• Psychopathology;
  • Cognition;
  • Language;
  • Complex partial seizure disorder;
  • Child


  1. Top of page
  2. Abstract
  6. Acknowledgments

Summary: Purpose: This study examined the role of cognition, language, seizure-related, and demographic variables in the psychopathology of children with complex partial seizure disorder (CPS) of average intelligence.

Methods: One-hundred one CPS and 102 normal children, aged 5.1 to 16.9 years, had a structured psychiatric interview and cognitive and language testing. Parents provided demographic, perinatal, and seizure-related information, as well as behavioral information through the Child Behavior Checklist (CBCL) and a structured psychiatric interview about the child.

Results: Significantly more CPS patients had psychopathology, cognitive deficits, and linguistic deficits than did those in the normal group. Among the patients, Verbal IQ predicted the presence of a psychiatric diagnosis, as well as CBCL scores in the borderline/clinical range. Seizure, linguistic, and demographic variables were unrelated to psychopathology. The cognitive and linguistic deficits of the CPS group, however, were predicted by seizure factors (e.g., prolonged seizures/febrile convulsions; seizure frequency/number of antiepileptic drugs) and demographic factors (e.g., minority status).

Conclusions: Because subtle verbal cognitive deficits predict behavioral disturbances in pediatric CPSs, the study's findings highlight the importance of assessing behavior, cognition, and language in these children. They also underscore the negative impact of prolonged seizures, febrile convulsions, seizure frequency, and antiepileptic drug polytherapy on cognition and language in pediatric CPSs.

Hermann et al. (1) proposed a multifactorial model to delineate the mechanisms underlying behavioral disturbances in children with epilepsy. Relevant factors included seizure-related variables, such as type of seizure disorder, age at onset, duration of illness, EEG pattern, and neuropsychological status; antiepileptic drug (AED) variables, such as number of drugs and types of drugs; psychosocial variables, such as parents' marital status; and demographic variables, such as gender and median family income. They found that seizure control was the most powerful predictor of behavioral disturbances by using parent-based Child Behavior Checklist (CBCL) in 183 children with epilepsy, 53% with partial and 40% with primarily generalized epilepsy.

Since then, several studies have shown that seizure-related variables, particularly seizure control (2–10), type of seizure disorder (6,11–15), severity of seizure disorder (10,16–18), age at onset (4,19), duration of illness (4,20), AEDs (4,11,21–24), and focal EEG findings (6) are associated with psychopathology in children with epilepsy. When controlling for the effects of IQ, however, presence of a psychiatric diagnosis (7) and behavioral outcome (23) are unrelated to seizure variables in children with epilepsy.

As in psychopathology, different aspects of cognition are associated with seizure-related variables both in children with epilepsy who have mental retardation (25) [see review in (26)] and in children with normal intelligence (4,9,10,27–34). Nevertheless, only few studies examined the relation among psychopathology, IQ, and seizure-related variables in children with epilepsy. Sturniolo and Galletti (33) showed that emotional maladjustment in children with idiopathic epilepsy is associated with poor school performance, and Schoenfeld et al. (9) found a significant correlation between total CBCL scores and overall neuropsychological status in children with CPS. These studies did not determine how seizure-related variables interacted with both psychopathology and cognition.

Psychopathology in childhood is related to language difficulties (35–38). Children seen in psychiatric clinics for both externalizing and internalizing behavioral difficulties have high rates of undiagnosed language disorders (36,38). Linguistic deficits also are found in children with attention deficit hyperactivity disorder (ADHD) (35,37).

Only one study examined language in children with epilepsy with average intellectual skills. Schoenfeld et al. (9) reported significantly lower language scores in word knowledge, category fluency, and response to commands of increasing length and complexity in children with CPSs with average intelligence compared with their siblings, even after controlling for between-group differences in overall intelligence. Age at onset correlated significantly with the overall language score of the CPS group. No studies to date, however, considered the interaction among linguistic deficits, seizure variables, and psychopathology in medically treated children with epilepsy of normal intelligence.

