Head impact severity measures for evaluating mild traumatic brain injury risk exposure

Abstract

Objective: The aims of this study were to quantify the sensitivity of various biomechanical measures (linear acceleration, rotational acceleration, impact duration, and impact location) of head impact to the clinical diagnosis of concussion in United States football players and to develop a novel measure of head impact severity combining these measures into a single score that better predicts the incidence of concussion.

Methods: On-field head impact data were collected from 449 football players at 13 organizations (n = 289,916) using in-helmet systems of six single-axis accelerometers. Concussions were diagnosed by medical staff and later associated with impact data. Principal component analysis and a weighting coefficient based on impact location were used to transform correlated head impact measures into a new composite variable, weighted principal component score (wPCS). The predictive power of linear acceleration, rotational acceleration, head injury criterion, and wPCS was quantified using receiver operating characteristic curves. The null hypothesis, that a measure was no more predictive than guessing, was tested (alpha = 0.05). In addition, receiver operating characteristic curves for wPCS and classical measures were directly compared to test the hypothesis that wPCS was more predictive of concussion than were classic measures (alpha = 0.05).

Results: When all of the impacts were considered, every biomechanical measure evaluated was statistically more predictive of concussion than guessing (P < 0.005). However, for the top 1 and 2% of impacts based on linear acceleration, a subset that consisted of 82% of all diagnosed concussions, only wPCS was significantly more predictive of concussion than guessing (P < 0.03); when compared with each other, wPCS was more predictive of concussion than were classical measures for the top 1 and 2% of all of the data (P < 0.04).

Conclusion: A weighted combination of several biomechanical inputs, including impact location, is more predictive of concussion than a single biomechanical measure. This study is the first to the authors' knowledge to quantify improvements in the sensitivity of a biomechanical measure to incidence of concussion when impact location is considered.

Figure 1

The Wayne State Tolerance Curve defines an injury threshold based on the onset of skull fracture which was thought to correlate with moderate to severe concussion.

Figure 2

Impacts were grouped into 5 bins based on impact location which is defined by azimuth (θ) and elevation (α). Azimuth (θ) is defined from −180° to 180° with 0° at the X axis and positive (θ) to the right side of the head. Elevation (α) is defined from 0° (horizontal plane passing through the head center of gravity, CG) to 90° (crown of the head at the Z axis). The XZ plane represents the midsagittal plane with the positive X corresponding to the caudal direction. The XY plane represents the coronal plane with positive Y to the right side of the head. Impacts with elevation > 65 were defined as top. The remaining impacts were grouped based on azimuth; −45° to 45° were back, ±45° to ±135° were side, and −135 ° to 135° were front.

Figure 3

Receiver Operator Characteristic (ROC) curves demonstrate the sensitivity of the independent measures (e.g. linear acceleration, rotational acceleration, HIC, weighted PCS) to the incidence of concussion. Each point on a curve represents a possible concussion threshold (for a given head impact severity measure) for which the correct prediction level and false response rate are calculated. The x-axis in the figure shown is zoomed in to the clinically relevant portion of the ROC curves. Each curve begins at (0,0) and ends at (1,1).

Figure 4

Values of the biomechanical measures that are associated with each correct prediction level (i.e. percentage of concussions correctly predicted).

Figure 5

Linear acceleration concussion probability function generated from NFL impacts reconstructed in the laboratory, as in Pellman, shown with 17 concussive impacts recorded in-vivo and a random sample of 100 controls (non-injurious impacts). In total there were 289,899 controls of which 3,476 were > 98.9g (the 75% concussion probability level based on NFL data).