Risk Factors Associated With Sustaining a Sport-related Concussion: An Initial Synthesis Study of 12,320 Student-Athletes

Risk Factors Associated With Sustaining a Sport-related Concussion: An Initial Synthesis Study of... Abstract Objective The empirical identification of risk factors associated with sport-related concussion (SRC) may improve the management of student-athletes. The current study attempted to identify and quantify bio-cognitive risk factors associated with sustaining a SRC. Methods Cross-sectional ambispective study; level of evidence, 3. Neurocognitive testing of 12,320 middle school, high school and collegiate athletes was completed at preseason baseline and post-SRC. Univariate and multivariable logistic regressions were used to determine which pre-injury variables accurately predicted the occurrence of SRC. A quantitative risk score for each variable was developed. Results Five of 13 variables maintained significance in the multivariable model with the associated weighted point scores: SRC history (21), prior headache treatment (6), contact sport (5), youth level of play (7), and history of ADHD/LD (2). Six stratified groups were formed based on probability of SRC, which produced an area under the curve (AUC) of 0.71 (95% CI 0.69–0.72, p < .001). Though the model was a significant predictor of SRC (X2 = 1,112.75, p < .001), the effect size was small and accounted for only 16% of the overall variance. Conclusions An initial aggregate model of weighted bio-cognitive factors associated with increased odds of sustaining a SRC was developed. Previously validated factors were confirmed, yet a large source of variance remained unexplained. These findings emphasize the need to expand the host factors studied when assessing SRC risk, and that the existing, empirically based bio-cognitive factors do not adequately quantify the risk of SRC. Head injury, Traumatic brain injury, Childhood brain insult Introduction Sport-related concussion (SRC) is a major public health concern that affects athletes of all ages and competition levels. For children 18-years or younger, approximately 1.1–1.9 million sports- and recreation-related concussions are reported annually in the US (Bryan, Rowhani-Rahbar, Comstock, Rivara, & Collaborative, 2016). Additionally, 10–15% of patients endure prolonged symptoms following the injury, termed post-concussion syndrome (PCS; Hou et al., 2012; Morgan et al., 2015; Zuckerman et al., 2016). While the potential long-term effects of this transient brain injury are less well-defined (Ban, Madden, Bailes, Hunt Batjer, & Lonser, 2016), the acute effects can significantly impact neurocognition, vestibular and oculomotor capacities, and daily functioning, including missed time from school and sport (McCrory et al., 2017). Recent trends reflect a rising incidence of SRC. In a study of high school student-athletes that included 25 public schools, Lincoln and co-authors (Lincoln et al., 2011) reported a 4.2-fold increase in SRC rates over an 11 year period. Further, the average annual concussion rate in high school athletes has more than doubled from the 2005/2006 to 2015/2016 school year (Yang, Comstock, Yi, Harvey, & Xun, 2017). A similar trend was found in collegiate student-athletes, where SRC rates doubled from 1988 to 2004 (Daneshvar, Nowinski, McKee, & Cantu, 2011). This upward trend has continued throughout 2013–2014 (O’Connor et al., 2017) following the National Collegiate Athletic Association (NCAA)’s 2011 adoption of a four-pronged concussion policy (Baugh et al., 2015; National Collegiate Athletic Association, 2014). Further highlighting the importance of this injury in younger athletes, the age group most vulnerable for sustaining a SRC is 9–22 years, when team sports are most popular (Zemek, Farion, Sampson, & McGahern, 2013). A recent report by the Institute of Medicine (IOM) called for more research surrounding SRC incidence and risk in athletes aged 5–21 years (Institute of Medicine, 2013). Being able to accurately identify those most at risk may improve concussion safety. The literature suggests that SRC occurs more frequently in females (Castile, Collins, McIlvain, & Comstock, 2012; Covassin, Moran, & Elbin, 2016; Lincoln et al., 2011; Marar, McIlvain, Fields, & Comstock, 2012; O’Connor et al., 2017), young athletes (Davis et al., 2017; Guskiewicz, Weaver, Padua, & Garrett, 2000; Knox, Comstock, McGeehan, & Smith, 2006), those with learning disorders/attention-deficit hyperactivity disorder (ADHD; Iverson et al., 2016; Nelson et al., 2016a), and prior SRC (Abrahams, Fie, Patricios, Posthumus, & September, 2014; Guskiewicz et al., 2007; Makdissi et al., 2013; Schick & Meeuwisse, 2003; Schulz et al., 2004). Although independent risk factors have been identified, heterogeneity in methodology across studies has resulted in the need to quantify the weighted risk of known factors associated with higher rates of SRC. The lack of an empirical method to quantify odds of SRC was the impetus for the current investigation. Drawing from a regional neurocognitive testing database of adolescent and young adult student-athletes, our objective was to synthesize previously identified bio-cognitive risk factors associated with sustaining a SRC by developing an aggregate model of risk (Pavlou et al., 2015). Methods Study Design and Overview A retrospective analysis of prospectively collected data (ambispective design) was conducted. Participants included 12,320 middle school, high school and collegiate student-athletes from middle Tennessee and the surrounding states of Kentucky, Georgia, and Alabama who underwent routine preseason and post-concussion neurocognitive testing (ImPACT, 2012). Consistent with CDC guidelines, new baseline data were obtained at least every 2 years (CDC, 2015). Institutional Review Board (IRB) approval was obtained prior to analysis (IRB# 120991). Data Collection Data in the current study were obtained from the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) evaluation, a commercially available computerized test platform that provides four neurocognitive indices, basic demographic and bio-cognitive information, and a 22-item self-report symptom inventory (ImPACT, 2012). Demographic and bio-cognitive data from each ImPACT evaluation were extracted, including self-reported age, gender, sport, and concussion history, as well as medical and neuropsychiatric history. Given prior research, the variables of headache treatment (Morgan et al., 2015), medication status (McCrory et al., 2013), and self-report of attention deficit hyperactivity disorder and/or learning disability (ADHD/LD; Alosco, Fedor, & Gunstad, 2014; Iverson et al., 2016; Nelson et al., 2016a) were also included. Sport contact level was determined by a pre-existing classification system: contact (football, ice hockey, wrestling, soccer, basketball, men’s lacrosse), limited contact (baseball, softball, women’s lacrosse, volleyball, gymnastics, field hockey), and non-contact (golf, cross-country/track, tennis, swimming; Rice, American Academy of Pediatrics Council on Sports, & Fitness, 2008; Shehata et al., 2009; Zuckerman et al., 2016). Selection of Participants Following written, informed consent by the student-athlete or his/her parent/guardian, all participants completed a baseline test as part of routine athletic care. Baseline ImPACT testing was conducted in group settings during the preseason and under the supervision of a sports medicine professional trained in the administration of ImPACT (Kuhn & Solomon, 2014). If a SRC occurred, the student-athlete completed post-injury ImPACT testing. For all student-athletes in the database, a SRC diagnosis was made in accordance with the definition provided by the international Concussion in Sport Group (CISG) guidelines (McCrory et al., 2013). Suspected concussive injury was assessed initially by certified athletic trainers (ATCs) and subsequently diagnosed by team physicians based on the following on-field or sideline signs or symptoms: (1) lethargy, fogginess, headache, dizziness, nausea, visual problems, photophobia, or phonophobia (2) alteration in mental status, (3) loss of consciousness, or (4) amnesia. Grading systems of concussion severity were not utilized, based on the aforementioned CISG guidelines (McCrory et al., 2013). On the basis of the presence of a post-injury ImPACT tests, student-athletes were operationally dichotomized into those who sustained a SRC and those who did not. Student-athletes without a post-injury test were defined as non-concussed. If a student-athlete sustained multiple SRCs, the first injury was analyzed as the “SRC event” to keep the variable of SRC history free of compounding. All student-athletes with a neurocognitive assessment during the 2011–2016 seasons were eligible for inclusion in the study. Of the total 13,927 individual student-athletes, those with missing data were excluded (1,607, 11.5%), resulting in a final sample of 12,320. The reported and missing data groups significantly differed by treatment history for psychiatric disorder, χ2(1) = 10.97, p < .01, gender χ2(1) = 16.83, p < .01, Level of play (no collegiate cases for exclusions), χ2(1) = 2,939.65, p < .01, and contact sport classification, χ2(1) = 40.74, p < .01. While groups differed significantly in the percentage of cases in each category, the very large N suggests the data are robust and likely not to be affected by the removal of cases with missing data (Hampel, Ronchetti, Rousseau, & Stahel, 1986). Statistical Analysis Variables within the model were treated as dichotomous (e.g., ADHD/LD diagnosis, yes or no) or categorical (e.g., youth, high school, and college sport participation). Univariate logistic regression methods were used to determine the relationship between each bio-cognitive variable with the outcome. Variables with a p-value of <.05 in their respective univariate regression analyses were included in a multivariable regression model. Collinearity was assumed for variables with a Pearson’s correlation coefficient of 0.60 or greater; however, no variables exceeded this commonly used cutoff (Berry & Feldman, 1985; Hinkle, Wiersma, & Jurs, 2002) To develop the SRC risk score, β coefficient values were calculated for each significant risk factor in the multivariable regression. Points were assigned to each factor, with higher points corresponding to stronger