Mild Traumatic Brain Injury and Behavior and Sleep Among 9- and 10-Year Old Children: Initial Findings From the Adolescent Brain Cognitive Development (ABCD) Study

Abstract

There has been concern about the potential sequelae of mild traumatic brain injury (mTBI) in children. This study used data from the Adolescent Brain Cognitive Development^SM (ABCD) study to investigate associations between mTBI and behavior and sleep in school-aged children. Generalized additive mixed models were run to examine the association between TBI and parent-reported Child Behavior Checklist and Sleep Disturbance Scale for Children scores. mTBI with or without loss of consciousness (LOC) in 9- and 10-year old children was associated with 1) higher internalizing, externalizing and total problems and 2) greater sleep disturbance scores on the CBCL. The study also demonstrated a higher incidence of mTBI with and without LOC in boys compared to girls. This study shows a statistically significant but modest association between mTBI and behavioral and sleep changes, suggesting that in a non-clinical, sociodemographically diverse community sample of school-aged children mTBI does not result in clinically significant behavioral or psychological sequelae.

Introduction

Traumatic brain injury (TBI) in children is associated with long-term adverse events, creating a significant public health concern (Babikian et al., 2015). The leading causes of TBI vary by age but falls, motor-vehicle crashes, and sports- and recreation-related injuries are the primary mechanisms of injury in children (Coronado et al., 2015). In 2013, there were approximately 640,000 TBI-related emergency department (ED) visits, 18,000 TBI-related hospitalizations, and 1500 TBI-related deaths among children aged 14 years and younger (Taylor et al., 2017). In a cohort study reporting TBI severity in children under the age of 18 seeking emergency medical care from hospitals (N = 2940), 84.5% had mild TBI (mTBI), 13.2% had moderate TBI, and 2.3% had severe TBI, suggesting that mTBI accounts for the majority of TBI in children (Rivara et al., 2011).

Symptoms of mTBI can include headaches and dizziness, as well as problems with thinking, memory, physical activities, emotions and moods, and sleep (Hung et al., 2014; Mittenberg et al., 1997). Longitudinal studies suggest that most children with mTBI recover from the initial symptoms within 6 weeks after injury, with approximately 60% having persistent symptoms at 1 month post-injury, 10% at 3 months post-injury, and less than 5% at 1 year post-injury (Barlow et al., 2015; Crowe et al., 2016). Studies have noted increased hyperactivity and reading impairments in some children following mTBI (Babikian et al., 2015). Up to a third of children with mTBI develop behavioral or psychological symptoms that persist beyond the initial injury recovery period, such as poor conduct and problems with empathy and peer relationships (Rimel et al., 1981). However, there has been an ongoing debate regarding the pre-injury contribution of behavioral symptoms to post mTBI clinical presentation and limited data have been available in a large, diverse childhood sample (Emery et al., 2016).

More recent research examining social behavior in children after mTBI found difficulties in social outcomes, including problems with emotional perception, social skills, social problem-solving, and social language use (Ryan et al., 2014). Similar results have been observed in school-aged children in terms of reduced emotional control, elevated emotional and affective symptoms, and conduct problems following mTBI (Keenan et al., 2018). Overall, some studies investigating the sequelae of mTBI have identified postinjury emotional and behavioral problems in children of diverse ages, although it is unclear if the onset of the problems definitively followed the head injury. Further, a recent review conducted by Emery and colleagues suggests that there is mixed evidence regarding psychiatric, psychological, and behavioral outcomes of mTBI in children and adolescents (Emery et al., 2016). Overall, the review concluded that few rigorous prospective studies have examined psychological, behavioral, and psychiatric outcomes following mTBI and more empirical data were needed to understand the temporal relationship among these domains.

A growing body of evidence also suggests that TBI in childhood is associated with changes in sleep patterns. Common sleep problems identified in childhood TBI include insomnia (i.e., difficulties falling asleep and staying asleep) and excessive daytime sleepiness (EDS) (i.e., tendency to inadvertently fall asleep during the day) (Botchway et al., 2019). Studies using subjective sleep measures generally reported insomnia symptoms in mTBI with rates ranging between 33.5% and 79% (Kaufman et al., 2001; Theadom et al., 2016). For example, in a cross-sectional study of 8–12 year-olds, Milroy and colleagues reported that according to parent report, children with mTBI showed greater sleep problems than children in the orthopedic injury control group (Milroy et al., 2008). In another longitudinal study in 8–16 year-olds, prevalence of sleep problems in the mTBI group peaked at 1 month post-injury (39%), increased until 6 months (30%), and reduced slightly between 6 and 12 months (28%) post-injury (Theadom et al., 2016). Compared with healthy control, the mTBI group reported poorer sleep quality and sleep efficiency at 12 months post-injury (Theadom et al., 2016).

In the absence of true reports of pre-injury problems, there is little evidence to suggest that psychological, behavioral, and sleep problems persist beyond the acute (injury through 3 days post-injury) and subacute period (4 days through 3 months post-injury) following an mTBI in children and adolescents. The literature to date is insufficient for proving a causal link between mTBI and behavioral and sleep changes. Asking youth or their parents to provide retrospective recall of preinjury functioning is subject to recall and retrospective biases. Thus, more research is needed to determine if behavioral and sleep problems are the result of an mTBI, altered by an mTBI, risk factors for sustaining an mTBI, or preinjury predictors of outcome.

