Abstract

Research suggests that IQ profiles identify subgroups of children with traumatic brain injury (TBI) based on sparing and impairment of cognitive abilities, but little information is available regarding whether these subgroups are differentiated on variables that are important for TBI outcome, such as behavioral functioning. The current study examined behavioral disturbances in 123 children with TBI in association with profiles of intellectual abilities identified using cluster analysis. On the basis of prior research, four clusters were hypothesized. Consistent with the hypothesis, cluster analysis identified four IQ clusters in the current sample. Comparisons among the clusters on behavior variables assessed from the Behavioral Assessment System for Children parent ratings indicated significant differences among the four IQ clusters, with the most impaired cluster exhibiting the severest disturbances. Results of the current study indicate that subgroups of children with TBI can be identified using IQ tests and that these subgroups are stable across different samples, and more importantly are moderately associated with behavioral disturbances that persist during the recovery period.

Introduction

An estimated 475,000 children sustain a traumatic brain injury (TBI) each year in the United States (Langlois, Rutland-Brown, & Thomas, 2005). TBI is one of the leading causes of death and disability among children (Hawley, 2004) and survivors often exhibit severe and persisting cognitive, behavioral, and emotional difficulties (Bishop, 2008; Fay et al., 2009; Taylor et al., 2008; Tonks, Williams, Frampton, Yates, & Slater, 2007). In addition, families of children with TBI are burdened with the additional stress and costs of caring for an injured child and often exhibit symptoms of increased parental stress and depression in direct correlation with the severity of their child's injury (Stancin, Wade, Walz, Yeates, & Taylor, 2008). A considerable body of literature has examined the neuropsychological and behavioral abnormalities typically associated with TBI (Babikian & Asarnow, 2009; Ganesalingam et al., 2008; Hawley, 2004; Holm, Schönberger, Poulsen, & Caetano, 2009; Worthington & Woods, 2008). However, TBI outcomes can be difficult to predict, in part due to heterogeneity of a number of factors including region of the brain insult, secondary complications, and premorbid functioning. As such, these differential outcomes present challenges for interventions designed to address behavioral and emotional problems present in TBI (Rees, Marshall, Hartridge, Mackie, & Weiser, 2007). This heterogeneity in outcome can be investigated using cluster analysis, a descriptive technique that identifies common attributes in large data sets and then combines these attributes into homogenous subtypes (Romesburg, 1984). Cluster analysis has been used to identify IQ-based cluster subtypes in both adult (Crawford, Garthwaite, & Johnson, 1997; Curtiss, Vanderploeg, Spencer, & Salazar, 2001) and child (Donders & Warschausky, 1997) populations with TBI, although these studies are few in number and the generalizability of findings from one clinical sample to another has not been extensively examined.

Subtypes derived through cluster analysis represent more homogenous TBI neuropsychological profiles that differ on level and pattern of performance on IQ index scores. However, there is little information available regarding these subtypes’ relationships with outcome variables, including behavioral functioning. Such relationships evaluate what has been termed the “external validity” of the cluster solution (Aldenderfer & Blashfield, 1984) and may serve to identify subgroups of children with TBI who exhibit differences in behavioral and emotional disturbances during the recovery period. Some associations have been identified between IQ and behavioral outcomes following pediatric TBI, with lower IQ both pre- and post-injury generally linked to increased aggression and school problems (Hawley, 2004; Wood & Liossi, 2006), whereas verbal IQ in particular may be important in executive functions related to behavioral regulation (Vriezen & Pigott, 2002). In addition, parent-reported measures have found associations between intellectual functioning and behavioral outcomes in clinical populations (de Bildt, Sytema, Kraijer, Sparrow, & Minderaa, 2005; Dietz, Lavigne, Arend, & Rosenbaum, 1997; Jansen et al., 2007) and there may be utility in comparing these measures to IQ functioning in pediatric TBI. The goal of the current study was to investigate the behavioral differences among children with TBI who demonstrate varying IQ profiles in an attempt to establish the external validity of the IQ clusters. A second goal of this study was to investigate whether these cluster-based subtypes may provide additional information that may aid clinicians and researchers in classifying TBI subtypes. A third goal was to examine generalizability of IQ clusters across study populations through comparisons of current findings to those reported by others (Donders & Warschausky, 1997).

TBI is currently classified via multiple means, including severity of the injury, length of coma or amnesia, and mechanism of injury (Smith, Barth, Diamond, & Giuliano, 1998). In broadest terms, injuries may be identified as open or closed, differentiated by the presence or the absence of a skull breach (Smith et al., 1998). Concomitant injuries such as neuronal shearing, hematoma, and infections can further complicate the prognostic outcome of TBI and contribute to its heterogeneous symptomatology in neurocognitive and behavioral domains (Mottram & Donders, 2006). Behavioral deficits associated with TBI include hyperactivity and conduct problems resulting from compromised intellectual development and reduced attentional capacity (Basson et al., 1991; Schwartz et al., 2003; Yeates et al., 2005). Because of the substantial variability presented across patients with TBI, additional diagnostic and prognostic information might be provided vis-à-vis the associations of IQ profiles to some of these behavioral deficits, which are often not considered when using traditional classification methods.

When severity classification is made using the Glasgow Coma Scale (GCS; Teasdale & Jennet, 1974) or other similar procedures, many children initially classified as having severe TBI do not demonstrate significant neurocognitive deficits when examined after a period of recovery, and other factors including age at injury and clinical expertise may account for more variance in neurocognitive outcomes (Fay et al., 2009; Lieh-Lah, Theodorou, & Sarnaik, 1992; Wells, Minnes, & Phillips, 2009). Indeed, the GCS has been described as only a gross predictor of TBI severity and functional outcome (Ghosh et al., 2009; Hackbarth et al., 2002; Hiekkanen, Kurki, Brandstack, Kairisto, & Tenovuo, 2009). Given the associations between many neuropsychology measures and severity of brain damage, as well as associations between neurocognitive function and functional outcomes, neurocognitive profiles differentiated by IQ cluster subtypes may provide a different, and possibly more robust, method of classification, which may also show stronger associations with long-term behavioral and functional outcomes. Some preliminary studies suggest that this is indeed a possibility. For example, one recent study of 150 children with TBI that cluster analyzed scores from the Test of Memory and Learning (Reynolds & Bigler, 1994) found behavioral differences among the clusters including increased hyperactivity, attention deficits, and conduct problems based on parent and teacher ratings of children in the most severely impaired memory cluster (Allen et al., in press). However, this cluster did not significantly differ on GCS scores compared with other clusters that exhibited selective deficits in visual and verbal memory domains, suggesting that it predicted certain outcome variables that were missed when considering the GCS alone.

