Epilepsia, 34(3):453468, 1993

Raven Press, Ltd., New York

0 International League Against Epilepsy

Incidence of Epilepsy and Unprovoked Seizures in

Rochester, Minnesota: 1935-1984

W. Allen Hauser, *John F. Annegers, and ?-Leonard T. Kurland

G . H . Sergievsky Center, College of Physicians and Surgeons, Columbia University, New York, New York; *School
of Public Health, University of Texas, Houston, Texas; and tSection of Medical Research, Mayo Clinic,
Rochester, Minnesota, U.S.A.

Summary: The incidence of epilepsy and of all unpro- isting condition. The cumulative incidence of epilepsy
voked seizures was determined for residents of Roches- through age 74 years was 3.1%. The age-adjusted inci-
ter, Minnesota U.S.A. from 1935 through 1984. Age- dence of all unprovoked seizures was 61 per 100,000 per-
adjusted incidence of epilepsy was 44 per 100,000 person- son-years. Age- and gender-specific incidence trends
years. Incidence in males was significantly higher than in were similar to those of epilepsy, but a higher proportion
females and was high in the first year of life but highest in of cases was of unknown etiology and was characterized
persons aged 375 years. Sixty percent of new cases had by generalized onset seizures. The cumulative incidence
epilepsy manifested by partial seizures, and two thirds of all unprovoked seizures was 4.1% through age 74
had no clearly identified antecedent. Cerebrovascular dis- years. With time, the incidence of epilepsy and of unpro-
ease was the most commonly identified antecedent, ac- voked seizures decreased in children and increased in the
counting for 11% of cases. Neurologic deficits from birth, elderly. Key Words: Epidemiology-Seizures-Epilepsy-
mental retardation and/or cerebral palsy, observed in 8% IncidenceEtiology-Seizure type-Time trends.
of cases, was the next most frequently identified preex-

Although studies of the prevalence of epilepsy (Hauser et al., 1990). A seizure was considered to
abound, there are few studies of the incidence of be the clinical manifestation of an abnormal and
epilepsy. Prevalence is important for determination excessive discharge of a set of neurons of the brain,
of health care needs and may allow hypothesis gen- including cortical cells. An unprovoked seizure was
eration, but also may be misleading if used for eti- considered a seizure occurring without an identified
ologic studies or to provide information on progno- proximate precipitant, thus excluding seizures as-
sis. Information regarding incidence is necessary sociated only with an acute insult to the central ner-
for appropriate evaluation of etiologic factors, and vous system (CNS) or a generalized systemic met-
incidence cohorts are the most appropriate group in abolic disturbance (acute symptomatic seizures).
which to evaluate prognosis. We used the facilities Seizures and epilepsies characterized by unique
of the Rochester, Minnesota, U.S.A., epidemio- sensitivity to external stimuli such as photoconvul-
logic project to identify all medical contacts for sei- sive seizures, sensory stimulus-sensitive epilepsy,
zures in a 50-year period. We report information auditory-induced epilepsy, or reading epilepsy were
regarding the incidence of epilepsy and of unpro- considered unprovoked.
voked seizures by age, gender, seizure type, and Epilepsy is defined as a condition characterized
etiology in this community. by recurrent unprovoked seizures. An individual
with a single unprovoked seizure was not catego-
METHODS rized as having epilepsy in the present study, but
such cases were identified and categorized sepa-
A general description of methods, definitions of
rately. A cluster of seizures occurring in a single
seizures and of epilepsy, and definitions of etiologic
24-h interval was considered a single seizure epi-
categories are reported in a companion paper
sode.
Etiology
Received July 1992; revision accepted July 1992.
Address correspondence and reprint requests to Dr. W. Allen Cases were categorized by etiology into two
Hauser at 630 W. 168th St., New York, NY 10032, U.S.A. broad groups: remote symptomatic and idiopathic.

453
454 W . A . HAUSER ET AL.

Remote symptomatic seizures (epilepsy) cases, the distinction between simple and complex
The remote symptomatic seizures (epilepsy) partial seizures could not be made from information
group included individuals with unprovoked sei- provided by history alone. In this situation, cases
zure(s) and a history of a CNS insult before (and were classified as partial epilepsy of “uncertain”
temporally remote from) the first unprovoked sei- type. Individuals with partial seizures but with
zure. This category is limited to antecedent condi- more than one clinical manifestation of seizures
tions well demonstrated to be causes of epilepsy. suggesting either the presence of multiple foci or
Major categories of remote symptomatic epilepsy different paths of excitation of a single focus were
for this presentation are as follows: head injury, also classified as partial.
cerebrovascular disease, CNS infection, CNS neo- Patients with presumed generalized onset sei-
plasms, neurologic deficits presumed present at zures were categorized into one of three seizure
birth [mental retardation (MR, IQ <70) and/or ce- types: generalized major motor, absence, or myo-
rebral palsy (CP), motor handicap or movement dis- clonic. The category of major motor generalized
order], degenerative neurologic disorders, and met- onset seizures included individuals with predomi-
abolic neurologic disorders. Criteria for inclusion nantly tonic, clonic, or tonic-clonic seizures
were described by Hauser et al. (1991). because recorded descriptions seldom allow dis-
tinction of these types. For individuals who experi-
Idiopathic seizures (epilepsy)
enced predominantly absence seizures or predomi-
The idiopathic seizures (epilepsy) group included
nantly myoclonic seizures, the occurrence of gen-
individuals with unprovoked seizures occurring in
eralized major motor seizures did not modify the
the absence of a historical insult demonstrated to
classification, although this information was ab-
increase greatly the risk of unprovoked seizures.
stracted. Seizure classification was based upon pre-
For the current presentation, the presence of an
dominant seizure type at diagnosis.
unexplained localized abnormality on neurologic
Patients were classified as having multiple sei-
examination (i.e., a reflex asymmetry) or identified
zure types only if there was evidence of both gen-
in the diagnostic workup [i.e., focal slowing or focal
eralized onset and partial onset seizures. Patients
spike on EEG examination, lesions of uncertain eti-
with multiple types of generalized onset seizures
ology identified on computed tomography (CT) or
such as absence seizures and generalized tonic-
magnetic resonance imaging (MRI)] are not exclu-
clonic (GTC) seizures were considered to have only
sions from this category. This policy has been
generalized onset seizures and were categorized ac-
adopted to allow comparison with epidemiologic
cording to their predominant seizure type.
field studies in which detailed neurologic examina-
The classification was based on the diagnostic im-
tions and/or sophisticated neurophysiologic and
pression of the study neurologist (W.A.H.) after re-
neuroimaging testing is not available. Data on these
view of all available histories. Most of the residents
variables were collected and will be the subject of a
of Rochester included in the study were evaluated
subsequent publication. Most cases in the idio-
at some time by staff neurologists at the Mayo
pathic category would be categorized as cryptoge-
Clinic and descriptions of premonitory symptoms
nic in the most recent revision of the International
(“auras”, ictal manifestations, and postictal symp-
League against Epilepsy (ILAE) Classification of
tomatology) were generally quite detailed. Seizures
the Epilepsies and Epileptic Syndromes, which re-
were not necessarily categorized in accordance
serves the term idiopathic for cases of presumed
with the diagnostic impression of the treating phy-
genetic origin (Commission, 1989).
sician since diagnostic terminology has not been
Seizure type constant in the 50 years covered by this study.
Seizures were classified from the history accord- The interictal EEG was not used to modify the
ing to the criteria established by the ILAE (Com- classification of seizure type. Although EEGs were
mission, 1981). Seizures were classified as partial first recorded at the Mayo Clinic in 1938, the test
when there was evidence of partial onset regardless was not used routinely until the late 1940s. Even
of whether they became secondarily generalized. when EEGs have been obtained, the information
According to the 1981 ILAE convention, epileptic will vary depending on the duration of recording
persons who maintain alertness and ability to inter- and the activation procedures used during the pro-
act appropriately with their environment are classi- cedure. Furthermore, individuals with partial sei-
fied as having simple partial seizures. Individuals zures may demonstrate an interictal generalized
with confusion or, in the absence of seizures be- spike and wave, just as focal spikes may be ob-
coming secondarily generalized, amnesia for the ic- served in the interictal recording of individuals with
tus were classified as complex partial. For some generalized onset seizures. Such data have been ab-

