Seizure 2001; 10: 306–316

doi:10.1053/seiz.2001.0584, available online at http://www.idealibrary.com on

CPD Education and self-assessment

The epidemiology of epilepsy: the size of the problem

G. S. BELL & J. W. SANDER

Institute of Neurology, University College London, Queen Square, London WC1N 3BG; National
Society for Epilepsy, Chalfont St Peter, Buckinghamshire SL9 0RJ

Correspondence to: Professor Ley Sander, 33 Queen Square, London WC1N 3BG

The prevalence of epilepsy is generally taken as between 5 and 10 cases per 1000 persons, and the overall incidence as about
50 cases per 100 000 persons. The rates are dependent on case ascertainment and on definitions used. The prognosis depends
on many factors, including the number of seizures at presentation, the seizure type and the use of anti-epileptic drugs. Epilepsy
carries an excess mortality; the cause of death can be unrelated to epilepsy, related to the underlying disease causing epilepsy,
or related to epilepsy itself.
c 2001 BEA Trading Ltd

Key words: epilepsy; epidemiology; incidence; prevalence; prognosis; mortality.

EPIDEMIOLOGY divided by the total person-time at risk during that

period. It is generally expressed as the number of cases
Epidemiology is the study of the distribution and per 100 000 people in the population per year. The
determinants of disease in human populations; the point prevalence is the number of diseased persons in
study of the dynamics of a medical condition in a defined population at one point in time, divided by
the community. It can be divided into three aspects: the number of persons in that population and time.
descriptive, analytical and experimental. Despite epilepsy being amongst the most common
Descriptive epidemiology concerns the incidence serious neurological conditions, the reported incidence
and prevalence, and natural history (prognosis and and prevalence figures vary widely. This can be due to
mortality) of a condition. Descriptive studies may be differences in case ascertainment, differences in age
thought of as observational, with no designed control groups studied, and differences in the location of the
group. Analytical epidemiology compares those with a study.
disease or risk factor with those without, for example
in cross-sectional, cohort and case control studies. Case ascertainment
Studies under conditions that allow an investigator
to control relevant factors constitute experimental Until the 1960s, studies of epilepsy were based on
epidemiology. The epidemiology of epilepsy is largely patients seen in tertiary referral centres1 . These tended
based on descriptive and analytical studies. to show that epilepsy was a chronic, progressive
and incurable disorder, as milder cases were under-
represented. Similarly, studies based on a retrospective
DEFINITIONS review of medical notes for seizures, anti-epileptic
drugs (AEDs) or diagnostic codings may lead to
The incidence rate of a condition is the number of inaccuracy and under-reporting2 . In a study of subjects
persons who become diseased during a defined period with epilepsy found during a field survey in Warsaw,

1059–1311/01/040306 + 11 $35.00/0 c 2001 BEA Trading Ltd

The epidemiology of epilepsy 307

one third had never been treated for epilepsy, over THE INCIDENCE AND PREVALENCE OF
a quarter had stopped treatment, and over a quarter EPILEPSY
had not been diagnosed as having epilepsy. Some did
not suspect that their transient symptoms could be The overall incidence of epilepsy, excluding febrile
epilepsy3 . convulsions and single seizures, is generally taken
Incidence figures vary because of differences in to be about 50 cases per 100 000 persons per
inclusion criteria. If febrile seizures, neonatal seizures year (range 40 to 70 per 100 000/year)2 in devel-
or single seizures are included the figures may be oped countries. In developing countries a range of
elevated several-fold. Similarly, acute symptomatic 100 to 190 per 100 000 per year is generally quoted2 .
seizures account for up to one third of all new- The prevalence of epilepsy is usually regarded as
onset seizures, and are often not classified as epilepsy. between 5 and 10 cases per 1000 persons, excluding
Clearly, the inclusion criteria should be specified when febrile convulsions, single seizures and inactive cases.
epidemiological rates are quoted. The lifetime prevalence of seizures is between 2
Prevalence data will vary according to whether only and 5%. Over two-thirds of patients enter long-term
active epilepsy is considered, and upon the definition remission, and subsequent relapse is uncommon.
of ‘active epilepsy’.

