Spaghetti plot of individuals (thin colored lines) and mean (heavy black line) symptom trajectories for participants without symptom spikes (A) and those with at least 1 symptom spike (B). PCSS indicates postconcussion symptom scale.
Spaghetti plot of individual symptom trajectories for 20 patients who had a symptom spike.
eAppendix. Activity-related symptom exacerbations after pediatric concussion.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Silverberg ND, Iverson GL, McCrea M, Apps JN, Hammeke TA, Thomas DG. Activity-Related Symptom Exacerbations After Pediatric Concussion. JAMA Pediatr. 2016;170(10):946–953. doi:10.1001/jamapediatrics.2016.1187
Recovery from concussion generally follows a trajectory of gradual improvement, but symptoms can abruptly worsen with exertion. This phenomenon is poorly understood.
To characterize the incidence, course, and clinical significance of symptom exacerbations (spikes) in children after concussion.
Design, Setting, and Participants
This secondary analysis of clinical trial data analyzes 63 eligible participants prospectively recruited from an emergency department who were asked to complete a postconcussion symptom scale and record their activities in a structured diary for the next 10 days. They completed standardized assessments of symptoms (postconcussion symptom scale), cognition (Immediate Post-Concussion Assessment and Cognitive Testing), and balance (Balance Error Scoring System) 10 days following the injury. Eligible participants were aged 11 to 18 years and sustained a concussion (according to the Centers for Disease Control and Prevention criteria) that did not result in an abnormal computed tomography scan or require hospital admission. The trial was conducted from May 2010 to December 2012, and the analysis was conducted from November 2015 to February 2016.
Main Outcome and Measure
The occurrence of symptom spikes, defined as an increase of 10 or more points on the postconcussion symptom scale over consecutive days.
Of the 63 participants, there were 41 boys (65.1%) and 22 girls (34.9%), and the mean (SD) age was 13.8 (1.8) years. Symptom spikes occurred in one-third of the sample (20 participants [31.7%]). Symptom spikes tended to partially resolve within 24 hours. An abrupt increase in mental activity (ie, returning to school and extracurricular activities) from one day to the next increased the risk of a symptom spike (relative risk, 0.81; 95% CI, 0.21-3.21), but most symptom spikes were not preceded by a documented increase in physical or mental activity. Patients with symptom spikes were initially more symptomatic in the emergency department and throughout the observation period but did not differ from the group without symptom spikes on cognition or balance 10 days following injury.
Conclusions and Relevance
Certain patients appeared susceptible to high and variable symptom reporting. Symptom spikes may not themselves be detrimental to recovery. The present findings support clinical guidelines for adolescents to return to school and activities gradually after concussion.
clinicaltrials.gov Identifier: NCT01101724
There is growing evidence that prolonged rest does not facilitate recovery from concussion1-7 and concern that it may be counterproductive to recovery.8-10 Contemporary clinical practice guidelines still caution that children might need to rest for longer and progress through stages of increased activity more slowly than adults.11,12 This appears motivated by concern that the developing brain may be more vulnerable to reinjury as well as excessive neurometabolic demands during recovery from concussion.9,12 It has long been observed that symptoms after concussion can be exacerbated by vigorous physical activity or resolve but then reemerge with physical exertion.13-17 Mental exertion (eg, sustained concentration) can also provoke postconcussive symptoms,18,19 which is particularly relevant to return to school planning after pediatric concussion.20
Little is currently known about the incidence, natural history, and clinical significance of activity-related symptom exacerbations after pediatric concussion. The present study aims to characterize symptom exacerbations (“spikes”) as well as their antecedents and potential consequences. We conducted a secondary analysis of data from a randomized clinical trial4 of prescribed rest vs usual care for pediatric concussion. Uniquely, this trial had participants keep a daily diary of their activities and symptoms for the first 10 days after concussion, permitting a fine-grained temporal analysis of changes in activity level and postconcussion symptoms.4 We predicted that high physical and mental activity would precede symptom exacerbations. We further hypothesized that symptom exacerbations would be transient and not raise the risk of poor clinical outcome. The present findings could help guide clinicians in advising patients when and how to return to their preinjury activities after concussion.
