Association of Attention-Deficit/Hyperactivity Disorder Diagnoses With Pediatric Traumatic Brain Injury: A Meta-analysis

Key Points

Question  Are traumatic brain injuries associated with increased risk for attention-deficit/hyperactivity disorder?

Findings  This meta-analysis of 24 studies including 12 374 children with traumatic brain injuries found that severe traumatic brain injuries appear to be associated with an increased risk for attention-deficit/hyperactivity disorder compared with noninjured and other injured controls; no association between attention-deficit/hyperactivity disorder and concussions and mild or moderate traumatic brain injury was identified. The rate of pretraumatic brain injury attention-deficit/hyperactivity disorder diagnoses was significantly greater than the incidence of attention-deficit/hyperactivity disorder in the general pediatric population.

Meaning  These findings show that clinicians should carefully review psychosocial and medical issues that antedate a traumatic brain injury that may need to be addressed to adequately treat attention-deficit/hyperactivity disorder symptoms in children with traumatic brain injuries.

Importance  There are conflicting accounts about the risk for attention-deficit/hyperactivity disorder (ADHD) following traumatic brain injury (TBI), possibly owing to variations between studies in acute TBI severity or when ADHD was assessed postinjury. Analysis of these variations may aid in identifying the risk.

Objective  To conduct a meta-analysis of studies assessing ADHD diagnoses in children between ages 4 and 18 years following concussions and mild, moderate, or severe TBI.

Data Sources  PubMed, PsycInfo, and Cochrane Central Register of Controlled Trials (1981-December 19, 2019) were searched including the terms traumatic brain injury, brain injuries, closed head injury, blunt head trauma, concussion, attention deficit disorders, ADHD, and ADD in combination with childhood, adolescence, pediatric, infant, child, young adult, or teen.

Study Selection  Limited to English-language publications in peer-reviewed journals and patient age (4-18 years). Differences about inclusion were resolved through consensus of 3 authors.

Data Extraction and Synthesis  MOOSE guidelines for abstracting and assessing data quality and validity were used. Odds ratios with 95% credible intervals (CrIs) are reported.

Main Outcomes and Measures  The planned study outcome was rate of ADHD diagnoses.

Results  A total of 12 374 unique patients with TBI of all severity levels and 43 491 unique controls were included in the 24 studies in this review (predominantly male: TBI, 61.8%; noninjury control, 60.9%; other injury control, 66.1%). The rate of pre-TBI ADHD diagnoses was 16.0% (95% CrI, 11.3%-21.7%), which was significantly greater than the 10.8% (95% CrI, 10.2%-11.4%) incidence of ADHD in the general pediatric population. Compared with children without injuries, the odds for ADHD were not significantly increased following concussion (≤1 year: OR, 0.32; 95% CrI, 0.05-1.13), mild TBI (≤1 year: OR, 0.56; 0.16-1.43; >1 year: OR, 1.07; 95% CrI, 0.35-2.48), and moderate TBI (≤1 year: OR, 1.28; 95% CrI, 0.35-3.34; >1 year: OR, 3.67; 95% CrI, 0.83-10.56). The odds for ADHD also were not significantly increased compared with children with other injuries following mild TBI (≤1 year: OR, 1.07; 95% CrI, 0.33-2.47; >1 year: OR, 1.18; 95% CrI, 0.32-3.12) and moderate TBI (≤1 year: OR, 2.34; 95% CrI, 0.78-5.47; >1 year: OR, 3.78; 95% CrI, 0.93-10.33). In contrast, the odds for ADHD following severe TBI were increased at both time points following TBI compared with children with other injuries (≤1 year: OR, 4.81; 95% CrI, 1.66-11.03; >1 year: OR, 6.70; 95% CrI, 2.02-16.82) and noninjured controls (≤1 year: OR, 2.62; 95% CrI, 0.76-6.64; >1 year: OR, 6.25; 95% CrI, 2.06-15.06), as well as those with mild TBI (≤1 year OR, 5.69; 1.46-15.67: >1 year OR, 6.65; 2.14-16.44). Of 5920 children with severe TBI, 35.5% (95% CrI, 20.6%-53.2%) had ADHD more than 1 year postinjury.

