Accident Analysis and Prevention 70 (2014) 17

Contents lists available at ScienceDirect

Accident Analysis and Prevention
j our nal homepage: www. el sevi er . com/ l ocat e/ aap
Bicycle helmets are highly effective at preventing head injury during
head impact: Head-form accelerations and injury criteria for
helmeted and unhelmeted impacts
Peter A. Cripton
a,b,c,d,
, Daniel M. Dressler
a,b,d
, Cameron A. Stuart
e
,
Christopher R. Dennison
a,b
, Darrin Richards
e
a
Department of Mechanical Engineering, University of British Columbia, Vancouver, Canada
b
International Collaboration on Repair Discoveries, University of British Columbia, Canada
c
Centre for Hip Health and Mobility, University of British Columbia, Canada
d
Orthopaedic and Injury Biomechanics Group, University of British Columbia, Canada
e
Synaptic Analysis Consulting Group, Vancouver, British Columbia, Canada
a r t i c l e i n f o
Article history:
Received 18 February 2013
Received in revised form8 January 2014
Accepted 19 February 2014
Keywords:
Brain injury
Concussion
Helmet
Bicycle
Injury prevention
a b s t r a c t
Cycling is a popular form of recreation and method of commuting with clear health benets. However,
cycling is not without risk. In Canada, cycling injuries are more common than in any other summer sport;
and according to the US National Highway and Trafc Safety Administration, 52,000 cyclists were injured
in the US in 2010. Head injuries account for approximately two-thirds of hospital admissions and three-
quarters of fatal injuries among injured cyclists. In many jurisdictions and across all age levels, helmets
have been adopted to mitigate risk of serious head injuries among cyclists and the majority of epidemio-
logical literature suggests that helmets effectively reduce risk of injury. Critics have raised questions over
the actual efcacy of helmets by pointing to weaknesses in existing helmet epidemiology including selec-
tion bias and lack of appropriate control for the type of impact sustained by the cyclist and the severity
of the head impact. These criticisms demonstrate the difculty in conducting epidemiology studies that
will be regarded as denitive and the need for complementary biomechanical studies where confounding
factors can be adequately controlled. In the bicycle helmet context, there is a paucity of biomechanical
data comparing helmeted to unhelmeted head impacts and, to our knowledge, there is no data of this
type available with contemporary helmets. In this research, our objective was to perform biomechanical
testing of paired helmeted and unhelmeted head impacts using a validated anthropomorphic test head-
form and a range of drop heights between 0.5 mand 3.0m, while measuring headform acceleration and
Head Injury Criterion (HIC). In the 2 m(6.3 m/s) drops, the middle of our drop height range, the helmet
reduced peak accelerations from 824 g (unhelmeted) to 181 g (helmeted) and HIC was reduced from 9667
(unhelmeted) to 1250 (helmeted). At realistic impact speeds of 5.4 m/s (1.5 m drop) and 6.3 m/s (2.0 m
drop), bicycle helmets changed the probability of severe brain injury from extremely likely (99.9% risk
at both 5.4 and 6.3 m/s) to unlikely (9.3% and 30.6% risk at 1.5 m and 2.0 m drops respectively). These
biomechanical results for acceleration and HIC, and the corresponding results for reduced risk of severe
brain injury show that contemporary bicycle helmets are highly effective at reducing head injury metrics
and the risk for severe brain injury in head impacts characteristic of bicycle crashes.
2014 Elsevier Ltd. All rights reserved.

Corresponding author at: Department of Mechanical Engineering, University of

British Columbia, 6250 Applied Science Lane, Vancouver, BC V6T 1Z4, Canada. Tel.:
+1 604 675 8835.
E-mail address: cripton@mech.ubc.ca (P.A. Cripton).
1. Introduction
Cycling is a popular form of recreation and it is used for com-
muting and other forms of transportation. It is generally safe and
the health benets of it are clear (Hamer and Chida, 2008; Wen and
Rissel, 2008), whichis insharpcontrast tomotorizedtransportation
of any type. However, cycling is also not without risk. In Canada,
cycling injuries are the most common injury occurring fromsum-
mer sports; over 4300 people were hospitalized due to a cycling
http://dx.doi.org/10.1016/j.aap.2014.02.016
0001-4575/ 2014 Elsevier Ltd. All rights reserved.
2 P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17
injury in 20092010 (Canadian Institute for Health Information,
2010). AccordingtotheNational HighwayandTrafc SafetyAdmin-
istration (NHTSA), between 600 and 800 cyclists are fatally injured
each year in the United States and 52,000 cyclists were injured in
the US in2010(NHTSATrafc Safety Facts 2010Data, 2010). Among
cyclists, head injuries account for approximately two-thirds of hos-
pital admissions and three-quarters of fatal injuries (Thompson
et al., 1999).
