Journal of Science and Medicine in Sport 18 (2015) 507–511
Contents lists available at ScienceDirect
Journal of Science and Medicine in Sport journal homepage: www.elsevier.com/locate/jsams
Original research
The underreporting of self-reported symptoms following sports-related concussion Timothy B. Meier a,∗ , Bradley J. Brummel b , Rashmi Singh a , Christopher J. Nerio c , David W. Polanski c , Patrick S.F. Bellgowan a,d a
Laureate Institute for Brain Research, Tulsa, OK, USA Department of Psychology, The University of Tulsa, Tulsa, OK, USA c Department of Athletics, The University of Tulsa, Tulsa, OK, USA d Faculty of Community Medicine, The University of Tulsa, Tulsa, OK, USA b
a r t i c l e
i n f o
Article history: Received 16 February 2014 Received in revised form 8 July 2014 Accepted 14 July 2014 Available online 24 July 2014 Keywords: Mild traumatic brain injury Post-concussion symptoms Neuropsychological testing
a b s t r a c t Objectives: This cohort study was conducted to examine patterns of symptom reporting in concussed athletes in two different testing environments. Design: A prospective cohort study was conducted with repeated measures. Methods: Self-reported symptoms collected by team athletic trainers using the ImPACT Post-Concussion Scale (PCS) were compared to symptoms collected in a confidential setting using structured interviews for depression and anxiety. Ratings were scaled to match scoring of the PCS and categorized into symptom-domains. Scores collected 2 days post-concussion were compared across different rating scales. Confidential self-report scores approximately 9 days post-concussion in cleared athletes were compared to PCS scores collected during return-to-play decisions. Finally, confidential self-report scores collected 9 days post-concussion were compared between cleared and not cleared athletes. Results: Athletes self-reported significantly fewer symptoms to team athletic trainers using the ImPACT test compared to self-reported symptoms collected in a confidential setting during the acute phase of concussion using standard psychiatric interviews. Athletes cleared to play continued to underreport symptoms 9 days post-concussion, particularly psychiatric symptoms. Finally, cleared athletes self-reported similar magnitude of symptoms than non-cleared athletes 9 days post-concussion in confidential research setting. Conclusions: The systematic underreporting of post-concussion symptoms may represent motivated behavior or differences in self-reporting data acquisition. By underreporting symptoms, many cleared athletes are still symptomatic over 1-week post-concussion. This study highlights the need for objective measures for somatic and psychiatric symptoms. © 2014 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
1. Introduction Concussion has a clinically heterogeneous symptomology that requires assessment across multiple symptom domains.1,2 Adding to diagnostic complexity, symptom severity is also modulated by “concussion modifiers” that include age, comorbidities, medication, style of play, and sport.2 The degree of symptom reporting is another modifier that can mislead clinical diagnoses, treatment
∗ Corresponding author. E-mail address: [email protected] (T.B. Meier).
strategies, and recovery decisions. Underreporting is at least partially motivated by a desire to continue playing, with awareness that self-reporting symptoms will prolong return-to-play decisions in high-school football players.3 In an attempt to standardize concussion diagnoses and to minimize clinical reliance on self-reported symptoms, diagnosis of sports-related concussion is often supplemented with computerized or onsite testing metrics.4–6 These metrics include objective measures such as neurocognitive testing; however, they also rely on athletes’ self-reported symptoms for psychiatric and some somatic symptoms. Thus, clinical judgment regarding return-to-play decisions may be biased toward symptoms assessed using objective measures obtained from neurological and
http://dx.doi.org/10.1016/j.jsams.2014.07.008 1440-2440/© 2014 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
508
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
neuropsychological examination. This may leave cleared athletes vulnerable to participating while still symptomatic in the psychiatric and somatic domains. International guidelines for return-to-play decisions following concussion have been established, despite the relative lack of evidence-based criteria.2,7 A graduated return-to-play protocol is recommended, with increasing physical activity occurring at each step after athletes are fully asymptomatic. Currently, there is no distinction regarding what specific symptoms should be favored over other symptoms during return-to-play decisions. Rather, the full resolution of the hallmark symptoms of concussion, which include cognitive, somatic, mood, and sleep-related symptoms, is considered. Typically, symptoms resolve within 7–10 days, though recovery can extend for longer periods of time.2 Previous retrospective studies have documented underreporting of concussive events in collegiate soccer, high school football, collegiate football, and a variety of collegiate women’s sports.3,8,9 In addition, underreporting of post-concussion symptoms compared to cognitive batteries has been reported in a variety of collegiate and high school sports, including football and cheerleading.10–12 This prospective study investigates the underreporting of postconcussive symptoms in the authentic athletic environment and in a controlled research environment. We compared self-reported symptoms collected with the ImPACT battery4 administered onsite at athletic facilities to scores on the Hamilton Anxiety Rating Scale (HAM-A)13 and the Hamilton Depression Rating Scale (HAM-D)14 collected in a confidential setting. First, we tested the hypothesis that athletes underreport cognitive, psychiatric, insomnia, and somatic symptoms in the days immediately post-injury. Secondly, we tested the hypothesis that self-reported symptoms collected during the last ImPACT test used in the decision making process to clear athletes to return-to-play are significantly lower than confidential depression and anxiety measures collected approximately 1-week post-concussion. Finally, we tested the hypothesis that athletes that have been cleared to return-to-play self-report fewer symptoms on depression and anxiety scales 1-week postconcussion than athletes that have not been cleared.