Conclusions and Relevance  This study noted a significant association between TBI severity and ADHD diagnosis. In children with severe but not mild and moderate TBI, there was an association with an increase in risk for ADHD. The high rate of preinjury ADHD in children with TBI suggests that clinicians should carefully review functioning before a TBI before initiating treatment.

Traumatic brain injury (TBI) is the most common cause of death and acquired neurologic disability in children and youths in the industrialized world.^1,2 More than 600 000 children visit US emergency departments yearly for TBIs.³ Survivors of pediatric TBI often develop impaired cognitive, academic, and behavioral functioning.^4-6 Pediatric TBI can also result in serious psychiatric problems that adversely affect psychosocial development, family and peer relationships, and academic achievement.^7,8

Many clinicians believe and some studies⁹ report that attention-deficit/hyperactivity disorder (ADHD) occurs frequently following pediatric TBI. However, other studies did not report increased rates of ADHD following TBI. The conflicting results may be due to variations between studies in 2 key moderator variables: TBI injury severity and the timing of assessment relative to the injury. In this meta-analysis, we assessed the association of severity and assessment timing with ADHD diagnoses after TBI. Because ADHD occurs frequently in the general pediatric population, a key issue is the extent to which the risk for ADHD increases after TBI.

One meta-analysis of ADHD diagnoses in children and adults with mild TBI included 5 studies,¹⁰ with the most recent study published in 2009; the authors noted an association between mild TBI and ADHD. The meta-analysis reported herein included 24 studies, the most recent published in 2020, of children with concussions and mild, moderate, and severe TBI that provide quantitative estimates of the association of acute brain injury severity with the risk for ADHD following TBI. Given the clinical significance of ADHD in pediatric practices and parental concerns about the serious effects of mild TBI/concussion stimulated by media reports, information about the risk for ADHD following TBI may be useful in managing children with TBI and counseling their parents.

To our knowledge, this is the first meta-analytic review that integrates findings across study designs (eg, cross-sectional and longitudinal) to examine the association of injury severity and time since injury with the development of ADHD diagnoses following pediatric TBI. We estimate the magnitude of the association of injury severity at 2 time intervals following TBI.

This search was developed with the Health Sciences librarian at the University of California, Los Angeles. We searched PubMed, PsycInfo, and Cochrane Central Register of Controlled Trials (1981-December 19, 2020). Key terms were traumatic brain injury, concussion, brain injuries, closed head injury, blunt head trauma, attention deficit disorders, ADHD, or ADD and were searched in combination with the key words childhood, adolescence, pediatric, infant, child, young adult, or teen. We reviewed reference lists, articles that cited included studies, and review articles to identify studies that may have been missed in database searches. The search was limited to English-language publications in peer-reviewed journals and patient age between 4 and 18 years. This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.

Titles and abstracts of all identified articles were independently reviewed by ³ of us (R.F.A., N.N., and E.S.). We eliminated studies that included duplicate data sets, review articles, book chapters, studies not involving TBI and/or concussions, and studies that included children younger than 2 years who experienced a TBI or inflicted injuries. At least 2 of the 3 authors who performed the initial screening reviewed all the remaining articles; differences about inclusion were resolved through consensus of all 3 authors. The same 3 authors rated the methodologic quality of each study using the Newcastle-Ottawa scale,¹¹ a rating scale of methodologic rigor for observational studies. Again, differences were resolved through consensus of these 3 authors. The Newcastle-Ottawa scale quality ratings for each study are included in Table 1, which also includes details on the methods (eg, clinician interview, medical records, or parent response on questionnaires) used in each study to diagnose ADHD.^8,9,12-33

Only studies reporting descriptive group statistics about ADHD diagnoses before or following TBI were included. We excluded articles reporting on identical participant pools to avoid duplicate data.