Epidemiological studies show that helmets are highly effective
at preventing head and brain injury amongst riders who crash. A
casecontrol study conducted by Thompson et al. (1989) in Seat-
tle over a period of 1 year found that bicycle helmets reduced the
risk of head and brain injury by 85% and 88%, respectively. In a
second larger casecontrol study by the same group (Thompson
et al., 1996), helmets decreased the risk of head injury by 69%,
brain injury by 65%, and severe brain injury by 74%. Helmets were
found to be equally effective in accidents involving motor vehi-
cles and those not involving motor vehicles. Furthermore, helmets
were found to provide substantial protection from head injuries
across all age groups. Amoros et al. (2012) recently conducted
a casecontrol study in France and studied helmet effectiveness
over more than 13,500 cyclist injuries. They concluded that hel-
mets were associated with a decreased risk of head injury in cyclist
trauma and this decrease seemed to be more pronounced for
severe headinjuries. Maimaris et al. (1994) studiedover a thousand
patients that sustained cycling-related injuries who were treated
at an emergency department in England. They concluded that hel-
mets reduced the risk of head injury by a factor of more than three.
Heng et al. (2006) found that helmet use signicantly reduced the
risk of head and facial injury in a 2006 study of cycling trauma in
Singapore.
Despite the protection provided by helmets, as demonstrated
by the epidemiological studies above, the safety benets offered by
helmets are not universallyaccepted. Manycities, towns, states and
provinces do not have helmet laws and many cyclists do not wear
helmets (Page et al., 2012). Anti-helmet groups state that helmets
are not effective and that, insome cases, due to the increasedsize of
a helmeted head compared to a bare head or due to the compliance
of the shell or presence of vent holes, helmets cancause rotational
injuries such as diffuse axonal injury (DAI). In the lay press, some
groups claim that helmets cause injuries by obstructing vision or
blocking sound. Researchers have also published articles, critical of
the many epidemiological studies (cited above) that showthat hel-
mets are highly effective at preventing headinjuries, accusing them
of bias andconicts of interest (Curnow, 2006; Elvik, 2011). Curnow
argued that bicycle helmets are not as effective as claimed because
previous epidemiological studies have not considered rotational
injury (Curnow, 2003). There is considerable debate on the merit
and limitations of the epidemiological evidence (Curnow, 2006,
2003; Elvik, 2011; Hagel and Barry Pless, 2006). One limitation of
theepidemiological approaches is that it infers helmet performance
duringtheimpact fromevidencecollectedafter theimpact andthus
the severity of the headimpact under study is never known. It is not
our purpose to debate the merit of the epidemiological literature.
Here we aim to explore the extent to which the epidemiological
evidence of helmet efcacy can be supported or contradicted by
a biomechanical study that allows study of helmet performance
during the impact.
Biomechanical investigations of helmet efcacy, andindeedhel-
met certication standards, simulate helmeted head impact by
dropping helmeted headforms onto prescribed impact surfaces.
In helmet certication standards, the primary metric to assess
impact management efcacy is linear headformacceleration mea-
sured during a drop test; helmets are considered to have met
the certication criteria if the helmeted headform acceleration is
belowaprescribedthreshold. Thethresholdvaries fromstandardto
standard (Table 1), and is not directly correlated to established risk
curves. The standards generally require that helmets be certied
using a magnesium headform. The range of drop heights associ-
ated with these standards is from 1.5m (EN1078) to 2.2m (Snell
B95A) (Table 1).
In biomechanical investigations, linear and rotational head
accelerations are measured during the impact and helmet ef-
cacy is determined by comparing these accelerations, and other
derived metrics such as the Head Injury Criterion (HIC), to injury
risk functions. For example, Mertz et al. have established head
injuryprobabilitycurves, interms of HICandlinear acceleration, for
the Hybrid III headform which was originally developed for auto-
motive crash testing (Mertz et al., 2003). Because injury tolerances
exist for this headform, the Hybrid III is increasingly applied in
biomechanical helmet and head impact studies (Beckwith et al.,
2012; Kendall et al., 2012; Pang et al., 2011; Pellman et al., 2003;
Scher, 2006; Scher et al., 2009; VianoandHalstead, 2012; Vianoand
Pellman, 2005). Overall, the biomechanical studies indicate that
helmets signicantly reduce head accelerations relative to unhel-
meted impacts (Benz et al., 1993; Hodgson, 1990; Mattei et al.,
2012; Scher, 2006) or to impacts with thin uncertied novelty
helmets (DeMarco et al., 2010; Scher et al., 2009). Furthermore,
because linear head acceleration is known to be monotonically cor-
relatedtoconcussionandskull fracturerisk(Greenwaldet al., 2008;
Mertz et al., 2003; Pellman et al., 2003) they are therefore known
to reduce the risk of sustaining head injury.
The biomechanical comparison that best matches the epidemi-
ological studies, and thus that would be best able to augment the
debate in that eld, is a comparison of helmeted and unhelmeted
head impact under identical impact conditions. Unfortunately,
these tests are difcult to perform because of limitations of the
magnesiumhead forms that are mandated in bicycle helmet stan-
dards and that have thus most often been used in bicycle helmet
impact tests. The magnesiumhead forms are at high risk of damage
if they are tested with no helmet and they have not been validated
to matchthe expectedhumanresponse for bare headimpacts. Thus
there is no test series available to our knowledge that contrasts
helmeted and unhelmeted impacts for contemporary bicycle hel-
mets under direct matchedimpact. Hodgsoncontrastedearly1990s
era helmets with an unhelmeted impact using a small humanoid
headform, Benz et al. dropped an unhelmeted Hybrid II headform
froma lower height than their helmeted impacts and Mattei et al.
dropped human cadaver skulls with and without helmets from
six and nine inch drop heights (Benz et al., 1993; Hodgson, 1990;
Mattei et al., 2012). All of these studies demonstrated a dramatic
decrease in head accelerations for the helmeted compared to the
Table 1
Comparison of several bicycle helmet standards.