2. Methods A self-selection sampling method was applied to consecutive cases of athletes with concussion from a NCAA Division I university athletic program. Inclusion criteria included any collegiate-aged athlete with a clinician-diagnosed sports-related concussion. Athletes with other, non-concussion injuries that prevented return-to-play were excluded. All participants provided written consent. This study followed procedures outline in the Belmont Report for protection of human participants,15 and was approved by the Western Institutional Review Board and by the University of Oklahoma Health Sciences Center Institutional Review. Medical professionals trained in sports medicine diagnosed concussions based on clinical symptoms. Forty concussed athletes were evaluated following concussion. The average participant age at the time of their concussion was 20.44 years (SD = 1.41) and with 13.08 years of education (SD = 1.05). Three participants were female soccer players, two were male soccer players, four were female basketball players, one was a male basketball player, one was a female volleyball player, and the remaining 29 were male football players. Athletes who participated in this study in or after 2012 were compensated monetarily following NCAA Bylaws 16.11.1.11.1 and 16.11.1.11.12. Athletes who participated prior to 2012 could not be monetarily compensated due to NCAA Bylaw 16.02.3. Athletes completed the ImPACT battery at their athletic facilities under the supervision of athletic trainers following a cliniciandiagnosed concussion. The ImPACT battery includes a series of
cognitive tests comprising of four composite scores and a 22item post-concussion symptom inventory (PCS).4,16 This inventory asks participants to self-report post-concussion symptoms on a seven-point likert scale, with 0 indicating symptom absence, 1–2 indicating mild, 3–4 indicating moderate, and 5–6 indicating severe symptoms. Initial ImPACT testing was performed on an average of 1.30 days post-injury (SD = 0.57). Athletes performed additional post-injury ImPACT testing until they had been cleared to return-to-play by clinicians trained in sports-medicine. Return-toplay decisions were based on clinical judgment following medical exams, aided by the use of the ImPACT battery by comparing neurocognitive task performance and PCS scores to baseline levels. Athletes returned-to-play on an average of 11.77 (SD = 7.83) days following concussion. The average number of clinician-diagnosed concussions prior to the current concussion was 0.75 (SD = 1.0). Thirty-seven of the athletes also performed neuropsychiatric testing in a confidential setting at the Laureate Institute for Brain Research on an average of 1.92 days (SD = 1.04) post-injury. Athletes were informed that their responses in this setting would remain confidential and that team officials would not have access to their test results. Experienced assessment professionals performed structure interviews for the Hamilton Depression Rating Scale with Atypical Depression Supplement (HAM-D)13,17,18 and the Hamilton Anxiety Rating Scale (HAM-A).14,19 The original seventeen-item HAM-D with individual questions scored on either a three-point or five-point scale, and the standard fourteen item HAM-A with individual questions score on a five-points scale were used to document post-concussion symptoms. Twenty-nine athletes participated in a confidential research testing session on an average of 9.14 days post-concussion (SD = 2.08), during which HAM-D and HAM-A structured interviews were collected. At the time of this session, nine of the twentynine athletes had been cleared to return-to-play using the criteria previously described. Scaling of responses was performed to compare the HAM-A and HAM-D scales to the PCS scale based on the number of response options in the scale and the anchors for those response options prior to examining any of the participant scores on the scales. For HAM-A items and five-point items of the HAM-D scores of 4 were converted to 6, scores of 3 were converted to 5, scores of 2 were converted to 3.5, scores of 1 were converted to 1.5, and scores of 0 remained as 0. For three-point HAM-D items scaling to PCS scale was as follows: scores of 0 remained as 0, scores of 1 were converted to 2.5, and scores of 2 were converted to 5.5. Symptoms were grouped into four symptom-domains for analyses comparing the reporting of symptoms from the HAM-D and HAM-A interviews to the PCS score. PCS symptoms were classified into one of four domains following previous classification schemes 20 : Insomnia, psychiatric, cognitive, and somatic (Supplementary Table 1). Individual items from the HAM-D and HAM-A were also grouped into the same symptom-domains. Two of the authors independently classified HAM-D and HAM-A items into symptomdomains using both a criterion of maximum item inclusion and of best symptom fit to PCS items. We chose to use the best symptom fit approach. Items that covered two domains were included in both (Supplementary Table 1). Domain scores for the HAM-D, HAM-A, and PCS were obtained by averaging the scores of the individual items in each domain. Supplementary Table 1 can be found, in the online version, at doi:10.1016/j.jsams.2014.07.008. Two two-way repeated-measures analysis of variances (ANOVA) were performed with the within-subjects factors of symptom-domain (cognitive, insomnia, psychiatric, and somatic) and rating scale (HAM-D, HAM-A, PCS), on self-reported symptom-scores. One ANOVA compared the scores collected approximately two days post-concussion for all athletes, and one
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
Fig. 1. (a) The average domain scores for the scaled HAM-D and HAM-A were compared to the average domain scores for the PCS across all participants 1–2 days post-concussion. (b) The average domain scores for the scaled HAM-D and HAMA collected 9 days post-concussion were compared to PCS scores collected during the last ImPACT session prior to the return-to-play decision. Error bars represent standard deviation.
ANOVA compared cleared athletes HAM-D and HAM-A scores collected approximately 9 days post-concussion to PCS scores collected by the athletic trainers at the last ImPACT test prior to being cleared. For both ANOVAs, planned contrasts were performed comparing the HAM-D and HAM-A scores to PCS scores in each symptom-domain to directly test the hypothesis that athletes underreport post-concussive symptoms to team affiliated athletic trainers. A mixed-design repeated-measures ANOVA with the withinsubjects factors of symptom-domain and rating scale, and the between-subjects factor of group (cleared and not-cleared) was performed to test the hypothesis that domain-specific symptom severity 9 days post-concussion depends on whether or not athletes had been cleared to return-to-play. Post hoc analyses with Bonferroni correction for multiple comparisons effects were performed for each ANOVA when appropriate. Effect sizes, calculated as Cohen’s d, were computed for each planned contrast to demonstrate the magnitude of the observed differences.
3. Results There was a significant main effect of rating scale, F(2,72) = 33.53, p < 0.01, and symptom-domain, F(3,108) = 7.52, p < 0.01, on self-reported scores collected approximately two-days post-concussion (Fig. 1a). The interaction between rating scale and symptom-domain was also significant, F(6,216) = 10.35, p < 0.01.
509
Post hoc pairwise comparisons with Bonferroni correction showed that athletes reported fewer symptoms across all four symptom-domains in the PCS than in the HAM-A (p < 0.01) and HAM-D (p < 0.01). Planned contrasts revealed significantly greater HAM-A than PCS scores for the somatic, F(1,36) = 21.30, p < 0.01, d = 0.65, cognitive, F(1,36) = 32.42, p < 0.01, d = 1.11, psychiatric, F(1,36) = 46.16, p < 0.01, d = 1.48, and insomnia symptoms, F(1,36) = 9.62, p < 0.01, d = 0.55. Planned contrasts also revealed significantly higher HAM-D than PCS scores for the somatic, F(1,36) = 80.77, p < 0.01, d = 1.64, cognitive, F(1,36) = 25.72, p < 0.01, d = 1.07, and psychiatric symptoms, F(1,36) = 36.39, p < 0.01, d = 1.27. There was not a significant difference in self-reported insomnia symptoms between the HAM-D and PCS. F(1,36) = 2.11, p = 0.16, d = 0.23. The initial ANOVA comparing HAM-D and HAM-A scores collected 9 days post-concussion in cleared athletes to PCS scores collected at the ImPACT prior to being cleared failed due to a lack of variation in the insomnia PCS scores, as no athletes reported insomnia symptoms at this time. Therefore, only the somatic, psychiatric, and cognitive domains were included in this analysis. There was a significant main effect of rating scale, F(2,16) = 7.68, p < 0.01. The main effect of symptom-domain was not significant, F(2,16) = 1.22, p = 0.32. There was a significant interaction of rating and symptom-domain, F(4,32) = 3.64, p = 0.02. Post hoc pairwise comparisons with Bonferroni correction showed that both HAM-A and HAM-D scores were higher than PCS scores (Fig. 1b). This difference was significant for the HAM-D (p = 0.05), but not for the HAM-A (p = 0.08). Planned contrasts showed that players reported significantly higher HAM-A than PCS psychiatric symptoms, F(1,8) = 9.19, p = 0.02, d = 1.34. The differences in self-reported HAM-A and PCS cognitive, F(1,8) = 4.58, p = 0.06, d = 1.04, and somatic scores, F(1,8) = 0.63, p = 0.45, d = 0.37, were not significant. Planned contrasts showed that HAM-D score were significantly higher than PCS scores for somatic symptoms, F(1,8) = 10.34, p = 0.01, d = 1.57, cognitive symptoms, F(1,8) = 7.07, p = 0.03, d = 1.30, and psychiatric symptoms, F(1,8) = 6.47, p = 0.04, d = 1.13. Self-reported symptoms on the HAM-D and HAM-A collected approximately 9 days following concussion were compared between cleared and not cleared athletes (Fig. 2). The main effect of group, F(1,27) = 1.20, p = 0.28, the main effect of rating scale, F(1,27) = 0.09, p = 0.77, and the main effect of symptom-domain, F(3,81) = 2.60, p = 0.06, were not significant. The interaction between symptom-domain and rating scale was significant, F(3,81) = 14.46, p < 0.01. However, there was not a significant interaction between group and rating scale, F(1,27) = 0.01, p = 0.94, or group and symptom-domain, F(3,81) = 0.72, p = 0.54. The three-way interaction between rating scale, symptom-domain, and group was not significant, F(3,81) = 0.75, p = 0.53. Thus, there were some differences in the number of symptoms reported by symptom-domain, but no differences in responses between cleared and not cleared athletes.
