Abstract

Traumatic brain injury (TBI) causes heterogeneous patterns of neurocognitive deficits. In an attempt to identify homogenous subgroups within this heterogeneity, cluster analysis was used to examine memory and attention abilities as measured by the Test of Memory and Learning (TOMAL) in 300 children, 150 with TBI and 150 matched nonbrain injured controls (standardization sample [SS]). Significant differences were present between the TBI and the SS groups on all TOMAL subscale and index scores, with the TBI groups performing approximately 1.3 SD below the SS. Factor analysis of the TOMAL indicated six factors that assessed various aspects of verbal and nonverbal learning and memory, as well as attention/concentration. Cluster analyses of TOMAL factor scores indicated that a four-cluster solution was optimal for the SS group, and a five-cluster solution for the TBI group. For the TBI clusters, differences were present for clinical, achievement, neurocognitive, and behavioral variables, providing some support for the validity of the cluster solution. These findings suggest that TBI results in unique patterns of neurocognitive impairment that are not accounted for by individual differences in test performance commonly observed in normal populations. Additionally, neurocognitive profiles identified using cluster analysis may prove useful for identifying homogeneous subgroups of children with TBI that are differentiated by a number of important clinical, cognitive, and behavioral variables associated with treatment and outcomes.

Introduction

Behavioral disturbances and neurocognitive deficits in attention, learning, and memory commonly occur following traumatic brain injury (TBI; Babikian & Asarnow, 2009; Roman et al., 1998; Schwartz, et al., 2003; Yeates et al., 2005). These neurocognitive deficits are associated with a number of important outcomes (Gil, 2003; Rassovsky et al., 2006). However, both the severity and patterns of neurocognitive impairment are heterogeneous. Heterogeneity of these neurocognitive and behavioral deficiencies are not reflected in more general approaches to TBI severity classification based on, for example, open and closed head injury, length of coma or post-traumatic amnesia, and persisting neurological signs (Reitan & Wolfson, 1993), and it may be that neurocognitive tests could provide more meaningful classification by identifying subgroups of children with TBI, who differ not only in the patterns of neurocognitive deficits, but also exhibit differences in other domains, such as behavioral abnormalities, academic achievement, and functional outcomes. Cluster analysis has been effectively used to identify subgroups of patients who exhibit differing patterns of neurocognitive impairment in a variety of psychiatric and neurological disorders, including TBI (Allen et al., 2000; Crawford, Garthwaite, & Johnson, 1997; Demery, Pedraza, & Hanlon, 2002; Wiegner & Donders, 1999). Studies of adults with TBI indicate the existence of discrete TBI subgroups that differ with regard to both pattern and severity of neurocognitive deficits (Chan, Hoosain, Lee, Fan, & Fong, 2003; Crosson, Greene, Roth, Farr, & Adams, 1990; Goldstein, Allen, & Caponigro, 2010; Malec, Machulda, & Smigielski, 1993; Millis & Ricker, 1994; van der Heijden & Donders, 2003). Fewer studies have examined neurocognitive heterogeneity in children with TBI, but provide evidence for similar heterogeneity as observed in adults, when the California Verbal Learning Test for Children and the Wechsler Intelligence Scale for Children are examined (Donders & Warschausky, 1997; Mottram & Donders, 2006).

Although providing substantive support for the presence of neurocognitive heterogeneity in TBI, the results of these studies are limited in a number of ways, including in some cases, the use of small sample size which limits generalizability, the use of IQ tests to develop clusters, because IQ tests have limited sensitivity to brain damage (Malec et al., 1993), and the relative absence of support for the “external validity” of cluster solutions through, for example, examination of cluster differences on neurocognitive, achievement, or behavioral variables that were not included in the cluster analysis. With regard to this latter point, associations between cluster membership and post-injury disability in TBI (Malec et al. 1993), as well as differences in neurocognitive abilities and affective symptomotology in HIV infected individuals (Murji et al., 2003), provide some evidence that neurocognitive subtypes may have utility in predicting post-injury disability, psychiatric status, and possibly other outcomes in pediatric TBI. A further limitation is that there has not been a direct comparison between clusters derived in TBI and nonbrain injured samples in order to establish whether TBI results in different profiles, even though the importance of such comparisons has been noted for many years (Crosson et al., 1990). Separate cluster analysis of TBI and standardization sample (SS) data for measures such as the CVLT and Wechsler Intelligence scales (Donders, 1996, 1999, 2008; Donders, Zhu, & Tulsky, 2001) provide support for such differences, but direct comparisons between the groups are made difficult due to differences among the TBI and normal groups in sample size, demographic variables, as well as other considerations. Additionally, there has been a tendency for cluster analytic studies to focus on a single ability (e.g., verbal memory), which constrains the identification of clusters that might reflect meaningful subtypes within TBI, because TBI may result in a variety of neurocognitive deficits. Finally, in some studies, cluster analysis is used to investigate the psychometric properties of specific tests by, for example, determining base rates of score profiles for a particular instrument in standardization samples. Although these studies provide valuable psychometric information about specific tests, they do not address broader questions, such as whether there are common neurocognitive profiles in children with TBI that identify TBI subgroups/subtypes, as established by external validity indicators, which in turn may predict important outcomes. Cluster analysis can be used as a taxometric procedure to address such questions in children with TBI, in order to develop a classification system that may prove useful in understanding not only the common types of neurocognitive profiles that result from TBI, but also cognitive, clinical, and behavioral abnormalities that characterize each neurocognitive subgroup. Such classification may, in turn, prove useful for predicting prognosis and outcomes, prescribing treatment, and directing rehabilitation. On balance, the existing studies do provide a strong base from which to hypothesize that meaningful subgroups can be identified in pediatric TBI using neuropsychological tests, even though the validity of these neurocognitive subgroups has not been examined extensively.

The purpose of the current study was to further investigate neurocognitive heterogeneity in pediatric TBI. We were particularly interested in providing a broader assessment of cognitive abilities than accomplished by using tests of a single ability and the Test of Memory and Learning (TOMAL; Reynolds & Bigler, 1994a, 1994b) for this purpose. The TOMAL includes tests of verbal and nonverbal memory, immediate and delayed recall, as well attention/concentration. Because attention and memory deficits are common sequelae of childhood TBI (Babikian & Asarnow, 2009), and memory deficits are predictive of educational needs and persists even after extended periods of recovery (Miller & Donders, 2003; Yeates et al., 2002), the TOMAL should be an excellent tool for examination of TBI subtypes. A second goal was to determine whether differing patterns of learning, memory, and attention/concentration would be evident between our sample of children with TBI and a normal control sample. A third goal was to determine if the clusters identified in the TBI group differed from each other on important clinical, cognitive, and behavioral variables, in a manner that might be anticipated based on prior research, as such differences would provide further support for cluster validity.

With regard to number of clusters, based on prior cluster analyses, we hypothesized that cluster analysis of the TOMAL in children with TBI would result in at least four clusters, two that were differentiated by the level of performance (average/low average and impaired) and two additional clusters differentiated by relative strengths in either verbal or nonverbal memory abilities. Also, because the TOMAL has an Attention/Concentration index (ACI), and given the presence of attention deficits in TBI, a fifth cluster might also emerge that exhibited selective impairment in this domain. For the normal comparison sample selected from the TOMAL SS, four clusters were predicted that would be differentiated primarily by the level of performance (superior, high average, average, low average).

With regard to the external validity of the TBI clusters, comparison among the clusters on demographic, clinical, achievement, IQ, neuropsychological, and behavioral variables were made in order to determine if the clusters identified in the TBI group differed in a manner that was consistent with prior research. We did not predict differences between the clusters on demographic variables, although it was hypothesized that the most severely impaired cluster would have lower Glasgow Coma Scale (GCS) scores, a shorter time interval between injury and assessment (Mottram & Donders, 2006), and a younger age at injury compared with the average cluster. We hypothesized younger age at injury in the impaired cluster because longitudinal studies suggest that earlier age of injury is associated with poorer outcomes including slower development of attention, motor, and verbal and spatial skills (Anderson, Catroppa, Morse, Haritou, & Rosenfeld, 2000; Anderson, Catroppa, Rosenfeld, Haritou, & Morse, 2000; Anderson & Moore, 1995; Anderson, Morse, Catroppa, Haritou, & Rosenfeld, 2004; Kriel, Krach, & Panser, 1989; Thompson et al., 1994). Significant differences between the clusters were also hypothesized for intellectual variables (Mottram & Donders, 2006), with perceptual organization abilities and speed of information processing demonstrating the largest differences among the clusters, and associated with overall levels of neurocognitive impairment. Neurocognitive variables assessing attention, perceptual organization, executive, language, and motor performance were predicted to exhibit differential patterns of impairment that were consistent with domain- and modality-specific patterns identified in the clusters. For example, individuals exhibiting selective deficits in verbal memory were predicted to also exhibit relative impairment on tests of verbal and language abilities, whereas nonverbal memory impairment was expected to exhibit associations with visuospatial and perceptual processing deficits (Allen, Thaler, Donohue, & Mayfield, 2010). We have reported preliminary evidence supporting generalized decrements across achievement test scores in children with TBI (Stolberg, Terranova, Jones, Allen, & Mayfield, 2009) and thus anticipated overall level of performance differences between clusters, with the most impaired cluster exhibiting the lowest achievement test scores, although differential patterns of impairment among the clusters may also be present. Finally, given that difficulties with social behavioral regulation are common following child TBI (Dennis, Guger, Roncadin, Barnes, & Schachar, 2001; Fay et al., 2009; Hawley, 2003; McKinlay, Grace, Horwood, Fergusson, & MacFarlane, 2009; Yeates et al., 2004) and that these problems are associated with severity of TBI (e.g., Dennis et al., 2001), we predicted that severity of impairment as indicated by cluster membership would be associated with attention, hyperactivity, and conduct problems in school and at home, as well as increased levels of self-reported psychological distress.

Materials and Methods

Participants

Participants included 300 children. Of these, 150 had sustained TBI and were selected from a consecutive series of cases referred for neuropsychological evaluation (TBI group). Definitive evidence of brain damage resulting from TBI was established in all cases using appropriate neuroimaging, laboratory, and examinational findings. GCS scores were available for 101 of the participants and indicated that on average, injuries were severe in nature (median = 7). Participants were included if they had a primary diagnosis of TBI, had been administered the TOMAL as part of a comprehensive neuropsychological evaluation, and did not have pre-injury neurological or neurodevelopmental disorders. Of the 150 participants selected, 56.3% were Caucasian, 23.4% were African American, 17.2% were Hispanic, 0.8% were Asian American, and 2.3% were of other ethnicities. The sample was 52.1% male with an average age of 11.7 years (SD = 3.7; range = 5–19). They were assessed an average of 6.9 months (SD = 3.1; range = 3–21) following injury. Ten (6%) participants had open head injuries, whereas the others had closed head injuries. The remaining 150 children (SS group) were selected from the TOMAL standardization smaple. SS cases were matched individually to the TBI participants on age and sex. When more than one member of the SS group matched a TBI participant on age and sex, random selection was used to choose the SS case. For 17 TBI cases, matches were not available, so these cases were randomly selected from the remaining SS cases. Participants in the SS group were on average 11.2 years old (SD = 3.1; range = 5–18) and 47.4% were male.

Measures

Test of Memory and Learning (Reynolds & Bigler, 1994a)

The TOMAL was developed for children ages 5–19 years and is composed of 14 subtests, 10 that are core and 4 that are supplementary. Core subtests include Memory for Stories, Facial Memory, Word Selective Reminding, Visual Selective Reminding, Object Recall, Abstract Visual Memory, Digits Forward, Visual Sequential Memory, Paired Recall, and Memory for Location. Supplementary subtests include Letters Forward, Digits Backward, Letters Backward, and Manual Imitation. Subtest scores are used to calculate a number of core index scores, including a Verbal Memory, Nonverbal Memory, Composite Memory, Delayed Recall, and Attention/Concentration Indexes. Index scores are standard scores with means of 100 and standard deviations of 15.

