Preseason Baseline Neurocognitive Performances and Symptom Reporting on Immediate Post-Concussion Assessment and Cognitive Testing: A Comparison of Adolescent Student-Athletes Tested in Spanish and English

Between 2009 and 2015, adolescent student-athletes (N = 47 606, age = 13–18 years) from approximately 100 middle schools and high schools across the state of Maine completed ImPACT as a component of the preseason baseline evaluation. From this larger sample, 182 student-athletes chose to complete ImPACT in Spanish. Participants were excluded if they had invalid test performance (n = 10) or self-reported past treatment for meningitis (n = 2) or seizures (n = 1). No participants reported a history of brain surgery or diagnosis of autism. This resulted in a sample of 169 participants who completed ImPACT in Spanish. A matched comparison group was drawn from student-athletes who completed ImPACT in English with valid test performances and no self-reported history of an autism diagnosis or treatment for seizures, brain surgery, or meningitis (N = 42951). Matching was conducted using case-control matching in SPSS (version 25; IBM Corp), which identified participants tested in English who matched participants tested in Spanish on demographic variables (ie, age, gender), academic history variables (ie, diagnosis of learning disability or dyslexia, history of special education), and health history variables (ie, concussion history, self-reported diagnosis of ADD or ADHD, previous treatment for a psychiatric condition such as depression or anxiety, and previous treatment for a substance use disorder, headaches, or migraines). This case-control matching procedure revealed exact matches tested in English for 167 participants tested in Spanish and 2 close matches that varied in age by ± 2 years. The final sample consisted of 169 participants tested in Spanish (age = 15.82 ± 1.11 years, range = 13–18, 55.0% boys) and 169 participants tested in English (mean = 15.82 ± 1.11 years, range = 13–18; 55.0% boys). Data regarding race and ethnicity were not available.

Participants completed a self-report survey of sport affiliation and health and academic history information, including questions on whether they had “been diagnosed with ADD/ADHD” or “had problems with ADD/hyperactivity,” been “diagnosed with dyslexia” or been “diagnosed with a learning disability,” attended special education classes, or repeated 1 or more years of school. They were also asked about the number of lifetime diagnosed concussions and past treatment for headaches, migraines, epilepsy or seizures, brain surgery, meningitis, substance or alcohol use, or a psychiatric condition (eg, anxiety or depression). All student-athletes completed ImPACT as a preseason baseline in a group setting in high school computer laboratories, administered by the local athletic training staff. Testing data were merged into a statewide data set. Institutional review board approval for use of this deidentified database was obtained from Colby College, and an exemption was obtained from Spaulding Rehabilitation Hospital for secondary use.

Immediate Post-Concussion Assessment and Cognitive Testing consists of 6 computer-administered neuropsychological test modules and the Post-Concussion Symptom Scale (PCSS). The ImPACT scores used as dependent variables in this study were the 5 neurocognitive composites (ie, verbal memory, visual memory, visual motor speed, reaction time, and impulse control) and the PCSS total symptom score.¹⁶  For verbal memory, visual memory, and visual motor speed, higher scores indicate better performances; and for reaction time and impulse control, lower scores indicate better performances. On the PCSS, a higher score indicates greater symptom severity. We also evaluated the 22 individual PCSS items for group differences in symptom endorsement (ie, rated as 1 or more on a 0–6 scale).

Independent-samples t tests were used to compare participants tested in Spanish with participants tested in English on the 5 ImPACT cognitive test composite raw scores and, due to a non-normal distribution, a Mann-Whitney U test was used to compare these groups on the PCSS total symptom score. Significance was set at P < .002 (ie, α = 0.05/18) based on a Bonferroni correction to control our familywise type I error rate, planning on 18 pairwise comparisons: 5 ImPACT cognitive composite scores and the PCSS total symptom score were compared between language groups for the full sample and each gender separately. Post hoc analyses compared total symptom severity between boys and girls within each language group.

Information regarding participants' academic and health history, including a history of learning disability or dyslexia diagnosis; ADHD diagnosis; special education; repeated academic grade; number of prior concussions; prior treatment for headache, migraine, substance use, or psychiatric conditions; and sport affiliation at the time of baseline testing is provided in Table 1. The student-athletes tested in Spanish were active in many different sports, such as soccer (42.0%), basketball (7.1%), and tennis (7.7%), at the time of baseline assessment. The most common sport affiliations reported by student-athletes tested in English were soccer (17.8%), American football (16.6%), and basketball (9.5%).