4. Discussion Despite the use of advanced computerized concussion assessment systems, concussion diagnosis still depends on self-reporting of psychiatric and somatic symptoms. Unfortunately, clinical decisions regarding return-to-play can be compromised due to patient underreporting of symptoms in these critical domains. Retrospective studies have demonstrated that youth ice hockey athletes and collegiate athletes from several sports underreport concussive events in order to prevent removal from play and due to lack of concussion awareness.9,21,22 Previous research has also found that self-reported post-concussive symptoms in collegiate athletes of
510
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511 Table 1 Percentage of cleared and not cleared athletes with at least mild symptoms at 1week post-concussion.
Fig. 2. There was not a significant difference in self-reported symptoms collected using the HAM-A (a) and HAM-D (b) 9 days post-concussion between cleared and not cleared athletes. Error bars represent standard deviation.
several sports, including football, are often not consistent with their diminished performance on cognitive tests.10–12 Consistent with previous retrospective studies, our data demonstrate that athletes underreport post-concussive symptoms to team medical staff in the days immediately following injury across the somatic, cognitive, and psychiatric symptom-domains. Furthermore, athletes cleared to return-to-play self-report more somatic, cognitive, and psychiatric symptoms 9 days post-concussion in a confidential setting than during testing immediately prior to being cleared. Critically, unlike the cognitive symptom domain, the psychiatric and portions of the somatic domains remain uniquely vulnerable to underreporting. Developing metrics to improve objective assessment in these domains would both reduce underreporting effectiveness and reduce the risk of premature return-to-play decisions. The potential for re-injury and increased risk of short- and long-term cognitive and affective dysregulation make it critical to overcome underreporting post-concussion symptoms by athletes. Unfortunately, the present results showed no differences in selfreported symptoms 9 days post-concussion between cleared and not cleared athletes. This result suggests that clinical judgments reliant on ImPACT neurocognitive testing, self-reports, and general clinical presentation may not accurately identify all symptomatic athletes. As depicted in Table 1, up to 60% of cleared athletes had at least mild symptoms in one of the symptom domains 9 days postconcussion. Thus, not only do athletes underreport post-concussive symptoms immediately following concussion, underreporting also results in premature return-to-play decisions for symptomatic athletes. Given our increasing knowledge of the short-term and longterm consequences of concussions, it is important to make clinical judgment regarding return-to-play using “true” metrics of symptoms across all symptom domains. The difference in confidentiality
Psychiatric
Cognitive
Somatic
Insomnia
HAM-A Not cleared (n = 20) Cleared (n = 9)
70 66.7
65 33.3
40 11.1
55 44.4
HAM-D Not cleared (n = 20) Cleared (n = 9)
75 55.6
75 44.4
75 66.7
45 33.3
between the clinical and research setting may partially account for the observed underreporting. Here, athletes were informed that information collected in the research setting was to remain confidential, and that no information would be shared with their athletic trainers, team officials, or teammates. A previous study using survey data of high school football players found that over 40% of athletes did not report a concussive event to prevent removal from play.3 The motivation to avoid being withheld from competition could explain some of the underreporting observed here, as the athletes were repeatedly informed that their responses collected in the research setting were confidential, and thus had no effect on their return-to-play decisions. Furthermore, the stigma associated with some of the post-concussive symptoms, especially the psychiatric symptoms, may prevent athletes from fully reporting the symptom severity in a non-confidential setting. The methods of self-report on the ImPACT test and in the structured HAM-D and HAM-A test may also account for differences in the degree of self-reporting. Items in the HAM-D and HAM-A scales have descriptive levels of severity, with specific examples to appropriately score each item and specific follow-up questions for the interviewer. In contrast, the ImPACT asks athletes to rate a single symptom as being absent, mild, moderate, or severe. This approach may allow for more unintentional underreporting as athletes are left to interpret these labels without behavioral anchors to guide them. One step in preventing the underreporting of post-injury symptoms is the education of players and medical professionals, including athletic trainers. When surveyed about the possible consequences of concussion, high school and collegiate football and soccer athletes readily highlight cognitive and somatic disturbances; however, psychiatric problems, such as changes in mood, irritability, anxiety, and depression are rarely mentioned.9,23 Here, over 50% of cleared athletes still had at least mild psychiatric symptoms 9 days post-concussion. It is possible that the increased mood symptoms were in response to the fact that the athletes had missed some competition time. However, psychiatric complaints are, in fact, common in traumatic brain injury regardless of injury severity.24,25 Educating players on the possible consequences of concussions, including psychiatric symptoms, and the risk associated with premature return-to-play could prevent some cases of underreporting. Underreporting of cognitive symptoms following sports-related concussion can be circumvented with the use of neurocognitive testing. However, no objective measures exist for psychiatric and somatic symptoms, so the assessment of these symptoms relies solely on self-reporting. Efforts have been made to develop objective measures for psychiatric symptoms, relying largely on the cognitive bias observed in depression, where depressed individuals are more likely to recall negatively valenced stimuli than positively valenced stimuli.26 Further validation of objective psychiatric measures is required to improve return-to-play decisions. The findings of this study should be interpreted within the limitations of the data. First, the total number of athletes included in
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
