2 : 1
3 : 1–––
<1, >60<1, >65>60
*Restricted to generalized convulsive status epilepticus.
804
R. F. M. Chin, B. G. R. Neville and R. C. Scott
? 2004 EFNS European Journal of Neurology 11, 800–810