Journal Logo

Original Studies

Diffusion Tensor Imaging Correlates of Resilience Following Adolescent Traumatic Brain Injury

Schmidt, Adam T. PhD*; Lindsey, Hannah M. PhD; Dennis, Emily PhD; Wilde, Elisabeth A. PhD; Biekman, Brian D. MA; Chu, Zili D. PhD; Hanten, Gerri R. PhD; Formon, Dana L. PhD§; Spruiell, Matthew S. BA; Hunter, Jill V. MD∥,¶; Levin, Harvey S. PhD

Author Information
Cognitive and Behavioral Neurology: December 2021 - Volume 34 - Issue 4 - p 259-274
doi: 10.1097/WNN.0000000000000283

Abstract

Moderate to severe traumatic brain injury (TBI) is a significant cause of morbidity and mortality within the adolescent population (Kraus, 1995; Langlois et al, 2003; Schmidt et al, 2014). Adolescence (ie, roughly the period of time between ages 11 and 17 for the purposes of the current study) represents a critical period of social development and a time of dynamic brain maturation (Casey et al, 2005, 2008; Steinberg, 2004, 2008, 2010). Many of the areas of the brain that are most vulnerable to TBI underlie complex abilities such as executive functioning and social interaction, resulting in significant problems in these domains after TBI (V Anderson et al, 2013; Hanten et al, 2008; Levin and Hanten, 2005; Ryan et al, 2016a, 2016b; Schmidt et al, 2010a, 2010c, 2012, 2013b, 2014).

Social difficulties manifest as deficits in social problem-solving, theory of mind, or pragmatic communication and can become some of the most challenging, long-term difficulties faced by adolescents and their families after a head injury (Rosema et al, 2012; Schmidt et al, 2010a, 2010c, 2012, 2013b, 2014; Yeates et al, 2007). Long-term difficulties with behavioral regulation, complex decision-making, and emotion recognition, as well as the emergence of novel psychiatric disorders, can also result from TBIs in adolescents (Ganesalingam et al, 2006; Levin et al, 2004; Max et al, 2012; Schmidt et al, 2010a, 2012, 2013b, 2014).

Nonetheless, some research suggests that persistent deficits after pediatric TBI are far from assured, despite significant injury to brain parenchyma (Fay et al, 2009). For example, Fay et al (2009) found that 60% of the children in their study who had sustained a severe TBI were functioning within the average range in at least three of four broad domains (eg, neuropsychological functioning, behavioral functioning, adaptive skills, and academic skills) at a mean of 4 years post injury. The authors speculated that premorbid abilities and environmental factors can have a significant effect on the trajectory of recovery after pediatric TBI. This speculation is supported by research demonstrating that positive parenting practices, family cohesion, and a stable home environment had beneficial effects on behavioral and psychosocial functioning in adolescents after TBI (Holland and Schmidt, 2015; Ryan et al, 2016b; Schmidt et al, 2010c, 2012; Yeates et al, 1997, 2010).

Resilience and Resilience-promoting Factors

The capacity to demonstrate good adaptation in the context of significant hardship is known as resilience (Masten, 2001; Masten et al, 1999). Researchers have conceptualized resilience as a process whereby an individual’s beliefs, cognitive abilities, and personality traits, in combination with family and community support, facilitate positive growth and development under conditions of significant adversity (Armstrong et al, 2005; Avci et al, 2013; Cicchetti and Rogosch, 2009; Masten, 2001; Masten et al, 1999; Schmidt et al, 2013a; Ungar, 2008; Waugh et al, 2008). However, individuals who are resilient in one situation may not be resilient across all contexts or outcomes (Masten, 2001; Masten et al, 1999).

The process of resilience should not be confused with traits such as hardiness or resiliency. Although these personality traits may be protective and may foster positive adaptation under many circumstances, they are not, in themselves, the same as the process of resilience, although they likely contribute to it (Masten, 2001; Masten et al, 1999).

Resilience is best conceptualized as a process and therefore cannot be measured directly in the same way one would measure a trait such as hardiness (Avci et al, 2013; Masten, 2001; Schmidt et al, 2013a; Ungar, 2008). To measure resilience, researchers must assess multiple resilience-promoting factors that span a variety of domains such as community and cultural resources, social and familial supports, and individual traits that are widely associated with positive adaptation across populations and outcomes (Ungar and Lienbenberg, 2011). Intact intellectual functioning, access to community resources, the presence of at least one warm and supportive caregiver, and prosocial peer influences are known resilience-promoting factors that have established relationships with the process of resilience (Armstrong et al, 2005; Masten, 2001; Ungar, 2004).

Differential outcomes following a TBI suggest that examining outcomes after TBI with respect to those variables that are considered important for resilience under other conditions of adversity holds promise for understanding postinjury adaptation and may lead to additional avenues for intervention. Indeed, some of the same factors that are related to increased resilience in high-risk populations such as individuals growing up under conditions of extreme poverty—such as family functioning and parenting behaviors—have been shown to be protective for adolescents who have sustained a TBI (Micklewright et al, 2012; Potter et al, 2011; Raj et al, 2014; Schmidt et al, 2010c; Wade et al, 2011; Yeates et al, 2010). Nonetheless, research examining the influence of known resilience-promoting factors such as community support, close interpersonal relationships, and individual traits on the recovery of individuals who have sustained a TBI is sparse (see Holland and Schmidt, 2015, for a review).

Previous research from our research group and others suggests that resilience-promoting factors that are known to facilitate the process of resilience are associated with fewer postconcussion and depressive/anxiety symptoms and less fatigue up to 1 year post injury in adults who have sustained a mild TBI (Losoi et al, 2015; McCauley et al, 2013). Similarly, Elliott et al (2015) found that military veterans demonstrating a greater number of resilience-related personality factors exhibited increased social functioning and cognitive flexibility and experienced better outcomes after TBI compared with veterans demonstrating fewer resilience-promoting personality factors.

In one of the few studies to examine resilience in adolescents after a TBI, Tonks et al (2011) found that the adolescents with TBI had fewer resilience-promoting factors than the adolescents without TBI, and that fewer resilience-promoting factors were related to increased levels of depression and anxiety. We are not aware of any studies that have integrated an examination of resilience-promoting factors that are known to promote the process of resilience with neurobiological indicators in a TBI population. This lack of information represents a notable shortcoming in our understanding of recovery following pediatric TBI because it limits our knowledge of potential biological mechanisms that may facilitate positive cognitive and behavioral development and may provide novel avenues to evaluate this population’s progress in treatment and rehabilitation.

Examining the Relationship Between White Matter Structure and Resilience

Over the past several years, there has been increasing interest in the role of neurobiological and cognitive factors contributing to the process of resilience in general (Burt et al, 2016; Galinowski et al, 2015; Kalisch et al, 2015; van der Werff et al, 2013). Several large-scale studies of high-risk adolescent populations have indicated that white matter (WM) tracts and cortical volume, especially in the right frontal regions of the brain, impact resilience (Burt et al, 2016; Galinowski et al, 2015). Advances in neuroimaging procedures have made it possible to investigate in detail the impact of subtle WM anomalies and focal lesions that may be present in individuals who have sustained a TBI.

One imaging technique that has proven to be highly sensitive to the impact of TBI and other acquired brain injuries on an individual’s brain is diffusion tensor imaging (DTI) (Hunter et al, 2012; Schmidt et al, 2010b, 2013b). This imaging technique examines WM structure by detecting the diffusion of water molecules along tracts in the CNS and making an inference about the integrity of the WM tissue making up the tracts (Hunter et al, 2012; Schmidt et al, 2010b, 2013b). Generally, tracts that are intact, homogenous, myelinated, or coherently aligned exhibit a pattern of molecular diffusion that is direction specific (ie, generally parallel to the orientation of fibers in a specific WM tract), whereas tracts that are less intact or that have more complex configurations (eg, crossing, bending fibers) may exhibit a pattern of molecular diffusion that occurs more equally in all directions.

There are several metrics that can be used to measure this diffusion, including fractional anisotropy (FA) and mean diffusivity (MD). FA and MD allow inferences about different aspects of WM integrity to be made by assessing the direction and rate of the diffusion of water molecules along the tract. FA is a measure of anisotropic diffusion; it provides information regarding diffusion direction but does not provide information related to orientation. MD is a measure of the isotropy in DTI; it provides information regarding the average weighted diffusion that is acquired from each voxel, or the average of the total diffusion from all directions. FA and MD reflect salient characteristics of WM microstructure throughout the CNS, including its relative density, diameter, and myelination (Hunter et al, 2012; Schmidt et al, 2010b, 2013b). Given its sensitivity to WM abnormalities, DTI is a powerful investigating technique for evaluating the relationship of WM structure to resilient functioning after TBI (Dennis et al, 2018; Hunter et al, 2012; Levin et al, 2008; Schmidt et al 2010b, 2013b).

Current Study

We used DTI to examine the relationship between FA and MD and resilience-promoting factors in a sample of adolescents after TBI or orthopedic injury (OI) at 1 year post injury. Given the vulnerability to brain regions that are involved in social cognition and executive functioning, we hypothesized that resilience-promoting factors would be related to FA and MD of WM tracts in, or projecting to, the frontotemporal areas of the brain that are implicated in social cognition and executive functioning (ie, cognitive abilities disrupted by TBI) as well as tracts that are implicated as being important for resilience after adversity (Burt et al, 2016; Levin and Hanten, 2005; Ryan et al, 2018). Specifically, we defined the genu of the corpus callosum and the right and left uncinate fasciculi, inferior fronto-occipital fasciculi, sagittal striatum, anterior limb of the internal capsule, and anterior corona radiata as our a priori regions of interest based on literature implicating these regions in social cognition and executive functioning (Grieve et al, 2007; Levin et al, 2008; Martino et al, 2010; Wilde et al, 2006).