Epilepsia, Vol. 34, No. 3, 1993

 INCIDENCE OF EPILEPSY 455

stracted but have not been used for stratification or using the 1970 total United States population as the
modification of assignment by etiology or seizure standard population (Rothman, 1986).
type in this report.
Patients were classified as having “unknown” RESULTS
seizure type when there was sufficient information Total age-adjusted incidence of epilepsy
in the medical records to document the occurrence In the 50-year period, 880 individuals (418 male,
of a seizure but inadequate information to allow fur- 462 female) came to medical attention and received
ther classification. a new diagnosis of epilepsy. The incidence of epi-
Case identification lepsy for the 50-year period age adjusted to the 1970
The diagnostic record system provided by the United States population was 441100,000 person-
Rochester Project was used to identify residents of years. Age-adjusted incidence was significantly
the city of Rochester, Minnesota between 1935 and higher for males than for females (49 and 41 per
1984 with a diagnosis of seizures, convulsions, or 100,000 person-years, respectively).
epilepsy (Kurland and Molgaard, 1981). A broad Incidence by ILAE seizure type
spectrum of diagnostic rubrics, including nonspe- Epilepsy manifested by partial seizures alone oc-
cific entries such as “syncope” and “loss of con- curred in 57% (503) of the incidence cases. The age-
sciousness,” were screened by a neurologist adjusted incidence per 100,000 person-years was 25
(W.A.H.) to identify potential cases. All residents partial seizures and 17 for epilepsy manifested by
considered to have experienced a seizure were cat- generalized onset seizures (Table 1). For both gen-
egorized by etiology and by seizure type. eralized and partial epilepsies, age-adjusted inci-
All individuals were further screened to deter- dence was higher for males than females. For par-
mine residency status (city resident or nonresident) tial epilepsies, the incidence was 28 and 24 per
at the time of first seizure on the date of first diag- 100,000 person-years for males and females, respec-
nosis of epilepsy. tively; for generalized epilepsies, age-adjusted inci-
Incidence calculation dence was 20 and 15 per 100,000 person-years for
Incidence was determined for epilepsy and for all males and females, respectively.
unprovoked seizures for the total study period and Among those with epilepsy characterized by gen-
for five 10-year intervals, with the midpoint being eralized onset seizures, epilepsy characterized by
the decennial census year. The denominator was absence seizures accounted for 13.3% of cases, and
the projected number of residents within age- and epilepsy characterized by myoclonic seizures ac-
gender-specific strata based on extrapolation of counted for 7.7% of cases. The age-adjusted inci-
data provided for each census year. Published de- dence of epilepsy Characterized primarily by absence
nominator figures were modified to exclude individ- seizures at diagnosis was 3 per 100,OOO person-years
uals who were considered residents by the United and was higher for females. The incidence of epilep-
States Census on a particular census day but who sies characterized by myoclonus was 1 per 100,OOO
failed to meet criteria for inclusion in epidemiologic TABLE 1. Incidence of epilepsy in Rochester,
studies in Rochester, Minnesota (Schroeder and Of- Minnesota, 1935-1984, age adjusted to the 1970
ford, 1982). Primary exclusions from official census U S . population
counts were individuals moving into the community Male Female Total
from other areas for medical reasons (i.e., long- Population (900,475) (1,002,882) (2,003,357)
term residents of the Rochester State Psychiatric All epilepsy 49 41 44
Hospital, a chronic care facility operational
Seizures
throughout most of the study period, and residents All partial 28 24 25
of skilled nursing care facilities). Simple I 5 6
Individuals were considered incidence cases of Complex 18 15 16
Unknown 3 3 3
epilepsy or of a single unprovoked seizure if they Secondarily generalized 17 13 15
were bonafide residents of the community at the All generalized 20 15 17
time of diagnosis. Although residency at onset was Absence 2 3 3
Myoclonic 1 1 1
determined for all cases, residency at diagnosis Generalized tonic-clonic 14 8 10
rather than residency at onset was used for inci- Other 3 3 3
dence calculations for epilepsy because a consider- Unknown 1 2 2
able interval may occur between initial seizure and Etiology
ultimate diagnosis. Idiopathic 31 28 29
Remote symptomatic 18 14 15
Incidence was age-adjusted by the direct method

Epilepsia, Vol. 34, N o . 3 , 1993

456 W . A . HAUSER ET AL.

person-years. The incidence of epilepsy characterized

by major motor seizures alone was 10 per 100,000
person-years and was higher for males (Fig. 1).
For the partial epilepsies, the majority of cases
were complex (64%) or of uncertain partial classifi-
cation (12%). Epilepsy characterized by simple par- Vascular 10.9%
tial seizures alone accounted for 24% of cases. The
age-adjusted incidence of complex partial epilepsy
was 16, that of simple partial epilepsy was 6, and
that of unclassified partial epilepsy was 5. Age-
Congenital 8.0%
adjusted incidence for each of the categories of par-
FIG. 2. Proportion of incidence cases (1935-1984) by etiol-
tial epilepsy was higher in males than in females. ogy of epilepsy (all ages).
Etiology of incidence cases
About two thirds (573) of all incidence cases were life, but was higher in males than females after this
classified as idiopathic. About 11% of all incidence age, with the greatest discrepancy by gender occur-
cases were attributed to cerebrovascular disease, ring in the oldest age groups.
and 8% were associated with neurologic deficits
Seizure type
from birth (MR and/or CP) with smaller proportions
Age-specific incidence showed different patterns
of cases attributable to other antecedents (Fig. 2).
for the two broad categories of epilepsy seizure
Proportions are similar whether based on age-
type. As a group, the generalized onset epilepsies
adjusted incidence or number of cases.
demonstrated the highest incidence (67) in the first
In the 50-year period, the age-adjusted incidence
year of life. Incidence decreased throughout child-
per 100,000 person-years was 29 for idiopathic/
hood and demonstrated a modest increase in the
cryptogenic epilepsy, and 15 for remote symptom-
oldest age groups. In contrast to this pattern, the
atic epilepsy. Age-adjusted gender-specific inci-
incidence of epilepsy characterized by partial sei-
dence was higher in males than in females for both
zures demonstrated a relatively constant rate of
idiopathic (31 vs. 27) and remote symptomatic (18
-20 throughout childhood and adult years (to age
vs. 14) categories.
65 years). In persons aged 3 6 5 , there was a dra-
Age-specific incidence matic increase in incidence of partial epilepsy, and
the incidence of partial epilepsy was 98 in those
Total group
aged >75, a fivefold increase over that observed at
The age-specific incidence was high in the first
earlier ages.
year of life (82) and decreased throughout child-
When age-specific incidences of generalized and
hood. Incidence was relatively stable (and low) in
partial epilepsies are compared (Fig. 4), the inci-
the adult years, and increased in each subsequent
dence of generalized epilepsies was higher than that
age group after age 55 years (Fig. 3). The greatest
of partial epilepsies in the first 5 years of life. The
incidence (139) occurred in persons aged a 7 5 years.
incidence of the two broad categories by seizure
The pattern of age-specific incidence was similar in
type was similar from age 6 through age 24 years,
males and females. The incidence was slightly
but after this age the incidence of partial epilepsy
higher in females than in males in the first 5 years of
was at least twice that of generalized onset epilepsy.

Complex partial
36.0%
Generalized TC

fi
23.0% Unclassified 3.0%
Myoclinic 3.0%

h 1 - 1 L- - L-
Simple partial ^.. _._Partial unknown 20 40 60 80
14.0% uiner gen 7.0% AGE
8.Oo/o
FIG. 3. Age- and gender-specific incidence/100,000 of epi-
FIG. 1. Proportion of all incidence cases by seizure type. Tc, lepsy in Rochester, Minnesota, 1935-1984. Total (solid cir-
tonic-clonic; gen, generalized. cles), male (plus signs), female (stars).