CLASSIFICATIONS OF EPILEPSY
Age groups
Epilepsy can be classified in a number of ways, and
The incidence of epilepsy is high in childhood, each has its uses and problems. The prognosis of
decreases in young people and rises again in the epilepsy is often determined to a large extent by its
elderly4 . About 50% of cases of epilepsy start in aetiology, and so a classification including this is to be
the two extremes of life, with half of those being encouraged.
under one year. Epilepsy has a high incidence in
the elderly and this may be because of demographic
changes in the population, with increasing numbers
of people surviving into old age, with significant Seizure type
morbidity. The major cause of epilepsy in the
elderly is cerebrovascular disease (CVD), although
paradoxically the incidence of CVD has decreased The seizure type is frequently classified according
over the last three decades. to the International Seizure Classification proposed
by the International League against Epilepsy (ILAE),
which is based on clinical and electroencephalo-
Location graphic (EEG) features of the seizure8 . This scheme
divides seizures into three groups: generalized, partial
Epilepsy occurs throughout the world, but higher (or localization-related) and unclassified. Generalized
incidence figures are generally found from studies in seizures are those in which epileptic discharges
developing countries. The reasons for this may be involve both cerebral hemispheres from the onset of
multiple. The incidence of epilepsy is particularly high the seizure, whereas partial seizures are those in which
in Latin America and in several African countries. the epileptic activity is confined to a focal area of the
Parasitic infections have been suggested as important brain. Generalized seizures are subdivided into tonic–
aetiological factors for epilepsy in these countries, as clonic, absence, myoclonic, atonic, tonic and clonic
have intracranial infections, perinatal brain damage seizures. Partial seizures are subdivided into simple
and hereditary factors5 . No consistent racial dif- partial seizures, where consciousness is preserved, and
ferences have been found, although several studies complex partial seizures where consciousness is im-
from the United States show higher incidence and paired. Partial seizures may also become secondarily
prevalence figures amongst black Americans when generalized if the epileptic activity spreads to involve
compared with white Americans. Epilepsy is usually both cerebral hemispheres.
found to be slightly more common in the lower socio- In general, partial seizures account for most
economic groups. Two recent studies in the UK and cases. The National General Practice Study of
the USA have shown lower prevalence and incidence Epilepsy (NGPSE), a prospective population-based
rates respectively in the South Asian population6, 7 . cohort of patients with newly diagnosed epileptic
However, there have been acknowledged sampling seizures, classified the seizure type of 564 subjects.
problems in both studies, and suggested bias includes At 6 months 11% were classified as having complex
selective migration and healthy worker effects. partial seizures, 3% simple partial seizures, 27%
308 G. S. Bell & J. W. Sander

secondarily generalized seizures, 35% primarily gen- age of onset, number of seizures at onset, the natural
eralized tonic–clonic seizures, and less than 1% each history of the condition and the influence of treatment.
generalized absence and myoclonus. Fourteen percent However, overall between 70 and 80% of people
were mixed (partial or generalized) and 9% were developing epilepsy will go into long-term remission,
unclassifiable4 . usually within the first 5 years.
This classification does not take into account the The most important consideration of prognosis in
aetiology or any anatomical features, and rarely gives epilepsy is the likelihood of seizure freedom, with or
any guide to prognosis. without the use of antiepileptic drugs (AEDs). As most
patients with a history of more than one seizure are
generally prescribed an AED, the prognosis has been
Aetiology more fully investigated in this group. Generally the 1
year remission rate is between 65 and 80%12a .
The aetiology of epilepsy is frequently multifactorial,
and exact attribution of cause is often impossible.
About 60% of epilepsies have no clear cause, although Recurrence after a first seizure
advances in magnetic resonance imaging (MRI) have
increased the number of patients in whom a positive Historically, single seizures have been regarded as
putative aetiological diagnosis is possible. A recent different from epilepsy, largely because early studies
community-based MRI study found that in newly suggested that most subjects with a first seizure have
diagnosed patients, a relevant putative aetiology could no further attacks. However, as early as 1975 it was
be found in 70% of those with partial onset seizures found that the incidence of epilepsy was much higher
and 30% of those with generalized seizures9 . The than that of single seizures, suggesting that many
increasing availability of high-resolution MRI should such patients will undergo seizure recurrence12 . In
improve the ability to determine the causes of epilepsy. a General Practice study in Tonbridge, UK, it was
found that seizures recurred in four out of five patients
after the first seizure13 . A recent retrospective study of
Epilepsy syndrome recurrence after a first untreated tonic–clonic seizure
in Hong Kong showed that 30% experienced another
In 1989 the ILAE suggested a new classification, now seizure within the first year, and a further 17% within
taking into account seizure type, EEG, and prognos- the next 3 years14 . However, this study ignored all
tic, pathophysiological and aetiological data10 . This but tonic–clonic seizures, which may have affected the
classification also divides epilepsies into seizure type recurrence rate, as the risk of recurrence is higher for
(localization-related, generalized or undetermined), other seizure types15 .
but further divides them into idiopathic, symptomatic, There are many problems involved in determining
or cryptogenic, according to the putative cause. Symp- the rate of recurrence after a first seizure, most
tomatic epilepsy is considered to be the consequence related to case ascertainment. Although most subjects
of a known disorder of the central nervous system, experiencing a first tonic–clonic seizure will seek
cryptogenic epilepsy is presumed to be due to an medical advice, other seizure types may not be
underlying but unidentified focal abnormality and immediately recognized as epilepsy. In a large study
idiopathic epilepsy is reserved for those syndromes in Australia 17% of those presenting with tonic–
which are presumed to be inherited. clonic seizures had had previous tonic–clonic seizures
Unfortunately, although this is a comprehensive which were unwitnessed, not medically assessed
classification, it is too complicated to be of utility or incorrectly diagnosed. Of 43 patients presenting
in clinical practice and epidemiological research, and with non-tonic–clonic seizures, 60% had experienced
takes no account of recent developments in neuro- similar episodes in the past and had not sought medical
imaging and neurogenetics11 . advice16 .
The risk of seizure recurrence is greater in the
first weeks or months after an initial seizure. If there
PROGNOSIS OF EPILEPSY is a long interval between the initial seizure and
recruitment into a study a second seizure may already
Prognosis in epilepsy is usually taken as the prospect have occurred, thus excluding the subject from the
of attaining complete seizure freedom once a pattern study. A large hospital-based study in London showed
of recurrent epileptic seizures has been established. the recurrence at 2 years to be 51% in those recruited
Seizures are a symptom of disease, and do not consti- in under a week, but only 15% in those recruited more
tute a single well-defined illness. As a consequence of than 8 weeks after the first seizure17 . Community-
this, the prognosis depends on many factors; aetiology, based studies are perhaps more likely to pick up
The epidemiology of epilepsy 309