Question How common and clinically significant are activity-related symptom exacerbations after concussion?
Findings In this secondary analysis of a randomized clinical trial, 63 children with a concussion were studied for 10 days after injury. Symptom exacerbations were common, occurring in 1 of 3 participants, within the first 10 days of a concussion but appeared largely transient, and they were not associated with impaired balance or cognition. A major increase in activity associated with returning to school and extracurricular activities doubled the risk of a symptom exacerbation; however, most symptom exacerbations were not preceded by mental or physical exertion.
Meaning Symptom exacerbations after concussion may not be clinically significant events.
This is a secondary analysis of a prospective randomized clinical trial4 of patients presenting to the Children’s Hospital of Wisconsin emergency department (ED). The study was approved by the Children’s Hospital of Wisconsin Institutional Review Board and registered with ClinicalTrials.gov (NCT01101724). Written informed consent and assent were obtained from caregivers and participants. Children were eligible if they were aged 11 to 22 years (mean [SD] age, 13.8 [1.8]; range, 11-18) and presented to the ED within 24 hours of a concussion (operationalized as direct or indirect forces to the head resulting in 1 or more acute signs or symptoms).21 Patients were excluded if they were non-English speaking, were admitted to the hospital, had no legal guardian present, had an intracranial injury, or had a condition that interfered with valid clinical assessments.
Adolescents underwent an initial screening to gather demographic information and injury details. In the ED, participants also underwent computerized neurocognitive testing and a standardized balance assessment by a research assistant. Participants were then randomized to strict rest, in which ED physicians advised restricted school and extracurricular work and physical exertion for 5 days, or to usual care, in which ED physicians were free to recommend activity restrictions as they saw fit. Participants completed activity and symptom diaries for the first 10 days after injury and had a follow-up appointment 10 days after their ED visit, which included the readministration of neurocognitive tests and balance assessments by a research assistant.
Participants completed a daily diary that assessed type and duration of physical and mental activities.22,23 Participants were asked to record their physical activity at specified times (eg, 15-minute intervals) throughout the day. Activities were categorized according to their average energy costs, from the lowest activity level, 1 (eg, sleep or rest in bed), to the highest level, 9 (eg, sport participation). The diaries were analyzed by adding the number of periods for each categorical value and then multiplying by the basal metabolic rate multiplier for each activity category22 and the predicted basal metabolic rate of the individual participant, using the sex and weight prediction formula.24 These values were then summed to yield the total energy expenditure for each day. Activity-related energy expenditure was then determined by subtracting the basal metabolic rate from the total energy expenditure. Participants also reported the time in hours spent on specific mental activities each day. There exists neither a criterion standard nor a validated scale to assess mental activity. Therefore, patients were asked to report the time spent on specific mental activities, specifically those activities patients were advised to limit on the Acute Concussion Evaluation form21: school work (ie, attending class, homework, studying, and taking tests) and extracurricular participation. Listening to music, watching television, and surfing the internet did not count as mental activity in the present study.
A 19-item postconcussion symptom scale (PCSS), adapted from Aubry et al,13 was included in the daily diary entry. Each symptom was graded by the subject from none (0) to severe (6) and was obtained daily for the first 10 days.
The Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT version 2.0; ImPACT Applications) computerized test battery was used as a primary neurocognitive assessment. This measure has been validated for use in the ED setting, and it reliably detects neurocognitive deficits after concussion.25,26 The test administers 6 neuropsychological test modules, and composite scores are reported for verbal memory, visual memory, reaction time, and processing speed.
The Balance Error Scoring System objectively assesses balance.27 The test consists of 3 stance conditions (ie, double leg, single leg, and tandem); each stance is performed with eyes closed on normal firm flooring and a medium-density foam surface. An inability to maintain the stance or keep eyes closed is deemed an error. Performance was scored by adding the error points for all 6 trials.
