Cycling infrastructure as a determinant of cycling for recreation and transportation in Montréal, Canada: a natural experiment using the longitudinal national population health survey.

Background: Cycling is associated with numerous health benefits. Evidence suggests that new cycling infrastructure leads to increases in cycling, though studies of network-level changes are lacking. The objective of this study was to determine the longitudinal effect of cycling infrastructure on cycling engagement among adults living in Montréal, Canada.

Methods: Using data from the National Population Health Survey (1994-2011), this study included adults who resided in the Montréal Census Metropolitan Area for a minimum of two survey cycles (N = 779). Outcomes included self-reported any cycling (transportation or recreation) and time in recreational cycling (minutes/week). Archival maps describing temporal changes in the cycling network for five-year intervals (1991-2011) were classified using the Canadian Bikeway Comfort and Safety Classification System (Can-BICS). Three cycling exposures were calculated from the centroid of each dissemination area: (1) distance to the nearest cycling path categorized by Can-BICS comfort-level (low, medium or high), (2) presence of cycle paths of each comfort level within distance thresholds (low = 321 m, medium = 623 m, high = 1790 m), and (3) density of cycle paths within a 1000 m buffer. Mixed effects logistic regression models estimated associations between cycling infrastructure and any cycling. Linear mixed effects models estimated associations between cycling infrastructure and time spent in recreational cycling.

Results: Over the study period, low- and medium-comfort cycle paths were more prevalent than high-comfort paths and cycling for recreation was more common than cycling for transportation. Exposure to high-comfort paths within an acceptable distance (< 1790 m) was associated with higher odds of any cycling (aOR = 1.28, 95% CI: 1.00-1.63). Cumulative exposure to medium-comfort paths within an acceptable distance (< 623 m) was associated with greater time spent in recreational cycling (β = 0.09, 95% CI: 0.03-0.16). Gender-stratified analyses suggested that cumulative exposures to low- and medium-comfort infrastructure within distance thresholds was associated with time spent in recreational cycling (low: β = 0.06, 95% CI: 0.00-0.12, medium: β = 0.13, 95% CI: 0.04-0.22,) among women. No significant effects were observed for distance to the nearest cycling infrastructure for either outcome. Density was not examined in models due to low variation with most buffers having no cycling infrastructure.

Conclusions: This research provides evidence that cycle paths, especially of higher comfort and safety, can promote cycling. Future work is needed to explore cumulative exposures to cycling infrastructure, taking into consideration connectivity of networks, integrated public transport, and accessibility to work.