Response to the London Assembly Call for Evidence: Bus Services

This is a response to the London Assembly Call for Evidence on Bus Services.

I am writing as an expert in cycling, and hence am only focusing on bus infrastructure/services as they relate to cycling. This is clearly only a part of the picture, but I believe one that matters and one that should be taken into account in planning for buses and for cycling.

Specifically, in relation to the safety topics, I am responding to questions:
3. What are the particular safety concerns for other road users (NB I am writing here in relation to cycling)
12. Are there any problems caused by bus and cycling infrastructure sharing road space (particularly kerb side) and how could these be resolved?

Finally, the response briefly touches on cycling as it relates to bus journey times (which is of interest for the service provision topic, though there is no question specifically on the issue).

The key points I would like to make are:

Please don’t only look at cycle-bus collisions. Also look at cycling injuries that take place where shared bus lanes are present – hundreds of cyclists are injured annually in London at such locations.
Perception of safety also matters – shared bus lanes are not a preferred form of cycle infrastructure.
When high volumes of cyclists and of buses are sharing lanes, this may have a negative impact on bus journey times, compared with separate provision for each mode.

Only a small minority of London cycle collisions involve buses (although the annual number of serious cycling injuries resulting from bus collisions and from HGV collisions are similar – both are large vehicles and when collisions do occur they can often have serious consequences, although HGVs kill more cyclists than do buses). However, to only examine the risk posed to cyclists by buses would be to miss the much wider question, implied in (12).

In the UK bus lanes have historically been considered a form of cycle infrastructure provision. This policy position implies that bus lanes should deliver some benefits for cyclists, including importantly in terms of safety, perceived safety and comfort. But do they?

While shared bus lanes may be seen as preferable to A roads with no cycle provision at all, London’s cyclists do not view bus lanes as a desirable form of cycle infrastructure. For instance a TfL (2012) stated preference survey found cyclists thought bus lanes were slightly worse than narrow advisory cycle lanes, while my own study found that across the UK, including in London, bus lanes were not seen as preferred infrastructure and were not seen as child-friendly.

The question of how far bus lanes protect cyclists from motor vehicle collisions has been very little investigated either in the UK or elsewhere. This is a shame, because London has a substantial volume of bus lanes, and therefore it would be feasible to conduct such investigation; given annual data on (a) estimated cycling flow across the network and (b) bus lane locations. I have requested (a) so I can carry out academic research looking more broadly at cycling risk and infrastructure, and subject to agreement, I hope to publish this research in an academic journal which can then be publically accessed.

At the time of writing, I haven’t got access to this data, though (and it will be a while before I can get an article published, when I do have it). Although in recent analysis I have used the Propensity to Cycle Tool to estimate cycling flow at a borough level and believe this to be sufficiently robust to estimate risks at borough level, I decided against using this to estimate volumes on specific route sections, because the PCT routes cyclists on the ‘most direct’ cycleable route, which does not necessarily mimic actual route choice behaviour (it is not intended to do this). Thus, the discussion here is necessarily limited, but I believe worth submitting, given the importance of the topic.

So, I do not have the ‘denominator’ data to reliably calculate the extent to which the presence of bus lanes might affect cycling injury rates. What I do have, is collision data from Stats19 on cycle injuries in London, and a map of the London bus lane network as it existed around 2015. Using QGIS software, I combined the two maps, attempting to identify which A road cycling injuries between 2012-5 took place on a road section with a with-flow bus lane. There was relatively little change to the bus lane network during these years, so I think this is reasonable, although annual bus lane data would be better. It was harder than I thought it would be to automatically assign each A road injury to a bus lane section or not, and not every injury is necessarily located accurately inside or outside a bus lane section.

However, looking at the results it seems generally good enough on a network-wide basis, as a first attempt. When I get a chance to do a more thorough analysis with better data, I’ll refine it.

I then explored the characteristics of A road cycling injuries happening where bus lanes were present or absent. I removed from the dataset the small (though disproportionate) amount of cycling collisions happening at junctions with other A roads, on the grounds that bus lanes in the UK do not – except in very unusual circumstances – continue in such locations. Thus it seemed misleading to attribute those collisions and their characteristics to ‘non bus lane’ A roads. In a small number of cases (with or without bus lanes) cycling infrastructure might have been present at a crash location; but given that these years largely predate the completion of the segregated superhighways, in the vast majority of all cases (bus lane or no) cyclists would be legally obliged to use the carriageway.

One thing that surprised me was the large number of non-bus vehicles involved in cycling injuries where bus lanes were present. Perhaps I should not have been surprised. Bus lanes are only paint-based protection, after all. And many are not in operation off-peak, while often car parking or at least loading and unloading (creating more hazardous interactions for cyclists) are allowed at least some of the time. Almost all London bus lanes allow taxis and many now allow motorcycles, which seems unlikely to change in the near future – in fact recently the Mayor has announced that more bus lanes will be opened up to taxis.

Perhaps more fundamentally, though, many of the classic causes of cycling injury in London – which, as noted above, usually involve non-bus vehicles – are unlikely to be avoided by well-enforced, full-time bus lanes that ban parking and even loading. Bus lanes do not offer protection against motor vehicles pulling out of side roads into a main road, or turning left or right into a side road across a cyclist’s path. As alluded to above, they can rarely provide any protection at major junctions, as they disappear in most cases. There are also specific hazards caused by cyclists having to share with buses on busy roads – in particular, interactions at bus stops where cyclists may overtake buses and be at risk from collisions with other vehicles.

