Organisation
University College London
Amount awarded
£148,182
Completed
2024
Uploaded to Knowledge Centre
10 February 2025
While fatal and serious injuries to cyclists are rare, near-misses, where cyclists or drivers take action to avoid a crash, occur frequently. Systematic analysis of these incidents can help reveal the root causes of unsafe events and prevent future crashes. Our research aimed to use mobile and video technology to track near misses, offering detailed data to improve safety measures and evaluate interventions.
This study aimed to pinpoint the safest infrastructure for cyclists and identify risk factors based on cyclist type, behaviour and travel environment. The objectives were to:
• Define “near miss” through stakeholder consultation.
• Create a user-friendly, affordable, and unobtrusive system for cyclists to report near misses.
• Determine individual and environmental predictors of near misses for focused safety improvements.
• Connect near miss data with individual cyclist characteristics.
The researchers recruited 60 adult cyclists who commuted in London. Each participant completed a survey about their cycling experience, use of protective clothing, use of cycling Apps, collision experience and demographic characteristics. Each cyclist was given a helmet with a GoPro camera mounted on it. Participants were asked to say ‘near-miss’ when they experienced a near miss. Voice recognition was used to extract the timestamp of instances of the phrase, which were in turn used to extract near-miss events. Participants were asked to record any near misses that they experienced over a two-week period.
A descriptive analysis was carried out of near-miss events recorded by participants, focusing on event types, timing, rider interactions, and demographics. The most common near-miss types were close passes, followed by left/right hooks and pulling out incidents. Near misses mostly occurred during afternoon peak hours, with fewer incidents in the morning. Rider speed before incidents was typically below 30 kph, and no clear link was found between speed and near-miss type. Cars were most often involved, while traffic was generally light or moderate during events.
Most near-misses occurred on roads without cycling infrastructure, though left/right hooks were more common on paths or segregated lanes. Close passes mainly happened on non-intersecting roads, while left/right hooks occurred more frequently at intersections or side roads. Male participants recorded more near misses, especially close passes, compared to females. Females experienced more pedestrian-related near misses, while males were involved in more pulling out incidents. Males also tended to be cycling at faster speeds during near-misses.
To better understand risk, near misses were considered relative to a rider’s exposure. The rate of near misses per 100km traveled was calculated. The near miss rate ranged from zero to 54, with a median of 3. Demographics and riding behaviour were included as factors. The model revealed several significant factors: More time spent on 30mph roads and riding during the AM peak (7-10am) raised the near miss rate whereas increased riding speed and time spent on shared lanes reduced near misses. Additionally, participants who had not been involved in a crash in the past year reported fewer incidents.
Finally, a case control study was undertaken which considered participants’ exposure to different conditions. This compared the circumstances of video clips of near-miss events to a random sample of those with no reported near miss event.
To read the full report, visit The Road Safety Trust website:
https://www.roadsafetytrust.org.uk/funded-projects/17/ucl-near-miss-incidents-for-cyclists