Exposure-based traffic crash injury rates by mode of travel in British Columbia.
Teschke, Kay ; Harris, M. Anne ; Reynolds, Conor C.O. 等
Unintentional injuries are the fifth leading cause of death in
Canada and the leading cause of death of those between the ages of 1 and
44. (1,2) Traffic crashes are the cause of about one quarter of these
fatalities (estimates for 2008: n=2,628, Statistics Canada; n=2,419,
Transport Canada). (3,4) Non-fatal injuries from traffic crashes affect
many more Canadians, with about 176,000 police-reported injuries
annually and between 12,000 and 20,000 hospitalizations (n=12,360 in
2008, Transport Canada; n=20,140 in 2005/6, Public Health Agency of
Canada). (4,5)
Despite the importance of these injuries to public health, we have
much to understand about traffic-related morbidity and mortality in
Canada. International agencies, including the Organisation for Economic
Co-operation and Development and the World Health Organization, have
focused attention on road safety and have developed profiles by country
to allow comparisons and provide opportunities to discover best
practices. (6,7) One way to compare data is to examine mortality and
morbidity rates for the population as a whole. Such rates are
straightforward to calculate using death and injury numerators and
census denominators, and Canadian contributions to international
databases have included these rates (e.g., Figure 1).
Injury risks vary by mode of transportation (e.g., car, bicycle,
walking, motorcycle, transit). (8-10) Understanding these differences is
important for prevention. However, since travel modes are not used
equally, injury rates calculated with a population denominator may
reflect differences in the share of trips rather than differences in
risk between modes. "Exposure-based" denominators take into
account proportions of trips or distances travelled by each mode, and
provide a better basis for comparisons. Some countries conduct national
travel surveys that provide such denominator data.6,8-10 In Canada,
there is no national travel survey. The long-form Census includes a
question querying the usual mode of travel to work, but it excludes
those not in the workforce and non-work trips of those who are. This
means that in international comparisons, we cannot benchmark rates of
injury and death for each mode of travel.
To begin to provide exposure-based rates in Canada and show their
value, here we use data from one province, British Columbia (BC), to
calculate crude injury and death rates for three road user classes:
motor vehicle occupants, pedestrians, and bicyclists. Incomplete data
are provided for motorcyclists and transit users.
METHODS
We gathered data about injuries, fatalities, and travel by each
mode. The data were used to calculate injury and fatality rates using
three commonly used denominators: (6,8,9) per 100,000 population; per
100 million person-trips; and per 100 million kilometres travelled.
We used three published data sources. Population-level injury data
were abstracted from the British Columbia Motor Vehicle Branch
"Traffic Collisions Statistics, Police-attended Injury and Fatal
Collisions". The three most recent years for which data were
available (2005, 2006, 2007) were used.11-13 The 2008 Regional Trip
Diary Survey (n=17,603 households) conducted by TransLink (the Metro
Vancouver regional transportation authority) was used to provide the
average number of trips by all modes, average distances travelled per
trip by each mode, and the proportions of all trips by each mode in
Metro Vancouver (including both work and non-work trips of employed
persons, and trips by youth, the elderly and others not in the
workforce). (14) The Census of Canada provided the population of BC in
2006. (15) Data from the long-form Census (20% of population) was used
to adjust the Metro Vancouver trip data to the province as a whole,
using the ratio of proportions of employed people travelling to work by
each mode at the Metro Vancouver and BC levels. (16)
Annual crude fatality and police-reported injury rates were
calculated by dividing the number of injury events, averaged over the
three years, by one of the three denominators: BC population;
person-trips; and km travelled. The latter two exposure-based
denominators were calculated as follows:
Person-trips, mode 1 =
[%[A.sub.1,V]x[T.sub.V]x[P.sub.BC]x%[W.sub.1,BC]]/%[W.sub.1,V] [1]
where
%[A.sub.1,V] = % of all Metro Vancouver trips by this mode;
[T.sub.V] = average number of trips of all types per person per
year in Metro Vancouver;
[P.sub.BC] = population of BC;
%[W.sub.1,BC] = % of BC working population usually using this mode
to travel to work;
%[W.sub.1,V] = % of Metro Vancouver working population usually
using this mode to travel to work.
