Annals of Biomedical Engineering

, Volume 45, Issue 8, pp 1974–1984 | Cite as

Age has a Minimal Effect on the Impact Performance of Field-Used Bicycle Helmets

  • Alyssa L. DeMarco
  • Craig A. Good
  • Dennis D. Chimich
  • Jeff A. Bakal
  • Gunter P. Siegmund
Article

Abstract

Helmet manufacturers recommend replacing a bicycle helmet after an impact or after anywhere from 2 to 10 years of use. The goal of this study was to quantify the effect of helmet age on peak headform acceleration during impact attenuation testing of field-used bicycle helmets. Helmets were acquired by donation from consumers and retail stores, and were included in the study if they were free of impact-related damage, had a legible manufacture date label, and were certified to at least one helmet standard. Helmets (n = 770) spanning 0–26 years old were drop tested to measure peak linear headform acceleration during impacts to the right and left front regions of the helmets at two impact speeds (3.0 and 6.2 m/s). General linear mixed models were used to assess the effect of age and three covariates (helmet style, size and certification impact speed) on peak acceleration. Overall, age was related to either no difference or a statistically significant but small increase (≤0.76 g/year of helmet age) in peak headform acceleration. Extrapolated across 20 years, age-related differences were less than both style- (traditional vs. BMX) and size-related differences. The age-related differences were also less than the variability observed between different helmets after accounting for style, size and certification effects. These findings mean that bicycle helmets (up to 26-year-old traditional helmets and 13-year-old BMX helmets) do not lose their ability to attenuate impacts with age; however, other helmet features that may change with age were not evaluated in this study.

Keywords

Helmet Head acceleration Impact attenuation Foam degradation Bicycle 

Introduction

Most bicycle helmets have foam liners that crush and crack during an impact and therefore need to be replaced after a significant impact. For helmets that do not experience an impact, there are few and varying guidelines on when they should be replaced and there is no standard governing their impact performance over time. Some manufacturers provide specific recommendations in their product manuals to replace the helmet every 2–10 years (e.g., Bern—2 years; Giro, CCM, Lazer, Supercycle, Schwinn—3 years; Specialized, Alpina, Trek—5 years; Nutcase—10 years) whereas other manufacturers provide no recommendations (e.g., some Bell, RED, Louis Garneau, and Supercycle models), or vague recommendations like replacing the helmet when it shows obvious signs of wear (e.g., some Bell, Giro and CCM models), or after a few years of careful use (e.g., Protec) (Table A1 in Appendix A). The Snell Memorial Foundation recommends that helmets be replaced after 5 years of use, or less if recommended by the manufacturer.22 The scientific basis for these various replacement recommendations is unclear.

The impact attenuation performance of old, used bicycle helmets is not well studied. McIntosh and Dowdell18 evaluated helmets subjected to real-world impacts between 2 and 60 months after manufacturing and observed no trend between impact performance and age. Williams26 used new helmets to reconstruct the impact severity of field-impacted helmets up to 6 years old and estimated that age differences introduced a potential error of 15% into his data; however, he attributed this error to variations in the molding consistency of the expanded polystyrene (EPS) foam liner rather than to degradation in helmet performance with age. More recently, we showed that the material properties of the EPS liner in 10 models of field-used helmets between 2 and 20 years old did not vary with age.17 To date, however, there has been no focused study examining how bicycle helmet impact performance varies with helmet age.

The goal of this study is to determine the effect of helmet age on peak headform acceleration during impact attenuation testing of bicycle helmets. Based on manufacturer recommendations that helmets be replaced every 2–10 years, we hypothesized that peak headform acceleration would increase with helmet age. We included helmet style, size and certification impact speed as covariates to better isolate age from other factors that could potentially affect peak headform acceleration. We looked for age-related performance changes at both the standard impact speed of 6.2 m/s (drop height of about 1.96 m) used by the Consumer Product Safety Commission11 and at a lower severity of 3.0 m/s (drop height of about 0.46 m). The lower severity was chosen to capture the lower bound of the perpendicular impact speeds reported for reconstructed bicycle head impacts.16,24