There are sex differences in the incidence of TBI as males are the majority of TBI cases across the age continuum and age-adjusted TBI-related hospitalization rates for males are consistently higher than females (Coronado et al., 2015). A number of factors have been identified that may contribute to the sex difference in epidemiology of TBI such as higher incidence of general injury among younger males, variance between males and females in traditional societal roles and activities, as well as differences in risk-taking behaviors. Results of a recent systematic review of 24 studies found that male workers working in the primary (e.g., agriculture, forestry, mining) or construction industries were more likely to sustain a work-related TBI (wrTBI) and that male workers were significantly more likely than female workers to be employed in the skilled agricultural and forestry occupational sectors at the time of injury, while female workers were more frequently occupied in the managerial, professional and associate professional-type roles. Thus, working in professions that are more prone to wrTBI could in part explain the higher incidence of TBI in males (Mollayeva et al., 2018). In a national, prospective, observational study in Ireland with 342 participants less than 17 years of age, it was observed that boys sustained injuries associated with a greater energy transfer, were less likely to use protective devices and more likely to be injured deliberately, which results in a different pattern of injury, higher levels of associated injury and a higher mortality rate (Collins et al., 2013). Risk factors for incurring mTBI also may differ according to sex: boys who have more difficulty with self-regulation, interaction and autonomy and girls who have worse adaptive functioning are more likely to incur a mTBI compared to their peers (Kaldoja & Kolk, 2015).

Emerging research suggests that long-term psychosocial and behavioral outcomes following TBI in childhood could be influenced by sex and/or gender. For example, a recent study evaluated sex differences in psychosocial outcomes in adulthood (assessed when they were between 18 and 31) following childhood TBI (injury when they were between 1 and 17 years). Researchers showed that females were significantly more likely than males to report internalizing problems such as depression and anxiety while males were more likely to report externalizing problems such as substance abuse or criminal behavior post-TBI (Scott et al., 2015). Another study in 210 children between the ages of 3 and 7 with TBI revealed that male sex (HR, 1.97; 95% CI, 1.03–3.77) was associated with an increased risk for Attention Deficit Hyperactivity Disorder (ADHD) secondary to the TBI, defined as T-scores higher than 65 on the DSM-Oriented Attention-Deficit/Hyperactivity Problems scale on the Child-Behavior Checklist (CBCL) (Narad et al., 2018). A recent study examining the association between reported concussion history and factors relating to cognitive, behavioral, and emotional health among a population-based sample of US high school–aged adolescents (ages 14–17) also found sex differences. Among male participants, concussions were associated with factors in the cognitive domain and factors within the behavioral domain related to substance use. Meanwhile, concussions among female participants were associated with factors in the emotional domain (suicidal thoughts and actions) and factors in the behavioral domain relating to decision making (Knell et al., 2020). Sex differences in sleep problems associated with mTBI has not been investigated extensively. In one of the few studies that explored this question, Tham and colleagues reported that controlling for depression, having a mTBI and male sex in an adolescent sample (ages between 12 and 18) were associated with poorer actigraphic sleep efficiency (Tham et al., 2015).

Despite the alarming prevalence of this neurological disorder, the research devoted to the investigation of TBI in children has only recently gained momentum. Furthermore, very few studies have focused on sex differences in overall outcomes. Interpretation of the research findings in this field is often affected by heterogeneous pathophysiology, severity criteria, and assessment instruments and practices as well as small sample sizes. For example, most studies solely focus on moderate or severe TBI or confound severity levels (Chapman et al., 2010), despite the fact that 70–90% of injuries are classified as mTBI (Cassidy et al., 2004).

The Adolescent Brain Cognitive Development^SM Study (ABCD Study^®)

(www.ABCDstudy.org) is a longitudinal, multi-site study that has recruited 9 and 10-year-old children with the goal of investigating a wide range of factors that influence developmental trajectories across childhood and adolescence. In addition to giving information about the prevalence and impact of TBI in youth, the ABCD study will be able to provide information on how such injuries affect mental and physical health outcomes (Volkow et al., 2018). The overall goal of the current analyses was to investigate the relationship between the incidence of head injury and behavioral/psychological and sleep impairment utilizing baseline data from participants in the ABCD study (data release 2.0.1). These data can be accessed from the National Institutes of Mental Health (NIMH) Data Archive (https://data-archive.nimh.nih.gov/abcd/query/abcd-annual-releases.html) after obtaining permissions.

Based on previously published literature, we hypothesized that children with a history of mTBI would exhibit more behavioral and psychological difficulties as assessed by the CBCL as well as sleep problems as assessed by the SDSC. Furthermore, the study also investigated sex differences in CBCL outcomes. Specifically, we hypothesized that girls may report more internalizing problems and boys may report more externalizing problems associated with mTBI.

Methods

Participants

The sample comprised of 11,875 participants from the ABCD study, a large-scale longitudinal study tracking 9–10-years-olds recruited from 21 research sites across the United States (Garavan et al., 2018). The cohort exhibits a large degree of socio-economic and demographic diversity, which makes this sample unique. Parents or caregivers provided written informed consent and permission and all children provided assent to participate in the study. All procedures were approved by a central institutional review board. The ABCD sample was largely recruited through public, private, and charter elementary schools. ABCD adopted a population neuroscience approach to recruitment (Paus, 2010) by employing epidemiologically informed procedures to ensure demographic variation in its sample that would mirror the variation in the US population of 9- and 10-year-olds (Garavan et al., 2018). Potential participants were excluded for the following reasons: child not fluent in English, MRI contraindication (e.g., irremovable ferromagnetic implants or dental appliances, claustrophobia, pregnant), major neurological disorder, gestational age less than 28 weeks or birthweight less than 1200 g, birth complications that resulted in hospitalization for more than 1 month, uncorrected vision, or current diagnosis of schizophrenia, autism spectrum disorder (moderate, severe), mental retardation/intellectual disability, or alcohol/substance use disorder. The ABCD sample included: non-twin siblings: 1600; Twins: 2100 (1050 pairs); 30 triplets (10 sets); 8150 singletons.

