Cycle safety in London – getting worse, not better

Transport for London’s latest Travel in London report is out – the fourth.

In the section on ‘Road Safety’, we are told that

The year 2010 was the target year for both national and more stringent London-specific targets for the reduction of road casualties to be assessed (in absolute terms, comparing total casualties rather than casualty rates) against the average for the period 1994-1998.

One of these ‘stringent’ targets was to reduce the absolute number of London pedal cycle KSIs (killed or seriously injured) by 50%, against that 1994-98 baseline.

This target has not been met, as TfL admit –

the actual reduction achieved was 18 per cent.

Some way short of that 50% reduction. But TfL do have an excuse –

However, the substantial growth in cycling, which had doubled since the early 1990s, implies a much higher reduction in the collision risk per trip.

Fair enough, you might say – more people are cycling, so it’s not surprising that the absolute number of people being killed or seriously injured has not declined by as much as we would hope, and the target has been missed.

But this begs the question – surely some growth in cycling was accounted for when the target was set? If it was not, then the required reduction in the absolute number of people killed or seriously injured while cycling could have been achieved simply by a 50% drop in the number of people using bicycles, compared to the 1994-98 baseline – without any genuine increase in safety. Indeed, when the target is an absolute number, it almost gives a perverse incentive to not encourage more people to cycle. Transport for London should have been aiming for growth in cycling, and for the target they have been set by the Department for Transport – it’s a bit weak to use partial success on one front as an excuse for failure on another. A target is a target – and it has not been met.

In any case, is cycling even getting safer in relative terms, as Transport for London would like to have us believe?

According to the same report, the number of cycling ‘journey stages’ in London increased from 0.51 million in 2009, to 0.54 million in 2010 – an increase of nearly 6% (Table 3.6, p.63).

But if we refer to Table 6.1 on page 146 of the report, we see that cycle KSIs have increased from 433 in 2009, to 467 in 2010. An 8% increase in casualties.

It’s worth noting two other standout features from this data.

  1. KSIs for all other mode users have decreased, in most cases by over 10%, while cycling KSIs, alone, have increased (pedestrian journey stages are up, overall, by 0.9% in 2009-10, while car journey stages fell 1% over the same period. See page 24 of the report, and the comments below).
  2.  ‘Slight casualties’ have also increased markedly, by 5% across London – up by 1250 over 2009-2010 – which to me is suggestive of an increased rate of ‘collisions’. Put this ‘background’ against the KSI pattern and it’s not hard to work out what’s happening.

But the standout feature is the 8% increase in the number of cycle casualties, which actually outstrips the increase in the number of journey stages over the same period. It’s clear that – for 2009-2010, at least – cycling was getting less safe, in both absolute and relative terms.

Given that we have just seen one of the worst years for cycle fatalities in some time, I’m not sure 2010-2011 is going to turn out any better.

Thanks to Jim Gleeson for bringing the report to my attention

This entry was posted in Boris Johnson, Department for Transport, London, Road safety, Targets. Bookmark the permalink.

24 Responses to Cycle safety in London – getting worse, not better

  1. This doesn’t surprise me in the least. They have been putting in bike lanes and bike paths. It’s exactly what the majority of the studies have been saying for the last 25 years – cycling infrastructure is unsafe.

    • ‘cycling infrastructure is unsafe.’

      The country with the best safety record for cyclists in the world has rather a lot of infrastructure, deliberately designed to minimize conflict between cyclists and the main source of hazard – motor vehicles.

      London, on the other hand, has very little, if any, ‘infrastructure’. Most of the deaths and serious injuries occur at places without it.

      Your assertion is bogus.

      • The country with the best safety record for cyclists in the world (I can only assume you mean the Netherlands) also has far lower speed limits and a cycling modal share that makes comparison with the UK useless. The massive numbers of cyclists on the road there, combined with lower traffic speeds, have a huge impact on safety. But that in no way means that the infrastructure makes the Netherlands safe for cyclists. Studies suggest that the Netherlands is safe for cyclists DESPITE its bicycle infrastructure.

        Peer-reviewed safety studies done in the Netherlands are conspicuous by their absence. The one piece of data I can find, comes from the Dutch Ministry of Transport ITE Conference in 1977, and says: “Safety problems of two-way cycle tracks at junctions almost insuperable”.