Materials and Methods

Field-used bicycle helmets were solicited from the general public through a collection campaign at local bike shops, cycling clubs, schools, and bicycling events in Vancouver, BC, Toronto, ON and Calgary, AB. All donated helmets (n ~ 1500) were inspected by one author (ALD) and excluded if they did not meet the following criteria: (i) legible labeling specifying date of manufacture (~90% of excluded helmets), (ii) no evidence of impact related damage—defined as visible cracking or residual compression of the energy absorbing liners (~10% of excluded helmets), and (iii) certification to at least one helmet standard requiring a minimum impact attenuation performance (<1% of excluded helmets). Helmets with deteriorated or missing comfort foam pads were not excluded and were tested as received. Of the 770 helmets included in the study, most (88%) were used and the others were unused. Manufacturing dates ranged from 1987 to 2013 and thus the helmets were 0–26 years old at the time of testing in 2013. Helmets were classified as traditional, BMX, kids bucket, kids with integrated EPS or PU visor, or full-face style (Fig. 1). BMX helmets consisted of a hard shell with limited venting and were differentiated here from full-face helmets that are commonly used for BMX competition or downhill mountain biking. Kids helmets that were of similar construction to adult traditional helmets were classified as traditional helmets. Helmet size was determined by finding the best fitting headform from the following: ISO A (extra-small, XS), ISO E (small, S), ISO J (medium, M), ISO M (large, L), and ISO O (extra-large, XL) (Half Magnesium K1A, Cadex Inc., Quebec, Canada). Most helmets (99%; n = 763) had EPS liners, with the remaining helmets having polyurethane (PU) liners (1%; n = 7). From a certification perspective, helmets were categorized into three groups based on the highest impact speed prescribed for a flat anvil test within the standards listed on their labels17,10,11,14,15,2123 (Table 1). Finally, helmets were grouped into matched sets (MS), which were defined as essentially identical helmets (determined by a thorough visual inspection and comparison of their design, geometry and construction), though of potentially different makes, models, colors and sizes.
Figure 1

Examples of the five different classifications of bicycle helmet styles. From left to right: traditional, BMX, kids bucket, kids visor, and full-face.

Table 1

Certification categories, the related certification standards, and the approximate impact speeds for the flat anvil tests.

Category

Certification

Flat anvil speed (m/s)

Cert 1

CPSC,11 Snell 198421/B9022/B90A,23 ASTM 14474/20405/19526/17512

~6.2

Cert 2

AS/NZS 2063,7 CE/EN 107715/1078,14 CSA D113.210

~5.5

Cert 3

ANSI Z90.4,1 ASTM 14923

~4.5

Test Procedures

Our impact attenuation testing protocol was adapted from the Safety Standard for Bicycle Helmets.11 A 3.8 m tall monorail and trolley assembly (Biokinetics and Associates, Ottawa, ON) was used for all drop tests, but was supplemented by a piston to prevent helmets from sustaining more than one impact per drop test (Fig. 2). Impact speed was measured by a speed trap located within 50 mm of impact with a resolution of ±0.12% at 3.0 m/s and ±0.25% at 6.2 m/s. A uni-axial linear accelerometer (Model 7264, ±2000 g, Endevco, San Juan Capistrano, CA) was mounted at the headform center of gravity. A CFC1000 hardware filter was present in the amplifier of the accelerometer. The total mass of the moving assembly for each headform was 5.0 kg without a helmet. All impacts were onto a flat steel anvil, as data from real-world bicycle helmet impacts indicate a flat impact surface is common.8,20,26
Figure 2

Test setup showing the headform, surrogate chin and the piston that extended upward after impact to prevent a second impact.

Speed trap and accelerometer signals were digitally sampled at 100 kHz and the accelerometer signals were low-pass filtered using a fourth-order, zero-lag Butterworth filter with a cutoff of 1650 Hz before extracting the peak headform acceleration, i.e., the maximum acceleration at the headform center of gravity during the impact (ASTM 14474) using custom code in Matlab (2013b, MathWorks, Natick, MA). Video of each test was acquired at either 500 Hz (PRI-F1, AOS Technologies, Baden, Switzerland) or 600 Hz (EX-F1, Casio, Tokyo, Japan). Instrument system checks, which consisted of dropping a spherical impactor onto a Modular Elastomer Programmer (MEP), were done at the start and end of each testing day according to the CPSC protocol.11 All testing was conducted under ambient temperature and humidity conditions.

Each helmet was tested twice: once on the left front and once on the right front, i.e., mirrored impact locations. These impact locations were determined for each headform size by finding the midpoint of the surface region bounded by the coronal, median (midsagittal) and reference planes (Fig. 3). The centers of the two impact locations on the surface of the helmets were more than 120 mm apart as required by the CPSC protocol.11 This impact location was chosen because the front/side of the helmet is a common impact site in real-world helmet impacts9,18, 19, 20,26 and this location could be held constant across the five headform sizes. Custom molds for both sides of each headform were used to repeatably orient the headforms relative to the ball arm.
Figure 3

The “X” depicts the location of the left side impacts. The right side impacts were a mirror image of this impact location (Figure adapted from CPSC11).

