Personality Change Due to Traumatic Brain Injury in Children and Adolescents: Neurocognitive Correlates

Personality change (PC) due to traumatic brain injury (TBI) is the most common and important acquired psychiatric disorder after brain trauma in children and adolescents.^1–5 We previously described the phenomenology of PC in detail; however, the neurocognitive correlates of PC have not been studied.⁵ The rationale for investigating the neurocognitive profile of PC is that this may shed light on the neurobiological systems or networks that are perturbed and may provide a link between neurobiology and observed behaviors.^6,7

PC has five specific subtypes: affectively labile, aggressive, disinhibited, apathetic, and paranoid. The first three subtypes are common and often co-occur, whereas the latter two subtypes are uncommon.⁵ Studies show that most, if not all, cases of PC are accounted for by having at least the affectively labile subtype with impairing irritability.³ The irritability has numerous potential antecedents, including a frustrated need for constant attention, sensitivity to criticism, concrete thinking, delay in gratification, unpredictability or change in routine, intellectual or concentration deficits increasing the effort in task completion, communication difficulties resulting in misunderstanding humor and instructions, and sensitivity to pain or accidental mild injury.⁵ The course of PC is continuous, rather than episodic. The temper tantrums or rages that are associated with PC may be interspersed with euthymia or with persistent irritability, varying case by case. PC manifests as a disorder of mood regulation with associated behavioral disruption. Consequently, PC rarely occurs alone as a new-onset psychiatric disorder after TBI. The comorbid new-onset disorders include anxiety disorders,⁸ depressive disorders,⁹ mania/hypomania,¹⁰ attention deficit hyperactivity disorder (ADHD),² and oppositional defiant disorder.² There are a number of misperceptions regarding the diagnosis of PC. Most importantly, PC is not a personality disorder. Rather, PC is a collection of clinically significant symptoms, which are described by its phenomenologically evocative named subtypes. Children, adolescents, and adults who develop PC are so impaired and radically altered as to be considered to have had a change in their personality.

PC occurs in up to 40% of cases of consecutively hospitalized children with severe TBI.² The disorder is significantly associated with measures of severity of injury, rather than psychosocial variables.^2–4 PC is associated with lesions within the superior frontal gyrus within the first year postinjury,^3,4 consistent with proposed models of affective dysregulation.^11–13 In the second year after injury, PC is associated with frontal white matter lesions and preinjury adaptive function, suggesting that function is limited by network (rather than by gray matter) damage as well as by the preinjury brain-behavioral reserve within each individual.⁴

In the absence of studies of the neurocognitive correlates of PC, our approach was to examine the relationship of PC with neurocognitive domains that are each known to be sensitive to the effects of TBI. These domains include attention, processing speed, verbal memory, intellectual function, working memory, executive function, naming, expressive language, and motor speed.¹⁴ Pediatric TBI studies generally have found that motor inhibition measured by the Stop Signal Task is not significantly related to brain injury and is only even inconsistently related to the diagnosis of postinjury new-onset (secondary) ADHD.¹⁵ We thus did not expect PC to be related to motor inhibition, although emotional and/or behavioral disinhibition is present in PC.

Methods

The methodology of this study was previously reported in detail.³

Participants

Children and adolescents (N=177) 5–14 years old were enrolled from consecutively hospitalized patients after a single TBI at five academic medical centers including three in Texas, one in San Diego, and one in Toronto. Enrollment ranged from severe to mild TBI at every center except San Diego, where recruitment was limited to severe to complicated mild TBI. Youth with preinjury schizophrenia, autistic disorder, mental deficiency, and injury from child abuse or a penetrating bullet injury were excluded. Children and adolescents with preinjury ADHD were excluded in San Diego. Parents or guardians of all children and adolescents provided written informed consent, and all children and adolescents signed an assent to participate in accordance with each site’s institutional review board. Table 1 shows the demographic and injury indices for participants.

Psychiatric Measures

The Neuropsychiatric Rating Schedule¹ is a semistructured interview to identify symptoms and subtypes of PC. Parents and children served as informants in the interviews that took place at baseline (within 1 month postinjury) and at 6 months postinjury. At baseline, the lifetime preinjury psychiatric history was the focus of inquiry; at the 6-month assessment, the focus was the period from injury to 6 months postinjury. The Neuropsychiatric Rating Schedule interview generated ratings defining the five major subtypes of PC. We waived the criterion for patients to have a 1-year duration of symptomatology, in order to allow us to monitor the course of the disorder for the first 6 months after injury and to assess the neurocognitive correlates of PC. The Neuropsychiatric Rating Schedule has been shown to provide reliable and valid diagnoses of the common subtypes of PC.¹ Good convergent validity and good discriminant validity have been demonstrated for PC subtypes by using subscales from validated parent- and teacher-completed questionnaires that measure lability, aggression, impaired social judgment/disinhibition, apathy, and psychotic symptoms. Interrater agreement for Neuropsychiatric Rating Schedule items is fair to excellent, test-retest reliability is fair to good, and sensitivity to change has been demonstrated.¹

Other DSM-IV psychiatric diagnoses¹⁶ were derived by using the Schedule for Affective Disorders and Schizophrenia for School-Aged Children–Present and Lifetime version.¹⁷ This is a parent-child interview that generates diagnoses based on a clinician synthesizing data collected from the parent and child separately, reviewing present and lifetime symptoms (at baseline) and symptoms present or past from injury to 6 months (at 6-month assessment).

