- Email: [email protected]
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic Geoffrey L. Heyer, MD1,2, Caroline E. Schaffer, BS3, Sean C. Rose, MD1,2, Julie A. Young, ATC4, Kelly A. McNally, PhD5,6, and Anastasia N. Fischer, MD4 Objective To identify the clinical factors that influence the duration of postconcussion symptoms among youth referred to a sports concussion clinic.
Study design A retrospective cohort study was conducted to evaluate several potential predictors of symptom duration via a Cox proportional hazards analyses. The individual postconcussion symptom scores were highly correlated, so these symptoms were analyzed in the statistical model as coefficients derived from principal component analyses. Results Among 1953 youth with concussion, 1755 (89.9%) had dates of reported symptom resolution. The remainder (10.1%) were lost to follow-up and censored. The median time to recovery was 18 days (range 1-353 days). By 30 days, 72.6% had recovered; by 60 days, 91.4% had recovered; and by 90 days, 96.8% had recovered. Several variables in a multivariate Cox model predicted postconcussion symptom duration: female sex (P < .001, hazard ratio [HR] = 1.28), continued activity participation (P = .02, HR = 1.13), loss of consciousness (P = .03, HR = 1.18), anterograde amnesia (P = .04, HR = 1.15), premorbid headaches (P = .03, HR = 1.15), symptom components from the day of concussion (emotion, P = .03, HR = 1.08), and the day of clinic evaluation (cognitivefatigue, P < .001, HR = 1.22; cephalalgic, P < .001, HR = 1.27; emotional, P = .05, HR = 1.08; arousal-stimulation, P = .003, HR = 1.1). In univariate analyses, greater symptom scores generally predicted longer symptom durations. Worsening of symptoms from the day of concussion to the day of clinic evaluation also predicted longer recovery (P < .001, HR = 1.59). Conclusions Several factors help to predict protracted postconcussion symptom durations among youth referred to a sports concussion clinic. (J Pediatr 2016;-:---). See editorial, p and related articles, p and p
C
oncussions are common among children, resulting in approximately 144 000 or more visits to the emergency department annually in the US.1 Many more youth with concussion are treated solely by primary care providers, outpatient specialists, or athletic trainers, or they do not seek medical care.2,3 Recovery after concussion is poorly understood. Studies in adults suggest that most patients recover within 7-10 days,4-6 but children may be more vulnerable to the concussion effects and may have longer durations of postconcussion symptoms.7 For example, high school athletes generally recover more slowly than collegiate and professional athletes.8,9 When a variety of study designs are considered, an estimated 2.8%-43% of youth continue to report postconcussion symptoms at 3 months.10-17 Inconsistent, and at times contradictory, evidence exists linking clinical factors with protracted postconcussion symptoms in these patients.18 Potential predictors include overall symptom burden,11,13,14,19-22 specific symptoms,12,23-28 previous concussions,14,17,29,30 loss of consciousness (LOC),13,22,31 absence of LOC,14 post-traumatic amnesia,11,13,22 age,12,14,15 female sex,15,32 premorbid symptoms,17,30,32-34 and hospital admission.12,31 We sought to identify factors that influence the duration of postconcussion symptoms among youth referred to a sports concussion clinic. We expand on current knowledge by evaluating previous and novel predictors from a large representative sample and determining their relative importance in an inclusive multivariate survival model. Understanding From the Division of Pediatric Neurology, Nationwide which factors lead to prolonged recovery can improve overall concussion manChildren’s Hospital; Department of Neurology, The Ohio State University, Columbus, OH; Department of Health agement by informing early subspecialty referral decisions, optimizing resource Sciences, Clemson University, Clemson, SC; Divisions of utilization, and helping clinicians to provide anticipatory guidance. Sports Medicine, and Pediatric Psychology and 1
2
3
4
5
Neuropsychology, Nationwide Children’s Hospital; and 6 Department of Pediatrics, The Ohio State University, Columbus, OH
HR LOC PCA
Hazard ratio Loss of consciousness Principal component analysis
The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2016.03.014
1
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Methods We conducted a retrospective cohort study of pediatric patients presenting to a sports concussion clinic at a large Midwestern children’s hospital between June 2012 and September 2014. The study was approved by the Institutional Review Board at Nationwide Children’s Hospital. Electronic medical records and clinic charting were analyzed. Concussion diagnoses were determined clinically for each patient, at the discretion of the managing physician, on the basis of the injury mechanism(s), history, and examination, and generally followed the operational definition of concussion set forth in the Consensus Statement on Concussion in Sport.35 Study inclusion criteria were age from 10 to 19 years and a concussion injury occurring within 30 days of clinic evaluation. The age range represents the referral limits of the clinic. The interval of 1-30 days between injury and evaluation was chosen arbitrarily to allow a balanced analysis within a representative clinic sample while minimizing the numbers of individuals with distant injuries and potentially excessive recall biases. Exclusion criteria were the absence of traumatic brain injury, clinical features consistent with moderate or severe traumatic brain injury, and inability or refusal to complete the postconcussion symptom questionnaire. When patients sustained more than 1 concussion during the 28-month study period, only the data that corresponded with the first documented injury were analyzed. None of the patients in this study had an acute concussion that overlapped with recovery of a previous concussion. As a standard component of the concussion clinic evaluation, all patients completed a written symptom questionnaire on arrival to the clinic but before their medical evaluations. The questionnaire addressed the presence and severity of 23 common postconcussion symptoms (Table I; available at www.jpeds.com), rated on a scale of 0 (not present) to 6 (severe). Separate scales were completed for current symptoms on the day of clinic evaluation and recalled symptoms from the day of concussion. Our questionnaire has not been validated. It was adapted from previously published symptom assessments.36-38 In addition, patients and their parents answered basic questions related to the concussion, including the date of injury, mechanism of injury, the presence of LOC, the presence of anterograde and retrograde amnesia, whether the patient continued participating in the activity immediately after injury, the number of previous concussions, and the presence of premorbid headaches. Families self-reported race and ethnicity. Managing physician, insurance status, and clinic follow-up appointment dates were extracted from the electronic clinical records. As a standard component of our charting template, patients were asked at each consecutive clinic visit whether postconcussion symptoms had resolved and, if so, when. Resolution of symptoms in this study was defined as patient report of absent symptoms, return to premorbid symptom status, or symptom improvement 2
Volume allowing return to play and without planned follow-up. When resolution occurred within 24 hours of injury, such was designated as day 1 (rather than day 0). Dates of symptom resolution and cancelled/missed appointments were also extracted from the clinical record. Statistical Analyses Descriptive statistics were calculated for demographic and concussion data. The c2 test was used to compare categorical variables, and the Student t test or the Mann-Whitney U test was used to compare continuous variables that were related to the assessment of clinical predictors. A multivariate Cox proportional hazards analysis was used to explore which factors influence the duration of postconcussion symptoms. The dependent variable was time to symptom resolution. When clinic follow-up appointments were missed or cancelled, and the date of symptom resolution was not available, the subject was censored on the date of the earliest missed appointment. All patients who completed follow-up had dates of symptom resolution. To address multicollinearity among postconcussion symptoms that are highly correlated, the 23 symptom scores from each time period (day of concussion and day of clinic evaluation) were reduced to 6 symptom components via a principal component analysis (PCA) with oblique rotation (Promax, kappa = 4).39 The PCA-derived principal components (symptom clusters) are listed in Table I; each cluster was labeled by us according to the symptoms it comprised. As expected, some differences existed between symptoms reported on the day of concussion and the day of clinic evaluation, but the resulting clusters represent logical patterns of symptom groupings for both analyses. The Bartlett method was used to estimate PCA score coefficients for the Cox analysis. Given the inconsistent results from previous studies, we entered predictor variables in hierarchical stages to explore the relative effects of covariates: (1) demographic information and clinic information; (2) concussion characteristics and premorbid headaches; (3) and then the principal component data (symptom clusters) on the day of concussion and the day of clinic evaluation. Interaction terms were not analyzed. The corresponding hazard ratios (HRs) were inverted to reflect the risk of continued postconcussion symptoms (rather than symptom resolution). In addition, univariate Cox proportional hazards analyses were used to determine whether total symptom scores at the time of concussion or the time of clinic evaluation influenced postconcussion symptom duration. The sums of the 23 symptom scores were divided by quartile (Table II), and pairs of consecutive quartiles were compared. We also calculated the differences between scores at clinic evaluation and at concussion (symptom score on clinic day minus symptom score on concussion day) and compared symptom durations among patients with positive and negative differences. A positive difference means that symptoms worsened from the day of concussion to the day Heyer et al
- 2016
ORIGINAL ARTICLES
Table II. Demographic and concussion data Characteristics
n
Data
Age in years, mean (SD) Male sex Ethnicity Non-Hispanic Hispanic/Latino Multiethnic Family declined Race White Black Asian Biracial/multiracial Other Family declined Insurance status Private Public Previous concussions 0 1 2 3 4 or more Mechanism of injury Head to body, ground, or equipment Helmet to body, ground, or equipment Other Don’t know Continued activity participation after concussion LOC Retrograde amnesia Anterograde amnesia Premorbid headaches Mean symptom score, day of concussion (SD) Concussion-day score, 25th percentile Concussion-day score, 50th percentile Concussion-day score, 75th percentile Mean symptom score, day of clinic (SD) Clinic-day score, 25th percentile Clinic-day score, 50th percentile Clinic-day score, 75th percentile
1953 1953 1953
14.1 (2.1) 63.4% 94.3% 2.8% 0.4% 0.4%
1923 80.6% 12.8% 0.4% 3.4% 1.9% 0.2% 1942 80.2% 19.8% 1953 68.9% 21.2% 6.0% 2.7% 1.2% 1953
1922 1923 1938 1935 1940 1953
1953
52.5% 26.4% 11.6% 9.4% 42.6% 16.3% 21.0% 20.3% 21.4% 42.2 (24.4) 22 40 59 21.5 (21.7) 4 15 34
of clinic visit. A Bonferroni correction for multiple testing was applied directly to all univariate P values so that a constant alpha value of .05 could be used across analyses. Kaplan-Meier plots were constructed to illustrate differences in symptom durations. All statistical analyses were performed using SPSS Version 21 (SPSS Inc, Chicago, Illinois). The significance threshold was set at 5%.
