Most individuals recover from a concussion within 7–10 days, but some do not recover for months or years after injury, and some never recover. Individuals who do not recover within the usual time are said to have postconcussion syndrome (PCS). Studies of PCS have reported major differences in the incidence of this syndrome, ranging from less than 5%17 to 58%,6 but the most commonly reported incidence is 10%–15% of concussions.19,33,34 One major reason for this large range is the lack of agreement about the definition of PCS. The International Classification of Diseases, 10th Revision (ICD-10) requires 3 symptoms or more from a restricted list of symptoms including headache, dizziness, fatigue, irritability, insomnia, concentration difficulty, memory difficulty, and reduced tolerance to stress, emotional excitement, or alcohol.42 In contrast, the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) requires 3 or more symptoms from a different list, including fatigue, sleep disturbance, headache, dizziness, irritability, affective disturbance, personality change, and apathy, and requires symptoms to last 3 or more months following injury. Also, DSM-IV requires the presence of cognitive deficits in attention and/or memory and a significant decline in social and/or occupational functioning.2 In DSM-V, PCS was replaced by “Major or Mild Neuro-cognitive Disorder Due to Traumatic Brain Injury,” and unfortunately the “diagnostic criteria” do not include the majority of patients with the diagnosis of PCS based on DSM-IV or ICD-10. In general, DSM-V only recognizes patients with loss of consciousness (LOC), posttraumatic amnesia, disorientation, or focal neurological signs such as hemiparesis or abnormal neuroimaging; thus, in our view, DSM-V is not useful for assessing patients with PCS.8 In our view, the ICD-10 and DSM definitions and changes are arbitrary and without a clear scientific basis with respect to the list of eligible symptoms, the 3-month requirement, and the recategorization. In view of the lack of unanimity about the definition of PCS, we used an inclusive definition of PCS as any 3 or more postconcussion symptoms persisting for at least 1 month. We excluded all head injuries that resulted in a hemorrhage, contusion, or a moderate to severe traumatic brain injury (TBI). The head injury must have been a concussion that conforms to the 4th International Consensus Conference on Concussion in Sport.27 The criteria for PCS according to the ICD-10, DSM-IV, and the definition used in this study are shown in Table 1.
Definitions of postconcussion syndrome
| Variable | WHO ICD-10 (1992)42,43 | DSM-IV2 | Our Definition |
|---|---|---|---|
| Essential components | Head trauma …
|
|
|
| No. of symptoms | At least 3 from the list below | At least 3 from the list below + difficulty in attention or memory | Any 3 symptoms from the list of 40 symptoms shown in Table 6 |
| Eligible symptoms | Headache | Headache | See Table 6 |
| Dizziness | Dizziness or vertigo | ||
| Fatigue | Fatigue | ||
| Irritability | Irritability or aggression on little provocation | ||
| Insomnia | Disordered sleep | ||
| Memory problems | Personality changes | ||
| Concentration issues | Apathy | ||
| Reduced tolerance to stress, emotional excitement, or alcohol | Anxiety or depression | ||
| May also have these symptoms: | Depression, anxiety | Visual or hearing impairments, anosmia, orthopedic &/or neurological complications, substance-related disorders | Not applicable |
| Duration of PCS | Does not specify the duration | At least 3 mos | At least 1 mo |
Table 2 lists all the potential predictors of PCS previously cited in the literature and categorized into preinjury, injury, and postinjury predictors. There is no unanimity about the validity of most of these predictors. One of the most commonly cited predictors is premorbid psychiatric illness,10,14,25,32,39 including depression,12,26 anxiety,12,19–21 and posttraumatic stress disorder (PTSD).26,39 Other predictors include learning disabilities,44 migraine headaches,19 females,1,4–6,10,12,13,26,32,39 older age,18,21,39 and children and adolescents.15,27 Lastly, those with previous concussions are believed to have an increased risk of PCS,3,7,13,19,27,32,44 although the exact number required to increase the risk is unknown.
Potential predictors of PCS cited in the literature*
| Predictor | Qualifier | References |
|---|---|---|
| Preinjury predictors of PCS | ||
| Age | Increased risk w/older age (above age 40, but the exact age of increased risk is unknown); also increased risk in children & adolescents | 8,15,18,21,27 |
| Learning disabilities & ADD/ADHD | Increased risk w/previous history of a learning disability or ADD/ADHD | 44 |
| Migraines | Increased risk w/history of migraine | 19 |
| No. of previous concussions | Increased risk w/previous concussions | 3,7,13,19,27,32,44 |
| Psychiatric conditions | Depression, anxiety, & PTSD | 10,12,14,19,21,25,26,32,39,44 |
| Sex | Females are at increased risk | 1,4,5,6,10,12,13,26,32,39 |
| Socioeconomic status | Those of lower income & less education are at increased risk | 5,39 |
| Substance abuse | Increased risk w/substance abuse of alcohol or drugs | 24 |
| Injury predictors of PCS | ||
| Amnesia &/or LOC | Increased risk if there was amnesia &/or LOC at time of injury | 19,41 |
| Extracranial injuries | Increased risk if extracranial injuries sustained at the time of injury | 5,18,33 |
| GCS score | Increased risk w/GCS score < 15 (13 or 14), & correlated w/abnormal brain CT/MR scans | 23,28 |
| Mechanism of injury | Increased risk w/motor vehicle accidents, falls, & assaults; decreased risk w/sports & recreation | 4,6,12,32 |
| Postinjury predictors of PCS | ||
| Litigation | Increased risk if patient involved in litigation about the injury | 9,11,31 |
The predictors are presented in alphabetical order, thus the order is not indicative of the strength of the predictor.
