Childhood absence epilepsy: Behavioral, cognitive, and linguistic comorbidities


Address correspondence to Rochelle Caplan, M.D., Semel Institute for Neuroscience and Human Behavior, 760 Westwood Plaza, Los Angeles, CA 90024, U.S.A. E-mail:


Purpose: Evidence for a poor psychiatric, social, and vocational adult outcome in childhood absence epilepsy (CAE) suggests long-term unmet mental health, social, and vocational needs. This cross-sectional study examined behavioral/emotional, cognitive, and linguistic comorbidities as well as their correlates in children with CAE.

Methods: Sixty-nine CAE children aged 9.6 (SD = 2.49) years and 103 age- and gender-matched normal children had semistructured psychiatric interviews, as well as cognitive and linguistic testing. Parents provided demographic, seizure-related, and behavioral information on their children through a semi-structured psychiatric interview and the child behavior checklist (CBCL).

Results: Compared to the normal group, 25% of the CAE children had subtle cognitive deficits, 43% linguistic difficulties, 61% a psychiatric diagnosis, particularly attention deficit hyperactivity disorder (ADHD) and anxiety disorders, and 30% clinically relevant CBCL broad band scores. The most frequent CBCL narrow band factor scores in the clinical/borderline range were attention and somatic complaints, followed by social and thought problems. Duration of illness, seizure frequency, and antiepileptic drug (AED) treatment were related to the severity of the cognitive, linguistic, and psychiatric comorbidities. Only 23% of the CAE subjects had intervention for these problems.

Conclusions: The high rate of impaired behavior, emotions, cognition, and language and low intervention rate should alert clinicians to the need for early identification and treatment of children with CAE, particularly those with longer duration of illness, uncontrolled seizures, and AED treatment.

Children diagnosed with childhood absence epilepsy (CAE) represent approximately 8% of cases of epilepsy among school-aged children (Pavone et al., 2001). Based on earlier studies, CAE has been considered a benign disorder with relatively easily attained seizure control and minimal involvement of cognition and behavior (Dieterich et  al., 1985; Covanis et al., 1992). However, more recent outcome studies report a varying rate of seizure control, associated generalized tonic–clonic seizures, progression to juvenile myoclonic epilepsy (Wirrell et al., 2001; Trinka et al., 2004; Grosso et al., 2005), as well as learning and cognitive difficulties (Pavone et al., 2001). In addition, young adults with a history of CAE have high rates of work and social difficulties, persistent difficulties in their relationships with family and friends, fewer regular social outings with friends or their partner, as well as psychiatric and emotional difficulties (Olsson & Campenhausen, 1993; Wirrell et al., 1997). These young adults are more likely to have required special educational help, had below average academic performance, and repeated a grade (Wirrell et al., 1997). The poor long-term vocational, educational, and social outcomes are found in subjects with and without adequate seizure control. However, the behavioral outcome is significantly worse in the patients with continued poor seizure control compared to those with good seizure control (Wirrell et al., 1997).

As CAE begins during childhood, there appears to be long-term unmet mental health, social, and vocational needs in patients with this “seemingly benign” epilepsy syndrome. Recent cross-sectional studies on small samples of CAE have, in fact, demonstrated that already during childhood and even at diagnosis these children have cognitive (Williams et al., 1996; Mandelbaum & Burack, 1997; Pavone & Niedermeyer, 2000; Henkin et al., 2005), linguistic (Caplan et al., 2001, 2002; Henkin et al., 2005), and behavioral/emotional problems (Williams et al., 1996; Mandelbaum & Burack, 1997; Caplan et al., 1998; Ott et al., 2001). The cognitive difficulties of these children involve visual sustained attention (Levav et al., 2002), visual spatial skills (Pavone et al., 2001), verbal and nonverbal attention (Henkin et al., 2005), as well as verbal (Nolan et al., 2004; Henkin et al., 2005; Hoie et al., 2006), and nonverbal memory (Pavone et al., 2001).