Epilepsia, Vol. 34, No. 3 , 1993

 INCIDENCE OF EPILEPSY 457

120 I
INCIDENCE/lOO,OOO
p ~
-.
~ -

r--
i
::i
I2O
100
100

40 -
60
8o I
J /
... I-- L- p - - p . L p p I
o
0 -
0 20 40 60 80
0 20 40 60 80
AGE
AGE
FIG. 6. Age- and gender-specific incidence/100,000 of par-
FIG. 4. Age-specific incidence of generalized onset (solid tial epilepsy in Rochester, Minnesota, 1935-1984. Male (plus
circles) and partial onset (plus signs) epilepsies in Rochester, signs), female (stars).
Minnesota, 1935-1984.

incidence was in early childhood. The age-specific

For the generalized epilepsies, age- and gender- incidence of epilepsy characterized by GTC alone
specific rates were higher in the first year of life in was relatively constant for all age groups through
females than in males and higher in males than in age 64 years, after which a modest increase in inci-
females after age 44 years (Fig. 5). dence was noted.
For epilepsy manifested by partial seizures, inci- The incidence of simple partial seizures was rel-
dence was similar for males and females for all age atively constant for all age groups. The incidence of
groups through the sixth decade. After age 65 years, complex partial and unclassified partial epilepsy
the incidence of partial seizures was higher in males was relatively stable throughout the childhood and
(Fig. 6). adult years but increased dramatically in the elderly
Even with more than 2,000,000 person-years of (Fig. 8). Seizures becoming secondarily generalized
observation in the 50-year period studied, the num- occurred in 60% of incidence cases with partial ep-
ber of cases in some seizure type categories was ilepsy. The proportion of incidence cases with sec-
small. Nonetheless, age-specific patterns of sub- ondarily generalized seizures was slightly lower
groups of generalized onset epilepsy are consistent among those with simple partial epilepsy (54%) than
with usual clinical impressions (Fig. 7). New cases those with complex and uncertain partial epilepsy
of epilepsy characterized by absence seizures were (61%), although these differences were not statisti-
diagnosed primarily during childhood, although cally significant.
some cases were first diagnosed in the third decade
Etiology
of life. Epilepsies characterized by myoclonic ac-
For each of the broad etiologic classifications (id-
tivity occur throughout life, although the highest
iopathickryptogenic and remote symptomatic), the
80

60 I\
60

40

20

0 u
& _- _t3
_ --.
0 20 40 60 80 0 20 40 60 80
AGE AGE
FIG. 5. Age- and gender-specific incidence of generalized FIG. 7. Age-specific incidence/100,000 of generalized onset
onset epilepsy by seizure type in Rochester, Minnesota, epilepsies by subcategory of seizure type in Rochester, Min-
1935-1 984. Total (solid circles), male (plus signs), female nesota, 1935-1984. Total (solid circles), absence (squares),
(stars). myoclonic (plus signs), generalized tonic-clonic (stars).

Epilepsia, Vol. 34, No. 3 , 1993

458 W . A . HAUSER ET AL.

120 7
symptomatic epilepsy was higher in males than in
females in the elderly.
loo i More than 50% of cases were assigned to the id-
iopathic category for all but the oldest age group
(Fig. 10). Only 45% of cases in the oldest age group
were of unknown etiology. Among children with a
presumed etiology, the greatest proportion of cases
40
was associated with neurologic deficits believed to

.' be present at birth (MR andlor CP) (Fig. 11). Among

the young adult population (age 15-34 years) with
an identified etiology, CNS infections, CNS tu-
0 20 40 60 80 mors, neurologic deficits from birth, and brain
AGE trauma were identified as antecedents with equal
FIG. 8. Age-specific incidence/l00,000 of partial epilepsy by frequency. In the adult population (age 35-64 years)
subcategory in Rochester, Minnesota, 1935-1984. Complex with a presumed cause, trauma and neoplasms were
partial (solid circles), simple partial (plus signs), partial un-
classified (stars), total (crosses). identified as antecedents with equal frequency, but
were less frequent antecedents than cerebrovascu-
lar disease. Cerebrovascular insults preceded the
age-specific incidence reflected a U-shaped pattern diagnosis of epilepsy in 15% of all cases diagnosed
with high incidence in the first year of life and high- in this age group and were the presumed cause in
est rates in the elderly (Fig. 9). The decrease in 35% of cases with an identified etiology. In the el-
incidence with advancing age in childhood was less derly (aged >65 years) the most frequently identi-
precipitous for idiopathic epilepsy than for remote fied antecedent also was cerebrovascular disease,
symptomatic epilepsy. Age-specific incidence for which accounted for 28% of all new cases in this age
idiopathic epilepsy was substantially greater than group and for two of three of all cases with a pre-
that for remote symptomatic epilepsy into the sumed etiology. About 20% of cases with presumed
fourth decade. After age 35 years, the incidence of etiology in the oldest age group were associated
epilepsy classified as idiopathic was almost identi- with degenerative CNS diseases.
cal to that of epilepsy classified as remote symp-
tomatic. Etiology/seizure-specificincidence
Age- and gender-specific rates were higher in Seventy percent of all generalized onset cases
males for both idiopathic and remote symptomatic and 64% of all partial cases were classified as idio-
epilepsy. For both males and females, age-specific pathic. These proportions were similar for both
incidence patterns showed the familiar U-shaped males and for females. Except for the oldest age
pattern for both idiopathic and remote symptomatic group, idiopathic cases accounted for most cases
epilepsy. The incidence of idiopathic epilepsy was for both partial and generalized onset seizures. In
higher in females than in males in the first year of
life, and the incidence of both idiopathic and remote 100
0 Idiopathic
LV Cangenital
Trauma

-3t
8D W
I Infection
0 Cerebrovascular
LV Tumor

-8 W Degenerative

's::
c
40

t
20

"
0-14 15-34 35-64 65 +
Age
FIG. 10. Proportion of cases of newly diagnosed epilepsy
assigned to specific etiologic categories within age groups,
including idiopathic/cryptogenic category. Area: idiopathic
AGE (gray cross-hatched), congenital (dashed), trauma (dotted),
FIG. 9. Age-specific incidence/l00,000 by broad etiologic trauma (widely dotted), infection (hatched), cerebrovascular
categories. Remote symptomatic (solid circles), idiopathic/ (closely dotted), tumor (black), degenerative (light cross-
cryptogenic (plus signs). hatch).

Epilepsia, Vol. 34, N o . 3 , 1993

 INCIDENCE OF EPILEPSY , 459

1M 1 Congenital
1 Trauma
1 Infection
80 1Cerebrovascular 50

2 I Tumor
I Degenerative
2s 40 ~

-
0
60

zs
30
40
n
t
20

0
0-14 15-34

Age
35-64 65 i o1
0
~ ~ 1
20
lp---
40

AGE
- i
60
-- 80

FIG. 11. Proportion of cases of newly diagnosed epilepsy

assigned to specific etiologic categories within age groups FIG. 13. Age-specific incidence/100,000 by broad etiologic
among those with assigned etiologies. Area: congenital categories for partial seizures in Rochester, Minnesota,
(dashed), trauma (widely dotted), infection (hatched), cere- 1935-1984. Partial idiopathic/cryptogenic (solid circles), par-
brovascular (closely dotted), tumor (black), degenerative tial remote symptomatic (stars).
(crosshatched).

