Academic Underachievement in Children with Epilepsy Wendy
G. Mitchell, MD;
John
M. Chavez, PhD;
Hang Lee, MS;
Bianca L. Guzman
Abstract Academic achievement was studied in 78 children with epilepsy, ages 5 to 13 years, to determine how seizures, treatment of seizures, and sociocultural factors influence academic achievement. Cognitive abilities were assessed with either the McCarthy Scales of Children’sAbilities or the Wechsler Intelligence Scale for Children-Revised. Achievement was measured with the Peabody Individual Achievement Tests. Achievement scores were corrected for cognitive ability (IQ), and underachievement was defined as achievement score 1/2 standard deviation or more below IQ. Information regarding seizure history (severity, duration) and treatment with anticonvulsant drugs was obtained. In addition, the family was extensively interviewed regarding the child’s environment, behavior, and demographic background. The Home Observation for Measurement of the Environment (HOME) scale was completed on a home visit. Underachievement was frequent, ranging from 16% (Reading Recognition) to 50% (General Knowledge), but there was no relationship between severity or duration of seizure disorder or total exposure to anticonvulsant medications and achievement. Major determinants of achievement included subscales of the HOME scale, age (older children more likely to be underachieving), and parental education. An equal proportion of newly diagnosed and/or untreated subjects were underachieving compared to those with longstanding epilepsy and anticonvulsant drug treatment. (J Child Neurol
1991;6:65-72).
has been associated and behavioral probwith Children with epilepsy have been noted to have increased frequency of underachievement, reading disabilities, behavior problems, overdependency, and psychiatric illness.6-9 Contributing to the complexity of these problems are interactions involv-
epilepsy long childhood academic, social, lems.l-5
ing neurologic disabilities, parental overprotection, social barriers, and medication effects. In addition, epilepsy disproportionately affects poor, inner-city, and immigrant children, who also may be more prone to academic failure for a variety of other rea-
sons.10
Severity Received
of seizures may be
a
poor
predictor
June 20, 1990. Received revised August 10,
Accepted for publication August
of
1990.
13, 1990.
From the Department of Pediatrics (Dr Mitchell) and Division of Neurology (Drs Mitchell and Chavez and Ms Guzman), Childrens Hospital of Los Angeles, and the Departments of Neurology (Dr Mitchell) and Preventive Medicine, Division of Biometry (Mr Lee), University of Southern California School of Medicine, Los Angeles, CA. The data contained in this paper were presented to the Child Neurology Society, San Antonio, TX, October 1989. Address correspondence to Dr Wendy G. Mitchell, Neurology Division—Box 42, Childrens Hospital of Los Angeles, PO Box 54700, Los Angeles, CA 90054.
academic and whose seizures
psychosocial disability. Children completely controlled or very infrequent perform poorly compared to their nonepileptic peers. Hodgman et al found that adolescents with well-controlled epilepsy had poor self-image, low achievement, and low expectations for the future compared to adolescents with uncontrolled or more severe epilepsy. 11 Hartlage and Greenl2 suggested that environmental factors, rather than specific neurologic disability, are major determinants of are
academic and social
in children with
epilepsy. Seidenberg et a113 studied academic achievement in children with epilepsy. They found that children with epilepsy made poor academic progress. This deficiency persisted after adjustment of academic expectations for IQ differences. Deficiencies were most marked in mathematics, followed by spelling, reading comprehension, and reading recognition. They found that a combination of demographic and clinical seizure variables were moderately predictive of success
academic underachievement in mathematics but not in reading or spelling. Neither the number nor the type of anticonvulsants were predictors of achievement.
65
Concern has been raised that treatment of epimay cause or contribute to academic difficulties. Freeman et al 14 have proposed that children with mild epilepsy not be treated, for fear of causing or exacerbating behavioral and academic difficulties. Although anticonvulsant drugs may influence cognitive function and occasional patients clearly exhibit drug-related impairment, anticonvulsant drug exposure per se has never been causally related to academic underachievement.l5-19 Sociodemographic, family, and environmental factors contribute to academic achievement, both in well and chronically ill children. Caldwell and Bradley2° developed an instrument to evaluate the home environment (The Home Observation for Measurement of the Environment, or HOME, scale). In a diverse and multiethnic group, HOME scores collected at 3 years of age correlated with both IQ and achievement test scores during the elementary years. HOME scores in infancy have moderately strong correlations with achievement at age 7 years, as well as with IQ at age 4 years. Several subscales of the elementary version of the HOME scale correlate with academic achievement at age 10 years in a socioeconomically and racially mixed group of children.21 Thus, we include evaluation of the home environment as an essential component in the understanding of the academic achievement of children with epilepsy.
lepsy
TABLE 1 Tests and Measures Used
66
We studied academic achievement in
a
group of
elementary-school aged children with epilepsy. We particularly asked whether epilepsy-related factors or sociocultural factors are the major determinants of school achievement. We expected that academic underachievement would be excessive in this group. However, we hypothesized that severity, duration, and treatment of the epilepsy would play a minor role in determining academic success, while family setting, parental attitudes, and underlying neurologic disability would play a major role.
Methods Patients aged 5 to 13 years were enrolled in the study at the time of their initial contact with the Children’s Hospital Los Angeles Neurology Service. Children were included if they had a definite diagnosis of epilepsy, regardless of duration. We did not enroll subjects with mental retardation, those with significant motor or sensory handicaps that would interfere with testing, or children who did not speak either English or Spanish. Informed consent was sought at the time of the first clinic visit, if the family indicated that they planned to continue to obtain their child’s neurologic care at Children’s Hospital. Enrollments continued over a period of 3 years. Initial evaluation of the child included psychometric, behavioral, and academic testing. Specific tests used are listed in Table 1. Children under age 7 years were tested with the McCarthy Scales of Children’s Abilities22; over age 7 the Wechsler Intelligence Scale for Children-Revised
(WISC-R) was used. 13 McCarthy General Cognitive Index (GCI) or WISC-R full-scale IQ were used as the cognitive
listed in Table 3. For patients with multiple seizure types, seizure severity score was summed. For some analyses, each epilepsy variable was further condensed: seizure severity scores of 1 or 2 versus 3 and above; seizure duration score of 1 (under 1 month) versus 3 (over 1 year); and medication duration scores of 1 (none or under a month) versus 3 or more (at least 1 year of medication). These scores represent the seizure and treatment history at the time of enrollment in our clinic. Detailed treatment histories, such as precise doses and serum levels, were rarely available.
