CONTEXT: Pediatric mild traumatic brain injury (mTBI) is a common and poorly understood injury. Neuroimaging indexes brain injury and outcome after pediatric mTBI, but remains largely unexplored.
OBJECTIVE: To investigate the differences in neuroimaging findings in children/youth with mTBI. Measures of behavior, symptoms, time since injury, and age at injury were also considered.
DATA SOURCES: A systematic review was conducted up to July 6, 2016.
STUDY SELECTION: Studies were independently screened by 2 authors and included if they met predetermined eligibility criteria: (1) children/youth (5–18 years of age), (2) diagnosis of mTBI, and (3) use of neuroimaging.
DATA EXTRACTION: Two authors independently appraised study quality and extracted demographic and outcome data.
RESULTS: Twenty-two studies met the eligibility criteria, involving 448 participants with mTBI (mean age = 12.7 years ± 2.8). Time postinjury ranged from 1 day to 5 years. Seven different neuroimaging methods were investigated in included studies. The most frequently used method, diffusion tensor imaging (41%), had heterogeneous findings with respect to the specific regions and tracts that showed group differences. However, group differences were observed in many regions containing the corticospinal tract, portions of the corpus callosum, or frontal white-matter regions; fractional anisotropy was increased in 88% of the studies.
LIMITATIONS: This review included a heterogeneous sample with regard to participant ages, time since injury, symptoms, and imaging methods which prevented statistical pooling/modelling.
CONCLUSIONS: These data highlight essential priorities for future research (eg, common data elements) that are foundational to progress the understanding of pediatric concussion.
- ADC —
- apparent diffusion coefficient
- CC —
- corpus callosum
- CT —
- computed tomography
- DTI —
- diffusion tensor imaging
- FA —
- fractional anisotropy
- fMRI —
- functional MRI
- MD —
- mean diffusivity
- MRS —
- magnetic resonance spectroscopy
- mTBI —
- mild traumatic brain injury
- rsfMRI —
- resting-state functional MRI
- SWI —
- susceptibility-weighted imaging
- TBI —
- traumatic brain injury
Pediatric concussion (ie, mild traumatic brain injury [mTBI]) is a common but poorly understood injury.1 Symptoms of mTBI are widespread, encompassing physical, behavioral, emotional, and cognitive aspects.2,3 Although in most children and/or youth, mTBI symptoms resolve within 1 month,4 ∼20% experience persistent symptoms months after the initial injury.5,6 These symptoms can disrupt a child’s everyday life, particularly activities involved in learning and social development.7 It is not known why some individuals experience persistent postconcussive symptoms and others do not.4
An assessment of pediatric mTBI includes self- or parent-reported questionnaires or behavioral assessments (eg, cognitive tests, neuropsychological tests, symptom reporting, and functional tests). Although useful in the immediate (ie, <3 days) and short-term (ie, 1 month) stages postinjury,4 these measures are not sensitive to detect differences among individuals in the long-term (ie, >1 month) stages postinjury.8 Additionally, these measures do not provide an index of the underlying neuropathology.9,10 It is possible that measures of brain structure or function could inform our understanding of persistent post-mTBI symptoms.
Investigating the neuropathological causes of pediatric mTBI is important for 2 reasons. Firstly, the inclusion of an assessment that can help explain the underlying pathology may be key to the stratification of individuals and identifying those who are more likely to experience persistent symptoms.11 Secondly, quantifying the neurobiological impact of brain injury on behavioral outcomes or the trajectory of recovery is an essential first step to designing effective interventions.12,13 Thus, together, this information could help advance our understanding of the “right person, right intervention” approach and progress toward personalized medicine after mTBI.
Brain biomarkers that reflect alterations in brain structure and function may help explain the recovery trajectory and neurobiology of an outcome after neurologic injury.12 There are a number of reviews in which researchers identify brain biomarkers in adult mTBI that collectively reveal alterations in white matter,10,14,15 connectivity,16 and neurophysiology.17,18 To date, there has been no systematic review in which researchers investigate all the neuroimaging methods used to determine brain biomarkers in pediatric mTBI.
Imaging of pediatric mTBI reveals neurologic alterations. Changes include an increase in fractional anisotropy (FA) on diffusion imaging in various regions of the brain at a single time point after injury compared with matched controls19–22 and increased and decreased activation in various regions, such as the prefrontal cortex.23–25 Yet, it is unclear how these changes relate to behavior (eg, cognitive ability, neuropsychological function, symptom reporting, and motor function) or recovery. The identification of the brain biomarkers of pediatric mTBI will enable the quantification of the impact of an injury and trajectory of recovery.13,26 This will directly inform future research and the impact of persistent post-mTBI symptoms, which may help refine return-to-activity decision-making and assist in the development of effective interventions.
