OBJECTIVE: To test the hypothesis that children with a previous history of concussion have a longer duration of symptoms after a repeat concussion than those without such a history.
METHODS: Prospective cohort study of consecutive patients 11 to 22 years old presenting to the emergency department of a children's hospital with an acute concussion. The main outcome measure was time to symptom resolution, assessed by the Rivermead Post-Concussion Symptoms Questionnaire (RPSQ). Patients and providers completed a questionnaire describing mechanism of injury, associated symptoms, past medical history, examination findings, diagnostic studies, and the RPSQ. Patients were then serially administered the RPSQ for 3 months after the concussion or until all symptoms resolved.
RESULTS: A total of 280 patients were enrolled over 12 months. Patients with a history of previous concussion had a longer duration of symptoms than those without previous concussion (24 vs 12 days, P = .02). Median symptom duration was even longer for patients with multiple previous concussions (28 days, P = .03) and for those who had sustained a concussion within the previous year (35 days, P = .007) compared with patients without those risk factors. In a multivariate model, previous concussion, absence of loss of consciousness, age ≥13, and initial RPSQ score >18 were significant predictors of prolonged recovery.
CONCLUSIONS: Children with a history of a previous concussion, particularly recent or multiple concussions, are at increased risk for prolonged symptoms after concussion. These findings have direct implications on the management of patients with concussion who are at high risk for repeat injuries.
- ADHD —
- attention-deficit/hyperactivity disorder
- ED —
- emergency department
- LOC —
- loss of consciousness
- mTBI —
- mild traumatic brain injury
- RPSQ —
- Rivermead Post-Concussion Symptoms Questionnaire
What’s Known On This Subject:
Although concussion is increasingly being diagnosed in the pediatric population, little is known about what factors lead to prolonged postconcussive symptoms in children. In particular, the effect of previous history of concussion on recovery from a repeat injury is unclear.
What This Study Adds:
Children with a history of previous concussion, particularly recent or multiple concussions, are at increased risk for prolonged symptoms after concussion. This suggests that repeat concussion, particularly within a vulnerable time window, may lead to longer duration of symptoms.
Approximately 144 000 children present each year to US emergency departments (EDs) with concussion,1,2 and millions more are treated by athletic trainers, primary care doctors, or outpatient specialists.3 Despite increasing scientific research into concussion, however, relatively little is known about what factors lead to prolonged postconcussive symptoms in children. This lack of prognostic data has led to significant uncertainty among patients, families, and health care providers as to which patients would benefit from specialist follow-up, extended academic accommodations, prolonged abstinence from athletic participation, and even permanent cessation of high-risk activity.
Although a large and expanding body of literature has examined concussion in adults,4–6 the data that exist on pediatric concussion is composed primarily of smaller case series and retrospective studies.7–9 Few studies have examined risk factors for prolonged symptom duration in children,10–14 and they have generated conflicting results on the modifying effect of previous concussion. Supporting an association between previous history of concussion and prolonged symptom duration, Moser et al7 showed that high school athletes with multiple previous concussions had persistent neurocognitive deficits indistinguishable from those in the acute recovery phase of concussion. Similarly, in an analysis of 119 children with mild traumatic brain injury (mTBI), Ponsford et al15 demonstrated that previous head injury was predictive of persistent concussive symptoms 3 months after the injury. However, 2 larger studies, a secondary analysis of an mTBI cohort presenting to a pediatric ED and a prospective study of concussed high school athletes, found no association between previous concussion and duration of postconcussive symptoms.16,17
Animal models of mTBI offer a possible explanation for this variation, suggesting that both timing and number of injuries modify the effects of previous injury. In particular, several studies in rodents suggest that the effect of multiple concussions is cumulative, and that the time elapsed between concussions to allow for return of normal physiologic function may be the key to the course of recovery.18–22
Despite this, no clinical study has characterized the vulnerable window during which repeat concussion results in worse outcome among children, or the critical number of injuries that confers increased risk of prolonged recovery. If reinjury within a window of vulnerability confers higher risk of long-term neurocognitive deficits, then abstinence from at-risk activities may ultimately improve outcomes. The objective of this study was to characterize the effect of previous concussion on recovery from concussion among children presenting to an ED. We hypothesized that those with a previous history of concussion would have a longer time to symptom resolution than those without such a history, and that those who sustained a concussion within the previous year would have a longer time to symptom resolution than those with more remote injuries.
