Address correspondence and reprint requests to Dr. K. J. Oostrom at Department of Child Neurology, Division of Neuropsychology (Hp. KG 01.327.1), University Medical Centre, Wilhelmina Children's Hospital, P.O. Box 85090, 3508 AB Utrecht, The Netherlands. E-mail: K.Oostrom@WKZ.AZU.nl
Summary: Purpose: To investigate relevant adults' perceptions of behavioral problems in 66 children with newly diagnosed “epilepsy only” and in 63 healthy gender-matched classmates.
Methods: Parents' and teachers' perceptions of the children's behavioral problems were quantified by using the Total Problem score of the Child Behavior Checklist (CBCL) and the Teacher's Report Form (TRF), after correction for epilepsy-related item ambiguity. Questionnaires were filled out immediately after diagnosis and 3 and 12 months later. Relations with demographic and educational variables, school attitudes, and interview-derived prior adversities were analyzed.
Results: As a group, children with cryptogenic rather than those with idiopathic epilepsy have more behavioral problems than do healthy classmates. Family troubles and long-standing behavioral and learning problems are associated with more behavioral problems. The child's adaptation to the adversity of epilepsy onset is important. No adverse effect of antiepileptic drug (AED) treatment was found. Although the percentages of patients with clinically relevant (mean of controls + 2 SD) behavioral problems are consistently 25% (parents) and 22% (teachers), at each assessment, different children contribute to these percentages. In not a single child did parents and teachers agree on the presence of clinically relevant behavioral problems.
Conclusions: (a) Behavioral problems are common in “epilepsy only,” but are not persistent. (b) Agreement between parent's and teacher's perceptions of behavior is low. (c) Behavioral problems are perceived to occur already in the earliest stage of the disease.
Children with epilepsy have a higher incidence of behavioral problems than have healthy children. This seems to hold even for schoolchildren with epilepsy of an idiopathic or cryptogenic etiology who have no other chronic illness, who go to mainstream schools, live at home, and visit the neurologist, a pediatrician, or a family physician only on an outpatient basis. We refer to this specific subset of children with epilepsy as children with “epilepsy only”(1).
The purpose of our study was to assess clinically relevant behavioral problems in children with newly diagnosed “epilepsy only.” This segment of the epilepsy population has a relatively favorable medical prognosis. Seventy percent of schoolchildren with “epilepsy only” who are treated with antiepileptic drugs (AEDs) become free of seizures; and two thirds of these children remain so after discontinuation of AEDs (2,3). In the past decade, numerous studies have reported behavioral problems in children with “epilepsy only”(4–18). However, the presence of behavioral problems is in itself not enough to justify medical attention or health care. Nearly all children have behavioral problems in some period of life. For clinical relevance, problems must be definable as inordinate (e.g., in degree or duration). Unfortunately, most studies in “epilepsy only” fail to consider the clinical relevance of behavioral problems and do not go beyond group-wise comparisons between children with epilepsy and healthy controls or children with other chronic illnesses.
Only a few authors define the clinical relevance of the problems by means of the criteria offered in the manual of the assessment instrument. Austin et al. (18) used the Child Behavior Checklist (19) and found behavioral problems to reach criteria for psychopathology in 25% of children with newly diagnosed epilepsy, and in 18% of nearest-to-age healthy siblings. Behavioral problems in the at-risk range were found to be present in 32% of the children with newly diagnosed epilepsy, in 23% of siblings, and in 32% of children with asthma (16). When interpreting these figures, one is confronted with the fact that the report of Austin et al. on etiology was not clear. Although it may be assumed that the majority of the children studied by these authors belong to the segment of “epilepsy only,” it cannot be excluded that an unknown proportion have less benign epilepsy conditions.
Another source of prevalence estimations was the structured psychiatric interview leading to diagnoses based on DSM-IV criteria (20). Caplan et al. (10) used this procedure and reported clinically relevant behavioral problems to be present in 55% of children with longer existing absence epilepsy and in 63% of children with complex partial seizures, compared with 18% in a group of healthy matched controls. Of children with other chronic illnesses (i.e., cancer, cystic fibrosis, inflammatory bowel disease, or diabetes), Canning (21) reported that 47% had DSM-IV–based behavioral pathology.
