Academic Achievement Following Childhood Brain Disease: Implications for the Concept of Learning Disabilities H. Gerry Taylor and Christopher Schatschneider
The present study examined three hypotheses regarding the consequences of early brain damage for academic achievement: First, early brain insults will have a negative impact on achievement, even in children with normal intelligence. Second, underachievement in these children will be at least partially independent of IQ (i.e., not fully accounted for by a lowering of IQ within the average range). Third, normally intelligent children with histories of brain insult will also manifest selective cognitive dysfunctions. To test these hypotheses, we compared two groups of children who had recovered from Haemophilus influenzae type b meningitis. The "complicated" group consisted of children who, as a consequence of having sustained neurologic complications during their illness, were more likely to have had brain insults. Children in the "uncomplicated" group did not have complications with their illness and were regarded as having escaped significant central nervous system (CNS) pathology. Only children with normal hearing and a prorated Full Scale IQ of at least 80 (WISC-R) were considered. Group differences on the Wide Range Achievement Test-Revised were consistent with the first two hypotheses. Although the two groups had similar Verbal IQs, the complicated group also had a lower mean Performance IQ and performed less well on perceptually demanding neuropsychological tasks. Findings suggest that learning disabilities may have selective, brain-related cognitive antecedents; but they challenge the practice of using IQ criteria for clinical diagnosis.
T
he practice of defining learning disabilities on the basis of discrepancies between academic achievement and IQ scores has come under increasing scrutiny in recent years. Recommendations have been made to abandon discrepancy formulae altogether (Algozzine & Ysseldyke, 1983; Share, McGee, & Silva, 1989; Siegel, 1988,1989; Taylor, 1989a). The major drawbacks to applying these formulae are that even relatively isolated cognitive deficiencies may depress IQ as well as achievement, and that poor achievement early in school influences subsequent learning and intellectual
development (Siegel, 1989; Stanovich, 1988; Taylor & Fletcher, 1990). In either case, children with learning problems of a relatively specific nature will fail to meet diagnostic criteria for learning disabilities. Findings indicating that poor readers who meet discrepancy formulae are not clearly distinguishable from other poor readers raise even further doubt in this regard (Fletcher et al., 1989; Scarborough, 1989; Share, McGee, McKenzie, Williams, & Silva, 1987; Share et al., 1989; Taylor, Satz, & Friel, 1979). It is important to emphasize, however, that fixed IQ cutoffs and IQ-
achievement discrepancies have been advocated primarily for the purposes of educational classification. Strict IQ limits are not detailed either in the original definition of learning disabilities or in subsequent revisions (Taylor, 1988). The intention of those who developed the definition, and of the tradition that led up to its formulation, seems to have been to focus attention on a particular subset of learning problems. This subset includes problems that are presumed to be CNS-related but cannot be explained in terms of general mental deficiency, environmental factors, or emotional disorder (Taylor, 1989b). Following the historical tradition from which the concept of learning disabilities emerged, IQ criteria were imposed primarily as a means of ruling out confounding variables and for assisting in the search for more specific cognitive antecedents (Taylor, 1984). The original intent of measuring IQ was not to provide for clinical diagnosis. Recent commentaries on the subject reinforce the view that IQ restrictions are useful in research, but highly problematic in determining educational placement (Torgesen, 1989). Observations of children with histories of brain insult have been a major
JOURNAL OF LEARNING DISABILITIES VOLUME 25,atNUMBER 10, DECEMBER 1992on February 28, 2015 Downloaded from ldx.sagepub.com MICHIGAN STATE UNIV LIBRARIES PAGES 630-638
VOLUME 25, NUMBER 10, DECEMBER 1992
stimulus for this research strategy, and have contributed in large part to the assumption that constitutional factors underlie specific forms of learning failure. According to reports dating as far back as the turn of the century, developmental problems frequently were noted in children who had survived neurological disease or injury with normal intelligence (Taylor & Fletcher, 1990). Although difficulties in impulse/ attentional control, social judgment, perception, and abstract reasoning became hallmarks of "the brain-injured child," poor educational progress was also stressed (Benton, 1962; Kresky, Buchbinder, & Greenberg, 1962; Myer & Byers, 1952). As Byers and Lord (1943) stated, these children "seem smart enough but do not learn" (p. 481). What is less clear, however, is whether the academic problems of nonretarded children with brain injuries are attributable to relatively isolated processing deficits or merely to a general lowering of cognitive abilities within the average range. Surprisingly, this question has rarely been asked; and relevant studies have yielded contradictory results. Support for a specific deficit was provided by Klein, Hack, and Breslau (1989), who found that low birthweight children obtained lower IQ-adjusted scores on an arithmetic test than a control group. Seidel, Chadwick, and Rutter (1975) made a similar discovery in comparing the reading skills of children with brain lesions to those of children with peripheral injuries. On the other hand, Chadwick, Rutter, Brown, Shaffer, and Traub (1981) found that academic difficulties observed in children after head injury corresponded with postinjury intellectual functioning. The fact that neurological disorders are more common in children whose reading problems are commensurate with their IQs than in children meeting IQachievement discrepancy criteria is further evidence against the specific deficit hypothesis (Ingram, Mason, & Blackburn, 1970; Rutter & Yule, 1975; Silva, McGee, & Williams, 1985).
