Clinical Reaction-Time Performance Factors in Healthy Collegiate Athletes

Between 1.6 million and 3.8 million sport-related concussions (SRCs) occur annually in the United States.¹ The symptom presentation of sport-related concussion symptoms is highly variable, so a multifaceted and multimodel assessment that supports the clinical examination is the recommended approach for diagnosing SRC and tracking recovery.^2,3 This assessment may include tests of mental status and cognition, oculomotor function, gross sensorimotor function, coordination, gait, vestibular function, balance, and reaction time (RT).^2–4 According to the 5th “Consensus Statement on Concussion in Sport,”² baseline testing may be useful but is not necessary for interpreting postinjury scores. In the absence of baseline testing, normative data are used to interpret postinjury scores.² For neurocognitive assessments (eg, Immediate Post-Concussion Assessment and Cognitive Testing [ImPACT]), clinical neuropsychologists use demographically adjusted normative data to interpret postinjury scores.^5,6 For other assessments (eg, RT), performance factors, such as demographics, medical and concussion history, self-reported symptoms, sleep, and sport-related features, have not been assessed. Without a comprehensive assessment of these performance factors, it is difficult to provide normative data that will aid in interpreting postinjury scores in the absence of baseline testing.

The measurement of RT typically relies on specialized computer programs as part of the neurocognitive evaluation, which limits its accessibility and translatability in many athletic settings. To enable all health care practitioners to assess simple RT, Eckner et al^7,8 developed a clinically feasible test of simple RT (RT_clin), which involved timing how long it took participants to catch a suspended vertical dowel by pinch grip. Clinical RT scores were moderately reliable (intraclass correlation coefficient = 0.645 over a 1-year test-retest interval),^9–11 valid (R = 0.445 compared with computerized RT testing, R = 0.725 compared with a functional head-protective task),^9,10 and sensitive (75%) and specific (68%) in SRC recognition.^12,13 However, performance factors associated with RT_clin have received limited attention in the literature. Understanding the performance factors associated with RT_clin will allow us to provide normative data that can be used in the absence of baseline testing.

Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT Applications, Inc, Coralville, IA) is the most widely used computerized neurocognitive assessment in SRC management and includes RT tests resulting in an RT composite score.^14,15 Performance factors influencing ImPACT RT composite scores in National Collegiate Athletic Association (NCAA) student-athletes include sex, sport type (ie, contact, limited contact, noncontact), and neurodevelopmental and concussion history, although the literature is mixed.^4,16–18 Composite ImPACT scores may differ from RT_clin performance as a result of different experimental setups and stimuli. Therefore, the purpose of our study was to explore the effects of performance factors, including demographics (eg, sex), medical (eg, neurodevelopmental) and concussion history, self-reported symptoms, sleep, and sport-related features (eg, level of contact), on baseline RT_clin performance. Based on computerized RT performance factors in NCAA student-athletes, we hypothesized that contact-sport athletes and those with neurodevelopmental disorders would have slower RT_clin^4,18 but that no differences based on demographic variables¹⁷ or concussion history would be present.¹⁶

METHODS

Clinical RT Testing

The RT_clin was performed as previously described (Figure 1).^{7–10,12,13,19,20} Briefly, participants were instructed to catch an 80-cm wooden dowel coated in high-friction tape and marked in 0.5-cm increments as quickly as possible. The dowel was embedded in a weighted rubber disk of diameter = 7.5 cm, height = 2.5 cm, and mass = 256 g. The participant sat with the dominant hand positioned at the edge of the table in an open C-shape position. The examiner initially held the dowel so that the weighted rubber disk was in line with the participant's first and second digits and then released the dowel at predetermined randomly assigned time intervals ranging from 2 to 5 seconds.^{7–10,12,13,19,20} The participant caught the dowel as quickly as possible. The distance the apparatus fell was determined from the marked increments on the dowel. The RT_clin values were calculated by converting distance to time, in milliseconds, using the formula for a body falling under the influence of gravity (D = 1/2 gt²). Two practice trials were followed by 8 data-collection trials. If a participant did not catch the dowel, then the examiner recorded the “dropped” trial and the participant continued with the next trial; dropped trials were not replaced. Mean RT_clin was then calculated for participants from all successfully completed trials.^{7–10,12,13,19,20}