We hypothesized that the TBI group would reveal lower FA and higher MD values compared with the OI group for all of the aforementioned WM tracts. Additionally, we predicted that FA and MD would be correlated with variables that are related to resilience across domains for these WM tracts. Because we were primarily interested in how variables related to resilience would function across the range of TBI, we initially conducted the analyses by combining individuals with a TBI into one group in order to make a direct comparison with individuals with an OI. However, we recognize that the severity of an injury plays a major role in predicting long-term outcomes for an adolescent and significantly influences his or her psychosocial functioning as well as the family system. Thus, we decided to examine if there were any differences in the pattern of relations based on TBI severity (ie, complicated mild [cmTBI] vs moderate or severe [msTBI]). We conducted a series of supplemental subgroup analyses to explore this question.

METHOD

Participants

We invited adolescents (11 through 17 years) with TBI and adolescents with an OI to participate in our study. Recruitment involved approaching each adolescent and his or her caregiver after the adolescent was medically stable. Researchers recruited participants at level 1 trauma centers in two large metropolitan areas in a large southern state in the United States. The OI group was intended to control for factors that predispose adolescents to significant injuries and for the process/experience of hospitalization (ie, stress, anxiety). The two groups were evaluated as part of a larger study on the long-term social–cognitive effects of TBI on maturation. Initial behavioral data (including resilience measurement and intellectual functioning) were collected shortly after injury once the adolescents were medically stable and were able to participate in postinjury baseline assessments. Imaging data and follow-up resilience measurements were collected between 10.6 and 19.4 months post injury (M  =  14.98, SD  =  0.5).

Inclusion criteria for the TBI group was a postresuscitation Glasgow Coma Scale score falling between the range of 3 and 15, with evidence of intracranial anomalies on the day-of-injury neuroimaging that had been performed as part of the clinical care (Teasdale and Jennett, 1974). Inclusion criteria for the OI group was having experienced mild to moderate physical injuries, overnight hospitalization (at a minimum), no evidence of injuries involving the head, and no report of other indicators of injury (eg, loss or alteration of consciousness, posttraumatic amnesia).

Exclusion criteria for both groups included history of a head injury, lack of fluency in English, and the presence of any significant neurodevelopmental or psychiatric diagnoses (eg, autism, intellectual disabilities, schizophrenia). All of the individuals also needed to be able to complete all of the measures immediately post injury (ie, baseline) and at the 1-year follow-up independently in accordance with standardized procedures and without significant modification or assistance.

We further divided the TBI group according to the individuals’ lowest postresuscitation Glasgow Coma Scale score. Individuals with a score between 13 and 15 were put in the cmTBI group, and individuals with a score between 9 and 12 (moderate) and 3 and 8 (severe) were put in the msTBI group.

The individuals’ socioeconomic status (SES) was measured using the Socioeconomic Composite Index (SCI; Yeates et al, 1997, 2004, 2010). The SCI is a composite measure that estimates SES by combining information regarding one’s occupational status, family income, and maternal education into a single variable by transforming each component score into a z score and then taking the average of the three values (Yeates et al, 1997, 2004, 2010).

The study protocol was approved by the institutional review board of Baylor College of Medicine, The University of Texas at Dallas, and all participating hospitals and was performed according to the ethical guidelines of the Declaration of Helsinki and its later amendments. All individuals provided informed written consent before enrolling in the study.

Measures

Baseline Intellectual Functioning

In order to control for premorbid differences in intellectual abilities that might influence the findings, we had all of the individuals complete the Wechsler Abbreviated Scale of Intelligence, Second Edition (Wechsler, 2011). This scale is a brief measure of intellectual abilities for individuals ranging in age from 6 to 90 years. The assessment takes ~30 minutes and provides standardized scores (M  =  100, SD  =  15) of verbal comprehension, perceptual reasoning, and general cognitive ability based on four subtests (ie, Vocabulary, Similarities, Block Design, and Matrix Reasoning). The Wechsler Abbreviated Scale of Intelligence, Second Edition is considered reliable and valid and has been shown to have convergent validity with other established measures of intelligence. Because we were only interested in overall intellectual functioning, we had our participants take just the Vocabulary and Matrix Reasoning subtests.

Resilience-promoting Factors

We measured resilience-promoting factors spanning caregiver, community, and individual domains at baseline and the 1-year follow-up using the Child and Youth Resilience Measure (CYRM–28; Unger, 2008; Ungar and Lienbenberg, 2011), which is a culturally sensitive, validated, and standardized measure of factors that are related to resilient outcomes (Ungar, 2004). The CYRM–28 was developed initially through a process of face-to-face interviews with youth and their caregivers across various cultural contexts regarding the types of obstacles youth face, how youth navigate these obstacles, and individual and community characteristics that help youth to thrive during adversity.

The CYRM–28 was developed through a norming process that involved 1451 youth across 14 communities and 11 countries (Ungar, 2008; Ungar and Lienbenberg, 2011). The CYRM–28 is a 28-item self-report questionnaire that takes ~7–10 minutes for an adolescent to complete. Responses are rated on a 5-point scale ranging from (1) almost never to (5) almost always, with higher scores indicating a greater number of resilience-promoting factors; thus, more resilient functioning.

The CYRM–28 assesses a variety of variables that have been linked to positive outcomes across cultures and contexts (Ungar, 2008; Ungar and Lienbenberg, 2011). The CYRM–28 yields a Total score (sum of responses on items 1–28) as well as three subscale scores (sum of items on each subscale). Subscales include Caregiver Factors (eg, “I feel that my parent(s) watch me closely” and “I feel that my parent(s) know a lot about me”), Community Factors (eg, “I feel supported by my friends,” “I know where to go in my community to get help”), and Individual Factors (eg, “Spiritual beliefs are a source of strength for me,” “I am aware of my own strengths”).

The CYRM–28 has been proven to be effective at identifying factors that contribute to the process of resilience and has been used by us to examine the cognitive sequelae of resilient functioning in adolescent samples (Avci et al, 2013; Schmidt et al, 2013a). It has cross-cultural validity and acceptable reliability, meaning that it has reasonable internal consistency and can be used to reliably measure resilience-promoting factors (α  =  0.754; Ungar, 2008; Ungar and Lienbenberg, 2011). Due to differences in the timing of when the CYRM–28 was included in the data collection, all of the participants did not complete the CYRM–28 at baseline, and individuals from one of the data collection sites were more likely to have completed the CYRM–28 at baseline. As such, we controlled for site differences in our analyses.

Procedure

MRI Acquisition and DTI Analysis

We used comparable Phillips 3T Achieva scanners with similar software releases to scan all of the participants. All scanners had been subjected to regular and rigorous quality assurance testing using standard phantoms and other procedures, and no issues were identified as part of this testing. We used transverse multislice spin echo, single-shot, echo-planar imaging sequences (10702.6 ms repetition time, 51 ms echo time, 2-mm-thick slices with 0-mm gap) over ~4½ minutes to acquire 70 slices. We employed a 224-mm field of view with a measured voxel size of 2.00 × 2.03 × 2.00 mm.

We collected 33 images per individual: 1 with no diffusion weighting and 32 diffusion weighted (b  =  1000 s/mm2). Additionally, we used T1-weighted imaging over 5.94 minutes to capture 170 slices (8.06 ms repetition time, 3.68 ms echo time, 1-mm thick slices with 0-mm gap). We visually inspected all of the data for the presence of significant distortion and other artifacts both before the postprocessing steps outlined in the next paragraph and after. All of the data contained in this report were determined to be of sufficient quality to use in the analysis following postprocessing.

T1-weighted images were automatically masked and were corrected for inhomogeneities using the BrainExtraction advanced normalization tool (ANT) with the Nathan Kline Institute Under 10 template and N4BiasFieldCorrection (Tustison et al, 2014). We visually checked the masks, and one of the authors with training in imaging analysis (E.L.D.) manually corrected them when necessary. Each individual’s masked, N4-corrected, T1-weighted image was registered to the ENIGMA DTI template using the ANT for both linear and nonlinear registration (Avants et al, 2008, 2009; Jahanshad et al, 2013; Pineda et al, 2014).

Six-parameter rigid-body transformation, affine registration, and SyN registration used a multilevel approach (ie, the moving and fixed images were successively less smoothed at each level), with a full-resolution registration occurring at the final level. We chose to use 1000, 500, 250, and 100 iterations at each level, with a Gaussian kernel smoothing sigma set to 3, 2, 1, and 0, respectively. We measured image similarity using the ANT implementation of mutual information for the rigid and affine registrations, and cross-correlation for SyN (Avants et al, 2011). We visually checked all of the registrations for quality. We corrected DTI volumes for eddy current distortions using the FMRIB Software Library (FSL) eddy correct tool (https://fsl.fmrib.ox.ac.uk/), and bvec files were rotated accordingly (JLR Anderson and Sotiropoulos 2016). We generated tensor maps using the FSL dtifit tool.

For each individual, we registered the resulting b0 files to their registered T1-weighted image, again using the ANT for linear and nonlinear registration. Researchers visually checked these files for quality. The resulting transformation matrix was then applied to each individual’s tensor images (FA and MD). Registered tensor maps were skeletonized using the FSL tbss skeleton tool using the ENIGMA DTI skeleton (Smith et al, 2006). FA and MD were then averaged across the whole skeleton and in each of five midline and 19 bilateral WM regions of interest from the Johns Hopkins University atlas. The end result was FA and MD averaged in 63 regions of interest (whole skeleton, five midline structures, 19 structures left/right/bilaterally averaged).

Statistical Analysis

We screened all of the outcome measures for outliers before analysis, and we fenced any variables with values exceeding ±1.5 IQR from the median (Tukey, 1977). Before all between-group comparisons, we evaluated the assumptions of homoscedasticity and normality using the Levene and Shapiro-Wilk tests, respectively. We compared demographic and injury characteristics between groups using the Pearson χ2 or Fisher exact test for discrete variables and independent t tests for continuous variables. We used the Wilcoxon Mann-Whitney U ranked sums tests to compare non-normal sample characteristics.