In a normal population, both have a mean of 100 standard deviation of 15. For Spanish-speaking children, Escela Inteligencia por Ninos Wechsler (EINW)24 or a translated McCarthy Scale&dquo; was used. Academic achievement was measured with the Peabody Individual Achievement Test (PIAT). 26 PIAT Subtests include Reading, Reading Comprehension, Spelling, Mathematics, and General Knowledge. Subjects whose primary classroom instruction was in Spanish were tested in Spanish, using translated instructions and reading material (J.M. Chavez, PhD, unpublished data, 1983). Behavior was assessed with a questionnaire regarding parent’s perception of the child’s attention, activity level, conduct, coordination, and sleep problems. The parent bringing the child to clinic, usually the mother, was interviewed regarding family attitudes (Family Environment Scale), life events (modified Coddington Life Events Scale), school history, and social supports (Family Routines Index)?7-29 A home visit was made, and the Elementary version of the HOME scale was completed for a detailed evaluation of the home environment. The HOME scale includes eight subscales, detailed in Table 2. Academic achievement scores were adjusted for intelligence quotient (IQ) by converting both to z-scores (ie, mean 0, standard deviation 1). The z-score for IQ was subtracted from that for the academic achievement (PIAT) subscale. Thus, a positive &dquo;adjusted achievement score&dquo; represents achievement above expectation for cognitive abilities and a negative one, relative underachievement. Compared to the method of adjusting for IQ described by McLeod3° and used by Seidenberg et al,~~ our procedure underestimates the number of underachievers when the median IQ of the group is below 100. By including a correction factor related to the expected correlation between the tests, McLeod’s method would increase the relative proportion of underachievers. Demographic variables included parental education, income, household composition (both parents, single parent, or other), language spoken at home, birthplace of parents and child (US-born versus immigrant), and length of residence in the United States for immigrant parents and children. Seizure history and previous treatment exposure was assessed from historical information provided by the parent at the first clinical visit. Severity and duration of seizures and total medication exposure, a combination of number of medications and duration of use, was rated as measure.
and
a
=
Statistical Analyses Initial statistical analysis using simple correlations and ttests was used to select independent variables for multiple regression analysis, excluding those without significant relationship with either PIAT or IQ. Variables with significant relationship to raw or adjusted PIAT scores or IQ scores were used in multiple regression analyses. When a portion of a test (ie, subscales of the HOME scale) were significantly related to IQ or PIAT scores, all of the subscales were included in the regressions. Multiple regression was used to determine the linear effect of the independent variables on the PIAT scores or the IQ-adjusted PIAT scores. Significant independent variables were selected by forward selection method, using the maximum r’ criterion.31 Significance level for inclusion was set at .10. Seizure-related variables were included in multiple regression analyses, even if insignificant on preliminary analyses.
=
TABLE 3 Scoring of Seizure Variables
TABLE 2 HOME Subscales
.
*Sum if more than one type of seizure. +Sum if more than one medication; do not record medication stopped over 5 years ago.
67
Missing Data and seizure data was examined, comparing the group with and without HOME scales, using X2 or ttests. Dependent variables (IQ and PIAT subscales) were separately analyzed for the entire group, excluding HOME scales, and for the group that completed the HOME scale. All demographic and sociocultural measures other than HOME scales were available for all subjects.
Demographic
Results Over 90% of parents
study. However,
agreed to enrollment in the not all enrolled subjects completed
all aspects of the evaluation. Subjects were excluded from the analysis if IQ or GCI was under 80 or if appropriate cognitive and achievement tests were not completed. Of 88 initial enrollees, a total of 78 subjects were available for analysis; 44 had the HOME scale completed. Subject demographics are described in Table 4. The only significant difference between subjects completing and those not completing the HOME scale was distance from clinic: no subjects living over 50 miles from the clinic had the HOME scale administered (four subjects). In all
TABLE 4
Description
of
Study Population other ways, the group completing the HOME scale comparable to the entire group. PIAT subscales for reading comprehension and general knowledge were not used for subjects under age 6 years, since most did not reach a basal score. These subjects are included in analyses of reading, math, and spelling portions of the PIAT. Subjects varied in their duration of seizure disorder and anticonvulsant exposure. Approximately one third of the subjects had experienced their first seizure within a month of enrollment. Half of the subjects had a seizure history of greater than a year. Over half had been treated for less than 1 month with anticonvulsant drugs. Subjects who had previously been treated with anticonvulsants received a variety of agents, usually including barbiturates. However, specific doses, compliance, and serum anticonvulsant levels were not available. Etiologic diagnoses were not generally available at enrollment. However, review of diagnostic information gathered over subsequent clinic visits revealed that most of the subjects had either primary was
generalized epilepsies (generalized convulsions, petit mal, or mixed in 27; juvenile myoclonic epi68
lepsy in two), or undetermined etiology (26). Six had benign focal epilepsy of childhood with central-temporal spikes. A smaller group had symptomatic or secondary epilepsies: cysticercosis (4), posttraumatic epilepsy (4), low-grade neoplasms (3), remote history of viral meningitis (1). Five patients had a mild hemiparesis, limb asymmetry or early handedness suggesting possible prenatal or perinatal onset of a focal cerebral lesion. Neurologic examination was considered to be abnormal in 13 subjects: five had mild hemiparesis or abnormal reflexes, and the remainder had nonspecific and nonlocalizing findings such as mild ataxia, poor fine-motor abilities, awkwardness, or apraxias. None had major focal neurologic deficits, and none used any adaptive equipment or assistive devices for mobility or activities of
daily living. In the group
as a whole, academic achievement poor, if not adjusted for IQ. For example, 55% were below the 25th percentile and 40% were below the 10th percentile for reading comprehension. Similarly, 41% were below the 25th and 18% were below the 10th percentile for mathematics. Median IQ was 95, with a range of 80 to 114. After adjusting expected achievement for IQ, 16% to 50% of the subjects were underachieving for each subscale of the PIAT, using a criteria of a 1/2 standard deviation difference between the PIAT and the IQ. As with unadjusted achievement scores, underachievement was greatest for reading comprehension and general knowledge. For both IQ-adjusted and uncorrected PIAT scores, achievement tended to be lower in older subjects. Underachievers were compared to achievers to see if any seizure variable (seizure duration or sever-
was
exposure) was related to achieve(Table 5). Achievement versus underachievement in reading recognition, reading comprehension, general knowledge, and spelling was
ity
or
medication
ment status
unrelated to seizure and medication variables. Underachievement in mathematics was slightly more frequent in the group with longer duration of seizure disorder, but this barely reached statistical significance (P .05). Stepwise regression was used to look at each IQadjusted PIAT subscale (Table 6). Final model? ranged from .37 for adjusted Reading (Adj-Reading) to .76 for adjusted Reading Comprehension (AdjReading Comp). Age had a significant negative correlation with Adj-Reading, Adj-Reading Comp, and adjusted Spelling (Adj-Spelling), ranging from r2 of .19 for Adj-Reading to r2 of .26 for Adj-Reading Comp. For all subscales except adjusted Mathematics (Adj-Math), sociocultural variables and not epilepsy-related variables entered the regressions at =
significant levels. Adj-Reading
was
positively
corre-
lated with two of the HOME subscales and father’s education and negatively correlated with parent reports of coordination problems. Adj-Reading Comp was positively correlated with HOME subscales 1, 2, 7, and 8 and negatively correlated with parental reports of overactivity. Adj-Spelling was negatively correlated with parental reports of attention and activity problems. Adjusted General Knowledge (AdjGen Knowledge) was associated with mother’s birthplace, with a relative advantage going to children of US-born mothers. Parents’ education also correlated positively to Adj-Gen Knowledge, as did several of the HOME subscales. Adj-Math behaved in a significantly different manner than the other
TABLE 5 Achievement Status Versus Seizure Variables
*A tUA
=
Achieving (zPIAT - zIQ > -0.5). Underachieving (zPIAT - zIQ :5 -0.5).