Standard Reference Drop height (m) Drop height (feet) Criteria (gs)
Consumer Product Safety Commission (CPSC) 16 CFR Part 1203 2 6.6 300
Snell Memorial Foundation (Snell) BF95 (1998 Revision) 2.2 7.2 300
American Society for Testing and Materials (ASTM) ASTMF1447F12 2 6.6 300
Canadian Standards Association (CSA) CSA D113 2FM89 (Reafrmed 2004) 1.6 5.2 250
European Standards (CEN) EN 1078 1.5 4.9 250
P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17 3
Fig. 1. Photograph showing helmeted Hybrid III headform (left) and unhelmeted Hybrid III headform (right) in contact with the steel anvil. The Hybrid III headform was
attached to the ball armwhich was mounted to a linear bearing on the monorail drop-tower.
unhelmeted impacts. However, to the best of our knowledge, there
have beennostudies that compare the crucial situationof helmeted
andunhelmetedimpacts usingcontemporarybicyclehelmets, with
a head formthat is validated for bare head impacts and fromdrop
heights that compare to bicycle helmet standards and real world
cycling falls. The data that would result from such a test series is
directlyrelevant andindeedcentral totheongoingdebateof bicycle
helmet efcacy.
The objective of this study was to assess the biomechanical ef-
cacy of bicycle helmets to reduce risk of head injury in simulated
head impacts from drop heights consistent with bicycle helmet
standards and real world cycling head impacts. We designed and
fabricated a custom-made test xture that allowed us to attach a
Hybrid III headformto a monorail drop tower and performed both
helmeted and unhelmeted drops. The Hybrid III head form can be
tested without a helmet and it is validated in bare head impacts
(Foster et al., 1977). Linear head acceleration was measured, and
HIC and injury risk were determined from these accelerations to
ascertain the efcacy of helmets to reduce risk of head injury.
2. Methods and materials
We simulatedheadimpacts using a monorail droptower similar
to those specied in helmet certication standards. We fabricated
a custom-made test xture that allowed us to attach a Hybrid III
headform(Humanetics Inc., Plymouth, MI, USA), that corresponded
to a 50th percentile male head, to the drop tower. A ball-armwas
mounted to a monorail drop tower that was purpose-built for this
application. Pairedtests wereperformedinorder tostudytheriskof
injuryinhelmetedandunhelmetedimpacts. Apairedtest is dened
as two drops onto anidentical anvil andfromidentical dropheights
both with and without a helmet. The impact surface for all drops
was a at, xed steel anvil.
Fig. 1 shows the anvil, helmeted and unhelmeted Hybrid III
headformand features of the bearing and guide rail. Translational
acceleration along the direction of impact was measured using a
single axis accelerometer (2000g range, Endevco model 7264C-
2000, Meggitt Sensing Systems, San Juan Capistrano, CA, USA),
which was mounted to the center of the ball-arm which in turn
was placed within the Hybrid III head close to the head center of
mass. The mass of the entire dropassembly, including the Hybrid III
headform, ball-arm, and linear bearing was 5.05kg. The mass of the
helmet was approximately 0.25kg and was considered additional
mass for the helmeted drops.
The helmeted and unhelmeted drops were conducted from
nominal heights starting at 0.5m to 3m in 0.5m increments. This
range brackets heights used in certication standards but exceeds
themaximumheight of typical standards (CSAD113.2-M89(1.7m),
CPSC (2.0m), ASTM F1447 (2.0m), EN1078 (1.5m) Snell B95A
(2.2m)) to allow study of higher energy impacts that can occur
in real-world cycling where falls can happen while traveling at
considerable speed. Our testing range also brackets the range of
perpendicular impacts documented for reconstructed bicycle falls
(Fahlstedt et al., 2012). Two drops were performed at 0.5m, 1m,
1.5m, 2.5mand 3m; one drop for a helmeted Hybrid III headform
and one unhelmeted. Six drops, three helmeted and three unhel-
meted, were performed from 2m. More drops were performed at
2mthan other heights to obtain drop speed and acceleration data
that would allowa limited investigation of the repeatability of the
experiment andtodosoat commondropheight usedinbicycle hel-
met standards. To assess repeatability, we calculatedthe maximum
difference in both peak acceleration and HIC and expressed these
differences as a percentage of the mean peak acceleration and HIC.
In total sixteen drops were conducted (8 helmeted, 8 unhelmeted).
The headformwas adjusted so that impacts took place to the fore-
head of the headform as seen in Fig. 1. Actual drop heights were
increased by approximately 5cm above the nominal drop height
to account for friction in the drop rail. Speed at impact was calcu-
lated using high-speed video and was found to be within 5% of the
expected velocity for each respective drop height.
