Attributes of environments supporting walking

Purpose. This study established a framework to audit environments supporting walking in neighborhoods. Design. Cross-sectional analysis using a telephone survey and 200 objective environmental variables. Setting. Urbanized King County, WA. Subjects. 608 randomly sampled adults. Measures. Walking measures constructed from survey questions, objective environmental measures taken from parcel-level databases in Geographic Information Systems. Analysis. Multinomial models estimated the odds of people engaging in moderate walking (< 149 min/wk) and in walking sufficiently to meet recommendations for health (150+ min/wk), relative to not walking; and in walking sufficiently, relative to walking moderately. A base model consisted of survey variables, and final models incorporated both survey and environmental variables. Results. Survey variables strongly associated with walking sufficiently to enhance health included household income, not having difficulty walking, using transit, perceiving social support for walking, walking outside of the neighborhood, and having a dog (p < .01). The models isolated 14 environmental variables associated with walking sufficiently (pseudo R-2 up to 0.46). Measures of distance to neighborhood destinations dominated the results: shorter distances to grocery stores/markets, restaurants, and retail stores, but longer distances to Offices or mixed-use buildings (p < .01 or .05). The density of the respondent's parcel was also strongly associated with walking sufficiently (p < .01). Conclusions. The study offered valid environmental measures of neighborhood walkability.