In the general population, the prevalence rate of clinically relevant behavioral problems amounts to 19%(22), when estimated by the CBCL and the parallel Teachers Report Form (TRF) (23), and 18% based on a psychiatric interview (21). As far as the course of behavioral problems is concerned, in the past decade, five studies applied the required longitudinal design (6,8,15–17), two of which started at diagnosis. Stores et al. (6) used Conners' scales (24) to study behavioral effects of AEDs. Unfortunately they did not show any empiric data to illustrate the results, but reported many significantly worse behaviors in children with epilepsy than in healthy classmates to exist before treatment. These differences had disappeared after 12 months of treatment. In contrast, Mandelbaum and Burack (8) found that, at diagnosis, behavioral patterns in children with “epilepsy only” were normal. Yet, over time, behavioral ratings became “somewhat higher” than those in the general population. Unfortunately, large sample attrition (72%) and a lack of repeated-measures analysis confound the significance of these findings. Because of similar methodologic shortcomings, results of a longitudinal controlled study of both children with new-onset and with established epilepsy that reported no behavioral differences between the two groups are of uncertain significance (17).
The nature of behavioral problems is variously characterized. Internalizing problems (8,17), externalizing problems (17), schizoid and uncommunicative behavior (11), disruptive behavior (10), and problems in different aspects of attention (6,13) have been mentioned. It remains unclear whether the spectrum of reported behavioral problems reflects either the true diversity of behavioral problems in patients with “epilepsy only” or semantic confusion, resulting from a lack of one generally accepted nomenclature (25).
It becomes increasingly clear that behavioral vulnerability in childhood epilepsy cannot be understood solely from illness parameters. Neither on their own nor in combination do seizure type (6,8), seizure frequency (5,6), seizure control (6,12), seizure severity (16), or adverse medication effects (15) explain behavioral problems in “epilepsy only.” Rather, factors not intrinsically related to illness might pertain to the risk of behavioral problems. Parents' yielding to their children's fear or anxiety is associated with a negative behavioral outcome (26). Loss of control is a feature of epilepsy that may have negative effects on feelings of responsibility and command, which may explain behavioral problems (27). In a similar vein, both patients and healthy children perceive epilepsy as particularly shame provoking (28). Recurrently experiencing shame is associated with psychopathology later in life (29,30). Austin et al. (5,16) found other contextual factors such as family stress, family mastery, external social support, and a child's attitudes to be predictors of psychopathology. Hodes et al. (31) cited negatively altered family relationships for the increased rates of psychopathology in children with epilepsy. If nonillness variables are causally related to the behavioral problems, situation-relatedness of the problems becomes more conceivable (6,17). The present study was designed with these various issues in mind. We contribute by reporting on a group of consecutively included children with idiopathic or cryptogenic epilepsy (32) and fulfilling the criteria of “epilepsy only”(1). To enhance knowledge on the course of behavioral problems, we started data gathering before initiation of AEDs, reassessed after 3 and 12 months, compared treated and untreated children, and controlled for normal development and received education by concurrent assessment of healthy classmates. We took up the matter of clinical relevance of behavioral problems but refrained from reporting on the nature of problems and stuck to the more global Total Problems scale after correcting for ambiguity (33). We analyzed children's demographic and educational data, school-related attitudes, and context-related and epilepsy variables as possibly explanations for both parents' and teachers' ratings of behavior.
Specific research questions were as follows:
1Do parents and teachers rate behavioral problems higher in children with “epilepsy only” than in healthy classmates?
2What percentage of children with “epilepsy only” compared with what percentage of healthy children is rated to have clinically relevant behavioral problems?
3What is the course of behavioral problems in the year after diagnosis?
4What is the relation of epilepsy-related, attitudinal, and context-related variables to behavioral problems in childhood “epilepsy only”?
