nilssonmodellingeffectinjuriesfatalities2017.md 71 KB
category: literaturenote citekey: nilssonmodellingeffectinjuriesfatalities2017 title: Modelling the effect on injuries and fatalities when changing mode of transport from car to bicycle authors: "Nilsson, Philip; Stigson, Helena; Ohlin, Maria; Strandroth, Johan" year: 2017 date: 2017-00-00 2017 doi: 10.1016/j.aap.2016.12.020 publication: "Accident Analysis & Prevention" url: "https://doi.org/10.1016/j.aap.2016.12.020" zotero_key: BTZVQMUT zotero_storage: E3UX9UVG collections: transpordi_planeerimine folder: 001_artiklid/Road Safety Indicators firstAuthor: "Nilsson, Philip"
status: converted
Contents lists available at ScienceDirect
Accident Analysis and Prevention
jour nal home page: www.elsevier.com/locate/aap
Modelling the effect on injuries and fatalities when changing mode of transport from car to bicycle
Philip Nilssona,∗, Helena Stigsona,b, Maria Ohlinb,c, Johan Strandrothb,d
- a Folksam Reasearch, Bohusgatan 14, 10660 Stockholm, Sweden
- b Department of Applied Mechanics, Chalmers University of Technology, Hörsalsvägen 7a, 96 Gothenburg, Sweden
- c Department of Food and Nutrition and Sport Science, University of Gothenburg, Läroverksgatan 5, 40530 Gothenburg, Sweden
- d Swedish Transport Administration, Röda vägen 1, 781 70 Borlänge, Sweden
a r t i c l e i n f o
Article history: Received 7 September 2016 Received in revised form 23 December 2016 Accepted 29 December 2016 Available online 10 January 2017
Keywords: Bicycle injuries Car injuries DALY Modelling Transport mode change
a b s t r a c t
Background: Several studies have estimated the health effects of active commuting, where a transport mode shift from car to bicycle reduces risk of mortality and morbidity. Previous studies mainly assess the negative aspects of bicycling by referring to fatalities or police reported injuries. However, most bicycle crashes are not reported by the police and therefore hospital reported data would cover a much higher rate of injuries from bicycle crashes. The aim of the present study was to estimate the effect on injuries and fatalities from traffic crashes when shifting mode of transport from car to bicycle by using hospital reported data.
Methods: This present study models the change in number of injuries and fatalities due to a transport mode change using a given flow change from car to bicycle and currentinjury and fatality risk per distance for bicyclists and car occupants.
Results: show that bicyclists have a much higher injury risk (29 times) and fatality risk (10 times) than car occupants. In a scenario where car occupants in Stockholm living close to their work place shifts transport mode to bicycling, injuries, fatalities and health loss expressed in Disability-Adjusted Life Years (DALY) were estimated to increase. The vast majority of the estimated DALY increase was caused by severe injuries and fatalities and it tends to fluctuate so that the number of severe crashes may exceed the estimation with a large margin.
Conclusion: Although the estimated increase of traffic crashes and DALY, a transport mode shift is seen as a way towards a more sustainable society. Thus, this present study highlights the need of strategic preventive measures in order to minimize the negative impacts from increased bicycling.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
The popularity of bicycling in urban areas has increased and in recent years more attention has been paid by stakeholders to create a safer road environmentfor bicyclists. Especially in big cities, bicycling is an important complementto reduce vehicle congestion and greenhouse gas emissions. Transportation bicycling also provides an opportunity for individuals to incorporate physical activity into daily life. Previous studies have shown that bicycling as a health enhancing physical activity can reduce the risk of mortality and morbidity (Matthews et al., 2007; Oja et al., 2011). A meta-analysis
E-mail addresses: Philip.k.nilsson@gmail.se
(P. Nilsson), Helena.stigson@folksam.se (H. Stigson), Maria.ohlin@gu.se (M. Ohlin), Johan.strandroth@trafikverket.se (J. Strandroth).
identified that bicycling reduces the risk of all-cause mortality by 10 percent (Kelly et al., 2014) and a Danish cohort study identified that bicycling to work reduces the risk of all-cause mortality by 28 percent(Andersenet al., 2000). Several studieshavehighlighted the positive health impacts of bicycling as active commuting, showing benefits in reducing risk for various cardio vascular diseases (Hu et al., 2007; Hamer and Chida, 2008). Furthermore, studies have shown that the health benefits of moving from cars to bicycles outweigh the cost of injury from road chrashes (Lindsay et al., 2011; Rabl and DE Nazelle, 2012).
