Redesign Grand Army Plaza

Article Text

Original research

Low Traffic Neighbourhoods in London reduce road traffic injuries: a controlled before-and-after analysis (2012–2024)

http://orcid.org/0000-0002-5730-5281 Jamie Furlong 1,
David Fevyer 1,
Ben Armstrong 2,
http://orcid.org/0000-0003-4431-8822 Phil Edwards 2,
http://orcid.org/0000-0001-7413-7251 Rachel Aldred 1,
http://orcid.org/0000-0001-9995-6659 Anna Goodman 2
1University of Westminster, London, UK
2London School of Hygiene & Tropical Medicine, London, UK
Correspondence to Dr Jamie Furlong; J.Furlong@westminster.ac.uk

Abstract

Background Between 2015 and 2024, 113 Low Traffic Neighbourhoods (LTNs) were implemented across Greater London, with 27 subsequently removed. We investigated their impacts on road traffic injuries inside LTNs and on ‘boundary roads’ immediately surrounding the LTNs.

Methods We matched police-recorded injuries from STATS19 data to Ordnance Survey road links that were spatially intersected with LTNs/boundary roads. Conditional fixed-effects Poisson regression models used the number of injuries per road link per quarter of each year (January 2012 to June 2024) to test whether LTN implementation was associated with changes in injury rates.

Results LTN implementation was associated with a 35% (95% CI 29% to 40%; p<0.001) decrease in all injuries and a 37% (95% CI 24% to 48%; p<0.001) decrease in people Killed or Seriously Injured (KSI). Injuries decreased across a range of casualty and LTN characteristics. However, there was evidence of a smaller benefit in LTNs implemented in Outer London since 2020. Following the removal of an LTN, injury numbers increased back to pre-intervention levels. On boundary roads, there was no evidence of a change in total injury numbers (estimate −2%, 95% CI −5% to +2%) or KSI injury numbers (estimate 0%, 95% CI −7% to +8%). This reflected decreased numbers of injuries on boundary roads for cyclists and motorcyclists, and no change for pedestrians and other motor vehicle users.

Conclusion LTNs in London reduced road traffic injuries among all road users inside the LTN areas, with no evidence of overall impact (and for cyclists and motorcyclists a benefit) on boundary roads.

Urban
Longitudinal
Geographical / Spatial analysis
Planning
Policy

Data availability statement

Data are available upon reasonable request. All data are available upon request to the authors.

https://creativecommons.org/licenses/by/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

https://doi.org/10.1136/ip-2024-045571

Statistics from Altmetric.com

See more details

Picked up by 7 news outlets

Posted by 44 X users

Reddited by 1

Referenced by 155 Bluesky users

17 readers on Mendeley

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Request permissions

WHAT IS ALREADY KNOWN ON THIS TOPIC

Existing research has found decreases in total injury numbers, and no evidence of boundary road impacts, associated with Low Traffic Neighbourhoods (LTNs), implemented (a) in the London Borough of Waltham Forest in 2015–2016 and (b) London-wide in 2020. However, due to small sample sizes and short follow-up periods, these studies did not have enough statistical power to exclude the possibility of modest impacts on boundary roads nor to measure impacts on Killed or Seriously Injured (KSI) injuries .

WHAT THIS STUDY ADDS

This comprehensive, well-powered study presents clear evidence of decreasing injury numbers, including decreasing KSIs, inside London’s LTNs. These benefits were observed regardless of casualty mode, sex or age, and are observed across a range of LTN characteristics. However, there was evidence that the benefit was smaller in LTNs implemented in Outer London since 2020. This study also finds no evidence of any overall increase in injuries (and for cyclists and motorcyclists a benefit) on boundary roads. This is reassuring given concerns that have previously been raised that LTNs could adversely affect safety on boundary roads due to traffic displacement.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Road traffic injuries remain a leading cause of death among children and young adults. This research indicates that LTNs could, alongside other measures, help London achieve its ‘Vision Zero’ commitment of zero deaths and serious injuries on its transport network in 2041. It is also an important contribution to both the national and international evidence base and has relevance for ongoing evaluations of area-wide traffic reduction schemes.