childhood, for both those with partial and with gen- mulative incidence was 1.8% for partial epilepsy
eralized onset epilepsy, idiopathic cases accounted and 1.3% for generalized onset epilepsy (Fig. 16).
for a higher proportion of cases than at other ages.
Time trends
In the oldest age groups, for both partial and for
Incidence was determined for each of five 10-year
generalized onset seizures, the preponderance of
intervals, each with a midpoint at the decennial cen-
cases was associated with a remote symptomatic sus year. Although there was some fluctuation in
etiology (Figs. 12 and 13).
incidence, there was no significant difference in the
Cumulative incidence age-adjusted incidence over time. This apparent
The cumulative incidence of all epilepsy was stability is misleading and camouflages important
1.2% through age 24, 3% through age 74, and 4.4% temporal patterns of age-specific incidence. There
through age 85 years. The cumulative incidence was was a significant trend for decreasing incidence in
significantly higher in males (3.4% through age 74 children with time (Fig. 17). The incidence in those
years) than in females (2.8% through age 74 years) aged <10 years decreased 40% between early and
(Fig. 14). The cumulative incidence of idiopathic latest time interval (Table 2). This decrease in inci-
epilepsy was -2% through age 74 years and was dence with time in children was compensated for by
1.1% for symptomatic epilepsy. For both etiologic an increasing incidence with time in the elderly (al-
classes, gender-specific cumulative incidence was most doubling in the 50-year study period). A more
similar through the adult years, but diverged in the detailed assessment of these time trends by gender,
sixth decade and afterward, with the risk being seizure type, and etiology will be the subject of a
higher for males (Fig. 15). By age 75 years, the cu- subsequent publication.

f 5 :
4
/

0 20 40 60 8C

AGE
FIG. 12. Age-specific incidence/l00,000 by broad etiologic AGE
categories for generalized onset seizures in Rochester, Min- FIG. 14. Cumulative incidence of epilepsy by gender (%) in
nesota, 1935-1984. Generalized idiopathic/cryptogenic (plus Rochester, Minnesota, 1935-1984. Males (plus signs), fe-
signs), generalized remote symptomatic (solid circles). males (stars).

Epilepsia, Vol. 34, No. 3 , 1993

460 W . A . HAUSER ET AL.

0 20 40 60 80
AGE AGE
FIG. 15. Cumulative incidence of epilepsy by broad etio- FIG. 17. Time trends in age-specific incidence/100,000 of
logic categories (Yo)
in Rochester, Minnesota, 1935-1984. Id- epilepsy in Rochester, Minnesota, 1935-1 984: 1935-1 944
iopathic/cryptogenic (solid circles), remote symptomatic (solid circles), 19451954 (plus signs), 1955-1964 (stars),
(plus signs). 1965-1974 (squares), 1975-1984 (open circles).

Incidence of first medical diagnosis of an medical attention for one or more unprovoked sei-
unprovoked seizure zures while they were residents of Rochester, Min-
Most individuals who receive a medical diagnosis nesota. The age-adjusted incidence of FUS during
of epilepsy will have had multiple unprovoked sei- the 50-year study period was 61/100,000 person-
zures before their first medical contact for epilepsy. years, -33% higher than the incidence of epilepsy
Individuals included as a case for determination of in this community (Table 3). The incidence was sig-
incidence of first medical diagnosis of unprovoked nificantly higher in males than in females (68 vs. 56,
seizure (FUS)differ from those with a first diagno- respectively) (Table 3).
sis of epilepsy by inclusion of individuals who have
Seizure type
had only one unprovoked seizure at the time of
Among those with a first unprovoked seizure, the
medical contact. Age-specific incidence may also
number of cases with one or more generalized sei-
differ because age at first seizure rather than age at
zures was equal to that of one or more partial sei-
diagnosis of epilepsy will be used for those medi-
zures. The age-adjusted incidence per 100,000 per-
cally evaluated for their first seizure. A population-
sons-years of both partial FUS and generalized on-
based study may have other differences: e.g., an
set FUS was 30. The incidence of unclassified FUS
individual excluded as an incident case of epilepsy
was 2 per 100,000 person-years.
as a nonresident may be included for incidence of
FUS if that individual is a resident with a medical Etiology
diagnosis at the time of the FUS.There were 1,208 Seventy percent of cases with a first unprovoked
individuals (584 male, 624 female) who came to seizure were classified as idiopathic. The age-
adjusted incidence of idiopathic FUS was 41 as
compared with 20 for remote symptomatic FUS.
I ?t

2.5 I Age-specific incidence

Trends in age-specific incidence of FUS were
similar to that of epilepsy: highest in the extremes
of life. Age- and gender-specific analysis showed a
higher incidence in males after age 45 years (Fig. 18).
Seizure type
The age-specific incidence of generalized onset
FUS was high in the first year of life, decreased in
the first decade, remained relatively stable through
Y C the fifth decade, and increased after age 55 years. In
0 20 40 60 80
contrast, the age-specific incidence of partial FUS
AGE was -20/100,000 person-years through the sixth de-
FIG. 16. Cumulative incidence by seizure type 1%) in Roch- cade and increased only after that age. The inci-
ester, Minnesota, 1935-1984. Partial epilepsy (plus signs), all
generalized onset epilepsy (stars), absence (solid circles), dence of generalized onset FUS was higher than
myoclonic (squares). that of partial FUS through the first decade. In the

Epilepsia, Vol. 34, NO.3 , 1993

 INCIDENCE OF EPILEPSY 461

adult years, the incidence of the two broad seizure son with the incidence of epilepsy in etiologic stud-
types were similar. Only in the oldest age groups ies. Incidence cohorts can also provide the most
did the incidence of partial FUS exceed that of gen- comprehensive information of prognosis and co-
eralized FUS (Fig. 19). morbidity because they include not only persons
Comparison of age-specific incidence of a gener- who will die shortly after diagnosis but also those
alized onset FUS with incidence of generalized on- with mild disease who will achieve remission (An-
set epilepsy showed a greater incidence of the negers et al., 1979; Hauser et al., 1980). Both these
former in all age groups (Fig. 20). This was also true categories of patients will be missed in many prev-
when generalized onset epilepsies and FUS were alence studies and among clinical series generated
analyzed by age and gender. from tertiary-care centers.
The incidence of partial FUS was virtually iden- We used the medical records linkage system of
tical to that of partial epilepsy for all age groups the Rochester project to estimate the incidence of
through the sixth decade. In the oldest age groups epilepsy and of one or more unprovoked seizures
(over aged >65 years), the incidence of partial FUS from a midwestern community of the United States
increased more dramatically than did that of partial in a 50-year period from 1935 to 1984. The popula-
epilepsy, and only in this oldest age group was the tion is predominantly white and middle class.
incidence greater than that of partial epilepsy (Fig. Age-adjusted incidence of epilepsy was 44 per
21). This age-specific pattern was also noted when 100,000 person-years and was significantly higher in
gender-specific incidences of unprovoked partial males than in females. Age-adjusted incidence of
seizures and epilepsy were compared. epilepsy manifested by partial seizures (25) was
Etiology
higher than that for epilepsy manifested by gener-
Age-specific incidence of idiopathic FUS ex- alized onset seizures (19). Two thirds of the inci-
ceeded that of remote symptomatic FUS through dence cases were of unknown etiology, and the
age 35 years. Incidences of idiopathic and of remote most frequently identified antecedents were neuro-
symptomatic FUS were similar through age 65 years. logic deficits from birth (important in the younger
After this age, the incidence of remote symptomatic age groups) and cerebrovascular disease (important
FUS exceeded that of idiopathic FUS (Fig. 22). in the elderly).
Age-specific incidence of idiopathic unprovoked Age-specific incidence was high in the first year
seizures was higher than that of idiopathic epilepsy of life, decreased during childhood, and was rela-
in all age groups (Fig. 23). The incidence of remote tively stable in the adult years. Incidence increased
symptomatic FUS diverged from that of epilepsy after age 54 years and was highest in the oldest age
only in the oldest age groups (Fig. 24). These ob- group. Age-specific incidence of generalized onset
servations held true when age/gender/etiology- epilepsy was higher than that of partial epilepsy in
specific incidence was determined. the first 5 years of life. The incidence of partial and
generalized onset seizures was similar through the
Cumulative incidence of all unprovoked seizures adult years, and in the elderly the incidence of ep-
The cumulative incidence of FUS through age 74 ilepsy manifested by partial seizures was consider-
years was 4.1%. Cumulative incidence in males and ably higher than that of epilepsy manifested by gen-
females was similar through age 54 years, after eralized seizures. The difference in incidence by
which the cumulative incidence increased more rap- gender is primarily accounted for by a higher inci-
idly in males than in females (Fig. 25). Cumulative dence in older adult males.
incidence of FUS through age 75 was significantly Age-adjusted incidence of a first diagnosis of an
higher in males (4.7%) than in females (3.7%). unprovoked seizure (including individuals with only
Time trends a single unprovoked seizure at first medical con-
As with epilepsy, age-adjusted incidence of all tact), was 61 per 100,000 person-years, a 33% in-
FUS appeared to show no significant variation with crease over the incidence of epilepsy. The differ-
time. Incidence in children decreased with time and ence in incidence of FUS (61) and of epilepsy (44) is
increased with time in the elderly. The changes in accounted for primarily by the addition of those
age-specific incidence for age groups at the ex- who experienced only one unprovoked seizure. The
tremes of life were even more dramatic than those clinical characteristics of the cohort are modified in
observed for the incidence of epilepsy (Fig. 26). accordance with what might be expected from what
is known about seizure recurrence after an FUS:
DISCUSSION The additional cases are more likely to have sei-
Incidence studies are important for hypothesis zures that are generalized in onset and of unknown
generation and provide baseline data for compari- etiology (Hauser et al., 1990; Shinnar et al., 1990).

