“Who kills whom” and the measurement of danger.

In our Charter we give a commitment to: “Find new measures to define the level of danger on our roads. These would more accurately monitor the use of and threat to benign modes.” This post is part of our work at doing that – hopefully it will contribute to debate. It is based on a document by PACTS given to the Transport Committee Active Travel enquiry in December 2018.

In previous posts and discussions, we have spent a lot of time talking about the need to have measures and targets for benign transport modes expressed with a measure of exposure – e.g. casualty rates per distance, or time, or number of trips travelled. Examples are here  and here . In this post we move on to look at the question of: Who Kills/Hurts/Endangers Whom?

(Note: The data is from Reported Road Casualties GB (RRCGB) 2017, Transport Statistics GB (TSGB) 2017 and Domestic Road Freight Stats GB (DRFSGB) 2017)

PACTS tweeted that their submission: ”Includes suggestions for better analysis of road casualty stats which would show harm as well as vulnerability and encourage danger reduction and active travel policies.” The purpose of this post is to build on their analysis to do just that. Here I refer to a key section of their document (I have abridged it – I strongly suggest you read the original).

“…in relation to understanding the risks and benefits of active travel, Reported Road Casualties Great Britain still has significant limitations:

 
• It does not distinguish benign modes from dangerous ones, i.e. those modes sustaining the casualties and those inflicting them.

• It assesses at length, with graphs, rates and commentary, the “vulnerability” of different road users, mainly of active travel modes.

• It does not present the other side of the equation, i.e. danger and casualties suffered by other road users in collisions with motorised modes – the number or rate of third party casualties associated with cars, buses and HGVs. Using the same RRCGB data, PACTS has done a preliminary analysis which shows an alternative picture of the risks by mode. (See graphs below.)

• It does not clearly convey that, for the vast majority of UK road users, they are much more likely to be killed in a car, or by car, than any other mode.

• It compares user fatality rates per distance travelled, concluding that walking and cycling are “high risk” modes whilst cars, buses and HGVs are low risk. This implies the trips are similar and interchangeable, which generally they are not. The average person travels over 5,000 miles by car/van per year. By comparison, the average person walks and cycles fewer than 300 miles per year. Even under the most optimistic CWIS (The Government’s Cycling and Walking Investment Strategy) scenario, there is no prospect that these trips or miles would be substituted on a like for like basis.

• Casualty numbers and risks are often measured in terms of those Killed and Seriously Injured (KSI). However, as there are approximately 13 reported serious injuries for every one fatality, KSI is really a measure of serious injuries. In addition, the UK definition of Serious Injury used in STATS19 is very broad. It ranges from cuts and minor fractures to dying more than 30 days after the collision. Whilst this may be a reasonable definition for road safety purposes, it makes KSIs a very crude assessment of risk. This has a bearing on assessing the risks of active travel. Cyclists have a much higher ratio of reported serious injuries to fatalities (36:1 in 2017) than car occupants (11:1) and the percentage of clinically serious (MAIS3+) injuries for cyclists admitted to hospital was lower (9%) than for any other road user group. In other words, cyclists may be involved in more KSIs but they are predominantly minor “serious injuries”.

So, below we have PACTS giving a basic illustration of casualties killed in collisions with the mode in question, as well as the users of that mode:

FIGURE ONE

(Incidentally, this is a good illustration of the point PACTS make that the vast majority of people killed on GB roads in 2017 died in collisions involving cars or vans.)

Teasing this out a little, take a look at the illustration below from SWOV in the Netherlands (with a population 3.9 times smaller), also for deaths in 2017.

FIGURE TWO
Transport mode victim (slachtoffer) against Transport mode third party (tegenpartij)

Of course, this does not imply legal responsibility is always that of the user of the third party. Nevertheless, I think this is a good indicator of “Who Kills Whom”.

Where this gets more interesting is in the next graphic from the PACTS document, where the number of non-users of the mode killed in collisions is expressed as a rate with exposure in terms of distance travelled by the mode:


FIGURE THREE
(Source: RAS 40004 P.235 -237 )

Figure Three gets closer to an understanding of which modes (or modes’ use) are implicated in the deaths of other road users. This in turn will lead us to assessing which modes pose more or less threat to others on the road (the two are different, albeit related, characteristics).
I want to refine this further by considering involvement of other modes in pedestrian deaths only, to allow for a more direct comparison. Furthermore, we see that according to Road type TSGB0702 TSGB0709 Percentage of road traffic and road length on different road types

LENGTH                                    TRAFFIC
Motorway                                    1%                                             21%
Urban A roads                            3%                                             15%
Rural A roads                              9%                                             29%
Urban minor roads                  33%                                             21%
Rural minor roads                    54%                                             14%

This means that 65% of traffic (i.e. motor traffic) is on roads used relatively less by pedestrians and cyclists.

