Risk in Cycling (Main Article)

Introduction

This article is based on several papers I co-authored, all published in peer-reviewed journals. More detailed information is provided in linked sub-pages. Links are also provided to the original papers (see “Further Information” at the bottom of this page). Some additional information has been included from other studies.

Risk here is defined as the chance of fatality per hour. Risks of serious injury, slight injury, or any crash, are not examined in depth. This is because non-fatal injuries are not recorded consistently enough to provide reliable comparisons. Risk is expressed in units that will appear esoteric to the newcomer. If you would like to know more about these units,  then please read here. For more details on sources of data, please read here.

The bulk of the article concerns England only, due to data availability and the need for a large base population when studying risk in cycling. However, some comment is made on the situation in Scotland.

Considerations of Risk

The risk in cycling can be examined according to age and sex, the road class, the year, or the country. Comparisons can be made with the risks of walking or driving. The UK has a particularly detailed record of personal transport habits on which good estimates of these risks may be based. A few other countries have similar quality data. This means some international comparisons can be made.

The high-level results for risk per hour for different age groups, males and females, walking, cycling or driving, are presented in the charts below. Note that these are based on the three years 2010-12.  Risks in travel change slowly over time. The trends and relative measures of risk in these results will not have significantly changed in the years since.

Males w c d F - MHU

Chart 1: Males, England, Risk in Walking, Cycling and Driving

Females w c d F - MHU

Chart 2: Females, England, Risk in Walking, Cycling and Driving

If the units of F/MHU are strange to you, read more here. As a rule of thumb, 0.1 F/MHU is “pretty safe”, while 1 F/MHU is still not an especially high risk from the individual perspective. However, even 0.1 F/MHU is too high a risk for transport planners to tolerate. Transport planners have to consider millions of people exposed to risk every day, so they must have a different viewpoint from the individual.

Charts 1 and 2 show that age has the greatest effect on risk in personal travel. Cyclists are safest when young, risk rises with age, the increase being especially steep after the age of about 70. A similar pattern is seen for pedestrians. This is in contrast to driver risk, which is highest in youth and falls with age, except for the elderly.

Male cyclists face higher risks than female cyclists. Note that there are comparatively few female cyclist deaths in England, so the error bars are large for most age groups (extending off the tops of the charts in a couple of cases).

Road class also has a significant effect. Cycling on quiet urban or rural roads incurs risks as low as typically experienced in cycle-friendly countries like the Netherlands or Denmark. Higher traffic speeds incur higher risks. Rural A-roads are the riskiest places to cycle.

DfT Risk by Road Class 2011 2012

Chart 3: Cycling risk versus road class, Great Britain 2011-12.

In this chart the risk is expressed in fatalities per billion KM, because that is the form of the original data. Roughly speaking, the risk of A-road cycling per hour would be 3 F/MHU.  It is worth noting that this result fluctuates quite a bit, for instance in 2016 the risk had fallen to 136 F/BnKM, although this is still a high risk for cycling. In many areas, especially mountainous ones, cyclists are forced to use A-roads due to lack of any alternatives.

All modes of travel have become safer over time. In the chart below I plot driver versus cyclist risk since 1965. Reliable data for walking are only available since the late 1990’s, so that mode is not included. The series are indexed from values in fatalities per billion KM, rather than F/MHU, in line with the original data.

Driver versus Cyclists Comparison

Chart 4: British Driver and Cyclist Risk Indexed to 1965

Note that the data series have been indexed so that both start from the same point in 1965. This does not mean the risks were equal in 1965. The purpose of indexing is to make the trends of both lines comparable; to show that cycling as well as driving have become safer during the last half-century. In 1965, the fatality rate for drivers was 11 F/BnKM and for cyclists 78 F/BnKM. At the end of the period, these risks are respectively about 1.5 F/BnKM and 22 F/BnKM.

The fall for drivers has been larger than for cyclists, although not dramatically so. Perceived risk in cycling has not fallen over time, it has increased. I discuss this paradoxical observation more on this page.

Cycling Risk in Different Countries

Researchers at the University of Zurich recently carried out a survey of the average risks of cycling in different countries. European countries generally have reliable figures for traffic deaths, but information on the amount of cycling by the population is much less consistent. Concerning those countries for which high quality information were available, the results were as follows:

Cyclist risk by Country

Chart 5: Average Risk in Cycling by Country 2011-2015 averages.

These figures should be treated with an element of caution. Although these are drawn from the results deemed most reliable by the survey, methods of measuring cycle use no doubt vary from country to country. In addition, the type of people involved in cycling varies from country to country. As widely noted, cycling in Britain is dominated by male enthusiasts, rather than a typical cross section of the population.

