Developmental Neurorehabilitation, February 2009; 12(1): 12–23

Predicting social and functional outcomes for individuals sustaining

paediatric traumatic brain injury

REBECCA WELLS1, PATRICIA MINNES2, & MARJORY PHILLIPS3

1
University of Waterloo, Psychology, Waterloo, Ontario, Canada, 2Queen’s University, Psychology, Kingston, Ontario,
Canada, and 3Integra Foundation, Toronto, Ontario, Canada

(Received 17 January 2009; accepted 21 January 2009)

Abstract
Objective: To determine the efficacy of models employing the classification of severity of injury, based on either GCS scores
or clinical findings using standardized test scores for Occupational Therapy, Physiotherapy and Psychology in predicting
outcomes associated with paediatric traumatic brain injury (TBI).
Method: Medical records were reviewed to obtain GCS scores and standardized tests used in the clinically based
classification of severity of injury for 30 individuals who sustained a paediatric TBI and were patients of a brain injury
treatment programme. Interviews were conducted with parents to obtain current data on social participation, cognitive
functioning and environmental factors.
Results: Three variables emerged as significant predictors of outcome: age at injury, clinical ratings of injury severity and
environmental factors.
Conclusion: Findings offer preliminary support for the idea that a combination of factors, including age at injury, clinical
expertise and the environment, provide the best estimate of long-term outcome.

Keywords: Traumatic brain injury, paediatrics, clinicians, quality of life, severity, outcomes

Introduction behavioural sequelae noted after moderate-to-severe

Empirically-based studies indicate that paediatric
brain injury is more devastating than was once
thought [1–4]. Unlike adults, the effects of paediatric
brain injury on brain function interact with the
maturation and development of the child. Skills that
are emerging or developing may be affected differ-
ently by brain injury than skills that are already
established. Because the child’s brain is still matur-
ing at the time of injury, the full impact of an insult
to the brain may not become evident for many
months or even years [1, 3, 4]. Moreover, the rate
and recovery of cognitive and behavioural deficits
vary as a function of age at injury, severity of injury
and pre- and co-morbid factors [5].
Accurate identification of the severity of the
traumatic brain injury (TBI) is important for the
development of timely and efficient rehabilitation
programmes. However, predicting which children
require acute and intensive rehabilitation, school and
community-based programming or monitoring is
an inexact science [6]. Severe TBI is most clearly
associated with longer-term impairments in cogni-
tive and behavioural functioning [7–10]. Of the
brain injury, those behaviours related to memory
and executive functioning appear to be the most
commonly affected [11, 12].
In contrast, the empirical data reflecting brain
injuries that are classified as ‘mild’ are less clear. In
a review of 40 research studies conducted on mild
head injury in children and adolescents between
1970–1995, Satz et al. [13] found inconsistent
evidence regarding neuropsychological, academic
and psychosocial outcomes. More specifically,
although 33% of the studies reported adverse
effects of mild brain injury, 45% reported no adverse
effects resulting from mild brain injury. The
remaining 22% of studies did not provide clear
contrasts between mild brain-injured groups and
controls or more severely injured groups. This
review mirrors the findings of other studies in the
TBI literature that highlight variability. In particular,
controlling for pre-injury characteristics, mild brain
injury has been associated with hyperactivity,
inattention and conduct disordered behaviours in
children between the ages of 10–13 years and
adverse academic sequelae [14–16]. In contrast,