Therefore, our main aim in this systematic review was to investigate neuroimaging studies using behavioral outcomes to identify potential brain biomarkers in children and/or youth with mTBI. We considered the following questions: What changes are evident on neuroimaging after mTBI by time point postinjury? Are differences evident on neuroimaging related to behavior? And are neuroimaging findings related to age at the time of mTBI, type of mTBI, or recovery pattern from mTBI?
A systematic review with a best-evidence synthesis was planned.27–29 Pooling data for a meta-analysis was intended (data permitting). If no data pooling was possible, the provision of means, SDs, and proportions was planned.
This review was registered on PROSPERO on July 6, 2016 (CRD42016041499). A literature search of English-language studies was conducted by using the following electronic databases up to July 6, 2016: Medline Ovid, Embase Ovid, the Cumulative Index to Nursing and Allied Health Literature EBSCO, and PsycINFO. The full search strategy for Medline Ovid and Embase Ovid is shown in Supplemental Table 6. The search strategy included keywords and expanded Medical Subjects Headings terms related to pediatrics, imaging methodology, and mTBI. The imaging methods targeted in our search strategy were MRI, functional MRI (fMRI), resting-state functional MRI (rsfMRI), diffusion tensor imaging (DTI), susceptibility-weighted imaging (SWI), EEG, transcranial magnetic stimulation, magnetoencephalography, magnetic resonance spectroscopy (MRS), and positron emission tomography. Definitions of key neuroimaging methods, the measures used, what they index, and how to interpret findings are outlined in Table 1.
Studies were included if they met the following predetermined eligibility criteria:
Population: Human children and youth aged 5 to 18 years diagnosed with mTBI at any stage postinjury as determined by the individual study criteria were included. Mixed age samples of group data were eligible provided that >50% of the sample was within the specified age range;
Intervention: The type of intervention did not influence eligibility. Studies that involved any nonpharmacological intervention (eg, rehabilitation, behavioral interventions, devices, and complementary and/or alternative medicine) or no intervention were included. If intervention studies were found, data were extracted from both pre- and postintervention time points. Only the preintervention time point data were used to pool with other single time point studies;
Comparator: Cohort studies with or without a comparison group were included. Studies in which researchers compared participants with mTBI to participants with more severe traumatic brain injury (TBI) (eg, moderate-to-severe TBI) were not included. Studies were excluded if an imaging method used (eg, computed tomography [CT]) was collected in the context of routine clinical diagnosis (eg, determining a skull fracture) because these provided primarily diagnostic information and did not quantify the nature of the neurologic injury or recovery from mTBI; and
Study type: All study types were included except for systematic reviews, literature reviews, and single-case studies.
From the initial search, all duplicate references were excluded by using the Endnote “find duplicates” filter and then by hand search of references (J.S.). Two authors (J.S. and K.S.H.) screened all reference titles and abstracts according to the predetermined eligibility criteria. Full texts of the remaining studies were independently screened by 2 authors (J.S. and K.S.H.). Disagreements regarding the inclusion of a study were resolved by discussion and criteria review between J.S. and K.S.H. If not resolved, a third author (K.E.B.) was involved to achieve a consensus. If still not resolved, an additional 2 authors (L.A.B. and J.G.Z.) reviewed the study. If majority agreement was not reached, it was documented. Reference lists of all eligible studies were hand searched by J.S. to identify potential studies that were not identified through the initial search process, along with a citation-tracking database, using Web of Science.
Data extraction was undertaken by 1 author (J.S.) and was independently verified by a second author (K.E.B.) by using a predetermined extraction form. Information extracted included the following: (1) study details (ie, authors, date, and study location), (2) participants (ie, age, sex, and characteristics of the injury), (3) imaging method used (ie, measure of brain injury and/or recovery, method of measurement, timing of measurement, and frequency of measurement), (4) clinical measures of mTBI symptoms (ie, symptom reporting, measures of cognition, physical function and mental health, timing of measurement, and frequency of measurement), (5) intervention (ie, no intervention or nonpharmacological interventions, such as behavioral interventions or rehabilitation), (6) comparator (ie, no comparator or children with no neurologic or orthopedic injury), (7) results (ie, means, SDs, coefficients, P values, and effect size), and (8) miscellaneous data that were viewed to be of potential importance to the research questions. Reports were reviewed to ensure that data were only included once in the review.
