Home > Attention > Can Variability In Concussion Testing Really Tell Us Something Important?

Can Variability In Concussion Testing Really Tell Us Something Important?


In the past decade, the negative consequences of traumatic brain injuries, more commonly referred to as concussions, have become highly publicized. Once brushed off as an innocent hit to the head, concussions are now taken much more seriously. Although concussions can occur for many reasons, due to their frequency, sports related concussions have become the target of concern. It is estimated that among the 38 million children and adolescents and 170 million adults participate in athletic activities in the US, there are as many as 3.8 million mild traumatic brain injuries that occur each year. Many of these go untreated (Giza et al., 2013).

Most of us have been hit on the head at some point in our lives, but for some, sports related concussions can have important consequences that affect a person’s everyday life.  Research looking at recovery from concussions in young athletes has shed light on short and long-term complications. Most symptoms are typically experienced briefly, including headaches, dizziness, fatigue, irritability, reduced concentration, sleep disturbance, loss of memory, anxiety, sensitivity to noise, blurred vision, light sensitivity, and depression (Master, Balcer, & Collins, 2014). Although most symptoms resolve quickly, a small percent of people have symptoms that last for 12 or more months. If you’ve ever had a concussion, you’ve probably had “good days and bad days” – some where you feel you can focus and others not as well. There can be a lot of variability in how you feel.

We’ve all see athletes who have hit their heads during a game being swarmed with trainers, asking them if they know their name and how many fingers they are holding up. Post concussion evaluation is a very important for predicting the course of recovery and the amount of time before athletes can return to play (Rosenthal, Foraker, Collins, & Comstock, 2014). Click here for more information on the diagnosis and management of sports related concussion. Identifying factors that predict post concussion outcomes is important since athletes who return to play too early can exacerbate their concussioncognitive symptoms, make them more susceptible to a subsequent concussions and increase the risk of long-term complications (Rosenthal et al., 2014).

You are probably asking yourself what is being done to minimize the impact of sports related concussions. Most academic institutions and professional athletic associations have implemented programs that assess concussions and monitor recovery by comparing preseason baseline and post concussion cognitive tests (Rosenthal et al., 2014). While this sounds like a great approach, one major challenge is separating the normal variability due to test reliability from differences in test scores related to the injury.  What exactly do I mean by test variability? If I were to test your IQ three days in a row, you might score a little higher on the first day than on the last day. It wouldn’t mean that your IQ was declining. There would just be some variability in how accurately the test measures your IQ. For an athlete with a head injury, we would not want to attribute this decline to a head injury if we knew that the test had test retest variability.

Why is this concept of test variability important to think about? In studies on aging, depression, anxiety, and schizophrenia, research has suggested that test variability in an individual may not be due solely to test variability but may actually represent a pathologic process (MacDonald, Li, & Bäckman, 2009). For example, studies have shown that if you administer cognitive testing to people as they age, more variability in the test results may be a better predictor of cognitive decline than their average level of performance (Lövdén, Li, Shing, & Lindenberger, 2007). Why is this such a cool concept? In the past, variability was considered noise in the data–it didn’t provide additional information. In this case, the variability is actually the information that provides clues to pathology (cause and effect of a disease/injury) and prognosis (likely outcome of the situation). Variability has suddenly become a useful tool.

Rabinowitz and Arnett decided to expand this variability concept to sports related concussion. They examined how much variability there was in cognitive testing before and after a sports related head injury, and if people with more variability were more likely to have serious problems (Rabinowitz & Arnett, 2013). They measured variability by calculating a standard deviation, which is kind of confusing so let me try to provide a simple explanation. Standard deviation measures the spread around the average of an individual’s test results. In other words, how tight your scores are to the average. The larger the standard deviation, the more spread there is from the mean. For example, if you measured how fast you ran a mile each day for 7 days and your time varies a lot, then your standard deviation would be large. If you want to learn more about the math behind it, click here.

The researchers decided to try to determine if injured athletes have more variability between test results than uninjured people. They reasoned that normal variability could be quantified by giving an uninjured person a battery or set of tests that each measure various aspects of cognitive function and then calculating the standard deviation between test results. You could do this same thing post injury. The difference in variability between normal people and injured athletes should reflect variability due to a pathologic process (aka something not so good). If this extra variability predicts worse long-term outcomes, then it would be a very important discovery.

Rabinowitz and Arnett studied 71 college varsity athletes and 42 controls who participated in intermural sports. The varsity athletes had testing done before injury and then again after the head injury. Controls were measured at baseline then one month later. Control data was used to determine what normal performance and performance variability would be expected in the battery of tests used. The test battery included 10 tests, which assessed cognitive (like memory and attention), physiological (like balance) and emotional functioning (like depression). The tests asked participants to do things like connect numbered dots in the correct order to test for visual attention or substituted a number for a geometric figure to test for brain dysfunction. They chose these tests because previous research had shown they were impacted by traumatic brain injury.

The researchers measured both cognitive performance and variability in cognitive performance. To do this, cognitive performance was measured by taking an average of all of the individual test scores for each time point, and variability was measured by calculating how much each of the individual’s test score differed from their average score for each time point. They hypothesized that the injured athletes would have lower performance scores and more variability from their average score. They also used a fancy approach that applies statistics to group people into high or a low variability groups based on their baseline testing data. If an individual’s test scores differed a lot from their average test score, they would be grouped into a high variability group and if it didn’t differ much, they would be grouped into a low variability group. In order to determine if an athlete had declined in performance, the researchers used data from the control group. If the athlete’s change in cognitive function after injury was outside what was expected, it was deemed an important decline.

