Autism Spectrum Disorders in Children with Seizures in the First Year of Life—A Population-Based Study

A high frequency of seizure disorder was an important factor in demonstrating that autism had a neurobiological etiology (Barton and Volkmar, 1998). According to one review (Tuchman and Rapin, 2002), the data on reported frequency of epilepsy in autism ranges widely (5%–38.3%). Schain and Yannet (1960) were probably first to provide data that supports this association; in their study, 21 of 50 severely impaired autistic participants had a history of one or more seizures. Of those, 10 had their seizure onset in the first year of life, and three were diagnosed with infantile spasms. In a more recent study of 246 children with autism spectrum disorder (ASD), the majority of those who also had epilepsy (13 of 16) had their seizure onset in the first year of life, and four had infantile spasms (Wong, 1993). Further evidence indicates a peak incidence of seizures in early childhood in individuals with ASD (Olsson et al., 1988; Volkmar and Nelson, 1990; Steffenburg et al., 2003; Danielsson et al., 2005).

Studies on the prevalence of ASD in children with epilepsy are scarce (Steffenburg et al., 1996, 2003; Clarke et al., 2005). Steffenburg and colleagues (2003) studied children aged 6–13 years (N = 90; 47 males, 43 females) with mental retardation (MR) and active epilepsy. This group was severely impaired; over 60% had an IQ <50 (Steffenburg et al., 1996). Diagnosis was standardized with multiple sources of information, including clinical examination, semistructured interviews, and behavioral checklists. The prevalence of ASD in their study was 38% (20 males, 14 females). Of ASD cases, 35% had onset of epilepsy in the first year of life. Clarke and colleagues (2005) looked for ASD in a group of children with epilepsy aged 2–18 years (N = 97; 59 males, 38 females) who were followed in a tertiary care epilepsy clinic. Assessment of probable ASD was done by having parents respond to the Autism Screening Questionnaire (Berument et al., 1999). Thirty-one children (32%) scored 15 points or above, the recommended cutoff indicative of ASD, but only nine had a previous diagnosis of ASD. Diagnostic assessment was not part of the study design and the cognitive level of the participants was not known.

Infantile spasms constitute a significant minority of unprovoked seizures and epilepsy with onset in the first year of life (Hrachovy and Frost, 2003; Olafsson et al., 2005). Infantile spasms are in several studies found to be associated with autistic behaviors (Riikonen and Amnell, 1981; Sidenvall and Eeg-Olofsson, 1995; Askalan et al., 2003; Saemundsen et al., in press), and the most recent one was done in the same population as the present study (Saemundsen et al., in press). Unprovoked seizures with onset in the first year of life, excluding infantile spasms, consist of various epileptic syndromes, classified as generalized or partial, with or without secondary generalization (Cavazutti et al., 1984; Kramer et al., 1997). It is unknown whether such seizures with onset in the first year of life present a risk for developing ASD.

Thus we considered it of interest to study the prevalence of ASDs in a population-based group of children with unprovoked seizures (excluding infantile spasm) with onset in the first year of life, using diagnostic instruments and appropriate cognitive tests.

METHODS

The present study is a part of a larger research project on the outcome of children who had seizures in the first year of life during the period between January 1, 1982 and December 31, 2000. The primary source of data was computer and paper hospital records from the Landspitali University Hospital and the Landakotsspitali, both in Reykjavík, and the Regional Hospital in Akureyri, which were the only pediatric inpatient facilities in the country during the study period.

Inclusion criteria were at least one unprovoked seizure during the age range between 28 days and 12 months (based on gestational age). Included in the study group were all cases with a convincing description of an epileptic seizure by an eyewitness, or a description of a seizure in the presence of other supportive evidence, i.e., epileptiform changes on EEG, CNS infection, stroke, or other cerebral pathologies known to be related to seizures, or a family history of seizures in the first year of life in a first-degree relative. Febrile seizures and infantile spasms were excluded. Altogether 357 children were found in the hospital records to have discharge diagnoses of convulsions, seizures, or fits in the first year of life (International Classification of Diseases-Ninth Revision (ICD-9) no. 345.0–345.9 and 780.3; International Classification of Diseases-Tenth Revision (ICD-10) no. G40.0–G40.9, G41.0–41.9, and R56.8). After a careful chart review, 102 were considered to have had unprovoked seizures fulfilling the criteria of the study. The other 255 either had febrile seizures, other nonepileptic fits, or infantile spasms. The parents of 84 children (82.4%) agreed to participate in the ASD study.