We evaluated group differences in the FA and MD of the WM tracts of interest and in the CYRM–28 subscale and Total scores using one-way analyses of covariance (ANCOVAs) with age at injury, time since injury, and study site included as covariates. The results of all of the ANCOVAs are accompanied by Cohen’s f estimates of effect size, where f  =  0.10, 0.25, and 0.40 are interpreted as small, medium, and large effects, respectively (Cohen, 1988).

We investigated the relationship between the FA/MD of each tract and the CYRM–28 subscale and Total scores in each group using partial correlation coefficients, with age at injury, time since injury, and study site included as covariates. To demonstrate the proportion of variance in the CYRM–28 scores that were uniquely associated with the FA/MD of a given tract, all partial correlations are accompanied by squared semipartial correlations, where r2Y(1.2)  =  0.02, 0.13, and 0.26 are interpreted as small, medium, and large effects, respectively.

To examine if there were any differences based on TBI severity, we performed the same analyses on the TBI subgroups. Then, we compared the FA, MD, and CYRM–28 scores between the cmTBI, msTBI, and OI groups. After removing the effects of age at injury, time since injury, and study site, we also investigated the relationship between the FA/MD and the CYRM–28 scores in the cmTBI and msTBI groups. We conducted pairwise comparisons using the Scheffe test following all significant overall ANCOVA models. Additionally, supplemental comparisons of baseline CYRM–28 scores, controlling for age at injury, time since injury (to baseline testing), and study site, were performed for a subset of the present sample (25 TBI, 10 OI).

For all of the analyses, we used the the Benjamini-Hochberg false discovery rate procedure (Benjamini and Hockberg, 1995) to control for multiple comparisons, and we set statistical significance at α  =  0.05. All statistical analyses were completed using STATA 16.1 (StataCorp LP).

RESULTS

Participants

We recruited 38 adolescents (26 males, 12 females) with cmTBI or msTBI between the ages of 9.6 and 18.0 years at the time of injury (M  =  14.31, SD  =  2.36) and 23 adolescents (14 males, 9 females) with an OI between the ages of 9.5 and 18.1 years at the time of injury (M  =  13.02, SD  =  2.48) for our study.

Between-group comparisons across demographic characteristics indicated that the two groups (ie, TBI vs OI) did not differ significantly in terms of study site, sex, ethnicity, age at evaluation, SES, or estimated IQ (Table 1). On average, the adolescents in the TBI group were injured at an older age than those in the OI group (MD  =  1.29, SED  =  0.64), and fewer months had elapsed since the injury for those with TBI compared with those with OI (MD  =  –0.82, SED  =  0.49). As expected, the adolescents in the TBI group sustained a greater number of high-velocity injuries (53%) than those in the OI group (13%). The TBIs had largely been sustained from falls (29%), motor vehicle accidents (39%), and recreational vehicle accidents (13%), and the OIs had largely been sustained during sports and play activities (52%) and falls (30%).

TABLE 1 - Study Participants’ Demographic and Injury Characteristics
Statistic
TBI (n = 38) n (%) OI (n = 23) n (%) χ2 df P
Study site 4.33 1 0.063
 Houston 16 (42.11) 16 (69.57)
 Dallas 22 (57.89) 7 (30.43)
Sex 0.36 1 0.587
 Male 26 (68.42) 14 (60.87)
 Female 12 (31.58) 9 (39.13)
Ethnicity 7.64 4 0.091
 Caucasian 18 (47.37) 7 (30.43)
 African American 2 (5.26) 4 (17.39)
 Hispanic 14 (36.84) 11 (47.83)
 Asian American 4 (10.53) 0 (0.00)
 Native American 0 (0.00) 1 (4.35)
Mechanism 13.01 5 0.015*
 Fall 11 (28.95) 7 (30.43)
 MVA–Passenger 13 (34.21) 2 (8.70)
 MVA–Pedestrian 2 (5.26) 0 (0.00)
 RVA 5 (13.16) 1 (4.35)
 Sports/Play 7 (18.42) 12 (52.17)
 Unknown 11 (28.95) 1 (4.35)
Velocity 9.56 1 0.003**
 High 20 (52.63) 3 (13.04)
 Low 18 (47.37) 20 (86.96)
Duration of PTA 49.29 4 0.000***
 None 2 (5.26) 22 (95.65)
 <1 hour 8 (21.05) 0 (0.00)
 1–24 hours 4 (10.53) 0 (0.00)
 >24 hours 13 (34.21) 0 (0.00)
 Unknown 11 (28.95) 1 (4.35)
M SD Range M SD Range z P
Age at evaluation 15.45 2.38 10.61–18.95 14.23 2.51 10.62–19.39 −1.92 0.554
Age at injury 14.31 2.36 9.61–17.97 13.02 2.48 9.52–18.14 −2.04 0.042*
TSI (months) 13.62 1.75 11.83–20.53 14.44 1.99 12.09–21.13 2.10 0.036*
t df P
SCI 0.17 0.95 –1.81, 1.74 0.01 1.04 –1.97, 2.16 −0.60 58 0.549
FSIQ 105.11 8.69 88–121 103.96 8.75 88–119 −0.50 59 0.620
GCS Low 8.00 5.43 3–15
One of the individuals with an OI did not complete all of the measures needed to compute this statistic.
*Significant at P < 0.05.
**Significant at P < 0.01.
***Significant at P < 0.001.
FSIQ = Full-scale IQ. GCS Low = lowest 24-hour Glasgow Coma Scale score. MVA = motor vehicle accident. OI = orthopedic injury. PTA = posttraumatic amnesia. RVA = recreational vehicle accident. SCI = Socioeconomic Composite Index. TBI = traumatic brain injury. TSI = time since injury.

TBI Group Analyses

Baseline CYRM–28 Analysis

Demographic and injury characteristics for the subset of TBI and OI groups with baseline CYRM–28 data (n  =  35) are presented in the supplemental digital content (SDC) as Table 1 (https://links.lww.com/CBN/A97). No differences were found between any demographic or injury characteristic of the two groups. After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, the TBI and OI groups did not differ significantly in scores for any CYRM–28 subscale or on the Total CYRM–28 at baseline (Table 2).

TABLE 2 - CYRM–28 Scores at Baseline and the 1-Year Follow-up for the OI Group, TBI Group, and TBI Subgroups
Group and Injury M SD M SD F P FDR0.05 f
Group baseline OI (n = 10) TBI (n = 25) (1, 30)
 Individual 35.20 5.29 34.28 6.90 0.05 0.825 0.050 0.04
 Caregiver 22.00 3.74 20.08 4.07 2.00 0.168 0.013 0.26
 Context 29.70 4.99 27.44 7.03 0.42 0.520 0.038 0.12
Total 86.90 13.08 81.80 15.48 0.59 0.448 0.025 0.14
Group follow-up OI (n = 23) TBI (n = 38) (1, 56)
 Individual 36.61 4.40 32.87 6.96 5.44 0.023* 0.025 0.31
 Caregiver 22.39 3.77 19.26 4.53 4.62 0.036* 0.038 0.29
 Context 30.13 6.15 25.63 7.79 4.41 0.040* 0.050 0.28
  Total 89.13 12.88 77.76 16.85 6.21 0.016 0.013 0.33
M SD M SD M SD F P FDR0.05 f
Subgroup baseline OI (n = 10) cmTBI (n = 12) msTBI (n = 13) (2, 28)
 Individual 35.20 5.29 34.08 8.23 34.46 5.74 0.03 0.966 0.050 0.05
 Caregiver 22.00 3.74 19.75 4.63 20.38 3.64 1.05 0.364 0.013 0.27
 Context 29.70 4.99 25.67 7.88 29.08 5.99 1.03 0.369 0.025 0.27
Total 86.90 13.08 79.50 17.59 83.92 13.62 0.55 0.582 0.038 0.20
Subgroup follow-up OI (n = 23) cmTBI (n = 15) msTBI (n = 23) (2, 54)
 Individual 36.61 4. 40 36.73 5.19 30.35 6.89 10.14 0.000*** 0.013 0.61
 Caregiver 22.39 3.77 20.93 4.93 18.17 3.98 5.13 0.009** 0.038 0.43
 Context 30.13 6.15 27.47 7.14 24.43 8.11 3.40 0.040* 0.050 0.35
 Total 89.13 12.88 85.13 14.43 72.96 16.85 7.34 0.002** 0.025 0.52
Significant differences that survived the multiple comparison correction (P < FDR < 0.05) are indicated. Small, medium, and large effect sizes are indicated by Cohen’s f ≥ 0.10, 0.25, and 0.40, respectively.
Some of the individuals were unavailable to take the CYRM–28 at baseline, resulting in a smaller n at baseline compared with the 1-year follow-up.
*Significant at P < 0.05.
**Significant at P < 0.01.
***Significant at P < 0.001.
cm = complicated mild. CYRM–28 = Child and Youth Resilience Measure. FDR = false discovery rate. ms = moderate to severe. OI = orthopedic injury. TBI = traumatic brain injury.

Group Differences in FA and MD

Two individuals had missing imaging data as a result of a movement artifact. The other 59 individuals were included in the DTI analysis. After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, one-way ANCOVAs (Table 3) demonstrated significantly decreased FA with medium to large effect sizes in the genu of the corpus callosum (f  =  0.37), the right sagittal striatum (f  =  0.31), and the right and left anterior corona radiata (fR  =  0.47; fL  =  0.52) of the adolescents with TBI compared with those with OI. Additionally, the ANCOVAs demonstrated decreased FA with medium effect sizes in the right uncinate fasciculus (f  =  0.26) and the left sagittal striatum (f  =  0.31) of the TBI group. Although no significant differences were seen in the MD of any tract between the two groups, medium effect sizes were seen for increased MD in the genu of the corpus callosum (f  =  0.34), the right uncinate fasciculus (f  =  0.32), the right and left sagittal striatum (fR  =  0.28; fL  =  0.28), and the right and left anterior corona radiata (fR  =  0.31; fL  =  0.29) of the TBI group.