We followed the Meta-analysis of Observational Studies in Epidemiology (MOOSE) reporting guideline for abstracting and assessing data quality and validity. Data from the selected studies were extracted and categorized based on 2 key moderator variables: injury severity and time since injury. One preinjury and 2 postinjury intervals were defined: (1) before injury, (2) 1 year or less postinjury (T1), and (3) more than 1 year postinjury (T2). We used the average time postinjury reported in the study when assigning the time to 1 of these categories; some studies reported results in both time intervals and we included both sets of results. Studies included in this review used either Diagnostic and Statistical Manual of Mental Disorders, 3rd ed, or Diagnostic and Statistical Manual of Mental Disorders, 4th ed, Revised criteria to diagnose ADHD. Classification of injury severity was typically based on Glasgow Coma Scale³⁴ (GCS) scores that were frequently confirmed by other clinical findings (eg, positive neuroimaging findings) along with concussion as a separate category. Although concussion is usually regarded as a form of mild TBI, we distinguished between mild TBI and concussion because differences in injury mechanism and patient ascertainment between these 2 conditions might result in subtle differences in injury severity. In the 2 studies of concussion, patients were ascertained from pediatric sports concussion clinics where they were first seen days to weeks after the injury. In studies of mild TBI, the patients had a variety of injury mechanisms and were diagnosed in emergency departments shortly after the injury. Consistent with convention, most studies reviewed used the following GCS score ranges to group samples on injury severity: mild (GCS, 13-15), moderate (GCS, 9-12), and severe (GCS, 3-8). Groups at baseline were noninjured controls (NICs), other injured controls (OICs), TBI groups of any severity, and concussion. In preliminary analyses, we found that preinjury rates of ADHD were not statistically significantly different between mild, moderate, and severe TBI, so we combined preinjury rates of ADHD across TBI severity. At follow-up, groups included OIC, concussion, mild TBI, moderate TBI, and severe TBI—all excluding groups of children with baseline ADHD or mild TBI including baseline ADHD. At T2 follow-up, we did not have data on concussion including baseline ADHD and mild TBI. The eMethods in the Supplement gives details of data extraction. For each group and interval reported in each study, we extracted (1) total number of participants in the group at each point; (2) number of ADHD diagnoses in each group; (3) breakdown of TBI severity if the group consisted of mixed severity; (4) whether the study controlled for preinjury ADHD; (5) GCS scores for TBI groups; (6) demographic data, such as age, sex, and race/ethnicity; and (7) diagnostic criteria used for ADHD.

Most studies reported the number of preinjury and novel (occurring only post TBI) ADHD diagnoses and total number of participants for each TBI severity level by time since injury group. In 7 studies,^{8,13,17,19,27,30,33} the number of diagnoses was reported for mixed severity levels. The eMethods in the Supplement details how we accommodated the data from the 7 studies.

A key methodologic issue in interpreting the results of individual studies was whether participants with preinjury ADHD were removed from the study sample before determining ADHD rates following TBI because, if studies did not exclude this cohort, interpretation of the results may overestimate the association between TBI and ADHD. Patients with preinjury ADHD were not included in analyses at T1 or T2, with the exception of 2 studies that included patients with preinjury ADHD at T1.^24,25 In those 2 studies, we subtracted the preinjury rate of ADHD pooled across TBI severity from the ADHD rate of ADHD at T1 in children with ADHD before mild TBI. For studies that excluded patients with preinjury ADHD, we did not need to subtract the pooled TBI preinjury ADHD rate because the reported rates were already of novel ADHD at T1 or T2.

