Effects of Neighbourhood Exposure Definition on the Association between Walkability and Transportation Physical Activity ( breakout presentation )

Abstract

Background The walkability of individuals’ neighbourhoods is believed to positively contribute to their participation in physical activity. Reported associations between walkability and physical activity, however, are highly variable. One explanation is that many studies only assess individuals’ home environments and do not account for their exposure to other neighbourhoods, potentially introducing bias into observed relationships. Aim Our aim was to assess how associations between neighborhood walkability and transportation physical activity vary when residential versus activity space definitions of individuals’ walkability environment are used. Methods University students attending schools in Toronto, Canada were sampled between September and December 2015 as part of the StudentMoveTO survey. Participants reported a one-day travel diary accounting for all trips taken over the previous 24 h, including travel mode (e.g. vehicle, public transportation, walking), destination location, and reason for travel. Transportation physical activity was defined as reporting any walking or cycling trips. Neighbourhood walkability (quintiles, Q5 highest, Q1 lowest) was assessed using a previously validated index shown to be associated with physical activity, obesity, and diabetes. Three definitions of individuals’ walkability exposure were used: (i) residential (neighbourhood around individuals’ homes), (ii) full activity space (average of all locations visited), (iii) restricted activity space (average of all routine neighbourhoods visited [home, work, school]). Information on potential confounders, including age, sex, area-level income, and self-selection into walkable neighbourhoods, was collected. Logistic regression was used to assess the adjusted association between walkability and physical activity. Sensitivity analyses accounting for spatial dependence between individuals were also performed. Results A total of 10415 individuals were included in the analyses (mean (SD) age = 23.8 (7.40), 67.6% female). Associations between walkability and physical activity were significant for all definitions, but varied in magnitude, with residential walkability exhibiting the weakest association (Q5 vs. Q1 OR =3.42, (95% CI: 2.78, 4.21)), compared with full (Q5 vs. Q1. OR = 8.96, (6.62, 12.41)) and restricted (Q5 vs. Q1 OR = 7.21, (5.54, 9.52)) activity space definitions. Conclusion Activity space walkability showed the strongest association with physical activity, compared with more restricted definitions. Exposure misclassification may contribute to the variability in built environment & health relationships.