Distance travelled, mode 1 =
[%[A.sub.1,V]x[T.sub.V]x[D.sub.1,V]x[P.sub.BC]x%[W.sub.1,BC]]/%[W.sub.1V] [2] where [D.sub.1,V] = average trip distance by this mode in Metro
Vancouver.
RESULTS
Table 1 presents the numerator data used for the rate calculations:
the annual average numbers of fatalities and police-reported injuries of
motor vehicle drivers and passengers, pedestrians, motorcyclists and
passengers, and bicyclists in British Columbia in the period 2005 to
2007. (11-13) Motor vehicle drivers and passengers had the highest
numbers of injuries, followed by pedestrians, motorcyclists and
passengers, then bicyclists.
Table 2 presents the denominator data used for the rate
calculations: the estimated annual number of person-trips and the
estimated annual distance travelled by motor vehicle drivers and
passengers, pedestrians, transit users, and bicyclists in British
Columbia. (14,16) Motor vehicle drivers and passengers had the highest
annual number of trips and distance travelled, followed by pedestrians
or transit users (depending on the measure), then bicyclists.
Table 3 presents the crude fatality and injury rates per 100,000
population, per 100 million person-trips and per 100 million km
travelled by travel mode. Using the population denominator suggests that
driving was the least safe travel mode, more dangerous even than
motorcycle travel. However, rates using the population denominator
reflect burden, not risk. Using the exposure-based denominators provides
risk estimates. Based on either person-trips or distance travelled,
motor vehicle occupants had the lowest fatality rates of the three
travel modes for which these denominators were available. Bicyclists and
pedestrians had similar fatality rates based on the person-trip
denominator, but bicyclists had a lower rate using the distance
denominator. For injuries, the picture is somewhat different:
pedestrians had the lowest rate and bicyclists the highest using the
person-trip denominator, whereas using the distance denominator, motor
vehicle occupants had the lowest rate and the other two modes had
similar rates.
DISCUSSION
This analysis allows comparisons of risk between modes of travel in
British Columbia. It makes clear that the reason motor vehicle occupants
have the highest numbers of injuries, followed by pedestrians, then
bicyclists is that the numbers of person-trips and the distances
travelled by these three modes follow that order. Once exposure-based
denominators are taken into account, the injury rates by mode of travel
are not ordered the same way for fatalities and injuries, nor for the
two different denominators.
The relative merits of the two exposure-based denominators can be
debated. If a trip is a prescribed distance (e.g., a trip to work or
school), then the distance denominator is likely best for comparing
risks. But if a trip destination is selected differently based on the
mode of travel (e.g., where to buy groceries), the trip denominator may
be more appropriate. Some commentators suggest that time spent using
each mode would be the most appropriate denominator. (9) Such data are
rarely collected, but rates based on a trip denominator would provide an
inter-modal comparison similar to a time denominator, since trips by
each mode are more comparable in time than in distance. A recent study
in Belgium provided rates in terms of years of life lost (YLL) and
disability adjusted life years (DALY) per distance travelled by each
mode. (10) It takes into account both age distribution and injury type.
Questions that might arise from our analysis include whether the
differences between modes are large or small, and how the rates in BC
compare to other jurisdictions. Since road infrastructure, vehicle
models, and mode shares are similar between the United States and
Canada, comparisons of our results to the US are a useful starting
point. A team at the Centers for Disease Control and Prevention
calculated injury rates for various modes of travel in the United States
over the period 1999-2003 inclusive, using a person-trip denominator.
(8) Figure 2 compares the BC and US rates (note the log scale). It
includes US data for two additional modes of travel: bus and motorcycle
(data were not sufficient to calculate rates for these modes in BC).
This comparison shows that injury and fatality rates for drivers and
passengers, pedestrians, and bicyclists were very similar in BC and the
US. Differences include a somewhat lower fatality rate for BC bicyclists
and a somewhat higher injury rate for BC pedestrians. Striking features
of the US analysis are that bus travel had a much lower fatality rate
than any other mode (over 20 times lower), and that motorcycle travel
had much higher fatality and injury rates than any other mode (over 25
and 7 times higher, respectively). These broader comparisons help to
situate fatality and injury rates for drivers and passengers,
pedestrians, and bicyclists as intermediate between a much safer and a
much more dangerous mode of travel. Such data are helpful for
individuals deciding between modes, and for public policy-makers trying
to promote safe and active travel.