Demographic and clinical information

In addition to demographic variables, severity of injury was measured by the GSC (Teasdale & Jennett, 1974). The GCS was administered either by first responders at the scene of the accident or in the emergency room when the child was transported to the hospital. The time interval between injury and assessment was used as an indicator of recovery, with a longer interval indicating more recovery.

Intellectual, achievement, and neurocognitive measures

Wechsler Intelligence Scales (WIS)

The WISs were used to assess intelligence. Because children were tested over a period of more than 7 years and were of different ages, some were administered the WISC-III (n = 116) or WISC-IV (n = 10), and others the Wechsler Adult Intelligence Scale-Revised (n = 9), or Wechsler Adult Intelligence Scale-Third Edition (n = 7). Given that versions of the Wechsler Scales share many common subtests, and subtests are designed to measure similar abilities across age groups, data were combined across the various versions. We analyzed individual subtests that were strong measures of their representative index scores (Reynolds & Ford, 1994), including Vocabulary (Verbal Comprehension Index), Block Design (Perceptual Organization Index), Digit Span (Working Memory Index), and Digit Symbol/Coding (Processing Speed Index). We also examined group differences on Full-Scale IQ.

The Woodcock–Johnson Psychoeducational Battery Tests of Achievement (Woodcock & Johnson, 1989)

Academic achievement was evaluated using the Revised or Third version of the Woodcock–Johnson Psychoeducational Battery Tests of Achievement. Broad Reading and Broad Math cluster scores were selected for analysis because these were completed by most participants and reflect two of the major components assessed by the tests.

Grooved Peg Board Test (Tiffin, 1948)

The Grooved Peg Board Test is a widely used measure of motor speed and dexterity. Scores included in the analyses were the time taken to place all of the pegs with the dominant and nondominant hands.

Beery–Buktenica Developmental Test of Visual-Motor Integration (Fifth Edition)

The Beery–Buktenica Developmental Test of Visual-Motor Integration (Beery & Beery, 2004) is a paper and pencil test of visual perception and visual-motor integration that takes approximately 10–15 min to administer and is comprised of 30 figures of increasing complexity that the test subject is instructed to copy.

Trail Making Test Parts A and B Intermediate Version (Reitan & Wolfson, 1993)

The Trail Making Test parts A and B were used to assess psychomotor speed, visual search and sequencing, and mental flexibility (Part B).

Conners' Continuous Performance Test II (Conners & MHS Staff, 2000)

The Conners' Continuous Performance Test II (CPT-II; Conners & MHS Staff, 2000) is a computer-administered task comprised of a series of letters presented intermittently on the computer screen, with time intervals of varying lengths occurring between the letters. Test subjects were asked to click the mouse when a letter appeared that was not the letter “X.”

The Oral and Written Language Scales (Carrow-Woolfolk, 1996)

The Oral and Written Language Scale (OWLS) is an individually administered assessment of receptive and expressive language for ages 3 through 21 years. The OWLS is comprised of three co-normed scales, including Listening Comprehension (LC), Oral Expression, and Written Expression. The LC scale was used in this study as a measure of receptive language.

The Behavior Assessment System for Children (Reynolds & Kamphaus, 1992)

The Behavior Assessment System for Children (BASC; Reynolds & Kamphaus, 1992) Parent Rating Scale (PRS), Teacher Rating Scale (TRS), and Self-Report of Personality (SRP) scales were used to assess adaptive and clinical (maladaptive) behaviors. Analyses focused on TRS, PRS, and SRP composite scores, but given the literature suggesting the presence of attention deficits, impulsivity, conduct problems, and learning difficulties following TBI, we examined these domains as assessed by individual BASC subscales.

Procedure and data analysis

Tests were administered according to standardized procedures by a pediatric neuropsychologist or doctoral level technicians who were supervised by the pediatric neuropsychologist. The TOMAL was routinely administered to all children, unless there was a reason not to administer the test (referral question, age of child, etc.). The BASC was completed by either the parent (PRS), child with TBI (SRP), or teacher (TRS).

To establish the sensitivity of the TOMAL to TBI, SS and TBI between group comparisons were made using multivariate analysis of variance (MANOVA) for the TOMAL subtest and Index scores, with post hoc univariate analyses used to examine specific subtest and composite scores differences between the groups when overall MANOVAs were significant. Principal component analysis (PCA) with varimax rotation was then used to identify groupings of TOMAL subtests to be used in the cluster analysis. PCA was conducted because prior investigations of the TOMAL have produced mixed results with regard to the number of factors (compare Alexander & Mayfield, 2005; Reynolds & Bigler, 1994b, 1996; and Thaler, Allen, McMurray, & Mayfield, 2010). The Kaiser–Guttman criterion was used to select the number of factors. Factor scores rather than TOMAL Index scores were used in the cluster analysis, because index scores are not independent of each other, as some TOMAL subtests contribute to more than one index score. Factor scores are also more reliable than individual subtest scores. Factor scores were calculated by deriving the mean of the standard scores for each factor's respective subtests (e.g., Allen, Strauss, Kemtes, & Goldstein, 2007).

Cluster analysis was then used to examine neurocognitive heterogeneity in the SS and TBI groups. We analyzed the data with four- and five-cluster solutions to determine which solution was optimal. TBI and SS groups were analyzed separately to determine optimal cluster solutions for each group. Ward's method of cluster analysis was used because it is consistent with the cluster analytic methodology of previous studies (Curtiss, Vanderploeg, Spencer, & Salazar, 2001; Mottram & Donders, 2006), it produces results that are consistent with other agglomerative clustering methods, and is less affected by outliers compared with those methods (Morris, Blashfield, & Satz, 1981). We used squared Euclidian distance as the dissimilarity measure in the analyses because it provides a direct, physical measure of distance in Euclidean space (Everitt, Landau, & Leese, 2001) and it is sensitive to pattern and level of performance differences among individuals. To identify the appropriate number of clusters, we followed the approach suggested by Aldenderfer and Blashfield (1984). Stability of the cluster solutions derived using Ward's method were examined using K-means iterative partitioning cluster analysis. For K-means, the number of clusters and initial cluster centers were specified based on Ward's method results. K-means derives clusters by iteratively adding cases to the group that has the mean value closest to the unassigned data point (Hair, Black, Babin, Anderson, & Tatham, 1992). Cohen's kappa was then used to assess the extent to which the K-means and Ward's method solutions agreed. On the basis of these various methods, along with consideration of the a priori hypotheses regarding number and type of clusters, the most appropriate solutions were identified for the SS and the TBI groups.

External validity of the optimal TBI cluster solution was then examined by comparisons on variables that were not included in the cluster analysis, but were theoretically and clinically relevant to TBI, including demographic, clinical, IQ, achievement, neurocognitive, and behavioral variables. In cases where interaction effects were hypothesized (e.g., IQ and neurocognitive variables), mixed-model ANOVAs were used to examine cluster differences, in which cluster membership served as a between-subjects variable and test scores as a within-subjects variable. When no interaction effects were predicted, chi-squared, univariate ANOVAs, or multivariate ANOVAs were used (as appropriate) to examine cluster differences.

Results

There were no significant differences between the TBI and the SS groups on age or sex, χ2(1) = 0.66, p = .42, indicating that the group selection and matching procedure was successful. MANOVAs indicated significant differences between the SS and the TBI sample for TOMAL subtest scores, F(68, 528) = 5.59, p < .001, forumla and Index scores, F(68, 528) = 12.4, p < .001, forumla Following the Bonferonni correction for multiple comparisons, significant differences were present between the TBI and the SS groups on all of the TOMAL scores. The SS group's performance on the TOMAL subtest and index scores approximated the SS mean, and the TBI group performed approximately 1.3 SD below the SS group on most TOMAL scores, suggesting that the TOMAL was sensitive to TBI (detailed results of these analyses may be obtained from the corresponding author).

Principal Component Analysis

PCA conducted for the SS group indicated six factors based on the Kaiser–Guttman criteria. This solution approached simple structure, with most TOMAL subtests exhibiting primary loadings greater than 0.70, and few cross loadings greater than 0.40. The six factors were interpretable and assessed attention/concentration, and verbal and nonverbal memory processes, with distinctions between the verbal and the nonverbal factors based to some extent on task requirements. For the TBI group, the Kaiser–Guttman criteria indicated four factors. However, the four-factor solution did not approach simple structure and some factors that were not interpretable. When a six-factor solution was specified for the TBI group, a solution comparable to the SS group was obtained that approached simple structure. A final PCA conducted for the combined samples generally improved the six-factor solution (Table 1) and is consistent with the original conceptualization of the TOMAL as a measure of verbal and nonverbal memory processes, and attention/concentration (Reynolds & Bigler, 1994a, 1996). On the basis of these PCA results, six-factor scores were derived and subsequently used in the cluster analyses (separate PCA results for the TBI and the SS groups may be obtained from the corresponding author).

Table 1.

Factor structure of the TOMAL combined SS and TBI groups (N = 300)

TOMAL Factors
 
 
LF 0.83 0.12 0.13 0.15 0.16 0.10 
DF 0.82 0.06 0.18 0.09 0.06 0.14 
LB 0.76 0.32 0.07 0.11 0.00 0.17 
DB 0.75 0.23 0.08 0.13 0.06 0.12 
VSM 0.48 0.11 0.28 0.35 0.22 0.04 
WSR 0.28 0.73 0.21 0.19 0.15 0.12 
PR 0.16 0.74 0.23 0.02 0.04 0.24 
OR 0.38 0.49 0.31 0.26 0.27 0.21 
WSRD 0.14 0.78 0.10 0.26 0.12 -0.05 
MFSD 0.19 0.27 0.86 0.15 0.16 0.08 
MFS 0.19 0.22 0.90 0.07 0.07 0.16 
VSRD 0.11 0.22 0.11 0.88 0.06 0.15 
VSR 0.37 0.20 0.10 0.76 0.07 0.20 
FM 0.19 0.18 0.26 0.21 0.72 0.10 
FMD 0.05 0.08 0.01 −0.03 0.88 0.03 
AVM 0.24 0.12 0.31 0.19 0.31 0.60 
MFL 0.21 0.14 0.06 0.16 −0.02 0.85 
Eigenvalue 7.1 1.5 1.2 1.1 1.0 0.8 
% Variance 41.6 8.9 7.0 6.4 5.8 4.7 
Chronbach's α 0.86 0.82 0.92 0.74 0.62 0.56 
TOMAL Factors
 
 
LF 0.83 0.12 0.13 0.15 0.16 0.10 
DF 0.82 0.06 0.18 0.09 0.06 0.14 
LB 0.76 0.32 0.07 0.11 0.00 0.17 
DB 0.75 0.23 0.08 0.13 0.06 0.12 
VSM 0.48 0.11 0.28 0.35 0.22 0.04 
WSR 0.28 0.73 0.21 0.19 0.15 0.12 
PR 0.16 0.74 0.23 0.02 0.04 0.24 
OR 0.38 0.49 0.31 0.26 0.27 0.21 
WSRD 0.14 0.78 0.10 0.26 0.12 -0.05 
MFSD 0.19 0.27 0.86 0.15 0.16 0.08 
MFS 0.19 0.22 0.90 0.07 0.07 0.16 
VSRD 0.11 0.22 0.11 0.88 0.06 0.15 
VSR 0.37 0.20 0.10 0.76 0.07 0.20 
FM 0.19 0.18 0.26 0.21 0.72 0.10 
FMD 0.05 0.08 0.01 −0.03 0.88 0.03 
AVM 0.24 0.12 0.31 0.19 0.31 0.60 
MFL 0.21 0.14 0.06 0.16 −0.02 0.85 
Eigenvalue 7.1 1.5 1.2 1.1 1.0 0.8 
% Variance 41.6 8.9 7.0 6.4 5.8 4.7 
Chronbach's α 0.86 0.82 0.92 0.74 0.62 0.56 

Notes: TOMAL = Test of Learning and Memory; SS = Standardization sample group; TBI = Traumatic brain injury group; LF = Letters Forward; DF = Digits Forward; LB = Letters Backwards; DB = Digits Backwards; VSM = Visual Sequential Memory; WSR = Word Selective Reminding; PR = Paired Recall; OR = Object Recall; WSRD = Word Selective Reminding Delayed; MFS = Memory for Stories; MFSD = Memory for Stories Delayed; VSRD = Visual Selective Reminding Delayed; VSR = Visual Selective Reminding; FM = Facial Memory; FMD = Facial Memory Delayed; AVM = Abstract Visual Memory; MFL = Memory for Location.