Each helmet experienced the low severity impact (3.01 ± 0.02 m/s) before the high severity impact (6.21 ± 0.02 m/s) and these impacts were separated by a minimum of 24 h. Low- and high-speed impacts were split nominally equally between the left and right sides. Helmets were tested in batches of about 20 helmets of similar size to minimize setup time, and the test order of the batches was randomized for size and left/right side. All helmets were tested without their removable visors. Helmets missing their fitting foam, comfort liner or thin exterior plastic shells were tested to capture the range of conditions present in field-used helmets. With the exception of full-face helmets, all helmets were secured to the headform using a 20 cm Velcro strap looped through the helmet straps and braced against a surrogate chin (Fig. 2). The surrogate chin allowed us to equalize the tension in both pairs of front and back straps using the under-ear sliders (when present), and thus to ensure proper and consistent orientation of the helmets on the headform. Helmets were positioned on the headform by centering the front and rear of the helmet with the median plane and by aligning the front bottom edge of the helmet with the reference plane. When present, ring fit systems were tightened to the headform. For the full-face helmets, the chin guards were removed to eliminate interference with the trolley and these helmets were secured to the headform/surrogate chin with duct tape.

Statistics and Data Analysis

The continuous variables in each group were summarized using means and standard deviations (SDs), and the categorical variables were summarized as frequencies and percentages. The characteristics were compared between groups using the t tests and χ2 tests, as appropriate. All analyses were conducted using SAS 9.2 (SAS Institute, Cary, NC). Statistical significance for all tests was set to p < 0.05.

To examine the relationship between age and peak headform acceleration, a series of general linear mixed models (GLMM) was developed that included covariates for helmet style (Traditional and BMX), headform size (ISO A, E, J, M, O), and certification group (Cert 1, Cert 2 and Cert 3). Both styles of the kids helmets and the full-face helmets were excluded from the statistical models due to their limited numbers, limited size ranges (Appendix B), and different construction topology. Data from each impact speed were fit using three models: the first model included all traditional and BMX helmets in the sample (All T + BMX model; n = 675 at 3.0 m/s, n = 674 at 6.2 m/s), the second model focused on helmets with ≥4 samples per matched set (MS ≥ 4 model, n = 369 at both speeds), and the third model focused on helmets with ≥8 samples per matched set (MS ≥ 8 model, n = 192 at both speeds). Of the seven PU helmets, three were included in the T + BMX model and the other four were excluded because they were a kid’s visor style. We excluded one helmet from the 6.2 m/s data set because the piston failed to catch it following its 3.0 m/s impact and thus it sustained multiple impacts as it bounced to rest. The statistical models containing only matched sets allowed for age-related changes within specific helmet designs to be better discerned and more likely to be identified. Both matched-set models also included only traditional and BMX helmets. The MIXED procedure was used for the models with the restricted maximum likelihood (REML) estimation and Satterthwaite method for estimating the degrees of freedom. Residuals from all models were analyzed and plotted to confirm the quality of model fit.

The form of the statistical models is shown in Eq. 1, where yij is the peak headform acceleration measured in g for the ith helmet within the jth matched set (MS), A is the overall intercept, uj0 is the MS-specific intercept for the jth matched set, B is the overall age-related slope, uj1 is the MS-specific age-related slope for the jth matched set, and ageij is the age of a helmet in years at the time of testing. C, D and E are the coefficients for the style (styleij), size (sizeij) and certification group (certij) respectively of the helmet, and εij is the residual for each helmet. The intercepts and age-related slopes were treated as random effects and the covariates were treated as fixed effects. To assess if adding a MS-specific intercept (uj0) and slope (uj1) improved a model’s fit, a χ2 test was first used to compare the −2*residual log likelihood between the reference model (no uj0 or uj1) and a model including just the MS-specific intercept (uj0), and then a second χ2 test was used to compare between the model including just the MS-specific intercept (uj0) and a model including both the MS-specific intercept and age slope (both uj0 and uj1).25
$$y_{ij} = \left( {A + u_{j}^{0} } \right) + \left( {B + u_{j}^{1} } \right) {\text{age}}_{ij} + C {\text{style}}_{ij} + D {\text{size}}_{ij} + E {\text{cert}}_{ij} + \varepsilon_{ij}$$
(1)

