Professional Documents
Culture Documents
Epidemiology of Epilepsy: Definitions
Epidemiology of Epilepsy: Definitions
Chapter 10
Epidemiology of epilepsy
S. ABRAMOVICI1* AND A. BAGIĆ
University of Pittsburgh Comprehensive Epilepsy Center (UPCEC), University of Pittsburgh Medical School, Pittsburgh, PA, USA
Abstract
Modern epidemiology of epilepsy maximizes the benefits of advanced diagnostic methods and sophisti-
cated techniques for case ascertainment in order to increase the diagnostic accuracy and representativeness
of the cases and cohorts studied, resulting in better comparability of similarly performed studies. Overall,
these advanced epidemiologic methods are expected to yield a better understanding of diverse risk factors,
high-risk populations, seizure triggers, multiple and poorly understood causes of epilepsy, including the
increasing and complex role of genetics, and establish the natural course of treated and untreated epilepsy
and syndromes – all of which form the foundation of an attempt to prevent epileptogenesis as the primary
prophylaxis of epilepsy. Although data collection continues to improve, epidemiologists still need to
overcome definition and coding variability, insufficient documentation, as well as the interplay of socio-
economic factors and stigma. As most of the 65–70 million people with epilepsy live outside of resource-
rich countries, extensive underdiagnosis, misdiagnosis, and undertreatment are likely. Epidemiology will
continue to provide the necessary information to the medical community, public, and regulators as the
foundation for improved health policies, targeted education, and advanced measures of prevention and
prognostication of the most common severe brain disorder.
1
Current address: Sergiu Abramovici, MD, Tel-Aviv Sourasky Medical Center, 6 Weizmann St., Tel-Aviv, 6423906, Israel.
Cell-phone number: +972-52-7360535, Email: sergiua@tlvmc.gov.il
*Correspondence to: S. Abramovici, University of Pittsburgh Comprehensive Epilepsy Center (UPCEC), University
of Pittsburgh Medical School, Suite 811, Kaufmann Medical Building, 3471 Fifth Ave, Pittsburgh PA 15213, USA.
Tel: +1-412-877-2582, E-mail: sergiu.a@gmail.com
160 S. ABRAMOVICI AND A. BAGIĆ
transient ischemic attacks, migraine-related transient Epilepsy
neurologic deficits or visual phenomena, syncope, acute
Epilepsy is defined as “a disorder of the brain charac-
psychosis-related hallucinations, parasomnias, espe-
terized by an enduring predisposition to generate
cially rapid-eye-movement sleep behavioral disorder,
epileptic seizures, and by the neurobiologic, cognitive,
hyperkinetic movement disorders, like tremor, chorea,
psychologic, and social consequences of this condition
and athetosis, metabolic and toxic derangements that
(Fisher et al., 2014).” A diagnosis of epilepsy should be
cause movement and awareness-related changes, and
made in any of the following conditions: (1) at least two
nonepileptic behavioral spells (Smith, 2012). The latter
unprovoked (or reflex) seizures occurring > 24 hours
is common in conversion disorders and was diagnosed
apart; (2) one unprovoked (or reflex) seizure and a prob-
in 25–26% of epilepsy monitoring unit (EMU) admis-
ability of further seizures similar to the general recurrence
sions (Salinsky et al., 2011). Misidentification of the
risk after two unprovoked seizures, occurring over the
specific episode may lead to false diagnosis and misdiag-
next 10 years (at least 60%); or (3) diagnosis of an epilepsy
nosis of epilepsy, thus skewing the data towards over- or
syndrome (Fisher et al., 2014). The definition of epilepsy,
underestimation of incidence and prevalence.
as adopted by the International League Against Epilepsy
(ILAE) in 2014, requires the occurrence of at least one epi-
Acute symptomatic seizures leptic seizure. This detail is extremely important in the cur-
rent and future study of epilepsy epidemiology, since
An acute symptomatic seizure, or provoked seizure, is a
previous definitions of epilepsy required two seizures at
seizure that was provoked by a direct insult to the brain.
