Author(s): Schwartz, Rothman, El Amiri, Cloutier, Hagel, Macarthur, Nettel-Aguirre, Winters, Macpherson, Fuselli, Howard Student Poster Competition: 1st Place
In Canada, road traffic crashes are a leading cause of death and serious injury among children and youth. Child pedestrians and cyclists are particularly vulnerable road-users as they have less experience navigating traffic environments and are of smaller stature and, therefore, are at higher risk of serious injury. Built environment and social factors are associated with child road-user collisions. Examining built and social environment correlates of child pedestrian and cyclist collisions across cities can help identify roadway environments that promote safety.
This cross-sectional study from the Child Active-transportation Safety and the Environment (CHASE) program of research examines associations between the social and built environment around schools and child vulnerable road-user (pedestrian or cyclist) motor vehicle collisions (VRMVC) in 5 Canadian municipalities/regions: Calgary, Laval, Montreal, Toronto, and Peel Region. This study will provide insight into the design of roadway environments in Canada that protect children and improve safety around schools.
All public elementary schools, excluding specialty schools or programs, were included across five cities (n=1030). Direct observation of active school transportation (AST) was collected at a subset of these schools from the CHASE study (n=389). VRMVC among children ages 1-12 between 2012-2018 were obtained from police reports. Covariates included built environment factors: i) traffic interventions (i.e., speed-humps, crosswalks, bike paths, one-way streets, traffic-calming features, and crossing guards per road km), ii) roadway environments (i.e., proportion of minor roads, intersection density, traffic signal density), iii) land-use features (i.e., multi-dwelling density, old home (prior to 1960s) density, proportion park land-use, proportion residential land-use), and iv) social environment features (After-tax low-income cut-off (ATLICO), and proportion recent immigrant - immigrated within five years) obtained from municipal data sources and the 2016 Canadian Census. These covariates were quantified for the areas within 1000m of schools.
All VRMVCs that occurred within 1000m of schools were assigned to schools to calculate a school ‘burden’ of VRMVC per year. Collisions that fell in multiple school buffer areas were assigned proportionally to schools based on inverse distance between the school and collision. All variables were analyzed descriptively, considering city-specific means and standard deviations (SD). Random effects negative binomial regression was used to estimate the association between the built and social environment and the outcome of school VRMVC burden per year among all schools and among the subset of schools with AST data, with city included as a random effect. City-specific models were also estimated. Here we focus on univariate results, as multivariable modeling is underway. The subset of CHASE schools were modeled separately to analyze associations between AST and school VRMVC burden as well as examine whether controlling for AST might change observed associations between the built and social environment and school burden of VRMVC.
Across all school areas 1,919 VRMVCs occurred, with 996 occurring around CHASE schools. The average VRMVC school burden across all schools was 0.27 per year (SD: 0.25); ranging from the highest in Montreal (mean 0.35, SD: 0.27) to the lowest in Peel Region (mean 0.13, SD: 0.15). Across all schools, in univariate models, higher school burden of VRMVC was associated with child population (Incidence Rate Ratio (IRR): 1.81, 95% Confidence Interval (CI): 1.67-1.95, per increase in 1000 children). Additionally, higher school VRMVC burden was associated with all traffic interventions with crossing guards showing the strongest association (IRR: 1.23, 95% CI: 1.17-1.29, per 10% increase per road km), ii) roadway environments, such as intersection density (IRR: 1.10, 95% CI: 1.04-1.16, per increase in intersection per road km) and traffic signal density, iii) land use features, including multi-dwelling density and old home density (IRR: 1.18, 95% CI: 1.12-1.24, per increase in 1000 homes per km2), and iv) social environment features, including proportion of people below ATLICO (IRR:1.29, 95% CI 1.21-1.37, per 10% increase) and proportion of recent immigrants (IRR: 1.85, 95% CI: 1.62-2.12, per 10% increase). Lower school VRMVC burden was associated with ii) roadway environment features, including the proportion of local roads (IRR: 0.91, 95% CI: 0.86-0.95, per 10% increase), and iii) land-use features, such as residential land-use (IRR: 0.97, 95% CI 0.94-0.99, per 10% increase) or park land-use. Among the subset of CHASE Schools, proportion of children using AST was associated with increased school collision burden (IRR: 1.22, 95% CI: 1.15-1.28, per 10% increase).
City-specific univariate models for all schools showed that several covariates had consistent associations with school burden. In most cities, higher school burden of VRMVC was associated with crossing guards, traffic signal density, proportion below ATLICO (except Peel Region), and the proportion of recent immigrants, while the proportion of local roads, and residential land-use (except Laval), were consistently associated with lower school VRMVC burden. Differences between cities were observed in associations between certain covariates and school VRMVC burden. For example, intersection density was negatively associated with school burden in Calgary, and Laval, but was associated with higher school burden in Peel Region and Toronto. In the subset of CHASE schools, more AST was significantly associated with school VRMVC burden in all cities, except for Laval.
This large multi-city study will provide a comprehensive understanding of built and social environmental correlates of child VRMVC in diverse urban settings across Canada. These preliminary univariate results indicate environmental characteristics that are associated with lower school VRMVC burden, including roadway environments (i.e., local roads) and land-uses (i.e., residential/ park land-use). This study also showed positive associations between AST and social environment features (i.e., low-income, recent immigration), and school VRMVC burden. Traffic interventions were associated with higher school VRMVC burden, though they are meant to reduce collisions. This is likely related to the cross-sectional nature of this analysis. Traffic intervention may be placed where there are more dangerous traffic circumstances. Previous work has also found that controlling for factors, such as population and housing density, allows for a clearer understanding of their association with collisions. Ongoing adjusted models will further clarify these relationships. Built environment correlates differed somewhat between cities, highlighting the need for local knowledge and interventions to address child VRMVC risk.