Measures

Modified Ohio State University TBI Screen-Short Version.

The ABCD study asks the parent or caregiver to report on the youth’s lifetime history of head injury using the Modified Ohio State University TBI Screen-Short Version (Bogner et al., 2017; Corrigan & Bogner, 2007). It queries parents/caregivers on whether their child had been to the emergency room due to an injury to the head or neck, whether the child had injured their neck in a fall or from being hit by something or from being in a fight or from a gunshot wound. A positive response to an occurrence question is followed up with questions to determine loss of consciousness (LOC), memory loss, and other details about the event. The age at which the injury occurred is also captured. A summary variable with the worst injury overall is generated with the following categorization: no TBI (responses to all head injuries are ‘no’ or response to at least one question about head injury is ‘yes’ but all responses to LOC and memory loss are “no”); mTBI without LOC (TBI without LOC but with memory loss); mTBI with LOC (TBI with LOC ≤ 30 minutes); moderate TBI (TBI with LOC between 30 minutes and 24 hours), severe TBI (TBI with LOC ≥ 24 hours). This modified form also includes indices such as the age at first head injury and periods with multiple or repeated injuries.

Child Behavior Checklist (CBCL).

The Child Behavior Checklist (Achenback, 1991) is a widely employed, 113-item standardized inventory used to assess the competencies and problems of children and adolescents aged 4–18 years. Parents or other caregivers are asked to complete a behavior problems inventory consisting of specific statements for which respondents are asked to choose “0: not true”, “1: somewhat or sometimes true”, or “2: very true or often true” with respect to the past 6 months. Eight syndrome scales are generated (i.e., Aggressive Behavior, Attention Problems, Delinquent Behavior, Social Problems, Somatic Complaints, Thought Problems, Anxious-Depressed, and Withdrawn). Scores on the Withdrawn, Anxious/Depressed, and Somatic Complaints syndromes contribute to the overall Internalizing summary score, whereas scores on the Aggressive and Delinquent syndromes comprise the Externalizing summary score. The Total Problems score, which is the sum of the scores of all the problem items, was also calculated.

Sleep Disturbance Scale for Children.

The Sleep Disturbance Scale for Children (SDSC) was used to assess sleep disturbance at the baseline visit. It was developed as a parent-report for children and adolescents between the ages of 6.5 and 15.3 and takes about 10 minutes to complete (Bruni et al., 1996). The SDSC consists of 26 questions assessing frequency of domains/subscales of disorders of initiating and maintaining sleep, sleep-disordered breathing disorders, disorders of arousal, sleep-wake transition disorders, disorders of excessive somnolence and sleep hyperhidrosis in the previous 6 months. The SDSC has demonstrated good test-retest validity and internal consistency despite the heterogeneity of the items. The following sleep-wake disturbances were assessed: insomnia (seven items, e.g., “the child has difficulty getting to sleep at night”, α = 0.73), sleep disordered breathing disturbances (three items, e.g., “the child has difficulty in breathing during the night”, α = 0.37), arousal disturbances (three items, e.g., “you have observed the child sleepwalking”, α = 0.56), sleep-wake transition disturbances (six items, e.g., “the child startles or jerks parts of the body while falling asleep”, α = 0.59), excessive sleepiness (five items, e.g., “the child experiences daytime sleepiness”, α = 0.72), and sleep hyperhidrosis (two items, e.g., “the child sweats excessively during the night”, α = 0.81). Responses subscale items were summed to generate a subscale score and responses to all items were summed to compute an overall sleep-wake disturbance symptom severity score (α = 0.65).

Statistical Analyses

The statistical analyses were performed in Data Exploration and Analysis Portal (DEAP), a software provided by the Data Analysis and Informatics Center of ABCD located at the UC San Diego with generous support from the National Institutes of Health and the Centers for Disease Control and Prevention under award number U24DA041123. The DEAP project information and links to its source code are available under the resource identifier RRID: SCR_016,158.

Generalized additive mixed models (GAMMs), using the R package GAMM4 (Simon Wood, Fabian Scheipl), which modelled family nested within site as random effects, were used to examine the statistical relationship between head injury and CBCL scores. Separate models were run for CBCL internalizing, externalizing, total problems composite, and overall sleep disturbance scores as dependent variables and TBI group as the independent variable. The following fixed effects covariates were used in all models: child’s age, caregiver’s report of the child’s sex at birth (2 levels: male or female), parental higher education (highest of any household member; five levels: less than High School Diploma; High School Diploma/GED; Some College; Bachelor; Post Graduate Degree), total combined family income (3 levels: < 50K, between 50K and 100 K, > 100 K) and parent marital status (2 levels: Yes/No). Effect size was calculated using the following formula: Cohen’s d = M1 – M2/pooled SD, whereas M1 = mean of group 1 and M2 = mean of group 2, pooled SD = pooled standard deviation. The Bonferroni correction was used to correct for multiple comparisons and the critical value was set to p < 0.05/10 = 0.005.