With regard to IQ clusters, four clusters were identified in a pediatric TBI sample (Donders & Warschausky, 1997) using the Wechsler Intelligence Scale for Children, Third Edition (WISC-III; Wechsler, 1991). Three of these clusters were differentiated primarily by level of performance, including an above average cluster, a second cluster that had average index scores, and a third that had impaired index scores. A fourth cluster was identified as differing on patterns of performance, with impaired scores on verbal and attention indexes and severely impaired scores on the nonverbal and processing speed indexes, suggesting that the latter two indexes are particularly sensitive to TBI in some patients. Significant differences were present among the clusters based on injury severity as defined by both the GCS and neuroimaging findings, indicating that the fourth cluster was associated with severest brain injury. No differences were found for demographic variables.

Although IQ clusters may be differentiated across measures of injury severity, little information is yet available on the clusters’ behavioral outcomes. Thus, a follow-up study comparing clusters across these outcomes is warranted. The present study attempted to achieve that goal using WISC-III data. The WISC-III was the edition available at the time of data collection, and its availability supports doing a cross-validation of the Donders and Warschausky (1997) study which used the WISC-III. Such use should have general applicability to the currently used WISC-IV given the similarity among the factor structures of the various Wechsler Intelligence Scales and their common use in the assessment of TBI.

Therefore, cluster analysis was used in the current study to examine the patterns of IQ performance in a sample of children with TBI via the WISC-III in order to: (i) cross-validate cluster subtypes based on tests of intellectual functioning in TBI identified using various measures from the WISC-III, and (ii) establish the external validity of these clusters through examination of demographic, clinical, and behavioral variables that were not included in the cluster analysis. With regard to cross-validation of the cluster subtypes, we hypothesized that four meaningful clusters would be identified in the current sample with similar profiles to those reported by Donders and Warschausky (1997). Confirmation of this hypothesis would demonstrate that these four clusters are stable and generalize across different pediatric populations with TBI. The importance of such cross-validation in children with TBI has long been noted. For the external validity of the cluster solutions, no differences among the clusters were hypothesized for demographic variables, although it was predicted that differences in injury severity as measured by the GCS would be present given previous findings (Donders & Warschausky, 1997). In addition, we anticipated that further differentiation will be identified by the relationships between cluster subtype and behavioral variables measured by the Behavioral Assessment System for Children (BASC; Reynolds & Kamphaus, 1992).

Method

Participants

Participants were children who sustained a TBI and who were selected from a consecutive series of 365 cases seen for clinical neuropsychological evaluation over a 7-year period at a children's medical facility. They were previously hospitalized for the consequences of their TBIs and were referred to evaluate cognitive strengths and weakness for educational planning purposes. Participants were included in the current study if they had sustained a TBI as determined by direct observation, neurological evaluation, and appropriate neuroimaging and laboratory tests. Individuals with records of preexisting mental retardation, pervasive developmental disorder, learning disability, or previous evaluation with the WISC-III were excluded, as were individuals who could not cooperate with the testing procedures. One of the participants had a comorbid diagnosis of ADHD hyperactive subtype diagnosed prior to TBI, a second participant had a preexisting seizure disorder, and three others had other neurological disorders not specified. Of the 123 participants who qualified, 58% were males and 81% were right-handed. The average age of the sample was 11.6 years (SD = 3.1, range = 6.0–16.92 years) and had a mean WISC-III FSIQ of 82.4 (SD =14.1) at the time of assessment. One hundred and six (86.2%) of the participants were used in a previous study that examined the associations between memory clusters and outcomes (Allen et al., in press).

The majority of participants were Caucasian (36%), followed by African American (17%), then Hispanic (11%), and Asian American (1%), whereas the remainder of the participants (35%) did not report their ethnicity. The mechanisms of injuries were as follows: motor vehicle accident (50%), struck by a motor vehicle (24%), gunshot (4%), fall (4%), four-wheeler accident (7%), bike accident (2%), skiing (2%), and other (7%). Children sustained their injuries an average 7.39 months (SD= 6.4, range = 3–60 months) prior to assessment. However, they were all outpatients who were clinically stable at the time of neuropsychological assessment and capable of cooperating for the tests. None of them were in a confusional state or apparently suffering from post-traumatic amnesia. Nine (7.3%) of our participants had open-head injuries. GCS scores were available for 83 of our participants and average score was 7.10 (SD = 2.7; median = 7; range = 3–15), indicating that the sample could be generally characterized as having severe injuries.

Measures

Intellectual ability was assessed using the WISC-III's 10 core subtests. The four index factors comprising the subtests included the Verbal Comprehension (VCI), Perceptual Organization (POI), Freedom from Distractibility (FFDI), and Processing Speed (PSI) index scores. These index scores were used in the data analysis and have a mean of 100 and a standard deviation of 15.

Behavioral functioning was measured through the parent rating scales (PRS) of the Behavioral Assessment System for Children (BASC). The BASC is composed of multimethod reports of behavior and self-perception that are evaluated across a number of domains. Two of the reports, the PRS and the Teacher Rating Scale (TRS), are descriptive assessments of the child's observable behavior, provided by parents and teachers. A third report is the Self-Report of Personality (SRP), in which children provide their own descriptions of self-perception and emotions. For the purposes of this study, we focused our analysis on the PRS because only limited data were available for the SRP and TRS. PRS data were available for 102 (82.9%) of our participants.

The PRS is composed of Hyperactivity, Aggression, Conduct Problems, Anxiety, Depression, Somatization, Attention Problems, Atypicality, and Withdrawal clinical subscales and the Adaptability, Social Skills, and Leadership adaptive subscales. Clinical composite scales include Externalizing Problems, Internalizing Problems, and the Behavioral Symptoms Index. An Adaptive Skills scale serves as an adaptive composite scale.

The severity of injury was evaluated with the GCS. The GCS classifies severity via three methods: Best Eye Response (Scores 1–4), Best Verbal Response (Scores 1–5), and Best Motor Response (1–6). The sum of these scores can be used to categorize TBI severity as either mild (13–15), moderate (9–12), or severe (3–8). The GCS data utilized were administered shortly after injury either by first responders to the accident or in the emergency room of the hospital.

