Abstract
Objective
External examination of a clinical risk score to predict persistent postconcussive symptoms (PPCS) in a pediatric emergency department (ED).
Methods
Prospective cohort study of 5- to 18-year-old patients diagnosed with an acute concussion. Risk factors were collected at diagnosis and participants (n = 85) were followed to determine PPCS 30 days postinjury. Univariate logistic regression analyses were completed to examine associations of risk factors with PPCS.
Results
Headache and total clinical risk score were associated with increased odds of PPCS in the univariate analyses, OR 3.37 (95% CI 1.02, 11.10) and OR 1.25 (95% CI 1.02, 1.52), respectively. Additionally, teenage age group, history of prolonged concussions, and risk group trended toward association with PPCS, OR 4.79 (95% CI 0.93, 24.7), OR 3.41 (95% CI 0.88, 13.20), and OR 2.23 (95% CI 0.88, 5.66), respectively.
Conclusion
Our study supports the use of multiple variables of a clinical risk score to assist with ED risk stratification for pediatric patients at risk for PPCS.
Keywords: Children and behavioral disorders, Head injury, Traumatic brain injury
Introduction
Pediatric concussion remains a critical public health concern. Children in the United States sustain as many as 1.8 million closed head injuries annually, accounting for approximately 1.3 million emergency department (ED) and outpatient visits per year (Bryan, Rowhani-Rahbar, Comstock, Rivara & on behalf of the Seattle Sports Concussion Research Collaborative, 2016; Mannix, O’Brien, & Meehan, 2013; Schutzman & Mannix, 2016). Recent studies estimate that one-third of patients experience symptoms beyond 1 month (Zemek et al., 2016). Persistent postconcussive symptoms (PPCS) include physical and cognitive dysfunction, sleep disturbances, and behavioral changes that can lead to missed days of school and work, impaired academic performance, mood changes, and decreased quality of life (Belanger & Vanderploeg, 2005; Karlin, 2011).
While limited by different definitions of PPCS, previous studies have found multiple factors associated with PPCS, including female sex (Henry, Elbin, Collins, Marchetti, & Kontos, 2016); higher acute symptom severity (Meehan, Mannix, Stracciolini, Elbin, & Collins, 2013); history of multiple concussions (Guskiewicz et al., 2003); teenage age (Babcock et al., 2013); higher premorbid symptom ratings (Olsson et al., 2013); presence of premorbid behavioral and learning problems (Babikian, McArthur, & Asarnow, 2013); higher parent distress (Olsson et al., 2013); lower academic achievement (Babikian et al., 2013); higher levels of somatization (Root et al., 2016); and symptom exaggeration (Kirkwood, Peterson, Connery, Baker, & Grubenhoff, 2014).
A recent multicenter cohort study from nine pediatric emergency departments in Canada examined 47 predictors to derive and internally validate a clinical risk score to predict pediatric patients at risk for PPCS in over 3000 children and adolescents (Zemek et al., 2016). This score is also known as the Predicting and Preventing Postconcussive Problems in Pediatrics (5P) clinical risk score and includes the risk factors of female sex; adolescent age; history of concussion with recovery beyond 1 week; history of physician-diagnosed migraines; specific physical signs and symptoms at ED initial visit (headache, fatigue, sensitivity to noise, and answering questions slowly); and four or more errors on the tandem stance of the Balance Error Scoring System (BESS) (Child SCAT3, 2013; Howell et al., 2018; Zemek et al., 2016). This 12-point risk score was able to predict which participants would have persistent symptoms at 4 weeks postconcussion with close to 70% accuracy in both derivation and validation samples and with better prediction than ED physician judgment. A high clinical risk score was highly specific to PPCS but not as sensitive: Those with high clinical risk scores were very unlikely to recover quickly but many individuals without high clinical risk scores also had PPCS (Zemek et al., 2016).
The primary objective of this study was to externally examine the 5P clinical risk score for PPCS in a pediatric ED population in the United States. We hypothesized that the 5P clinical risk score and its component variables would be associated with PPCS in our patient population.