In a telephone interview, parents were asked several questions regarding the child's seizures, development, and behavior. Children with a known neurodevelopmental disorder or parental concern regarding developmental skills or behavior were contacted again by mail, followed by a telephone call, and asked to participate in a further study on autistic behaviors. As an initial test of autistic behaviors, parents were asked to complete the Social Communication Questionnaire (SCQ) (Rutter et al., 2003a), except in the cases of severely handicapped children.

The SCQ, previously named Autism Screening Questionnaire (ASQ) (Berument et al., 1999), is a screening measure for autistic behaviors developed from the ADI-R (see below). It contains 40 items that require yes or no responses from parents. There are two versions, a lifetime version and current version. In the present study, the lifetime version was used, which “…is completed with reference to the individual's entire developmental history and produces results that are pertinent to referral for more complete diagnostic workup” (Rutter et al., 2003a, p. 1). A score of 15 points on the SCQ is the recommended cutoff to distinguish between ASDs and other diagnoses. However, in order to minimize the possibility of missing ASD in cases with milder sympomatology or in high functioning individuals, a score of 10 points was chosen as cutoff in this study. Diagnostic instruments were applied when behavioral scores reached or exceeded this cutoff.

The Autism Diagnostic Interview-Revised (ADI-R) (Lord et al., 1994) is a standardized, semistructured, investigator-based interview for caregivers of individuals with suspected autism. For a diagnosis of autism, abnormality in development has to be evident at or before 36 months and the cutoff must be reached or exceeded in all three symptom domains: (a) impaired social interaction, (b) impaired communication, and (c) stereotyped and repetitive actions. In this study, the proposed ADI-R algorithm for ICD-10, revised October 1994, was used. The Autism Diagnostic Observational Schedule (ADOS) (Lord et al., 2001) is a semistructured, standardized assessment of social interaction, communication, and repetitive behaviors, developed to accompany the ADI-R when autism or ASDs are suspected. ADOS provides algorithms and cutoffs for both autism and ASD. It consists of four modules, the selection of which depends on the verbal status and age of the individual to be assessed. The psychologists administering ADI-R and ADOS had been trained in their application at sites recognized by the authors of the instruments. The Childhood Autism Rating Scale (CARS) (Schopler et al., 1988) was also used for rating autistic behaviors. CARS was developed as an observational measure and consists of 15 items that are scored on a seven-point scale, with all items equally weighted in the total score. The proposed cutoff score for autism is 30 points.

All the participants investigated for ASD were also assessed with an appropriate cognitive test. Various tests were used, including the Wechsler Preschool and Primary Scale of Intelligence-Revised (WPPSI-R) (Wechsler, 1989), the Wechsler Intelligence Scale for Children-Third Edition (WISC-III) (Wechsler, 1992), and the Wechsler Adult Intelligence Scale-Third Edition (WAIS-III) (Wechsler, 1997). The cognitive tests were used to obtain deviation standard score or intelligence quotients (IQ), i.e., verbal IQ, performance IQ, and full scale IQ. If the child had been recently or repeatedly tested, those test results were accepted as valid and not repeated for the purpose of the study. In the case of severe handicap (n = 4), the Vineland Adaptive Behavior Scales-Survey Form (VABS) (Sparrow et al., 1984) was used to estimate the level of cognitive functioning. The VABS is a general assessment of adaptive behavior. It consists of four scales or domains that assess information regarding communication skills, daily living skills, socialization, and motor skills as well as a standard score, the adaptive behavior composite, which sums up the performance in the four domains. The VABS manual also provides “age equivalents” corresponding to domain raw scores.