In terms of psychopathology, prior studies describe attentional deficits (Levav et al., 2002; Nolan et al., 2004; Henkin et al., 2005) and attention deficit hyperactivity disorder (ADHD), mainly inattentive type (Dunn et al., 2003), as well as affective/anxiety disorder diagnoses (Caplan et al., 2005b) in CAE. Using the child behavior checklist (CBCL), Williams et al. (1996) reported thought and attention problems, as well as withdrawn behavior in CAE children.

Of the few childhood studies that examined the relationship of seizure variables with cognition, language, and behavior in CAE, some show that the onset before age of 4  years is significantly associated with more global cognitive and nonverbal memory impairment (Pavone et al., 2001) as well as learning deficits (Lagae et al., 2001). Others find that poor seizure control is related to learning and behavioral difficulties (Williams et al., 1996) and to impaired use of language to organize and formulate thoughts (i.e., discourse deficits) in younger CAE children. However, duration of illness is associated with discourse deficits in older CAE patients (Caplan et al., 2006). The presence of a psychiatric diagnosis and the type of diagnosis in CAE are unrelated to seizures variables (Caplan et al., 1998; Ott et al., 2001).

We previously reported psychopathology findings and treatment rate of a relatively small CAE sample (Caplan et al., 1998; Ott et al., 2003). The goal of the observational study presented in this paper was to describe the cognitive, linguistic, and behavioral comorbidities and their correlates in a large sample of children with CAE. We hypothesized that significantly more children with CAE would have cognitive deficits, impaired language, and psychopathology compared to normal age and gender matched children. In addition, they would have significantly more problems with attention, thinking, and withdrawn behavior compared to normal age and gender matched subjects. Within the CAE group, we posited that significantly more children would meet diagnostic criteria for ADHD than for psychiatric diagnoses, such as depression and anxiety disorders. In terms of the correlates of the comorbidities in the CAE group, we hypothesized that seizure-related variables would be associated with IQ and linguistic deficits, and explored the role played by seizure, cognitive, linguistic, and demographic variables in the predicted psychopathology findings.



The study included 69 CAE children, aged 6.7–11.2 years with IQ scores of 70 and higher and 103 normal children. Table 1 presents demographic features of the sample. We determined socioeconomic status (SES) using the Hollingshead 2 factor index (Hollingshead, 1973), based on parental occupational and educational status. There were no significant differences in the demographic variables of the CAE and normal groups. Sixty-six subjects were recruited during 1994–1998 and 73 during 1999–2005.

Table 1.   Demographic features of study groups
Age (years)9.64 (2.49)9.9 (2.33)
Socioeconomic status  
 High (i, ii)19%25%
 Low (iii–v)81%75%

We recruited 35% of the CAE subjects from tertiary centers (i.e., UCLA and USC based clinics) and 65% from community services (i.e., Los Angeles and Anaheim Kaiser Permanente, the Los Angeles and San Diego Chapters of the Epilepsy Foundation of America, private practices). There were no significant differences in the recruitment source of the new (tertiary 30%, community 70%) and old cohorts (tertiary 38%, community 62%).

The primary study inclusion criterion for each subject was that he/she had a diagnosis of CAE and at least one seizure in the year prior to participation in the study. A pediatric neurologist at each recruitment site made a diagnosis of CAE according to the International Classification of Epilepsy (Commission, 1989). As described in this classification, all CAE patients had EEG evidence of 3 Hz spike and wave in addition to absence seizures induced by hyperventilation. We excluded patients with a mixed seizure disorder, previous epilepsy surgery, atypical spike and wave complexes, juvenile myoclonic epilepsy, generalized tonic clonic seizures, a neurological illness other than epilepsy, chronic medical illness, imaging evidence for structural brain abnormalities, a metabolic disorder, a hearing disorder, mental retardation based on school/classroom placement, and bilingual speakers of American English who did not attend English speaking schools or speak English at home.