View Figure

• Download as Powerpoint
• Download Figure

Statistical Analyses

Generalized linear modeling was used to establish performance factors that might be associated with RT_clin. Potential performance factors (self-reported) were entered in the following order: sex (male, female)¹⁷; race (American Indian or Alaska Native, Asian, black or African American, Native Hawaiian or other Pacific Islander, white, multiple races, unknown or not reported)²¹; ethnicity (Hispanic or Latino, not Hispanic or Latino, unknown or not reported)²¹; dominant hand (right, left, ambidextrous)²²; sport type (contact, limited contact, noncontact)⁴; number of prior concussions, including both diagnosed and undiagnosed (0, 1–2, >3)¹⁶; presence of anxiety (yes, no)²³; learning disability (yes, no)^18,23; attention-deficit disorder or attention-deficit/ hyperactivity disorder (yes, no)^18,23; depression (yes, no)²³; migraine headache (yes, no)²³; self-reported sleep the night before the test (<7, 7–9, >9 hours, not reported) as self-reported on the ImPACT²⁴; mass (continuous)²⁵; height (continuous)²⁵; age (continuous)²²; total number of SCAT-3 symptoms (continuous)²⁶; total SCAT-3 symptom burden (weighted score on the graded symptom checklist, continuous).²⁶ We fit the generalized linear model for RT_clin based on a normal (Gaussian) distribution (Figure 2). Significance was defined a priori as P < .05. This analysis was conducted using SPSS (version 24; IBM Corp, Armonk, NY). Effect sizes are included to illustrate the clinical meaningfulness of the findings.

View Figure

• Download as Powerpoint
• Download Figure

RESULTS

The overall RT_clin was 202.0 ± 25.0 milliseconds, with a breakdown by subgroup provided in Table 1. The generalized linear model was a suitable fit to the data (omnibus test: likelihood ratio χ² = 256.496, P < .001). In this model, sex, race, sport type, and height were the only significant predictors of RT_clin (Table 2). Female sex (B = 8.6 milliseconds, P < .001, Cohen d = 0.54 relative to male sex), black or African American race (B = 5.3 milliseconds, P = .001, Cohen d = 0.08 relative to white race), and limited-contact (B = 4.2 milliseconds, P < .001, Cohen d = 0.30 relative to contact) and noncontact (B = 5.9 milliseconds, P < .001, Cohen d = 0.38 relative to contact) sport participation were associated with slower RT_clin when all other factors were controlled (Table 1). Being taller was associated with a faster RT_clin, although this association was weak (B = −0.7 milliseconds, P < .001; Table 2). No other predictors were significant (Table 2). When adjustments are made for sex and sport type, the following normative data may be considered (mean ± standard deviation): female, noncontact (211.5 ± 25.8 milliseconds), limited contact (212.1 ± 24.3 milliseconds), contact (203.7 ± 21.5 milliseconds); male, noncontact (199.4 ± 26.7 milliseconds), limited contact (196.3 ± 23.9 milliseconds), contact (195.0 ± 23.8 milliseconds; Table 3).

Table 1 Frequency Statistics for Nominal- and Ordinal-Level Predictors

View Fullscreen

Table 2 Test of Model Effects for Generalized Linear Model

View Fullscreen

Table 3 Clinical Reaction-Time Test: Normative Data by Sex and Sport Type

View Fullscreen

DISCUSSION

Several performance factors may be associated with RT_clin in NCAA student-athletes, and adjustments should be made for these performance factors when providing normative baseline data. The purpose of our study was to explore the effects of performance factors, including demographics (eg, sex), medical (eg, neurodevelopmental) and concussion history, self-reported symptoms, sleep, and sport-related features (eg, level of contact), on baseline RT_clin performance. Female sex, black or African American race, limited-contact or noncontact-sport participation, and being shorter were associated with slower RT_clin. After evaluating effect sizes, we suggest that female sex (Cohen d = 0.54 relative to male sex) and sport type (limited contact: Cohen d = 0.30 relative to contact; noncontact: Cohen d = 0.38 relative to contact) result in potentially clinically relevant differences. We therefore have provided normative data based on these 2 performance factors, which can be considered in the absence of baseline testing.