Correspondence: Ms Rebecca Wells, MA, University of Waterloo, Psychology, 200 University Ave W, Waterloo, Ontario, N2L3G1 Canada.
E-mail: remwells@uwaterloo.ca
ISSN 1751–8423 print/ISSN 1751–8431 online/09/010012–12 ß 2009 Informa Healthcare USA, Inc.
DOI: 10.1080/17518420902773109
 Predicting social and functional outcomes for individuals sustaining paediatric TBI
arguing that associations between mild TBI and
deficits reflects problems in methodology, research-
ers have failed to find differences between children
with mild TBI and a control group on measures of
intellectual ability, receptive language and memory
capacity [17–24].
Classification of TBI into mild, moderate or
severe groups has traditionally relied upon measure-
ment of coma following insult to the brain [25].
In particular, the Glasgow Coma Scale (GCS) was
developed by Teasdale and Jennett [26] for an adult
population to classify injuries by assessing three
determinants: eye opening, verbal responses and
motor response. Individuals are considered to have
experienced a mild brain injury when their GCS
score is 13–15; moderate TBI (9–12); and severe
TBI (<8). Of note, the GCS was intended to be used
serially over a period of time in order to capture
changes related to both the primary injury
(i.e. contusions, lacerations, diffuse axonal injury)
and secondary injury (i.e. oedema, intracranial
infections) [27].
Although the GCS is widely used by medical
personnel and in the bulk of research studies
investigating brain injury to determine severity of
injury, research into its reliability and validity for
this purpose has offered inconsistent results [28–32].
There are few studies available that specifically
examine the utility of the GCS in assessing severity
of injury in paediatric populations [27, 33].
Furthermore, the majority of studies using severity
of injury as a predictor of outcome use only one GCS
score, including the best, the worst or the initial
GCS, to determine injury severity. This practice is
problematic not only because there is inconsistency
across studies with regard to which GCS score is
used, but also in light of the fact that recent studies
have reported that low GCS does not necessarily
mean poor prognosis [34]. In fact, several studies
have reported that secondary injury factors may be
more important in predicting outcomes than the
primary injury [27, 33, 34].
In those studies that attempt to predict outcomes
following paediatric brain injury, a number of
variables have been suggested as possible confounds
for the ability to make predictions based on severity
of injury. Unlike TBI during adulthood, paediatric
TBI occurs during a time of ongoing physical and
cognitive development [35]. Thus, in examining
outcomes related to paediatric TBI, it is important
to consider variables such as age at injury which
may impact findings regarding outcomes for these
individuals [4].
Another set of potentially confounding variables
involves the influence of demographic and environ-
mental factors including gender, ethnicity and
parental factors such as maternal education,
13
socioeconomic status (SES) and parental stress
levels. In a study of 109 school-aged children with
TBI, Donders and Nesbit-Greene [36] found that
demographic variables, including ethnicity and SES,
accounted for a significant portion of the variability
in performance on neuropsychological tests such as
the Wechsler Intelligence Scale for Children (WISC)
and the Children’s Category Test. Similarly, Hawley
[37] and Taylor et al. [10] found that parental
marital status and the family environment were
associated with behavioural problems up to 4 years
post-injury in samples of school-aged children
sustaining TBI.
One recent direction that is being taken in the
study of paediatric TBI involves the use of neurop-
sychological assessments to predict outcomes.
Rather than relying solely on indices of severity of
injury such as the GCS or Post Traumatic Amnesia,
several studies have employed neuropsychological
test scores during the initial post-injury assessment
phase to predict short- and long-term outcomes
following a brain injury [9, 38]. In a study of 58
children with TBI, Miller and Donders [9] found
that T-scores below 45 on the California Verbal
Learning Test-Children’s Version (CVLT-CV)
resulted in an increase of up to 13 times that
children would be placed in special education classes
within 2 years following injury. Kinsella et al. [38]
reported similar results in their study of 29 school-
aged children sustaining a TBI. More specifically,
lower scores on the Rey’s Auditory Verbal Learning
Test and the Controlled Oral Word Association Test
were predictive of placement in special education at
2 years post-injury.
In addition to the trend toward the utilization
of standardized testing in determining outcomes,
there is currently a movement toward evidence-
based practice in the health care system. Although
not extensively tested, there is evidence to suggest
that clinical expertise is important in determining
outcome [39]. The usefulness of clinical expertise in
predicting outcomes is based on the idea that experts
are trained and experienced in recognizing mean-
ingful patterns when assessing an individual.
Additionally, experts acquire extensive knowledge
and organize it in ways that allow for the retrieval of
relevant knowledge automatically. It is easy to see
how these abilities would be useful in the assessment
and treatment of brain injury when one considers
how complex the relationship between injury char-
acteristics and outcomes can be. It is necessary to be
familiar with the variety of patterns that can emerge
following an injury to the brain and to be able to
recognize variables that may directly or indirectly
impact future functioning.
In addition to exploring alternate methods of
operationalizing predictors, researchers have looked
14 R. Wells et al.
at measures of outcome. Typically, standardized
scores on tests of neurocognitive functioning are
used to operationalize outcome [10]. However, more
recent approaches to understanding health and
disability view outcome from a functional perspec-
tive and include quality of life as an important aspect
of functioning. Although quality of life can be
defined in a number of ways, it typically involves
being an active and contributing member of society
including at least some level of involvement in social
activities [40]. As such, it is important to consider an
individual’s level of participation in social contexts
when examining outcomes. Unfortunately, social
competence is also an area in which survivors
of paediatric brain injury tend to do poorly. This
difficulty is due primarily to the cognitive and
behavioural challenges that TBI survivors face,
including difficulties with memory and language as
well as their impaired ability to inhibit inappropriate
behaviours. Unfortunately, individuals who have
sustained a paediatric TBI continue to face restric-
tions in their social participation for years following
injury [40]. Areas of social participation include
peer socialization, family activities such as decision-
making, structured community activities such as
sports leagues and school-based activities with other
children in the classroom. In their study of 60
school-aged children up to 6.7 years post-injury,
