Facebook Google Menu Linkedin lock Pinterest Search Twitter

Advertisement

Cycling

Jun 24, 2014

Is the risk of getting killed while cycling on roads increasing?

Cycling on roads is riskier than travelling in a vehicle, but claims that cyclists are at increasing risk of being fatally injured are unconvincing. It seems more likely cycling is getting less risky

Share

Absolute number of cycling fatalities on roads, Australia (source data: BITRE Australian Road Deaths Database)

The Guardian ran an article yesterday warning that cycling accidents are rising in Australia (1). The writer, Nick Evershed, notes there’s been “a spate of accidents involving cyclists in the last couple of weeks” and says it’s worth taking a look at the safety statistics:

Are things getting better or worse? In the past year and a half cyclists have increased their share in the road toll, due to a proportional increase in deaths in 2013 and 2014…In 2014 there were 26 deaths between January and May. There were 17 in the same period last year. So the trend points to a further increase in annual deaths.

He draws his data on cycling fatalities from the Bureau of Infrastructure, Transport and Regional Development’s (BITRE) Australian road deaths database. I’ve used the same source to create the exhibit, but I’m not persuaded it makes a convincing case that the risk of dying is rising.

It’s evident from the exhibit that the long term trend over the 24 year period from 1989 to 2013 was a significant decline in the absolute number of cycling deaths.

There was a dramatic drop in fatalities in the early years (1989 to 1992), coinciding with the introduction of the mandatory helmet law. Whether that was primarily because the law reduced head injuries, or reduced the number of cyclists, is contested territory (and isn’t germane to this discussion).

Even when those first four years are put to one side, the trend over the period starting after introduction of the helmet law (i.e. from 1992 to 2013) was also downward.

It’s true there was a substantial jump in 2013 in the number of fatalities but it’s a big call in the absence of corroborating evidence to say one year’s figure signals the start of a sustained upward trend.

The numbers move around from year to year. In fact just the year before (2012), the number of on-road cycling fatalities was the second lowest since 1989.

It’s also true the number of fatalities is running at a high rate this year. As Mr Evershed notes, there’ve been 26 deaths to the end of May 2014, compared to 17 over the same period in 2013. (2)

However care is needed in assuming there’s a trend based on such a short period. There were almost as many deaths (24) by the end of May in 2010, but the death toll at the end of that year was 38. That was fewer deaths than the annual average for the entire 1992-2013 period and much the same as the annual average since 2000. (3)

In any event, it needs to be remembered that the BITRE data shows the absolute number of deaths; it’s not adjusted for changes in the number of cyclists on the road. While Mr Evershed is right to say we don’t have good data on exposure, anecdotal evidence suggests the number of cyclists has started to accelerate in line with improvements in on-road infrastructure.

That surmise is consistent with Census data showing the number of commuters travelling to work in capital cities by bicycle increased by a whopping 37% between 2006 and 2011 (compared to 67% over 1991-2011). The BITRE data for the corresponding periods shows the number of cycling fatalities fell -17% between 2006 and 2011 (and by -41% over 1991-2011). (4)

It’s also worth noting that Olivier et al found the per capita rate of serious arm and head injuries from cycling crashes in NSW started to decline from around 2006 and continued until at least 2010 (the last year they looked at). They attribute the change largely to improvements in cycling infrastructure.

It seems to me the risk of dying while cycling on public roads in Australia is very likely improving in ‘real’ terms. I agree with Mr Evershed, though, that cycling is still a much riskier way of travelling on roads than driving.

As I noted this time last year (see Is cycling more dangerous than driving?), Garrard et al estimated the relative risk of being killed while cycling on Sydney’s roads over 2002-05 was around 11-19 times higher than it was in a car. It’s probably improved since then, but there’s still a pressing need for better infrastructure and reform of road law to support cycling.

___________________________________

  1. Perhaps The Guardian’s sub editors are too busy to check, but the title of the article, Cycling accidents rising in Australia, is a little misleading. The article doesn’t discuss the trend in injuries; it only considers cycling fatalities, drawing on BITRE’s Australian Road Deaths Database.
  2. If this rate were sustained for the remainder of the year, 2014 would finish with around 60 cyclists killed on the roads. Fatalities haven’t been experienced on that scale since 1991, when there were 58.
  3. The period matters; in both 2010 and 2014 there were 21 cycling road deaths at the end of April.
  4. Bicycle Industries Australia says annual sales of bicycles exceed one million units. Garrard et al estimated that aggregate kilometres of cycling in Sydney increased from 488,000 per day in 2002 to 630,000 in 2005 (although the year-on-year variations were substantial).
Alan Davies — Editor of The Urbanist

Alan Davies

Editor of The Urbanist

The Urbanist is edited by Dr Alan Davies, a principal of Melbourne-based economic and planning consultancy, Pollard Davies Consulting.