The study was part of a multicenter epilepsy project of the Dutch Study of Epilepsy in Childhood (DuSECh). All children participated on the basis of written informed consent from their parents, and, if 12 years and older, on their own agreement. The study was approved by the Ethics and Research Committees of the participating hospitals.
The sample under study consisted of 71 children from 10 hospitals who were consecutively included immediately after epilepsy was diagnosed (Table 1). Inclusion took place between January 1997 and October 1998. Inclusion criteria were two or more unprovoked seizures; an idiopathic, cryptogenic,1 or unclassifiable nonsymptomatic etiology [diagnosed by the child's neurologist according to the standards set forth by the International League Against Epilepsy (ILAE) (32)]; age between 5 and 16 years; and receiving normal education. Exclusion criteria were any associated neurologic disorder (by history, physical examination, or neuroimaging); other chronic illness (e.g., diabetes mellitus or asthma), or previous use of AEDs. One year after diagnosis, two senior neurologists (A.C.B.P., Cv. Donselaar) double-checked the epilepsy classifications in retrospect and excluded two children: one because of the simultaneous presence of a different chronic illness of which no one was initially aware, and one because the diagnosis was revised to nonepileptic psychogenic seizures. All 69 children with epilepsy participated in all three examinations.
Table 1. Age (yr), intellectual levels, and treatment (yes), according to ILAE epilepsy syndromes classificationa
Effects of retesting and of normal development were controlled for by comparing the behavioral scores with those of a control group, comprising 66 healthy classmates matched to the children with epilepsy for age, ethnicity, and gender. They were asked to cooperate by the patients and their parents. Three of the children with epilepsy could not find a suitable classmate within the short time between diagnosis and start of AED treatment. One control child refused to participate in the second and third examinations. Two other control children did not participate in the third examination.
Six children with epilepsy and four of their control children were not Dutch by birth. Table 1 shows age, gender, and intellectual levels of both patients and controls. Intelligence quotients (IQs) were derived from the computerized version of the Coloured Progressive Matrices (35) in children younger than 11 years and the Standard Progressive Matrices (36) in children older than 11 years (37,38).
Children's demographic and educational data
Age, gender, intellectual level, and whether the child had repeated a year.
Etiology and seizure type were classified by the children's neurologists according to the DuSECh-protocol based on the classification set forth by the Commission on Classification and Terminology of the ILAE (39). Data on AED treatment and on seizure remission of ≥6 months when examined 12 months after diagnosis were provided according to a fixed protocol by the children's neurologists. Epilepsy features are presented in Table 1. The time span between first recognized seizure and diagnosis varied from <48 h to 8 years (median, 38.5 days).
Attitudes of the children
At each assessment, children aged between 9 and 16 years rated their motivational attitude toward school, social and emotional well-being at school, and self-concept with regard to own academic and social capacities on the School Attitude Questionnaire (SAQ) (40), which consists of 80 items that are rated by using 3-point scales (0, not true; 1, do not know; 2, true). The scale has sufficient reliability and validity (41). Stratified standardized scores are available for girls and boys. The SAQ provides five composite scales:
2Well-being (enjoyment at school, social acceptance, relationship with teachers),
3Self-concept (expressive skills, self-confidence in examinations, social skills),
4Total (motivation, well-being, self-concept), and
To obtain the percentage of children with clinically relevant negative attitudes, the numbers of children with stanine scores in the “extremely strong,”“very strong,” and “strong” negative attitude range were calculated for the three composite scales, Motivation, Well-being, and Self-concept.
Within 48 h after diagnosis, the parents of the children with epilepsy underwent a semistructured interview focusing on concerns (a) over the period well before seizures set in or were recognized as such, and (b) over the current period up until diagnosis and neuropsychological assessment. The five major topics were as follows:
Parents' perceptions of continuity in parenting when facing adversity caused by the onset of epilepsy;
Parents' perceptions of the child's adaptation to its changed situation;
Presence or absence of long-standing behavioral problems (i.e., problems perceived as already present before the first signs of epilepsy);
Presence or absence of long-standing learning problems (i.e., problems perceived as already present before the first signs of epilepsy); and
Presence or absence of family troubles, such as marital distress, divorce, psychopathology in another member of the family.