UCLA IRB approved recruitment flyers were available for parents of CAE children at each recruitment site. Parents who decided to enter their children into the study contacted the study coordinator who provided information about the study and used a UCLA IRB approved telephone script to determine if the children met the study’s inclusionary but none of the exclusionary criteria. The study coordinator also contacted the child’s pediatric neurologist to confirm the child’s diagnosis and to rule out exclusionary criteria. One UCLA pediatric neurology investigator (WDS.) reviewed the history, EEG records, and diagnosis of each CAE subject from the different recruitment sites. If he did not concur with the diagnosis or EEG findings, the child was not included in the study.

The parents and child’s medical records provided information on seizure variables. The mean age of onset and duration of illness in the CAE group were 6.15 (SD 2.52) and 3.54 (SD 2.82) years, respectively. Twenty-five percent of the CAE group had at least one seizure, 4% had 2–10, and 71% had 10 or more seizures during the year prior to participation in the study. Twelve percent of the CAE group received no antiepileptic drugs (AEDs), 76% were on AED monotherapy (51% on valproic acid, 35% on ethosuximide, 14% other), and 13% on AED polytherapy. Ten percent of the children had experienced febrile convulsions and 4% prolonged febrile seizures (i.e., >5 min). Six CAE subjects had slowing on EEG. Comparison of the new and old CAE cohorts on seizure variables revealed significantly higher mean seizure frequency in the new cohort [t (66) = 2.47, p < 0.02] but no other significant differences between groups.

To include children from a wide range of ethnic and SES backgrounds similar to that of the CAE group, we recruited the normal control subjects from four public and two private schools in the Los Angeles community. The study coordinator screened potential participants for neurological, psychiatric, language, and hearing disorders through a telephone conversation with a parent. We excluded children with diagnoses of these disorders in the past from the study.


Written informed consent and assent were obtained from the parents and children, respectively, in accordance with policies of the Human Subject Protection Committees of the UCLA after the procedures were fully explained.



The Wechsler Intelligence Scale for Children-Revised (WISC-R), (Wechsler, 1974) given to children tested from 1994 to 1998, and the Wechsler Intelligence Scale for Children-3rd edition (WISC-III), (Wechsler, 1991) administered to children tested from 1999 to 2005, generated full scale, verbal, and performance IQ scores (see detailed description in Caplan et al., 2005b).


The Test of Language Development [TOLD (Newcomer & Hammil, 1988)]

As presented in (Caplan et al., 2006), the spoken language quotient (SLQ) was derived from the three forms of this test, the TOLD-2 primary, the TOLD-2 intermediate, and the Test of Adolescent Language (TOAL), administered to 38%, 50%, and 12%, respectively, of the CAE and normal subjects.


The Kiddie Schedule for Affective Disorders and Schizophrenia (K-SADS)

The Schedule for Affective Disorders and Schizophrenia for School-Age Children, Epidemiologic Version (K-SADS-E) (Orvaschel & Puig-Antich, 1987) during 1994–1998 and the Present and Lifetime Version (K-SADS-PL) (Kaufman et al., 1997) during 1998–2005, were administered separately to each child and parent by RC or a research assistant trained in the administration of the interview, as detailed in (Caplan et al., 2005b).

The Child Behavior Checklist

Parents completed the 118 behavior problem items (Achenbach, 1991), as elaborated in Caplan et al. (2005b). The cutoff point for borderline/clinically significant pathology in this study was based on the age and gender normed T score of 60 for the broad band and 67 for the narrow band scales (Achenbach, 1991).


Parents provided information on whether the child was undergoing or had received any type of treatment for cognitive, linguistic, or behavioral difficulties.