631
The objective of the present investi- rott, Taylor, & Montes, 1991; Taylor, gation was to subject the latter hypoth- 1987). esis to more rigorous examination, using data from a study of the consequences of Haemophilus influenzae type Method b meningitis in school-age children. A special virtue of this data set is that Sample relatively comprehensive assessments were made of the children's socialThe sample consisted of children environmental status and of neuropsy- who had been hospitalized earlier in chological, behavioral, and school out- life for Haemophilus influenzae type b comes. Previous analyses suggested meningitis. Children were recruited that reading and spelling skills were from a survey of medical records from adversely affected by the disease, and three children's hospitals (Montreal that sequelae were evident only for Children's Hospital, Children's Hosindex (i.e., postmeningitis) cases in pital of Eastern Ontario in Ottawa, which there were neurologic complica- and the Hospital for Sick Children in tions during the acute-phase illness Toronto). Eligibility criteria were age at (Taylor et al., 1990; Taylor, Schatsch- testing from 6 to 14 years, English as neider, & Rich, 1991). The absence of the major language of schooling, abany detectable sequelae in children sence of neurologic disorder other than without complications made it possi- meningitis, and residence within the ble to divide the postmeningitis sam- greater metropolitan areas served by ple into two groups, based on the the three admitting hospitals. likelihood of significant CNS insult. A total of 127 children who had had Considering only children from the meningitis were recruited in the larger larger study with broadly average IQs study, together with 97 siblings (Taylor and with normal hearing, we then et al., 1991). The sample of interest compared the "complicated" and "un- for the present purposes consisted of complicated" groups in terms of aca- the 107 postmeningitis cases with nordemic achievement and other skills. mal hearing and with prorated Full Three hypotheses were tested. First, Scale IQs on the Wechsler Intelligence we expected that the complicated Scale for Children-Revised (WISC-R) group would perform more poorly (Wechsler, 1974) of at least 80 at the than the uncomplicated group on tests time of testing. of academic achievement, and that a Division of the sample into those greater number of the children who with and without acute-phase neurohad had complications would display logic complications yielded 41 complisubstandard achievement. Second, we cated cases (22 males and 19 females) predicted that it would not be possible and 66 uncomplicated cases (32 males to fully account for any differences in and 34 females). Mean age at testing achievement in terms of IQ, and that did not differ between groups (9.7 the incidence of IQ-achievement dis- years, SD=2.2, for the complicated crepancies would be higher in the group; 9.8 years, SD=2.4, for the uncomplicated group. Finally, we hy- complicated group). Mean age at illpothesized that the differences in ness was approximately IV2 years for achievement would be accompanied both groups. The vast majority of chilby differences in cognitive functioning dren in this Canadian sample were and behavior. Based on previous re- from white, two-parent families. search, we anticipated that the differOnly four of the children (all in the ences would be selective rather than complicated group) had neurologic disglobal, and that the pattern of differ- orders persisting more than 6 weeks ences would be suggestive of factors after discharge from the hospital. One responsible for the achievement limi- child had residual paresis of the sixth tations of the complicated group (Per- cranial nerve, which resolved within
Downloaded from ldx.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on February 28, 2015
632
JOURNAL OF LEARNING DISABILITIES
6 months of the illness, and the other three children had seizure disorders (only one was on anticonvulsant medications at the time of testing).
summarizing the child's recent life history.