The interviews were transcribed and processed according to Chi (42), and described in more detail elsewhere (43).
Child Behavior Checklist (CBCL)
The Dutch version of the CBCL (44) was completed by parents at the time of the neuropsychological assessment.
Teacher's Report Form (TRF)
After parents had given their permission (for patients, n = 69; for controls, n = 65), teachers were contacted by telephone. All teachers appeared to have been informed on the epilepsy of the child and on that basis, the teachers' cooperation was requested. The Dutch version of the TRF (45) was sent to and was completed by teachers, who returned it by using a stamped reply envelope.
Because of ambiguity of the phrasing of some items of the CBCL and TRF, raters run the risk of confusing symptoms of epilepsy with those of behavioral problems (33,46). Therefore we controlled for item ambiguity by treating ambiguous items in the CBCL and TRF as “missing values”(33). After repeated rating, the questionnaires were processed according to the manual, by using the computer program provided (47). Another adaptation was that at pretreatment baseline, parents and teachers were instructed to rate the child's current behavior (i.e., over the 4 weeks preceding the diagnosis until today). Parents and teachers of controls were instructed to rate the child's behavior over the same period (i.e.,the past 4 weeks). At 3 and 12 months, parents and teachers of patients and controls were asked to rate their child's behavior from the time of prior assessment to avoid overlap. The standard CBCL form specifies the past 6 months as the period to be covered, which at the second assessment would have been inappropriate.
Raw scores rather than T scores were used for data analysis to maximize differences within the normal range (46), and only the first-order Total Problems scale was studied, as urged by recent concerns about the construction of the CBCL and TRF scales (25).
As a result of these methodologic issues, clinical criteria as set forth by the Dutch manual (44,45) were no longer relevant. We set the criterion for clinical relevance on z-score ≥2 (Patient's Total Problems raw score ≥ mean of the control sample + 2 SD of the control sample).
Data management and statistical analysis
Data were analyzed in the statistical package S-PLUS (2000). Linear mixed effects (LME) analyses of variance (ANOVAs) were used to answer research questions 1, 3, and 4. Individuals were modeled by a random cluster variable to account for dependency of repeated measurements, thus making it possible to include children with incomplete CBCL and TRF data. The effects of child's demographics, education, attitudes, and epilepsy features on CBCL and TRF were analyzed with each of the explanatory factors as independent variable. Because of wide scatter, data on seizure frequency and time since first recognized seizure resisted statistical analysis. In each analysis, “time after diagnosis” was entered as an additional independent variable. In the SAQ ratings, no statistically significant change over time being found, the ratings at diagnosis were used as independent variables. Where appropriate, significant effects were further analyzed by using pair-wise comparisons. Context variables were analyzed as possible explanatory variables within the sample of children with epilepsy. Heterogeneity of variance in both the CBCL and TRF outcome measures was corrected for by applying a square-root transformation to the original values. Because analyses using LME ANOVAs showed a significant relation between IQ and TRF, IQ was included as a covariate in all analyses involving the TRF. Differences in frequency distributions (research question 2) between patients and controls were assessed by using nonparametric tests.
All tests were performed with a significance level of alpha = 0.05, two-sided. If a p value was between 0.05 and 0.10, a tendency to statistical significance was presumed and will be reported.
The small group of three patients, two girls and one boy (mean age, 9.7 ± 0.8 years), with unclassifiable etiology had a considerable impact on several relations. After follow-up, it was decided they had been included incorrectly. All three had nonepileptic, psychogenic seizures, as concluded after repeated EEG investigations, some with simultaneous video registration of behavior. These three children functioned at the lower end of the IQ distribution, and they were perceived by their parents as having severe long-standing behavioral problems. The mean CBCL Total Problems score in these children was 71.38 (±10.24), and the mean TRF Total Problems score was 41.00 (±14.44). The analyses as presented here were carried out after removal of the three children and their controls, which reduced the samples to 66 children with epilepsy and 63 classmates. Table 2 shows the number of returned questionnaires together with the behavioral outcome measures. With regard to demographic, epilepsy, or context-derived variables, no statistically significant difference was found between children whose parent(s) or teacher did and children whose parent(s) or teacher did not return the questionnaire.
