Abstract
To examine initial and longitudinal health-related quality of life (HRQOL) in adolescent sports-related concussion (SRC) patients, a prospective observational case-series study was conducted among adolescent SRC patients who were evaluated at a multi-disciplinary pediatric concussion program. Health-related quality of life was measured using the child self-report Pediatric Quality of Life Inventory (PedsQL) generic score scale (age 13–18 version) and the PedsQL Cognitive Functioning scale. Initial and longitudinal HRQOL outcomes were compared between patients who did and did not develop post-concussion syndrome (PCS). A total of 63 patients met the inclusion criteria during the study period. The mean age of the cohort was 14.57 years (standard deviation, 1.17) and 61.9% were male. The median time from injury to initial consultation was 6.5 days (interquartile range, 5, 11). At initial consultation, impairments in physical and cognitive HRQOL but not social or emotional HRQOL were observed. Initial symptom burden and length of recovery were associated with greater impairment in physical and cognitive HRQOL. Patients who went on to develop PCS had significantly worse physical and cognitive HRQOL at initial consultation and demonstrated a slower rate of recovery in these domains, compared with those who recovered in less than 30 days. Adolescent SRC was associated with HRQOL impairments that correlated with clinical outcomes. No persistent impairments in HRQOL were detected among patients who achieved physician-documented clinical recovery. Future studies are needed to evaluate the clinical utility of HRQOL measurement in the longitudinal management of adolescent SRC and PCS patients.
Introduction
O
Health-related quality of life (HRQOL) is a patient-reported outcome that aims to measure the effect of a medical condition on a patient's perception of their physical, mental, and social health. Specifically, HRQOL includes dimensions of quality of life that are within the scope of healthcare services and has emerged as an important component of pediatric clinical trials and healthcare utilization studies conducted across a wide range of medical conditions. 9,10 Previous studies have demonstrated HRQOL impairments among children and adolescents with mild to severe TBI 11 –16 ; however, the results of a systematic review indicates that there are limited studies that have examined HRQOL outcomes among pediatric SRC patients. 17 Although more recent work has aimed to shed light on this subject, 18,19 there remains an unmet need for studies that evaluate the effects of adolescent SRC on patient-reported HRQOL measured throughout the clinically supervised recovery process.
The objectives of this study were three-fold: 1) to determine initial HRQOL and clinical predictors of initial HRQOL among pediatric SRC patients referred to a multi-disciplinary pediatric concussion program; 2) to determine longitudinal changes in HRQOL during recovery among this clinical population; and 3) to compare HRQOL outcomes among SRC patients who did and did not develop PCS.
Methods
Study design and participants
A prospective case-series was conducted and additional demographic and clinical information was extracted from medical charts. Study participants were recruited from the Pan Am Concussion Program, a provincial government-funded multi-disciplinary pediatric concussion program that accepts referrals for pediatric acute concussion and PCS patients from emergency departments and primary care, pediatric, and sports medicine physicians throughout Manitoba and Canada. Study inclusion criteria included: 1) age 13–18 years of age; 2) diagnosis of a SRC; and 3) consultation less than 30 days from the date of injury. The age range of 13–18 years was chosen because this allowed all study participants to complete the same version of the HRQOL measure. Patients with an initial diagnosis of PCS, non-sports-related injuries, with moderate or severe traumatic brain injuries, who presented with traumatic spine injuries, cranial neuropathies, and seizures, and those with traumatic abnormalities detected on clinical neuroimaging studies were excluded. This study was approved by the institutional ethics review board at the University of Manitoba.
Definitions
The diagnosis of concussion was made according to the International Consensus on Concussion in Sport Group and defined as an injury caused by transmission of biomechanical forces to the brain leading to clinical symptoms affecting multiple domains of physical, cognitive, sleep, and neurobehavioral functioning. 2 Patients were diagnosed with PCS if they endorsed three or more post-concussion symptoms identified by the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision definition 20 and remained symptomatic for 1 month post-injury (30 days) or longer. Patients were classified as fully recovered when they were asymptomatic at rest according to clinical interview and self-reported symptoms (Post Concussion Symptom Scale [PCSS]), were participating in full-time school activities without symptoms, and had a normal neurological examination. Patients returning to sports also were managed using the International Consensus on Concussion in Sport Group return-to-play guidelines. 2
Clinical assessments
At initial medical consultation all patients completed a standardized data collection form that included demographic and injury data and past medical history. All patients also completed the PCSS, a symptom inventory consisting of 22 symptoms that are rated on a 7-point (0–6) Likert scale with a maximum score of 132 (i.e., 6 × 22). All patients underwent clinical history, complete physical examination, and medical clearance by one neurosurgeon. Follow-up appointments were scheduled as clinically indicated and not based on a pre-determined research schedule.
