Exploring Walking Behavior, Access, and Health Outcomes: Insights from the All of Us Health Datasets

Abstract: This study presents the first application of the All of Us Research Program data to transportation research, as known by the authors. The All of Us program offers the largest and most diverse health dataset in the United States, providing a fresh perspective on understanding walking behavior, access, and health outcomes. After excluding participants who reported difficulty in walking, we included data from 17,034 participants in the analysis, covering 670 zip code areas.
Using multilevel models and multinomial logistic regression, we investigated the influence of individual- and community-level factors on walking behavior, accounting for zip code-level random effects such as average income and education levels. Our study uniquely combined health data, revealing how factors such as age, gender, perceived facility accessibility, perceived neighborhood safety influence walking and, in turn, health outcomes. In particular, we examined how walking time correlates with participants’ perceived physical and mental health, as well as their surrounding environmental factors.
After analyzing the overall model, we conducted a gender-based subgroup analysis. Specifically, we performed separate logistic regression on male and female subgroups to gain a deeper insight into how gender shapes neighborhood safety perception, walking behavior, and health outcomes. This analysis deepened our understanding of transportation equity by highlighting gender-specific influences.
Our findings reveal that walking behavior is affected by multi-level factors, including individual health perceptions and access to walkable areas. Notably, there are significant differences in the impact across different gender groups. Participants who rated their physical and mental health positively were more likely to engage in frequent walking, especially in areas with accessible facilities and lower safety concerns. These results emphasize the importance of considering gender-specific and health-related factors when developing policies that promote walking and improve public health. Our approach also demonstrates the potential of large-scale health data in advancing transportation research.