association between the variable and outcome. The population was subsequently divided into categories of different SRC odds through a Classification and regression tree (CART) analysis of injury outcome, based on total risk score of aggregate relevant factors. A bootstrap analysis was performed (1,000 samples) in order to attempt to control for sample bias and more accurately reflect population parameters of risk (Pavlou et al., 2015). Results Injury Analysis and Risk-Scoring System A total of 12,320 unique neurocognitive assessments were included in the final analysis. Of total sample, 1,517 (12.3%) student-athletes sustained a SRC. Student-athlete demographics and bio-cognitive characteristics are summarized in Table 1, along with univariate logistic regression analyses assessing the ability of different risk factors to predict the occurrence of SRC. Table 1. Population characteristics and associations of potential sport-related concussion risk factors Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  aNumber of and percentages of subjects indicating the presence of a condition or group. bp Values and odds ratios, with corresponding 95% confidence intervals (CIs) were calculated by binary logistic regression analysis and are for the comparison of those who sustained SRC and athletes who were uninjured. cCategorization of contact sport according to a classification system put forth in a policy statement by the American Academy of Pediatrics (2008). dReference group for weighted comparisons within factor categories. View Large Table 1. Population characteristics and associations of potential sport-related concussion risk factors Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  aNumber of and percentages of subjects indicating the presence of a condition or group. bp Values and odds ratios, with corresponding 95% confidence intervals (CIs) were calculated by binary logistic regression analysis and are for the comparison of those who sustained SRC and athletes who were uninjured. cCategorization of contact sport according to a classification system put forth in a policy statement by the American Academy of Pediatrics (2008). dReference group for weighted comparisons within factor categories. View Large The nine significant variables/sub-categories were entered into logistic regression model by descending order of significance. Five variables maintained their prognostic significance (Table 2). The final multivariable logistic regression model included history of SRC, a positive treatment history for headaches, level of play/age, participation in a contact sport, and a diagnosis of ADHD/LD. Table 2. Logistic regression analysis and development of risk model Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  aAssignment of points to prognostic factors was based on a linear transformation of the corresponding β regression coefficient. bCategory was used as the reference variable (REF) for comparisons with other variables. View Large Table 2. Logistic regression analysis and development of risk model Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  aAssignment of points to prognostic factors was based on a linear transformation of the corresponding β regression coefficient. bCategory was used as the reference variable (REF) for comparisons with other variables. View Large Conversion of β coefficients to risk score points involved a linear transformation (Sullivan, Massaro, & D’Agostino, 2004) by dividing each variable β by 0.21 (the lowest β value, corresponding with ADHD/LD diagnosis), multiplying by a constant (2), and rounding to the nearest integer (Table 3A). Composite scores were calculated for each patient. The classification and regression tree analysis yielded six groups based on the odds of sustaining a SRC, which included very low (0), low (1–5), moderate (6–8), intermediate (9–14), high (15–24), and very high (25–41) with increasing SRC prevalence rates from 4.7% to 54.3% (Table 3B). Likelihood of SRC by classification group produced an area under the curve (AUC) of 0.71 (95% CI 0.69–0.72, p < .001) (Fig. 1). The risk model was a significant predictor of SRC occurrence (X2 = 1,112.75, p < .001); however, the effect size was small at 16% (Nagelkerke’s R2 = 0.16; Table 3C), meaning 84% of the variance remained unaccounted in the current model. Table 3. A. Points corresponding with significant factors; B. Risk categories with corresponding prevalence of SRC per category; C. Overall model containing all variables examining SRC score and odds of SRC. A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  aThe odds category was calculated by adding the points for each factor. bThe difference in the probability of injury in high- and the low-risk groups calculated by the formula (Phigh – Plow) ÷ 100. cThe C statistic/Area under the curve is reported and significant at the p < .001 level. View Large Table 3. A. Points corresponding with significant factors; B. Risk categories with corresponding prevalence of SRC per category; C. Overall model containing all variables examining SRC score and odds of SRC. A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  aThe odds category was calculated by adding the points for each factor. bThe difference in the probability of injury in high- and the low-risk groups calculated by the formula (Phigh – Plow) ÷ 100. cThe C statistic/Area under the curve is reported and significant at the p < .001 level. View Large Fig. 1. View largeDownload slide Receiver operating characteristic (ROC) curve of sport-related concussion (SRC) model. Area under the curve (AUC) = 0.71 (95% CI; 69–72), p < .001. Fig. 1. View largeDownload slide Receiver operating characteristic (ROC) curve of sport-related concussion (SRC) model. Area under the curve (AUC) = 0.71 (95% CI; 69–72), p < .001. Discussion From a sample of 12,320 student-athletes, the current study synthesized five empirically based medical and sport-specific variables associated with sustaining a SRC. While these five variables have been previously associated with increased SRC risk, the current study replicated prior analogous findings using one of the largest sample sizes to date. Additionally, the current study extended prior findings by quantifying and synthesizing the five identified SRC risk factors in a single model. From these variables and corresponding risk categories, it was our hope that practitioners could provide student-athletes and families with further objective information on SRC risk, as complex sport and scholastic decisions are made. However, the use of bio-cognitive variables poorly quantified the risk of SRC, signifying that new areas to assess SRC risk are needed. Significant Factors The current risk score was comprised of five previously reported SRC risk factors: SRC history, prior headache treatment, sport contact level, age, and history of ADHD/LD. Previous concussion was the strongest predictor in our analysis, a finding confirmed by Abrahams and colleagues (2014) systematic review of SRC risk factors. Concussion history as a risk factor has been further supported in a Level 1 study by Zemper et al. (2003) of 572 high school and collegiate football players, a Level 2 study by Hollis et al. (2009) of 347 amateur rugby players, and a Level 2 study by Marshall et al. (2015) of 8,905 student-athletes playing seven sports at the high school and college level. Given that recurrent concussion rates have significantly declined nationally approximately 2.6 years following the implementation of concussion laws in high school sports (Yang et al., 2017), this association may be weakened in future, similar studies. Contact sports were also found to be significantly associated with SRC. These results are fairly intuitive, and have been confirmed by a systematic review of youth sports concussion, where rugby, hockey, and football had the highest rates compared to all contact and non-contact sports (Pfister, Pfister, Hagel, Ghali, & Ronksley, 2016). We also found that a history of prior headache treatment increased the odds of SRC, which is in agreement with a previous study (Gordon, Dooley, & Wood, 2006). Interestingly, a history of headache treatment remained significant when entered into the multivariable model, but migraine, which was significant independent of other variables, did not. This suggests that a self-reported history of headache treatment may place student-athletes at higher odds for SRC, whereas a self-reported history of more severe migraines may not. It could be that headaches alone, not their severity or type, is the relevant factor. However, the self-reported nature of both diagnoses limits our interpretation. Lastly, previous research supports increased SRC odds in those with ADHD/LD (Nelson et al., 2016a). Non-Significant Factors With respect to gender, some evidence suggests that females may be at increased risk for SRC, as demonstrated by three large high school studies, each comprised of over approximately 2,000 high school student-athletes (Castile et al., 2012; Lincoln et al., 2011; Marar et al., 2012). Little evidence exists to suggest males are at increased risk, except for one large study of 4,468 youth American football players (Nation, Nelson, Yard, Comstock, & McKenzie, 2011) and a survey study of 50,352 Canadian citizens ages 12–24 years who suffered sport/recreational concussions (Gordon et al., 2006). In light of these mixed findings, the most recent CISG guidelines concluded that gender was not a risk factor for SRC. The previously mentioned systematic review (Abrahams et al., 2014) concluded the same based on several large studies (Koh & Cassidy, 2004; Yang et al., 2008). Some studies have illustrated that psychiatric histories may place student-athletes at an increased risk for prolonged symptoms, but to our knowledge, with respect to concussion incidence, the area has been relatively understudied. Of note, psychiatric history was a significant predictor of SRC independent of other variables, but no longer remained significant within the multivariable model, suggesting a potential underlying relationship where one of, or multiple variables may have mediated the association between psychiatric history and SRC incidence. Utility of These Findings & Future Directions The current SRC risk model and scores has potential to improve counseling and management of the 36 million middle school, high school, and