If you are math oriented, curious and want to know more about the ingenious approach the researchers took, read on; otherwise, skip to the next paragraph. You might be wondering how the researchers combined tests that were all measured on a different scale, making it impossible to combine or compare tests. The researchers standardized each test thus, creating a standardized score. This isn’t really fancy; it is done for IQ tests. The important point is that it allowed them to combine all of the test scores to create an overall performance score by averaging the various test scores at baseline and again post concussion. This overall performance score is a bit like measuring how agile an athlete is by running a set of agility tests that measure various aspects of agility and then coming up with an average score that reflects their overall agility. They also wanted to know how much a participant’s individual test scores differed from their average score. To do this, they calculated the standard deviation of the tests scores before and again after the concussion. Using the agility example, they were asking how much each of the agility tests scores differed from the individual’s average score. It’s a bit of a brain teaser but ingenious.

There was a lot of individual variability in testing in everyone, although they didn’t say how the groups differed. This is no great surprise since we know that we rarely get the same score on tests and that this is a common problem with tests. In controls, the performance increased and variability decreased between baseline and month 1. This is also not surprising since practice probably helped improve their scores and make them more consistent.

In athletes, they found that at baseline and post concussion, the higher the overall performance the less performance variability. This means that if perform better on a test, you are more likely to perform consistently. They also found that there was no difference in performance or performance variability between the baseline and post injury assessment. In other words, concussion had no impact.

The baseline low variability athlete group looked pretty similar to the low variability control group. They had a decrease in performance variability and an increase in performance. The baseline high variability athlete group looked different than controls though. Their performance decreased with a concussion, and their performance variability increased. Importantly, individuals in the high variability group were more likely than those in the low variability group to have reliable decline on one or more cognitive test.

How the athlete feels... And maybe you too

How the athlete feels… And maybe you too

So what does this all mean? There is significant variability in test performance in healthy people, and it did not appear to be related to someone’s intellectual capacity. In other words, when measuring cognitive performance in different ways, it is normal to get a pretty wide range of scores. However, a lower overall performance and more variability in performance went hand in hand for both healthy and injured athletes. Interestingly, test performance was similar before and after testing as was the amount of variability.  In other words, head injuries didn’t change performance or result in cognitive tests with more varied results. I know it might not sound like it, but this was quite surprising. It is possible that that only some of the tests were affected by concussion and that by combining all of the cognitive tests together it diluted the effects and prevented them from seeing important differences that might have been seen if only a subset of the tests were used. Interestingly, when the researchers grouped people into high and low variability groups, they did find that those in the high variability group were more likely to have poorer post concussion performance, greater performance variability, and post concussion declines in individual tests. This data suggest that athletes with more variability at baseline may be more likely to suffer consequences from a concussion. Why might this be? Athletes who have had prior concussions may have more variability at baseline from prior injury and may also be more susceptible to future concussions. The researchers didn’t tell us if this was the case. I don’t think this was due to differences in how seriously the athletes took the baseline test (which might have resulted in more or less variability). Athletes had no way of knowing that they were going to later have a concussion, and so you would not predict that the concussion group would have more people that don’t take the test seriously at baseline. There are some comparability issues with the control and athlete groups. The control group had repeat testing one month after baseline. Although it is possible that some of the athletes had their injury and thus second test within one month of their preseason baseline testing, many likely had their injury more than a month later. This was not clarified by the researchers. It is possible longer testing intervals would not result in improved performance and decreased variability observed with the controls.

Frustrating as it may be, the results did not show that performance variability was different after head injury as hoped. However, the research did show that if an athlete scores very differently on each test in a battery of cognitive tests, this might be a signal for coaches to keep a special eye on these athletes if they end up suffering a concussion. This suggests that intraindividual variability is important and may reflect something different about an athlete’s brain, but we still have a lot to learn to really understand what it means for these concussed athletes. Because sustaining a concussion can lead to some seriously adverse outcomes, it is very important to take them seriously, continue to implement baseline and post concussion test, and keep exploring the many unanswered questions around how to identify and monitor concussions.

Full article can be read by clicking here!

Here is another very interesting concussion articles check it out!


Giza, C. C., Kutcher, J. S., Ashwal, S., Barth, J., Getchius, T. S., Gioia, G. A., . . . Zafonte, R. (2013). Summary of evidence-based guideline update: evaluation and management of concussion in sports: report of the Guideline Development Subcommittee of the American Academy of Neurology. Neurology, 80(24), 2250-2257. doi: 10.1212/WNL.0b013e31828d57dd

Lövdén, M., Li, S.-C., Shing, Y. L., & Lindenberger, U. (2007). Within-person trial-to-trial variability precedes and predicts cognitive decline in old and very old age: Longitudinal data from the Berlin Aging Study. Neuropsychologia, 45(12), 2827-2838. doi: 10.1016/j.neuropsychologia.2007.05.005

MacDonald, S. W. S., Li, S.-C., & Bäckman, L. (2009). Neural underpinnings of within-person variability in cognitive functioning. Psychology and Aging, 24(4), 792-808. doi: 10.1037/a0017798

Master, C. L., Balcer, L., & Collins, M. (2014). Concussion. Ann Intern Med, 160(3), Itc2-1. doi: 10.7326/0003-4819-160-3-201402040-01002

Rabinowitz, A. R., & Arnett, P. A. (2013). Intraindividual cognitive variability before and after sports-related concussion. Neuropsychology, 27(4), 481-490. doi: 10.1037/a0033023

Rosenthal, J. A., Foraker, R. E., Collins, C. L., & Comstock, R. D. (2014). National High School Athlete Concussion Rates From 2005-2006 to 2011-2012. Am J Sports Med. doi: 10.1177/0363546514530091

  1. No comments yet.
You must be logged in to post a comment.