Bedell and Dumas [40] found that 73% of their
sample experienced restrictions in social participa-
tion in at least one area of participation. Other
investigations focusing on individuals with develop-
mental disabilities or acquired brain injuries have
employed the AIMS Interview which examines
social participation from an acculturation perspec-
tive (i.e. identifying disability-related needs requiring
support in order to promote community participa-
tion) [41–44]. More specifically, the AIMS examines
whether individuals’ needs are being identified and/
or supported in a way that promotes community
involvement. In a study of children and adolescents
with acquired brain injuries, Alvarez et al. [41] found
that only 64% of the sample of 28 children and
adolescents had their needs identified and sup-
ported. In a similar study of 19 children and
adolescents with TBI, Vilela et al. [44] found that
although 74% of the sample was participating in
social activities, only 16% of the sample had their
social participation needs identified and supported.
Additionally, 26% of the sample did not have their
needs identified and were not participating in social
activities.
One goal of the present study was to determine the
efficacy of additive models employing the classifica-
tion of severity of injury based on either GCS scores,
a combination of standardized test scores or clinical
expertise in predicting functional outcomes
associated with paediatric brain injury. In light of
the recent shift toward the inclusion of clinical
expertise in determining outcomes and the lack of
consistent support for the utility of the GCS in
predicting outcomes, it was hypothesized that
severity of injury classifications based on clinical
expertise (Clinical Team Severity Rating) would
account for more variance in long-term social
participation outcomes than injury severity ratings
based on either standardized test scores
(Standardized Test Severity Rating) or GCS scores
(Glasgow Coma Scale Severity Rating). Similarly, it
was hypothesized that the Clinical Team Severity
Rating (CTSR) would account for more variance in
long-term cognitive outcomes than either the
Standardized Test Severity Rating (STSR) or the
Glasgow Coma Scale Severity Rating (GCSR).
Additionally, it was hypothesized that age at injury
and environmental factors would account for a large
portion of the variance not accounted for by severity
of injury in the long-term social participation and
cognitive outcomes of paediatric brain injury.
Method
Participants
Participants were 30 pairs, each consisting of a
parent (27 female) and his/her child (12 female) who
had sustained a TBI between birth and 10 years of
age (mean age at injury ¼ 5.32 years, SD ¼ 2.75
years). Each child was a current or former patient
at a Paediatric ABI Programme in Southeastern
Ontario. The time since injury ranged from 3.92–
16.61 years (mean time since injury ¼ 10.05,
SD ¼ 2.93 years).
Measures
Archival data. A review of consenting patients’
medical records allowed for the collection of data
obtained at the time of injury including available
GCS, psychological test data including standardized
scores on memory tests, physiotherapy data consist-
ing of scores on the Bruininks-Oseretsky Test of
Motor Proficiency and occupational therapy data
consisting of scores on the Beery Developmental
Test of Visual-Motor Integration.
Glasgow Coma Scale Severity Ratings (GCSR)
were assigned based on the first available GCS score
obtained through the review of medical files. Based
on the protocol established by Teasdale and Jennett
[26] and employed by numerous researchers, indi-
viduals were classified as having a mild brain injury
when their score was between 13–15. A score of
9–12 was considered to indicate a moderate brain
injury and a score of 8 or less was classified as a
 Predicting social and functional outcomes for individuals sustaining paediatric TBI
severe brain injury. Although only 15 participants
had actual GCS scores recorded in their files, the
other 15 participants were not assessed using the
GCS because their injuries did not warrant assess-
ment. These were cases in which there was no loss
of consciousness or memory disturbance noted.
Thus, for the purposes of the present study, these
participants were deemed to have had a GCS score
of 15 and were classified as having mild injuries.
Psychological, occupational therapy and physical
therapy assessment data collected within 1 year of
the date of injury were combined to determine the
Standardized Test Severity Rating (STSR). More
specifically, based on clinical practices of members
of the ABI Programme team, standardized test
scores falling more than 2 SD below the mean
indicated impairment in the area being measured
(i.e. intellectual functioning, gross motor ability and
fine motor ability). Furthermore, a lack of impair-
ment in any area was classified as a mild injury,
impairment in any one area was classified as a
moderate injury and impairment in two or more
areas was classified as a severe injury.
Memory is one of the most commonly reported
deficits associated with TBI [36] and standardized
memory tests were among the most consistently used
measures during the psychological assessment of the
children and adolescents in this study. Impairment
of intellectual functioning was therefore determined
using the ‘general memory’ standardized score from
either the Wide Range Assessment of Memory and
Learning (WRAML) [45] or the Children’s Memory
Scale (CMS) [46]. Psychometric testing of both
measures indicated a high level of internal consis-
tency (WRAML: Cronbach’s  ¼ 0.93; CMS:
Spearman-Brown split half reliability ¼ 0.91) and
good test–re-test reliability (WRAML: mean interval
49 days; r ¼ 0.81; CMS: mean interval 60 days;
r ¼ 0.84).
Standardized scores from the Bruininks-Oseretsky
Test of Motor Proficiency (BOTMP) [47] were used
to determine impairment in participants’ gross motor
functioning and standardized scores on the Beery
Developmental Test of Visual Motor Integration
(VMI) [48] were used to determine fine motor
impairment, as they were the most commonly
reported scores in the physiotherapy and occupa-
tional therapy reports, respectively. For the purposes
of the present study, the gross motor composite score
as measured by the physiotherapist was used to assess
gross motor impairment. Psychometric testing of the
BOTMP indicated a high level of internal consistency
(Cronbach’s  ¼ 0.86). The fine motor composite
VMI score as measured by the occupational therapist
was used to determine fine motor impairment.
Psychometric testing of the VMI indicated a high
15
level of internal consistency (Cronbach’s  ¼ 0.82)
and good test–re-test reliability (r ¼ 0.89).
The Clinical Team Severity Ratings (CTSR) were
determined within 1 year of the injury by the ABI
Programme team, which included a physician,
a psychologist, a physiotherapist, an occupational
therapist and a speech-language pathologist. To
determine severity of injury, each team member
independently rated the patient according to the
severity of the team member’s unique area of
function (i.e. cognitive, language, gross motor,
fine motor/skills of independent living, behaviour).
Severity was interpreted in the context of a perceived
change in function from estimated pre-accident
levels. The team then met together to review the