Results

Sample characteristics at baseline

The current analyses included 10,544 participants with complete data on head injury (missing = 9), CBCL scale (missing = 11), sleep disturbance (missing = 40) and relevant covariates: age (missing = 0), sex (missing = 1), parent education (missing = 345), household income (missing = 874), and parent marital status (missing = 40). To examine whether missing data associated with socio-demographic factors may bias results, study covariates were used in binary logistic regressions predicting missingness for dependent variables. No study covariate significantly predicted probability of missing data. There were 10,142 participants who had no history of TBI, 285 with a mTBI without LOC and 117 with a mTBI with LOC in their lifetime. There were also four participants with a lifetime history of moderate TBI and three participants with a lifetime history of severe TBI. Since the sample sizes for moderate and severe TBI were so low, these participants were excluded from analyses. Low rates of moderate and severe TBI were expected given that one of the exclusionary criteria for the study was having sustained a head injury that led to loss of consciousness for more than 30 minutes and/or memory impairment for more than 24 hours. Only a small portion of the ABCD sample endorsed having periods of multiple or repeated head injuries (n = 107), making it unfeasible to examine correlates of multiple injuries. Out of the 107 participants, 103 reported no LOC and 95 reported no impairment in memory as a result of those injuries. The average age of first injury was 5.6 years (standard deviation = 2.9, standard error of mean = 0.3). For each injury question, the breakdown was as follows a) child ever been hospitalized or treated in an ER following an injury to head or neck (mean = 4.98 years, 56.7% sustained injury at or before age 5) b) child injured head or neck in a car accident or from crashing some moving vehicle (mean = 6.61 years, 26.5% sustained injury at or before age 5) c) child injured head or neck in a fall or from being hit by something (mean = 5.86 years, 41.2% sustained injury at or before age 5) d) child injured head or neck in a fight, from being hit by something, or shaken violently or shot in the head (mean = 6.53 years, 35.5% sustained injury at or before 5 years). Overall, majority of injuries were sustained after 5 years of age according to parent report.

At baseline, participants had a mean age of 9.9 years. The sample consisted of an almost equal distribution of the sexes (48% female) and a majority of non-Hispanic participants (81.1%). Moreover, 66.2% of the participants were white, and 62.3% of the sample had parental education status of a bachelor’s/professional degree or higher. On average, participants’ scores on the internalizing, externalizing, and total problem CBCL scales as well as the total sleep disturbance scale scores were in the normal range. Means and standard deviations for general demographics of the sample can be found in Table 1. Table 2 demonstrates the mean and standard deviations of the demographic variables divided by group. There were no significant differences between groups with regard to sociodemographic variables except sex. Males had a higher incidence of mTBI without LOC (61.1% vs. 38.9%) and mTBI with LOC (60.7% vs. 39.3%) than females after controlling for other socio-demographic variables (Table 2, p = 0.01).

Table 1.

Demographics, Child Behavior Checklist (CBCL) and Sleep Disturbance Scores for The Baseline ABCD Sample.

Measure	Frequency		Percent
Sex
Female	5058		48.0
Male	5486		52.0
Ethnicity
Hispanic	1990		18.9
Not-Hispanic	8554		81.1
Race
White	6983		66.2
Black	1541		14.6
Asian	235		2.2
Other/Mixed	1785		16.9
Parental education
Less than high school	393		3.7
High school/GED	877		8.3
Some college	2704		25.6
Graduate/professional degree	2797		26.5
Postgraduate degree	3773		35.8
Household Income
<50 K	3042		28.9
≥50 K & <100 K	3009		28.5
>100 K	4493		42.6
Measure	Mean	Range	SD
Age (months)	118.96	108–131	7.5
Internalizing (CBCL t-score)	48.4	33–93	10.6
Externalizing (CBCL t-score)	45.7	33–84	10.3
Total problems (CBCL t-score)	45.7	24–83	11.2
Total sleep disturbance score	36.4	26–75	7.7

Table 2.

Demographic characteristics of youth with no TBI compared to youth with mTBI without loss of consciousness (LOC) and mTBI with LOC.

	Level	No TBI	mTBI without LOC	mTBI with LOC
Age (mean (SD))		118.9 (7.5)	120.1 (7.3)	119.5 (7.9)
Sex (%)	F	4901 (48.3)	111 (38.9)	46 (39.3)
	M	5241 (51.7)	174 (61.1)	71 (60.7)
Race (%)	White	6689 (66.0)	211 (74.0)	83 (70.9)
	Black	1504 (14.8)	27 (9.5)	10 (8.5)
	Asian	229 (2.3)	5 (1.8)	1 (0.9)
	Other/Mixed	1720 (17.0)	42 (14.7)	23 (19.7)
Highest education	<HS Diploma	382 (3.8)	7 (2.5)	4 (3.4)
	HS Diploma/GED	857 (8.5)	14 (4.9)	6 (5.1)
	Some College	2619 (25.8)	58 (20.4)	27 (23.1)
	Bachelor	2679 (26.4)	81 (28.4)	37 (31.6)
	Post-graduate	3605 (35.5)	125 (43.9)	43 (36.8)
Hispanic (%)	No	8213 (81.0)	251 (88.1)	90 (76.9)
	Yes	1929 (19.0)	34 (11.9)	27 (23.1)
Household income (%)	<50 K	2955 (29.1)	57 (20.0)	30 (25.6)
	≥50K & <100 K	2896 (28.6)	84 (29.5)	29 (24.8)
	>100 K	4291 (42.3)	144 (50.5)	58 (49.6)
Married (%)	No	3092 (30.5)	81 (28.4)	37 (31.6)
	Yes	7050 (69.5)	204 (71.6)	80 (68.4)

Association between TBI and CBCL scores

After adjusting for sociodemographic factors (age, sex, parent education, household income, and parent marital status), site, and twin/sibling status, there was a statistically significant association between mTBI and CBCL scores such that participants with a lifetime history of mTBI without LOC and mTBI with LOC had higher internalizing, externalizing and total problems CBCL scores compared to youth without TBI. Table 3 demonstrates the model statistics for the association between mTBI and CBCL scores. Detailed parameter tables with estimates, Wald statistics (t-value) used to test the significance of the variable in the regression model, and used to compute the p-values assuming asymptotically normal coefficient estimates can be found in Table 4. Effect sizes for the GAMM model used to estimate the relationship between TBI and CBCL scores were modest (internalizing: Cohen’s d = 0.39, % variance = 3.6%; externalizing: Cohen’s d = 0.31, % variance = 2.3%; total problems: Cohen’s d = 0.40, % variance = 3.8%)

Table 3.