=
69
HOME subscales 4 and 6, father’s education, and testing in English. Negative predictors included age and reported attention problems. As the HOME scale was completed on only 44 of the 78 subjects, the analysis was repeated on the entire group, excluding the HOME scores. Parent education and household composition (two parent versus single parent) tended to increase in influence, and family income had a positive influence on IQ score. As with the smaller group with completed HOME scales, seizure-related variables had minimal influences on IQ-adjusted achievement. In only one analysis (Adj-Reading Comp) does medication duration enter a regression, and then with a partial r2 of only .06.
TABLE 6
Multiple Regression Examination
of
Adjusted
Achievement Scores
Discussion As expected, academic difficulties
were common
in
this group of children with epilepsy. Even after adjusting for cognitive abilities, 38% of the group were
underachieving
in
reading comprehension,
31% in
mathematics, and 32% in spelling. Underachievement in
extremely high our subjects were from a poor, inner-city population, with uneducated, non-English-speaking immigrant parents. Not all of our cultural biases apply, however. Immigrant children performed lower on IQ measures, but not on achievement tests after adjusting for IQ, and not on mathematics tests even prior to adjustment for IQ. The best predictors of IQ were subscales of (50%).
general knowledge
was
It must be noted that many of
the HOME scale that assessed educational materials in the household and family participation in devel-
*Regressions HOME Scale tP < .005.
were performed completed.
for 44
subjects
with IQ >80
and
tp < .01. ~P < .05. subscales. Adj-Math score was significantly negatively related to IQ (more underachievers among higher IQ subjects). The only other correlate of AdjMath was birthplace, with immigrant children performing better than US-born subjects. Notably, no epilepsy-related variable correlated with adjusted achievement scores. IQ score was examined with multiple regression as well. Significant positive predictors included
70
opmentally stimulating activities, as well as parental education. Although we have no nonepileptic control group in this study, this relationship was equally strong in newly diagnosed as in chronic epileptic subjects. It reemphasizes the importance of socioeconomic and cultural variables as potential confounders in studies of ill populations. Subjects for studies of epilepsy, including this one, are often drawn from teaching centers that serve disadvantaged populations to a disproportionate degree. This may serve to bias our view of the effects of epilepsy and its treatment: cognitive and academic deficits may reflect the influence of low socioeconomic level and cultural differences, rather than be attributable to seizures or medication effects. Dodrill et a132 have raised this issue in examining psychosocial consequences of epilepsy in adults. In his study, subjects selected from private practice settings or dues-paying members of voluntary organizations appeared to
be doing better than subject populations drawn from clinics or social service organizations. Parents report attention and coordination problems more frequently in children who are underachieving in reading, spelling, and general knowledge. Parents are probably aware that their child is not doing as well as others and report this; alternatively, the reported problems are symptomatic of underlying neurologic dysfunction. None of the reported problems with attention or coordination correlated with severity, duration, or treatment of epilepsy, however. Our findings contrast with those of Seidenberg et a1.13 The studies differ in that Seidenberg and coworkers used the Wide Range Achievement Test (WRAT) for reading recognition, spelling, and arithmetic, and the PIAT for reading comprehension, while we used the PIAT for all achievement tests. The PIAT also includes the additional general knowledge scale, not used by Seidenberg. PIAT reading, arithmetic, and spelling subscales correlate highly with the WRAT, both in normal and learningdisabled populations, although grade-equivalent scores on the arithmetic subscales tend to be higher on the PIAT.33,34 Using similar regression methods, Seidenberg et al entered only basic demographic variables, along with seizure-related variables. In their analysis, only a small portion of the variance of achievement is explained by these limited variables (6% for word recognition to 17% for arithmetic), despite a total of 122 subjects. In contrast, by including more extensive descriptions of the environment, sociocultural background, and economic status of the family, we attain cumulative r2 values ranging from 35% to 67%. Although the rates of underachievement are similar, Seidenberg and coworkers found a relationship between academic achievement in arithmetic and both seizure type and lifetime seizure frequency, which we did not find. However, we used different ratings of seizure frequency, severity, and treatment. Arithmetic achievement differed from other academic areas in both studies. In our subjects, arithmetic achievement was higher in immigrant children and furthest below expectations in the subjects with the highest cognitive abilities. Arithmetic achievement showed no significant relationship with any other sociocultural or environmental variable or with any seizure variable in our subjects. Reading comprehension, in contrast, was significantly positively related to several aspects of the home environment and negatively related to
age.
