Published by Oxford University Press on behalf of the International Epidemiological Association International Journal of Epidemiology 2011;40:794–803

ß The Author 2011; all rights reserved. Advance Access publication 9 March 2011 doi:10.1093/ije/dyr040

Effectiveness of different types of

motorcycle helmets and effects of their
improper use on head injuries

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

Wen-Yu Yu,1,2 Chih-Yi Chen,3 Wen-Ta Chiu3 and Mau-Roung Lin3*
1
Department of Emergency Medicine, Taipei Medical University Hospital, Taipei, Taiwan, R.O.C., 2Department of Emergency
Medicine, China Medical University Hospital, Taichung, Taiwan, R.O.C. and 3Institute of Injury Prevention and Control,
College of Public Health and Nutrition, Taipei Medical University, Taipei, Taiwan, R.O.C.
*Corresponding author. Institute of Injury Prevention and Control, Taipei Medical University, 250 Wu-Hsing Street, Taipei 110,
Taiwan, R.O.C. E-mail: mrlin@tmu.edu.tw

Accepted 15 February 2011

Background Differences among three helmet types and the ineffectiveness of
improper helmet use in preventing head injuries are speculated
about but are seldom explored with evidence. A case–control
study was conducted to examine how different helmet types and
improper helmet use affected protection against head injuries
among motorcyclists in Taiwan.
Methods Case motorcyclists comprised 435 persons who sought emergency
care due to head injuries at a medical centre in west-central Taiwan
over an 8-month period and 23 motorcyclists who died from head
injuries at the scene of the crash; 458 motorcyclists who had
non-head injuries were used as the control group, and their crashes
occurred within 1 hour earlier or later than the corresponding cases.
Information on helmet type was validated by interviewing motor-
cyclists who were refuelling at petrol stations.
Results A conditional logistic regression analysis showed that compared
with helmeted motorcyclists, non-helmeted motorcyclists were
more than four times as likely to have head injuries [odds ratio
(OR) 4.54; 95% confidence interval (CI) 1.25–16.5] and ten times as
likely to have brain injuries (OR 10.4; 95% CI 1.82–59.2). Compared
with motorcyclists wearing full-face helmets, those wearing half-
coverage helmets were more than twice as likely to have head
injuries (OR 2.57; 95% CI 1.50–4.40) and brain injuries (OR 2.10;
95% CI 1.01–4.38). Compared with motorcyclists with firmly
fastened helmets, those with loosely fastened helmets increased
their risk of head injury (OR 1.94; 95% CI 1.33–2.82) and were
more than twice as likely to have brain injuries (OR 2.50; 95% CI
1.47–4.25).
Conclusions Of the three helmet types, half-coverage helmets provided motor-
cyclists the least protection from head injuries. Furthermore, wear-
ing a loosely fastened helmet may compromise any potential
protection.
Keywords Head injury, helmet, motorcycle, case–control study, crash severity,
epidemiology

794
 HEAD INJURIES: HELMET TYPES AND THEIR IMPROPER USE 795

Introduction motorcycle injuries were eligible for this study.

Cases were selected from among motorcyclists who
Head injuries lead to high mortality rates and incur sustained head injuries with the presence of at least
enormous economic costs;1,2 they are the most serious one of the following conditions: lacerations, abrasions
public health problem for motorcycle riders, particu- and bruises to the scalp, forehead and ears; fracture
larly in developing countries.3,4 In the USA, motor- of the skull and face; and brain injury, since a helmet
cycles are generally used for recreation and comprise might reasonably be expected to protect these areas
2% of registered motor vehicles.5 In contrast, in of the head. Brain injury was defined as a diagnosis
many Asian countries, motorcycles are one of the of concussion, cerebral contusions and lacerations,