CBCL scores in youth with no TBI compared to youth with mTBI without LOC and mTBI with LOC. Overall as well as CBCL scores divided by sex are shown. Values are shown as Mean (S.D.).

	No TBI	mTBI without LOC	mTBI with LOC	F-Value	p-value
Internalizing
Overall	48.3 (10.5)	52.1 (10.8)	52.8 (12.4)	19.3	<1e-6***
Males	49.2 (10.5)	52.0 (11.1)	53.6 (12.4)
Females	47.4 (10.4)	52.3 (10.5)	51.5 (12.3)
Externalizing
Overall	45.5 (10.2)	48.3 (11.1)	49.0 (12.1)	15.7	<1e-6***
Males	46.3 (10.6)	48.4 (11.3)	49.8 (13.0)
Females	44.7 (9.8)	48.2 (10.8)	47.9 (10.5)
Total problems
Overall	45.7 (11.1)	49.8 (11.5)	50.6 (13.2)	22.4	<1e-6***
Males	46.7 (11.4)	49.9 (11.9)	51.6 (13.9)
Females	44.6 (10.8)	49.7 (11.0)	49.2 (12.1)

Table 4.

Parameter table showing model statistics for relationship between TBI and CBCL scores. Reference category is no TBI.

		Beta co-efficient	Std.Error	t wald’s	p-value
Internalizing	mTBI without LOC	2.84	0.6	4.74	2.2e-06
	mTBI with LOC	3.8	0.9	4.1	4.1e-05
Externalizing	mTBI without LOC	2.6	0.58	4.39	1.1e-05
	mTBI with LOC	3.2	0.90	3.57	0.00036
Total problems	mTBI without LOC	3.2	0.61	5.17	<1e-06
	mTBI with LOC	4.1	0.95	4.35	1.4e-05

Association between TBI and sleep disturbance scores

After adjusting for sociodemographic factors (age, sex, parent education, household income, and parent marital status), site, and twin/sibling status, there was a statistically significant association between mTBI and sleep disturbance scores such that participants with a lifetime history of mTBI without LOC and mTBI with LOC had higher total sleep disturbance scores compared to youth without TBI. Specifically, a lifetime history of mTBI without LOC and mTBI with LOC was associated with increased scores of disorders of initiating and maintaining sleep, excessive somnolence, and sleep wake transition disorders. Table 5 demonstrates the model statistics for the association between mTBI and total sleep disturbance scores as well as specific sleep-wake disturbance scores. Detailed parameter tables with estimates, Wald statistics (t-value) used to test the significance of the variable in the regression model, and used to compute the p-values assuming asymptotically normal coefficient estimates can be found in Table 6. Effect sizes for the GAMM model used to estimate the relationship between TBI and sleep disturbance scores were modest (Disorders of initiating and maintaining sleep: Cohen’s d = 0.26, % variance = 1.7%; disorders of excessive somnolence: Cohen’s d = 0.28, % variance = 1.9%; sleep wake transition disorder: Cohen’s d = 0.55, % variance = 7%; total sleep disturbance score: Cohen’s d = 0.43, % variance = 4.4%)

Table 5.

Sleep disturbance scores in youth with no TBI compared to youth with mTBI without LOC and mTBI with LOC. Overall as well as sleep disturbance scores divided by sex are shown. Values are shown as Mean (S.D.).

	No TBI	mTBI without LOC	mTBI with LOC	F-Value	p-value
Disorder of initiating and maintaining sleep
Overall	11.7 (3.7)	12.7 (4.4)	12.6 (3.9)	10.9	1.8e-05***
Males	11.8 (3.8)	12.4 (4.7)	12.7 (4.1)
Females	11.7 (3.6)	13.1 (4.4)	12.3 (3.7)
Disorders of excessive somnolence
Overall	6.9 (2.4)	7.6 (2.9)	7.6 (2.7)	14.1	<1e-6***
Males	6.9 (2.4)	7.4 (2.7)	7.6 (2.8)
Females	7.0 (2.4	7.9 (3.2)	7.4 (2.6)
Disorders of arousal
Overall	3.4 (0.9)	3.6 (1.7)	3.6 (0.8)	5.0	0.005
Males	3.4 (0.9)	3.7 (1.3)	3.5 (0.7)
Females	3.4 (0.9)	3.6 (1.2)	3.7 (1.0)
Sleep breathing disorders
Overall	3.7 (1.2)	3.7 (1.1)	3.9 (1.1)	0.8	NS
Males	3.8 (1.2)	3.8 (1.3)	3.9 (1.1)
Females	3.7 (1.2)	3.6 (0.9)	3.8 (1.1)
Sleep hyperhydrosis disorder
Overall	2.4 (1.2)	2.6 (1.4)	2.5 (1.1)	2.6	NS
Males	2.5 (1.3)	2.7 (1.7)	2.6 (1.2)
Females	2.3 (0.9)	2.3 (0.9)	2.2 (0.8)
Sleep wake transition disorder
Overall	8.1 (2.6)	8.9 (2.9)	9.6 (3.5)	18.0	<1e-6***
Males	8.3 (2.6)	8.9 (2.7)	9.8 (3.6)
Females	8.0 (2.5)	8.8 (3.1)	9.2 (3.3)
Total sleep disturbance score
Overall	36.3 (7.6)	38.9 (9.4)	39.6 (8.6)	19.3	<1e-6***
Males	36.7 (8.2)	39.1 (10.1)	40.2 (8.6)
Females	36.1 (7.8)	39.4 (10.1)	38.6 (8.6)

Table 6.

