Epilepsia, 46(6):918–923, 2005
Blackwell Publishing, Inc.
C ?2005 International League Against Epilepsy
Epilepsy in Young Adults with Autism: A Prospective
Population-based Follow-up Study of 120 Individuals Diagnosed
∗Susanna Danielsson, †I. Carina Gillberg, †Eva Billstedt, †‡Christopher Gillberg,
Departments of∗Pediatrics and †Child and Adolescent Psychiatry, Queen Silvia Children’s Hospital, G¨ oteborg, Sweden;
and ‡St. George’s Hospital Medical School, London, England
Summary: Purpose: Little is known about the long-term out-
come of epilepsy in autism and the epilepsy characteristics of
conducted in an attempt to point out differences on a group ba-
sis between adults with autism with or without epilepsy, and
to describe the occurrence, the seizure characteristics, and the
outcome of epilepsy in autism.
Methods: One hundred eight of 120 individuals with autism
diagnosed in childhood and followed up prospectively for a pe-
riod of 13–22 years were reevaluated at ages 17–40 years. As
adults, the majority had mental retardation and autistic disorder
or autistic-like condition. Interviews were performed with the
caretakers of 42 of 43 individuals with a history of epilepsy, and
their medical records were reviewed.
a severely disabled group. On a group basis, both the cognitive
level and the adaptive behavior level were lower in the epilepsy
group than in the nonepilepsy group (p < 0.05). In all, 38%
had epilepsy. One third had epilepsy onset before age 2 years.
Remission of epilepsy was seen in 16%. Partial seizures with or
without secondarily generalized seizures were the dominating
Conclusions: In a community sample of individuals with
autism followed up from childhood through to adult age,
one of three had epilepsy since childhood/adolescence. Se-
vere mental retardation and autism are significantly associated
with epilepsy, especially in female patients. Seizure frequency
has a great impact on the individuals’ lives. Specialist med-
ical care is needed in this severely communication-disabled
Key Words: Epilepsy—Seizures—Autism—
in reciprocal social interaction and communication abili-
ties and with a restricted pattern of behavior since early
childhood. Epilepsy is more common in people with AD
than in the general population (2,3), and AD and autistic-
like conditions (ALCs) are more common in people with
tematic investigations of autism and epilepsy diagnosed
on the basis of specific international diagnostic criteria,
(ILAE), have been undertaken (5,6).
The purpose of the present study was to describe the
occurrence and outcome of epilepsy and the epilepsy
Accepted February 12, 2005.
Address corresponding and reprint requests to Dr. S. Daniels-
son at Queen Silvia Children’s Hospital, Child Neuropsychiatry
(BNK), Otterh¨ allegatan 12A, S-411 18 G¨ oteborg, Sweden. E-mail:
characteristics of young adults with autism followed up
from childhood. We have the unique opportunity to point
out differences on a group basis between individuals with
for services and specialist care.
This study relates to individuals considered typical (or
the results cannot and should not be generalized to apply
to all individuals with autistic-spectrum disorders.
Epileptic seizures, epilepsy, active epilepsy, and remis-
sion of epilepsy defined according to the ILAE (7,8).
Autistic disorder (AD): diagnosed in accordance with
the DSM-IV criteria (1).
EPILEPSY IN YOUNG ADULTS WITH AUTISM919
or more, but not full DSM-IV criteria for AD; severity
of the autistic symptoms of the same degree as in the
AD group, even though the pattern of symptoms was not
sufficiently typical to assign the diagnosis.
Mental retardation (MR)/Learning disability: IQ <70,
resulting in functional impairment.
Mild mental retardation (MMR): IQ 50–69.
Severe mental retardation (SMR): IQ <50.
All individuals born in 1962 through 1984 in G¨ oteborg
and its neighboring rural area were screened for AD and
viously (9–11). Diagnostic instruments used at diagnosis
in childhood were in-depth clinical interviews, the Handi-
Autism Rating Scale (CARS) (13), and the Autistic Be-
havior Checklist (14). A total of 120 individuals with AD
or ALC were found in these three epidemiologic studies,
and they had all been examined by experts in the field of
with AD met the criteria of the DSM-III-R, and 42 had
ALC (six or more of the 18—but not full—DSM-III-R
criteria for AD).
These 120 individuals were clinically followed up in
not want to participate. The remaining 108 (77 male and
31 female subjects; mean age at study, 25.5 years; range,
17–40 years) had a psychiatric/medical assessment (I. C.
Gillberg) and a neuropsychological evaluation (E. Bill-
and/or the Vineland Adaptive Behavior Scales (15). The
Diagnostic Interview for Social and Communication dis-
orders (DISCO) (16) was also conducted. All individuals
but one were diagnosed as still having AD or ALC. The
relatives and/or nursing staff were asked about the life-
time occurrence of seizures, including absences. Further
details of this study have been published elsewhere (17).