So, to make a better comparison of involvement in pedestrian deaths I have shown in the next figure pedestrian deaths by involvement with other road users (per billion vehicle kilometres) in urban areas only, again for 2017.

Bicycle                                                    0.8

Motorcycle                                            4.9

Car                                                          1.4

Bus/Coach                                             0.7

Van                                                          0.6

HGV                                                         13

All                                                            1.6

Pedestrians killed in collision with other road users on urban roads per billion vehicle kilometres
FIGURE FOUR

(Apologies for tabular representation only: problems with uploading the graphic)

Source: RAS 30018 p.114 

Discussion: Involvement in deaths by mode…

Firstly, I have used deaths for uniformity and as these are the most reliable statistics, with fewer problems from non-reporting. However, this does mean that analysis is limited by some very small numbers, such as the three pedestrians killed in 2017 in collisions with pedal cycles. A different picture can emerge if using the less reliable Serious Injuries figures as well.

Secondly, there are two “stand-out” vehicle classes for involvement in pedestrian deaths – and these are similar to those for all other 3rd parties as shown in Figure Three, namely HGVs and, to a lesser extent, motorcycles.

Thirdly, I have separated out cars from vans in Figure Four: vans have the lowest involvement per billion vehicle kilometres travelled. I have to say I don’t know the reason for this. (Please feel free to send in comments on this and other issues below.)

Fourthly, buses and coaches seem to be very low for involvement in both Figures Three and Four.

Fifthly, while bicycles are at a very low level of involvement, they are not as low as pedestrians in Figure Three.

I now want to make a further step in the analysis of these figures which looks at what lies behind these levels of involvement. This is where we look at potential lethality of different modes.

…and potential lethality of modes

So: why is it that we see these differences in involvement?

1. HGVs. For some time we have known that HVs are heavily represented (compared to their modal share) in urban cyclist deaths – approximately 50% of cyclist deaths in London for the last couple of decades, despite lorries being some 5% of traffic. The figures for pedestrians are similar. Attention has been on the “blind spots” for HGV drivers that make it difficult for them to see pedestrians or cyclists near their vehicle, and the gap between vehicle body and tarmac, which is large enough for pedestrians and cyclists to be run over by what are very heavy vehicles. So far, so straightforward.

2. Buses and coaches. There are design features on modern buses which reduce danger to pedestrians: unlike HGVs there is Direct Vision on many, and generally a smaller gap between bus body and tarmac. Bus drivers are professionally trained: but so are HGV drivers, and bus drivers are under pressure to reduce “headway” and have other pressures on them. So the position of buses and coaches in both figures three and four is very low, and indeed lower than we might expect.

3. Motorcycles. The involvement rate is high here. We know that motorcycle users have a far higher user KSI rate than other modes by some way. But why the especially high figure for involvement with other users generally and pedestrians, as shown in figures three and four?

4. Cars. Again, although the figure for cars is higher than buses/coaches and bicycles, it is still a lot lower than motorcycles. Cars have a wider frontal area than motorcycles and can be used at similar speeds – so why is their involvement so much lower than motorcycles?

I want to answer these questions by referring to two key factors which have to be considered as we work backwards from involvement rates to a proper measure of danger:

Potential kinetic energy dispersed on impact

This is the key element of what we mean by “road danger”. It is essentially a product of the MASS of the vehicle (what most people understand as “weight”) and the SPEED it is travelling at when it hits a pedestrian. (When it hits another vehicle, the speed and mass of both vehicles have to be considered).

It’s why we in the Road Danger Reduction movement spend so much time seeing motor vehicle usage as the problem: it isn’t because of the personalities or intentions of drivers, it’s because of the physics involved. (This point is made nicely by the journalist Peter Walker  ).

It’s why anybody concerned with pedestrian safety has been urging communities to accept that motor vehicles should go no more than 20 mph if there are pedestrians in the vicinity. Partly this is because of the biomechanics of injury sustained by pedestrians – 20 mph is at the upper limit – but also because as speed goes up the energy potentially released increases exponentially: the usual figures quoted for pedestrian death rates in collisions with motor vehicles are 10% (or lower) at 20 mph; 50% at 30 mph and right up to 90% at 40 mph.