        Studies have been done in another country that has a huge cycling population – Denmark. The studies found that cycling infrastructure of all kinds reduced safety. See:

        Click to access bicycle%20tracks%20and%20lanes.pdf


        Click to access agerholm_et_al._bicycle_paths.pdf

        The fact that cycling infrastructure poses a greater risk to cyclists should not come as a shock to cycling advocates. It is disingenuous to just label the assertion that cycling infrastructure is unsafe ‘bogus’. Studies have been saying that bike infrastructure reduces safety for decades. If you have studies that back up your claim that the assertion is bogus, let’s see it. Sure, Lusk 2011 is there for you, but is fatally flawed by selection bias, as is Moritz 1997.

      • Both these Danish papers you mention reference the 10% increase in the number of injuries/accidents on roads in Copenhagen, following the introduction of bike paths – comparing ‘before’ and ‘after’ periods.

        This of course coincided with an 18% increase in cycling on those same roads.

        Indeed the author of the paper – Søren Underlien Jensen – makes clear that he is in favour of the construction of bicycle paths. See here, for instance.

        Peer-reviewed safety studies done in the Netherlands are conspicuous by their absence. The one piece of data I can find, comes from the Dutch Ministry of Transport ITE Conference in 1977, and says: “Safety problems of two-way cycle tracks at junctions almost insuperable”.

        You really think the Dutch have not done any research on safety since 1977? Look harder.

      • Look harder? You’re the one who makes claims with no sources. Post links!

    • Stefan Ertmann says:

      As a Copenhagerner I find that argument laughable. look at some of the statistics from the latest bicycle account:


      Seriously injured cyclists per year: in 1996 it was 252 in 2010 it was down to 92

      Cycled kilometeres per serious accident 1.2 million km in 1996 is now up to 4.4 million cycled km in 2010.

      All the while the network of cycle tracks has increased by 17% (294 to 346 km)

      Could it perhaps be that modern bicycle infrastructure design with pre greens, advanced stoplines for bikes and removal parking to improve orientation is working?

      And while we are all at bending statistics to our own advantage this one is interresting. in 2010 there were 3 bike fatalaties in Copenhagen, compared with 467 in London, with roughly 7% of its population, that horrific number should be 32. Uh, and whats the modal share of Bikes compared between those two cities again?

  2. Ian Brett Copper – I hope you mean that the infra they’ve been putting in until now is unsafe (which is mostly true).

    • Most of the studies suggest that moving cyclists out of the main traffic lane tends to place them in more danger, especially at intersections. The quality of the infrastructure does not seem to make much difference. So far, no infrastructure method that has been studied properly has been shown to reduce injuries or fatalities, and in fact the opposite seems to be true.

      I can cite a long list of studies showing this to be the case, from Aultman-Hall in 1998 to the recent study in 2011 by Reid. The studies supposedly showing a benefit to bicycle infrastructure all show significant flaws – the most recent one being the appalling Lusk study that was done in Montreal, which had selection bias that you could (to mix metaphors) drive a truck through.

      • And yet the safest place to cycle in the world – the Netherlands – consistently takes cyclists out of the main traffic lane, particularly on the busiest roads.

      • On the other hand we have the example of London where cyclists are exposed to danger along the whole route (not only intersections) and looking at statistics and personal experence you can clearly see that it’s definitely worse than segregated infra. Plus it’s a nobrainer that intersectons will always be a problem clearly because the segregated path stops being really segregated when it crosses a motor traffic road – another point for segregation it would seem.
        To be honest having seen Dutch success (and Paris, and CPH and Munich and many others) and having seen 30 years of failed VC advocacy in UK I don’t think there’s room for any doubt as to what brings in mass cycling.

  3. Mike says:

    I agree with you that an absolute measure is an odd choice, and that the metrics should focus on relative rates. Obviously, as a cyclist myself I want to see things become safer out on the roads.

    This data looks very similar to that which Jenny Jones was quoting a couple of weeks ago. My stats are a bit rusty, so someone with stronger maths should check that the following conclusions are valid.