Data Analysis

Ward's method, a form of hierarchical cluster analysis, was applied to the four index scores of the WISC-III. Ward's method was used because of its consistency with the cluster analytic methodology of previously conducted studies in this area of research (Curtiss et al., 2001; Mottram & Donders, 2006). Furthermore, Ward's method produces results that are consistent with other agglomerative clustering methods and is less affected by outliers than other clustering methods. Squared Euclidian distance was used as the dissimilarity measure in the analyses, as it is the most commonly used of all such measures. It provides a direct, physical measure of distance in Euclidean space (Everitt, Landau, & Leese, 2001). The stability of the clusters was then evaluated with the K-means iterative classification process, a form of partitional analysis that would theoretically agree with a hierarchical analysis if the solution is stable. The centers of the WISC-III index scores derived by Ward's method were used as the starting points for the iterative cluster analysis. Cohen's Kappa was used to measure the agreement between the K-means and Ward's method solutions. A discriminant function analysis (DFA) was performed to confirm cluster membership. The clusters were then compared across a number of demographic, clinical, and behavioral variables.

Given previous findings on the BASC PRS (Allen et al., in press), we looked at the PRS hyperactivity, attention problems, and conduct problems subscales and all the PRS composites. In order to evaluate the hypotheses regarding behavioral differences between clusters, separate MANOVAs were used to evaluate the BASC PRS subscales and the PRS composites. In these analyses, the BASC ratings were dependent variables and cluster membership served as the between-subjects factor. When subscale or composite comparisons were significant, the Scheffe tests were used to examine cluster differences. Estimates of effect size were evaluated with forumla.

Results

Cluster Analysis

Ward's method was used as the hierarchical cluster analysis. Inspection of the cluster dendogram indicated that a four-cluster solution was reasonable, in that it was possible to clearly identify four clusters in the dendogram. Inspection of this four-cluster solution suggested that the clusters were separated by level and pattern of performance across the four WISC-III index scores. One cluster had average performance, the second cluster had below average performance, the third cluster had below average performance on the VCI and FFDI and poor performance on the POI and PSI, and the fourth cluster had average performance on the VCI, POI, and FFDI and poor performance on the PSI. The iterative partitioning clustering method had a kappa coefficient of 0.74 with the four-cluster solution, a level of agreement that is considered substantial (Landis & Koch, 1977). This level of agreement suggests that four clusters are stable and thus an acceptable solution for our sample, replicating the previous cluster finding on other TBI samples (Donders, 1996; Donders & Warshausky, 1997). DFA indicated that these clusters were adequately separated in discriminant function space (Wilks’ lambda < 0.01; Box's test < 0.01). DFA correctly classified 94.3% of the total cases (Cluster 1 = 93%, Cluster 2 = 93%, Cluster 3 = 100%, Cluster 4 = 100%). The profiles for the clusters are presented in Fig. 1.

Fig. 1.

Four-cluster solution of WISC-III index scores. VCI = Verbal Comprehension Index; POI = Perceptual Organization Index; FFDI = Freedom From Distractability Index; PSI = Processing Speed Index.

Fig. 1.

Four-cluster solution of WISC-III index scores. VCI = Verbal Comprehension Index; POI = Perceptual Organization Index; FFDI = Freedom From Distractability Index; PSI = Processing Speed Index.

The four clusters were identified using norms established by Heaton, Grant, and Matthews (1991) as “average,” “moderately impaired,” moderately impaired with “severe POI/PSI,” and average with mildly “impaired PSI.” Of the 123 participants, 45% were in the average cluster, 34% were in the impaired cluster, 12% were in the severe POI/PSI cluster, and 9% were in the impaired PSI cluster. These percentages are comparable to the WISC-III cluster study by Donders and Warschausky (1997), who found that 39% of their participants were in the average cluster, 25% were in the impaired cluster, and 18% were in the severe POI/PSI cluster. Additionally, 18% of their sample fell into a high average cluster. A high average cluster did not emerge in our study, possibly because they had a mix of mild, moderate, and severe injuries while most of the children in the present study sustained severe injuries.

Children in our Average cluster performed at or near the standardization mean of all four index scores, although a small number of children scored above the mean in some of the index scores and subtests. Children in the impaired cluster were approximately 1.6 SD below the standardization mean across the IQ and factor index scores. The children in the severe POI/PSI cluster performed about 1.6 SD below the mean on the VCI and FFDI, but nearly 3 SD below on the POI and PSI. Finally, the children in the impaired PSI cluster were near mean performance on the VCI, POI, and FFDI but almost 2SD below the mean on the PSI. See Table 1 for the WISC-III subtest and index scores of our four clusters and the entire sample.

Table 1.

Four-cluster solution for WISC-III traumatic brain injury data

WISC-III scores Clusters
 
Total (N = 123)
 
 Average (n = 55)
 
Impaired (n= 42)
 
Severe POI/PSI (n = 15)
 
Impaired PSI (n = 11)
 
  
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Subtests 
 IN 9.3 2.6 5.8 2.0 4.7 2.6 9.8 2.0 7.6 3.1 
 SM 9.5 2.1 6.4 2.1 6.0 2.9 9.9 2.1 8.0 2.7 
 AR 10.0 2.5 6.7 2.1 5.2 2.1 9.5 2.9 8.2 3.0 
 VO 8.9 2.4 5.4 2.6 5.1 2.4 9.7 1.5 7.3 3.0 
 CO 7.8 2.4 5.0 2.3 3.5 2.3 7.4 2.9 6.3 2.9 
 DS 9.8 2.4 6.9 2.2 5.9 1.9 8.7 2.6 8.2 2.8 
 PC 9.1 2.8 6.8 2.2 3.3 2.7 10.2 2.2 7.7 3.2 
 CD 8.8 2.7 5.5 2.4 1.8 1.3 4.4 2.4 6.4 3.4 
 PA 8.4 3.6 6.3 2.4 2.3 1.5 7.1 3.7 6.8 3.6 
 BD 9.4 2.7 5.8 2.7 2.8 1.8 9.6 2.7 7.4 3.5 
 OA 8.4 3.6 5.8 2.8 2.3 1.5 8.2 2.0 6.8 3.6 
 SS 10.1 2.3 6.2 2.6 2.2 1.6 4.4 2.2 7.3 3.6 
Indexes 
 VIQ 95.0 9.7 76.7 8.2 71.3 10.3 95.9 8.3 85.9 13.6 
 PIQ 92.8 12.0 75.5 6.3 54.9 4.6 87.0 10.6 81.8 15.8 
 FSIQ 93.2 9.4 74.1 5.4 60.0 6.9 90.8 3.0 82.4 14.1 
 VCI 94.4 9.6 76.6 9.1 72.5 10.4 96.0 8.2 85.8 13.4 
 POI 93.6 14.1 77.9 7.3 57.3 6.0 93.2 11.1 83.8 16.4 
 FFDI 100.4 10.3 82.6 9.5 75.5 9.7 95.8 10.3 90.9 13.9 
 PSI 98.1 10.2 79.3 10.6 57.5 6.5 71.1 9.3 84.3 17.2 
WISC-III scores Clusters
 