least 24 hours apart. As many cohorts studied after this
This insult may be focal, such as stroke, encephalitis, a
declaration are expected to include more cases diagnosed
tumor, or any focal process exerting direct irritation over
with epilepsy, the future incidence and prevalence of epi-
the cortex. Other insults may be generalized and exert
lepsy are expected to rise. Nevertheless, there are some cir-
their effect indirectly by a systemically diffuse toxic,
cumstances under which recurrent seizures do not
metabolic, inflammatory, or paraneoplastic process. In
necessarily mandate the diagnosis of epilepsy and are sel-
a review of several studies conducted between 1990
dom considered as such. A cluster of multiple seizures
and 2005, on diverse populations from different parts
occurring in a period of 24 hours, a single episode of status,
of the world, the incidence of acute symptomatic seizures
or febrile seizures should not be considered epilepsy indef-
ranged between 20 and 39 cases per 100 000 persons per
initely (Commission on Epidemiology and Prognosis,
year (Loiseau et al., 1990; Hauser et al., 1993; Forsgren
1993). While epilepsy can begin in the neonatal period,
et al., 1996; Jallon et al., 1997; MacDonald et al., 2000;
neonatal seizures are usually considered to be reactive to
Olafsson et al., 2005).
an acute cerebral insult and usually do not persist beyond
a few weeks after onset.
Unprovoked seizure
An unprovoked seizure is a seizure without any identi- Drug-resistant epilepsy
fied focal or systemic immediate cause, although it fre- Drug-resistant epilepsy, also named intractable epilepsy
quently has such an event as an antecedent, serving as or refractory epilepsy, is defined as failure of adequate
a risk factor. A classic example is the emergence of sei- trials of two tolerated, appropriately chosen and used
zures after a significant latent period of several months or antiepileptic drug schedules (whether as monotherapy
even years following a stroke or encephalitis. In an anal- or in combination) to achieve sustained seizure freedom
ysis of several studies conducted between 1990 and (Kwan et al., 2009). This definition requires expert epi-
2005, on diverse populations from different parts of lepsy care, since there are numerous cases of patients
the world, the incidence of isolated unprovoked seizures using multiple agents in combination but often inade-
ranged between 42.3 and 61 cases per 100 000 persons quate dosing, which have as yet, or unrightfully, been
per year (Loiseau et al., 1990; Hauser et al., 1993; declared as drug-resistant epilepsy cases. This, of course,
Forsgren et al., 1996; Jallon et al., 1997; MacDonald has a direct implication on quality of care and timely
et al., 2000; Olafsson et al., 2005). In general, these stud- referral to presurgical evaluation of adequate cases.
ies revealed that the incidence of acute symptomatic sei-
zures, although variable, tends to be lower than that of
Active epilepsy and epilepsy in remission
epilepsy. Acute symptomatic seizures are predominant
in men, in patients younger than 1 year, and patients older Active epilepsy and epilepsy in remission are opposite
than 65 years. The variability in results regarding the but elusive and poorly defined terms. These terms, how-
incidence of acute symptomatic seizures may be ever, become important for understanding that different
explained by different methodologies in case validation types of epilepsy behave in different ways and a patient’s
and definitions used (Hauser and Beghi, 2008). epilepsy may be active or in remission, depending on an
EPIDEMIOLOGY OF EPILEPSY 161
elaborate matrix of factors. It is complicated to define of the investigated susceptible population. This is usually
when to consider a patient’s epilepsy as in remission, expressed as number of new cases per 100 000 persons
but it is clear that these patients may not be included in per year.
some epidemiologic studies.
Public health surveillance
Epilepsy resolved
Public health surveillance is defined as the ongoing sys-
Epilepsy resolved is a term adopted recently by the ILAE tematic collection, analysis, and interpretation of health
Task Force to define patients with age-dependent epilepsy data, necessary for the design, implementation, and eval-
syndromes that have passed the applicable age and who no uation of public health prevention programs (German
longer have seizures, as well as those who remained et al., 2001). This ongoing process provides researchers
seizure-free for over 10 years, of which at least 5 years with a dynamic view of incidence, prevalence, morbidity,
were without any seizure medication (Fisher et al., 2014). and mortality related to epilepsy, utilizing sources such
Therefore, since epilepsy represents a heterogeneous as national mortality registries and hospital discharge
group of diseases with variable duration, typical age of data. This study permits constant monitoring of health-
onset (and resolution times in age-limited entities), sever- care program effectiveness and medical policy success.