We used a bayesian random-effects meta-analysis model fit using Markov chain Monte Carlo³⁵ with JAGS³⁶ in R, version 3.6.3³⁷ to analyze the resulting data. The model included indicators for preinjury, T1, T2, TBI severity level, control groups (injured or noninjured), whether the study controlled for preinjury ADHD, TBI severity level by follow-up interactions, and random effects for study and group within the study. The eMethods in the Supplement gives a complete presentation of our statistical model, including the model and prior distribution. We estimated odds ratios (ORs) and 95% credible intervals (CrIs) comparing each group at each measurement point with the NIC at baseline and OIC at the same point. We calculated the estimated rates with 95% CrI of ADHD for each group at each point. All reported P values are 2 sided, with P < .05 considered statistically significant.

We conducted 3 sets of sensitivity analyses. First, we omitted data from groups in which we estimated TBI proportions in studies that reported ADHD diagnoses in pooled severity groups. Second, we considered an alternative prior distribution for the fixed effects with variances 9 times larger than those used in our primary model. Third, we refit our model 23 times, omitting data from exactly one study at each refit plus 1 refit omitting 2 studies.

We included 24 studies in this review from the 378 studies screened (Figure 1) with 12 374 unique patients with TBI of all severity levels and 43 491 unique controls (NIC, 43 192; and OIC, 299). Males were the largest sex cohort (TBI, 61.8%; NIC, 60.9%; and OIC, 66.1%). Table 1 summarizes age, sex, NOS scale scores, and other characteristics of included studies.

Table 2 reports the numbers of studies and participants for each group at each measurement point and ORs for being diagnosed with ADHD compared with baseline NIC and OIC levels at the same point. eTable 4 in the Supplement reports raw data input into the meta-analysis models, eTable 1 in the Supplement reports ORs comparing TBI groups with the NIC or OIC groups, and eTable 3 in the Supplement reports pairwise OR values at each time. eTable 2 and eFigure 1 in the Supplement present results omitting studies in which the breakdown of TBI severity was not directly reported. Figure 2 illustrates the estimated rates of ADHD with 95% CrIs at preinjury, T1, and T2 for each group and the prevalence of ADHD in the pediatric population. eFigure 2 in the Supplement presents the same data using results from a less informative prior distribution. Estimated rates of ADHD at T2 are given in eFigure 3 in the Supplement for mild TBI and eFigure 4 in the Supplement for severe TBI after single-study deletion. eFigures 5-7 in the Supplement are forest plots of individual study and meta-analysis results.

Odds for a preinjury ADHD diagnosis were not significantly greater in concussion and TBI cohorts than in the NIC or OIC cohorts. However, the rates of ADHD were significantly greater in the TBI but not the concussion, NIC, or OIC groups compared with the pediatric population base rate. The rate of preinjury ADHD diagnoses was 16.0% (95% CrI, 11.3%-21.7%) across TBI injury severity. This rate was significantly greater than the general pediatric population base rate of 10.8% (95% CrI,10.2%-11.4%).

Compared with children without injuries, the odds for ADHD were not significantly increased following concussion (<1 year: OR, 0.32; 95% CrI, 0.05-1.13), mild TBI (≤1 year: OR, 0.56; 95% CrI, 0.16-1.43; >1 year: OR, 1.07; 95% CrI, 0.35-2.48), and moderate TBI (≤1 year: OR, 1.28; 95% CrI, 0.35-3.34; >1 year: OR, 3.67; 95% CrI, 0.83-10.56). Compared with children with other injuries, the odds for ADHD were not significantly increased following mild TBI (≤1 year: OR, 1.07; 95% CrI, 0.33-2.47; >1 year: OR, 1.18; 95% CrI, 0.32-3.12), and moderate TBI (≤1 year: OR, 2.34; 95% CrI, 0.78-5.47; >1 year: OR, 3.78; 95% CrI, 0.93-10.33). In contrast, the odds for ADHD following severe TBI were increased at both time points following TBI compared with children with other injuries (≤1 year: OR, 4.81; 95% CrI, 1.66-11.03; >1 year: OR, 6.70; 95% CrI, 2.02-16.82) and noninjured controls (≤1 year: OR, 2.62; 95% CrI, 0.76-6.64; >1 year: OR, 6.25; 95% CrI, 2.06-15.06). Of 5920 children with severe TBI, 35.5% (95% CrI, 20.6%-53.2%) were diagnosed with ADHD more than 1 year after TBI.