On the other hand, bicyclists and other vulnerable road users have higher risk of being injured or fatally injured compared to car occupants. The number of bicyclists killed per passenger-kilometer has been reported to be nine times higher than for car occupants (Björketun and Nilson, 2006). According to hospital reported crashes and injuries, bicyclists account for a higher proportion than any other road user in Sweden. In 2014, bicyclists and car occupants
∗ Corresponding author.
represented 45% and 33% of all hospital reported injured, respectively (STA, 2015). The injury consequences of bicycle crashes are primarily correlated to non-fatal injuries (de Hartog et al., 2010) and they statistically represent around 6% of all road fatalities in Sweden (STA, 2015). In total almost 80% of all bicycle crashes in Sweden are single crashes (Rizzi et al., 2013), which is similar to results based on hospital data from other countries (Schepers et al., 2014).
Several studies have tried to estimate the total health effect of active commuting, claiming that the health benefits of bicycling outweigh the risks in terms of increased injuries and fatalities related to bicycle crashes (Holm et al., 2012; de Hartog et al., 2010). However, in previous studies the assessments of the negative aspects of bicycling mainly refers to fatalities or police reported injuries. Since only around 7% of bicycle crashes in Sweden are police reported, the negative impact may not be adequately described using police reported data (Rizzi et al., 2013). Hospital data cover a much larger share of bicyclistinjuries and are therefore more suitable when analysing bicyclist crashes and injuries.
The aim of the present study was to estimate the effect on injuries and fatalities due to traffic crashes when shifting mode of transport from car to bicycle.
2. Method and materials
In this study, the health effect on injuries and fatalities due to a change in transport mode from car to bicycle was modelled by estimating the change in bicycle and car crashes, based on a scenario where car occupants in Stockholm change mode of transport to bicycle given that they can reach their work place within 30 min by bicycle. Given this scenario, a flow change from car to bicycle was estimated to 135 million km per year (111,487 new bicyclists) by Schantz (2015). At a national level, this corresponds to an increase of bicycle traffic volume by 5.9% and a decrease of car traffic volume by 0.2%.
2.1. Modelling approach
Denote the number of bicyclists and car occupants being injured or killed as a random variate of Xi,j, where i ∈ bicycle, car and j ∈ injured, killed . Also, let the health loss for each individual be a random variate of Yi,j. Then,the aggregated health effect on a population, e.g. injured bicyclists, is as a random variable Zi,j defined as
$$Z{i,j} = \sum{k=1}^{X{i,j}} Y{i,j,k} \tag{1}$$
The variable Xi,j was modelled as a random Poisson variable with expected value equal to the injury or fatality rate multiplied with the traffic flow change. The injury and fatality rates for bicyclists and car occupants were given in risk per distance and estimated using data from STRADA (Swedish Traffic Accident Data Acquisition; Swedish Transport Agency, 2016). STRADA contains information on crashes and injuries occurring in the Swedish road transport system as reported by the police, and medical data injuries as reported by emergency care hospitals. Traffic mileage data using the Swedish national travel survey RVU (Transport Analysis, 2014). Mileage, injury and fatality data were collected from year 2014. Suicide was not included in the data. In order to make the data represent active commuting to work, crash data involving bicyclists aged 18–65 years were used. The characteristics of the crashes for this group were similar to crashes occurring on work time hours. Furthermore, the total national traffic mileage from the RVU was used. The risk per distance at a national level was almost identical to the risk per distance in Stockholm. The national risk was preferred over the city-specific risk for two reasons. Firstly, the sample from national data was larger and hence, more stable results was yielded. Secondly, the city-specific risk might be biased since bicyclists that are involved in crashes outside Stockholm may still visit a hospital in Stockholm, resulting in upward biased risks.