this study was relatively small for studies comparing self-reporting tendencies. However, this study was prospective in nature, examining actual concussive event as they occurred. Additionally, the observed effect sizes (Cohen’s d) for our planned comparisons were quite large. The power to detect an effect size of 0.5 with 37 participants is 0.84 for a two-tailed paired t-test. In this study we observed effect sizes as large as 1.64. Similarly, for the secondary analyses in cleared athletes, the power to detect an effect size of 1.1 with 9 participants is 0.83. Thus, the post hoc comparisons performed in this study have sufficient power, supporting the generalizability of the findings. 5. Conclusion This is the first prospective study to demonstrate that athletes significantly underreport post-concussive symptoms to their athletic trainers compared to symptoms reported in a confidential setting. Specifically, athletes underreported somatic, psychiatric, and cognitive symptoms in the days immediately following concussion. Underreporting was also observed over 1-week postconcussion in cleared athletes, suggesting that underreporting results in the clearing of symptomatic athletes. Finally, confidentially collected self-reported measures 9 days post-concussion did not differ between athletes who had and had not been cleared to return-to-play, implying that neurocognitive testing does not identify all symptomatic athletes. While neurocognitive testing can circumvent the problem of underreporting of cognitive symptoms, our finding is particularly concerning for psychiatric and somatic symptoms, as no validated objective measures currently exist. Educating players and medical professionals about the risks and signs of underreporting, as well as the development of more objective measures, are needed to assess symptom severity following sportsrelated concussion. 6. Practical implications • The supervisory nature of semi-structured interviews may make them more reliable than self-reported symptom scales for sportsrelated concussion management. • Clinicians must be knowledgeable about the prevalence of psychiatric symptoms following sports-related concussion, as they are dependent on self-report. • The development of objective measures for psychiatric and somatic symptoms would reduce premature return-to-play decisions. Acknowledgements The authors would like to thank the Laureate Institute for Brain Research psychiatric assessment team for conducting the structured interviews and Dr. Sheldon Preskorn, M.D. for his insight on this project. This research was conducted using internal funds from the Laureate Institute for Brain Research, which is supported by The William K. Warren Foundation.
511
References 1. McCrory P, Meeuwisse WH, Echemendia RJ et al. What is the lowest threshold to make a diagnosis of concussion? Br J Sports Med 2013; 47(5):268–271. 2. McCrory P, Meeuwisse WH, Aubry M et al. Consensus statement on concussion in sport: the 4th international conference on concussion in sport held in Zurich November 2012. Br J Sports Med 2013; 47(5):250–258. 3. McCrea M, Hammeke T, Olsen G et al. Unreported concussion in high school football players: implications for prevention. Clin J Sport Med 2004; 14(1):13–17. 4. Maroon JC, Lovell MR, Norwig J et al. Cerebral concussion in athletes: evaluation and neuropsychological testing. Neurosurgery 2000; 47(3):659–669, discussion 669–672. 5. Cernich A, Reeves D, Sun W et al. Automated neuropsychological assessment metrics sports medicine battery. Arch Clin Neuropsychol 2007; 22(Suppl. 1):S101–S114. 6. McCrory P. Sport concussion assessment tool 2. Scand J Med Sci Sports 2009; 19(3):452. 7. Giza CC, Kutcher JS, Ashwal S et al. Summary of evidence-based guideline update: evaluation and management of concussion in sports: report of the Guideline Development Subcommittee of the American Academy of Neurology. Neurology 2013; 80(24):2250–2257. 8. Delaney JS, Lacroix VJ, Leclerc S et al. Concussions among university football and soccer players. Clin J Sport Med 2002; 12(6):331–338. 9. Kaut KP, DePompei R, Kerr J et al. Reports of head injury and symptom knowledge among college athletes: implications for assessment and educational intervention. Clin J Sport Med 2003; 13(4):213–221. 10. Lovell MR, Solomon GS. Neurocognitive test performance and symptom reporting in cheerleaders with concussions. J Pediatr 2013; 163(4):1192–1195. 11. Van Kampen DA, Lovell MR, Pardini JE et al. The “value added” of neurocognitive testing after sports-related concussion. Am J Sports Med 2006; 34(10):1630–1635. 12. Fazio VC, Lovell MR, Pardini JE et al. The relation between post concussion symptoms and neurocognitive performance in concussed athletes. Neurorehabilitation 2007; 22(3):207–216. 13. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23:56–62. 14. Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959; 32(1):50–55. 15. Register F. In: Department of Health and Welfare, editor. Protection of human subjects: Belmont report – ethical principals and guidelines for the protection of human subjects of research, vol. 44. 1979, p. 23192–23197. 16. Lovell MR, Iverson GL, Collins MW et al. Measurement of symptoms following sports-related concussion: reliability and normative data for the post-concussion scale. Appl Neuropsychol 2006; 13(3):166–174. 17. Williams JB. A structured interview guide for the Hamilton Depression Rating Scale. Arch Gen Psychiatry 1988; 45(8):742–747. 18. Williams JBW, Terman M. Structured interview guide for the hamilton depression rating scale with atypical depression supplement (SIGH-ADS), New York, New York State Psychiatric Institute, 2003. 19. Shear MK, Vander Bilt J, Rucci P et al. Reliability and validity of a structured interview guide for the Hamilton Anxiety Rating Scale (SIGH-A). Depress Anxiety 2001; 13(4):166–178. 20. Lau B, Lovell MR, Collins MW et al. Neurocognitive and symptom predictors of recovery in high school athletes. Clin J Sport Med 2009; 19(3):216–221. 21. Williamson IJ, Goodman D. Converging evidence for the under-reporting of concussions in youth ice hockey. Br J Sports Med 2006; 40(2):128–132, discussion 128–132. 22. Llewellyn T, Burdette GT, Joyner AB. Concussion reporting rates at the conclusion of an intercollegiate athletic career. Clin J Sport Med 2014; 24(1):76–79. 23. Chrisman SP, Quitiquit C, Rivara FP. Qualitative study of barriers to concussive symptom reporting in high school athletics. J Adolesc Health 2013; 52(3):330–335, e333. 24. Konrad C, Geburek AJ, Rist F et al. Long-term cognitive and emotional consequences of mild traumatic brain injury. Psychol Med 2011; 41(6):1197–1211. 25. Kreutzer JS, Seel RT, Gourley E. The prevalence and symptom rates of depression after traumatic brain injury: a comprehensive examination. Brain Inj 2001; 15(7):563–576. 26. Ramanathan DM, Rabinowitz AR, Barwick FH et al. Validity of affect measurements in evaluating symptom reporting in athletes. J Int Neuropsychol Soc 2012; 18(1):101–107.