All helmets used in this work were CCM V15 Backtrail bicycle
helmets (Reebok-CCMHockey, Montreal, QC, Canada). The helmets
were constructed with a micro-shell and an expanded polystyrene
liner. The helmets conformed to the standards set out by the Con-
sumer Product Safety Commission (CPSC) (CPSC, 1998). In impacts
where helmets were used, the helmet was placed on the Hybrid
III headformin a standardized fashion. The orientation of the head-
formwas heldconstant for all dropheights using angle andposition
landmarks drawn on the Hybrid III headform (Fig. 1). The chin
retention strap was tightened to secure the helmet to the Hybrid III
headform(Fig. 1). The helmets were also equipped with a ratchet-
ing tension systemthat is designed to pass inferior to the occipital
protuberance. This was tightened prior to all drop tests. A check
for helmet t and secure attachment to the Hybrid III headform
involved manipulation of the helmet on the head to ensure no vis-
ible relative motion. Each helmet was used for a single drop and
then replaced with a newhelmet.
An Analog Devices (Analog Devices Inc., Norwood, MA) data
acquisition system was used to collect the data, with the acceler-
ation signal sampled at 39kHz and hardware anti-alias ltered to
complywithSAEJ211-1(SAEJ211Instrumentationfor Impact Test
Part 1: Electronic Instrumentation). In addition, the accelerom-
eter data were low-pass ltered at 1650Hz (CFC1000) during
post-processing as per SAE J211-1.
4 P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17
Fig. 2. Typical acceleration data plotted versus time for both a helmeted and unhel-
meted Hybrid III. Data shown is for the 2m drop height. Acceleration is expressed
in g, where one g corresponds to 9.81m/s
2
.
Peak head accelerations were used as a biomechanical metric of
helmet efcacy and were compared for the paired helmeted and
unhelmeted tests. In order to assess the risk of injury associated
with these impacts, head accelerations were also compared to the
Injury Assessment Reference Value (IARV), and probability curves
published by Mertz et al. (2003). For presenting peak acceleration
data, we use a 5% risk threshold for skull fracture based on peak
acceleration (180g).
The Head Injury Criterion (hereafter HIC) was calculated during
post-processingusingEq. (1). TheHICquanties headimpact sever-
ity by incorporating time of acceleration exposure and acceleration
magnitude.
HIC
15
=

1
t
2
t
1

t
2
t
1
a(t)dt

2.5
(t
2
t
1
)

max
(1)
For this analysis, a(t) is the head acceleration, in g, as measured
by the single axis accelerometer, and the time interval (t
2
t
1
)
was chosen to maximize HIC over a maximum duration of 15ms
(Eppinger et al., 1999). In the subject testing the sensing axis of
the accelerometer was aligned with the direction of impact and
thus captured the resultant acceleration. Similar to the accelera-
tionanalysis, HIC
15
values were comparedto the IARVof 700 which
Mertz et al. have reported corresponds to a 5% risk of AIS 4 brain
injury for the adult population (Mertz et al., 2003).
3. Results
Repeatability was evaluated by analyzing multiple drops at 2m
(3 helmeted drops and 3 unhelmeted drops). The maximuminter-
drop difference in peak acceleration was 1.5% and 3.3% (percentage
of mean peak accelerations for 2m drop) for the helmeted and
unhelmetedHybridIII headform, respectively. Similarly, maximum
inter-drop differences in HIC were 6.0% and 5.0% (percentage of
mean HIC), respectively.
Fig. 2 shows typical acceleration curves plotted over the time
of the impact event for both the unhelmeted and helmeted Hybrid
III headform. In general, the acceleration magnitudes plotted over
time exhibited a single abrupt increase in acceleration, which con-
tinues to the peak acceleration, followed by an abrupt decrease in
acceleration. Following this, the accelerations uctuate (e.g. Fig. 2,
after 5ms for the unhelmeted data) and these uctuations cor-
respond to head/helmetanvil interactions that are secondary to
the initial head/helmet-to-anvil impact (i.e. the head bounces
off of the anvil). In general, peak accelerations, HIC and injury
Fig. 3. Peak accelerations for both helmeted and unhelmeted drops. Numbers over
bars indicate peak acceleration. For 2m drop height, results stated are the mean
value calculated fromthree drops. Horizontal dashed line indicates the IARVof 180g
(5% chance of skull fracture) for a midsize male.
probability were all of smaller magnitude in drops where the
Hybrid III was helmeted. The duration of the impact pulse was
larger inhelmeted drops relative to unhelmeted. As the head decel-
erated, a small amount (approximately less than 5mm) of sliding
outward occurred between the helmet shell and the impact surface
as the helmet shell and liner deformed during impact.
Peak accelerations (Fig. 3) were smaller in helmeted drops rel-
ative to unhelmeted drops, for all drop heights. On average and
considering all drops, the peak accelerations for helmeted drops
were smaller by a factor of 4.2 relative to unhelmeted. For the
severity of impacts tested, peak accelerationexhibiteda linear rela-
tionship with drop height. In the unhelmeted situation, the head
accelerations were above the IARV of 180g for every drop from
0.5mto 3m. For drop heights of 0.51.5m, helmets decreased the
peak accelerations to a value belowthe IARV (Fig. 3).