Cluster Analyses for TBI and SS Groups

Cluster analysis of the TOMAL subtests was accomplished specifying four- and five-cluster solutions for the TBI and the SS groups. For both groups, a preliminary graphical inspection of the cluster solutions based on the factor scores was not particularly informative because the number of factors made interpretation difficult, so Index score profiles were examined. Factor score profiles were largely consistent with the Index score profiles, although Index score profiles were easier to interpret and allowed for examination of profiles as they would occur in clinical settings (i.e., clinicians will not typically interpret factor scores).

Examination of index score profiles for the SS cluster solutions indicated that the clusters were differentiated primarily by the level of performance. The four-cluster solution consisted of a low average (C1), average (C2), and two above average clusters, one that obtained scores slightly above 100 (C3), and another with superior performance on the ACI (mean = 120.5) as well as elevated scores on the Verbal and Nonverbal Memory Indexes (C4; Fig. 1, Frame A). The five-cluster solution further divided the average cluster (C2) into two clusters, which did not markedly differ from each other. Thus, preliminary inspection of the cluster solutions indicated that the four-cluster solution was the better of the two solutions for the SS group.

Fig. 1.

Cluster profiles of TOMAL Indexes for the standardization sample group four-cluster solution (Frame A), and TBI group five-cluster solution (Frame B). Frame A: SS group; Frame B: TBI group. VMI = Verbal Memory Index; NMI = Non-verbal Memory Index; ACI = Attention/Concentration Index; CMI = Composite Memory Index; DRI = Delayed Memory Index.

Fig. 1.

Cluster profiles of TOMAL Indexes for the standardization sample group four-cluster solution (Frame A), and TBI group five-cluster solution (Frame B). Frame A: SS group; Frame B: TBI group. VMI = Verbal Memory Index; NMI = Non-verbal Memory Index; ACI = Attention/Concentration Index; CMI = Composite Memory Index; DRI = Delayed Memory Index.

For the TBI group, clusters were differentiated by both level and pattern of performance. The four-cluster solution consisted of a low-scoring cluster that exhibited a relative increase on the Delayed Recall Index (DRI) and ACI, a middle cluster exhibiting a relatively lower score on the VMI, an average cluster that exhibited its highest performance on the Nonverbal Memory Index, and an average cluster that obtained its lowest Index scores on the ACI and NMI. As seen in Fig. 1, Frame B, the five-cluster solution included an Impaired cluster, that obtained Index scores ranging from 56.7 to 66.7, and obtained the highest scores on the DRI and ACI, with relatively lower scores on the VMI and NMI. There was also a Verbal cluster that obtained its lowest score on the VMI (70.8), with NMI, DRI, and ACI scores approximating 80.0. A Visual cluster is also apparent that exhibits the lowest performance on the NMI (65.6) with scores on the VMI, DRI, and ACI of approximately 80.0. Two additional clusters exhibit performance across all Indexes in the low average to average range, with the first cluster obtaining average scores on all indexes (Average cluster), and the second exhibiting relative decreases on the ACI and possibly the NMI (Attention cluster). Thus, preliminary inspection of the four- and five-cluster solutions in the TBI group suggest that the four- and five-cluster solutions provide clear indication of level and pattern of performance differences, with this being the case particularly for the five-cluster solution. The five-cluster solution also appears to be the most theoretically and clinically relevant of the solutions, based on the a priori hypothesis.

Stability of the four- and five-cluster solutions examined using the K-means iterative partitioning method indicated that the Cohen's kappas for the SS group were 0.86 and 0.77, respectively, and 0.78 and 0.87, respectively, for the TBI group. Kappas at or above a 0.80 level are considered to indicate excellent agreement (Landis & Koch, 1977) and indicate that four-cluster SS solution and five-cluster TBI solution are stable and have a high-level agreement when derived using Ward's and K-means methods of cluster analysis. Therefore, the five-cluster solution appeared optimal for TBI, and the four-cluster solution appeared optimal for SS.

TOMAL Index score profiles for the four-cluster SS solution and five-cluster TBI solutions are presented in Fig. 1. In the figure, the Verbal Memory, Nonverbal Memory, and Attention/Concentration Indexes are presented first, followed by the Composite and Delayed Recall Memory Indexes, as these latter indexes combined verbal, nonverbal, and attention/concentration subscale scores, and so are not useful in determining group differences across discrete abilities. Comparison of the SS and TBI cluster solutions indicated not only overall differences in the level of performance, with the SS clusters generally obtaining higher scores than the TBI clusters, but also that the TBI clusters were differentiated by both the level and the pattern of performance, whereas the SS clusters tended to be differentiated by the level of performance, with pattern differences in the SS group mostly determined by DRI scores. In contrast, for the TBI group, there were verbal and nonverbal memory discrepancies among the clusters, as well as a cluster that exhibited relative impairment in the attention domain (figures plotting the factor scores for all cluster solutions may be obtained from the corresponding author).

Given that the five-cluster solution for the TBI sample was most consistent with the hypothesis, in combination with its increased stability and theoretical interest, we further evaluated the external validity of the five-cluster solution in the TBI group. Descriptive statistics for the TBI five-cluster solution on the TOMAL subtest and Index scores are presented in Table 2 (tables containing descriptive statistics for the TOMAL subtests for all TBI and the SS cluster solutions may be obtained from the corresponding author).

Table 2.

Demographic, Clinical, and TOMAL subtest and Index scores for the five-cluster solution in the TBI group

Variables Clusters
 
 Average (n = 38)
 
Attention (n = 28)
 
Visual (n = 14)
 
Verbal (n = 54)
 
Impaired (n = 16)
 
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Age (years) 12.9 3.6 11.3 3.5 11.1 4.0 12.1 2.9 11.4 3.2 
Age at Injury (years) 12.8 3.6 11.2 3.4 10.9 3.9 11.8 3.0 11.3 3.2 
Time Elapsed (months) 7.0 2.5 6.3 2.4 7.0 3.5 8.2 5.0 6.9 2.2 
GCS 7.5 3.2 8.5 3.5 6.1 2.4 6.2 2.8 6.0 2.1 
Gender (% men) 55.3 53.6 61.1 42.9 68.8 
TBI Type (% closed) 92.6 97.4 92.9 93.8 89.3 
TOMAL Subtests 
 MFS 8.9 2.7 11.6 1.6 9.2 1.5 6.1 2.0 5.4 2.2 
 WSR 10.2 2.4 10.2 2.6 6.7 2.5 6.5 2.9 3.6 2.4 
 OR 8.5 3.3 7.7 2.1 5.1 2.0 4.2 2.3 1.9 1.7 
 DF 7.8 3.0 7.0 2.2 7.0 2.2 6.4 2.5 4.3 1.2 
 PR 10.4 2.4 9.3 3.2 7.7 3.1 5.5 3.2 3.4 3.0 
 LF 8.1 3.0 7.0 1.8 6.3 2.2 6.5 2.6 4.2 2.2 
 DB 10.0 2.4 8.5 1.8 8.3 1.6 7.5 1.7 5.8 1.8 
 LB 9.7 2.8 8.5 1.6 7.6 1.3 7.1 2.3 5.3 2.0 
 FM 9.7 2.7 8.5 2.7 6.4 2.7 6.5 2.4 5.5 2.4 
 VSR 8.4 3.1 6.9 2.3 3.9 1.9 6.7 2.9 2.7 1.4 
 AVM 10.5 2.7 8.3 2.2 3.4 2.0 6.9 2.6 3.3 2.6 
 VSM 9.9 2.6 9.4 1.8 8.0 2.8 7.4 2.6 5.6 1.7 
 MFL 12.1 3.4 7.5 3.0 3.3 2.0 6.8 3.4 4.1 3.1 
 MFSD 7.6 2.8 10.3 1.5 7.9 1.7 4.6 2.5 3.2 1.8 
 FMD 10.3 1.7 9.0 2.6 7.9 2.9 8.5 2.5 7.8 1.5 
 WSRD 10.0 2.2 9.4 1.8 7.4 2.3 6.6 2.5 3.9 1.7 
 VSRD 9.9 1.4 9.1 2.1 6.4 1.3 8.2 2.2 5.7 1.7 
TOMAL Indexes           
 VMI 94.0 11.7 94.0 8.3 80.4 7.9 70.8 8.3 56.7 7.7 
 NMI 100.7 8.6 86.9 6.8 65.6 8.4 78.6 8.5 60.6 7.4 
 CMI 97.4 8.4 90.2 5.3 71.9 8.1 73.6 6.6 56.9 5.0 
 DRI 96.2 8.8 96.2 8.1 82.6 9.7 79.7 8.5 67.7 5.5 
 ACI 92.6 13.5 84.5 7.9 82.8 10.2 79.3 11.8 66.5 5.8 
Variables Clusters
 
 Average (n = 38)
 
Attention (n = 28)
 
Visual (n = 14)
 
Verbal (n = 54)
 
Impaired (n = 16)
 
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Age (years) 12.9 3.6 11.3 3.5 11.1 4.0 12.1 2.9 11.4 3.2 
Age at Injury (years) 12.8 3.6 11.2 3.4 10.9 3.9 11.8 3.0 11.3 3.2 
Time Elapsed (months) 7.0 2.5 6.3 2.4 7.0 3.5 8.2 5.0 6.9 2.2 
GCS 7.5 3.2 8.5 3.5 6.1 2.4 6.2 2.8 6.0 2.1 
Gender (% men) 55.3 53.6 61.1 42.9 68.8 
TBI Type (% closed) 92.6 97.4 92.9 93.8 89.3 
TOMAL Subtests 
 MFS 8.9 2.7 11.6 1.6 9.2 1.5 6.1 2.0 5.4 2.2 
 WSR 10.2 2.4 10.2 2.6 6.7 2.5 6.5 2.9 3.6 2.4 
 OR 8.5 3.3 7.7 2.1 5.1 2.0 4.2 2.3 1.9 1.7 
 DF 7.8 3.0 7.0 2.2 7.0 2.2 6.4 2.5 4.3 1.2 
 PR 10.4 2.4 9.3 3.2 7.7 3.1 5.5 3.2 3.4 3.0 
 LF 8.1 3.0 7.0 1.8 6.3 2.2 6.5 2.6 4.2 2.2 
 DB 10.0 2.4 8.5 1.8 8.3 1.6 7.5 1.7 5.8 1.8 
 LB 9.7 2.8 8.5 1.6 7.6 1.3 7.1 2.3 5.3 2.0 
 FM 9.7 2.7 8.5 2.7 6.4 2.7 6.5 2.4 5.5 2.4 
 VSR 8.4 3.1 6.9 2.3 3.9 1.9 6.7 2.9 2.7 1.4 
 AVM 10.5 2.7 8.3 2.2 3.4 2.0 6.9 2.6 3.3 2.6 
 VSM 9.9 2.6 9.4 1.8 8.0 2.8 7.4 2.6 5.6 1.7 
 MFL 12.1 3.4 7.5 3.0 3.3 2.0 6.8 3.4 4.1 3.1 
 MFSD 7.6 2.8 10.3 1.5 7.9 1.7 4.6 2.5 3.2 1.8 
 FMD 10.3 1.7 9.0 2.6 7.9 2.9 8.5 2.5 7.8 1.5 
 WSRD 10.0 2.2 9.4 1.8 7.4 2.3 6.6 2.5 3.9 1.7 
 VSRD 9.9 1.4 9.1 2.1 6.4 1.3 8.2 2.2 5.7 1.7 
TOMAL Indexes           
 VMI 94.0 11.7 94.0 8.3 80.4 7.9 70.8 8.3 56.7 7.7 
 NMI 100.7 8.6 86.9 6.8 65.6 8.4 78.6 8.5 60.6 7.4 
 CMI 97.4 8.4 90.2 5.3 71.9 8.1 73.6 6.6 56.9 5.0 
 DRI 96.2 8.8 96.2 8.1 82.6 9.7 79.7 8.5 67.7 5.5 
 ACI 92.6 13.5 84.5 7.9 82.8 10.2 79.3 11.8 66.5 5.8 

Notes: TOMAL = Test of Learning and Memory; TBI = Traumatic brain injury group; GCS = Glasgow Coma Scale; MFS = Memory for Stories; WSR = Word Selective Reminding; OR = Object Recall; DF = Digits Forward; PR = Paired Recall; LF = Letters Forward; DB = Digits Backwards; LB = Letters Backwards; FM = Facial Memory; VSR = Visual Selective Reminding; AVM = Abstract Visual Memory; VSM = Visual Sequential Memory; MFL = Memory for Location; MFSD = Memory for Stories Delayed; FMD = Facial Memory Delayed; WSRD = Word Selective Reminding Delayed; VSRD = Visual Selective Reminding Delayed; VMI = Verbal Memory Index; NMI = Non-Verbal Memory Index; CMI = Composite Memory Index; DRI = Delayed Recall Index; ACI = Attention/Concentration Index.