Materials and Methods
Patients diagnosed with an acute concussion in our tertiary care pediatric ED or satellite community ED were prospectively recruited to participate from June 2017 to May 2019. Participants were considered eligible if they were greater than 5 years and less than 19 years old and had an acute concussion within 48 hr of presentation (Root, Sady, Gai, Vaughan, & Madati, 2020). A concussion was defined as head trauma with associated signs and symptoms of a concussion, such as answering questions slowly, headache, nausea or vomiting, blurry vision, and/or difficulty concentrating (Zuckerbraun, Atabaki, Collins, Thomas, & Gioia, 2014). Patients were ineligible if they had a major psychiatric diagnosis (such as schizophrenia or bipolar disorder), a history of intellectual disability, intracranial pathology such as ventriculoperitoneal shunt, positive findings of traumatic brain injury on a computed tomography scan of head (if obtained), or a Glasgow Coma Scale score ≤ 13 at time of diagnosis (Root et al., 2020).
All enrolled participants completed a questionnaire collecting the nine risk factors identified by Zemek and colleagues (2016) as well as the validated Postconcussion Symptom Inventory (PCSI) to assess their current symptom burden (Gioia, Vaughan, & Sady, 2019; Sady, Vaughan, & Gioia, 2014). The PCSI is an age-based survey ranging from 5 to 20 questions that asks participants to rate their preinjury and current concussion symptoms on an age-appropriate Guttman scale (Fig. 1; online). A research assistant scored each patient on the tandem stance of the BESS, where the number of times the child moved out of position in 20 s was recorded while standing stand heel-to-toe with the nondominant foot in back (Child SCAT3, 2013). Individual ED providers were asked to prospectively predict whether their patient would have PPCS in order to compare physician prediction of PPCS to the 5P clinical risk score.
Fig. 1.
Postconcussion Symptom Inventory (PCSI) Scales, divided by age group.
Participants received discharge instructions on their acute concussion from their individual ED provider. Research coordinators then contacted participants 28–32 days postdischarge via phone or email to complete the PCSI again. If research assistants were unable to reach patients for follow-up via phone, at least three additional follow-up phone calls were made or emails were sent over the course of 2 weeks in order to maximize study participant retention. This project was approved by the hospital’s institutional review board. Parents of participants less than 18 provided consent for participation, and participants over age 12 provided assent.
Statistical Analysis
As simple change algorithms such as those used by Zemek and colleagues (2016) have recently been shown to misclassify concussions and have poor intermethod agreement (Mayer et al., 2020), participants were classified as having the primary outcome of persistent concussion symptoms (PPCS) if their follow-up preinjury-adjusted concussion symptom score on the PCSI was outside the reliable change index (RCI)-based 80% confidence interval, indicating that current symptoms were significantly higher than preinjury symptoms (Gioia et al., 2019; Root et al., 2020). The cut-points were ≥2 points difference between preinjury and current symptom totals for 5- to 7-year-olds, ≥4 points for 8- to 12-year-olds, and ≥6 points for 13- to 18-year-olds.
A 12-point 5P clinical risk score was calculated using the nine known risk factors for PPCS according to Zemek and colleagues (2016): age 8–12 years old, a history of prior concussions with symptoms ≥1 week, a history of physician-diagnosed migraines, symptoms at initial ED visit of answering questions slowly, headache, and sensitivity to noise, and a BESS tandem stance error score ≥4 were each worth one point. Age group 13–18 years old, female sex and feeling fatigued were worth two points each (Zemek et al., 2016). Using the PPCS risk categories presented in Zemek and colleagues (2016), we also divided the sample into low (0–3 points), medium (4–8 points), and high (9–12 points) risk of PPCS.
In order to assess the individual variable associations with PPCS, a series of univariate logistic regression analyses were performed by regressing each of the variable components of the 5P clinical risk score, the risk groups identified by Zemek and colleagues (2016), and the total 5P clinical risk score on the binary outcome of PPCS.
In order to compare physician prediction to the 5P clinical risk score, two receiver operating curves (ROCs) were generated by plotting sensitivity and specificity over threshold values of the 5P clinical risk score and physician prediction in relation to PPCS, and accompanying areas under the curve (AUCs) were calculated. The AUCs for the 5P clinical risk score and physician prediction were then compared using chi-square analysis.
This study was part of a larger study examining risk factors for PPCS, as physician prediction information was added after the original study began (Root et al., 2020). Sample size estimates were determined a priori based on the number of patients needed for the larger study (Root et al., 2020). We estimated a 30% prevalence of PPCS based on Zemek and colleagues (2016) and Peduzzi, Concato, Kemper, & Holford (1996) recommended 10 times the number of predictors, k, taking into account the proportion, p, of successes. With four covariates, the minimum numbers of cases would be 133 and assuming a 25% loss to follow-up, we aimed to recruit 180 patients in the larger study. (Root et al., 2020).