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

most important means of transportation, and riders intracranial haemorrhage, loss of consciousness or
have especially high rates of injury.6 For example, post-traumatic amnesia. Furthermore, according to
motorcycle crashes account for 81% of head injuries medical examiner records, 23 motorcyclists who
in Vietnam7 and 450% of head injuries in Malaysia8 were citizens of Taichung died from brain injuries at
and Taiwan.9 the crash scene during the study period. Riders who
Empirical evidence strongly supports the conclusion were not operating a motorcycle—i.e. those who were
that wearing a helmet protects motorcycle riders from riding a minibike, a bicycle or a tricycle or wore a
the high risk of head injuries and death.10–13 This safety helmet for construction or were involved in a
evidence has facilitated legislation requiring manda- crash outside the city of Taichung—were excluded
tory motorcycle helmet use in many states in the USA from the study.
and in many other countries.14–19 Wearing a helmet For each case, a control was matched to the crash
effectively reduces head injuries among motorcyclists; time, selected by incidence density sampling from
even so, a substantial proportion of motorcycle riders injured motorcyclists who sought care at the same
who wear helmets still sustain head injuries in emergency room due to a crash resulting in injury
crashes.20,21 Three helmet types, full-face, open-face to the body but not the head. The crash time of the
and half-coverage, are commonly used by motorcycle control patient occurred within 1 h before or after that
riders; their effectiveness in preventing head injuries of the case. Matching was presumed to exclude con-
may differ and some types might even be inadequate. founding effects from environmental factors such as
It is not uncommon to observe a helmet becoming weather, road conditions and traffic volume. A poten-
detached in a motorcycle crash;21,22 for example, tial control was excluded if she/he had a history of a
nearly one-fourth of helmets came off during crashes previous head injury.
in Thailand and 5% in Los Angeles, CA, USA.23 In total, 458 case–control pairs participated in the
Moreover, head injuries seem to occur more fre- study. With the significance level set to 0.05 and
quently and are more severe for riders who wear a the power to 0.80, a sample size of 388 pairs was
non-standard helmet than those who wear a standard required if the prevalence of wearing half-coverage
helmet.24 The use of borrowed and poorly fitting hel- helmets was 0.60 in case motorcyclists and 0.50
mets is widely reported in many developing countries in control motorcyclists,27 and 356 pairs were re-
and more than one-third of riders exhibit improper quired if the prevalence of improper helmet use (e.g.
helmet use, such as wearing it on the back of the loosely fastened) was 0.40 and 0.3 in case and control
head and having a loose chin strap.25,26 motorcyclists,28 respectively. This research was
To our knowledge, potential differences in the approved by the Institutional Review Board of Taipei
effectiveness of various helmet types have not been Medical University, and informed consent was ob-
adequately examined, and no study has explored the tained from participants or the family of deceased
effects of improper use of helmets on head injuries
riders.
among motorcycle riders. To address these issues, a
case–control study was conducted to examine the ef-
fectiveness of different helmet types and improper hel- Exposure measurement
met use in protecting against head injuries among Exposure data were collected from medical records
motorcyclists. and personal interviews. Information on demo-
graphics such as age, gender, height, weight and edu-
cational level was extracted from medical records. In
Materials and Methods addition, a research nurse in the emergency room
interviewed participants using a standardized ques-
Study subjects tionnaire. For those who were fatally injured or
China Medical University Hospital is a 2200-bed, unable to communicate during their stay in the emer-
level-I trauma medical centre and can manage 50% gency room, their companions or paramedics were
of emergency patients in the city of Taichung, west- interviewed.
central Taiwan. During an 8-month period from 1 During the interview, information was collected
January to 31 August 2008, patients aged 515 years about helmet-related characteristics (see below),
who lived in Taichung and visited the emergency motorcycle licensure, safety-related traffic violations
room at China Medical University Hospital due to in the past year, riding speed, alcohol consumption
796 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

at the time of the crash, motorcycle engine volume Validation of exposure information
(450, 70–110 and 5125 cc), collision type (rear-end, To validate whether the distribution of helmet types
head-on or single-vehicle crash) and collision object worn by the emergency-room controls represented the
(moving motorcycle, moving car, other moving object, prevalence of those helmets on motorcyclists on the
stationary object and no object hit). roads, information on helmet type was collected from
motorcyclists who refuelled at petrol stations in
Taichung City. For each case motorcyclist, 10 petrol-
Helmet-related characteristics station motorcyclists were asked whether they had
had a crash in the past year, and if they answered posi-