ity, and manifestations, there is great variability in the Public health surveillance involves the systematic iden-
registration and inclusion of patients in various epidemi- tification of epilepsy cases at multiple points of contact
ologic studies. Variability and unpredictability of epi- (hospitals, clinics, physician offices, electroencephalog-
lepsy dynamics are important factors responsible for raphy (EEG) laboratories, schools, etc.) using a broad
the skewing of the incidence and prevalence of epilepsy definition, or inclusion criteria matrix, with high sensitiv-
and must be taken into account when studying epilepsy ity, followed by a more specialized and refined criteria
epidemiology. matrix for high specificity, thus assuring broad and cor-
rect identification of cases (Banerjee et al., 2009).
EPIDEMIOLOGIC METHODOLOGIES
AND CHALLENGES Epidemiologic research
A recent study estimated that there are roughly 65 million Epidemiologic research is typically a more elaborate and
people living with epilepsy worldwide (Ngugi et al., 2010); detailed study, conducted over a relatively short period of
however, since then, the number has surely increased. In time, usually for deliberate testing of a novel hypothesis
developed countries, the annual incidence of epilepsy regarding a specific healthcare paradigm. These studies
was estimated to be around 50 per 100 000 population describe the incidence rate, prevalence, and mortality
and the prevalence was estimated around 700 per rate of epilepsy, as well as defining high-risk groups,
100 000 population (Hirtz et al., 2007). In the developing associated risk factors, comorbidities, causes, severity,
world, with middle and low income, the numbers are esti- and outcome (Thurman et al., 2011).
mated to be higher (Hauser, 1995; Ngugi et al., 2010). A successful epidemiologic study or surveillance pro-
Proper study and analysis of epilepsy and epidemiologic gram should be economic and avoid collecting unneces-
characteristics pose an enormous challenge, partially due sary data. It should be accepted and easily tolerated by
to heterogeneous definitions, different methods of compar- subjects. It should be accurate, with sensitivity, specific-
ison, and different resources. There are two mechanisms ity, and positive predictive value for measured informa-
used for the epidemiologic study of epilepsy: public health tion as high as possible. It should reliably represent the
surveillance and epidemiologic research. general population of patients – since some epileptic
patients will never be admitted to hospital, basing col-
Point prevalence lected data solely on discharge registries will poorly rep-
Point prevalence is calculated as the total number of resent that population (Thurman et al., 2011).
patients diagnosed with active epilepsy on the prevalence There are several commonly used data sources for sei-
day, divided by the total population under study on that zures and epilepsy-related surveillance and epidemio-
prevalence day. It is of note that active prevalence differs logic studies:
from lifetime prevalence, in that the latter includes
patients with a past history of epilepsy, epilepsy in remis- DIRECT POPULATION SURVEYS
sion, or even resolved epilepsy (Banerjee et al., 2009).
