Safe RESIDential Environments? A longitudinal analysis of the influence of crime-related safety on walking

Sarah Foster, Paula Hooper, Matthew Knuiman, Hayley Christian, Fiona Bull, Billie Giles-Corti

Numerous cross-sectional studies have investigated the premise that the perception of crime will cause residents to constrain their walking; however the findings to date are inconclusive. In contrast, few longitudinal or prospective studies have examined the impact of crime-related safety on residents walking behaviours. This study used longitudinal data to test whether there is a causal relationship between crime-related safety and walking in the local neighbourhood. Participants in the RESIDential Environments Project (RESIDE) in Perth, Australia, completed a questionnaire before moving to their new neighbourhood (n = 1813) and again approximately one (n = 1467), three (n = 1230) and seven years (n = 531) after relocating. Self-report measures included neighbourhood perceptions (modified NEWS items) and walking inside the neighbourhood (min/week). Objective built environmental measures were generated for each participant's 1600 m neighbourhood at each time-point, and the count of crimes reported to police were generated at the suburb-level for the first three time-points only. The impact of crime-related safety on walking was examined in SAS using the Proc Mixed procedure (marginal repeated measures model with unrestricted variance pattern). Initial models controlled for demographics, time and self-selection, and subsequent models progressively adjusted for other built and social environment factors based on a social ecological model. For every increase of one level on a five-point Likert scale in perceived safety from crime, total walking within the local neighbourhood increased by 18.0 min/week (p = 0.000). This relationship attenuated to an increase of 10.5 min/week after accounting for other built and social environment factors, but remained significant (p = 0.008). Further analyses examined transport and recreational walking separately. In the fully adjusted models, each increase in safety from crime was associated with a 7.0 min/week increase in recreational walking (p = 0.009), however findings for transport walking were non-significant. All associations between suburb-level crime and walking were non-significant. This study provides longitudinal evidence of a potential causal relationship between residents' perceptions of safety from crime and recreational walking. Safety perceptions appeared to influence recreational walking, rather than transport-related walking. Given the popularity of recreational walking and the need to increase levels of physical activity, community social and physical environmental interventions that foster residents' feelings of safety are likely to increase recreational walking and produce public health gains.