External Validity of the TBI Clusters

Demographic and clinical differences among the TBI clusters

Descriptive statistics for the demographic and clinical variables are presented in Table 2. Chi-square analyses indicated no significant differences among the five clusters with regard to gender, χ2(4) = 2.60, p = .63, open or closed TBI, χ2(4) = 1.79, p = .78, type of accident that caused the TBI, χ2(28) = 0.29, p = .40, or whether the children were appropriately restrained in the case of motor vehicle accidents, χ2(4) = 4.74, p = .32. One-way ANOVAs indicated that the clusters did not differ significantly with regard to age, F(4,145) = 1.66, p = .16, age at injury, F(4,145) = 1.70, p = .15, or time elapsed between TBI and assessment, F(4,145) = 0.88, p = .48. However, analysis of ranked GCS scores indicated significant differences among the groups, F(4,96) = 2.5, p < .05. Post hoc comparisons indicated that the Attention cluster had significantly higher GCS scores than the Verbal cluster, although comparable but nonsignificant differences were also present for the Impaired and Visual clusters.

IQ, achievement, and neuropsychological differences among clusters

Descriptive statistics for IQ, achievement, and neuropsychological variables are presented in Table 3, as are ANOVA results and post hoc comparisons examining differences among the clusters. For Full-Scale IQ, ANOVA indicated a significant difference between the clusters (p < .001). The Average and Attention clusters did not significantly differ from each other, but obtained significantly higher scores than the Verbal and Visual clusters, which in turn performed significantly better than the Impaired cluster. Mixed-model ANOVA examining potential differences on subtests selected from the four Wechsler Index scores indicated significant effects for Cluster, F(4,134) = 35.64, p < .001, forumla and for Wechsler subtest, F(3,402) = 8.53, p < .001, forumla as well as a significant Cluster by Wechsler subtest interaction effect, F(12,402) = 2.42, p = .005, forumla The interaction appears to be caused primarily by decrements in performance on the Digit Symbol/Coding (DC) subtest for the Average cluster, as well as diminished performance on the Vocabulary (VO) and Block Design (BD) subtests by the Visual Cluster. Significant differences among the clusters were present for each subtest, with post hoc analysis indicating a general pattern in which the Average and Attention clusters performed significantly better than the other three clusters (Table 3). Consistent with modality-specific effects, the Normal cluster performed significantly better than the Verbal and Visual clusters on the Vocabulary and Block Design subtests, respectively.

Table 3.

Cluster differences on the Wechsler IQ, the Woodcock–Johnson, and neurocognitive test performances

 Clusters
 
Total
 
F-value p-value Tukey forumla 
 Average
 
Attention
 
Verbal
 
Visual
 
Impaired
 
    
 M SD M SD M SD M SD M SD M SD     
Intelligence 
 n 37 27 47 13 15 139     
 VO 9.0 3.4 9.4 2.4 6.5 2.6 5.9 3.2 4.7 2.1 7.4 3.2 12.0 .000 AtAv, VrVsI .259 
 BD 9.8 2.7 9.0 3.0 7.8 3.1 5.0 2.8 3.0 2.0 7.8 3.5 20.0 .000 AvAtVr, VsI .361 
 DS 10.4 3.1 8.8 2.0 7.3 2.5 7.9 2.5 5.1 1.4 8.2 3.0 15.3 .000 AvAt, AtVrVs, I .312 
 DC 8.0 2.8 8.7 3.3 5.3 2.9 6.2 3.1 3.1 2.2 6.5 3.4 13.6 .000 AtAv, AvVs, VsVr, VrI .289 
 IQ (n = 146) 93.1 13.0 93.9 10.8 78.5 10.7 76.6 10.3 63.9 8.3 83.4 14.7 28.5 .000 AtN, VrVs, I .447 
Achievement 
 n 37 26 42 12 15 132     
 Math 102.6 16.5 101.6 10.9 87.8 13.7 88.1 11.8 72.3 18.5 92.9 17.4 15.9 .000 AvAt, VsVr, I .334 
 Reading 98.2 13.4 101.4 9.0 87.1 14.3 87.6 12.6 75.3 17.6 91.7 15.6 12.5 .000 AtAv, AvVsVr, I .283 
Attention 
 n 31 25 42 13 14 125     
 Commissions 47.9 7.2 50.4 11.4 49.4 11.0 50.4 8.3 50.2 10.7 49.4 9.9 0.3 .88 – .010 
 Hit Rate 51.3 9.4 50.4 15.8 50.1 19.8 48.4 16.2 45.1 21.6 49.7 16.7 0.4 .83 – .013 
 Hit Rate SE 56.5 10.0 57.5 9.8 65.3 13.9 57.1 10.7 72.3 15.5 61.5 13.2 6.2 .000 AvVsAtVr, VrI .172 
 Variability 54.2 10.2 56.0 9.3 60.1 13.3 53.0 9.4 65.6 12.9 57.7 11.9 3.6 .009 VsAvAtVr, AtVrI .106 
 Detectability 51.3 7.6 52.5 9.5 53.2 9.7 52.4 10.1 58.0 10.2 53.0 9.3 1.3 .26 – .043 
 Perseveration 56.6 16.1 53.7 11.9 64.6 24.5 53.7 12.5 70.8 20.2 60.0 19.5 3.1 .02 AtVsAvVr, AvVrI .092 
Language 
 n 36 27 50 14 14 141     
 OWLS 100.3 17.0 95.3 10.9 85.3 12.2 86.1 13.6 77.9 15.6 90.4 15.6 10.3 .000 AvAt, AtVsVr, VsVrI  
Visuopercept. 
 n 34 27 50 13 15 139     
 BVMI 90.5 12.8 88.9 10.5 87.9 13.7 79.0 11.9 78.5 13.5 86.9 13.2 3.8 .006 AvAtVr, AtVrVsI  
Motor 
 n 37 26 44 12 13 132     
 GPBD 74.1 15.0 76.2 30.0 89.3 35.0 86.0 27.7 96.3 51.6 83.1 32.1 1.9 .11 – .057 
 GPBND 80.9 22.8 83.2 26.3 99.2 46.1 114.2 55.5 131.3 74.3 96.9 46.4 4.4 .002 AtAvVrVs, VrVsI .123 
Executive 
 n 28 18 29 11 10 96     
 TMA 25.1 9.6 31.2 15.2 34.6 15.8 31.1 12.2 55.0 55.3 32.9 22.7 3.6 .009 AvVsAtVr, VrI .137 
 TMB 46.2 17.6 61.9 24.0 79.0 40.9 79.1 37.9 150.1 110.6 73.7 53.6 10.0 .000 AvAtVrVs, I .305 
 Clusters
 
Total
 
F-value p-value Tukey forumla 
 Average
 
Attention
 
Verbal
 
Visual
 
Impaired
 
    
 M SD M SD M SD M SD M SD M SD     
Intelligence 
 n 37 27 47 13 15 139     
 VO 9.0 3.4 9.4 2.4 6.5 2.6 5.9 3.2 4.7 2.1 7.4 3.2 12.0 .000 AtAv, VrVsI .259 
 BD 9.8 2.7 9.0 3.0 7.8 3.1 5.0 2.8 3.0 2.0 7.8 3.5 20.0 .000 AvAtVr, VsI .361 
 DS 10.4 3.1 8.8 2.0 7.3 2.5 7.9 2.5 5.1 1.4 8.2 3.0 15.3 .000 AvAt, AtVrVs, I .312 
 DC 8.0 2.8 8.7 3.3 5.3 2.9 6.2 3.1 3.1 2.2 6.5 3.4 13.6 .000 AtAv, AvVs, VsVr, VrI .289 
 IQ (n = 146) 93.1 13.0 93.9 10.8 78.5 10.7 76.6 10.3 63.9 8.3 83.4 14.7 28.5 .000 AtN, VrVs, I .447 
Achievement 
 n 37 26 42 12 15 132     
 Math 102.6 16.5 101.6 10.9 87.8 13.7 88.1 11.8 72.3 18.5 92.9 17.4 15.9 .000 AvAt, VsVr, I .334 
 Reading 98.2 13.4 101.4 9.0 87.1 14.3 87.6 12.6 75.3 17.6 91.7 15.6 12.5 .000 AtAv, AvVsVr, I .283 
Attention 
 n 31 25 42 13 14 125     
 Commissions 47.9 7.2 50.4 11.4 49.4 11.0 50.4 8.3 50.2 10.7 49.4 9.9 0.3 .88 – .010 
 Hit Rate 51.3 9.4 50.4 15.8 50.1 19.8 48.4 16.2 45.1 21.6 49.7 16.7 0.4 .83 – .013 
 Hit Rate SE 56.5 10.0 57.5 9.8 65.3 13.9 57.1 10.7 72.3 15.5 61.5 13.2 6.2 .000 AvVsAtVr, VrI .172 
 Variability 54.2 10.2 56.0 9.3 60.1 13.3 53.0 9.4 65.6 12.9 57.7 11.9 3.6 .009 VsAvAtVr, AtVrI .106 
 Detectability 51.3 7.6 52.5 9.5 53.2 9.7 52.4 10.1 58.0 10.2 53.0 9.3 1.3 .26 – .043 
 Perseveration 56.6 16.1 53.7 11.9 64.6 24.5 53.7 12.5 70.8 20.2 60.0 19.5 3.1 .02 AtVsAvVr, AvVrI .092 
Language 
 n 36 27 50 14 14 141     
 OWLS 100.3 17.0 95.3 10.9 85.3 12.2 86.1 13.6 77.9 15.6 90.4 15.6 10.3 .000 AvAt, AtVsVr, VsVrI  
Visuopercept. 
 n 34 27 50 13 15 139     
 BVMI 90.5 12.8 88.9 10.5 87.9 13.7 79.0 11.9 78.5 13.5 86.9 13.2 3.8 .006 AvAtVr, AtVrVsI  
Motor 
 n 37 26 44 12 13 132     
 GPBD 74.1 15.0 76.2 30.0 89.3 35.0 86.0 27.7 96.3 51.6 83.1 32.1 1.9 .11 – .057 
 GPBND 80.9 22.8 83.2 26.3 99.2 46.1 114.2 55.5 131.3 74.3 96.9 46.4 4.4 .002 AtAvVrVs, VrVsI .123 
Executive 
 n 28 18 29 11 10 96     
 TMA 25.1 9.6 31.2 15.2 34.6 15.8 31.1 12.2 55.0 55.3 32.9 22.7 3.6 .009 AvVsAtVr, VrI .137 
 TMB 46.2 17.6 61.9 24.0 79.0 40.9 79.1 37.9 150.1 110.6 73.7 53.6 10.0 .000 AvAtVrVs, I .305 

Notes: IQ = Wechsler Full-Scale IQ; VO = Wechsler Vocabulary; BD = Wechsler Block Design; DS = Wechsler Digit Span; DC = Wechsler Digit Symbol/Coding; Reading = Woodcock Johnson Broad Reading; Math = Woodcock Johnson Broad Math; CPT = Continuous Performance Test; Commissions = CPT Commissions; Hit Rate = CPT Hit Rate; Hit Rates SE = CPT Hit Rate Standard Error; Variability = CPT Variability of the Standard Errors; Detectability = CPT Detectability; Perseverations = CPT Perseverations; OWLS = Oral and Written Language Scale; BVMI = Beery–Buktenica Developmental Test of Visual-Motor Integration-5th Edition; GPBD = Grooved Peg Board dominant hand; GPBND = Grooved Peg Board nondominant hand; TMA = Trail Making Test Part A; TMB = Trail Making Test Part B; Av = Average Cluster; At = Attention Cluster; Vr = Verbal Cluster; Vs = Visual Cluster; I = Impaired Cluster.

