Evaluating the “Threshold theory”: Can head impact indicators help?

Research Team Members: Jason Mihalik, PhD, ATC, Robert Lynall, PhD, ATC (HMSC PhD Student now at University of Georgia), Erin Wasserman, PhD (EXSS Gfeller Center Postdoc now at Datalys Center), Kevin Guskiewicz, PhD, ATC, Steve Marshall, PhD

Why did you do this study?

concussion-blog-featured-imageAs many as 1.6 – 3.8 million sport and recreation traumatic brain injuries (TBI) occur in the US on an annual basis. The direct and indirect costs for managing all forms of TBI exceed %56B annually. Proper detection and management of sport related concussion continues to challenge clinicians working with athletes. A number of options are available to clinicians, but mostly rely on subjective and clinical expertise. One example is the Sport Concussion Assessment Tool Version 3 (includes symptom inventories, mental status tests, and balance assessments). These acute injury screening tools are typically administered only after the clinician has sufficient evidence to suspect a concussion diagnosis. In the absence of obvious concussion signs (e.g., loss of consciousness, staggered gait, etc.), clinicians must rely solely on subjective symptoms reported by athletes. Research has documented a large portion of athletes either underreport concussion symptoms or fail to report them entirely. Thus, the medical field has looked to emerging technologies to fill this shortfall and provide heightened objectivity to the dilemma.

Technological advances have resulted in the emergence of commercially available head impact measurement devices. These devices typically serve two broad functions: 1) collect data for research-based inquiry, and 2) signal to clinical staff the occurrence of high-level impacts in near real-time during sports participation. Head impact indicators—the latter function—seek to identify athletes who have sustained pronounced head impacts so that they can be evaluated for symptomology. These products are usually worn directly on the head or affixed to a helmet, and are designed to indicate to medical personnel, players, coaches, and parents when a head impact magnitude has exceeded a pre-programmed threshold. There are no fewer than 20 different products that have permeated the marketplace in the last decade, and many use differing thresholds (some unknown to the user!).


Head impact indicators are believed to identify athletes who otherwise would elect not to report symptoms to the clinical staff. If an ‘alert’ is triggered, some of the manufacturers recommend the athlete be removed from activity and evaluated for a head injury, regardless of whether or not the athlete is exhibiting signs or reporting symptoms consistent with concussion. Our own work here at UNC suggests that a single impact injury threshold is not obvious, which question the clinical utility of these head impact indicators.

What did you do and what did you find in this study?

The purpose of this study was to investigate the clinical utility of head impact magnitude thresholds employed by various commercially available head impact indicators to positively predict concussion among American football players. We hypothesized these tools, by themselves, would be limited in helping clinicians make informed decisions regarding head injury during athletics due to the inherent variability of biomechanical values observed in concussed individual and the low incidence of concussion even at very high measured impact levels.

Over the last 10 years, we have collected hundreds of thousands of head impact biomechanics from hundreds of football players. A multidisciplinary clinical team independently made concussion diagnoses during this same time period (n=24). We dichotomized each impact using diagnosis (‘yes’ they were injured, ‘no’ they were not), and across a range of plausible impact indicator thresholds (10g increments beginning with a resultant linear head acceleration of 50g and ending with 120g). We then performed computations to determine the sensitivity, specificity, negative predictive value and positive predictive value, which are common measures used to assess the clinical utility of any diagnostic or screening assessment.


How do these findings impact the public?

In particular, any head impact indicator must demonstrate that it has predictive value; that is, it is an efficient use of time and resources and that it yields a practical frequency of identified concussions to be clinically useful. All thresholds we studied had low positive predictive value (<0.4%). Even when conservatively adjusting the frequency of diagnosed concussions by a factor of 5 to account for unreported/undiagnosed injuries, the positive predictive value of head impact indicators at any threshold was no greater than 1.94%. Simply put, fewer than 4 out of 1000 trigger alerts would result in a diagnosed concussion. Or, looking at it from the vantage of clinician time resources, 996 sideline evaluations would be done in vain and at the possible detriment of distracting the sideline medical personnel from observing and intervening in other emergencies and injuries during that time.