Introduction

Road traffic collisions are an important cause of serious injury and are the leading cause of death among 5–29-year-olds globally. 1 Perceptions of road danger deter people from walking 2 and cycling, 3 thereby reducing active travel and associated health and environmental benefits. 4

Like many cities, London has adopted a ‘Vision Zero’ commitment, aiming to eliminate all traffic deaths and serious injuries from its transport network by 2041. 5 Beyond its intrinsic value, city leadership views reducing road danger as a necessary precondition for achieving the target that 80% of all journeys should be made by walking, cycling or using public transport. Some progress has been made, with a 24% fall in Killed or Seriously Injured (KSI) casualties in 2023 compared with the ‘Vision Zero’ 2010–2014 baseline. 6 Nevertheless, pedestrians and cyclists still represent a high share of total KSIs, accounting for 57 fatalities (60% of all fatalities) and 2158 serious injuries (60% of all serious injuries) in 2023.

Low Traffic Neighbourhoods (LTNs) are area-wide schemes that restrict motor vehicle access to residential streets, aiming to reduce traffic and create safer, more pleasant environments for walking and cycling. LTNs have been implemented in the London Borough of Waltham Forest since 2015 as part of the ‘mini-Hollands’ programme 7 and were then implemented more widely across London during the COVID-19 pandemic in 2020. 8 The rate of implementation has since slowed, in part, as they are no longer being introduced as ‘pandemic emergency’ measures but instead follow a more standard design and consultation process. In addition, some LTNs implemented since 2020 have been modified to allow some through traffic and others have been removed altogether, typically after public protest or negative public consultation responses (Some ‘removed’ LTNs are reversed entirely while other ‘removed’ LTNs involve some measures being retained but, in our judgement, the area no longer meets the criteria for being an LTN. Note that LTNs are implemented by local authorities as trials, typically lasting 18 months. ‘Reversing’ an LTN therefore often involves deciding not to make the trial scheme permanent).

London’s LTNs vary in size (from under 7000 m2 to over 1 000 000 m2) and in how they restrict traffic—some use physical barriers, others use number plate recognition cameras as virtual barriers. In addition, the 2015–2019 Waltham Forest LTNs were more likely than the 2020–2024 London-wide LTNs to be accompanied by boundary road improvements (eg, new cycle tracks) and by internal permanent physical changes (eg, new planting, play areas or public seating in areas that used to be roadway).

LTNs reduce motor vehicle traffic volumes on residential streets, 9 which would generally be expected to reduce injury risk on those streets (see e.g., 7). However, LTN boundary roads—that is, the typically larger roads bordering new LTNs—see small average increases in traffic volumes (+1% mean, +4% median). 9 This has led to concerns that traffic injury numbers could rise on LTN boundary roads. Previous empirical studies have found substantial reductions in traffic injuries inside new LTNs implemented in Waltham Forest in 2015–2016 10 and implemented London-wide in 2020, 11 with no evidence of changes on boundary roads. However, these studies were limited by small sample sizes and short follow-up durations: moderate or modest changes in traffic injuries on boundary roads would plausibly not have been detected. Similarly, these studies were not sufficiently powered to detect LTN impacts on KSI injuries specifically, nor to examine whether LTN impacts vary across different types of road user or across different types of LTN.

We aimed to improve on these previous studies by providing a comprehensive and statistically well-powered assessment of the impacts of all LTNs implemented in London between January 2012 and June 2024.

Methods

Overview

Our study included all road links (sections of road between two junctions) in London. Across the years 2012–2024, some road links became inside an LTN or became part of an LTN boundary road, while others did not and remained in our control group. Our analysis estimated the before-versus-after change in injury numbers after the implementation of each LTN, using our control group to adjust for background changes in injury numbers over time.