Finally, demographic variables (e.g., age, gender, socioeconomic status) have been variably related to psychopathology in children with epilepsy. Earlier studies indicate that boys, particularly those with CPSs, have more behavioral problems (1,11,15,39). More-recent studies suggest that girls with epilepsy are more behaviorally impaired (40,41), and others found no gender effect (17,18,23,42). Age (17,18,40,43), socioeconomic status (1,44,45), and social acculturation (23) also have been related inconsistently to psychopathology in children with epilepsy.

Based on the findings of the previously reviewed studies, the study presented here examined how seizure factors, cognition, language, and demographic variables play a role in the behavioral disturbances of a large group of 101 children with CPSs. In adults, an association of temporal lobe epilepsy with hippocampal sclerosis, history of febrile seizures, and status epilepticus is found (46,47), which might reflect underlying perinatal abnormalities involving the hippocampus rather than the consequence of seizures. In animal models, early febrile seizures alter hippocampal connectivity even in the absence of cell loss or neurogenesis (48), and prolonged seizures produce hippocampal and extrahippocampal injury (49–52). Therefore in addition to the previously reviewed seizure-related variables, we also investigated the association of psychopathology measures with a history of prolonged seizures, febrile seizures, and perinatal abnormalities (e.g., pregnancy, delivery).

We first compared IQ, language, and psychopathology measures in the CPS group with those of children without epilepsy, controlling for demographic (e.g., age, gender, ethnicity, socioeconomic status) and perinatal differences between the groups. We hypothesized that, compared with the normal children, significantly more patients would have a psychiatric diagnosis, CBCL scores in the clinical range, higher mean CBCL scores, cognitive deficits, and linguistic deficits.

Because of the interrelation of seizure variables, such as age at onset, duration, seizure frequency, number of AEDs, history of prolonged seizures, history of febrile convulsions, and EEG patterns, we included these variables in a principal components analysis to determine independent seizure components. We then developed a model to examine the role played by seizure, cognitive, linguistic, demographic, and perinatal variables in the psychopathology of the CPS group. We hypothesized that IQ, language, demographic, and perinatal variables but not seizure variables would predict psychopathology in the CPS group. In contrast, we posited that seizure, demographic, and perinatal variables would predict both IQ and language measures in these children.


  1. Top of page
  2. Abstract
  6. Acknowledgments


The study included 101 CPS and 102 normal children, aged 5.1 to 16.9 years (Table 1). No significant differences were noted between the CPS and normal groups in demographic and perinatal variables (see Table 1). We recruited and tested 129 subjects in 1994 through 1998 and 74 in 1999 through 2003.

Table 1. Demographic and perinatal features of study groups
  1. CPS, complex partial seizure disorder.

Age (yr)10.7 (2.82)10.5 (2.46)
Socioeconomic status
 High (i, ii)26%27%
 Low (iii-v)74%73%
Delivery problems
Pregnancy problems

Table 2 presents the seizure-related features of the patients, 60% of whom were recruited from tertiary centers (e.g., UCLA- and USC-based clinics) and 40% from community sources (i.e., Los Angeles and Anaheim Kaiser Permanente, the Los Angeles and San Diego Chapters of the Epilepsy Foundation of America, private practices). A pediatric neurologist at each recruitment site made the diagnosis of CPS based on clinical history and EEG findings according to the International Classification of Epilepsy (53). As described in this classification, we also included patients with a clinical history of CPS, but no EEG evidence for focal epileptic activity.

Table 2. Epilepsy-related features of CPS group
Seizure variablesCPS
  1. CPS, complex partial seizure disorder; AED, antiepileptic drug.

Seizure frequency
 Prolonged seizures46%
 Febrile convulsions26%
 None 5%
 Age at onset (yr)5.5 (3.62)
 Duration (yr)5.2 (3.36)
EEG epileptic activity
Spikes, sharp waves, slowing
 Temporal, frontal,57%
Other, none
 Left, bilateral43%
 None, right57%

One UCLA pediatric neurology investigator (W.D.S.) reviewed the history, EEG records performed at about the time of the child's diagnosis, and diagnosis of each epilepsy 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. We excluded patients with a mixed seizure disorder, previous epilepsy surgery, magnetic resonance imaging (MRI) or computed tomography (CT) evidence of brain abnormality other than hippocampal sclerosis, atypical spike-and-wave complexes, juvenile myoclonic epilepsy, a metabolic disorder, a hearing disorder, and bilingual speakers of American English who did not attend English-speaking schools or speak English at home.