The data in Table 3 can be used to calculate the numbers of trips
or distances travelled for each fatality or injury. For example, in BC
from 2005 to 2007, one car driver or passenger died per 10,416,667
person-trips by that mode, 1 pedestrian died per 6,802,721 person-trips
and 1 bicyclist died per 7,246,377 person-trips. Using the US data for
the other two travel modes, 1 bus passenger died per 250,000,000
person-trips and 1 motorcyclist died per 186,220 person-trips. (8)
Broader international comparisons are more difficult because few
countries have calculated rates by mode.6 Reports from the United
States, Belgium, and the Netherlands, like ours, found that injury rates
for motor vehicle occupants are usually lower than those for bicyclists
and pedestrians, supporting the designation of the latter as
"vulnerable road users". (6,8,10) The data in Figure 1,
comparing crude fatality rates per 100,000 population, suggest that
Canada has an opportunity to reduce traffic deaths by examining
strategies used in safer jurisdictions like the Netherlands.7 Data from
2004 to 2008 in the Netherlands indicate that there were 0.3 fatalities
per 100 million km among motor vehicle occupants (vs. 0.97 in BC) and
1.1 fatalities per 100 million km among bicyclists (vs. 2.6 in BC). (6)
Achieving these lower rates would result in dramatic reductions in
traffic crash deaths in BC (using average numbers from Table 1, a
reduction of over 200 deaths per year).
To our knowledge, this is the first study to calculate
exposure-based injury rates by travel mode in a Canadian jurisdiction.
The analysis had a number of limitations deriving from the availability
of data. Analyses could not be done for the country as a whole because
of the lack of trip diary data. Differences between the Census and
TransLink Trip Diary data for Metro Vancouver show that Census
travel-to-work data overestimates transit trips (16.5% vs. 11.5%,
respectively) and underestimates walking trips (6.3% vs. 10.5%,
respectively). (14,16) This may be because non-work trips are likely to
be closer to home so walking is used instead of transit. Without full
trip data, the risk of pedestrian trips would be overestimated. Here we
used Metro Vancouver data to calculate denominators, using adjustments
based on Census data to allow extension to the province as a whole. The
Metro Vancouver area includes over half the population of the province,
but the extrapolation has unknown errors given potential differences in
numbers and lengths of trips, and availability of sidewalks, bike
routes, and transit between the Metro area and the province. Despite the
problems with the data, we were encouraged by the similarity of the BC
and US rates based on person-trips. (8)
Some data problems were not possible to overcome. Both pedestrians
and bicyclists may have traffic crashes that do not involve motor
vehicles and these are less likely to be recorded unless a fatality is
involved. (6,7) We were unable to calculate rates for motorcycle travel
because denominator data were not available. We were unable to calculate
rates for bus or transit travel because numerator data were not
available. Finally, the data available to us were published summary
data. Denominators were not available by age or sex, preventing
adjustment by or calculation of specific rates for these variables.
Similarly, confidence intervals around the injury rates could not be
calculated because raw data from the TransLink Travel Diary were not
available.14 The annual injury data were at the population level, and
the proportions of trips by each mode of travel were calculated based on
large samples (~200,000 employed persons and 17,603 households), so
variances should be low. (11-15)
It is important to note that this analysis examines injury risks of
transportation modes but does not consider other health outcomes
associated with travel. Studies have consistently shown that active
modes of transportation such as walking and bicycling have important
health benefits (e.g., reduced ischaemic heart disease, cerebrovascular
disease, depression, dementia, and diabetes) that greatly outweigh
injury risks. (17)
In summary, this analysis shows that in British Columbia, motor
vehicle occupants have lower traffic-crash fatality rates than
pedestrians and bicyclists, two groups often designated as vulnerable
road users. Differences between pedestrians and bicyclists depended on
whether person-trips or distance travelled was used as a denominator and
whether fatalities or injuries were considered. Bicyclists had a much
lower fatality rate using the distance denominator and pedestrians had
the lowest injury rate using the person-trip denominator. International
comparisons suggest that bus travel is much safer and motorcycle travel
much more dangerous than driving, walking and bicycling. (8,10) In
addition, such comparisons suggest there are opportunities for dramatic
reductions in fatalities from traffic crashes in British Columbia. Given
the importance of traffic-related injuries to public health and the
potential for large reductions in such injuries, it would be valuable to
institute a national trip diary survey to allow injury rate calculations
on a countrywide basis for all modes of travel.