Table 2. Summary of empiric data for the CBCL and TRF (raw scores) per assessment
Children with “epilepsy only” and healthy control children were found not to be statistically significantly different as far as age or IQ was concerned. At intake, the mean age of the patient sample was 9.16 years [95% confidence interval (CI), 8.45–9.87; SD, 2.81) and in controls 9.15 years (95% CI, 8.48–9.82; SD, 2.71). The mean IQ of the patients was 99.93 (95% CI, 95.74–104.12; SD, 16.50), and of the controls, 98.50 (95% CI, 94.30–102.70; SD, 16.93). At intake, 14 (21%) of the children with “epilepsy only” and 10 (16%) controls had repeated a year at school. Related to age requirements, repeaters with epilepsy tended to select classmates who had also repeated a year as their controls. Over the year, six more children with epilepsy, including one boy for the second time, and only two controls had to repeat a year.
At diagnosis, 25 patient–control couples fulfilled age requirements to complete the SAQ. This number increased to 26 at the second and 28 at the third assessment. Children who had repeated a year entertained less-favorable attitudes toward school than did children who had never repeated a year (motivation, F = 5.52, p ≤ 0.01; self-concept, F = 6.99, p ≤ 0.05; well-being, F = 11.14, p ≤ 0.05). Children with epilepsy had a worse motivational attitude toward school (F = 6.25, p ≤ 0.05) and felt socially less accepted at school (F = 4.48, p ≤ 0.05) than healthy classmates did. No statistically significant difference in self-concept was found between children with epilepsy and controls. None of the SAQ attitudes changed statistically significantly with time. No statistically significant interaction between medical status and time after diagnosis was found.
Parents' ratings (CBCL Total Problems scale) and relations with self-ratings of the children.
Parents rated children with epilepsy as having more problems than healthy controls (F = 18.56, p ≤ 0.001). More specifically, children with a cryptogenic etiology (F = 8.67, p ≤ 0.001) were rated as having more behavioral problems than were healthy children, whereas those with idiopathic etiology could not be distinguished from controls. A significant main effect of AED use also was found (F = 10.53, p < 0.001): in contrast to children in whom AED treatment was not initiated (n = 17), those who were treated with AEDs were rated as having more behavioral problems than did healthy children. However, this cannot be interpreted as an adverse medication effect, because no interaction was found with time: the elevated CBCL scores of children who were eventually treated already existed at pretreatment baseline. The problem ratings by parents of patients and those by parents of controls decreased with time (F = 5.46, p ≤ 0.001). Differences between children with epilepsy and controls existed at baseline and persisted without significant change, irrespective of epilepsy features or children's attitudes. Parents' ratings of the children's behavior were not related to IQ, age, gender, or whether a child had ever repeated a year.
For the children who had completed the SAQ (n = 25), a statistically significant relation was found between their motivational attitude toward school at diagnosis and parents' ratings of behavior (F = 6.65, p ≤ 0.01). Parents continued to rate more behavioral problems in children with a poor motivational attitude to school. No statistically significant relation was found between the children's social well-being or self-concept and parents' ratings. No statistically significant interaction effect of medical status by any of the SAQ scales was found.
Teachers' ratings (TRF Total Problems scale) and relations with self-ratings of the children.
Teachers rated children with epilepsy as having more behavioral problems on the Total Problems scale than did healthy controls (F = 25.36, p ≤ 0.00). Children with a seizure remission of ≥6 months obtained higher TRF Total Problems scores than did healthy children (F = 2.23, p < 0.05). No statistically significant difference was found between TRF problem scores of children who continued to have seizures and those of controls.