The health loss expressed in terms of Disability-Adjusted Life Years (DALY) from a crash was modelled by the random variable Yi,j. The distribution of Yi,j was constructed using empirical data of the health loss from bicycle or car crashes (Appendix A). Injury and fatality data for the empirical distributions were collected from the years 2010–2014. Again, suicide was excluded and only individuals aged 18–65 were included.
DALY is a measure that combines disability and mortality expressed as years of life lost due to premature death (YLL) and years lived with disability (YLD) (Eq. (2)). The YLL is the expected life years lost for a fatally injured individual (Eq. (3)). The life years lost due to an injury are equal to the duration of the injury multiplied with a disability weight relating to the specific injury (Eq. (4)) (Salomon et al., 2013). The health effect associated with a change of transport mode from car to bicycle was modelled as the DALY increase due to an increase of bicyclists and a DALY decrease because of less car occupants (Eq. (5)). Note that injury rate, fatality rate and the characteristics of health loss are based on the present situation. This implies that the model assumes that the distribution of age, gender and other factors that might influence the risk or the severity of crashes are somewhat constant in time.
$$DALY = YLD + YLL \tag{2}$$
$$YLL = E [life \ expectancy] - age$$ (3)
$$YLD = disability \ weigth * duration$$ (4)
$$\Delta DALY = YLD{bicycle} + YLL{bicycle} - (YLD{car} + YLL{car})$$ (5)
The distribution of the random variable Yi,j was constructed using empirical YLD and YLL of bicycle and car crashes. AIS coded injuries from STRADA were translated into the injury types used in the Burden of Disease study (Salomon et al., 2013) by the metric used in Tainio et al. (2014) (Table B1). This transformation had to be done in order to compute the YLD and YLL caused by the crashes. In the case of multi trauma, only the injury with the highest YLD was used. Injuries with a lifelong duration were computed as the difference between the expected life time and the age of the individual. The life expectancy table used was collected from Statistics Sweden (SCB) (Table B2).
A distribution of the DALY change due to a transport mode change was obtained using the Monte Carlo technique and independent random samples of Xi,j and Yi,j were obtained using the inverse method (Appendix A). An estimate and confidence interval (CI) of the DALY change were obtained using the percentiles of its distribution. For each simulation run, 1000 samples of the DALY change were simulated. In order to reduce the influence of random variation, each assessment was averaged over 50 simulation runs. This procedure was repeated to ensure convergence.
3. Results
The number of injured bicyclists and car occupants per million km were estimated to 5.2 and 0.2, respectively. The fatality rates were estimated to 14.3 ∗ 10−3 fatalities per million km driven for bicyclists and 1.4 ∗ 10−3 fatalities per million km driven for car occupants. The injury rate was 29 timeshigher for bicyclists thanfor car occupants and the fatality rate was 10 times higher for bicyclists than for car occupants.
The average YLD of a bicycle and car injury were 0.19 and 0.28 years, respectively. That is, the average life time lost due to an injury was 69 days for bicyclists and 104 days for car occupants.
Fig. 1. Probability distribution of the DALY change, 50.000 simulations.
Fig. C1. Distribution of YLD for bicyclists.
Table 1 Simulation results showing mean values and corresponding 95% CI.
| Bicycle [CI] | Car [CI] | |
|---|---|---|
| Injuries | 695 [644;748] | 23 [15;34] |
| Fatalities | 2 [0;5] | 0 [0;1] |
| YLD | 129 [55;225] | 0.2 [0.1;45] |
| YLL | 56 [0;173] | 0 [0;63] |
| DALY change | 176 [53;324] |
The empiricalinjurydistributions of bicycle andcar injuries showed that the majority of the injured received a low YLD; 60% of the bicyclists and 80% of the car occupants received an YLD of less than one week (Figs. C1 & C2).