Parameter table showing model statistics for relationship between TBI and sleep disturbance scores. Reference category is no TBI.

		Beta co-efficient	Std.Error	t wald’s	p Value
Disorders of initiating and maintain sleep	mTBI without LOC	0.89	0.22	4.1	4.86e-05
	mTBI with LOC	0.80	0.33	2.4	0.017
Disorders of excessive somnolence	mTBI without LOC	0.63	0.14	4.5	8.2e-06
Disorders of excessive somnolence	mTBI with LOC	0.65	0.22	3.0	0.0029
Sleep wake transition disorder	mTBI without LOC	0.53	0.15	3.5	0.00045
	mTBI with LOC	1.14	0.23	4.9	<1e-6
Total sleep disturbance score	mTBI without LOC	2.04	0.44	4.7	2.9e-06
	mTBI with LOC	2.8	0.67	4.2	2.9e-05

Sex differences in associations between TBI and clinical scales

Given prior reports of sex differences in CBCL outcomes associated with TBI, we investigated whether sex interacted with TBI diagnosis to predict CBCL outcomes. There was no significant interaction between sex and diagnosis suggesting that the higher CBCL scores associated with TBI were independent of sex (internalizing: F = 1.6, p = 0.21; externalizing: F = 0,11, p = 0.89; total: F = 0.48, p = 0.62). However, there was a main effect of sex with males demonstrating higher internalizing (49.3 vs. 47.4), externalizing (46.5 vs. 44.9) and total problem scores (46.9 vs. 44.7) as compared to females independent of TBI. As part of an exploratory analysis, we also investigated whether sex interacted with TBI group to predict sleep pattern changes. There was no significant interaction between sex and TBI suggesting that the higher total sleep disturbance score as well as the specific sleep wake disturbance scores in the mTBI groups were independent of sex (Total sleep disturbance: F = 0.6, p = 0.55; disorders of initiating and maintaining sleep: F = 1.7, p = 0.17; disorders of excessive somnolence: F = 1.0, p = 0.35; sleep-wake transition disorders: F = 0.25, p = 0.77). Overall, males demonstrated higher scores for sleep hyperhydrosis (2.5 vs. 2.3) and sleep-wake transition disorders (8.3 vs. 8.1) as compared to females. Means and standard deviations according to sex as well as the model statistics for sex as the independent variable and CBCL or the SDSC scores as the dependent variables (with sociodemographic factors as covariates) are provided in Tables 7 and 8 respectively.

Table 7.

CBCL and sleep disturbance scores by sex. Values are shown as Mean (S.D.). Scales with significant association with sex are depicted.

	Males	Females	F-Value	p-value
Internalizing	49.3 (10.7)	47.4 (10.5)	68.3	<1e-6***
Externalizing	46.5 (10.7)	44.9 (9.9)	56.0	<1e-6***
Total problems	46.9 (11.6)	44.7 (10.9)	88.3	<1e-6***
Sleep hyperhydrosis disorder	2.5 (1.3)	2.3 (0.9)	120.1	<1e-6***
Sleep wake transition disorder	8.3 (2.7)	8.1 (2.5)	18.1	1.39e-05***

Table 8.

Parameter table showing model statistics for relationship between sex and CBCL and sleep disturbance scores. Reference category is females.

		Estimate	Std.Error	t wald’s	Pr>(|t|)
Internalizing	Sex:M	1.67	0.2	8.3	<1e-6
Externalizing	Sex:M	1.46	0.19	7.5	<1e-6
Total problems	Sex:M	1.97	0.21	9.4	<1e-6
Sleep hyperhydrosis disorder	Sex:M	0.25	0.022	11.0	<1e-6
Sleep wake transition disorder	Sex:M	0.22	0.051	4.4	1.39e-05

Discussion

The goal of this study was to examine the association of a history of mTBI and internalizing and externalizing problem behaviors as well as sleep changes in a sample of 9- and 10-year old children. Drawing on a large sample of parents/caregivers and 9- and 10-year old children approximating the demographic diversity of the US population, this study showed that a history of mTBI with LOC or mTBI without LOC in children was associated with higher scores on the internalizing, externalizing, total problem CBCL and sleep disturbance scales after adjusting for key sociodemographic factors, site, and twin/sibling status.

Increasing societal awareness of the risk for TBI during childhood has made this a specific and pressing question, not only for researchers but also for parents, educators, and coaches. The effects of mTBI in childhood differ from those of mTBI in adulthood owing to the differences in brain biomechanisms and the neurodevelopmental changes (Daneshvar et al., 2011). The developing brain undergoes many changes in the first few years of life characterized by functional networks becoming more distributed and developing longer-range connections, increased synaptic pruning and increased myelination (Fair et al., 2009; Gilmore et al., 2018; Huttenlocher & Dabholkar, 1997). Disruption to any of these neurobiological processes can alter the development of structural and functional connections that are necessary for emotional and cognitive control. Injuries of any severity to the developing brain can negatively impact children’s behavior and cognitive skills as they grow, placing them at risk for significant changes to their developmental trajectory across multiple domains. In fact, previous research has suggested that children with mild head injuries and more severely injured children were equally likely to demonstrate post-injury psychosocial challenges such as delinquency and attention problems (Hayman-Abello et al., 2003).