Sex and sport-contact type were associated with RT_clin, whereby males (196 ± 24 milliseconds) had faster RT_clin than females (209 ± 24 milliseconds, B = 8.6 milliseconds, Cohen d = 0.54) and contact-sport participants (197 ± 23 milliseconds) had faster RT_clin than both limited-contact (205 ± 25 milliseconds, B = 4.2 milliseconds, Cohen d = 0.30) and noncontact (207 ± 27 milliseconds, B = 5.9 milliseconds, Cohen d = 0.38) participants. Although these findings differ from those for computerized RT testing,^4,17 RT_clin is strongly correlated with a task designed to simulate a natural head-protective response in a sport-related environment and may be a better indicator of functional RT.¹⁹ Males and contact-sport participants may perform better on functional RT testing as a result of faster processing speed and muscle composition, or contact-sport participants may elect to participate in these sports because they have faster RT.

Although statistically, participants of black or African American race (204 ± 27 milliseconds, B = 5.3 milliseconds, Cohen d = 0.08 relative to white race) had slower RT_clin, these differences were small and may be clinically insignificant. We aimed to identify common performance factors in RT_clin that were consistent with previous RT and neurocognitive assessments, but some performance factors (eg, race) had categories with low frequencies (ie, 80% of all participants were white). This is a limitation, though these data likely represent the NCAA athletic population. Considering that only 11.7% of our population reported being African American, these findings may be a result of the small sample size and should be investigated further in the future.

Many factors have been associated with RT in previous work. Age is typically associated with RT. Specifically, RT decreases from infancy into the late 20s and then begins to slow.²⁷ Considering that the age range of our participants was 17 to 28 years, our sample may have lacked enough variation in age to show this relationship. Similarly, RT increases with fatigue; however, 1 group²⁸ induced fatigue with 24+ hours of sleep deprivation, and our participants averaged 6.9 ± 1.5 hours of sleep the night before the test. Moreover, fatigue is multifaceted, and as such, we cannot say that the number of hours of sleep reported the night before the test was a measure of fatigue. Although we were able to evaluate a number of potential performance factors, additional factors such as arousal, fasting, alcohol consumption, effort and motivation, personality type, exercise, stimulant medications, and intelligence should be considered in future research.

According to the 5th “Consensus Statement on Concussion in Sport,”² baseline testing may be useful but is not necessary to interpret neurocognitive performance after SRC. We have provided normative data adjusting for sex and contact-sport type that can be used in interpreting postinjury scores in the absence of baseline testing. Previous authors^12,13 suggested that a 0-millisecond cutoff score compared with each athlete's own baseline score during the preseason maximized the sensitivity and specificity of SRC diagnosis, so these normative data must be tested with regard to postinjury scores to determine their sensitivity and specificity for SRC management. Therefore, future investigators should determine whether clinicians may consider using these normative data in lieu of individualized baseline measures.

Although we identified several significant performance factors for RT_clin and provided normative data for interpreting postinjury scores in the absence of baseline testing, our study had several limitations. Some participants (n = 53) had a RT_clin <150 milliseconds, which is physiologically unlikely,^29,30 considering that the stimulus-detection time alone is approximately 120 milliseconds, or the latency of early cortical components of the visual-evoked potentials.²² These unusually fast RT_clin times suggest that participants were likely able to anticipate the apparatus drop. However, they were not statistical outliers, and such an RT_clin may occur in the context of typical baseline testing, so these individuals' data were not removed from analyses. To minimize the likelihood of anticipation, examiners were instructed to drop the apparatus at randomly assigned time intervals ranging from 2 to 5 seconds, but some participants may still have been able to predict timing of the drop. Additionally, an athlete's motivation may influence baseline RT_clin, though previous research suggested that athletes appeared to be more motivated during RT_clin testing than during computerized RT testing.²⁰ It is certainly possible that an athlete might purposely perform poorly (ie, sandbagging) at preseason baseline testing in an attempt to mask deficits after a suspected concussion. Finally, we recruited only NCAA student-athletes, and thus, these findings may not be applicable to other populations.

Our aim was to explore factors associated with baseline RT_clin performance and to provide normative data based on these performance factors. We saw potentially clinically relevant differences in RT_clin for sex and contact-sport type, whereby females and noncontact and limited-contact athletes had slower RT_clin performance. When adjustments are made for sex and sport type, the following normative data may be considered (mean ± standard deviation): female, noncontact (211.5 ± 25.8 milliseconds), limited contact (212.1 ± 24.3 milliseconds), contact (203.7 ± 21.5 milliseconds); male, noncontact (199.4 ± 26.7 milliseconds), limited contact (196.3 ± 23.9 milliseconds), contact (195.0 ± 23.8 milliseconds).