For the Woodcock–Johnson achievement test data, mixed-model ANOVA indicated a significant effect for Cluster, F(4,127) = 17.50, p < .001, forumla but not for WJ Composite Score, F(1,127) = 0.08, p = .78, forumla or for the Cluster by WJ Composite Score interaction effect, F(4,127) = 1.19, p = .32, forumla Post hoc comparisons indicated the level of performance differences with the Normal and Attention clusters obtaining higher scores than the Visual and Verbal clusters, who in turn had higher scores than the Impaired cluster (Table 3).

For the attention domain, CPT variables were entered into the mixed-model ANOVA, and results indicated a significant main effect for Cluster, F(4,120) = 2.83, p = .03, forumla and for CPT score, F(5,600) = 22.40, p < .001, forumla as well as a significant Cluster by CPT score interaction effect, F(20,600) = 2.22, p = .002, forumla Expected differences were present between groups on a number of CPT measures, with the Impaired cluster typically performing worse than the others, with significant differences among the clusters for CPT Standard Error, Variability, and Perseverations (Table 3).

Significant differences were also present between the clusters on the visuomotor (BVMI) and language (OWLS) domains. Post hoc analyses indicated that for language, the Average cluster exhibited the best performance but did not differ from the Attention cluster, whereas the Impaired cluster exhibited the worst performance, but did not differ from the Verbal or Visual clusters. For visuomotor integration, the Average cluster again exhibited the best performance, although it did not differ from the Attention or Verbal cluster, while the Impaired cluster exhibited the worst performance, but did not differ from the Visual, Verbal, or Attention clusters.

For the motor domain, results for the Grooved Peg Board dominant and nondominant hands indicated that there were significant main effects for Cluster, F(4,127) = 3.74, p = .007, forumla and Hand, F(1,127) = 30.95, p < .001, forumla and a significant Cluster by Hand interaction effect, F(4,127) = 2.83, p = .028, forumla The dominant hand was faster than the nondominant hand in all groups, although a larger than expected discrepancy is present for the Impaired cluster, with the nondominant hand being slower than expected. Significant differences among the clusters were present for only the nondominant hand, and post hoc analyses indicated that the Normal and the Attention clusters performed similar to each other and significantly better than the Visual and Verbal clusters, who in turn performed better than the Impaired cluster.

Finally, for the executive function domain, mixed-model ANOVA indicated a significant effect for Trail, F(1,91) = 8.9, p < .001, forumla Cluster, F(1,91) = 141.7, p < .001, forumla and a significant Trail by Cluster interaction effect, F(1,91) = 8.5, p < .001, forumla As can be seen from Table 3, the interaction effect was caused primarily by the larger than expected discrepancy between Parts A and B for the Impaired cluster compared with the others. Post hoc analysis indicated that in general, the Impaired cluster differed from the other clusters on Parts A and B, although the other clusters did not differ from each other.

Behavioral differences among the TBI clusters

Descriptive statistics for the BASC subscales are presented in Table 4. Separate MANOVAs were conducted for the Parent, Teacher, and Self-report ratings, where cluster membership served as the between-subjects variable and BASC scores were the dependent variables. Additionally, scale and composite scores were analyzed separately given the overlap between composite and scale scores. MANOVA for the PRSs indicated an overall significant effect, F(12,387) = 2.95, p = .01, forumla although the MANOVA for the composite scores was not significant, F(16,528) = 1.41, p = .13, forumla The largest effect was for the Hyperactivity and Conduct scales. Post hoc analyses indicated the Impaired cluster exhibited more behavioral problems than the other clusters, who did not differ from each other.

Table 4.

Parent, teacher, and self-report behavioral reports of the five-cluster solution

BASC Scores Clusters
 
Total
 
F-value p-value Tukey forumla 
 Average
 
Attention
 
Visual
 
Verbal
 
Impaired
 
    
 Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD     
Parent Rating 
 Subscales                 
  n 37 26 14 42 15 134     
  Hyperactivity 53.5 16.2 52.4 13.4 60.4 18.9 56.7 16.0 63.9 19.7 56.2 16.5 1.7 .16 – .050 
  Attention 58.4 12.8 55.0 13.2 61.9 11.7 59.0 10.7 72.4 9.6 59.9 12.6 5.7 .000 AtAvVrVs, I .149 
  Conduct 52.8 15.5 51.2 15.1 57.7 18.7 51.1 13.6 64.8 22.5 53.8 16.5 2.4 .05 VrAtAvVs, VsI .071 
 Composites                 
  n 37 25 14 47 14 137     
  Externalizing 55.0 17.1 52.7 15.0 60.1 20.2 54.6 14.6 66.4 21.7 56.1 17.0 1.9 .11 – .055 
  Internalizing 53.9 12.0 51.9 10.9 55.9 17.2 55.1 14.2 50.6 11.5 53.8 13.1 0.5 .72 – .016 
  Behavioral 56.1 15.1 53.7 13.5 60.2 20.1 57.9 16.3 64.4 16.6 57.6 16.0 1.2 .32 – .035 
  Adaptive Skills 44.7 11.9 45.2 13.9 44.0 9.5 44.0 11.1 41.8 7.0 44.2 11.3 0.2 .93 – .007 
Teacher Rating 
 Subscales                 
  n 17 44     
  Hyperactivity 50.1 12.8 45.8 8.8 58.5 13.0 48.1 7.8 63.2 11.6 51.5 11.4 3.8 .01 AtVrAv, VsI .283 
  Attention 50.3 7.0 50.9 7.2 58.5 11.3 58.2 9.9 75.0 4.1 58.0 11.3 8.4 .000 AvAtVrVs, I .464 
  Conduct 47.7 7.3 46.4 2.8 44.5 2.3 45.7 4.8 66.0 16.7 48.8 10.0 9.5 .000 VsVrAtAv, I .493 
  Learning 47.0 4.6 50.5 10.5 58.2 8.3 56.5 8.9 79.2 7.7 57.2 12.6 13.8 .000 AvAtVsVr, I .587 
 Composites                 
  n 17 44     
  Externalizing 49.0 10.6 45.6 6.0 50.5 6.3 46.4 5.7 64.7 13.6 49.7 9.9 6.4 .000 AtVrVsAv, I .395 
  Internalizing 48.4 8.6 50.0 12.2 63.5 12.0 51.0 7.3 51.3 9.7 52.2 10.2 2.6 .052 – .209 
  School 48.6 5.4 50.6 8.3 58.5 9.7 57.7 9.5 78.3 5.8 57.9 12.1 12.6 .000 AvAtVsVr, I .564 
  Behavioral 49.7 11.6 48.0 9.5 59.5 12.3 50.9 6.3 62.8 6.4 53.0 9.9 3.8 .01 AtAvVr, I .281 
  Adaptive Skills 50.7 8.9 49.8 10.4 43.8 8.7 47.1 10.0 31.2 6.2 45.5 10.8 4.7 .004 AtAvVrVs, I .323 
Self-Report 
 Composites                 
  n 33 23 11 33 12 112     
  School 47.2 9.5 49.2 9.7 54.6 10.7 50.4 11.0 57.3 12.0 50.4 10.7 1.3 .26 – .048 
  Clinical 48.8 12.0 48.4 7.7 50.7 10.4 48.5 7.9 46.4 7.4 48.5 9.3 0.2 .93 – .011 
  Personal 50.9 11.0 51.5 7.8 49.0 10.6 46.3 11.0 46.1 7.6 49.0 10.2 1.5 .21 – .054 
  Emotional 48.0 11.1 48.5 8.7 51.9 12.3 51.1 9.1 49.3 6.8 49.5 9.7 0.6 .64 – .023 
BASC Scores Clusters
 
Total
 
F-value p-value Tukey forumla 
 Average
 
Attention
 
Visual
 
Verbal
 
Impaired
 
    
 Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD     
Parent Rating 
 Subscales                 
  n 37 26 14 42 15 134     
  Hyperactivity 53.5 16.2 52.4 13.4 60.4 18.9 56.7 16.0 63.9 19.7 56.2 16.5 1.7 .16 – .050 
  Attention 58.4 12.8 55.0 13.2 61.9 11.7 59.0 10.7 72.4 9.6 59.9 12.6 5.7 .000 AtAvVrVs, I .149 
  Conduct 52.8 15.5 51.2 15.1 57.7 18.7 51.1 13.6 64.8 22.5 53.8 16.5 2.4 .05 VrAtAvVs, VsI .071 
 Composites                 
  n 37 25 14 47 14 137     
  Externalizing 55.0 17.1 52.7 15.0 60.1 20.2 54.6 14.6 66.4 21.7 56.1 17.0 1.9 .11 – .055 
  Internalizing 53.9 12.0 51.9 10.9 55.9 17.2 55.1 14.2 50.6 11.5 53.8 13.1 0.5 .72 – .016 
  Behavioral 56.1 15.1 53.7 13.5 60.2 20.1 57.9 16.3 64.4 16.6 57.6 16.0 1.2 .32 – .035 
  Adaptive Skills 44.7 11.9 45.2 13.9 44.0 9.5 44.0 11.1 41.8 7.0 44.2 11.3 0.2 .93 – .007 
Teacher Rating 
 Subscales                 
  n 17 44     
  Hyperactivity 50.1 12.8 45.8 8.8 58.5 13.0 48.1 7.8 63.2 11.6 51.5 11.4 3.8 .01 AtVrAv, VsI .283 
  Attention 50.3 7.0 50.9 7.2 58.5 11.3 58.2 9.9 75.0 4.1 58.0 11.3 8.4 .000 AvAtVrVs, I .464 
  Conduct 47.7 7.3 46.4 2.8 44.5 2.3 45.7 4.8 66.0 16.7 48.8 10.0 9.5 .000 VsVrAtAv, I .493 
  Learning 47.0 4.6 50.5 10.5 58.2 8.3 56.5 8.9 79.2 7.7 57.2 12.6 13.8 .000 AvAtVsVr, I .587 
 Composites                 
  n 17 44     
  Externalizing 49.0 10.6 45.6 6.0 50.5 6.3 46.4 5.7 64.7 13.6 49.7 9.9 6.4 .000 AtVrVsAv, I .395 
  Internalizing 48.4 8.6 50.0 12.2 63.5 12.0 51.0 7.3 51.3 9.7 52.2 10.2 2.6 .052 – .209 
  School 48.6 5.4 50.6 8.3 58.5 9.7 57.7 9.5 78.3 5.8 57.9 12.1 12.6 .000 AvAtVsVr, I .564 
  Behavioral 49.7 11.6 48.0 9.5 59.5 12.3 50.9 6.3 62.8 6.4 53.0 9.9 3.8 .01 AtAvVr, I .281 
  Adaptive Skills 50.7 8.9 49.8 10.4 43.8 8.7 47.1 10.0 31.2 6.2 45.5 10.8 4.7 .004 AtAvVrVs, I .323 
Self-Report 
 Composites                 
  n 33 23 11 33 12 112     
  School 47.2 9.5 49.2 9.7 54.6 10.7 50.4 11.0 57.3 12.0 50.4 10.7 1.3 .26 – .048 
  Clinical 48.8 12.0 48.4 7.7 50.7 10.4 48.5 7.9 46.4 7.4 48.5 9.3 0.2 .93 – .011 
  Personal 50.9 11.0 51.5 7.8 49.0 10.6 46.3 11.0 46.1 7.6 49.0 10.2 1.5 .21 – .054 
  Emotional 48.0 11.1 48.5 8.7 51.9 12.3 51.1 9.1 49.3 6.8 49.5 9.7 0.6 .64 – .023 

Notes: Av = Average Cluster; At = Attention Cluster; Vr = Verbal Cluster; Vs = Visual Cluster; I = Impaired Cluster.