The purposes of the present study were to examine the demographic and etiological features of PCS in a large retrospective cohort and to identify any predictors. We also aimed to determine whether there were any differences in predictors if the ICD-10 or DSM-IV definitions were used. Our long-term aims are to develop strategies to prevent and treat PCS and to determine whether PCS is an antecedent to chronic traumatic encephalopathy (CTE).
Methods
Study Design and Patients
This study was a retrospective chart review approved by the Research Ethics Board of the University Health Network in Toronto, Canada. All consecutive patients seen in consultation by 1 neurosurgeon (C.H.T.) with a specific interest in PCS were considered for this study. All patients diagnosed with a concussion from January 1997 to June 2013 were identified from clinical records by a single researcher (H.S.D.). Data extraction from medical records was also performed by this researcher. All information including definitive diagnoses was clearly outlined in the clinical notes and other findings from imaging and other tests.
This study defined concussion according to the 4th International Consensus Conference on Concussion in Sport.27 Although most of the cases evaluated in the clinic antedated this conference, this definition of concussion was rigorously used to select patients for inclusion. It should be noted that the Glasgow Coma Scale (GCS) score is not a factor in this definition of concussion. Moreover, for the present study of PCS, patients were excluded if they had any lesions on routine MRI or CT scans because such lesions indicate the presence of more severe injuries such as cerebral contusion or intracranial hemorrhage.
As noted above, patients were considered to have PCS if they had any 3 or more postconcussion symptoms persisting for at least 1 month. Patients were excluded if they did not meet this specific definition of PCS or if they had a more severe brain injury. Of the 284 consecutive patients with concussions, 63 were excluded for the reasons stated in Table 3, leaving a sample of 221. It is interesting to note that 21 (7.4%) of the 284 cases referred as concussions were excluded due to focal neurological deficits, which indicated brain injuries more severe than concussion (Table 3), and these cases were excluded on the basis of imaging. One hundred thirty-eight (62.4%) of the 221 accepted cases sustained their most recent concussion in sports and recreation. They are included in the present study, but certain features of this group were recently reported.38
List of patients excluded and reasons for exclusion
| Reason for Exclusion | Value (%) |
|---|---|
| Fewer than 3 symptoms | 15 (23.8) |
| Symptoms for less than 1 mo | 16 (25.4) |
| Fewer than 3 symptoms & symptoms for less than 1 mo (not included in previous 2 categories) | 11 (17.5) |
| Intracerebral hemorrhage | 3 (4.8) |
| Cerebral contusion | 6 (9.5) |
| Subarachnoid hemorrhage & cerebral contusion | 2 (3.2) |
| Subdural &/or epidural hematoma | 3 (4.8) |
| Intracerebral hematoma & cerebral contusion | 1 (1.6) |
| No PCS & brain injury more severe than concussion (not included in any of the above categories) | 6 (9.5) |
| Total exclusions | 63 (100) |
Data Collection
The data collected from the patients' clinical charts were from many sources including clinician's notes, neuropsychological reports, imaging reports, referring physicians' notes, other physicians' notes, Sport Concussion Assessment Tools versions 2 and 3, and patient self reports. A total of 50 different symptoms were recorded from these sources, along with demographic information, previous medical history, imaging performed, duration of symptoms, medications, and other treatments. A total of 77 different variables were recorded.
Statistical Analysis
Chi-square tests were used to compare the prevalence of incidental findings in the patients versus the general population. Statistical significance was set at p ≤ 0.05 before the analysis was conducted. This statistical test was conducted using Microsoft Excel.
To evaluate this heterogeneous, multivariate data set in a principled way, the data were analyzed as follows. Variables were separated into a set of predictors and outcomes, with predictors consisting of age, sex, number of previous concussions, prior history of psychiatric conditions, prior history of migraines, cause of injury (sports or otherwise), presence of extracranial injuries, occurrence of amnesia or LOC, and the patient's involvement in litigation. Outcomes consisted of the number of reported symptoms, and the duration of PCS. The number of symptoms played a dual role, acting as an outcome to be predicted in one analysis and as an additional predictor for the analysis of PCS duration. The variables age, number of previous concussions, and number of symptoms were all heavily skewed toward lower values, so these data were log-transformed to better approximate normally distributed variables. Analyses conducted with or without these log-transformations produced qualitatively similar results.
Four multivariate analyses were performed to explore the correlational structure among the predictors, and to evaluate their capacity to predict the two outcomes. The first analysis searched for the presence of statistically significant pairwise associations among all predictors (including the number of symptoms). The nonparametric Spearman's rank correlation was used for pairs of continuous variables, the chi-square test for pairs of categorical variables, and the nonparametric Wilcoxon rank-sum test for continuous/categorical variable pairs. Compensation for multiple comparisons (10 × (10 − 1)/2 = 45 tests) was achieved by controlling for a false discovery rate of p = 0.05.
The second analysis explored the clustering of variables implied by the pairwise associations with a principal components analysis. All predictors were standardized to have zero mean and unit variance so that their component scores could be meaningfully compared regardless of measurement units. The variance captured by each component and the component scores were examined, and the significance of components was established via permutation testing (10,000 iterations).