first seizures than those based in hospitals; in the study, subjects presenting within 7 days of a first
NGPSE, when all seizure types were considered, 67% unprovoked tonic–clonic seizure were randomized to
of subjects had a recurrence within 12 months of the receive AEDs immediately or only in the event of
first seizure, and 78% within 36 months, all seizure seizure recurrence21 . The overall risk of recurrence
types are considered18 . was greatest initially. Starting AED treatment after the
The first seizure is also not a clinical entity in itself, first seizure, compared with starting treatment after
and different aetiologies may have different natural seizure recurrence, reduced the risk of relapse over
histories. the first 2 years, but did not affect the long-term
probability of 1 or 2 year remission.

Seizure type
AED usage
Seizure type has been thought to be an important
predictor of recurrence risk. Partial seizures have a Several studies have looked at the prognosis with use
poor prognosis for remission: in studies of American of different AEDs. An early study showed improved
adults complex partial seizures were controlled in seizure control with carbamazepine or phenytoin,
only 23 to 26% of patients, whereas secondarily compared with phenobarbitone or primidone22 . In
generalized attacks were controlled in 48 to 55% a more recent study of 525 patients with newly
at 1 year19 . The early findings of the NGPSE also diagnosed epilepsy, patients were either given an
suggested that recurrence rates were much higher appropriate AED for the type of seizures and other
with partial seizures than with tonic–clonic seizures, clinical indications, or were randomized to receive an
and with remote symptomatic seizures than with unknown AED. There was no significant difference in
acute symptomatic seizures18 . However, later review the seizure-free rate between those taking established
showed no significant difference for chances of AEDs and those taking newer ones23 . Amongst the
remission between primary generalized tonic–clonic patients who had never before received an AED, 47%
seizures, partial seizures only, any partial seizures, became seizure-free on their first AED and a further
secondarily generalized seizures or any generalized 14% became seizure-free on a second or third drug.
tonic–clonic seizures20 . This study showed that the Additionally, 3% were controlled by a combination
predominant clinical factor that seemed to predict of two AEDs. These results suggest that the response
prognosis was the number of seizures occurring in to the first AED is a powerful prognostic factor.
the first 6 months after presentation. Other factors, Furthermore, the reason for failure of the first AED is
including seizure type, seemed to be unimportant or prognostically important; of those who changed drugs
to be variables associated with this single important because of lack of efficacy, 11% became seizure-free
prognostic factor. The authors suggest that this may on a subsequent drug. Where the first drug failed
be because epilepsy in any individual patient has an because of intolerable side-effects or idiosyncratic
inherent severity and response to treatment, and that reactions, 41 and 55% respectively, became seizure-
severely affected patients are difficult to control from free on a second AED.
early in their condition. They also note that a potential The cause of drug resistance in epilepsy is poorly
confounding factor is the differential mortality for understood. Early work has suggested there may
patients with severe underlying pathological findings be non-tumoural overexpression of a multidrug-
who were followed for shorter periods because of resistance protein in some cases of focal cortical
death. The finding of the prognostic significance of dysplasia24 . Whether this can be extended to other
the number of early seizures is in accordance with causes of epilepsy is unclear.
other studies showing that most patients who entered
remission did so in the first 2 years, and that the
prospect of entering remission decreased as time Refractory epilepsy
elapsed13 .
Community and hospital studies of newly diagnosed
epilepsy have shown that 20–30% of patients do not
Time of starting AEDs enter remission. The prevalence of persistent seizures
seems to be higher in subjects with symptomatic or
Those who have had two or more unprovoked seizures cryptogenic epilepsy than in those with idiopathic
are usually started on AEDs. Patients with just one epilepsy23 .
unprovoked seizure are not routinely treated in the However, between 10 and 20% of patients referred
UK, perhaps in view of the fact that not all these to clinics with ‘refractory epilepsy’ may have been
will go on to develop epilepsy. In a hospital-based misdiagnosed2, 25 . In the latter study, of 92 patients
310 G. S. Bell & J. W. Sander