Acknowledgement: We thank Arno Schortinghuis and Ron Van Der Eerden
for inspiring this work.
Conflict of Interest: None to declare.
Received: September 6, 2012 Accepted: November 23, 2012
REFERENCES
(1.) Statistics Canada. Leading causes of death, by sex, 2008.
Available at: http://www.statcan.gc.ca/
tables-tableaux/sum-som/l01/cst01/hlth36aeng.htm (Accessed July 11,
2012).
(2.) Canadian Institute of Health Information. Trauma and Injuries.
Available at: http://www.cihi.ca/CIHI-ext-
portal/internet/EN/TabbedContent/types+of+care/specialized+services/trauma+and+injuries/cihi010639 (Accessed July 11, 2012).
(3.) Statistics Canada. Motor Vehicle Accidents Causing Death, by
Sex and Age Group. Available at:
http://www.statcan.gc.ca/tables-tableaux/sumsom/l01/cst01/health112a-eng.htm (Accessed July 11, 2012).
(4.) Transport Canada. Canadian Motor Vehicle Traffic Collision
Statistics: 2008. TP 3322, 2010. Available at:
http://www.tc.gc.ca/eng/roadsafety/tp-tp33222008-1144.htm (Accessed July
11, 2012).
(5.) Public Health Agency of Canada. Leading Causes of Injury
Hospitalizations in Canada, 2005/2006. Available at:
http://dsol-smed.phac-aspc.gc.ca/dsolsmed/is-sb/c_hosp_matrix-eng.php
(Accessed July 11, 2012).
(6.) International Traffic Safety Data and Analysis Group. IRTAD
Road Safety 2010 Annual Report. OECD/ITF. 2011.
(7.) World Health Organization. Global Status Report on Road
Safety: Time for Action. Switzerland, Geneva: WHO, 2009.
(8.) Beck LF, Dellinger AM, O'Neil ME. Motor vehicle crash
injury rates by mode of travel, United States: Using exposure-based
methods to quantify differences. Am J Epidemiol 2007;166:212-18.
(9.) McAndrews C. Traffic risks by travel mode in the metropolitan
regions of Stockholm and San Francisco: A comparison of safety
indicators. Inj Prev 2011;17:204-7.
(10.) Dhondt S, Macharis C, Terryn N, Van Malderen F, Putman K.
Health burden of road traffic accidents, an analysis of clinical data on
disability and mortality exposure rates in Flanders and Brussels. Accid
Anal Prev 2013;50:659-66.
(11.) Motor Vehicle Branch. Traffic Collision Statistics.
Police-attended Injury and Fatality Collisions, British Columbia, 2005.
Available at: http://www.icbc.com/road-safety/safety-research/collisions-2005.pdf (Accessed December 18, 2011).
(12.) Motor Vehicle Branch. Traffic Collision Statistics.
Police-attended Injury and Fatality Collisions, British Columbia, 2006.
Available at: http://www.icbc.com/road-safety/safety-research/collisions-2006.pdf (Accessed December 18, 2011).
(13.) Motor Vehicle Branch. Traffic Collision Statistics.
Police-attended Injury and Fatality Collisions, British Columbia, 2007.
Available at: http://www.icbc.com/road-safety/safety-research/traffic-coll-stats-2007.pdf (Accessed December 18, 2011).
(14.) Mustel Group Market Research. TransLink's 2008 Regional
Trip Diary Survey: Final Report. Vancouver, BC: Mustel Group, 2010.
(15.) Statistics Canada. Population and Dwelling Counts, for
Canada, Provinces and Territories, 2006 and 2001 censuses--100% data.
Available at: http://www12.statcan.ca/census-recensement/2006/dp-pd/hlt/97550/Index.cfm?TPL=P1C&Page=RETR&LANG=Eng&T=101 (Accessed
July 11, 2012).
(16.) Statistics Canada. Commuting Patterns and Places of Work of
Canadians, 2006 Census. Catalogue no. 97-561-X, 2008. Available at:
http://www12.statcan.ca/census-recensement/2006/as-sa/97561/tables-tableaux-notes-eng.cfm (Accessed December 18, 2011).
(17.) Woodcock J, Edwards P, Tonne C, Armstrong BG, Ashiru O,
Banister D, et al. Public health benefits of strategies to reduce
greenhouse-gas emissions: Urban land transport. Lancet 2009;374:1930-43.