Fig. 4 shows maximumHIC for helmeted and unhelmeted drops
fromall heights. For eachdrop height, helmets reduced HICrelative
to unhelmeted drops. The mean interval required to maximize HIC
for unhelmeted and helmeted drops was 1.0ms and 5.0ms, respec-
tively. The increased HIC interval for helmeted drops is consistent
withthe considerably wider (inthe time domain) accelerationpeak
for helmeted drops relative to unhelmeted shown in Fig. 2.
Fig. 5 shows the calculated risk of a severe brain injury (AIS 4+)
for all helmetedandunhelmeteddrops. Overall, thehelmeteddrops
dramatically reduced the risk across all drop heights. For drops of 1
meter andgreater, the unhelmetedconditionresultedinessentially
Fig. 4. Head Injury Criterion (HIC) calculated using HIC
15
convention for both hel-
meted and unhelmeted drops. Numeric values over bars indicate HIC values and
long dashed line indicates IARV based on HIC.
P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17 5
Fig. 5. Calculated probability of sustaining a severe brain injury (i.e. AIS 4+) based
on the HIC values for the helmeted and unhelmeted drops. Numeric values over bars
indicate risk magnitude.
100% chance of an AIS 4+ brain injury. For all drops of 2m or less,
the helmeted condition resulted in a risk of below 35%. However,
at the 3m drop height the risk of an AIS 4+ brain injury exceeded
90% for the helmeted condition.
For each of the helmeted drops, various levels of helmet defor-
mation and helmet shell and foam damage were evident. The
expanded polystyrene (EPS) foam cracked near the impact loca-
tion in all but the lowest drop height (0.5m) and the micro-shell
fractured in the 3m drop. Upon further inspection, all of the test
scenarios resulted in plastic deformation of the EPS foamliner near
the impact location.
4. Discussion
The overarching objective of this study was to assess the ef-
cacy of certied contemporary bicycle helmets to mitigate skull
and brain injury risk in head impacts with characteristic velocities
matching impact velocities fromhelmet certicationstandards and
also matching those fromthe biomechanical literature. We used a
purpose-built drop tower with a Hybrid III test headformand a uni-
axial accelerometer aligned to the direction of impact to measure
linear head accelerations during both helmeted and unhelmeted
drops. Linear head acceleration is one accepted mechanical mea-
sure that canbe relatedtobothskull injury risk andbraininjury risk
(Mertz et al., 2003; Pellman et al., 2003) both directly and through
HIC(Mertz et al., 2003), andtherefore accelerationwas our primary
biomechanical measure to assess efcacy.
We have characterized the ability of one typical contemporary
bicycle helmet to reduce the severity of a head impact and reduce
the risk of severe life-threatening skull and brain injury, compared
to not wearing a helmet, in matched impact tests where impact
severities (i.e. dropheight andpre-impact headvelocity) wereiden-
tical for the case of the helmeted and unhelmeted headform. The
tested helmet dramatically decreased peak linear head accelera-
tion (Fig. 3), HIC
15
(Fig. 4) and the potential for severe brain injury
(Fig. 5) in all impacts. Considering peak acceleration (Fig. 3), the
helmeted headform experienced accelerations below the IARV of
180g (Mertz et al., 2003) indrops from0.5m, 1.0m, and1.5mwhile
the unhelmeted headformexperienced acceleration well above the
IARV in drops from all heights. In drops from 2.0m up to 3.0m,
the helmeted headformexperienced accelerations above the IARV,
but helmeted headformaccelerations were at least 4 times smaller
thanthose of the unhelmetedheadform(Fig. 3). Evaluationof HIC
15
(Fig. 4) was consistent with the acceleration results when com-
paredwiththe IARVof 700. Probabilityof skull fracture (not shown)
and severe brain injury (Fig. 5) is reduced, for all drop heights, for
the helmeted headformrelative to the unhelmeted headform. This
biomechanical evidence clearly indicates that contemporary bike
helmets are highly effective at reducing injury risk through paired
helmeted and unhelmeted impacts with realistic drop heights and
impact speeds. For example, the helmets reduced the head peak
acceleration from 824g to 181g for drops of 2.0m reducing the
risk of skull fracture from99.9%+ to 5%.
The reductions in head acceleration and HIC
15
described above
demonstrate that certied helmets signicantly reduce the risk of
sustaining severe and even fatal injuries. It is worth noting that
common helmet standards do not seem to be designed to pre-
vent head accelerations fromexceeding the IARV values for severe
skull and brain injuries published by Mertz et al. (2003), although
it is noted that the helmets reduced the accelerations to a value
well belowthat specied in the standard. The helmeted drop from
2.0mresulted in a HIC of 1250 that corresponds to a 34% chance of
severe brain injury. It may be necessary to re-evaluate the helmet
standards and contemplate lowering the allowable accelerations in
some standards and to require testing frommultiple drop heights
in order to decrease the potential for serious and severe skull and
brain injuries fromfalls of various drop heights.