referred to a single consultant with a diagnosis of of these children had a decelerating pattern of seizures;
‘refractory seizures’, 13% did not have epilepsy, that is the child became free of seizures without AEDs
and a further 44% were significantly improved by or the time intervals between seizures successively
the optimal use of AEDs or epilepsy surgery. The increased29 . This study included only children with
authors found that incomplete history-taking and tonic–clonic seizures, and only those with idiopathic
misinterpretation of the EEG appeared to be the or remote symptomatic seizures.
principal causes for misdiagnosis. In the Warsaw field study, almost one-third of those
who had never been treated were free of seizures
for more than 5 years3 . Some of these had had
Withdrawal of AEDs frequent generalized seizures in the past. A small study
in Finland, looking at untreated epilepsy, found the
If long-term seizure control is achieved with AED use, probability of remission to be 42% by 10 years after
the decision has to made whether or not to withdraw the onset of epilepsy30 .
the AEDs. For each patient the risk–benefit ratio has to Epilepsy is less likely to be treated in the developing
be considered; AEDs are not without side-effects, but world, and may provide an estimate of the prognosis
employment and driving risks ensue if seizures return. of untreated epilepsy. However, the aetiologies may
Numerous studies have been undertaken to try to be very different from those in the developed world,
establish the risk of recurrence after AED withdrawal. and this may affect the prognosis. If epilepsy rarely
A meta-analysis found the risk to be about 25% at 1 remits when untreated, then the prevalence rates for
year and 29% at 2 years26 . There appeared to be an epilepsy would be expected to be much higher in the
increased risk of relapse associated with adolescent- developing world. Although this has sometimes been
onset epilepsy; childhood-onset epilepsy has the most found to be the case it is by no means always so. The
favourable risk figure. The large Medical Research lack of higher prevalence rate in the developing world
Council study compared slow withdrawal of AEDs could be explained in part by the higher mortality rate
with continuation of AEDs, and found the risk of of epilepsy or the shorter life expectancy. Although
recurrence to be substantially larger at 2 years in case ascertainment could be different in the two areas,
those randomized to slow withdrawal compared with it is more likely that cases in remission would be
those randomized to continuation (41% and 22% missed in the developing world, rather than those with
respectively)27 . By 2 to 4 years after the start, patients active epilepsy. A more likely explanation is that many
still seizure free may have a higher risk of relapse untreated cases enter spontaneous remission.
when randomized to continue medication than those A study in a rural area of Northern Ecuador in
randomized to withdrawal. The reasons for this are not the early 1990s identified over 1000 people with
certain; patients randomized to continue medication epilepsy31 . Only 37% of these had ever received
may have withdrawn after one or two years, or those AEDs. Despite this, over 40% were free of seizures,
who were still seizure free taking no AEDs may of whom almost two-thirds had never received AEDs.
have remitted, whereas those still seizure free on Additionally, a treatment programme was instituted
medication may only have been so because of their amongst almost 200 people with active untreated
AEDs. This study found that the most important epilepsy. The effectiveness of new treatment with
predictors of risk were longer seizure-free periods, AEDs was as good as that seen in new cases in
which reduced the risk, and the use of more than developed countries.
one AED before remission and a history of tonic–
clonic seizures or myoclonic seizures, which increased
the risk. Further follow-up of the subjects in this Prognosis according to aetiology and response
study who had a seizure recurrence showed that the to treatment
prognosis is still relatively good whether or not AEDs
were withdrawn28 . It is seen that the prognosis of epilepsy is based
on multiple factors including the number of seizures
at presentation, seizure type and AED usage. The
Untreated epilepsy syndromic classification, although complicated, dif-
ficult to apply in many forms of clinical practice
In the developed world, most people are started on and nowadays somewhat out of date, nonetheless is
AEDs at the time of their second seizure or earlier. a useful basis for grouping seizures into prognostic
It is hard to conduct studies comparing the course groups. People with seizures can be categorized into
of treated epilepsy with that of untreated epilepsy. four prognostic groups19, 32 :
A study in Holland of children with untreated newly
diagnosed tonic–clonic seizures found that about 40% (1) Excellent prognosis. This group comprises
The epidemiology of epilepsy 311

about 20–30% of all people who develop defined as the observed number of deaths in the
epileptic seizures. Usually only a few seizures study population divided by the expected number of
occur and spontaneous remission is the rule. deaths in the study population, if the age- and sex-
Conditions include benign idiopathic or famil- specific mortality rates were the same as those of the
ial neonatal convulsions33 , the benign partial standard population. Thus the SMR measures how
epilepsies34 , benign myoclonic epilepsy of much more, or less, likely a person is to die in the
infancy and acute symptomatic seizures35 . study population compared with someone of the same
age and sex in the standard population. Despite an
(2) Good prognosis. This group comprises
overall good prognosis for seizure control, epilepsy is
about 30–40% of people who develop epilepsy.
a potentially life-threatening condition and carries an
Seizures are easily controlled with AEDs
excess mortality. The SMR for epilepsy is 2–3, and
and, once remission is achieved, it is usually
may be higher in males than females.
permanent. Conditions include childhood
absence epilepsy34 , epilepsy with generalized
tonic–clonic seizures on awakening35 , non-
Cause of death
specific generalized tonic–clonic seizures in
patients with no abnormal neurological signs,
and some of the localization-related epilepsies. The cause of death in people with epilepsy can
be unrelated to epilepsy, related to the underlying
(3) AED-dependant prognosis. In this group, com- disease causing epilepsy, or related to epilepsy; the
prising some 10–20% of people who develop distinction between these is not always clear-cut in
epilepsy, there is a long-term tendency to an individual case37 . In similar ways to the other
seizures. AEDs suppress seizures and patients epidemiological variables, mortality rates in epilepsy
may achieve remission, but may relapse if AEDs can be hard to quantify, particularly in retrospective
are stopped. Juvenile myoclonic epilepsy is studies. Death certificates do not always mention
well known for this behaviour34 , which also epilepsy as the direct or underlying cause of death,
occurs in the majority of localization-related and, as noted earlier, people with mild epilepsy may
epilepsies. Some patients with localization- not have consulted the medical profession. Prospective
related epilepsies may be treated by surgery, studies based in the community are the most likely to
with change of prognostic group. provide an accurate estimate of SMRs, but the studies
need to be large enough to be able to assess less
(4) Bad prognosis. Up to 20% of people with
frequent causes of death, such as certain neoplasms
epilepsy may be in this group, in whom AEDs
and suicide38 .
are palliative rather than suppressive of seizures.
A large study in Sweden of mortality in patients
Despite intensive treatment with AEDs, seizures
once hospitalized for epilepsy (4001 deaths)38 ,
tend to occur. A few patients may respond to
a smaller community based study in the UK
novel AEDs or to surgical treatment. Conditions
(214 deaths)39 and a hospital based study in the
with this poor prognosis include seizures associ-
Netherlands (404 deaths)40 together provide some
ated with neurological deficit present from birth
statistics. Despite the large differences in case ascer-
(tuberous sclerosis, Sturge–Weber syndrome,
tainment, for many causes of death there is remarkable
malformations, cerebral palsy, etc), epilep-
agreement (Table 1). The difference in overall SMR
sia partialis continua34 , progressive myoclonic
may be that mild cases are never hospitalized and so
epilepsies and other progressive neurological
do not contribute to the deaths in the Swedish study.
diseases, West syndrome36 , Lennox–Gastaut
In the NGPSE, with a median follow-up period of
syndrome34 and others in which atonic/tonic
11.8 years, SMRs were raised in patients with acute
seizures are a prominent feature, partial seizures
symptomatic epilepsy, remote symptomatic epilepsy
associated with gross structural lesions and
and epilepsy due to congenital neurological deficits39 .
some of the localization-related cryptogenic
This increase was most prominent in the initial years
epilepsies.
following diagnosis.