Simulation result showed that both injuries and fatalities increased in a scenario where car occupants in Stockholm within 30 min to work shifts transport mode to bicycle (Table 1). Injuries contributed to about two thirds of the total DALY increase while the last third was derived from the increase in fatalities (Table 1). Severe injuries, here defined as injuries having an YLD exceeding one year, represented the great majority of the health loss derived fromthe injuries. Thedistributionofthe DALYchange was shiftedto the right of zero, meaning that the health loss due to traffic injuries is estimated to increase as the transport mode shifted from car to bicycle (Fig. 1). Also, the probability distribution is slightly rightskewed and has more probability mass in the right tail than in the left (Fig. 1).
). The average YLL was 32.8 years for bicyclists and 44.13 years for car occupants, showing that car occupants died at a younger age compared to bicyclists given a fatal crash (Figs. C3 & C4
4. Discussion
This study estimated the health impact on injuries and fatalities from a transport mode shift from car to bicycle. In a scenario where car occupants in Stockholm living close to their work place shift transport mode, injuries, fatalities and health loss expressed in DALY were estimated to increase. Fatalities and severe injuries caused the vast majority of the simulated DALY increase. Also, the DALY distribution was right skewed with a long right tail, indicating that the number of severe crashes may exceed the estimation with a large margin.
The present study highlights the negative impactfrom increased injuries and fatalities when changing mode of personal travel from car to bicycle. Previous studies have shown that the positive health impacts from bicycling outweigh the risks. However, the negative effect on injuries might have been underestimated since mainly police reported crash data (Holm et al., 2012) and fatality data (de
Fig. C2. Distribution of YLD for car occupants.
Fig. C3. Distribution of YLL for bicyclists.
Hartog et al., 2010) have been used. Very few bicycle crashes are known by the police (Rizzi et al., 2013) and thereby it is essential to use hospital data to make realistic estimations. In the current study we were able to use hospital reported data on injuries that are representative for Sweden (in 2014 all but one emergency care hospital in Sweden were included in STRADA) making it possible to more accurately estimate the impact from several types of injuries.
Studieshave showngreat benefits fromhaving anactive lifestyle (Hu et al., 2007, Andersen et al., 2000; Hamer and Chida, 2008). By shifting transport mode from car to bicycle, a large population is estimated to decrease their risk of several diseases related to physicalinactivity, and also to avoid orpostponepremature death(Holm et al., 2012). Instead, this study shows that the health loss due to traffic crashes is derived from crashes with severe outcomes and not from an accumulated effect of minor injuries. Also, a transport mode shift increases the overall injury and fatality risk at the same time as the shift occurs, meaning that the health loss will have an immediate impact rather than lagged until the end of an individual's life time. Thus,thepresent study shows thatthe characteristics of the health loss due to traffic crashes differ from some of the benefits gained by having an active lifestyle. Computationally, DALY treats health losses and health benefits the same. However, the use of DALY as a measure for comparing potential health losses and gains might lessen the focus on severe injuries and fatalities.
Trading mortality risk reduction against an increase in fatalities and injuries or to claim that health benefits in one population outweigh health losses in another is problematic from a Vision Zero perspective, where the only acceptable targetfor fatalities and severe injuries is zero. However, a transport mode shift from car to bicycle is seen as a seen as a way towards a more sustainable society and could thereby be an important contributor in reaching the UN Sustainable Development Goals for 2030 (UN, 2016). Knowing this, the comparison between health loss and health benefits from a transport mode shift might become less relevant. Instead, it becomes even more relevant to start addressing the issues connected to an increase of bicyclists. Based on the finding from this study, it is suggested that there is a need to consider and prevent the negative impacts from increased bicycling.
The number of electric bikes sold in Europe has increased rapidly. The mean age as well as the injury severity of motorassisted pedal cyclists are documented to be slightly higher compared to conventional bicyclists (Otte and Facius, 2016). The proportion of electric bikes in the present study is unknown, but could be assumed to be very low. However, taking into consideration a future scenario with a higher proportion of electric bikes the result further emphasizes the need of preventive measures in order to cope with the consequences from increased bicycling.
Fig. C4. Distribution of YLL for car occupants.