Results Our cohort comprised 1953 youth with concussion. Demographic and concussion data are summarized in Table II. The mean interval between concussion injury and the initial clinical evaluation was 10.1 days (median 9 days, SD 6.32). The majority of patients had no previous concussions (68.9%) or 1 previous concussion (21.2%); a very small proportion reported $4 previous concussions (1.2%). Among patients with documented postconcussion symptom recovery, defined as symptom resolution (n = 1755, 89.9%), the median time to recovery was
18 days (range 1-353 days). By 30 days, 72.6% had recovered; by 60 days, 91.4% had recovered; and by 90 days, 96.8% had recovered. Patients lost to follow-up (n = 198, 10.1%) had a median time to their missed or cancelled appointment of 36 days, which differed from the median time to recovery among those with known symptom resolution (P < .001). These patients also reported greater total symptom scores on the day of concussion (mean 50.7 vs 41.2; P = .003) and at the day of clinic evaluation (mean 32 vs 20.4; P < .001) than those not lost to follow-up. Several variables predicted postconcussion symptom duration. Table III shows the results from the full multivariate Cox model. Female sex had a significant effect when added in the first stage of the hierarchical model (P < .001; HR 1.4; 95% CI 1.26-1.55). This effect remained significant in the full model, in which being female increased the risk of protracted concussion symptoms by 28%. Outcomes did not differ in relation to age, ethnicity, insurance status, or the managing physician. Continuing activity participation after injury, LOC, anterograde amnesia, and premorbid headaches all contributed to the prediction of prolonged recovery after concussion, with HR effect sizes ranging from 1.13 to 1.18. LOC was not significant when initially added in stage 2 of the hierarchical model, but it became significant when the symptom clusters were added in the final stage. The symptoms reported on the day of clinic evaluation had greater predictive value for overall recovery than the symptom scores recalled from the day of concussion. The emotional-symptoms cluster (sadness, feeling more emotional, nervousness, and irritability) was the only group of symptoms on concussion day that significantly predicted symptom duration, increasing the risk of protracted symptom duration by 8%. In contrast, 4 groups of symptoms reported on the day of clinic evaluation predicted protracted recovery: cognitive-fatigue, cephalalgic, arousal-stimulation, and emotional. Analyses of total symptom scores demonstrated that each consecutive quartile of scores led to longer symptom durations, with the exception of quartile 1 vs quartile 2 scores recalled from the day of concussion (Table IV and Figure; Figure available at www.jpeds.com). The effects of these changes (in univariate analyses) were especially prominent on the day of clinic evaluation with large effect sizes (HRs 1.48-1.81). Notably, the rate of cancelled and missed appointments increased in kind with the increases in reported symptom burden (Table IV). Lastly, when we compared symptom scores from concussion day and clinic day, 228 patients (11.7%) reported greater scores (ie, worse symptoms) on clinic day. Interval worsening of symptoms was associated with an increase in the risk of protracted recovery by 59% (in univariate analysis), Table IV. This finding was not related to a difference in evaluation times; the mean intervals between concussion day and clinic day did not differ between groups (mean of 10.1 days vs mean of 10 days, P = .81). Patients with worsening of symptoms recalled lower mean symptom
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
3
THE JOURNAL OF PEDIATRICS
Volume -
www.jpeds.com
Table III. Predictors of protracted concussion recovery based on the full statistical model Predictors
P value
HR*
95% CI
Age Female sex Ethnicity Insurance status Attending physician Previous concussions 0 1 2 3 4 or more Mechanism of injury Don’t know Head to body, ground, or equipment Helmet to body, ground, or equipment Other Continued participation LOC Anterograde amnesia Retrograde amnesia Premorbid headaches Day of concussion symptom clustersz Cognitive-fatigue component Cephalalgic component Emotional component Somatic component Arousal-stimulation component Vomiting component Day of clinic symptom clustersz Cognitive-fatigue component Cephalalgic component Emotional component Somatic component Arousal-stimulation component Vomiting component
.16 <.001† NS .18 NS
1.02 1.28
0.99-1.04 1.14-1.44
0.92
0.81-1.04
Reference .72 .7 .12 .59
0.98 1.04 1.27 1.14
0.87-1.1 0.85-1.28 0.94-1.71 0.71-1.83
Reference .72 .07 .92 .02† .03† .04† .89 .03†
0.98 1.2 0.99 1.13 1.18 1.15 0.99 1.15
0.87-1.1 0.99-1.46 0.79-1.23 1.02-1.25 1.02-1.37 1.01-1.31 0.87-1.13 1.02-1.29
.09 .72 .03† .24 .38 .06
1.07 1.01 1.08 1.04 0.98 0.95
0.99-1.17 0.94-1.09 1.01-1.16 0.97-1.1 0.92-1.04 0.9-1.01
<.001† <.001† .05† .09 .003† .54
1.22 1.27 1.08 0.95 1.1 1.02
1.11-1.33 1.18-1.37 1.01-1.16 0.89-1.01 1.03-1.17 0.96-1.08
NS, not significant. *HRs inverted to represent probabilities of continued symptoms, ie, exp(B)1. †Statistically significant. zPrincipal components from symptom scores on concussion day and clinic day (Table I).
scores from the day of concussion (31.7 vs 43.6, P < .001) and reported greater mean scores at the day of clinic evaluation (46.8 vs 18.2, P < .001).