Epilepsy was suspected in 48 of 108. In two cases, the
described manifestations were not epileptic seizures, but
cluded because only one epileptic seizure had occurred
during the lifetime and no treatment with antiepileptic
drugs (AEDs) had been given. The parents of one woman
with active epilepsy who participated in the psychiatric
follow-up study declined participation in the epilepsy
study. Thus 42 individuals with epilepsy were included
(25 men, 17 women; mean age, 26 years; range, 18–38
years). Parents or caretakers were interviewed according
to a systematically applied protocol (S. Danielsson, avail-
able from author on request). Special care was taken to
analyzed concerned: age at onset; the occurrence of sta-
tus epilepticus; seizure frequency; duration and course of
epilepsy; antiepileptic and psychotropic drugs, and medi-
cal services supplied. Medical records were reviewed, in-
had been performed in 33 of 42 [10 computerized tomog-
raphy (CT) and magnetic resonance imaging (MRI); 23
CT]. The result from at least one interictal EEG record-
ing was available for all cases but one. In each case, the
cussed with an experienced neuropediatrician (I. Olsson),
and seizure types were classified according to the ILAE
(8). Seizures were divided into four main categories: (a)
the partial seizure group included simple partial and com-
eralized seizures. If a history of obvious focal semiology
preceding a generalized seizure was found, it was classi-
fied as partial, or if the seizures with or without impaired
tonomic symptoms, or automatisms. (b) the generalized
seizure group included cases with myoclonic, clonic, to-
nic, atonic, or tonic–clonic seizures without partial onset,
partial and generalized seizure group. (d) In the unclassi-
fied group, insufficient description of seizure semiology
was available, mainly because of the difficulty in disting-
ly mentally retarded patients with severe autistic features.
Chi-square tests were used for contingency tables.
Fisher’s exact test was used in the event of small num-
bers. Mann–Whitney U test was used for comparison of
the medians of two unpaired groups.
The study was approved by the local Medical Ethics
Committee. The subjects participated after informed con-
sent was obtained from the closest relative.
At least 45 (38%) of 120 individuals with AD or ALC
had epilepsy at some point in their lives. We know from
the medical records that two of the six who had died had
participate had epilepsy. Figure 1 shows the age at onset
had SMR in childhood and 18 had MMR or average/near-
average cognitive ability. Among those with SMR, the
median age at epilepsy onset was 3.5 years, and in those
with MMR or average/near-average cognitive ability, the
median age at onset was 7.2 years, a nonsignificant (n.s.)
Epilepsia, Vol. 46, No. 6, 2005
920S. DANIELSSON ET AL.
FIG. 1. Cumulative percentage of age at onset of epilepsy in 45
out severe mental retardation (SMR) in childhood and 27 with
SMR in childhood.
The characteristics of all individuals in the psychiatric
follow-up study (n = 108) and individuals with epilepsy
(n = 43, including the woman not participating in the
in Table 1. Epilepsy was more common in women, 58%
(18 of 31), versus 32% (25 of 77) in men (p = 0.01).
The cognitive level was lower (p < 0.01) in the adults
with epilepsy than in those without. Among those with
SMR, 48% (37 of 77) had a history of epilepsy; in MMR,
20% (five of 25); and in cases without MR, 17% (one of
group compared with the nonepilepsy group (p < 0.05).
epilepsy group), and three did not (5% of the nonepilepsy
group; p < 0.01).
TABLE 1. Characteristics of all cases and cases with or without epilepsy
All cases (n = 108) Cases with epilepsy (n = 43) Cases without epilepsy (n = 65)
AD or ALC at follow-upb
>3 yr 11 mo
<3 yr 11 mo
AD, autistic disorder; ALC, autistic-like condition; SMR, severe mental retardation; MMR, mild mental retardation; MR, mental retardation.
ap ≤ 0.01; cases with epilepsy versus those without.
bAt follow-up, one individual did not fulfill the criteria for AD or ALC.
cp < 0.05; cases with epilepsy versus those without.
Among the 77 adults with SMR, there were 37 with
epilepsy and 40 without. Almost all had AD, 97% and
almost as many males as females (1.2:1), whereas almost
6 times more men than women were in the nonepilepsy
group (5.7:1; p < 0.01). The adaptive behavior level was
low in both groups: 84% in the epilepsy group had an
adaptive behavior <4-year age equivalent level and 82%
in the nonepilepsy group. The corresponding figures for
1-year age equivalent level were eight (22%) of 37, com-
pared with three (8%) of 40 (p = 0.11). In the epilepsy
dren, compared with 21 (52%) of 40 in the nonepilepsy
Of the subjects with epilepsy (n = 43, including the
woman not participating in the epilepsy study) associated
rofibromatosis type 1 (n = 2), tuberous sclerosis (n = 2),
fragile X syndrome (n = 2), Rett syndrome (n = 1), atyp-
ical Rett syndrome (n = 1), Moebius syndrome (n = 1),
sponding figures in those without epilepsy were 13 (20%)
of 65: neurofibromatosis type 1 (n = 2), fragile X syn-
Biedl syndrome (n = 1).