    I actually think that both TFL’s and your own interpretations of the data are wrong. If you run a test for statistical significance (t-test or chi-squared, 95% confidence level) then the difference in accident rates between the two years (2009 & 2010) per number of journey stages could well be a chance variation. You simply can’t say with that much confidence if the changes are “real”, or random perturbations.

    Your readers can’t draw any inference at all from any of the other numbers, as the journey stages aren’t supplied for them in your post. It may be that it’s actually getting more dangerous to walk, motor or ride a motorbike (in relative, not absolute terms), but without an indication of the number of journeys per transport mode for each year who’s to say?

    I want cycling to be as safe as possible, but I also want to make sure that the arguments we’re using to chase that goal stand up to scrutiny. Get @steinsky (Joe Dunckley) on the case, he’s a scientist, I think.

    • Mike, you’re absolutely right. The change could be accounted for by natural fluctuation. I’ll do some digging on the validity of any conclusions this evening.

      On your second point, according to TfL, the number of walking stages increased by 0.9% between 2009 and 2010, while journey stages by car fell by 1% (page 24). I’ll add those figures into the body of my post.

  4. Mark Ortiz says:

    The whole idea of evaluating infrastructure simply from outcome statistics is flawed even if you have good numbers.

    Why? Because people compensate behaviorally for danger, in both voluntary and legally compelled ways.

    In the US at least, large trucks have better per-mile safety statistics than ordinary cars, because they are obviously more dangerous, and people compensate for that with better driver training, more stringent licensing, and more caution when driving. Yet one could argue purely from outcome statistics that we’d all be safer driving huge vehicles that are hard to see out of, have poor brakes, roll over easily, and are hinged in the middle. That would be absurd of course, but to see that, you have to look at the dynamics of the vehicles themselves, not just the outcome statistics.

    This doesn’t mean that outcome statistics are totally irrelevant or shouldn’t be examined, but it does mean that there’s more to the picture, and any evaluation of whether a design, of anything, is safe or not needs to be based primarily on how the thing in question functions in use, rather than outcome statistics in a particular application.

    I don’t know about the UK, but in North America, bike and bike-ped paths are not legally roads, and crashes occurring on them do not show up as road safety statistics. They are considered off-road crashes. This means it is very difficult even to get comparable outcome statistics.

    Moreover, one of the biggest problems with separated paths is that they promote collisions at intersections. But these collisions occur in the intersecting street; the path ends at the edge of the street and resumes after the street. Consequently, crashes occurring at the intersection show up as statistics for the street, not the path.

    Finally, there are the complexities and uncertainties of determining distance travelled for bicycles. Numbers for miles or kilometers travelled are at best educated guesses, and statistics “per trip” are even more uncertain and ambiguous.

    Combine these problems in the statistics themselves with the shortcomings of using even very good outcome statistics as a dependable method of safety evaluation, and you have a situation where not much at all can be reliably inferred from outcome statistics.

  5. Ian Brett Cooper needs to stop looking at micro-studies and look hard at overall cycling safety rates for different countries. He will see that these show that provision of dedicated cycle infrastructure makes cycling safer. The micro-studies, if they show something else, must be giving an unreal picture, for one reason or another. The reality is in the overall national safety and cycle mileage statistics.

    The assertion that the Netherlands is safe DESPITE its cycle infrastructure is quite simply loony, as anyone who goes and cycles there will discover.

  6. Mike’s point is correct, you cannot draw any valid conclusions from looking at one year’s changes in this level of casualty statistic. TfL’s conclusion that cycling has become absolutely and relatively safer since the 1994-98 baseline is sound.
    From this year national and London casualty stats are meant to be measured as rates, not absolute numbers, which was the norm for the last century. This is something most transport user groups have been lobbying for over many decades. For cycling stats there is a real problem that no one has reliable usage rates. London is better than the rest of UK, they have reasonable growth data on the Mayors road network but patchy data on other roads. The journey stages data is probably only reliable for the last two years, before that it was based on small samples and even guesswork influenced by the demands of the Mayors’ or TfL’s press office.
    The figures put out by Jenny Jones and referred to in the GLA debate don’t prove a worsening in casualty rates, they just show how TfL and the Mayors have guessed some of the cycle data until very recently.
    So if the absolute casualty numbers have improved by 18% against the base line then the relative casualty rate has improved by somewhere between 30% and 50%, we do not have the data to be more accurate.
    One of the actions for TfL’s Cycle Safety Working Group in 2012 is to get a decent understanding of the statistical problems in the casualty data.