Get a free trial to post comments
More from Alan Davies

Advertisement

We recommend

From around the web

Powered by Taboola

11 comments

Leave a comment

11 thoughts on “Is the risk of getting killed while cycling on roads increasing?

  1. David Penington

    The change from 2012 to 2013 is not statistically significant. This sort of random event has a standard deviation of roughly the square route of the average value, so about 7 for this data, and + or – two standard deviations is 5% probable.
    Are there separate figures for road cycling training on high speed roads (80kmh +) vs transport cycling on lower speed roads? These seem to be two very different exposures – high speed rear end collisions vs car dooring and low speed side-swipes. The high speed roads produce the multiple casualty horror crashes.
    Overall it seems to me there are several modes of cycle use with possibly quite different risk levels.

  2. Strewth

    Linda Ward #9: Research has documented a decline in head injury rates among cyclists (and also pedestrians) in many parts of Australia through the 1980s. For example, there was a “Safe roads for children” inquiry in Victoria in 1987 that provides some useful data. It also found high levels of support for bike helmets of course, but this was at a time when use was voluntary. It was also a time when cycling rates were increasing.

    Similar figures for the 1980s are reproduced in MUARC’s report no.32, “Evaluation of the Bicycle Helmet Wearing Law in Victoria During its First 12 Months”. Again the data show a sustained decline through the 1980s both in cyclist head injuries and in severe cycling injuries as a whole (though not so much in total cycling injuries).

    For the early 1990s there are figures for child cyclist admissions at three major Melbourne trauma hospitals from the Victorian Injury Surveillance System, reported in Robinson: “Head Injuries and Bicycle Helmet Laws” (Accident Analysis and Prevention, 1996). The total number of child cycling injuries (under 15) is reported as 809, 628, 604 and 633 for the years 1990 through 1993 respectively. The total number of head injuries is reported as 88, 60, 58 and 63 for the same years, indicating the number of head injuries as a proportion of total injuries remained constant at around 10 per cent. As Robinson also found, there is evidence for a roughly 30% drop in total child cyclist numbers between 1990 and 1992 which would explain the entire drop in both total injuries and head injuries.

    Of course, you can get different results by looking at injuries of different severity, because the mitigating effect of parallel measures such as drink-driving and speeding enforcement and traffic calming is known to be greater on severe injuries than on mild ones. So a reduction of 60% in the most severe head injuries is not surprising – but again for consistency, one should also look at the reduction in pedestrian head injuries of comparable severity. The MUARC studies from the 1990s didn’t look at this.

    To keep with the thrust of Alan’s article, it would really be of benefit to understand better the factors underlying the reducing trend in severe cycling injuries during the 1980s despite the growth in cycling at that time. It appears researchers at the time sought good explanations for this but could not identify any save for the general improvements in traffic calming and road rule enforcement.

  3. Linda Ward

    Strewth #1

    Re your assertion that “the encouragement of voluntary helmet use was having the desired effect and didn’t need to be reinforced by compulsion”.

    According to http://www.monash.edu.au/miri/research/reports/muarc045.pdf
    – just prior to the Vic law in 1990, the adult wearing rate was 36%, and the teenage rate was 21%
    – after the helmet law, there was “a significant post-law increase in helmet wearing rates in all age-groups’

    On what evidence do you base your assertion that the encouragement of voluntary helmet use was having the desired effect?

