Not Only a Matter of Epilepsy: Early Problems of Cognition and Behavior in Children With “Epilepsy Only”—A Prospective, Longitudinal, Controlled Study Starting at Diagnosis
Objective. To understand early educational and behavioral predicament in childhood “epilepsy only.”
Methods. A multicenter, prospective, longitudinal study was conducted of 51 outpatient schoolchildren with newly diagnosed idiopathic or cryptogenic epilepsy and 48 sex-matched classmate control subjects. All children underwent neuropsychological assessment 3 times within the first year after diagnosis; parents and teachers completed behavior questionnaires, and patients’ parents were interviewed to inventory contextual adversity. Principal components analysis of cognition and behavior disclosed 6 major components that were related with the interview data (repeated measures analysis of variance).
Results. Despite similar intelligence and educational background, significantly more patients (51%) than control subjects (27%) required special educational assistance. Patients obtained worse scores across components of cognition and behavior. Parents and teachers perceived patients to have more behavioral problems. Differences between groups existed at pretreatment baseline. Over time, notwithstanding stable percentages of poor scores in both groups, nonpersistence of poor scores was impressive (each time other children scored poorly in other domains). Rather than epilepsy characteristics, contextual adversities were significant risk factors.
Conclusion. Already in the earliest stage of the illness, children with epilepsy are liable to vicissitudes in cognitive and behavioral functioning. Contextual variables are all-important.
Having epilepsy is more than suffering from a purely medical condition. Even when the medical prognosis is favorable, as in “epilepsy only,”1 the patient may still experience psychosocial adversity years after the diagnosis. Children with “epilepsy only” have idiopathic or cryptogenic (according to the International League Against Epilepsy,2 presumed to be symptomatic but no cause could be identified, by magnetic resonance imaging, history, or otherwise) epilepsy; receive mainstream education; visit the pediatrician or (child) neurologist only on an outpatient basis; and are hardly, if at all, restricted in daily life activities. Having followed children with “epilepsy only” well into adulthood, Sillanpää et al3 reported even among patients who had been seizure-free for years an unduly large number of people with only primary education. Presumably related with this was the excessively frequent occurrence of unemployment and low-prestige professions among adults who had at one time had idiopathic or cryptogenic epilepsy. Similar findings were reported in Great Britain4 and the United States.5 All 3 of these large-scale studies were unable to identify causative factors. One of the questions that needs to be addressed is when educational and psychosocial deprivation starts to develop.
To contribute to an explanation, we focused on the earliest stage of the disease. Starting immediately after diagnosis and 3 and 12 months thereafter, we thoroughly studied cognition and behavior of children with “epilepsy only.” The controlled study focused on 2 levels: 1) the factual level of whether children required special educational assistance (SEA) and 2) the possibly explanatory level of illness variables, cognitive and/or behavioral deviations, and contextual adversity. The queries were as follows: 1) Do children with “epilepsy only” and healthy classmate control children differ with respect to educational hard facts, ie, repeating years at school and requiring SEA? 2) Can the groups be distinguished with respect to measures of cognition and/or behavior? 3) What proportion of the children obtain abnormally poor scores, and in which domains of cognition and behavior do they do so? 4) Which epilepsy variables or contextual variables are associated with eventual differences?
This study is a part of the neuropsychological branch of the multicenter Dutch Study of Epilepsy in Childhood (DuSECh),6 a group of closely collaborating Dutch (child) neurologists. It was approved by the ethics and research committees of the participating hospitals.
From January 1997 to November 1998, 69 patients were included in the study. Inclusion criteria were having had at least 2 unprovoked nonfebrile seizures with a time interval of at most 12 months, idiopathic or cryptogenic cause,7 age between 5 and 16 years, and attendance of a mainstream school. Exclusion criteria were having any associated neurologic disorder (identified by history, physical examination, or neuroimaging), having received a diagnosis of having another chronic illness, or previous use of antiepileptic drugs (AEDs). For the purpose of the present study, data obtained from the 15 children who were younger than 7 years were omitted because the Dutch education system is organized in such a way that a majority of children under this age have not yet received formal education on academic skills such as reading, writing to dictation, and arithmetic. After follow-up, it was decided that 3 children with unclassifiable epilepsy had to be removed from the analyses because they had nonepileptic, psychogenic seizures. This decision was based on repeated electroencephalogram investigations, some with simultaneous video registration of behavior. We therefore report on 51 patients (Table 1).