A symptom spike was defined as an increase of 10 or more points on the PCSS over consecutive days, which corresponded to a statistically reliable change (ie, exceeding the upper limit of the 95% CI for test-retest change in healthy controls) on a very similar postconcussion symptom questionnaire (eAppendix in the Supplement).28 We restricted the analyses to symptom spikes between days 2 and 9 postinjury so that we could examine antecedents (eg, activity level on the day prior) and consequences (eg, activity level on the day after) within the 10-day observation period. The 63 participants (64% of original sample) in the Thomas et al4 trial who contributed a full 10 days of diary data were eligible for the present study. Note that participants who were excluded for missing data did not differ from eligible participants (with or without symptom spikes) with regard to age, sex, randomization, or PCSS scores in the ED (P > .05).
Linear mixed-effects models were used to test the null hypotheses that symptoms, physical activity, and mental activity all remained stable over the 10-day observation period. Time (in days) was the covariate of interest. Sex and randomization allocation (control vs strict rest advice) were included as covariates in all mixed models. In all cases, models with both a random intercept and slope achieved a superior model fit (based on the Akaike Information Criterion) compared with those with a random intercept only. Therefore, the reported models contain random effects for the intercept and slope.
Groups with and without symptom spikes were compared on demographic and clinical variables with χ2 tests for proportions and independent samples with t tests for continuous variables. Relative risk values were computed to compare the rate of symptom spikes in participants who had usual vs unusually high levels (>90th percentile for the full sample) of activity and activity change over consecutive days. Linear regression was used to determine if participants with at least 1 symptom spike had worse outcomes at the 10-day in-person assessment, after controlling for potential confounders.
Symptoms decreased over the 10 days following injury in the full sample (t = −7.50; P < .001). The 20 participants with symptom spikes reported more symptoms over the observation period than the 43 participants without symptoms spikes (t = 4.90; P < .001). As can be seen in Figure 1, participants with symptom spikes started out with higher PCSS scores and had a less steep symptom improvement trajectory, whereas those with no symptom spikes followed a more linear, uninterrupted symptom recovery course. Participants who went on to have at least 1 symptom spike were initially more symptomatic in the ED than participants without symptom spikes (mean [SD] PCSS score, 40.6 [21.0] vs 29.7 [18.2]; t60 = 2.07; P = .04; Cohen d, 0.54 [medium effect size]). The individual trajectories for participants who had symptom spikes were also more variable throughout the observation period, with more prominent peaks and valleys. The SD for the daily PCSS score in the group without symptom spikes gradually decreased but remained high in the group with symptom spikes; the SD was twice as large in the group with symptom spikes by day 6 (20.7 vs 9.4) and thrice as large by day 9 (21.5 vs 7.3).
Next, we examined time trends for return to physical and mental activity. Physical activity did not significantly increase over the observation period (t = −0.29; P = .77). Mental activity increased markedly over this period (t = 7.97; P < .001). Note that the coefficients for sex and randomization assignment (control vs strict rest advice) were nonsignificant in all models (P > .05), suggesting that these factors did not influence symptom trajectories.
Of the 20 participants with at least 1 symptom spike, 4 participants had a second symptom spike, and no participant had more than 2. The first symptom spike occurred on average on day 5 (interquartile range [IQR], 3-6). The first spike was characterized by a mean (SD; IQR) increase of 50.3% (21.6; 36.9%-62.2%) or 17.1 points (11.9; 11.0-18.0) on the PCSS, followed by a mean (SD; IQR) decrease of 27.9% (36.6; 10.3%-52.8%) or 9.3 points (11.3; 3.3-15.0). By the day after the first symptom spike, the PCSS score returned to or fell below the PCSS score on the day prior to the symptom spike in 12 children (60.0%). In the remaining 8 children (40.0%) whose symptom exacerbation did not completely resolve, the PCSS score decreased somewhat (ie, partially resolved exacerbation) the day after the symptom spike in 4 children (50.0%). The percentage change in PCSS score from the day of the first symptom spike to the following day are reported in Table 1. The course of participants’ first symptom spikes is also illustrated in Figure 2. To better understand the nature of symptom spikes, Table 1 also includes the specific postconcussive symptoms that increased at the individual-patient level. These data suggest that symptom spikes were not characterized by a particular symptom cluster. Rather, there was tremendous between-subject variability in the type of symptoms that contributed to a symptom spike.