Previous research has shown benefits of combining the effect of different preventive measures. Ohlin et al. (2016) showed that in crashes where bicyclists were struck by cars, helmet use, in combination with lower speed restrictions and well-designed car fronts reduced the risk ofimpairing injuries on bicyclists by approximately 79%. As for now, shortcomings and defects in the road infrastructure cause approximately 60% ofthe singe bicycle crashes in Sweden (Niska et al., 2013). By making strategic infrastructural changes, current crash rates as well as the risk of getting severely injured or killed may be reduced. Furthermore, studies have shown that head injuries in comparison to other injuries have a high risk of result in sever disability (Rizzi et al., 2013) and that bicycle helmets are effective in preventing head injuries including severe head injuries (Olivier and Creighton, 2016; Thompson et al., 1999). Considering that the vast majority of the estimated DALY increase in this present study is derived from severe injuries and fatalities, it could be argued that an effective solution to prevent the negative impacts from increased bicycling would be to increase the helmet use.
This study explores a simulation approach in an attempt to modelhealtheffects connectedto a transportmode change.Amajor advantage of applying the Monte Carlo technique rather than a more simple method is that knowledge about the distribution of the health effect is obtained instead of a point estimate only. Also, the use of empirical distributions of the health loss from bicycle and car crashes is seen as a good way to utilize existing data and exploitation of the properties of the Poisson distribution has been made frequently in the field of traffic safety (Forsman, 2008; Lord et al., 2005; Ossiander and Cummings, 2002). Furthermore, the model of use may be further developed in order to capture possible dependency structures such as a safety by number effect and effects of having a large shift in transport modes. The safety in numbers effect has been observed in previous studies (Jacobsen, 2003) and might possibly reduce the negative consequences of an increase of bicyclists. Marshall and Garrick (2011) have shown that the overall fatality risk for all road users was lower in cities with a high bicycling rate compared to the other cities. Also cars, as opposed to bicycles, often causes injuries to vulnerable road users, indicating that a decrease in the number of cars might lead to a decreased risk of bicyclists getting injured.
Decisions of how to best utilize existing data have been made, which may be of interest for future works to consider. In order to reduce the impact of volatile data, national injury and fatality rates have been used instead of having risk rates for urban areas only. In the present study AIS coded injuries were translated into disability weights developed by Tainio et al. (2014). To fully take the effect on injuries into account in further work, there is a need of a more extended scale including a greater number of injuries with disability weights. By mapping injury data to a larger set of diagnoses, the empirical injury distribution would replicate the true distribution in a better way and thus increase the reliability of the study. Also, only the most severe injury for each individual was taken into account when constructing the injury distributions, leading to an underestimation of the health loss due to traffic crashes. However, if every injury of an individual were to be considered, the life years lost due to being disabled may exceed the actual duration of being disabled yielding unreasonable results. This issue could be solved by accumulating and weighting injuries in multi traumas such that the health loss cannot address infeasible values.
Appendix A.
Empirical Distribution Function
The empirical distribution function is the cumulative distribution function of a sample. Consider a random variable V and let F denote its probability distribution. Then, the empirical distribution function Fˆ is a step function that puts 1⁄n probability mass at each of the data points *v*1, *v*2,. . ., *v*n:
$$\hat{F}(v) = \frac{1}{n} \sum_{i=1}^{n} I(v_i \le v),$$
where I is the indicator function.
The Monte Carlo Technique
The Monte Carlo technique works as follows: assume that we want to compute the probability -V ∈ A for a random variable V and some set A. Using the Monte Carlo approach, independent and identically distributed samples is drawn from V. Then, the frequency of hitting the set A is the Monte Carlo estimate of P (V ∈ A). By simulating the V many times, an estimate of its distribution is obtained.
The inverse method
The inverse method may be used to generate independent samples from a random variable. Consider a random variable V. and let F denote its probability distribution. Assume that U is the standard uniform distribution that can be generated. Then, W = F−1 (U) has the same distribution as V, where F−1 = min v|F (v) ≥ p .
Appendix B.