between severity or duration of seizure disortotal exposure to anticonvulsant drugs and underachievement. An equal proportion of newly diagnosed and/or untreated subjects were underachieving compared to patients with longstanding seizure disorder and anticonvulsant drug treatment. If, in fact there is a relationship between epilepsy and underachievement, underlying neurologic abnormalities may cause both the epilepsy and the learning disabilities seen in these children, rather than either the seizures or the treatment causing the academic difficulties. Without a nonepileptic, demographically matched control group, we cannot state that there is any causal relationship between epilepsy, or the underlying neurologic conditions causing epilepsy, and academic underachievement. We believe that our findings have important implications in the treatment of children with epilepsy. First of all, despite the lack of a causal link between seizure or treatment history and academic achievement, underachievement is common in this population. Assessment of academic achievement, attention to school placement, and appropriate intervention, when necessary, is an important aspect of the comprehensive treatment of children with epilepsy, particularly in populations such as this one. We encourage the use of academic screening, either by use of brief testing in the clinic, or by sending for school records, in all children enrolled in our clinics. Even in overburdened inner-city schools, a letter from &dquo;the doctor&dquo; encouraging the school to provide a complete assessment and remedial services often brings action from the school in mustering appropriate resources. On the other hand, we feel that our data is reassuring, in that neither severity, duration, nor treatment of epilepsy appeared to be a risk factor for underachievement. Parents whose children are doing well academically at the time of initial evaluation probably can be reassured that the diagnosis of epilepsy does not imply that the child will underachieve in the future. The cross-sectional nature of the study make our conclusions tentative. Longitudinal follow-up of the same subjects will provide more detailed data regarding seizure control and treatment and allow us to look for a more subtle relationship between academic underachievement and epilepsy or its treat-
ship der
or
ment.
Acknowledgments funded by a grant from The Robert Wood Johnson Foundation. The authors wish to thank Drs S. Robert SnodThe
Most
important,
there appears to be
no
relation-
study
was
71
grass and Stanley P. Azen for their help in the planning and execution of the study and their suggestions in preparing the
16.
manuscript.
17.
18.
References Pond DA: Conduct disorders in epileptic children. J Neurol Neurosurg Psychiatry 1957;20:65-68. 2. Rutter M, Graham P, Yule W: A Neuropsychiatric Study in Childhood. London, Spastics International Medical, 1970. 3. Bagley CR: Educational performance of children with epi1.
Grunberg F,
lepsy. BrJ Educ Psychol 1970;40:82-83. L, Whitmore K: A study of children with epilepsy attending ordinary schools: I. Seizure patterns, progress
4. Holdsworth
and behavior in school. Dev Med Child Neurol 1974; 16:746-758. 5. Stores G, Hart J: Reading skills of children with generalized or focal epilepsy attending ordinary school. Dev Med Child
1976;18:705-716. Neurol 6. Stores G, Piran N: Dependency of different types of children with epilepsy. Psychol Med 1978;8:441-445. 7. Long CB, Moore JR: Parental expectations for their epileptic . J Child Psychol Psychiatry 1979;20:299-312. children 8. Stores G: Problems of learning and behavior in children with epilepsy, in Reynolds EH, Trimble MJ (eds): Epilepsy and Psychiatry. Edinburgh, Churchill Livingstone, 1981, pp 33-48. 9. Farwell JR, Dodrill CB, Batzel LW: Neuropsychological abilities of children with epilepsy. Epilepsia 1985;26:395-400. 10. Cowan LD, Bodenstein JB, Leviton A, Doherty L: Prevalence of the epilepsies in children and adolescents. Epilepsia
1989;30:94-106. 11.
Hodgman CH, McAnarney ER, Myers GJ, et al: Emotional complications of adolescent grand mal epilepsy. J Pediatr 1979;95:309-312.
Green JB: The relation of parental attitudes to academic and social achievement in epileptic children. Epilepsia 1972;13:21-26. 13. Seidenberg M, Beck N, Geisser M, et al: Academic achievement in children with epilepsy. Epilepsia 1986;27:753-759. 14. Freeman JM, Tibbles J, Camfield C, Camfield P: Benign epilepsy of childhood: A speculation and its ramifications. Pediatrics 1987;79:864-868. 15. Vining EPG, Mellits ED, Dorsen M, et al: Psychologic and behavioral effects of anticonvulsant drugs in children. A double blind comparison of phenobarbital and valproate. Pediatrics 12.
Hartlage LC,
1987;80:165-174.
72
19.
Reynolds EH, Trimble MR (eds): Epilepsy and Psychiatry. Edinburgh, Churchill Livingstone, 1981, pp 27-32. Thompson PJ, Trimble MR: Anticonvulsant serum levels: Relationship to impairments of cognitive functioning. J Neurol Neurosurg Psychiatry 1983;46:227-233. Trimble MR, Thompson PJ: Anticonvulsant drugs, cognitive function and behavior. Epilepsia 1983;24(Suppl 1):S55-S63. Mitchell WG, Chavez JM: Carbamazepine versus phenobarbital for partial onset seizures in children. Epilepsia 1987;
28:56-60. 20. Caldwell BM, Bradley RH: Home Observation for Measurement of the Environment. Little Rock, University of Arkansas, 1978. 21. Bradley RH, Rock SL: The HOME Inventory. Its relation to school failure and development of an elementary-age version, in Frankenberg WK, Emde RN, Sullivan JW (eds): Early Identification of Children at Risk. New York, Plenum Press, 1985, pp 159-173. 22. Kaufman AS, Kaufman NL: Clinical Evaluation of Young Children With the McCarthy Scales. New York, Grune & Stratton, 1977. 23. Wechsler D: Manual for the Wechsler Intelligence Scale for Children-Revised. New York, Psychological Corporation, 1974. 24. Wechsler D: Escala de Inteligencia Para Niños. New York, Psychological Corporation, 1983. 25. Valencia RR: Predicting academic achievement of Mexican American children: Preliminary analyses of McCarthy Scales. Educ Psychol Meas 1982;42:1269-1278. 26. Dunn LM, Markwardt FC: Peabody Individual Achievement Test Manual. Circle Pines, American Guidance Service, 1970. 27. Moos R: Manual for the Family Environment Scale. Palo Alto, CA, Consulting Psychologist Press, 1974. 28. Coddington RD: The significance of life events as etiologic factors in the disease of children. J Psychosom Res 1972; 16:205-214. 29. Boyce WT: The Family Routines Inventory: Theoretical origins. Soc Sci Med 1983;17:193-200. 30. McLeod J: Educational underachievement: Toward a defensible psychometric definition. J Learn Disabil 1979;12:42-50. 31. Bentler PM: Multivariate analysis with latent variables: Causal modeling. Ann Rev Psychol 1980;332:419-456. 32. Dodrill CB, Beier R, Kasparick M, et al: Psychosocial problems among adults with epilepsy. Epilepsia 1984;25:168-175. 33. Harmer WR, Williams F: The Wide Range Achievement Test and the Peabody Individual Achievement Test: A comparative study. J Learn Disabil 1978;11:667-670. 34. Scull JW, Brand LH: The WRAT and the PIAT with learning disabled children. J Learn Disabil 1980;13:350-352.
