Development of the STAR Evaluation System for Assessing Bicycle Helmet Protective Performance

Abstract

Cycling is a leading cause of mild traumatic brain injury in the US. While bicycle helmets help protect cyclists who crash, limited biomechanical data exist differentiating helmet protective capabilities. This paper describes the development of a bicycle helmet evaluation scheme based in real-world cyclist accidents and brain injury mechanisms. Thirty helmet models were subjected to oblique impacts at six helmet locations and two impact velocities. The summation of tests for the analysis of risk (STAR) equation, which condenses helmet performance from a range of tests into a single value, was used to summarize measured linear and rotational head kinematics in the context of concussion risk. STAR values varied between helmets (10.9-25.3), with lower values representing superior protection. Road helmets produced lower STAR values than urban helmets. Helmets with slip planes produced lower STAR values than helmets without. This bicycle helmet evaluation protocol can educate consumers on the relative impact performance of various helmets and stimulate safer helmet design.

Keywords: Biomechanics; Concussion; Cycling; Impact; Injury risk.

Figure 1

Custom impact rig used for STAR testing. A helmeted headform is dropped onto an angled anvil to generate an oblique impact.

Figure 2

The six impact locations selected for testing (top), along with required orientations of the helmeted headform in the support ring to impact each location (bottom). Locations 1, 2, and 6 represent body-driven impacts, in which the head leads the body, locations 3 and 5 represent skidding-type impacts, and location 4 represents an impact from flipping over the handlebars.

Figure 3

CDF of digitized data from helmet damage replication studies. STAR test velocities were chosen based on the 50th and 90th percentiles (3.4 and 5.2 m/s normal velocities, respectively). A 0.5 m/s range from either velocity was then mapped to the CDF (shaded region), and the number of impacts occurring within that range was divided evenly by the six impact locations to determine weighting factors for the STAR equation. The 3.4 ± 0.5 m/s range encompassed 38.0 impacts, so all low speed impacts were weighted by 6.33. The 5.2 ± 0.5 m/s range encompassed 9.4 impacts, so all high speed impacts were weighted by 1.57.

Figure 4

PLA, PRV, and concussion risk distributions for each impact configuration. Concussion risk was computed using a bivariate function including both PLA and PRV.

Figure 5

Effect of kinematic results on STAR values. Shown are PLA and PRV results averaged across impact locations for every helmet. Each helmet is represented by one circle per velocity, with its shade determined by its overall STAR value (e.g., the Ballista MIPS produced the lowest STAR value of 10.9 and is represented by a white circle near the lower left corner of each plot). STAR values increase with increasing kinematics.

Figure 6

Range in STAR values across all helmet models. Lower STAR values indicate reduced incidence of concussion, and thereby enhanced protection. Error bars represent 95% CI ranges.

Figure 7

STAR value distributions based on style and MIPS. Road helmets with MIPS produced the lowest STAR values and thereby offer the greatest protection. MIPS helmets generated lower STAR values than non-MIPS helmets for both styles.

Figure 8

Average PLA (left) and PRV (right) values per helmet at 7.3 m/s vs. 4.8 m/s. Each point represents one helmet’s average values. Shaded regions represent the 95% prediction interval. Strong correlations were observed, suggesting that a helmet’s ability to reduce impact kinematics relative to other helmets is similar at both velocities.