Contents lists available at ScienceDirect
Journal of Science and Medicine in Sport journal homepage: www.elsevier.com/locate/jsams
Original research
The underreporting of self-reported symptoms following sports-related concussion Timothy B. Meier a,∗ , Bradley J. Brummel b , Rashmi Singh a , Christopher J. Nerio c , David W. Polanski c , Patrick S.F. Bellgowan a,d a
Laureate Institute for Brain Research, Tulsa, OK, USA Department of Psychology, The University of Tulsa, Tulsa, OK, USA c Department of Athletics, The University of Tulsa, Tulsa, OK, USA d Faculty of Community Medicine, The University of Tulsa, Tulsa, OK, USA b
a r t i c l e
i n f o
Article history: Received 16 February 2014 Received in revised form 8 July 2014 Accepted 14 July 2014 Available online 24 July 2014 Keywords: Mild traumatic brain injury Post-concussion symptoms Neuropsychological testing
a b s t r a c t Objectives: This cohort study was conducted to examine patterns of symptom reporting in concussed athletes in two different testing environments. Design: A prospective cohort study was conducted with repeated measures. Methods: Self-reported symptoms collected by team athletic trainers using the ImPACT Post-Concussion Scale (PCS) were compared to symptoms collected in a confidential setting using structured interviews for depression and anxiety. Ratings were scaled to match scoring of the PCS and categorized into symptom-domains. Scores collected 2 days post-concussion were compared across different rating scales. Confidential self-report scores approximately 9 days post-concussion in cleared athletes were compared to PCS scores collected during return-to-play decisions. Finally, confidential self-report scores collected 9 days post-concussion were compared between cleared and not cleared athletes. Results: Athletes self-reported significantly fewer symptoms to team athletic trainers using the ImPACT test compared to self-reported symptoms collected in a confidential setting during the acute phase of concussion using standard psychiatric interviews. Athletes cleared to play continued to underreport symptoms 9 days post-concussion, particularly psychiatric symptoms. Finally, cleared athletes self-reported similar magnitude of symptoms than non-cleared athletes 9 days post-concussion in confidential research setting. Conclusions: The systematic underreporting of post-concussion symptoms may represent motivated behavior or differences in self-reporting data acquisition. By underreporting symptoms, many cleared athletes are still symptomatic over 1-week post-concussion. This study highlights the need for objective measures for somatic and psychiatric symptoms. © 2014 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
1. Introduction Concussion has a clinically heterogeneous symptomology that requires assessment across multiple symptom domains.1,2 Adding to diagnostic complexity, symptom severity is also modulated by “concussion modifiers” that include age, comorbidities, medication, style of play, and sport.2 The degree of symptom reporting is another modifier that can mislead clinical diagnoses, treatment
∗ Corresponding author. E-mail address: [email protected] (T.B. Meier).
strategies, and recovery decisions. Underreporting is at least partially motivated by a desire to continue playing, with awareness that self-reporting symptoms will prolong return-to-play decisions in high-school football players.3 In an attempt to standardize concussion diagnoses and to minimize clinical reliance on self-reported symptoms, diagnosis of sports-related concussion is often supplemented with computerized or onsite testing metrics.4–6 These metrics include objective measures such as neurocognitive testing; however, they also rely on athletes’ self-reported symptoms for psychiatric and some somatic symptoms. Thus, clinical judgment regarding return-to-play decisions may be biased toward symptoms assessed using objective measures obtained from neurological and
http://dx.doi.org/10.1016/j.jsams.2014.07.008 1440-2440/© 2014 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
508
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
neuropsychological examination. This may leave cleared athletes vulnerable to participating while still symptomatic in the psychiatric and somatic domains. International guidelines for return-to-play decisions following concussion have been established, despite the relative lack of evidence-based criteria.2,7 A graduated return-to-play protocol is recommended, with increasing physical activity occurring at each step after athletes are fully asymptomatic. Currently, there is no distinction regarding what specific symptoms should be favored over other symptoms during return-to-play decisions. Rather, the full resolution of the hallmark symptoms of concussion, which include cognitive, somatic, mood, and sleep-related symptoms, is considered. Typically, symptoms resolve within 7–10 days, though recovery can extend for longer periods of time.2 Previous retrospective studies have documented underreporting of concussive events in collegiate soccer, high school football, collegiate football, and a variety of collegiate women’s sports.3,8,9 In addition, underreporting of post-concussion symptoms compared to cognitive batteries has been reported in a variety of collegiate and high school sports, including football and cheerleading.10–12 This prospective study investigates the underreporting of postconcussive symptoms in the authentic athletic environment and in a controlled research environment. We compared self-reported symptoms collected with the ImPACT battery4 administered onsite at athletic facilities to scores on the Hamilton Anxiety Rating Scale (HAM-A)13 and the Hamilton Depression Rating Scale (HAM-D)14 collected in a confidential setting. First, we tested the hypothesis that athletes underreport cognitive, psychiatric, insomnia, and somatic symptoms in the days immediately post-injury. Secondly, we tested the hypothesis that self-reported symptoms collected during the last ImPACT test used in the decision making process to clear athletes to return-to-play are significantly lower than confidential depression and anxiety measures collected approximately 1-week post-concussion. Finally, we tested the hypothesis that athletes that have been cleared to return-to-play self-report fewer symptoms on depression and anxiety scales 1-week postconcussion than athletes that have not been cleared.