TABLE 3 - Group Differences in FA and MD, Controlling for Age at Injury, Time Since Injury, and Site
OI (n = 22) TBI (n = 37)
M SD M SD F 1,54 P FDR.05 f
FA
 Genu of CC 0.630 0.032 0.580 0.061 7.32 0.009** 0.017 0.37
 Right UF 0.445 0.055 0.407 0.055 3.71 0.059 0.033 0.26
 Left UF 0.475 0.043 0.441 0.064 1.60 0.211 0.039 0.17
 Right SS 0.496 0.031 0.476 0.043 5.20 0.027* 0.028 0.31
 Left SS 0.497 0.032 0.475 0.035 5.23 0.026 0.022 0.31
 Right ACR 0.414 0.015 0.376 0.040 11.85 0.001** 0.011 0.47
 Left ACR 0.419 0.016 0.378 0.039 14.79 0.000*** 0.006 0.52
 Right ALIC 0.479 0.025 0.473 0.030 0.03 0.862 0.050 0.02
 Left ALIC 0.487 0.023 0.478 0.033 1.40 0.242 0.044 0.16
MD
 Genu of CC 0.764 0.030 0.805 0.048 6.25 0.016 0.006 0.34
 Right UF 0.802 0.029 0.829 0.037 5.46 0.023 0.011 0.32
 Left UF 0.786 0.019 0.806 0.046 1.07 0.306 0.050 0.14
 Right SS 0.754 0.031 0.780 0.053 4.17 0.046 0.033 0.28
 Left SS 0.813 0.025 0.834 0.039 4.18 0.046 0.028 0.28
 Right ACR 0.753 0.026 0.789 0.050 5.02 0.029 0.017 0.31
 Left ACR 0.755 0.025 0.786 0.045 4.38 0.041 0.022 0.29
 Right ALIC 0.709 0.024 0.726 0.038 2.01 0.162 0.039 0.19
 Left ALIC 0.706 0.018 0.721 0.024 1.66 0.203 0.044 0.18
FDR corrections were applied separately for FA and MD (9 comparisons each). Significant differences that survived the multiple comparison correction (P < FDR < 0.05) are indicated. Small, medium, and large effect sizes are indicated by Cohen’s f ≥ 0.10, 0.25, and 0.40, respectively.
One individual had missing imaging data as a result of a movement artifact.
*Significant at P < 0.05.
**Significant at P < 0.01.
***Significant at P < 0.001.
ACR = anterior corona radiata. ALIC = anterior limb of the internal capsule. CC = corpus callosum. FA = fractional anisotropy. FDR = false discovery rate. MD = mean diffusivity. OI = orthopedic injury. SS = sagittal stratum. TBI = traumatic brain injury. UF = uncinate fasciculus.

Group Differences in CYRM–28 Scores at Follow-up

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, significant group differences with medium effect sizes were seen in the 1 year post injury Individual (f  = 0.31), Caregiver (f  =  0.29), and Context (f  =  0.28) subscales of the CYRM–28, where adolescents within the TBI group scored significantly lower than adolescents within the OI group (Table 2). Although the difference in CYRM–28 Total scores did not survive the multiple comparison correction, a medium effect size was present for lower total 1 year post injury CYRM–28 scores (f =0.33) in the TBI group.

Relationship Between FA/MD and CYRM–28 Scores

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, partial correlations demonstrated no relationship between the FA or MD of any tract with the CYRM–28 subscale or Total scores in the OI group (SDC Table 2, https://links.lww.com/CBN/A98).

In the TBI group, however, several significant relationships with medium effect sizes were found between the CYRM–28 scores and the MD of various tracts (Table 4). Specifically, increased scores on the CYRM–28 Caregiver subscale were associated with decreased MD in the right anterior corona radiata (r2Y(1.2)  =  0.17) and the right (r2Y(1.2)  =  0.23) and left (r2Y(1.2)  =  0.21) anterior limb of the internal capsule. Moreover, increased scores on the CYRM–28 Context subscale were associated with decreased MD in the genu of the corpus callosum (r2Y(1.2)  =  0.20) and the right (r2Y(1.2)  =  0.20) and left (r2Y(1.2)  =  0.21) anterior corona radiata. Increased scores on the Total CYRM–28 were associated with decreased MD in the genu of the corpus callosum (r2Y(1.2)  =  0.19), the right (r2Y(1.2)  =  0.20) and left (r2Y(1.2)  =  0.23) anterior corona radiata, and the right (r2Y(1.2)  =  0.14) and left (r2Y(1.2)  =  0.17) anterior limb of the internal capsule. Additionally, we demonstrated medium effect sizes despite nonsignificant associations between increased scores on the CYRM–28 Individual subscale and increased FA of the left sagittal striatum (r2Y(1.2)  =  0.14) and the left anterior limb of the internal capsule (r2Y(1.2)  =  0.15). We also observed decreased MD of the left anterior corona radiata (r2Y(1.2)  =  0.18) and increased scores on the CYRM–28 Context subscale and decreased MD in the left anterior limb of the internal capsule (r2Y(1.2)  =  0.14).

TABLE 4 - Partial Correlations Between FA/MD and CYRM–28 Scores in the TBI Group, Controlling for Age at Injury, Time Since Injury, and Study Site
Individual Caregiver Context Total
TBI (n = 37) r Y1.2 r 2 Y(1.2) P FDR.05 r Y1.2 r 2 Y(1.2) P FDR.05 r Y1.2 r 2 Y(1.2) P FDR.05 r Y1.2 r 2 Y(1.2) P FDR.05
FA
 Genu of CC 0.29 0.08 0.100 0.028 0.21 0.04 0.233 0.017 0.34 0.11 0.049 0.006 0.33 0.11 0.058 0.011
 Right UF 0.14 0.02 0.434 0.033 −0.09 0.01 0.627 0.044 0.04 0.00 0.812 0.050 0.05 0.00 0.772 0.050
 Left UF 0.32 0.09 0.070 0.022 0.18 0.03 0.319 0.033 0.32 0.10 0.069 0.011 0.32 0.10 0.066 0.017
 Right SS 0.32 0.09 0.067 0.017 0.24 0.06 0.166 0.011 0.27 0.07 0.128 0.028 0.32 0.10 0.069 0.022
 Left SS 0.39 0.14 0.022 0.011 0.19 0.04 0.279 0.028 0.29 0.08 0.097 0.017 0.34 0.11 0.048 0.006
 Right ACR 0.12 0.01 0.497 0.044 0.25 0.06 0.156 0.006 0.28 0.08 0.111 0.022 0.24 0.06 0.166 0.033
 Left ACR 0.07 0.00 0.700 0.050 0.20 0.04 0.261 0.022 0.21 0.04 0.232 0.039 0.18 0.03 0.314 0.039
 Right ALIC 0.13 0.02 0.450 0.039 0.12 0.01 0.505 0.039 0.11 0.01 0.528 0.044 0.14 0.02 0.439 0.044
 Left ALIC 0.40 0.15 0.020 0.006 0.07 0.01 0.689 0.050 0.24 0.06 0.165 0.033 0.29 0.08 0.095 0.028
MD
 Genu of CC −0.35 0.11 0.046 0.011 −0.36 0.13 0.035 0.022 −0.45 0.20 0.008** 0.017 −0.44 0.19 0.009** 0.017
 Right UF −0.14 0.02 0.428 0.044 −0.22 0.05 0.217 0.039 −0.18 0.03 0.305 0.044 −0.20 0.04 0.261 0.044
 Left UF −0.05 0.00 0.784 0.050 0.06 0.00 0.724 0.050 −0.09 0.01 0.611 0.050 −0.04 0.00 0.804 0.050
 Right SS −0.30 0.08 0.088 0.028 −0.27 0.07 0.118 0.033 −0.34 0.11 0.049 0.033 −0.35 0.12 0.043 0.033
 Left SS −0.32 0.10 0.065 0.022 −0.16 0.02 0.377 0.044 −0.35 0.12 0.040 0.028 −0.33 0.11 0.054 0.039
 Right ACR −0.34 0.10 0.052 0.017 −0.41 0.17 0.015* 0.017 −0.46 0.20 0.006** 0.011 −0.46 0.20 0.007** 0.011
 Left ACR −0.44 0.18 0.009 0.006 −0.35 0.12 0.041 0.028 −0.47 0.21 0.005** 0.006 −0.49 0.23 0.003** 0.006
 Right ALIC −0.26 0.06 0.140 0.039 −0.48 0.23 0.004** 0.006 −0.32 0.10 0.067 0.039 −0.38 0.14 0.027* 0.028
 Left ALIC −0.29 0.08 0.101 0.033 −0.46 0.21 0.006** 0.011 −0.38 0.14 0.026 0.022 −0.42 0.17 0.015* 0.022
Significant correlations that survived the multiple comparison correction (P < FDR < 0.05) are indicated. FDR corrections were applied separately for FA and MD across each of the CYRM–28 scales (9 comparisons each). Squared semipartial correlations are interpreted as small, medium, and large effect sizes when r2Y(1.2) = 0.02, 0.13, and 0.26, respectively.
One individual had missing imaging data as a result of a movement artifact.
*Significant at P < 0.05.
**Significant at P < 0.01.
ACR = anterior corona radiata. ALIC = anterior limb of the internal capsule. CC = corpus callosum. CYRM28 = Child and Youth Resilience Measure. FA = fractional anisotropy. FDR = false discovery rate. MD = mean diffusivity. SS = sagittal striatum. TBI = traumatic brain injury. UF = uncinate fasciculus.