ASD is an umbrella term for life-long developmental disorders of brain function with autism representing the more severe end of the spectrum (Tuchman and Rapin, 2002). This term was deemed more suitable for the purpose of the present study than the more encompassing term pervasive developmental disorder, which includes Rett's syndrome and other childhood disintegrative disorders. Otherwise, the ICD-10 (World Health Organization, 1992, 1993) was used for classification, except in the case of attention-deficit hyperactivity disorder (ADHD). The subclassification of mental retardation based on measured IQ was as follows: mild 50–69, moderate 35–49, severe 20–34, and profound below 20. Epilepsy was defined as recurrent episodes of unprovoked, paroxysmal seizure activity (Cowan, 2002). Active epilepsy was defined as seizure within 5 years or still on antiepileptic medication. Children were classified cryptogenic if they had normal neurological and developmental history, normal neurological examination, no known associated etiological factor, and negative diagnostic evaluation. All other cases were classified as symptomatic. All diagnoses presented in this study are subsumed under the term “neurodevelopmental disorder” (NDD), which is used for genetic or acquired developmental disorders affecting the nervous system at the time of rapid development and causing various forms of neurological dysfunction.

Age at assessment ranged from 4 to 20 years with a mean age of 11 years and 1 month (standard deviation 6 years and 1 month). The clinical diagnosis of a major NDD was made by an interdisciplinary team at a tertiary center specialized in developmental disabilities. Other NDDs were diagnosed by pediatricians and psychologists. All diagnoses were reviewed for the purpose of the study. Prevalence was calculated as the percentage of cases in the studied group and the exact 95% confidence intervals were found assuming binomial distribution (Armitage and Berry, 1991). Pearson chi-square was used for categorical data and Fisher's exact test when a cell had an expected count of less than five. For the calculation of sensitivity and specificity of the SCQ, the Wilson's method was used (Altman, 2003). For the Wechsler IQ data, a one-sample t-test was used to test the deviance from the mean of 100, and the Wilcoxon test for comparing paired verbal and performance measures, because the dispersion of test scores was not of normal distribution.

All the parents gave their written informed consent, as did the older and more able children. The study was approved by the Data Protection Authority and the National Bioethics Committee in Iceland.

RESULTS

One hundred two children (42 boys and 60 girls) had experienced at least one unprovoked seizure in the first year of life during the period under study. Of those, consent was given for 84 children (82.4%), composed of 28 boys and 56 girls. The gender proportion is similar among the participants as among the total sample of children with seizure in the first year of life. Seventy-two participants (85.7%) had had more than one seizure. Fifty-three children were without parental concern regarding development or behavior and 31 had parental concern and were investigated for possible ASD (Fig. 1). Twenty-four children, or 28.6% (24/84), had at least one NDD: MR 14.3%; cerebral palsy (CP) 8.3% (6.0% dual diagnosis MR and CP); ADHD 6.0%; disorders of scholastic skills 3.6%; and receptive language disorder 2.4% (Table 1). More boys than girls had NDD (32.1% vs. 26.8%), although the difference was not statistically significant (p = 0.608).

Of the 31 children (12 boys and 19 girls) who were investigated for possible ASD, all but two (cases no. 8 and 20) had more than one seizure. Seventy-one percent (22/31) of these children were classified as cryptogenic. Twenty-nine parents answered the SCQ, but in two cases the parents were not asked to complete the SCQ because the children had severe CP and profound MR. Twelve children scored at or above the cutoff of 10 points on the SCQ. The SCQ was incompletely filled out by the parent of case no. 2 but this child was included in the diagnostic stage because, in conversation with the parent, it was clear that the child would have scored above the cutoff. Hence, 13 children were examined further for ASD. One was tested with ADOS without being further considered suspect of ASD. All the others received ADI-R (n = 12) and ADOS (n = 10), and/or CARS (n = 5) testing.

Six (6/84) children including four girls and two boys or 7.1% (95% confidence interval 2.7–14.9) were diagnosed with ASD, according to ICD-10 (Table 2). All were mentally retarded: two had mild MR, two had moderate MR, and two had profound MR. Three of the children with ASD had congenital brain abnormalities: one had a porencephalic cyst, malformation of the cerebellum, and spastic hemiplegia; one had congenital toxoplasmosis and blindness; and one had trisomy 21. Cases no. 2, 6, and 16 received their ASD diagnosis within the study, but cases no. 22, 23, and 24 had been diagnosed previously. In view of the principles of use in the ADI-R manual that the interview is appropriate for individuals of mental age above 24 months (Rutter et al., 2003b), one of the children (case no. 24) should be excluded. With this restriction, the prevalence of ASD becomes 6.0% (95% confidence interval 2.0–13.4).