Total (N = 123)
 
 Average (n = 55)
 
Impaired (n= 42)
 
Severe POI/PSI (n = 15)
 
Impaired PSI (n = 11)
 
  
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Subtests 
 IN 9.3 2.6 5.8 2.0 4.7 2.6 9.8 2.0 7.6 3.1 
 SM 9.5 2.1 6.4 2.1 6.0 2.9 9.9 2.1 8.0 2.7 
 AR 10.0 2.5 6.7 2.1 5.2 2.1 9.5 2.9 8.2 3.0 
 VO 8.9 2.4 5.4 2.6 5.1 2.4 9.7 1.5 7.3 3.0 
 CO 7.8 2.4 5.0 2.3 3.5 2.3 7.4 2.9 6.3 2.9 
 DS 9.8 2.4 6.9 2.2 5.9 1.9 8.7 2.6 8.2 2.8 
 PC 9.1 2.8 6.8 2.2 3.3 2.7 10.2 2.2 7.7 3.2 
 CD 8.8 2.7 5.5 2.4 1.8 1.3 4.4 2.4 6.4 3.4 
 PA 8.4 3.6 6.3 2.4 2.3 1.5 7.1 3.7 6.8 3.6 
 BD 9.4 2.7 5.8 2.7 2.8 1.8 9.6 2.7 7.4 3.5 
 OA 8.4 3.6 5.8 2.8 2.3 1.5 8.2 2.0 6.8 3.6 
 SS 10.1 2.3 6.2 2.6 2.2 1.6 4.4 2.2 7.3 3.6 
Indexes 
 VIQ 95.0 9.7 76.7 8.2 71.3 10.3 95.9 8.3 85.9 13.6 
 PIQ 92.8 12.0 75.5 6.3 54.9 4.6 87.0 10.6 81.8 15.8 
 FSIQ 93.2 9.4 74.1 5.4 60.0 6.9 90.8 3.0 82.4 14.1 
 VCI 94.4 9.6 76.6 9.1 72.5 10.4 96.0 8.2 85.8 13.4 
 POI 93.6 14.1 77.9 7.3 57.3 6.0 93.2 11.1 83.8 16.4 
 FFDI 100.4 10.3 82.6 9.5 75.5 9.7 95.8 10.3 90.9 13.9 
 PSI 98.1 10.2 79.3 10.6 57.5 6.5 71.1 9.3 84.3 17.2 

Notes: WISC-III = Wechsler Intelligence Scale for Children, Third Edition; IN = Information; SM = Similarities; AR = Arithmetic; VO = Vocabulary; CO = Comprehension; DS = Digit Span; PC = Picture Completion; CD = Coding; PA = Picture Arrangement; BD = Block Design; OA = Object Assembly; SS = Symbol Search; VIQ = Verbal IQ; PIQ = Performance IQ; FSIQ = Full-Scale IQ; VCI = Verbal Comprehension Index; POI = Perceptual Organization Index; FFDI = Freedom from Distractibility Index; PSI = Processing Speed Index

Comparisons Across Demographic and Clinical Variables

Chi-square analyses indicated no significant differences among the clusters associated with gender, χ2(3) = 2.02, p = .57, ethnicity, χ2(3) = 0.43, p = .93, handedness, χ2(3) = 4.98, p = .17, open versus closed injury, χ2(3) = 3.79, p = .29, or circumstances of injury, χ2(3) = 3.26, p = .35. One-way ANOVAs also indicated no significant differences among clusters for age in months, F(3, 119) = 1.18, p = .32, months since time of injury, F(3, 118) = 0.60, p = .62, and Glasgow Coma Scores, F(3, 79) = 1.02, p = .39. The average cluster had a mean Glasgow Coma Score of 7.6 (SD =2.8), the impaired PSI cluster had a mean of 7.1 (SD =4.2), the impaired cluster had a mean of 6.9 (SD =2.2), and the severe POI/PSI cluster had a mean of 5.9 (SD =1.6). These mean scores are not significantly different from each other, but are in the expected direction. See Table 2 for a breakdown of demographic and clinical variables among the four clusters. In summary, demographic and clinical variables were not found to be associated with significant differences among clusters.

Table 2.

Descriptive and clinical variables of the four-cluster solution

Variables Clusters
 
Total 
 Average
 
Impaired Severe POI/PSI Impaired PSI   
Gender (n
 Male 31 22 10 71 
 Female 24 20 52 
Ethnicity (n
 Caucasian 18 15 44 
 African American 10 21 
 Hispanic 14 
 Asian American 
 Other 
Injury (n
 Open 
 Closed 53 39 13 114 
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Age in Years 11.3 3.1 11.5 2.9 11.7 3.2 13.2 3.1 11.6 3.1 
Months since injury 7.7 8.9 6.8 3.5 9.0 3.2 6.2 2.6 7.4 6.4 
GCS 7.6 2.8 6.9 2.2 5.9 1.6 7.1 4.2 7.1 2.7 
Variables Clusters
 
Total 
 Average
 
Impaired Severe POI/PSI Impaired PSI   
Gender (n
 Male 31 22 10 71 
 Female 24 20 52 
Ethnicity (n
 Caucasian 18 15 44 
 African American 10 21 
 Hispanic 14 
 Asian American 
 Other 
Injury (n
 Open 
 Closed 53 39 13 114 
 Mean SD Mean SD Mean SD Mean SD Mean SD 
Age in Years 11.3 3.1 11.5 2.9 11.7 3.2 13.2 3.1 11.6 3.1 
Months since injury 7.7 8.9 6.8 3.5 9.0 3.2 6.2 2.6 7.4 6.4 
GCS 7.6 2.8 6.9 2.2 5.9 1.6 7.1 4.2 7.1 2.7 

Notes: GCS = Glasgow Coma Scale; POI = Perceptual Organization Index; PSI = Processing Speed Index.