individual ratings and to achieve a consensual rating
of severity for each patient. Thus, the CTSR were
partially based on the standardized test scores, used
to determine STSR, as well as GCS scores; however,
additional test data and the clinicians’ expertise in
the area of brain injury were also used to provide
additional information that these scores alone might
not provide.
Current data. An interview was conducted with one
parent of each of the children included in the study.
The interview was completed over the telephone
following the collection of archival data from patient
medical records. Questionnaires administered in
the telephone interview included the Demographic
Questionnaire, the Child and Adolescent Scale of
Environment (CASE), the Child and Adolescent
Factors Inventory (CAFI) and the Child and
Adolescent Scale of Participation (CASP) [49].
Demographic data collected included the age and
sex of the brain injury survivor and parent, the date
and type of injury, the annual household income and
the parent’s highest level of education completed.
The Child and Adolescent Scale of Environment
(CASE) consists of 18 items pertaining to problems
experienced with physical, social and attitudinal
aspects of the environment as well as problems
related to the quality and availability of services or
assistance that children with disabilities may need
[49]. The CASE was developed as an inventory
of related and unrelated items pertaining to the
environment. A factor analysis conducted by
G. Bedell (personal communication, 20 June 2006)
identified six factors in the CASE (school support/
attitudes, home and community support/attitudes,
family stress/finances, physical design of buildings,
crime/violence and transportation) accounting for
79% of the variance explained (Cronbach’s
 ¼ 0.86). For the present study, only those items
related to school, home and community support
and attitudes as well as family stress and finances
16 R. Wells et al.
factors (i.e. Questions 4–7, 9, 10 and 12–15) were
included in the analyses, as the remaining factors
were not highly endorsed by participants. Questions
are worded ‘Does your child currently experience
any problems with . . .’. Each item is rated on a 3-
point scale: no problem (1), little problem (2) or big
problem (3). Additionally, each item has a ‘not
applicable’ response option. Psychometric testing
of the CASE indicates a high level of internal
consistency (Cronbach’s  ¼ 0.91).
The Child and Adolescent Scale of Participation
(CASP) assesses social functioning of children and
youth within the context of the WHO ICF [49].
The CASP consists of 20 items pertaining to home,
school and community life for children and youth
[40, 49]. Questions are worded ‘Compared to other
children your child’s age, what is your child’s current
level of participation in . . .?’ and include items
pertaining to social, play or leisure activities, family
chores, academic activities and money management.
Each item is rated on a 4-point scale: age expected
(4), somewhat limited (3), very limited (2) or
unable (1). Additionally, each item has a ‘not
applicable’ response option. A factor analysis con-
ducted by G. Bedell (personal communication,
20 June 2006) identified two factors (social/leisure/
communication (SLC) and activities of daily living/
personal care) accounting for 68% of the variance
explained (Cronbach’s  ¼ 0.96). As the purpose of
the current study was examining social participation,
only those questions pertaining to the first factor,
SLC (i.e. questions 1, 2, 6–8, 10, 12 and 15), were
included in the analyses. Psychometric testing of the
CASP indicates a high level of internal consistency
(Cronbach’s  ¼ 0.98).
The Child and Adolescent Factors Inventory
(CAFI) assesses cognitive and behavioural function-
ing of children and youth within the context of the
WHO ICF [49]. The CAFI consists of 15 items
pertaining to impairments and child-related factors
that are typically of concern for family caregivers of
children with TBI [40, 49]. The CAFI was devel-
oped as an inventory of related and unrelated items
pertaining to cognitive and behavioural functioning.
A factor analysis conducted by G. Bedell (personal
communication, 20 June 2006) identified three
factors (cognitive/behavioural/speech (CBS), move-
ment/vision/health and hearing) accounting for
62% of the variance explained (Cronbach’s
 ¼ 0.86). For the present study, only items pertain-
ing to CBS (i.e. questions 1–7) were included in
the analyses, including problems with attention
and concentration, motivation and judgement.
Questions are worded ‘Does your child currently
have any difficulty . . .’. Each item is rated on a
3-point scale: no problem (1), little problem (2) or
big problem (3).
All questionnaires were scored according to the
procedures determined by the original authors.
CASP, CAFI and CASE summary scores were
calculated by adding the scores from each applicable
item and dividing by the total maximum possible
score based on the number of items rated. These
numbers were then multiplied by 100 to conform to
a 0–100 point scale. Higher CASP scores indicate a
greater degree of age-expected participation, higher
CAFI scores indicate more child-related problems, a
greater impact of problems or a combination of the
two and higher CASE scores reflect more environ-
ment problems, a greater impact of problems or
a combination of the two.
Results
Demographic data indicated that 25 participants
were living in a 2-parent home with annual house-
hold incomes ranging from $15,000–$200,000
with a mean of $68,965.52 (SD ¼ $40,247.87).
Ten parents had completed high school, 12 had
completed college, three had obtained an under-
graduate degree and one had obtained a graduate
degree. Only four parents had not completed high
school.
Table I provides the overall means and standard
deviations for all continuous variables as well as the
means and standard deviations for each continuous
variable broken down by injury severity within each
severity of injury rating method: Glasgow Coma
Scale Severity Rating (GCSR), Standardized Test
Severity Rating (STSR) and Clinical Team Severity
Rating (CTSR). One Way ANOVAs revealed no
significant group differences between mean scores
on any of the measures.
In calculating the STSR it was determined that
24 children experienced impairments in cognitive
functioning, 22 experienced impairments in gross
motor functioning and 26 experienced impairments
in fine motor functioning. An examination of CASP
scores revealed that 80% of the sample was not
participating in age-expected levels of social partici-
pation across settings (i.e. home, school or commu-
nity). Additionally, 100% of the sample was found
to be experiencing difficulties in two or more areas
of cognitive functioning (CAFI), with memory and
learning being the most commonly reported diffi-
culties. Similarly, 100% of the sample reported
experiencing problems with support and attitudes
in the environment (CASE) with the most common
difficulties being attitudes of people at school and
in the community.
A correlational analysis was conducted to examine
the relationships between the dependent variables
(Child and Adolescent Scale of Participation (CASP)
 Predicting social and functional outcomes for individuals sustaining paediatric TBI 17