Data analysis

Prior to statistical analyses, all data were inspected for outliers, skewness, kurtosis, and homogeneity of variance to ensure their appropriateness for parametric statistical tests. ANCOVAs, controlling for demographic variables (namely, age, gender, SES, and ethnicity), were used to compare the CAE and normal groups on cognition and SLQ, and logistic regressions were used to compare the two groups on the presence of a psychiatric diagnosis and CBCL scores in the clinical/borderline range. In addition to demographic variables, IQ was included as a covariate in the logistic regressions. To compare the two groups on their mean CBCL scores, ANCOVAs with demographic and IQ variables in the model were estimated. Given the relationship of cognition, language, and psychopathology with age, gender, SES, and ethnicity, we controlled for these demographic measures in the between-group comparisons. All tests were two-tailed and a significance level of 0.05 was adopted. As all our analyses were hypothesis-driven, we did not correct for multiple tests.

We then determined which of the demographic and seizure variables were predictive of cognition, language, and psychopathology in the CAE group. The seizure variables that were used as predictors included age of seizure onset, duration of illness (time from age of onset to participation in study), number of seizures (log-transformed), history of prolonged seizures, history of febrile convulsions, and the number of AEDs (subdivided into no AEDs, monotherapy, and polytherapy). For IQ and SLQ scores, the demographic and seizure-related variables were included in the model while for psychopathology, the above as well as IQ scores were utilized as the starting point.

We first included all the variables as predictors in a  general linear model. We then used a combination of a stepwise strategy (in which the variables are selected for either inclusion or exclusion from the model in a sequential fashion based solely on statistical criteria) and inclusion or  exclusion of variables based on careful scrutiny of the resulting model. Thus, following the fit of the model from stepwise selection, the importance of each variable included in the model was verified. We also checked for variables whose coefficients change markedly in magnitude when other variables were excluded. This process of deleting, refitting, and verifying was performed until a final model was obtained which explained the data.

For these models, the variance explained was calculated in the case of continuous outcome measures. For discrete outcome measures, the area under the receiver operating-characteristic (ROC) curve (receiver operating-characteristic curve), was then estimated. Area under the ROC curve measures the ability of the predictor variables to correctly classify the subjects into one or the other of the discrete states; hence, area under the curve (AUC) is a measure of the discrimination of the predictors.


This section presents the between group IQ, SLQ, and psychopathology findings, followed by the within CAE modeling findings, secondary analyses, and treatment rate.

Between group differences

IQ and SLQ scores (Table 2)

Table 2.   Mean IQ and SLQ scores and distribution of scores in the CAE and normal groups
  Distribution of scores  
 Means (SD)CAENormal  
 CAENormalF(df)p<1SD (%)Mean (%)>1SD (%)<1SD (%)Mean (%)>1SD (%)X2(2)p
  1. SD, standard deviation; FSIQ, full scale IQ; VIQ, verbal IQ; PIQ, performance IQ; SLQ, speech language.

FSIQ101 (15.61)111 (13.22)19.71 (1,166)<0.00012745276435119.09 0.0001
VIQ100 (17.46)112 (15.37)19.21 (1,166)<0.00012939329365515.11 0.0005
PIQ101 (15.35)108 (11.85) 8.53 (1,166) 0.00122492974944 9.580.008
SLQ 94 (17.16)104 (13.31)16.78 (1,155)<0.000143381915533218.06 0.0001

The mean IQ and SLQ scores of the CAE group were significantly lower than those of the normal group. Significantly more CAE children had below average IQ and SLQ scores than the normal group. The SLQ scores of the old CAE cohort were significantly lower than those of the new cohort [t (59) = 2.13, p < 0.04).


Psychiatric diagnosis (Table 3)

Table 3.   Psychopathology in the CAE and normal groups
  1. ADHD, attention deficit disorder; NS, not significant.