See Tables B1 and B2
Table B1 Injury distribution for bicyclists and car occupants.
| Primary injury | Bicycle (n = 5837) |
Car (n = 6431) |
Disability Weight |
Duration |
|---|---|---|---|---|
| Amputations, arm (life long) |
0 | 0 | 0.257 | Lifelong |
| Amputations, foot, leg (life long) |
0 | 0 | 0.3 | Lifelong |
| Amputations, thumb, finger (life long) |
0 | 1 | 0.134 | Lifelong |
| Amputations, toe (life long) |
0 | 0 | 0.102 | Lifelong |
| Burns < 20% | 0 | 9 | 0.158 | 0.083 |
| Burns < 20% (life long) |
0 | 1 | 0.001 | Lifelong |
| Burns > 20% | 0 | 0 | 0.441 | 0.279 |
| Burns > 20% (life long) |
0 | 0 | 0.255 | Lifelong |
| Crushing | 0 | 0 | 0.218 | 0.094 |
| Dislocations other dislocation |
47 | 5 | 0.074 | 0.019 |
| Dislocations shoulder, elbow or hip |
132 | 8 | 0.074 | 0.034 |
| Fractured ankle | 69 | 6 | 0.196 | 0.096 |
| Fractured clavicle, scapula or humerus |
527 | 44 | 0.143 | 0.112 |
| Fractured face bones |
296 | 68 | 0.223 | 0.118 |
| Fractured femur | 64 | 21 | 0.372 | 0.14 |
| Fractured femur (life long) |
3 | 9 | 0.272 | Lifelong |
| Fractured foot bones |
94 | 23 | 0.077 | 0.073 |
| Fractured hand bones |
433 | 61 | 0.1 | 0.07 |
| Fractured patella, tibia or fibula |
170 | 31 | 0.271 | 0.09 |
| Fractured pelvis | 44 | 27 | 0.247 | 0.126 |
| Fractured radius or ulna |
746 | 56 | 0.18 | 0.112 |
| Fractured rib or sternum |
196 | 118 | 0.199 | 0.115 |
| Fractured skull | 22 | 11 | 0.431 | 0.107 |
| Fractured skull (life long) |
3 | 0 | 0.361 | Lifelong |
| Fractured vertebral column |
85 | 228 | 0.226 | 0.14 |
| Injured nerves (life long) |
2 | 2 | 0.064 | Lifelong |
| Injured spinal cord (life long) |
7 | 31 | 0.725 | Lifelong |
| Injury to eyes | 2 | 4 | 0.108 | 0.019 |
| Injury to eyes (life long) |
0 | 2 | 0.299 | Lifelong |
| Internal injuries | 24 | 41 | 0.208 | 0.042 |
| Intracranial injuries |
602 | 531 | 0.359 | 0.067 |
| Intracranial injuries (life long) |
62 | 35 | 0.361 | Lifelong |
| Open wound | 1564 | 810 | 0.108 | 0.024 |
| Poisoning | 0 | 0 | 0.909 | 0.008 |
| Sprains | 643 | 4248 | 0.064 | 0.038 |
Table B2 Life expectancy table.
| Age | Male | Female | Age | Male | Female |
|---|---|---|---|---|---|
| 18 | 62.7 | 66.3 | 42 | 39.6 | 42.7 |
| 19 | 61.7 | 65.3 | 43 | 38.6 | 41.8 |
| 20 | 60.7 | 64.3 | 44 | 37.7 | 40.8 |
| 21 | 59.8 | 63.4 | 45 | 36.7 | 39.8 |
| 22 | 58.8 | 62.4 | 46 | 35.8 | 38.9 |
| 23 | 57.8 | 61.4 | 47 | 34.8 | 37.9 |
| 24 | 56.9 | 60.4 | 48 | 33.9 | 36.9 |
| 25 | 55.9 | 59.4 | 49 | 32.9 | 36.0 |
| 26 | 55.0 | 58.4 | 50 | 32.0 | 35.0 |
| 27 | 54.0 | 57.5 | 51 | 31.1 | 34.1 |
| 28 | 53.1 | 56.5 | 52 | 30.1 | 33.1 |
| 29 | 52.1 | 55.5 | 53 | 29.2 | 32.2 |
| 30 | 51.1 | 54.5 | 54 | 28.3 | 31.3 |
| 31 | 50.2 | 53.5 | 55 | 27.4 | 30.3 |
| 32 | 49.2 | 52.5 | 56 | 26.5 | 29.4 |
| 33 | 48.2 | 51.6 | 57 | 25.6 | 28.5 |
| 34 | 47.3 | 50.6 | 58 | 24.8 | 27.6 |
| 35 | 46.3 | 49.6 | 59 | 23.9 | 26.7 |
| 36 | 45.4 | 48.6 | 60 | 23.0 | 25.8 |
| 37 | 44.4 | 47.6 | 61 | 22.2 | 24.9 |
| 38 | 43.4 | 46.7 | 62 | 21.3 | 24.0 |
| 39 | 42.5 | 45.7 | 63 | 20.5 | 23.1 |
| 40 | 41.5 | 44.7 | 64 | 19.7 | 22.3 |
| 41 | 40.5 | 43.7 | 65 | 18.9 | 21.4 |
Appendix C.