How many cycling injuries take place in bus lanes? Excluding those A road/A road junctions, I think that just over one in five A road cycling casualties between 2012-5 took place where a with-flow bus lane section was present. London’s A roads have around 275 km of bus lane, compared to just under 1,700 km of A roads. Note that the former measures one-way sections of bus lane, so it’s not the case that 15% of London’s A road length has a bus lane – it’s likely to be more like 8-9%, as that A road length will be mostly (not entirely) made up of two-way A roads – i.e. becoming more like 3,000 km when we separate out directions.

So 21% of casualties injured in bus lane sections, versus around 8-9% of road length with A road sections. Does this mean bus lanes are more dangerous than A roads without bus lanes, then? I don’t think so, because (a) Inner London boroughs have more cycling, and also (in almost all cases) have a higher proportion of bus lane than Outer London boroughs, and (b) people might in any case be more likely to cycle on A roads with bus lanes, rather than bus lanes with no cycle infrastructure at all, given the choice. However, in many cases people wouldn’t have a simple choice between ‘A road with bus lane’ and ‘A road without bus lane’ in that way, because often bus lanes are intermittent across the length of an A road – even in Inner London.

Anyhow, as I don’t currently have good enough data to estimate levels of cycling in bus lane vs. non bus lane sections, this is somewhat speculative. To return to the question of cycling injuries on A roads where bus lanes are, or are not, present, between 2012-15 (excluding as I mentioned the minority of collisions at A road-A road intersections), I found the following:

• For 4.7% of bus lane cycle casualties, a motorcycle was involved, compared to 3.0% of non-bus lane casualties
• For taxis/PHVs, the figures were 7.0% and 7.3%, a difference that was not statistically significant
• For 65.3% of bus lane cycle casualties, a car was involved, compared to 68.9% of non bus lane casualties
• For 11.0% of bus lane cycle casualties, a van was involved, compared to 10.0% of non bus lane casualties, a difference that was not statistically significant
• For 3.6% of bus lane casualties, a bus was involved, compared to 3.0% of non bus lane casualties, a difference that was not statistically significant
• For 1.8% of bus lane casualties, an HGV was involved, compared to 2.6% of non bus lane casualties
• For 0.9% of bus lane casualties, a goods vehicle of unknown/unrecorded weight was involved, compared to 0.5% of non bus lane casualties

The bus lane sections had a higher proportion of fatal and slight casualties and a lower proportion of serious injuries, but these differences were not statistically significant.

NB that the above doesn’t tell us if someone is at higher or lower risk of being hit by any of the above vehicles in a bus lane vs. a non bus lane section; it just gives us information about the types of collision/casualty that occur in each context.

Lambeth has the highest proportion of A road cycle casualties (41.2% of all A road cycle casualties between 2012-5, excluding A road-A road collisions), and in Bromley and Sutton there were no A road cycle injuries between 2012-5 in a bus lane section. This underlines the point about the importance of exposure – the most important factor here is the presence of bus lanes, I’d think. There’s an awful lot more bus lanes in Lambeth than there in Bromley or Sutton.

And overall, I found, the risk per km cycling in Inner London boroughs like Lambeth is often several times lower than it is in many Outer London boroughs including Bromley and Sutton. Borough-level analysis is fraught with confounders and I’m wary of trying to derive general explanations at this level. Plausible suggestions could include motor traffic speeds (which tend to be lower in Inner London) or driver culture in Outer vs. Inner London (possibly related to cycling levels). It’s also possible that infrastructure that existed in 2012-5, including potentially bus lanes, could be a contributor (although the Inner London borough that stands out as having relatively few bus lanes, Kensington and Chelsea, doesn’t seem that different to other Inner London boroughs in terms of cycling injuries per kilometre). As I say, I’m hoping I’ll soon be able to explore this further at a route rather than area level, which I feel is more appropriate.

In any case, what we can say so far is that the pattern of cycling injuries taking place in A road bus lane sections so far doesn’t seem to be radically different from those taking place in non-A road bus lane sections – for one thing, they’re both relatively unlikely to involve an actual bus. This suggests we need to broaden our thinking about cycles and buses – to think about the level of protection bus lanes might or might not provide cyclists, rather than only focusing on bus-cycle injuries (of course, that’s not to say those aren’t important). Hundreds of Londoners are injured every year cycling where there are bus lanes; so we need to take a closer look at risk and injury in these. This post and submission to the inquiry is a start.

Finally, a word on cycling and bus services. Again here, I think we need to take a broader look at shared bus lanes. Research I have carried out suggests that at the kind of cycling volumes we are getting in some Inner London locations, shared bus and cycle lanes are inevitably going to have an impact on bus journey times. This is something that seems intuitive when you think about it, yet is not factored into planning. Instead, we assume that the impact that cyclists have on bus journey times is only through the building of cycle tracks and other ways of allocating separate space to cyclists.

A research paper based on a study that I was involved in (currently under second stage peer review; I am happy to send a copy on request but it cannot be publically shared until accepted for publication) found cyclists having a significant impact on bus journey times across London Bridge. Hence, re-allocating separate space for cyclists can – in some circumstances, depending upon where the space comes from, what the current and likely future levels of cycling is – potentially have positive impacts on bus journey times. However, in modelling this is generally not taken into account yet, partly because the high cycling flows in bus lanes that London is seeing are pretty unprecedented in the UK (and probably anywhere in fact, as most high-cycling countries don’t generally make cyclists share with buses, particularly where there are high volumes of both modes).