Discussion In a large cohort of youth with concussion, several factors influenced postconcussion symptom duration: female sex,
continued activity participation, LOC, anterograde amnesia, premorbid headaches, overall symptom burden, worsening of symptoms over time, and specific types of symptoms on the day of concussion and the day of clinic evaluation. The present study assessed the clinical factors that predict postconcussion symptom duration among youth referred to a sports concussion clinic. The available evidence from previous studies has been inconsistent about which clinical features of concussion injury lead to rapid resolution of symptoms. When assessed, greater symptom burden (ie, more overall symptoms or greater symptom ratings) during the acute concussion period has been the most consistent predictor of longer symptom durations.11,13,14,19-22 For example, Meehan et al19 categorized symptom duration as #28 days and >28 days and found that total symptom score was the only factor independently associated with symptoms lasting longer than 28 days. Barlow et al13 created subcategories of injury severity that incorporated symptom reporting, LOC, and amnesia. Each stratum of increasing severity led to roughly corresponding increases in symptom duration. Similarly, symptom burden in our study was a significant predictor of recovery time, whether measured as total symptom scores compared across quartiles or specific symptom clusters analyzed in a broadly inclusive survival model. Unfortunately, the high correlations among all postconcussion symptoms prevented us from analyzing each symptom independently. The cephalalgic and cognitive-fatigue clusters at the time of clinic evaluation had the largest effect sizes. These findings resemble those of previous studies in which acute postconcussion headache, dizziness (part of our clinic-day cephalalgic cluster), and the symptom of fogginess predicted protracted recovery.12,23-28 Patients who reported premorbid headaches in our cohort also had a greater probability of prolonged recovery. Emotional symptoms on concussion day and on clinic day each had predictive value in our study. Because we did not assess premorbid emotional symptoms, we do not know whether noninjury factors contributed to these postconcussion effects. While analyzing the associations between symptom score quartiles and symptom duration, we found that patients with greater overall reported symptom burdens were also lost to follow up in higher numbers. It is not clear why this would occur, but it may be related to fatigue from the longer recovery times and
Table IV. Postconcussion symptom durations differ according to initial symptom scores Predictors of concussion duration z
Recalled symptom scores from day of concussion, Q1 vs Q2 Recalled symptom scores from day of concussion, Q2 vs Q3 Recalled symptom scores from day of concussion, Q3 vs Q4 Symptom scores on day of clinic evaluation, Q1 vs Q2 Symptom scores on day of clinic evaluation, Q2 vs Q3 Symptom scores on day of clinic evaluation, Q3 vs Q4 Clinic day symptoms worse than concussion day symptoms
First comparator, n
Total, n
Censored (%), total
P value*
HR†
95% CI
450 524 485 475 478 508 228
974 1009 979 953 986 1000 1953
8.2 8.2 12.1 6.8 8.4 13.3 10.1
.29 .002 <.001 <.001 <.001 <.001 <.001
1.15 1.27 1.36 1.81 1.59 1.48 1.59
1.01-1.31 1.11-1.44 1.19-1.56 1.58-2.06 1.39-1.82 1.29-1.7 1.37-1.85
Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile. *Bonferroni correction applied. †HRs inverted to represent probabilities of continued symptoms, ie, exp(B)1. zQ1 vs Q2 refers to symptom score.
4
Heyer et al
- 2016 multiple follow-up visits. Noninjury factors also may have played a role. The variations in study designs have made comparisons across pediatric studies difficult. Female sex had a large effect in our study but was found to be predictive in only 2 previous studies.15,32 In contrast, previous head injuries14,17,29,30 and age12,14,15 each influenced recovery times in other studies but were not significant in our model. Investigators have published conflicting results about the effects of LOC. Eisenberg et al14 found that LOC was protective against protracted concussion recovery, and other groups demonstrated its negative effects.13,22,31 In our analysis, LOC was not significant when entered initially (stage 2) in the statistical model, but this changed with the addition of symptom clusters in the final model (stage 3), highlighting the potential complexities related to comparisons across differently structured studies. Not all postconcussion symptoms develop immediately after injury; some can take several hours or several days to emerge.35,36 A group of patients in our cohort reported greater symptom scores on the day of clinic evaluation than on the day of concussion, which was not related to the timing of the initial clinic visit. Compared with the patients who improved during the acute concussion period, those who had this positive difference in scores, or “positive delta,” also had a greater probability of protracted postconcussion symptoms. The factors related to the worsening of symptoms during the week or so after concussion have not been well characterized among pediatric or adult patients. Morgan et al30 found that delayed symptom onset of at least three hours following injury was associated with prolonged symptom durations, but worsening of symptoms over a longer period was not assessed. It is possible that our patients with interval worsening of symptoms had different activity levels than those who did better. Symptom burden is important in predicting concussion duration, yet comparison of absolute symptom scores across studies requires that identical instruments be used. The presence or absence of a positive delta, theoretically, could be determined from a variety of postconcussion symptom scales, and it does not require knowledge of premorbid symptoms. Given its predictive value in the present study, we encourage further investigation of the positive delta. We acknowledge several limitations of our study. First, our patients were highly selected. Selection factors included referral to a concussion clinic and postconcussion symptom duration of up to 30 days at the time of initial evaluation. Although the patients in this study may be representative of those seen in tertiary-care concussion clinics, the symptom durations in our cohort generally are not representative of all concussions. Indeed, the reason for referral of some patients was persistence of symptoms beyond one week. Second, we relied on patient report for the determination of resolution of symptoms. Patient report is a meaningful outcome. However, even though objective measures (eg, neurocognitive testing and balance measures) were used as clinically indicated, we did not incorporate their results in this study
ORIGINAL ARTICLES because of variations in timing and use of each objective test. In addition, at the time of reported return to symptom baseline not all patients had returned to full activity or unrestricted play, which is a more practical definition of concussion resolution. Some patients were near their premorbid symptom baseline when reporting symptom resolution, which underestimates overall recovery times. Third, the questionnaire used has not been validated. Finally, postconcussion symptoms are nonspecific. Some patients may have incorrectly attributed baseline symptoms as postconcussion symptoms, falsely lengthening recovery times. We did not collect premorbid symptom scores for comparison, and we relied on recall for the concussion-day symptoms. Despite these limitations, the present study adds to the current knowledge about the factors that predict symptom duration among youth with concussion. Several factors were predictive. Symptom reports on the day of clinic evaluation appear to be better predictors of recovery than recollection of symptom burden on the day of injury. Our findings have direct clinical relevance because all the factors explored in this study are readily available from a basic concussion history and assessment, without the need for specialized testing or prior knowledge of premorbid symptoms. Future research should apply these factors prospectively to youth concussion cohorts to corroborate their predictive validity. n We thank Jingzhen Yang, PhD, for her helpful recommendations regarding manuscript preparation, and Caitlin Schmittauer, RN, for her assistance with data collection. Submitted for publication Nov 15, 2015; last revision received Jan 21, 2016; accepted Mar 2, 2016. Reprint requests: Geoffrey L. Heyer, MD, Departments of Pediatrics and Neurology, Nationwide Children’s Hospital and The Ohio State University, 700 Children’s Drive, ED-5, Columbus, OH 43205. E-mail: geoffrey.heyer@ nationwidechildrens.org
References 1. Meehan WP III, Mannix R. Pediatric concussions in United States emergency departments in the years 2002 to 2006. J Pediatr 2010;157:889-93. 2. Langlois JA, Rutland-Brown W, Wald MM. The epidemiology and impact of traumatic brain injury: a brief overview. J Head Trauma Rehabil 2006;21:375-8. 3. Meehan WP III, d’Hemecourt P, Comstock RD. High school concussions in the 2008-2009 academic year: mechanism, symptoms, and management. Am J Sports Med 2010;38:2405-9. 4. Pellman EJ, Powell JW, Viano DC, Casson IR, Tucker AM, Feuer H, et al. Concussion in professional football: epidemiological features of game injuries and review of the literature—part 3. Neurosurgery 2004; 54:81-94. 5. McCrea M, Guskiewicz KM, Marshall SW, Barr W, Randolph C, Cantu RC, et al. Acute effects and recovery time following concussion in collegiate football players: the NCAA Concussion Study. JAMA 2003;290:2556-63. 6. Bleiberg J, Cernich AN, Cameron K, Sun W, Peck K, Ecklund PJ, et al. Duration of cognitive impairment after sports concussion. Neurosurgery 2004;54:1073-8. 7. Giza CC, Hovda DA. The neurometabolic cascade of concussion. J Athl Train 2001;36:228-35.