collegiate student-athletes who participate in organized sports every year (NFHS, 2014; Brain, 2015) However, we view our findings as preliminary, and recommend it be used cautiously as one tool amongst many strategies to counsel patients regarding SRC risk. While the aggregate score cannot predict absolute risk of SRC, the included factors can be taken under consideration by parents, and clinicians. Despite its concise format and simplicity, the current stratification tool, if cross-validated in further research, should serve as one piece of an individualized and multi-disciplinary management process. More importantly, the small effect size (16%) of the current model tells a larger, more pressing story. Despite a large sample size and validated scoring methods, the use of baseline bio-cognitive variables poorly quantified the risk of SRC. It is incumbent on the SRC community to learn from this poorly performing model and expand the current paradigm to be more inclusive of a diverse range of factors that account for the other 86% of risk potential risk. A natural progression may be to include the assessment of baseline symptoms, balance scores, and/or neurocognitive scores into a newer model, either alone or in conjunction with the established bio-cognitive variables. Other factors worthy of investigation include position (Pellman et al., 2004), mechanism (Zuckerman et al., 2016), and competition type (practice or games) (Dompier et al., 2015). The use of video analysis to identify objective signs can also be used to quantify in-competition factors (Echemendia et al., 2017). Further, genetics (Finnoff, Jelsing, & Smith, 2011), playing behaviors, and protective equipment may prove to be prognostic variables (Abrahams et al., 2014). It is our belief that SRC researchers need to rigorously pursue these difficult to answer questions, such risk of SRC and/or prolonged recovery, while reporting the effect sizes of significant results. This is especially true when considering the limitations inherent in reporting only p-values and not effect sizes, or the magnitude at which factors place people at risk (Sullivan et al., 2004) Limitations The current study is not without limitation. Our synthesis model had an AUC of 0.71, which is in the fair range (Metz, 1978). However, given the frequent uncertainty of SRC diagnosis, we believe this is clinically relevant, and also in line with similar SRC-related risk scores (Zemek et al., 2016). Further, it is possible that student-athletes in our database saw non-institution affiliated providers post-injury, which would lead to imprecise estimations of injury odds. This would occur if the student-athletes transferred schools, broke protocol and did not have a post-injury test, or was not seen by an institutional provider. However, if the concussion was sustained while playing a scholastic sponsored sport, the likelihood of capturing the SRC episode was high. Moreover, all comorbidities were self-reported by the student-athlete and/or guardian and not corroborated with medical record data. Differentiating between higher-level medical diagnoses, such as a headache versus migraine, may be above the medical knowledge level of patients and their families. As stated, the current model accounted for 16% of the variance in outcomes, suggesting that 84% of factors that account for the odds of SRC were not included in this model. Other sources of variance potentially may include baseline symptom measures, some of which have been shown to place student-athletes at a higher odds of concussion (Lau, Collins, & Lovell, 2012). However, these measures were purposefully excluded, so the score could be applied to all student-athletes, not just those with baseline neurocognitive and symptom scores, as baseline symptom assessment and neuropsychological testing may be beyond the resources of many sports and individuals. Furthermore, because the SRC risk model was developed from only student-athletes who underwent neurocognitive testing and symptom assessment, these results may not be applicable to dissimilar settings. Lastly, it is important to note the probabilities generated from this model do not represent absolute risk or relative risk of sustaining SRC, but rather, represent the relative odds of injury-based trends within a dataset relative to other subjects and factors. As such, if further research validates and extends this initial effort, then consideration of the related factors could be used as a complementary tool in the management of SRC. Conclusions An empirically developed risk model was generated to synthesize bio-cognitive factors associated with the risk of sustaining a SRC in American youth, high school, and collegiate student-athletes. If validated by further independent research, assigning student-athletes into one of six stratified groups may provide more informed preseason counseling, which could translate into individualized in-season care. It should be emphasized that this initial model is not a stand-alone tool, should be used in conjunction with additional patient-specific information, and awaits cross-validation, refinement, and extension. It is our hope that the current model will spark researchers to expand their study of SRC risk predictors outside the commonly studied demographic and bio-cognitive factors that to date, have poorly quantified the risk of SRC. Previous Presentations None. Funding This study did not receive funding from any institution or grant. Conflict of Interest None declared. References Abrahams, S., Fie, S. M., Patricios, J., Posthumus, M., & September, A. V. ( 2014). Risk factors for sports concussion: an evidence-based systematic review. British Journal of Sports Medicine , 48, 91– 97. doi:10.1136/bjsports-2013-092734. Google Scholar CrossRef Search ADS PubMed  Alosco, M. L., Fedor, A. F., & Gunstad, J. ( 2014). Attention deficit hyperactivity disorder as a risk factor for concussions in NCAA division-I athletes. Brain Injury , 28, 472– 474. doi:10.3109/02699052.2014.887145. Google Scholar CrossRef Search ADS PubMed  Ban, V. S., Madden, C. J., Bailes, J. E., Hunt Batjer, H., & Lonser, R. R. ( 2016). The science and questions surrounding chronic traumatic encephalopathy. Neurosurgical Focus , 40, E15. doi:10.3171/2016.2.FOCUS15609. Google Scholar CrossRef Search ADS PubMed  Baugh, C. M., Kroshus, E., Daneshvar, D. H., Filali, N. A., Hiscox, M. J., & Glantz, L. H. ( 2015). Concussion management in United States college sports: Compliance with national collegiate athletic association concussion policy and areas for improvement. The American Journal of Sports Medicine , 43, 47– 56. doi:10.1177/0363546514553090. Google Scholar CrossRef Search ADS PubMed  Berry, W. D., & Feldman, S. ( 1985). Multiple regression in practice (quantitative applications in the social sciences) . Thousand Oaks. CA: SAGE Publications. Google Scholar CrossRef Search ADS   Brain, S. ( 2015). Youth Sports Statistics. Retrieved from http://www.statisticbrain.com/youth-sports-statistics/ Bryan, M. A., Rowhani-Rahbar, A., Comstock, R. D., Rivara, F., & Collaborative, S. S. C. R. ( 2016). Sports- and recreation-related concussions in US youth. Pediatrics , 138. 10.1542/peds.2015-4635. Castile, L., Collins, C. L., McIlvain, N. M., & Comstock, R. D. ( 2012). The epidemiology of new versus recurrent sports concussions among high school athletes, 2005-2010. British Journal of Sports Medicine , 46, 603– 610. doi:10.1136/bjsports-2011-090115. Google Scholar CrossRef Search ADS PubMed  Center for Disease Control and Prevention (CDC). ( 2015, February 16). FAQs about baseline testing.  Retrieved from https://www.cdc.gov/headsup/basics/baseline_testing.html Covassin, T., Moran, R., & Elbin, R. J. ( 2016). Sex differences in reported concussion injury rates and time loss from participation: An update of the National Collegiate Athletic Association Injury Surveillance Program From 2004-2005 Through 2008-2009. Journal of Athletic Training , 51, 189– 194. doi:10.4085/1062-6050-51.3.05. Google Scholar CrossRef Search ADS PubMed  Daneshvar, D. H., Nowinski, C. J., McKee, A. C., & Cantu, R. C. ( 2011). The epidemiology of sport-related concussion. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 30 ( 1), 1– 17 vii. doi:10.1016/j.csm.2010.08.006. Davis, G. A., Anderson, V., Babl, F. E., Gioia, G. A., Giza, C. C., Meehan, W., et al.  . ( 2017). What is the difference in concussion management in children as compared with adults? A systematic review. British Journal of Sports Medicine , 51, 949– 957. doi:10.1136/bjsports-2016-097415. Google Scholar CrossRef Search ADS PubMed  Dompier, T. P., Kerr, Z. Y., Marshall, S. W., Hainline, B., Snook, E. M., Hayden, R., et al.  . ( 2015). Incidence of concussion during practice and games in youth, high school, and collegiate American Football Players. JAMA Pediatric , 169, 659– 665. doi:10.1001/jamapediatrics.2015.0210. Google Scholar CrossRef Search ADS   Echemendia, R. J., Bruce, J. M., Meeuwisse, W., Hutchison, M. G., Comper, P., & Aubry, M. ( 2017). Can visible signs predict concussion diagnosis in the National Hockey League? British Journal of Sports Medicine . doi:10.1136/bjsports-2016-097090. Finnoff, J. T., Jelsing, E. J., & Smith, J. ( 2011). Biomarkers, genetics, and risk factors for concussion. PM&R , 3, S452– S459. Google Scholar CrossRef Search ADS PubMed  Gordon, K. E., Dooley, J. M., & Wood, E. P. ( 2006). Is migraine a risk factor for the development of concussion? British Journal of Sports Medicine , 40, 184– 185. doi:10.1136/bjsm.2005.022251. Google Scholar CrossRef Search ADS PubMed  Guskiewicz, K. M., Marshall, S. W., Bailes, J., McCrea, M., Harding, H. P., Jr., Matthews, A., et al.  . ( 2007). Recurrent concussion and risk of depression in retired professional football players. Medicine and Science in Sports and Exercise , 39, 903– 909. doi:10.1249/mss.0b013e3180383da5. Google Scholar CrossRef Search ADS PubMed  Guskiewicz, K. M., Weaver, N. L., Padua, D. A., & Garrett, W. E., Jr. ( 2000). Epidemiology of concussion in collegiate and high school football players. The American Journal of Sports Medicine , 28, 643– 650. Google Scholar CrossRef Search ADS PubMed  Hampel, F. R., Ronchetti, E. M., Rousseau, P. J., & Stahel, W. A. ( 1986). Robust statistics, Wiley Series in probability and mathematical statistics: Probability and mathematical statistics . New York: John Wiley & Sons, Inc. Hinkle, D. E., Wiersma, W., & Jurs, S. G. ( 2002). Applied statistics for the behavioral sciences . Andover, United Kingdom: Cengage Learning. Hou, R., Moss-Morris, R., Peveler, R., Mogg, K., Bradley, B. P., & Belli, A. ( 2012). When a minor head injury results in enduring symptoms: A prospective investigation of risk factors for postconcussional syndrome after mild traumatic brain injury. Journal of Neurology, Neurosurgery and Psychiatry , 83, 217– 223. doi:10.1136/jnnp-2011-300767. Google Scholar CrossRef Search ADS   Immediate Post-concussion Assessment and Cognitive Testing. ( 2012). Immediate post-concussion assessment and cognitive testing (ImPACT): Technical manual. Retrieved from https://www.impacttest.com/pdf/ImPACTTechnicalManual.pdf Institute of Medicine. ( 2013). Sports-related concussions in youth: Improving the science, changing the culture. Retrieved from http://www.iom.edu/~/media/Files/Report/Files/2013/Concussions/concussions-RB.pdf Iverson, G. L., Wojtowicz, M., Brooks, B. L., Maxwell, B. A., Atkins, J. E., Zafonte, R., et al.  . ( 2016). High school athletes with ADHD and learning difficulties have a greater lifetime concussion history. Journal of Attention Disorders . doi:10.1177/1087054716657410. Knox, C. L., Comstock, R. D., McGeehan, J., & Smith, G. A. ( 2006). Differences in the risk associated with head injury for pediatric ice skaters, roller skaters, and in-line skaters. Pediatrics , 118, 549– 554. doi:10.1542/peds.2005-2913. Google Scholar CrossRef Search ADS PubMed  Koh, J. O., & Cassidy, J. D. ( 2004). Incidence study of head blows and concussions in competition taekwondo. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 14, 72– 79. Google Scholar CrossRef Search ADS PubMed  Kuhn, A. W., & Solomon, G. S. ( 2014). Supervision and computerized neurocognitive baseline test performance in high school athletes: An initial investigation. Journal of Athletic Training , 49, 800– 805. doi:10.4085/1062-6050-49.3.66. Google Scholar CrossRef Search ADS PubMed  Lau, B. C., Collins, M. W., & Lovell, M. R. ( 2012). Cutoff scores in neurocognitive testing and symptom clusters that predict protracted recovery from concussions in high school athletes. Neurosurgery , 70, 371– 379. doi:10.1227/NEU.0b013e31823150f0discussion 379. Google Scholar CrossRef Search ADS PubMed  Lincoln, A. E., Caswell, S. V., Almquist, J. L., Dunn, R. E., Norris, J. B., & Hinton, R. Y. ( 2011). Trends in concussion incidence in high school sports: A prospective 11-year study. The American Journal of Sports Medicine , 39, 958– 963. doi:10.1177/0363546510392326. Google Scholar CrossRef Search ADS PubMed  Makdissi, M., Davis, G., Jordan, B., Patricios, J., Purcell, L., & Putukian, M. ( 2013). Revisiting the modifiers: How should the evaluation and management of acute concussions differ in specific groups? British Journal of Sports Medicine , 47, 314– 320. doi:10.1136/bjsports-2013-092256. Google Scholar CrossRef Search ADS PubMed  Marar, M., McIlvain, N. M., Fields, S. K., & Comstock, R. D. ( 2012). Epidemiology of concussions among United States high school athletes in 20 sports. The American Journal of Sports Medicine , 40, 747– 755. doi:10.1177/0363546511435626. Google Scholar CrossRef Search ADS PubMed  Marshall, S. W., Guskiewicz, K. M., Shankar, V., McCrea, M., & Cantu, R. C. ( 2015). Epidemiolgy of sports- related concussion in seven US high school and collegiate sports. Injury Epidemiology , 2 ( 1), 1– 10. Google Scholar CrossRef Search ADS   McCrory, P., Meeuwisse, W., Aubry, M., Cantu, B., Dvorak, J., Echemendia, R., et al.  . ( 2013). Consensus statement on Concussion in Sport – The 4th international conference on concussion in sport held in Zurich, November 2012. Physical Therapy in Sport: Official Journal of the Association of Chartered Physiotherapists in Sports Medicine , 14, e1– e13. doi:10.1016/j.ptsp.2013.03.002. Google Scholar CrossRef Search ADS PubMed  McCrory, P., Meeuwisse, W., Dvorak, J., Aubry, M., Bailes, J., Broglio, S., et al.  . ( 2017). Consensus statement on concussion in sport – The 5(th) international conference on concussion in sport held in Berlin, October 2016. British Journal of Sports Medicine . doi:10.1136/bjsports-2017-097699. Metz, C. E. ( 1978). Basic principles of ROC analysis. Seminars in Nuclear Medicine , 8, 283– 298. Google Scholar CrossRef Search ADS PubMed  Morgan, C. D., Zuckerman, S. L., Lee, Y. M., King, L., Beaird, S., Sills, A. K., et al.  . ( 2015). Predictors of postconcussion syndrome after sports-related concussion in young athletes: A matched case-control study. Journal of Neurosurgery. Pediatrics , 15, 589– 598. doi:10.3171/2014.10.PEDS14356. Google Scholar CrossRef Search ADS PubMed  Nation, A. D., Nelson, N. G., Yard, E. E., Comstock, R. D., & McKenzie, L. B. ( 2011). Football-related injuries among 6- to 17-year-olds treated in US emergency departments, 1990-2007. Clinica Pediatrica (Phila) , 50, 200– 207. doi:10.1177/0009922810388511. Google Scholar CrossRef Search ADS   National Collegiate Athletic Association. ( 2014). 2014-15 Sports Medicine Handbook. Retrieved from http://www.ncaapublications.com/DownloadPublication.aspx?download=MD15.pdf National Federation of State High School Assoction (NFHS). ( 2014). High School Participation Increases for 25th Consecutive Year. Retrieved from https://www.nfhs.org/articles/high-school-participation-increases-for-25th-consecutive-year/ Nelson, L. D., Guskiewicz, K. M., Marshall, S. W., Hammeke, T., Barr, W., Randolph, C., et al.  . ( 2016a). Multiple self-reported concussions are more prevalent in athletes with ADHD and learning disability. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 26, 120– 127. doi:10.1097/JSM.0000000000000207. Google Scholar CrossRef Search ADS PubMed  O’Connor, K. L., Baker, M. M., Dalton, S. L., Dompier, T. P., Broglio, S. P., & Kerr, Z. Y. ( 2017). Epidemiology of sport-related concussions in high school athletes: National Athletic Treatment, Injury and Outcomes Network (NATION), 2011-2012 through 2013-2014. Journal of Athletic Training , 52, 175– 185. doi:10.4085/1062-6050-52.1.15. Google Scholar CrossRef Search ADS PubMed  Pavlou, M., Ambler, G., Seaman, S. R., Guttmann, O., Elliott, P., King, M., et al.  . ( 2015). How to develop a more accurate risk prediction model when there are few events. BMJ , 351, h3868. doi:10.1136/bmj.h3868. Google Scholar CrossRef Search ADS PubMed  Pellman, E. J., Powell, J. W., Viano, D. C., Casson, I. R., Tucker, A. M., Feuer, H., et al.  . ( 2004). Concussion in professional football: Epidemiological features of game injuries and review of the literature--part 3. Neurosurgery , 54, 81– 94. discussion 94-86. Google Scholar CrossRef Search ADS PubMed  Pfister, T., Pfister, K., Hagel, B., Ghali, W. A., & Ronksley, P. E. ( 2016). The incidence of concussion in youth sports: A systematic review and meta-analysis. British Journal of Sports Medicine , 50, 292– 297. doi:10.1136/bjsports-2015-094978. Google Scholar CrossRef Search ADS PubMed  Rice, S. G., American Academy of Pediatrics Council on Sports Medicine and Fitness. ( 2008). Medical conditions affecting sports participation. Pediatrics , 121, 841– 848. doi:10.1542/peds.2008-0080. Google Scholar CrossRef Search ADS PubMed  Schick, D. M., & Meeuwisse, W. H. ( 2003). Injury rates and profiles in female ice hockey players. The American Journal of Sports Medicine , 31, 47– 52. Google Scholar CrossRef Search ADS PubMed  Schulz, M. R., Marshall, S. W., Mueller, F. O., Yang, J., Weaver, N. L., Kalsbeek, W. D., et al.  . ( 2004). Incidence and risk factors for concussion in high school athletes, North Carolina, 1996-1999. American Journal of Epidemiology , 160, 937– 944. doi:10.1093/aje/kwh304. Google Scholar CrossRef Search ADS PubMed  Shehata, N., Wiley, J. P., Richea, S., Benson, B. W., Duits, L., & Meeuwisse, W. H. ( 2009). Sport concussion assessment tool: Baseline values for varsity collision sport athletes. British Journal of Sports Medicine , 43, 730– 734. doi:10.1136/bjsm.2009.059832. Google Scholar CrossRef Search ADS PubMed  Sullivan, L. M., Massaro, J. M., & D’Agostino, R. B., Sr. ( 2004). Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Statistics in Medicine , 23, 1631– 1660. doi:10.1002/sim.1742. Google Scholar CrossRef Search ADS PubMed  Yang, J., Comstock, R. D., Yi, H., Harvey, H. H., & Xun, P. ( 2017). New and recurrent concussions in high-school athletes before and after traumatic brain injury laws, 2005-2016. American Journal of Public Health , 107, 1916– 1922. doi:10.2105/AJPH.2017.304056. Google Scholar CrossRef Search ADS PubMed  Yang, J., Phillips, G., Xiang, H., Allareddy, V., Heiden, E., & Peek-Asa, C. ( 2008). Hospitalisations for sport-related concussions in US children aged 5 to 18 years during 2000-2004. British Journal of Sports Medicine , 42, 664– 669. doi:10.1136/bjsm.2007.040923. Google Scholar CrossRef Search ADS PubMed  Zemek, R., Barrowman, N., Freedman, S. B., Gravel, J., Gagnon, I., & McGahern, C., Pediatric Emergency Research Canada Concussion Team. ( 2016). Clinical risk score for persistent postconcussion symptoms among children with acute concussion in the ED. JAMA , 315, 1014– 1025. doi:10.1001/jama.2016.1203. Google Scholar CrossRef Search ADS PubMed  Zemek, R. L., Farion, K. J., Sampson, M., & McGahern, C. ( 2013). Prognosticators of persistent symptoms following pediatric concussion: A systematic review. JAMA Pediatric , 167, 259– 265. doi:10.1001/2013.jamapediatrics.216. Google Scholar CrossRef Search ADS   Zemper, E. D. ( 2003). Two-year prospective study of relative risk of a second cerebral concussion. American Journal of Physical Medicine & Rehabilitation , 82, 653– 659. Google Scholar CrossRef Search ADS PubMed  Zuckerman, S. L., Yengo-Kahn, A. M., Buckley, T. A., Solomon, G. S., Sills, A. K., & Kerr, Z. Y. ( 2016). Predictors of postconcussion syndrome in collegiate student-athletes. Neurosurgical Focus , 40, E13. doi:10.3171/2016.1.FOCUS15593. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Archives of Clinical Neuropsychology Oxford University Press

Risk Factors Associated With Sustaining a Sport-related Concussion: An Initial Synthesis Study of 12,320 Student-Athletes

Loading next page...
 
/lp/ou_press/risk-factors-associated-with-sustaining-a-sport-related-concussion-an-fHgNBxJsy2
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0887-6177
eISSN
1873-5843
D.O.I.
