SaferActive Report 4: Trends in road safety and potential causal factors

Institute for Transport Studies

Source: vignettes/report4.Rmd

1 Introduction

The Saferactive project is funded by the Department for Transport. It aims to investigate spatial and temporal patterns in road safety for active travel. This is the fourth and final report of the project, which provides an overview of project findings and recommendations for next steps and research and data collection priorities. The previous reports covered data on road safety (Report 1), methods of geographic data analysis and scenarios of change (Report 2), and modelling uptake of active modes and safety nationwide (Report 3). These reports are open access and can be found online.

Datasets on road traffic casualties that were reported to the policy are available from the open access STATS19 database. Published by the Department for Transport, STATS19 is a comprehensive database of road collision statistics, mainly as reported by police officers on the scene of the collision, with information at accident, casualty, and vehicle levels.

However, to estimate collision risk, we must also know the exposure. Exposure can be measured in different ways, including the number of people walking and cycling through a particular point or junction on the road network and total distance travelled along a particular segment or within a particular area. The exposure acts as the denominator in any estimate of road safety risk, allowing safety to be calculated as a rate such as number of people killed and seriously injured per billion kilometres (KSI/bkm) (Feleke et al. 2018). The degree of uncertainty associated with estimates of exposure is high, and this uncertainty will be a key focus of the report.

The majority of the report focuses on cycling rather than walking. Firstly, this is because better data is available for cycle journeys, in particular journeys to work. Travel to work is recorded in the 2011 Census, and according to the NTS survey in 2011 it comprised around 38% of all cycle journeys but only 7% of journeys on foot. Therefore we decided it is acceptable to use travel to work as a proxy for cycling but not for walking. Networks of cycle counters from the Department for Transport (DfT), Transport for London (TfL) and others are also available, providing data on annual changes in cycle flows, while comparable data does not exist for walking.

A second reason for focusing on cycling is that since 2020 there have been large changes in how we use our streets, with unprecedented increases in cycling during the lockdown periods, accompanied in some cases by a rise in cycle casualties. It is important to take these changes in context and to recognise that an increase in casualties does not necessarily mean an increase in risk. This is why it is important to focus on exposure and the denominators of road safety.

Initially, in Section 2, we explore the input data used. Sections 3 and 4 of this report focus on changes in casualty rates at national and regional scales, over the last decade, using data that covers all journey purposes. Section 5 interrogates changes at sub-regional scales. This is where data availability forces us to limit the scope to commuter cycling within England and Wales. In Section 6 we again narrow the geographic scope and look at how cycle and pedestrian casualties are associated with different types of road, including a generalised definition of Low Traffic Neighbourhoods, in a West Yorkshire case study area.

2 Input data

We have used a range of nationally available datasets to estimate and visualise road safety, including Stats19 collision data, DfT traffic counts, NTS survey returns and 2011 Census commute data.

We have focused mainly on the decade from 2010 to 2020, but there is some variability in the years covered by different datasets. We accessed 2020 data for collisions and DfT traffic counts, but not for all of the other datasets. Where possible, we include data from England, Wales and Scotland. The sub-regional analyses are only available for England and Wales since they are based on the Propensity to Cycle Tool. NTS data are available for England only.

Stats19 is a comprehensive dataset of British road collision statistics, published by the DfT. The data are collected by police forces, mostly being recorded by officers at the scene of collisions. Casualties are classed as slight, serious or fatal. Collisions resulting in slight injuries may not always be reported, but reporting rates for serious and fatal injuries are very high. The key yardstick of road safety is the number of people killed or seriously injured (KSI). In recent years there have been changes in the methodology used by many police forces to categorise injuries as slight versus serious. Adjustment factors are used to standardise and account for these methodological changes. The Stats19 data for 2020 was only available in summary form at the time this report was written.

The DfT traffic counts are a series of manual traffic counters. Counter locations are typically surveyed from 07:00 - 19:00 on a single date per year, with seasonal adjustments applied to calculate an Annual Average Daily Flow (AADF). However, the same locations are not surveyed each year. Only 7% of counter locations see annual surveys 2010-2019, and very few of these locations were also surveyed in 2020. Careful analysis is therefore required to accurately represent temporal trends.

NTS surveys have been conducted annually since 1988. Results are available for each English region but not at higher geographic resolutions. The self-reported household survey takes the form of a seven-day travel diary, and includes journeys of all purposes. It is completed by around 15,000 individual respondents.