2. Methods A self-selection sampling method was applied to consecutive cases of athletes with concussion from a NCAA Division I university athletic program. Inclusion criteria included any collegiate-aged athlete with a clinician-diagnosed sports-related concussion. Athletes with other, non-concussion injuries that prevented return-to-play were excluded. All participants provided written consent. This study followed procedures outline in the Belmont Report for protection of human participants,15 and was approved by the Western Institutional Review Board and by the University of Oklahoma Health Sciences Center Institutional Review. Medical professionals trained in sports medicine diagnosed concussions based on clinical symptoms. Forty concussed athletes were evaluated following concussion. The average participant age at the time of their concussion was 20.44 years (SD = 1.41) and with 13.08 years of education (SD = 1.05). Three participants were female soccer players, two were male soccer players, four were female basketball players, one was a male basketball player, one was a female volleyball player, and the remaining 29 were male football players. Athletes who participated in this study in or after 2012 were compensated monetarily following NCAA Bylaws 16.11.1.11.1 and 16.11.1.11.12. Athletes who participated prior to 2012 could not be monetarily compensated due to NCAA Bylaw 16.02.3. Athletes completed the ImPACT battery at their athletic facilities under the supervision of athletic trainers following a cliniciandiagnosed concussion. The ImPACT battery includes a series of
cognitive tests comprising of four composite scores and a 22item post-concussion symptom inventory (PCS).4,16 This inventory asks participants to self-report post-concussion symptoms on a seven-point likert scale, with 0 indicating symptom absence, 1–2 indicating mild, 3–4 indicating moderate, and 5–6 indicating severe symptoms. Initial ImPACT testing was performed on an average of 1.30 days post-injury (SD = 0.57). Athletes performed additional post-injury ImPACT testing until they had been cleared to return-to-play by clinicians trained in sports-medicine. Return-toplay decisions were based on clinical judgment following medical exams, aided by the use of the ImPACT battery by comparing neurocognitive task performance and PCS scores to baseline levels. Athletes returned-to-play on an average of 11.77 (SD = 7.83) days following concussion. The average number of clinician-diagnosed concussions prior to the current concussion was 0.75 (SD = 1.0). Thirty-seven of the athletes also performed neuropsychiatric testing in a confidential setting at the Laureate Institute for Brain Research on an average of 1.92 days (SD = 1.04) post-injury. Athletes were informed that their responses in this setting would remain confidential and that team officials would not have access to their test results. Experienced assessment professionals performed structure interviews for the Hamilton Depression Rating Scale with Atypical Depression Supplement (HAM-D)13,17,18 and the Hamilton Anxiety Rating Scale (HAM-A).14,19 The original seventeen-item HAM-D with individual questions scored on either a three-point or five-point scale, and the standard fourteen item HAM-A with individual questions score on a five-points scale were used to document post-concussion symptoms. Twenty-nine athletes participated in a confidential research testing session on an average of 9.14 days post-concussion (SD = 2.08), during which HAM-D and HAM-A structured interviews were collected. At the time of this session, nine of the twentynine athletes had been cleared to return-to-play using the criteria previously described. Scaling of responses was performed to compare the HAM-A and HAM-D scales to the PCS scale based on the number of response options in the scale and the anchors for those response options prior to examining any of the participant scores on the scales. For HAM-A items and five-point items of the HAM-D scores of 4 were converted to 6, scores of 3 were converted to 5, scores of 2 were converted to 3.5, scores of 1 were converted to 1.5, and scores of 0 remained as 0. For three-point HAM-D items scaling to PCS scale was as follows: scores of 0 remained as 0, scores of 1 were converted to 2.5, and scores of 2 were converted to 5.5. Symptoms were grouped into four symptom-domains for analyses comparing the reporting of symptoms from the HAM-D and HAM-A interviews to the PCS score. PCS symptoms were classified into one of four domains following previous classification schemes 20 : Insomnia, psychiatric, cognitive, and somatic (Supplementary Table 1). Individual items from the HAM-D and HAM-A were also grouped into the same symptom-domains. Two of the authors independently classified HAM-D and HAM-A items into symptomdomains using both a criterion of maximum item inclusion and of best symptom fit to PCS items. We chose to use the best symptom fit approach. Items that covered two domains were included in both (Supplementary Table 1). Domain scores for the HAM-D, HAM-A, and PCS were obtained by averaging the scores of the individual items in each domain. Supplementary Table 1 can be found, in the online version, at doi:10.1016/j.jsams.2014.07.008. Two two-way repeated-measures analysis of variances (ANOVA) were performed with the within-subjects factors of symptom-domain (cognitive, insomnia, psychiatric, and somatic) and rating scale (HAM-D, HAM-A, PCS), on self-reported symptom-scores. One ANOVA compared the scores collected approximately two days post-concussion for all athletes, and one
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
Fig. 1. (a) The average domain scores for the scaled HAM-D and HAM-A were compared to the average domain scores for the PCS across all participants 1–2 days post-concussion. (b) The average domain scores for the scaled HAM-D and HAMA collected 9 days post-concussion were compared to PCS scores collected during the last ImPACT session prior to the return-to-play decision. Error bars represent standard deviation.
ANOVA compared cleared athletes HAM-D and HAM-A scores collected approximately 9 days post-concussion to PCS scores collected by the athletic trainers at the last ImPACT test prior to being cleared. For both ANOVAs, planned contrasts were performed comparing the HAM-D and HAM-A scores to PCS scores in each symptom-domain to directly test the hypothesis that athletes underreport post-concussive symptoms to team affiliated athletic trainers. A mixed-design repeated-measures ANOVA with the withinsubjects factors of symptom-domain and rating scale, and the between-subjects factor of group (cleared and not-cleared) was performed to test the hypothesis that domain-specific symptom severity 9 days post-concussion depends on whether or not athletes had been cleared to return-to-play. Post hoc analyses with Bonferroni correction for multiple comparisons effects were performed for each ANOVA when appropriate. Effect sizes, calculated as Cohen’s d, were computed for each planned contrast to demonstrate the magnitude of the observed differences.
3. Results There was a significant main effect of rating scale, F(2,72) = 33.53, p < 0.01, and symptom-domain, F(3,108) = 7.52, p < 0.01, on self-reported scores collected approximately two-days post-concussion (Fig. 1a). The interaction between rating scale and symptom-domain was also significant, F(6,216) = 10.35, p < 0.01.
509
Post hoc pairwise comparisons with Bonferroni correction showed that athletes reported fewer symptoms across all four symptom-domains in the PCS than in the HAM-A (p < 0.01) and HAM-D (p < 0.01). Planned contrasts revealed significantly greater HAM-A than PCS scores for the somatic, F(1,36) = 21.30, p < 0.01, d = 0.65, cognitive, F(1,36) = 32.42, p < 0.01, d = 1.11, psychiatric, F(1,36) = 46.16, p < 0.01, d = 1.48, and insomnia symptoms, F(1,36) = 9.62, p < 0.01, d = 0.55. Planned contrasts also revealed significantly higher HAM-D than PCS scores for the somatic, F(1,36) = 80.77, p < 0.01, d = 1.64, cognitive, F(1,36) = 25.72, p < 0.01, d = 1.07, and psychiatric symptoms, F(1,36) = 36.39, p < 0.01, d = 1.27. There was not a significant difference in self-reported insomnia symptoms between the HAM-D and PCS. F(1,36) = 2.11, p = 0.16, d = 0.23. The initial ANOVA comparing HAM-D and HAM-A scores collected 9 days post-concussion in cleared athletes to PCS scores collected at the ImPACT prior to being cleared failed due to a lack of variation in the insomnia PCS scores, as no athletes reported insomnia symptoms at this time. Therefore, only the somatic, psychiatric, and cognitive domains were included in this analysis. There was a significant main effect of rating scale, F(2,16) = 7.68, p < 0.01. The main effect of symptom-domain was not significant, F(2,16) = 1.22, p = 0.32. There was a significant interaction of rating and symptom-domain, F(4,32) = 3.64, p = 0.02. Post hoc pairwise comparisons with Bonferroni correction showed that both HAM-A and HAM-D scores were higher than PCS scores (Fig. 1b). This difference was significant for the HAM-D (p = 0.05), but not for the HAM-A (p = 0.08). Planned contrasts showed that players reported significantly higher HAM-A than PCS psychiatric symptoms, F(1,8) = 9.19, p = 0.02, d = 1.34. The differences in self-reported HAM-A and PCS cognitive, F(1,8) = 4.58, p = 0.06, d = 1.04, and somatic scores, F(1,8) = 0.63, p = 0.45, d = 0.37, were not significant. Planned contrasts showed that HAM-D score were significantly higher than PCS scores for somatic symptoms, F(1,8) = 10.34, p = 0.01, d = 1.57, cognitive symptoms, F(1,8) = 7.07, p = 0.03, d = 1.30, and psychiatric symptoms, F(1,8) = 6.47, p = 0.04, d = 1.13. Self-reported symptoms on the HAM-D and HAM-A collected approximately 9 days following concussion were compared between cleared and not cleared athletes (Fig. 2). The main effect of group, F(1,27) = 1.20, p = 0.28, the main effect of rating scale, F(1,27) = 0.09, p = 0.77, and the main effect of symptom-domain, F(3,81) = 2.60, p = 0.06, were not significant. The interaction between symptom-domain and rating scale was significant, F(3,81) = 14.46, p < 0.01. However, there was not a significant interaction between group and rating scale, F(1,27) = 0.01, p = 0.94, or group and symptom-domain, F(3,81) = 0.72, p = 0.54. The three-way interaction between rating scale, symptom-domain, and group was not significant, F(3,81) = 0.75, p = 0.53. Thus, there were some differences in the number of symptoms reported by symptom-domain, but no differences in responses between cleared and not cleared athletes.