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

The study had two hypotheses: full-face, open-face
and half-coverage helmets perform differently in pre- tively, the helmet type and fastening status were deter-
venting head injuries; and improper helmet use re- mined. The times of the interviews at the petrol stations
duces its effectiveness in preventing head injuries. were matched to the crash time of the cases (i.e. the
Helmet-related characteristics consisted of helmet same day of the week and the same time of the day),
use (helmeted or non-helmeted), helmet type, helmet and a petrol station nearest to the case crash site was
ownership (driver or other), helmet fit (good or poor selected. Of 4580 motorcyclists interviewed at 23 petrol
fit) and helmet cost (<300, 300–600 and 4NT$600) stations, 377 had experienced a motorcycle crash in the
past year.
(the exchange rate at the time of the study was
To validate the self-reported helmet fit and alcohol
US$1.00 ¼ NT$32.00), the manner of wearing the
consumption at the time of the crash, the occipital–
helmet (covering the entire head, worn on the back
frontal and inner helmet circumferences and blood
of the head or worn in reverse), fastening status
alcohol concentrations of the case and control motor-
(firmly or loosely fastened), helmet visor (pulled
cyclists were assessed. A helmet was considered to fit
down, not pulled down or without a visor) and helmet poorly when a difference between the two circumfer-
fixation during the crash (fixed on the head, dis- ences exceeded 3 cm. A blood alcohol concentration of
placed but still on the head or had come off). 550 mg/dl was considered to indicate recent alcohol
Improper helmet use that might impair the maximum consumption. In total, the occipital–frontal and inner
protection in a crash was ascertained in terms of a helmet circumferences of 336 motorcyclists who car-
borrowed helmet, a poorly fitting helmet, a poor- ried their helmets into the emergency room were
quality helmet (indicated by helmet cost <NT$300), assessed, and the blood alcohol concentrations of
an incorrect manner of wearing the helmet, being 679 riders were measured.
loosely fastened, no use of the visor and the helmet
having come off during the crash.
Statistical analysis
Distributions of helmet-related characteristics and
Crash severity other categorical variables among the case and control
groups were compared using Pearson’s chi-squared
Crash severity is an important confounder but seldom tests. Furthermore, a conditional logistic regression
controlled for in studies of helmet use and head inju- model was applied to investigate the independent
ries. Several variables such as the collision type, object relationships among head injury, helmet type and
of the collision, cost of repairs to the motorcycle and improper helmet use after adjusting for potential con-
Injury Severity Score (ISS) for injuries other than founders such as age, crash severity and behavioural
those to the head (1–8, 9–15 and 516 points)29,30 characteristics. To avoid large type II errors in variable
were used to indicate the crash severity. The cost of selection, variables with a P-value of <0.25 in the
repairs for motorcycle damage (<NT$2000, 2000–4999 bivariable logistic analysis were included in the initial
and 55000) was estimated by the manager of a local multivariable analysis.32 Moreover, relationships of
motorcycle body shop. When a subject was recruited, helmet type and improper helmet use with severe
the manager was notified to examine the involved head injury (i.e. brain injury) were further examined.
motorcycle by visiting the scene of the crash, a motor- Since use of an indicator value for missing values to a
cycle body shop or the subject’s home in order to regression model may introduce bias,33 our data ana-
estimate the repair cost of the motorcycle. An emer- lyses were limited to complete pairs in which both
gency physician (W-Y.Y) reviewed injured body re- cases and controls had exposure measurements.
gion(s) and computed the Abbreviated Injury Scale Distributions of helmet types among the emergency-
(AIS) and ISS for each subject. The AIS is an ordinal room controls and motorcyclists interviewed at petrol
scale of injury severity ranging from 1 (minor injury) stations were compared using Pearson’s chi-squared
to 6 (unsurvivable) for each of six body regions (head, tests. Kappa statistics were calculated to estimate
face, chest, abdomen, extremities/pelvis and external the measurement reliability of self-reported data on
skin).31 The AIS scores of the three most severely inju- helmet fit and alcohol consumption at the time of
red body regions other than the head were squared the crash. Statistical Analysis Software version 9.1
and added together to produce the non-head ISS (SAS Institute, Cary, NC, USA) was used for all stat-
scores. istical analyses.
 HEAD INJURIES: HELMET TYPES AND THEIR IMPROPER USE 797

Results at the time of the crash, and cost of repairs for motor-
cycle damage, compared with helmeted motorcyclists,
Of 4191 motorcyclists who sought treatment at the non-helmeted motorcyclists were more than four
emergency room during the study period, 1428 were times as likely to have head injuries (OR 4.54; 95%
diagnosed as having a head injury, of whom 535 sus- CI 1.25–16.5) and more than ten times as likely to
tained brain injuries. After matching the crash time, have brain injuries (OR 10.4; 95% CI 1.82–59.2).
458 motorcyclists with head injuries (23 drawn from Compared with motorcyclists who wore full-face hel-
medical examiner records), including 290 with brain mets, those wearing the open-face helmets did not
injuries, were successfully matched as the controls. have a significantly increased risk of head injury