Epilepsia, Val. 34, No. 3 , 1993

462 W. A . HAUSER ET AL.

TABLE 2. Age-speci9c incidence of epilepsy in Rochester, Minnesota, for each 10-year time interval 1935-1984
Age group
1935-1944 1945-1 954 1955-1 964
Total Population n Incidence Population n Incidence Population n Incidence
Total population
0-1 3,715 3 81 6,640 9 136 10,522 11 105
14 15,623 15 96 23,126 22 95 37,834 21 56
5-9 17,182 11 64 19,981 11 55 38,391 21 55
10-14 17,109 9 53 16,583 7 42 31,131 9 29
15-19 20,354 8 39 21,458 9 42 29,617 16 54
20-24 27,526 7 25 26,346 14 53 30,424 11 36
25-29 26,104 5 19 25,021 8 32 28,295 10 35
30-34 23,253 5 22 22,708 7 31 27,571 9 33
35-39 18,340 4 22 19,181 11 57 23,556 8 34
4044 16,860 2 12 18,559 6 32 21,602 9 42
4549 5,118 1 7 17,161 6 35 20,435 6 29
50-54 11,658 3 26 15,886 5 31 18,502 9 49
55-59 9,930 3 30 14,197 7 49 16,616 4 24
60-64 7,291 6 82 11,605 2 17 15,034 11 73
65-69 6,067 5 82 9,340 2 21 12,629 8 63
70-74 4,105 2 43 6,454 7 108 9,717 6 62
75-79 2,533 3 118 4,271 3 70 6,689 7 105
80-84 1,531 1 65 2,517 5 199 3,771 5 133
85 + 1,070 1 93 1,418 1 71 2,545 6 236
Total 245,969 94 38 282,452 142 50 384,881 187 49
Age-adjusted 44 47 45
Male
0-1 1,976 2 101 3,506 4 114 5,600 6 107
1-4 8,288 6 72 11,607 10 86 19,309 11 57
5-9 8,583 3 35 9,874 9 91 19,619 11 56
10-14 8,351 8 96 8,591 2 23 15,737 2 13
15-19 8,539 0 0 7,734 3 39 10,429 9 86
20-24 9,197 3 33 8,242 4 49 9,325 3 32
25-29 10,457 2 19 11,092 4 36 13,216 4 30
30-34 10,290 1 10 10,793 2 19 13,872 7 50
35-39 8,407 2 24 8,274 8 91 11,272 2 18
40-44 7,795 1 13 8,339 2 24 10,311 7 68
4549 7,115 0 0 7,284 3 41 8,987 3 33
50-54 5,620 2 36 1,057 4 57 8,321 4 48
55-59 4,273 2 47 6,246 3 48 6,885 2 29
60-64 3,179 4 126 5,369 1 19 6,175 2 32
65-69 2,598 4 154 3,745 1 27 5,193 3 58
70-74 2,004 2 100 2,666 3 113 3,975 3 75
75-79 1,247 1 80 1,659 2 121 2,282 3 131
80-84 747 0 0 902 2 222 1,196 2 167
85 i- 328 1 305 469 1 213 813 2 246
Total 108,994 44 40 123,449 68 55 172,517 86 50
Age adjusted 47 56 50
Female
0-1 1,739 1 58 3,134 5 160 4,922 5 102
1 4 7,335 9 123 11,519 12 104 18,525 10 54
5-9 8,599 8 93 10,107 2 20 18,772 10 53
10-14 8,758 1 11 7,992 5 63 15,394 7 45
15-19 11,815 8 68 13,724 6 44 19,188 7 36
20-24 18,329 4 22 18,104 10 55 21,099 8 38
25-29 15,647 3 19 13,929 4 29 15,079 6 40
30-34 12,963 4 31 11,915 5 42 13,699 2 15
35-39 9,933 2 20 10,907 3 28 12,284 6 49
40-44 9,065 1 11 10,220 4 39 11,291 2 18
4549 8,003 1 12 9,877 3 30 11,448 3 26
50-54 6,038 1 17 8,829 1 11 10,181 5 49
55-59 5,657 1 18 7,951 4 50 9,731 2 21
60-64 4,112 2 49 6,236 1 16 8,859 9 102
65-69 3,469 1 29 5,595 1 18 7,436 5 67
70-74 2,701 0 0 3,788 4 106 5,742 3 52
75-79 1,286 2 156 2,612 1 38 4,407 4 91
80-84 784 1 128 1,615 3 186 2,575 3 117
85 + 742 0 0 949 0 0 1,732 4 23 1
Total 136,975 50 37 159,003 74 47 212,364 101 48
Age adjusted 41 47 47

Epilepsia, Vol. 34, No. 3, 1993

 INCIDENCE OF EPILEPSY 463

TABLE 2-(Continued)