Psychiatric diagnosis
 Yes 61%15%23.071  <0.0001
 No 39%85% 
Type of diagnosis
 ADHD 26%6%13.5910.0002
 Affective/anxiety 20%7%6.4210.01
 ADHD + Affective/anxiety 11%2%5.9910.01
 Other 4%0%2.221NS
 Somatic complaints34.4%5.9%20.421  <0.0001
 Social problems23.4%5.9%6.0710.01

Controlling for IQ and demographic variables, significantly more CAE subjects had a psychiatric diagnosis compared to the normal subjects. ADHD and affective/anxiety disorder diagnoses were the most frequent diagnoses in the CAE group. Twenty-one children had ADHD, 10 with hyperactivity and 11 with inattention, but no hyperactivity. Of the 20 CAE subjects with affective/anxiety disorder diagnoses, 15 had anxiety disorders, 4 had depression, and 1  had both anxiety disorder and depression. There were no significant differences in the number of children with psychiatric diagnoses and in the type of diagnosis in the old and new CAE cohorts.


With IQ and demographic variables in the model, significantly more CAE subjects had clinically relevant parent based broad band and narrow band CBCL factor scores, other than anxious/depressed, delinquent, and aggressive factor scores, compared to the normal subjects (Table 3). Their mean CBCL scores, except the anxious/depressed and delinquent scores, were also significantly higher than those of the normal subjects with IQ and demographic variables in the model (Fig. 1).

Figure 1.

 Mean broad and narrow band CBCL scores for CAE and normal groups, *p < 0.05.

Of note, the most commonly found CBCL narrow band factor scores in the clinical/borderline range in the CAE children were attention problems in 37.5% and somatic problems in 34.4%. Clinically relevant social and thought problems were found in 23.4% and 20.3%, respectively, of the CAE sample. The mean scores of these four narrow band CBCL factor scores were all significantly higher than in the normal subjects (Fig. 1).

Modeling of cognition, language, and psychopathology in CAE

IQ and SLQ

Modeling IQ and SLQ scores in the CAE group with demographic and seizure-related variables as predictors revealed a significant effect of increased duration of illness for lower verbal IQ [F (1,63) = 5.30, p < 0.02] and SLQ scores [F (1,56) = 12.54, p < 0.0008] and of AED treatment (both monotherapy and polytherapy vs. no AEDs) for each of the IQ [F (2,63) = 5.50 – 7.54, p < 0.005) and SLQ scores [F (2,56) = 6.85, p < 0.002]. Computing these analyses with type of AEDs (valproic acid vs. ethosuximide) in the model indicated no significant effect. Ethnicity (non-Caucasian < Caucasian) was associated with full scale IQ, verbal IQ, and SLQ scores [F (1,63) = 4.86 – 7.48, p < 0.03].

Seizure variables accounted for 19% of the variance of full scale and performance IQ scores, 16% of the variance of both verbal IQ scores, and 24% of the variance of SLQ scores. Demographic variables accounted for 1–5% of the variance of these scores.


Psychiatric diagnosis

Longer duration of illness [X2 (1) = 5.44, p < 0.02) and seizure frequency (X2 (1) = 4. 71, p < 0.03) were significantly associated with the presence of a psychiatric diagnosis. For every additional year of having CAE, a child had a 1.35 times greater chance (95% CI: 1.05–1.73) of having a psychiatric diagnosis. Similarly, a CAE subject with an increase of seizure frequency by 10 or more seizures per year was 1.2 times (95% CI: 1.02–1.38) more likely to have a psychiatric diagnosis. These two seizure variables were also significantly related to having an ADHD (duration: X2 (1) = 7.02, p < 0.008; seizure frequency: X2 (1)  = 3.84, p < 0.05) and anxiety disorder diagnosis (duration: X2 (1) = 4.61, p < 0.03; seizure frequency: X2 (1) = 5.05, p < 0.02). In addition, girls with CAE were 5.8 times more likely than boys (95% CI: 1.05–31.95) to have an anxiety disorder diagnosis. The logistic regression for psychiatric diagnosis indicated that the area under the ROC curve was 0.72 with seizure frequency and duration of illness as predictors.


