Professional Documents
Culture Documents
Articulo 1 PDF
Articulo 1 PDF
a r t i c l e i n f o a b s t r a c t
Article history: Background: The acceptance and usage of electric bicycles has rapidly increased in Switzerland in the
Received 24 March 2014 last years. Hence this topic has been addressed by policy makers with the aim to facilitate new transport
Received in revised form 14 July 2014 modes and, moreover, to improve their safety.
Accepted 18 July 2014
Methods: Police-recorded accidents of the years 2011 and 2012 involving a total of 504 e-bikers and 871
Available online 28 August 2014
bicyclists were analysed. National figures were compared with those of a rural and an urban environment.
Results: Most e-bikers who were involved in accidents were 40–65 years old. It was found that most e-
Keywords:
bikers sustained single accidents and that helmet usage was higher in the investigated rural environment
Electric bicycle
E-bike
than in the investigated urban area. The evaluation of the injury severity of e-bikers, particularly compared
Accident characteristics to bicyclists, lead to diverging results.
Injury severity Conclusions: The findings presented in this study are intended to serve as a benchmark since basic infor-
mation on characteristics of e-bike accidents is provided. With respect to differences between the injury
severity of e-bikers and bicyclists to-date no clear statement can be drawn. It is suggested to regularly
evaluate e-bike accidents to show trends and/or identify changes.
© 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.aap.2014.07.020
0001-4575/© 2014 Elsevier Ltd. All rights reserved.
48 T. Weber et al. / Accident Analysis and Prevention 73 (2014) 47–52
Table 2
Age distribution of e-bikers and bicyclists who sustained an accident. The totals of the percentages given in brackets may differ from 100% due to rounding effects.
E-bike [n] Bicycle [n] E-bike [n] Bicycle [n] E-bike [n] Bicycle [n]
3.2. Type of accident were taken into consideration. By conducting both methods dif-
ferences between the distributions of e-bikers and bicyclists were
All age groups were considered in the investigation of the type revealed to be statistically significant.
of accident. The analysis of the whole country and the investi- In the rural and urban environments most e-bikers and bicy-
gated rural environment revealed e-bikers to be involved in single clists were also slightly injured, but in contrast to the distribution
accidents most frequently. In contrast, in the whole country most of the whole country, e-bikers were uninjured to the same extent
bicyclists sustained crossing accidents and in the rural environment and in bicyclists the uninjured formed the second largest group.
most bicyclists sustained crossing- and single accidents, both being Moreover, only one bicyclist was killed. The distributions between
equally distributed. In the investigated urban environment e-bikers e-bikers and bicyclists were in both investigated areas not found to
sustained other accidents and bicyclists crossing accidents most be statistically significantly different, even when considering the
frequently (see Table 3). Thus differences between the distributions methods described above. Eventually, differences between the dis-
of e-bikers and bicyclists in the whole country and of e-bikers and tributions of e-bikers in the whole country and e-bikers in the rural
bicyclists in the investigated urban area were found to be statisti- area (p = 1.74E − 06), of e-bikers in the whole country and e-bikers
cally significant. E-bikers in the urban area sustained less single and in the urban area (p = 0.0004) as well as of bicyclists in the whole
crossing accidents and more turning and other accidents compared country and bicyclists in the rural area (p = 0.014) were found to be
to e-bikers in the rural area. The corresponding distributions were statistically significant.
also found to be statistically significant (p = 5.28E − 04). Moreover,
significantly different distributions were discovered between bicy-
3.5. Helmet usage and type of accident
clists in the whole country and in the rural area (p = 0.014) as well
as between bicyclists in the rural and in the urban environment
For the investigation of relations between the helmet usage
(p = 0.049).