Procedure
Social-Environmental Status
Tests administered included the Reading, Spelling, and Arithmetic subtests of the Wide Range Achievement Test-Revised (WRAT-R) (Jastak & Wilkinson, 1984), the Gilmore Oral Reading Test (Gilmore & Gilmore, 1968), six subtests from the WISC-R (Similarities, Vocabulary, Digit Span, Block Design, Object Assembly, and Coding), and a battery of neuropsychological procedures assessing language abilities, psychomotor skills, memory, attention, and abstract reasoning (see Taylor et al., 1990, for a more complete description). Tests were given in counterbalanced order across children. Behavior and school performance were evaluated by means of the Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983), the Teacher's Report Form (TRF) (Achenbach & Edelbrock, 1986), and the Vineland Adaptive Behavior Scales (Sparrow, Balla, & Ciccetti, 1984). Parents were asked to rate each child's school progress for each year of schooling on a 0-to-2 scale (poor, fair, good), and each child's average rating was computed. Family socioeconomic status (SES) was measured by means of the Hollingshead Two-Factor Index (Hollingshead, 1965). Other potential social and environmental influences on child outcomes were assessed by having the child's parent complete the Home Environment Questionnaire (Sines, 1983) and the Life Events Questionnaire (Garmezy, Tellegen, & Dervine, 1981). The former scale provides a parentbased rating of several behaviorally relevant characteristics of the home environment, including an alternative assessment of family SES. The latter scale requires parents to indicate which of 50 events have transpired in the child's life during the previous 12 months. Positive and negative events are tallied separately as a means of
Data summarizing the socialenvironmental status of the two groups are given in Table 1. According to the results of t tests, the complicated group had higher scores than the uncomplicated group on the Press for Supervision scale of the Home Environment Questionnaire. The two groups were otherwise well matched in terms of social-environmental background factors. (Due to the exploratory nature of the study and given the limited statistical power, alpha for significance testing was set at the conventional .05 level for these and other comparisons.)
Results
WISC-R
Scores
Group differences on the WISC-R subtests and on prorated Verbal, Per-
formance, and Full Scale IQ are summarized in Table 2. Verbal IQ was prorated on the basis of Similarities, Vocabulary, and Digit Span; and Performance IQ on the basis of Block Design, Object Assembly, and Coding. Full Scale IQ was derived using prorated raw score totals for the Verbal and Performance scales. According to the results of t tests, individual subtests that distinguished the two groups were Block Design, t(105)=2.4, p < .05; Object Assembly, f(105)=2.1, p
The present study examined three hypotheses regarding the consequences of early brain damage for academic achievement: First, early brain insults will have a negative impact on achievement, even in children with normal intelligence. Second, underachievement in these children will be at least partially independent of IQ (i.e., not fully accounted for by a lowering of IQ within the average range). Third, normally intelligent children with histories of brain insult will also manifest selective cognitive dysfunctions. To test these hypotheses, we compared two groups of children who had recovered from Haemophilus influenzae type b meningitis. The "complicated" group consisted of children who, as a consequence of having sustained neurologic complications during their illness, were more likely to have had brain insults. Children in the "uncomplicated" group did not have complications with their illness and were regarded as having escaped significant central nervous system (CNS) pathology. Only children with normal hearing and a prorated Full Scale IQ of at least 80 (WISC-R) were considered. Group differences on the Wide Range Achievement Test-Revised were consistent with the first two hypotheses. Although the two groups had similar Verbal IQs, the complicated group also had a lower mean Performance IQ and performed less well on perceptually demanding neuropsychological tasks. Findings suggest that learning disabilities may have selective, brain-related cognitive antecedents; but they challenge the practice of using IQ criteria for clinical diagnosis.
