Effects of Concussion on Attention and Executive Function in Adolescents

David Howell; Louis Osternig; Paul Van Donkelaar; Ulrich Mayr; Li-Shan Chou


Med Sci Sports Exerc. 2013;45(6):1030-1037. 

In This Article


Forty high school students participating in school sports at three local high schools (36 men/4 women) were identified and recruited for testing. Twenty of the participants were identified by specialized health professionals (certified athletic trainer/physician) as suffering a concussion consequent to sport participation. Concussion was defined according to the 3rd International Statement on Concussion in Sport as an injury caused by a direct blow to the head, face, neck, or elsewhere in the body with an impulsive force transmitted to the head resulting in impaired neurologic function and acute clinical symptoms.[28] Each concussed subject in the study was matched with a healthy control subject (n = 20) by sex, height, mass, age, and sport ( Table 1 ). Prospective control subjects were identified by certified athletic trainers at the high school from which the matched concussed subject was a student. Matching criteria were confirmed at the laboratory test site.

Each concussed subject was removed from the injury site on the day of injury and did not return to preinjury levels of physical activity until cleared by a physician in accordance with state law. Exclusion criteria for concussed and control subjects included the following: 1) lower extremity deficiency or injury, which may affect normal gait patterns; 2) history of cognitive deficiencies, such as permanent memory loss or concentration abnormalities; 3) history of three or more previous concussions; 4) loss of consciousness from the concussion lasting more than 1 min; 5) history of attention deficit hyperactivity disorder; or 6) a previously documented concussion within the past year. Individuals with a history of three or more previous concussions were ineligible to participate to ensure, to the extent possible, the exclusion of subjects experiencing chronic mTBI. Subjects who suffered loss of consciousness for more than 1 min were excluded from participation in the study because that sign is believed to play a role in concussion management modification.[28]

Before data collection, the institutional review board reviewed and approved the protocol of the current study. All subjects and parent/guardian (if younger than 18 yr) provided informed consent. Permission was also granted by the respective school districts to conduct testing with student participants.

A prospective, repeated-measures design was used in which each subject reported to the laboratory and was tested within 72 h of sustaining a concussion as well as on four subsequent testing days at the following time increments: 1 wk, 2 wk, 1 month, and 2 months postinjury. Control subjects were similarly tested according to the same time schedule. The Attentional Network Test (ANT[12] and the Task-Switching Test (TST), adapted from Mayr,[25] were administered separately and individually to each study participant in a visually enclosed space free from distracting noise and other people. The components of cognitive function each test measures have been shown to be sensitive to the effects of concussion.[4,15] In addition, to better understand the clinical presentation of each subject, a concussion symptom checklist was administered, which assessed 22 symptoms on a six-point Likert scale adapted from McCrory et al.[28]

The ANT was originally designed to evaluate abnormalities arising in cases of brain injury, stroke, schizophrenia, and attention deficit disorder.[12] It probes the efficiency of three distinct attentional components: alerting (alerting effect), spatial orientation (orienting effect), and executive function (conflict effect) by assessing the relative change in reaction time (RT) to differing precue and stimulus configurations.[12] Event-related functional magnetic resonance imaging has been used to explore the brain areas involved in the three attention systems targeted by the ANT.[11] The results suggested that the functional contrasts within the ANT tended to differentially activate three separable anatomical networks related to the components of attention. The conflict effect has been shown to be significantly affected by mTBI up to 1 month postinjury, and the orienting effect has been shown to be affected within the first 2 d postinjury in young adults,[15] but these effects have yet to be systematically studied in concussed adolescents.

For the past 10 yr, versions of the ANT have been used in more than 60 publications dealing with a wide range of topics and methods, including development, neuroimaging, pharmacology, genetics, psychiatric disorders, brain damage, and individual differences.[17] Recently, Ishigami and Klein[17] tested the reliability of the ANT on healthy young adults for the 10 testing sessions. They observed learning effects during the first few sessions for the executive component (conflict effect) and reported that reliability improved as more sessions were included in the analysis. Between sessions 1 and 2, they found correlations of -0.02, 0.57, and 0.86 and when combining sessions 1–10 (using a modified split-half correlation analysis) of 0.80, 0.65, and 0.93 for the alerting, orienting, and executive components, respectively. It was concluded that the ANT was robust after multiple sessions and suitable for applications requiring repeated testing. Because of the noted practice effects, it was suggested that controls are warranted in some designs.