The language groups differed on 1 of the 5 ImPACT neurocognitive composites. Student-athletes tested in Spanish performed at a lower level on the visual motor speed composite score than those tested in English (P < .001, d = 0.51, medium effect size; Table 2). The participants tested in Spanish were slower in reaction time composite score, and the group difference approached the adjusted significance threshold (P = .004, d = 0.33, small effect size). When divided by gender, language-group differences on visual motor speed were observed for boys (P = .001, d = 0.49, medium effect size) and girls (P = .001, d = 0.54, medium effect size). No group differences were present between language groups for either gender in reaction time, although boys showed a small to medium magnitude of difference (P = .012, d = 0.42); those tested in Spanish performed more slowly than those tested in English.

The groups differed in the PCSS total symptom score (P < .001, r = 0.21, small effect size), with participants tested in Spanish endorsing a greater severity of symptoms. When separated by gender, the language groups displayed similar small effect sizes for boys (P = .003, r = 0.22) and girls (P = .008, r = 0.21), and both genders approached the adjusted threshold for statistical significance. We examined gender within each language group via post hoc analysis, which identified no differences in the PCSS total symptom score for participants tested in English (U = 2969.5, P = .067, r = 0.14) or Spanish (U = 3018, P = .101, r = 0.13).

We assessed differences in endorsement rates of individual symptoms between the language groups; the results are provided in Table 3 with stratification by gender. Girls tested in Spanish endorsed 10 of the 22 individual symptoms at a greater rate than girls tested in English, and these differences were associated with predominantly large ORs for each symptom (OR range = 1.95–20.00). These symptoms included physical (ie, headaches, vomiting, nausea, balance problems, dizziness, numbness or tingling, and visual problems) and affective (ie, nervousness, sadness, and feeling more emotional) symptoms. The girls tested in English endorsed only 1 symptom (ie, irritability, OR = 2.84, large effect size) at a greater rate than the girls tested in Spanish. Compared with boys tested in English, those tested in Spanish also endorsed 10 of the 22 symptoms at greater rates, all of which were associated with large ORs (OR range = 2.52–7.39). These symptoms included many of the same physical (ie, headache, vomiting, nausea, dizziness, and numbness or tingling) and affective (ie, sadness and feeling more emotional) symptoms as the girls but also included cognitive symptoms (ie, difficulty concentrating and feeling slowed down) and a single sleep-arousal symptom (ie, sleeping more than usual). No symptoms were endorsed more frequently by boys tested in English.

Previous researchers^3,8  have compared baseline neurocognitive performances and symptom reporting of athletes tested in English or Spanish and found consistently lower levels of neurocognitive performances and greater symptom severity among those tested in Spanish. We compared a sample of adolescent student-athletes who completed baseline ImPACT testing in Spanish with a sample of student-athletes, matched on age, gender, and academic and health history, who completed their baseline testing in English. The results were in partial alignment with earlier investigations^3,8 : Spanish speakers demonstrated lower performances for the visual motor speed composite score (d = 0.51), slower performances for the reaction time composite score (d = 0.33), and greater symptom severity (r = 0.22), all of which were associated with small to medium effect sizes. The magnitude and direction of these group differences were similar when stratified by gender for visual motor speed (boys: d = 0.49; girls: d = 0.54) and total symptom severity (boys: r = 0.22; girls: r = 0.21) but showed a greater difference between language groups in reaction time for boys (d = 0.42) than girls (d = 0.21). Two of our authors are bilingual in English and Spanish and identified certain lexical and grammatical concerns in the Spanish-language instructions for the cognitive tests but none that were substantial enough to lead to a misunderstanding of task demands. Group differences observed in neurocognitive composites are not likely attributable to challenges in translation of the ImPACT instructions from English to Spanish.

In comparison with previous work,^1,2,5,7  the magnitudes of the effect-size differences between participants tested in English versus Spanish were smaller for all neurocognitive composites (see Table 4), which could be attributable to the close matching across language groups on numerous demographic characteristics (eg, age, gender), academic history variables (eg, learning disability or dyslexia, history of special education), and health history variables (eg, concussion history, ADD or ADHD, previous treatment for a psychiatric condition, substance use disorder, headaches, or migraines) that have been associated with group differences in cognitive functioning or symptom reporting. Earlier authors matched or excluded participants based on age, ADD or ADHD, learning disability,³  and educational achievement (ie, high school diploma, college degree).⁸  Matching by education led to a reduction in the effect-size differences between language groups,⁸  indicating that variables other from language may explain some differences in neurocognitive performance.