Table I. Means and standard deviations of dependent and independent variables.

Injury age (years) CASE CASP CAFI

n M SD M SD M SD M SD

Overall sample 30 5.32 2.75 44.72 13.92 88.89 10.73 60.56 15.81
GCSR
Mild 17 5.09 2.58 45.58 15.90 90.07 10.55 58.82 16.59
Moderate 3 7.82 2.51 38.89 9.62 91.67 11.83 65.28 15.77
Severe 10 4.95 2.97 45.00 11.92 86.04 11.30 62.08 15.65
STSR
Mild 20 5.27 2.43 45.42 15.17 90.31 10.43 58.33 17.36
Moderate 4 7.60 1.20 45.83 14.43 87.50 7.22 68.75 7.98
Severe 6 3.95 3.76 41.67 10.54 85.07 13.99 62.50 13.69
CTSR
Mild 8 4.72 1.02 43.75 13.18 93.36 6.98 53.13 18.33
Moderate 11 5.41 2.75 42.42 12.05 90.06 9.46 59.85 15.40
Severe 11 5.67 3.33 47.73 16.70 84.47 13.09 66.67 12.91

Table II. Correlations among dependent and independent
variables (n ¼ 30).
CASP
GCSR 0.17
STSR 0.20
CTSR 0.34
CASE 0.64**
Number of parents in home 0.05
Household income 0.04
Parent education 0.13
Child age at injury 0.37*
Time since injury 0.16
**Significant at the 0.01 level (2-tailed); *Significant at the
0.05 level (2-tailed).
and Child and Adolescent Factors Inventory (CAFI))
and the independent variables (Child and Adolescent
Scale of Environment (CASE), Glasgow Coma Scale
Severity Rating (GCSR), Standardized Test Severity
Rating (STSR), Clinical Team Severity Rating
(CTSR) and age at injury). The results of this ana-
lysis are presented in Table II. Results indicated that
age at injury was positively correlated with CASP
scores. Thus, younger age at injury was related to
lower current social participation. Additionally,
CASE scores were negatively correlated with CASP
scores and positively correlated with CAFI scores.
More specifically, better environments were related
to higher levels of social participation and lower
levels of cognitive difficulty. Conversely, parent
education, household income and injury severity
ratings were not significantly related to either of the
outcome measures. However, there were trends with
respect to the CTSR with both CASP and CAFI
such that greater severity of injury was associated
with lower levels of social participation and higher
levels of cognitive functioning difficulties.
Forced entry hierarchical regression analyses were
conducted to test all hypotheses. In each analysis,
age at injury was entered in the first block; a rating
of severity of injury (GCSR, STSR or CTSR) was
CAFI entered in the second block and environment
(CASE) scores were entered in the third block.
0.10 Current environment was used as an independent
0.15
0.34
variable as it is a factor external to the child and
0.55** environment has been shown to impact on children’s
0.06 functioning [37]. Similarly, as the focus of the
0.12 present study was examining children’s functional
0.10
outcomes, social participation and cognitive func-
0.15
0.03 tioning were used as dependent variables. The
outcome variable was dependent on the hypothesis
in question and was either social participation
(CASP) scores or cognitive functioning (CAFI)
scores.
Severity of injury ratings as predictors
of social participation
It was hypothesized that Clinical Team Severity
Rating (CTSR) would account for more variance
in social participation than the Standardized Test
Severity Rating (STSR) and Glasgow Coma Scale
Severity Rating (GCSR). Forced entry hierarchical
regression analyses were conducted with Child and
Adolescent Scale of Participation (CASP) scores
entered as the dependent variable. R2 change scores
for the GCSR (R2 ¼ 0.028, p ¼ 0.347) and STSR
(R2 ¼ 0.026, p ¼ 0.369) were non-significant (see
Tables III and IV). The R2 change score for CTSR
(R2 ¼ 0.153, p ¼ 0.023), however, was significant,
with CTSR accounting for 15.3% of the variance in
social participation outcomes (see Table V). Further
examination of the regression equation revealed that
age at injury was acting as a suppressor. That is, even
though the bivariate correlation between CTSR and
CASP was non-significant, the inclusion of age at
injury removed some of the error variance in CTSR
18 R. Wells et al.

Table III. Social participation as predicted by injury age, Glasgow coma scale severity rating (GCSR)
and child and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 89.181 0.138 0.138 0.043 0.372

2 Injury age 84.578 0.166 0.028 0.347 0.372
GCSR 0.168
3 Injury age 107.838 0.519 0.353 <0.001 0.270
GCSR 0.187
CASE 0.603

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

Table IV. Social participation as predicted by injury age, standardized test severity rating (STSR) and
child and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 81.181 0.138 0.138 0.043 0.372

2 Injury age 84.809 0.164 0.026 0.369 0.353
STSR 0.162
3 Injury age 110.182 0.538 0.374 <0.001 0.240
STSR 0.236
CASE 0.625

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

Table V. Social participation as predicted by injury age, clinical team severity rating (CTSR) and child
and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 81.181 0.138 0.138 0.043 0.372

2 Injury age 91.160 0.291 0.153 0.023 0.425
CTSR 0.395
3 Injury age 109.900 0.577 0.286 <0.001 0.321
CTSR 0.312
CASE 0.549

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

that was not correlated with CASP scores, thereby
increasing the multiple correlation in this model.
Severity of injury ratings as predictors of long-term
cognitive outcomes
It was hypothesized that Clinical Team Severity
Rating (CTSR) would account for more variance in
long-term cognitive outcomes than the Standardized
Test Severity Rating (STSR) and Glasgow Coma
Scale Severity Rating (GCSR), R2 change scores
from the first block (age at injury) to the second
block (severity rating) were compared for each
rating system when Child and Adolescent Factors
Inventory (CAFI) scores were entered as the
dependent variable. Again, R2 change scores for
the GCSR (R2 ¼ 0.011, p ¼ 0.585) and STSR
(R2 ¼ 0.018, p ¼ 0.484) were non-significant (see
Tables VI and VII); however, the R2 change score
for the CTSR (R2 ¼ 0.135, p ¼ 0.048) was signifi-
cant, accounting for 13.5% of the variance in CAFI
scores (see Table VIII). Further examination of the
regression equation revealed that age at injury was
acting as a suppressor. As with the CASP model,
even though the bivariate correlation between CTSR
and CAFI was non-significant, the inclusion of age
at injury increased the multiple correlation in the
model, suggesting that age at injury was suppressing
some of the error variance in CTSR that was not
correlated with CAFI scores.
 Predicting social and functional outcomes for individuals sustaining paediatric TBI 19

Table VI. Cognitive functioning as predicted by injury age, Glasgow coma scale severity rating (GCSR)
and child and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 64.986 0.021 0.021 0.445 0.145

2 Injury age 61.868 0.032 0.011 0.585 0.145
GCSR 0.105
3 Injury age 30.885 0.320 0.288 0.003 0.053
GCSR 0.122
CASE 0.545

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

Table VII. Cognitive functioning as predicted by injury age, standardized test severity rating (STSR)
and child and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 64.986 0.021 0.021 0.445 0.145

2 Injury age 60.536 0.039 0.018 0.484 0.130
STSR 0.135
3 Injury age 26.728 0.345 0.306 0.002 0.027
STSR 0.201
CASE 0.565

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

Table VIII. Cognitive functioning as predicted by injury age, clinical team severity rating (CTSR) and
child and adolescent scale of environment (CASE).

Model Predictor B R2 R2 p 

1 Injury age 64.986 0.021 0.021 0.445 0.145

2 Injury age 51.201 0.156 0.135 0.48 0.195
CTSR 0.370
3 Injury age 26.265 0.389 0.233 0.004 0.101
CTSR 0.295
CASE 0.496

R2 reflects the amount of variance in the dependent variable accounted for by the variable entered
on that block. R2 reflects the amount of variance in the dependent variable accounted for by all variables
entered to that point.