Comparisons across PRS BASC scores

The MANOVA produced an overall significant effect for PRS subscales, F(9, 291) = 2.44, p= .01, forumla, but not for the composites, F(12, 282) = 1.45, p= .15. Results of the MANOVA are presented in Table 3. There were significant differences among the clusters for the Hyperactivity and Attention Problems subscales, but not for the Conduct Problems subscale. The subsequent Scheffe tests found no significant differences among clusters for the hyperactivity subscale. The impaired cluster had significantly higher Attention Problems scores than the average cluster, whereas the severe POI/PSI cluster had significantly higher Attention Problems scores than the average and impaired PSI clusters.

Table 3.

Behavior Assessment System for Children PRS scores across the four-cluster solution

Variables Clusters
 
F-value p-value forumla 
 Average (n = 46)
 
Impaired (n = 32)
 
Severe POI/PSI (n = 13)
 
Impaired PSI (n = 10)
 
   
 Mean SD Mean SD Mean SD Mean SD    
PRS subtest 
 Hyperactivity 54.5 15.5 64.5 19.9 60.8 17.4 48.8 9.0 3.46 .02 0.097 
 Conduct problems 51.1 12.9 58.4 19.4 56.2 22.5 50.0 11.6 1.53 .21 0.045 
 Attention problems 56.4 12.3 64.3 11.1 68.7 11.0 55.4 6.0 6.03 <.01 0.157 
PRS composite 
 Externalizing 53.6 13.9 63.6 20.8 58.7 20.8 51.4 11.7 2.70 .05 0.079 
 Internalizing 54.0 13.2 57.4 18.2 53.3 9.9 51.4 12.9 .63 .60 0.019 
 Behavioral symptoms 55.3 14.8 65.6 20.7 61.1 14.3 52.7 8.7 3.19 .03 0.092 
 Adaptive skills 47.7 13.0 41.5 10.8 43.3 8.2 44.3 7.7 1.80 .15 0.054 
Variables Clusters
 
F-value p-value forumla 
 Average (n = 46)
 
Impaired (n = 32)
 
Severe POI/PSI (n = 13)
 
Impaired PSI (n = 10)
 
   
 Mean SD Mean SD Mean SD Mean SD    
PRS subtest 
 Hyperactivity 54.5 15.5 64.5 19.9 60.8 17.4 48.8 9.0 3.46 .02 0.097 
 Conduct problems 51.1 12.9 58.4 19.4 56.2 22.5 50.0 11.6 1.53 .21 0.045 
 Attention problems 56.4 12.3 64.3 11.1 68.7 11.0 55.4 6.0 6.03 <.01 0.157 
PRS composite 
 Externalizing 53.6 13.9 63.6 20.8 58.7 20.8 51.4 11.7 2.70 .05 0.079 
 Internalizing 54.0 13.2 57.4 18.2 53.3 9.9 51.4 12.9 .63 .60 0.019 
 Behavioral symptoms 55.3 14.8 65.6 20.7 61.1 14.3 52.7 8.7 3.19 .03 0.092 
 Adaptive skills 47.7 13.0 41.5 10.8 43.3 8.2 44.3 7.7 1.80 .15 0.054 

Note: PRS = parent rating scale.

Discussion

This study examined parent-reported behavioral differences among IQ cluster subtypes that were identified in a sample of children with TBI. Similar cluster studies on TBI samples have been accomplished for children (Donders & Warschausky, 1997) also using the WISC-III and adults (van der Heijden & Donders, 2003). Consistent with these previous findings, we found that processing speed deficits, as reflected in lower PSI scores, contributed to differences identified among the clusters. The hypothesis that a four-cluster solution would be identified was supported, replicating Donders and Warschausky's (1997) cluster analysis of a different pediatric TBI sample. These findings suggest that children with TBI have at least four cognitive subgroups that differ on Wechsler levels and patterns of performance. This result does not appear to be attributable to differences in injury type among the clusters, since the great majority of participants had closed head injuries, which are not associated with focal pathology but with diffuse cerebral injury varying largely in severity (Levin, Benton, & Grossman, 1982). Further, there was not a significant difference among the clusters in frequency of open head injury.

The current results provide additional support for the generalizability and validity of IQ clusters in children with TBI, given that they were identified in two different TBI samples including the one examined in the current investigation and the one reported by Donders and Warchausky (1997). The similarities between the clusters identified here and those previously reported are striking. In both cases, each group was characterized with a severe POI/PSI cluster displaying particular impairments in POI and PSI scores, an average cluster performing at an average level across all four index scores, and an impaired cluster performing below the standardization sample mean by more than 1SD. These clusters had similar proportions of participants to the Donders and Warschausky study, providing evidence that the cluster profiles have consistent prevalence rates. Additionally, we extracted an impaired PSI cluster that fell within the average range on three of the index scores, demonstrating impairment only on the PSI.

Our clusters did not differ across demographic variables, suggesting that cluster solutions may be generalized to both boys and girls of different ethnicities and ages and also that the differences in emotional and behavioral functioning among the clusters is not easily accounted for by these variables. However, it should be noted that employment outcome measures in adults with TBI show racial differences (Gary et al., 2009) and female children have a higher risk of brain injury from falls than do men (Love, Tepas, Wiudyka, & Masnita-Iusan, 2009). Additionally, there appears to be a higher risk of poor outcome with increasing age in adults (Hukkelhoven et al., 2003), so caution must be exercised in generalizing these clusters. The clusters also did not differ by months since injury, possibly because a majority of our sample had a relatively short interval (within 1 year) between injury and time of assessment. Whether or not similar clusters would emerge following more extended time period of recovery was not directly addressed in our study, although given the lack of differences among clusters, it does not appear that time since injury has a major impact on cluster membership. Also, Donders and Warschausky's (1997) sample was overall assessed within an even shorter time period (∼2–4 months) and identified similar clusters, although their overall level of IQ was higher than it was in the present study (our mean FSIQ = 82.4, Donders & Warschausky's mean FSIQ = 91.1). This may also explain our Impaired PSI cluster, which has not been previously identified in other Wechsler cluster studies and may represent a unique finding within severely injured children.