Participants with at least 1 symptom spike were similar to participants without symptom spikes with respect to age, sex, randomization assignment, presence of loss of consciousness, and history of previous concussion(s). However, participants with symptom spikes were initially more symptomatic in the ED and on the following day, before we began recording symptom spikes.
Activity-related energy expenditure was used to measure physical activity. Mental activity was measured in hours per day spent engaged in school and extracurricular activity. We used the 90th percentile of the full sample (see eAppendix in the Supplement for raw score cutoffs) to define an unusually high level or change in activity, and we calculated relative risk values for a symptom spike based on physical or mental activity on the day the symptom spike occurred or over the preceding 24 hours. Relative risk values greater than 1.0 indicate that participants with unusually high activity were more likely to experience a symptom spike. The results are presented in Table 2. The only significant predictor of a symptom spike was an unusually high increase in mental activity the day prior; a large and nongraduated increase in mental activity from one day to the next was associated with an increased risk of experiencing a symptom spike on that day. However, of the 20 participants with at least 1 symptom spike, only 5 (25.0%) demonstrated an activity-related symptom spike. Most instances (38 of 43 [88.4%]) of an unusually large increase in mental activity were not followed by a symptom spike, and most symptom spikes (20 of 24 [83.3%]) were not preceded by an unusually high increase in mental activity.
In a series of regression models, we examined whether having at least 1 symptom spike between 2 and 9 days after injury was associated with symptoms, cognition, and balance in the clinical assessment conducted 10 days after injury. Participants’ randomization assignment (1, control; 2, strict rest advice) and initial score (obtained in the ED) on the outcome measure of interest were included as covariates in each model. The regression model results are presented in Table 3. The group that experienced at least 1 symptom spike reported more severe subjective symptoms in the 10-day postinjury assessment. In contrast, the groups did not differ on objective, performance-based cognitive, and balance testing in the 10-day postinjury assessment.
The present study characterized symptom exacerbations during the first 10 days after pediatric concussion. There were several key findings. Symptom spikes occurred in nearly a third of patients. They were more likely to occur in patients who reported a high symptom burden when initially evaluated in the ED but not in those with a loss of consciousness, prior concussion(s), or certain demographic features (eg, age and sex). We observed that symptom spikes were not characterized by a specific symptom (eg, headache) or symptom cluster and tended to diminish or completely resolve within 24 hours. Contrary to our hypothesis, symptom spikes were not associated with physical activity or the absolute mental activity level on the day of the symptom spike or the day prior. However, they were associated with a sharp increase in mental activity over the preceding 24 hours (eg, returning to school and extracurricular activities relatively abruptly). Finally, symptom spikes were associated with higher symptom reporting but not cognitive function or balance 10 days after injury.
To our knowledge, the present study is the first to demonstrate a temporal association between increased mental activity and symptom exacerbations outside of the laboratory. Youths who sharply increased their school participation from one day to the next were more likely to experience a symptom spike the following day. However, it should be emphasized that most children were able to return to school (even abruptly) without experiencing a symptom spike and that most symptom spikes were not related to activity. The diversity of symptoms that increased in each symptom spike further suggests that factors other than exertion (such as stress29 and sleep30,31) likely contribute to daily fluctuations in postconcussion symptoms.
Many questions about activity-related symptom exacerbations remain unanswered. First, the time course of increasing activity and escalating symptoms is not clear. Symptoms and activity were measured on a daily basis, and some association between those 2 variables was found. However, the association may have been stronger if we measured both variables with greater granularity (eg, hourly). Second, because our last data point was obtained at 10 days following injury, we do not know whether acute symptom exacerbations were associated with worse longer-term outcomes or whether the association was a causal one. The finding that patients who had a symptom spike were more symptomatic before the symptom spike and had more variable trajectories throughout the observation period leads us to hypothesize that unmeasured injury or patient characteristics might predispose some individuals to both high and variable postconcussion symptom reporting at any time. Finally, the mechanism underlying activity-related symptom exacerbations is unknown. Persistent impairments in neurometabolism17 and autonomic cerebrovascular regulation32 have both been proposed but not verified empirically.