We conducted a prospective cohort study of consecutive patients aged 11 to 22 years who presented to the ED of a tertiary care children’s hospital within 72 hours of a concussion from September 1, 2011, to August 31, 2012.
Concussion was defined as a blunt injury to the head resulting in either (1) alteration in mental status or (2) any of the following symptoms that started within 4 hours of the injury and were not present before the injury: headache, nausea, vomiting, dizziness/balance problems, fatigue, drowsiness, blurred vision, memory difficulty, or difficulty concentrating in the absence of indications for head imaging or without intracranial hemorrhage when imaging was obtained. Although there has been controversy regarding the definition of concussion for clinical research,23–26 we chose this broad definition so as to include both those who had alteration of consciousness and those who presented with typical postconcussive symptoms following a head injury.6,27,28
Patients were excluded from the study if any of the following were present: (1) Glasgow Coma Score <13 on arrival to the ED, (2) coexisting skull or long-bone fracture, (3) coexisting injury to intra-abdominal or intrathoracic organ or spinal cord, (4) cognitive or developmental disability preventing patient from completing the questionnaire, or (5) involvement of either law enforcement or ED social workers for victims of an assault. These criteria were designed to distinguish postconcussive symptoms from symptoms related to other injuries or psychological stresses related to major trauma or assault.
The primary outcome was time to resolution of postconcussive symptoms assessed via the Rivermead Post-Concussion Symptoms Questionnaire (RPSQ), a 16-item concussion symptom inventory checklist.29 The RPSQ has been used extensively in both adult and pediatric studies of mTBI,10,30,31 shown a high degree of interrater and test-retest reliability,29,32 and been shown to be valid and unbiased in young children.31,33 The questionnaire was available to study participants in both English and Spanish. Patients were considered symptom-free when all inventories of the RPSQ were scored a 0 (symptom not present) or 1 (symptom present at preinjury baseline). Patients filled out the survey independently or with assistance of parents at the discretion of the patient and family.
Enrollment and Consent
Study participants were enrolled during their ED visit by trained research coordinators after informed consent (and assent for patients <18 years) was obtained. Eligible patients who were not contacted during their ED visit were offered enrollment the following day if they were still within 72 hours of the injury. On enrollment, patients completed an electronic questionnaire asking for demographic information (including self-reported race and ethnicity), mechanism of injury, associated symptoms, such as loss of consciousness (LOC) and amnesia, relevant past medical history, and the RPSQ. Providers completed a questionnaire with physical examination findings, diagnostic studies, and information on interventions, disposition, and discharge instructions.
An online follow-up questionnaire containing the RPSQ was sent electronically to patients 1, 2, 4, 6, 8, and 12 weeks after their ED visit or until they met criteria for symptom resolution. Patients who reported resolution of symptoms were prompted to provide the last date on which symptoms occurred. Patients with incomplete or inconsistent data were called to resolve these issues. Patients were considered lost to follow-up if they failed to respond to 2 consecutive questionnaires. Study data were collected and managed using REDCap (Research Electronic Data Capture, Nashville, TN) electronic data capture tools hosted at Boston Children’s Hospital.34
Statistical analysis was performed by using PASW Statistics 18 (IBM SPSS Statistics, IBM Corporation, Chicago, IL). We estimated that 250 patients would be required to demonstrate a 5-day difference in time to resolution of symptoms between those with and without previous concussion, using a 2-tailed α of 0.05 and power of 90%.
Data were analyzed by Student’s t-test or Wilcoxon Rank Sum test as appropriate. Our primary outcome, time to symptom resolution, was assessed as a continuous variable ranging from 0 to 90 days. Potential predictors that were continuous in nature were dichotomized before analysis according to their median value. The only exception to this was age; 13 years or older was chosen rather than the median as a proxy for pubertal status.
In univariate analysis of time to symptom resolution, Kaplan-Meier analysis with log-rank tests of significance was used to best account for censored data. To adjust for confounders, we constructed a multivariate Cox regression model using predictors with P < .05 on univariate screening. Where Kaplan-Meier analysis was performed, “median” refers to median survival time (where “survival” signifies persistence of symptoms), a measure that takes into account expected duration of symptoms for censored patients. Values were considered statistically significant if P < .05. The institutional review board approved this study before onset of data collection.