MORTALITY IN EPILEPSY Cause of death unrelated to epilepsy

Definition Clearly, people with epilepsy are subject to the same

intercurrent illnesses as the rest of the population.
The mortality rate for a disease is generally quoted However, some studies have shown that the risk of
as the standardized mortality ratio (SMR), which is dying from ischaemic heart disease may be increased
312 G. S. Bell & J. W. Sander

Table 1: Cause specific deaths in 3 large cohorts

Stockholm study39 UK NGPSE40 Heemstede study41

Follow-up (years) 3–12 10–14 2–40
Age 15+ 95% CI All 95% CI All 95% CI
Overall SMR 3.6 3.5–3.7 2.1 1.8–2.4 2.8 2.5–3.1
Malignancies 2.6 2.4–2.8 2.6 1.9–3.4 1.7 1.3–2.1
Malignancies except brain tumours 2.0 1.9–2.2 1.9 1.3–2.6 1.5 1.1–1.9
Neoplasm lung 2.7 2.2–3.2 2.7 1.5–4.6 1.9 1.1–2.9
CNS neoplasms 29.9 25.3–35.1 5.4 1.9–11
Ischaemic heart disease 2.5 2.3–2.7 1.1 0.7–1.6 1.2 0.8–1.6
Cerebrovascular disease 5.3 4.9–5.8 3.2 2.2–4.4 2.5 1.6–3.7
Pneumonia 4.2 3.6–4.8 5.9 4.1–8.0 8.8 4.6–14
Suicide 3.5 2.6–4.6 1.7 0.6–3.4

in epilepsy38 , while others have not shown this39–41 . Phenytoin hypersensitivity can lead to haemopoietic
and immune complications, which may be fatal,
and ethosuximide can also cause haematological
Cause of death related to underlying disease disorders. Valproate can cause hepatic failure and
pancreatitis and lamotrigine may cause ‘Steven’s
As seen in the table, the SMR is greatly increased Johnson syndrome’, toxic epidermal necrolysis or
for CNS tumours. Other underlying diseases which disseminated intravascular coagulation. Felbamate is
increase the mortality rate include cerebrovascular associated with idiosyncratic aplastic anaemia and
disease39 , cerebral infective agents and inherited with liver failure. Idiosyncratic drug reactions are
disorders. unusual.
Most studies of epilepsy surgery report that one
to five percent of patients have moderate to severe
Cause of death related to epilepsy adverse sequelae, including, occasionally, death45 .
Vagal nerve stimulation does not appear to carry a
Epilepsy related deaths can be divided among seizure large risk of excess death.
related deaths (e.g. status epilepticus, accidents
including drowning), treatment related deaths, and
Sudden Unexpected Death in Epilepsy. Suicide
Status epilepticus affects approximately 50 patients
per 100 000 population annually, and is associated
The suicide rate has been found to be increased in
with significant mortality rates of 15 to 20% in adults
some studies of epilepsy38, 46 , but not in others40, 47 .
and 3 to 15% in children42 . Death due to accidents,
Early studies showed the suicide rate was particularly
most commonly drowning43 , is unfortunately not rare.
increased in those with temporal lobe epilepsy, and in
Drowning most commonly occurs whilst swimming
the early years of the condition46 .
or in the bath. Supervision is essential, but swimming
should not be discouraged. A recent report suggested
that those with a history of tonic seizures are at
particular risk44 . Other accidental causes of death SUDEP
include head trauma and burns43 .
Sudden Unexpected Death in Epilepsy (SUDEP) is
a rare cause of death overall, but is more common
Treatment related deaths in those with chronic epilepsy. The rate of sudden
unexpected death is over 20 times higher in those
Treatment related deaths in epilepsy can be drug with epilepsy than in the general population48 . It
related, or related to surgical procedures, but these are occurs in both symptomatic epilepsies and primary
rare events. generalized epilepsy, is rare in remission49 , and is
Drug related deaths are either dose related, and often though to be a seizure-related event. It has been
therefore usually occur when taken in overdosage, or suggested that SUDEP is associated with subthera-
idiosyncratic. Phenobarbitone overdosage can cause peutic post-mortem serum AED levels50 , suggesting
respiratory depression and coma, as can massive non-compliance. However, a study comparing levels
overdose with valproate. Overdosage with phenytoin in subjects dying from SUDEP with those in patients
may cause respiratory and circulatory depression. who had epilepsy but died from other causes, found
The epidemiology of epilepsy 313