Results

Most of the 770 bicycle helmets were traditional (78.1%), followed by BMX (9.6%), kids bucket (1.2%), kids visor (9.0%) and full-face styles (2.1%) (Appendix B—Table B1, Fig. 4a). There were no BMX helmets older than 13 years and most BMX helmets (92%; n = 68) were ≤9 years old. Most kids helmets (78%; n = 61) were also ≤9 years old, and all of the full-face helmets were ≤11 years old. The most common size in the sample was ISO J (medium, 45.3%), followed by ISO E (small, 29.5%), ISO M (large, 14.8%), ISO A (extra-small, 9.1%) and ISO O (extra-large, 1.3%) (Appendix B—Table B1, Fig. 4b). Helmets certified to our Cert 1 category dominated the sample (87.5%), with Cert 2 (10.7%) and Cert 3 (1.8%) both occurring relatively infrequently (Appendix B—Table B1; Fig. 4c). All of the Cert 3 helmets were ≥15 years old.
Figure 4

Distribution of helmets by (a) style, (b) size, and (c) certification group versus helmet age at the time of testing.

Peak headform accelerations for all styles of helmets combined were higher at the 6.2 m/s impact speed and showed considerable variability at both impact speeds, even within helmets of the same age (Fig. 5; Appendix B—Tables B2, B3). Of the 769 helmets tested at 6.2 m/s, four helmets exceeded the 300 g limit currently mandated by CPSC at this impact speed. One of these four helmets was 4 years old and certified to a standard requiring a 6.2 m/s impact speed (Cert 1), whereas the other three were over 20 years old and certified to a standard requiring a 4.5 m/s impact speed (Cert 3).
Figure 5

Peak headform acceleration versus helmet age at 3.0 m/s (n = 768) and 6.2 m/s (n = 769). Note that four helmets exceeded a peak acceleration of 300 g. This figure shows data from all traditional, BMX, kids bucket, kids visor and full-face helmets.

Peak headform acceleration varied with age, style, size and certification, although not all of these effects were statistically significant in each of the six models (Tables 2, 3). Including the MS-specific intercept (uj0) improved all of the models (p < 0.0001), whereas including the MS-specific age slope (uj1) improved only the two All T + BMX models (p ≤ 0.048). Thus the coefficients reported for the MS ≥ 4 and MS ≥ 8 models in Tables 2 and 3 were for statistical models that did not include the MS-specific slope term. For each of the statistical models, the conditional residuals were well behaved.
Table 2

Statistical results for the 3.0 m/s impact speed showing the population statistics and fixed model coefficients for all traditional and BMX helmets (All T + BMX), and matched sets with 4 or more helmets (MS ≥ 4) and 8 or more helmets (MS ≥ 8).

Statistical model

All T + BMX

MS ≥ 4

MS ≥ 8

Helmets (n)

675

369

192

Groups of unique helmet designs

260

49

15

Mean peak acceleration [g]

92.4

91.7

91.1

Standard deviation [g]

12.4

11.9

11.2

p-value for adding MS-specific intercept (uj0)

<0.0001

<0.0001

<0.0001

 Range of uj0 (g)

−26.0 to 20.9

−22.5 to 19.6

−22.3 to 15.6

 SD of uj0 (g)

7.5

9.4

9.1

p-value for then adding MS-specific slope (uj1)

0.048

0.086

0.106

 Range of uj1 (g/year)

−0.514 to 0.454

††

††

 SD of uj1 (g/year)

0.133

††

††

Coefficients

 A: Intercept [g]

86.8*

89.9*

90.5*

 B: Age Slope [g/year]

0.353*

0.0489

−0.182

 C: BMX v Trad [g]

22.54*

15.82*

14.70

 D: ISO A v ISO J (XS v M) [g]

−7.14*

−9.04*

−4.00

 D: ISO E v ISO J (S v M) [g]

−1.85*

−2.32*

−0.50

 D: ISO M v ISO J (L v M) [g]

2.84*

2.83*

3.38*

 D: ISO O v ISO J (XL v M) [g]

2.11

2.71

−0.92

 E: Cert 2 v Cert 1 [g]

1.58

4.85*

5.28*

 E: Cert 3 v Cert 1 [g]

−3.01

−6.01

−5.78

*Significantly different from zero, p < 0.05

0.05 < p < 0.10

††Model does not contain the MS-specific age-related slope term (uj1)

Table 3

Statistical results for the 6.2 m/s impact speed showing the population statistics and fixed model coefficients for all traditional and BMX helmets (All T + BMX), and matched sets with 4 or more helmets (MS ≥ 4) and 8 or more helmets (MS ≥ 8).