A total of 302 patients were approached for the study and 280 (93%) were enrolled over a 12-month period (Fig 1); 207 patients (74%) completed the study, 28 (10%) completed at least 1 follow-up questionnaire but did not complete the study, and 45 (16%) were lost to follow-up. Nonwhite patients, Hispanic patients and patients with attention-deficit/hyperactivity disorder (ADHD) were more likely to be lost to follow-up (Table 1).
Most patients (66.0%) were enrolled in the study on the day their concussion occurred, with 24.7% enrolled the following day, 7.2% enrolled 2 days later, and 1.7% 3 days later.
A majority of (63.8%) patients were injured playing a sport, with the most common sports being hockey (14%), soccer (9.4%), football (8.5%), and basketball (8.1%). The most common presenting symptoms (RPSQ score ≥2) were headache (85.1%), fatigue (64.7%), and dizziness (63.0%). An abnormal physical finding was noted in 10.6% of patients, with the most common abnormalities being altered gait or balance (4.3%) and altered mental status (2.4%). Among the 20.8% of patients who had neuroimaging performed, there were no abnormalities identified related to trauma. On discharge, 65.9% of patients were prescribed a period of cognitive rest, 92.4% were recommended to take time off from athletics, and 63.8% were advised to follow-up with their primary care doctor, 45.5% in a sports concussion clinic, and 6.2% with another specialist.
Time to Recovery
Median time to symptom resolution and percentage of patients symptomatic 7, 28, and 90 days after the concussion are shown in Table 2. In univariate analysis, history of previous concussion (Fig 2A), age ≥13, initial RPSQ score >18, female gender, history of depression, absence of LOC, and abnormal neurologic examination on presentation were all predictive of a longer time to symptom resolution (Table 2). Among patients with a previous concussion, those who had a concussion in the past year had nearly 3 times the median duration of symptoms compared with those who had no previous concussion or whose most recent concussion occurred >1 year previous (Fig 2B). Similarly, patients with 2 or more previous concussions had more than double the median symptom duration compared with patients with 0 or 1 previous concussion (Fig 2C).
In the multivariate model, patients with a history of previous concussion, absence of LOC, age ≥13, and RPSQ score >18 had significantly longer symptom duration than patients without these risk factors (Table 3).
Our study demonstrates that previous concussion is predictive of a longer time to symptom resolution after pediatric concussion. Importantly, we found that the effect of previous concussion on symptom duration was strongly influenced by both the number of previous concussions and the time elapsed since the most recent previous concussion. Both study participants with multiple previous concussions and those who had sustained a concussion within the previous year had a markedly greater duration of symptoms than those with no previous concussion. Conversely, patients who had only a single previous concussion that occurred more than a year before their current injury had no statistical difference in duration of symptoms from children without a previous concussion. These findings, which suggest both temporal vulnerability and a dose-response effect of previous injuries, support previous research in animal models of concussive brain injury.18–22
Previous studies in rodents have demonstrated cumulative effects of repetitive mild head injuries.19,21 Additionally, several of these studies suggest a temporal window of vulnerability wherein the repeat trauma has a more pronounced effect. One study observed that whereas mice concussed at monthly intervals performed similarly to noninjured mice on tests of learning and memory, animals concussed at weekly and daily intervals developed persistent cognitive deficits, with the daily concussion group showing ongoing effects up to 1 year after the injury compared with controls.22 Other animal models have offered possible biological mechanisms for this temporal vulnerability. Longhi et al18 showed evidence of axonal injury and cytoskeletal damage that was significantly greater in mice that received a second concussion within 3 to 5 days of an initial head injury than those who had been subjected to only a single concussion. Vagnozzi et al20 found a similar effect in rats and correlated it to reversible impairments in mitochondrial enzymes.
Our study demonstrates this temporally sensitive effect of previous concussion for the first time in humans. This has direct implications on the management of athletes and other at-risk individuals who sustain concussions, supporting the concept that sufficient time to recover from a concussion may improve long-term outcomes. However, we did not find an association between physician-advised cognitive or physical rest and duration of symptoms, which may reflect the limitations of our observational study. Previous studies looking at the effect of cognitive rest and abstinence from sports have been inconclusive, limited by either observational or retrospective study design.35,36 A randomized control trial will likely be necessary to address the utility of this intervention in patients.