TBI Subgroup Analyses

Demographic and injury characteristics for the cmTBI, msTBI, and OI groups are presented in SDC Table 3 (https://links.lww.com/CBN/A99). No differences were found in terms of sex, ethnicity, age, age at injury, time since injury, or SES between the three groups; however, differences were present in study site and IQ between the three groups, where the mean IQ of the cmTBI subgroup was higher than that of both the msTBI subgroup (MD  =  9.41) and the OI group (MD  =  6.84).

Subgroup Differences in Baseline CYRM–28

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, one-way ANCOVAs were performed to assess differences in CYRM–28 scores at baseline between the cmTBI, msTBI, and OI groups (Table 2). The results demonstrated no significant differences between the CYRM–28 subscale or Total scores between the three groups, although medium effect sizes were observed for the Caregiver (f  =  0.27) and Context (f  =  0.27) subscales, where the cmTBI subgroup scored 3 and 4 points lower, respectively, than the OI group.

Subgroup Differences in FA and MD

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, one-way ANCOVAs (Table 5) demonstrated significant differences, with large effect sizes, in the FA in the genu of the corpus callosum (f  =  0.51), the right and left sagittal striatum (fR  =  0.46; fL  =  0.64), and the right and left anterior corona radiata (fR  =  0.51; fL  =  0.57). Medium effect sizes were also present for differences in the FA in the right and left uncinate fasciculus (fR  =  0.26; fL  =  0.22) and the right and left anterior limb of the internal capsule (fR  =  0.20; fL  =  0.32).

TABLE 5 - Group Differences in FA and MD, Controlling for Age at Injury, Time Since Injury, and Study Site
OI (n = 22) cmTBI (n = 14) msTBI (n = 23)
M SD M SD M SD F (2, 53) P FDR0.05 f
FA
 Genu of CC 0.630 0.032 0.605 0.041 0.564 0.067 6.77 0.002** 0.022 0.51
 Right UF 0.445 0.055 0.409 0.056 0.406 0.055 1.82 0.172 0.039 0.26
 Left UF 0.475 0.043 0.454 0.060 0.432 0.066 1.30 0.280 0.044 0.22
 Right SS 0.496 0.031 0.496 0.036 0.464 0.043 5.49 0.007** 0.028 0.46
 Left SS 0.497 0.032 0.501 0.023 0.460 0.031 10.79 0.000*** 0.006 0.64
 Right ACR 0.414 0.015 0.386 0.038 0.370 0.041 6.97 0.002** 0.017 0.51
 Left ACR 0.419 0.016 0.389 0.033 0.371 0.041 8.71 0.001** 0.011 0.57
 Right ALIC 0.479 0.025 0.481 0.025 0.468 0.033 1.00 0.373 0.050 0.20
 Left ALIC 0.487 0.023 0.492 0.027 0.469 0.035 2.76 0.073 0.033 0.32
MD
 Genu of CC 0.764 0.030 0.789 0.039 0.815 0.051 6.57 0.003** 0.022 0.50
 Right UF 0.802 0.029 0.820 0.038 0.834 0.037 3.97 0.025* 0.044 0.39
 Left UF 0.786 0.019 0.802 0.039 0.809 0.050 0.61 0.548 0.050 0.15
 Right SS 0.754 0.031 0.752 0.042 0.798 0.052 8.29 0.001** 0.011 0.56
 Left SS 0.813 0.025 0.807 0.032 0.851 0.035 9.98 0.000*** 0.006 0.61
 Right ACR 0.753 0.026 0.773 0.041 0.798 0.053 5.34 0.008** 0.028 0.45
 Left ACR 0.755 0.025 0.767 0.031 0.798 0.049 6.65 0.003** 0.017 0.50
 Right ALIC 0.709 0.024 0.719 0.039 0.731 0.037 4.23 0.020* 0.033 0.40
 Left ALIC 0.706 0.018 0.716 0.022 0.724 0.026 4.14 0.021* 0.039 0.40
FDR corrections were applied separately for FA and MD (9 comparisons each). Only significant differences that survived the multiple comparison correction (P<FDR<0.05) are indicated. Small, medium, and large effect sizes are indicated by Cohen’s f≥0.10, 0.25, and 0.40, respectively.
One individual had missing imaging data as a result of a movement artifact.
*Significant at P < 0.05.
**Significant at P < 0.01.
***Significant at P < 0.001.
ACR = anterior corona radiata. ALIC = anterior limb of the internal capsule. CC = corpus callosum. cmTBI = complicated mild traumatic brain injury. FA = fractional anisotropy. MD = mean diffusivity. msTBI = moderate or severe traumatic brain injury. OI = orthopedic injury. SS = sagittal striatum. UF = uncinate fasciculus.

Follow-up pairwise comparisons (SDC Table 4, https://links.lww.com/CBN/A100) revealed significantly decreased FA in the genu of the corpus callosum and the left sagittal striatum of the msTBI subgroup compared with both the cmTBI subgroup and the OI group, as well as decreased FA in the right sagittal striatum and the right and left anterior corona radiata of the msTBI subgroup relative only to the OI group. The cmTBI subgroup did not differ from the OI group in the FA of any tract.

Findings demonstrated significant differences, with medium to large effect sizes, in the MD of the genu of the corpus callosum (f  =  0.50), the right uncinate fasciculus (f  =  0.39), the right and left sagittal striatum (fR  =  0.56; fL  =  0.61), the right and left anterior corona radiata (fR  =  0.45; fL  =  0.50), and the right and left anterior limb of the internal capsule (fR  =  0.40; fL  =  0.40). We also observed a medium effect size for differences in the MD in the left uncinate fasciculus (f  =  0.15).

Follow-up pairwise comparisons (SDC Table 5, https://links.lww.com/CBN/A101) revealed significantly increased MD in the right and left sagittal striatum and the left anterior corona radiata of the msTBI subgroup compared with both the cmTBI subgroup and the OI group, as well as increased MD in the msTBI subgroup relative only to the OI group in the genu of the corpus callosum, the right uncinate fasciculus, and the right anterior corona radiata, and increased MD in the msTBI subgroup relative only to the cmTBI subgroup in the left anterior limb of the internal capsule. The cmTBI subgroup did not differ from the OI group in the MD of any tract.

Group and Subgroup Differences in CYRM–28 Scores at Follow-up

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, significant group differences with medium to large effect sizes were seen at the 1-year follow-up across the Individual (f  =  0.61), Caregiver (f  =  0.43), and Context (f  =  0.35) subscales as well as the CYRM–28 Total score (f  =  0.52; Table 2).

Follow-up pairwise comparisons (SDC Table 6, https://links.lww.com/CBN/A102) revealed significantly decreased Individual subscale and Total scores in the msTBI subgroup compared with both the cmTBI subgroup and the OI group, and significantly decreased Caregiver and Context subscale scores in the msTBI subgroup relative only to the OI group. The cmTBI subgroup did not differ from the OI group on any CYRM–28 subscale or Total scores.

Relationship Between FA/MD and CYRM–28 Scores

After removing the effects of age at injury, time since injury, and study site, and controlling for multiple comparisons, partial correlations demonstrated no significant relationship between the FA or MD of any tract with CYRM–28 subscale or Total scores between either TBI subgroup (SDC Tables 7 and 8, https://links.lww.com/CBN/A103 and https://links.lww.com/CBN/A104). In each subgroup, however, we observed several significant relationships with medium to large effect sizes. Specifically, in the cmTBI subgroup, an increased score on the CYRM–28 Individual subscale was associated with increased FA of the left sagittal striatum (r2Y(1.2)  =  0.16), an increased score on the CYRM–28 Caregiver subscale was associated with decreased MD of the left anterior corona radiata (r2Y(1.2)  =  0.16) and right (r2Y(1.2)  =  0.19) and left (r2Y(1.2)  =  0.26) anterior limb of the internal capsule, and an increased CYRM–28 Total score was associated with decreased MD of the right anterior limb of the internal capsule (r2Y(1.2)  =  0.13).

In the msTBI subgroup, no associations with FA were found; however, an increased score on the CYRM–28 Individual subscale was associated with decreased MD of the left anterior corona radiata (r2Y(1.2)  =  0.15); an increased score on the CYRM–28 Caregiver subscale was associated with decreased MD of the right anterior corona radiata (r2Y(1.2)  =  0.16); an increased score on the CYRM–28 Context subscale was associated with decreased MD of the genu of the corpus callosum (r2Y(1.2)  =  0.20), left sagittal striatum (r2Y(1.2)  =  0.15), and right (r2Y(1.2)  =  0.26) and left (r2Y(1.2)  =  0.20) anterior corona radiata; and an increased CYRM–28 Total score was associated with decreased MD of the genu of the corpus callosum (r2Y(1.2)  =  0.16) and the right (r2Y(1.2)  =  0.20) and left (r2Y(1.2)  =  0.17) anterior corona radiata. The relationship between FA/MD and CYRM–28 scores between the cmTBI, msTBI, and OI groups is represented graphically in Figure 1.

FIGURE 1
FIGURE 1:
Relationship between FA/MD and CYRM–28 scores between the cmTBI, msTBI, and OI groups. cm = complicated mild. CYRM–28 = Child and Youth Resilience Measure. FA = fractional anisotropy. MD = mean difficulty. ms = moderate or severe. OI = orthopedic injury. TBI = traumatic brain injury.

DISCUSSION

We endeavored to determine if there was a relationship between resilience-promoting factors that are known to promote resilience and WM microstructure 1 year after cmTBI or msTBI that had been sustained by adolescents. We supported our main hypothesis. That is, individuals in the overall TBI group demonstrated reduced WM integrity of structures throughout various regions of the brain, including the genu of the corpus callosum, the anterior corona radiata, and the anterior limb of the internal capsule. These findings are consistent with previous studies from our group and other groups (Grieve et al, 2007; Hunter et al, 2012; Levin et al, 2008; Martino et al, 2010; Schmidt et al, 2013a; Wilde et al, 2006). In addition, adolescents in the TBI group exhibited fewer resilience-promoting factors at 1 year post injury compared with matched controls who had sustained only an OI. These differences appeared most striking in the msTBI individuals despite a lack of baseline group differences in terms of resilience-promoting factors or important covariates.