Statistical model

All T + BMX

MS ≥ 4

MS ≥ 8

Helmets (n)

674

369

192

Groups of unique helmet designs

260

49

15

Mean peak acceleration [g]

215.5

214.6

216.6

Standard deviation [g]

23.1

19.1

16.6

p-value for adding MS-specific intercept (uj0)

<0.0001

<0.0001

<0.0001

 Range of uj0 (g)

−42.8 to 46.8

−25.2 to 22.0

−13.9 to 13.3

 SD of uj0 (g)

14.6

12.5

9.1

p-value for then adding MS-specific slope (uj1)

0.0212

0.411

0.328

 Range of uj1 (g/year)

−0.866 to 1.640

††

††

 SD of uj1 (g/year)

0.389

††

††

Coefficients

 A: Intercept [g]

204.2*

204.4*

207.1*

 B: Age Slope [g/year]

0.763*

0.598*

0.658*

 C: BMX v Trad [g]

39.96*

34.82*

24.62*

 D: ISO A v ISO J (XS v M) [g]

−13.68*

−15.99*

−11.98*

 D: ISO E v ISO J (S v M) [g]

−4.76*

−5.19*

−4.49*

 D: ISO M v ISO J (L v M) [g]

6.42*

6.25*

7.00*

 D: ISO O v ISO J (XL v M) [g]

7.90*

6.63

3.50

 E: Cert 2 v Cert 1 [g]

1.09

3.08

1.46

 E: Cert 3 v Cert 1 [g]

11.56*

0.09

−1.50

* significantly different from zero, p < 0.05

0.05 < p < 0.10

†† Model does not contain the MS-specific age-related slope term (uj1)

Overall, the effect of helmet age on peak headform acceleration was small (Tables 2, 3). At 3.0 m/s, peak headform acceleration increased 0.353 g/year of helmet age (95th percentile confidence interval (CI): 0.166–0.540 g/year; p = 0.0002) in the All T + BMX model, but did not vary significantly with age in either the MS ≥ 4 model (p = 0.69) or the MS ≥ 8 model (p = 0.29). At 6.2 m/s, peak headform acceleration increased with helmet age for all three models (All T + BMX: 0.763 g/year, CI 0.380–1.146 g/year, p = 0.0001; MS ≥ 4: 0.598 g/year, CI 0.163–1.032 g/year, p = 0.0072; and MS ≥ 8: 0.658 g/year, CI 0.0474–1.268 g/year, p = 0.0349). Within the MS-specific age slopes of both T + BMX models, only one matched set at 3.0 m/s had a MS-specific age slope that was significantly different from the overall slope (B), and its net slope (B + uj1) showed no age effect. At 6.2 m/s, four MS-specific age slopes were significantly different from the overall slope; however, all four were in the Cert 3 group and three of the four helmets exceeded 300 g.

Helmet style and size had a larger effect on peak headform acceleration than helmet age. BMX helmets generated higher peak headform accelerations than traditional helmets (14.70–22.54 g at 3.0 m/s, and 24.62–39.96 g at 6.2 m/s; Tables 2, 3). Peak headform accelerations also varied with helmet size (7.4–11.9 g range at 3 m/s, and 19.0–22.6 g range at 6.2 m/s) and generally increased with increasing helmet size, although not all helmet sizes were significantly different from one another (Tables 2, 3). Certification-related differences were less consistent than either style- or size-related differences, and varied between no effect and increasing peak headform acceleration by 11.56 g at 6.2 m/s in the All T + BMX model.

Discussion

We evaluated the effect of age on the impact attenuation performance of field-used bicycle helmets up to 26 years old and found that peak headform acceleration increased by less than 1 g per year of helmet age. Although this age effect was statistically significant, it was small compared to the effects we observed with the covariates for helmet style, size and certification. Even for 20 years of aging, the increase in peak headform acceleration within each statistical model was less than both the difference between traditional and BMX helmet styles and the maximum difference between helmet sizes. Thus from an impact performance perspective, we found little or no support for replacing a helmet every 2–10 years as recommended by some bicycle helmet manufacturers or every 5 years as recommended by the Snell Memorial Foundation,21 and currently cannot recommend an optimum replacement period for either traditional or BMX bicycle helmets. These findings support our prior work17 showing that the dynamic mechanical properties of bicycle helmet foam do not deteriorate with age and extends this prior work by showing that the impact attenuation performance of the whole helmet system does not seriously deteriorate with age. It is important to highlight that prior to testing we discarded any helmet that showed evidence of prior impact related damage to the foam or shell, and therefore our results do not justify wearing an old helmet with pre-existing damage. We also did not evaluate how age-related deterioration of the comfort foam or retention system affected the helmet’s fit, stability or roll-off potential, but instead focused solely on the impact attenuation performance of a proper-fitting helmet that was in position at the time of impact. Within this context, our findings do not support discarding or avoiding the use of an old, undamaged and properly fitting helmet simply because it is old.