Study quality and risk of bias were independently appraised by 2 authors (J.S. and K.S.H.). We used a modified version of the Case Control Study Checklist or the Cohort Study Checklist, using items that are pertinent to quality rating, developed by the Critical Appraisal Skills Program46 and the risk of bias tool from the Cochrane Handbook.47 If disagreement occurred, resolution was sought through discussion and review of the study and appraisal checklist. If not resolved, a third reviewer (K.E.B.) was involved to achieve a consensus. If still not resolved, an additional 2 authors (S.B. and J.G.Z.) were asked to review the study and appraisal checklist. If consensus was not reached, it was documented. A best-evidence synthesis was planned by using the best set of studies determined by the quality rating score (at least 8 of 10) and risk of bias (a rating of 3 of 3) to draw conclusions.27
Flow of Studies Through the Review
Through the search, we identified 1421 studies, with 717 remaining after duplicate removal. In Fig 1, we outline the flow of studies. All inclusion criteria disagreements were resolved through discussion. Twenty-two studies met all eligibility criteria. In Table 2, we describe each study. Notably, there has been a progressive increase in imaging studies conducted over time (Fig 2).
Characteristics of Included Studies
Seven different imaging methods were identified, including DTI (41%),19,21,22,48,53,55,57,58 fMRI (27%),23–25,56,60,61 SWI (27%),22,48,55 EEG (14%),44,49,59 anatomic MRI (14%),51,54,55 rsfMRI (5%),52 and MRS (5%)55 (Table 2). Time postinjury ranged from 1 day to 5 years. Researchers in all but 2 studies44,54 reported behavioral data (378 participants; Table 2) using 75 different tests and/or subtests. Researchers in 10 studies did not analyze the relationship between brain imaging and behavior.* Researchers in 6 studies (n = 90) reported individual patient data from 5 different brain imaging methods21,23,52,55,57,60; data were collected between 24 hours and 1 year postinjury. Most studies (n = 14 of 22 studies; 64%) were conducted within ∼1 month postinjury (Fig 3).
Participants and Injury Characteristics
In total, 931 participants (448 with mTBI: average of 13 years old at injury [range 9–16 years]) were included (Fig 4). The cause of injury was documented in 18 studies (220 participants)†: sport-related injury (70%), falls (18%), kicked and/or struck by object (6%), and motor vehicle crash (5%). Imaging methods were diverse when pooling participants with common causes of injury (Fig 5). Researchers in 5 studies (177 participants)22,50,51,53,54 reported Glasgow Coma Scale scores (average of 14.7 ± 0.4 out of 15, indicating minimal disruption in consciousness).62 Loss of consciousness (<30 minutes) was reported in 6 studies (131 participants).20,24,48,50,53,55 Researchers in 7 studies used standardized symptom scales (112 participants)19,21–25,48 and in 2 studies (80 participants)50,59 recorded the number of symptoms that participants experienced.
Quality and Risk of Bias
Quality scores ranged from 2 to 7 out of 8 (average of 5.1; SD of 1.5), revealing moderate quality. Risk of bias ranged from 0 to 3 out of 3 (average of 1.8; SD of 1.2), revealing a low risk of bias. Results regarding quality appraisal and risk of bias are provided in Table 3.
Question 1: What Differences Are Evident on Neuroimaging After mTBI by Time Postinjury?
Researchers in all the studies provided data on neuroimaging by comparing participants with mTBI to controls. Because of the heterogeneity in imaging type, time postinjury, and age of participants, it was not possible to conduct a best-evidence synthesis. In Table 4, we describe each imaging method based on time postinjury. Most of these data reflect a single time point; only 3 studies included longitudinal data.20,55,58 As such, we were not able to include a statistical analysis of differences on time postinjury. However, below is a description of the findings from each imaging method used in the included studies.