The normal children, recruited from four public and two private schools in the community, were screened for neurologic, psychiatric, language, or hearing disorders through a telephone conversation with the parent. We excluded children from the normal sample if they manifested symptoms of these disorders either at the time of the study or in the past.

We obtained information on seizure frequency during the past year, current AEDs, age at onset of seizures, duration of illness, as well as history of the number of prolonged seizures (i.e., >5 min) and of febrile convulsions from the parents and the child's medical records. Of the 101 CPS patients, 28 had nonfocal EEG findings; 21 had a left focus; 19, a right focus; and 19, bilateral foci. EEGs were unavailable for 14 CPS patients. Regarding focal EEG findings, 28 patients had interictal spikes in the temporal lobe; 10, in the frontal lobe; 10, in the frontal and temporal lobes; and seven, in other areas. Eleven children had secondary generalization, and 24 had background slowing.

We determined socioeconomic status by using the Hollingshead two-factor index (54), based on parental occupational and educational status. Hollingshead levels I and II were classified as high, and levels III–V, as low socioeconomic status. Perinatal data on number of pregnancy and delivery complications were collected from the children's mothers by using a questionnaire modified from the Yale Neuropsycho-educational Assessment Scales (55).


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 University of California, Los Angeles, after the procedures were fully explained. All study procedures were cross-sectional and administered once to each subject.

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

The epidemiologic version for school-age children, the K-SADS-E (56), was administered during 1994 through 1998, and the Present and Lifetime Version, the K-SADS-PL (57), during 1999 through 2003 by R.C. or a trained research assistant separately to each child and parent. Given the large number of diagnoses relative to the number of subjects in each diagnostic group, we grouped the diagnoses as follows. “Disruptive” disorders included attention deficit disorder, oppositional defiant disorder, and conduct disorder. “Affective/anxiety” disorders included any mood or anxiety disorder. Children with a “comorbid” diagnosis had both “disruptive” and “affective/anxiety” disorders.

Because the child or parent often talked about the child's seizures during the interview, these interviewers were not blinded with regard to the child's seizure disorder. The second clinician reviewed videotapes of the child interviews and audiotapes of the parent interviews and a consensus Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV (58) diagnosis was reached. Where a diagnostic consensus was not reached, the child was excluded from the study.

Childhood Behavioral Checklist (CBCL)

Parents completed the CBCL (59), which consists of 20 social competence and 113 behavioral problem items. Although the CBCL generates broad-band (i.e., externalizing, internalizing) and narrow-band behavioral scales (i.e., aggression, depression, hyperactivity), only broad-band scores are presented in this study because of space limitations. The selected cut point for borderline/clinically significant pathology in this study was 60 (59).


The Wechsler Intelligence Scale for Children–Revised (WISC-R) (60), administered to children tested from 1994 through 1998, and the WISC-III (61), administered to children tested from 1999 through 2003, generated Full-Scale, Verbal, and Performance IQ scores. Although the Full-Scale, Verbal, and Performance IQ scores of both instruments are highly correlated (62), children score lower on the WISC-III compared with the WISC-R.


All subjects underwent testing with the Test of Language Development (TOLD, which has three forms: the TOLD-2 Primary, normed for children aged 4 to 8 years; the TOLD-2 Intermediate, normed for children aged 8 to 12 years; and the TOAL, normed for adolescents aged 12 to 18 years (63). Each form of the TOLD-2 consists of a series of subtests through which it assesses both vocabulary and grammar. We administered the TOLD Primary to 27%, the TOLD Intermediate to 47%, and the TOAL to 27% of the children in the study. Spoken language quotient (SLQ) derived from each of these tests was used as an independent variable in the study's data analysis. Language data were missing in 15 CPS patients because of problems ascertaining ceiling (n = 1) and basal levels (n = 9), as well as errors in test administration (n = 5).