and the type of accident again all age groups were considered. It
was found that e-bikers with or without helmet sustained single
3.3. Helmet usage accidents most frequently in the whole country as well as in the
investigated rural environment. Thus the distributions between
Again all age groups were taken into consideration. In the whole e-bikers with and without helmet were not statistically signifi-
country and in the investigated rural environment most of the cantly different. Most bicyclists who sustained an accident in the
accident-involved e-bikers wore helmets. In contrast, a clearly less whole country with or without helmet were involved in crossing or
frequent helmet usage was found in accident-involved e-bikers rather other accidents. In the investigated rural area most bicyclists
in the urban area. In comparison to e-bikers, however, bicyclists with helmet sustained single accidents and most bicyclists with-
wore helmets clearly less frequently in all investigated areas (see out helmet sustained crossing accidents. Differences between the
Table 4). Differences between the distributions of e-bikers and bicy- distributions of bicyclists with and without helmet were in both
clists in the whole country as well as of e-bikers and bicyclists investigated areas not statistically significant which is similar to
in the investigated rural area were statistically significant; in the what was found in e-bikers. Table 6 illustrates that in the urban
urban area helmet usage was comparatively low in both, e-bikers environment the distributions between e-bikers with and without
and bicyclists. Besides, statistically significant differences were helmet are apparently different; the same holds true for bicyclists.
determined between the distributions of e-bikers in the rural and However, due to small numbers the chi-square test was not applied
e-bikers in the urban area (p = 0.031) as well as of bicyclists in the to the analysis of the urban environment.
whole country and bicyclists in the urban environment (p = 0.046).
3.6. Injury severity and type of accident
3.4. Injury severity
For the investigation of relations between the injury severity
To eliminate age effects related to injury severity (i.e. elderly and the type of accident only two types of accidents (single acci-
may suffer from more severe injuries than younger people) only dents and collisions) and two age groups (0–39 and 40+) were
accidents with e-bikers and bicyclists of the age group 40–65 years selected. The age groups were specifically considered in order to
of age were analysed. determine differences between younger and older e-bikers and
Nationwide more than half of the accident-involved e-bikers bicyclists. This selection, however, only allowed a statistical eval-
and bicyclists were slightly injured. The second largest group was uation of the nationwide distributions (see Table 7). In the rural
severely injured and two e-bikers as well as three bicyclists were and urban environments the selection applied lead to even smaller
fatally injured. However, the distribution was statistically not sig- total numbers of cases and was thus neglected.
nificantly different (see Table 5). To ensure sufficiently large groups Younger e-bikers who sustained single accidents were in most
as required by the chi-square test (a) severely and fatally injured cases slightly or severely injured and younger e-bikers who were
were put into one group and (b) only slightly and severely injured involved in collisions were most frequently slightly injured or
50 T. Weber et al. / Accident Analysis and Prevention 73 (2014) 47–52
Table 3
Type of accident of e-bikers and bicyclists. The totals of the percentages given in brackets may differ from 100% due to rounding effects.
Table 4
Protection system of e-bikers and bicyclists who sustained an accident. The totals of the percentages given in brackets may differ from 100% due to rounding effects. The
exclusion of cases in which “belt worn” was reported explains variations in the total numbers.
Table 5
Injury severity of 40–65 year old e-bikers and bicyclists who sustained an accident. The totals of the percentages given in brackets may differ from 100% due to rounding
effects.
uninjured. Differences between these groups were not found to be Moreover, the distributions between younger and older e-bikers
statistically significant. who sustained collisions significantly differed (p = 0.008); in this
Older e-bikers who sustained single accidents or collisions were regard older e-bikers were more often severely and fatally injured
in most cases slightly or severely injured, but the rate of severely compared to younger e-bikers.
injured (older) e-bikers who were involved in single accidents was Younger and older bicyclists exhibit similar distributions as
higher than of those who sustained collisions. Thus distributions younger and older e-bikers. However, younger bicyclists who sus-
between these groups were found to be statistically significantly tained single accidents proportionally suffered from more severe
different. and fatal injuries than younger bicyclists who were involved in
Table 6
Type of accident related to helmet usage for e-bikers and bicyclists. The totals of the percentages given in brackets may differ from 100% due to rounding effects.
p-Value
p-Value
p-Value
p-Value
p-Value
E-bike [n] Bicycle [n] E-bike [n] Bicycle [n] E-bike [n] Bicycle [n]
0.959
0.798
0.401
n.a.
n.a.