Although our effect sizes may be smaller than those of past studies, the differences observed for visual motor speed and reaction time were consistent with many earlier findings^3,4,8,20,21  on linguistic and racial or ethnic differences in baseline ImPACT performances. Previous investigators^3,8  compared participants tested in English and Spanish and found the visual motor speed composite score was associated with the largest group difference. After matching by education, the only group differences between English and Spanish first-language speakers were noted for the visual motor speed composite (for athletes with both high school and college educations) and reaction time composite (for athletes with high school educations only) scores.⁸  Group differences in baseline visual motor speed (d = 0.43) and reaction time (d = 0.42) have also been identified between African American and White collegiate athletes, with the former athletes exhibiting lower performances on both composites⁴ ; however, another examination²⁰  of participants matched on age, gender, concussion history, and sport affiliation revealed no group differences at baseline between African American and White collegiate athletes. A baseline ImPACT comparison of student-athletes tested in English and Mandarin reflected a small to medium group difference for the visual motor speed composite score (d = 0.37), with higher performances by the participants tested in Mandarin.²¹  Among the ImPACT neurocognitive composite scores, visual motor speed appears to differ most consistently across racial or ethnic and linguistic groups, albeit with varying magnitudes of difference. The group differences for the visual motor speed composite score may be related to cultural differences in the value placed on fast performances, such that Spanish-speaking cultures may value response speed less than English speakers in the United States do.²² 

Regarding total symptom reporting, the differences between student-athletes tested in English versus Spanish was significant but of small magnitude (r = 0.21). Earlier researchers^3,8  conducted parametric comparisons of symptom reporting despite a likely skewed distribution. Although calculation of a Cohen d effect size may be imprecise and less appropriate for non-normal symptom data, effect sizes for baseline symptom severity differences between athletes tested in Spanish and English in previous studies are provided in Table 4, showing negligible to medium differences (d range = 0.01–0.54). Student-athletes tested in English or Spanish were also compared for the frequencies with which they endorsed individual symptoms. Greater symptom endorsement by student-athletes tested in Spanish was evident for numerous physical (ie, headache, vomiting, nausea, dizziness, numbness or tingling) and affective (ie, sadness, feeling more emotional) symptoms across genders. Within the American Latinx community, physical symptoms have been associated with psychological distress;^23,24  if we infer that the student-athletes tested in Spanish were of predominantly pan-ethnic Latinx heritage (ie, 92.5% of US residents who speak Spanish at home identify pan-ethnically as Hispanic²⁵), greater physical symptoms at baseline could reflect a greater degree of baseline psychological distress among the student-athletes tested in Spanish, especially when considered in combination with greater endorsement frequencies of affective symptoms. Within primary care settings, Latinx patients tend to report greater physical and affective symptoms; the latter have been associated with perceived discrimination.²⁶  Our student-athletes who preferred to communicate in Spanish within a largely English-speaking cultural environment may also have perceived greater discrimination (although we did not measure this factor), which could have contributed to greater psychological distress. The differences in symptom endorsement between groups did not appear to be attributable to the translation of symptoms from English to Spanish. Our 2 bilingual authors (M.A.G.B. and J.M.M.) independently reviewed the symptoms and translated them back to English. Blind to each other's review of the symptom names, neither author found significant discrepancies in meaning between the English-language and Spanish-language symptoms.

A key strength of our work was the focus on a large sample of student-athletes tested in English and Spanish who were matched almost exactly on demographic, academic history, and health history variables that can affect ImPACT performance and symptom reporting.^1,2,5–7  Thus, we controlled for many confounding variables that could have contributed to observed group differences. Still, our investigation also involved limitations. The athletes were not matched on sport, and as shown in Table 1, group differences were present. The most common affiliation for both language groups was soccer, but student-athletes tested in Spanish participated in soccer (42.0%) at a greater rate than those tested in English (17.8%). Authors²⁷  have begun discussing environmental variables associated with symptom reporting, which may include the culture of a specific sport. Additional limitations pertain to the sampling of student-athletes tested in Spanish and the absence of contextual racial, ethnic, or linguistic data. Although we lacked data on participants' racial and ethnic identities, nearly all US residents from a Spanish-speaking household identify pan-ethnically as Latinx.²⁵  The full sample of athletes completed ImPACT as part of a larger program of baseline testing at schools in the state of Maine, and the athletes chose to complete the ImPACT battery in Spanish. The reason for this decision cannot be determined based on the available data, but we suspect it was related to a preference for communicating in Spanish rather than English. However, we did not know many characteristics that could have affected the student-athletes' level of proficiency in Spanish, including whether Spanish was their first language and whether they were formally educated in Spanish. We also did not know their proficiency in English. These factors can affect performances on cognitive testing in Spanish, and bilingualism has been specifically associated with better neurocognitive performances on ImPACT and less symptom reporting compared with monolingual Spanish speakers.³ 