Every child with epilepsy or her or his parents asked a healthy classmate to participate. The healthy control children, matched to the children with epilepsy for age, sex, and educational level, provided the data to control for effects of retesting and of normal development. Three of the 51 children with epilepsy could not find a suitable classmate in time. There was no sample attrition.
The study was prospective, longitudinal, and controlled. Consecutive inclusion of subjects reduced the chance of sampling bias. Because of the requirement of assessing patients within 48 hours after diagnosis in a standard examination setup, a van was converted into a neuropsychological laboratory with 2 examination rooms. Differences in fatigue resulting from differing traveling times to 1 central assessment center were prevented as each child with epilepsy and his or her control child were examined on the compound of the hospital where the child with epilepsy was treated. At diagnosis, indicators of epilepsy were recorded by the children’s neurologists, according to standard protocol, formulated by the DuSECh6 (Table 2).
Teachers provided protocoled information on the children’s school careers. At the first assessment, a well-trained psychologist held a semistructured interview with the patients’ parents,10 focusing on their and their child’s adaptation to the onset of epilepsy and on previous adversities not intrinsically related to epilepsy (Table 2).
The children were assessed shortly (within 48 hours) after having received the diagnosis and before they started AED treatment (if necessary). All children were reassessed 3 and 12 months after the first assessment. The comprehensive neuropsychological assessment covered major domains of cognition, academic skills, and mental and motor speed (Table 2). As many available tools in child neuropsychology are downward extensions of adult psychological tests that are not suitable for children, existing tests were adapted and some new tasks were developed, addressing learning and memory (Word Span and Location Learning12), several aspects of attention (Balloon Piercing13), and behavior regulation (Color Trails13). At every assessment, parents and teachers completed behavior checklists (Table 2). The ratings were corrected for epilepsy-related ambiguity.15,16
Data Management and Statistical Analysis
Analyses were conducted with SPSS 10.0. The data set was completed by imputing missing data (391 of 5252 [7.4%]) by means of regression scores from a missing value analysis.17 Principal component analysis with Varimax rotation was used for data reduction.18 Principal component analysis is a statistical procedure to transform dependent variables into a new set of linear variables in such a way that a maximum of variance is extracted from the data set. The resulting linear combinations of cohering variables are called components. According to the criterion of Kaiser,19 only components with eigenvalues >1 were retained. A solution of 6 normally distributed (Kolmogorov-Smirnov One-Sample Tests) components emerged. The cumulative percentage of explained variance was 70.4. Variables with weights ≥0.4 and ≤−0.4 were considered to contribute saliently to a component (Table 3).18 The components could be adequately characterized by the labels Attention, Reaction Times, Intelligence including Arithmetic, Academic Skills, Location Learning, and Behavior. “Percentage of calculation errors” weighed in the component Academic Skills (0.282) but fulfilled the criterion for salience of the component Intelligence.
The 6 components were used as dependent variables in repeated measures analysis of variance with “time after diagnosis” as the within-subject factor. For multiple comparisons, the Bonferroni correction was applied. When relevant (4 of 6 components), age was a co-variable. Transversal nominal data were studied using Fisher exact test. Longitudinal nominal data were studied using log linear analysis.
Statistical tests were 2-sided, with a 5% significance level. For P values between .05 and .10, a tendency to statistical significance was presumed and reported.
As individual differences were to be expected, in the case-by-case investigation the criterion for “abnormally poor score” was set on 2 standard deviations worse than the mean component score in the control sample. When an abnormally poor score in a specific component persisted throughout 3 assessments, it was qualified as a “deficit.”
At study entrance, 24% (n = 12) of the patients had at some time repeated a year at school, more than usual in The Netherlands (11%)20 but not significantly more than in the control subjects (n = 10) who were matched for age and sex. Moreover, SEA was required for significantly more children with epilepsy (n = 26; 51%) than healthy control subjects (n = 13; 27%; χ2 4.15, df1, P ≤ .05). This difference between patients and control subjects did not statistically significantly change over the year after diagnosis.
Children (patients and control subjects taken together) who required SEA had worse scores in all components of cognition but not in the component Behavior than children who did not require SEA. Sizes of subsamples were too small for reliable statistical analysis, yet comparison of the 4 groups (patients requiring SEA, patients not requiring SEA, control subjects requiring SEA, and control subjects not requiring SEA) suggested differences. Within patients, the difference between those who required SEA (n = 26) and those who did not require SEA (n = 25) pertained to all components of cognition as well as to the component Behavior. Within the control group, the difference between those who required SEA (n = 13) and those who did not require SEA (n = 35) was limited to the component Academic Skills (size of effect [SE]: 13.8, to the advantage of those not requiring SEA).