10.1093/arclin/acy006
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective The empirical identification of risk factors associated with sport-related concussion (SRC) may improve the management of student-athletes. The current study attempted to identify and quantify bio-cognitive risk factors associated with sustaining a SRC. Methods Cross-sectional ambispective study; level of evidence, 3. Neurocognitive testing of 12,320 middle school, high school and collegiate athletes was completed at preseason baseline and post-SRC. Univariate and multivariable logistic regressions were used to determine which pre-injury variables accurately predicted the occurrence of SRC. A quantitative risk score for each variable was developed. Results Five of 13 variables maintained significance in the multivariable model with the associated weighted point scores: SRC history (21), prior headache treatment (6), contact sport (5), youth level of play (7), and history of ADHD/LD (2). Six stratified groups were formed based on probability of SRC, which produced an area under the curve (AUC) of 0.71 (95% CI 0.69–0.72, p < .001). Though the model was a significant predictor of SRC (X2 = 1,112.75, p < .001), the effect size was small and accounted for only 16% of the overall variance. Conclusions An initial aggregate model of weighted bio-cognitive factors associated with increased odds of sustaining a SRC was developed. Previously validated factors were confirmed, yet a large source of variance remained unexplained. These findings emphasize the need to expand the host factors studied when assessing SRC risk, and that the existing, empirically based bio-cognitive factors do not adequately quantify the risk of SRC. Head injury, Traumatic brain injury, Childhood brain insult Introduction Sport-related concussion (SRC) is a major public health concern that affects athletes of all ages and competition levels. For children 18-years or younger, approximately 1.1–1.9 million sports- and recreation-related concussions are reported annually in the US (Bryan, Rowhani-Rahbar, Comstock, Rivara, & Collaborative, 2016). Additionally, 10–15% of patients endure prolonged symptoms following the injury, termed post-concussion syndrome (PCS; Hou et al., 2012; Morgan et al., 2015; Zuckerman et al., 2016). While the potential long-term effects of this transient brain injury are less well-defined (Ban, Madden, Bailes, Hunt Batjer, & Lonser, 2016), the acute effects can significantly impact neurocognition, vestibular and oculomotor capacities, and daily functioning, including missed time from school and sport (McCrory et al., 2017). Recent trends reflect a rising incidence of SRC. In a study of high school student-athletes that included 25 public schools, Lincoln and co-authors (Lincoln et al., 2011) reported a 4.2-fold increase in SRC rates over an 11 year period. Further, the average annual concussion rate in high school athletes has more than doubled from the 2005/2006 to 2015/2016 school year (Yang, Comstock, Yi, Harvey, & Xun, 2017). A similar trend was found in collegiate student-athletes, where SRC rates doubled from 1988 to 2004 (Daneshvar, Nowinski, McKee, & Cantu, 2011). This upward trend has continued throughout 2013–2014 (O’Connor et al., 2017) following the National Collegiate Athletic Association (NCAA)’s 2011 adoption of a four-pronged concussion policy (Baugh et al., 2015; National Collegiate Athletic Association, 2014). Further highlighting the importance of this injury in younger athletes, the age group most vulnerable for sustaining a SRC is 9–22 years, when team sports are most popular (Zemek, Farion, Sampson, & McGahern, 2013). A recent report by the Institute of Medicine (IOM) called for more research surrounding SRC incidence and risk in athletes aged 5–21 years (Institute of Medicine, 2013). Being able to accurately identify those most at risk may improve concussion safety. The literature suggests that SRC occurs more frequently in females (Castile, Collins, McIlvain, & Comstock, 2012; Covassin, Moran, & Elbin, 2016; Lincoln et al., 2011; Marar, McIlvain, Fields, & Comstock, 2012; O’Connor et al., 2017), young athletes (Davis et al., 2017; Guskiewicz, Weaver, Padua, & Garrett, 2000; Knox, Comstock, McGeehan, & Smith, 2006), those with learning disorders/attention-deficit hyperactivity disorder (ADHD; Iverson et al., 2016; Nelson et al., 2016a), and prior SRC (Abrahams, Fie, Patricios, Posthumus, & September, 2014; Guskiewicz et al., 2007; Makdissi et al., 2013; Schick & Meeuwisse, 2003; Schulz et al., 2004). Although independent risk factors have been identified, heterogeneity in methodology across studies has resulted in the need to quantify the weighted risk of known factors associated with higher rates of SRC. The lack of an empirical method to quantify odds of SRC was the impetus for the current investigation. Drawing from a regional neurocognitive testing database of adolescent and young adult student-athletes, our objective was to synthesize previously identified bio-cognitive risk factors associated with sustaining a SRC by developing an aggregate model of risk (Pavlou et al., 2015). Methods Study Design and Overview A retrospective analysis of prospectively collected data (ambispective design) was conducted. Participants included 12,320 middle school, high school and collegiate student-athletes from middle Tennessee and the surrounding states of Kentucky, Georgia, and Alabama who underwent routine preseason and post-concussion neurocognitive testing (ImPACT, 2012). Consistent with CDC guidelines, new baseline data were obtained at least every 2 years (CDC, 2015). Institutional Review Board (IRB) approval was obtained prior to analysis (IRB# 120991). Data Collection Data in the current study were obtained from the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) evaluation, a commercially available computerized test platform that provides four neurocognitive indices, basic demographic and bio-cognitive information, and a 22-item self-report symptom inventory (ImPACT, 2012). Demographic and bio-cognitive data from each ImPACT evaluation were extracted, including self-reported age, gender, sport, and concussion history, as well as medical and neuropsychiatric history. Given prior research, the variables of headache treatment (Morgan et al., 2015), medication status (McCrory et al., 2013), and self-report of attention deficit hyperactivity disorder and/or learning disability (ADHD/LD; Alosco, Fedor, & Gunstad, 2014; Iverson et al., 2016; Nelson et al., 2016a) were also included. Sport contact level was determined by a pre-existing classification system: contact (football, ice hockey, wrestling, soccer, basketball, men’s lacrosse), limited contact (baseball, softball, women’s lacrosse, volleyball, gymnastics, field hockey), and non-contact (golf, cross-country/track, tennis, swimming; Rice, American Academy of Pediatrics Council on Sports, & Fitness, 2008; Shehata et al., 2009; Zuckerman et al., 2016). Selection of Participants Following written, informed consent by the student-athlete or his/her parent/guardian, all participants completed a baseline test as part of routine athletic care. Baseline ImPACT testing was conducted in group settings during the preseason and under the supervision of a sports medicine professional trained in the administration of ImPACT (Kuhn & Solomon, 2014). If a SRC occurred, the student-athlete completed post-injury ImPACT testing. For all student-athletes in the database, a SRC diagnosis was made in accordance with the definition provided by the international Concussion in Sport Group (CISG) guidelines (McCrory et al., 2013). Suspected concussive injury was assessed initially by certified athletic trainers (ATCs) and subsequently diagnosed by team physicians based on the following on-field or sideline signs or symptoms: (1) lethargy, fogginess, headache, dizziness, nausea, visual problems, photophobia, or phonophobia (2) alteration in mental status, (3) loss of consciousness, or (4) amnesia. Grading systems of concussion severity were not utilized, based on the aforementioned CISG guidelines (McCrory et al., 2013). On the basis of the presence of a post-injury ImPACT tests, student-athletes were operationally dichotomized into those who sustained a SRC and those who did not. Student-athletes without a post-injury test were defined as non-concussed. If a student-athlete sustained multiple SRCs, the first injury was analyzed as the “SRC event” to keep the variable of SRC history free of compounding. All student-athletes with a neurocognitive assessment during the 2011–2016 seasons were eligible for inclusion in the study. Of the total 13,927 individual student-athletes, those with missing data were excluded (1,607, 11.5%), resulting in a final sample of 12,320. The reported and missing data groups significantly differed by treatment history for psychiatric disorder, χ2(1) = 10.97, p < .01, gender χ2(1) = 16.83, p < .01, Level of play (no collegiate cases for exclusions), χ2(1) = 2,939.65, p < .01, and contact sport classification, χ2(1) = 40.74, p < .01. While groups differed significantly in the percentage of cases in each category, the very large N suggests the data are robust and likely not to be affected by the removal of cases with missing data (Hampel, Ronchetti, Rousseau, & Stahel, 1986). Statistical Analysis Variables within the model were treated as dichotomous (e.g., ADHD/LD diagnosis, yes or no) or categorical (e.g., youth, high school, and college sport participation). Univariate logistic regression methods were used to determine the relationship between each bio-cognitive variable with the outcome. Variables with a p-value of <.05 in their respective univariate regression analyses were included in a multivariable regression model. Collinearity was assumed for variables with a Pearson’s correlation coefficient of 0.60 or greater; however, no variables exceeded this commonly used cutoff (Berry & Feldman, 1985; Hinkle, Wiersma, & Jurs, 2002) To develop the SRC risk score, β coefficient values were calculated for each significant risk factor in the multivariable regression. Points were assigned to each factor, with higher points corresponding to stronger association between the variable and outcome. The population was subsequently divided into categories of different SRC odds through a Classification and regression tree (CART) analysis of injury outcome, based on total risk score of aggregate relevant factors. A bootstrap analysis was performed (1,000 samples) in order to attempt to control for sample bias and more accurately reflect population parameters of risk (Pavlou et al., 2015). Results Injury Analysis and Risk-Scoring System A total of 12,320 unique neurocognitive assessments were included in the final analysis. Of total sample, 1,517 (12.3%) student-athletes sustained a SRC. Student-athlete demographics and bio-cognitive characteristics are summarized in Table 1, along with univariate logistic regression analyses