Data analysis

To compare cognition and SLQ in the CPS and normal groups, we computed analyses of variance (ANOVAs) with demographic (i.e., age, gender, socioeconomic status, ethnicity) and perinatal variables (i.e., delivery problems, pregnancy problems) in the model. Logistic regression was used to compare the two groups on the presence of a psychiatric diagnosis and CBCL scores in the clinical range. Demographic, perinatal, IQ, and SLQ variables were included in the logistic regression. To compare the two groups on their mean CBCL scores, analyses of covariance (ANCOVAs) with demographic, perinatal, IQ, and SLQ variables in the model were estimated. All tests were two-tailed, and a significance level of 0.05 was adopted. Because all our analyses were hypothesis driven, we did not correct for multiple tests.

A principal-components analysis of seizure variables within the CPS group revealed four components with the following loadings: a duration (0.90)/onset (–0.88) component; an EEG component localization (0.81), lateralization (0.77), severity (.53); a prolonged seizures (0.90)/febrile convulsions (0.77) component; and a seizure frequency (0.79)/number of AEDs (0.79) component. In the EEG component, localization of epileptic activity was categorized as frontal, temporal, frontotemporal, or other, and lateralization as left, right, or bilateral focal epileptic activity. The EEG severity rating included presence or absence of spikes, sharp waves, secondary generalization, and slowing. The number of AEDs was subdivided into no AEDs, monotherapy, and polytherapy.

We used the following model-building strategy to determine which of the demographic, perinatal variables, and seizure components were predictive of the outcome measures of interest (psychopathology, IQ, SLQ) in the CPS group. We first included all these 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 after 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 changed markedly in magnitude when other variables were excluded. This process of deleting, refitting, and verifying was performed until we obtained a final model that explained the data. For IQ and SLQ scores, the seizure components, demographic variables, and perinatal factors were included in the model, whereas for psychopathology, all of the above as well as IQ and SLQ scores were used as the starting point.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Between-group differences

The CPS group had significantly lower Full-Scale IQ, Verbal IQ, Performance IQ, and SLQ scores than the normal group, while controlling for demographic (i.e., age, gender, socioeconomic status, ethnicity) and perinatal variables (Table 3). Across both the CPS and normal groups, children with lower socioeconomic status (p < 0.02) and nonwhite children (p < 0.005) had significantly lower Full-Scale, Verbal, and Performance IQ scores, as well as SLQ scores than did children with higher socioeconomic status and white children.

Table 3. Cognition,a language,a and psychopathologyb by group
  1. NA, Statistical analysis not meaningful for type of diagnosis because of small sample size.

  2. aAge, gender, socioeconomic status, ethnicity, and perinatal variables in model.

  3. bAge, gender, socioeconomic status, ethnicity, perinatal, IQ, and language in model.

Cognition, mean (SD)
 Full-Scale IQ93 (18.32)111 (13.11)83.951, 1970.0001
 Verbal IQ93 (18.81)111 (15.23)73.641, 1970.0001
 Performance IQ94 (18.09)109 (11.90)52.031, 1970.0001
Language, mean (SD)
 SLQ90 (18.45)104 (14.54)41.701, 1740.0001
 Psychiatric diagnosis56%13%15.0910.001 
 Disruptive17% 6%NA 
 Affective/Anxiety14% 3%NA 
 Comorbid23% 3%NA 
 Other 2% 1%NA 
CBCL, mean (SD)
 Total57.4 (11.10)   47 (12.04)17.051, 1870.0001
 Internalizing57.5 (11.10)  49.1 (11.45) 15.501, 1870.0001
 Externalizing51 (11.57)46.6 (10.16)  2.821, 187NS
CBCL T >65

The CPS and normal groups also differed significantly in the number of children with a psychiatric diagnosis and total and internalizing CBCL scores in the clinical range, as well in the mean total and internalizing CBCL scores when demographic, cognitive, SLQ, and perinatal variables were in the model (see Table 3). No significant changes were found in the rate and distribution of psychiatric diagnoses or in the mean IQ and SLQ scores in the children tested in 1994 through 1998 compared with those tested in 1999 through 2003 (Table 4).