Models to predict social participation For the STSR (Table IV), adding severity of injury
in the second block resulted in a non-significant
It was hypothesized that age at injury and Child and
R2 change score; however, adding CASE scores in
Adolescent Scale of Environment (CASE) scores
the third block accounted for an additional 37.4%
would account for a large portion of the variance in
of the variance in CASP scores.
Child and Adolescent Scale of Participation (CASP)
Finally, for the model employing CTSR
scores not accounted for by severity of injury ratings.
(Table V), adding severity of injury ratings in the
As can be seen in Tables III–V, age at injury added
second block accounted for an additional 15.3% of
in the first block uniquely accounted for 13.8%
the variance in CASP scores. Additionally, adding
of the variance in CASP scores, regardless of which
CASE scores in the third block accounted for an
severity of injury rating system was employed.
additional 28.6% of the variance in CASP scores.
However, as can be seen in Table III, for the
model employing the GCSR, adding the severity
Models to predict cognitive functioning
of injury rating in the second block and CASE scores
in the third block resulted in non-significant R2 It was hypothesized that age at injury and Child and
change scores for both. Adolescent Scale of Environment (CASE) scores
20 R. Wells et al.
would account for a large portion of the variance
in Child and Adolescent Factors Inventory (CAFI)
scores not accounted for by severity of injury ratings.
As can be seen in Tables VI–VIII, age at injury
added in the first block did not significantly account
for any of the variance in CAFI scores, regardless
of the severity of injury rating system employed.
For the GCSR model (Table VI), adding severity
of injury ratings in the second block resulted in a
non-significant R2 change score. However, adding
CASE scores in the third block accounted for 28.6%
of the variance.
Similarly, for the STSR model (Table VII),
adding severity of injury ratings in the second
block resulted in a non-significant R2 change score;
however, adding CASE scores in the third block
accounted for 30.6% of the variance.
Conversely, for the CTSR model (Table VIII),
adding severity of injury ratings in the second block
accounted for an additional 13.5% of the variance
in CAFI scores. Furthermore, adding CASE scores
in the third block accounted for an additional 23.3%
of the variance in CASP scores.
Discussion
The purpose of the present study was two-fold: The
first goal was to examine the amount of variance in
functional outcomes accounted for by each of three
methods of classifying severity of injury: GCS scores
(GCSR), Standardized Test scores (STSR) and
Clinical Team ratings (CTSR). The second goal
was to determine the efficacy of additive models,
each employing one of the three methods of
classification, in predicting long-term social and
cognitive outcomes associated with paediatric TBI.
The results of this study indicated that clinical
team ratings accounted for more variance in both
social and cognitive outcomes than more standar-
dized ratings (i.e. GCSR or STSR). This finding
is important in light of the evidence that decisions
regarding the allocation of services and supports
following an injury are often made in the early stages
of recovery and are predicated on the basis of
severity of injury. This finding also lends support
to the recent move toward employing clinicians’
expertise in predicting health care outcomes [39].
Furthermore, despite the fact that the GCS was
developed to be used serially in adult populations
to track the depth and duration of coma and
impaired consciousness following an injury to the
brain, single GCS scores have consistently been used
as the measure of severity of injury in both treatment
planning and research with children [13]. However,
the results of the present study, in which GCS
scores did not account for any variance in social or
cognitive outcomes, do not support reliance on a
single GCS score as a valid predictor of future social
participation and cognitive outcome for children.
Findings from the present study also support
the idea that outcomes are not merely the product
of the severity of the primary physical injury itself,
but that additional factors play a role in determining
long-term functional outcomes [4, 13]. For example,
age at injury appears to play an important role in
predicting outcomes. This makes sense when con-
sidering that child and adolescent brains are still
developing at the time of injury. Thus, the injury
disrupts both current abilities and ongoing develop-
ment. As the findings of the present study suggest,
the older children are when they sustain their brain
injury, the fewer restrictions they experience in social
participation. There are several reasons that this may
be the case. First, a child who sustains an injury
at an older age has had more of an opportunity to
learn appropriate social behaviours than a toddler
or infant who sustains an injury. Although some
of the social skills may be lost or impaired, the older
child may still retain some knowledge and/or skill
related to social conduct, whereas the toddler or
infant would have to learn these skills. Additionally,
the acquisition of social skills may be negatively
affected by the cognitive and behavioural impair-
ments typically associated with TBI. More specifi-
cally, children with TBI often experience both
expressive and receptive language delays [22]. For
example, children with TBI may have difficulty
interpreting words with more than one meaning or
may have difficulty communicating their thoughts.
They also frequently experience difficulty controlling
impulsive behaviour, which could lead to difficulties
in social situations [23].
In addition to age at injury, the environment in
which the child lives appears to be related to social
participation and cognitive functioning. That is, if
a child is being reared in a nurturing, supportive
environment and is accepted by members of the
community they are more likely to be receiving the
support necessary to participate in age-appropriate
social activities as well as assistance with improving
cognitive functioning that may have been impaired
through the brain injury.
Consistent with the findings of Bedell and
Dumas [40], 80% of the participants in the present
study experienced at least some level of restriction
in their social participation, with the most common
restrictions being faced in social activities with same-
aged peers. Additionally, all participants reported
experiencing some level of cognitive impairment,
with motivation and judgement being the most
frequently reported problems. These findings lend
support to the call for increased efforts in addressing
social participation needs and cognitive functioning