The situation in a given country will also vary dramatically. From personal experience, I can say that cycling around Brussels was downright hair-raising (or it was ten years ago). Whereas, cycling around western Belgium was quite civilized. Similar comments apply in Britain: try mountain trunk road versus lowland lanes.

The full presentation to the International Transport Forum is available here.

Risks Associated with Vehicle Use (Total Risk)

The total risks of operating a vehicle type has been studied for England, covering the period 2005 to 2013. A neutral measure is taken from fatal crash involvements of a given vehicle type. For instance, during 2011 to 2013, a total of 4,013 road users were killed in crashes involving at least one car. Of this number, 1,327 were car drivers and 2,686 were others (HGV drivers, car passengers, motorcyclists, pedestrians or cyclists). In the same years, a total of 331 road users were killed in crashes involving at least one cyclist. Of this number, 301 were cyclists, and 30 were others (Note: not all of the “others” were killed due to the impact of the bicycle, since some collisions involved more than one vehicle, perhaps driver+cyclist+pedestrian in which the pedestrian was killed by the impact of the car).

After taking into account the greater number of hours driven relative to hours cycled, the total risk for cycling was 1.5 to 2 times higher than that for driving. However, for those under the age of about 30, the total risk was higher for drivers. For the 17-20 y.o. male group specifically, the total risk was 3 times greater for drivers than cyclists.

The charts below show the total risk in F/MHU for males and females, for two time periods: 2005-07 and 2011-13 (confidence intervals have been omitted for clarity):

Males Total Risk By Age

Chart 6: Total Risk for Males by Age (England)

The confidence intervals for older cyclists were large, as suggested by the erratic results between the time periods. The risks had generally fallen in the later time period. The good news is that the fall for young male drivers was especially large, although total risk is still much higher than for cyclists of that age/sex group.

Females Total Risk By Age

Chart 7: Total Risk for Females by Age (England)

Results for females show generally much lower risks overall. In the 2011-13 data for cyclists some points were omitted due to lack of meaningful results. A “problem” in evaluating risk for female cyclists is that few women cycle, especially in older age groups, which means there are few casualties to base the estimate on.

The comparatively high number of others killed in car crashes relative to driver deaths is a measure of the inherent capacity to inflict fatal injury (on anyone) associated with the use of motor cars. Cyclists impose minimal risk on others. It is important to appreciate there is no implication of blame in these figures.

Limitations of Risk Estimates

It must be noted that these studies concerned travel only on public roads. They excluded riding on cycle paths off-highway. This means the results for cycling are pessiminstic in some degree. Other factors leading to over-estimation of cycling risk are: drivers spend more time stuck in jams, which reduces the risk estimate by increasing trip durations; drivers enjoy low-risk conditions on motorways and trunk roads, for which no equivalent is included for cycling; most UK cyclists are male; many UK cyclists are sporting riders not typical of utility cyclists. None of these effects has been allowed for due to lack of information.

Conclusions on Risk in Cycling in England and Scotland

Broadly speaking, risk in cycling follows intuitively sensible trends. The risks are low where motor traffic speeds are low; risks increase as traffic speeds rise. This means that a UK cyclist may face risks as low as walking most of the time, but experience “spikes” of high risk. This may explain why cycling is widely regarded as dangerous, despite average levels of risk that are low for most age groups.

Taking a broad view of the risks of walking, cycling and driving, the effect of age is much greater than differences in risk between the modes of travel.

Risk estimates should include all road users killed in crashes involving a given vehicle type (“Total Risk”). This gives a fuller picture than just risks to drivers or cyclists. Cyclists impose minimal risk on other road users. This is not so for drivers, especially young ones.

Considering national data, the average fatality rate for cyclists in Scotland is higher than that for Great Britain as a whole (dominated by England). I discuss possible reasons for this here.

Low-risk cycling akin to Dutch or Danish experience already exists in the UK in places where the conditions are benign. The challenge is to spread these low-risk conditions over a national network. Anyone getting on a bike should be confident of a journey – even a long one – that is easy to navigate, comfortable and feels safe.

For young people, cycling should be promoted over driving as the preferred choice. This supports road danger reduction measures, quite apart from all the other benefits of cycling.

Further Information

See the following papers for more details of the research on which this page is based:

Mindell J, Lesley L, Wardlaw M. “Exposure-Based, ‘Like-for-Like’ Assessment of Road Safety by Travel Mode Using Routine Health Data” PLOS One 2012. [Link]

Feleke R, Scholes S, Wardlaw M, Mindell J. “Comparative fatality risk for different travel modes by age, sex, and deprivation” Journal of Transport and Health 2017 [Link]

Scholes S, Wardlaw M, Anciaes P, Heydecker B, Mindell J. “Fatality Rates Associated with Driving and Cycling for All Road Users in Great Britain 2005-2013”. Journal of Transport and Health 2018 [Link]