Does a walking intervention work equally in low and high walkable areas? A large-scale incentive program in Japan

Abstract Background Research investigating whether walking promotion programs work equally at different levels of walkability needs to be scaled up. This study examined the associations of neighbourhood walkability attributes with adult participants’ withdrawal from an incentive-based walking promotion program, and changes in step counts among non-withdrawing participants. Methods We used data from 30,530 participants aged 40–74 years enrolled in an incentive walking program in Yokohama City, Japan, in which uploading step-count data was required to enter a raffle for rewards. The follow-up period was up to 48 months. Withdrawal was defined as not uploading step counts during the last six months of the follow-up. Neighbourhood walkability attributes included population density, intersection density, destination diversity, and their composite measure (the Japan Postcode-level Walkability Index). We fitted multilevel Cox proportional hazards models to examine the associations between walkability attributes and program withdrawal, and mixed-effects models to examine the interaction between walkability attributes and time on step counts. Results During the follow-up period (median: 28 months), 62% of participants withdrew from the program. Compared to those living in the lowest intersection density areas, those in areas with the highest, the second-highest, and the third-highest intersection density had 7% (95% CI: 0.88, 0.98), 6% (0.89, 0.99), and 7% (0.89, 0.98) lower hazard ratios of withdrawal from the program, respectively. The third-highest population density areas showed a 5% (1.00, 1.10) higher hazard ratio of withdrawal, relative to the lowest. Step counts, which exceeded 7,000 steps/day during the study period, declined between the first and second years but slightly increased from the second to third year. There were no statistically significant interactions between the neighbourhood walkability attributes and time in relation to changes in step counts. Conclusion Living in areas with higher street connectivity was associated with a higher retention in the program, whereas the program showed almost comparable effectiveness regarding changes in step counts for continuing participants living in areas with different levels of walkability. These findings suggest the need for initiatives to support adherence to walking-incentive programs, particularly among those who live in areas with less connected streets.