T
he practice of defining learning disabilities on the basis of discrepancies between academic achievement and IQ scores has come under increasing scrutiny in recent years. Recommendations have been made to abandon discrepancy formulae altogether (Algozzine & Ysseldyke, 1983; Share, McGee, & Silva, 1989; Siegel, 1988,1989; Taylor, 1989a). The major drawbacks to applying these formulae are that even relatively isolated cognitive deficiencies may depress IQ as well as achievement, and that poor achievement early in school influences subsequent learning and intellectual
development (Siegel, 1989; Stanovich, 1988; Taylor & Fletcher, 1990). In either case, children with learning problems of a relatively specific nature will fail to meet diagnostic criteria for learning disabilities. Findings indicating that poor readers who meet discrepancy formulae are not clearly distinguishable from other poor readers raise even further doubt in this regard (Fletcher et al., 1989; Scarborough, 1989; Share, McGee, McKenzie, Williams, & Silva, 1987; Share et al., 1989; Taylor, Satz, & Friel, 1979). It is important to emphasize, however, that fixed IQ cutoffs and IQ-
achievement discrepancies have been advocated primarily for the purposes of educational classification. Strict IQ limits are not detailed either in the original definition of learning disabilities or in subsequent revisions (Taylor, 1988). The intention of those who developed the definition, and of the tradition that led up to its formulation, seems to have been to focus attention on a particular subset of learning problems. This subset includes problems that are presumed to be CNS-related but cannot be explained in terms of general mental deficiency, environmental factors, or emotional disorder (Taylor, 1989b). Following the historical tradition from which the concept of learning disabilities emerged, IQ criteria were imposed primarily as a means of ruling out confounding variables and for assisting in the search for more specific cognitive antecedents (Taylor, 1984). The original intent of measuring IQ was not to provide for clinical diagnosis. Recent commentaries on the subject reinforce the view that IQ restrictions are useful in research, but highly problematic in determining educational placement (Torgesen, 1989). Observations of children with histories of brain insult have been a major
JOURNAL OF LEARNING DISABILITIES VOLUME 25,atNUMBER 10, DECEMBER 1992on February 28, 2015 Downloaded from ldx.sagepub.com MICHIGAN STATE UNIV LIBRARIES PAGES 630-638
VOLUME 25, NUMBER 10, DECEMBER 1992
stimulus for this research strategy, and have contributed in large part to the assumption that constitutional factors underlie specific forms of learning failure. According to reports dating as far back as the turn of the century, developmental problems frequently were noted in children who had survived neurological disease or injury with normal intelligence (Taylor & Fletcher, 1990). Although difficulties in impulse/ attentional control, social judgment, perception, and abstract reasoning became hallmarks of "the brain-injured child," poor educational progress was also stressed (Benton, 1962; Kresky, Buchbinder, & Greenberg, 1962; Myer & Byers, 1952). As Byers and Lord (1943) stated, these children "seem smart enough but do not learn" (p. 481). What is less clear, however, is whether the academic problems of nonretarded children with brain injuries are attributable to relatively isolated processing deficits or merely to a general lowering of cognitive abilities within the average range. Surprisingly, this question has rarely been asked; and relevant studies have yielded contradictory results. Support for a specific deficit was provided by Klein, Hack, and Breslau (1989), who found that low birthweight children obtained lower IQ-adjusted scores on an arithmetic test than a control group. Seidel, Chadwick, and Rutter (1975) made a similar discovery in comparing the reading skills of children with brain lesions to those of children with peripheral injuries. On the other hand, Chadwick, Rutter, Brown, Shaffer, and Traub (1981) found that academic difficulties observed in children after head injury corresponded with postinjury intellectual functioning. The fact that neurological disorders are more common in children whose reading problems are commensurate with their IQs than in children meeting IQachievement discrepancy criteria is further evidence against the specific deficit hypothesis (Ingram, Mason, & Blackburn, 1970; Rutter & Yule, 1975; Silva, McGee, & Williams, 1985).
631
The objective of the present investi- rott, Taylor, & Montes, 1991; Taylor, gation was to subject the latter hypoth- 1987). esis to more rigorous examination, using data from a study of the consequences of Haemophilus influenzae type Method b meningitis in school-age children. A special virtue of this data set is that Sample relatively comprehensive assessments were made of the children's socialThe sample consisted of children environmental status and of neuropsy- who had been hospitalized earlier in chological, behavioral, and school out- life for Haemophilus influenzae type b comes. Previous analyses suggested meningitis. Children were recruited that reading and spelling skills were from a survey of medical records from adversely affected by the disease, and three children's hospitals (Montreal that sequelae were evident only for Children's Hospital, Children's Hosindex (i.e., postmeningitis) cases in pital of Eastern Ontario in Ottawa, which there were neurologic complica- and the Hospital for Sick Children in tions during the acute-phase illness Toronto). Eligibility criteria were age at (Taylor et al., 1990; Taylor, Schatsch- testing from 6 to 14 years, English as neider, & Rich, 1991). The