Modeling the mean broad band CBCL scores in the CAE group revealed that the children on AED monotherapy had significantly higher mean total [F (2,62) = 4.22, p < 0.02] and internalizing scores [F (2,62) = 3.76, p < 0.03] than those on no AEDs. Longer duration of illness predicted mean externalizing scores in the clinical/borderline range [F (1,61) = 8.65, p < 0.005]. Seizure variables accounted for 12%, 9%, and 10% of the mean total, internalizing, and externalizing CBCL scores, respectively.

In terms of the CBCL narrow band scores, an ADHD diagnosis predicted increased attention problems [F (4,56) = 4.66, p < 0.03]. Children with later age of onset had significantly higher mean somatic complaint scores than those with earlier onset [F (1,59) = 3.78, p < 0.05]. The presence of a psychiatric diagnosis [F (1,59) = 5.71, p < 0.02] and type of diagnosis [i.e., ADHD, F (4,56) = 2.98, p < 0.03] were related to significantly higher thought problem scores. The CAE children with lower mean full scale IQ scores had significantly more social problems than those with higher mean IQ scores [F (1,62) = 4.86, p < 0.03].

Secondary analyses

Age of onset, duration of illness, chronological age, and seizure frequency are interrelated variables. Although the 18 CAE subjects with onset at or before age 4 years tended to be younger [t (66) = 1.78, p < 0.08] and had longer duration of illness [t (66) = –4.17, p < 0.0001] than the 50 CAE subjects with onset after age 4 years, there were no significant age of onset related differences in seizure frequency [t (66) = –0.82, p < 0.4]. The study’s between and within group findings were also unrelated to the presence of slowing in the EEG of six CAE.

Although 43% of the 16 CAE children with anxiety disorder diagnoses had clinically relevant somatic complaint scores, there were no significant differences compared to those without a psychiatric diagnosis (33%) and those with other diagnoses (24%) [X2 (2) = 3.59, p < 0.2]. Among the 11 CAE children with anxiety/depression CBCL factor scores in the clinical/borderline range, 4 had ADHD,  1 had affective/anxiety diagnoses, 4 had both ADHD and affective/anxiety disorder diagnoses, and 2 had other diagnoses.


Only 23% of the subjects received some form of intervention (current, past) for these comorbidities.


Supporting the findings of prior studies on smaller CAE samples (Caplan et al., 1998; Dunn & Austin, 1999; Lagae et al., 2001; Ott et al., 2001; Pavone et al., 2001; Henkin et al., 2003, 2005), we found a broad range of untreated cognitive, linguistic, and behavioral/emotional comorbidities in CAE. Thus, despite mean average IQ and SLQ scores, one fourth of the CAE children had subtle cognitive difficulties, just under a half had linguistic deficits, almost two-thirds had psychiatric diagnoses, particularly ADHD and anxiety disorder diagnoses, and about one third had CBCL broad band and narrow band scores (i.e., somatic complaints, social problems, thinking, attention) in the clinical/borderline range.

The association of seizure variables (i.e., duration of illness, AED treatment, seizure frequency) with IQ and SLQ scores, the presence and type of psychiatric diagnosis, as well as CBCL scores emphasize that CAE is not a “benign” disorder. Treatment of these comorbidities in only 23% of the children together with the reported poor functional outcome (Olsson & Campenhausen, 1993; Wirrell et al., 1997) of CAE further emphasize the importance of timely detection and treatment of these problems.

In terms of the study’s psychopathology findings, the role seizure control played predicting the presence of a psychiatric diagnosis is similar to Wirrell et al’s follow-up findings (Wirrell et al., 1997) of a higher rate of psychiatric diagnoses in the CAE subjects with poor seizure control compared to those with good seizure control. The increased frequency of ADHD, problems with attention, thinking, and social problems, and association of ADHD with problems in attention and thinking are also comparable to findings in other children with epilepsy (Dunn & Austin, 1999; Hesdorffer et al., 2004; Rodenburg et al., 2005; Hermann et al., 2007; Jones et al., 2007). However, we had not predicted the high rate of anxiety disorder diagnoses and somatic complaints found in our CAE sample.