4. Discussion Despite the use of advanced computerized concussion assessment systems, concussion diagnosis still depends on self-reporting of psychiatric and somatic symptoms. Unfortunately, clinical decisions regarding return-to-play can be compromised due to patient underreporting of symptoms in these critical domains. Retrospective studies have demonstrated that youth ice hockey athletes and collegiate athletes from several sports underreport concussive events in order to prevent removal from play and due to lack of concussion awareness.9,21,22 Previous research has also found that self-reported post-concussive symptoms in collegiate athletes of
510
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511 Table 1 Percentage of cleared and not cleared athletes with at least mild symptoms at 1week post-concussion.
Fig. 2. There was not a significant difference in self-reported symptoms collected using the HAM-A (a) and HAM-D (b) 9 days post-concussion between cleared and not cleared athletes. Error bars represent standard deviation.
several sports, including football, are often not consistent with their diminished performance on cognitive tests.10–12 Consistent with previous retrospective studies, our data demonstrate that athletes underreport post-concussive symptoms to team medical staff in the days immediately following injury across the somatic, cognitive, and psychiatric symptom-domains. Furthermore, athletes cleared to return-to-play self-report more somatic, cognitive, and psychiatric symptoms 9 days post-concussion in a confidential setting than during testing immediately prior to being cleared. Critically, unlike the cognitive symptom domain, the psychiatric and portions of the somatic domains remain uniquely vulnerable to underreporting. Developing metrics to improve objective assessment in these domains would both reduce underreporting effectiveness and reduce the risk of premature return-to-play decisions. The potential for re-injury and increased risk of short- and long-term cognitive and affective dysregulation make it critical to overcome underreporting post-concussion symptoms by athletes. Unfortunately, the present results showed no differences in selfreported symptoms 9 days post-concussion between cleared and not cleared athletes. This result suggests that clinical judgments reliant on ImPACT neurocognitive testing, self-reports, and general clinical presentation may not accurately identify all symptomatic athletes. As depicted in Table 1, up to 60% of cleared athletes had at least mild symptoms in one of the symptom domains 9 days postconcussion. Thus, not only do athletes underreport post-concussive symptoms immediately following concussion, underreporting also results in premature return-to-play decisions for symptomatic athletes. Given our increasing knowledge of the short-term and longterm consequences of concussions, it is important to make clinical judgment regarding return-to-play using “true” metrics of symptoms across all symptom domains. The difference in confidentiality
Psychiatric
Cognitive
Somatic
Insomnia
HAM-A Not cleared (n = 20) Cleared (n = 9)
70 66.7
65 33.3
40 11.1
55 44.4
HAM-D Not cleared (n = 20) Cleared (n = 9)
75 55.6
75 44.4
75 66.7
45 33.3
between the clinical and research setting may partially account for the observed underreporting. Here, athletes were informed that information collected in the research setting was to remain confidential, and that no information would be shared with their athletic trainers, team officials, or teammates. A previous study using survey data of high school football players found that over 40% of athletes did not report a concussive event to prevent removal from play.3 The motivation to avoid being withheld from competition could explain some of the underreporting observed here, as the athletes were repeatedly informed that their responses collected in the research setting were confidential, and thus had no effect on their return-to-play decisions. Furthermore, the stigma associated with some of the post-concussive symptoms, especially the psychiatric symptoms, may prevent athletes from fully reporting the symptom severity in a non-confidential setting. The methods of self-report on the ImPACT test and in the structured HAM-D and HAM-A test may also account for differences in the degree of self-reporting. Items in the HAM-D and HAM-A scales have descriptive levels of severity, with specific examples to appropriately score each item and specific follow-up questions for the interviewer. In contrast, the ImPACT asks athletes to rate a single symptom as being absent, mild, moderate, or severe. This approach may allow for more unintentional underreporting as athletes are left to interpret these labels without behavioral anchors to guide them. One step in preventing the underreporting of post-injury symptoms is the education of players and medical professionals, including athletic trainers. When surveyed about the possible consequences of concussion, high school and collegiate football and soccer athletes readily highlight cognitive and somatic disturbances; however, psychiatric problems, such as changes in mood, irritability, anxiety, and depression are rarely mentioned.9,23 Here, over 50% of cleared athletes still had at least mild psychiatric symptoms 9 days post-concussion. It is possible that the increased mood symptoms were in response to the fact that the athletes had missed some competition time. However, psychiatric complaints are, in fact, common in traumatic brain injury regardless of injury severity.24,25 Educating players on the possible consequences of concussions, including psychiatric symptoms, and the risk associated with premature return-to-play could prevent some cases of underreporting. Underreporting of cognitive symptoms following sports-related concussion can be circumvented with the use of neurocognitive testing. However, no objective measures exist for psychiatric and somatic symptoms, so the assessment of these symptoms relies solely on self-reporting. Efforts have been made to develop objective measures for psychiatric symptoms, relying largely on the cognitive bias observed in depression, where depressed individuals are more likely to recall negatively valenced stimuli than positively valenced stimuli.26 Further validation of objective psychiatric measures is required to improve return-to-play decisions. The findings of this study should be interpreted within the limitations of the data. First, the total number of athletes included in
T.B. Meier et al. / Journal of Science and Medicine in Sport 18 (2015) 507–511