All studies in which researchers used DTI (n = 9) revealed differences in diffusion data between the mTBI and control groups. Researchers in all but 1 study55 reported an increase in FA19–22,48,53,57,58 (88% of the studies). Researchers in 2 studies investigated an apparent diffusion coefficient (ADC). Here, there was a decrease in values in the mTBI group compared with the controls.22,53 Mean diffusivity (MD) was explored in 4 studies and generally decreased in the mTBI group compared with controls: 2 studies revealed significantly decreased values,21,48 1 study revealed a nonsignificant trend of decreased values,19 and 1 study revealed no significant differences.58 Studies (n = 7) in which researchers reported radial diffusivity values had mixed findings: 4 studies revealed decreased values,22,48,53,57 and 3 studies revealed no difference between groups.19,55,58 Studies (n = 6) in which researchers reported axial diffusivity values also yielded mixed results: 1 study revealed decreased values,19 1 study revealed increased values,48 and 4 studies revealed no significant differences.53,55,57,58 There was a large amount of heterogeneity among the included studies with respect to the specific regions and tracts that showed group differences; however, many contained the corticospinal tract, components of the corpus callosum (CC), or frontal white-matter regions.
Studies in which researchers used fMRI (n = 6) revealed mixed patterns of brain activity. Two reported decreased activation in areas that included the dorsolateral prefrontal cortex and premotor and/or supplementary motor areas during a verbal working memory task23 and in various regions (including the cerebellum, basal ganglia, and thalamus) during an auditory orienting task.61 Two studies revealed increased activation, specifically in the cerebellum during an inhibitory control component of a task24 and in various clusters during a working memory task.60 Two studies revealed both higher and lower activation, primarily in the dorsolateral prefrontal cortex during various conditions of a working memory task56 and a navigational memory task.25 Researchers in 1 study used rsfMRI at 1 month postinjury, revealing altered functional connectivity in 3 resting-state networks (eg, the default mode network, executive function network, and ventral attention network) in the mTBI compared with the control group.52
Other imaging methods included in this review were SWI, EEG, MRI, and MRS. Four of the 5 studies using SWI generally revealed no differences between those with mTBI compared with matched controls.22,48,55,61 One study revealed a lower volume and number of covert lesions in participants with uncomplicated mTBI (eg, no abnormalities on clinical MRI and/or CT) compared with complicated mTBI (eg, imaging abnormalities on clinical MRI and/or CT).50 Two of the 3 EEG studies revealed lower P3b amplitude after mTBI compared with controls.44,49,59 Three studies in which researchers used anatomic MRI revealed no differences in participants with and without mTBI.51,54,55 Yet, 1 revealed that participants with mTBI had decreased volume in the CC white matter compared with controls.51 Researchers in 1 study employed proton MRS, finding no differences between the mTBI and control groups in concentrations of N-acetyl aspartate, N-acetyl aspartate creatine, and phosphocreatine.55
Question 2: Are Neuroimaging Findings Related to Behavior?
There were 20 studies that provided data on behavior, with or without noting a relationship to brain imaging data (Table 5). Most studies (n = 14 of 17; 82%) in which researchers used neurocognitive assessments did not reveal any differences between the mTBI and control groups.‡ Researchers in most studies (n = 7 of 9; 78%) that provided data on postconcussive symptoms used a standardized measure, with the majority of these (n = 6 of 9; 67%) reporting a significant difference between the mTBI and control groups.19,23–25,48,52
Researchers in 12 studies§ analyzed correlations between brain imaging and behavioral data. Results are described in Table 4, grouped by the type of imaging method. It was not possible to conduct a best-evidence synthesis because of the heterogeneity of the imaging type, time postinjury, participant ages, and measures of behavior. In summary, most DTI studies (n = 5 of 7; 71%) revealed significant correlations, with alterations in diffusivity relating to symptom reporting,22,53 emotional distress,22,53 arithmetic problem-solving,57 and concussion outcome scores.19,21 All EEG studies (n = 2) revealed significant correlations, with alterations in EEG data relating to symptom reporting.49,59 One of 2 fMRI studies revealed a significant relationship between brain activation and neurocognitive assessments and symptom reporting.24 A minority of studies (n = 4 of 12; 33%) with brain-behavior analysis revealed no significant correlations or associations, including studies in which researchers used DTI and neurocognitive assessment20 and concussion assessment scale,55 fMRI and memory task,60 and anatomic MRI volumetric measures and neurocognitive assessments.51
Question 3: Are Neuroimaging Findings Related to Age at the Time of the Injury, Type of Injury, or Recovery Pattern From Injury?
Age at the Time of mTBI
Because of the heterogeneity of the data, including imaging methods and behavioral measures, a statistical analysis accounting for age was not possible. Additionally, no individual study provided a statistical analysis on data for these metrics.
Cause of Injury
Because of the heterogeneity of studies, specifically the time postinjury, age at the time of the injury, and imaging method used, it was not possible to perform a meta-analysis of these data.
Recovery Time After mTBI
Because of the large variability in the time frame of the postinjury assessments and lack of data regarding recovery patterns, it was not possible to statistically analyze these data.