Of the three covariates we examined, helmet style had the largest effect on peak headform acceleration. Compared to traditional helmets, BMX-style helmets had peak accelerations that were 15–23 g higher at 3.0 m/s and 25–40 g higher at 6.2 m/s. A prior study of new bicycle helmets also found that BMX helmets do not attenuate peak headform acceleration as well as traditional helmets, and attributed this finding to a thinner foam liner typically installed in these helmets.13 We have observed increased popularity of BMX-style helmets over the past decade, suggesting that some consumers prefer these helmets, perhaps for their aesthetics, hard shell, and greater coverage; however, they may not be aware that these helmets generally perform worse than traditional helmets, at least at the impact location used for these tests.

Helmet size had the next largest effect on peak headform acceleration, varying by 7–12 g at 3 m/s and by 19–23 g at 6.2 m/s. Helmet size is presumably dictated by a consumer’s head size, and based on our findings, consumers with small heads are better protected by well-fitting bicycle helmets than consumers with large heads. It remains unclear whether the helmet-size effect we observed is an artifact of using a 5 kg drop mass for all helmets sizes. Smaller heads typically weigh less, and therefore may experience higher accelerations than we observed. While some standards include different drop masses for different size headforms, the most common standard in North America (CPSC11) does not. While not addressed in the testing conducted for this study, we believe that a helmet should be of the correct size and of proper fit for the user with a functional retention system adjusted in accordance with the manufacturer’s recommendations in order for it to provide optimum protection.

Helmet certification had the smallest and least consistent effect on peak headform acceleration of the three covariates studied here. Compared to the helmets certified at impact speeds of 6.2 m/s (Cert 1), the helmets certified to 5.5 m/s (Cert 2) generated up to 5 g of additional head acceleration during the 3.0 m/s tests, but no difference during the 6.2 m/s tests. In contrast, the helmets certified to 4.5 m/s (Cert 3) were no different from Cert 1 helmets during the 3.0 m/s tests, but the acceleration was up to 12 g higher during the 6.2 m/s tests. Given that all Cert 3 helmets in our sample were 15 years or older and that three of the four helmets that exceeded 300 g were Cert 3 helmets, some confounding of age and certification may be present in the statistical models examined here. While consumers still wearing Cert 3 helmets should consider upgrading to newer Cert 1 helmets, we were encouraged to find that 11 of the 14 Cert 3 helmets tested here met the current CPSC standard for impact attenuation.

The statistical models included MS-specific intercepts and slopes to capture the potentially different age-related changes between different helmet designs and constructions. None of the statistical models containing 4 or more helmets within each matched set yielded significant MS-specific age slopes, although helmets within these matched sets only spanned an average of 3.9–5.1 years and a maximum 11 years rather than the full 26 years of the helmets in the full models. In contrast, the MS-specific intercepts were significant in all six models we evaluated. Moreover, in the full model, the MS-specific intercepts spanned 47 g at 3.0 m/s and 90 g at 6.2 m/s. These large ranges mean that helmet design variations other than style, size and certification accounted for most of the variability present in our test data (Fig. 5). Helmet foam density and liner thickness are two possible explanations for this variability and further work is needed to correlate these variables directly to the impact performance of field-used helmets. Although impact location was kept constant here, differences between helmets in the placement of ribs and/or vents could also explain some of the variability between helmet designs.

Four helmets in our sample exceeded the 300 g ceiling prescribed by CPSC for peak headform acceleration at a 6.2 m/s impact speed (Fig. 6). Three of these helmets were certified to a standard with a lower 4.5 m/s impact speed (Cert 3, Figs. 6a–6c) and the fourth helmet was previously reported to exceed the CPSC requirement when it was new.12
Figure 6

Four helmets that exceeded 300 g at 6.2 m/s. (a) 1989 Vetta, (b) 1987 Vetta Supercorsa, (c) 1991 Troxel ProAction, (d) 2009 Nutcase URS-011S.