Notably, there were several other risk factors associated with prolonged time to recovery in our study. Patients with more severe symptoms at presentation, represented by an RPSQ score >18, had more than double the median duration of symptoms compared with patients with RPSQ scores below this threshold, consistent with the findings of multiple previous studies.17,37,38 Additionally, our study suggests that patients 13 years and older had a longer duration of symptoms than younger children. Most previous studies examining the modifying effect of age on concussion recovery have compared high school age patients with young adults.39 Only Babcock et al16 previously compared preadolescent children to adolescents, likewise finding that younger age predicted a faster recovery. It is uncertain whether the difference between younger children and adolescents reflects differing neurobiology between the 2 groups or more severe mechanisms of injury in sports such as ice hockey and football, where games between older children involve more contact and higher-force impacts.
In our ED cohort, we demonstrate an association between LOC and shorter duration of symptoms. This is in contrast to previous studies that have shown that LOC was a risk factor for prolonged recovery or had no effect at all.12,16,17 Our findings may reflect a referral bias, in which patients with LOC are more likely present to the ED on the basis of this factor alone, whereas those without LOC may be referred to the ED only when their symptoms or injury mechanism are severe. Alternatively, it is possible that patients who lose consciousness as a result of their injury may be more likely to comply with medical recommendations of physical and cognitive rest than those who did not lose consciousness, thus speeding recovery from their injury. We cannot, however, eliminate the possibility that there is a biological basis to this finding, in which mechanisms of injury responsible for LOC and those responsible for long-term neurocognitive sequelae may differ.
Some of the potential predictors of prolonged recovery that did not have a statistical association with duration of symptoms in our study are also noteworthy. In the multivariate analysis, female gender did not predict prolonged recovery, although there appeared to be a difference between the genders in univariate analysis. This is nearly identical to the results of Babcock et al,16 and likely because female patients had more severe symptoms at presentation in our study (mean initial RPSQ 21.3 vs 17.0 in male patients, P = .02), thus representing a confounding variable. Whether this finding is indicative of the fact that female patients have more severe symptoms from concussion in general, as suggested in several previous studies,40–42 or is due to referral bias in which female individuals preferentially present to the ED when symptoms are more severe than male counterparts cannot be ascertained from our data.
We also observed that patients with a history of migraine headaches, depression, ADHD, and anxiety did not have prolonged symptoms after concussion; however, small sample size limits our ability to detect relationships in these specific patient populations.
Patients in our study had a longer duration of symptoms than most previous reports in the literature, which have frequently shown symptoms resolving by 7 to 10 days after the injury.5,17,43 This may be because of differences in study populations, as children presenting to an ED may represent a more severe subset of concussed patients than the outpatient cohorts previously studied. The RPSQ may also be more sensitive to ongoing symptoms than the self-assessment tools or neurocognitive testing used in many other studies.
There are several limitations to our study. Although we enrolled 93% of patients approached to participate, there were additional patients who were not contacted by the research coordinators. The most common reasons for this were arrival to the ED during a time when no research coordinator was available and short length of ED stay, leading to discharge before the patient could be contacted about enrollment. Second, by using survival analysis to best account for censored data, we assumed that there was no difference between patients who were censored and those who were not. Although this assumption appears to be valid based on similar demographics and symptom scores of censored versus noncensored patients, it is possible that patients ultimately failed to complete the study because of more bothersome symptoms or, conversely, more rapid symptom resolution. A final limitation is the accuracy of self-reported symptoms: certain subgroups of the population, notably athletes, may have had an incentive to report symptom resolution to expedite return-to-play; others may report symptoms that are no longer present to avoid schoolwork; still others may have a difficult time understanding the questions of the RPSQ.
Despite these limitations, our study had several strengths, including a large sample size and prospective methodology. We enrolled a high percentage of eligible patients and most patients continued participation until symptom resolution. By serially surveying patients at short time intervals and asking for the specific date of symptom resolution within that interval, our estimates of symptom duration are likely to be more accurate than if we used a single follow-up questionnaire months after the injury.
Finally, unlike previous studies that focused on particular subgroups of patients, such as participants in a particular sport, our study examined all eligible patients who presented to a tertiary care ED. As a result, we studied a diverse group in terms of both demographics and mechanisms of injury, potentially making our study more generalizable to medical providers practicing in the primary care and ED settings than samples of selected populations.