Kay Teschke, PhD, [1] M. Anne Harris, PhD, [2] Conor C.O. Reynolds,
PhD, [3] Hui Shen, PhD, [1] Peter A. Cripton, PhD, [4] Meghan Winters,
PhD [5] Author Affiliations
[1.] School of Population and Public Health, University of British
Columbia, Vancouver, BC
[2.] School of Occupational and Public Health, Ryerson University,
Toronto, ON
[3.] Liu Institute for Global Issues, University of British
Columbia, Vancouver, BC
[4.] Mechanical Engineering, University of British Columbia,
Vancouver, BC
[5.] Faculty of Health Sciences, Simon Fraser University, Burnaby,
BC
Correspondence: Kay Teschke, School of Population and Public
Health, 2206 East Mall, University of British Columbia, Vancouver, BC
V6T 1Z3, Tel: 604-822-2041, E-mail:
[email protected]
Table 1. Annual Numbers of Fatalities and Police-reported
Injuries (means and standard errors (SE)) in Traffic
Crashes in British Columbia Over the Period 2005
to 2007 Inclusive, by Road User Class * (8-10)
Fatalities Injuries
Mean (SE) Mean (SE)
Drivers and passengers 300.7 (16.2) 22,274 (624)
Pedestrians 70.7 (1.8) 1880 (18)
Motorcyclists and passengers 46.0 (1.5) 1061 (17)
Bicyclists 9.7 (1.5) 982 (46)
* Data not available for transit riders.
Table 2. Estimated Percent of Trips, Annual Numbers of Trips,
Average Trip Distances and Annual Distances Travelled, by Road User
Class *, Data From 2008 Trip Diary Survey11 and 2006 Census (13)
Percent of Annual Number
All Trips of Trips
Drivers and 78.6 3,125,479,000
passengers
Pedestrians 12.1 479,347,000
Transit riders 7.87 312,948,369
Bicyclists 1.76 70,214,000
Average Trip Annual Distance
Distance (km) Travelled (km)
Drivers and 10.0 31,107,465,000
passengers
Pedestrians 2.0 958,694,000
Transit riders 12.0 3,755,380,434
Bicyclists 5.3 372,132,000
* Separate data not available for motorcyclists and passengers.
Table 3. Estimated Crude Traffic Crash Fatality and Injury
Rates in British Columbia, by Road User Class *, With Population,
Person-trip and Distance Travelled Denominators
Exposure-based
Fatality and
Injury Rates
Annual Fatalities Fatalities per
per 100,000 100 Million
Population ([dagger]) Person-trips
Drivers and 7.31 9.6
passengers
Pedestrians 1.72 14.7
Motorcyclists 1.12 --
and passengers
Bicyclists 0.24 13.8
Exposure-based Fatality and Injury Rates
Injuries per
Fatalities per 100 Million Injuries per
100 Million km Person-trips 100 Million km
Drivers and 0.97 713 72
passengers
Pedestrians 7.37 392 196
Motorcyclists -- -- --
and passengers
Bicyclists 2.60 1,398 264
* Numerator data not available for transit riders, so no
rates could be calculated.
([dagger]) Population of British Columbia, 2006 Census = 4,113,487.12
- Denominator data not available for exposure-based rate calculations.
Figure 1. Crude road traffic fatality rates per 100,000
population, all transportation modes, selected
countries (7)
Crude Fatality Rate per 100,000 Population
Netherlands 4.8
Switzerland 4.9
Norway 5.0
Sweden 5.2
United Kingdom 5.4
Germany 6.0
Finland 7.2
France 7.5
Australia 7.8
Portugal 8.0
Austria 8.3
Ireland 8.5
Canada 8.8
Spain 9.3
Italy 9.6
New Zealand 10.1
Belgium 10.2
United States 13.9
Note: Table made from bar graph.
Figure 2. Fatality and injury rates per 100 million person-trips
by road user class, British Columbia and the United
States (14)
Fatality and injury rates per 100 million person-trips
[Note log scale]
BC US BC US
fatalities fatalities injuries injuries
Drivers and
passengers 9.6 9.2 713 803
Pedestrians 14.7 13.7 392 216
Bicyclists 13.8 21.0 1398 1461
Bus passengers 0.4 161
Motorcyclists 537 10337
Note: Table made from bar graph.