G. Mitchell, MD;
John
M. Chavez, PhD;
Hang Lee, MS;
Bianca L. Guzman
Abstract Academic achievement was studied in 78 children with epilepsy, ages 5 to 13 years, to determine how seizures, treatment of seizures, and sociocultural factors influence academic achievement. Cognitive abilities were assessed with either the McCarthy Scales of Children’sAbilities or the Wechsler Intelligence Scale for Children-Revised. Achievement was measured with the Peabody Individual Achievement Tests. Achievement scores were corrected for cognitive ability (IQ), and underachievement was defined as achievement score 1/2 standard deviation or more below IQ. Information regarding seizure history (severity, duration) and treatment with anticonvulsant drugs was obtained. In addition, the family was extensively interviewed regarding the child’s environment, behavior, and demographic background. The Home Observation for Measurement of the Environment (HOME) scale was completed on a home visit. Underachievement was frequent, ranging from 16% (Reading Recognition) to 50% (General Knowledge), but there was no relationship between severity or duration of seizure disorder or total exposure to anticonvulsant medications and achievement. Major determinants of achievement included subscales of the HOME scale, age (older children more likely to be underachieving), and parental education. An equal proportion of newly diagnosed and/or untreated subjects were underachieving compared to those with longstanding epilepsy and anticonvulsant drug treatment. (J Child Neurol
1991;6:65-72).
has been associated and behavioral probwith Children with epilepsy have been noted to have increased frequency of underachievement, reading disabilities, behavior problems, overdependency, and psychiatric illness.6-9 Contributing to the complexity of these problems are interactions involv-
epilepsy long childhood academic, social, lems.l-5
ing neurologic disabilities, parental overprotection, social barriers, and medication effects. In addition, epilepsy disproportionately affects poor, inner-city, and immigrant children, who also may be more prone to academic failure for a variety of other rea-
sons.10
Severity Received
of seizures may be
a
poor
predictor
June 20, 1990. Received revised August 10,
Accepted for publication August
of
1990.
13, 1990.
From the Department of Pediatrics (Dr Mitchell) and Division of Neurology (Drs Mitchell and Chavez and Ms Guzman), Childrens Hospital of Los Angeles, and the Departments of Neurology (Dr Mitchell) and Preventive Medicine, Division of Biometry (Mr Lee), University of Southern California School of Medicine, Los Angeles, CA. The data contained in this paper were presented to the Child Neurology Society, San Antonio, TX, October 1989. Address correspondence to Dr Wendy G. Mitchell, Neurology Division—Box 42, Childrens Hospital of Los Angeles, PO Box 54700, Los Angeles, CA 90054.
academic and whose seizures
psychosocial disability. Children completely controlled or very infrequent perform poorly compared to their nonepileptic peers. Hodgman et al found that adolescents with well-controlled epilepsy had poor self-image, low achievement, and low expectations for the future compared to adolescents with uncontrolled or more severe epilepsy. 11 Hartlage and Greenl2 suggested that environmental factors, rather than specific neurologic disability, are major determinants of are
academic and social
in children with
epilepsy. Seidenberg et a113 studied academic achievement in children with epilepsy. They found that children with epilepsy made poor academic progress. This deficiency persisted after adjustment of academic expectations for IQ differences. Deficiencies were most marked in mathematics, followed by spelling, reading comprehension, and reading recognition. They found that a combination of demographic and clinical seizure variables were moderately predictive of success
academic underachievement in mathematics but not in reading or spelling. Neither the number nor the type of anticonvulsants were predictors of achievement.
65
Concern has been raised that treatment of epimay cause or contribute to academic difficulties. Freeman et al 14 have proposed that children with mild epilepsy not be treated, for fear of causing or exacerbating behavioral and academic difficulties. Although anticonvulsant drugs may influence cognitive function and occasional patients clearly exhibit drug-related impairment, anticonvulsant drug exposure per se has never been causally related to academic underachievement.l5-19 Sociodemographic, family, and environmental factors contribute to academic achievement, both in well and chronically ill children. Caldwell and Bradley2° developed an instrument to evaluate the home environment (The Home Observation for Measurement of the Environment, or HOME, scale). In a diverse and multiethnic group, HOME scores collected at 3 years of age correlated with both IQ and achievement test scores during the elementary years. HOME scores in infancy have moderately strong correlations with achievement at age 7 years, as well as with IQ at age 4 years. Several subscales of the elementary version of the HOME scale correlate with academic achievement at age 10 years in a socioeconomically and racially mixed group of children.21 Thus, we include evaluation of the home environment as an essential component in the understanding of the academic achievement of children with epilepsy.
lepsy
TABLE 1 Tests and Measures Used
66
We studied academic achievement in
a
group of
elementary-school aged children with epilepsy. We particularly asked whether epilepsy-related factors or sociocultural factors are the major determinants of school achievement. We expected that academic underachievement would be excessive in this group. However, we hypothesized that severity, duration, and treatment of the epilepsy would play a minor role in determining academic success, while family setting, parental attitudes, and underlying neurologic disability would play a major role.
Methods Patients aged 5 to 13 years were enrolled in the study at the time of their initial contact with the Children’s Hospital Los Angeles Neurology Service. Children were included if they had a definite diagnosis of epilepsy, regardless of duration. We did not enroll subjects with mental retardation, those with significant motor or sensory handicaps that would interfere with testing, or children who did not speak either English or Spanish. Informed consent was sought at the time of the first clinic visit, if the family indicated that they planned to continue to obtain their child’s neurologic care at Children’s Hospital. Enrollments continued over a period of 3 years. Initial evaluation of the child included psychometric, behavioral, and academic testing. Specific tests used are listed in Table 1. Children under age 7 years were tested with the McCarthy Scales of Children’s Abilities22; over age 7 the Wechsler Intelligence Scale for Children-Revised
(WISC-R) was used. 13 McCarthy General Cognitive Index (GCI) or WISC-R full-scale IQ were used as the cognitive
listed in Table 3. For patients with multiple seizure types, seizure severity score was summed. For some analyses, each epilepsy variable was further condensed: seizure severity scores of 1 or 2 versus 3 and above; seizure duration score of 1 (under 1 month) versus 3 (over 1 year); and medication duration scores of 1 (none or under a month) versus 3 or more (at least 1 year of medication). These scores represent the seizure and treatment history at the time of enrollment in our clinic. Detailed treatment histories, such as precise doses and serum levels, were rarely available.
