Behavior and social competency in idiopathic and cryptogenic childhood epilepsy Anne T Berg* PhD, Department of Biology, Northern Illinois University, IL; Barbara G Vickrey MD MPH, Department of Neurology, University of California Los Angeles, Los Angeles, CA; Francine M Testa MD; Susan R Levy MD, Departments of Pediatrics and Neurology, Yale Medical School, New Haven, CT; Shlomo Shinnar MD PhD, Departments of Neurology and Pediatrics, Montefiore Medical Center, New York, NY; Francis DiMario MD, Departments of Pediatrics and Neurology, Connecticut Children’s Medical Center, Hartford, CT, USA. *Correspondence to first author at Department of Biology, Northern Illinois University, DeKalb, IL 60115, USA. E-mail:
[email protected]
Behavioral and related disorders are frequently reported in association with childhood epilepsy but the reasons for this are unclear. In a long-term prospective, community-based study of newly-diagnosed childhood epilepsy, behavioral assessments (Child Behavior Checklist) were performed in children 8 to 9 years after the initial diagnosis of epilepsy to determine the impact of remission and medication status on behavioral problems. Children with epilepsy were also compared with sibling controls. A total of 226 children (108 females, 118 males; mean age 13y 1mo [SD 2y 8mo], range 8–17y) with idiopathic or cryptogenic epilepsy were included in the analyses. One hundred and twenty-eight matched pairs were included in analyses of case–sibling differences. Lack of remission and current medication use were associated with worse behavioral problem and competency scores. Lack of remission generally had a greater effect than medication use, except for attention problems; medication status had the more deleterious effect (p70 when available, or a clinical or educational assessment) were eligible for these analyses. In addition, cases with cryptogenic forms of one of the epileptic encephalopathies were excluded because they are generally associated with severe impairments17 and our objective was to focus on children with epilepsy without other impairments that can often be associated with behavioral problems. Remission status was classified as being seizure-free for at least 5 years or not at the time of the assessment. Medication status was classified as either taking or not taking an AED at the time of the assessment. The ethics committees of all participating institutions approved all procedures used in this study. Written informed consent was obtained from the parent and written assent from the child, both to participate in the original study and separately to participate in the assessments performed 8 to 9 years later. STATISTICAL ANALYSIS
Primary outcomes were the overall Internalizing and Externalizing CBCL scales. We examined individual (or narrower) scales to identify whether effects observed on the main outcome scales were due to selective effects in a few domains or whether they represented more pervasive influences. In the first set of analyses, we examined factors associated with scores indicating higher levels of problems within the epilepsy group. The primary focus of the analyses was the T score. To determine whether differences in mean T scores resulted in potentially important clinical differences, we also studied the dichotomous clinical indicator for each scale. Analyses were performed using SAS (version 9.1). Within the case group, bivariate analyses of T scores were performed with the use of t-tests and one-way analyses of variance. Multivariable analyses were performed with multiple linear regression. For analysis of the clinical indicators, we used χ2 tests for unpaired control. In matched analyses, we examined differences between cases and matched siblings. We first constructed a case minus
488
Developmental Medicine & Child Neurology 2007, 49: 487–492
sibling difference score. Matched t-tests were used to test case–sibling differences. Multiple linear regression, with the difference score as the outcome, was used to test multiple determinates of case–sibling differences simultaneously. For the clinical indicators, associations were summarized with a matched odds ratio (OR) and statistical significance determined with McNemar’s χ2 or an exact method.18 Confidence intervals (CI) for matched ORs were calculated according to the methods described by Miettinen.19 Results Of the original 613 children in the cohort, 305 (50%) were eligible for this analysis because they had idiopathic or cryptogenic epilepsy and their syndromes did not fall within the category of epileptic encephalopathy. In addition, they were of normal intelligence and less than 18 years old at the time they were included or would have been eligible for inclusion in this assessment. Thirty-four (11%) children were no longer in active follow-up at the time this assessment was performed. Of the 271 who were actively followed-up, 226 (83%) participated in the behavioral assessment with the CBCL. Eligible cases who were actively followed but who did not participate in the CBCL behavioral assessment (n=45) were highly comparable with cases who did participate (n=226) with respect to epilepsy syndrome (p=0.93), sex (p=0.80), probability of being at least 5 years seizure-free (p=0.74), and probability of having intractable epilepsy (p=0.47). There was a trend for participating cases to be younger than non-participating cases by about 1 year at the time of first seizure (p=0.02) and by 9 months at time of diagnosis (p=0.05). Sibling controls were obtained for 128 cases. When controls were not obtained it was usually because one was not available (only child), the only available sibling also had epilepsy, or the control child was too young or too old for the CBCL. Two additional 18-year-old cases with younger sibling controls were also included for CBCL analyses. There were 108 (48%) females and 118 (52%) males. Mean age at time of assessment was 13 years 1 month (SD 2y 8mo, range 8–18y). Of this group, 158 (70%) had been seizure-free for 5 years and 170 (75%) were no longer taking AEDs when they and their parents participated in the assessment. Only six children were taking multiple AEDs. Epilepsy syndrome groups included: idiopathic localization-related (n=39, 17%); cryptogenic focal (n=109, 48%); idiopathic generalized (n=60, 27%); and other (usually unclassified; n=18, 8%). CBCL scores, already adjusted for sex and age in the population, did not vary by these factors in our study group. Syndrome group was not associated with any of the CBCL scores and findings for this factor are not presented or further discussed. Compared with children who who had been seizure-free for 5 years, children who were less than 5 years seizure-free had scores indicative of higher behavioral problem levels for the Total and Externalizing scores, as well as for five of the eight narrow scales and four of the six DSM-oriented scales. They also had lower scores for three of the four competence scales. Being on medication was associated with higher (worse) Total (p=0.02) and Externalizing scores (p=0.02). Of the eight narrower scores, medication use was associated only with higher attention problem scores for both CBCL (p=0.003) and DSMoriented (p=0.01) scales. Medication status was also associated with lower (worse) School (p