Common injuries coded AIS 4 (severe) and above include pen-
etrating skull injuries leading to brain injury, large contusions
(e.g., coupe-contrecoup, intermediate, andgliding) andhematomas
(e.g., subdural, subarachnoid, and intracerebral), as well as diffuse
axonal injury with associated loss of consciousness for a period
exceeding six hours. The severity of these injuries when coded
as 4+ range from severe to maximum (usually fatal) (Gennarelli
and Wodzin, 2005). Considering a realistic bicycle accident sce-
nario documented in the literature (Fahlstedt et al., 2012) where
a cyclist was thrown at 20km/h (i.e. 5.6m/s which corresponds to
a drop height of approximately 1.5m), our analysis indicates that
a helmeted cyclist in this situation would have a 9% chance of sus-
taining the severe brain and skull injuries noted above whereas an
unhelmeted cyclist would have sustained these injuries with 99.9%
certainty. In other words, a helmet would have reduced the prob-
ability of skull fracture or life threatening brain injury from very
likely to highly unlikely.
Evaluation of the 3mdrops demonstrate that helmets only offer
a nite amount of protection. However, the 7.7m/s impact speed
is not representative of most real-world bicycle impacts (Fahlstedt
et al., 2012). At impact speeds of this velocity the energy manage-
ment capability of the helmet is saturated (Newman, 2002) and
the EPS liner bottoms out. The range of drop heights that we tested
was consistent with the range of impact speeds that has been doc-
umented as plausible for cyclist impact scenarios. Our 3.0m, 2.5m,
and 2.0m drops were consistent with the resultant cyclist head
velocity in the Fahlstedt study that was oblique to the ground (i.e.
glancing off it). However, if the cyclist hit a curb or the wheel of
a car at the point of glancing off the ground then there would be
head impact of similar severity and impact direction to the higher
impact severity tests that we carried out.
The biomechanical results for acceleration and HIC
15
, and the
corresponding results for reduced risk of severe skull and brain
injury, complement epidemiological studies that have sought to
assess the protective efcacy of bicycle helmets. The majority of
the epidemiological studies suggest that helmets are effective at
reducing injury risk in a range of sports and across both adult and
youth segments of the population. However, concerns have been
raised in recent years over issues of selection-bias, failure to com-
pensate for time-trending and public policy, and lack of control for
confounding aspects of the input data for statistical studies includ-
ing type and mechanism of head/brain injury and severity of the
head impact (Curnow, 2006; Elvik, 2011). A fundamental reason
for this study is that the lack of appropriate knowledge of and
6 P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17
statistical control for the severity of the impact confounds the
results and decreases the signicance of results below what they
would be if appropriate controls were applied. As a consequence,
the results in some studies suggest helmets are only slightly effec-
tive at preventing head injuries or, in extreme cases, actually cause
head injuries (Elvik, 2011). However, the results of our biomechan-
ical study, where impact severity was controlled in helmeted and
unhelmeted impacts, strongly refute the epidemiological studies
that suggest helmets are marginally effective. Indeed, in all impacts
the risk of sustaining injury was reduced when the Hybrid III head-
form was helmeted, and in the case of a realistic bicycle head
impact (5.5m/s impact speed which corresponds to a drop height
of approximately 1.5m), a helmetedcyclist wouldhave a 9%chance
of sustaining AIS 4+ injuries whereas an unhelmeted cyclist would
almost certainly sustain these injuries.
The results of this study are in good agreement with previous
biomechanical research on the protective efcacy of bicycle hel-
mets. For example, Benz et al. (1993) conducteda study of childand
adolescent helmets in drops from1mto 1.5musing an unspecied
child headform (Benz et al., 1993) and a Hybrid II headform for
adolescent helmets, both dropped onto a at anvil, and reported
an overall threefold decrease in HIC as a result of protecting the
headformwith a helmet. Hodgson (1990) conducted a study using
a Hodgson-WSUheadformdropped from1mand 2monto at and
convex surfaces (Hodgson, 1990). For 1m drops, the unhelmeted
headformwas at increased risk of injury (7099% of the population
would be injured) relative to the helmeted headform (<1% of the
population would be injured) as indicated by the severity index for
head impact and results were similar for drops of 2m (Hodgson,
1990). Mattei et al. (2012) used pediatric skulls, both helmeted and
unhelmeted, in drops from 6 inches (0.15m) to 9 inches (0.23m)
onto a at surface and showed a maximum87% reduction in mean
acceleration as a result of skull protection through helmet use. The
results of our study were overall reductions in peak acceleration
(helmeted accelerations on average 4 times smaller), HIC
15
(hel-
meted HIC
15
were on average 6 times smaller) and risk of skull
and AIS 4+ brain injury, which are in overall agreement with the
previous literature. Direct comparison of our work to the previous
literature is difcult because our study has several key strengths
that are lacking in previous work. Unlike previous studies, we used
a Hybrid III headform that is validated for bare-head impacts and
is capable of surviving bare-head impacts from drops in excess
of 3m. Therefore, we were able to perform paired helmeted and
unhelmeted drops over a greater range of drop heights than that
of previous work, up to 3m, and were further able to document
the protective efcacy of bicycle helmets over this entire range.
We also used contemporary off-the-shelf helmets and tested over
a much broader range of impact severities than the previous stud-
ies and our drop experiments bracketed and incorporated the drop
heights prescribed in contemporary bicycle helmet standards.
