|

Exposure-Adjusted Bicycling Crash Rates and Risk Factors in Seven European Cities

Author(s): Branion-Calles, Nelson, Winters, Gotschi

Slidedeck Presentation Only:

6C_Branion-Calles

Abstract:

Background/Context: Exposure-based crash rates for bicyclists are rare and prone to bias, due to underreporting of crashes and a dearth of disaggregated exposure data. The Physical Activity through Sustainable Transport Approaches project followed a dynamic cohort of over 10,000 individuals in seven European cities over two years, collecting self-report crash data, as well as exposure data in duration (Global Physical Activity Questionnaire (GPAQ)) over two years.

Aims/Objectives: We aimed to quantify crash rates and individual risk factors for a crash across seven European cities.

Methods/Targets: Overall crude crash incidence was calculated by combining recorded crashes with exposure data as the number of crashes per (i) 10,000 hours. We also calculated incidence rates by city and a range of sociodemographic, attitudinal, and built environment characteristics.
Bicyclist crash risk factors were explored using Generalized Linear Models with negative binomial error structures and logarithmic links. Important predictors were assessed using a step-wise procedure.

Results/Activities: Overall, the crash rate was 13.35 crashes per 10,000 hrs (95% CI, 12.10 - 14.71) or 1 crash per 749 hrs. We found that crash rates were highest in London and lowest in Orebro.
Adjusting for city and exposure, risk factors for a crash included being male, living in a neighbourhood of very high or very low building density, while protective factors included perceiving that bicycling was well regarded in one's neighbourhood. We found a non-linear relationship between exposure and crash risk, with the risk of a crash decreasing with increasing exposure.

Discussion/Deliverables: This study analysed prospectively collected data in a large cohort of bicyclists across seven European cities, providing a unique dataset with exposure-adjusted crash rates at the individual-level. We show that crash risks include city, neighbourhood and individual level factors.

Conclusions: We assessed crash rates across seven European cities and found London to be the riskiest city and Orebro to the safest. Our crash risk models show that individual crash rates decreased with increasing bicycling frequency and were altered by a mix of sociodemographic, attitudinal and built environment factors. Our findings demonstrate that one contributing factor to “safety in numbers” might be a lower number of inexperienced or infrequent bicyclists.