Data sources

We used three sources of data:

Road injury data from January 2012 to June 2024 from the Department for Transport’s (DfT) 12 ‘STATS19’ Road Safety Data release. These data, as recorded by attendant police officers, included the easting-northing geographical location of collisions (The location of a crash is usually recorded as the point at which the first impact that resulted in an injury occurred (see 13 for more details), the road classification of the crash site (Motorway, A-road, B-road or Unclassified), and the age, sex, travel mode and injury severity of casualties.
University of Westminster 14 shapefiles (for Geographic Information System (GIS) software) of LTN areas and boundary roads implemented between September 2015 and June 2024, including date of implementation and, where applicable, removal. These LTNs are shown in figure 1. We know of no LTNs implemented in London between January 2012 and August 2015.
Ordnance Survey (OS) 13 MasterMap Highways Network—a road network file containing the geography and road classification characteristics of each road link.

Figure 1

Map of Low Traffic Neighbourhoods (LTNs) in London and their associated boundary roads (as of end June 2024).

Dataset creation

First, each road link in Greater London was spatially intersected with the LTN areas and boundary road lines (which had previously been drawn unmatched to OS roads). This determined, for each quarter of each year, whether that link was inside an LTN, on a boundary road, or neither. A road link could not be simultaneously inside an LTN and on a boundary road. For each quarter, a road link was defined as inside an active LTN if it had that status for at least two of the relevant 3 months (eg, an LTN implemented during January and present during February and March would be considered ‘present’ for that quarter). Otherwise, the link was defined as ‘pre-LTN control’ (if never an active LTN) or ‘removed’ (if ever previously an active LTN). The same approach was used for boundary roads.

Second, road traffic injuries were matched to the nearest OS road link with the same classification, if the injury to road link distance was ≤50 m. This was done separately for road classes ‘Motorway’, ‘A’, and ‘B’, and for a combined category of ‘C or Unclassified’. Injuries at junctions were assigned based on the road class designated ‘first’ in STATS19—the first road class is assigned based on the road where the crash occurred (or the road with priority, if at a junction), as recorded by the reporting police officer. 15 Where an injury was >50 m from the nearest road link of the same classification, but ≤100 m from the nearest road link of any class, it was matched to the nearest road of any class. Injuries >100 m from any road link were removed from the analysis.

We joined the (a) road links to injuries and (b) road links to LTNs datasets to generate a dataset for analysis. Figure 2 outlines the full process of data cleaning and dataset creation in more detail. Additional details can also be found in online supplemental material 1.

Supplemental material

[ip-2024-045571supp001.pdf]

Figure 2

Number of injuries and road links used in analysis. †Motorway road links excluded because motorways are a rare and atypical road type in London, and no motorways were ever inside an LTN or on a boundary road.

Statistical modelling

To estimate the change in injury rates after LTN introduction, we used a controlled before-and-after design with the rest of London included in the control group. Specifically, we fit conditional fixed-effect Poisson models, similar to earlier studies of road traffic injuries. 16–18 We fit the models to injury counts on each road link for each quarter of each year. By using road link as a fixed effect, we could compare injury rates on the same road links before and after LTN implementation. This removed the possibility of injury numbers being impacted by unobserved, time-invariant road characteristics (eg, road width).

To account for background trends in injury rates (unrelated to LTN implementation), we included terms for between-quarter changes in injury rates across all road links, including in the model the ‘control’ links that were always controls (ie, London road links that were never inside or on the boundary of an LTN). Due to the COVID-19 pandemic, these changes across time were not smooth, so unlike earlier studies, we made no assumption of smoothness. Instead, we allowed for step changes between year-quarters. As patterns of this background change differed by: (a) road type (A/B vs C/Unclassified) and (b) Inner vs Outer London, we fit interaction terms to account for patterns of between-quarter change in each of these four road classification-area combinations. We thus estimated changes in injury rates associated with the introduction of LTNs, allowing for changes that occurred on other similar roads. The coefficients for LTN and boundary road implementation were converted into relative percentage changes (eg, rate ratio 0.8 becomes −20%).