Results
Four-hundred and twenty patients were eligible for participation and one-hundred and eighty-three (43.6%) were enrolled in the larger parent study (Root et al., 2020). One hundred and twenty-five patients were enrolled in the current study—when physician prediction information was available—and 85 (68%) completed follow-up. Twenty-two of the eighty-five subjects (25.9%) had PPCS.
Patient demographics, injury mechanism, and initial ED signs and symptoms of the 85 patients who completed this study, compared to the 40 patients who did not complete 1-month follow-up, are shown in Table 1 (online). Subjects lost during follow-up were more likely to have had a concussion as a result of a fall or bike injury/motor vehicle collision, while those who completed the follow-up were more likely to have a sports-related concussion.
Table 1.
Demographic data of enrolled versus lost to follow-up
| Variable name | Enrolled (n = 85) | Lost to F/U (n = 40) | Overall | p-value | |
|---|---|---|---|---|---|
| Age, mean | 11.1 | 11.0 | 11.1 | 0.82 | |
| Age group, % (n) | 0.79 | ||||
| 5–7 years old | 17.6% (15) | 20.0% (8) | 18.4% (23) | ||
| 8–12 years old | 43.5% (37) | 47.5% (19) | 44.8% (56) | ||
| 13–18 years old | 38.8% (33) | 32.5% (13) | 36.8% (46) | ||
| Gender, % (n) | 0.17 | ||||
| Male | 49.4% (42) | 62.5% (25) | 53.6% (67) | ||
| Female | 50.6% (43) | 37.5% (15) | 46.4% (58) | ||
| Mechanism of injury, % (n) | 0.002 | ||||
| Fall | 44.7% (38) | 62.5% (25) | 50.4% (63) | ||
| Sports-related | 31.8% (27) | 15.0% (6) | 26.4% (33) | ||
| Bike/MVC | 2.4% (2) | 15.0% (6) | 6.4% (8) | ||
| Other | 21.2% (18) | 7.5% (3) | 16.8% (21) | ||
| History of prolonged concussion, % (n) (symptoms ≥ 1 week) | 0.61 | ||||
| Yes | 11.8% (10) | 15.0% (6) | 12.8% (16) | ||
| No | 88.2% (75) | 85.0% (34) | 87.2% (109) | ||
| History of migraines, % (n) | 0.91 | ||||
| (physician-diagnosed) | Yes | 9.6% (8) | 10.3% (4) | 9.8% (12) | |
| No | 90.4% (75) | 89.7% (35) | 90.2% (110) | ||
| Missing | n = 2 | n = 1 | |||
| Headache, % (n) | 0.12 | ||||
| Yes | 63.5% (54) | 77.5% (31) | 68.0% (85) | ||
| No | 36.5% (31) | 22.5% (9) | 32.0% (40) | ||
| Answering questions slowly, % (n) | 0.80 | ||||
| Yes | 37.6% (32) | 40.0% (16) | 38.4% (48) | ||
| No | 62.4% (53) | 60.0% (24) | 61.6% (77) | ||
| Fatigue, % (n) | 0.20 | ||||
| Yes | 55.3% (47) | 67.5% (27) | 59.2% (74) | ||
| No | 44.7% (38) | 32.5% (13) | 40.8% (51) | ||
| Sensitivity to noise, % (n) | 0.47 | ||||
| Yes | 25.9% (22) | 20.0% (8) | 24.0% (30) | ||
| No | 74.1% (63) | 80.0% (32) | 76.0% (95) | ||
| Balance error scoring system, % (n) | 0.53 | ||||
| <4 | 75.3% (64) | 70.0% (28) | 73.6% (92) | ||
| ≥4 | 24.7% (21) | 30.0% (12) | 26.4% (33) | ||
| Initial PCSI, median | 9.0 | 8.0 | 9.0 | 0.90 |
Note: Bolded values indicate a p-value < .05.
The univariate logistic regression analyses are shown in Table 2. Headache at presentation and total 5P clinical risk score were associated with increased odds of PPCS, OR 3.37 (95% CI 1.02, 11.10), and OR 1.25 (95% CI 1.02, 1.52), respectively. The risk factors of teenage age group and history of concussion with symptoms ≥1 week had increased odds of PPCS, but the relationships did not reach statistical significance, OR 4.79 (95% CI 0.93, 24.70) and OR 3.41 (95% CI 0.88, 13.20), respectively.
Table 2.