no difference in serum AED levels51 . Recent work 5. Senanayake, N. and Roman, G. C. Epidemiology of epilepsy
suggests that both central and obstructive apnoea may in developing countries. Bulletin of the World Health
Organisation 1993; 71: 247–258.
be important in the genesis of SUDEP52 . 6. Wright, J., Pickard, N., Whitfield, A. and Hakin, N. A
population-based study of the prevalence, clinical character-
istics and effect of ethnicity in epilepsy. Seizure 2000; 9:
Other causes of death 309–313.
7. Annegers, J. F., Dubinsky, S., Coan, S. P., Newmark, M. E. and
Roht, L. The incidence of epilepsy and unprovoked seizures
Whilst death in epilepsy can be attributed directly or in multiethnic, urban health maintenance organizations.
indirectly to the cause or effects of epilepsy in some Epilepsia 1999; 40: 502–506.
cases, other causes for elevated SMRs remain unclear. 8. Proposal for revised clinical and electroencephalographic
It is likely that death due to cancer is more frequent in classification of epileptic seizures. The commission for
classification and terminology of the International League
people with epilepsy than in the general population. Against Epilepsy. Epilepsia 1981; 22: 489–501.
It is postulated that the increase in death due to 9. Everitt, A. D., Birnie, K. D., Stevens, J. M., Sander, J. W.,
pneumonia might be secondary to aspiration during Duncan, J. S. and Shorvon, S. D. A prospective MRI study of
seizures; this hypothesis has not yet been tested. the aetiology of epileptic seizures in a large community based
cohort. Journal of Neurology, Neurosurgery and Psychiatry
1998; 65: 417.
10. Proposal for revised classification of epilepsies and epileptic
CONCLUSION syndromes. Commission on classification and terminology of
the International League Against Epilepsy. Epilepsia 1989; 30:
389–399.
• Epilepsy is a common condition and affects 1 in 11. Everitt, A. D. and Sander, J. W. Classification of the
200 people of the UK population at any one time. epilepsies: time for a change? A critical review of the
• The highest incidence is at the two extremes of International Classification of the Epilepsies and Epileptic
Syndromes (ICEES) and its usefulness in clinical practice
life, with 50% of cases being under age one or and epidemiological studies of epilepsy. European Neurology
over 60. 1999; 42: 1–10.
• Recent advances in MRI neuroimaging have 12. Hauser, W. A. and Kurland, L. T. The epidemiology
of epilepsy in Rochester, Minnesota, 1935 through 1967.
increased the percentage of people in whom a
Epilepsia 1975; 16: 1–66.
putative aetiology can be found. 12a. Sander, J. W. and Sillanpaa, M. Natural history and prognosis.
• The majority of people with epilepsy have good In: Epilepsy: A Comprehensive Textbook (Eds J. Engel and
seizure control with AEDs. T. A. Pedley). Philadelphia, Lippincott-Raven Publishers,
1997: pp. 69–86.
• The risk of seizure recurrence is greater in the first 13. Goodridge, D. M. and Shorvon, S. D. Epileptic seizures in
weeks or months after an initial seizure. a population of 6000. II: treatment and prognosis. British
• The prognosis is largely determined by the Medical Journal 1983; 287: 645–647.
14. Hui, A. C., Tang, A., Wong, K. S., Mok, V. and Kay, R. Recur-
background aetiology. rence after a first untreated seizure in the Hong Kong Chinese
• A large number of seizures at an early stage may population. Epilepsia 2001; 42: 94–97.
indicate a guarded prognosis. 15. So, N. K. Recurrence, remission, and relapse of seizures.
Cleveland Clinic Journal of Medicine 1993; 60: 439–444.
• The mortality rate is slightly but significantly 16. King, M. A., Newton, M. R., Jackson, G. D., Fitt, G. J.,
increased in those with epilepsy. Mitchell, L. A., Silvapulle, M. J. et al. Epileptology of the
• In newly diagnosed epilepsy death is usually first-seizure presentation: a clinical, electroencephalographic,
and magnetic resonance imaging study of 300 consecutive
attributed to the underlying aetiology. patients. Lancet 1998; 352: 1007–1011.
• In chronic epilepsy, a significant cause of death is 17. Hopkins, A., Garman, A. and Clarke, C. The first seizure in
SUDEP. adult life. Value of clinical features, electroencephalography,
and computerised tomographic scanning in prediction of
seizure recurrence. Lancet 1988; 1: 721–726.
18. Hart, Y. M., Sander, J. W., Johnson, A. L. and
Shorvon, S. D. National general practice study of epilepsy:
REFERENCES recurrence after a first seizure. Lancet 1990; 336: 1271–1274.
19. Mattson, R. H., Cramer, J. A. and Collins, J. F. Prognosis
1. Berg, A. T. and Shinnar, S. The contributions of epidemiology for total control of complex partial and secondarily gener-
to the understanding of childhood seizures and epilepsy. alized tonic–clonic seizures. Department of Veterans Affairs
Journal of Child Neurology 1994; 9(Suppl. 2): 19–26. Epilepsy Cooperative Studies No. 118 and No. 264 Group.
2. Sander, J. W. and Shorvon, S. D. Epidemiology of the Neurology 1996; 47: 68–76.
epilepsies. Journal of Neurology, Neurosurgery and Psychiatry 20. MacDonald, B. K., Johnson, A. L., Goodridge, D. M.,
1996; 61: 433–443. Cockerell, O. C., Sander, J. W. and Shorvon, S. D. Factors
3. Zielinski, J. J. Epileptics not in treatment. Epilepsia 1974; 15: predicting prognosis of epilepsy after presentation with
203–210. seizures. Annals of Neurology 2000; 48: 833–841.
4. Sander, J. W., Hart, Y. M., Johnson, A. L. and 21. Musicco, M., Beghi, E., Solari, A. and Viani, F. Treatment of
Shorvon, S. D. National General Practice Study of Epilepsy: first tonic–clonic seizure does not improve the prognosis of
newly diagnosed epileptic seizures in a general population. epilepsy. First Seizure Trial Group (FIRST Group). Neurology
Lancet 1990; 336: 1267–1271. 1997; 49: 991–998.
314 G. S. Bell & J. W. Sander