Best-estimate psychiatric diagnoses¹⁸ were generated by the interviewer after integrating the accounts of the parent and the patient from the Neuropsychiatric Rating Schedule and the Schedule for Affective Disorders and Schizophrenia for School-Aged Children–Present and Lifetime version interviews and, when available, from the Survey Diagnostic Instrument¹⁹ completed by the teacher.

TBI Classification

The classification of severity of TBI was based on the lowest postresuscitation score on the Glasgow Coma Scale,²⁰ which was recorded from the clinical notes. The Glasgow Coma Scale is the standard measure of severity of acute brain injury associated with TBI. The scale measures eye opening, motor, and verbal responsiveness. Scores range from 3 (unresponsive) to 15 (normal).

Socioeconomic Status

Socioeconomic status assessment was derived from the Four Factor Index.²¹ Classification depends on scores generated from a formula involving both the maternal and paternal occupational and educational levels. Scores range from 8 to 66, with lower scores indicating lower educational and occupational levels and lower socioeconomic status.

Neurocognitive Measures

Attention.

Attentional processes in single- and dual-task performance were measured with the Divided Attention Task,²² which assesses an individual’s ability to allocate attentional resources when he or she is simultaneously engaged in performing two independent tasks. Timed comparisons were evaluated for performing the single task of finger tapping versus simultaneously performing the dual tasks of right finger tapping and reciting a nursery rhyme. The outcome variable is number of words correct.

Processing speed.

The symbol search and coding subtests of the WISC-III²³ measure visual scanning, tracking ability, and psychomotor processing speed. These tasks require interpreting whether a target symbol appears in a row of symbols and in digit-symbol codes. Correct responses minus the number of errors completed within the allotted 120 seconds for each subtest were noted. Together, these measures provide a processing speed index.

Verbal memory.

The California Verbal Learning Test–Children’s Version²⁴ was administered and assesses verbal learning and memory abilities. Children were instructed to learn 15 words in three categories across five learning trials and one distraction trial. Verbal memory for long-delay free recall was assessed and expressed as a z score.

Intellectual function.

The Wechsler Abbreviated Scale of Intelligence²⁵ was administered to all children at 6 months postinjury. The Wechsler Abbreviated Scale of Intelligence is a test of intelligence that consists of the following four subtests: vocabulary, similarities, block design, and matrix reasoning. These four components evaluate a person’s verbal and nonverbal knowledge and reasoning as well as general cognitive functioning, and together, they produce a full-scale IQ, which was the variable reported here as a standard score.

Verbal working memory.

Working memory was evaluated with a computerized N-back experimental task.²⁶ The letter identity task has three levels of memory load: one back, two back, and three back. In addition, there is a zero-back condition, which imposes minimal memory load while controlling for attention to the task. Children were instructed to match the same alphabetic letters printed in different cases. For each level, there were 40 trials in which a string of 40 letters appeared one at a time for 2 seconds on the computer screen. The child responded by pressing a button with the preferred hand when a match occurred or, in the zero-load condition, when a designated target appeared. Training and practice occurred before experimental trials. The percentage of hits (i.e., detection of targets) was recorded.

Executive function.

A modified version of the Tower of London test²⁷ was used to assess planning skills that involve the ability to look ahead, follow rules, conceive of alternative solutions to the problem, and weigh and make choices. The test required that children arrive at the most direct, fewest move solution by determining the order of moves necessary within set rules (e.g., “pick up one bead at a time”) to rearrange three colored beads on pegs of three disks of differing heights. We report data on the total solution time (i.e., the time, measured in seconds, taken to come to the final solution for that trial).

Naming.

The Rapid Automatized Naming task²⁸ was administered by asking the participant to rapidly name line drawings of five common objects, which are reproduced 10 times each and interspersed on a board. This task is related to processing speed and reading. The time required to complete the task was measured and expressed as a z score.

Expressive language.

Expressive language was assessed with the Clinical Evaluation of Language Fundamentals–Third Edition²⁹ formulated sentence subtest, consisting of 22 items. Children were shown an image with a target word/phrase, and they were instructed to construct a sentence in response. Standard scores were recorded and analyzed.