The third test examined the ability to predict the number of symptoms reported using multiple linear regression over the remaining variables. Again, standardized predictors were used so the regression coefficient values could be compared. In addition to the analysis of the full model incorporating all predictors, a stepwise analysis was also conducted to identify the most parsimonious model with the fewest predictors.
The fourth analysis attempted to forecast the duration of PCS using all predictors, including the number of reported symptoms. This analysis is complicated by the fact that for patients who have recovered, we know exactly how long their PCS lasted, but for those who have not yet recovered, we know only that their PCS has lasted at least as long as the time from injury to the last consultation. The Cox proportional hazards model was selected to assess the effect of predictor variables on the rate at which an event of interest occurs. In this study, the event of interest is recovery from concussion, and the outcome variable is the time elapsed from concussion to recovery. If recovery had not yet occurred at the time of the last consultation, then the time elapsed from injury to consultation date was recorded and the data were flagged as censored. All analyses were performed using Matlab and its Statistics Toolbox Release 2014b (The MathWorks, Inc.).
Finally, because the criteria for PCS used in this work are novel, the above analyses were repeated for the subsets of patients corresponding to the ICD-10 and DSM-IV definitions and the results compared.
Results
Demographic Features
The cohort consisted of 127 males (57.5%) and 94 females (42.5%; Table 4). The mean age at the time of the most recent concussion was 27.0 years, and the median age was 20 years, with a range of ages from 10 to 74 years.One hundred patients (45.2%) were between the ages of 10 and 19 (Fig. 1). The average number of total concussions (including the most recent that was the cause of the consultation) was 3.3, and the median number of total concussions was 3 (Fig. 2). Fifty-eight patients (26.2%) had a history of 1 of the following comorbidities prior to their concussion: previous migraine headaches, psychiatric illness, learning disability, and/or attention-deficit disorder (ADD)/attention-deficit hyperactivity disorder (ADHD). Twenty-one patients (9.5%) were involved in litigation related to the concussion.
Demographic characteristics in 221 patients
| Demographic Variable | Value (%) |
|---|---|
| Sex | |
| Male | 127 (57.5) |
| Female | 94 (42.5) |
| Comorbidity | |
| Migraines | 21 (9.5) |
| Depression | 18 (8.1) |
| Anxiety | 17 (7.7) |
| Learning disability | 9 (4.1) |
| ADHD | 6 (2.7) |
| Sleep disorder | 5 (2.3) |
| ADD | 3 (1.4) |
| Eating disorder | 2 (0.9) |
| Postpartum depression | 2 (0.9) |
| Substance abuse | 2 (0.9) |
| Bipolar disorder | 1 (0.5) |
| Dysgraphia | 1 (0.5) |
| Hypochondriasis | 1 (0.5) |
| OCD | 1 (0.5) |
| PTSD | 1 (0.5) |
OCD = obsessive-compulsive disorder.
Graph showing the ages of patients at the time of their most recent concussion.
Graph of the total number of concussions per patient, including the most recent concussion.
Concussion Details
The median time interval from the last concussion to consultation was 4 months, with a range of 4 days to 26 years (Fig. 3). The mean and median duration of PCS were 14.7 and 7 months, respectively, with a range of 1 month to 26 years (Fig. 4). In 11.8% the PCS lasted more than 2 years, and 23.1% with PCS had only 1 concussion. In only 25 patients (11.3%), the PCS had resolved completely at the time of the last examination. Forty-three patients (19.5%) lost consciousness in the last concussion, and 72 patients (32.6%) had amnesia. Fifty-seven patients (25.8%) were prescribed medications to manage their symptoms prior to the consultation, 36 (63.2%) of whom were prescribed antidepressants.
Graph of the time interval from the most recent concussion until consultation with the neurosurgeon (C.H.T.). m = months, y = years.
Graph of the duration of PCS symptoms from the most recent concussion. In many patients, the total duration would be longer because the PCS had not resolved completely at the time of their last examination. m = months, y = years.
With respect to the etiology of the most recent concussion, 138 (62.4%) occurred in sports and recreation, with 72 (52.2%) in ice hockey (Table 5). Motor vehicle crashes caused 23 (10.4%) of the most recent concussions, falls caused 19 (8.6%), and school-related activity caused 17 (7.7%).
Cause of most recent concussion
| Concussion Cause | Value (%) |
|---|---|
| Sports & recreation | 138 (62.4) |
| Winter sports | 87 (39.4) |
| Ice hockey* | 72 (32.6) |
| Skiing* | 8 (3.6) |
| Snowboarding* | 4 (1.8) |
| Field sports | 22 (10.0) |
| Soccer* | 12 (5.4) |
| Rugby* | 3 (1.4) |
| Floor sports | 8 (3.6) |
| Basketball* | 4 (1.8) |
| Equestrian | 6 (2.7) |
| Water sports | 5 (2.3) |
| Bicycling | 3 (1.4) |
| Motor sports | 3 (1.4) |
| Snowmobile* | 2 (0.9) |
| Miscellaneous | 2 (0.9) |
| Racquet sports | 1 (0.5) |
| Summer sports | 1 (0.5) |
| Motor vehicle crash | 23 (10.4) |
| Driver | 13 (5.9) |
| Passenger | 4 (1.8) |
| Cyclist | 1 (0.5) |
| Pedestrian | 1 (0.5) |
| Motorcycle | 1 (0.5) |
| Falls | 19 (8.6) |
| School-related | 17 (7.7) |
| Team sport – university/college varsity† | 8 (3.6) |
| Team sport – elementary/high school† | 3 (1.4) |
| Roughhousing | 2 (0.9) |
| Fighting | 1 (0.5) |
| Gym class† | 1 (0.5) |
| Lunch/recess | 1 (0.5) |
| Work-related | 7 (3.2) |
| Assault | 6 (2.7) |
| Falling object | 4 (1.8) |
| Striking object | 4 (1.8) |
| Roughhousing | 2 (0.9) |
| Other | 1 (0.5) |
| Total | 221 |
Specific sport/recreational activity was listed if more than 1 concussion occurred in this activity.