The findings from this study add to a growing body of literature that has found evidence of behavioral, affective and sleep problems associated with childhood TBI. For example, a large prospective cohort study found increased psychosocial difficulties in a sample of children and adolescents between 2.5 and 15 years of age with TBI as compared to children with orthopedic injury and that these difficulties persist for at least 12 months (Keenan et al., 2018). Children of all TBI severities, including mild had higher scores on the affective and anxiety CBCL subscales (Keenan et al., 2018). In a large sample of children and adolescents from the Project of Human Development in Chicago Neighborhoods, results showed that mTBI may act as a predictor of general psychopathology across the internalizing and externalizing dimensions (Connolly & McCormick, 2019; McCormick et al., 2021). A longitudinal study conducted in Australia in children and adolescents between the ages of 5 and 18 recruited from a pediatric ED within 48 hours of sustaining a concussion found preliminary evidence of greater behavioral disturbances, measured using the total problems CBCL score 2 weeks after the concussion (compared to those who sustained an orthopedic injury), which were likely to improve considerably over time. (Gornall et al., 2020). In addition, there are a number of cross sectional and longitudinal studies demonstrating an association between childhood TBI and disturbances in sleep using both objective and subjective sleep measures (Botchway et al., 2019). A systematic review of 16 studies investigating the prevalence and type of sleep disturbances in childhood TBI showed that some studies reported an increase in sleep disturbances between 1 and 3 and 6 months postinjury (Schmidt et al., 2015; Theadom et al., 2016), while others reported persistent sleep disturbances at 12 months (Tham et al., 2015), 3 years (Kaufman et al., 2001), and 6 years (Pillar et al., 2003) following mTBI in childhood. Studies have attributed sleep disturbances in childhood TBI to injuries involving sleep and wake-contingent brain areas such as the hypothalamus and brainstem (Baumann et al., 2005; Jang & Kwon, 2016) as well as injury–related psychological (e.g., anxiety and stress) and clinical (e.g., pain, headache, and medication) factors (Pillar et al., 2003). Given the cross-sectional nature of the current study, it is not possible to establish the direction of causality. Thus, more prospective longitudinal studies are needed to disentangle the problem of causality.

Research on childhood TBI has reported limited findings on sex differences related to outcome variables. To our knowledge, the present study is among the first to examine not only mTBI and clinical sequelae but also the interaction of sex by mTBI in a large sample of youth. As such, this study also investigated the potential role of sex in moderating the effect of TBI on CBCL and SDSC scores. Although males were more likely to report having sustained a TBI, the effect of mTBI on CBCL and SDSC scores was independent of sex. Interestingly, prior literature on the role of sex in TBI outcomes has been mixed. For example, sex was not shown to be a prognostic indicator for postconcussive symptoms (PCS) after mild TBI in one study (Cancelliere et al., 2016); however, a population-based Swedish cohort study found sex differences in PCS with a higher percentage of females showing PCS than males after mild TBI (Styrke et al., 2013). Furthermore, a longitudinal study reported that as age increased, greater disparity in internalizing and total behavior problems was observed between males and females in concussed participants with males’ total and internalizing problems decreasing as age increased and rates of internalizing and total behavior problems increasing with age in females (Gornall et al., 2020). A more recent study reported sex differences in the relative risk of developing internalizing and externalizing behaviors after childhood TBI, which was obtained independently of the effect of TBI (Scott et al., 2015). Females were more likely to report internalizing problems and males were more likely to report externalizing problems (Scott et al., 2015). In contrast, our results show that males reported higher externalizing, internalizing and total problems on the CBCL independent of TBI. An important difference between the two studies is the age range of the study sample. Scott and colleagues reported on a sample that was between 18 and 30 years of age who had sustained a TBI during their childhood while the current study examined findings collected in 9- and 10-year old children. Finally, we observed a sex difference in the incidence of mTBI with a higher percentage of males sustaining mTBI (with or without LOC). This finding is in agreement with previously published literature that has shown majority of TBI cases occurring in males across the age continuum (Coronado et al., 2015).

Although mTBI was statistically associated with increased scores on the CBCL and SDSC, the mean scores between groups differed in general by only a few points (e.g., internalizing behaviors subscale no TBI = 48.3, mTBI without LOC = 52.1, mTBI with LOC = 52.8). These scores on average were not in the borderline or clinically significant range, suggesting that the post-mTBI behavioral and sleep perturbances in this age group are modest. This finding is a positive one, as it may be indicative of resilience at this age in terms of mTBI not necessarily being related to significant behavioral or sleep impairment. This point is particularly important given the overall prevalence of mTBI in childhood, especially in comparison to more severe TBI. If mTBI is found to have minimal effects on internalizing and externalizing behavior, in addition to sleep, other factors may be related to these challenges, which may require a different therapeutic approach. Relative resilience post-mTBI in youth may be due to neuroplasticity and synaptic pruning protecting children from developing severe behavioral and sleep complications subsequent to mTBI. ABCD is a longitudinal study that will last a minimum of 10 years post-baseline; as such, it will be interesting to investigate whether these relatively minor behavioral and sleep changes in the mTBI group put them at a higher risk of developing psychopathology and more serious problems during adolescence or whether statistical differences between the TBI-no TBI groups decrease over time. Further, the lack of association of mTBI with disorders of arousal, sleep breathing disorders, and sleep hyperhydrosis disorder may suggest that neural circuits and mechanisms underlying these disorders are relatively spared by mTBI or that developmental processes cause faster recovery in these circuits. However, it is also possible that because the ABCD cohort was a relatively healthy community sample at baseline, we did not have enough range with regard to the scores in these subdomains to detect significant differences.

Strengths and limitations

ABCD is the largest long-term study of cognitive brain development and child physical and mental health in the US. The large number of participants, sociodemographic diversity of the youth and their families, and national representativeness of the sample provide significant generalizability to these findings. Moreover, given the recognition that traumatic brain injury is a chronic health condition with a dynamic presentation over time the analyses presented here can be built upon in subsequent years to examine whether changes in internalizing, externalizing, and sleep behaviors are altered over time (Corrigan & Bogner, 2007). In addition, the CBCL, the SDSC, and modified OSU-TBI are well-established standardized assessments that have demonstrated good validity and reliability. Further, the availability of many important potential covariates strengthens the findings from the analysis. An additional strength of this work is the fact that youth behavior and symptoms were ascertained from parent report which included not only a history of mTBI occurrence but also collected information about loss of consciousness (Cook et al., 2022). Reliance on parent report is a common strategy in pediatric research. However, the ABCD study is planning to collect electronic health records (EHR) in the near future.

