We conducted a retrospective study to examine sports-related concussion in relation to cognitive-related symptoms among children, following institutional review board (IRB) approval. To assess the response or outcome variable, namely cognitive-related symptom, in relation to health disparities indicators, a retrospective cohort design (case-only) was used. This design is effective, in spite of its retrospective nature, in obtaining evidence on the assessed relationship with risk ratio effect size. Using preexisting data from our electronic medical records, we examined all cases with concussion as our cohort.
Study Population
The study population comprised children aged 2–19 years, who were diagnosed with sports-related concussion between January 2007 and June 2014. This consecutive sample consisted of all children regardless of race, ethnicity, and sex who met our eligibility criteria. To be included in this sample, participants were (a) diagnosed with concussion or TBI between 2007 and 2014, (b) children 2–19 years of age, (c) residing in DE Valley, and (d) underwent SCAT evaluation. We utilized consecutive sampling technique, thus including all children with sports-related concussion during the study period, between 2007 and mid-2014. Following the eligibility criteria, there were n = 1429 in our sample, with racial distribution indicating the following: White 1146 (80.2 %), Black/African American 170 (11.9 %), and other 113 (7.9 %).
Study Variables
Concussion which is derived from Latin “concutere” refers to violent shaking of the brain, and represents the most common and less severe or mild traumatic brain injury (mTBI). Concussion reflects the disturbance in brain function as a result of direct or indirect force to the head. A direct impact to the brain is not required for concussion to occur, given the nature of this condition to be functional rather than structural injury due to shear stress to brain tissue as an outcome of rotational or angular forces [18]. The impact from concussion may result in bruising, blood vessels damage, and injury to the nerves. The third International Conference on Concussion in sports characterized concussion as a complex pathophysiologic process affecting the brain, induced by traumatic biomechanical forces [18].
Sport-Related Concussion Ascertainment
Concussion as related above is mild traumatic brain injury and could be characterized as invisible injury due to rapid acceleration or deceleration of brain tissues within the skull. These impacts may result in changes in the shape of the brain, brain stretch as well as brain cell damage. Additionally, the impulsive force may be transmitted beyond the head involving the face, neck, or elsewhere. Participants were involved in organized sports at school (football, baseball, softball, basketball, cheerleading, gymnastics, soccer) and were assessed using the symptom checklist. Girls who played soccer, softball, and basketball while boys who played football presented with majority of mTBI. To assist in the diagnosis of concussion, we used many tools to increase the sensitivity and specificity of the ascertainment. Specifically, we applied the post-concussion symptom scale, and concussion symptom inventory (CSI). Of these two instruments, we based our ascertainment on CSI since this is an empirically derived symptom checklist. Additionally, we used sideline assessment tools mainly Sport Concussion Assessment Tool (SCAT2) since it combines multiple assessment tools as well as physical examination to determine the possibility of concussion as mTBI. Therefore, our final assessment was based on SCAT2, the source for our data on CRS.
Children as patients provided the information during history taking and physical examination clinical encounter. However, for the every young ones, parents and caregivers serve as proxies to the children as respondents. The mean duration between injury and assessment (rehabilitation admission) at our facility was 7.6 days, SD 4.3.
Other variables studied included cognitive-related symptom, race, ethnicity, sex, age, insurance coverage, and length of hospitalization, as well as other variables that were available but not included in the final analysis, namely language, comorbidity, means of hospital arrival, and year of concussion.
Cognitive-Related Symptom (CRS) Ascertainment
Cognitive-related symptom was available as a string variable and was characterized as memory, orientation, and attention deficit. Specifically, we grouped symptoms and manifestation based on the variables in the medical records that reflected cognitive-related symptoms including a feeling of mental fogginess, problems concentrating, problems remembering, visual disturbance, amnesia, dizziness, a feeling of being slowed down, a feeling of increased emotion, and balance problems. We extracted these variables and used a binary scale to describe cognitive-related symptoms (0 = absence, 1 = presence). Cognitive-related symptom, based on our approach, indicated several attributes for a given case indicative of magnitude or degree of symptoms.
Other Variables Ascertainment (Race, Sex, Insurance, Age at Injury)
Race and ethnicity were self-reported. We collected data on those who self-identified as White, Black/African American, and some other race. Additional races represented in our data consisted of American Indian/Alaska Native, Asian, and Hawaiian Native/Other Pacific Islander. Similarly, participants self-identified as non-Hispanic/Latino or Hispanic/Latino. Self-reported variables have been utilized and deemed to be reliable with at least an estimated 80 % confidence [19]. Sex was available as male and female and was measured on a nominal scale. As recommended by the American Medical Association, we used boys and girls to represent the sex of our patients [20]. Furthermore, sex was critical to stratify sports-related concussion and was included in prevalence risk ratio test and adjusted risk ratio test. We examined insurance as both a proxy for income and access to available care following concussion. The insurance variable was classified into commercial (private), public (Medicaid), and uninsured/self-pay. We assessed the effect of insurance on cognitive-related symptoms using private/commercial insurance as the reference group. Data from our medical records were available for age on a continuous scale. To assess the effect of age on CRS as well as concussion, we utilized age as a continuous variable and transformed the age variable into categories, namely 2–9 years, 10–14 years, and 15–19 years. These ages at injury were treated as continuous variables which were later categorized to assess the impact of age on concussion as well as cognitive-related symptom.
Statistical Analysis
The categorical and nominal variables including the prevalence of CRS were summarized using frequency and percentages, while the continuous scale variable was summarized using median and interquartile range (IQR) depending on the normality assumption. The chi-squared statistic, Pearson, and where necessary Fisher’s exact were used to characterize the categorical and nominal variables with concussion and cognitive-related symptoms. To examine the association between race, ethnicity sex, insurance, age, and age group with CRS, univariable and multivariable log binomial regression models were used. We built a multivariable log binomial regression model to examine the simultaneous effect of race, sex, insurance, age, and age group on CRS. We used forward loading and backward elimination and included all variables that were significant at 0.25 type I error, potential confounding variables with 10 % difference between the crude and stratified association and those that had a biologic or clinical relevance, such as age and sex. The significance level for the univariable model was 0.05 (95 % confidence interval), while the significance level for the multivariable model was 0.01 (99 % confidence interval) as to minimize type I error that may result from multiple comparison in the multivariable log binomial regression model. All tests were two-tailed, and the entire analyses were performed using Stata version 13.0 (Stata Corporation, College Station, Texas) [21].