Children with a history of a previous concussion, particularly those with recent or multiple concussions, are at increased risk for prolonged symptoms after concussion. These findings have direct implications on the management of concussion patients, particularly those at high risk for future concussive injuries, such as athletes.
Michael Monutaeux, ScD, provided statistical guidance and support. Rich Bachur, MD, and Mark Neuman, MD, provided mentorship and manuscript review. Chris Landrigan, MD, assisted with manuscript review. Mark Berry, MA, provided programming and administrative support. The research coordinator team (Elizabeth Paulsen, BS, Lucy Abernethy, BA, Kaitlin Morris, BA, Hillary Chu, BA, and Jessica LeSage, BS) enrolled and followed up with patients and provided administrative support.
- Accepted March 27, 2013.
- Address correspondence to Matthew A. Eisenberg, MD, Division of Emergency Medicine, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115. E-mail:
Dr Eisenberg was responsible for study conception and design, data acquisition and analysis, and drafting and review of the article; Mr Andrea was responsible for data acquisition and analysis, and drafting and review of the article; Dr Meehan was responsible for study conception, study design, and article review; Dr Mannix was responsible for study conception, study design, data analysis, and article review; and all authors approved the final version of this manuscript.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding. Expenses related to use of research coordinators were paid by the Division of Emergency Medicine, Boston Children’s Hospital.
- Meehan WP III,
- d’Hemecourt P,
- Comstock RD
- ↵Iverson G. Predicting slow recovery from sport-related concussion: the new simple-complex distinction. Clin J Sport Med. 2007;17(1):31–37
- Yeates KO,
- Taylor HG,
- Rusin J,
- et al
- Zemek RL, Farion KJ, Sampson M, McGahern C. Prognosticators of persistent symptoms following pediatric concussion: a systematic review. JAMA Pediatr 2013;167(3):259–265
- ↵Ponsford J, Cameron P, Fitzgerald M, Grant M, Mikocka-Walus A, Schonberger M. Predictors of postconcussive symptoms 3 months after mild traumatic brain injury. Neuropsychology. 2012;26(3):304–313
- ↵McCrea M, Guskiewicz K, Randolph C, 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(1):22–33
- ↵Longhi L, Saatman KE, Fujimoto S, et al. Temporal window of vulnerability to repetitive experimental concussive brain injury. Neurosurgery. 2005;56:364–374; discussion 364–374
- Meehan WP III,
- d’Hemecourt P,
- Collins CL,
- Comstock RD
- McCrory P, Meeuwisse W, Johnston K, et al. Consensus statement on concussion in sport 3rd International Conference on Concussion in Sport held in Zurich, November 2008. Clin J Sport Med 2009;19:185–200
- McCrory P,
- Meeuwisse W,
- Johnston K,
- et al
- ↵International Symposium on Concussion in Sport. Abstracts. Br J Sports Med. 2001;35(5):367–377pmid:11579081
- Halstead ME,
- Walter KD,
- Council on Sports Medicine and Fitness
- Meehan WP III,
- Bachur RG
- ↵Faux S, Sheedy J, Delaney R, Riopelle R. Emergency department prediction of post-concussive syndrome following mild traumatic brain injury—an international cross-validation study. Brain Inj 2011;25(1):14–22
- Eyres S,
- Carey A,
- Gilworth G,
- Neumann V,
- Tennant A
- ↵Lannsjo M, Borg J, Bjorklund G, Af Geijerstam JL, Lundgren-Nilsson A. Internal construct validity of the Rivermead Post-Concussion Symptoms Questionnaire. J Rehabil Med 2011;43(11):997–1002
- Stulemeijer M,
- van der Werf S,
- Borm GF,
- Vos PE
- De Kruijk JR,
- Leffers P,
- Menheere PP,
- Meerhoff S,
- Rutten J,
- Twijnstra A
- Colvin AC,
- Mullen J,
- Lovell MR,
- West RV,
- Collins MW,
- Groh M
- ↵Preiss-Farzanegan SJ, Chapman B, Wong TM, Wu J, Bazarian JJ. The relationship between gender and postconcussion symptoms after sport-related mild traumatic brain injury. PM R 2009;1(3):245–253
- Copyright © 2013 by the American Academy of Pediatrics