TBI and Resilience-promoting Factors

The pattern of findings in the overall TBI group are consistent with other research demonstrating a high likelihood of changes in family and caregiver functioning after injury of a child; thus, it is reasonable to expect that the present findings of decreased resilience-promoting factors in the TBI group reflect postinjury changes in the family system and the functioning of the adolescent in the broader community (Langlois et al, 2003; Schönberger et al, 2010; Wade et al, 2011).

Although no significant subgroup differences were observed in baseline resilience-promoting factors between the cmTBI, msTBI, and OI groups, 1-year follow-up results indicated striking changes in the msTBI subgroup when compared with both the cmTBI subgroup and the OI group. That is, the msTBI subgroup exhibited fewer resilience-promoting factors across all of the domains of the CYRM–28 at follow-up compared with the OI group. Notable, but less dramatic, differences were observed between the cmTBI and msTBI subgroups. The cmTBI subgroup demonstrated moderate (but not significant) differences in caregiver and context resilience-promoting factors at baseline, which may have impacted the patterns that were observed at the follow-up assessment.

These findings suggest that an msTBI is much more likely than a cmTBI to have a detrimental impact on the family and social functioning of adolescents (Ryan et al, 2016b; Schmidt et al, 2010c). The decreases in adaptive functioning, social skills, cognitive abilities, and personality changes that are the hallmarks of severe TBI, combined with increased medical expenses and caregiver burden, likely underlie these differences (Ryan et al, 2016b; Schmidt et al, 2010c; Yeates et al, 1997, 2004); however, the current findings highlight that TBIs, most especially moderate and severe injuries, undermine the very social capital and caregiver support that is necessary for a resilient recovery following a brain injury (Fay et al, 2009; Schmidt et al, 2010c, 2014).

Resilience-promoting Factors, TBI Subgroups, and WM Integrity

Although our present findings indicated that resilience-promoting factors were related to WM integrity in the overall TBI group, subgroup analyses revealed more striking effects in the msTBI subgroup in terms of both disruptions of FA/MD and WM relationships with resilience-promoting factors, suggesting that the msTBI subgroup may have been driving most of the effects that were seen in the overall TBI group. Interestingly, the individuals in the cmTBI subgroup did not differ from the OI individuals in terms of FA and MD in any tract. Nonetheless, there were significant relationships between the DTI metrics and the CYRM–28 scores in the cmTBI subgroup that were not present in the OI group, suggesting some specificity of the findings to the postinjury period in the cmTBI subgroup. That is, significant and unique relationships emerged between CYRM–28 Caregiver and Total scores and the anterior limb of the internal capsule. Other than issues related to injury type, the precise reasons for these unique relationships in the cmTBI subgroup are unclear, although previous research has indicated that the anterior limb of the internal capsule may be vulnerable to long-term disruption following a concussive injury (Terry et al, 2019).

The msTBI subgroup demonstrated a greater number of significant correlations between resilience-promoting factors as assessed by the CYRM–28 and specific WM tracts—a finding that is consistent with the increased WM disruptions that are associated with more severe injuries. Moreover, individuals in the msTBI subgroup had a greater likelihood of sustaining a high-velocity injury compared with individuals in the cmTBI subgroup (SDC Table 8, https://links.lww.com/CBN/A104); this finding may also account for the greater number of significant relationships as well as the greater involvement of larger tracts such as the genu of the corpus callosum that were observed in the msTBI subgroup. The observed differences between individuals sustaining cmTBIs and those sustaining msTBIs underscores the necessity to account for injury severity and to exercise caution when combining cm and ms injuries together in the same analyses (Levin et al, 2004, 2008; Levin and Hanten, 2005; Yeates et al, 2010).

WM Tracts and Resilience-promoting Factors

Previous research indicates that each of the tracts with the most robust WM findings as related to resilience-promoting factors (ie, genu of the corpus callosum, corona radiata, and internal capsule) are vulnerable to disruption after TBI (McDonald et al, 2018; Raikes et al, 2018; Schmidt et al, 2013b; Terry et al, 2019; Yin et al, 2019). Research has also indicated that frontotemporal WM integrity broadly relates to social cognition abilities in individuals with TBI (Levin et al, 2011; Ryan et al, 2018), whereas other research specifically implicated these specific structures in social cognitive processing in TBI and other neurologic populations (Batista et al, 2017; Cheng et al, 2010; McDonald et al, 2018; Mike et al, 2013; Radoeva et al, 2012; Schmidt et al, 2010b, 2013b).

Finally, other studies have indicated that psychosocial factors such as high SES protect against age-related WM degeneration and support the possibility that environmental factors impact WM structure and function (Johnson et al, 2013). In the context of these previous investigations, the fact that our current findings suggest that the integrity of specific WM tracts are related to resilience-promoting factors facilitating the process of resilience through social relationships and caregiver and community support is not surprising.

Potential Explanations of Current Findings

It is possible that these brain–behavior relationships themselves are both independent indicators of improved trajectories after TBI. That is, the correlation between WM tracts and the protective effects of parenting and context may represent generally better premorbid functioning/recovery in those individuals without being a causative agent. Thus, it is possible that the present findings represent correlation without any causative relationships. However, the lack of baseline differences in the CYRM–28 scores between the TBI and OI groups, lack of a similar pattern emerging in the OI group, and our inclusion of an extensive variety of covariates including estimated intellectual functioning greatly increases our confidence that the current pattern of findings is not the result of spurious correlations.

Although our analyses controlled for age, the TBI group was slightly older than the OI group. It is possible that relationships between resilience-promoting factors and WM integrity emerge at later stages in development, and a similar pattern of findings would have been observed in the OI group had they been closer to the TBI individuals in chronological age at the time of their injury. However, the groups did not differ statistically in age or time since injury, and we were able to control for age and time since injury in our analyses. Unfortunately, our OI group was not sufficiently large enough to conduct a separate analysis examining the correlations between CYRM–28 scores and specific WM tracts across different age groups, although this would be an informative topic for future studies to address.

Another intriguing possibility is that the resilience-promoting factors that are represented by the CYRM–28 scores may directly lead to neural plasticity after TBI occurring in adolescence. That is, more psychosocial resilience-promoting factors may have helped individuals in the TBI group (especially the msTBI subgroup) to maintain or rebuild WM tracts. It is true that many of the variables measured by the CYRM–28 have been linked to more positive psychosocial outcomes after TBI in adolescents (Ryan et al, 2016; Schmidt et al, 2010c, 2014; Yeates et al, 2007, 2010); however, evidence of causative relationships and mechanisms linking these factors to specific neuroplastic changes are lacking.

Although it is tempting to claim that the positive correlations between resilience-promoting factors and WM integrity are evidence of a direct protective effect of these psychosocial influences operating in the TBI group, caution is warranted. That is, confidence in this speculation would be greatly bolstered by the inclusion of baseline imaging data and multiple extended follow-up assessments resulting in data that are amenable to more sophisticated longitudinal analyses. Nonetheless, it is difficult to account for the lack of significant findings in the OI group because these correlations would be expected if resilience-promoting factors merely indexed better overall brain health and development.

Our findings may represent evidence that well-functioning postinjury psychosocial environments (caregiver and context factors especially) serve to buffer an adolescent from neurologic damage and/or foster recovery following TBI. This notion is consistent with previous research suggesting that parenting and family functioning play a role in promoting recovery following brain injury, but this concept has not been demonstrated with regard to neurobiological indicators of recovery following TBI (Ryan et al, 2016b; Schmidt et al, 2010c; Yeates et al, 1997, 2010). For example, previous research has indicated that caregiver influences including family functioning and parental behaviors, such as high warmth and low negativity, are related to better outcomes after pediatric TBI (Schmidt et al, 2010c; Wade et al, 2011). Thus, it is possible that the current observations reflect the neuroprotective effect of these resilience-promoting factors following a pediatric head injury.

Finally, it is possible that the current findings represent real differences in the trajectory of recovery for those individuals with higher numbers of resilience-promoting factors, and this difference represents changes in postinjury psychosocial stress. For example, it may be that in circumstances in which the brain is not challenged (ie, as in the OI group), there is little significant relationship/interaction between psychosocial support and WM integrity in the regions of interest. However, this relationship may emerge under situations of neurobiological stress, as was seen in the TBI group. This speculative explanation rests on the supposition that in addition to direct damage to brain areas, TBIs induce a series of downstream neurobiological cascades and psychosocial risks (eg, family discord, financial hardship) that directly or indirectly may trigger other adverse neuroplastic changes (Clausen et al, 2012; Kreber and Griesbach, 2016; Srinivas et al, 2010). As such, the current findings may represent a protective process wherein individuals with strong psychosocial support are partially protected from these indirect effects and may evidence more intact structures in long-term follow-ups. This mechanism may help to explain the relative lack of correlations in the OI group and in the presence of significant results in both subgroups of TBI individuals, even if the findings in the cmTBI subgroup were less robust.

Study Limitations

Our relatively small sample size did not allow us to break groups into more precise age ranges. Moreover, we acknowledge that site differences can be problematic in multisite studies due to differences in scanner hardware, software, and other issues (Burt et al, 2016; Hunter et al, 2012). Although harmonization techniques for imaging data have been proposed, no single method has been widely accepted at this time. We note that the hardware and software from both sites were comparable, and that the same acquisition parameters were used at the two sites. Further, the current investigation matched individuals on important characteristics for TBI research; however, a different pattern of findings may have emerged if variables that are more typical of child development and developmental psychopathology studies had been used to recruit and match the individuals.