1965-1974 1975-1 984 1935- 1984

Population n Incidence Population n Incidence Population n lncidence

10,723 5 47 10,117 8 79 41,717 36 86

39,791 17 43 35,554 19 53 151,928 94 62
55,206 22 40 42,704 22 52 173,464 87 50
49,053 16 33 43,249 20 46 157,125 61 39
43,299 17 39 48,423 21 43 163,151 71 44
45,174 18 40 59,848 23 38 189,318 73 39
43,891 13 30 59,163 19 32 182,474 5.5 30
38,268 5 13 49,156 11 22 160,956 37 23
28,566 7 25 37,394 12 32 127,037 42 33
26,556 5 19 31,303 8 26 114,880 30 26
24,431 5 20 26,559 6 23 103,704 24 23
22,622 4 18 24,642 8 32 93,310 29 31
20,668 6 29 22,860 7 31 84,271 27 32
18,446 1 5 19,980 9 45 72,356 29 40
15,251 8 52 17,110 10 58 60,397 33 55
13,048 11 84 15,211 17 112 49,135 43 88
10,234 14 137 12,467 13 104 36,194 40 111
6,793 5 74 8,772 21 239 23,384 37 158
4,883 3 61 7,814 21 269 17,730 32 180
516,903 I82 35 573,152 275 48 2,003,357 880 44
35 47 44
5,789 3 52 5,323 2 38 22,194 17 77
20,298 9 44 18,086 10 55 77,588 46 59
28,386 14 49 21,603 10 46 88,065 47 53
24,952 7 28 22,142 12 54 79,773 31 39
18,450 7 38 22,571 12 53 67,723 31 46
14,339 9 63 23,211 11 47 64,314 30 47
20,013 7 35 28,141 12 43 82,919 29 35
19,275 3 16 23,784 8 34 78,014 21 27
14,603 1 7 18,366 5 27 60,922 18 30
12,788 3 23 15,116 1 7 54,349 14 26
1 1,460 3 26 13,049 3 23 47,895 12 25
10,522 3 29 11,573 2 17 43,093 15 35
8,573 2 23 10,542 5 47 36,519 14 38
7,762 1 13 8,915 5 56 31,400 13 41
5,685 6 106 6,697 4 60 23,918 18 75
4,629 6 130 5,258 8 152 18,532 22 119
3,384 6 177 3,986 3 75 12,558 15 119
2,163 0 0 2,364 10 423 7,372 14 I90
1,303 2 153 1,761 5 284 4,674 11 235
234,374 92 39 261,141 128 49 900,475 418 46
41 51 48
4,934 2 41 4,794 6 125 19,523 19 97
19,493 8 41 17,468 9 52 74,340 48 65
26,820 8 30 21,101 12 57 85,399 40 47
24,101 9 37 21,107 8 38 77,352 30 39
24,849 10 40 25,852 9 35 95,428 40 42
30,835 9 29 36,637 12 33 125,004 43 34
23,878 6 25 31,022 7 23 99,555 26 26
18,993 2 11 25,372 3 12 82,942 16 19
13,963 6 43 19,028 7 37 66,115 24 36
13,768 2 15 16,187 7 43 60,531 16 26
12,971 2 15 13,510 3 22 55,809 12 22
12,100 1 8 13,069 6 46 50,217 14 28
12,095 4 33 12,318 2 16 47,752 13 27
10,684 0 0 11,065 4 36 40,956 16 39
9,566 2 21 10,413 6 58 36,479 15 41
8,419 5 59 9,953 9 90 30,603 21 69
6,850 8 117 8,481 10 118 23,636 25 106
4,630 5 108 6,408 11 172 16,012 23 144
3,580 1 28 6,053 16 264 13,056 21 161
282.529 90 32 312.01 1 147 47 1,102,882 462 42
31 45 42

Epilepsia, Vol. 34, No. 3, 1993

464 W . A . HAUSER ET AL.

TABLE 3. Incidence of FUS in Rochester, Minnesota,

1935-1984, age-adjusted to the 1970

Population
U S . population
Male
900,475
Female
1,002,882
Total
2,003,357
:i
80

All epilepsy
All partial
All generalized
Idiopathic
Remote symptomatic
68
34
34
4s
23
56
28
28
37
19
61
30
30
40
21
mm
20

0 ‘
0
I
20
I
40 60
I
80
I

Whether we evaluate the incidence of epilepsy or AGE

of an FUS, we note significant changes in the age- FIG. 19. Age-specific incidence/l00,000 of first unprovoked
seizure by broad seizure type in Rochester, Minnesota, 1935-
specific incidence with time. The dramatic increase 1984. Generalized first unprovoked seizure (stars), partial
in the elderly could be related in part to better case first unprovoked seizure (solid circles).
ascertainment in the later periods of the study.
ities. The increasing incidence in the elderly seems
Some sources of medical care available to the com-
paradoxic because the incidence of cerebrovascular
munity in the first 15 years covered by this study
were not included in the medical record index of the disease has decreased dramatically in this same pe-
Rochester project. Furthermore, patterns of medi- riod in this community, and we had hypothesized a
cal care use by the elderly may have changed with decreasing incidence of epilepsy because cerebro-
vascular disease is the presumed mechanism under-
time. Nonetheless, incidence in the oldest age
lying most epilepsy in elderly persons in whom a
group continued to demonstrate an increase across
cause is identified and “silent stroke” is hypothe-
the last three time intervals when indexing (and thus
sized to account for much of the idiopathic/
case identification) was more complete. By the
same token, we might expect an underenumeration cryptogenic cases in this age group (Shinton RA et
al., 1987; Roberts et al., 1988; Broderick et al., 1989).
of cases of epilepsy in children in the earliest time
interval. Therefore, potential underascertainment Comparisons with other studies
could only strengthen the apparent time-trend find- Because of the difficulty in conducting incidence
ings in children. studies, few studies report the incidence of epilepsy
We have no ready explanation for the changes in for all ages by seizure type and etiology. There is
either age group. For both the young and the elder- substantial variation in reported incidence, but di-
ly, changes in incidence appear to predate major rect comparisons across studies can be misleading
diagnostic or therapeutic breakthroughs. The inci- because of differences in methods of case ascertain-
dence in children is lowest in the 1965-1974 period, ment as well as differences in definitions. When the
a period of introduction of high-technologyneonatal definition of incidence includes single seizures (an
monitoring units, but the trend to decreasing inci- FUS) or some or all acute symptomatic seizures, a
dence started before implementation of these activ- higher incidence is expected (Loiseau et al.,
1990a,b; Sander et al., 1990). The extent to which
variation in reported incidence can be attributed to
differences in definitions and methods of case as-
200
120

150
loo R
100

50

0 20 40 60 80 0 ‘ I I I I
0 20 40 60 80
AGE
FIG. 18. Age- and gender-specific incidence/100,000 of all AGE
first unprovoked seizures in Rochester, Minnesota, 1935- FIG. 20. Age-specific incidence/lOO,OOOof generalized epi-
1984. All cases (solid circles), male (plus signs), female lepsy (solid circles) and generalized first unprovoked sei-
(stars). zures (stars) in Rochester, Minnesota, 1935-1984.

Epilepsia, Vol. 34, No. 3, 1993

 INCIDENCE OF EPILEPS Y 465

140, 1w 7- ~ .

120 1 R t;

I I I I 0 i- I
20
I
40
L-
60
’ 80
0 20 40 60 80
AGE
AGE FIG. 23. Age-specific incidence/100,000 of idiopathic first
FIG. 21. Age-specific incidence/l00,000 of partial epilepsy unprovoked seizures and epilepsy in Rochester, Minnesota,
(plus signs) and of a partial first unprovoked seizure (solid 1935-1984. ldiopathiclcryptogenic epilepsy (plus signs), id-
circles) in Rochester, Minnesota, 1935-1984. iopathic/cryptogenic unprovoked (stars).

certainment as opposed to true differences in inci- acute symptomatic seizures separately should allow
dence is not clear (Hauser and Hesdorffer, 1990). restructuring of current incidence to allow compar-
Seizures occurring in the context of acute insults isons with incidence from most other studies. Inci-
represent a measure of the frequency of the under- dence of acute symptomatic seizures will parallel
lying illness rather than epilepsy. Furthermore, the age/gender/time trend patterns for underlying
acute symptomatic seizures are generally self- illness. Geographic variations may be expected de-
limited, and appropriate treatment is directed to the pending on the incidence of the underlying cause.
underlying condition. In contrast, epilepsy as dis- This may be particularly true of febrile seizures,
cussed in the present report represents a chronic which have shown clear but as yet unexplained geo-
condition for which long-term therapy for seizure graphic variation. The following comparisons are
control is generally warranted. limited to studies using definitions similar to that
Incidence appears to be slightly more consistent used in the current study.
in studies limited to children, although variation still
Total population studies
is considerable. A multitude of definitions of epi-
In general, the incidence reported in the present
lepsy are provided across all studies, and drawing
study is higher than that reported for epilepsy in
conclusions through direct comparisons is tenuous.
studies using similar definitions that include all age
Most troublesome in the childhood studies is inclu-
groups. Most contemporary studies, including the
sion of febrile seizures with epilepsy. Although
present study, show the incidence to be higher in
comparisons of incidence in different geographic ar-
males than in females (Gundmundsson, 1966; Juul-
eas may allow hypothesis generation, consistency
Jensen and Foldspang, 1983; Granieri et al., 1983;
of definitions is necessary. We believe that the
Joenson, 1986; Li et al., 1986). The preponderance
strategy of tabulating epilepsy, single seizures, and
of cases with partial seizures among patients with
epilepsy noted in the present study is similar to that
120

100

0 20 40 60 80
AGE 0 20 40 60 80

FIG. 22. Age-specific incidence1100,OOO of a first unpro- AGE

voked seizure by broad etiologic categories (idiopathic/ FIG. 24. Age-specific incidence/100,000 of remote symp-
cryptogenic, stars; remote symptomatic, solid circles) in tomatic epilepsy (plus signs) and first unprovoked remote
Rochester, Minnesota, 1935-1984. symptomatic seizures (solid circles).