These studies are more advantageous in small popula-
tions, where data can be gathered via door-to-door
Incidence
interview of patients and their proxies. In recent years
Incidence of epilepsy is defined as the number of new these studies have been replaced in part by telephone-
cases of epilepsy over a specific time period, as part and web-based questionnaires; however, as modern life
162 S. ABRAMOVICI AND A. BAGIĆ
is ever more demanding, people do not find the time to ostracism,” with well-known negative consequences:
participate and samples may be underrepresentative shame, withdrawal, isolation, limited access to
quantitatively as well as qualitatively. Another disadvan- resources, and even various forms of discrimination
tage of this data source is occasional cognitive difficulty prevalent even in modern society (Link and Phelan,
the patients may have. This factor may cause poor recall 2006; Hatzenbuehler et al., 2013). Ironically, what mod-
of important information like age of onset and date of ern society focuses most on – “good medical and surgical
last seizure. There is also limited detection of nonconvul- treatment” – represents only a smaller part of the Lennox
sive seizures, which leads to skewed reports. pyramid (Lennox, 1960). Stigma as a sociopsychologic
construct was originally conceptualized, best defined,
EXISTING CODED DATA and most profoundly studied by the sociologist Erving
Goffman (1922–1982), who defined it as “the process
This information may be based on International Classi- by which the reaction of others spoils normal identity”
fication of Diseases, ninth edition (ICD-9), ICD-10, and (Goffman, 1986). The concept has been further devel-
Systematized Nomenclature of Medicine – Clinical oped (Link and Phelan, 2001), in particular with a more
Terms (SNOMED-CT)-coded discharge diagnosis. The comprehensive understanding of its public health impli-
data comes from national or regional registries, hospital cations (Link and Phelan, 2006) and recognition as
discharge registries, and coronary registries. Studies “a fundamental cause of population health inequalities”
based on coded data are prone to error due to occasional (Hatzenbuehler et al., 2013). Just two short decades ago,
lack of training and individual coding interpretation and it was eloquently stated that “the history of epilepsy can
coder knowledge. There is often limited and poor iden- be summarized as 4000 years of ignorance, superstition,
tification of nonepilepsy behavioral spells, due to lack and stigma followed by 100 years of knowledge, super-
of readily available EMUs (Salinsky et al., 2011), leading stition, and stigma” (Kale, 1997). Considering the pro-
to erroneous diagnosis of epilepsy that may skew registry found effects of stigma on public and overall visibility
data as well. of affected persons, modern epidemiologic studies have
to incorporate these realizations in order to achieve more
EXISTING UNCODED DATA accurate results.
These data may be gathered from additional nonmedi-
cal figures such as teachers, religious leaders, and DESCRIPTIVE EPIDEMIOLOGY
traditional healers in some populations. Pharmacy OF EPILEPSY
prescription records regarding antiseizure medication
represent another data source. However, since there is The epidemiologic study of epilepsy mandates the con-
growing use of antiseizure medication in the treatment stant gathering of information regarding patient demo-
of depression, migraine, and some pain disorders, a graphics, incidence, prevalence, seizure types, etiology,
growing number of patients without epilepsy are being and risk factors. Cross-checking information from mul-
prescribed and may be erroneously identified. EEG lab- tiple sources may improve study accuracy, by increasing
oratories represent another commonly used source of case ascertainment. Epidemiologists increasingly deploy
information. advanced automated data-mining algorithms and data-
base analytics for better case ascertainment and enhanced
data accuracy, although case ascertainment and valida-
STIGMA
tion still represent major challenges. Nevertheless, epi-
Special emphasis should be placed on the role of stigma lepsy was estimated to affect 65 million people
in the skewing of epidemiologic studies and treatment worldwide, with roughly 80% of these people living in
gaps. What we understand nowadays as stigma leads the developing world (Ngugi et al., 2010; Thurman
to the regrettable, unnecessary, and preventable suffering et al., 2011). According to the fact sheet on epilepsy
of patients with epilepsy due to social factors which we (no. 999) published by the World Health Organization
all experience. Over 50 years ago, William G. Lennox in May 2015, epilepsy accounts for 0.75% of the global
(1884–1960) conceptualized his infamous “epilepsy burden of disease (http://www.who.int/mediacentre/
pyramid” (Lennox, 1960). The foundation and main factsheets/fs999/en/), as measured by combining years
structure of the pyramid are built of “social-emotional of life lost due to premature mortality and years lived
conditions” created by society and family that may in less than full health (Fig. 10.1). In 2012 epilepsy alone
be healthy and lead to “public understanding and was responsible for 20.6 million disability-adjusted life-
acceptance,” along with family support instilling years (DALYs) lost, and was comparable to DALY of
“security, courage and hope.” Alternatively, the structure breast cancer in woman and lung cancer in men
may be broken, leading to “misconceptions and (Murray et al., 2013).
EPIDEMIOLOGY OF EPILEPSY 163
Fig. 10.1. Disability-adjusted life-year rates from epilepsy by country (per 100 000 persons) in 2004. (Reproduced from https://
commons.wikimedia.org/wiki/File:Epilepsy_world_map_-_DALY_-_WHO2002.svg).