Teachers' ratings showed a statistically significant increase in behavior problems with time (F = 5.46, p ≤ 0.00), both for patients and controls. Differences between children with epilepsy and controls existed at baseline, persisted, and did not change significantly.
Teachers' ratings of the children's behavior were statistically significantly related to IQ (F = 12.19, p ≤ 0.00) but neither to age of the children nor to having repeated a year. Teachers tended to rate more behavioral problems in boys than in girls (F = 2.94, p < 0.10).
A statistically significant relation was found between the children's motivational attitude toward school at diagnosis and the teachers' ratings of behavior (F = 11.14, p ≤ 0.001). No statistically significant relation was found between the children's social well-being or self-concept and the teachers' ratings.
However, statistically significant [medical status × SAQ selfconcept] and [medical status × SAQ motivation] interaction effects were found. Particularly, patients with inferior self-concepts were rated as having more behavioral problems than were patients with normal self-concepts (F = 4.33, p ≤ 0.05). Especially patients with a poor motivation toward school tended to be rated as having more behavioral problems than did patients with a good motivational attitude toward school (F = 3.63, p ≤ 0.10). These relations of poor self-concept and poor motivation with behavioral problems were not significantly present in controls.
Children with “epilepsy only”
No statistically significant relations were found between the epilepsy variables.
Within the group of children who completed the SAQ, none of the epilepsy features was statistically significantly associated with the attitudinal scores. Two relations with context variables were found: Children perceived by their parents as having long-standing behavior problems (F = 6.23, p ≤ 0.05) and children perceived as having long-standing learning problems (F = 4.21, p ≤ 0.05) felt socially worse at school than did other patients. Patients' attitudes toward school did not change with time. No other statistically significant relations among the independent variables were found.
Parents' ratings (CBCL).
Relations of children's demographics, education, attitudes, and epilepsy features with the CBCL Total Problems scores were not statistically significantly different from those in the total sample (data not shown).
Context variables revealed that parents who perceived their child as maladapting to the adversities of the epilepsy onset rated more behavior problems (F = 9.90, p ≤ 0.01) than parents who perceived their child as adapting positively to the onset of epilepsy. Children in troubled families (F = 17.39, p ≤ 0.00), and children perceived as having long-standing behavioral problems (F = 14.65, p ≤ 0.00) were rated as having statistically significantly more behavioral problems than did children in nontroubled families and than did children perceived as not having long-standing behavioral problems. Neither parents' own perception of discontinuation of parenting, nor their perception of long-standing learning problems in their child was related to ratings of behavior problems in their child. No statistically significant interaction with time after diagnosis was found.
Teachers' ratings (TRF).
Relations of children's demographics, education, and attitudes with the TRF Total Problems scores were not statistically significantly different from those in the total sample (data not shown).
Immediately after diagnosis, teachers' ratings in children who eventually became seizure free and in children who did not become seizure free coincided. After 3 months, the teachers rated the problems of the children who eventually became seizure free as slightly more severe and those in the children who did not become seizure free as less severe than before. After 1 year, the ratings coincided again (F = 7.84, p ≤ 0.05). None of the context variables was statistically significantly related to TRF-rated behavioral problems in children with epilepsy.
Summated over questionnaires and assessments, 41 patients and nine control children fulfilled the criteria for clinical relevance of the behavioral problems, at either one or more assessments in the year after diagnosis; 18 patients and 15 control children fulfilled the criteria for clinical relevance of negative attitudes. Consistency, or rather inconsistency of clinically relevant abnormality over measures as well as over time is depicted in Fig. 1.
Judged upon the ratings by parents (CBCL), immediately after diagnosis 16 (25% of 65 returned questionnaires) children with epilepsy and three healthy classmates (6% of 54 returned questionnaires) had clinically relevant behavioral problems [≥(mean + 2 SD) control sample]. Three months later, these percentages were 21% in patients and 6% in healthy controls. One year after diagnosis, 22% of patients and 4% of controls had clinically relevant behavioral problems
Judged from the ratings by teachers (TRF), immediately after diagnosis 13 (22% of 60 returned questionnaires) children with epilepsy and two (3%) healthy controls had clinically relevant behavioral problems. Three months later, these percentages were 30% in patients and 6% in healthy controls, and 1 year after diagnosis, 19% in patients and 1% in controls.