MANOVA for the Teacher scale ratings indicated an overall significant effect, F(16,156) = 3.48, p < .001, forumla as did the MANOVA for the Teacher composite scores, F(20,152) = 2.95, p < .001, forumla Significant group differences were present among the clusters on all of the scale scores, and all but one of the composite scores (internalizing problems), with the Impaired cluster again exhibiting more behavioral disturbances than the other clusters.

MANOVA for the Self-Report composite scores indicated no significant differences among the clusters, F(12,318) = 1.74, p = .06, forumla

Discussion

The current study provides a number of findings relevant to neurocognitive functioning in children with TBI, particularly with regard to heterogeneity. Most notably, results support the notion that TBI produces marked heterogeneity in neurocognitive function, so that there is no profile that best typifies these children and adolescents, although there do appear to be subgroups within this diverse population that might be characterized by performance profiles from neuropsychological tests that assess memory and attention/concentration abilities. Evidence supporting the validity of these subgroups was provided not only by comparisons on clinical, neuropsychological, and behavioral variables, but also through comparison with clusters identified in a matched normal sample. With regard to the latter point, the need for direct comparisons of TBI and normal cluster solutions has long been suggested (Crosson et al., 1990), in order to determine if there are profiles unique to TBI that represent more than normal or expected variation in test performance among individuals, and thus would be attributable to brain injury. Results indicated that the TBI clusters differed in patterns of performance which was accounted for largely by discrepancies in verbal and nonverbal memory abilities, or a relative decrease in attention/concentration. There was also evidence for a group that exhibited severe impairment, as well as another that exhibited average performance. These TBI clusters stood in contrast to the four clusters identified in the SS group, which were differentiated primarily by level of performance, with pattern of performance differences accounted for primarily by the tendency toward reduced variability in the SS group on the DRI compared with the TBI group (SD's = 8.8 and 12.9, respectively; Table 1). Thus, the current findings support the contention that TBI results in patterns of memory and attention deficits that vary markedly from those observed in normal populations and these patterns are not easily accounted for by intra-individual variation in test performance that is often observed in nonbrain injured populations. Differences in profile and number of clusters are also consistent with prior studies that have undertaken separate TBI and SS analyses on, for example, the Wechsler Intelligence Scales (e.g., compare Donders & Warschausky, 1997, and Donders, 1996).

Although the number of clusters and profile differences are strongly influenced by the measures that are entered into the cluster analysis (Malec et al., 1993), there is some consistency across studies of adults and children with moderate to severe TBI which may support the utility of neuropsychological tests in classifying injury severity. For example, most studies identify average and impaired clusters. Other studies that have examined verbal memory abilities find that some children with TBI exhibit deficits in various aspects of memory functioning and others do not. Studies of attention also identify clusters of children with selective deficits in some aspects of attention. In this study, nonverbal memory tests were included and a group was also identified with selective deficits in this area. However, one objection to the results of cluster analytic studies is that the replicability and thus generalizability of cluster solutions are not typically assessed. One study that compared cluster solutions derived from two different TBI samples that were assessed using the same tests provided strong support for the stability and generalizability of neurocognitive clusters identified across different TBI groups (Curtiss et al., 2001). Another study that investigated the stability of clusters derived using different tests (abstraction vs. IQ) in 221 patients with schizophrenia identified four clusters for each set of tests, and there was a significant degree of association between the solutions, although this association was weak (Goldstein, Allen, & Seaton, 1998). Thus, although there is support for the stability of clusters across populations and possibly across tests, test selection remains a significant determining factor in the number and profile of the obtained clusters, which highlights the need for selection of tests that are directly relevant to the types of cognitive deficits exhibited by the population under investigation. In the current study, measures of memory and attention/concentration were selected for cluster analysis because deficits in these areas are commonly reported following TBI.

With regard to the specific clusters, the Average cluster obtained TOMAL Index scores ranging from 94.0 to 100.7. We do not refer to the Average cluster as “normal” even though their performance is in the average range, because it is likely that some reduction in neurocognitive abilities resulting from TBI occurred in these children, and so they should not be considered normal. Donders and Warschausky (1997) found that their highest performing WISC-III TBI cluster had a higher socioeconomic background, which may indicate that while still average in their performance, some decline from premobid levels of functioning had occurred. A similar observation has been made for some individuals with schizophrenia who perform in the average range on neuropsychological tests (Palmer et al. 1997), but nevertheless exhibit subtle neurocognitive deficits consistent with the presence of schizophrenia (Allen, Goldstein, & Warnick, 2003). These patients have been referred to as “neuropsychologically normal” or alternatively “high cognitive functioning,” akin to terminology used to describe autism (e.g., Goldstein et al., 2008). These high cognitive functioning patients with schizophrenia are of some interest since preserved neurocognitive function is associated with better outcomes, and less severe impairment is associated with decreased neuroanatomic abnormalities (Allen et al., 2000; Wexler et al., 2009). The Average cluster in this study may also exhibit more subtle signs of brain damage, as indicated by the relative decrease in processing speed inferred by relatively poorer performance on the Digit Symbol/Coding subtest compared to the other WISC subtests (Allen et al., 2010; Donders & Janke, 2008). Whether the nomenclature of “high functioning” applied to schizophrenia and autism would also be a useful descriptor in classifying childhood TBI awaits further investigation, although it is apparent that more severe deficits have been linked to poorer outcomes in TBI, in this study and in others.

The Impaired cluster obtained TOMAL Index scores ranging between 56.7 and 67.7, and consistently performed more poorly than the other clusters on many of the IQ, neuropsychological, and achievement variables, and also exhibited more severe behavioral disturbances and poorer adaptive functioning than the other clusters. There were no marked differential patterns of impairment across the neuropsychological domains assessed as external validity variables, which is consistent with the overall picture of severe impairment. This impairment was apparent not only on neurocognitive measures but also measures of achievement, which are typically considered less susceptible to decline following TBI and other acquired brain disorders, and can be used as an indicator of premorbid function in some cases. However, the current findings suggest that for this impaired cluster performance on achievement tests is also diminished, which is consistent with a preliminary report of 170 children with TBI, who performed significantly worse on many of the WJ III scales compared to the SS mean (Stolberg et al., 2009). Elevations in parent ratings of attention and conduct problems were also apparent. Similarly, teacher ratings indicated increased problems with hyperactivity, attention, conduct and learning, as well as increased externalizing, school and behavioral problems, and decreased adaptive skills. Thus, the Impaired cluster identified in this study and exhibits disturbances across a broad range of neurocognitive, achievement, and behavioral variables and is also expected to require the most intensive treatment and rehabilitative services during recovery.

There was also a cluster that obtained its lowest score on the Attention Concentration Index, although it also exhibited a decrease in performance on the TOMAL Nonverbal Memory Index. This cluster was similar to the Average cluster in many respects, including performance on the IQ and achievement tests, as well as on tests of motor speed and executive function. Expected decrements in performance on tests of attention, particularly the CPT and Wechsler Digit Span, were not apparent. Also, expected deficits in attention on the behavioral ratings were not apparent, which may reflect a more general lack of association between neuropsychological and behavioral measures of attention (Dennis et al., 2001; Naglieri, Goldstein, Delauder, & Schwebach, 2005; Young & Gudjonsson, 2005). It did appear that compared with the Average cluster, the Attention cluster exhibited somewhat poorer performance on tests of language and visuoperceptual abilities. This later finding may be associated with the relative decrease in performance on the TOMAL Nonverbal Memory Index, which has previously been shown to have significant associations with the Perceptual Reasoning Index from the WISC-IV (Allen et al., 2010), suggesting its reliance on visuoperceptual and visual reasoning processes. However, given the lack of associated decrements in attention on neuropsychological and behavioral measures, and the decrease in the TOMAL Nonverbal Memory index, this cluster should not be viewed as a “pure” attention cluster, and its external validity is questionable.

Verbal and Nonverbal clusters were also identified, with the Verbal Cluster obtaining index scores ranging from 70.8 (Verbal Memory Index) to 78.6, and the Visual cluster obtaining index scores ranging from 65.6 (Nonverbal Memory Index) to 82.8. Verbal memory tests have been the most extensively examined in cluster analytic studies in children with TBI, and the current results provide further evidence for a cluster that is differentiated by this ability. Measures of nonverbal learning and memory have not been included in prior cluster analytic studies of children with TBI, but based on the current results there also appears to be a subgroup of children with TBI who are characterized by relatively poor performance on measures of this domain. With regard to external validity, the Verbal and Visual clusters typically performed worse than the Average and Attention clusters and better than the Impaired cluster on the neurocognitive variables. In contrasting the Verbal and Visual clusters on the IQ variables, the Verbal cluster exhibits the expected dissociation between Vocabulary and Block Design, although this pattern in less pronounced in the Visual cluster. However, the Verbal cluster did not differ from the Visual cluster on the OWLS, which is a measure of listening comprehension. Consistent with neurocognitive data, the Visual cluster scored lower on the perceptual organization domain, without any distinct discrepancies in other areas. Behaviorally, few differences were present, although the Visual cluster did exhibit elevations on BASC parent ratings of Conduct and teacher ratings of Hyperactivity, similar to those observed in the Impaired cluster. These differences between the Visual and Verbal clusters provide preliminary support for their external validity, although some differences were not as robust as might have expected.

With regard to external validity of the TBI clusters on clinical and demographic variables, similar to prior studies, we found that only severity of coma significantly differentiated clusters, and clusters were not differentiated on gender, ethnicity, cause of injury, injury-testing interval, or age at testing (Crosson et al., 1990; Donders & Warschausky, 1997). Contrary to hypotheses, age at injury and time elapsed from injury to assessment did not differ among clusters, which appears inconsistent with longitudinal studies indicating a developmental influence on how TBI effects outcome (e.g., Verger et al., 2001). Few studies have investigated associations between age or developmental level at the time of TBI and outcome, but existing longitudinal studies suggest that earlier age of injury is associated with poorer outcomes including slower development of attention, motor, and verbal and spatial skills (Anderson, Catroppa, Morse, et al., 2000, 2004; Anderson & Moore, 1995; Duval et al. 2008; Kriel et al., 1989; Thompson et al., 1994). Failure to find age differences among our clusters was likely a result of the relatively brief time between injury and assessment compared with other studies (e.g., Hawley, 2004; Millis et al., 2001). Whether clusters identified in this study will be useful in predicting longitudinal outcomes awaits future research that incorporates longer time periods between injury and assessment. However, one longitudinal study of neurocognitive, adaptive, behavioral, and academic functioning of children with moderate to severe TBI who were classified according to the longitudinal patterns of outcomes, found that those with severe TBI had an increased likelihood of experiencing poorer outcomes, functional deficits, and behavioral deterioration an average of 4 years after injury (Fay et al., 2009). It also remains to be seen whether injury severity classified using neurocognitive tests such as the TOMAL would provide better prediction of long-term outcomes than severity classification based on the GCS or other similar techniques, or whether comparable variability in course and outcome would be noted.