With regard to severity of injury, it should be noted that GCS scores were not significantly different among the clusters though they were in the expected direction with the average cluster obtaining the highest GCS and the severe POI/PSI cluster having the lowest score. Nevertheless, severity of injury as assessed with the GCS was not a robust predictor of cluster differences. Several BASC domains were spared even within the lowest functioning cluster. However and as would be expected, the impaired cluster exhibited the greatest behavioral deficiencies, with higher parental ratings of the attention problems and hyperactive subscales, with medium to large effect sizes. In comparison, Allen and colleagues (in press) compared TOMAL cluster subtypes and also found elevated Attention Problems and Conduct Problems, but not Hyperactivity. However, Attention Problems, Conduct Problems, and Hyperactivity all fall under the Externalizing Problems Composite and provide support that externalizing behavioral problems are a signature symptom associated with severe IQ and memory deficits. These results also provide evidence that TBI profiles characterized by global impaired functioning across Wechsler index scores are associated with greater externalizing behavioral difficulties than other TBI profile types. The severe POI/PSI cluster was associated with higher PRS Attention Problem ratings than the average cluster. This finding is consistent with the literature, which reports significant attention problems primarily in individuals with severe TBI (Mathias & Wheaton, 2007).

In summary, these results indicate the existence of differentiated IQ clusters in children with TBI. Moreover, the successful identification of an impaired subtype within an overall severe TBI sample marked by numerous behavioral problems adds to the external validity of cluster analysis in dividing heterogeneous samples of patients with TBI into smaller, more homogeneous groups with unique outcome variables. An additional finding of interest was the discovery that not all children with TBI exhibit processing speed deficits as reflected by lower PSI scores. This is an important finding as the PSI has previously been identified as the index score most sensitive to TBI (Calhoun & Mayes, 2005; Mayes & Calhoun, 2004). It may be that selective PSI deficits, therefore, are not as invariably associated with TBI as has traditionally been thought. Further, children with TBI who perform within the normal range on the WISC-III may have better behavioral and emotional outcomes than do children with TBI who have other WISC-III profile patterns. Evidence in support of this suggestion was found when the average cluster was compared with the impaired cluster. Additional studies should attempt to replicate these findings as well as more thoroughly investigate the way in which other variables differ across the PSI/POI impaired cluster in light of the fact that this cluster has emerged in separate samples of children with TBI.

Given the current results that suggested differences between IQ clusters on important domains associated with outcomes, longitudinal studies would be useful in determining the predictive value of the identified clusters with regard to a long-term academic, psychosocial, and functional outcome. However, although the point of the current study was not to evaluate such outcomes, some studies suggest that longitudinal and cross-sectional studies provide similar findings in TBI (Baguley, Cooper, & Felmingham, 2006; Hanten et al., 2009; Mathias, Bowden, Bigler, & Rosenfeld, 2007), so the current results may have some value in that regard. A further study of interest might combine PRS composite scores and IQ index scores in a second cluster analysis to see how mixed cognitive/behavioral profiles might be associated with additional external variables, such as neuroimaging data or other measures of severity.

Though our IQ clusters were associated with PRS outcomes, caution must be noted in using measures of intelligence to predict behavioral outcome in TBI. Though some studies have found associations between IQ and functional outcome (Hawley, 2004; Wood & Liossi, 2006), other studies suggest that IQ may not be as robust a predictor as other cognitive domains such as memory and executive functioning (Kirkwood et al., 2000; Tramontana, Hooper, & Nardolillo, 1988). Indeed, executive functioning has been found to be a strong predictor of outcome (Muscara, Catroppa, & Anderson, 2008; Nadebaum, Anderson, & Catroppa, 2007), whereas the recent study by Allen and colleagues (in press) found significant and differential associations among attention and memory clusters, neurocognitive domains, and behavioral outcomes. As such, subtyping TBI studies will likely move away from solely using IQ clusters and focus more on specific cognitive and behavioral domains that may provide more detailed information about clusters. On a further note, such clusters should be considered primarily for their associations with other variables of interest (e.g., outcome variables) and may not be useful in diagnostic classification, as they often fluctuate when different clustering variables and methods are used.

Several limitations to this study exist. Our sample consisted primarily of children with moderate to severe TBI. It is difficult to generalize these clusters or their outcomes to populations with milder injuries. Also, some caution in the interpretation of clusters is warranted, as attentional difficulties in this study were assessed by the way of parental observation and verbal report as opposed to objective testing. A number of studies have indicated a lack of correspondence between neuropsychological measures and behavioral ratings of attention (Barney et al., under review; Dennis, Guger, Roncadin, Barnes, & Schachar, 2001; Naglieri, Goldstein, Delauder, & Schwebach, 2005; Young & Gudjonsson, 2005). Also, we relied on a clinically referred sample of convenience for study participants, which may limit the generalizability of the present findings. However, similar clusters have been independently identified in other samples of TBI survivors (e.g., Donders and Warschausky, 1997; van der Heijden & Donders, 2003), which provide support for the generalizability of the current results. Regarding severity, as only 83 of our sample had GCS scores and as our study lacked neuroimaging data, the full nature of our sample's injuries were unknown. Our outcome variables were derived from the BASC, which is an outdated version of the BASC-2 (Reynolds & Kamphaus, 2004). Finally, although our findings support the validity of subtyping of TBI based on intellectual test performance, the clusters were based on the WISC-III, which has been revised to WISC-IV (Wechsler, 2003), and some differences between the two versions have been noted (Allen, Thaler, Donohue, & Mayfield, 2010; Donders & Janke, 2008). Thus, similar investigation of potential cluster subtypes using the WISC-IV in children with TBI is warranted. However, given studies identifying average and impaired clusters in children with TBI using various assessment instruments, it is likely that the average and impaired clusters identified here would emerge for the WISC-IV and that the latter cluster would exhibit the severest behavioral difficulties.

Conflict of Interest

None declared.

References

Aldenderfer
M. S.
Blashfield
R. K.
Cluster analysis
 , 
1984
Beverly Hills, CA
Sage
Allen
D. N.
Leany
B.
Thaler
N. S.
Cross
C.
Sutton
G.
Mayfield
J.
Attention and memory profiles in pediatric traumatic brain injury
Archives of Clinical Neuropsychology
  
in press
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
)
Babikian
T.
Asarnow
R.
Neurocognitive outcomes and recovery after pediatric TBI: Meta-analytic review of the literature
Neuropsychology
 , 
2009
, vol. 
23
 
3
(pg. 
283
-
296
)
Baguley
L.
Cooper
J.
Felmingham
K.
Aggressive behavior following traumatic brain injury: How common is common?
The Journal of Head Trauma Rehabilitation
 , 
2006
, vol. 
21
 