The odds for ADHD following severe TBI were significantly increased compared with concussion (≤1 year: OR, 12.89; 95% CrI, 2.13-43.09), mild TBI (≤1 year: OR, 5.69; 95% CrI, 1.46-15.67; >1 year: OR, 6.65; 95% CrI, 2.14-16.44). At both time points, the odds of ADHD did not differ between concussion, mild TBI, and moderate TBI.

Sensitivity analyses are reported in the eMethods in the Supplement. The sensitivity analyses resulted in minor changes except when the studies by Narad et al²⁹ or Yang et al⁹ were omitted. Excluding these studies resulted in opposite changes in the rate of ADHD at T2 in patients with severe TBI. Omitting the Yang et al study increased the rate of ADHD from 35.9% to 51.6% (95% CrI, 32.3%-70.6%); omitting the Narad et al study decreased the rate of ADHD from 36.0% to 26.2% (95% CrI, 13.4%-43.4%). Omitting both studies left the rate estimate and interval slightly higher than retaining both studies (42.0%; 95% CrI, 21.8%-64.4%).

Across TBI injury severity, the rate of preinjury ADHD diagnoses was 16.0%, which is significantly greater than the general pediatric population base rate of 10.8% in the most recent Centers for Disease Control and Prevention survey.³⁸ This result underscores the importance of controlling for preinjury ADHD in studies attempting to determine the association between TBI and ADHD. To the extent that patients with ADHD diagnoses before TBI are not removed from analysis, the association with TBI may be overestimated. Although the odds of preinjury diagnosis of ADHD in patients with TBI is almost double the odds in the NIC cohort, this difference was not statistically significant with the wide 95% CrIs. The odds of preinjury ADHD in patients with TBI do not significantly differ from the odds in the OIC cohort. These results suggest that comparisons between OIC and TBI may be one way to control for preinjury ADHD, but comparison with a noninjured sample does not adequately control for preinjury ADHD. Given the higher-than-expected rate of preinjury ADHD in patients with TBI, studies that remove preinjury cases from analysis have the most probative value in estimating the effect of TBI on ADHD diagnoses. The increase in rate of preinjury ADHD compared with the population base rate is consistent with studies that reported an increase in risk for TBI in children with ADHD.^39-41 Thus, ADHD appears to be a risk factor for TBI.

Figure 2 depicts an association between injury severity and odds for diagnosis of ADHD at T2 following a TBI. At T1 and T2, the odds for ADHD after concussion, mild TBI, and moderate TBI were not significantly increased compared with the odds in the NIC and OIC cohorts. In contrast, at T1, the odds for ADHD were significantly greater in patients with severe TBI than in the OICs. At T2, the odds for ADHD were significantly greater for patients with severe TBI than for the NIC and OIC cohorts: 35.5% (95% CrI, 20.6%-53.2%) of 5920 children with severe TBI had ADHD at T2. Although the association of severe TBI with an increase in risk for ADHD appeared to be relatively robust, there was a wide CrI for of moderate TBI, particularly at T2.

There may be less of an association of severe TBI at T1 than T2. There are a limited number of longitudinal studies in our analysis. The lack of association might reflect differences between the sample characteristics and diagnostic methods in the studies at T1 and T2 rather than true changes over time. Longitudinal studies with large representative samples are needed to evaluate the development of ADHD following TBI.

Sensitivity analyses generally resulted in minor changes in the estimates with slightly widened 95% CrIs (eMethods in the Supplement). Important exceptions were deletion of studies of Narad et al²⁹ and Yang et al⁹ on the estimated rate of ADHD after severe TBI at T2. Yang et al used different diagnostic criteria and ascertainment methods that resulted in a much lower rate of ADHD diagnoses (6% after 9 years) than other studies. Narad et al had a much higher rate of ADHD diagnoses (75%) than other studies, possibly as a result of broader diagnostic criteria.

