Is cycling dangerous?

It goes without saying that a law compelling protective equipment can only be justified if the activity being regulated is actually dangerous. Is this the case with cycling?

There are several ways to define dangerous, of course. I’ll look at some of them below, and when you’re done you might like to try the amusing “test your safety perceptions” quiz at Bicycling Life.

Risk per hour

Risk per hour is difficult to determine, but as far as we can make out the risk per for cyclists is about the same as it is for pedestrians. Failure Analysis Associates in America went further: they computed that the fatality risk for cycling and being a passenger in a motor vehicle were about the same, and both were lower than the risk for walking alongside a road. But that is in the USA.

It’s a tragic irony that Ken Kifer, one of the most popular sources of information on cycling (e.g. his page on cycling risk), was himself killed by a drunk driver while cycling. This illustrates a fundamental problem in human assessment of risk: separating the specific form the aggregate. You might be entirely confident that your child will be safer if all children walk and cycle to school (and you would be absolutely right) but this does not help if your child is the one who falls off and is injured.

Malcolm Wardlaw has written an informative guide to assessing the real risks of cycling. Malcolm’s also the one who wrote “three lessons for a better cycling future” in the BMJ. De Clarke also has a discussion of cycling risks.

Risk per mile

Risk per mile is also about the same as for pedestrians. The source for this is the DfT’s transport statistics at http://www.dft.gov.uk. There’s a summary at http://www.dft.gov.uk/pgr/statistics/datatablespublications/modal/tsgbchapter1passtrans1861.pdf, but it does not tell the whole story.

According to DfT, then, for 2006 the risk per billion kilometres travelled was 36 fatal, 371 KSI for pedestrians, 31 killed and 527 KSI for cyclists. That is based on an average distance travelled of 39 miles per year by bike and 201 miles walking, for the entire population. If you compare these with the KSI figures for cyclists and pedestrians from STATS19 data – 2442 cyclists and 7051 pedestrians – you can cross-check the numbers by calculating the population. The cyclist figures give a population of 74 million, the pedestrian figures 59 million. The latter is about right, so if you factor the cyclist figure down to match you end up with a risk ratio of 419:371 – which would make cycling just over 10% more dangerous than being a pedestrian.

But there’s another problem here: the STATS19 data on which these figures are based include all serious or fatal injuries where the police are called and which involve a vehicle on the roads. So any cyclist collision – including potentially single-vehicle crashes – where the emergency services attend, will be counted. But a pedestrian injury with no vehicle involved (such as tripping on a paving stone) is not included and in fact even pedestrian injuries involving motor traffic are known to be under-reported (see “Road accident casualties: a comparison of STATS19 data with Hospital Episode Statistics” [PDF]). An accurate comparison is difficult because “bike” is ambiguous in reporting and STATS19 and HES data do not tally on a number of levels. DfT are trying to reduce this but they have not managed yet.

On top of that, the figures for exposure for pedestrians are recognised as being terribly inaccurate, so much so that you probably can’t trust them to be right within a factor of two or three at best.

The inescapable conclusion, as noted by economist Tim Harford in the BBC’s More Or Less broadcast in August 2010, is that cycling is not especially dangerous – unless you consider being a pedestrian to be especially dangerous.

Risk to others

The risk posed by cyclists to others is extremely low. The majority of cyclist miles are in areas where there are pedestrians, but cyclists kill on average under one pedestrian per year – drivers kill around seventy pedestrians on the pavement every year, and over 3,000 people in total. When cyclists are injured in car v. bike collisions the car driver is usually at fault (in between 2/3 and 7/8 of cases depending on your source)

Government figures show only a handful of cases of serious or fatal injury caused to pedestrians by cyclists in a year, and these are massively outweighed by the numbers of pedestrians killed and injured on the footway by motor vehicles. Pedestrians are at vastly more risk from motor vehicles than from cyclists even when they are on the pavement. I have some more thoughts about the politics of this.

Vulnerability

Like pedestrians, cyclists are vulnerable to motor traffic, but unlike pedestrians, cyclists spend a lot of time on the roads in amongst that traffic. Between four and five times as many pedestrians are killed and injured by cars every year as are cyclists, and there are an average of around 40 fatalities and a few hundred serious injuries of pedestrians by motor vehicles ion the footway annually (compared with an average of zero fatalities and a few tens of serious injuries from bikes on footways). Any way you look at it, the thing that’s dangerous is the motor vehicle.

One interesting fact is that cyclists are less likely than motor drivers to be to blame when they are hit, and are the most likely of all types of road users to be the victims of hit-and-run.

Does that make cycling dangerous? Or does it highlight that driving is dangerous and cyclists are vulnerable?

Explore more

Figures are hard to come by, and most of them are based on guesstimates of varying accuracy, but it is quite hard to make a compelling case for cyclists being at higher risk than pedestrians. Why does nobody suggest compulsory walking helmets?

DfT have published a really interesting discussion of cycling issues, you can read it at http://www.dft.gov.uk/162259/162469/221412/221549/227755/328843/pedalcyclistfsheet07.pdf (it may move, it has moved before, if necessary this Google search should find it)

Road accident casualties: a comparison of STATS19 data with Hospital Episode Statistics