The 2011 Census provides comprehensive nationwide coverage. Questions include method of travel to work, which can be used together with home and work locations to provide full origin-destination (OD) data for commuter journeys. For England and Wales, the Propensity to Cycle Tool combines this data with routing algorithms created by CycleStreets.net, in particular the fast route algorithm which identifies the fastest cycle route from origin to destination, accounting for the impact of hills.

6 Low Traffic Neighbourhoods and rat-runs

Traffic levels on minor roads have increased dramatically in recent years, by 26% nationwide compared with only 9% on A roads, according to DfT’s official statistics. Many factors influence road safety, but since the coronavirus pandemic modal filters and other traffic calming interventions in residential areas have come to the fore, explaining our focus on residential interventions in this section.

Although the phrase ‘Low Traffic Neighbourhood’ is a recent invention, the concept is not new. Modal filters preventing vehicle movement along a street while allowing passage to pedestrians and cyclists have existed for many decades. They can take a wide range of forms, such as bollards, gaps, and continuations of pavement across the entrances of side streets.

It can also be argued that any street, or any network of streets, which does not provide a vehicle connection between two main roads, is in effect part of a low traffic neighbourhood. These streets do not provide worthwhile opportunities for ratrunning, so they are likely to be used mainly by residents and those wishing to access destinations on the street itself. This includes cul-de-sacs and wider networks of streets for which all of the access points are on the same main road.

Using this definition of low traffic neighbourhoods, CycleStreets.net have created a nationwide beta map of LTNs. All roads are assigned as main roads, low traffic neighbourhood roads, ratruns, or ratruns with traffic calming measures. Main roads are those classified as ‘A’, ‘B’ or ‘C’ roads. Data on traffic calming measures are obtained from OSM.

This map extract from Leeds gives an example of how roads are classified.

Extract from CycleStreets.net showing a classification of roads in north Leeds

Figure 6.1: Extract from CycleStreets.net showing a classification of roads in north Leeds

We have used this beta map with West Yorkshire as a case study region, to assess the impact of Low Traffic Neighbourhoods on road safety.

6.1 Active travel KSI, road length and commuter cycling in West Yorkshire

For this case study we have divided the West Yorkshire road network into a set of 1km grid cells. We see a clear correlation between total road length within each 1km cell and the number of pedestrian KSI over the period 2010-2019.

Pedestrian KSI over the years 2010-2019 against road length in km. Each point represents a 1km grid cell

Figure 6.2: Pedestrian KSI over the years 2010-2019 against road length in km. Each point represents a 1km grid cell

There is also a positive association between road length and cycle KSI but it is weaker, likely reflecting the fact that the cycling distribution is more uneven than the pedestrian distribution.

Cycle KSI over the years 2010-2019 against road length in km. Each point represents a 1km grid cell

Figure 6.3: Cycle KSI over the years 2010-2019 against road length in km. Each point represents a 1km grid cell

We can see a positive correlation between distance cycled for travel to work and cycle KSI.

Cycle KSI over the years 2010-2019 against km cycled for travel to work according to the 2011 Census, using the fast route network derived from the PCT. Each point represents a 1km grid cell, with colour weighted by total road length within the grid cell

Figure 6.4: Cycle KSI over the years 2010-2019 against km cycled for travel to work according to the 2011 Census, using the fast route network derived from the PCT. Each point represents a 1km grid cell, with colour weighted by total road length within the grid cell

Switching to maps, we can see how these patterns play out across West Yorkshire. The greatest concentration of cycle and walking KSI are found in the urban areas such as Leeds and Bradford. Cycle commute km are strongly concentrated within Leeds.

Cycle and pedestrian KSI for the period 2010-2019 and km cycled for travel to work in West Yorkshire. The map of km cycled only includes cells that form part of the PCT-derived fast route network for the 2011 Census

Figure 6.5: Cycle and pedestrian KSI for the period 2010-2019 and km cycled for travel to work in West Yorkshire. The map of km cycled only includes cells that form part of the PCT-derived fast route network for the 2011 Census

Now looking at the distribution of KSI risk, we see that for walking, KSI risk per km road remains highest in the urban areas, especially Bradford. For cycling, KSI risk per km road has a slightly more scattered distribution with high values in Leeds. However, when we use Bkm cycled as the denominator, the map is very different. Now some of the lowest rates are found in Leeds. This map shows very high KSI rates in some of the Pennine fringes of West Yorkshire, but these are probably artefacts of the fact that we are using commuter cycling as the denominator, while leisure and sport cycling dominate in these areas. A more reliable finding is that KSI risk appears higher in Bradford than in Leeds.