Self-ratings of the children revealed the following picture. Immediately after diagnosis, seven children with epilepsy and three healthy children had clinically relevant negative motivational attitudes toward school. Three and 12 months later, these figures were five and seven in patients and three and three in healthy controls. Immediately after diagnosis, six children with epilepsy and six healthy children had clinically relevant negative feelings of social well-being. Three and 12 months later, these figures were four and five in patients and four and two in healthy controls. Immediately after diagnosis, five children with epilepsy and six healthy controls had a clinically relevant negative self-concept. Three and 12 months later, these figures were five and two in patients and five and three in healthy children.
School-children with epilepsy have more behavioral problems than do healthy classmates. Immediately after diagnosis, 26% of children with “epilepsy only” have clinically relevant behavioral problems, based on parents' ratings. This percentage fluctuates little over the subsequent year and matches previous findings well. Austin et al. (5,18) found 25% of children to have had clinically relevant behavioral problems before the first recognized seizure, and 27% of children who had had epilepsy for ≥1 year. Based on teachers' ratings, 22% of our patients have clinically relevant behavioral problems at diagnosis. This percentage also does not change significantly over the year. As yet, the literature does not offer percentages of clinically relevant teacher-based behavioral problems with which to compare our findings.
What no earlier study has demonstrated is the nonpersistence of behavioral problems in the majority of children with “epilepsy only.” Group differences between children with epilepsy and healthy controls as well as percentages of children with clinically relevant behavioral problems may change little with time; the composition of the group with clinically relevant behavioral problems does change. All children have, at times, difficult behavior. Because the population-based indicators of severity, as offered by the manuals of CBCL and TRF, could not be used for the present study, we had to abandon these indicators. In agreement with general practice, we defined the clinical relevance of the problems in terms of a cut-off score (i.e., two or more standard deviations beyond the mean of the control group). In this way, the CBCL singled out seven children and the TRF singled out five children as having persistent and clinically relevant problems of behavior. The low percentage of persistently present problems corroborates an earlier suggestion that only “some” children with epilepsy have a behavioral disorder (5,6,8,48). Furthermore, we tried to catch the significance of the problems by determining the coherence between the data obtained from different sources. Not in a single child did parents and teachers agree on persistent clinically relevant behavior problems. It remains difficult to interpret this finding. Either the problems are situation bound rather than general, or the raters' frames of reference differ considerably, or both. Different informants may yield complementary information (49).
The present study underscores the importance of variables not intrinsically related to epilepsy. Preexisting adversity, such as family trouble and long-standing problems with learning or behavior, and difficulty adapting to the upheaval and uncertainty in the period leading to the diagnosis of epilepsy are factors that are related to and perhaps determine the higher ratings in the CBCL of children with recently diagnosed “epilepsy only.”
Furthermore, attitudes of the children themselves appear to be relevant. Patients and classmates who expressed a negative attitude toward school were rated both by parents and by teachers as having more behavioral problems than were children with positive attitudes toward school. Children with “epilepsy only” expressed a more negative attitude toward school than did their healthy classmates. Moreover, children with “epilepsy only” felt less socially accepted at school than their healthy classmates. This finding is in concordance with our earlier work (28). Using the method of structured projection, we showed that epilepsy-related situations are experienced as more shame provoking than are situations caused by other illnesses. Shame is a social emotion that, when experienced recurrently, leads to a proneness to shame, which is associated with a self-concept of an “unwanted identity.” The present finding of a normal self-concept supports the view that children with newly diagnosed “epilepsy only” have not (yet) developed a proneness to shame and underscores the necessity of timely counseling to prevent proneness, which is known to be an important precursor of depression (29,30,50).