Our two primary outcomes were the total number of injuries of any severity by any travel mode and the total number of KSIs by any travel mode.

In total, 113 LTNs covering 40 km2 were included in the analysis. As of the end of June 2024, 27 of these LTNs had been removed (see figure 1). Road links inside or on the boundary of an LTN that was later removed were excluded from the main analyses. For our two primary outcomes, however, we included such road links in a separate analysis of the effect of moving into the category ‘LTN removed’, compared with the category ‘pre LTN’.

After running models for our two primary outcomes, we ran models stratified by:

Casualty travel mode (all modes and subdivided into pedestrians, cyclists, motorcyclists and other motor vehicle occupants). Travel mode was our pre-specified principal stratification of interest.
Casualty age (0–17 years, 18+ years).
Casualty sex (male, female).
LTN implementation timing and location (implemented in Waltham Forest 2015–2019; Inner London 2020–2024; and Outer London 2020–2024).
Road classification (‘A’ or ‘B’ (ie, trunk and distributor roads) vs ‘C’ or ‘Unclassified’ (ie, more minor, local roads)) (‘A roads’ are major roads accommodating substantial volumes of through-traffic. Almost all the Strategic Road Network in London is made up of A roads. ‘B roads’ are generally somewhat smaller and distribute traffic from A roads to minor roads. ‘C’ roads and ‘Unclassified’ roads are both minor roads intended for local traffic only).
LTN size (<50 000 m2, 50–99 000 m2, ≥100 000 m2).
LTN modal filter type (<25% modal filters in the LTN use physical barriers, 25%–75% of filters use physical barriers, >75% of filters use physical barriers. Non-physical modal filters use signs to prohibit motor vehicles, typically enforced by number plate recognition cameras).
Collision location distance from edge of LTN (≤100 m, >100 m). Only applicable to collisions inside the LTN.

In online supplemental material 1, further details of the analysis are specified in annotated core Stata code, and further information is given regarding model specification (including correcting for clustering and overdispersion) and sensitivity analyses.

Results

Table 1 shows the distribution of road links according to LTN status and also the pre-intervention injury numbers and rates. 98% of road links ever inside an LTN are on ‘C’ and ‘Unclassified’ roads (n=16 084/16 386), while 77% of boundary road links are on ‘A’ and ‘B’ roads (8000/10 396). It also shows that in the years 2012–2014, before any LTNs were implemented, injury rates were relatively similar between the ‘ever inside LTN’ and ‘always control’ groups after stratifying by road type. Injury rates were, however, higher on roads that would in the future become a boundary road ( table 1), and this generally remained true after stratification by Inner/Outer London status and after further disaggregation by road class (see online supplemental table 2.1). During these pre-LTN years, the total number of injuries on future boundary roads was seven times larger than on roads inside future LTNs. This highlights the potential for modest relative changes on boundary roads to have large absolute impacts, emphasising the need for this analysis to have sufficient power to examine boundary road effects.

View this table:

Table 1

Distribution of road classes by LTN and boundary road status, and pre-intervention injury rates in 2012–2014

Table 2 reports the impact of a road link becoming ‘inside an LTN’ or ‘on a boundary road’ on numbers of injuries. Overall, LTN implementation was associated with a 35% (95% CI 29% to 40%; p<0.001) decrease in all injuries and a 37% (95% CI 24% to 48%; p<0.001) decrease in the number of KSIs. On boundary roads, there was no evidence of an impact on the number of all-severity injuries (estimate −2%, 95% CI −5% to +2%) or KSIs (estimate 0%, 95% CI −7% to +8%).