this study was relatively small for studies comparing self-reporting tendencies. However, this study was prospective in nature, examining actual concussive event as they occurred. Additionally, the observed effect sizes (Cohen’s d) for our planned comparisons were quite large. The power to detect an effect size of 0.5 with 37 participants is 0.84 for a two-tailed paired t-test. In this study we observed effect sizes as large as 1.64. Similarly, for the secondary analyses in cleared athletes, the power to detect an effect size of 1.1 with 9 participants is 0.83. Thus, the post hoc comparisons performed in this study have sufficient power, supporting the generalizability of the findings. 5. Conclusion This is the first prospective study to demonstrate that athletes significantly underreport post-concussive symptoms to their athletic trainers compared to symptoms reported in a confidential setting. Specifically, athletes underreported somatic, psychiatric, and cognitive symptoms in the days immediately following concussion. Underreporting was also observed over 1-week postconcussion in cleared athletes, suggesting that underreporting results in the clearing of symptomatic athletes. Finally, confidentially collected self-reported measures 9 days post-concussion did not differ between athletes who had and had not been cleared to return-to-play, implying that neurocognitive testing does not identify all symptomatic athletes. While neurocognitive testing can circumvent the problem of underreporting of cognitive symptoms, our finding is particularly concerning for psychiatric and somatic symptoms, as no validated objective measures currently exist. Educating players and medical professionals about the risks and signs of underreporting, as well as the development of more objective measures, are needed to assess symptom severity following sportsrelated concussion. 6. Practical implications • The supervisory nature of semi-structured interviews may make them more reliable than self-reported symptom scales for sportsrelated concussion management. • Clinicians must be knowledgeable about the prevalence of psychiatric symptoms following sports-related concussion, as they are dependent on self-report. • The development of objective measures for psychiatric and somatic symptoms would reduce premature return-to-play decisions. Acknowledgements The authors would like to thank the Laureate Institute for Brain Research psychiatric assessment team for conducting the structured interviews and Dr. Sheldon Preskorn, M.D. for his insight on this project. This research was conducted using internal funds from the Laureate Institute for Brain Research, which is supported by The William K. Warren Foundation.
511
References 1. McCrory P, Meeuwisse WH, Echemendia RJ et al. What is the lowest threshold to make a diagnosis of concussion? Br J Sports Med 2013; 47(5):268–271. 2. McCrory P, Meeuwisse WH, Aubry M et al. Consensus statement on concussion in sport: the 4th international conference on concussion in sport held in Zurich November 2012. Br J Sports Med 2013; 47(5):250–258. 3. McCrea M, Hammeke T, Olsen G et al. Unreported concussion in high school football players: implications for prevention. Clin J Sport Med 2004; 14(1):13–17. 4. Maroon JC, Lovell MR, Norwig J et al. Cerebral concussion in athletes: evaluation and neuropsychological testing. Neurosurgery 2000; 47(3):659–669, discussion 669–672. 5. Cernich A, Reeves D, Sun W et al. Automated neuropsychological assessment metrics sports medicine battery. Arch Clin Neuropsychol 2007; 22(Suppl. 1):S101–S114. 6. McCrory P. Sport concussion assessment tool 2. Scand J Med Sci Sports 2009; 19(3):452. 7. Giza CC, Kutcher JS, Ashwal S et al. Summary of evidence-based guideline update: evaluation and management of concussion in sports: report of the Guideline Development Subcommittee of the American Academy of Neurology. Neurology 2013; 80(24):2250–2257. 8. Delaney JS, Lacroix VJ, Leclerc S et al. Concussions among university football and soccer players. Clin J Sport Med 2002; 12(6):331–338. 9. Kaut KP, DePompei R, Kerr J et al. Reports of head injury and symptom knowledge among college athletes: implications for assessment and educational intervention. Clin J Sport Med 2003; 13(4):213–221. 10. Lovell MR, Solomon GS. Neurocognitive test performance and symptom reporting in cheerleaders with concussions. J Pediatr 2013; 163(4):1192–1195. 11. Van Kampen DA, Lovell MR, Pardini JE et al. The “value added” of neurocognitive testing after sports-related concussion. Am J Sports Med 2006; 34(10):1630–1635. 12. Fazio VC, Lovell MR, Pardini JE et al. The relation between post concussion symptoms and neurocognitive performance in concussed athletes. Neurorehabilitation 2007; 22(3):207–216. 13. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23:56–62. 14. Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959; 32(1):50–55. 15. Register F. In: Department of Health and Welfare, editor. Protection of human subjects: Belmont report – ethical principals and guidelines for the protection of human subjects of research, vol. 44. 1979, p. 23192–23197. 16. Lovell MR, Iverson GL, Collins MW et al. Measurement of symptoms following sports-related concussion: reliability and normative data for the post-concussion scale. Appl Neuropsychol 2006; 13(3):166–174. 17. Williams JB. A structured interview guide for the Hamilton Depression Rating Scale. Arch Gen Psychiatry 1988; 45(8):742–747. 18. Williams JBW, Terman M. Structured interview guide for the hamilton depression rating scale with atypical depression supplement (SIGH-ADS), New York, New York State Psychiatric Institute, 2003. 19. Shear MK, Vander Bilt J, Rucci P et al. Reliability and validity of a structured interview guide for the Hamilton Anxiety Rating Scale (SIGH-A). Depress Anxiety 2001; 13(4):166–178. 20. Lau B, Lovell MR, Collins MW et al. Neurocognitive and symptom predictors of recovery in high school athletes. Clin J Sport Med 2009; 19(3):216–221. 21. Williamson IJ, Goodman D. Converging evidence for the under-reporting of concussions in youth ice hockey. Br J Sports Med 2006; 40(2):128–132, discussion 128–132. 22. Llewellyn T, Burdette GT, Joyner AB. Concussion reporting rates at the conclusion of an intercollegiate athletic career. Clin J Sport Med 2014; 24(1):76–79. 23. Chrisman SP, Quitiquit C, Rivara FP. Qualitative study of barriers to concussive symptom reporting in high school athletics. J Adolesc Health 2013; 52(3):330–335, e333. 24. Konrad C, Geburek AJ, Rist F et al. Long-term cognitive and emotional consequences of mild traumatic brain injury. Psychol Med 2011; 41(6):1197–1211. 25. Kreutzer JS, Seel RT, Gourley E. The prevalence and symptom rates of depression after traumatic brain injury: a comprehensive examination. Brain Inj 2001; 15(7):563–576. 26. Ramanathan DM, Rabinowitz AR, Barwick FH et al. Validity of affect measurements in evaluating symptom reporting in athletes. J Int Neuropsychol Soc 2012; 18(1):101–107.