Cycle and pedestrian KSI risk for the period 2010-2019 and KSI per Bkm cycled for travel to work in West Yorkshire, assuming 35% of cycle journeys are commutes (NTS mean for England in 2010-2019). The third map only includes cells that form part of the PCT-derived fast route network for the 2011 Census

Figure 6.6: Cycle and pedestrian KSI risk for the period 2010-2019 and KSI per Bkm cycled for travel to work in West Yorkshire, assuming 35% of cycle journeys are commutes (NTS mean for England in 2010-2019). The third map only includes cells that form part of the PCT-derived fast route network for the 2011 Census

6.2 KSI and road types in West Yorkshire

The next step is to investigate how these patterns relate to the distribution of road types. Unsurprisingly, the total road length within each 1km grid cell varies greatly across the region, being highest in towns and cities, but so too does the distribution of types of road within each grid cell.

Weighting by the road length within a grid cell, we look at how pedestrian and cyclist KSI vary with the proportion of roads of different types. We expect that KSI rates will increase as the proportion of ratruns increases, and decrease as the proportion of LTNs increases. This does seem to occur, as borne out by investigatory GLM models, although the relationship is weak as many other factors are at play. It would be interesting to study these relationships across wider spatial scales. Another option is to use a function to more precisely define street networks of local neighbourhoods that share common features, and use these as the unit of measurement rather than using 1km grid cells.

7 Discussion

There are many questions on the topic of road safety that deserve further investigation. A long standing focus of interest is the relationship of road safety risk with deprivation and inequality. It is possible for example to investigate the identity of casualties and of other people involved in collisions, together with the characteristics of the areas where collisions occur, assessing these using the Index of Multiple Deprivation. We might expect higher casualty rates for people from more deprived neighbourhoods, but interactions also take place with for example people who live in less deprived neighbourhoods being involved in collisions in more deprived neigbourhoods, and vice versa.

Another topic of interest is safety in numbers - the idea that in places where more people cycle, the risk of injury is lower. This appears to be borne out by the spatial trends shown in Section 5.3, as well as in temporal trends - for example when cycling increased across England and Wales in 2020 the risk reduced. However, we must still question the role of causality. Is it the higher numbers of cyclists that make the roads safer, or are more people willing to cycle in places and at times when other factors (such as the design of the road environment) mean it is inherently safe? Potential reasons why increased cycling might in itself lead to reduced injury risk could include changes in driver behaviour, in that drivers are more likely to actively look out for cyclists in locations where the number of cyclists is higher. A more in-depth study would be required to tease out these causal relations.

A further question is whether places with low casualty rates and falling risk share a set of common features or policies, that can be applied to other regions to improve their safety. Scotland has seen some of the greatest reductions in injury risk in the UK over the last decade, and has a comprehensive Road Safety Framework to 2030. A focus on safe speeds is one of the five pillars of this strategy, with others focusing on road design, vehicle design, road user behaviour, and post-crash response. Lancashire is another area in which initially high collision rates have improved greatly. The county appears to have an active Road Safety Partnership.

The road environment has changed greatly over the last two years. We have seen unprecedented lockdown and an increase in cycling, as well as opportunities for speeding, on relatively traffic-free roads.
Now traffic levels have returned to normal faster than public transport patronage, potentially increasing risk and congestion.
We also have new types of road user such as legal and illegal e-scooters, which are becoming more widespread. Meanwhile, among car drivers, there is a trend towards heavier and larger vehicles, such as SUVs, which may react differently in the event of a collision. These are just some of the trends that could influence road safety in future.

7.1 Assessing risk

Much of this report has focused on the question of how to measure road safety risk, but rather than having one simple answer it is clear again that there are many different ways of approaching this topic. We have focused mainly on risk related to cycle commuting, which has some of the best data quality, but even here different data sources will provide different interpretations. This uncertainty must be understood in the context that to assess risk, we need measures of both casualty rates and exposure.

In the casualty data, it is important to be aware of issues related to the way in which the data is collected, such as differences between data recorded at the scene of a collision and data that has been self-reported online. Contributory factors that are recorded may be the interpretation of the reporting police officer. There is also the need to adjust for changing methodologies in determining injury severity. Notwithstanding these points, the Stats19 casualty data is comprehensive and its core components (such as number of casualties) are highly reliable. Therefore it is in the exposure, i.e. the distance cycled on the roads, that the greatest degree of uncertainty lies.