    • Jim says:

      If the journey stages data is reliable for the last two years then doesn’t the argument made in this blog post still stand? More broadly, a number of different data sources (TfL’s journey stages, TfL’s trips, DfT’s distance cycled) all point to similar conclusions – that cycling casualty rates are higher than for other modes and that they have at best flatlined in recent years after falling prior to that. The exception is the TLRN flows data, but that covers only a small number of points on main roads mostly concentrated in inner London, so is the most likely to be skewed. So while more robust data would of course be nice, we already know enough to be very concerned.

  7. Fonant says:

    The Dutch situation, is, I think, of great interest. They have some of the safest roads for vulnerable roads users, consistent design to discourage car use and especially “rat running”, and very high levels of cycling as an ordinary mode of transport. And yet they are STILL building, and re-building, top-quality car-free cycle lanes. They spend quite a lot of money on this, even after some thirty years of investment in cycling, and even with mass cycling a reality. They clearly don’t consider that “soft measures”, or “safety in numbers”, are good enough: the only way to achieve genuine safety (such that 8-year-olds can cycle to and from school on their own) is to remove the danger from motor vehicles as completely as you can.

  8. Pingback: Dahab arraigned, CA cops under no obligation to investigate or report, and Santa left a bunch of links « BikingInLA

  9. Opus the Poet says:

    For Mr. Cooper, Infrastructure for bicycles is not merely the physical installations on the roads (or not on them but separate) but also the laws and enforcement of those laws. In the US (and from what I read the UK) those laws are minimal at best and rarely enforced in such a way as to increase a cyclist’s safety. In the Netherlands those laws place pedestrian and cyclist safety (in that order) above moving motor vehicles quickly and without impediment, and place the onus of safety on the operator of the more dangerous vehicle in a conflict. Where cyclists and pedestrians come in conflict the cyclist bears the burden of yielding to safety, in areas where it’s cyclists and motor vehicles it is the motor vehicle operators who bear the burden of yielding to safety.

    In the US we have the safest roads in the world, and the drivers do their best to test that safety by being among the stupidest in the world because we tend to treat driving as a right rather than a privilege. I’m trying to get that mindset changed, but I’m only one small voice clamoring in the wilderness, and it doesn’t carry very far.

  10. Iwin1961 says:

    I agree that cycling is definitely safer than it was before 2000, the point of Jenny’s article was to show that it was heading in the right direction until around 2008. Since then the figures have not maintained that trend and whether you put it down to statistical variation or not, the Mayor can’t claim that cycling has got safer since he took office. And it has probably become more dangerous.

    Having looked at the revised TfL figures for journey stages , they confirm this pattern.

    This may not fit with what some cycle campaigners want to hear, but the reality has to be faced and a dangerous Mayoral policy (smoothing traffic flow) has to change.

  11. PaulM says:

    Seems to me that the whole argument above neatly illustrates Mark Twain’s quote from Disraeli that “there are three types of lies: lies, damned lies, and statistics”.

    Different contributors to the discussion above have used substantially the same data to support diametrically opposed viewpoints about whether the Netherlands, for example, is safer than somewhere else on the basis of its cycling infrastructure. This all ignores the objective/subjective point that cycling is actually quite safe there and in other places to which it is compared, and the big question of subjective safety.

    On the former, even in the UK, infrastructure or not, cycling is arguably safer than walking beside roads, or tennis, or golf, or lots of other things (depends on statistics, of course). On the latter, that is not at all how people in general perceive things. Just as people worry about flying when the real danger is in the car journey to the airport, which they don’t worry about, people avoid cycling because they think it is dangerous. Anything which makes them change their mind is worth exploring.

    Anything, that is, apart from creating shit like the smurf lanes around London which we have seen positively lead cyclicts into places of danger. If that is what Ian Brett Cooper is talking about, I can only agree.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.