Associations of land-use diversity with walking for transport

Abstract

INTRODUCTION: Evidence is mixed on land-use diversity (LUD), which may provide a mix of different potential destinations for walking or cycling, as a determinant of physical activity. This study examined associations of walking for transport with LUD measures calculated accounting for spatial scale and excluding land-uses of low relevance to walking for transport. METHODS: The Physical Activity in Localities and Community Environment (PLACE) study collected questionnaire data from 2506 adults living in Adelaide on socio-demographics, perceived availability of destinations and walking for transport (minutes per day and sessions). The following four entropy-based measures of LUD were computed using Geographic Information Systems for 154 urban Census Collection Districts (CCDs): (1) the original PLACE LUD score, incorporating commercial, residential, recreational, industrial, and other land uses, (2) the original PLACE LUD score corrected for the size of the CCD, (3) a revised LUD score excluding recreational and other land uses and (4) a revised LUD score corrected for the size of the CCD. Generalized linear models with robust variance estimation (STATA) were used to test associations between decile rankings of the four LUD scores, walking and perceived destination availability, adjusted for other area characteristics and socio-demographic attributes. RESULTS: The original LUD score had no association with walking for transport in minutes per day (β = 0.003, 95% CI −0.61, 0.61) and in sessions (β = 0.01, 95% CI −0.01, 0.02). However the revised LUD was a significant predictor of walking for transport (β = 0.28, 95% CI −0.29, 0.85 for minutes, β = 0.02, 95% CI 0.01, 0.04 for sessions). After area corrections, significant associations became evident using both the original score (β = 1.06, 95% CI 0.33–1.78 for minutes, β = 0.04, 95% CI 0.03, 0.05 for sessions) and the revised LUD score (β = 0.75, 95% CI 0.22, 1.28 for minutes, β = 0.04, 95% CI 0.03, 0.05 for sessions). It was also found that revised LUD and area-corrected LUD showed stronger associations with perceived destination availability, compared with the original LUD score. CONCLUSIONS: Land-use diversity is associated with walking for transport, but associations can be obscured by imprecise measurement. Measures of land-use diversity at the Census Collection District level can be improved by accounting for the geographic dispersion of land uses within an area and by excluding land-uses that are not of theoretical relevance to walking for transport.