This is the first systematic review in which researchers investigate all neuroimaging methods in pediatric mTBI. The identified studies represent a heterogeneous cohort pertaining to age, time postinjury, imaging method, and behavioral data. Despite this, the included studies provide pertinent information on emerging areas for future research and gaps in our current knowledge that warrant discussion.
Firstly, with respect to DTI findings, which indicate variability in the diffusion properties of water molecules, differences were revealed between the mTBI and control groups. Specifically, the mTBI group generally had increased FA values, a decreased ADC, and decreased MD values up to 6 months postinjury (see Table 1 for definition and explanation). It is thought that these alterations occur immediately postinjury (eg, within 2 days)63 because of cytotoxic edema compressing intracellular space between fibers, and thus restricting diffusion to a uniform direction.63 Surprisingly, findings revealed that FA values in participants with mTBI were elevated even at 6 months postinjury. These longer-term alterations may be due to prolonged subtle cytotoxic edema, which may be more prevalent in a developing brain and, therefore, only apparent in the pediatric population.22 These findings contrast with research in the adult population, in whom many studies reveal decreased FA in the mTBI group64–66; others show increased FA,67,68 and 1 reveals both increased and decreased FA at 2 different time points in the early phase of recovery (ie, <8 days) postinjury.69 In summary, findings from the DTI studies show altered diffusion properties.
Secondly, although there was heterogeneity in the specific regions and tracts that showed differences, many of these regions contained the corticospinal tract components of the CC or frontal white-matter regions. This finding could reflect limited sensitivity in DTI as a method because only major white-matter tracts reliably show changes.30 Nevertheless, this trend is interesting in the pediatric context: myelination commences early (eg, in the third trimester)70 and does not become electrophysiologically complete until adolescence (ie, ∼13 years of age).71 Future researchers could investigate a younger population to understand whether the involvement of the corticospinal pathway is consistent with normal age-related maturation or is due to changes post-mTBI.
Finally, many studies were conducted in the short-term postinjury, which provides relevant information to indicate how neuroimaging may be related to the manifestation of early symptoms (eg, 2 days) and the typical resolution of behavioral symptoms (eg, 1 month postinjury). Notably, studies with a younger average age at injury often had a longer time from injury to assessment, which creates difficulty with data pooling. More longitudinal studies are necessary to pool data.
Clinically, current best practice in concussion management relies on a variety of tools for diagnosis and assessment. Imaging is not currently sensitive or reliable for this purpose. To provide a complete and accurate assessment and build the post-mTBI management plan, brain imaging combined with behavioral measures and other emerging biological biomarkers (eg, genetics and blood)11 may represent best practice.
Some imaging methods identified in this review did not reveal significant differences between participants with and without mTBI (eg, anatomic MRI, SWI, and MRS) nor correlations with behavior. It may be that these imaging methods are viable as potential biomarkers for pediatric mTBI, but investigation in larger samples is required to show group differences.26 The data presented here suggest that they may not provide useful information with which to predict recovery from mTBI in children and/or adolescents.
This systematic review had several limitations, demonstrating the need for future research. First, because of the heterogeneity of the studies, small number of publications, and low proportion of individual patient-reported data (6 studies), we were not able to conduct a best-evidence synthesis or meta-analysis. Pooling individual patient data can depict a different summary than can aggregated group data.72 Therefore, future research using imaging in pediatric mTBI would benefit from publishing individual patient data as well as common data elements73 to allow for the pooling of data. In this way, the scope of a meta-analysis will broaden and possibly be refined on the basis of imaging type (eg, fMRI or DTI), time postinjury, and age at the time of the injury as well as inform clinically ready brain biomarkers.
Second, there were limitations in both the population included in this review and the data reported in individual studies. For example, this review did not include studies with children <5 years of age. Younger children are a neurologically distinct population because of structural immaturities (eg, a lack of myelination)74 and functional differences (eg, differing levels and areas of activation during task performance).75 Thus, the younger age group requires specific investigation.
Additionally, there was heterogeneity in the mechanism of concussive injury (eg, diffuse axonal injury). This, coupled with differences in brain development among children and/or youth, may limit the reliability of currently reported findings. For example, even animal models with specific experimental control of biomechanical forces and homogeneity of age and/or development reveal heterogeneous patterns of diffuse axonal injuries despite the fiber bundles being exposed to nearly identical forces.76,77 As such, future researchers should report on the mechanism of injury and account for this in analyses.