The number of seizures in the first year of life (1 vs. >1) did not seem to be associated with ASD (p = 0.587). At the time of follow-up 14 children had active epilepsy and of the six children with ASD, three had active epilepsy. Twenty-one children were known to have benign infantile familial convulsions and none of these had ASD. Electroencephalogram (EEG) was performed at least once in 77 children (91.7%), either at the time of the first seizure or later. Of these, 12 (15.6%) were abnormal. Of the children later diagnosed with ASD, five had EEG, four of which were abnormal (generalized epileptiform: 1; focal epileptiform: 2; epileptiform but unknown if focal or generalized: 1). Of the non-ASD children, 72 had EEG, eight of which were abnormal (nonspecific slowing: 1; generalized epileptiform: 1; focal epileptiform: 6).

Of the 13 children examined for ASD, seven did not receive an ASD diagnosis. Their SCQ scores ranged from 10 to 23. The lower cutoff score of 10 points on the SCQ applied in this study did not give additional cases of ASD. The child with the lowest SCQ score had CP and MR. The child with the highest score on the SCQ was the only child in this group who exceeded the cutoff without neurodevelopmental diagnosis. Developmental history and the family background were complicated and this child had subclinical attention deficit, a history of difficulties learning to read, and perfectionist tendencies. Of the children with intermediate scores on the SCQ (12–16 points), three had ADHD, one had a receptive language disorder, and one had a specific reading disorder. Counting the case with an incomplete SCQ as negative and using the cutoff of 10 points on the SCQ resulted in seven false positives with sensitivity of 0.83 and specificity of 0.70. Using the recommended cutoff of 15 points resulted in unchanged sensitivity but higher specificity of 0.87.

Scores for those tested with the Wechsler intelligence tests (27/31) deviated significantly from 100 in the negative direction for both the verbal (mean = 85.50; p = 0.002) and the performance (mean = 89.88; p = 0.031) scales. No difference was found between mean verbal and mean performance scores in either direction for the group as a whole (p = 0.396). The same was true for the children (18/31) who tested in the normal range (≥70) and had a mean full scale IQ of 98.28 (p = 0.191).

DISCUSSION

To our knowledge, this is the first population-based study on ASD in children who had unprovoked seizures with onset in the first year of life, other than infantile spasms. The prevalence of ASDs found in this group was 6.0%–7.1%, depending on the inclusion of one participant with mental age below 24 months (Rutter et al., 2003b). The lower 95% confidence limits for both figures (2.0 and 2.7) exceed the estimated 0.6%–1% prevalence of ASDs in the general population (Chakrabarti and Fombonne, 2005; Baird et al., 2006). Our results are not directly comparable with the study of Steffenburg and colleagues (2003), since all their participants were mentally retarded and had active epilepsy, but only 14.3% of the children in the present study had MR. Neither are the results of Clarke and colleagues (2005) comparable with our results, since diagnostic assessment and cognitive measures were not part of their study design.

In the present study, all of the children with ASD were mentally retarded and had previously received the diagnosis of epilepsy. Three had congenital brain abnormalities, and three were without known associated neurological factors. It is interesting to note that previously, the prevalence of associated medical conditions with known or suspected etiologic relationship with autism has been estimated at 10%–15% (Barton and Volkmar, 1998; Kielinen et al., 2004). Hence, children who have epilepsy in the first year of life and who later develop ASD may constitute a subgroup with higher frequency of congenital brain abnormalities than reported for the group of other children with ASD (Olsson et al., 1988).