Concussions that occurred in a school setting were categorized under school-related, even if they occurred from sports and recreation activities.
Persistent Symptoms
The average number of persistent symptoms was 8.1, with a range of 3 to 23 persistent symptoms; the most common were headaches, memory deficits, concentration difficulties, imbalance, and dizziness (Table 6).
Most common persisting symptoms in PCS
| Persistent Symptoms | No. (%) |
|---|---|
| Headaches | 197 (89.1) |
| Memory deficits | 136 (61.5) |
| Concentration difficulties | 124 (56.1) |
| Imbalance | 115 (52.0) |
| Dizziness | 114 (51.6) |
| Fatigue | 101 (45.7) |
| Nausea | 95 (43.0) |
| Sensitivity to light | 86 (38.9) |
| Sleeping problems (insomnia, sleeping too much &/or too little) | 72 (32.6) |
| Irritability | 70 (31.7) |
| Depression | 68 (30.8) |
| Vision changes (blurry vision, double vision, unspecified) | 63 (28.5) |
| Anxiety | 51 (23.1) |
| Mental fogginess (“don't feel right,” dazed, feeling in a fog) | 50 (22.6) |
| Sensitivity to noise | 49 (22.2) |
| Neck pain | 40 (18.1) |
| Tinnitus | 37 (16.7) |
| Increased emotionality | 36 (16.3) |
| Pressure in the head | 29 (13.1) |
| Lightheadedness | 27 (12.2) |
| Personality changes | 24 (10.9) |
| Vertigo | 22 (10.0) |
| Confusion | 21 (9.5) |
| Numbness | 21 (9.5) |
| Feeling slowed down | 17 (7.7) |
| Learning difficulties | 17 (7.7) |
| Loss of appetite | 14 (6.3) |
| Vomiting | 10 (4.5) |
| Panic attacks | 8 (3.6) |
| Thinking time increased | 8 (3.6) |
| Frustrated | 7 (3.2) |
| Increased sensitivity to alcohol | 7 (3.2) |
| Speech problems | 7 (3.2) |
| Seizures | 3 (1.4) |
| Aggression | 2 (0.9) |
| Problem-solving difficulties | 2 (0.9) |
| Response speed slowed | 2 (0.9) |
| Restlessness | 2 (0.9) |
| Stomach ache | 2 (0.9) |
| Apathy | 1 (0.5) |
| Total | 221 |
Imaging and Other Diagnostic Tests Prior to or After the Concussion
Thirty-two patients (14.5%) had a CT scan prior to consultation, and 71 (32.1%) underwent MRI before consultation. As a result of the consultation, an additional 69 (31.2%) had MRI. Twelve (5.4%) underwent MRI both before and after consultation. A total of 184 patients (83.3%) had MRI and/or CT, either before or after the consultation. Some incidental findings were discovered by the imaging, including an arachnoid cyst in 8 (3.6%) and Chiari malformation in 10 (4.5%). One patient required surgical decompression for the Chiari malformation after the concussion. A significantly greater proportion of PCS sufferers in this study had an arachnoid cyst and Chiari malformation compared with the prevalence of these conditions in the general population: 4.5% of patients had a Chiari malformation, while the prevalence of Chiari malformation in the general population is 0.00078% (p < 0.0000005).29,35 Arachnoid cysts were present in 3.6% of patients, whereas the prevalence in the general population ranges from 0.006% to 1.7% (p < 0.00005 and p = 0.09, respectively). Thirty-eight patients (17.2%) underwent neuropsychological testing to evaluate cognitive deficits: 10 (26.3%) had results within the normal ranges, 25 (65.8%) had cognitive deficits, and in 3 (7.9%) the results are unknown.
Associations Between Predictors
The results from the analysis of associations between predictors are illustrated in Table 7. The table has 1 entry for each unique pair of predictors. Since associations are symmetric, only the lower triangle of the matrix is included. The positive symbols (“+”) indicate positive associations, while the 1 negative symbol (“−”) indicates a negative association, where the sign of the association is determined with reference to the coding of the variables described in the table caption. For example, the significant positive association between the number of reported symptoms and cause of concussion (coded as 0 for sports and 1 for other causes), indicates that concussions suffered during sporting activity are associated with fewer reported symptoms compared with other causes. In contrast, the significant negative association between the number of previous concussions and sex (coded as 0 for males and 1 for females) indicates that females reported fewer concussions than males prior to the key injury that led to the consultation. Of 45 tests of association, 17 were found to be statistically significant after controlling the false discovery rate (at p = 0.05).