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
5
THE JOURNAL OF PEDIATRICS
8. Field M, Collins MW, Lovell MR, Maroon J. Does age play a role in recovery from sports-related concussion? A comparison of high school and collegiate athletes. J Pediatr 2003;142:546-53. 9. Pellman EJ, Lovell MR, Viano DC, Casson IR. Concussion in professional football: recovery of NFL and high school athletes assessed by computerized neuropsychological testing—Part 12. Neurosurgery 2006;58:263-74. 10. Meehan WP III, d’Hemecourt P, Collins CL, Comstock RD. Assessment and management of sport-related concussions in United States high schools. Am J Sports Med 2011;39:2304-10. 11. Yeates KO, Taylor HG, Rusin J, Bangert B, Dietrich A, Nuss K, et al. Longitudinal trajectories of postconcussive symptoms in children with mild traumatic brain injuries and their relationship to acute clinical status. Pediatrics 2009;123:735-43. 12. Babcock L, Byczkowski T, Wade SL, Ho M, Mookerjee S, Bazarian JJ. Predicting postconcussion syndrome after mild traumatic brain injury in children and adolescents who present to the emergency department. JAMA Pediatr 2013;167:156-61. 13. Barlow KM, Crawford S, Stevenson A, Sandhu SS, Belanger F, Dewey D. Epidemiology of postconcussion syndrome in pediatric mild traumatic brain injury. Pediatrics 2010;126:e374-81. 14. Eisenberg MA, Andrea J, Meehan W, Mannix R. Time interval between concussions and symptom duration. Pediatrics 2013;132:8-17. 15. Blume HK, Vavilala MS, Jaffe KM, Koepsell TD, Wang J, Temkin N, et al. Headache after pediatric traumatic brain injury: a cohort study. Pediatrics 2012;129:e31-9. 16. Falk AC, Cederfjall C, von WL, Klang B. Are the symptoms and severity of head injury predictive of clinical findings three months later? Acta Paediatr 2006;95:1533-9. 17. Ponsford J, Willmott C, Rothwell A, Cameron P, Ayton G, Nelms R, et al. Cognitive and behavioral outcome following mild traumatic head injury in children. J Head Trauma Rehabil 1999;14:360-72. 18. Zemek RL, Farion KJ, Sampson M, McGahern C. Prognosticators of persistent symptoms following pediatric concussion: a systematic review. JAMA Pediatr 2013;167:259-65. 19. Meehan WP III, Mannix R, Monuteaux MC, Stein CJ, Bachur RG. Early symptom burden predicts recovery after sport-related concussion. Neurology 2014;83:2204-10. 20. Chrisman SP, Rivara FP, Schiff MA, Zhou C, Comstock RD. Risk factors for concussive symptoms 1 week or longer in high school athletes. Brain Inj 2013;27:1-9. 21. Iverson G. Predicting slow recovery from sport-related concussion: the new simple-complex distinction. Clin J Sport Med 2007;17: 31-7. 22. McCrea M, Guskiewicz K, Randolph C, Barr WB, Hammeke TA, Marshall SW, et al. Incidence, clinical course, and predictors of prolonged recovery time following sport-related concussion in high school and college athletes. J Int Neuropsychol Soc 2013;19:22-33. 23. Lau BC, Kontos AP, Collins MW, Mucha A, Lovell MR. Which onfield signs/symptoms predict protracted recovery from sport-related concussion among high school football players? Am J Sports Med 2011;39:2311-8. 24. Collins MW, Field M, Lovell MR, Iverson G, Johnston KM, Maroon J, et al. Relationship between postconcussion headache and neuropsycho-
6
Volume -
www.jpeds.com
25.
26.
27.
28.
29. 30.
31.
32.
33.
34.
35.
36. 37.
38. 39.
logical test performance in high school athletes. Am J Sports Med 2003; 31:168-73. Iverson GL, Gaetz M, Lovell MR, Collins MW. Relation between subjective fogginess and neuropsychological testing following concussion. J Int Neuropsychol Soc 2004;10:904-6. Kontos AP, Elbin RJ, Lau B, Simensky S, Freund B, French J, et al. Posttraumatic migraine as a predictor of recovery and cognitive impairment after sport-related concussion. Am J Sports Med 2013; 41:1497-504. Lau BC, Collins MW, Lovell MR. Sensitivity and specificity of subacute computerized neurocognitive testing and symptom evaluation in predicting outcomes after sports-related concussion. Am J Sports Med 2011;39:1209-16. Lau B, Lovell MR, Collins MW, Pardini J. Neurocognitive and symptom predictors of recovery in high school athletes. Clin J Sport Med 2009;19: 216-21. Moser RS, Schatz P, Jordan BD. Prolonged effects of concussion in high school athletes. Neurosurgery 2005;57:300-6. Morgan CD, Zuckerman SL, Lee YM, King L, Bearid S, Sills AK, et al. Predictors of postconcussion syndrome after sports-related concussion in young athletes: a matched case-control study. J Neurosurg Pediatr 2015;15:589-98. Taylor HG, Dietrich A, Nuss K, Wright M, Rusin J, Bangert B, et al. Postconcussive symptoms in children with mild traumatic brain injury. Neuropsychology 2010;24:148-59. Yeates KO, Taylor HG, Rusin J, Bangert B, Dietrich A, Nuss K, et al. Premorbid child and family functioning as predictors of post-concussive symptoms in children with mild traumatic brain injuries. Int J Dev Neurosci 2012;30:231-7. McNally KA, Bangert B, Dietrich A, Nuss K, Rusin J, Wright M, et al. Injury versus noninjury factors as predictors of postconcussive symptoms following mild traumatic brain injury in children. Neuropsychology 2013;27:1-12. Olsson KA, Lloyd OT, Lebrocque RM, McKinlay L, Anderson VA, Kenardy JA. Predictors of child post-concussion symptoms at 6 and 18 months following mild traumatic brain injury. Brain Inj 2013;27: 145-57. McCrory P, Meeuwisse WH, Aubry M, Cantu B, Dvorak J, Echemendia R, 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;47:250-8. Kelly JP, Rosenberg JH. Diagnosis and management of concussion in sports. Neurology 1997;48:575-80. Aubry M, Cantu R, Dvorak J, Graf-Baumann T, Johnston K, Kelly J, et al. Summary and agreement statement of the First International Conference on Concussion in Sport, Vienna 2001. Recommendations for the improvement of safety and health of athletes who may suffer concussive injuries. Br J Sports Med 2002;36:6-10. Lovell MR, Collins MW. Neuropsychological assessment of the college football player. J Head Trauma Rehabil 1998;13:9-26. Heyer GL, Young JA, Rose SC, McNally KA, Fischer AN. Posttraumatic headaches correlate with migraine symptoms in youth with concussion. Cephalalgia 2015 Jun 8. pii: 0333102415590240. [Epub ahead of print].
Heyer et al
- 2016
ORIGINAL ARTICLES
Figure. Kaplan-Meier plots illustrate the duration of postconcussion symptoms, in days, stratified by total symptom scores from A, the day of concussion and B, the day of clinic evaluation. Greater concussions symptom scores generally led to longer durations of symptoms.
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
6.e1
Concussion day
Factor loadings >.3
Clinic day
Factor loadings >.3
Postconcussion symptoms
Cognitive-fatigue Emotional Cephalalgic Somatic Arousal-stimulation Vomiting
Postconcussion symptoms
Cognitive-fatigue Cephalalgic Emotional Arousal-stimulation Somatic Vomiting
Feeling mentally foggy Drowsiness Feeling slowed down Difficulty remembering Difficulty concentrating Fatigue
.858
Fatigue
.932
.821 .766
Drowsiness Feeling mentally foggy Feeling slowed down Difficulty concentrating Sleeping more than usual Difficulty remembering Balance problems
.913 .858
Irritability
.441
.862 .818
Neck pain Headache
.407
.708 .459
Sensitivity to light Sensitivity to noise
.745 .699 .692
Sleeping more than usual Dizziness
.572
.436
.421
Sadness Feeling more emotional Nervous Irritability Headache Sensitivity to light
.789 .778
Sensitivity to noise Neck pain
.762 .564
Numbness in extremities Weakness in extremities Sleeping less
.316
Nausea Dizziness
.500 .406 .815 .722 .716 .716 .396
.426 .417
0.386
.861 .687
Sleeping less
.912
.900
Trouble sleeping
.670
.764
Numbness in extremities Weakness in extremities Vomiting
Vomiting
.921
Nausea
.646 8
.592
Sadness Feeling more emotional Nervous
.500
Trouble sleeping
Eigenvalues
.492
.619
www.jpeds.com
.588
.631
Balance problems
.788
1.57 1.49 1.33 KMO 0.921; Bartlett Test of Sphericity, P < .001.
1.11
1
Eigenvalues
THE JOURNAL OF PEDIATRICS
6.e2
Table I. Symptom clusters derived from PCAs, day of concussion, and day of clinic evaluation
.899 .852 .737
.856 .363
.651 .960
9.97 1.43 1.35 1.13 KMO 0.949; Bartlett Test of Sphericity, P < .001.