Epilepsia, Vol. 34, No. 3, 1993

466 W . A . HAUSER ET AL.

300 I- ~

spang, 1983). Incidence in this Danish study was

similar to that in the present study only through the
250
200
! first decade of life. After that age, the cumulative
incidence in the study of Juul-Jensen and Foldspang
(1983) was -50% of that reported in the present
150 study.
100 Incidence studies in selected age groups
Several studies report incidence of epilepsy as
50
defined in the present study within selected age
groups. In several studies limited to adult popula-
0 20 40 60 80 tions, incidence was also shown to increase with
AGE advancing age. Using the same definitions as those
FIG. 25. Time trends in age-specific incidence/100,000 of used in the current study, an age-adjusted incidence
first unprovoked seizures in Rochester, Minnesota, 1935- of 24 was reported in a Finnish population aged > 15
1984: 1935-1944 (solid circles), 1945-1954 (plus signs),
1955-1 964 (stars), 1966-1974 (crosses), 1975-1 984 years (Keranen et al., 1989). After adjustment to the
(squares). 1970 United States population, incidence for the
same age group in Rochester is 41 as compared with
reported in Denmark, the Faroes, and France (Juul- 23 for the Finnish population. In this Finnish study,
Jensen and Foldspang, 1983; Joenson, 1986; the incidence in males was almost twice that in fe-
Loiseau et al., 1990a,b). Only in the study of epi- males, and incidence increased with advancing age
lepsy in Copporo did generalized onset seizures in males but not in females.
predominate (Granieri et al., 1983). In a Swedish study in adults, incidence of unpro-
Age-specific incidence in the present study varies voked seizures age adjusted to the 1970 U.S. pop-
from that reported in most other total population ulation was 34/100,000 person-years (Forsgren,
studies (Gundmundsson, 1966; Granieri et al., 1983; 1990).
Juul-Jensen and Foldspang, 1983; Joenson, 1986; Li The incidence was higher in males than females,
et al., 1986). Most report the highest incidence in increased with advancing age, and demonstrated
early childhood and do not demonstrate an increase the greatest gender-specific difference in the oldest
in incidence in the elderly. It is the greater propor- age groups. Partial seizures accounted for 60% of
tion of cases identified in the elderly in the Roches- cases, and a cause was unknown in 51%. Age-
ter population that appears to account for the dif- adjusted incidence of unprovoked seizures for the
ference in overall incidence demonstrated in the same age groups in Rochester were 55, with 66% of
present study and in other recent studies using sim- cases being of unknown etiology and 56% of cases
ilar definitions. Age-specific incidence in children in having partial seizures.
Copparo is higher than that in the present study In a Danish study of epilepsy in the elderly, age-
(Granieri et al., 1983). adjusted incidence in those aged >60 years was 86
Cumulative incidence has been reported from one for epilepsy defined as recurrent seizures and 96 for
other total population study (Juul-Jensen and Fold- all first seizures (Luhdorf et al., 1986). Whether in-
dividuals with acute symptomatic seizure are in-
cluded is not clear, however. Age-adjusted inci-
dence in Rochester for these same age groups was
71 /+ 70 for epilepsy and 100 for all unprovoked seizures.

6 :
5 /4 The incidence in children in the present study is
-40% less than that reported in two studies of in-
cidence in children in Italy that used similar defini-
tions (Cavazzuti, 1980; Benna et al., 1984). The cu-
mulative incidence in the present report through age
14 years is similar to that reported in a study in
Japan, although the cumulative incidence for one or
0 20 40 60 80
more unprovoked seizure is less in Rochester than
that for one or more unprovoked seizure in cohort
AGE
studies from both Japan and in Oakland (Vanden
FIG. 26. Cumulative incidence of all unprovoked seizures
by gender (“h)in Rochester, Minnesota, 1935-1984. Male un- Berg and Yerushalmy, 1969; Tsuboi, 1988). Time
provoked (plus signs), female unprovoked (stars). trends have been evaluated in only one other study;