    Strewth #4
    Re you assertion/s that “we also know from analysis done in the 1990s that among cyclists, the decline in non-head injuries over this period was as great or greater than the decline in head injuries. Miraculously, we also see declines in pedestrian head and non-head injuries over this period. Unless we want to claim some magical effect of bike helmets in protecting the entire body and that of nearby pedestrians from injury, this suggests other factors were more instrumental in the decline . . . data sources suggest one of the most significant factors really was a marked decline in cycling coinciding with the introduction of mandatory helmets”,

    Injury data in the Williams study (http://www.bicycleinfo.nsw.gov.au/downloads/cycle_research/evaluation_of_nsw_introduction_of_compulsory_bicycle_helmet_legislation.pdf)
    shows that in NSW
    – serious casualties for all road users dropped by 15% between 1990 and 1993
    – in 1991-93, compared to 1988-90, adult cyclist non-head injury hospital admissions decreased by 4%, and adult cyclist head injury admissions dropped by 32%
    – in 1991-93, compared to 1988-90, child cyclist non-head injury hospital admission reduced by 13%, and child cyclist head injury admissions dropped by 37%

    Hospital admission data in the Carr study (http://www.monash.edu.au/miri/research/reports/muarc076.pdf) shows that in Victoria, the number of
    – pedestrian head injury hospital admissions dropped by 20%
    – cyclist non-head injury admissions dropped by 20%
    – serious/severe (AIS3/4) cyclist head/brain injuries dropped by 60%

    The Hendrie study (http://www.ors.wa.gov.au/Documents/Cyclists/ors-cyclists-report-helmets-evaluation.aspx) found that that after accounting for the long(er) term trends, the proportion cyclist head injuries dropped by considerably more than the proportion of pedestrian head injuries (prior to the WA helmet law, the proportion of bicyclists with a head injury was on average 6% higher than the proportion of pedestrians with a head injury, from 1992 onwards the proportion of bicyclists with a head injury was on average 16% less that the proportion of pedestrians with a head injury).

    Injury data in the Marshall study (http://www.bicyclenetwork.com.au/media/vanilla/file/SA%20Helmet%20eval%201994%20SA%20Marshall.pdf) shows that in SA
    – cyclist non-head injuries dropped by 9%, and cyclist head injuries excluding concussion dropped by 41%
    – cyclist non-head injuries did not drop by any more than injuries for other road users
    – there was a 27% reduction in non-cyclist concussion admissions, and a 54% reduction in cyclist concussion admissions

    On what evidence do you base your assertions that cyclist non-head injures declined by at least as much as cyclist head injuries, and that the most significant factor in injury reductions was a marked decline in cycling coinciding with the introduction of mandatory helmets?

  4. Alan Davies

    Nik Dow #7:

    Your chart of cyclist vs pedestrian fatalities is an absolute shocker. At first I was mystified how you could claim the absolute number of ped deaths was of the same order as cyclist deaths given the data shows 5-8 times as many pedestrians died on the roads as cyclists each year over the period when the helmet law was introduced. Then I saw this in the text underneath:

    “Pedestrian deaths have been scaled to sit on top of the cyclist death line for comparison.”

    I don’t think I’ve ever seen a more misleadingly constructed chart. Beats anything on The Checkout. There’s not even the customary second scale used when comparing wildly different orders of magnitude.

    The thing about charts is their purpose is to convey information visually; any “fancy stuff” like this should be clearly indicated on the chart itself. It’s not sufficient to include a disclaimer in the accompanying article (and certainly not one that doesn’t even explain how the scaling was done).

    What the data actually shows is the number of pedestrian fatalities fell -30% nationally from 1989 to 1992, the same as the average of all modes and much the same as the drop for motorcyclists (-34%) and occupants of cars (-27%). However fatalities for cyclists fell by a much larger percentage than any other road user type, i.e. -58%.

    If you instead take 1990 to 1992 as your comparison period, the difference is even more stark. The average change across all road user types was -15%, while that for pedestrians was -17%. The change for cyclists was -49%.

    The much larger drop for cyclists than other road users (esp motorists!) can’t be put down to the introduction of speed cameras. The most plausible hypothesis is the mandatory helmet law (although, as I noted in the article, this data doesn’t indicate whether that’s due to reduced head injuries or a drop in the number of cyclists).

  5. Nik Dow

    Strewth makes an important point about helmet law not explaining the drop in the early ’90s. A fact-based explanation is linked (see “detailed explanation”) http://www.freestylecyclists.org/do-helmet-laws-make-much-difference/
    and covers the introduction of demerit points and ramping up of speed and red-light cameras. The number and severity of collisions involving motor vehicles decreased due to decreases in speed and red-light running by motor vehicles. That is why pedestrians benefited as much as cyclists, but over and over we see reference to decreases in KSI figures being the result of helmet law.
    The greater decrease in non-head injuries probably reflects a change in the mix of cycling activities towards sport cycling (including mountain biking) as transport cyclists were discouraged by helmet law. Faster cycling carries a greater relative risk of head injuries.