In the ANT, the subject fixates on a cross in the center of a computer screen and responds as quickly as possible by pressing one of two arrow keys, indicating the direction (left or right) of a central arrow presented either directly above or below the cross (Figs. 1A and 1B). The alerting effect is examined by determining the RT difference between trials in which a warning cue (asterisk) precedes the arrow stimulus versus trials in which the warning cue does not precede the arrow stimulus (Fig. 1C). The orienting effect is examined by the RT difference between trials, in which the warning cue indicates the location of the arrow stimulus (above or below the fixation cross) versus trials in which the warning cue does not provide such spatially relevant information (Fig. 1D). Finally, the conflict effect is assessed by the RT difference between trials in which the arrow stimulus is accompanied on either side by two congruent flanker arrows (i.e., arrows pointing in the same direction) versus trials in which the arrow stimulus is accompanied on either side by two incongruent flanker arrows (i.e., arrows pointing in the opposite direction; Fig. 1E). Thus, the effect is measured by the RT difference between the two conditions presented for each of the three networks (alerting, orienting, and conflict). A greater RT difference score between groups/testing days indicates poorer performance on that attentional network. Participants first completed a series of 24 practice trials with visual accuracy feedback; they then completed two blocks of experimental trials made up of 96 trials (4 precue conditions × 2 target locations × 2 target directions × 3 flanker conditions × 2 repetitions) for a total of 192 experimental trials.

Figure 1.

ANT: the sequence of events during trials. Subjects focus on the fixation cross and are instructed to respond by pressing the keyboard arrow (left or right) corresponding to the direction of a target arrow when presented directly above or below the cross (A and B). The asterisk is a precue (C and D), which gives information about when (alerting) or where (orienting) the target will appear. Flanker arrows (E) may be presented in configurations which test conflict and nonconflict effects.

The TST uses a paradigm that specifically tests the ability to flexibly switch between competing task or stimulus–response rules.[33] The primary dependent variable is the switch cost, which is the difference score between the response time from trials on which the task changes (i.e., switch trials) and trials on which the task stays the same (i.e., no-switch trials). Switch costs across different task pairs correlate highly with each other,[25,30] suggesting that these costs (a) can be assessed reliably and (b) are independent of the primary task. Further, numerous studies have found that switch costs reflect an executive function that is largely independent of other executive abilities, such as the resolution of conflict or working memory.[30] Previous work has found that task-switching ability is highly sensitive to mTBI in adults[4] but has not been studied in the concussed adolescent population. The task-switching paradigm has been used to study executive functions in the context of cognitive development, cognitive aging, and brain imaging.[29] In addition, task switching has been used in studies of a wide array of clinical disorders, including attention deficit hyperactivity disorder, Parkinson's disease, and frontal lobe injury.[29] The specific paradigm used within the current study design has been documented as a valid way to examine executive functioning and is reliable across testing sessions.[33] Rogers and Monsell[33] demonstrated that extra practice did little to increase subject performance when repeat testing occurred, indicating test stability between testing days.

For the current TST, subjects were required to switch between responding congruently and incongruently to the position of a visual stimulus on every second trial in a sequence.[25] The subjects responded to the position of a circle presented in a horizontally configured rectangular box (Fig. 2) by pressing the right or left arrow key on a standard computer keyboard as quickly and accurately as possible. In the congruent case, the subject indicated the left or right position of the circle by pressing the corresponding arrow key. In the incongruent case, the subject pressed the opposite key (i.e., left arrow key for right target and vice versa). The subject alternated between congruent and incongruent responses on every second trial throughout the sequence of four blocks of 52 trials (208 total trials). If an incorrect response was generated, a visual display was presented to remind the subject which rule (congruent or incongruent) was required to respond correctly. The dependent variable of interest was the "switch cost." This is defined as the difference in RT between "stay" trials, in which the subject did not switch from responding congruent or incongruently (or vice versa), and "switch" trials, in which the subject did switch from responding congruently to incongruently (or vice versa). Therefore, the switch cost is a measurement of the difference in response time during trials where the rule switched compared with trials where the rule stayed the same. A higher switch cost indicates greater difficulty in adhering to the congruent/incongruent rules. Only trials in which an accurate response was generated were included in the calculation of the switch cost.

Figure 2.

TST: The subject responds to the position of the circle within the rectangle by pressing the arrow key in either a congruent or incongruent manner, alternating every two trials. The switch cost is calculated as the mean accurate response time difference of trials, which repeat congruent/incongruent responses (no switch) and those which switch between congruent/incongruent responses (switch). The performance differences between trial types are commonly used as measures of executive function ability.

Data were analyzed by two-way, mixed-effects ANOVA to determine the effect of group (concussed vs controls), time (72 h, 1 wk, 2 wk, 1 month, and 2 months), and the interaction effects on each of the dependent variables (alerting, orienting, and conflict effects and switch cost). For all omnibus tests, significance was set at P < 0.05. Follow-up pairwise comparisons were then examined using the Bonferroni procedure to control family-wise type I error. All statistical analyses were performed using the Statistical Package for the Social Sciences (version 20; SPSS Inc., Chicago, IL).