22. Mattson, R. H., Cramer, J. A., Collins, J. F., Smith, D. B., study of more than 9,000 patients once hospitalized for
Delgado-Escueta, A. V., Browne, T. R. et al. Comparison of epilepsy. Epilepsia 1997; 38: 1062–1068.
carbamazepine, phenobarbital, phenytoin, and primidone in 39. Lhatoo, S. D., Johnson, A. L., Goodridge, D. M., MacDon-
partial and secondarily generalized tonic–clonic seizures. New ald, B. K., Sander, J. W. and Shorvon, S. D. Mortality in
England Journal of Medicine 1985; 313: 145–151. epilepsy in the first 11 to 14 years after diagnosis: multivariate
23. Kwan, P. and Brodie, M. J. Early identification of refractory analysis of a long-term, prospective, population-based cohort.
epilepsy. New England Journal of Medicine 2000; 342: Annals of Neurology 2001; 49: 336–344.
314–319. 40. Shackleton, D. P., Westendorp, R. G., Trenite, D. G. and
24. Sisodiya, S. M., Lin, W. R., Squier, M. V. and Thom, Vandenbroucke, J. P. Mortality in patients with epilepsy:
M. Multidrug-resistance protein 1 in focal cortical dysplasia. 40 years of follow up in a Dutch cohort study. Journal of
Lancet 2001; 357: 42–43. Neurology, Neurosurgery and Psychiatry 1999; 66: 636–640.
25. Smith, D., Defalla, B. A. and Chadwick, D. W. The 41. Hauser, W. A., Annegers, J. F. and Elveback, L. R. Mortality
misdiagnosis of epilepsy and the management of refractory in patients with epilepsy. Epilepsia 1980; 21: 399–412.
epilepsy in a specialist clinic. Quarterly Journal of Medicine
42. Fountain, N. B. Status epilepticus: risk factors and complica-
1999; 92: 15–23.
tions. Epilepsia 2000; 41(Suppl. 2): S23–S30.
26. Berg, A. T. and Shinnar, S. Relapse following discontinuation
of antiepileptic drugs: a meta-analysis. Neurology 1994; 44: 43. Spitz, M. C. Injuries and death as a consequence of seizures
601–608. in people with epilepsy. Epilepsia 1998; 39: 904–907.
27. Randomised study of antiepileptic drug withdrawal in patients 44. Besag, F. M. Tonic seizures are a particular risk factor for
in remission. Medical Research Council Antiepileptic Drug drowning in people with epilepsy. British Medical Journal
Withdrawal Study Group. Lancet 1991; 337: 1175–1180. 2001; 322: 975–976.
28. Chadwick, D., Taylor, J. and Johnson, T. Outcomes after 45. Berg, A. T. and Vickrey, B. G. Outcome measures. In:
seizure recurrence in people with well-controlled epilepsy and Epilepsy: A comprehensive textbook (Eds J. Engel and
the factors that influence it. The MRC Antiepileptic Drug T. A. Pedley). Philadelphia, Lippincott-Raven Publishers,
Withdrawal Group. Epilepsia 1996; 37: 1043–1050. 1997: pp. 1891–1899.
29. van Donselaar, C. A., Brouwer, O. F., Geerts, A. T., Arts, W. F., 46. Barraclough, B. M. The suicide rate of epilepsy. Acta
Stroink, H. and Peters, A. C. Clinical course of untreated Psychiatrica Scandinavia 1987; 76: 339–345.
tonic–clonic seizures in childhood: prospective, hospital based 47. Cockerell, O. C., Johnson, A. L., Sander, J. W., Hart, Y. M.,
study. British Medical Journal 1997; 314: 401–404. Goodridge, D. M. and Shorvon, S. D. Mortality from epilepsy:
30. Keranen, T. and Riekkinen, P. J. Remission of seizures in results from a prospective population-based study. Lancet
untreated epilepsy. British Medical Journal 1993; 307: 483. 1994; 344: 918–921.
31. Placencia, M., Sander, J. W., Roman, M., Madera, A., 48. Ficker, D. M., So, E. L., Shen, W. K., Annegers, J. F.,
Crespo, F., Cascante, S. et al. The characteristics of epilepsy O’Brien, P. C., Cascino, G. D. et al. Population-based study
in a largely untreated population in rural Ecuador. Journal of of the incidence of sudden unexplained death in epilepsy.
Neurology, Neurosurgery and Psychiatry 1994; 57: 320–325. Neurology 1998; 51: 1270–1274.
32. Shorvon, S. D. Medical assessment and treatment of chronic 49. Nashef, L., Garner, S., Sander, J. W., Fish, D. R. and
epilepsy. British Medical Journal 1991; 302: 363–366. Shorvon, S. D. Circumstances of death in sudden death
33. Miles, D. K. and Holmes, G. L. Benign neonatal seizures. in epilepsy: interviews of bereaved relatives. Journal of
Journal of Clinical Neurophysiology 1990; 7: 369–379. Neurology, Neurosurgery and Psychiatry 1998; 64: 349–352.
34. Wallace, S. J. Seizures in children. In: A Textbook of Epilepsy
50. Earnest, M. P., Thomas, G. E., Eden, R. A. and Hos-
(Eds J. Laidlaw, A. Richens and D. Chadwick). Churchill
sack, K. F. The sudden unexplained death syndrome in
Livingstone, 1993: pp. 77–164.
epilepsy: demographic, clinical, and postmortem features.
35. Chadwick, D. Seizures and epilepsy in adults. In: A Textbook
Epilepsia 1992; 33: 310–316.
of Epilepsy (Eds J. Laidlaw, A. Richens and D. Chadwick ).
Churchill Livingstone, 1993: pp. 165–204. 51. Opeskin, K., Burke, M. P., Cordner, S. M. and
36. Wong, M. and Trevathan, E. Infantile spasms. Pediatric Berkovic, S. F. Comparison of antiepileptic drug levels
Neurology 2001; 24: 89–98. in sudden unexpected deaths in epilepsy with deaths from
37. Nashef, L. and Shorvon, S. D. Mortality in epilepsy. Epilepsia other causes. Epilepsia 1999; 40: 1795–1798.
1997; 38: 1059–1061. 52. Langan, Y., Nashef, L. and Sander, J. W. Sudden unexpected
38. Nilsson, L., Tomson, T., Farahmand, B. Y., Diwan, V. and death in epilepsy: a series of witnessed deaths. Journal of
Persson, P. G. Cause-specific mortality in epilepsy: a cohort Neurology, Neurosurgery and Psychiatry 2000; 68: 211–213.
The epidemiology of epilepsy 315