View this table:

Table 2

Impact on numbers of road traffic casualties of becoming inside an LTN or of becoming an LTN boundary road

Further analysis (not tabulated) indicated that these beneficial effects of becoming inside an LTN were reversed entirely if an LTN was removed. The estimated effect of a road link being a ‘removed’ LTN (vs the pre-LTN control) was +3% (95%CI −12% to +20%) for total number of injuries and +11% (95%CI −25%, +64%) for KSI injuries.

In total, 331 injuries, including 44 KSIs, were observed on roads in a former LTN that had been removed. From the observed effect sizes, we can estimate that (331*0.35)=116 fewer injuries, including (44*0.37)=16 fewer KSIs, would have been expected to occur if the removed LTNs had instead been retained. Likewise, in total, 1139 injuries, including 171 KSIs, were observed on roads inside active LTNs. Our observed effect sizes suggest that the presence of these LTNs may have averted [1139*(1/(1–0.35))−1139]=613 injuries, including [171*(1/(1–0.37))−171]=100 KSIs. See online supplemental material 2 for sensitivity analyses around these estimates and for estimates of the number of injuries averted per km2 per year of LTN implementation.

Stratification by casualty travel mode indicated decreases in injuries inside LTNs for all modes, and generally, no evidence of impact on the boundary roads ( table 2). The exceptions were (a) a decrease in cyclist and motorcyclist injuries on boundary roads for injuries of all severities; and (b) a decrease in cyclist KSI injuries on boundary roads. In addition, the reduction in injuries inside LTNs was smaller for cyclists than for other modes.

To understand the relative contribution of these effects, we examined the combined intervention effect of becoming either an LTN or a boundary road ( table 3). We found that the total number of injuries by any mode fell by 7% (95% CI −10% to −4%) for all severities. For KSIs, there was a borderline-significant decrease of 6% (95% CI −13% to +1%, p=0.08). There was no evidence of any differences by mode, as shown by the p values for heterogeneity by casualty mode in table 3 (p=0.30 for all injuries, p=0.14 for KSIs). In other words, despite a smaller reduction in cyclist injury numbers inside LTNs, the overall benefit was similar for cyclists and for other modes because of the reduction in injuries on boundary roads.

View this table:

Table 3

Impact on numbers of road traffic casualties of becoming inside an LTN or of becoming an LTN boundary road, if these are combined into a single category

We also examined whether the impact of LTN implementation varied according to casualty, road, area and LTN characteristics (see table 4). There was no convincing evidence of variation with respect to casualty age, casualty sex, road type, LTN size, LTN filter type or (among within-LTN injuries) crash distance from the LTN boundary. There was, however, evidence that the effect varied in relation to timing and area of implementation. Specifically, the benefits of being inside an LTN implemented in Outer London since 2020 were notably smaller and only borderline statistically significant (point estimate −18% for all injury severities, p=0.03, with a similar trend seen for KSIs).

View this table:

Table 4

Impacts of LTNs on numbers of injuries by casualty, road, LTN and area characteristics

None of the analyses in table 4 found statistically significant evidence of an increase in injury numbers on boundary roads.

Discussion

Road traffic injuries fell by around a third inside new LTNs while remaining unchanged on new LTN boundary roads. These findings held true for both overall injuries and KSIs, with effects generally observed irrespective of casualty travel mode, age or sex, road class, LTN size or modal filter types. The most notable instances of variation of impact were that (a) cyclist and motorcyclist injuries reduced on boundary roads as well as inside the LTNs; and (b) benefits inside LTNs implemented post-2020 in Outer London were smaller than those seen for Inner London or earlier LTNs.

The smaller benefits of post-2020 LTNs in Outer London may in part reflect the fact that these schemes have on average experienced somewhat smaller reductions in motor vehicle traffic within LTNs, and somewhat larger traffic increases on boundary roads, as compared with Inner London LTNs. 9

Important strengths of this study include its comprehensive coverage of London LTNs and its extended follow-up, permitting well-powered examination of impacts on KSIs and on boundary roads. Other ongoing international evaluations of area-wide traffic reduction schemes have highlighted the importance of measuring road traffic injuries, 19–22 but none have yet published their results. This study therefore represents an important contribution to both the national and international evidence base.