1.02
0.78
Gray shading represents grouping by symptom clusters (principal components).
Volume -
Heyer et al
Study design A retrospective cohort study was conducted to evaluate several potential predictors of symptom duration via a Cox proportional hazards analyses. The individual postconcussion symptom scores were highly correlated, so these symptoms were analyzed in the statistical model as coefficients derived from principal component analyses. Results Among 1953 youth with concussion, 1755 (89.9%) had dates of reported symptom resolution. The remainder (10.1%) were lost to follow-up and censored. The median time to recovery was 18 days (range 1-353 days). By 30 days, 72.6% had recovered; by 60 days, 91.4% had recovered; and by 90 days, 96.8% had recovered. Several variables in a multivariate Cox model predicted postconcussion symptom duration: female sex (P < .001, hazard ratio [HR] = 1.28), continued activity participation (P = .02, HR = 1.13), loss of consciousness (P = .03, HR = 1.18), anterograde amnesia (P = .04, HR = 1.15), premorbid headaches (P = .03, HR = 1.15), symptom components from the day of concussion (emotion, P = .03, HR = 1.08), and the day of clinic evaluation (cognitivefatigue, P < .001, HR = 1.22; cephalalgic, P < .001, HR = 1.27; emotional, P = .05, HR = 1.08; arousal-stimulation, P = .003, HR = 1.1). In univariate analyses, greater symptom scores generally predicted longer symptom durations. Worsening of symptoms from the day of concussion to the day of clinic evaluation also predicted longer recovery (P < .001, HR = 1.59). Conclusions Several factors help to predict protracted postconcussion symptom durations among youth referred to a sports concussion clinic. (J Pediatr 2016;-:---). See editorial, p and related articles, p and p
C
oncussions are common among children, resulting in approximately 144 000 or more visits to the emergency department annually in the US.1 Many more youth with concussion are treated solely by primary care providers, outpatient specialists, or athletic trainers, or they do not seek medical care.2,3 Recovery after concussion is poorly understood. Studies in adults suggest that most patients recover within 7-10 days,4-6 but children may be more vulnerable to the concussion effects and may have longer durations of postconcussion symptoms.7 For example, high school athletes generally recover more slowly than collegiate and professional athletes.8,9 When a variety of study designs are considered, an estimated 2.8%-43% of youth continue to report postconcussion symptoms at 3 months.10-17 Inconsistent, and at times contradictory, evidence exists linking clinical factors with protracted postconcussion symptoms in these patients.18 Potential predictors include overall symptom burden,11,13,14,19-22 specific symptoms,12,23-28 previous concussions,14,17,29,30 loss of consciousness (LOC),13,22,31 absence of LOC,14 post-traumatic amnesia,11,13,22 age,12,14,15 female sex,15,32 premorbid symptoms,17,30,32-34 and hospital admission.12,31 We sought to identify factors that influence the duration of postconcussion symptoms among youth referred to a sports concussion clinic. We expand on current knowledge by evaluating previous and novel predictors from a large representative sample and determining their relative importance in an inclusive multivariate survival model. Understanding From the Division of Pediatric Neurology, Nationwide which factors lead to prolonged recovery can improve overall concussion manChildren’s Hospital; Department of Neurology, The Ohio State University, Columbus, OH; Department of Health agement by informing early subspecialty referral decisions, optimizing resource Sciences, Clemson University, Clemson, SC; Divisions of utilization, and helping clinicians to provide anticipatory guidance. Sports Medicine, and Pediatric Psychology and 1
2
3
4
5
Neuropsychology, Nationwide Children’s Hospital; and 6 Department of Pediatrics, The Ohio State University, Columbus, OH
HR LOC PCA
Hazard ratio Loss of consciousness Principal component analysis
The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2016.03.014
1
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Methods We conducted a retrospective cohort study of pediatric patients presenting to a sports concussion clinic at a large Midwestern children’s hospital between June 2012 and September 2014. The study was approved by the Institutional Review Board at Nationwide Children’s Hospital. Electronic medical records and clinic charting were analyzed. Concussion diagnoses were determined clinically for each patient, at the discretion of the managing physician, on the basis of the injury mechanism(s), history, and examination, and generally followed the operational definition of concussion set forth in the Consensus Statement on Concussion in Sport.35 Study inclusion criteria were age from 10 to 19 years and a concussion injury occurring within 30 days of clinic evaluation. The age range represents the referral limits of the clinic. The interval of 1-30 days between injury and evaluation was chosen arbitrarily to allow a balanced analysis within a representative clinic sample while minimizing the numbers of individuals with distant injuries and potentially excessive recall biases. Exclusion criteria were the absence of traumatic brain injury, clinical features consistent with moderate or severe traumatic brain injury, and inability or refusal to complete the postconcussion symptom questionnaire. When patients sustained more than 1 concussion during the 28-month study period, only the data that corresponded with the first documented injury were analyzed. None of the patients in this study had an acute concussion that overlapped with recovery of a previous concussion. As a standard component of the concussion clinic evaluation, all patients completed a written symptom questionnaire on arrival to the clinic but before their medical evaluations. The questionnaire addressed the presence and severity of 23 common postconcussion symptoms (Table I; available at www.jpeds.com), rated on a scale of 0 (not present) to 6 (severe). Separate scales were completed for current symptoms on the day of clinic evaluation and recalled symptoms from the day of concussion. Our questionnaire has not been validated. It was adapted from previously published symptom assessments.36-38 In addition, patients and their parents answered basic questions related to the concussion, including the date of injury, mechanism of injury, the presence of LOC, the presence of anterograde and retrograde amnesia, whether the patient continued participating in the activity immediately after injury, the number of previous concussions, and the presence of premorbid headaches. Families self-reported race and ethnicity. Managing physician, insurance status, and clinic follow-up appointment dates were extracted from the electronic clinical records. As a standard component of our charting template, patients were asked at each consecutive clinic visit whether postconcussion symptoms had resolved and, if so, when. Resolution of symptoms in this study was defined as patient report of absent symptoms, return to premorbid symptom status, or symptom improvement 2
Volume allowing return to play and without planned follow-up. When resolution occurred within 24 hours of injury, such was designated as day 1 (rather than day 0). Dates of symptom resolution and cancelled/missed appointments were also extracted from the clinical record. Statistical Analyses Descriptive statistics were calculated for demographic and concussion data. The c2 test was used to compare categorical variables, and the Student t test or the Mann-Whitney U test was used to compare continuous variables that were related to the assessment of clinical predictors. A multivariate Cox proportional hazards analysis was used to explore which factors influence the duration of postconcussion symptoms. The dependent variable was time to symptom resolution. When clinic follow-up appointments were missed or cancelled, and the date of symptom resolution was not available, the subject was censored on the date of the earliest missed appointment. All patients who completed follow-up had dates of symptom resolution. To address multicollinearity among postconcussion symptoms that are highly correlated, the 23 symptom scores from each time period (day of concussion and day of clinic evaluation) were reduced to 6 symptom components via a principal component analysis (PCA) with oblique rotation (Promax, kappa = 4).39 The PCA-derived principal components (symptom clusters) are listed in Table I; each cluster was labeled by us according to the symptoms it comprised. As expected, some differences existed between symptoms reported on the day of concussion and the day of clinic evaluation, but the resulting clusters represent logical patterns of symptom groupings for both analyses. The Bartlett method was used to estimate PCA score coefficients for the Cox analysis. Given the inconsistent results from previous studies, we entered predictor variables in hierarchical stages to explore the relative effects of covariates: (1) demographic information and clinic information; (2) concussion characteristics and premorbid headaches; (3) and then the principal component data (symptom clusters) on the day of concussion and the day of clinic evaluation. Interaction terms were not analyzed. The corresponding hazard ratios (HRs) were inverted to reflect the risk of continued postconcussion symptoms (rather than symptom resolution). In addition, univariate Cox proportional hazards analyses were used to determine whether total symptom scores at the time of concussion or the time of clinic evaluation influenced postconcussion symptom duration. The sums of the 23 symptom scores were divided by quartile (Table II), and pairs of consecutive quartiles were compared. We also calculated the differences between scores at clinic evaluation and at concussion (symptom score on clinic day minus symptom score on concussion day) and compared symptom durations among patients with positive and negative differences. A positive difference means that symptoms worsened from the day of concussion to the day Heyer et al
- 2016
ORIGINAL ARTICLES
Table II. Demographic and concussion data Characteristics
n
Data
Age in years, mean (SD) Male sex Ethnicity Non-Hispanic Hispanic/Latino Multiethnic Family declined Race White Black Asian Biracial/multiracial Other Family declined Insurance status Private Public Previous concussions 0 1 2 3 4 or more Mechanism of injury Head to body, ground, or equipment Helmet to body, ground, or equipment Other Don’t know Continued activity participation after concussion LOC Retrograde amnesia Anterograde amnesia Premorbid headaches Mean symptom score, day of concussion (SD) Concussion-day score, 25th percentile Concussion-day score, 50th percentile Concussion-day score, 75th percentile Mean symptom score, day of clinic (SD) Clinic-day score, 25th percentile Clinic-day score, 50th percentile Clinic-day score, 75th percentile
1953 1953 1953
14.1 (2.1) 63.4% 94.3% 2.8% 0.4% 0.4%
1923 80.6% 12.8% 0.4% 3.4% 1.9% 0.2% 1942 80.2% 19.8% 1953 68.9% 21.2% 6.0% 2.7% 1.2% 1953
1922 1923 1938 1935 1940 1953
1953
52.5% 26.4% 11.6% 9.4% 42.6% 16.3% 21.0% 20.3% 21.4% 42.2 (24.4) 22 40 59 21.5 (21.7) 4 15 34
of clinic visit. A Bonferroni correction for multiple testing was applied directly to all univariate P values so that a constant alpha value of .05 could be used across analyses. Kaplan-Meier plots were constructed to illustrate differences in symptom durations. All statistical analyses were performed using SPSS Version 21 (SPSS Inc, Chicago, Illinois). The significance threshold was set at 5%.