HRQOL outcome measures
Health-related quality of life was measured using the child self-report Pediatric Quality of Life Inventory (PedsQL) generic score scales (age 13–18 version) and the PedsQL Cognitive Functioning scale. The PedsQL generic score scale is a 23-item questionnaire that is completed by the patient and includes items that provide an assessment of physical, emotional, social, and school functioning. The PedsQL generates six individual scores including a total scale score (based on all 23 items), physical health score (eight items), psychosocial health score (15 items), which is comprised of scores for emotional, social, and school functioning (5 items per score), with scores ranging from 0 (poor outcome) to 100 (good outcome). 10 The PedsQL Cognitive Functioning scale includes six items and generates a Cognitive Functioning score ranging from 0 (poor outcome) to 100 (good outcome). 21 The PedsQL generic scores scale has been found to provide a valid and reliable assessment of HRQOL in healthy children and across numerous pediatric health conditions. 10,22,23 Previous research has established minimal clinically important difference (MCID) point changes for the PedsQL total and subscale scores. 22 The PedsQL Cognitive Functioning scale also has been used to measure HRQOL in a number of acute and chronic health conditions in children and correlates highly with other measures of cognitive functioning in pediatric TBI. 12 At present, the authors are not aware of any established MCID point changes for the PedsQL Cognitive Functioning scale.
Data collection
A research assistant obtained parental informed consent and patient assent after their initial clinical appointment. All study participants completed the patient-reported PedsQL generic and PedsQL Cognitive Functioning scales at the time of initial consultation. Together, these instruments take approximately 5–10 min to complete in pen and paper format. At subsequent follow-up appointments, including the final visit for medical clearance, all patients completed the PCSS and PedsQL generic and Cognitive Functioning Scales prior to seeing the neurosurgeon.
Clinical variables extracted from medical charts included age, sex (male/female), history of prior concussion (yes/no), history of headaches or migraine (yes/no), history of attention-deficit hyperactivity disorder (ADHD; yes/no), history of depression (yes/no), loss of consciousness at time of the injury (yes/no), duration of symptoms prior to initial consultation (days), and initial PCSS score. After they were deemed clinically recovered, patients were asked if they thought they had received adequate or inadequate school accommodations during their recovery process (yes/no). School accommodations in place prior to concussion were not assessed in this study.
Statistical analysis
The distributions of baseline characteristics and injury data for SRC patients included in the study were summarized using proportions for dichotomous/polytomous characteristics and means with standard deviations (SDs) for continuous characteristics. If a continuous variable was not normally distributed, it was summarized as medians with interquartile range (IQR). Characteristics between patients who did and did not subsequently develop PCS were compared using chi-square test for dichotomous data, t-test for normally distributed continuous data and rank sum test for non-normally distributed continuous data.
To determine clinical predictors of initial HRQOL, only initial physical, emotional, social, and cognitive functioning HRQOL sub-scores captured at the first patient visit were included. Linear regression using backwards elimination approach was used to determine clinical predictors of HRQOL. A priori predictors included sex, age, previous history of concussion, history of migraine or headache, initial PCSS score, previous history of ADHD, previous history of depression, adequate school accommodations, and loss of consciousness at injury. Clinical predictors that were not statistically significant were removed from the model one at a time. The clinical predictors with the highest p value were removed from the model, providing the clinical predictor did not confound any of the other associations between the outcome and remaining risk factors. This was repeated until only statistically significant clinical predictors remained.
Days between clinic visits were calculated to account for different times between visits and HRQOL measurements. Mixed generalized linear modeling using maximum likelihood estimation was used to assess predictors of changes in HRQOL scores over time. The mixed generalized linear models used an unstructured covariance matrix and included terms for random intercepts to account for varying initial HRQOL scores and random slopes representing differing rates of improvement in HRQOL scores over time. This method was necessary to account for the correlated nature of the repeated measures of HRQOL within each individual. Each domain of HRQOL was analyzed separately because not all patients were attending school during the study due to severe symptoms or school not being in session (summer or winter break). Because PedsQL school functioning scores were absent for a proportion of the study participants, total PedsQL generic scores were not assessed. Due to significant effect modification by subsequently developing PCS, results were presented as those who did and did not develop PCS. The potential confounding effects of age, sex, and initial PCSS score (transformed as the square root of initial PCSS to account for skewedness) were examined. Using backwards elimination modeling approach, each potential confounder was removed one at a time. Whichever potential confounder changed the estimate of HRQOL improvement per week the least (providing <15%) was removed. This was repeated until either all potential confounders were removed from the model or the remaining potential confounder changed the HRQOL improvement estimate by more than 15%.