Univariate logistic regressions of risk factors for persistent postconcussive symptoms
| Variable name | OR of PPCS (95% CI)* | p-value | |
|---|---|---|---|
| Age group | |||
| 5–7 years old | Reference | ||
| 8–12 years old | 1.26 (0.22, 7.07) | 0.79 | |
| 13–18 years old | 4.79 (0.93, 24.70) | 0.06 | |
| Gender | |||
| Male | Reference | ||
| Female | 0.97 (0.37, 2.56) | 0.95 | |
| History of concussion with symptoms ≥ 1 week | |||
| No | Reference | ||
| Yes | 3.41 (0.88, 13.20) | 0.08 | |
| History of migraines | |||
| (physician-diagnosed) | No | Reference | |
| Yes | 3.17 (0.72, 14.00) | 0.13 | |
| Headache | |||
| No | Reference | ||
| Yes | 3.37 (1.02, 11.10) | 0.05 | |
| Answering questions slowly | |||
| No | Reference | ||
| Yes | 1.55 (0.58, 4.16) | 0.38 | |
| Fatigue | |||
| No | Reference | ||
| Yes | 1.59 (0.59, 4.32) | 0.36 | |
| Sensitivity to noise | |||
| No | Reference | ||
| Yes | 1.10 (0.37, 3.30) | 0.86 | |
| Balance error scoring system | |||
| Tandem stance | |||
| Errors <4 | Reference | ||
| Errors ≥4 | 2.20 (0.76, 6.35) | 0.15 | |
| Total 5P clinical risk score | 1.25 (1.02, 1.52) | 0.03 | |
| Risk group | 2.23 (0.88, 5.66) | 0.09 | |
*PPCS = persistent postconcussive symptoms, OR = odds ratio; bolded values indicate a p-value < .1
Additionally, risk group (low, medium, and high) based on Zemek and colleagues (2016) trended toward association with PPCS. Higher risk groups had 2.23 times higher odds of PPCS (95% CI 0.88, 5.66). As expected, the proportion of patients with PPCS rose with each risk category: 13.6% (3/22) of low-risk, 28.6% (16/56) of medium-risk, and 42.9% (3/7) of high-risk patients had PPCS at 1 month.
The ROC for the 5P clinical risk score in relation to predicting PPCS is shown in Fig. 2 (online). The 5P clinical risk score ROC had an AUC of 0.67, (95% CI 0.52–0.80). The ROC for the physician prediction in relation to predicting PPCS is shown in Fig. 3 (online). The physician prediction ROC had an AUC of 0.57 (95% CI 0.48–0.67). Compared to physician prediction, the 5P clinical risk score had an increase in AUC of 0.09 (95% CI −0.07 to 0.25), which was not statistically significant.
Fig. 2.
Receiver operating curve (ROC) for 5P clinical risk score and persistent postconcussive symptoms (PPCS); area under curve was 0.67 (95% CI 0.52–0.80).
Fig. 3.
Receiver operating curve (ROC) for physician prediction and persistent postconcussive symptoms (PPCS); area under curve was 0.57 (95% CI 0.48–0.67).
Discussion
The 5P clinical risk score is an ED-based score of nine variables based on demographics, medical history, and acute concussion physical signs and symptoms that was derived and internally validated to identify Canadian pediatric patients at risk for persistent postconcussive symptoms (Zemek et al., 2016). As a substantial minority of pediatric patients experience PPCS, it is important for ED providers to have a method to rapidly and accurately assess those at risk for PPCS in order to provide more accurate expectations for families and tailor acute interventions for prevention (Howell et al., 2018; Zemek et al., 2016). In our study population, just over 25% of participants had PPCS, aligning with previous research (Babikian et al., 2013; Meehan et al., 2013; Root et al., 2016; Zemek et al., 2016).
Our study is the first ED-based study aimed to prospectively examine the 5P criteria in an external population. We found that both headache at initial ED presentation and the total 5P clinical risk score were associated with increased odds of PPCS in our population. Additionally, while the relationships did not reach statistical significance, a prior history of concussion with recovery ≥1 week and a history of physician-diagnosed migraines each had over three times increased odds of PPCS and teenage age group had over four times increased odds of PPCS. Furthermore, the proportion of patients with PPCS rose with each of the risk categories identified by Zemek and colleagues (2016), and Zemek’s risk group trended toward association with PPCS.