Motor speed.

The Grooved Pegboard Test assesses fine motor coordination and manual dexterity.³⁰ Participants are instructed to insert 25 grooved pegs into a pegboard using one hand at a time as quickly as possible. Time to completion expressed as a z score for the dominant hand was recorded and used in the analysis.

Motor inhibition.

The Stop Signal Task assesses the ability to inhibit ongoing response in a choice reaction time task.³¹ The stop signal reaction time was used in the analyses. A prolonged stop signal reaction time has been identified as one of the signature executive deficits associated with ADHD.

Data Analysis

Effect sizes were used to compare neurocognitive scores between the groups with PC versus those without PC. However, because these scores could be influenced by severity of injury,¹⁴ socioeconomic status,³² and preinjury ADHD,⁶ linear regressions were conducted for each neurocognitive measure of interest controlling for these variables. The alpha value was kept at 0.05 because each of the regressions tested an independent hypothesized relationship between the neurocognitive measure and PC, guided by previous findings that each neurocognitive domain is sensitive to the effects of TBI.¹⁴

Results

Of the original 177 children and adolescents, 141 (80%) returned for the 6-month psychiatric assessment. The returning participants were not significantly different from those who did not return, in regard to age, race, gender, Glasgow Coma Scale score, or socioeconomic status. PC occurred in 31 (22%) of 141 patients at some point from the time of injury to the 6-month assessment. However, in five cases, the symptoms of PC remitted before the 6-month assessment, yielding 26 (18%) of 141 unremitted cases of PC. The affectively labile subtype occurred in 23 of the 26 cases. Of the three children with PC who did not meet criteria for the affectively labile subtype, two were diagnosed with the aggressive subtype of PC, and one was diagnosed with the disinhibited subtype. A detailed list including each child diagnosed with PC, along with their PC subtypes and specific brain lesions, was previously published.³ As in previous reports, comorbidity is common.² Unremitted PC was significantly associated with the following unremitted new-onset disorders: 1) new-onset ADHD was present in six of 19 children with PC versus 11 of 96 children with no PC (Fisher’s exact test=0.035); and 2) new-onset oppositional defiant disorder/conduct disorder/disruptive behavior disorder, not otherwise specified, was present in seven of 24 children with PC versus four of 110 children with no PC (Fisher’s exact test=0.001). PC was not significantly associated with new-onset unremitted depressive disorder (major depression/dysthymic disorder/depressive disorder, not otherwise specified), which was present in only six children, including three of 25 children with PC versus three of 113 with no PC (Fisher’s exact test=0.073). The denominators differ for the above disorders, depending on preinjury diagnoses (e.g., children with preinjury ADHD would not be eligible to develop new-onset ADHD).

Table 2 and Table 3 provide details on the relationship between each neurocognitive domain and PC. Table 2 demonstrates large effect sizes for processing speed (d=0.89) and full-scale IQ (d=1.14); moderate effect sizes for verbal memory (d=0.66), verbal working memory (d=0.73), expressive language (d=0.73), executive function (d=0.68), and attention (d=0.54); small effect sizes for naming (d=0.36); and trivial effect sizes for motor speed (d=0.13) and motor inhibition (d=0.17).

Table 3 illustrates in detail the linear regression analyses of the 10 neurocognitive domains and their respective relationships with PC, controlling for injury severity as measured by the Glasgow Coma Scale score, socioeconomic status, and preinjury ADHD (present versus absent). PC was independently related to full-scale IQ (p=0.04), divided attention (p=0.02), processing speed (p=0.01), verbal memory (p=0.01), verbal working memory (p=0.02), and executive function (p=0.01), but PC was not significantly related to naming, expressive language, motor speed, or motor inhibition. Consistent with extensive literature, naming/reading was significantly related to socioeconomic status.³³ Expressive language was significantly related to injury severity and socioeconomic status. Motor speed was significantly related to injury severity. Motor inhibition was related to ADHD at a trend level.

All statistical analyses reported in Table 2 and Table 3 were repeated comparing only the participants with the affective lability subtype of PC (N=23) versus all other participants. The results were essentially unchanged.

Discussion

The most important finding from this investigation is that PC was significantly associated with deficits in important neuropsychological domains, including intellectual function, processing speed, divided attention, verbal memory, working memory, expressive language, and executive function. Furthermore, these associations (except for expressive language) remained significant even when severity of brain injury, socioeconomic status, and preinjury ADHD were taken into account.