Associations among predictor variables*
| Variable | Age | Sex | No. of Previous Concussions | Previous Psychiatric Conditions | Migraine History | Sports/Other | Extracranial Injuries | Amnesia/LOC | Litigation |
|---|---|---|---|---|---|---|---|---|---|
| Sex | +† | ||||||||
| No. of previous concussions | × | −† | |||||||
| Previous psychiatric conditions | +† | × | × | ||||||
| Migraine history | +† | +† | × | +† | |||||
| Sports/other | +† | × | × | +† | × | ||||
| Extracranial injuries | × | × | × | × | × | +† | |||
| Amnesia/LOC | × | × | × | × | × | × | × | ||
| Litigation | +† | × | × | × | × | +† | × | × | |
| No. of symptoms | × | +† | × | +† | × | +† | +† | +† | +† |
+ = positive association, − = negative association, × = insignificant association.
The table lists the results of statistical tests of association between each pair of predictor variables. The sign of the association is with respect to the following coding of the variables: sex (male = 0, female = 1); number of previous concussions; prior history of psychiatric conditions (no = 0, yes = 1); migraine history (no = 0, yes = 1); sports/other (cause of injury; sports = 0, other = 1); extracranial injuries (absent = 0, present = 1); amnesia/LOC (no = 0, yes = 1); litigation (patient involved in litigation; no = 0, yes = 1); and number of reported symptoms. Thus, for example, the positive association between sex and migraine history reflects the more frequent reporting of migraines among the females in this data set.
Values significant following false discovery rate correction (p = 0.05).
Principle Components Analysis
The principal components analysis revealed only a weak overall correlation among the predictors, with the strongest 3 components accounting for 22.4%, 12.8%, and 11.4% of the variance among the standardized predictors, and the remaining 7 components falling between 9.4% and 5.3%. Visual inspection of a 3D scatterplot of the first 3 components revealed no discernible clustering pattern among the predictors. Only the first principal component was statistically significant at p < 0.05 as assessed by per mutation testing, with all component scores positive except for number of previous concussions (Table 8). Thus, the single axis that explained the greatest variance—although still relatively little at 22.4%—corresponded in one direction to older females with a history of migraines and previous psychiatric conditions, reporting many symptoms following a concussion that included extracranial injuries and either amnesia or LOC, sustained in a nonsporting activity, followed by litigation, but reporting fewer prior concussions. In the opposite direction, the axis describes younger males with no history of migraines or previous psychiatric conditions, reporting fewer symptoms following a concussion that did not include extracranial injuries, amnesia, or an LOC, sustained during a sporting activity, not resulting in litigation, but reporting a larger number of prior concussions.
Coefficient scores for the one significant principal component*
| Variable | Component Score |
|---|---|
| Age | 0.441 |
| Cause of injury (sports = 0, other = 1) | 0.407 |
| No. of symptoms | 0.354 |
| Prior history of migraines (no = 0, yes = 1) | 0.332 |
| Sex (male = 0, female = 1) | 0.305 |
| Extracranial injuries (no = 0, yes = 1) | 0.290 |
| Involved in litigation (no = 0, yes = 1) | 0.284 |
| Prior history of psychiatric condition (no = 0, yes = 1) | 0.278 |
| Amnesia/LOC reported (no = 0, yes = 1) | 0.196 |
| No. of previous concussions | −0.178 |
This component accounted for 22.4% of the variance among the standardized predictors. The variables are listed in order of decreasing component score magnitude.
Predicting the Number of Symptoms
The results of the multiple linear regression analysis aimed at predicting the number of reported symptoms is shown in Table 9. Of the 9 predictors tested for the ability to predict the number of symptoms, only the patient involvement in litigation was statistically significant (p = 0.006), with the presence of extracranial injuries (p = 0.054) and the occurrence of amnesia or LOC (p = 0.063) trending toward significance. A stepwise regression procedure yielded a model with the same 3 predictors in addition to sex as statistically significant in the most parsimonious model. The results suggest that involvement in litigation, presence of extracranial injuries, occurrence of amnesia and/or LOC, and being female were the most important predictors for reporting a larger number of symptoms. These results are consistent with the correlations reported in Table 7.
Results of multiple linear regression to predict the number of symptoms*
| Variable Name | Standardized Regression Coefficient ± SE | Statistical Significance | |
|---|---|---|---|
| Full Model | Stepwise Regression | ||
| Litigation | 0.091 ± 0.033 | 0.006 | 0.002 |
| Extracranial Injuries | 0.064 ± 0.033 | 0.054 | 0.023 |
| Amnesia/LOC | 0.060 ± 0.032 | 0.063 | 0.034 |
| Prior history of psychiatric conditions | 0.054 ± 0.033 | 0.104 | – |
| Sex | 0.053 ± 0.034 | 0.117 | 0.032 |
| Cause of injury (sports or other) | 0.033 ± 0.035 | 0.347 | – |
| Prior history of migraines | 0.029 ± 0.034 | 0.401 | – |
| No. of previous concussions | −0.022 ± 0.033 | 0.501 | – |
| Age | −0.017 ± 0.037 | 0.635 | – |
SE = standard error.
The list of symptoms is sorted in order of decreasing statistical significance in the full model. The standardized coefficients from the stepwise regression were not significantly different from those in the full model.