Our findings were limited by not having complete baseline data regarding resilience measurement and by having only a single follow-up period at 1 year post injury. However, there were no significant group differences at baseline in terms of resilience variables, and most functional recovery can be expected to occur in the first 12 months following pediatric TBI (Levin and Hanten 2005; Yeates et al, 2007). Nonetheless, our study would have been strengthened by more comprehensive measurement of resilience at baseline and by additional long-term follow-up assessments because recovery up to 2 years or even longer post injury is possible.

Finally, we were limited by our lack of other outcome measures and our use of a single imaging modality (ie, DTI); thus, it is unclear to what extent our individuals with TBI actually demonstrated clinically significant and observable recovery of function or to which similar findings would be obtained using structural imaging, spectroscopy, or physiological measurement of the brain and neuronal function.

CONCLUSION

We are not aware of any studies that have used measures of brain structure to clearly demonstrate the protective role of psychosocial variables. As such, although the explanations of mechanisms and the causative attributions of the present investigation were limited by various methodological issues, the current study suggests that the protective effects of resilience-promoting factors (a) may result in less disruption to WM tracts and (b) nicely correspond to previous studies demonstrating improvements in cognitive and behavioral outcomes with higher levels of family and caregiver support.

Given the relatively small sample size and the lack of multiple data points to evaluate change in CYRM–28 scores and WM, the current findings are best viewed as exploratory in nature. Thus, larger studies that examine a wider range of resilience-promoting factors and that collect neuroimaging and CYRM–28 data on multiple occasions in the first 2 years after TBI will be necessary to disentangle our explanations of the current results.

Our study highlights the importance of supportive community and caregiving environments for fostering the continued development and recovery of adolescents who have sustained TBI. It also hints at potential neurobiological protective mechanisms that may explain these beneficial outcomes.

