Background Walking to school is a way to increase daily physical activity; however the risk of injury must also be considered so that walking does not lead to an increase in pedestrian injuries. Risk factors associated with the environment around schools with high child pedestrian motor vehicle collision (PMVC) rates were examined.

Methods Child PMVCs from 2000–2013, ages 4–12 years, were mapped within elementary school attendance boundaries in Toronto, Canada. Case and control schools were defined as those with the highest and lowest quartile of PMVC rates, calculated using census data. Potential risk factors included built and non-built environment variables obtained from municipal data sources as well as via direct observational counts done in the spring, 2015, to measure the proportion of children walking to school. Logistic regression was used to compare case versus control schools stratified by geographic location (downtown vs inner suburbs).

Results The mean PMVC rate in case schools (n = 50) was 13.4/10,000/year and in controls (n = 50) was 1.75/10,000/year. Walking was not associated with high PMVC rates after adjustment for the built environment and school social disadvantage. Overall, lower residential (OR 0.56, 95% CI: 0.37, 0.86) and higher one-way street densities (OR 4.00, 95% CI: 1.76, 9.08), school crossing guards (OR 3.65, 95% CI: 1.10, 12.20) and higher social disadvantage (OR 1.37, 95% CI: 1.11, 1.70) were associated with high PMVC schools, Similar associations of high PMVC schools with built environment features were found in the inner suburbs; however, there was a stronger association with school social disadvantage downtown.

Conclusions Walking to school was unrelated to high PMVC rates after controlling for the built environment. The built environment and school disadvantage were associated with higher PMVC rates with possible differences by geographic location.