7.4 Impact of data scarcity

Data scarcity can cause issues when there are either limited numbers of counter locations or few collisions per year within a Local Authority. An empirical Bayes approach could counteract this with use of an informative prior which has the effect of normalising the results for LAs towards the mean to different degrees, dependent on the sample sizes they are based on. Another potential approach is the use of a multilevel model, with counter location as a random effect.

7.5 Further research

When investigating road safety, it is important to take into account both exposure and risk to avoid faulty assumptions arising from ‘denominator neglect’. Headlines around spikes in cyclist fatalities in London, with scant measure of the broader picture of increasing cycling rates that suggest an overall improvement in safety, demonstrate this point. Conversely, areas that see both decreases in active travel but a flatlining in casualties, should be cause for concern and interventions should be prioritised. Evidence on which interventions are most effective is still emerging but it is clear that they should tackle the root causes of unsafe roads for the modes that pose the least threat to other road users and who have the most to loose from high impact collisions motor vehicles.

This project demonstrates that ‘safety’ can be quantified using existing datasets that are available nationwide. Specifically, the metric of number of people killed and injured per billion km walked or cycled, accounts for both changes exposure and collisions. Our analysis clearly shows that some areas across Great Britain have successfully increased walking and cycling levels, while simultaneously reducing absolute numbers of people hurt on the roads while walking and cycling (not to mention wheeling on e-scooters and other mobility devices not explored in this project). We recommend further research specifically into areas that have made the greatest reductions in KSI/bkm rates, while succeeding in enabling more people to walk and cycle for everyday trips, reducing reliance on cars. We recommend regular reporting of road safety outcomes alongside reporting of open access estimates of active travel, as part of wider strategies that have a long term goal such as ‘vision zero’ policies akin the zero carbon objectives with evidence-based regular targets.

Despite an increase in the number of cyclists killed and seriously injured on British roads in 2020 in the context of the COVID pandemic, the risk in terms of KSI/Bkm actually fell by around 30%. This is an area that is worthy of further research: which places saw the greatest increases in cycling and improvements in safety, why, and how can lessons learned from these places be applied elswhere? There are many factors that influence risk and and further research, building on our findings, could help identify which conditions and types of intervention (e.g. LTNs) are most effective in different types of settlement.

The time series analysis presented in our work enables policy makers to track change in road safety over time, for different places, and different groups. For example, the finding that road safety inequalities for cycling seems to be increasing, especially for children (Vidal Tortosa et al. 2021), should be cause for urgent investigation and the prioritisation of resources to support the levelling up agenda.

More broadly, our work creates a foundation on which further road safety research focussing on active travel can build. The estimates of cycling uptake and collision rates each year over the past decade, for every local authority, is a free and open resource. Furthermore the code underlying our work is open source and we encourage others to build on the analysis contained the saferactive repos on GitHub: https://github.com/saferactive/ This is an important and emerging field in need of consilidation and comparison of diverse data sources. There is a need for national reporting and provision of open data at high temporal and spatial resolution: monthly reporting of road safety and active travel outcomes at local levels could, for example, be used to monitor, assess and support intervention such as experimental TROs of the type used to implement LTNs.

References

Feleke, Robel, Shaun Scholes, Malcolm Wardlaw, and Jennifer S. Mindell. 2018. “Comparative Fatality Risk for Different Travel Modes by Age, Sex, and Deprivation.” Journal of Transport & Health 8 (March): 307–20. https://doi.org/10.1016/j.jth.2017.08.007.

Laverty, Anthony A., Rachel Aldred, and Anna Goodman. 2021. “The Impact of Introducing Low Traffic Neighbourhoods on Road Traffic Injuries.” Findings, January, 18330. https://doi.org/10.32866/001c.18330.

Lovelace, Robin, Joseph Talbot, Malcolm Morgan, and Martin Lucas-Smith. 2020. “Methods to Prioritise Pop-up Active Transport Infrastructure.” Transport Findings, July, 13421. https://doi.org/10.32866/001c.13421.

Vidal Tortosa, Eugeni, Robin Lovelace, Eva Heinen, and Richard Mann. 2021. “Socioeconomic Inequalities in Cycling Safety: An Analysis of Cycling Injury Risk by Residential Deprivation Level in England.” Journal of Transport and Health.