 Predicting social and functional outcomes for individuals sustaining paediatric TBI
in individuals who have sustained a TBI [40].
As expected, mean social participation scores
decreased as the severity of injury based on clinical
team ratings increased. Similarly, when examining
severity of injury based on clinical team ratings,
individuals experienced greater frequency and/or
intensity of cognitive difficulties when their injuries
were rated as more severe. However, this was not the
case when severity of injury was based on Glasgow
Coma Scale scores or standardized test ratings,
suggesting that the addition of clinical expertise
may provide a better indication of future outcomes.
Limitations and future directions
This type of study has not been conducted pre-
viously and provides a foundation for future
research. The findings are interesting but prelimin-
ary and therefore must be interpreted with caution.
Additionally, there are several limitations that must
be taken into consideration when interpreting the
results.
Sample size and power. All participants in this study
were recruited from the same treatment facility and
live within a small geographical area. Furthermore,
participants were from a population that had actively
sought information and resources to assist them in
the rehabilitation process. Unfortunately, there are
many other families living with the effects of TBI
who are unaware of the resources available to assist
them. Thus, the sample may not be representative
of the general population of children and adolescents
with TBI. Additionally, only 43% of the potential
sample was included in the analyses. It is possible
that those individuals who declined participation
were feeling overwhelmed by their current situation
and participation in this study may have been
perceived as an additional burden that they did
not need.
Although the sample size is comparable to other
clinical research samples in the TBI research
literature [50, 51] the sample size was small
(n ¼ 30), yielding relatively low power (0.713). It is
possible, therefore, that smaller effects were not
detected. Obtaining larger, truly representative
samples is difficult given the nature of research;
however, future research should attempt to obtain
larger samples and include those individuals not
receiving information and support as well as those
individuals who are doing well but believe they have
nothing to offer to the study.
Design issues (i.e. archival data). Although data
regarding the injury and impairments provided a
basis for examining differences in the classification
21
of severity of injury, there are a number of variables
thought to be involved in outcomes that could
not be assessed. For example, research has shown
that it is important to consider data related to an
individual’s pre-injury functioning as it permits more
accurate measures of the effects of injury by allowing
an opportunity to control for pre-existing confounds
[4, 52]. As the present study relied on the availability
of data collected at the time of (or shortly after)
injury there was no opportunity to gather valid and
reliable data regarding pre-injury functioning.
Similarly, the present study employed a measure
of the current environment. Although current
environment is an important factor to be considered
when examining issues in social participation, several
studies have reported significant environmental
changes, including divorce and income loss, within
the months and years following TBI [38]. Thus,
including a measure of environment at or shortly
after injury may allow for the development of a
more predictive model of outcome following TBI.
Measuring social participation and environmental
change over time might allow for a more in depth
understanding of factors that contribute to children
and adolescents’ engagement in social activities.
Conclusion
Overall the findings of the present study offer
preliminary support for the idea that a single GCS
score or even a given set of neuropsychological tests
do not reliably predict outcomes for children and
adolescents who sustain a TBI. Rather, it appears
necessary to apply clinical expertise in the context
of a biopsychosocial model in order to determine
how to provide supports and resources that will
allow for the best possible quality of life following
a paediatric TBI.
Declaration of interest: The authors report no
conflicts of interest. The authors alone are respon-
sible for the content and writing of the paper.
References
1. Bond Chapman S, McKinnon L. Discussion of developmental
plasticity: Factors affecting cognitive outcome after pediatric
traumatic brain injury. Journal of Communication Disorders
2000;33:333–334.
2. Brooks BM, Rose FD, Johnson DA, Andrews TK,
Gulamali R. Support for children following traumatic brain
injury: The views of educational psychologists. Disability and
Rehabilitation 2003;25:51–56.
3. Gil AM. Neurocognitive outcomes following pediatric
brain injury: A developmental approach. Journal of School
Psychology 2003;41:337–353.
22 R. Wells et al.
4. Taylor HG, Alden J. Age-related differences in outcomes
following childhood brain insults: An introduction and
overview. Journal of the International Neuropsychological
Society 1997;3:555–567.
5. Beaulieu CL. Rehabilitation and outcome following pediatric
traumatic brain injury. Surgical Clinics of North America
2002;82:1–14.
6. Cassidy LD, Potoka DA, Adelson PD, Ford HR.
Development of a novel method to predict disability after
head trauma in children. Journal of Pediatric Surgery
2003;38:482–485.
7. Braga LW, Da Paz AC, Ylvisaker M. Direct clinician-
delivered versus indirect family-supported rehabilitation of
children with traumatic brain injury: A randomized con-
trolled trial. Brain Injury 2005;19:819–831.
8. Ewing-Cobbs L, Fletcher JM, Levin HS, Francis DJ,
Davidson K, Miner ME. Longitudinal neuropsychological
outcome in infants and preschoolers with traumatic brain
injury. Journal of the International Neuropsychological
Society 1997;3:581–591.
9. Miller LJ, Donders J. Prediction of educational outcome after
pediatric traumatic brain injury. Rehabilitation Psychology
2003;48:237–241.
10. Taylor HG, Yeates KO, Wade SL, Drotar D, Stancin T,
Minich N. A prospective study of short- and long-term
outcomes after traumatic brain injury in children: Behavior
and achievement. Neuropsychology 2002;16:15–27.
11. Slomine BS, Gerring JP, Grados MA, Vasa R, Brady KD,
Christensen JR, Denckla MB. Performance on measures of
‘executive function’ following pediatric traumatic brain
injury. Brain Injury 2002;16:759–772.
12. Wassenberg R, Max JE, Lindgren SD, Schatz A. Sustained
attention in children and adolescents after traumatic brain
injury: Relation to severity of injury, adaptive functioning,
ADHD and social background. Brain Injury 2004;18:
751–764.
13. Satz P, Zaucha K, McCleary C, Light R. Mild head injury