assessing the ability of different risk factors to predict the occurrence of SRC. Table 1. Population characteristics and associations of potential sport-related concussion risk factors Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  aNumber of and percentages of subjects indicating the presence of a condition or group. bp Values and odds ratios, with corresponding 95% confidence intervals (CIs) were calculated by binary logistic regression analysis and are for the comparison of those who sustained SRC and athletes who were uninjured. cCategorization of contact sport according to a classification system put forth in a policy statement by the American Academy of Pediatrics (2008). dReference group for weighted comparisons within factor categories. View Large Table 1. Population characteristics and associations of potential sport-related concussion risk factors Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  Total sample: N = 12,320  All patients  Injured  Non-injured  p Valueb  Odds Ratio (95% CI)  Demographic & medical historya   History of SRC  884 (7.2)  460 (30.3)  424 (3.9)  <.001  10.65 (9.20–12.33)   Female sex  8,112 (65.8)  505 (33.3)  3,703 (34.3)  .45  1.05 (0.93–1.17)   ADHD/LD diagnosis  2,239 (18.3)  350 (23.1)  1,889 (17.5)  <.001  1.42 (1.24–1.61)   Psychotropic medication  507 (5.8)  88 (17.4)  419 (3.9)  .69  1.06 (0.81–1.37)  Treatment history   Headaches  1,341 (10.9)  315 (20.8)  1,026 (9.5)  <.001  2.50 (2.17–2.87)   Migraines  905 (7.4)  195 (12.9)  710 (6.6)  <.001  2.09 (1.77–2.48)   Epilepsy/seizures  113 (0.9)  14 (0.9)  99 (0.9)  .99  1.00 (0.57 –1.76)   Brain surgery  15 (0.1)  0 (0.0)  15 (0.1)  .99  0.00 (0.0–0.0)   Meningitis  63 (0.5)  13 (0.9)  50 (0.5)  .05  1.84 (1.0–3.40)   Substance/alcohol abuse  26 (0.2)  3 (0.2)  23 (0.2)  .90  0.92 (0.28–3.08)   Psychiatric  398 (3.3)  76 (5.0)  322 (3.0)  <.001  1.72 (1.32–2.20)  Age   Youthd  851 (6.9)  189 (12.5)  662 (6.1)  REF  REF   High school  9,704 (78.8)  1,074 (70.8)  8,630 (79.9)  <.001  0.44 (0.36–0.52)   College  1,765 (14.3)  254 (16.7)  1,511 (14.0)  .15  0.89 (0.76–1.05)  Contact classificationc   No contactd  661 (5.4)  31 (2.0)  630 (5.8)  REF  REF   Limited contact  2,330 (18.9)  213 (14.0)  2,117 (19.6)  .26  1.09 (0.94–1.27)   Contact sport  9,329 (75.7)  1,273 (83.9)  8,056 (74.6)  <.001  1.72 (1.50–1.96)  aNumber of and percentages of subjects indicating the presence of a condition or group. bp Values and odds ratios, with corresponding 95% confidence intervals (CIs) were calculated by binary logistic regression analysis and are for the comparison of those who sustained SRC and athletes who were uninjured. cCategorization of contact sport according to a classification system put forth in a policy statement by the American Academy of Pediatrics (2008). dReference group for weighted comparisons within factor categories. View Large The nine significant variables/sub-categories were entered into logistic regression model by descending order of significance. Five variables maintained their prognostic significance (Table 2). The final multivariable logistic regression model included history of SRC, a positive treatment history for headaches, level of play/age, participation in a contact sport, and a diagnosis of ADHD/LD. Table 2. Logistic regression analysis and development of risk model Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  aAssignment of points to prognostic factors was based on a linear transformation of the corresponding β regression coefficient. bCategory was used as the reference variable (REF) for comparisons with other variables. View Large Table 2. Logistic regression analysis and development of risk model Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  Original model  Risk factor  Prevalence  Odds Ratio (95% CI)  p Value  SE  β  Pointsa  SRC history  884 (7.2)  9.14 (7.86–10.62)  <.001  0.08  2.21  21  Tx history headaches  1,341 (10.9)  1.87 (1.56–2.16)  <.001  0.08  0.63  6  ADHD/LD diagnosis  2,239 (18.3)  1.24 (1.07–1.42)  <.01  0.07  0.21  2  Contact classification   No contact sportb  661 (5.4)  REF  REF  REF  REF  0   Limited contact sport  2,330 (18.9)  1.13 (0.96–1.3)  .13  0.08  0.12  0   Contact sport  9,329 (75.7)  1.64 (1.47–1.89)  <.001  0.07  0.49  5  Level of play/age   Youthb  851 (6.9)  REF  REF  REF  REF  7   High school  9,704 (78.8)  0.47 (0.39–0.57)  <.001  0.09  –0.75  0   College  1,765 (14.3)  0.98 (0.82–1.17)  .84  0.09  0.002  0  aAssignment of points to prognostic factors was based on a linear transformation of the corresponding β regression coefficient. bCategory was used as the reference variable (REF) for comparisons with other variables. View Large Conversion of β coefficients to risk score points involved a linear transformation (Sullivan, Massaro, & D’Agostino, 2004) by dividing each variable β by 0.21 (the lowest β value, corresponding with ADHD/LD diagnosis), multiplying by a constant (2), and rounding to the nearest integer (Table 3A). Composite scores were calculated for each patient. The classification and regression tree analysis yielded six groups based on the odds of sustaining a SRC, which included very low (0), low (1–5), moderate (6–8), intermediate (9–14), high (15–24), and very high (25–41) with increasing SRC prevalence rates from 4.7% to 54.3% (Table 3B). Likelihood of SRC by classification group produced an area under the curve (AUC) of 0.71 (95% CI 0.69–0.72, p < .001) (Fig. 1). The risk model was a significant predictor of SRC occurrence (X2 = 1,112.75, p < .001); however, the effect size was small at 16% (Nagelkerke’s R2 = 0.16; Table 3C), meaning 84% of the variance remained unaccounted in the current model. Table 3. A. Points corresponding with significant factors; B. Risk categories with corresponding prevalence of SRC per category; C. Overall model containing all variables examining SRC score and odds of SRC. A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  aThe odds category was calculated by adding the points for each factor. bThe difference in the probability of injury in high- and the low-risk groups calculated by the formula (Phigh – Plow) ÷ 100. cThe C statistic/Area under the curve is reported and significant at the p < .001 level. View Large Table 3. A. Points corresponding with significant factors; B. Risk categories with corresponding prevalence of SRC per category; C. Overall model containing all variables examining SRC score and odds of SRC. A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  A. Risk factor  Points  SRC history  21  Youth level of play (vs. HS/college)  7  History of headache treatment  6  Contact sport  5  ADHD/LD  2  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  B. Odds categorya  SRC cases  Very low (0) N = 2,180  103 (4.7%)  Low (1–5) N = 6,151  512 (8.3%)  Moderate (6–8) N = 1,642  193 (11.8%)  Intermediate (9–14) N = 1,395  226 (16.2%)  High (15–24) N = 176  62 (35.2%)  Very high (25–41) N = 776  421 (54.3%)  Difference in probability of injuryb  0.50  C Statistic/AUC (95%CI)a  0.71 (0.69–0.72)  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  C. SRC score overall modelc  Value  Chi-squared  Sig. (p value)  Nagelkerke’s R2  1,112.75  <.001  0.16  aThe odds category was calculated by adding the points for each factor. bThe difference in the probability of injury in high- and the low-risk groups calculated by the formula (Phigh – Plow) ÷ 100. cThe C statistic/Area under the curve is reported and significant at the p < .001 level. View Large Fig. 1. View largeDownload slide Receiver operating characteristic (ROC) curve of sport-related concussion (SRC) model. Area under the curve (AUC) = 0.71 (95% CI; 69–72), p < .001. Fig. 1. View largeDownload slide Receiver operating characteristic (ROC) curve of sport-related concussion (SRC) model. Area under the curve (AUC) = 0.71 (95% CI; 69–72), p < .001. Discussion From a sample of 12,320 student-athletes, the current study synthesized five empirically based medical and sport-specific variables associated with sustaining a SRC. While these five variables have been previously associated with increased SRC risk, the current study replicated prior analogous findings using one of the largest sample sizes to date. Additionally, the current study extended prior findings by quantifying and synthesizing the five identified SRC risk factors in a single model. From these variables and corresponding risk categories, it was our hope that practitioners could provide student-athletes and families with further objective information on SRC risk, as complex sport and scholastic decisions are made. However, the use of bio-cognitive variables poorly quantified the risk of SRC, signifying that new areas to assess SRC risk are needed. Significant Factors The current risk score was comprised of five previously reported SRC risk factors: SRC history, prior headache treatment, sport contact level, age, and history of ADHD/LD. Previous concussion was the strongest predictor in our analysis, a finding confirmed by Abrahams and colleagues (2014) systematic review of SRC risk factors. Concussion history as a risk factor has been further supported in a Level 1 study by Zemper et al. (2003) of 572 high school and collegiate football players, a Level 2 study by Hollis et al. (2009) of 347 amateur rugby players, and a Level 2 study by Marshall et al. (2015) of 8,905 student-athletes playing seven sports at the high school and college level. Given that recurrent concussion rates have significantly declined nationally approximately 2.6 years following the implementation of concussion laws in high school sports (Yang et al., 2017), this association may be weakened in future, similar studies. Contact sports were also found to be significantly associated with SRC. These results are fairly intuitive, and have been confirmed by a systematic review of youth sports concussion, where rugby, hockey, and football had the highest rates compared to all contact and non-contact sports (Pfister, Pfister, Hagel, Ghali, & Ronksley, 2016). We also found that a history of prior headache treatment increased the odds of SRC, which is in agreement with a previous study (Gordon, Dooley, & Wood, 2006). Interestingly, a history of headache treatment remained significant when entered into the multivariable model, but migraine, which was significant independent of other variables, did not. This suggests that a self-reported history of headache treatment may place student-athletes at higher odds for SRC, whereas a self-reported history of more severe migraines may not. It could be that headaches alone, not their severity or type, is the relevant factor. However, the self-reported nature of both diagnoses limits our interpretation. Lastly, previous research supports increased SRC odds in those with ADHD/LD (Nelson et al., 2016a). Non-Significant Factors With respect to gender, some evidence suggests that females may be at increased risk for SRC, as demonstrated by three large high school studies, each comprised of over