Table 7
Injury severity related to the type of accident for specific age groups of e-bikers and bicyclists. The totals of the percentages given in brackets may differ from 100% due to
rounding effects.
p-Value
p-Value
E-bike [n] Bicycle [n]
0.010*
Fatally injured 0 (0%) 1 (2%)
0.121
Total 27 64
Collision Uninjured 9 (15%) 30 (13%)
Slightly injured 44 (73%) 165 (71%)
Severely injured 7 (12%) 36 (16%)
Fatally injured 0 (0%) 0 (0%)
Total 60 231
40+ Single Uninjured 5 (3%) 4 (4%)
Slightly injured 75 (50%) 47 (50%)
3.29E − 05*
Severely injured 66 (44%) 41 (44%)
0.005*
Fatally injured 4 (3%) 2 (2%)
Total 150 94
Collision Uninjured 10 (6%) 23 (11%)
Slightly injured 109 (64%) 150 (68%)
Severely injured 46 (27%) 43 (20%)
Fatally injured 6 (4%) 3 (1%)
Total 171 219
*
p-Value indicating a statistically significant difference.
collisions; the corresponding distributions revealed statistically found that about 80% of the accident-involved e-bikers were older
significant differences. The same was found in older bicyclists. than 40 years of age and more than 50% were older than 61 years of
age. Moreover, we determined that accident-involved e-bikers and
4. Discussion bicyclists tend to be younger in an urban than in a rural environ-
ment. Contrary to bicyclists only few accidents with e-bikers below
For the first time an evaluation of e-bike accidents in 23 years of age were reported. Thus it can be assumed that in urban
Switzerland was conducted with the aim to compare the find- areas only few young people own e-bikes and, hence, bicycles are
ings with bicycle accidents. Accidents of the years 2011 and 2012 used more frequently which might have an impact on the distri-
involving 504 e-bikers and 871 bicyclists were analysed. Lawinger butions analysed. Investigations taking the exclusion of school-age
and Bastian (2013), for instance, studied 126 and Otte et al. (2014) children (i.e. below 14 years of age) or rather the youngest age group
investigated 30 accidents with e-bikers involved. The total number (i.e. below 23 years of age) into account, however, revealed no sta-
of reported e-bike cases in this study is higher than in compara- tistical effects. Since the amount of school-age bicyclists was found
ble studies and thus provides a good basis to illustrate relevant to be nationwide only about 13% it was decided to not specifically
characteristics related to accidents. exclude them from the analyses.
It must be considered that the investigated database only The analysis of the type of accident revealed e-bikers to be
includes accidents which were recorded by police forces, i.e. if, for involved in single accidents most frequently, however, bicyclists
instance, a helmet prevented an injury the accident might not be sustained crossing accidents in most cases. Again these findings
officially reported. Hence the bias of under-reporting or rather the are similar to what was observed by Lawinger and Bastian (2013)
(currently unknown) estimated number of unreported cases, par- and Otte et al. (2014). In both studies it is assumed that the higher
ticularly in e-bikers, might be regarded as a limitation of this study. number of single accidents in e-bikes might be related to higher
In this context Thoma (1990) investigated road traffic accidents in velocities. First naturalistic driving studies (e.g. Twisk et al., 2013;
Switzerland and evaluated estimated numbers of unreported cases Dozza et al., 2013; Dozza and Fernandez, 2013) emphasize these
in different types of vehicles. In bicyclists, for example, he found findings by indicating higher velocities in e-bikers compared to
that the number of injured should be higher than actually reported. bicyclists.
A factor of 6.3 for slightly injured, of 4.0 for severely injured and of Nationwide 52% of the accident-involved e-bikers wore helmets.