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

The distribution of demographic and behavioural (OR 1.40; 95% CI 0.78–2.50) or brain injury (OR
characteristics for the 458 cases and 458 controls are 1.03; 95% CI 0.44–2.43), and those wearing half-
shown in Table 1. Compared with case motorcyclists, coverage helmets were more than twice as likely to
control motorcyclists were more likely to be younger or have head injuries (OR 2.57; 95% CI 1.50–4.40) and
older (P ¼ 0.06) and unlicensed (P < 0.001) and to have brain injuries (OR 2.10; 95% CI 1.01–4.38). Compared
lower education levels (P ¼ 0.007), higher riding speeds with motorcyclists whose helmets were firmly fas-
(P < 0.001) and positive alcohol consumption at the tened, those with loosely fastened helmets had an
time of the crash (P < 0.001). Furthermore, based on increased risk of head injury (OR 1.94; 95% CI 1.33–
the results of the bivariable logistic regression analyses, 2.82) and were more than twice as likely to have
unadjusted odds ratios (ORs) and 95% confidence brain injuries (OR 2.50; 95% CI 1.47–4.25).
intervals (CIs) of these characteristics as they related Of 377 motorcyclists interviewed at petrol stations
to the occurrence of head injuries are also given. who had experienced a motorcycle crash in the past
Table 2 shows the distributions of crash severity for year, 1.9% were not wearing a helmet at the time of
case and control motorcyclists. Compared with the the crash. Among helmeted motorcyclists, 15.7% were
controls, case motorcyclists were more likely to have wearing full-face helmets, 28.1% were wearing open-
had a head-on collision (P ¼ 0.007), to have higher face helmets and 56.2% were wearing half-coverage
repair costs for motorcycle damage (P < 0.001) and helmets. The distributions of helmet use (P ¼ 0.394)
higher non-head ISS scores (P < 0.001). and helmet type (P ¼ 0.07) among these motorcyclists
Distributions of helmet-related characteristics for did not differ from those of the emergency room con-
the case and control groups are shown in Table 3. trols. Of the petrol-station motorcyclists, 21.5% had
Compared with control motorcyclists, motorcyclists loosely fastened helmets; this differed from the
in the case group were less likely to have been wear- 33.8% of emergency room controls (P < 0.01) who
ing a helmet (P ¼ 0.01) and more frequently wore had loosely fastened helmets.
half-coverage helmets (P < 0.001). Among both case To test the measurement reliability, kappa values
and control motorcyclists, half-coverage helmets were were 0.72 for helmet fitness and 0.65 for alcohol
most commonly used, followed by full-coverage hel- use, indicating a substantial strength of agreement.34
mets. Motorcyclists in the case group more frequently Distributions of the two measures were similar be-
used a helmet that cost <NT$300 (P < 0.001), had tween the case (0.71 and 0.68) and control (0.73
a loosely fastened strap (P < 0.001) and lacked a and 0.63) groups.
helmet visor (P ¼ 0.005) than control motorcyclists.
Compared with control motorcyclists, helmets of
case motorcyclists were more likely to have been
displaced or to have come off in the crash Discussion
(P < 0.001). It was noted that missing values The results of this case–control study indicate that of
occurred more frequently in the case than in the con- the three commonly used helmet types, half-coverage
trol motorcyclists for helmet-related characteristics, helmets provide the least protection against head
particularly for the fixation status during the crash injuries for motorcycle riders when a crash occurs,
(13.7% vs 2.5%). and no significant difference in the protection
Table 4 shows the distributions of helmet type by was detected between full- and open-face helmets.
helmet-related characteristics. The three helmet Moreover, improper helmet use may affect hel-
types—full-face, open-face and half-coverage—dif- met fixation in a motorcycle crash and thus reduce
fered in terms of helmet cost (P < 0.001), helmet dam- the helmet’s effectiveness for preventing head
age prior to the crash (P ¼ 0.041), manner of wearing injuries.
the helmet (P < 0.001), fastening status (P ¼ 0.021), The better performance of full-face helmets in pre-
helmet visor (P < 0.001) and fixation status during venting head injuries over other helmet types may
the crash (P < 0.001). result from coverage of the entire head and the pres-
Table 5 shows the results of the multivariable con- ence of a chin bar. Usually, a motorcycle helmet con-
ditional logistic regression analyses respectively treat- sists of a hard shell of fibreglass or thermoplastic to
ing all head injuries and only brain injuries as the reduce the force of a direct blow to the skull, an
dependent variable. After controlling for age, motor- energy-absorbing foam liner to dissipate deceleration
cycle licensure, riding speed and alcohol consumption forces and a retention system consisting of a chin
798 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 1 Comparison of demographic and behavioural characteristics between case and control
motorcyclists

Cases (n ¼ 458) Controls (n ¼ 458)

Characteristic N (%) N (%) ORa (95% CI)
Age in years at the time of injury
<18 24 (5.2) 13 (2.8) 2.11 (1.06–4.21)
18–35 236 (51.5) 275 (60.2) 1.00
36–55 128 (28.0) 111 (24.3) 1.36 (1.00–1.87)

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

455 70 (15.3) 58 (12.7) 1.42 (0.92–2.18)
Gender
Female 248 (54.1) 273 (59.6) 1.00
Male 210 (45.9) 185 (40.4) 1.26 (0.96–1.67)
2
Body mass index (kg/m )
Underweight (<18.5) 61 (14.5) 46 (10.6) 1.37 (0.90–2.09)
Ideal (18.5–23.9) 228 (54.2) 245 (56.6) 1.00
Overweight (24.0–26.9) 75 (17.8) 100 (23.1) 0.82 (0.57–1.16)
Obese (527.0) 57 (13.5) 42 (9.7) 1.44 (0.94–2.19)
Educational level
College or above 109 (25.6) 152 (34.0) 1.00
Senior high 215 (50.5) 222 (49.7) 1.35 (0.98–1.85)
Junior high 45 (10.5) 28 (6.2) 2.14 (1.25–3.68)
Elementary or below 57 (13.4) 45 (10.1) 1.82 (1.11–2.96)
Motorcycle licensure
Licence for heavy motorcycles 278 (61.9) 354 (77.6) 1.00
Licence for light motorcycles 113 (25.2) 82 (18.0) 1.75 (1.26–2.44)
No motorcycle licence 58 (12.9) 20 (4.4) 3.68 (2.13–6.35)
Safety-related traffic violation in the past year
Yes 239 (56.2) 272 (60.0) 1.00
No 186 (43.8) 181 (40.0) 1.24 (0.94–1.62)
Riding speed at the time of the crash (km/h)
430 98 (23.1) 155 (34.3) 1.00
31–50 201 (47.4) 220 (48.7) 1.45 (1.06–2.00)
450 125 (29.5) 77 (17.0) 2.57 (1.73–3.82)
Alcohol consumption at the time of crash
Yes 35 (7.7) 3 (0.7) 19.8 (4.47–87.6)
No 421 (92.3) 454 (99.3) 1.00
Engine volume (cm3)
450 145 (33.9) 133 (29.4) 1.21 (0.88–1.67)
70–110 109 (25.5) 129 (28.5) 0.93 (0.67–1.29)
5125 174 (40.6) 190 (42.1) 1.00
a
OR and CI from bivariable analysis results of conditional logistic regression.