For all statistical tests, a p value of <0.05 was deemed statistically significant. HRQOL results are presented as point increase (out of 100) per week.
Results
Patient characteristics
From September 1, 2013 to October 1, 2014, 84 patients aged 13–18 years from the Pan Am Concussion Program were identified for the study. A total of 17 patients did not meet the inclusion criteria (two were not injured while playing sports and 15 presented more than 1 month after their concussion) and four patients did not consent. In total, 63 patients consented to participate. Overall, 61.9% were male with a mean age of 14.57 years (SD: 1.17). The median time from injury to initial consultation was 6.5 days (IQR: 5, 11). Table 1 describes additional clinical features.
PCS, post-concussion syndrome; SD, standard deviation; ADHD, attention-deficit hyperactivity disorder; PCSS, post-concussion symptoms scale; IQR, interquartile range.
Overall, 31 (49.21%) patients met the definition for PCS during their clinical course. The median length of recovery among SRC patients who did not develop PCS was 14 days (IQR: 10, 19.5). Among those who developed PCS, the median length of recovery was 57 days (IQR: 41, 210). The median number of clinical visits among patients who achieved complete recovery was 4 (IQR: 3, 5) and was not significantly different by subsequently developing PCS (p = 0.10). Of the 63 patients, 55 (87.30%) were medically cleared and discharged from the concussion program, 3 (4.76%) remained in treatment, and 5 (7.94%) were lost to follow-up.
HRQOL at initial clinical presentation
Initial HRQOL outcomes are summarized in Table 2. At initial consultation, the physical and cognitive domains of HRQOL were found to be most impaired for SRC patients who did develop PCS, as well as for those who did not develop PCS, compared with other domains. In general, those who were subsequently diagnosed with PCS initially presented with worse HRQOL in all measured domains. Although the mean initial physical Peds QL sub-scores were below the MCID for the entire cohort and those who developed PCS, scores were not below this threshold among those who did not develop PCS in comparison to published health normative data. Mean initial emotional and social PedsQL sub-scores did not fall below the MCID for the entire cohort or among those who did or did not develop PCS. 22
Normative values and minimal clinically important differences as described by Varni and colleagues. 21,22
CI, confidence interval; HRQOL, health-related quality of life; PCS, post-concussion syndrome; SD, standard deviation.
Clinical predictors of initial HRQOL
Table 3 summarizes the relationship between clinical variables and initial HRQOL scores. A pre-injury history of ADHD was associated with better initial physical HRQOL while a higher initial PCSS score was associated with worse initial physical HRQOL. Male sex was associated with better initial emotional HRQOL while a higher initial PCSS score was associated with worse initial emotional HRQOL. Higher initial PCSS score also was significantly associated with worse initial social HRQOL.
Example of calculation: =
CI, confidence interval; HRQOL, health-related quality of life; PCSS, post-concussion symptoms scale; ADHD, attention deficit hyperactivity disorder.
Longitudinal assessment of HRQOL
Longitudinal HRQOL outcomes for the SRC patients are summarized in Table 4. There was significant effect modification by subsequently developing PCS for cognitive functioning (p = 0.001), social functioning (p = 0.010), and emotional functioning (p = 0.015), and nearly for physical functioning (p = 0.084). Therefore, the mean child-reported scores for each PedsQL sub-score were presented separately. There was no evidence of effect modification by age, sex, initial PCSS score or concussion history for any of the four HRQOL domains and these were not clinical predictors of improvement in HRQOL over time. Age and initial PCSS score and concussion history did not confound any of the relationships between HRQOL and time (measured as weeks). However, sex confounded the relationship between physical HRQOL improvement over time and having adequate school accommodations confounded both emotional and social HRQOL improvement over time. Physical, cognitive, and emotional HRQOL demonstrated the most marked improvement over subsequent weeks, particularly for those who did not develop PCS. Social HRQOL was less impaired at initial assessment and therefore improved at a slower rate, compared with the other domains.
adjusted for sex.
adjusted for school accommodations.
CI, confidence interval; HRQOL, health-related quality of life; PCS, post-concussion syndrome.
Discussion
In the present study, we assessed initial and longitudinal patient-reported HRQOL outcomes in adolescent SRC patients who were evaluated and followed to clinical recovery at a multi-disciplinary pediatric concussion program.