Like all biomechanical studies relying on anthropomorphic test
devices (i.e. the Hybrid III headform), there are limitations of this
work. Inour studydropheight, impact speed, impact locationonthe
head/helmet, and helmet t on the head were all controlled, which
is in contrast to the real world where these parameters vary from
impact to impact and from individual to individual in the helmet
wearing population. Depreitere et al. (2004) showed in a sample of
86 bike accidents that impacts to the front of the head were 30% of
the total andthe front of the headwas the secondmost likely region
of impact. Therefore, our experimental protocol is representative of
commonrealistic impact scenarios that occur incycling, despite the
fact that it does not simulate all possible impacts. We conducted
isolated headform impacts in this study to simulate short dura-
tion head impacts. It is possible that, in some cycling impacts, the
effective mass of the head would be larger than the head mass we
tested here because some of the neck mass would couple to the
head. However, torso mass generally has little effect on the head
acceleration because of lag in time between head acceleration and
neck loading (Nightingale et al., 1997). A limitation of this work is
that we used only one model of helmet and thus were not able to
evaluate variation in the helmet mechanics as a function of design
features. However, this allowed us to minimize the variation in our
results and to study the repeatability of our tests. Also, the helmets
that we tested were certied to the CPSC standard and represent a
common design widely used in North American cities.
Bicycle helmets are effective at reducing peak translational
acceleration and HIC values; parameters that have been correlated
with risk of skull fracture and severe brain injury. For a 1.5mhel-
meteddrop, the riskof severe braininjurywas reducedfrom99.9%+
to9.3%. Thus, for realistic impact speeds (Fahlstedt et al., 2012) bicy-
clehelmets changedtheprobabilityof severebraininjuryfromvery
likely to highly unlikely. A contemporary helmet can transform a
headimpact that wouldresult insevere braininjury(whichinsome
cases could result in lasting disability) into an impact with little
potential for skull fracture or severe brain injury.
Acknowledgments
We gratefully acknowledge the Natural Sciences and Engi-
neering Research Council of Canada (NSERC - Research Tools and
Instruments Program) and the Canada Foundation for Innovation
(NewOpportunities Fund) for providing funding to purchase some
of themeasurement equipment usedinthis research. We alsothank
NSERC for supporting author CD with a postodoctoral fellowship.
References
Amoros, E., Chiron, M., Martin, J.-L., Thlot, B., Laumon, B., 2012. Bicycle helmet
wearing and the risk of head, face, and neck injury: a French casecontrol study
based on a road trauma registry. Inj. Prev. 18, 2732.
Beckwith, J.G., Greenwald, R.M., Chu, J.J., 2012. Measuring head kinematics in foot-
ball: correlation between the head impact telemetry system and Hybrid III
headform. Ann. Biomed. Eng. 40, 237248.
Benz, G., McIntosh, A., Kallieris, D., Daum, R., 1993. A biomechanical study of bicycle
helmets effectiveness in childhood. Eur. J. Pediatr. Surg. 3, 259263.
Canadian Institute for Health Information, 2010. Cycling Injury Hospitalizations in
Canada, 20092010.
Consumer Product Safety Commission, Safety Standard for Bicycle Helmets; Final
Rule, 16 CFR Part 1203, Federal Register, Vol.63 No.46, Tuesday March 10, 1998.
Curnow, W.J., 2003. The efcacy of bicycle helmets against brain injury. Accid. Anal.
Prev. 35, 287292.
Curnow, W.J., 2006. Bicycle helmets: lack of efcacy against braininjury. Accid. Anal.
Prev. 38, 833834.
DeMarco, A.L., Chimich, D.D., Gardiner, J.C., Nightingale, R.W., Siegmund, G.P., 2010.
The impact response of motorcycle helmets at different impact severities. Accid.
Anal. Prev. 42, 17781784.
Depreitere, B., Van Lierde, C., Maene, S., Plets, C., Vander Sloten, J., Van Audekercke,
R., Van der Perre, G., Gofn, J., 2004. Bicycle-related head injury: a study of 86
cases. Accid. Anal. Prev. 36, 561567.
Elvik, R., 2011. Publication bias and time-trend bias in meta-analysis of bicycle hel-
met efcacy: a re-analysis of Attewell, Glase and McFadden, 2001. Accid. Anal.
Prev. 43, 12451251.
Eppinger, R., Sun, E., Bandak, F., Haffner, M., Khaewpong, N., Maltese, M., Kuppa, S.,
Nguyen, T., Takhounts, E., Tannous, R., Zhang, A., Saul, R., 1999. Development of
Improved Injury Criteria for the Assessment of Advanced Automotive Restraint
Systems II. NHTSA Report.
Fahlstedt, M., Baeck, K., Halldin, P., Vander Sloten, J., Gofn, J., Depreitere, B., Kleiven,
S., 2012. Inuence of impact velocity and angle in a detailed reconstruction
of a bicycle accident. In: Proc. of the 2012 International IRCOBI Conf. on the
Biomechanics of Impacts, pp. 787799.
Foster, K.J., Kortge, J.O., Wolanin, M.J., 1977. Hybrid III-a biomechanically-based
crash test dummy. In: SAE International 770938.