Examination of variables associated with anxiety disorder diagnoses revealed that, as reported in children without epilepsy with anxiety disorders, girls with CAE were more likely to have anxiety disorder diagnoses than boys (Lewinsohn et al., 1998). The association of anxiety disorders with both increased seizure frequency and duration of illness in the CAE subjects implies that repeated episodes of loss of consciousness over a prolonged period might be associated with a feeling of a lack of control and increased anxiety. In fact, Dunn et al. (1999) have shown that external locus of control, perceiving others as socially powerful, was associated with the anxiety/depression factor score of the Youth Self Report (Achenbach, 1991) in adolescents with epilepsy.

In addition, the significant relationship of increased somatic complaints with later age of onset might also imply that the uncertainty of unpredicted repeated episodes of brief loss of consciousness could be more anxiety provoking in older children. Children with anxiety disorders without epilepsy, particularly older rather than younger children, have frequent somatic complaints (Ginsburg et al., 2006). Alternatively, parents might confuse the manifestations of CAE with behavior problems (Oostrom et al., 2001). The high parent based CBCL somatic complaint scores might, therefore, reflect parental misinterpretation of brief absence episodes as the child or adolescent “not feeling well” (i.e., a somatic complaint).

From the biological perspective, the frontal lobe, thalamus, and impaired 5HT metabolism are implicated in both absence seizures (Holmes et al., 2004; Meeren et al., 2005; Betting et al., 2006a, 2006b; Midzyanovskaya et al., 2006) and anxiety disorders (Jakus et al., 2004; Bercovici et al., 2006; Monk et al., 2006). More specifically, EEG recordings demonstrate that absence seizures begin with discrete spikes that are often unilateral in the dorsolateral frontal and orbital frontal regions and then evolve to engage orbital frontal and mesial frontal regions during the repeating spike and wave cycles (Holmes et al., 2004). MRI studies reveal increased gray matter concentration in the superior mesiofrontal region (Betting et al., 2006b) and larger anterior thalamus volumes (Betting et al., 2006a) in young adults with absence seizures. Rats show an association between their response to stress and the propensity to have absence seizures, reduced 5HT metabolism in the thalamus, and increased metabolism in the prefrontal cortex in contrast to those without absence seizures (Midzyanovskaya et al., 2006).

Furthermore, involvement of ventrolateral prefrontal cortex (Monk et al., 2006) and impaired 5HT metabolism (Jakus et al., 2004; Bercovici et al., 2006) are also found in anxiety disorders unrelated to epilepsy. Our clinical findings of high rates of anxiety disorder diagnoses and the association with increased seizure variables in CAE, therefore, highlight the importance of examining the possible biological basis of anxiety in pediatric CAE.

Similarly, the role of the frontal lobe in CAE (Holmes et al., 2004; Betting et al., 2006a, 2006b) might also underlie the high rate of ADHD in this epilepsy syndrome. Studies in children with ADHD without epilepsy show decreased cortical thickness (Shaw et al., 2006), volume reductions (Sowell et al., 2003; Durston et al., 2004), and reduced activation in the rostral mesial prefrontal region during tasks related to decision making/response selection aspects of the inhibitory processes (Smith et al., 2006). A recent study in children with new onset epilepsy which included both generalized and localization related epilepsy demonstrated increased rather than decreased gray matter volume in the frontal lobes of the children with ADHD compared to those without ADHD (Hermann et al., 2007).