1.0 for fatally injured bicyclists was determined; for very slightly In the rural area helmet usage was higher and in the urban environ-
injured bicyclists an even higher factor of up to 32 was suggested. ment it was less, while bicyclists wore helmets less frequently in all
In this regard also Elvik and Mysen (1999) analysed data from 13 investigated areas. Based on these findings it could be assumed that
countries and showed that reporting was incomplete in all levels e-bikers might have another risk awareness than bicyclists. How-
of injury severity. Reporting levels tended to be the highest in car ever, it also needs to be considered that this result might (only)
occupants and the lowest in cyclists. To what extent the findings reflect differences in exposure. Since exclusively police-reported
of these two studies are transferable to accident-involved e-bikers data was analysed, no information on exposure is available which
currently remains unclear. can be regarded as another limitation. Lawinger and Bastian (2013)
In the analysis of the injury severity it might be regarded as a reported a helmet usage rate in accident-involved e-bikers of about
limitation that only one age group was analysed. However, this was 50% which is similar to the overall helmet usage rate presented
specifically done to eliminate age effects. In this study we found above. In contrast, Otte et al. (2014) determined a clearly less fre-
that most e-bikers who were involved in an accident were 40–65 quent helmet usage rate in accident-involved e-bikers of only about
years old and thus older than most accident-involved bicyclists. 17%. The small number of investigated e-bike cases in comparison
This result is in line with what was observed by Lawinger and to the underlying study as well as to the study of Lawinger might
Bastian (2013) and Otte et al. (2014). Otte et al. (2014), for instance, explain this rather big difference.
52 T. Weber et al. / Accident Analysis and Prevention 73 (2014) 47–52
The methods and selections applied to the analysis of the injury are intended to serve as a benchmark since basic information on
severity of e-bikers and bicyclists lead to diverging results. At this e-bike accidents is provided. Thus it is recommended to regularly
point it remains unclear which of the methods described in the investigate e-bike accidents in order to show trends and/or identify
results section is more appropriate. In this context the conclusions changes. Improvements are proposed in terms of the optimisation
of Lawinger and Bastian (2013) and Otte et al. (2014) also differ. of police reports. In this regard the authors suggest an implemen-
With regard to injury severity another study analysing the tation of the localisation of injuries or rather to classify the injury
same Swiss accident data for 2011 did assign AIS (abbreviated severity by (M)AIS.
injury scale) codes for all casualties who were admitted to hos-
pital (Schmitt et al., 2014). AIS codes were derived from diagnosis Acknowledgements
codes (ICD) of the hospital statistics. No statistically significantly
different distribution between the AIS codes of e-bikers and bicy- We highly appreciate the support by the Police of the Canton
clists was found. Otte et al. (2014) also assigned (M)AIS codes for of Zurich and the Police of the City of Zurich as well as by the
all casualties and stated that the injury severity of e-bikers was Federal Roads Office (FEDRO). This study was funded by the State
not more severe than the injury severity of bicyclists. In contrast, Secretariat for Education, Research and Innovation (SERI) (Grant no.
Lawinger and Bastian (2013) evaluated the injury frequency and C12.0027) and conducted under the scope of COST Action TU1101
severity by using the descriptions of interviewed persons as well (“Towards safer bicycling through optimization of bicycle helmets
as descriptions written in police reports. They found that e-bikers and usage”).
were not involved in accidents more often than bicyclists, but it
was assumed that consequences from an accident could be worse References
for e-bikers than for bicyclists. As can be seen the results differ and
are not conclusive yet. Hence analyses of the distributions of the bfu, 2013. Status-Report 2013—Statistik der Nichtberufsunfälle und des Sicherheit-
following years and/or the use of medical data as a basis for further sniveaus in der Schweiz, 14–34.
Dozza, M., Werneke, J., Mackenzie, M., 2013. E-BikeSAFE: a naturalistic cycling
investigations is suggested. study to understand how electrical bicycles change cycling behaviour and
For the investigation of relations between injury severity and influence safety. Proceedings of the International Cycling Safety Conference,
type of accident a statistical evaluation was only established for 2013.