Because we relied on donated helmets, the specific exposure history of the helmets was unknown. Environmental and use factors may be expected to influence the material properties of the constituent helmet materials. While we cannot quantify these exposures, the helmet population we tested may have been in poorer condition than the average helmet in active public use since the people who donated them believed they were not worth keeping, or alternatively in better condition because the people who donated them did not use them. Also, about half of the donated helmets were excluded from the analysis, mostly because we could not identify a date of manufacture. While knowing this date was an obvious requirement of the study, the exclusion of these helmets may have biased our sample. We also do not know what design or manufacturing changes occurred over the range of manufacturing years we tested or across different helmet sizes we tested. And despite a relatively large sample, there were combinations of helmet style and size where no helmets were tested. Our tests also focused on a single impact location on a flat anvil and all tests were conducted at room temperature and humidity. Further work is needed to examine the effect of these variables on age-related changes to bicycle helmet impact performance.

In summary, the effect of age on the peak headform acceleration during drop tests of field-used bicycle helmets up to 26 years old was less than 1 g per year of helmet age. Extrapolated over 20 years, this age-related difference was less than the differences we observed between traditional and BMX-style helmets, between different sizes of helmets, and between different helmet designs. Thus within the limitations of our sample and test conditions, we observed little or no support for age-related deterioration of bicycle helmet impact performance that would justify replacing a helmet in good condition every 2–10 years as recommended by some bicycle helmet manufacturers.

Notes

Acknowledgments

The authors thank Justin Lam, Jeff Nickel and Mircea Oala-Florescu for their help in conducting the tests. We also thank the organizations, companies, and individuals who donated their helmets.

Conflict of interest

Authors ALD, CAG, DDC and GPS are forensic consultants who occasionally work on cases related to bicycle helmet effectiveness. No external funding was received for this study: all funding and support was provided by MEA Forensic Engineers & Scientists (employer of ALD, DDC, GPS) and Collision Analysis (employer of CAG). DDC and GPS are shareholders of MEA Forensic and CAG owns Collision Analysis.