The relationship between PC and neurocognitive function is clearly not uniform, as evidenced by effect sizes ranging from large to small. This is not surprising, given previous brain lesion-behavior correlates implicating the dorsal frontal area, especially the superior frontal gyrus, in PC. The weak relationship between PC and reading and expressive language may be because these neurocognitive domains are relatively crystallized and are more closely related to socioeconomic status than to a specific pattern of brain damage.³³ Neurocognitive processes including executive function, verbal memory, working memory, and attention, which are mediated by frontal networks,¹⁴ might be expected to be deficient in children with disrupted affective regulation caused by damage to similar or overlapping neuronal networks. As anticipated, our negative Stop Signal Task findings suggest that problematic motor inhibition is distinct from the disinhibition or overreactivity of emotional response characteristic of PC.

Individuals with TBI are known to have difficulty understanding negative emotions such as anger, sadness, and fearfulness compared with positive emotions such as happiness,^34,35 and this may be associated with marked increased reactivity to negative emotional stimuli manifested verbally or behaviorally in children with PC. Furthermore, children with TBI exhibit impairments in understanding a form of affect regulation involving social suppression of emotional expressions.^34,36 Ecologically, better understanding of affect regulation predicts less rejection-victimization in the classroom.³⁷

The relationship between PC and the neuropsychological domains studied is strikingly similar to the corresponding relationship previously reported for another condition characterized by significant affective dysregulation, namely bipolar disorder.⁶ For example, the respective effect sizes for bipolar disorder and PC are as follows for the specific neuropsychological domains: verbal memory (d=0.77 versus d=0.66), attention (d=0.62 versus d=0.54), executive function (d=0.60 versus d=0.68), working memory (d=0.60 versus d=0.73), naming/reading (d=0.40 versus d=0.36), and motor speed (d=0.33 versus d=0.13). The corresponding relationship between bipolar disorder and PC with full-scale IQ (d=0.32 versus d=1.14) was notably different, a finding driven most likely as a result of the association of IQ and PC with greater severity of injury.² Another less striking difference from a recent study suggested a significant relationship with a small effect size between bipolar disorder and motor inhibition on the Stop Signal Task, in contrast with the trivial relationship in this study.³⁸

It is difficult to outline a clear mechanism whereby affective dysregulation, which is the core feature of PC (and an important feature of bipolar disorder), is related to the neurocognitive findings. One possibility is that the pattern of brain network damage leads independently to both PC and neurocognitive dysfunction. Clinically, this seems likely because affective dysregulation and neurocognitive problems are evident within the first few days of brain injury in children. Another possibility is that regulation of affect is modified by multiple neurocognitive processes. For example, individuals with slow processing, problematic divided attention, and poor memory may become overwhelmed and frustrated by environmental and interpersonal stimuli. Deficits in their working memory, planning, and problem-solving ability may lead to selection of more angry or aggressive responses because of difficulties working adaptively with new challenges. A third possibility is that the affective dysregulation leads to the array of neurocognitive problems, although this is much less likely because the child’s explosive irritability, although frequent, is not constant and is generally not present during the formal neurocognitive testing.

Our findings must be appreciated within the context of limitations of this study. First, the neurocognitive measures administered in this study included a broad array of domains known to be sensitive to brain injury. However, more specific neurocognitive measures—as well as psychophysiological and functional brain imaging modalities targeting recognition and understanding of negative emotion, suppression of emotional expression, and executive function with emotional distractors—were not used. This limited the potential for a more in-depth understanding of mechanisms underlying the expression of PC. Second, a continuous measure of affective lability might shed more light on the relationship with neurocognitive domains. Third, attrition was approximately 20%. However, participants were not significantly different from nonparticipants with respect to age, race, gender, injury severity, and socioeconomic status. Fourth, this study examined only the short-term (6-month) outcome of PC and its relationship with neurocognitive measures. We intend to examine whether the relationships reported here are sustained in regard to PC persisting to 12 and 24 months postinjury in the same pediatric TBI cohort.

Strengths of the study should also be appreciated. The cohort studied is a large sample of nonreferred consecutively hospitalized children and adolescents, with semistructured psychiatric interviews and standardized neuropsychological tests encompassing multiple domains of function. Data reported in this study extend previously published findings that focused on psychosocial and lesion correlates of PC in the same cohort. The investigation of the neuropsychology of PC is a unique aspect of this study.

Conclusions and Implications

Ultimately, the purpose of understanding the neural and psychological mechanisms inherent in children and adolescents with PC is to develop treatment strategies. In theory, stimulation of brain networks whose function is to regulate the expression of affect (e.g., dorsal frontal areas) or partial inhibition of networks responsible for generating affect (e.g., ventral frontal area and amygdala) could be helpful. Clinical trials of mood stabilizers are very difficult to accomplish, especially for children with PC, but these studies should be done. Cognitive rehabilitation targeting processing speed, attention, problem solving, and memory may enhance cognitive control (a form of executive function) over emotional expression and should be tested.³⁹ It is conceivable that more specific emotional probes such as those utilized in social cognition and affect recognition studies will further clarify important mechanisms in children with PC and possibly other disorders of affective dysregulation.^{34,36,40–42}