The sensitivity to less informative prior distributions resulted in some changes in the estimates with wider 95% CrIs overall (eMethods in the Supplement). The odds of ADHD after concussion excluding baseline ADHD were lower than in the NIC cohort at T1 (OR, 0.17; 95% CrI, 0.01-0.83).

The results of this meta-analysis contradict the conclusion in the earlier meta-analysis that mild TBI was associated with an increase in risk for ADHD.¹⁰ Since that review, additional studies with more participants have been published. The meta-analysis in the earlier review included 5 studies. Only 2 of those studies reported the rates of ADHD after mild TBI; the other 3 studies did not specify whether ADHD occurred before or after TBI.

The risk for ADHD following TBI parallels earlier analyses on the association between TBI and cognitive and academic skills. Immediately after mild TBI, there was an acute, mild reduction in cognitive⁴ and academic⁵ function, but those functions recovered by T2. In contrast, severe TBI resulted in persistent impairments in cognitive function and academic skills. The persistence of functional impairments in children with severe TBI may be associated with the loss of white matter volume and organization^42,43 that worsens over the first 2 years post TBI. Similarly, in patients with ADHD following TBI, decreased white matter organization appears to be a central feature⁴⁴ of the pathophysiologic changes.

Despite the differences in the cause, children with moderate and severe TBI have similarities to children with primary ADHD in reduced white matter integrity⁴⁴ and appear to show enhanced cognitive function in response to methylphenidate, as in children with primary ADHD.⁴⁵

A limitation of this meta-analysis was the small number of studies and patients in the OIC and concussion groups. When using the estimated rates of ADHD contained in Figure 2, the 95% CrIs need to be considered. In addition, the soundness of the conclusions of this meta-analysis are limited by the methodologic quality of the studies included in our analysis.

There are individual exceptions to the aggregate association with injury severity in meta-analyses. This is particularly relevant to clinical practice. There may be some patients with mild TBI who present with severe ADHD symptoms and diagnoses, and some patients with severe TBI may not have ADHD symptoms and diagnoses. By uncovering generalizable information from multiple studies, meta-analytic reviews focus additional scrutiny on exceptions to those generalizations. When confronted with claims that a mild TBI causes persistent, severe ADHD symptoms, clinicians should carefully review how the child was functioning before the TBI before concluding that the child developed ADHD as a result of a TBI. For example, parental educational level (eg, high school or college) and school performance before injury estimated which children with mild TBI had elevated rates of behavior problems⁴⁶ and poor performance on neuropsychological tests of attention.⁴⁷ There may be psychosocial and medical issues that antedated the TBI that need to be addressed to adequately treat a child’s ADHD symptoms.

Accepted for Publication: April 8, 2021.

Published Online: July 12, 2021. doi:10.1001/jamapediatrics.2021.2033

Corresponding Author: Robert F. Asarnow, PhD, Department of Psychiatry, University of California, Los Angeles, 48-240 Semel Institute, 760 Westwood Plaza, Los Angeles, CA 90024 (rasarnow@mednet.ucla.edu).

Author Contributions: Drs Asarnow and Weiss had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Asarnow, Newman, Weiss.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Weiss, Su.

Obtained funding: Asarnow.

Supervision: Asarnow, Newman, Weiss.

Conflict of Interest Disclosures: Dr Su reported receiving grants from Della Martin Foundation during the conduct of the study. No other disclosures were reported.

Funding/Support: This research was supported by a gift from the Della Martin Foundation to Dr Asarnow. The gift provided a fellowship to Dr Su..

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Assistance with study selection was provided by librarian Bethany A. Myers, MS (University of California, Los Angeles). There was no compensation outside of university salary.