The results of this study suggest that adolescent SRC is associated with impairments in HRQOL that affect certain domains more than others. In this cohort of adolescent SRC patients who underwent initial assessment on average less than a week following injury, significant impairments in physical and cognitive HRQOL were observed while measures of emotional and social HRQOL remained similar to healthy adolescent samples. 10,21,22 Although the initial physical PedsQL sub-scores for the entire cohort and those who developed PCS exceeded the MCID for this sub-scale, scores among those who did not develop PCS did not exceed the MCID. 22 The mean PedsQL physical score for the entire cohort at initial consultation was 65.86, which is lower than mean scores observed among children with chronic health conditions, such as asthma, end-stage renal disease, cancer, 10 and migraine. 24 These scores also were substantially worse than those observed in a previous study of HRQOL in adolescent SRC patients that found mean PedsQL physical scores of 80.7 and 93.1 at 4 and 10 days post-injury, respectively. 18 Likewise, the mean PedsQL physical score in this cohort was worse than that observed among pediatric concussion patients evaluated 1 month post-injury (mean = 87.3). 16 In contrast to the PedsQL generic instrument, the PedsQL Cognitive Functioning Scale has been far less studied but has been used to examine cognitive functioning in children with chronic medical conditions, including pediatric TBI. 12,21 Among our cohort, PedsQL Cognitive Functioning scores at initial consultation also were substantially lower than those observed in healthy subjects. 21 These values were worse than those observed in pediatric concussion patients measured 1 month post-injury 25 and similar to values observed among pediatric patients with more severe TBIs evaluated three and 12 months post-injury. 12
Previous studies have suggested that injury severity is an important determinant of HRQOL across the pediatric TBI spectrum 11 ; however, objective measures of injury severity in concussion are lacking. Two outcome measures that may be reflective of injury severity are initial symptom burden and length of recovery. Although higher PCSS scores at initial presentation have been found to be associated with an elevated risk of prolonged recovery and PCS, 26 –29 concussion symptoms are also frequently reported among healthy adolescents and those with migraine headaches and psychiatric conditions. 30 In this study, higher initial PCSS scores were associated with worse initial physical, emotional, and social PedsQL sub-scores. Impairments in selected HRQOL domains at initial consultation also were predictive of longer recovery. Specifically, those patients who went on to develop PCS had significantly lower PedsQL physical, cognitive, and emotional sub-scores at initial consultation, compared with those patients who did not develop PCS. These findings are in agreement with one previous study that observed worse post-injury HRQOL among pediatric mTBI patients who endorsed a greater number of concussion symptoms. 14 However, these authors also found that pre-injury HRQOL was a stronger predictor of post-injury HRQOL than post-concussion symptom burden. Additional work has confirmed that injury factors are strong contributors to concussion symptoms within the first month following pediatric mTBI but that non-injury factors account more for symptoms that persist beyond 1 month. 31 Taken together, these findings suggest that there are numerous factors that impact concussion symptom burden and that future work is needed to evaluate the pre-injury and post-injury clinical variables that impact HRQOL changes following adolescent SRC.
Longitudinal assessment of HRQOL throughout the recovery process offered important insight into the heterogeneity of clinical outcomes following adolescent SRC. Although improvements were observed for a majority of HRQOL domains throughout the recovery process, the rate of HRQOL improvement was strongly related to length of clinical recovery. Interestingly, there was no evidence of effect modification by age, sex, initial PCSS score, or previous concussion history. Our results are consistent with those of a more recent multi-institutional study of pediatric concussion patients that underwent initial clinical evaluation in the emergency department setting and were longitudinally assessed through telephone and web-based surveys. 19 This study observed worse total, physical, emotional, school, and social PedsQL scores at 4 weeks post-injury among patients that endorsed a 1-point or greater worsening of three or more concussion symptoms, compared with baseline (persistent post-concussion symptoms [PPCS]), compared with those patients who did not meet this criteria at 4 weeks. Patients who did not meet the criteria for PPCS at 4 weeks demonstrated normal physical, emotional, and social HRQOL but impaired school-related HRQOL that persisted at 8 and 12 weeks post-injury. The authors suggested that persistent impairments in school HRQOL in patients without PPCS may have been attributable to persistent cognitive impairment or the result of stress related to accumulated schoolwork. In our study of patients that underwent longitudinal clinical follow-up, we found that physical, emotional, and social PedsQL scores returned to normal among patients that achieved physician-documented neurological recovery. Unfortunately, the fact that some of the patients were not attending school at the time of our study prohibited uniform assessment of school-related HRQOL and total PedsQL scores in our cohort. However, since Cognitive Functioning PedsQL scores also were observed to return to normal among clinically recovered patients in our study, these results suggest that future studies are needed to identify alternative factors that may contribute to school dysfunction at different time-points following pediatric concussion.