In the epilepsy study (n = 42), febrile convulsions had
preceded epilepsy in seven (17%) cases. Seizure type at
onset was partial in 23, generalized in 10, mixed par-
tial and generalized in six, and unclassified in three. One
boy had childhood absence epilepsy and was successfully
treated. Two girls and one boy (7%) had infantile spasms.
One of them became seizure free during the first year of
medically intractable epilepsy at ages 10 and 12.5 years.
Epilepsia, Vol. 46, No. 6, 2005
EPILEPSY IN YOUNG ADULTS WITH AUTISM 921
TABLE 2. Seizure types at follow-up or before remission (n = 42)
n% Seizure types EEG interictal expression
Partial only, with or without secondarily
Mixed partial and generalized
23 55Focal epileptiform discharges 16 of 23
Generalized epileptiform discharges 4 of 9
Both focal and generalized epileptiform discharges 3 of 5
Focal epileptiform discharges 4 of 5
Both focal and generalized epileptiform discharges 1 of 5
One with a cortical malformation continued to have in-
tractable epilepsy, and the other with tuberous sclerosis
became seizure free.
No change in seizure type was seen during the follow-
up time in 86% of cases. Table 2 shows seizure type at
the time of study or before remission and in how many
Hospitalization due to status epilepticus was reported
in 16 (38%) of 42 cases, and five of these had repeated
episodes. Severe trauma secondary to epileptic seizures
fractures or head trauma).
Epilepsy onset was before age 18 years in 38 of 42,
and remission of epilepsy was seen in six (16%) of these.
Their median age at onset of epilepsy was 4 years (range,
0.5–5.5 years), and the mean duration of epilepsy was 3.9
years (range, 0.5–10 years; median, 4.5 years). Three had
ondarily generalized seizures, and one was unclassified.
Two additional cases were seizure free, but with AED
monotherapy was used in 19 cases: 15 were taking car-
bamazepine (CBZ), three were taking lamotrigine (LTG),
and one was taking valproate (VPA). In 10 cases, two of
these AEDs were used in combination, or one was used
chotropic drugs (antipsychotics, antidepressants, and/or
lithium) were used in 16 of 34.
partial and generalized in five, and unclassified in four.
The majority, 25 (73%) of 34, had partial seizures with
or without secondarily generalized seizures or a mixture
of partial and generalized seizures. Two of the 34 cases
in the active epilepsy group had been subject to epilepsy
surgery: one with infantile spasms, mentioned earlier, and
one with a malformation of cortical development who un-
derwent temporal lobectomy and was seizure free with
Ten individuals (29%) of 34 had fewer than one seizure
per year. Among these, eight had had at least two seizures
more than one seizure per year in 24 (71%) of 34. Seven-
teen had more than one seizure every month, and, among
these, 11 (32%) of 34 had more than one seizure per week
epilepsy were seen by a neurologist once or twice a year.
In three cases, only two different AEDs in monotherapy
had ever been tried out, and one had no pharmacotherapy.
Median age at epilepsy onset was 3 years, in comparison
to 8 years in the nonintractable group (n.s.). No signifi-
cant differences were found in comparing the intractable
and the nonintractable groups concerning gender, cogni-
tive level, or seizure types.
said epilepsy had the greatest impact on the person’s daily
life as an adult; more frequently in those with intractable
p < 0.001). In the latter subgroup, autism and/or learning
disability was considered the most burdensome.
This study is based on the follow-up study by Billstedt
et al. (17) which is the longest and largest prospective
follow-up study ever published of a community-based
hood into adulthood. The sample of cases included is rep-
Only one population-based series of epilepsy in 16 chil-
dren with autism has previously been published (18), and
14 of the individuals from that study who survived into
adulthood are included in this study.