Table 4. Psychiatric diagnosis, IQ, and SLQ in CPS and normal subjects tested in 1994–1998 and 1999–2003
Psychiatric diagnosis61%48%16%6%
 Full-Scale93 (17.42)92 (19.91)111 (12.84)110 (13.77)
 Verbal92 (18.47)94 (19.56)111 (15.43)112 (15.30)
 Performance95 (17.25)92 (19.42)109 (10.85)106 (13.68)
 SLQ87 (16.75)93 (20. 18)108 (13.87)103 (15.29)

Modeling of psychopathology, cognition, and language in CPS

The psychopathology findings of the 14 CPS patients without EEG data and the 15 patients without language scores were similar to those of the remaining CPS patients. Those without EEG data had more boys [χ2 (1) = 5.49; p < 0.01] and a higher rate of febrile convulsions [χ2 (1) = 5.41; p < 0.02] than did those with EEG data. The children without language data had significantly more delivery problems (χ2= 4.12; p < 0.04), children with prolonged seizures (χ2= 4.44; p < 0.04), and lower mean IQ scores (t93= 2.42–3.03; p < 0.02) than did the patients with language data.


Modeling the presence of a psychiatric diagnosis and CBCL scores in the clinical range with demographic, cognitive, SLQ, perinatal, and seizure-related factors as predictor variables in a logistic regression model revealed a significant effect for Verbal IQ (Psychiatric diagnosis: χ2 (1) = 4.33; p < 0.03; Total CBCL: χ2 (1) = 6.11; p < 0.01; Internalizing CBCL: χ2 (1) = 3.76; p < 0.05; Externalizing CBCL: χ2 (1) = 5.09; p < 0.02). The CPS patients with a psychiatric diagnosis and borderline/clinically relevant CBCL scores had lower IQ scores compared with those without a psychiatric diagnosis or normal CBCL scores. It is notable that these findings were consistent with and without seizure factors in the model.

For mean CBCL scores, with seizure-related, demographic, cognitive, SLQ, and perinatal factors in the model, only the effect of Verbal IQ reached significance for the total (F1, 93= 9.71; p < 0.002) and externalizing CBCL scores (F1, 93= 6.85; p < 0.01). Children with higher total and externalizing CBCL scores had lower Verbal IQ scores. None of the demographic, cognitive, linguistic, and seizure variables was found to be a significant predictor when modeling the type of psychiatric diagnosis (i.e., disruptive, affective/anxiety, comorbid diagnoses).


In the model for IQ scores (Table 5, Fig. 1), the effects of prolonged seizures/febrile convulsions, seizure frequency/AEDs, and ethnicity were significant. Thus lower IQ scores in children were associated with more prolonged seizures and febrile convulsions, higher seizure frequency, more AEDs, and being nonwhite.

Table 5. Modeling of IQa and SLQb scores in CPS group
  1. aFSIQ, Full-Scale IQ; VIQ, Verbal IQ; PIQ, Performance IQ.

  2. bSLQ, Spoken language quotient.

 Prolonged/Febrile 7.391, 760.008  5.711, 760.01  8.211, 760.005 
 Frequency/AEDs 7.371, 760.008  5.351, 760.02  3.901, 760.05 9. 541, 640.003
 Ethnicity13.861, 760.000421.951, 760.00015.501, 760.02 9.321, 640.003

Figure 1. Modeling of IQ scores in complex partial seizure disorder: relation with seizure components. FSIQ, Full-Scale IQ; AED, antiepileptic drug.

Download figure to PowerPoint

Seizure components and demographic factors accounted for the variance of the IQ scores as follows: Full-Scale IQ (Seizure, 12%; Demographic, 11%), Verbal IQ (Seizure, 8%; Demographic, 18%), and Performance IQ scores (Seizure, 12%; Demographic, 4%).