strap.35 Most half-coverage helmets do not have a however, they are less likely to meet national safety
visor, whereas full-face and open-face helmets do standards.36 There is still much that can be done to
have a visor or faceguard to cover the face. Full-face reduce head injuries and deaths among helmeted
helmets also incorporate a chin bar that extends up- riders, particularly in countries where half-coverage
wards to a height just below the lips. Half-coverage helmets are commonly used (e.g. 460% of
helmets are cheaper than other helmet types; Taiwanese in this study).
 HEAD INJURIES: HELMET TYPES AND THEIR IMPROPER USE 799

Table 2 Comparison of crash severity between case and control motorcyclists

Cases (n ¼ 458) Controls (n ¼ 458)

Characteristic N (%) N (%) ORa (95% CI)
Collision type
Rear-end 189 (45.3) 253 (56.1) 1.00
Head-on 172 (41.3) 149 (33.0) 1.51 (1.11–2.04)
Single vehicle 56 (13.4) 49 (10.9) 1.50 (0.97–2.33)

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

Collision object
Moving motorcycle 94 (21.8) 115 (25.2) 1.00
Moving car 236 (54.8) 249 (54.5) 1.12 (0.80–1.59)
Other moving objects 17 (3.9) 22 (4.8) 0.96 (0.45–2.05)
Stationary objects 28 (6.5) 21 (4.6) 1.50 (0.78–2.88)
No object hit 56 (13.0) 50 (10.9) 1.40 (0.86–2.28)
Non-head ISS (points)
1–8 404 (88.2) 438 (95.6) 1.00
9–15 27 (5.9) 17 (3.7) 1.73 (0.92–3.76)
516 27 (5.9) 3 (0.7) 8.93 (2.69–29.7)
Cost of motorcycle repair (NT$)b
<2000 139 (33.3) 239 (53.7) 1.00
2000–4999 90 (21.6) 114 (25.6) 1.36 (0.95–1.96)
55000 109 (26.2) 78 (17.5) 2.43 (1.66–3.55)
Not repairable 79 (18.9) 14 (3.2) 6.86 (3.49–13.5)
a
OR and CI from bivariable analysis results of conditional logistic regression.
b
The exchange rate was US$1.00 ¼ NT$32.00 at the time of this study.

Our results also reflect the importance of correct hel- severity for motorcycle crashes. In this study, crash
met use for maximal protection against head injuries. severity was indicated by the cost of repairs to the
Given the same crash severity, the fixation of a damaged motorcycle, the type of collision and the
helmet during a crash was most closely related to object with which the motorcyclist had collided.
the retention system. The result of a further subgroup Motorcycle repair costs might be affected by social
analysis showed that helmet detachment was and economic factors, and collision type and collision
correlated with the helmet being loosely fastened object might not be able to discriminate severity levels
[point biserial correlation coefficient (rpb) ¼ 0.64]. among various collision objects or among single-
Nevertheless, helmet retention systems are rarely vehicle crashes. A commonly used alternative is a
tested in the laboratory or regulated by safety stand- modification of the ISS that calculates injuries to
ards, and very few empirical data are available. In body regions other than the head.30 However, the
many Asian countries, more than one-third of motor- non-head ISS index could no better explain the risk
cycle riders were found to wear a helmet improperly, of head injury in the regression model, and more im-
such as wearing it unfastened or loosely fastened in portantly, it had to assume that the occurrence of
order to exhibit ‘token’ compliance with helmet-use injuries to body regions other than the head was in-
laws, and some even put on a helmet only when the dependent of the incidence of head injuries or the use
police are nearby.25 Furthermore, the police seldom of helmets. In reality, this assumption might not be
enforce proper helmet use.26 In sum, this study dem- valid.39
onstrates for riders, policemen and policymakers that There are several limitations to the present study.
wearing a helmet improperly might not be sufficient First, a referral pattern might exist if case and control
to prevent head injuries. motorcyclists were not selected from the same popu-
Crash severity is seldom controlled for in studies of lation base. Motorcyclists who had serious injuries
helmet use and head injuries. Without controlling were probably more likely to have been referred to
for it, such studies implicitly assume similar distri- the study hospital than those with minor or no inju-
butions of crash severities between helmeted and ries. For instance, the emergency room controls had a
non-helmeted riders, and this assumption is often different distribution of helmet-fastening status
violated.37,38 There are no standardized measures of compared with motorcyclists at petrol stations.
800 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 3 Comparison of helmet-related characteristics between case and control motorcyclists

Cases (n ¼ 458) Controls (n ¼ 458)

Characteristic N (%) N (%) OR (95% CI)
Helmet usea
Helmeted 436 (95.2) 452 (98.7) 1.00
Non-helmeted 22 (4.8) 6 (1.3) 3.50 (1.29–9.54)
Helmet type