This is all shown above by comparing motorcycle and bicycle involvement rates: essentially motorcycles can be used at much higher speeds than bicycles (particularly well above 20 mph) and also have much greater mass.

But it doesn’t explain why cars, vans and buses are as low as they are. Buses tend not to go that much over 20 mph on routes in London, but cars and vans can and do. What is the reason for these levels of involvement being lower than we might expect just from looking at potential kinetic energy released on impact?

Pedestrian behaviour

A crucial element of the Road Danger Reduction approach is to accept that human beings constantly adapt their behaviour to their perceptions of danger. After some time being exposed to motor traffic – and the instructions from anxious parents and others – pedestrians tend to develop an awareness of what may pose a threat to their safety. This will affect how they actually behave in the highway environment.

So, I would suggest the following possible explanations for the data shown above:
1. Despite some awareness of the danger posed by HGVs, there is a problem with non-Direct Vision HGV cabs, and the consequences of making even small mistakes are very extreme. The figures for Serious Injuries (SIs) are less than twice as high as those for Killed – whereas the average for all vehicles is over ten times as high.

 2. Buses are very large and generally coloured red in London, taking up predictable positions in traffic, frequently in specific bus lanes. This would suggest a high level of pedestrian awareness of the threat, and accordingly a higher level of care taken.

3. By contrast with 1 and 2. above, motorcycles have a much narrower profile and are less easy for pedestrians to predict. I would argue this is a key factor in the high level of motorcycle involvement in pedestrian deaths.

4. Cars are a lot easier to see than motorcycles, and along with vans always make up a significant part of the traffic on roads in urban areas. People expect them. Therefore their involvement rate is lower than motorcycles, and not much higher than bicycles.

5. So why are bicycles at the level of involvement that they are? Their lack of mass and speed puts them far below motorcycles, with which they have a similar profile. But why are they not that far below cars (and possibly even at the same sort of level as vans)? I would suggest that pedestrians are not only less likely to be aware of bicycles because of the narrower profile, but also that pedestrians are aware that the danger level (because of lower speed and mass) is so much lower. This may be amplified by the fact that bicycling tends not to be seen as a “serious” form of transport in the UK. (I would urge readers who both drive and cycle in an urban area to consider this: do you find that pedestrians are more willing to step out in front of you when you are cycling or driving?)

Conclusion

I have attempted to present collision involvement rates, but also look at what lies behind them. This should indicate where practitioners should focus their attention when it comes to reducing danger. As ever, the indications are that attention should be paid to motor vehicles of all types – with particular issues around each of the motorised modes – before cyclists. Finally, we note that this type of discussion is unusual among transport professionals, and hope that it sparks comment and debate.

Dr Robert Davis 11th March 2019

 

…and also: Professor Jenny Mindell reminds us of her work here  (Note that measure of exposure is by time rather than vehicle kilometres travelled) 13/03/2019

13 thoughts on ““Who kills whom” and the measurement of danger.

    1. rdrfuk Post author

      There were 3 non-users (pedestrians) killed in collisions with cyclists in 2017: as said in text, it’s very small, but the larger number of Serious Injuries has it’s own problems of reliability, and I wanted uniformity with the figures in the PACTS document which was based on deaths.

      Reply
  1. David Davies

    Pleased to see that the pacts analysis was of interest and that you have developed it further and with a discussion if the reasons underlying the numbers. There is lot more to be done here. A great start.

    Reply
  2. David Davies

    Pleased to see that the pacts analysis was of interest and that you have developed it further and with a discussion if the reasons underlying the numbers. There is lot more to be done here. A great start.

    Reply
  3. John Davis

    Good to see lots of Davis/Davies mentions. Two thoughts on your conclusions from a London cycling commuter perspective. 1) noise, bicycles are almost the only silent mode, not a good combination with phone distracted pedestrians. 2) bicycles and motorcycles both have much more lane to lane mobility, i.e. they can be less predictable than other modes.

    Reply
    1. rdrfuk Post author

      I think those are two good points. Comparing motorcycles with bicycles as having lane-to-lane mobility and similar profiles, despite the relative silence of the bicycle, motorcyclists have a FAR greater involvement in pedestrian deaths per distance travelled: this (again) highlights the point about kinetic energy released on impact, with motorcycling having far more, due to greater mass and speed.