Regarding cognitive comorbidities, our findings support those of previous studies demonstrating impaired cognition and learning in CAE (Williams et al., 1996; Mandelbaum & Burack, 1997; Pavone et al., 2001; Henkin et al., 2005). Similar to earlier findings in a combined group of children with CAE and those with cryptogenic epilepsy with complex partial seizures (Caplan et al., 2005a) and in the general population of children (see review in Harlan Drewel & Caplan, 2007), the CAE children with lower IQ scores had significantly more social difficulties.

In terms of linguistic comorbidities, smaller sample size and use of different instruments might underlie discrepant linguistic findings of impaired language skills in our study compared to (Pavone et al., 2001; Henkin et al., 2003). Auditory evoked potential findings of longer P3 latency during semantic processing and a lack of difference in P3 amplitudes over the left and right scalp during phonetic and semantic processing suggested semantic deficit and different patterns of language organization in CAE compared to normal children (Henkin et al., 2003). These findings, evidence for the role of the frontal lobe and thalamus in cognition, attention, and language (Posner & DiGirolamo, 1998; Radanovic et al., 2003; Szaflarski et al., 2006; Stringaris et  al., 2007), and the association of seizure variables with impaired cognition and language in CAE highlight the need to focus future studies on involvement of these brain regions in the cognitive and linguistic comorbidities of CAE.

Several limitations restrict generalizability of the study’s findings to other CAE populations. First, regarding seizure control, as each CAE subject had to have had at least one seizure in the year prior to participation in the study, 71% of our CAE sample had 10 or more absence seizures during that period. Thus, the association of a psychiatric diagnosis with seizure frequency could reflect the study’s inclusionary criterion. Second, Grosso et al. (2005) suggest that the wide range of seizure control in CAE outcome studies, 33–79%, reflects differences across studies in the inclusion of patients with generalized tonic–clonic seizures during the active stage of CAE, myoclonic seizures, eyelid myoclonia, and EEG features of atypical CAE. Although we excluded children who had generalized tonic–clonic seizures from the data analysis, we had no information on the presence or absence of eyelid and perioral myoclonia which, like generalized tonic–clonic seizures, predict an unfavorable prognosis (Grosso et al., 2005) or different course (Incorpora et al., 2002). Our secondary analyses, however, suggest that the presence of slowing on EEG and onset at or prior to age 4 years did not drive the study's findings.

Third, the modeling findings indicating a significant association between the number of AEDs and mean SLQ scores might reflect a cohort effect given the higher seizure frequency in the new CAE cohort and lower mean SLQ scores in the old cohort. Of note, as described in the Methods section, there were no significant differences in the recruitment sources (tertiary vs. community) of the new and old cohorts.

Fourth, SLQ scores based on three age related forms of the language instrument, the TOLD-P, TOLD-I, and TOAL do not rule out the possible role of undiagnosed linguistic deficits as an associated factor. Fifth, although we used different IQ instruments (i.e., WISC-R, WISC-III), we found no significant differences in the mean IQ scores of the children tested with the WISC-R and the WISC-III. Sixth, the normal group had significantly higher mean IQ score than the CAE subjects. However, we controlled for these differences in the between group analyses of language and psychopathology. In addition, these between group IQ differences played no role in the modeling analyses conducted within the CAE group. Seventh, while we computed multiple statistical tests, they were hypothesis driven, and we reported findings with a p-level below 0.05.

In conclusion, our findings alert clinicians to the importance of identifying and treating the broad spectrum of comorbidities in CAE, particularly in children with longer duration of illness, ongoing seizures, and AED treatment. They also point to the need to elucidate the role of the underlying pathology and psychosocial variables in the comorbid anxiety and ADHD diagnoses, impaired cognition, and linguistic deficits of pediatric CAE.


This study was supported by grant NS32070 (RC). We appreciate the technical assistance of Amy Mo, Caroline Bailey, Ph.D., Kimberly Smith, M.A., Joanna Wu, and Sona Hovsepian.

Conflict of interest: We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. The authors have no conflicts of interest to declare regarding the study presented in this paper and preparation of the manuscript.