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

Full-face 50 (11.6) 73 (17.0) 1.00
Open-face 106 (24.7) 149 (34.6) 1.05 (0.65–1.69)
Half-coverage 274 (63.7) 208 (48.4) 2.06 (1.34–3.17)
Helmet ownership
Driver 367 (90.8) 402 (93.7) 1.00
Other 37 (9.2) 27 (6.3) 1.43 (0.85–2.42)
Helmet fit
Good fit 379 (94.0) 396 (92.3) 1.00
Poor fit 24 (6.0) 33 (7.7) 0.69 (0.39–1.23)
b
Helmet cost (NT$)
<300 194 (57.9) 156 (42.5) 1.96 (1.22–3.16)
300–599 102 (30.5) 150 (40.9) 1.08 (0.65–1.78)
5600 39 (11.6) 61 (16.6) 1.00
Manner of wearing helmet
Covering the whole head 293 (72.9) 302 (70.6) 1.00
Worn on the back of the head 90 (22.4) 104 (24.3) 0.87 (0.63–1.21)
Worn in reverse 19 (4.7) 22 (5.1) 0.81 (0.42–1.58)
Fastening status
Firmly fastened 222 (51.6) 286 (66.5) 1.00
Loosely fastened 208 (48.4) 144 (33.5) 1.98 (1.47–2.66)
Helmet visor
Pulled down 136 (33.9) 185 (43.7) 1.00
Not pulled down 63 (15.7) 73 (17.3) 1.17 (0.78–1.77)
Without a visor 202 (50.4) 165 (39.0) 1.70 (1.23–2.36)
Fixation status during the crash
Fixed on the head 214 (56.9) 351 (82.6) 1.00
Displaced but still on the head 54 (14.4) 47 (11.1) 2.07 (1.27–3.37)
Had come off 108 (28.7) 27 (6.3) 8.12 (4.66–14.1)
a
Data on non-helmeted motorcyclists are not presented other than ‘Helmet use’.
b
The exchange rate was US$1.00 ¼ NT$32.00 at the time of this study.

Nevertheless, the biased direction of the referral pat- bias might exist because eligible subjects who had
tern may have tended to underestimate the effect of serious head injuries were less likely to respond to
improper helmet use. Second, riders who wore our interviews, and case motorcyclists were more
full-face helmets may have been more aware of traffic likely than control motorcyclists to have proxy infor-
safety than those wearing half-coverage or open-face mation and missing values. Accordingly, differential
helmets. Personality characteristics such as ascertainments of exposure information such as the
risk-taking were not directly measured and controlled helmet type and fastening status between the case
for in the study, although they may have differed and control groups might have occurred, and a pro-
among riders who wore different helmet types and spective cohort study to validate these results is war-
with different fastening statuses. Third, information ranted. Fourth, to estimate the population-based
 HEAD INJURIES: HELMET TYPES AND THEIR IMPROPER USE 801

Table 4 Distribution of helmet type with other helmet-related characteristicsa

Helmet type
Characteristic Full-face (n ¼ 123) Open-face (n ¼ 255) Half-coverage (n ¼ 482) P-value
N (%) N (%) N (%)
Helmet ownership
Driver 112 (94.9) 222 (91.4) 435 (92.0) 0.483
Other 6 (5.1) 21 (8.6) 37 (8.0)

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

Helmet fit
Good fit 108 (91.5) 224 (92.2) 443 (94.1) 0.484
Poor fit 10 (8.5) 19 (7.8) 28 (5.9)
Helmet cost (NT$)b
<300 1 (1.0) 52 (25.4) 297 (75.4) <0.001
300–599 42 (40.4) 118 (57.5) 92 (23.6)
5600 61 (58.6) 35 (17.1) 4 (1.0)
Manner of wearing helmet
Covering the whole head 105 (89.0) 190 (78.2) 300 (63.8) <0.001
Worn on the back of the head 11 (9.3) 37 (15.2) 146 (31.1)
Worn in reverse 2 (1.7) 16 (6.6) 23 (5.1)
Fastening status
Firmly fastened 80 (65.0) 163 (63.9) 265 (53.8) 0.022
Loosely fastened 43 (35.0) 92 (36.1) 217 (46.2)
Helmet visor
Pulled down 69 (59.5) 138 (57.3) 114 (24.3) <0.001
Not pulled down 36 (31.0) 43 (17.8) 57 (12.1)
Without a visor 11 (9.5) 60 (24.9) 296 (63.6)
Fixation status during the crash
Fixed on the head 88 (75.2) 186 (79.2) 291 (64.8) <0.001
Displaced but still on the head 8 (6.8) 28 (11.9) 65 (14.6)
Had come off 21 (18.0) 21 (8.9) 93 (20.6)
a
Excluding 28 case–control pairs in which case or control motorcyclists did not wear a helmet.
b
The exchange rate was US$1.00 ¼ NT$32.00 at the time of this study.