In a normal population, both have a mean of 100 standard deviation of 15. For Spanish-speaking children, Escela Inteligencia por Ninos Wechsler (EINW)24 or a translated McCarthy Scale&dquo; was used. Academic achievement was measured with the Peabody Individual Achievement Test (PIAT). 26 PIAT Subtests include Reading, Reading Comprehension, Spelling, Mathematics, and General Knowledge. Subjects whose primary classroom instruction was in Spanish were tested in Spanish, using translated instructions and reading material (J.M. Chavez, PhD, unpublished data, 1983). Behavior was assessed with a questionnaire regarding parent’s perception of the child’s attention, activity level, conduct, coordination, and sleep problems. The parent bringing the child to clinic, usually the mother, was interviewed regarding family attitudes (Family Environment Scale), life events (modified Coddington Life Events Scale), school history, and social supports (Family Routines Index)?7-29 A home visit was made, and the Elementary version of the HOME scale was completed for a detailed evaluation of the home environment. The HOME scale includes eight subscales, detailed in Table 2. Academic achievement scores were adjusted for intelligence quotient (IQ) by converting both to z-scores (ie, mean 0, standard deviation 1). The z-score for IQ was subtracted from that for the academic achievement (PIAT) subscale. Thus, a positive &dquo;adjusted achievement score&dquo; represents achievement above expectation for cognitive abilities and a negative one, relative underachievement. Compared to the method of adjusting for IQ described by McLeod3° and used by Seidenberg et al,~~ our procedure underestimates the number of underachievers when the median IQ of the group is below 100. By including a correction factor related to the expected correlation between the tests, McLeod’s method would increase the relative proportion of underachievers. Demographic variables included parental education, income, household composition (both parents, single parent, or other), language spoken at home, birthplace of parents and child (US-born versus immigrant), and length of residence in the United States for immigrant parents and children. Seizure history and previous treatment exposure was assessed from historical information provided by the parent at the first clinical visit. Severity and duration of seizures and total medication exposure, a combination of number of medications and duration of use, was rated as measure.
and
a
=
Statistical Analyses Initial statistical analysis using simple correlations and ttests was used to select independent variables for multiple regression analysis, excluding those without significant relationship with either PIAT or IQ. Variables with significant relationship to raw or adjusted PIAT scores or IQ scores were used in multiple regression analyses. When a portion of a test (ie, subscales of the HOME scale) were significantly related to IQ or PIAT scores, all of the subscales were included in the regressions. Multiple regression was used to determine the linear effect of the independent variables on the PIAT scores or the IQ-adjusted PIAT scores. Significant independent variables were selected by forward selection method, using the maximum r’ criterion.31 Significance level for inclusion was set at .10. Seizure-related variables were included in multiple regression analyses, even if insignificant on preliminary analyses.
=
TABLE 3 Scoring of Seizure Variables
TABLE 2 HOME Subscales
.
*Sum if more than one type of seizure. +Sum if more than one medication; do not record medication stopped over 5 years ago.
67
Missing Data and seizure data was examined, comparing the group with and without HOME scales, using X2 or ttests. Dependent variables (IQ and PIAT subscales) were separately analyzed for the entire group, excluding HOME scales, and for the group that completed the HOME scale. All demographic and sociocultural measures other than HOME scales were available for all subjects.
Demographic
Results Over 90% of parents
study. However,
agreed to enrollment in the not all enrolled subjects completed
all aspects of the evaluation. Subjects were excluded from the analysis if IQ or GCI was under 80 or if appropriate cognitive and achievement tests were not completed. Of 88 initial enrollees, a total of 78 subjects were available for analysis; 44 had the HOME scale completed. Subject demographics are described in Table 4. The only significant difference between subjects completing and those not completing the HOME scale was distance from clinic: no subjects living over 50 miles from the clinic had the HOME scale administered (four subjects). In all
TABLE 4
Description
of
Study Population other ways, the group completing the HOME scale comparable to the entire group. PIAT subscales for reading comprehension and general knowledge were not used for subjects under age 6 years, since most did not reach a basal score. These subjects are included in analyses of reading, math, and spelling portions of the PIAT. Subjects varied in their duration of seizure disorder and anticonvulsant exposure. Approximately one third of the subjects had experienced their first seizure within a month of enrollment. Half of the subjects had a seizure history of greater than a year. Over half had been treated for less than 1 month with anticonvulsant drugs. Subjects who had previously been treated with anticonvulsants received a variety of agents, usually including barbiturates. However, specific doses, compliance, and serum anticonvulsant levels were not available. Etiologic diagnoses were not generally available at enrollment. However, review of diagnostic information gathered over subsequent clinic visits revealed that most of the subjects had either primary was
generalized epilepsies (generalized convulsions, petit mal, or mixed in 27; juvenile myoclonic epi68
lepsy in two), or undetermined etiology (26). Six had benign focal epilepsy of childhood with central-temporal spikes. A smaller group had symptomatic or secondary epilepsies: cysticercosis (4), posttraumatic epilepsy (4), low-grade neoplasms (3), remote history of viral meningitis (1). Five patients had a mild hemiparesis, limb asymmetry or early handedness suggesting possible prenatal or perinatal onset of a focal cerebral lesion. Neurologic examination was considered to be abnormal in 13 subjects: five had mild hemiparesis or abnormal reflexes, and the remainder had nonspecific and nonlocalizing findings such as mild ataxia, poor fine-motor abilities, awkwardness, or apraxias. None had major focal neurologic deficits, and none used any adaptive equipment or assistive devices for mobility or activities of
daily living. In the group
as a whole, academic achievement poor, if not adjusted for IQ. For example, 55% were below the 25th percentile and 40% were below the 10th percentile for reading comprehension. Similarly, 41% were below the 25th and 18% were below the 10th percentile for mathematics. Median IQ was 95, with a range of 80 to 114. After adjusting expected achievement for IQ, 16% to 50% of the subjects were underachieving for each subscale of the PIAT, using a criteria of a 1/2 standard deviation difference between the PIAT and the IQ. As with unadjusted achievement scores, underachievement was greatest for reading comprehension and general knowledge. For both IQ-adjusted and uncorrected PIAT scores, achievement tended to be lower in older subjects. Underachievers were compared to achievers to see if any seizure variable (seizure duration or sever-
was
exposure) was related to achieve(Table 5). Achievement versus underachievement in reading recognition, reading comprehension, general knowledge, and spelling was
ity
or
medication
ment status
unrelated to seizure and medication variables. Underachievement in mathematics was slightly more frequent in the group with longer duration of seizure disorder, but this barely reached statistical significance (P .05). Stepwise regression was used to look at each IQadjusted PIAT subscale (Table 6). Final model? ranged from .37 for adjusted Reading (Adj-Reading) to .76 for adjusted Reading Comprehension (AdjReading Comp). Age had a significant negative correlation with Adj-Reading, Adj-Reading Comp, and adjusted Spelling (Adj-Spelling), ranging from r2 of .19 for Adj-Reading to r2 of .26 for Adj-Reading Comp. For all subscales except adjusted Mathematics (Adj-Math), sociocultural variables and not epilepsy-related variables entered the regressions at =
significant levels. Adj-Reading
was
positively
corre-
lated with two of the HOME subscales and father’s education and negatively correlated with parent reports of coordination problems. Adj-Reading Comp was positively correlated with HOME subscales 1, 2, 7, and 8 and negatively correlated with parental reports of overactivity. Adj-Spelling was negatively correlated with parental reports of attention and activity problems. Adjusted General Knowledge (AdjGen Knowledge) was associated with mother’s birthplace, with a relative advantage going to children of US-born mothers. Parents’ education also correlated positively to Adj-Gen Knowledge, as did several of the HOME subscales. Adj-Math behaved in a significantly different manner than the other
TABLE 5 Achievement Status Versus Seizure Variables
*A tUA
=
Achieving (zPIAT - zIQ > -0.5). Underachieving (zPIAT - zIQ :5 -0.5).