For SLQ, the model revealed significant effects of seizure frequency/number of AEDs and ethnicity (see Table 5, Fig. 2). Increased seizures frequency, number of AEDs, and being nonwhite were related to lower SLQ scores. Seizure frequency/AEDs accounted for 10%, and ethnicity, for 10% of the variance of SLQ scores.


Figure 2. Modeling of language scores in complex partial seizure disorder: relation with seizure components. SLQ, Spoken Language Quotient; AED, antiepileptic drug.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgments

This is the first study to examine the role of independent seizure factors, cognition, and language in the behavioral disturbances of a large sample of children with CPS by using both child- and parent-based measures of psychopathology, as well as a control group of normal subjects. Compared with the normal group, significantly more CPS patients had a psychiatric diagnosis and CBCL scores (e.g., total, internalizing) in the clinical range. The patients also had significantly higher mean total and internalizing CBCL scores but lower mean IQ and SLQ scores than did the normal subjects.

Within the CPS group, Verbal IQ predicted the presence of a psychiatric diagnosis and CBCL scores in the borderline/clinical range. Seizure factors were unrelated to measures of psychopathology. They were, however, associated with IQ and SLQ scores. The prolonged seizure/febrile convulsions and the seizure frequency/AEDs components were significantly related to IQ, and the seizure frequency/AEDs component, to SLQ. Although demographic factors accounted for 11 to 18% of the variance of Full-Scale, Verbal IQ, and SLQ scores, they were unrelated to psychopathology. Perinatal factors were not associated with the study's psychopathology, IQ, or SLQ measures.

In contrast to studies that found that seizure variables predict behavioral problems in children with epilepsy (1,4,9,10,16–18), when independent seizure factors, cognition, language, and demographic factors are in the model, Verbal IQ is the most robust predictor of psychopathology in pediatric CPS. Similarly, Mitchell et al. (23) found that seizure variables were unrelated to the 18-month behavioral outcome of children with epilepsy by using structural equation modeling that included seizure, cognition, and demographic variables.

In our study, because seizure components accounted for only 8% of the variance of the Verbal IQ scores, the association between psychopathology measures and Verbal IQ does not simply represent the effect of seizure-related variables. The large number of children with one type of seizure disorder and the exclusion of children with mental retardation could account for the difference in our findings compared with those of others.

Furthermore, as demonstrated in this study, parent-based internalizing CBCL scores in the clinical range were a consistent finding in prior studies of children with epilepsy (12,64–68), unlike externalizing CBCL scores (9). Oostrom et al. (13) found no relation of parent-based CBCL internalizing scores with child-based IQ and seizure-related variables. They suggested that parental concerns, misinterpretation of seizure-related behaviors, or other psychosocial factors might underlie this finding (13,67). By demonstrating that seizure, demographic, perinatal, and linguistic variables were unrelated to the internalizing CBCL scores, our findings highlight the importance of determining how psychosocial factors, such as parental depression, coping difficulties, overinterpretation of behaviors as seizure-related phenomena, lack of community support, marital strife, and sociocultural factors (2,3,23,67,69) might be related to psychopathology in children with CPS.

In addition to psychosocial factors, the underlying neuropathology also might contribute to the psychopathology of children with CPSs. A complex relation exists between the underlying neuropathology, the impact of seizures on the developing brain, the development of cognition, and the severity of epilepsy in children (70). The association between verbal IQ and psychopathology might, therefore, reflect the fact that both Verbal IQ and psychopathology are determined by the underlying pathology.

Evidence from recent animal studies suggests that prolonged seizures/febrile convulsions and their impact on cognition might represent the effects of the underlying pathology. In an animal model of febrile seizures in very immature animals (postnatal day 10, P10), Bender et al. (48) demonstrated hippocampal circuit rearrangement even in the absence of cell loss or neurogenesis. Weanling rat pups (P21) subjected to the lithium–pilocarpine model of status epilepticus demonstrated hippocampal and extrahippocampal cell loss, reactive neurogenesis, and mossy fiber sprouting accompanying epilepsy (49–51). Roch et al (52) confirmed lesions by MRI and found an almost identical incidence of epilepsy in P21 rats subjected to lithium–pilocarpine seizures. Cilio et al. (71) found that such injury and reorganization resulting from seizures in immature animals had a cognitive consequence when tested in the Morris water maze.