Table 5 Results of a conditional logistic regression analysis prevalences of helmet types and fastening statuses,
treating all head injuries and only brain injuries as de- data from petrol station motorcyclists at the time of
pendent variables, respectively crash would be more valid than at the study time.
However, direct observations at petrol stations may
All head injuriesa Only brain injurya help avoid recall errors, memory lapses and defensive
Characteristic OR (95% CI) OR (95% CI)
responses. Fifth, missing data on some helmet-related
Helmet use characteristics such as fixation status during the crash
Helmeted 1.00 1.00 were unbalanced between case and control motorcyc-
Non-helmeted 4.54 (1.25–16.5) 10.4 (1.82–59.2) lists; in addition, only 36.7% of motorcyclists (39.3%
Helmet type of cases and 32.3% of controls) carried their helmets
Full-face 1.00 1.00 into the emergency room. Further analyses showed
that motorcyclists who had missing values for the fix-
Open-face 1.40 (0.78–2.50) 1.03 (0.44–2.43)
ation status during the crash or those who did not
Half-coverage 2.57 (1.50–4.40) 2.10 (1.01–4.38)
carry their helmets tended to have higher non-head
Fastening status ISS scores or higher repair costs for motorcycle
Firmly fastened 1.00 1.00 damage (i.e. higher levels of crash severity). Since
Loosely fastened 1.94 (1.33–2.82) 2.50 (1.48–4.25) the repair cost for motorcycle damage was controlled
a
Controlled for age, motorcycle licensure, riding speed and
for in the statistical analysis, a potential bias in
alcohol consumption at the time of the crash, and cost of the result due to this selecting factor should have
motorcycle repairs. been avoided. Finally, there was a possibility
802 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

of misdiagnosing motorcyclists who consumed alcohol wear safer helmets (e.g. full-face helmets) as well as
as having head injuries. Nevertheless, the proportion to fasten the helmets properly.
of motorcyclists who consumed alcohol at the time of
the crash was very low in the study.
This study demonstrates that the type and fas- Funding
tening status of a helmet can affect the helmet’s ef-
fectiveness in preventing head injuries. For a long National Science Council (NSC97-2314-B-038-012-
time, injury prevention workers have focused on MY3) and National Health Research Institute
whether or not motorcycle riders wear a helmet, (NHRI-EX98-9805PI and NHRI-EX100-9707PI),

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

rather than on which helmet type should be worn Taiwan, R.O.C.
or how the helmet is worn. Now it is high time to Conflict of Interest: None declared.
promote legislation that requires motorcycle riders to

KEY MESSAGES
 Among full, open-face and half-coverage helmets, the latter gave motorcyclists the least protection
from head injuries.
 Improper helmet use may affect helmet fixation in a crash and thus reduce the helmet’s effectiveness
for preventing or reducing head injuries.
 It is time to promote legislation that requires motorcyclists to wear safer helmets and to fasten the
helmets properly.

11
References Norvell DC, Cummings P. Association of helmet use with
1
death in motorcycle crashes: a matched-pair cohort study.
Langlois JA, Rutland-Brown W, Thomas KE. Traumatic Am J Epidemiol 2002;156:483–87.
Brain Injury in the United States: Emergency Department 12
Offner PJ, Rivara FP, Maier RV. The impact of motorcycle
Visits, Hospitalizations, and Deaths. Atlanta, GA: Centers of helmet use. J Trauma 1992;32:636–41.
Disease Control and Prevention, National Center for 13
Injury Prevention and Control, 2006. Rowland J, Rivara F, Salzberg P, Soderberg R, Maier R,
2 Koepsell T. Motorcycle helmet use and injury outcome
Lee ST, Lui TN, Chang CN, Wang DJ, Heimburger RF,
and hospitalization costs from crashes in Washington
Fai HD. Features of head injury in a developing
State. Am J Public Health 1996;86:41–45.
country—Taiwan (1977–1987). J Trauma 1990;30:194–99. 14
3 Sosin DM, Sacks JJ. Motorcycle helmet-use laws and
Ankarath S, Giannoudis PV, Barlow I, Bellamy MC,
head injury prevention. JAMA 1992;267:1649–51.
Matthews SJ, Smith RM. Injury patterns associated 15
with mortality following motorcycle crashes. Injury Kraus JF, Peek C, McArthur DL et al. The effect of the
2002;33:473–77. 1992 California motorcycle helmet use law on motorcycle
4 crash fatalities and injuries. JAMA 1994;272:1506–11.
Lin MR, Kraus JF. A review of risk factors and pat- 16
terns of motorcycle injuries. Accid Anal Prev 2009;41: Ferrando J, Plasencia A, Oros M, Borrell C, Kraus JF.
710–22. Impact of a helmet law on two wheel motor vehicle
5 crash mortality in a southern European urban area. Inj
National Highway Traffic Safety Administration. Traffic
Prev 2000;6:184–88.
Safety Facts 2005: Motorcycles. Washington, DC: National 17
Highway Traffic Safety Administration, 2007. Guillen GF, Aguinaga OI, Aguinaga OE. Helmet use by
6
Ameratunga S, Hijar M, Norton R. Road-traffic injuries: drivers and passengers of motorcycles in Pamplona
confronting disparities to address a global-health prob- (Spain), 1992. Eur J Epidemiol 1995;11:87–89.
18
lem. Lancet 2006;367:1533–40. Servadei F, Begliomini C, Gardini E, Giustini M, Taggi F,
7
Hung DV, Stevenson MR, Ivers RQ. Motorcycle helmets Kraus J. Effect of Italy’s motorcycle helmet law on trau-
in Vietnam: ownership, quality, purchase price, and af- matic brain injuries. Inj Prev 2003;9:257–60.
19
fordability. Traffic Inj Prev 2008;9:135–42. Chiu WT, Kuo CY, Hung CC, Chen M. The effect of
8
Radin-Umar RS. Motorcycle safety programmes in the Taiwan motorcycle helmet use law on head injuries.
Malaysia: how effective are they? Int J Inj Control Safety Am J Public Health 2000;90:793–96.
20
Prom 2006;13:71–79. Sarkar S, Peek C, Kraus JF. Fatal injuries in motor-
9
Ministry of Transportation and Communications (R.O.C.). cycle riders according to helmet use. J Trauma 1995;38:
Monthly Statistics of Transportation and Communications. Taipei, 242–45.
21
Taiwan: Ministry of Transportation and Communications, Richter M, Otte D, Lehmann U et al. Head injury mech-
Executive Yuan, 2007. anisms in helmet-protected motorcyclists: prospective
10
Liu B, Ivers R, Norton R, Blows S, Lo SK. Helmets for multicenter study. J Trauma 2001;51:949–58.
22
preventing injury in motorcycle riders. Cochrane Database Richards PG. Detachment of crash helmets during motor-
Syst Rev 2004;2:CD004333. cycle accidents. BMJ 1984;288:758.
 HEAD INJURIES: HELMET TYPES AND THEIR IMPROPER USE 803