To our knowledge, this is the first study to examine differences in behavior among neurocognitive clusters in children and adolescents with TBI using a well-standardized multimethod behavioral assessment, despite the observation that behavioral disturbances are common following TBI (McKinlay et al., 2009; Yeates et al., 2004). The results do not generally support a profile of behavioral disturbance that is differentiated based on membership in neurocognitive clusters in most cases. Rather, severity of neurocognitive impairment was more generally associated with disturbances in behavior and poorer adaptation, with the most severely impaired neurocognitive cluster exhibiting clinical elevations on a number of the BASC subscale and composite scores. Interestingly, there were differences between the type of behavioral report, with teacher reports consistently demonstrating the most differences among clusters. Although the findings are limited by a relatively small number of teacher ratings (n = 44), results are consistent with a recent preliminary study of children with TBI using the BASC-2 that found increased teacher endorsement of externalizing and school-related problems more frequently than internalizing problems, with the greatest elevations on the hyperactivity, attention, and learning problems subscales (Thaler, Terranova, Reynolds, Mayfield, & Allen 2010). Also, although behavioral problems following pediatric TBI are typically evaluated via parent reports (Yeates & Taylor, 2006), with less research available for teacher ratings, there do appear to be inconsistencies between parent and teacher ratings (Gross, Fogg, Garvey, & Julion, 2004; Lee, Elliott, & Barbour, 1994; McAuley, 2000; McConaughy, Stanger, & Achenbach, 1992). These inconsistencies are likely influenced by contextual factors relating to setting (e.g., hyperactivity may cause less disruption in the home than in the classroom), attribution of problem behaviors to internalizing or externalizing factors (Max et al., 1998), or by differences between teachers and parents referent comparisons to normally developing children. Because of these differences, teachers may be more accurate in rating hyperactivity, whereas parents are most reliable in rating conduct disorders, oppositional behavior, and internalizing behaviors (Loeber, Green, & Lahey, 1990; Yeates & Taylor, 2006). Thus, the suggestion that teacher reports provide unique insights to childhood behavioral disorders (Randazzo, Landsverk, & Ganger, 2003; Thaler, Terranova, et al., 2010) is consistent with the current findings and highlights the importance of supplementing parental reports with those of teachers when evaluating behavioral disturbances in children with TBI.

The current study has a number of limitations including that some clusters contained a limited number of participants, which may have decreased statistical power to detect cluster differences on some external validity variables. Also, because this was a study of a convenience sample referred for clinical evaluation, there was incomplete data for some tests (e.g., BASC TRS), which may limit generalizability of findings. In a prospective investigation, alternative measures may have been selected and some domains would have been more thoroughly evaluated (e.g., executive function). Additionally, children with mild TBI were not included. Their inclusion may have produced a high average cluster that would contain those children with above average abilities premorbidly, who did not experience a substantial decrease in neurocognitive abilities as a result of TBI. Finally, while a revised version of the TOMAL has been recently published (TOMAL-2; Reynolds & Voress, 2007), similar patterns of performance are anticipated on the TOMAL-2, because no new items or subtests were introduced, and the child SS remained largely the same. Future studies may attempt to determine stability of this cluster solution across different populations of children with TBI, as well as whether neurocognitive classifications of TBI, such as the one reported here, will be more useful in predicting longer-term outcomes in comparison to currently existing and commonly used classification methods such as the GCS. However, the current results do suggest that deficits in memory and attention abilities, particularly for patients who exhibit severe impairment in these areas, are associated with disturbances across a wider range of neurocognitive and behavioral domains, and so may prove to be clinically useful in predicting outcomes, anticipating adjustment problems, and planning educational and rehabilitative interventions designed to remediate deficits and improve functional outcomes following TBI in children.

Conflict of Interest

None declared.

Acknowledgements

Standardization Sample (SS) data was from the Test of Memory and Learning (TOMAL) standardization sample. Copyright 1994 by Pro-Ed. Used with permission. All rights reserved. We thank the TOMAL publisher for allowing access to the standardization data for this study.

References

Aldenderfer
M.
Blashfield
R.
Cluster analysis.
 , 
1984
Newbury Park
Sage Publications
Alexander
A. I.
Mayfield
J.
Latent factor structure of the Test of Memory and Learning in a pediatric traumatic brain injured sample: Support for a general memory construct
Archives of Clinical Neuropsychology
 , 
2005
, vol. 
20
 (pg. 
587
-
598
)
Allen
D. N.
Goldstein
G.
Warnick
E.
A consideration of neuropsychologically normal schizophrenia
Journal of the International Neuropsychological Society
 , 
2003
, vol. 
9
 (pg. 
56
-
63
)
[PubMed]
Allen
D. N.
Seaton
B. E.
Goldstein
G.
Sanders
R. D.
Gurklis
J. A.
Jr
Peters
J. L.
, et al.  . 
Neuroanatomic differences among cognitive and symptom subtypes of schizophrenia
Journal of Nervous and Mental Disease
 , 
2000
, vol. 
188
 (pg. 
381
-
384
)
Allen
D. N.
Strauss
G. P.
Kemtes
K. A.
Goldstein
G.
Hemispheric contributions to nonverbal abstract reasoning and problem solving
Neuropsychology
 , 
2007
, vol. 
21
 (pg. 
713
-
720
)
Allen
D. N.
Thaler
N. S.
Donohue
B.
Mayfield
J.
WISC-IV profiles in children with traumatic brain injury: Similarities to and differences from the WISC-III
Psychological Assessment
 , 
2010
, vol. 
22
 
1
(pg. 
57
-
64
)
Anderson
V.
Catroppa
C.
Morse
S.
Haritou
F.
Rosenfeld
J.
Recovery of intellectual ability following TBI in childhood: Impact of injury severity and age at injury
Pediatric Neurosurgery
 , 
2000
, vol. 
32
 (pg. 
282
-
290
)
Anderson
V.
Moore
C.
Age at injury as a predictor of outcome following pediatric head injury: A longitudinal perspective
Child Neuropsychology
 , 
1995
, vol. 
1
 (pg. 
187
-
202
)
Anderson
V. A.
Catroppa
C.
Rosenfeld
J.
Haritou
F.
Morse
S. A.
Recovery of memory function following traumatic brain injury in pre-school children
Brain Injury
 , 
2000
, vol. 
14
 (pg. 
679
-
692
)
[PubMed]
Anderson
V. A.
Morse
S. A.
Catroppa
C.
Haritou
F.
Rosenfeld
J. V.
Thirty month outcome from early childhood head injury: A prospective analysis of neurobehavioural recovery
Brain: A Journal of Neurology
 , 
2004
, vol. 
127
 (pg. 
2608
-
2620
)
[PubMed]
Babikian
T.
Asarnow
R.
Neurocognitive outcomes and recovery after pediatric TBI: Meta-analytic review of the literature
Neuropsychology
 , 
2009
, vol. 
23
 (pg. 
283
-
296
)
Beery
K. E.
Beery
N. A.
The Beery–Buktenica Developmental Test of Visual-Motor Integration: Administration, scoring, and teaching manual
 , 
2004
5th ed.
Minneapolis, MN
NCS Pearson
Carrow-Woolfolk
E.
Oral and Written Language Scales.
 , 
1996
Circle Pines, MN
American Guidance Service
Chan
R. C. K.
Hoosain
R.
Lee
T. M. C.
Fan
Y. W.
Fong
D.
Are there sub-types of attentional deficits in patients with persisting post-concussive symptoms? A cluster analytical study
Brain Injury
 , 
2003
, vol. 
17
 (pg. 
131
-
148
)
Conners
C. K.
MHS
Staff
Conners' Continuous Performance Test II: Computer program for windows technical guide and software manual.
 , 
2000
North Tonwanda, NY
Multi-Health Systems
Crawford
J. R.
Garthwaite
P. H.
Johnson
D. A.
WAIS-R subtest pattern clusters in closed-head-injured and healthy samples
Clinical Neuropsychologist
 , 
1997
, vol. 
11
 (pg. 
249
-
257
)
Crosson
B.
Greene
R. L.
Roth
D. L.
Farr
S. P.
Adams
R. L.
WAIS-R pattern clusters after blunt-head injury
The Clinical Neuropsychologist
 , 
1990
, vol. 
4
 (pg. 
253
-
262
)
Curtiss
G.
Vanderploeg
R. D.
Spencer
J.
Salazar
A. M.
Patterns of verbal learning and memory in traumatic brain injury
Journal of the International Neuropsychological Society
 , 
2001
, vol. 
7
 (pg. 
574
-
585
)
Demery
J. A.
Pedraza
O.
Hanlon
R. E.
Differential Profiles of Verbal Learning in Traumatic Brain Injury
Journal of Clinical and Experimental Neuropsychology
 , 
2002
, vol. 
24
 
6
(pg. 
818
-
827
)
Dennis
M.
Guger
S.
Roncadin
C.
Barnes
M.
Schachar
R.
Attentional-inhibitory control and social-behavioral regulation after childhood closed head injury: Do biological, developmental, and recovery variables predict outcome?
Journal of the International Neuropsychological Society
 , 
2001
, vol. 
7
 (pg. 
683
-
692
)
Donders
J.
Cluster subtypes in the WISC-III standardization sample: Analysis of factor index scores
Psychological Assessment
 , 
1996
, vol. 
8
 (pg. 
312
-
318
)
Donders
J.
Cluster subtypes in the standardization sample of the California Verbal Learning Test—Children's version
Developmental Neuropsychology
 , 
1999
, vol. 
16
 (pg. 
163
-
175
)
Donders
J.
Subtypes of learning and memory on the California Verbal Learning Test—Second Edition (CVLT-II) in the standardization sample
Journal of Clinical and Experimental Neuropsychology
 , 
2008
, vol. 
30
 
7
(pg. 
741
-
748
)
[PubMed]
Donders
J.
Janke
K.
Criterion validity of the Wechsler Intelligence Scale for Children—Fourth Edition after pediatric traumatic brain injury
Journal of the International Neuropsychological Society
 , 
2008
, vol. 
14
 (pg. 
651
-
655
)
[PubMed]
Donders
J.
Warschausky
S.
WISC-III factor index score patterns after traumatic head injury in children
Child Neuropsychology
 , 
1997
, vol. 
3
 (pg. 
71
-
78
)
Donders
J.
Zhu
J.
Tulsky
D.
Factor index score patterns in the WAIS-III standardization sample
Assessment
 , 
2001
, vol. 
8
 (pg. 
193
-
203
)
Duval
J.
Braun
C. M. J.
Montour-Proulx
I.
Daigneault
S.
Rouleau
I.
Begin
J.
Brain lesions and IQ: Recovery versus decline depends on age of onset
Journal of Child Neurology
 , 
2008
, vol. 
23
 (pg. 
663
-
668
)
Everitt
B. S.
Landau
S.
Leese
M.
Cluster analysis
 , 
2001
4th ed.
London
Arnold
Fay
T. B.
Yeates
K. O.
Wade
S. L.
Drotar
D.
Stancin
T.
Taylor
H. G.
Predicting longitudinal patterns of functional deficits in children with traumatic brain injury
Neuropsychology
 , 
2009
, vol. 
23
 (pg. 
271
-
282
)
Gil
A. M.
Neurocognitive outcomes following pediatric brain injury: A developmental approach
Journal of School Psychology
 , 
2003
, vol. 
41
 (pg. 
337
-
353
)
Goldstein
G.
Allen
D. N.
Caponigro
J. M.
A retrospective study of heterogeneity in neurocognitive profiles associated with traumatic brain injury
Brain Injury
 , 
2010
, vol. 
24
 
4
(pg. 
625
-
635
)
[PubMed]
Goldstein
G.
Allen
D. N.
Minshew
N. J.
Williams
D. L.
Volkmar
F.
Klin
A.
, et al.  . 
Structure of intelligence in children and adults with high functioning autism
Neuropsychology
 , 
2008
, vol. 
22
 (pg. 
301
-
311
)
Goldstein
G.
Allen
D. N.
Seaton
B. E.
A comparison of clustering solutions for cognitive heterogeneity in schizophrenia
Journal of the International Neuropsychological Society
 , 
1998
, vol. 
4
 