The present study has important limitations. First, physical activity was measured using diary entries. It would have been preferable to measure physical activity objectively, such as with actigraphy. However, objectively measuring mental activity in daily life is not feasible. Second, our sample size was modest, especially for participants with activity-related exacerbations. We likely did not have sufficient statistical power to detect small effect sizes. The small sample size also raises the risk of spurious results. Our novel findings require replication in future larger studies. Third, participants in the present study were randomized to receive different ED discharge instructions about when to start gradually resuming their preinjury activities.4 Although compliance with prescribed rest was variable for physical activity and moderate for mental activity,4 we cannot consider the present cohort to be strictly observational. To mitigate the potential effect of prescribed rest instructions, we included the randomization allocation as a covariate in our analyses, which was consistently nonsignificant. Fourth, limiting our sample to participants with complete diary data enabled us to accurately identify symptom spikes during the observation period but may have introduced bias. Although eligible participants and participants who were ineligible because of missing diary data did not differ on baseline variables, systematic group differences cannot be ruled out.
The present study is an important first step toward characterizing the incidence, natural history, and clinical significance of symptom spikes after pediatric concussion. We tentatively conclude that symptom exacerbations from one day to the next are common, largely transient, and not specific to a particular symptom domain. Returning to full days of school raises the risk of a symptom spike on the following day. However, symptom spikes may not be clinically significant events. Further research is needed to determine the causes and consequences of symptom spikes. In the interim, our findings support continuing to advise children to return to activities gradually and in a manner that does not significantly exacerbate symptoms,11 because even a transient worsening might provoke anxiety and interfere with school reintegration. The present findings do not provide a rationale for delaying when children can start to gradually return to activity after concussion (ie, prolonged rest). When a child or adolescent presents to medical attention with concern about a sudden increase in postconcussion symptoms, clinicians are encouraged to inquire about changes in daily activity as well as other potential triggers, such as stress and sleep. Until the mechanisms and long-term consequences of symptom exacerbations are better understood, it may be prudent to monitor these youths more regularly while being mindful that most will resume a trajectory of improving symptoms within 24 hours and not have worsened cognition or balance.
Corresponding Author: Danny G. Thomas, MD, MPH, Department of Pediatrics, Emergency Medicine, Children’s Hospital of Wisconsin Corporate Center, 999 N 92nd St, Ste C550, Milwaukee, WI 53226 (email@example.com).
Accepted for Publication: April 20, 2016.
Published Online: August 1, 2016. doi:10.1001/jamapediatrics.2016.1187.
Author Contributions: Dr Thomas had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Silverberg, Iverson, Apps, Hammeke, Thomas.
Drafting of the manuscript: Silverberg, Iverson, McCrea.
Critical revision of the manuscript for important intellectual content: Iverson, McCrea, Apps, Hammeke, Thomas.
Statistical analysis: Silverberg, Iverson.
Obtained funding: Thomas.
Administrative, technical, or material support: Iverson, McCrea, Apps, Hammeke, Thomas.
Study supervision: Thomas.
Conflict of Interest Disclosures: Dr Iverson has been reimbursed by the government, professional scientific bodies, and commercial organizations for discussing or presenting research relating to mild traumatic brain injury and sport-related concussion at meetings, scientific conferences, and symposiums. He has a clinical and consulting practice in forensic neuropsychology involving individuals who have sustained mild traumatic brain injuries (including professional athletes). He has received research funding from several test publishing companies, including ImPACT Applications, CNS Vital Signs, and Psychological Assessment Resources. He has not received research support from a test publishing company in the past 5 years.
Funding/Support: Dr Thomas received the Advancing A Healthier Wisconsin Seed Grant from the Injury Research Center of the Medical College of Wisconsin. Dr Silverberg received salary support from the Vancouver Coastal Health Research Institute. Dr Iverson received funding from the Mooney-Reed Charitable Foundation.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We wish to thank M. D. Mahsin, MSc (University of British Columbia, Vancouver, British Columbia, Canada), for his assistance with the statistical analyses and figures. Mr Mahsin was compensated for his contribution.