approximately 2,000 high school student-athletes (Castile et al., 2012; Lincoln et al., 2011; Marar et al., 2012). Little evidence exists to suggest males are at increased risk, except for one large study of 4,468 youth American football players (Nation, Nelson, Yard, Comstock, & McKenzie, 2011) and a survey study of 50,352 Canadian citizens ages 12–24 years who suffered sport/recreational concussions (Gordon et al., 2006). In light of these mixed findings, the most recent CISG guidelines concluded that gender was not a risk factor for SRC. The previously mentioned systematic review (Abrahams et al., 2014) concluded the same based on several large studies (Koh & Cassidy, 2004; Yang et al., 2008). Some studies have illustrated that psychiatric histories may place student-athletes at an increased risk for prolonged symptoms, but to our knowledge, with respect to concussion incidence, the area has been relatively understudied. Of note, psychiatric history was a significant predictor of SRC independent of other variables, but no longer remained significant within the multivariable model, suggesting a potential underlying relationship where one of, or multiple variables may have mediated the association between psychiatric history and SRC incidence. Utility of These Findings & Future Directions The current SRC risk model and scores has potential to improve counseling and management of the 36 million middle school, high school, and collegiate student-athletes who participate in organized sports every year (NFHS, 2014; Brain, 2015) However, we view our findings as preliminary, and recommend it be used cautiously as one tool amongst many strategies to counsel patients regarding SRC risk. While the aggregate score cannot predict absolute risk of SRC, the included factors can be taken under consideration by parents, and clinicians. Despite its concise format and simplicity, the current stratification tool, if cross-validated in further research, should serve as one piece of an individualized and multi-disciplinary management process. More importantly, the small effect size (16%) of the current model tells a larger, more pressing story. Despite a large sample size and validated scoring methods, the use of baseline bio-cognitive variables poorly quantified the risk of SRC. It is incumbent on the SRC community to learn from this poorly performing model and expand the current paradigm to be more inclusive of a diverse range of factors that account for the other 86% of risk potential risk. A natural progression may be to include the assessment of baseline symptoms, balance scores, and/or neurocognitive scores into a newer model, either alone or in conjunction with the established bio-cognitive variables. Other factors worthy of investigation include position (Pellman et al., 2004), mechanism (Zuckerman et al., 2016), and competition type (practice or games) (Dompier et al., 2015). The use of video analysis to identify objective signs can also be used to quantify in-competition factors (Echemendia et al., 2017). Further, genetics (Finnoff, Jelsing, & Smith, 2011), playing behaviors, and protective equipment may prove to be prognostic variables (Abrahams et al., 2014). It is our belief that SRC researchers need to rigorously pursue these difficult to answer questions, such risk of SRC and/or prolonged recovery, while reporting the effect sizes of significant results. This is especially true when considering the limitations inherent in reporting only p-values and not effect sizes, or the magnitude at which factors place people at risk (Sullivan et al., 2004) Limitations The current study is not without limitation. Our synthesis model had an AUC of 0.71, which is in the fair range (Metz, 1978). However, given the frequent uncertainty of SRC diagnosis, we believe this is clinically relevant, and also in line with similar SRC-related risk scores (Zemek et al., 2016). Further, it is possible that student-athletes in our database saw non-institution affiliated providers post-injury, which would lead to imprecise estimations of injury odds. This would occur if the student-athletes transferred schools, broke protocol and did not have a post-injury test, or was not seen by an institutional provider. However, if the concussion was sustained while playing a scholastic sponsored sport, the likelihood of capturing the SRC episode was high. Moreover, all comorbidities were self-reported by the student-athlete and/or guardian and not corroborated with medical record data. Differentiating between higher-level medical diagnoses, such as a headache versus migraine, may be above the medical knowledge level of patients and their families. As stated, the current model accounted for 16% of the variance in outcomes, suggesting that 84% of factors that account for the odds of SRC were not included in this model. Other sources of variance potentially may include baseline symptom measures, some of which have been shown to place student-athletes at a higher odds of concussion (Lau, Collins, & Lovell, 2012). However, these measures were purposefully excluded, so the score could be applied to all student-athletes, not just those with baseline neurocognitive and symptom scores, as baseline symptom assessment and neuropsychological testing may be beyond the resources of many sports and individuals. Furthermore, because the SRC risk model was developed from only student-athletes who underwent neurocognitive testing and symptom assessment, these results may not be applicable to dissimilar settings. Lastly, it is important to note the probabilities generated from this model do not represent absolute risk or relative risk of sustaining SRC, but rather, represent the relative odds of injury-based trends within a dataset relative to other subjects and factors. As such, if further research validates and extends this initial effort, then consideration of the related factors could be used as a complementary tool in the management of SRC. Conclusions An empirically developed risk model was generated to synthesize bio-cognitive factors associated with the risk of sustaining a SRC in American youth, high school, and collegiate student-athletes. If validated by further independent research, assigning student-athletes into one of six stratified groups may provide more informed preseason counseling, which could translate into individualized in-season care. It should be emphasized that this initial model is not a stand-alone tool, should be used in conjunction with additional patient-specific information, and awaits cross-validation, refinement, and extension. It is our hope that the current model will spark researchers to expand their study of SRC risk predictors outside the commonly studied demographic and bio-cognitive factors that to date, have poorly quantified the risk of SRC. Previous Presentations None. Funding This study did not receive funding from any institution or grant. Conflict of Interest None declared. References Abrahams, S., Fie, S. M., Patricios, J., Posthumus, M., & September, A. V. ( 2014). Risk factors for sports concussion: an evidence-based systematic review. British Journal of Sports Medicine , 48, 91– 97. doi:10.1136/bjsports-2013-092734. Google Scholar CrossRef Search ADS PubMed  Alosco, M. L., Fedor, A. F., & Gunstad, J. ( 2014). Attention deficit hyperactivity disorder as a risk factor for concussions in NCAA division-I athletes. Brain Injury , 28, 472– 474. doi:10.3109/02699052.2014.887145. Google Scholar CrossRef Search ADS PubMed  Ban, V. S., Madden, C. J., Bailes, J. E., Hunt Batjer, H., & Lonser, R. R. ( 2016). The science and questions surrounding chronic traumatic encephalopathy. Neurosurgical Focus , 40, E15. doi:10.3171/2016.2.FOCUS15609. Google Scholar CrossRef Search ADS PubMed  Baugh, C. M., Kroshus, E., Daneshvar, D. H., Filali, N. A., Hiscox, M. J., & Glantz, L. H. ( 2015). Concussion management in United States college sports: Compliance with national collegiate athletic association concussion policy and areas for improvement. The American Journal of Sports Medicine , 43, 47– 56. doi:10.1177/0363546514553090. Google Scholar CrossRef Search ADS PubMed  Berry, W. D., & Feldman, S. ( 1985). Multiple regression in practice (quantitative applications in the social sciences) . Thousand Oaks. CA: SAGE Publications. Google Scholar CrossRef Search ADS   Brain, S. ( 2015). Youth Sports Statistics. Retrieved from http://www.statisticbrain.com/youth-sports-statistics/ Bryan, M. A., Rowhani-Rahbar, A., Comstock, R. D., Rivara, F., & Collaborative, S. S. C. R. ( 2016). Sports- and recreation-related concussions in US youth. Pediatrics , 138. 10.1542/peds.2015-4635. Castile, L., Collins, C. L., McIlvain, N. M., & Comstock, R. D. ( 2012). The epidemiology of new versus recurrent sports concussions among high school athletes, 2005-2010. British Journal of Sports Medicine , 46, 603– 610. doi:10.1136/bjsports-2011-090115. Google Scholar CrossRef Search ADS PubMed  Center for Disease Control and Prevention (CDC). ( 2015, February 16). FAQs about baseline testing.  Retrieved from https://www.cdc.gov/headsup/basics/baseline_testing.html Covassin, T., Moran, R., & Elbin, R. J. ( 2016). Sex differences in reported concussion injury rates and time loss from participation: An update of the National Collegiate Athletic Association Injury Surveillance Program From 2004-2005 Through 2008-2009. Journal of Athletic Training , 51, 189– 194. doi:10.4085/1062-6050-51.3.05. Google Scholar CrossRef Search ADS PubMed  Daneshvar, D. H., Nowinski, C. J., McKee, A. C., & Cantu, R. C. ( 2011). The epidemiology of sport-related concussion. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 30 ( 1), 1– 17 vii. doi:10.1016/j.csm.2010.08.006. Davis, G. A., Anderson, V., Babl, F. E., Gioia, G. A., Giza, C. C., Meehan, W., et al.  . ( 2017). What is the difference in concussion management in children as compared with adults? A systematic review. British Journal of Sports Medicine , 51, 949– 957. doi:10.1136/bjsports-2016-097415. Google Scholar CrossRef Search ADS PubMed  Dompier, T. P., Kerr, Z. Y., Marshall, S. W., Hainline, B., Snook, E. M., Hayden, R., et al.  . ( 2015). Incidence of concussion during practice and games in youth, high school, and collegiate American Football Players. JAMA Pediatric , 169, 659– 665. doi:10.1001/jamapediatrics.2015.0210. Google Scholar CrossRef Search ADS   Echemendia, R. J., Bruce, J. M., Meeuwisse, W., Hutchison, M. G., Comper, P., & Aubry, M. ( 2017). Can visible signs predict concussion diagnosis in the National Hockey League? British Journal of Sports Medicine . doi:10.1136/bjsports-2016-097090. Finnoff, J. T., Jelsing, E. J., & Smith, J. ( 2011). Biomarkers, genetics, and risk factors for concussion. PM&R , 3, S452– S459. Google Scholar CrossRef Search ADS PubMed  Gordon, K. E., Dooley, J. M., & Wood, E. P. ( 2006). Is migraine a risk factor for the development of concussion? British Journal of Sports Medicine , 40, 184– 185. doi:10.1136/bjsm.2005.022251. Google Scholar CrossRef Search ADS PubMed  Guskiewicz, K. M., Marshall, S. W., Bailes, J., McCrea, M., Harding, H. P., Jr., Matthews, A., et al.  . ( 2007). Recurrent concussion and risk of depression in retired professional football players. Medicine and Science in Sports and Exercise , 39, 903– 909. doi:10.1249/mss.0b013e3180383da5. Google Scholar CrossRef Search ADS PubMed  Guskiewicz, K. M., Weaver, N. L., Padua, D. A., & Garrett, W. E., Jr. ( 2000). Epidemiology of concussion in collegiate and high school football players. The American Journal of Sports Medicine , 28, 643– 650. Google Scholar CrossRef Search ADS PubMed  Hampel, F. R., Ronchetti, E. M., Rousseau, P. J., & Stahel, W. A. ( 1986). Robust statistics, Wiley Series in probability and mathematical statistics: Probability and mathematical statistics . New York: John Wiley & Sons, Inc. Hinkle, D. E., Wiersma, W., & Jurs, S. G. ( 2002). Applied statistics for the behavioral sciences . Andover, United Kingdom: Cengage Learning. Hou, R., Moss-Morris, R., Peveler, R., Mogg, K., Bradley, B. P., & Belli, A. ( 2012). When a minor head injury results in enduring symptoms: A prospective investigation of risk factors for postconcussional syndrome after mild traumatic brain injury. Journal of Neurology, Neurosurgery and Psychiatry , 83, 217– 223. doi:10.1136/jnnp-2011-300767. Google Scholar CrossRef Search ADS   Immediate Post-concussion Assessment and Cognitive Testing. ( 2012). Immediate post-concussion assessment and cognitive testing (ImPACT): Technical manual. Retrieved from https://www.impacttest.com/pdf/ImPACTTechnicalManual.pdf Institute of Medicine. ( 2013). Sports-related concussions in youth: Improving the science, changing the culture. Retrieved from http://www.iom.edu/~/media/Files/Report/Files/2013/Concussions/concussions-RB.pdf Iverson, G. L., Wojtowicz, M., Brooks, B. L., Maxwell, B. A., Atkins, J. E., Zafonte, R., et al.  . ( 2016). High school athletes with ADHD and learning difficulties have a greater lifetime concussion history. Journal of Attention Disorders . doi:10.1177/1087054716657410. Knox, C. L., Comstock, R. D., McGeehan, J., & Smith, G. A. ( 2006). Differences in the risk associated with head injury for pediatric ice skaters, roller skaters, and in-line skaters. Pediatrics , 118, 549– 554. doi:10.1542/peds.2005-2913. Google Scholar CrossRef Search ADS PubMed  Koh, J. O., & Cassidy, J. D. ( 2004). Incidence study of head blows and concussions in competition taekwondo. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 14, 72– 79. Google Scholar CrossRef Search ADS PubMed  Kuhn, A. W., & Solomon, G. S. ( 2014). Supervision and computerized neurocognitive baseline test performance in high school athletes: An initial investigation. Journal of Athletic Training , 49, 800– 805. doi:10.4085/1062-6050-49.3.66. Google Scholar CrossRef Search ADS PubMed  Lau, B. C., Collins, M. W., & Lovell, M. R. ( 2012). Cutoff scores in neurocognitive testing and symptom clusters that predict protracted recovery from concussions in high school athletes. Neurosurgery , 70, 371– 379. doi:10.1227/NEU.0b013e31823150f0discussion 379. Google Scholar CrossRef Search ADS PubMed  Lincoln, A. E., Caswell, S. V., Almquist, J. L., Dunn, R. E., Norris, J. B., & Hinton, R. Y. ( 2011). Trends in concussion incidence in high school sports: A prospective 11-year study. The American Journal of Sports Medicine , 39, 958– 963. doi:10.1177/0363546510392326. Google Scholar CrossRef Search ADS PubMed  Makdissi, M., Davis, G., Jordan, B., Patricios, J., Purcell, L., & Putukian, M. ( 2013). Revisiting the modifiers: How should the evaluation and management of acute concussions differ in specific groups? British Journal of Sports Medicine , 47, 314– 320. doi:10.1136/bjsports-2013-092256. Google Scholar CrossRef Search ADS PubMed  Marar, M., McIlvain, N. M., Fields, S. K., & Comstock, R. D. ( 2012). Epidemiology of concussions among United States high school athletes in 20 sports. The American Journal of Sports Medicine , 40, 747– 755. doi:10.1177/0363546511435626. Google Scholar CrossRef Search ADS PubMed  Marshall, S. W., Guskiewicz, K. M., Shankar, V., McCrea, M., & Cantu, R. C. ( 2015). Epidemiolgy of sports- related concussion in seven US high school and collegiate sports. Injury Epidemiology , 2 ( 1), 1– 10. Google Scholar CrossRef Search ADS   McCrory, P., Meeuwisse, W., Aubry, M., Cantu, B., Dvorak, J., Echemendia, R., et al.  . ( 2013). Consensus statement on Concussion in Sport – The 4th international conference on concussion in sport held in Zurich, November 2012. Physical Therapy in Sport: Official Journal of the Association of Chartered Physiotherapists in Sports Medicine , 14, e1– e13. doi:10.1016/j.ptsp.2013.03.002. Google Scholar CrossRef Search ADS PubMed  McCrory, P., Meeuwisse, W., Dvorak, J., Aubry, M., Bailes, J., Broglio, S., et al.  . ( 2017). Consensus statement on concussion in sport – The 5(th) international conference on concussion in sport held in Berlin, October 2016. British Journal of Sports Medicine . doi:10.1136/bjsports-2017-097699. Metz, C. E. ( 1978). Basic principles of ROC analysis. Seminars in Nuclear Medicine , 8, 283– 298. Google Scholar CrossRef Search ADS PubMed  Morgan, C. D., Zuckerman, S. L., Lee, Y. M., King, L., Beaird, S., Sills, A. K., et al.  . ( 2015). Predictors of postconcussion syndrome after sports-related concussion in young athletes: A matched case-control study. Journal of Neurosurgery. Pediatrics , 15, 589– 598. doi:10.3171/2014.10.PEDS14356. Google Scholar CrossRef Search ADS PubMed  Nation, A. D., Nelson, N. G., Yard, E. E., Comstock, R. D., & McKenzie, L. B. ( 2011). Football-related injuries among 6- to 17-year-olds treated in US emergency departments, 1990-2007. Clinica Pediatrica (Phila) , 50, 200– 207. doi:10.1177/0009922810388511. Google Scholar CrossRef Search ADS   National Collegiate Athletic Association. ( 2014). 2014-15 Sports Medicine Handbook. Retrieved from http://www.ncaapublications.com/DownloadPublication.aspx?download=MD15.pdf National Federation of State High School Assoction (NFHS). ( 2014). High School Participation Increases for 25th Consecutive Year. Retrieved from https://www.nfhs.org/articles/high-school-participation-increases-for-25th-consecutive-year/ Nelson, L. D., Guskiewicz, K. M., Marshall, S. W., Hammeke, T., Barr, W., Randolph, C., et al.  . ( 2016a). Multiple self-reported concussions are more prevalent in athletes with ADHD and learning disability. Clinical Journal of Sport Medicine: Official Journal of the Canadian Academy of Sport Medicine , 26, 120– 127. doi:10.1097/JSM.0000000000000207. Google Scholar CrossRef Search ADS PubMed  O’Connor, K. L., Baker, M. M., Dalton, S. L., Dompier, T. P., Broglio, S. P., & Kerr, Z. Y. ( 2017). Epidemiology of sport-related concussions in high school athletes: National Athletic Treatment, Injury and Outcomes Network (NATION), 2011-2012 through 2013-2014. Journal of Athletic Training , 52, 175– 185. doi:10.4085/1062-6050-52.1.15. Google Scholar CrossRef Search ADS PubMed  Pavlou, M., Ambler, G., Seaman, S. R., Guttmann, O., Elliott, P., King, M., et al.  . ( 2015). How to develop a more accurate risk prediction model when there are few events. BMJ , 351, h3868. doi:10.1136/bmj.h3868. Google Scholar CrossRef Search ADS PubMed  Pellman, E. J., Powell, J. W., Viano, D. C., Casson, I. R., Tucker, A. M., Feuer, H., et al.  . ( 2004). Concussion in professional football: Epidemiological features of game injuries and review of the literature--part 3. Neurosurgery , 54, 81– 94. discussion 94-86. Google Scholar CrossRef Search ADS PubMed  Pfister, T., Pfister, K., Hagel, B., Ghali, W. A., & Ronksley, P. E. ( 2016). The incidence of concussion in youth sports: A systematic review and meta-analysis. British Journal of Sports Medicine , 50, 292– 297. doi:10.1136/bjsports-2015-094978. Google Scholar CrossRef Search ADS PubMed  Rice, S. G., American Academy of Pediatrics Council on Sports Medicine and Fitness. ( 2008). Medical conditions affecting sports participation. Pediatrics , 121, 841– 848. doi:10.1542/peds.2008-0080. Google Scholar CrossRef Search ADS PubMed  Schick, D. M., & Meeuwisse, W. H. ( 2003). Injury rates and profiles in female ice hockey players. The American Journal of Sports Medicine , 31, 47– 52. Google Scholar CrossRef Search ADS PubMed  Schulz, M. R., Marshall, S. W., Mueller, F. O., Yang, J., Weaver, N. L., Kalsbeek, W. D., et al.  . ( 2004). Incidence and risk factors for concussion in high school athletes, North Carolina, 1996-1999. American Journal of Epidemiology , 160, 937– 944. doi:10.1093/aje/kwh304. Google Scholar CrossRef Search ADS PubMed  Shehata, N., Wiley, J. P., Richea, S., Benson, B. W., Duits, L., & Meeuwisse, W. H. ( 2009). Sport concussion assessment tool: Baseline values for varsity collision sport athletes. British Journal of Sports Medicine , 43, 730– 734. doi:10.1136/bjsm.2009.059832. Google Scholar CrossRef Search ADS PubMed  Sullivan, L. M., Massaro, J. M., & D’Agostino, R. B., Sr. ( 2004). Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Statistics in Medicine , 23, 1631– 1660. doi:10.1002/sim.1742. Google Scholar CrossRef Search ADS PubMed  Yang, J., Comstock, R. D., Yi, H., Harvey, H. H., & Xun, P. ( 2017). New and recurrent concussions in high-school athletes before and after traumatic brain injury laws, 2005-2016. American Journal of Public Health , 107, 1916– 1922. doi:10.2105/AJPH.2017.304056. Google Scholar CrossRef Search ADS PubMed  Yang, J., Phillips, G., Xiang, H., Allareddy, V., Heiden, E., & Peek-Asa, C. ( 2008). Hospitalisations for sport-related concussions in US children aged 5 to 18 years during 2000-2004. British Journal of Sports Medicine , 42, 664– 669. doi:10.1136/bjsm.2007.040923. Google Scholar CrossRef Search ADS PubMed  Zemek, R., Barrowman, N., Freedman, S. B., Gravel, J., Gagnon, I., & McGahern, C., Pediatric Emergency Research Canada Concussion Team. ( 2016). Clinical risk score for persistent postconcussion symptoms among children with acute concussion in the ED. JAMA , 315, 1014– 1025. doi:10.1001/jama.2016.1203. Google Scholar CrossRef Search ADS PubMed  Zemek, R. L., Farion, K. J., Sampson, M., & McGahern, C. ( 2013). Prognosticators of persistent symptoms following pediatric concussion: A systematic review. JAMA Pediatric , 167, 259– 265. doi:10.1001/2013.jamapediatrics.216. Google Scholar CrossRef Search ADS   Zemper, E. D. ( 2003). Two-year prospective study of relative risk of a second cerebral concussion. American Journal of Physical Medicine & Rehabilitation , 82, 653– 659. Google Scholar CrossRef Search ADS PubMed  Zuckerman, S. L., Yengo-Kahn, A. M., Buckley, T. A., Solomon, G. S., Sills, A. K., & Kerr, Z. Y. ( 2016). Predictors of postconcussion syndrome in collegiate student-athletes. Neurosurgical Focus , 40, E13. doi:10.3171/2016.1.FOCUS15593. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Archives of Clinical NeuropsychologyOxford University Press

Published: Feb 17, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off