There are multiple possibilities as to why limited predictors were associated with PPCS in our analysis. The most obvious reason is that we had a limited sample size of 85 patients. It is possible that with increased patients, additional predictors would have been significantly associated with PPCS (e.g., history of prolonged concussion had a large effect size, but nonsignificant p-values). It is also important to point out that we had wide confidence intervals of our effect size in our final model, limiting interpretability of our results. This is also likely directly related to our overall small sample of 22 participants with PPCS.
Additionally, the demographics of our patient population in an urban tertiary care pediatric ED in the United States differ significantly from the population of patients from nine pediatric hospitals across Canada. Although our study did not account for fall-related sports injuries, almost half of the concussions in our sample were a result of a fall, while in the Canadian 5P study, over half of the concussions resulted from sports or recreational play. While close to one-third of patients were lost to follow-up, enrolled subjects had similar initial PCSI scores compared to those lost to follow-up, (9.0 and 8.0, respectively, Table 1), indicating similar initial symptom burden between both groups.
In relation to predicting PPCS, there was not a statistically significant difference between the AUC of the 5P clinical risk score, 0.67 (95% CI 0.52–0.80), and the AUC of the physician prediction, 0.57 (95% CI 0.48–0.67). However, these numbers are similar to the AUC of the 5P clinical score (0.68) and the AUC of the physician prediction (0.55) in the internal validation cohort of Zemek and colleagues (2016). With a larger sample, it is also possible that the improvement of the 5P clinical risk score over physician predication may become statistically significant.
Overall, our findings have similarities to Howell and colleagues (2018) who conducted a retrospective validation of the 5P clinical risk score in an outpatient clinic population. In their population of patients from a sports medicine clinic, two of nine predictors (fatigue and tandem stance BESS errors ≥4) were independently associated with PPCS, and the 5P clinical risk score was associated with increased odds of PPCS and total symptom duration (Howell et al., 2018).
Given the widespread prevalence of PPCS, it is critically important to be able to predict those patients who will be at high risk for prolonged recovery in order to intervene early with potential therapies. Our study supports the use of the total 5P clinical risk score and multiple individual variables of the risk score to assist with acute emergency department risk stratification for those at risk for PPCS. Large-scale studies are necessary to further examine the most useful components of the clinical risk score in external acute concussion populations.
Funding
This work was supported by the Clinical and Translational Science Institute at Children’s National Winter 2020 Voucher Award, supported by Award Numbers UL1TR001876 and KL2TR001877 from the NIH National Center for Advancing Translational Sciences, and the Clinical and Translational Science Institute at Children’s National Winter 2018 Voucher Award, supported by Award Numbers UL1TR001876 and KL2TR001877 from the NIH National Center for Advancing Translational Sciences.
Conflict of Interest
None declared.
Acknowledgments
Dr. James Bost, Research Division Chief of Biostatistics and Study Methodology at the Children’s National Research Institute, was instrumental in statistical analyses. Gerald Gioia, Division Chief of Pediatric Neuropsychology at Children’s National Health System, assisted with study planning and design.
References
- Babcock, L., Byczkowski, T., Wade, S. L., Ho, M., Mookerjee, S., & Bazarian, J. J. (2013). Predicting postconcussion syndrome after mild traumatic brain injury in children and adolescents who present to the emergency department. JAMA Pediatrics, 167(2), 156–161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Babikian, T., McArthur, D., & Asarnow, R. F. (2013). Predictors of 1-month and 1-year neurocognitive functioning from the UCLA longitudinal mild, uncomplicated, pediatric traumatic brain injury study. Journal of International Neuropsychological Society, 19(2), 145–154. [DOI] [PubMed] [Google Scholar]
- Belanger, H. G., & Vanderploeg, R. D. (2005). The neuropsychological impact of sports-related concussion: A meta-analysis. Journal of International Neuropsychological Society, 11(4), 345. [DOI] [PubMed] [Google Scholar]