Gennarelli, T.A., Wodzin, E., 2005. Abbreviated Injury Scale 2005.
Greenwald, R.M., Gwin, J.T., Chu, J.J., Crisco, J.J., 2008. Head impact severity measures
for evaluating mild traumatic brain injury risk exposure. Neurosurgery 62, 789.
Hagel, B.E., Barry Pless, I., 2006. A critical examination of arguments against bicycle
helmet use and legislation. Accid. Anal. Prev. 38, 277278.
Hamer, M., Chida, Y., 2008. Active commuting and cardiovascular risk: a meta-
analytic review. Prev. Med. 46, 913.
Heng, K.W.J., Lee, A.H.P., Zhu, S., Tham, K.Y., Seow, E., 2006. Helmet use and
bicycle-related trauma in patients presenting to an acute hospital in Singapore.
Singapore Med. J. 47, 367372.
P.A. Cripton et al. / Accident Analysis and Prevention 70 (2014) 17 7
Hodgson, V.R., 1990. Impact, Skid and Retention Tests on a Representative Group
of Bicycle Helmets to Determine Their Head-neck Protective Characteristics.
Wayne State University, Department of Neurosurgery.
Kendall, M., Walsh, E.S., Hoshizaki, T.B., 2012. Comparison between Hybrid III and
HodgsonWSUheadforms by linear andangular dynamic impact response. Proc.
Inst. Mech. Eng. P: J. Sports Eng. Technol. 226, 260265.
Maimaris, C., Summers, C.L., Browning, C., Palmer, C.R., 1994. Injury patterns in
cyclists attending an accident and emergency department: a comparison of
helmet wearers and non-wearers. BMJ 308, 15371540.
Mattei, T.A., Bond, B.J., Goulart, C.R., Sloffer, C.A., Morris, M.J., Lin, J.J., 2012. Perfor-
mance analysis of the protective effects of bicycle helmets during impact and
crush tests in pediatric skull models. J. Neurosurg.: Pediatr., 18.
Mertz, H.J., Irwin, A.L., Prasad, P., 2003. Biomechanical and scaling bases for frontal
and side impact injury assessment reference values. Stapp Car Crash J. 47, 155.
Newman, J.A., 2002. Biomechanics of head trauma: head protection. In: Acciden-
tal Injury: Biomechanics and Prevention. Springer Verlag, New York, USA, pp.
303323.
NHTSA Trafc Safety Facts 2010 Data, 2010. Bicyclists and Other Cyclists.
Nightingale, R.W., Richardson, W.J., Myers, B.S., 1997. The effects of padded surfaces
on the risk for cervical spine injury. Spine 22, 23802387.
Page, J.L., Macpherson, A.K., Middaugh-Bonney, T., Tator, C.H., 2012. Prevalence of
helmet use by users of bicycles, push scooters, inline skates and skateboards in
Toronto and the surrounding area in the absence of comprehensive legislation:
an observational study. Inj. Prev. 18, 9497.
Pang, T.Y., Thai, K.T., McIntosh, A.S., Grzebieta, R., Schilter, E., Dal Nevo,
R., Rechnitzer, G., 2011. Head and neck responses in oblique motorcycle
helmet impacts: a novel laboratory test method. Int. J. Crashworthiness 16,
297307.
Pellman, E.J., Viano, D.C., Tucker, A.M., Casson, I.R., Waeckerle, J.F., 2003. Concussion
in professional football: reconstruction of game impacts and injuries. Neuro-
surgery 53, 799812 (discussion 812814).
SAE J211 Instrumentation for Impact Test Part 1: Electronic Instrumentation.
Scher, I., 2006. Head injury in snowboarding: evaluating the protective role of hel-
mets. J. ASTMInt. 3.
Scher, I.S., Harley, E.M., Richards, D., Thomas, R., 2009. Likelihood of brain injury in
motorcycle accidents: a comparison of novelty and DOT-approved helmets. In:
SAE Technical Paper 01-0248.
Thompson, D.C., Rivara, F., Thompson, R., 1999. Helmets for preventing head and
facial injuries in bicyclists. In: The Cochrane Collaboration, Rivara, F. (Eds.),
Cochrane Database of Systematic Reviews. John Wiley & Sons, Ltd., Chichester,
UK.
Thompson, D.C., Rivara, F.P., Thompson, R.S., 1996. Effectiveness of bicyclesafetyhel-
mets in preventing head injuries. A casecontrol study. JAMA 276, 19681973.
Thompson, R.S., Rivara, F.P., Thompson, D.C., 1989. Acasecontrol study of the effec-
tiveness of bicycle safety helmets. N. Engl. J. Med. 320, 13611367.
Viano, D.C., Halstead, D., 2012. Change in size and impact performance of football
helmets fromthe 1970s to 2010. Ann. Biomed. Eng. 40, 175184.
Viano, D.C., Pellman, E.J., 2005. Concussion in professional football: biomechanics of
the striking playerPart 8. Neurosurgery 56, 266280 (discussion 266280).
Wen, L.M., Rissel, C., 2008. Inverse associations between cycling to work, public
transport, and overweight and obesity: ndings froma population based study
in Australia. Prev. Med. 46, 2932.