Results Our cohort comprised 1953 youth with concussion. Demographic and concussion data are summarized in Table II. The mean interval between concussion injury and the initial clinical evaluation was 10.1 days (median 9 days, SD 6.32). The majority of patients had no previous concussions (68.9%) or 1 previous concussion (21.2%); a very small proportion reported $4 previous concussions (1.2%). Among patients with documented postconcussion symptom recovery, defined as symptom resolution (n = 1755, 89.9%), the median time to recovery was
18 days (range 1-353 days). By 30 days, 72.6% had recovered; by 60 days, 91.4% had recovered; and by 90 days, 96.8% had recovered. Patients lost to follow-up (n = 198, 10.1%) had a median time to their missed or cancelled appointment of 36 days, which differed from the median time to recovery among those with known symptom resolution (P < .001). These patients also reported greater total symptom scores on the day of concussion (mean 50.7 vs 41.2; P = .003) and at the day of clinic evaluation (mean 32 vs 20.4; P < .001) than those not lost to follow-up. Several variables predicted postconcussion symptom duration. Table III shows the results from the full multivariate Cox model. Female sex had a significant effect when added in the first stage of the hierarchical model (P < .001; HR 1.4; 95% CI 1.26-1.55). This effect remained significant in the full model, in which being female increased the risk of protracted concussion symptoms by 28%. Outcomes did not differ in relation to age, ethnicity, insurance status, or the managing physician. Continuing activity participation after injury, LOC, anterograde amnesia, and premorbid headaches all contributed to the prediction of prolonged recovery after concussion, with HR effect sizes ranging from 1.13 to 1.18. LOC was not significant when initially added in stage 2 of the hierarchical model, but it became significant when the symptom clusters were added in the final stage. The symptoms reported on the day of clinic evaluation had greater predictive value for overall recovery than the symptom scores recalled from the day of concussion. The emotional-symptoms cluster (sadness, feeling more emotional, nervousness, and irritability) was the only group of symptoms on concussion day that significantly predicted symptom duration, increasing the risk of protracted symptom duration by 8%. In contrast, 4 groups of symptoms reported on the day of clinic evaluation predicted protracted recovery: cognitive-fatigue, cephalalgic, arousal-stimulation, and emotional. Analyses of total symptom scores demonstrated that each consecutive quartile of scores led to longer symptom durations, with the exception of quartile 1 vs quartile 2 scores recalled from the day of concussion (Table IV and Figure; Figure available at www.jpeds.com). The effects of these changes (in univariate analyses) were especially prominent on the day of clinic evaluation with large effect sizes (HRs 1.48-1.81). Notably, the rate of cancelled and missed appointments increased in kind with the increases in reported symptom burden (Table IV). Lastly, when we compared symptom scores from concussion day and clinic day, 228 patients (11.7%) reported greater scores (ie, worse symptoms) on clinic day. Interval worsening of symptoms was associated with an increase in the risk of protracted recovery by 59% (in univariate analysis), Table IV. This finding was not related to a difference in evaluation times; the mean intervals between concussion day and clinic day did not differ between groups (mean of 10.1 days vs mean of 10 days, P = .81). Patients with worsening of symptoms recalled lower mean symptom
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
3
THE JOURNAL OF PEDIATRICS
Volume -
www.jpeds.com
Table III. Predictors of protracted concussion recovery based on the full statistical model Predictors
P value
HR*
95% CI
Age Female sex Ethnicity Insurance status Attending physician Previous concussions 0 1 2 3 4 or more Mechanism of injury Don’t know Head to body, ground, or equipment Helmet to body, ground, or equipment Other Continued participation LOC Anterograde amnesia Retrograde amnesia Premorbid headaches Day of concussion symptom clustersz Cognitive-fatigue component Cephalalgic component Emotional component Somatic component Arousal-stimulation component Vomiting component Day of clinic symptom clustersz Cognitive-fatigue component Cephalalgic component Emotional component Somatic component Arousal-stimulation component Vomiting component
.16 <.001† NS .18 NS
1.02 1.28
0.99-1.04 1.14-1.44
0.92
0.81-1.04
Reference .72 .7 .12 .59
0.98 1.04 1.27 1.14
0.87-1.1 0.85-1.28 0.94-1.71 0.71-1.83
Reference .72 .07 .92 .02† .03† .04† .89 .03†
0.98 1.2 0.99 1.13 1.18 1.15 0.99 1.15
0.87-1.1 0.99-1.46 0.79-1.23 1.02-1.25 1.02-1.37 1.01-1.31 0.87-1.13 1.02-1.29
.09 .72 .03† .24 .38 .06
1.07 1.01 1.08 1.04 0.98 0.95
0.99-1.17 0.94-1.09 1.01-1.16 0.97-1.1 0.92-1.04 0.9-1.01
<.001† <.001† .05† .09 .003† .54
1.22 1.27 1.08 0.95 1.1 1.02
1.11-1.33 1.18-1.37 1.01-1.16 0.89-1.01 1.03-1.17 0.96-1.08
NS, not significant. *HRs inverted to represent probabilities of continued symptoms, ie, exp(B)1. †Statistically significant. zPrincipal components from symptom scores on concussion day and clinic day (Table I).
scores from the day of concussion (31.7 vs 43.6, P < .001) and reported greater mean scores at the day of clinic evaluation (46.8 vs 18.2, P < .001).