The findings of two recent clinical studies also point to a possible role for the underlying pathology. In children with new-onset epilepsy, Austin et al. (60) found behavioral disturbances before the onset of their seizures. In an MRI study of adults with CPS, Hermann et al. (72) demonstrated an association between reduced white matter volume in the frontal, temporal, and parietal lobes and the severity of neuropsychological deficits, and between increased amygdala volume and depression (73). In an ongoing MRI study we are investigating how the underlying neuropathology is related to psychopathology, Verbal IQ, and the independent seizure components in children with CPS.

Our current findings have five important clinical implications. First, the robust association of Verbal IQ with psychopathology, together with the reported relation between total CBCL scores and overall intelligence in CPS (9), and between emotional maladaptation and school performance in children with epilepsy (33) underscore the importance of ruling out subtle learning disorders in CPS children who have behavioral disturbances. Second, in light of evidence for a high rate of unmet mental health need in children with CPS and childhood absence epilepsy (74), this finding also implies the need for behavioral assessments in CPS children with subtle verbal IQ deficits.

Third, unlike the general population of children (36,37), psychopathology in pediatric CPS is unrelated to language deficits. However, the association of linguistic deficits with seizure frequency/number of AEDs and ethnicity further emphasizes the need for language assessment in CPS children with normal intelligence whose seizures are poorly controlled, who receive AED polytherapy, and who are minorities.

Fourth, similar to CPS in adults (72), childhood CPS affects broad areas of cognitive function. Seizure frequency, number of AEDs, prolonged seizures, and febrile convulsions contribute to the variance of IQ in pediatric CPS. These findings emphasize the need for prevention of prolonged seizures and febrile convulsions, as well as prompt intervention for ongoing seizures with fewer rather than more AEDs.

In terms of the study's limitations, the seizure-related information collected primarily from the parents might be subject to memory bias and subjectivity. Although we used different IQ instruments (i.e., WISC-R, WISC-III) on the children in the study, we found no significant differences in the mean IQ scores of the children tested with the WISC-R and the WISC-III, respectively.

In addition, computing SLQ scores from three different age-related forms of the language instrument—the TOLD primary, the TOAL Intermediate, and the TOAL—does not rule out undiagnosed linguistic deficits in the CPS and normal groups. Although missing language data in 15 patients did not affect the study's psychopathology findings, they might have affected the SLQ modeling findings. Significantly lower IQ scores and a higher rate of a history of prolonged seizures/febrile convulsions in this subgroup of CPS patients suggest that prolonged seizures/febrile convulsions rather than only seizure frequency and number of AEDs, as demonstrated in the study, might affect both cognitive and linguistic competence in children with CPS. Similarly, missing EEG data in 14 subjects does not rule out possible EEG effects on the study's measures.

With these limitations in mind, from the methodologic perspective, the study's findings underscore the need to control for the effects of cognitive, linguistic, and demographic variables when comparing psychopathology in children with CPS to that in normal children. From the clinical perspective, they demonstrate that subtle Verbal IQ deficits increase the vulnerability of these patients for behavioral disturbances and that multiple aspects of CPS adversely affect cognition and language in children. The study's findings highlight the importance of evaluating behavior, cognition, and language, as well as the negative impact of prolonged seizures, febrile seizures, seizure frequency, and AED polytherapy on cognition and language in pediatric CPS.


  1. Top of page
  2. Abstract
  6. Acknowledgments

Acknowledgment:  This study was supported by grant NS32070 (R.C.). We thank Ken Zaucha, Ph.D., Robert Asarnow, Ph.D., and Susan Curtis, Ph.D., for assistance in data acquisition. We appreciate the technical assistance of Amy Mo, Alexander Kaminski, Kimberley Smith, Jaclyn Sagun, Narod Simciyan, Lorrie Shiota, Shawn Zink, R.N., Natasha Wheeler, D. Psy, and Steven Huizar.


  1. Top of page
  2. Abstract
  6. Acknowledgments