Figs. C1 and C2 show the empirical distribution of YLD for bicyclists and car occupants, respectively. In the same manner, Figs. C.3 and C.4 show the empirical distributions of YLL. The bars represent the number of times a specific event, e.g. bicyclist having an YLL of 50, occur in the data set. Using this, the empirical cumulative distributions were constructed.
References
Andersen, L.B., Schnohr, P., Schroll, M., Hein, H.O., 2000. All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch. Int. Med. 160, 1621–1628.
Björketun, U., Nilson, G., 2006. Skaderisker för Motorcyklister [Accident Risks for Motorcyclists] VTI report 11-2006. Swedish National Road and Transport Research Institute (VTI), Linköping, Sweden.
de Hartog, J., Boogaard, H., Nijland, H., Hoek, G., 2010. Do the health benefits of cycling outweigh the risks? Environ. Health Perspect. 118, 1109–1116. Forsman, Å., 2008. Forecasting Models of the Number of Fatalities in Road Traffic. An Evaluation of a Poisson Regression Model with Serially Correlated Residuals. VTI, Linköping, Sweden.
Hamer, M., Chida, Y., 2008. Active commuting and cardiovascular risk: a meta-analytic review. Prev. Med. 46, 9–13.
Holm, A.L., Glumer, C., Diderichsen, F., 2012. Health Impact Assessment of increased cycling to place of work or education in Copenhagen. BMJ Open 2. Hu, G., Tuomilehto, J., Borodulin, K., Jousilahti, P., 2007. The joint associations of occupational, commuting, and leisure-time physical activity, and the Framingham risk score on the 10-year risk of coronary heart disease. Eur. Heart J. 28, 492–498.
Jacobsen, P.L., 2003. Safety in numbers: more walkers and bicyclists, safer walking and bicycling. Inj. Prev. 9 (3), 205–209.
Kelly, P., Kahlmeier, S., Götschi, T., Orsini, N., Richards, J., Roberts, N., Scarborough, P., Foster, C., 2014. Systematic review and meta-analysis of reduction in all-cause mortality from walking and cycling and shape of dose response relationship. Int. J. Behav. Nutr. Phys. Act. 11 (1).
Lindsay, G., Macmillan, A., Woodward, A., 2011. Moving urban trips from cars to bicycles: impact on health and emissions. Aust. N. Z. J. Public Health 35 (1), 54–60.
Lord, D., Washington, S.P., Ivan, J.N., 2005. Poisson, Poisson-gamma and zero-inflated regression models of motor vehicle crashes: balancing statistical fit and theory. Accid. Anal. Prev. 37 (1), 35–46.
Marshall, W.E., Garrick, N.W., 2011. Evidence on why bike-friendly cities are safer for all road users. Environ. Pract. 13 (01), 16–27.
Matthews, C.E., Jurj, A.L., Shu, X.O., Li, H.L., Yang, G., Li, Q., Gao, Y.T., Zheng, W., 2007. Influence of exercise, walking, cycling, and overall nonexercise physical activity on mortality in Chinese women. Am. J. Epidemiol. 165, 1343–1350.