      Reply
  4. Pete Owens

    Some interesting questions.
    Actually, in terms of comparing benign transport modes you should really quote the figures in terms of deaths per billion passenger km rather than vehicle km – on this measure a bus passenger is less lethal than a pedestrian.
    The issue of kinetic energy is more complex. When a massive object collides with a smaller one it does not transfer all its KE – it accelerates the smaller object to its own speed, So it the KE that the smaller object gains rather than the original KE of the larger object that is relevant. If a pedestrian is hit by a vehicle the KE will be proportional to the mass of the pedestrian rather than the vehicle – it doesn’t matter whether that vehicle is a motorbike, a car, a truck … or an ocean liner. Of course speed is critical – if a pedestrian is hit by a vehicle travelling at 40mph they will gain 4 times the KE as they would if hit by a vehicle travelling at 20mph, whatever the masses of the vehicles.
    Another consideration is the design of the front of vehicles. Bonnets of cars are designed to cushion the impact of a low-speed collision with a pedestrian. If we could get speeds down to 20mph in towns then cars would become almost non-lethal (at least if we ignore the pollution). Cycles & motorbikes have sharp, unforgiving fronts, so will tend to be more damaging at the same speed. To understand this, imagine you are trapped by fire in a 2nd floor office with your only means of escape to jump from one of two windows. Below one window is a parked car and below the other is a bike rack with several parked cycles. Which window do you choose?
    Better design of truck cabs for visibility should make some difference, but I think the most important feature explaining the excess lethality of HGVs is crush injuries cause by falling under the wheels.

    Reply
    1. rdrfuk Post author

      Good points IMO.
      Kinetic energy dispersed on impact is a bit more complex, but I think mass is important. Of course, you’re right about vehicle design: cars and vans have bonnets designed to make impacts les destructive than they would have been say, 50 years ago (or less).
      I don’t know if the jumping out of the window analogy is helpful: there sharp bits on bicycles, but the key issue would be the “give” in vehicle roof on the one hand, and the fact that the bicycles will fall over when hit on the other.

      Reply
    2. rdrfuk Post author

      …and also on Kinetic Energy, Susie Morrow of Wandsworth Living Streets has commented on twitter that: “re kinetic energy – wording implies KE = mass x velocity – or could be read as such. It’s the fact that KE = 1/2 mass x (velocity x velocity) that makes speed, and speed control, so critical.”

      Reply
  5. Jitensha Oni 🇪🇺🇳🇴🇯🇵 (@jitensha_oni)

    Minor point – from personal experience, I’d add to your list of pedestrian behaviours the “Invisible Gorilla” effect. A recreation of the experiment is given at:

    A discussion of how psychology generally affects drivers not seeing people cycling or motorcyclists is given on the London Cyclist blog:

    https://www.londoncyclist.co.uk/raf-pilot-teach-cyclists/

    But this will presumably also affect how pedestrians register people cycling. For example, I’ve been stationary straddling my bike looking at something in a shop window in a shared use area and had a person walk towards me until they were very close, then suddenly clutch their chest with a look of confusion, then shock, followed bizarrely by “you need to slow down”. At least their companion saw what had really happened, and asked them to look where they were going. But, whatever, this happens – something unexpected is just invisible to some people, so some pedestrians may walk into riders going at any speed, or even standing still, to their detriment. The vast majority don’t, even at the low cycling levels typical of the UK, and I wouldn’t like to guess how significant the “Invisible Gorilla’ effect is to interpreting your unexpectedly (?) high cycling danger figure. It may or may not only take a few in a population to be “cycle-blind” to cause some or all of the KSI rate you observe, but it’s not something easily recorded.

    Reply
    1. rdrfuk Post author

      Thanks for this Jitensha – I think that’s important. I suggest that one would expect with increasing levels of cycling one might get a “Safety in Numbers” effect whereby pedestrians become increasingly likely to watch out for cyclists. Obviously we need a significant increase in numbers of cyclists to find out.

      Reply
  6. rdrfuk Post author

    This tweet from the Urban Cycling Institute is relevant: Want to change the conversation from who is KILLED to who is KILLING on our roads? Just translate our tool: 🔩 Open source coding 👩‍👩‍👧‍👦 Crowdsourced data (harvesting local news articles) 🚗 Easy to categorize 🏥 Can also be used for traffic injuries Info: https://www.hetongeluk.nl

    Reply

Leave a Reply

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

WordPress.com Logo

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

Google photo

You are commenting using your Google 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