in children and adolescents: A review of studies (1970–1995).
Psychological Bulletin 1997;122:107–131.
14. Brown G, Chadwick O, Shaffer P, Rutter M, Traub M.
A prospective study of children with head injuries.
Psychological Medicine 1981;11:63–78.
15. Horowitz I, Costeff H, Sadan N, Abraham E, Geyer S,
Najenson T. Childhood head injuries in Israel: Epidemiology
and outcome. International Rehabilitation Medicine 1983;
4:32–36.
16. Segalowitz SJ, Brown D. Mild head injury as a source of
developmental disabilities. Journal of Learning Disabilities
1991;24:551–559.
17. Anderson V, Catroppa C, Morse S, Haritou F, Rosenfeld J.
Outcome from mild head injury in young children: A
prospective study. Journal of Clinical & Experimental
Neuropsychology 2001;23:705–717.
18. Bijur PE, Haslum M, Golding J. Cognitive and behavioural
sequelae of mild head injury in children. Pediatrics
1990;86:337–344.
19. Butterbaugh G. Traumatic brain injuries in adolescents:
A psychosocial and neuropsychological review. In: Singer M,
Singer L, Anglin T, editors. Assessing adolescents at risk.
Lexington: Lexington Publishing; 1993. pp.293–332.
20. Casey R, Ludwig S, McCormick MC. Morbidity
following minor head trauma in children. Pediatrics 1986;
78:497–502.
21. Knights RM, Ivan LP, Ventureyra EC, Bentivoglio C,
Stoddart C, Winogron W, Bawden HN. The effects of head
injury in children on neuropsychological and behavioural
functioning. Brain Injury 1991;5:339–351.
22. Leahy LF, Holland AL, Frattalli CM. Persistent deficits
following closed-head injury. Journal of Clinical and
Experimental Neuropsychology 1987;9:34–41.
23. Light R, Asarnow R, Satz P, Zaucha K, McCleary C,
Lewis R. Mild closed-head injury in children and adolescents:
Behavior problems and academic outcomes. Journal of
Consulting and Clinical Psychology 1998;66:1023–1029.
24. Winogron HW, Knights RM, Bawden HN. Neuropsycho-
logical deficits following head injury in children. Journal of
Clinical Neuropsychology 1984;6:269–286.
25. Degeneffe CE. Family caregiving and traumatic brain injury.
Health and Social Work 2001;26:257–269.
26. Teasdale G, Jennett B. Assessment of coma and impaired
consciousness: A practical scale. Lancet 1974;2:81–83.
27. Lieh-Lai MW, Theodorou AA, Sarnaik AP. Limitations of
the Glasgow Coma Scale in predicting outcome in children
with traumatic brain injury. Journal of Pediatrics 1992;120:
195–199.
28. Coppens P. Subpopulations in closed-head injury:
Preliminary results. Brain Injury 1995;9:195–208.
29. Gupta A, Ghai D. Memory and information processing
capacity after closed head injury. Indian Journal of Clinical
Psychology 1996;23:170–177.
30. McMillan TM, Jongen EL, Greenwood RJ, Assessment
of post-traumatic amnesia after severe closed head injury:
Retrospective or prospective? Journal of Neurology,
Neurosurgery & Psychiatry 1996;60:422–427.
31. Nelson LD, Adams KM. Challenges for neuropsychology in
the treatment and rehabilitation of brain-injured patients.
Psychological Assessment 1997;9:368–373.
32. Vilkki J, Ahola K, Holst P, Öhman J. Prediction of
psychosocial recovery after head injury with cognitive tests
and neurobehavioral ratings. Journal of Clinical &
Experimental Neuropsychology 1994;16:325–338.
33. Grewal M, Sutcliffe AJ. Early prediction of outcome
following head injury in children: An assessment of the
value of Glasgow Coma Scale score trend and abnormal
plantar and papillary light reflexes. Journal of Pediatric
Surgery 1991;26:1161–1163.
34. Hackbarth RM, Rzeszutko KM, Sturm G, Donders J,
Kuldanek AS, Sanfilippo DJ. Survival and functional out-
come in pediatric brain injury: A retrospective review and
analysis of predictive factors. Critical Care Medicine
2002;30:1630–1635.
35. Levin HS. Predicting the neurobehavioral sequelae of closed
head injury. In: Wood RL, editor. Neurobehavioural sequelae
of traumatic brain injury. Philadelphia: Taylor & Francis;
1990. pp 89–109.
36. Donders J, Nesbit-Greene K. Predictors of neuropsychologi-
cal test performance after pediatric traumatic brain injury.
Assessment 2004;11:275–284.
37. Hawley CA. Behaviour and school performance after brain
injury. Brain Injury 2004;18:645–659.
38. Kinsella G, Ong B, Murtagh D, Prior M, Sawyer M. The
role of the family for behavioral outcome in children and
adolescents following traumatic brain injury. Journal of
Consulting and Clinical Psychology 1999;67:116–123.
39. American Psychological Association. Evidence-based practice
in psychology. American Psychologist 2006;61:271–285.
40. Bedell GM, Dumas HM. Social participation of children and
youth with acquired brain injuries discharged from inpatient
rehabilitation: A follow-up study. Brain Injury 2004;
18:65–82.
41. Alvarez A, Benn K, Minnes P. Community integration of
children with developmental disabilities and children with
acquired brain injury: Caregivers’ perspectives. Poster pre-
sented at Ontario Association on Developmental Disabilities
Conference 2003; Barrie, ON.
 Predicting social and functional outcomes for individuals sustaining paediatric TBI
42. Minnes PM, Buell KB, Nolte ML, McColl MA, Carlson P,
Johnston J. Defining community integration of persons
with brain injuries as acculturation: A Canadian perspective.
Neurorehabilitation 2001;14:1–8.
43. Solish A, Minnes P, Kupferschmidt A. Community integra-
tion of children with developmental disabilities in social
activities. Journal on Developmental Disabilities 2003;
10:115–122.
44. Vilela T, Phillips M, Minnes P. Is the grass necessarily
greener? What families of children with ABI tell us about
stress and coping with and without access to third party
funding. Paper presented at the Fifth InterUrban Brain Injury
Conference 2005; Peterborough, ON.
45. Sheslow D, Adams W. Wide range assessment of memory
and learning. Wilmington, DE: Jastak Associates; 1990.
46. Cohen MJ. Children’s memory scale (CMS). San Antonio,
TX: Psychological Corporation; 1997.
47. Bruininks RH. Bruininks-Oseretksy test of motor proficiency.
Minnesota: American Guidance Service; 1978.
23
48. Beery KE. The Beery developmental test of visual-motor
integration, 4th ed. New Jersey: Modern Curriculum
Press; 1997.
49. Bedell GM. Developing a follow-up survey focused on
participation of children and youth with acquired brain
injuries after discharge from inpatient rehabilitation.
NeuroRehabilitation 2004;19:191–205.
50. McAllister TW, Sparling MB, Flashman LA, Guerin SJ,
Mamourian AC, Saykin SJ. Differential working memory
load effects after mild traumatic brain injury. NeuroImage
2001;14:1004–1012.
51. Watt S, Shores EA, Kinoshita S. Effects of reducing
attentional resources on implicit and explicit memory after
severe traumatic brain injury. Neuropsychology 1999;13:
338–349.
52. Arroyos-Jurado E, Paulsen JS, Merrell KW, Lindgren SD,
Max JE. Traumatic brain injury in school-age children:
Academic and social outcome. Journal of School Psychology
2000;38:571–587.