Niska, A., Gustafsson, S., Nyberg, J., ERIKSSON, J., 2013. Cyklisters singelolyckor – Analys av olycks- och skadedata samt djupintervjuer [Single bicycle accidents Analysis of hospital injury data and interviews]. VTI, Linköping, Sweden.
- Ohlin, M., Strandroth, J., Tingvall, C., 2016. The combined effect of vehicle frontal design, speed reduction, autonomous emergency braking and helmet use in reducing real life bicycle injuries. Saf. Sci. 92, 338–344.
- Oja, P., Titze, S., Bauman, A., DE Geus, B., Krenn, P., Reger-Nash, B., Kohlberger, T., 2011. Health benefits of cycling: a systematic review. Scand. J. Med. Sci. Sports 21 (4), 496–509.
- Olivier, J., Creighton, P., 2016. Bicycle injuries and helmet use: a systematic review and meta-analysis. Int. J. Epidemiol., http://dx.doi.org/10.1093/ije/dyw153.
- Ossiander, E.M., Cummings, P., 2002. Freeway speed limits and traffic fatalities in Washington State. Accid. Anal. Prev. 34 (1), 13–18.
- Otte, D., Facius, T., 2016. Situation of Pedelecs and comparison to conventional Bicycles. In: Proceedings of International Cycling Saftey Conference, Bologna, Italy.
- RVU Sweden. Transport Analysis. 2014. Stockholm, Sweden. Available at: http:// www.trafa.se/globalassets/statistik/resvanor/rvu sverige 2015.pdf (Accessed 15 July 2016).
- Rabl, A., DE Nazelle, A., 2012. Benefts of shift from car to active transport. Transp. Policy 19 (1), 121–131.
- Rizzi, M., Stigson, H., Krafft, M., 2013. Cyclist injuries leading to permanent medical impairment in Sweden and the effect of bicycle helmets. In: Int. IRCOBI Conf. on the Biomechanics of Injury, Gothenburg, Sweden.
- STA, 2015. Analysis of Road Safety Trends–Management by Objectives for Road Safety Work Towards the 2020 Interim Targets. STA Publication 2014:129. Swedish Transport Administration, Borlänge, Sweden.
STRADA Swedish Traffic Accident Data Acquisition, 2016. Swedish Transport Agency. Norrköping, Sweden. Available at: https://www.transportstyrelsen.se/ sv/vagtrafik/statistik-och-register/STRADA-informationssystem-for-olyckorskador/ (Accessed 15 July 2016).
Salomon, J.A., Vos, T., Hogan, D.R., Gagnon, M., Naghavi, M., Mokdad, A., Farje, M.R., 2013. Common values in assessing health outcomes from disease and injury: disability weights measurement study for the Global Burden of Disease Study 2010. Lancet 380 (9859), 2129–2143.
Schantz, P., 2015. Om gång och cykling, hälsa och en hållbar utveckling [About walking and bicycling, health and a sustainable development]. The Swedish School of Sport and Health Sciences, Stockholm, Available at: http://www. diva-portal.org/smash/record.
Schepers, P., Agerholm, N., Amoros, E., Benington, R., Bjornskau, T., Dhondt, S., Geus, D.E., Hagemeister, B., Loo, C., Niska, B.P., 2014. An international review of the frequency of single-bicycle crashes (SBCs) and their relation to bicycle modal share. Inj. Prev. 21 (e1), e138–e143, electronic source.
Tainio, M., Olkowicz, D., Teresinski, ´ G., DE Nazelle, A., Nieuwenhuijsen, M.J., 2014. Severity of injuries in different modes of transport, expressed with disability-adjusted life years (DALYs). BMC Public Health 14 (1), 1.
Thompson, D.C., Rivara, F.P., Thompson, R., 1999. Helmets for preventing head and facial injuries in bicyclists. Cochrane Database Syst. Rev. 4, http://dx.doi.org/ 10.1002/14651858.CD001855, CD001855.
UN, 2016. UN Sustainable Development Goals. New York, Available at: http:// www.un.org/sustainabledevelopment/cities/ (Accessed 18 July 2016).