=
69
HOME subscales 4 and 6, father’s education, and testing in English. Negative predictors included age and reported attention problems. As the HOME scale was completed on only 44 of the 78 subjects, the analysis was repeated on the entire group, excluding the HOME scores. Parent education and household composition (two parent versus single parent) tended to increase in influence, and family income had a positive influence on IQ score. As with the smaller group with completed HOME scales, seizure-related variables had minimal influences on IQ-adjusted achievement. In only one analysis (Adj-Reading Comp) does medication duration enter a regression, and then with a partial r2 of only .06.
TABLE 6
Multiple Regression Examination
of
Adjusted
Achievement Scores
Discussion As expected, academic difficulties
were common
in
this group of children with epilepsy. Even after adjusting for cognitive abilities, 38% of the group were
underachieving
in
reading comprehension,
31% in
mathematics, and 32% in spelling. Underachievement in
extremely high our subjects were from a poor, inner-city population, with uneducated, non-English-speaking immigrant parents. Not all of our cultural biases apply, however. Immigrant children performed lower on IQ measures, but not on achievement tests after adjusting for IQ, and not on mathematics tests even prior to adjustment for IQ. The best predictors of IQ were subscales of (50%).
general knowledge
was
It must be noted that many of
the HOME scale that assessed educational materials in the household and family participation in devel-
*Regressions HOME Scale tP < .005.
were performed completed.
for 44
subjects
with IQ >80
and
tp < .01. ~P < .05. subscales. Adj-Math score was significantly negatively related to IQ (more underachievers among higher IQ subjects). The only other correlate of AdjMath was birthplace, with immigrant children performing better than US-born subjects. Notably, no epilepsy-related variable correlated with adjusted achievement scores. IQ score was examined with multiple regression as well. Significant positive predictors included
70
opmentally stimulating activities, as well as parental education. Although we have no nonepileptic control group in this study, this relationship was equally strong in newly diagnosed as in chronic epileptic subjects. It reemphasizes the importance of socioeconomic and cultural variables as potential confounders in studies of ill populations. Subjects for studies of epilepsy, including this one, are often drawn from teaching centers that serve disadvantaged populations to a disproportionate degree. This may serve to bias our view of the effects of epilepsy and its treatment: cognitive and academic deficits may reflect the influence of low socioeconomic level and cultural differences, rather than be attributable to seizures or medication effects. Dodrill et a132 have raised this issue in examining psychosocial consequences of epilepsy in adults. In his study, subjects selected from private practice settings or dues-paying members of voluntary organizations appeared to
be doing better than subject populations drawn from clinics or social service organizations. Parents report attention and coordination problems more frequently in children who are underachieving in reading, spelling, and general knowledge. Parents are probably aware that their child is not doing as well as others and report this; alternatively, the reported problems are symptomatic of underlying neurologic dysfunction. None of the reported problems with attention or coordination correlated with severity, duration, or treatment of epilepsy, however. Our findings contrast with those of Seidenberg et a1.13 The studies differ in that Seidenberg and coworkers used the Wide Range Achievement Test (WRAT) for reading recognition, spelling, and arithmetic, and the PIAT for reading comprehension, while we used the PIAT for all achievement tests. The PIAT also includes the additional general knowledge scale, not used by Seidenberg. PIAT reading, arithmetic, and spelling subscales correlate highly with the WRAT, both in normal and learningdisabled populations, although grade-equivalent scores on the arithmetic subscales tend to be higher on the PIAT.33,34 Using similar regression methods, Seidenberg et al entered only basic demographic variables, along with seizure-related variables. In their analysis, only a small portion of the variance of achievement is explained by these limited variables (6% for word recognition to 17% for arithmetic), despite a total of 122 subjects. In contrast, by including more extensive descriptions of the environment, sociocultural background, and economic status of the family, we attain cumulative r2 values ranging from 35% to 67%. Although the rates of underachievement are similar, Seidenberg and coworkers found a relationship between academic achievement in arithmetic and both seizure type and lifetime seizure frequency, which we did not find. However, we used different ratings of seizure frequency, severity, and treatment. Arithmetic achievement differed from other academic areas in both studies. In our subjects, arithmetic achievement was higher in immigrant children and furthest below expectations in the subjects with the highest cognitive abilities. Arithmetic achievement showed no significant relationship with any other sociocultural or environmental variable or with any seizure variable in our subjects. Reading comprehension, in contrast, was significantly positively related to several aspects of the home environment and negatively related to
age.