In short, nonillness adversity appears to contribute significantly to the risk of being perceived as a behaviorally difficult child with epilepsy. The effect of epilepsy-related variables cannot, however, be ruled out. CBCL-based ratings indicate that children with idiopathic epilepsy, and among them particularly those with benign childhood epilepsy with centrotemporal spikes (BCECTS), have fewer behavioral problems than do children with cryptogenic epilepsy.
Two enigmatic findings must be discussed. First, children using AEDs were rated by parents as well as by teachers as having more behavioral problems than did children in whom AED treatment was not initiated. This finding should not lead to erroneous conclusions regarding deleterious effects of AEDs, as the difference of behavior between treated and untreated children was already present at pretreatment baseline. Rather, one might suggest that having more behavioral problems is related to the underlying epilepsy process causing symptoms that compel to initiation of AED treatment. Otherwise, the suggestion that children with behavioral problems are simply more likely to be treated cannot be rejected on the basis of the present data. A third explanation would be, as earlier noted, that dividing the patient group into treated and nontreated children results in dividing the group according to seizure type (8,51). The second enigmatic finding was a slightly diverging and later converging course of the problems as rated by teachers in children with terminal remission (n = 38) and in those who did not become seizure free (n = 28). The statistical difference between the groups is hard to interpret and requires longer follow-up, under way in our department. Nevertheless, it is clear that children with ongoing seizures do not necessarily have more behavioral problems than do children in whom remission is achieved.
Three methodologic shortcomings of the present study must be mentioned. First, the lack of valid illness-specific behavior instruments forced us to use questionnaires that were not perfectly suited to the population and queries under study. As there is insufficient evidence to support the syndrome scales of the questionnaires CBCL and TRF, we followed the advice of Hartman et al. (25) and abandoned syndromal analysis. Rather, the Total Problems scale, being judged to meet minimal validity requirements (25), was used to make a global inventory of behavioral problems. A second limitation is that we were not able to gain standardized information on the younger children's attitudes and self-concept. Furthermore, not unlike the majority of similar studies, the one presented here would have benefited from larger patient and control groups. The small group sizes impeded adequate analysis of possible effects of separate epilepsy variables. Larger samples are needed for a proper evaluation of mediating and/or moderating effects of illness and contextual variables, as has been shown for other illnesses (52). Nevertheless, our standard protocol and double-checking led to a careful composition of the patient groups consisting of children with newly diagnosed “epilepsy only.”
In conclusion, this 1 year follow-up provides clear evidence for the fact that relevant adults perceive more behavioral problems in children with “epilepsy only” than in healthy classmates. As a group, about one fourth of children with epilepsy are rated as having clinically relevant behavioral problems. Context problems offer reasonable explanations for the behavioral problems in newly diagnosed “epilepsy only” patients. Even so, that children with cryptogenic epilepsy have more behavioral problems than children with idiopathic epilepsy leaves the possibility open that cerebral dysfunction contributes to behavioral problems. However, in the case of cryptogenic epilepsy, the announcement of necessary (imaging) examinations, not necessary in the case of idiopathic epilepsy, may cause additional emotional upheaval leading to behavioral problems. Nonpersistence of these problems and noncoherence between instruments are the main messages of our study.
Acknowledgment: We are most grateful to the children with epilepsy and to their classmates who participated in our study. We also thank their parents and their teachers.
The study was subsidized by the Dutch Epilepsy Foundation (NEF), JANIVO Foundation, and Peugeot Holland NV. We thank Prof. Dr. F. Jennekens for his critical comments and Mrs. B. Vollers-King for language editing.
Participants in the Dutch Study of Epilepsy in Childhood (DuSECh): W. F. M. Arts, J. H. Begeer, O. F. Brouwer, C. A. van Donselaar, A. T. Geerts, E. A. J. Peeters, H. Stroink, G. Hageman, R. ten Houten, J. F. de Rijk-van Andel, L. M. E. Smit, and M. J. Wennekens.
Cryptogenic epilepsy according to ILAE (34): Presumed to be symptomatic, but no aetiology could be identified (including a normal MRI).