The reported rate of epilepsy in autism has varied from
7% to 42%, with more recent studies suggesting that in
about one third of children with autism, epilepsy devel-
ops (18–23). In our community-based cohort, the lifetime
occurrence of epilepsy was ≥38%.
had more severe cognitive impairment than did adults
with autism and no epilepsy. Epilepsy is considered to
be a negative prognostic factor for the outcome of autism
(24). Follow-up studies of autism show that a majority of
individuals affected have poor social functioning ability
in adulthood, and few live independently or are capable
of employment (17,24), and the autistic symptoms most
often persist throughout life (17,25). Among those with
SMR in our sample, almost half had epilepsy by early
Epilepsia, Vol. 46, No. 6, 2005
922S. DANIELSSON ET AL.
adult life, whereas epilepsy in MMR occurred in 20%. In
a population-based study (26), active epilepsy was found
in 20% of persons with MR, of all ages. Epilepsy was
more common in those with SMR, as described by others
Most studies on the epidemiology of epilepsy in indi-
autism in Forsgren’s study (26) if subjects had been as-
sessed with this diagnosis in mind. The diagnostic criteria
have changed, the concept of autistic spectrum disorders
has developed, and a growing awareness and knowledge
exist about autism among specialists and the general pub-
lic alike (28). For a long time, physicians in charge of
the treatment of epilepsy have reasoned that there were
already “appropriate diagnoses” (epilepsy and MR) (29),
neglecting the high rate of behavioral disturbances. Stef-
and MR had AD or ALC, with most cases not previously
ders are common in individuals with MR (30), and more
common in individuals with SMR than in MMR (31).
Our results imply that epilepsy is more common in fe-
male subjects with AD or ALC than in male patients,
which has been observed in other studies (20,32). It has
ity of associated MR in female patients with a diagnosis
Our data indicate that the risk for epilepsy is highest in
the first years of life and decreases from puberty and on-
2 years. However, in ≥10%, epilepsy will develop after
age 18 years. Because many individuals in this study only
just reached adult age, our figures on epilepsy occurrence
in adults with autism must be seen as a minimum. The
bimodal distribution of epilepsy onset with a peak in both
Rossi et al. (5) found that two thirds of 18 cases with AD
and MR had epilepsy onset after age 12 years, whereas
our corresponding number was slightly above one in four.
In contrast with Rossi, we have not excluded cases with
AD secondary to congenital or acquired encephalopathy,
which might explain the difference. In a long-term val-
idation study by Mouridsen et al. (21), the mean age at
onset of epilepsy was 8.1 years—similar to our finding of
before their 11th birthday.
Childhood-onset epilepsy often remits. Camfield (33),
in a population-based study, found that 55% of childhood
epilepsy cases remit. Predictors of remission have been
shown to be the occurrence of only one type of seizure,
normal mental and neurologic development, and the ab-
sence of a detectable cause (34). The remission rate of
epilepsy in children with autism and MR has not previ-
ously been described. In our study, it was only 16%.
It has previously been shown that a majority of cases
with epilepsy in autism have partial seizures. In the study
by Olsson et al. (18), three fourths of all children with
autism and epilepsy had partial seizures only or in com-
bination with other seizure types, as did three fourths of
toms of seizures and information from medical records
were used for the classification. We had to rely on the his-
tory from parents and caretakers, because we did not have
access to video-EEG or ictal EEG recordings, except in
those who had had presurgical assessments. It is a chal-
lenge trying to classify seizures in a group of adults with
autism and MR. Subclassification of seizures as simple
or complex partial could not be done. People with autism
and MR can only very rarely report on somatosensory,
special-sensory, or affective seizure symptoms. Another
problem is that a focal brain lesion may be the epilepto-
genic lesion triggering the seizure event, despite the lack
of focal seizure symptoms due to rapid generalization.
Interictal EEGs were used to support the clinical classifi-
cation. We preferred not to make further subclassification
because of the uncertainty just mentioned. Classification
of seizures was not possible in a few cases. This is not
surprising, considering that it is hard even for close rel-
atives or for professionals to assess what constitutes an
epileptic seizure and what is a marked lack of awareness
of the existence and intentions of others. A recent review
by Camfield and Camfield (35) addresses the inaccuracy
in the diagnosis of seizures and epilepsy syndromes in
It would have been interesting to have more data on
the etiologic background. In altogether 10 cases, an asso-
ciated medical condition was known. The figure did not
differ from that in the nonepilepsy group. MRI is the neu-
roimaging method to be preferred in etiologic diagnosis
of epilepsy. This had been performed in only one fourth
in our series, which does not allow us to draw conclusions
for the best possible care and to respect the aloofness of
some individuals with autism and not put these persons
in stressful and frightening situations. Conversely, such
respect must never be an excuse for not investigating, di-
agnosing, and treating epilepsy in people with autism and
MR. This is supported by the fact that two individuals of
epilepsy surgery with good results.
Seizure frequency was more than once a year in two
thirds in our study, as in Forsgren’s study (26) on the
prevalence of epilepsy in MR. Of the 11 adults who had
intractable epilepsy, there was some concern that the indi-
viduals did not have truly medically intractable epilepsy,
given that some had tried only a very limited number
of AEDs, and five did not have regular contact with a
Epilepsia, Vol. 46, No. 6, 2005