between severity or duration of seizure disortotal exposure to anticonvulsant drugs and underachievement. An equal proportion of newly diagnosed and/or untreated subjects were underachieving compared to patients with longstanding seizure disorder and anticonvulsant drug treatment. If, in fact there is a relationship between epilepsy and underachievement, underlying neurologic abnormalities may cause both the epilepsy and the learning disabilities seen in these children, rather than either the seizures or the treatment causing the academic difficulties. Without a nonepileptic, demographically matched control group, we cannot state that there is any causal relationship between epilepsy, or the underlying neurologic conditions causing epilepsy, and academic underachievement. We believe that our findings have important implications in the treatment of children with epilepsy. First of all, despite the lack of a causal link between seizure or treatment history and academic achievement, underachievement is common in this population. Assessment of academic achievement, attention to school placement, and appropriate intervention, when necessary, is an important aspect of the comprehensive treatment of children with epilepsy, particularly in populations such as this one. We encourage the use of academic screening, either by use of brief testing in the clinic, or by sending for school records, in all children enrolled in our clinics. Even in overburdened inner-city schools, a letter from &dquo;the doctor&dquo; encouraging the school to provide a complete assessment and remedial services often brings action from the school in mustering appropriate resources. On the other hand, we feel that our data is reassuring, in that neither severity, duration, nor treatment of epilepsy appeared to be a risk factor for underachievement. Parents whose children are doing well academically at the time of initial evaluation probably can be reassured that the diagnosis of epilepsy does not imply that the child will underachieve in the future. The cross-sectional nature of the study make our conclusions tentative. Longitudinal follow-up of the same subjects will provide more detailed data regarding seizure control and treatment and allow us to look for a more subtle relationship between academic underachievement and epilepsy or its treat-
ship der
or
ment.
Acknowledgments funded by a grant from The Robert Wood Johnson Foundation. The authors wish to thank Drs S. Robert SnodThe
Most
important,
there appears to be
no
relation-
study
was
71
grass and Stanley P. Azen for their help in the planning and execution of the study and their suggestions in preparing the
16.
manuscript.
17.
18.
References Pond DA: Conduct disorders in epileptic children. J Neurol Neurosurg Psychiatry 1957;20:65-68. 2. Rutter M, Graham P, Yule W: A Neuropsychiatric Study in Childhood. London, Spastics International Medical, 1970. 3. Bagley CR: Educational performance of children with epi1.
Grunberg F,
lepsy. BrJ Educ Psychol 1970;40:82-83. L, Whitmore K: A study of children with epilepsy attending ordinary schools: I. Seizure patterns, progress
4. Holdsworth
and behavior in school. Dev Med Child Neurol 1974; 16:746-758. 5. Stores G, Hart J: Reading skills of children with generalized or focal epilepsy attending ordinary school. Dev Med Child
1976;18:705-716. Neurol 6. Stores G, Piran N: Dependency of different types of children with epilepsy. Psychol Med 1978;8:441-445. 7. Long CB, Moore JR: Parental expectations for their epileptic . J Child Psychol Psychiatry 1979;20:299-312. children 8. Stores G: Problems of learning and behavior in children with epilepsy, in Reynolds EH, Trimble MJ (eds): Epilepsy and Psychiatry. Edinburgh, Churchill Livingstone, 1981, pp 33-48. 9. Farwell JR, Dodrill CB, Batzel LW: Neuropsychological abilities of children with epilepsy. Epilepsia 1985;26:395-400. 10. Cowan LD, Bodenstein JB, Leviton A, Doherty L: Prevalence of the epilepsies in children and adolescents. Epilepsia
1989;30:94-106. 11.
Hodgman CH, McAnarney ER, Myers GJ, et al: Emotional complications of adolescent grand mal epilepsy. J Pediatr 1979;95:309-312.
Green JB: The relation of parental attitudes to academic and social achievement in epileptic children. Epilepsia 1972;13:21-26. 13. Seidenberg M, Beck N, Geisser M, et al: Academic achievement in children with epilepsy. Epilepsia 1986;27:753-759. 14. Freeman JM, Tibbles J, Camfield C, Camfield P: Benign epilepsy of childhood: A speculation and its ramifications. Pediatrics 1987;79:864-868. 15. Vining EPG, Mellits ED, Dorsen M, et al: Psychologic and behavioral effects of anticonvulsant drugs in children. A double blind comparison of phenobarbital and valproate. Pediatrics 12.
Hartlage LC,
1987;80:165-174.
72
19.
Reynolds EH, Trimble MR (eds): Epilepsy and Psychiatry. Edinburgh, Churchill Livingstone, 1981, pp 27-32. Thompson PJ, Trimble MR: Anticonvulsant serum levels: Relationship to impairments of cognitive functioning. J Neurol Neurosurg Psychiatry 1983;46:227-233. Trimble MR, Thompson PJ: Anticonvulsant drugs, cognitive function and behavior. Epilepsia 1983;24(Suppl 1):S55-S63. Mitchell WG, Chavez JM: Carbamazepine versus phenobarbital for partial onset seizures in children. Epilepsia 1987;
28:56-60. 20. Caldwell BM, Bradley RH: Home Observation for Measurement of the Environment. Little Rock, University of Arkansas, 1978. 21. Bradley RH, Rock SL: The HOME Inventory. Its relation to school failure and development of an elementary-age version, in Frankenberg WK, Emde RN, Sullivan JW (eds): Early Identification of Children at Risk. New York, Plenum Press, 1985, pp 159-173. 22. Kaufman AS, Kaufman NL: Clinical Evaluation of Young Children With the McCarthy Scales. New York, Grune & Stratton, 1977. 23. Wechsler D: Manual for the Wechsler Intelligence Scale for Children-Revised. New York, Psychological Corporation, 1974. 24. Wechsler D: Escala de Inteligencia Para Niños. New York, Psychological Corporation, 1983. 25. Valencia RR: Predicting academic achievement of Mexican American children: Preliminary analyses of McCarthy Scales. Educ Psychol Meas 1982;42:1269-1278. 26. Dunn LM, Markwardt FC: Peabody Individual Achievement Test Manual. Circle Pines, American Guidance Service, 1970. 27. Moos R: Manual for the Family Environment Scale. Palo Alto, CA, Consulting Psychologist Press, 1974. 28. Coddington RD: The significance of life events as etiologic factors in the disease of children. J Psychosom Res 1972; 16:205-214. 29. Boyce WT: The Family Routines Inventory: Theoretical origins. Soc Sci Med 1983;17:193-200. 30. McLeod J: Educational underachievement: Toward a defensible psychometric definition. J Learn Disabil 1979;12:42-50. 31. Bentler PM: Multivariate analysis with latent variables: Causal modeling. Ann Rev Psychol 1980;332:419-456. 32. Dodrill CB, Beier R, Kasparick M, et al: Psychosocial problems among adults with epilepsy. Epilepsia 1984;25:168-175. 33. Harmer WR, Williams F: The Wide Range Achievement Test and the Peabody Individual Achievement Test: A comparative study. J Learn Disabil 1978;11:667-670. 34. Scull JW, Brand LH: The WRAT and the PIAT with learning disabled children. J Learn Disabil 1980;13:350-352.