The participants in our study elected to complete ImPACT in their preferred language, which may have meant that the participants tested in Spanish were less comfortable communicating in English than their same-aged peers. This preference for communicating in Spanish and taking this baseline battery of tests in Spanish might make our sample fairly unique within the US population. Data collection occurred from 2009 to 2015. In 2011, 15.6% of adolescents between the ages of 15 and 19 in the United States reported speaking Spanish at home, and of those, 82.7% reported speaking English “very well,” meaning that only 2.7% of the total adolescents surveyed were Spanish speakers without a high level of self-reported English proficiency.²⁵  These individuals may have had different acculturative experiences, which have been associated with performance on neuropsychological tests.^10–12  For example, among Spanish-speaking adults in the United States, acculturative experiences outside of the United States, such as being born or educated internationally, were associated with reduced fluid ability scores but higher crystallized ability scores when evaluated in Spanish.²⁸ 

The current findings revealed small to medium differences in ImPACT neurocognitive performances and symptom reporting between student-athletes tested in English versus Spanish, who were matched on age, gender, and academic and health history. Although small to medium in magnitude, the group differences were not negligible and could have meaning for clinical practice. First, regarding symptom reporting, symptom severity was modestly different between groups, but the rates of individual symptom endorsement appeared quite different, especially for physical and affective symptoms. Clinicians involved in the management of concussions in student-athletes who prefer to communicate in Spanish should be mindful that greater symptom reporting may occur among this group and that more physical symptoms could be related to psychological distress. Second, student-athletes tested in Spanish also had lower mean performances on all neurocognitive composite scores, but the only significant group differences were for visual motor speed and reaction time. The group differences in visual motor speed equated to 0.5 SD in magnitude (d = 0.51), which could lead to different interpretations of performances in clinical practice: a student-athlete performing at the 31st percentile (ie, average) and another performing at the 16th percentile (ie, low average) are separated by 0.5 SD. When interpreting visual motor speed and reaction time performances, clinicians should recognize that performances may be, on average, lower than expected among student-athletes tested in Spanish. Consistent with previous research comparing White and Latinx athletes,³  White and African American athletes,⁴  and athletes tested in English versus Mandarin,²¹  the largest group difference between athletes tested in English versus Spanish was for visual motor speed, indicating that this composite score may be the most sensitive to racial or ethnic and linguistic differences among athletes.

These data were gathered as part of the Maine Concussion Management Initiative (MCMI) under the direction of the principal investigator Paul D. Berkner, DO. We thank the Maine Athletic Trainers' Association for their collaboration with the MCMI. Grant L. Iverson, PhD, has been reimbursed by the government, professional scientific bodies, and commercial organizations for discussing or presenting research relating to mild traumatic brain injury (mTBI) and sport-related concussion at meetings, scientific conferences, and symposiums. He has a clinical practice in forensic neuropsychology, including expert testimony, involving individuals who have sustained mild TBIs (including athletes). He has received honoraria for serving on research panels that provided scientific peer review of programs. He is a coinvestigator, collaborator, or consultant on grants relating to mild TBI funded by the federal government and other organizations. He has received research support from test publishing companies in the past, including ImPACT Applications Systems, Psychological Assessment Resources, and CNS Vital Signs. He has received research support from the Harvard Integrated Program to Protect and Improve the Health of National Football League Players Association and a grant from the National Football League. He serves as a scientific advisor for BioDirection, Inc; Sway Medical, LLC; and Highmark Inc. He acknowledges unrestricted philanthropic support from the National Rugby League, Mooney-Reed Charitable Foundation, and ImPACT Applications, Inc.