Predicting the Duration of PCS
The Cox proportional hazards model identified the reported number of symptoms as the only statistically significant predictor of PCS duration (p = 0.0045), with a linear coefficient of −0.302, amounting to a 25% reduction in the recovery rate for each additional symptom. The number of symptoms was not log-transformed for this analysis, as a better fit was achieved using the original linear scale. Even with this predictor removed, none of the other predictors approached statistical significance. However, the weak performance of the other predictors is understandable given the under-constrained condition of the model, with only 25 (11.6%) of the 215 patients having a known recovery time and all others with ongoing PCS that may or may not resolve at some future unknown date.
Predictors of ICD-10 and DSM-IV
When the analyses reported above were repeated using only the subsets of the data corresponding to the ICD-10 (146 patients) and DSM-IV (101 patients) definitions of PCS, the most significant trend was toward a general reduction in statistical significance commensurate with the substantially smaller numbers of patients. For the analysis of predictor associations, only 6 of the 45 associations were significant for the ICD-10 set, and only 1 for the DSM-IV set. For the principal components analysis, the results were qualitatively identical to the main analysis, with some minor variations in the magnitudes of the factor scores. For the models aimed at predicting the number of symptoms and the PCS duration, the relative ordering of importance of the variables was largely preserved with one interesting exception: for both predictive models, for both the ICD-10 and DSM-IV sets, the one qualitative difference observed was an increase in the importance of the number of previous concussions. This variable had not been among the highest ranked in any of the main analyses using our new definition of PCS, but was the most important predictor of the number of symptoms and the PCS duration (though still not statistically significant) for both the ICD-10 and DSM-IV subsets.
Discussion
The present study is unique because a uniform definition of concussion was applied to all cases based on the 4th International Consensus Conference on Sports Concussions27 and because a new definition of PCS was used consisting of any 3 symptoms lasting at least 1 month. This is a novel, working definition of PCS that is based on the internationally accepted definition of concussion, which has not been well integrated into research. Also, we specifically excluded brain injuries more severe than concussion, such as intracranial hemorrhages, contusions, and focal neurological deficits. Table 3 shows that 7.4% of cases were excluded because of focal intracranial pathology. Most previous studies of PCS used other definitions of concussion and PCS and included a heterogeneous population of cases, including those with focal pathological lesions and a GCS score of 13–15 and labeled mild TBI.
As observed in other studies, the most common symptom of PCS in the present study was headache.16,19,22 However, unlike other studies, our patients had a much higher prevalence of previous concussions, with 76.9% in the current series having at least 1 previous concussion (Fig. 2), compared with a range of 22%–53% in other studies.6,12,15,30,32 Many of these studies were based on emergency department visits, or visits to the offices of pediatricians or general practitioners, and thus focused on a different population of patients compared with the present study. The differences in the incidence of previous concussions may also relate to the variations in definition of concussion. Our aim was to achieve complete documentation of all previous concussions because we considered this essential to the long-term goal of our project, which is to determine the role of previous concussions and PCS as antecedents of CTE.
Many other studies have shown an increased risk of PCS in females,4,6,7,13,26,32 and this trend is noted in our results as well. In the present series, females experienced significantly more persistent symptoms than males. However, being female was not a strong predictor of the duration of PCS. These data may suggest that PCS was more severe in females because they had more symptoms than males, but females reported fewer previous concussions, which are associated with prolonged recovery and an increased risk of PCS.6,7,19,27,44 Thus, it is possible females are more susceptible to PCS as fewer concussions can trigger PCS. Females also were more likely to have prior migraine headaches, which have previously been linked to an increased risk of PCS.19 However, whether these factors increased the risk of developing PCS in this study is uncertain.
There are studies that support a decreased risk of PCS after concussions in sports and recreation.4,6,32 In the present study, the cause of injury did not predict the number of symptoms from PCS sufferers or the duration of PCS.
Only some of the factors previously reported to predict PCS predicted the number of symptoms and the duration of PCS. Involvement in litigation, presence of extracranial injuries, amnesia and/or LOC at the time of concussion, and being female were more likely to report a larger number of symptoms in this study and are variables that have been previously described to be associated with an increased risk of developing PCS (Table 2). Additionally, a prior history of psychiatric conditions or migraines, cause of injury, number of previous concussions, and age did not significantly predict the symptom number. Further comment is limited on this finding as there may not have been sufficient statistical power.
Another interesting finding is that the number of symptoms was the sole significant predictor for the duration of PCS. For each additional symptom experienced, the recovery rate is reduced by 25%, which is indeed a major factor. This is a crucial finding as all 3 PCS definitions (ICD-10, DSM-IV, and the definition used in this study) require 3 symptoms to fulfill the criteria of the diagnosis of PCS. Thus, it is likely that these criteria need to be reconsidered and these definitions exclude many other patients who are experiencing less than 3 postconcussive symptoms after their concussion. However, the average number of persistent symptoms experienced was approximately 8, which is well above the required 3.
When comparing the PCS definition used in this study to the ICD-10 and DSM-IV, the predictive models were largely preserved, except the number of previous concussions was of higher importance in ICD-10 and DSM-IV in predicting the number of symptoms. It is possible that these definitions are biased toward more severe forms of concussions as multiple concussions can prolong recovery.27
The findings in this study show distinct differences when different definitions of PCS are used. The DSM-IV definition is no longer in practice and the validity of the ICD-10 definition is questionable. The evidence that supports this definition is lacking. As noted in Table 1, the ICD-10 states that the head trauma that causes PCS is “… usually sufficiently severe to result in loss of consciousness.”42 However, less than 15% of concussions result in LOC.37,38 Consequently, it is important for future research to refine the definition of PCS. We are currently working on a follow-up study on this population to determine other factors that may contribute to persistent PCS.