Dozza, M., Fernandez, A., 2013. Understanding bicycle dynamics and cyclist behavior
the nationwide distribution. Based on our findings it can be con- from naturalistic field data. IEEE Trans. Intell. Transp. Syst. Volume 15 Issue 1,
cluded that e-bikers aged 40+ are at higher risk of sustaining severe 376–384.
injuries than e-bikers aged 0–39. In a previous study with focus on Elvik, R., Mysen, A.B., 1999. Incomplete accident reporting: meta-analysis of studies
made in 13 countries. Transp. Res. Rec. 1665, 133–140.
bicycle accidents Otte et al. (2013) reported similar results in bicy-
Elvik, R., 2011. Publication bias and time-trend bias in meta-analysis of bicycle
clists. They determined a significant increase in the severity of head helmet efficacy. Accid. Anal. Prev. 43, 1245–1251.
injuries in bicyclists (without helmet) aged 40+. It was concluded Karkhaneh, M., Rowe, B.H., Saunders, L.D., Voaklander, D.C., Hagel, B.E., 2011. Bicycle
that with increasing age the injury severity of bicyclists (without helmet use four years after the introduction of helmet legislation in Alberta,
Canada. Accid. Anal. Prev. 43, 788–796.
helmet) in a head impact also increases. Kim, J., Kim, S., Ulfarsson, G., Porrello, L., 2007. Bicyclist injury severity in bicycle-
With the underlying data an investigation of relations between motor vehicle accidents. Accid. Anal. Prev. 39, 238–325.
injury severity and helmet usage was inappropriate since to-date’s Lawinger, T., Bastian, T., 2013. Neue Formen der Zweiradmobilität—Eine empirische
Tiefenanalyse von Pedelec-Unfällen in Baden-Württemberg. Zeitschrift für
police reports do not provide any information on the localisation Verkehrssicherheit 59 (2), 99–106.
of injuries. Therefore it is impossible to evaluate if a helmet had Otte, D., Facius, T., Wiese, B., 2013. Einflüsse auf das Verletzungsrisiko des Kopfes
a suitable effect. To resolve this problem, the authors suggest to von Radfahrern und Nutzen von Radhelmen zur Vermeidung und Minderung
von Verletzungen. Verkehrsunfall und Fahrzeugtechnik, 298–309, VKU 9/
implement the injury localisation in police reports and, again, to 2013.
classify the injury severity via (M)AIS. Both methods were already Otte, D., Facius, T., Müller, C., 2014. Pedelecs im Unfallgeschehen und Vergle-
applied in the investigations of Otte et al. (2014) and allowed ich zu konventionellen nicht motorisierten Zweirädern. Verkehrsunfall und
Fahrzeugtechnik, 48–60, VKU 2/2014.
assessing relations between injury severity and helmet usage as Schmitt, K.U., Baumgartner, L., Furter, K., Weber, T., Gubler, A., Scholz, S., Lüber,
well as other criteria. Hence in terms of the evaluation of injuries B., Thomas, P., 2014. Forschungspaket VeSPA, Teilprojekt 5: Medizinische
an enormous profit is expected. Folgen des Strassenunfallgeschehens: Schlussbericht (Medical outcomes of traf-
fic accidents: final report). Report, Swiss Federal Road Office, Available on
The results of this study provide an initial status in the evalu-
http://www.mobilityplatform.ch.
ation of e-bike accidents in Switzerland and complement existing Thoma, J., 1990. Das gesamte Ausmass der Strassenverkehrsunfälle, Pilotstudie.
studies. Schweizerische Beratungsstelle für Unfallverhütung, Bern.
Twisk, D.A.M., Boele, M.J., Vlakveld, W.P., Christoph, M., Sikkema, R., Remij,
R., Schwab, A.L., 2013. Preliminary results from a field experiment on
5. Conclusions e-bike safety: speed choice and mental workload for middle-aged and
elderly cyclists. Proceedings of the International Cycling Safety Conference,
In this study the first comprehensive evaluation of e-bike acci- 2013.
www.velosuisse.ch/de/statistik aktuell.html, (accessed 20.03.2014).
dents in Switzerland was conducted. The findings presented here