An important distinction between quality of life and HRQOL assessment is the inclusion of health dimensions that are within the scope of or that can be addressed by healthcare services. 9 Although there are limited studies that have examined the multi-disciplinary healthcare needs of pediatric SRC patients, 5 the modern management of PCS has evolved towards a more active approach employing rehabilitation strategies that target the pathophysiological mechanisms governing persistent concussion symptoms. 6,32 –34 The results of this study suggest that the incorporation of HRQOL measures may provide novel clinical information that could contribute to a more comprehensive assessment of patient outcomes and functioning following adolescent SRC. Previous studies have identified several obstacles to the routine clinical incorporation of HRQOL outcomes in out-patient clinical practice, including the added time and personnel required to administer and score these instruments, as well as the perception that they do not add information beyond that already available through conventional patient interviews. 35 However, other work has shown that incorporating HRQOL at initial consultation can help influence multi-disciplinary clinical decision making in conditions such as pediatric cardiac and rheumatological disease. 36,37 Despite increasing research attention paid to pediatric acute SRC and PCS, these conditions continue to remain poorly understood and may indeed effect domains of patient functioning that cannot be adequately captured by symptom inventories, clinical history, and physical examination findings. This study demonstrates that HRQOL instruments can be easily incorporated into the longitudinal management of patients at a high volume pediatric concussion program. However, future studies are needed to examine whether HRQOL measures can help aid in the early identification of patients who are at risk of worse outcomes and whether multi-disciplinary targeted rehabilitation strategies can help improve HRQOL outcomes in these populations.
The results of this study must be interpreted in light of several important limitations. First, this study recruited patients from a tertiary pediatric concussion program that likely selected for patients with more severe injuries who were more likely to develop PCS and demonstrate greater impairment in HRQOL. Second, since we completed this study we have demonstrated that clinical factors such the development of vestibulo-ocular dysfunction, post-injury psychiatric outcomes, and post-traumatic migraine headaches can impact recovery and treatment responses following SRC in our clinical population. 26,32,38 These outcomes were not collected in this cohort but should be considered in future studies. Third, and as previously discussed, because a substantial proportion of the patients were not attending school during the study period, PedsQL school sub-scores and total PedsQL generic scale scores could not be uniformly assessed thus limiting comparisons to other studies.
Fourth, this study only examined self-reported and not parent-reported measures of patient HRQOL. Previous studies have shown that proxy reported measures of HRQOL are not equivalent to patient-reported measures, especially for social and emotional HRQOL domains. 39 Concordance among parent- and self-reported measurements of HRQOL also has been found to be poor among children with mTBI. 25 Indeed, poor recognition and under-reporting of concussion symptoms remains a persistent concern with respect to pediatric SRC patients that can have an impact on clinical management and the results of studies in this population. As such, future studies are needed to examine whether parent-reported outcomes can identify impairments in HRQOL that may go unrecognized or unreported by pediatric SRC patients. Fifth, although we examined the effect of previous concussion on HRQOL outcomes, the number of previous concussions and the length of recovery for these injuries were not collected and should be considered in future studies. Indeed, previous studies have demonstrated lower baseline HRQOL scores among athletes with a history of concussion. 40,41 Lastly, the present study did not include a control group and therefore it is difficult to assess to what extent HRQOL impairments following SRC were a consequence of TBI and not just injury alone. This limitation will be addressed by a prospective case-control study of HRQOL outcomes in adolescents with SRC and orthopedic injuries that is underway at the authors' institution.
In conclusion, adolescent SRC is associated with impairments in patient-reported HRQOL that improve over time and are impacted by initial symptom burden and length of recovery. Patients who underwent longitudinal clinical follow-up did not demonstrate any persistent impairment in physical, emotional, social, or cognitive HRQOL at physician-documented neurological recovery. Future prospective longitudinal case-control studies are needed to examine the TBI- and non–TBI-related factors that impact HRQOL following injury. Additional studies are also needed to examine the added value of HRQOL measurement in the longitudinal clinical management of adolescent SRC and PCS patients.
Footnotes
Acknowledgments
All phases of this study were supported by a grant from University Research Grants Program from the University of Manitoba.
Author Disclosure Statement
No competing financial interests exist.