23 31
Ouellet JV, Kasantikul V. Motorcycle helmet effect on a Association for the Advancement of Automotive Medicine
per-crash basis in Thailand and the United States. Traffic (AAAM). The Abbreviated Injury Scale—1990 Revision. Des
Inj Prev 2006;7:49–54. Plain, IL: AAAM, 1990.
24 32
Peek-Asa C, McArthur DL, Kraus JF. The prevalence of Mickey RM, Greenland S. The impact of confounder se-
non-standard helmet use and head injuries among lection criteria on effect estimation. Am J Epidemiol 1989;
motorcycle riders. Accid Anal Prev 1999;31:229–33. 129:125–37.
25 33
Li LP, Li GL, Cai QE, Zhang AL, Lo SK. Improper motor- Rothman KJ, Greenland S, Lash TL (eds). Modern
cycle helmet use in provincial areas of a developing coun- Epidemiology. 3 rd edn. Philadelphia, PA: Lippincott
try. Accid Anal Prev 2008;40:1937–42. Williams and Wilkins, 2008.

Downloaded from https://academic.oup.com/ije/article-abstract/40/3/794/746475 by University of Tehran user on 25 February 2020

26 34
Conrad P, Bradshaw YS, Lamsudin R, Kasniyah N, Landis JR, Koch GG. The measurement of observer agree-
Costello C. Helmets, injuries, and cultural definitions: ment for categorical data. Biometrics 1977;33:159–74.
motorcycle injury in urban Indonesia. Accid Anal Prev 35
Hopes PD, Chinn BP. Helmets: a new look at design and
1996;28:193–200. possible protection. International Conference on the
27
Liang RC, Lin YH. Knowledge on Protection and Use of Biomechanics of Injury. Stockholm, 1989; 39–54.
Motorcycle Helmets in College Students. The 15th Meeting 36
Bureau of Standard, Metrology, and Inspection (R.O.C.).
of the Ergonomics Society of Taiwan; 2008 March 15. Taipei, Criterion Modification of Motorcycle Helmets and Enforcement of
Taiwan, R.O.C. Helmet Quality to Protect Children Riders. http://www.bsmi
28
Lin MR, Kraus JF. A review of risk factors and patterns of .gov.tw/wSite/public/data/f122275470185.doc (15 March
motorcycle injuries. Accident Anal Prev 2009;41:710–22. 2009, date last accessed).
29 37
Baker SP, O’Neill B, Haddon W et al. The injury severity Lin MR, Kraus JF. Methodological issues in motorcycle
score: a method for describing patients with multiple injury epidemiology. Accid Anal Prev 2008;40:1653–60.
injuries and evaluating emergency care. J Trauma 1974; 38
Prasad AK. CRASH3 Damage Algorithm Reformulation for
14:187–96. Front and Rear Collisions. Warrendale, PA: Society for
30
Rutledge R, Stutts J. The association of helmet use Automotive Engineering, 1990.
with the outcome of motorcycle crash injury when con- 39
Lin MR, Hwang HF, Kuo NW. Crash severity, injury pat-
trolling for crash/injury severity. Accid Anal Prev 1993;25: terns, and helmet use in adolescent motorcycle riders.
347–53. J Trauma 2001;50:24–30.