Discussion In a large cohort of youth with concussion, several factors influenced postconcussion symptom duration: female sex,
continued activity participation, LOC, anterograde amnesia, premorbid headaches, overall symptom burden, worsening of symptoms over time, and specific types of symptoms on the day of concussion and the day of clinic evaluation. The present study assessed the clinical factors that predict postconcussion symptom duration among youth referred to a sports concussion clinic. The available evidence from previous studies has been inconsistent about which clinical features of concussion injury lead to rapid resolution of symptoms. When assessed, greater symptom burden (ie, more overall symptoms or greater symptom ratings) during the acute concussion period has been the most consistent predictor of longer symptom durations.11,13,14,19-22 For example, Meehan et al19 categorized symptom duration as #28 days and >28 days and found that total symptom score was the only factor independently associated with symptoms lasting longer than 28 days. Barlow et al13 created subcategories of injury severity that incorporated symptom reporting, LOC, and amnesia. Each stratum of increasing severity led to roughly corresponding increases in symptom duration. Similarly, symptom burden in our study was a significant predictor of recovery time, whether measured as total symptom scores compared across quartiles or specific symptom clusters analyzed in a broadly inclusive survival model. Unfortunately, the high correlations among all postconcussion symptoms prevented us from analyzing each symptom independently. The cephalalgic and cognitive-fatigue clusters at the time of clinic evaluation had the largest effect sizes. These findings resemble those of previous studies in which acute postconcussion headache, dizziness (part of our clinic-day cephalalgic cluster), and the symptom of fogginess predicted protracted recovery.12,23-28 Patients who reported premorbid headaches in our cohort also had a greater probability of prolonged recovery. Emotional symptoms on concussion day and on clinic day each had predictive value in our study. Because we did not assess premorbid emotional symptoms, we do not know whether noninjury factors contributed to these postconcussion effects. While analyzing the associations between symptom score quartiles and symptom duration, we found that patients with greater overall reported symptom burdens were also lost to follow up in higher numbers. It is not clear why this would occur, but it may be related to fatigue from the longer recovery times and
Table IV. Postconcussion symptom durations differ according to initial symptom scores Predictors of concussion duration z
Recalled symptom scores from day of concussion, Q1 vs Q2 Recalled symptom scores from day of concussion, Q2 vs Q3 Recalled symptom scores from day of concussion, Q3 vs Q4 Symptom scores on day of clinic evaluation, Q1 vs Q2 Symptom scores on day of clinic evaluation, Q2 vs Q3 Symptom scores on day of clinic evaluation, Q3 vs Q4 Clinic day symptoms worse than concussion day symptoms
First comparator, n
Total, n
Censored (%), total
P value*
HR†
95% CI
450 524 485 475 478 508 228
974 1009 979 953 986 1000 1953
8.2 8.2 12.1 6.8 8.4 13.3 10.1
.29 .002 <.001 <.001 <.001 <.001 <.001
1.15 1.27 1.36 1.81 1.59 1.48 1.59
1.01-1.31 1.11-1.44 1.19-1.56 1.58-2.06 1.39-1.82 1.29-1.7 1.37-1.85
Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile. *Bonferroni correction applied. †HRs inverted to represent probabilities of continued symptoms, ie, exp(B)1. zQ1 vs Q2 refers to symptom score.
4
Heyer et al
- 2016 multiple follow-up visits. Noninjury factors also may have played a role. The variations in study designs have made comparisons across pediatric studies difficult. Female sex had a large effect in our study but was found to be predictive in only 2 previous studies.15,32 In contrast, previous head injuries14,17,29,30 and age12,14,15 each influenced recovery times in other studies but were not significant in our model. Investigators have published conflicting results about the effects of LOC. Eisenberg et al14 found that LOC was protective against protracted concussion recovery, and other groups demonstrated its negative effects.13,22,31 In our analysis, LOC was not significant when entered initially (stage 2) in the statistical model, but this changed with the addition of symptom clusters in the final model (stage 3), highlighting the potential complexities related to comparisons across differently structured studies. Not all postconcussion symptoms develop immediately after injury; some can take several hours or several days to emerge.35,36 A group of patients in our cohort reported greater symptom scores on the day of clinic evaluation than on the day of concussion, which was not related to the timing of the initial clinic visit. Compared with the patients who improved during the acute concussion period, those who had this positive difference in scores, or “positive delta,” also had a greater probability of protracted postconcussion symptoms. The factors related to the worsening of symptoms during the week or so after concussion have not been well characterized among pediatric or adult patients. Morgan et al30 found that delayed symptom onset of at least three hours following injury was associated with prolonged symptom durations, but worsening of symptoms over a longer period was not assessed. It is possible that our patients with interval worsening of symptoms had different activity levels than those who did better. Symptom burden is important in predicting concussion duration, yet comparison of absolute symptom scores across studies requires that identical instruments be used. The presence or absence of a positive delta, theoretically, could be determined from a variety of postconcussion symptom scales, and it does not require knowledge of premorbid symptoms. Given its predictive value in the present study, we encourage further investigation of the positive delta. We acknowledge several limitations of our study. First, our patients were highly selected. Selection factors included referral to a concussion clinic and postconcussion symptom duration of up to 30 days at the time of initial evaluation. Although the patients in this study may be representative of those seen in tertiary-care concussion clinics, the symptom durations in our cohort generally are not representative of all concussions. Indeed, the reason for referral of some patients was persistence of symptoms beyond one week. Second, we relied on patient report for the determination of resolution of symptoms. Patient report is a meaningful outcome. However, even though objective measures (eg, neurocognitive testing and balance measures) were used as clinically indicated, we did not incorporate their results in this study
ORIGINAL ARTICLES because of variations in timing and use of each objective test. In addition, at the time of reported return to symptom baseline not all patients had returned to full activity or unrestricted play, which is a more practical definition of concussion resolution. Some patients were near their premorbid symptom baseline when reporting symptom resolution, which underestimates overall recovery times. Third, the questionnaire used has not been validated. Finally, postconcussion symptoms are nonspecific. Some patients may have incorrectly attributed baseline symptoms as postconcussion symptoms, falsely lengthening recovery times. We did not collect premorbid symptom scores for comparison, and we relied on recall for the concussion-day symptoms. Despite these limitations, the present study adds to the current knowledge about the factors that predict symptom duration among youth with concussion. Several factors were predictive. Symptom reports on the day of clinic evaluation appear to be better predictors of recovery than recollection of symptom burden on the day of injury. Our findings have direct clinical relevance because all the factors explored in this study are readily available from a basic concussion history and assessment, without the need for specialized testing or prior knowledge of premorbid symptoms. Future research should apply these factors prospectively to youth concussion cohorts to corroborate their predictive validity. n We thank Jingzhen Yang, PhD, for her helpful recommendations regarding manuscript preparation, and Caitlin Schmittauer, RN, for her assistance with data collection. Submitted for publication Nov 15, 2015; last revision received Jan 21, 2016; accepted Mar 2, 2016. Reprint requests: Geoffrey L. Heyer, MD, Departments of Pediatrics and Neurology, Nationwide Children’s Hospital and The Ohio State University, 700 Children’s Drive, ED-5, Columbus, OH 43205. E-mail: geoffrey.heyer@ nationwidechildrens.org
References 1. Meehan WP III, Mannix R. Pediatric concussions in United States emergency departments in the years 2002 to 2006. J Pediatr 2010;157:889-93. 2. Langlois JA, Rutland-Brown W, Wald MM. The epidemiology and impact of traumatic brain injury: a brief overview. J Head Trauma Rehabil 2006;21:375-8. 3. Meehan WP III, d’Hemecourt P, Comstock RD. High school concussions in the 2008-2009 academic year: mechanism, symptoms, and management. Am J Sports Med 2010;38:2405-9. 4. Pellman EJ, Powell JW, Viano DC, Casson IR, Tucker AM, Feuer H, et al. Concussion in professional football: epidemiological features of game injuries and review of the literature—part 3. Neurosurgery 2004; 54:81-94. 5. McCrea M, Guskiewicz KM, Marshall SW, Barr W, Randolph C, Cantu RC, et al. Acute effects and recovery time following concussion in collegiate football players: the NCAA Concussion Study. JAMA 2003;290:2556-63. 6. Bleiberg J, Cernich AN, Cameron K, Sun W, Peck K, Ecklund PJ, et al. Duration of cognitive impairment after sports concussion. Neurosurgery 2004;54:1073-8. 7. Giza CC, Hovda DA. The neurometabolic cascade of concussion. J Athl Train 2001;36:228-35.