A recent review of prevalence surveys of autism found that the mean gender ratio was more than four males to every female (Fombonne, 2003). Conversely, the autistic group of the present study comprised two boys and four girls, which reflects the gender ratio of all participants and those investigated further for possible ASD. Thus, the gender ratio of the autistic group in the present study may be regarded as additional evidence supporting the view that children with ASD who had epilepsy in the first year of life may differ as a group from other children with ASD. Other studies have concluded that severe cognitive impairment and ASD are associated with epilepsy and the female gender (Tuchman et al., 1991; Elia et al., 1995; Danielsson et al., 2005).

A notable portion of children diagnosed with ASD and having a history of epilepsy did have their seizure onset in the first year of life (Wong, 1993; Steffenburg et al., 2003; Danielsson et al., 2005). This is not unexpected, as the age-specific incidence of epilepsy in childhood is highest during a child's first year (Tsuboi, 1988; Hauser et al., 1993; Sidenvall et al., 1993; Camfield et al., 1996; Olafsson et al., 2005). Epilepsy with onset in the first year of life is more often considered symptomatic than epilepsy with onset later in childhood (Czochańska et al., 1994; Kramer et al., 1997; Kramer, 1999; Datta and Wirrell, 2000). If symptomatic seizures in the first year of life are associated with MR (Cavazzutti et al., 1984, Kramer et al., 1997) and MR is associated with ASD (Nordin and Gillberg, 1996; Fombonne, 2003; de Bildt et al., 2005), it follows that there is an association between symptomatic seizures in the first year of life and ASDs (Olsson et al., 1988; Askalan et al., 2003; Pavone et al., 2004).

The association between epilepsy, MR, and ASD is complicated and the existence of a common causal factor or factors is of course a possibility (Elia et al., 1995; Steffenburg et al., 2003; Pavone et al., 2004). However, since the epilepsy is often found prior to diagnosis of ASD and MR, as is the case in present study, the temporal requirement for possible causation is met, and thus the influence of seizures themselves on the developing brain seems likely (Tuchman and Rapin, 1997; Asano et al., 2001; Clarke et al., 2005).

This study does have limitations. First, although the material is population based only children with known NDD or parental concern received the SCQ as an initial test of autistic behaviors. A stronger design would have been to give the SCQ to all those willing to participate (N = 84), and then take a random sample from the group that scored below the cutoff for diagnostic assessment, in view of possible adjustment of the prevalence found. Second, it was not possible to keep the professionals who were engaged in the diagnostic process blind to the results of the SCQ or other sources of information.

The study has five principal strengths. First, the data on seizure cases was collected prior to the diagnosis of ASD and independently of it, which minimizes the possibility of case-ascertainment bias. Second, the use of data from the comprehensive health care system in Iceland to identify children with seizures in the first year of life through a 19-year period ensured the inclusion of seizure cases and the recognition of epilepsy cases. Third, the systematic use of SCQ as the initial step in the search for autistic behaviors helped in the ascertainment of ASD cases. Fourth, the diagnosis of ASD was based on current diagnostic instruments, which take age into consideration, and clinical judgment, by trained and experienced professionals, ensuring reliable diagnosis. Fifth, the systematic use of ICD-10 minimizes misclassification of the NDD diagnoses.

In conclusion, the results of the present study suggest that the estimated prevalence of ASD in children with a history of seizure in the first year of life exceeds that of the general population. All the children with ASD had MR and epilepsy. There are indications that support the view that children with ASD and history of seizure in the first year of life have higher prevalence of congenital brain abnormalities and are more often female, than other children with ASD. The clinical implication is that a history of seizures in the first year of life should alert health care service providers to the possibility of accompanying NDD, which should generally become apparent during the preschool years if not already detected, and if MR is a concomitant feature of the seizures, the possibility of ASD should be evoked. Further research on the association between seizures in the first year of life and the risk of developing ASD should be investigated according to different types of epilepsy.

Acknowledgments

Acknowledgments:  The authors want to extend their gratitude toward the children and their parents for their participation in the study. Also, the authors thank the many professionals who assisted in the data gathering process: the staff at the State Diagnostic and Counseling Center, Pall Magnusson, Petur Maack, the staff at Family Services in Akureyri, other service providers, and teachers. This work was supported in part by the Memorial Fund of Helga Jonsdottir and Sigurlidi Kristjansson, and the Freemasons Fund of the Icelandic Order of Freemasons.