Associations Between Minute-Level Smartphone-Derived Exposure to Walkability and Consumer Wearable-Derived Physical Activity in the US-based Nurses’ Health Study 3 Cohort 2018-2020

Background and Aim: Walkable environments have been linked to increased physical activity and improved health outcomes. However, most studies measure exposure based on residential address alone, not accounting for mobility patterns. We aimed to explore the association of smartphone mobility-derived walkability exposure and wearable-based physical activity in a group of US women. Methods: We quantified associations of minute observation level smartphone GPS-based walkability with daily Fitbit wearable accelerometry data. The NHS3 mhealth study followed 348 participants who completed four seasonal seven-day sampling periods. GPS points collected every 10 minutes were linked to a census tract-level walkability index calculated by a summed z-score of road intersection density from 2018 TIGER/Line shapefiles, 2018 Infogroup business density, and 2015-2019 5-year American Community Survey population density estimates. Mean Fitbit-derived steps per minute was calculated for each 10-minute period (n=639,364). Generalized Additive Mixed Models accounting for correlated data within individuals and enabling penalized splines were used to explore associations. We adjusted for GPS based socioeconomic status, daily GPS temperature and precipitation, satellite-based greenness in the 30m around each GPS point, and individual factors. Results: Mean minute-level walkability was 0.24 (SD = 3.07) and mean steps per minute was 7.04 (SD = 14.98). Associations were non-linear in nature, trending steeply upward and plateauing at the highest levels of walkability. We found an increase of 2.24 (95% CI: 1.40, 3.09) mean steps per minute for each 1-point increase in walkability scores from -1.66 to 5.50. Conclusions: We used data at fine spatial-temporal scales to present novel estimates on time-intensive longitudinal associations between walkability and physical activity, while accounting for within- and between-participant differences. Participants were more likely to be physically active in more walkable areas, suggesting improvements in the built environment may improve health outcomes downstream of physical activity. Keywords: Wearables, Walkability, Physical Activity, Mixed Models