1
(pg. 
45
-
56
)
Barney
S. J.
Allen
D. N.
Park
B. S.
Thaler
N. S.
Strauss
G. P.
Mayfield
J.
Neuropsychological and behavioral measures of attention assess different constructs in children with traumatic brain injury
The Clinical Neuropsychologist
  
under review
Basson
M. D.
Guinn
J. E.
McElligott
J.
Vitale
R.
Brown
W.
Fielding
L. P.
Behavioral disturbances in children after trauma
Journal of Trauma
 , 
1991
, vol. 
31
 (pg. 
1363
-
1368
)
Bishop
D.
Behaviour after cerebral lesions in children and adults (1962)
Developmental Medicine and Child Neurology
 , 
2008
, vol. 
50
 
5
(pg. 
324
-
324
)
Calhoun
S. L.
Mayes
S. D.
Processing speed in children with clinical disorders
Psychology in the Schools
 , 
2005
, vol. 
42
 
4
(pg. 
333
-
343
)
Crawford
J. R.
Garthwaite
P. H.
Johnson
D. A.
WAIS-R subtest pattern clusters in closed-head injured and healthy samples
The Clinical Neuropsychologist
 , 
1997
, vol. 
11
 (pg. 
249
-
257
)
Curtiss
G.
Vanderploeg
R.
Spencer
J.
Salazar
A.
Patterns of verbal learning and memory in traumatic brain injury
Journal of the International Neuropsychological Society
 , 
2001
, vol. 
7
 
5
(pg. 
574
-
585
)
de Bildt
A.
Sytema
S.
Kraijer
D.
Sparrow
S.
Minderaa
R.
Adaptive functioning and behaviour problems in relation to level of education in children and adolescents with intellectual disability
Journal of Intellectual Disability Research
 , 
2005
, vol. 
49
 
9
(pg. 
572
-
681
)
Dennis
M.
Guger
S.
Roncadin
C.
Barnes
M.
Schachar
R.
Attention-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
)
Dietz
K.
Lavigne
J.
Arend
R.
Rosenbaum
D.
Relation between intelligence and psychopathology among preschoolers
Journal of Clinical Child Psychology
 , 
1997
, vol. 
26
 
1
(pg. 
99
-
107
)
Donders
J.
Cluster subtypes in the WISC-III standardization sample: Analysis of factor index scores
Psychological Assessment
 , 
1996
, vol. 
8
 
3
(pg. 
312
-
318
)
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
)
Donders
J.
Warschausky
S.
WISC-III factor index score patterns after traumatic head injury in children
Child Neuropsychology
 , 
1997
, vol. 
3
 
1
(pg. 
71
-
78
)
Everitt
S. E.
Landau
S.
Leese
M.
Cluster analysis
 , 
2001
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
)
Ganesalingam
K.
Yeates
K.
Ginn
M.
Taylor
H.
Dietrich
A.
Nuss
K.
, et al.  . 
Family burden and parental distress following mild traumatic brain injury in children and its relationship to post-concussive symptoms
Journal of Pediatric Psychology
 , 
2008
, vol. 
33
 
6
(pg. 
621
-
629
)
Gary
K. W.
Arango-Lasprilla
J. C.
Ketchum
J. M.
Kreutzer
J. S.
Copolillo
A.
Novack
T. A.
, et al.  . 
Racial differences in employment outcome after traumatic brain injury at 1, 2, and 5 years postinjury
Archives of Physical Medicine and Rehabilitation
 , 
2009
, vol. 
90
 
10
(pg. 
1699
-
1707
)
Ghosh
A.
Wilde
E.
Hunter
J.
Bigler
E.
Chu
Z.
Li
X.
, et al.  . 
The relation between Glasgow Coma Scale score and later cerebral atrophy in paediatric traumatic brain injury
Brain Injury
 , 
2009
, vol. 
23
 
3
(pg. 
228
-
233
)
Hackbarth
R. M.
Rzeszutko
K. M.
Sturm
G.
Donders
J.
Kuldanek
A. S.
Sanfilippo
D. J.
, et al.  . 
Survival and functional outcome in pediatric traumatic brain injury: A retrospective review and analysis of predictive factors
Critical Care Medicine
 , 
2002
, vol. 
30
 (pg. 
1630
-
1635
)
Hanten
G.
Li
X.
Newsome
M.
Swank
P.
Chapman
S.
Dennis
M.
, et al.  . 
Oral reading and expressive language after childhood traumatic brain injury: Trajectory and correlates of change over time
Topics in Language Disorders
 , 
2009
, vol. 
29
 
3
(pg. 
236
-
248
)
Hawley
C. A.
Behavior and school performance after brain injury
Brain Injury
 , 
2004
, vol. 
18
 
7
(pg. 
645
-
659
)
Heaton
R.
Grant
C.
Matthews
C.
Comprehensive norms for an expanded Halstead–Reitan Battery
 , 
1991
Odessa, FL
Psychological Assessment Resources
Hiekkanen
H.
Kurki
T.
Brandstack
N.
Kairisto
V.
Tenovuo
O.
Association of injury severity, MRI-results and ApoE genotype with 1-year outcome in mainly mild TBI: A preliminary study
Brain Injury
 , 
2009
, vol. 
23
 
5
(pg. 
396
-
402
)
Holm
S.
Schönberger
M.
Poulsen
I.
Caetano
C.
Patients' and relatives' experience of difficulties following severe traumatic brain injury: The sub-acute stage
Neuropsychological Rehabilitation
 , 
2009
, vol. 
19
 
3
(pg. 
444
-
460
)
Hukkelhoven
C. W.
Steyerberg
E. W.
Rampen
A. J.
Farace
E.
Habbema
J. D.
Marshall
L. F.
, et al.  . 
Patient age and outcome following severe traumatic brain injury: an analysis of 5600 patients
Journal of Neurosurgery
 , 
2003
, vol. 
99
 
4
(pg. 
666
-
673
)
Jansen
P. W.
Duijiff
S. N.
Beemer
F. A.
Vorstman
J. A.
Klaassen
P. W.
Morcus
M. E.
, et al.  . 
Behavioral problems in relation to intelligence in children with 22q11.2 deletion syndrome: A matched control study
American Journal of Medical Genetics
 , 
2007
, vol. 
143
 
6
(pg. 
574
-
580
)
Kirkwood
M.
Janusz
J.
Yeates
K.
Taylor
H.
Wade
S.
Stancin
T.
, et al.  . 
Prevalence and correlates of depressive symptoms following traumatic brain injuries in children
Child Neuropsychology
 , 
2000
, vol. 
6
 