4
(pg. 
353
-
362
)
[PubMed]
Gross
D.
Fogg
L.
Garvey
C.
Julion
W.
Behavior problems in young children: An analysis of cross-informant agreements and disagreements
Research in Nursing and Health
 , 
2004
, vol. 
27
 (pg. 
413
-
425
)
Hair
J. F.
Black
B.
Babin
B.
Anderson
R. E.
Tatham
R.
Multivariate data analysis with reading
 , 
1992
6th ed.
New Jersey
Prentice Hall
Hawley
C. A.
Reported problems and their resolution following mild, moderate and severe traumatic brain injury amongst children and adolescents in the UK
Brain Injury
 , 
2003
, vol. 
17
 (pg. 
105
-
129
)
Hawley
C. A.
Behaviour and school performance after brain injury
Brain Injury
 , 
2004
, vol. 
18
 (pg. 
645
-
659
)
Kriel
R. L.
Krach
L. E.
Panser
L. A.
Closed head injury: Comparison of children younger and older than 6 years of age
Pediatric Neurology
 , 
1989
, vol. 
5
 
5
(pg. 
296
-
300
)
Landis
J. R.
Koch
G. G.
The measurement of observer agreement for categorical data
Biometrics
 , 
1977
, vol. 
33
 
1
(pg. 
159
-
174
)
Lee
S. W.
Elliott
J.
Barbour
J. D.
A comparison of cross-informant behavior ratings in school-based diagnosis
Behavioral Disorders
 , 
1994
, vol. 
19
 
2
(pg. 
87
-
97
)
Loeber
R.
Green
S. M.
Lahey
B. B.
Mental health professionals' perception of the utility of children, mothers, and teachers as informants on childhood psychopathology
Journal of Clinical Child Psychology
 , 
1990
, vol. 
19
 
2
(pg. 
136
-
143
)
Malec
J. F.
Machulda
M. M.
Smigielski
J. S.
Cluster analysis of neuropsychological test results among patients with traumatic brain injury (TBI): Implications for a model of TBI-related disability
Clinical Neuropsychologist
 , 
1993
, vol. 
7
 (pg. 
48
-
58
)
Max
J.
Koele
S.
Smith
W.
Sato
Y.
Lindgren
S.
Robin
D.
, et al.  . 
Psychiatric disorders in children and adolescents after severe traumatic brain injury: A controlled study
Journal of the American Academy of Child and Adolescent Psychiatry
 , 
1998
, vol. 
37
 
8
(pg. 
832
-
840
)
McAuley
C.
Children's adjustment over time in foster care: Cross-informant agreement, stability and placement disruption
British Journal of Social Work
 , 
2000
, vol. 
30
 (pg. 
91
-
107
)
McConaughy
S. H.
Stanger
C.
Achenbach
T. M.
Three-year course of behavioral/emotional problems in a national sample of 4- to 16-year-olds: I. Agreement among inform-ants
Journal of the American Academy of Child and Adolescent Psychiatry
 , 
1992
, vol. 
31
 (pg. 
932
-
940
)
McKinlay
A.
Grace
R.
Horwood
J.
Fergusson
D.
MacFarlane
M.
Adolescent psychiatric symptoms following preschool childhood mild traumatic brain injury: Evidence from a birth cohort
The Journal of Head Trauma Rehabilitation
 , 
2009
, vol. 
24
 (pg. 
221
-
227
)
Miller
L. J.
Donders
J.
Prediction of educational outcome after pediatric traumatic brain injury
Rehabilitation Psychology
 , 
2003
, vol. 
48
 (pg. 
237
-
241
)
Millis
S. R.
Ricker
J. H.
Verbal learning patterns in moderate and severe traumatic brain injury
Journal of Clinical and Experimental Neuropsychology
 , 
1994
, vol. 
16
 (pg. 
498
-
507
)
Millis
S. R.
Rosenthal
M.
Novack
T. A.
Sherer
M.
Nick
T. G.
Kreutzer
J. S.
, et al.  . 
Long-term neuropsychological outcome after traumatic brain injury
The Journal of Head Trauma Rehabilitation
 , 
2001
, vol. 
16
 (pg. 
343
-
355
)
Morris
R.
Blashfield
R. K.
Satz
P.
Neuropsychology and cluster analysis: Potentials and problems
Journal of Clinical Neuropsychology
 , 
1981
, vol. 
3
 (pg. 
79
-
99
)
Mottram
L.
Donders
J.
Cluster subtypes on the California Verbal Learning Test—Children's version after pediatric traumatic brain injury
Developmental Neuropsychology
 , 
2006
, vol. 
30
 (pg. 
865
-
883
)
Murji
S.
Rourke
S. B.
Donders
J.
Carter
S. L.
Shore
D.
Rourke
B. P.
Theoretically derived CVLT subtypes in HIV-1 infection: Internal and external validation
Journal of the International Neuropsychological Society
 , 
2003
, vol. 
9
 (pg. 
1
-
16
)
[PubMed]
Naglieri
J. A.
Goldstein
S.
Delauder
B. Y.
Schwebach
A.
Relationships between the WISC-III and the Cognitive Assessment System with Conners' rating scales and continuous performance tests
Archives of Clinical Neuropsychology
 , 
2005
, vol. 
20
 (pg. 
385
-
401
)
Palmer
B. W.
Heaton
R. K.
Paulsen
J. S.
Kuck
J.
Braff
D.
Harris
M. J.
, et al.  . 
Is it possible to be schizophrenic yet neuropsychologically normal?
Neuropsychology
 , 
1997
, vol. 
11
 (pg. 
437
-
446
)
Randazzo
K.
Landsverk
J.
Ganger
W.
Three informants' reports of child behavior: Parents, teachers, and foster parents
Journal of the American Academy of Child and Adolescent Psychiatry
 , 
2003
, vol. 
42
 (pg. 
1343
-
1350
)
Rassovsky
Y.
Satz
P.
Alfano
M. S.
Light
R. K.
Zaucha
K.
McArthur
D. L.
, et al.  . 
Functional outcome of TBI I: Neuropsychological, emotional, and behavioral mediators
Journal of Clinical and Experimental Neuropsychology
 , 
2006
, vol. 
28
 (pg. 
567
-
580
)
Reitan
R. M.
Wolfson
D.
The Halstead-Reitan Neuropsychological Test Battery: Theory and clinical interpretation
 , 
1993
2nd ed.
Tucson, AZ
Neuropsychology
Reynolds
C. R.
Bigler
E. D.
Test of Memory and Learning
 , 
1994
Austin, TX
Pro-Ed
Reynolds
C. R.
Bigler
E. D.
Test of Memory and Learning: Examiner's manual
 , 
1994
Austin, TX
Pro-Ed
Reynolds
C. R.
Bigler
E. D.
Factor structure, factor indexes, and other useful statistics for interpretation of the Test of Memory and Learning (TOMAL)
Archives of Clinical Neuropsychology
 , 
1996
, vol. 
11
 (pg. 
29
-
43
)
Reynolds
C. R.
Ford
L.
Comparative Three-Factor Solutions of the WISC-III and WISC-R at 11 Age Levels Between 6-l/2 and 16-l/2 Years
Archives of Clinical Neuropsychology
 , 
1994
, vol. 
9
 (pg. 
553
-
570
)
[PubMed]
Reynolds
C. R.
Kamphaus
R. W.
Behavior Assessment System for Children.
 , 
1992
Circle Pines, MN
American Guidance Service
Reynolds
C. R.
Voress
J. K.
Test of Memory and Learning
 , 
2007
2nd ed.
Austin, TX
Pro-Ed
Roman
M. J.
Delis
D. C.
Willerman
L.
Magulac
M.
Demadura
T. L.
de la Pena
J. L.
, et al.  . 
Impact of pediatric traumatic brain injury on components of verbal memory
Journal of Clinical and Experimental Neuropsychology
 , 
1998
, vol. 
20
 (pg. 
245
-
258
)
Schwartz
L.
Taylor
H. G.
Drotar
D.
Yeates
K. O.
Wade
S. L.
Stancin
T.
Long-term behavior problems following pediatric traumatic brain injury: Prevalence, predictors, and correlates
Journal of Pediatric Psychology
 , 
2003
, vol. 
28
 (pg. 
251
-
263
)
Stolberg
P.
Terranova
J.
Jones
P. W.
Allen
D. N.
Mayfield
J.
Performance on the Woodcock–Johnson Test of Achievement of individuals with traumatic brain injury [Abstract]
Archives of Clinical Neuropsychology
 , 
2009
, vol. 
24
 
5
pg. 
476
 
Teasdale
G.
Jennett
B.
Assessment of coma and impaired consciousness: A practical scale
Lancet
 , 
1974
, vol. 
2
 
7872
(pg. 
81
-
84
)
[PubMed]
Thaler
N. S.
Allen
D. N.
McMurray
J.
Mayfield
J.
Sensitivity of the Test of Memory and Learning to attention and memory deficits in children with ADHD
Clinical Neuropsychologist
 , 
2010
, vol. 
24
 
2
(pg. 
246
-
264
)
Thaler
N. S.
Terranova
J.
Reynolds
C. R.
Mayfield
J.
Allen
D. N.
Teacher reported behavioral disturbances in children with TBI: An Examination of the BASC-2. Presented at the American College of Professional Neuropsychology Annual Meeting, Feb
2010
 
26–28, Las Vegas, NV
Thompson
N. M.
Francis
D. J.
Stuebing
K. K.
Fletcher
J. M.
Ewing-Cobbs
L.
Miner
M. E.
, et al.  . 
Motor, visual-spatial, and somatosensory skills after closed head injury in children and adolescents: A study of change
Neuropsychology
 , 
1994
, vol. 
8
 (pg. 
333
-
342
)
Tiffin
J.
Purdue Pegboard Test
 , 
1948
Lafayette, IN
Lafayette Instrument Company
van der Heijden
P.
Donders
J.
WAIS-III factor index score patterns after traumatic brain injury
Assessment
 , 
2003
, vol. 
10
 (pg. 
115
-
122
)
Verger
K.
Junque
C.
Levin
H. S.
Jurado
M. A.
Perez-Gomez
M.
Bartres-Faz
D.
, et al.  . 
Correlation of atrophy measures on MRI with neuropsychological sequelae in children and adolescents with traumatic brain injury
Brain Injury
 , 
2001
, vol. 
15
 (pg. 
211
-
221
)
[PubMed]
Wexler
B. E.
Zhu
H.
Bell
M. D.
Nicholls
S. S.
Fulbright
R. K.
Gore
J. C.
, et al.  . 
Neuropsychological near normality and brain structure abnormality in schizophrenia
The American Journal of Psychiatry
 , 
2009
, vol. 
166
 (pg. 
189
-
195
)
Wiegner
S.
Donders
J.
Performance on the California Verbal Learning Test after traumatic brain injury
Journal of Clinical and Experimental Neuropsychology
 , 
1999
, vol. 
21
 (pg. 
159
-
170
)
Woodcock
R. W.
Johnson
M. B.
Woodcock–Johnson psycho-educational battery – revised
 , 
1989
Chicago
Riverside Publishing
Yeates
K. O.
Armstrong
K.
Janusz
J.
Taylor
H. G.
Wade
S.
Stancin
T.
, et al.  . 
Long-term attention problems in children with traumatic brain injury
Journal of the American Academy of Child and Adolescent Psychiatry
 , 
2005
, vol. 
44
 
6
(pg. 
574
-
584
)
Yeates
K. O.
Swift
E.
Taylor
H. G.
Wade
S. L.
Drotar
D.
Stancin
T.
, et al.  . 
Short- and long-term social outcomes following pediatric traumatic brain injury
Journal of the International Neuropsychological Society
 , 
2004
, vol. 
10
 (pg. 
412
-
426
)
[PubMed]
Yeates
K. O.
Taylor
H. G.
Behavior problems in school and their educational correlates among children with traumatic brain injury
Exceptionality
 , 
2006
, vol. 
14
 (pg. 
141
-
154
)
Yeates
K. O.
Taylor
H. G.
Wade
S. L.
Drotar
D.
Stancin
T.
Minich
N.
A prospective study of short- and long-term neuropsychological outcomes after traumatic brain injury in children
Neuropsychology
 , 
2002
, vol. 
16
 (pg. 
514
-
523
)
Young
S.
Gudjonsson
G.
Neuropsychological correlates of the YAQ-S and YAQ-I self- and informant-reported ADHD symptomatology, emotional and social problems and delinquent behaviour
British Journal of Clinical Psychology
 , 
2005
, vol. 
44
 (pg. 
47
-
57
)