References

  1. 1.
    American National Standards Institute. American national standard for protective headgear for bicyclists. ANSI Z90.4. Washington, DC, 1984.Google Scholar
  2. 2.
    American Society for Testing and Materials International. Standard Specification for Helmets Used in Recreational Roller Skating, F 1751-00. West Conshohocken, PA, 2000.Google Scholar
  3. 3.
    American Society for Testing and Materials International. Standard Specifications for Helmets Used in Skateboarding and Trick Roller Skating, F 1492-00. West Conshohocken, PA, 2000.Google Scholar
  4. 4.
    American Society for Testing and Materials International. Standard Specification for Helmets Used in Recreational Bicycling or Roller Skating, F 1447-06. West Conshohocken, PA, 2006.Google Scholar
  5. 5.
    American Society for Testing and Materials International. Standard Specification for Helmets Used for Recreational Snow Sports, F 2040-06. West Conshohocken, PA, 2006.Google Scholar
  6. 6.
    American Society for Testing and Materials International. Standard Specification for Helmets Used for Downhill Mountain Bicycle Racing, F 1952-09. West Conshohocken, PA, 2009.Google Scholar
  7. 7.
    Australian/New Zealand Standard®. Pedal cycle helmets. AS/NZS 2063:1996. Homebush NSW Australia, Wellington New Zealand, 1996.Google Scholar
  8. 8.
    Cameron, M. H., C. F. Finch, and A. P. Vulcan. The protective performance of bicycle helmets introduced at the same time as the bicycle helmet wearing law in Victoria. Monash University Accident Research Centre Report No. 59, 1994.Google Scholar
  9. 9.
    Ching, R. P., D. C. Thompson, R. S. Thompson, D. J. Thomas, W. C. Chilcott, and F. P. Rivara. Damage to bicycle helmets involved with crashes. Accid. Anal. Prev. 29(5):555–562, 1997.CrossRefPubMedGoogle Scholar
  10. 10.
    Canadian Standards Association. National standard of Canada—cycling helmets, CAN/CSA-D113.2-M89. Toronto, ON, 1989.Google Scholar
  11. 11.
    Consumer Product Safety Commission. Safety Standard for Bicycle Helmets; Final Rule. 16 CFR Part 1203. Consumer Product Safety Commission, Bethesda, Maryland, 1998.Google Scholar
  12. 12.
    Consumer Reports, 2012. Best bike helmets www.consumerreports.org. Yonkers, N.Y.
  13. 13.
    DeMarco, A. L., D. D. Chimich, J. C. Gardiner, and G. P. Siegmund. The impact response of traditional and BMX-style bicycle helmets at different impact severities. Accid. Anal. Prev. 92:175–183, 2016.CrossRefPubMedGoogle Scholar
  14. 14.
    European Committee for Standardization, 1997. Helmets for pedal cyclists and for users of skateboards and roller skates, EN 1078. Brussels.Google Scholar
  15. 15.
    European Committee for Standardization, 2007. Helmets for alpine skiers and snowboarders, EN 1077. Brussels.Google Scholar
  16. 16.
    Fahlstedt M., K. Baeck, P. Halldin, J. Vander Sloten, J. Goffin, B. Depreitere, and S. Kleiven. Influence of impact velocity and angle in a detailed reconstruction of a bicycle accident. In: Proceedings of the International IRCOBI Conference on the Biomechanics of Impact, IRCOBI Secretariat, Dublin, Ireland, 2012, pp. 787–799.Google Scholar
  17. 17.
    Kroeker, S. G., S. J. Bonin, A. L. DeMarco, C. A. Good, and G. P. Siegmund. Age does not affect the material properties of expanded polystyrene liners in field-used bicycle helmets. J. Biomech. Eng. 138:041005, 2016.CrossRefPubMedGoogle Scholar
  18. 18.
    McIntosh, A. S., and B. Dowdell. A field and laboratory study of the performance of pedal cycle helmets in real accidents. In: Proceedings of the International IRCOBI Conference on the Biomechanics of Impact, IRCOBI Secretariat, Zurich, Switzerland, 1992, pp. 51–72.Google Scholar
  19. 19.
    McIntosh, A. S., D. Kallieris, R. Mattern, N. L. Svensson, and B. Dowdell. An evaluation of pedal cycle helmet performance requirements (952713). In: Proceedings of 39th Stapp Car Crash Conference, Society of Automotive Engineers, Warrendale, PA, 1995, pp. 111–119.Google Scholar
  20. 20.
    Smith, T. A., D. Tees, D. R. Thom, and H. H. Hurt Jr. Evaluation and replication of impact damage to bicycle helmets. In: Proceedings of 37th Annual AAAM Conference, San Antonio, TX, 1993, pp. 143–155.Google Scholar
  21. 21.
    Snell Memorial Foundation. Standard for Protective Headgear for Use in Bicycling. North Highlands, CA: Snell Memorial Foundation, 1984.Google Scholar
  22. 22.
    Snell Memorial Foundation. Standard for Protective Headgear for Use in Bicycling, B90. North Highlands, CA: Snell Memorial Foundation, 1990.Google Scholar
  23. 23.
    Snell Memorial Foundation. 1998 Augmentation to the 1990 Standard for Protective Headgear for Use in Bicycling, B90A. North Highlands, CA: Snell Memorial Foundation, 1998.Google Scholar
  24. 24.
    Verschueren, P. Biomechanical analysis of head injuries related to bicycle accidents and a new bicycle helmet concept. PhD Thesis, Katholieke Universiteit Leuven, Belgium, 2009.Google Scholar
  25. 25.
    West, B. T., K. B. Welsh, and A. T. Gałecki. Linear Mixed Models—A Practical Guide Using Statistical Software. Chapman & Hall/CRC, Taylor & Francis Group, Boca Raton, FL, 2007.Google Scholar
  26. 26.
    Williams, M. The protective performance of bicyclists’ helmets in accidents. Accid. Anal. Prev. 23:119–131, 1991.CrossRefPubMedGoogle Scholar

Copyright information

© Biomedical Engineering Society 2017

Authors and Affiliations

  • Alyssa L. DeMarco
    • 1
  • Craig A. Good
    • 2
    • 3
  • Dennis D. Chimich
    • 1
  • Jeff A. Bakal
    • 4
  • Gunter P. Siegmund
    • 1
    • 5
  1. 1.MEA Forensic Engineers & ScientistsRichmondCanada
  2. 2.Collision AnalysisCalgaryCanada
  3. 3.Schulich School of EngineeringUniversity of CalgaryCalgaryCanada
  4. 4.Alberta Health ServicesUniversity of AlbertaEdmontonCanada
  5. 5.School of KinesiologyUniversity of British ColumbiaVancouverCanada

Personalised recommendations