Components of Cognition and Behavior
Differences Between Patients and Control Subjects
Children with epilepsy obtained statistically significantly worse scores than control subjects in the components Behavior, Attention, Reaction Times, and Location Learning and tended to score worse in Academic Skills (Fig 1). The differences between patients and control subjects could not be ascribed to any of the epilepsy characteristics.
Differences Within the Group of Patients
None of the epilepsy features explained any differences among patients.
Patients who required SEA obtained worse scores on the components Attention (SE: 17.4), Academic Skills (SE: 9.1), and Reaction Times (SE: 8.5) than patients who did not require SEA.
Reactions to the onset of epilepsy.
Patients with parents who had had difficulty continuing their habitual parenting style once epilepsy had set in obtained worse scores in Reaction Times (SE: 4.9), Location Learning (SE: 4.5), and Attention (SE: 4.5) and tended to obtain worse scores in Academic Skills (SE: 3.2) than patients with parents who adapted well to the adversity of epilepsy onset. Patients who, themselves, did not adequately react to the diagnosis obtained worse scores in Behavior than patients who adapted well to the onset of epilepsy (SE: 4.4).
Previous adversities not intrinsically related to epilepsy.
Comparing patients whose parents reported previous adversities to patients without a history of difficulties led to the following findings: Patients with family problems obtained worse scores in Behavior (SE: 7.5) and Location Learning (SE: 4.5) than patients from happy families. Patients with preexisting problems of academic acquisition obtained worse scores in Academic Skills (SE: 5.6), and tended to obtain worse scores in Intelligence and Arithmetic (SE: 3.8) and in Attention (SE: 2.9) than children who progressed normally at school. Patients who had behavioral problems that antedated the first signs and symptoms of epilepsy tended to obtain worse scores in Location Learning (SE: 3.3) and Behavior (SE: 3.0) than children without a history of behavior problems.
Changes over the year after diagnosis were found within the group of patients only: a decline in Academic Skills was found in the group of patients whose parents had complained of problems in academic skill acquisition, dating from before the onset of epilepsy (SE: 4.4), and in the small group of patients who required SEA.
Summated over the year, more patients (n = 23; 45%) than control subjects (n = 11; 23%) had at least 1 abnormally poor score (likelihood ratio χ2 7.25, P ≤ .01). The proportions of patients and control subjects with abnormally poor scores remained stable over time, but the children of whom this proportion was composed changed (Fig 2).
Abnormally poor scores occurred in 1 or a few components rather than in all at the same time. The component Behavior drew more children with abnormally poor scores (14 with epilepsy, 4 control subjects) than any of the other components (likelihood ratio χ2 20.73, P ≤ .00). In the component Behavior, 5 patients qualified for a “deficit,” and in the component Reaction Times, 1 other patient did so. One control child had a deficit in both Attention and Academic Skills. Another control child had a deficit in Location Learning.
The major findings of the present study focusing on early features of the educational and behavioral predicament in childhood “epilepsy only” are 3-fold: First, the school career of children with newly diagnosed “epilepsy only” is already at risk in the very earliest stages of the disease. Notwithstanding average intelligence levels, half of the school children aged 7 years and older needed SEA. Second, as a group, children with “epilepsy only” obtained worse scores in principal components of cognition and behavior than healthy age- and sex-matched control subjects. As the latter were classmates of the children with epilepsy, an explanation by differences in educational background was ruled out. Third, psychosocial context rather than characteristics of the epilepsy were related to the patients’ performances on measures of cognition and behavior. These are findings that need understanding, for, if not remedied, these early educational and behavioral predicaments may end in psychosocial and vocational burden in adulthood.3,4
To estimate the clinical significance of group findings, we looked at the performance of patients and control children individually. On the basis of assumptions in clinical practice, we considered scores worse than the cutoff of 2 standard deviations from the mean score in the control sample to be clinically relevant. This classified 45% of the children with epilepsy, compared with 23% of the healthy children, in the range of clinically relevant problems of cognition and behavior. However, as illustrated by the scattering in Fig 2, clinically relevant poor cognitive performances are persistent in only 1 patient (patient 1, in the component Reaction times) and in 2 control subjects (child 1, in the components Attention and Academic Skills; child 2 in the component Location Learning). The Behavior component, assessed by means of ratings by parents and teachers, yielded the largest difference between the groups. No fewer than 5 patients (10%) were qualified as having a (persistent) behavior “deficit,” whereas none of the control children was qualified as such.