  6. Alan Davies

    Joshua Sanders #5:

    My “year-on-year” reference is to the variation in average daily distance cycled from calendar year to calendar year. Agree it’s not clear; will amend to clarify. And you’re right about heavy vehicles, they’re already over-represented. Interestingly, in London, female cyclists are much more at risk of death from heavy vehicles than male riders, according to a report I saw the other day.

  7. Joshua Saunders

    I was a bit thrown by the reference to 488,000 kilometres by bike in Sydney per annum. I myself do 13,000 plus in a year, and the 100 or so people I follow on Strava (a tiny minority of total Sydney cyclists) exceed 488,000 km per annum. Even though it’s 2002 and 2005 when usage was lower than today, it’s a gross underestimate.

    When I looked at the report, it also had motor vehicle use at 121M. Given there are ~4 million people in Sydney – that’s about 30 km each. Then it clicked, looking at the relevant Table – it’s an estimate of total *daily* use.

    Anecdotal evidence suggests things slowly improve, with much more total distance covered by bike in Sydney now than 9 years ago. But even with increased usage, if the present pattern continues, there will be a definite statistical blip in 2014 for Sydney, both total fatalities and IMO fatalities per km ridden. And it’s likely heavy vehicles will be over-represented in the statistics as the “other vehicle” in collisions causing fatalities.

  8. Strewth

    Alan #3: It’s possible, if you look at the figures state-by-state and aggregated six-monthly rather than annually, to discern an effect of the helmet law in Victoria and NSW, but not anywhere else. But we also know from analysis done in the 1990s that among cyclists, the decline in non-head injuries over this period was as great or greater than the decline in head injuries. Miraculously, we also see declines in pedestrian head and non-head injuries over this period. Unless we want to claim some magical effect of bike helmets in protecting the entire body and that of nearby pedestrians from injury, this suggests other factors were more instrumental in the decline.

    This does now have some relevance for the current debate, because if there were other measures (aside from a reduction in cyclist numbers) that brought about such an improvement 25 years ago, that could point to strategies that would be more effective today, if only we could identify them. Unfortunately, other data sources suggest one of the most significant factors really was a marked decline in cycling coinciding with the introduction of mandatory helmets, which won’t be of any use to us today.

  9. Alan Davies

    Strewth #1:

    Nothing glib about my reference to the helmet law. It took effect in the most populous states prior to 1992 i.e. in Victoria in July 1990; NSW Jan 1991 (adults); Tasmania Jan 1991; Qld July 1991; SA July 1991. It took effect from the start of 1992 (1 Jan) in WA and NT. Only the ACT was later i.e. July 1992. I think 1989 and 1992 are the right years to use.

    You may be right that encouragement of voluntary helmet wearing would’ve been enough and compulsion wasn’t necessary, but while it’s an interesting point, it’s not the subject of this article.

    Dylan Nicholson #2:

    The point of noting that the BITRE data records the absolute number of deaths is to emphasise that it takes no account of changes in the level of exposure i.e. the number of cyclists has increased. Hence cycling is getting safer in ‘real’ terms. Agree not enough is being done to make cycling safer.

  10. Dylan Nicholson

    “In any event, it needs to be remembered that the BITRE data shows the absolute number of deaths…”

    Sure, but the absolute number of car fatalities has been dropping steadily virtually every year for decades, even while the number of trips was increasing.

    Personally, I would say that graph shows 6 years of statistics demonstrating that nowhere near enough is being done to make cycling safer.

  11. Strewth

    Alan, agree that the helmet issue is not germane to recent trends, but then why make a glib statement at the outset about the effect of compulsory helmets that’s entirely unsupported by evidence?

    The decline between 1989 and 1992 can’t be the result of compulsory helmets, because the law was only introduced in most states between mid-1991 and 1992. (Victoria led the others but even they only introduced it in mid-1990.) The dramatic decline in fatalities at this time was actually seen across all modes of transport and is generally attributed to a surge in random breath testing and the wider use of mobile speed cameras, adding to a decline in travel with the early-90s recession.

    To the extent bike helmets contributed to the decline in fatalities from the late 1980s it suggests that the encouragement of voluntary helmet use was having the desired effect and didn’t need to be reinforced by compulsion.

Leave a comment