3
(pg. 
195
-
208
)
Landis
J. R.
Koch
G. G.
The measurement of observer agreement for categorical data
Biometrics
 , 
1977
, vol. 
33
 (pg. 
159
-
174
)
Langlois
J. A.
Rutland-Brown
W.
Thomas
K. E.
The incidence of traumatic brain injury among children in the United States: Differences by race
Journal of Head Trauma Rehabilitation
 , 
2005
, vol. 
3
 (pg. 
229
-
238
)
Levin
H. S.
Benton
A. L.
Grossman
R. G.
Neurobehavioral consequences of closed head injury
 , 
1982
New York
Oxford University Press
Lieh-Lai
M. W.
Theodorou
A. A.
Sarnaik
A. P.
Limitations of the Glasgow Coma Scale in predicting outcome in children with traumatic brain injury
Journal of Pediatrics
 , 
1992
, vol. 
120
 (pg. 
195
-
199
)
Love
P. F.
Tepas
J. J.
3rd
Wiudyka
P. S.
Masnita-Iusan
C.
Fall-related pediatric brain injuries: The role of race, age, and sex
Journal of Trauma
 , 
2009
, vol. 
67
 
1 Suppl.
(pg. 
S12
-
S15
)
Mathias
J.
Bowden
S.
Bigler
E.
Rosenfeld
J.
Is performance on the Wechsler Test of Adult Reading affected by traumatic brain injury?
British Journal of Clinical Psychology
 , 
2007
, vol. 
46
 
4
(pg. 
457
-
466
)
Mathias
J. L.
Wheaton
P.
Changes in attention and information-processing speed following severe traumatic brain injury: A meta-analytic review
Neuropsychology
 , 
2007
, vol. 
21
 
2
(pg. 
212
-
223
)
Mayes
S. D.
Calhoun
S. L.
Similarities and differences in Wechsler Intelligence Scale for Children – Third Edition (WISC-III) profiles: Support for subtest analysis in clinical referrals
The Clinical Neuropsychologist
 , 
2004
, vol. 
18
 (pg. 
559
-
572
)
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
 
3
(pg. 
865
-
883
)
Muscara
F.
Catroppa
C.
Anderson
V.
The impact of injury severity on executive function 7–10 years following pediatric traumatic brain injury
Developmental Neuropsychology
 , 
2008
, vol. 
33
 
5
(pg. 
623
-
636
)
Nadebaum
C.
Anderson
V.
Catroppa
C.
Executive function outcomes following traumatic brain injury in young children: A five year follow-up
Developmental Neuropsychology
 , 
2007
, vol. 
32
 
2
(pg. 
703
-
728
)
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
)
Rees
L.
Marshall
S.
Hartridge
C.
Mackie
D.
Weiser
M.
Cognitive interventions post acquired brain injury
Brain Injury
 , 
2007
, vol. 
21
 
2
(pg. 
161
-
200
)
Reynolds
C. R.
Bigler
E. D.
Test of Memory and Learning
 , 
1994
Austin, TX
Pro-Ed
Reynolds
C. R.
Kamphaus
R.W.
Behavior Assessment System for Children
 , 
1992
Circle Pines, MN
American Guidance Service
Reynolds
C. R.
Kamphaus
R. W.
BASC-2: Behavioral Assessment System for Children Manual
 , 
2004
2nd ed.
Circle Pines, MN
AGS
Romesburg
C. H.
Cluster analysis for researchers
 , 
1984
Belmont, CA
Lifetime Learning Publications
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
 
4
(pg. 
251
-
263
)
Smith
R. J.
Barth
J. T.
Diamond
R.
Giuliano
A. J.
Goldstein
G.
Nussbaum
P. D.
Beers
S. R.
Evaluation of head trauma
Neuropsychology
 , 
1998
New York
Plenum Press
(pg. 
135
-
170
. In
Stancin
T.
Wade
S.
Walz
N.
Yeates
K.
Taylor
H.
Traumatic brain injuries in early childhood: Initial impact on the family
Journal of Developmental and Behavioral Pediatrics
 , 
2008
, vol. 
29
 
4
(pg. 
253
-
261
)
Taylor
H.
Swartwout
M.
Yeates
K.
Walz
N.
Stancin
T.
Wade
S.
Traumatic brain injury in young children: Postacute effects on cognitive and school readiness skills
Journal of the International Neuropsychological Society
 , 
2008
, vol. 
14
 
5
(pg. 
734
-
745
)
Teasdale
G.
Jennett
B.
Assessment of coma and impaired consciousness: A practical scale
Lancet
 , 
1974
, vol. 
2
 
7872
(pg. 
81
-
84
)
Tonks
J.
Williams
W.
Frampton
I.
Yates
P.
Slater
A.
The neurological bases of emotional dys-regulation arising from brain injury in childhood: A ‘when and where’ heuristic
Brain Impairment
 , 
2007
, vol. 
8
 
2
(pg. 
143
-
153
)
Tramontana
M.
Hooper
S.
Nardolillo
E.
Behavioral manifestations of neuropsychological impairment in children with psychiatric disorders
Archives of Clinical Neuropsychology
 , 
1988
, vol. 
3
 
4
(pg. 
369
-
374
)
van der Heijden
P.
Donders
J.
WAIS-III factor index score patterns after traumatic brain injury
Assessment
 , 
2003
, vol. 
10
 
2
(pg. 
115
-
122
)
Vriezen
E.
Pigott
S.
The relationship between parental report on the BRIEF and performance-based measures of executive function in children with moderate to severe traumatic brain injury
Child Neuropsychology
 , 
2002
, vol. 
8
 
4
(pg. 
296
-
303
)
Wechsler
D.
Manual: Wechsler Intelligence Scale for Children-III
 , 
1991
New York
Psychological Corporation
Wechsler
D.
WISC-IV technical and interpretive manual
 , 
2003
San Antonio, TX
Psychological Corporation
Wells
R.
Minnes
P.
Phillips
M.
Predicting social and functional outcomes for individuals sustaining paediatric traumatic brain injury
Developmental Neurorehabilitation
 , 
2009
, vol. 
12
 
1
(pg. 
12
-
23
)
Wood
R.
Liossi
C.
Neuropsychological and neurobehavioral correlates of aggression following traumatic brain injury
The Journal of Neuropsychiatry and Clinical Neurosciences
 , 
2006
, vol. 
18
 
3
(pg. 
333
-
341
)
Worthington
A.
Wood
R.
Behaviour problems
Psychological approaches to rehabilitation after traumatic brain injury
 , 
2008
Malden, Leicester, England
Blackwell Publishing
(pg. 
227
-
259
)
Yeates
K. O.
Armstrong
K.
Janusz
J.
Taylor
G. H.
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
)
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
)