Among children with “epilepsy only,” cause, epilepsy syndrome, use of AED, and course of the epilepsy in the first year after diagnosis were not significantly related to the cognitive or behavioral findings. Contextual variables (listed in Table 2), however, had significant effects on cognition and behavior. Most consistent, having parents who are thrown off balance in the time preceding the diagnosis and who fail to continue their habitual parenting affected the patients’ cognitive and behavioral functioning. However, belonging to a troubled family or having preexisting problems of behavior and of acquiring academic skills was also associated with scoring poorly at tests that assess reaction times, location learning, and attention. Also, children who could not adaptively integrate the adversity of epilepsy onset obtained worse scores in behavior. This is of particular relevance. We reported previously that healthy children and patients alike perceive epilepsy as different from other illnesses in the sense that epilepsy provokes more shame than other illnesses.21 Current views on risk factors for psychopathology hold that recurrently experiencing negative emotions induces a proneness to those emotions and that this proneness distorts cognition and biases action and perception. For shame, it has been demonstrated that shame-proneness is a potent predictor of depression later in life.22 The main message of the present study is that already at diagnosis, children who do not adaptively react to the adversity of epilepsy have an increased risk of negative reactions.
The present study clearly shows that cognitive and behavioral sequelae arise from multiconditional vulnerability rather than from strictly medical aspects of epilepsy and that these problems already become apparent at the very earliest stages of “epilepsy only.” As problems of cognition and behavior may have far-reaching consequences, physicians should be attentive not only to the child’s school history and/or behavioral problems but also to the child’s and parents’ upset as a result of the epilepsy. In case of complaints, help should be called in without delay.
The study was funded by the National Epilepsy Foundation (grant 97-04), the JANIVO Foundation, and Peugeot Holland NV.
Participants in 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, A.C.B. Peters, J.F. de Rijk-van Andel, L.M.E. Smit, and M.J. Wennekes.
We thank all of the patients, classmates, parents, and teachers who took part in the study. We thank Professor F.G.I. Jennekens for critical comment on the article and Marguerite Schinkel, BA, for language editing.
- ↵Engel J Jr. A proposed diagnostic scheme for people with epileptic seizures and with epilepsy: report of the ILAE task force on classification and terminology. Epilepsia.2000;42 :1– 8
- ↵Morgan CLI, Ahmed Z, Kerr MP. Social deprivation and prevalence of epilepsy and associated health usage. J Neurol Neurosurg Psychiatry.2000;69 :13– 17
- ↵Preter M. Social deprivation and prevalence of epilepsy and associated health usage. J Neurol Neurosurg Psychiatry.2000;69 :837
- ↵Raven JC, Court JH, Raven J. Coloured Progressive Matrices (1990 edition). Oxford, UK: Oxford Psychologists Press; 1990
- ↵Raven JC, Court JH, Raven J. Standard Progressive Matrices (1992 edition). Oxford, UK: Oxford Psychologists Press; 1992
- ↵Bruyn EEJ de, Steene G van der, Haassen PP van. Wechsler Intelligence Scale for Children-Revised. Nederlandse Uitgave. Lisse: Swets & Zeitlinger; 1986
- ↵Perrin EC, Stein REK, Drotar D. Cautions in using the Child Behavior Checklist/4–18. J Pediatr Psychol.1991;16 :411– 421
- ↵Little RJA, Rubin DB. Statistical Analysis With Missing Data. New York, NY: Wiley; 1987
- ↵Stevens J. Exploratory and confirmatory factor analysis. In: Stevens J, ed. Applied Multivariate Statistics for the Social Sciences. Mahwah, NJ: Lawrence Erlbaum Associates; 1996:362–428
- ↵Kaiser HF. The application of electronic computers to factor analysis. Educ Psychol Measures.1960;20 :141– 151
- ↵van de Grift W. Hoe kan ik de opbrengst van mijn school bepalen? [English translation: How to assess the output of my school?] School Manag.2000;13 :12– 22
- ↵Tangney JP, Burggraf SA, Wagner PE. Shame-proneness, guilt-proneness, and psychological symptoms. In: Tangney JP, Fischer KW, eds. Self Conscious Emotions: The Psychology of Shame, Guilt, Embarrassment and Pride. New York, NY: The Guilford Press; 1995:175–197
- Copyright © 2003 by the American Academy of Pediatrics