- Bryan, M. A., Rowhani-Rahbar, A., Comstock, R. D., Rivara, F., & on behalf of the Seattle Sports Concussion Research Collaborative (2016). Sports-and recreation-related concussions in US youth. Pediatrics, 138(1), e20154635. [DOI] [PubMed] [Google Scholar]
- Child SCAT3 (2013). British Journal of Sports Medicine, 47(5), 263–263. [PubMed] [Google Scholar]
- Gioia, G. A., Vaughan, C. G., & Sady, M. D. (2019). Technical manual for the Post Concussion Symptom Inventory™–2 (PCSI-2). FL: Lutz. [Google Scholar]
- Guskiewicz, K. M., McCrea, M., Marshall, S. W., Cantu, R. C., Randolph, C., Barr, W. et al. (2003). Cumulative effects associated with recurrent concussion in collegiate football players: The NCAA concussion study. Journal of the American Medical Association, 290(19), 2549–2555. [DOI] [PubMed] [Google Scholar]
- Henry, L. C., Elbin, R. J., Collins, M. W., Marchetti, G., & Kontos, A. P. (2016). Examining recovery trajectories after sport-related concussion with a multimodal clinical assessment approach. Neurosurgery, 78(2), 232–241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Howell, D. R., Zemek, R., Brilliant, A. N., Mannix, R. C., Master, C. L., & Meehan, W. P. (2018). Identifying persistent postconcussion symptom risk in a pediatric sports medicine clinic. American Journal of Sports Medicine, 46(13), 3254–3261. [DOI] [PubMed] [Google Scholar]
- Karlin, A. M. (2011). Concussion in the pediatric and adolescent population: Different population, different concerns. PM & R, 3, S369–S379. [DOI] [PubMed] [Google Scholar]
- Kirkwood, M. W., Peterson, R. L., Connery, A. K., Baker, D. A., & Grubenhoff, J. A. (2014). Postconcussive symptom exaggeration after pediatric mild traumatic brain injury. Pediatrics, 133(4), 643–650. [DOI] [PubMed] [Google Scholar]
- Mannix, R., O’Brien, M. J., & Meehan, W. P. (2013). The epidemiology of outpatient visits for minor head injury: 2005 to 2009. Neurosurgery, 73(1), 129–134. [DOI] [PubMed] [Google Scholar]
- Mayer, A. R., Stephenson, D. D., Dodd, A. B., Robertson-Benta, C. R., Pabbathi Reddy, S., Shaff, N. A. et al. (2020). Comparison of methods for classifying persistent post-concussive symptoms in children. Journal of Neurotrauma, 37(13), 1504–1511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Meehan, W. P., Mannix, R. C., Stracciolini, A., Elbin, R. J., & Collins, M. W. (2013). Symptom severity predicts prolonged recovery after sport-related concussion, but age and amnesia do not. Journal of Pediatrics, 163(3), 721–725. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Olsson, K. A., Lloyd, O. T., LeBrocque, R. M., McKinlay, L., Anderson, V. A., & Kenardy, J. A. (2013). Predictors of child post-concussion symptoms at 6 and 18 months following mild traumatic brain injury. Brain Injury, 27(2), 145–157. [DOI] [PubMed] [Google Scholar]
- Peduzzi, P., Concato, J., Kemper, E., Holford, T. R., & Feinstein, A. R. (1996). A simulation study of the number of events per variable in logistic regression analysis. Journal of Clinical Epidemiology, 49(12), 1373–1379. [DOI] [PubMed] [Google Scholar]
- Root, J. M., Sady, M. D., Gai, J., Vaughan, C. D., & Madati, P. J. (2020). Effect of cognitive and physical rest on persistent post-concussive symptoms following a pediatric head injury. Journal of Pediatrics, 227, 184–190.e4. [DOI] [PubMed] [Google Scholar]
- Root, J. M., Zuckerbraun, N. S., Wang, L., Winger, D. G., Brent, D., Kontos, A. et al. (2016). History of somatization is associated with prolonged recovery from concussion. Journal of Pediatrics, 174, 39–44.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sady, M. D., Vaughan, C. G., & Gioia, G. A. (2014). Psychometric characteristics of the postconcussion symptom inventory in children and adolescents. Archives of Clinical Neuropsychology, 29(4), 348–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schutzman, S. A., & Mannix, R. (2016). Pediatric head trauma. In Shaw, K. N., & Bachur, R. G. (Eds.), Fleisher and Ludwig’s textbook of pediatric emergency medicine (, pp. 247–253). Philadelphia, PA: Wolters Kluwer. [Google Scholar]
- Zemek, R., Barrowman, N., Freedman, S. B., Gravel, J., Gagnon, I., McGahern, C. et al. (2016). Clinical risk score for persistent postconcussion symptoms among children with acute concussion in the ED. Journal of the American Medical Association, 315(10), 1014–1025. [DOI] [PubMed] [Google Scholar]
- Zuckerbraun, N. S., Atabaki, S., Collins, M. W., Thomas, D., & Gioia, G. A. (2014). Use of modified acute concussion evaluation tools in the emergency department. Pediatrics, 133(4), 635–642. [DOI] [PubMed] [Google Scholar]