While the study has a number of strengths, it is not without limitations. One limitation of cross-sectional research is inference regarding the direction of the relationship between TBI and CBCL and SDSC scores cannot be drawn. The ABCD study is a longitudinal study and as future waves of data become available, it will be important to investigate how the relationship between TBI and behavior and sleep by sex evolve over time especially after the onset of puberty. Secondly, the magnitude of the effect size is modest with significance in part driven by the large sample size. Nevertheless, the presence of the predicted relationship in a large community sample in contrast to a clinical sample is noteworthy. It has been suggested that small effect sizes from large sample size studies are the most likely to reflect the true state of nature (Funder & Ozer, 2019). Specifically, it was proposed that an effect-size r of 0.10 indicates an effect that is still small at the level of single events but potentially more ultimately consequential and an effect-size r of 0.20 indicates a medium effect that is of some explanatory and practical use even in the short run and therefore even more important (Funder & Ozer, 2019). Finally, although SDSC is well-validated and frequently used in research to measure sleep problems (Bruni et al., 2021; Moo-Estrella et al., 2021; Romeo et al., 2021), measures such as polysomnography and actigraphy are gold standards and provide more objective measures of sleep. Given the scale and the large sample size of the ABCD study, conducting full objective, overnight measures of sleep such as polysomnography has significant challenges both in terms of cost and logistics.

Conclusions and Future Directions

Overall, our study suggests that youth with mTBI do not on average display clinically significant behavioral and psychological sequelae. Findings did supports three areas of interest: 1) A history of a mTBI without LOC or mTBI with LOC in 9- and 10-year old children is associated with a modest higher internalizing, externalizing, total problems score as assessed by the CBCL as well as higher sleep disturbance scores on the SDSC; 2) Higher percentage of males had mTBI with and without LOC than females; 3) Males demonstrated higher scores on the internalizing, externalizing and total problem scales independent of TBI as well as sleep hyperhydrosis and sleep wake transition disorders. Although previous studies have explored associations of behavioral and sleep variables with mTBI in youth, they have not collected data on these measures simultaneously in a general population sample of youth during middle childhood rather than in a clinical sample.

The results from the current study have important implications. The current findings do not suggest that behavioral disorders indexed by the CBCL are the consequence of an mTBI, as few participants displayed CBCL scores in the “abnormal range.” Rather, our findings suggest that sleep and behavior should be closely monitored in children after mTBI and psychosocial interventions targeting sleep and behavior can be employed. Further, it may be worthwhile to include assessment of sleep and behavior while evaluating whether the youth is ready to return to school and/or sports.

Future longitudinal studies will aim to clarify the relationship between mTBI and sleep problems and behavior. For example, there is evidence that daytime sleepiness and poor sleep hygiene as well as an ADHD diagnosis are risk factors for sustaining a mTBI. Poor sleep and conduct problems are also commonly reported post-concussive symptoms. Thus, disentangling the causality between these variables will be a crucial aim for future studies. As participants of the ABCD study age, they may have a high probability of sustaining mTBIs (especially sports-related concussions). Future research should track the impact of additional concussions on developmental trajectories since it may provide important implications for policy and intervention given the prevalence of sports-related concussion during adolescence.

Biographies

Author Biographies

Chandni Sheth, PhD, is a Research Assistant Professor in the Department of Psychiatry at the University of Utah. She obtained her PhD in Pharmacology and Toxicology from the University of Utah in 2016. With an undergraduate degree in pharmacy and graduate training in the neurobiology of alcohol abuse, she has a broad background in the neuropharmacology of psychiatric illnesses as well as traumatic brain injury and chronic pain. Her interests include investigating neurobiological alterations underlying psychopathology using multimodal neuroimaging and using this knowledge towards finding better therapeutic targets for intervention.

Rebekah Huber, PhD is a Research Assistant Professor in the Department of Psychiatry at the University of Utah School of Medicine. Her research focuses on understanding risk correlates of psychiatric disorders and suicide, especially in youth.

Perry Renshaw, MD, PhD, MBA, completed his PhD in Biophysics at the University of Pennsylvania. His training as a biophysicist and psychiatrist has led to a primary research interest in the use of multinuclear magnetic resonance spectroscopy (MRS) neuroimaging to identify changes in brain chemistry associated with psychiatric disorders and substance abuse.

Deborah Yurgelun-Todd, PhD, Professor and Vice Chair of Research in the Department of Psychiatry at the University of Utah, completed her doctorate in Psychology from Harvard University and subsequently trained at McLean Hospital and Harvard Medical School. Her training in neuroscience and psychopathology has led to the application of multimodal imaging approaches aimed at identifying neurobiological bases of major psychiatric disorders.

Erin McGlade, PhD, is an Associate Professor in the department of Psychiatry at the University of Utah whose research focuses on brain differences in females compared to males. She seeks to better understand sex differences in psychiatric symptoms through the use of clinical interviews, neuro-psychological assessments, standardized symptom measures, and neuroimaging. Her primary work is on sex differences in brain development of adolescents and young adults, suicide risk in adolescents, and suicide prevention in female veterans.

Data Availability Statement

The ABCD data repository grows and changes over time. The ABCD data used in this report came from NIMH Data Archive Digital Object Identifier (DOI) 10.15154/1518688. DOIs can be found at https://dx.doi.org/10.15154/1518688.