Self-assessment questions

Question 1.

The incidence of epilepsy:

(1) is defined as the number of diseased persons in a defined population at one point in time, divided by the
number of persons in that population and time.
(2) is about 50 per 100 000 per year in developed countries.
(3) is lower in Latin America than in the UK.
(4) is lowest in the elderly.
(5) varies according to case ascertainment and location of the study.

Question 2.

The prognosis of epilepsy:

(1) epilepsy is a chronic, progressive and incurable disorder.
(2) the prognosis is independent of aetiology.
(3) the prevalence of persistent seizures is lower in those with idiopathic epilepsy.
(4) overall, a high proportion of people developing epilepsy will go into long-term remission.
(5) the risk of seizure recurrence is lowest in the first weeks or months after an initial seizure.

Question 3.

Mortality of epilepsy:
(1) the mortality rate in epilepsy is usually expressed as the BMI.
(2) epilepsy does not carry an increased risk of death.
(3) most people with epilepsy will die from SUDEP.
(4) the cause of death in people with epilepsy is always related to the epilepsy itself.
(5) surgery is a rare cause of death in epilepsy.

Question 4.

Drug treatment in epilepsy:

(1) is the major determinant of prognosis.
(2) may occasionally cause death.
(3) is curative.
(4) is always indicated after a first seizure.
(5) idiosyncratic drug reactions are unusual in epilepsy.
316 G. S. Bell & J. W. Sander

Answers

Question 1.

The incidence of epilepsy:

(1) false—that is the definition of point prevalence.
(2) true.
(3) false—the incidence of epilepsy is particularly high in Latin America.
(4) false—the incidence is high in the elderly.
(5) true.

Question 2.

The prognosis of epilepsy:

(1) false—this was thought to be the case until the 1960s, but was based on observations of patients seen in
tertiary referral centres.
(2) false—the prognosis of epilepsy depends on many factors, including the aetiology.
(3) true.
(4) true.
(5) false—the risk of seizure recurrence is highest in the first weeks or months after an initial seizure.

Question 3.

Mortality in epilepsy:
(1) false—the mortality rate is usually expressed as the SMR, the standardised mortality ratio.
(2) false—the SMR in epilepsy is between 2 and 3.
(3) false—although SUDEP is a significant cause of death in chronic epilepsy, it is still rare overall.
(4) false—the cause of death in people with epilepsy can be related to the epilepsy, related to the underlying
disease causing epilepsy, or unrelated to epilepsy.
(5) true.

Question 4.

Drug treatment in epilepsy:

(1) false—although AED usage can influence the likelihood of becoming seizure free, studies have shown that
a minority of people with untreated epilepsy will enter spontaneous remission.
(2) true—although unusual.
(3) false—AEDs can suppress seizures.
(4) false—in the UK patients with just one unprovoked seizure are not routinely treated.
(5) true—idiosyncratic drug reactions are very unusual, but can nonetheless occur.