Given that the investigation of pediatric mTBI is in its infancy, our inclusion criteria were designed to maximize the number of studies included. As such, some work outlined in this review includes comparisons between mTBI and other clinical injuries (eg, orthopedic injuries). As data accumulate, researchers in pediatric brain biomarker identification studies should compare mTBI and control groups with no history of neurologic injury or report on the severity and effect of other nonhead injuries. Although comparisons to participants with recent injuries not involving the head are common to account for subtle, nonspecific effects of an injury (eg, anxiety and distress), nonhead-injured control groups may have behavioral differences that limit data pooling.
There were only 3 longitudinal studies included in which researchers collected data up to 6 months postinjury. Long-term studies are vital to indicating predictive biomarkers for persistent impairments in behavior. Findings could help determine which individuals are more at risk for persistent impairments, and thus require more intensive monitoring or comprehensive clinical assessments.78 Furthermore, it could guide individualized mTBI management, including intervention plans and return-to-activity decisions.79 Longitudinal studies are particularly important for pediatric mTBI because of the variability of normal developmental74 and individual outcomes after mTBI.13 Future researchers should include multiple time points after injury and manage individuals for at least 12 months.
There are a number of potentially appropriate imaging and/or brain stimulation methods that were not identified in this review, which have been established in the adult mTBI population. These include transcranial magnetic stimulation17 and near-infrared spectroscopy80 methods, which may yield important data that allow for the detection of subtle brain changes that link the brain and behavior after mTBI.81
This systematic review represents an important step forward in mTBI research and will inform future work. It has been challenging to identify brain biomarkers in pediatric mTBI, in part because of the dynamic time course of changes postinjury16 and heterogeneity across the experiments performed to date. The most frequently used imaging method, DTI, had equally heterogeneous findings, although FA was increased in all but 1 of the included studies. Collectively, this study represents the critical first step in enabling pediatric concussion brain biomarker researchers to overcome the inherent lag behind advancements in the adult population.
- Accepted February 12, 2018.
- Address correspondence to Julia Schmidt, OT, PhD, Department of Physical Therapy, University of British Columbia, 212-2177 Wesbrook Mall, Vancouver, BC, Canada V6T 1Z3. E-mail:
This trial has been registered on PROSPERO (identifier: CRD42016041499).
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Funded by a grant from the Jakeway Family Foundation. Dr Schmidt receives salary support from the Michael Smith Foundation for Health Research (MSFHR); Dr Hayward is funded by Australia’s National Health and Medical Research Council (1088449) and the MSFHR (15980); Dr Brown was supported by the Natural Sciences and Engineering Research Council of Canada; and Dr Zwicker is funded by the MSFHR, Canadian Child Health Clinician Scientist Program, the BC Children’s Hospital Research Institute, the Sunny Hill Foundation, and the Canadian Institutes of Health Research.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Willer B,
- Dumas J,
- Hutson A,
- Leddy J
- Davis GA,
- Anderson V,
- Babl FE, et al
- McCrea M; American Academy of Clinical Neuropsychology
- McCrory P,
- Meeuwisse W,
- Dvořák J, et al
- Kamins J,
- Bigler E,
- Covassin T, et al
- Arfanakis K,
- Haughton VM,
- Carew JD,
- Rogers BP,
- Dempsey RJ,
- Meyerand ME
- McCrea M,
- Meier T,
- Huber D, et al
- Mayer AR,
- Ling JM,
- Yang Z,
- Pena A,
- Yeo RA,
- Klimaj S
- Wilde EA,
- McCauley SR,
- Hunter JV, et al
- Yeates KO,
- Beauchamp M,
- Craig W, et al; Pediatric Emergency Research Canada
- Rosenblum WI
- Levine B,
- Kovacevic N,
- Nica EI, et al
- Hillary FG
- Raichle ME,
- MacLeod AM,
- Snyder AZ,
- Powers WJ,
- Gusnard DA,
- Shulman GL
- Fiser DH,
- Long N,
- Roberson PK,
- Hefley G,
- Zolten K,
- Brodie-Fowler M
- Critical Appraisal Skills Programme
- Chu Z,
- Wilde EA,
- Hunter JV, et al
- Maugans TA,
- Farley C,
- Altaye M,
- Leach J,
- Cecil KM
- Mayer AR,
- Ling J,
- Mannell MV, et al
- Sarnat HB
- Greer JE,
- McGinn MJ,
- Povlishock JT
- Copyright © 2018 by the American Academy of Pediatrics