One limitation of our study is that the STATS19 dataset does not capture all road traffic casualties in Greater London, as some non-fatal casualties are not reported to the police. 23 However, this is unlikely to bias our results, as there is no reason to believe the rate of reporting differs between casualties within LTNs and those outside LTNs.

A further limitation is that our approach captures changes in injury numbers rather than change in risk per trip. However, we can make some inferences from other data. A systematic review of local authority monitoring data found that, relative to background trends, motor vehicle volumes fell by 46% (mean)/32% (median) on LTN roads and increased by 1% (mean)/4% (median) on boundary roads. 9 Thus, the observed ≈40% decrease in motor vehicle injury numbers inside LTNs may not correspond to any change in risk per motor vehicle trip.

Conversely, existing evidence suggests that residents within LTNs tend to increase the amount of walking and cycling. 24 Therefore, it is plausible that the observed reduction in injury numbers inside LTNs corresponds to even more favourable changes in injury risk per trip. This aligns with findings that LTN implementation enhances residents’ perceptions of local road safety, regarding cycling in general and walking safety for young children. 25

Regarding the reduction in cyclist injury numbers on boundary roads, we do not believe this is likely to reflect boundary road infrastructural improvements (eg, cycle tracks). The post-2020 LTNs were not typically accompanied by such measures, and it is these LTNs rather than the pre-2020 Waltham Forest LTNs that have a trend towards the most favourable changes in boundary road injuries. One possible contributing factor is cyclist diversion, with some cyclists who used to cycle on the boundary road now travelling through the LTN instead. This is unlikely to be the full explanation, however, as the limited available evidence points towards an increase in cyclist numbers inside LTNs without any reduction in volumes of cyclists on boundary roads relative to background trends. 11 It is plausible, therefore, that part of the observed decline in cycle injuries reflects reduced risk per trip on boundary roads, perhaps due to fewer motor vehicles turning into and out of the LTN side roads. This explanation is consistent with the fact that a reduction in total injury numbers (although not KSIs) was also seen for motorcyclists, who, like cyclists, are particularly vulnerable to collisions caused by motor vehicles turning across them at T-junctions. 26 27

It is reassuring that our study found no evidence of increased injury numbers on boundary roads. Nonetheless, in the pre-implementation period (2012–2014), there were six injuries on future boundary roads for every one injury on roads inside future LTNs. As a result, the 35% decrease in total injuries inside LTNs translated to only a 7% drop in total injury numbers when LTNs and boundary roads were pooled together. We therefore conclude that while LTNs have successfully reduced numbers of road traffic injuries and have probably reduced injury risk for pedestrians and cyclists, achieving Vision Zero will require additional complementary measures to reduce traffic crashes on surrounding major boundary roads.

Data availability statement

Data are available upon reasonable request. All data are available upon request to the authors.

Ethics statements

Not applicable.

Ethics approval

This research was conducted as part of ‘The Low Traffic Neighbourhoods in London: a mixed methods study of benefits, harms, and experiences’ project, funded by the National Institute for Health and Social Care Research (NIHR) (NIHR135020). Ethical approval for all aspects of the project was granted by the Westminster University Research Ethics Committee, ETH2021-0902.

References

↵
1. WHO

. Road traffic injuries. 2018. Available: https://www.who.int/news-room/fact-sheets/detail/road-traffic-injuries [Accessed 15 Aug 2024].

Abstract

Data availability statement

Statistics from Altmetric.com

Request Permissions

WHAT IS ALREADY KNOWN ON THIS TOPIC

WHAT THIS STUDY ADDS

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Introduction

Methods

Overview

Data sources

Dataset creation

Supplemental material

Statistical modelling

Results

Discussion

Data availability statement

Ethics statements

Patient consent for publication

Ethics approval

References

Footnotes

Read the full text or download the PDF:

Log in using your username and password