Another important difference was that a significantly greater proportion of PCS sufferers in the present study had incidental findings of arachnoid cysts and Chiari malformations compared with the prevalence in the general population. We are uncertain about the nature of the relationship between these incidental findings and PCS. However, the relationship between trauma and the onset of symptoms from Chiari malformation has been previously recognized.40 Due to the mass effect on the brainstem caused by Chiari malformations and the mass effect on the adjacent brain caused by arachnoid cysts, it is possible that the brain moves differently in response to rotational concussive forces in patients with these conditions, and is at greater risk for concussion.
Limitations of the Study
One of the major limitations in a retrospective study of this nature is a bias in the information that is collected. For example, those with more information in their patient chart had more symptoms recorded, and therefore, the number of persistent symptoms may not accurately represent the severity of PCS. Another limitation is that many patients had only 1 office visit, and therefore the actual duration of the PCS in those patients was longer than recorded. Prospective longitudinal studies on PCS are needed to improve the understanding of this condition, such as its potential role as an antecedent condition leading to CTE.36 We have instituted a long-term follow-up study of the patients described in the present report. Lastly, this study was a retrospective cohort review of patients from 1 concussion specialist. The criteria for referral to this concussion specialist are not known, but it is likely that more complicated and persistent cases of PCS were referred. Therefore, a selection bias exists and the patients in this study would not be representative of the general population of concussed patients. Also, as noted, the use of a new definition of PCS should be considered as a potential factor affecting generalizability, and therefore caution must be taken when generalizing these results.
Acknowledgments
Funds for this study were provided by the Canadian Concussion Centre at the Toronto Western Hospital, which is funded by the Toronto General and Western Hospital Foundation.
References
- 1↑
Alves W, , Macciocchi S, & Barth J: Postconcussive symptoms after uncomplicated mild head injury. J Head Trauma Rehabil 8:48–59, 1993
- 2↑
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders ed 4 Washington, American Psychiatric Association, 1994
- 3↑
American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders ed 5 Arlington, American Psychiatric Publishing, 2013
- 4↑
Bazarian JJ, & Atabaki S: Predicting postconcussion syndrome after minor traumatic brain injury. Acad Emerg Med 8:788–795, 2001
- 5↑
Bazarian JJ, , Blyth B, , Mookerjee S, , He H, & McDermott MP: Sex differences in outcome after mild traumatic brain injury. J Neurotrauma 27:527–539, 2010
- 6↑
Bazarian JJ, , Wong T, , Harris M, , Leahey N, , Mookerjee S, & Dombovy M: Epidemiology and predictors of postconcussive syndrome after minor head injury in an emergency population. Brain Inj 13:173–189, 1999
- 7↑
Benson BW, , Meeuwisse WH, , Rizos J, , Kang J, & Burke CJ: A prospective study of concussions among National Hockey League players during regular season games: the NHL-NHL-PA Concussion Program. CMAJ 183:905–911, 2011
- 8↑
Binder LM: Persisting symptoms after mild head injury: a review of the postconcussive syndrome. J Clin Exp Neuropsychol 8:323–346, 1986
- 9↑
Binder LM, & Rohling ML: Money matters: a meta-analytic review of the effects of financial incentives on recovery after closed-head injury. Am J Psychiatry 153:7–10, 1996
- 10↑
Carroll LJ, , Cassidy JD, , Holm L, , Kraus J, & Coronado VG: Methodological issues and research recommendations for mild traumatic brain injury: the WHO Collaborating Centre Task Force on Mild Traumatic Brain Injury. J Rehabil Med 43:43 Suppl 113–125, 2004
- 11↑
Chan RC: How severe should symptoms be before someone is said to be suffering from postconcussion syndrome? An exploratory study with self-reported checklist using Rasch analysis. Brain Inj 19:1117–1124, 2005
- 12↑
Dischinger PC, , Ryb GE, , Kufera JA, & Auman KM: Early predictors of postconcussive syndrome in a population of trauma patients with mild traumatic brain injury. J Trauma 66:289–297, 2009
- 13↑
Edna TH, & Cappelen J: Late postconcussional symptoms in traumatic head injury. An analysis of frequency and risk factors. Acta Neurochir (Wien) 86:12–17, 1987
- 14↑
Evered L, , Ruff R, , Baldo J, & Isomura A: Emotional risk factors and postconcussional disorder. Assessment 10:420–427, 2003
- 15↑
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 142:546–553, 2003
- 16↑
Guskiewicz KM, , Weaver NL, , Padua DA, & Garrett WE Jr: Epidemiology of concussion in collegiate and high school football players. Am J Sports Med 28:643–650, 2000
- 18↑
Jacobs B, , Beems T, , Stulemeijer M, , van Vugt AB, , van der Vliet TM, & Borm GF, et al.: Outcome prediction in mild traumatic brain injury: age and clinical variables are stronger predictors than CT abnormalities. J Neurotrauma 27:655–668, 2010