REFERENCES

Anderson JLR, Sotiropoulos SN. 2016. An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage. 125:1063–1078. doi:10.1016/j.neuroimage.2015.10.019
Anderson V, Beauchamp MH, Yeates KO, et al. 2013. Social competence at 6 months following childhood traumatic brain injury. J Int Neuropsychol Soc. 19:539–550. doi:10.1017/S1355617712001543
Armstrong MI, Birnie-Lefcovitch S, Ungar MT. 2005. Pathways between social support, family well being, quality of parenting, and child resilience: what we know. J Child Fam Stud. 14:269–281. doi:10.1007/s10826-005-5054-4
Avants BB, Epstein CL, Grossman M, et al. 2008. Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Med Image Anal. 12:26–41. doi:10.1016/j.media.2007.06.004
Avants BB, Tustison N, Song G. 2009. Advanced normalization tools (ANTs). Insight J. 2:1–35.
Avants BB, Tustison NJ, Song G, et al. 2011. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage. 54:2033–2044. doi:10.1016/j.neuroimage.2010.09.025
Avci G, Hanten G, Schmidt AT, et al. 2013. Cognitive contributors to resilience in youth from underserved populations: a brief report. J Public Ment Health. 12:165–170. doi:10.1108/JPMH-02-2013-0005
Batista S, Alves C, d’Almeida OC, et al. 2017. Disconnection as a mechanism for social cognition impairment in multiple sclerosis. Neurology. 89:38–45. doi:10.1212/WNL.0000000000004060
Benjamini Y, Hockberg Y. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol. 57:289–300.
Burt KB, Whelan R, Conrod PJ, et al. 2016. Structural brain correlates of adolescent resilience. J Child Psychol Psychiatry. 57:1287–1296. doi:10.1111/jcpp.12552
Casey BJ, Galvan A, Hare TA. 2005. Changes in cerebral functional organization during cognitive development. Curr Opin Neurobiol. 15:239–244. doi:10.1016/j.conb.2005.03.012
Casey BJ, Jones RM, Hare TA. 2008. The adolescent brain. Ann N Y Acad Sci. 1124:111–126. doi:10.1196/annals.1440.010
Cheng Y, Chou KH, Chen IY, et al. 2010. Atypical development of white matter microstructure in adolescents with autism spectrum disorders. Neuroimage. 50:873–882. doi:10.1016/j.neuroimage.2010.01.011
Cicchetti D, Rogosch FA. 2009. Adaptive coping under conditions of extreme stress: multilevel influences on the determinants of resilience in maltreated children. Published online Summer 2009. New Dir Child Adolesc Dev. 2009:47–59. doi:10.1002/cd.242
Clausen F, Marklund N, Lewén A, et al. 2012. Interstitial F(2)-isoprostane 8-iso-PGF(2α) as a biomarker of oxidative stress after severe human traumatic brain injury. J Neurotrauma. 29:766–775. doi:10.1089/neu.2011.1754
Cohen J. 1988. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York, New York: Lawrence Erlbaum Associates.
Dennis EL, Babikian T, Giza CC, et al. 2018. Neuroimaging of the injured pediatric brain: methods and new lessons. Neuroscientist. 24:652–670. doi:10.1177/1073858418759489
Elliott TR, Hsiao YY, Kimbrel NA, et al. 2015. Resilience, traumatic brain injury, depression and posttraumatic stress among Iraq/Afghanistan war veterans. Rehabil Psychol. 60:263–276. doi:10.1037/rep0000050
Fay TB, Yeates KO, Wade SL, et al. 2009. Predicting longitudinal patterns of functional deficits in children with traumatic brain injury. Neuropsychology. 23:271–282. doi:10.1037/a0014936
Galinowski A, Miranda R, Lemaitre H, et al. 2015. Resilience and corpus callosum microstructure in adolescence. Psychol Med. 45:2285–2294. doi:10.1017/S0033291715000239
Ganesalingam K, Sanson A, Anderson V, et al. 2006. Self-regulation and social and behavioral functioning following childhood traumatic brain injury. J Int Neuropsychol Soc. 12:609–621. doi:10.1017/S1355617706060796
Grieve SM, Williams LM, Paul RH, et al. 2007. Cognitive aging, executive function, and fractional anisotropy: a diffusion tensor MR imaging study. AJNR Am J Neuroradiol. 28:226–235.
Hanten G, Wilde EA, Menefee DS, et al. 2008. Correlates of social problem solving during the first year after traumatic brain injury in children. Neuropsychology. 22:357–370. doi:10.1037/0894-4105.22.3.357
Holland JN, Schmidt AT. 2015. Static and dynamic factors promoting resilience following traumatic brain injury: a brief review. Published online August 4. Neural Plast. 2015:902802. doi:10.1155/2015/902802
Hunter JV, Wilde EA, Tong KA, et al. 2012. Emerging imaging tools for use with traumatic brain injury research. J Neurotrauma. 29:654–671. doi:10.1089/neu.2011.1906
Jahanshad N, Kochunov PV, Sprooten E, et al. 2013. Multi-site genetic analysis of diffusion images and voxelwise heritability analysis: a pilot project of the ENIGMA-DTI working group. Neuroimage. 81:455–469. doi:10.1016/j.neuroimage.2013.04.061
Johnson NF, Kim C, Gold BT. 2013. Socioeconomic status is positively correlated with frontal white matter integrity in aging. Age (Dordr). 35:2045–2056. doi:10.1007/s11357-012-9493-8
Kalisch R, Müller MB, Tüscher O. 2015. Advancing empirical resilience research. Published online September 2. Behav Brain Sci. 38:e128. doi:10.1017/S0140525X15000023
Kraus JF Broman SH, Michel ME. 1995. Epidemiological features of brain injury in children: occurrence, children at risk, causes and manner of injury, severity, and outcomes. Traumatic Head Injury in Children. New York, New York: Oxford; 22–39.
Kreber LA, Griesbach GS. 2016. The interplay between neuropathology and activity based rehabilitation after traumatic brain injury. Brain Res. 1640:152–163. doi:10.1016/j.brainres.2016.01.016
Langlois JA, Kegler SR, Butler JA, et al. 2003. Traumatic brain injury-related hospital discharges. Results from a 14-state surveillance system, 1997. MMWR Surveill Summ. 52:1–20.
Levin HS, Hanten G. 2005. Executive functions after traumatic brain injury in children. Pediatr Neurol. 33:79–93. doi:10.1016/j.pediatrneurol.2005.02.002
Levin HS, Hanten G, Zhang L, et al. 2004. Selective impairment of inhibition after TBI in children. J Clin Exp Neuropsychol. 26:589–597. doi:10.1080/13803390409609783
Levin HS, Wilde EA, Chu Z, et al. 2008. Diffusion tensor imaging in relation to cognitive and functional outcome of traumatic brain injury in children. J Head Trauma Rehabil. 23:197–208. doi:10.1097/01.HTR.0000327252.54128.7c
Levin HS, Wilde EA, Hanten G, et al. 2011. Mental state attributions and diffusion tensor imaging after traumatic brain injury in children. Dev Neuropsychol. 36:273–287. doi:10.1080/87565641.2010.549885
Losoi H, Silverberg ND, Wäljas M, et al. 2015. Resilience is associated with outcome from mild traumatic brain injury. J Neurotrauma. 32:942–949. doi:10.1089/neu.2014.3799
Martino J, Brogna C, Robles SG, et al. 2010. Anatomic dissection of the inferior fronto-occipital fasciculus revisited in the lights of brain stimulation data. Cortex. 46:691–699. doi:10.1016/j.cortex.2009.07.015
Masten AS. 2001. Ordinary magic: resilience processes in development. Am Psychol. 56:227–238. doi:10.1037/0003-066X.56.3.227
Masten AS, Hubbard JJ, Gest SD, et al. 1999. Competence in the context of adversity: pathways to resilience and maladaptation from childhood to late adolescence. Dev Psychopathol. 11:143–169. doi:10.1017/s0954579499001996
Max JE, Wilde EA, Bigler ED, et al. 2012. Psychiatric disorders after pediatric traumatic brain injury: a prospective, longitudinal, controlled study. J Neuropsychiatry Clin Neurosci. 24:427–436. doi:10.1176/appi.neuropsych.12060149
McCauley SR, Wilde EA, Miller ER, et al. 2013. Preinjury resilience and mood as predictors of early outcome following mild traumatic brain injury. J Neurotrauma. 30:642–652. doi:10.1089/neu.2012.2393
McDonald S, Rushby JA, Dalton KI, et al. 2018. The role of abnormalities in the corpus callosum in social cognition deficits after traumatic brain injury. Soc Neurosci. 13:471–479. doi:10.1080/17470919.2017.1356370
Micklewright JL, King TZ, O’Toole K, et al. 2012. Parental distress, parenting practices, and child adaptive outcomes following traumatic brain injury. J Int Neuropsychol Soc. 18:343–350. doi:10.1017/S1355617711001792
Mike A, Strammer E, Aradi M, et al. 2013. Disconnection mechanism and regional cortical atrophy contribute to impaired processing of facial expressions and theory of mind in multiple sclerosis: a structural MRI study. Published online December 13. PLoS One. 8:e82422. doi:10.1371/journal.pone.0082422
Pineda RG, Neil J, Dierker D, et al. 2014. Alterations in brain structure and neurodevelopmental outcome in preterm infants hospitalized in different neonatal intensive care unit environments. J Pediatr. 164:52–60. doi:10.1016/j.peds.2013.08.047
Potter JL, Wade SL, Walz NC, et al. 2011. Parenting style is related to executive dysfunction after brain injury in children. Rehabil Psychol. 56:351–358. doi:10.1037/a0025445
Radoeva PD, Coman IL, Antshel KM, et al. 2012. Atlas-based white matter analysis in individuals with velo-cardio-facial syndrome (22q11.2 deletion syndrome) and unaffected siblings. Published online August 1. Behav Brain Funct. 8:38. doi:10.1186/1744-9081-8-38
Raikes AC, Bajaj S, Dailey NS, et al. 2018. Diffusion tensor imaging (DTI) correlates of self-reported sleep quality and depression following mild traumatic brain injury. Published online June 20. Front Neurol. 9:468. doi:10.3389/fneur.2018.00468
Raj SP, Wade SL, Cassedy A, et al. 2014. Parent psychological functioning and communication predict externalizing behavior problems after pediatric traumatic brain injury. J Pediatr Psychol. 39:84–95. doi:10.1093/jpepsy/jst075
Rosema S, Crowe L, Anderson V. 2012. Social function in adolescents after traumatic brain injury: a systematic review 1989–2011. J Neurotrauma. 29:1277–1291. doi:10.1089/neu.2011.2144
Ryan NP, Catroppa C, Godfrey C, et al. 2016a. Social dysfunction after pediatric traumatic brain injury: a translational perspective. Neurosci Biobehav Rev. 64:196–214. doi:10.1016/j.neubiorev.2016.02.020
Ryan NP, van Bijnen L, Catroppa C, et al. 2016b. Longitudinal outcome and recovery of social problems after pediatric traumatic brain injury (TBI): contribution of brain insult and family environment. Int J Dev Neurosci. 49:23–30. doi:10.1016/j.ijdevneu.2015.12.004
Ryan NP, Genc S, Beauchamp MH, et al. 2018. White matter microstructure predicts longitudinal social cognitive outcomes after paediatric traumatic brain injury: a diffusion tensor imaging study. Psychol Med. 48:679–691. doi:10.1017/S0033291717002057
Schmidt AT, Hanten GR, Li X, et al. 2010a. Emotion recognition following pediatric traumatic brain injury: longitudinal analysis of emotional prosody and facial emotion recognition. Neuropsychologia. 48:2869–2877. doi:10.1016/j.neuropsychologia.2010.05.029
Schmidt A, Martin R, Ozturk A, et al. 2010b. Neuroimaging and neuropsychological follow-up study in a pediatric brain tumor patient treated with surgery and radiation. Neurocase. 16:74–90. doi:10.1080/13554790903329133
Schmidt AT, Orsten KD, Hanten GR, et al. 2010c. Family environment influences emotion recognition following paediatric traumatic brain injury. Brain Inj. 24:1550–1560. doi:10.3109/02699052.2010.523047
Schmidt AT, Hanten GR, Li X, et al. 2012. Decision making after pediatric brain injury: trajectory of recovery and relationship to age and gender. Int J Dev Neurosci. 30:225–230. doi:10.1016/j.ijdevneu.2011.11.003
Schmidt AT, Hanten G, Avci G, et al. 2013a. Relating resilience factors and decision-making skills in two groups of underserved adolescents: implications for intervention. J Juv Justice. 2:69–75.
Schmidt AT, Hanten G, Li X, et al. 2013b. Emotional prosody and diffusion tensor imaging in children after traumatic brain injury. Brain Inj. 27:1528–1535. doi:10.3109/02699052.2013.828851
Schmidt A, Beers S, Levin H Albright L, Pollack EF, Adelson PD. 2014. Rehabilitation and outcome of head injuries. Principles and Practice of Pediatric Neurosurgery. New York, New York: Thieme Medical; 737–749.
Schönberger M, Ponsford J, Olver J, et al. 2010. A longitudinal study of family functioning after TBI and relatives’ emotional status. Neuropsychol Rehabil. 20:813–829. doi:10.1080/09602011003620077
Smith SM, Jenkinson M, Johansen-Berg H, et al. 2006. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage. 31:1487–1505. doi:10.1016/j.neuroimage.2006.02.024
Srinivas R, Brown SD, Chang YF, et al. 2010. Endocrine function in children acutely following severe traumatic brain injury. Childs Nerv Syst. 26:647–653. doi:10.1007/s00381-009-1038-9
Steinberg L. 2004. Risk taking in adolescence: What changes, and why? Ann N Y Acad Sci. 1021:51–58. doi:10.1196/annals.1308.005
Steinberg L. 2008. A social neuroscience perspective on adolescent risk-taking. Dev Rev. 28:78–106. doi:10.1016/j.dr.2007.08.002
Steinberg L. 2010. A dual systems model of adolescent risk-taking. Dev Psychobiol. 52:216–224. doi:10.1002/dev.20445
Teasdale G, Jennett B. 1974. Assessment of coma and impaired consciousness. A practical scale. Lancet. 2:81–84. doi:10.1016/S0140-6736(74)91639-0
Terry DP, Mewborn CM, Miller LS. 2019. Repeated sport-related concussion shows only minimal white matter differences many years after playing high school football. J Int Neuropsychol Soc. 25:950–960. doi:10.1017/S1355617719000754
Tonks J, Yates P, Frampton I, et al. 2011. Resilience and the mediating effects of executive dysfunction after childhood brain injury: a comparison between children aged 9–15 years with brain injury and non-injured controls. Brain Inj. 25:870–881. doi:10.3109/02699052.2011.581641
Tukey J. 1977. Exploratory Data Analysis. Reading, Massachusetts: Addison-Wesley.
Tustison NJ, Cook PA, Klein A, et al. 2014. Large-scale evaluation of ANTs and FreeSurfer cortical thickness measurements. Neuroimage. 99:166–179. doi:10.1016/j.neuroimage.2014.05.044
Ungar M. 2004. The importance of parents and other caregivers to the resilience of high-risk adolescents. Fam Process. 43:23–41. doi:10.111/j.1545-5300.2004.04301004.x
Ungar M. 2008. Resilience across cultures. Br J Soc Work. 38:218–235. doi:10.1093/bjsw/bcl343
Ungar M, Lienbenberg L. 2011. Assessing resilience across cultures using mixed methods: construction of the child and youth resilience measure. J Mix Methods Res. 5:126–149. doi:10.1177/1558689811400607
van der Werff SJ, Pannekoek JN, Stein DJ, et al. 2013. Neuroimaging of resilience to stress: current state of affairs. Hum Psychopharmacol. 28:529–532. doi:10.1002/hup.2336
Wade SJ, Cassedy A, Walz NC, et al. 2011. The relationship of parental warm responsiveness and negativity to emerging behavior problems following traumatic brain injury in young children. Dev Psychol. 47:119–133. doi:10.1037/a0021028
Waugh CE, Wager TD, Fredrickson BL, et al. 2008. The neural correlates of trait resilience when anticipating and recovering from threat. Soc Cogn Affect Neurosci. 4:322–332. doi:10.1093/scan/nsn024
Wechsler D. 2011. Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI–II). San Antonio, Texas: NCS Pearson.
Wilde EA, Chu Z, Bigler ED, et al. 2006. Diffusion tensor imaging in the corpus callosum in children after moderate to severe traumatic brain injury. J Neurotrauma. 23:1412–1426. doi:10.1089/neu.2006.23.1412
Yeates KO, Taylor HG, Drotar DD, et al. 1997. Pre-injury family environment as a determinant of recovery from traumatic brain injuries in school-age children. J Int Neuropsychol Soc. 3:617–630.
Yeates KO, Swift E, Taylor HG, et al. 2004. Short-and long-term social outcomes following pediatric traumatic brain injury. J Int Neuropsychol Soc. 10:412–426. doi:10.1017/S1355617704103093
Yeates KO, Bigler ED, Dennis M, et al. 2007. Social outcomes in childhood brain disorder: a heuristic integration of social neuroscience and developmental psychology. Psychol Bull. 133:535–556. doi:10.1037/0033-2909.133.3.535
Yeates KO, Taylor HG, Walz NC, et al. 2010. The family environment as a moderator of psychosocial outcomes following traumatic brain injury in young children. Neuropsychology. 24:345–356. doi:10.1037/a0018387
Yin B, Li DD, Huang H, et al. 2019. Longitudinal changes in diffusion tensor imaging following mild traumatic brain injury and correlation with outcome. Published online May 7. Front Neural Circuits. 13:28. doi:10.3389/fncir.2019.00028
Keywords:

traumatic brain injury; diffusion tensor imaging; resilience-promoting factors; neuroimaging; resilience

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.