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
5
THE JOURNAL OF PEDIATRICS
8. Field M, Collins MW, Lovell MR, Maroon J. Does age play a role in recovery from sports-related concussion? A comparison of high school and collegiate athletes. J Pediatr 2003;142:546-53. 9. Pellman EJ, Lovell MR, Viano DC, Casson IR. Concussion in professional football: recovery of NFL and high school athletes assessed by computerized neuropsychological testing—Part 12. Neurosurgery 2006;58:263-74. 10. Meehan WP III, d’Hemecourt P, Collins CL, Comstock RD. Assessment and management of sport-related concussions in United States high schools. Am J Sports Med 2011;39:2304-10. 11. Yeates KO, Taylor HG, Rusin J, Bangert B, Dietrich A, Nuss K, et al. Longitudinal trajectories of postconcussive symptoms in children with mild traumatic brain injuries and their relationship to acute clinical status. Pediatrics 2009;123:735-43. 12. Babcock L, Byczkowski T, Wade SL, Ho M, Mookerjee S, Bazarian JJ. Predicting postconcussion syndrome after mild traumatic brain injury in children and adolescents who present to the emergency department. JAMA Pediatr 2013;167:156-61. 13. Barlow KM, Crawford S, Stevenson A, Sandhu SS, Belanger F, Dewey D. Epidemiology of postconcussion syndrome in pediatric mild traumatic brain injury. Pediatrics 2010;126:e374-81. 14. Eisenberg MA, Andrea J, Meehan W, Mannix R. Time interval between concussions and symptom duration. Pediatrics 2013;132:8-17. 15. Blume HK, Vavilala MS, Jaffe KM, Koepsell TD, Wang J, Temkin N, et al. Headache after pediatric traumatic brain injury: a cohort study. Pediatrics 2012;129:e31-9. 16. Falk AC, Cederfjall C, von WL, Klang B. Are the symptoms and severity of head injury predictive of clinical findings three months later? Acta Paediatr 2006;95:1533-9. 17. Ponsford J, Willmott C, Rothwell A, Cameron P, Ayton G, Nelms R, et al. Cognitive and behavioral outcome following mild traumatic head injury in children. J Head Trauma Rehabil 1999;14:360-72. 18. Zemek RL, Farion KJ, Sampson M, McGahern C. Prognosticators of persistent symptoms following pediatric concussion: a systematic review. JAMA Pediatr 2013;167:259-65. 19. Meehan WP III, Mannix R, Monuteaux MC, Stein CJ, Bachur RG. Early symptom burden predicts recovery after sport-related concussion. Neurology 2014;83:2204-10. 20. Chrisman SP, Rivara FP, Schiff MA, Zhou C, Comstock RD. Risk factors for concussive symptoms 1 week or longer in high school athletes. Brain Inj 2013;27:1-9. 21. Iverson G. Predicting slow recovery from sport-related concussion: the new simple-complex distinction. Clin J Sport Med 2007;17: 31-7. 22. McCrea M, Guskiewicz K, Randolph C, Barr WB, Hammeke TA, Marshall SW, et al. Incidence, clinical course, and predictors of prolonged recovery time following sport-related concussion in high school and college athletes. J Int Neuropsychol Soc 2013;19:22-33. 23. Lau BC, Kontos AP, Collins MW, Mucha A, Lovell MR. Which onfield signs/symptoms predict protracted recovery from sport-related concussion among high school football players? Am J Sports Med 2011;39:2311-8. 24. Collins MW, Field M, Lovell MR, Iverson G, Johnston KM, Maroon J, et al. Relationship between postconcussion headache and neuropsycho-
6
Volume -
www.jpeds.com
25.
26.
27.
28.
29. 30.
31.
32.
33.
34.
35.
36. 37.
38. 39.
logical test performance in high school athletes. Am J Sports Med 2003; 31:168-73. Iverson GL, Gaetz M, Lovell MR, Collins MW. Relation between subjective fogginess and neuropsychological testing following concussion. J Int Neuropsychol Soc 2004;10:904-6. Kontos AP, Elbin RJ, Lau B, Simensky S, Freund B, French J, et al. Posttraumatic migraine as a predictor of recovery and cognitive impairment after sport-related concussion. Am J Sports Med 2013; 41:1497-504. Lau BC, Collins MW, Lovell MR. Sensitivity and specificity of subacute computerized neurocognitive testing and symptom evaluation in predicting outcomes after sports-related concussion. Am J Sports Med 2011;39:1209-16. Lau B, Lovell MR, Collins MW, Pardini J. Neurocognitive and symptom predictors of recovery in high school athletes. Clin J Sport Med 2009;19: 216-21. Moser RS, Schatz P, Jordan BD. Prolonged effects of concussion in high school athletes. Neurosurgery 2005;57:300-6. Morgan CD, Zuckerman SL, Lee YM, King L, Bearid S, Sills AK, et al. Predictors of postconcussion syndrome after sports-related concussion in young athletes: a matched case-control study. J Neurosurg Pediatr 2015;15:589-98. Taylor HG, Dietrich A, Nuss K, Wright M, Rusin J, Bangert B, et al. Postconcussive symptoms in children with mild traumatic brain injury. Neuropsychology 2010;24:148-59. Yeates KO, Taylor HG, Rusin J, Bangert B, Dietrich A, Nuss K, et al. Premorbid child and family functioning as predictors of post-concussive symptoms in children with mild traumatic brain injuries. Int J Dev Neurosci 2012;30:231-7. McNally KA, Bangert B, Dietrich A, Nuss K, Rusin J, Wright M, et al. Injury versus noninjury factors as predictors of postconcussive symptoms following mild traumatic brain injury in children. Neuropsychology 2013;27:1-12. Olsson KA, Lloyd OT, Lebrocque RM, McKinlay L, Anderson VA, Kenardy JA. Predictors of child post-concussion symptoms at 6 and 18 months following mild traumatic brain injury. Brain Inj 2013;27: 145-57. McCrory P, Meeuwisse WH, Aubry M, Cantu B, Dvorak J, Echemendia R, 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;47:250-8. Kelly JP, Rosenberg JH. Diagnosis and management of concussion in sports. Neurology 1997;48:575-80. Aubry M, Cantu R, Dvorak J, Graf-Baumann T, Johnston K, Kelly J, et al. Summary and agreement statement of the First International Conference on Concussion in Sport, Vienna 2001. Recommendations for the improvement of safety and health of athletes who may suffer concussive injuries. Br J Sports Med 2002;36:6-10. Lovell MR, Collins MW. Neuropsychological assessment of the college football player. J Head Trauma Rehabil 1998;13:9-26. Heyer GL, Young JA, Rose SC, McNally KA, Fischer AN. Posttraumatic headaches correlate with migraine symptoms in youth with concussion. Cephalalgia 2015 Jun 8. pii: 0333102415590240. [Epub ahead of print].
Heyer et al
- 2016
ORIGINAL ARTICLES
Figure. Kaplan-Meier plots illustrate the duration of postconcussion symptoms, in days, stratified by total symptom scores from A, the day of concussion and B, the day of clinic evaluation. Greater concussions symptom scores generally led to longer durations of symptoms.
Specific Factors Influence Postconcussion Symptom Duration among Youth Referred to a Sports Concussion Clinic
6.e1
Concussion day
Factor loadings >.3
Clinic day
Factor loadings >.3
Postconcussion symptoms
Cognitive-fatigue Emotional Cephalalgic Somatic Arousal-stimulation Vomiting
Postconcussion symptoms
Cognitive-fatigue Cephalalgic Emotional Arousal-stimulation Somatic Vomiting
Feeling mentally foggy Drowsiness Feeling slowed down Difficulty remembering Difficulty concentrating Fatigue
.858
Fatigue
.932
.821 .766
Drowsiness Feeling mentally foggy Feeling slowed down Difficulty concentrating Sleeping more than usual Difficulty remembering Balance problems
.913 .858
Irritability
.441
.862 .818
Neck pain Headache
.407
.708 .459
Sensitivity to light Sensitivity to noise
.745 .699 .692
Sleeping more than usual Dizziness
.572
.436
.421
Sadness Feeling more emotional Nervous Irritability Headache Sensitivity to light
.789 .778
Sensitivity to noise Neck pain
.762 .564
Numbness in extremities Weakness in extremities Sleeping less
.316
Nausea Dizziness
.500 .406 .815 .722 .716 .716 .396
.426 .417
0.386
.861 .687
Sleeping less
.912
.900
Trouble sleeping
.670
.764
Numbness in extremities Weakness in extremities Vomiting
Vomiting
.921
Nausea
.646 8
.592
Sadness Feeling more emotional Nervous
.500
Trouble sleeping
Eigenvalues
.492
.619
www.jpeds.com
.588
.631
Balance problems
.788
1.57 1.49 1.33 KMO 0.921; Bartlett Test of Sphericity, P < .001.
1.11
1
Eigenvalues
THE JOURNAL OF PEDIATRICS
6.e2
Table I. Symptom clusters derived from PCAs, day of concussion, and day of clinic evaluation
.899 .852 .737
.856 .363
.651 .960
9.97 1.43 1.35 1.13 KMO 0.949; Bartlett Test of Sphericity, P < .001.
1.02
0.78
Gray shading represents grouping by symptom clusters (principal components).
Volume -
Heyer et al