- 20
King NS: Emotional, neuropsychological, and organic factors: their use in the prediction of persisting postconcussion symptoms after moderate and mild head injuries. J Neurol Neurosurg Psychiatry 61:75–81, 1996
- 21↑
King NS, & Kirwilliam S: Permanent postconcussion symptoms after mild head injury. Brain Inj 25:462–470, 2011
- 22↑
Lau BC, , Kontos AP, , Collins MW, , Mucha A, & Lovell MR: Which on-field signs/symptoms predict protracted recovery from sport-related concussion among high school football players?. Am J Sports Med 39:2311–2318, 2011
- 23↑
Levin HS, , Hanten G, , Roberson G, , Li X, , Ewing-Cobbs L, & Dennis M, et al.: Prediction of cognitive sequelae based on abnormal computed tomography findings in children following mild traumatic brain injury. J Neurosurg Pediatr 1:461–470, 2008
- 24↑
Lishman WA: Physiogenesis and psychogenesis in the ‘postconcussional syndrome’. Br J Psychiatry 153:460–469, 1988
- 25↑
Luis CA, , Vanderploeg RD, & Curtiss G: Predictors of postconcussion symptom complex in community dwelling male veterans. J Int Neuropsychol Soc 9:1001–1015, 2003
- 26↑
McCauley SR, , Boake C, , Levin HS, , Contant CF, & Song JX: Postconcussional disorder following mild to moderate traumatic brain injury: anxiety, depression, and social support as risk factors and comorbidities. J Clin Exp Neuropsychol 23:792–808, 2001
- 27↑
McCrory P, , Meeuwisse WH, , Aubry M, , Cantu B, , Dvorák J, & Echemendia RJ, 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 47:250–258, 2013
- 28↑
McCullagh S, , Oucherlony D, , Protzner A, , Blair N, & Feinstein A: Prediction of neuropsychiatric outcome following mild trauma brain injury: an examination of the Glasgow Coma Scale. Brain Inj 15:489–497, 2001
- 29↑
Meadows J, , Kraut M, , Guarnieri M, , Haroun RI, & Carson BS: Asymptomatic Chiari Type I malformations identified on magnetic resonance imaging. J Neurosurg 92:920–926, 2000
- 30↑
Mooney G, , Speed J, & Sheppard S: Factors related to recovery after mild traumatic brain injury. Brain Inj 19:975–987, 2005
- 31↑
Paniak C, , Reynolds S, , Toller-Lobe G, , Melnyk A, , Nagy J, & Schmidt D: A longitudinal study of the relationship between financial compensation and symptoms after treated mild traumatic brain injury. J Clin Exp Neuropsychol 24:187–193, 2002
- 32↑
Ponsford J, , Willmott C, , Rothwell A, , Cameron P, , Kelly AM, & Nelms R, et al.: Factors influencing outcome following mild traumatic brain injury in adults. J Int Neuropsychol Soc 6:568–579, 2000
- 33↑
Prigatano GP, & Gale SD: The current status of postconcussion syndrome. Curr Opin Psychiatry 24:243–250, 2011
- 34↑
Shenton ME, , Hamoda HM, , Schneiderman JS, , Bouix S, , Pasternak O, & Rathi Y, et al.: A review of magnetic resonance imaging and diffusion tensor imaging findings in mild traumatic brain injury. Brain Imaging Behav 6:137–192, 2012
- 35↑
Speer MC, , Enterline DS, , Mehltretter L, , Hammock P, , Joseph J, & Dickerson M, et al.: Chiari Type I malformation with or without syringomyelia: prevalence and genetics. J Genet Couns 12:297–311, 2003
- 36↑
Tator CH: Chronic traumatic encephalopathy: how serious a sports problem is it?. Br J Sports Med 48:81–83, 2014
- 37↑
Tator CH: Concussions and their consequences: current diagnosis, management and prevention. CMAJ 185:975–979, 2013
- 38↑
Tator CH, & Davis H: The postconcussion syndrome in sports and recreation: clinical features and demography in 138 athletes. Neurosurgery 75:Suppl 4 S106–S112, 2014
- 39↑
Wall PL: Posttraumatic stress disorder and traumatic brain injury in current military populations: a critical analysis. J Am Psychiatr Nurses Assoc 18:278–298, 2012
- 40↑
Wan MJ, , Nomura H, & Tator CH: Conversion to symptomatic Chiari I malformation after minor head or neck trauma. Neurosurgery 63:748–753, 2008
- 41↑
Wenden FJ, , Crawford S, , Wade DT, , King NS, & Moss NE: Assault, post-traumatic amnesia and other variables related to outcome following head injury. Clin Rehabil 12:53–63, 1998
- 42↑
World Health Organization: ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines Geneva, World Health Organization, 1992
- 43↑
World Health Organization: ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic Criteria for Research Geneva, World Health Organization, 1992
- 44↑
Zemek RL, , Farion KJ, , Sampson M, & McGahern C: Prognosticators of persistent symptoms following pediatric concussion: a systematic review. JAMA Pediatr 167:259–265, 2013
Disclosures
The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Author Contributions
Conception and design: Tator, HS Davis. Acquisition of data: HS Davis. Analysis and interpretation of data: Tator, HS Davis, Dufort, Hiploylee. Drafting the article: Tator, HS Davis, Dufort, Hiploylee. Critically revising the article: Tator, HS Davis, Tartaglia, KD Davis, Ebraheem, Hiploylee. Reviewed submitted version of manuscript: Tator, HS Davis, Hiploylee. Approved the final version of the manuscript on behalf of all authors: Tator. Statistical analysis: HS Davis, Dufort. Administrative/technical/material support: Hiploylee. Study supervision: Tator.