absence of the major language of schooling, abany detectable sequelae in children sence of neurologic disorder other than without complications made it possi- meningitis, and residence within the ble to divide the postmeningitis sam- greater metropolitan areas served by ple into two groups, based on the the three admitting hospitals. likelihood of significant CNS insult. A total of 127 children who had had Considering only children from the meningitis were recruited in the larger larger study with broadly average IQs study, together with 97 siblings (Taylor and with normal hearing, we then et al., 1991). The sample of interest compared the "complicated" and "un- for the present purposes consisted of complicated" groups in terms of aca- the 107 postmeningitis cases with nordemic achievement and other skills. mal hearing and with prorated Full Three hypotheses were tested. First, Scale IQs on the Wechsler Intelligence we expected that the complicated Scale for Children-Revised (WISC-R) group would perform more poorly (Wechsler, 1974) of at least 80 at the than the uncomplicated group on tests time of testing. of academic achievement, and that a Division of the sample into those greater number of the children who with and without acute-phase neurohad had complications would display logic complications yielded 41 complisubstandard achievement. Second, we cated cases (22 males and 19 females) predicted that it would not be possible and 66 uncomplicated cases (32 males to fully account for any differences in and 34 females). Mean age at testing achievement in terms of IQ, and that did not differ between groups (9.7 the incidence of IQ-achievement dis- years, SD=2.2, for the complicated crepancies would be higher in the group; 9.8 years, SD=2.4, for the uncomplicated group. Finally, we hy- complicated group). Mean age at illpothesized that the differences in ness was approximately IV2 years for achievement would be accompanied both groups. The vast majority of chilby differences in cognitive functioning dren in this Canadian sample were and behavior. Based on previous re- from white, two-parent families. search, we anticipated that the differOnly four of the children (all in the ences would be selective rather than complicated group) had neurologic disglobal, and that the pattern of differ- orders persisting more than 6 weeks ences would be suggestive of factors after discharge from the hospital. One responsible for the achievement limi- child had residual paresis of the sixth tations of the complicated group (Per- cranial nerve, which resolved within
Downloaded from ldx.sagepub.com at MICHIGAN STATE UNIV LIBRARIES on February 28, 2015
632
JOURNAL OF LEARNING DISABILITIES
6 months of the illness, and the other three children had seizure disorders (only one was on anticonvulsant medications at the time of testing).
summarizing the child's recent life history.
Procedure
Social-Environmental Status
Tests administered included the Reading, Spelling, and Arithmetic subtests of the Wide Range Achievement Test-Revised (WRAT-R) (Jastak & Wilkinson, 1984), the Gilmore Oral Reading Test (Gilmore & Gilmore, 1968), six subtests from the WISC-R (Similarities, Vocabulary, Digit Span, Block Design, Object Assembly, and Coding), and a battery of neuropsychological procedures assessing language abilities, psychomotor skills, memory, attention, and abstract reasoning (see Taylor et al., 1990, for a more complete description). Tests were given in counterbalanced order across children. Behavior and school performance were evaluated by means of the Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983), the Teacher's Report Form (TRF) (Achenbach & Edelbrock, 1986), and the Vineland Adaptive Behavior Scales (Sparrow, Balla, & Ciccetti, 1984). Parents were asked to rate each child's school progress for each year of schooling on a 0-to-2 scale (poor, fair, good), and each child's average rating was computed. Family socioeconomic status (SES) was measured by means of the Hollingshead Two-Factor Index (Hollingshead, 1965). Other potential social and environmental influences on child outcomes were assessed by having the child's parent complete the Home Environment Questionnaire (Sines, 1983) and the Life Events Questionnaire (Garmezy, Tellegen, & Dervine, 1981). The former scale provides a parentbased rating of several behaviorally relevant characteristics of the home environment, including an alternative assessment of family SES. The latter scale requires parents to indicate which of 50 events have transpired in the child's life during the previous 12 months. Positive and negative events are tallied separately as a means of
Data summarizing the socialenvironmental status of the two groups are given in Table 1. According to the results of t tests, the complicated group had higher scores than the uncomplicated group on the Press for Supervision scale of the Home Environment Questionnaire. The two groups were otherwise well matched in terms of social-environmental background factors. (Due to the exploratory nature of the study and given the limited statistical power, alpha for significance testing was set at the conventional .05 level for these and other comparisons.)
Results
WISC-R
Scores
Group differences on the WISC-R subtests and on prorated Verbal, Per-
formance, and Full Scale IQ are summarized in Table 2. Verbal IQ was prorated on the basis of Similarities, Vocabulary, and Digit Span; and Performance IQ on the basis of Block Design, Object Assembly, and Coding. Full Scale IQ was derived using prorated raw score totals for the Verbal and Performance scales. According to the results of t tests, individual subtests that distinguished the two groups were Block Design, t(105)=2.4, p < .05; Object Assembly, f(105)=2.1, p