Epilepsia, Vol. 34, N o . 3, 1993

 INCIDENCE OF EPILEPSY 46 7

no significant changes were identified (Granieri et of epilepsy in adults in Eastern Finland. Epilepsia 1989;30:
413-21.
al., 1983); this study covered a much shorter time Kurland LT, Molgaard CA. The patient record in epidemiology.
interval, however. Sci A m 1981;245:54-63.
These incidence data provide important compar- Li S , Schoenberg BE, Wang C, et al. Epidemiology of epilepsy
in urban areas of the Peoples Republic of China. Epilepsia
ison data for our own cohort studies of potential 1985;26:39 1-4.
risk factors, a source of incidence cases for case Loiseau J, Loiseau P, DuchC B, Guyot M, Dartigues J, Aublet B.
control studies, and an incidence cohort for evalu- A survey of epileptic disorders in southwest France: Seizures
in elderly patients. Ann Neurol 1990a;27:232-7.
ation of long-term prognosis. The present report Loiseau J, Loiseau P, Guyot M, DuchC B, Dartigues J, Aublet B.
also provides a description of the patterns of occur- Survey of seizure disorders in the French Southwest. I. In-
rence of epilepsy by gender, age, seizure type, and cidence of epileptic syndromes. Epilepsia 19906;31:391-6.
Luhdorf K, Jensen LK, Plesner AM. Epilepsy in the elderly:
etiology with time, which may be useful for hypoth- incidence, social function, and disability. Epilepsia 1986;27:
esis generation for other investigators. 135-41.
Roberts RC, Shorvon SD, Cox TCS, Gilliatt RW. Clinically un-
suspected cerebral infarction revealed by computed tomog-
Acknowledgment: This w o r k was supported in part b y raphy scanning in late onset epilepsy. Epilepsia 1988;29:
Grant No. NS 16308. Patricia Perkins and Sally Book 190-4.
provided technical assistance throughout this project. Rothman KJ. Modern epidemiology. Boston: Little Brown,
Dale Hesdorffer provided programming assistance f o r all 1986.
data analysis as well as critical comments. Sander JWA, Hart YM, Johnson AL, Shorvon SD. National
general practice study of epilepsy: newly diagnosed epileptic
REFERENCES seizures in a general population. Lancet 1990;336: 1267-71.
Schroeder DJ,Offord KP. A SAS macro which utilizes local and
Annegers JF, Hauser WA, Elveback LR. Remission of seizures reference population counts appropriate f o r incidence, prev-
and relapse in patients with epilepsy. Epilepsia 1979;20:729- alence, and mortality rate calculations in Rochester and
37. Olmsted County Minnesota. Technical Report Series 20.
Benna P, Ferrero P, Bianco C, et al. Epidemiologic aspects of Mayo Clinic, Section of Medical Research Statistics, Roch-
epilepsy in the children of a Piedmontese district (Alba-Bra). ester Minnesota, August 1982;5-7.
Pan Med 1984;26:113-8. Shinnar S, Berg D, MoshC SL, et al. The risk of seizure recur-
Broderick JP, Phillips SJ, Whisnant JP, O’Fallon WM, Berg- rence following a first unprovoked seizure in childhood: a
stralh EJ. Incidence rates of stroke in the eighties: the end of prospective study. Pediatrics 1990;85:1076-85.
the decline in stroke? Stroke 1989;20:577-82. Shinton RA, Gill JS, Zezulka AV, Bevers DG. The frequency of
Cavazutti GB. Epidemiology of different types of epilepsy in epilepsy preceding stroke: case control study in 230 patients.
school-age children of Modena, Italy. Epilepsia 1980;21:57- Lancet 1987;l:ll-2.
62. Tsuboi T.Prevalence and incidence of epilepsy in Tokyo. Epi-
Commission on Classification and Terminology of the Interna- lepsia 1988;29:103-10.
tional League Against Epilepsy. Proposal for revised clinical Van den Berg BJ, Yerushalamy J. Studies on convulsive disor-
and electroencephalographic classification of epileptic sei- ders in young children. I. Incidence of febrile and nonfebrile
zures. Epilepsia 1981;22:489-501. convulsions by age and other factors. Pediatr Res 1969;3:
Commission on Classification and Terminology of the Interna- 298-304.
tional League Against Epilepsy. A revised proposal for the
classification of epilepsy and epileptic syndromes. Epilepsia IUhUME
1989;30:268-278.
Forsgren L. Prospective incidence study and clinical character- L’incidence de I’Cpilepsie et de toutes les crises non provo-
ization of seizures in newly referred adults. Epilepsia 1990; quCes a CtC CvaluCe parmi les residents de Rochester, Minnesota,
3 1:292-301. USA, entre 1935 et 1984 inclus. L’incidence de l’kpilepsie
Granieri E, Rosati G, Tola R, et al. A descriptive study of epi- ajustCe a I’bge etait de 44 pour 100,000 personnes par annCe.
lepsy in the district of Copparo, Italy, 19641978. Epilepsia L’incidence chez les hommes Ctait significativement plus forte
1983;24:502-14. que chez les femmes, elle Ctait ClevCe dans la premi2re annCe de
Gudmundsson G. Epilepsy in Iceland. A clinical and epidemio- vie mais le maximum Ctait constat6 chez les sujets de 75 ans et
logic investigation. Acta Neurol Scand 1966;43(suppl 25): 1- plus. Soixante pour cent des nouveaux cas avaient une Cplepsie
124. qui se manifestait par des crises partielles, les deux-tiers ne prC-
Hauser WA, Annegers JF, Elveback LR. Mortality in patients sentaient pas d’antCcCdents Cvidents. La maladie cCrCbrovascu-
with epilepsy. Epilepsia 1980;21:3994 12. laire Ctait I’ant6cCdent le plus souvent identifie, elle expliquait
Hauser WA, Annegers JF, Kurland LT. The incidence of epi- 11% des cas. Des dCficits neurologiques lies a une pathologie
lepsy in Rochester, Minnesota 1935-79. Epilepsia 1984;25: pCrinatale (retard mental, IMC) ont CtC constates dans 8% des
666. cas, et constituaient le second facteur le plus souvent identifiC.
Hauser WA, Rich SS, Annegers JF, Anderson VE. Seizure re- L’incidence cumulative de I’Cpilepsie jusqu’a I’bge de 74 ans
currence after a first unprovoked seizure: an extended fol- Ctait de 3.1%. L’incidence ajustCe B I’Age de toutes les crises non
low-up. Neurology 1990;40:1163-70. provoquCes Ctait de 61 pour 100,000 personnes par annCe. L’in-
Hauser WA, Hesdorffer DC. Epilepsy: frequency, causes, and cidence Age spkcifique et sexe spCcifique avait les memes ten-
consequences. New York: Demos Press, 1990:21-28. dances que celles pour I’Cpilepsie, mais une plus forte proportion
Hauser WA, Annegers JF, Kurland LT. Prevalence of epilepsy de cas Ctait sans Ctiologie connue, et les crises Ctaient le plus
in Rochester, Minnesota, 1940-1980. Epilepsia 1991;31:429- souvent gCn6ralisCes d’emblCe. L’incidence cumulative de
45. toutes les crises non provoquCes Ctait de 4,1% jusqu’a 74 ans. Au
Joensen P. Prevalence, incidence and classification of epilepsy in fil des annCes, I’incidence de I’Cpilepsie et des crises non provo-
the Faroes. Actu Neurol Scand 1986;76:150-5. quCes a diminuC chez les enfants et a augment6 chez les sujets
Juul-Jensen P, Foldspang A. Natural history of epileptic sei- bgCs.
zures. Epilepsia 1983;24:297-3 12.
Karanen T, Rekkiner PJ, Sillampal V. Incidence and prevalence (P. Genton, Marseille)

Epilepsia, Vol. 34, No. 3 , 1993

468 W . A . HAUSER ET AL.

RESUMEN ZUSAMMENFASSUNG
Se ha determinado la incidencia de la epilepsia para 10s resi-
Die Inzidenz von Epilepsie und samtlicher nicht provozierter
dentes de Rochester, Minnesota, U.S.A., de todos 10s ataques
no provocados, desde 1935 a 1984. La incidencia ajustada de Anfalle wurde fur Bewohner von Rochester, Minnesota, USA
zwischen 1935 und 1984 bestimmt. Die Alterskorrelierte lnzidi-
epilepsfa fue de 44 por 100,000 personaslano. La incidencia en
varones fue significativamente m6s elevada que en mujeres y fue enz von Epilepsie lag bei 44 pro 100,000 Personen pro Jahr. Bei
elevada en el primer aiio de la vida per0 el nivel m6s alto se Mannern war sie signifikant hoher als bei Frauen. Sie war in den
ersten Lebens-jahren hoch aber am hochsten im Alter von 75
alcanz6 por encima de 10s 75 anos. El 70% de 10s nuevos casos
la epilepsta se manifestaba por ataques parciales y en dos tercios Jahren und dariiber. 60% der neuen Falle hatte Partidanfalle als
no se identific6 un antecedente claro. La enfermedad cerebro- erste Epilepsiemanifestation, 2/3 zeigten keine sicher zu identi-
vascular fue de antecedente m6s comunmente identificado y con- fizierende Ursachen. Cerebrovaskulare Erkrankungen stellten
stituy6 el 11% de 10s casos. Los deficits neurol6gicos como con- die haufigsten erkennbaren Ursachen in 11% der Falle dar. Ge-
secuencia del parto, retraso mental ylor parhlisis cerebral se en- burtstraumata wie mentale Retardierung oder Cerebralparesen
contraron en 8% de 10s casos y fue la segunba condici6n fanden sich in 8% der Falle-sie stellten die zweithaufigste Ursa-
preexistente identificada. La incidencia acumulada de epilepsia che dar. Die kumulative Inzidenz der Epilepsie lag bis zu einem
hasta 74 anos fue de 3.1%. La incidencia ajustada a la edad de Alter von 74 Jahren bei 3.1%. Die altersberechnete Inzidenz aller
todos 10s ataques no provocados fue de 61 por 100,000 personas1 nicht provozierter Anfalle lag bei pro 100,000 Personen pro Jahr.
afio. Las tendencias de la incidencia espectficas para edad y sex0 Alters- und geschlechtsspezifische Inzidenz-Trends waren denen
fueron semejantes a las de la epilepsia pero una proporci6n mhs der Epilepsie ahnlich. Ein hoherer Anteil von Fallen bot eine
elevada de 10s casos fue de etiologia desconocida y se caracter- ungeklarte Atiologie und war durch generalisiert beginnende An-
iz6 por ataques generalizados. La incidencia acumulada para falle gekennzeichnet. Die kumulative Inzidenz aller nicht provo-
todos 10s ataques no provocados fue de 4.1% hasta la edad de 74 zierten Anfalle lag bei 4.1% bis zum Alter von 74 Jahren. Im
anos. A lo largo del tiempo la incidencia de epilepsfa y de 10s Verlauf der Zeit nahm die Inzidenz fur Epilepsie und nicht pro-
ataques no provocados se redujo en nirios y se increment6 en 10s vozierte Anfalle wahrend der Kindheit ab um im Alter wieder
ancianos. anzusteigen.
(A. Portera-Sknchez, Madrid) (C. G. Lipinski, HeidelberglNeckargemund)

Epilepsia, Vol. 34, NO. 3 , 1993