What Characteristics Modify the Relation of Neighborhood Walkability and Walking Behavior in Older Adults?

Background and objectives: Neighborhood walkability can influence walking behaviors in older adults. However, its associations of walkability with walking may differ by demographic, socioeconomic, health, social, and residential characteristics due to factors such as increased vulnerability to environmental factors or increased resilience.

Research design and methods: In a sample of older adults (n = 493, median age = 82 [range 78-89], 56% female, 31% Black), a walkability index was derived from audits of Google Street View images of participants' immediate neighborhoods. Walking was self-reported in the past week. Effect modifiers by demographic (age, race, gender, marital status), socioeconomic (education, income), health (gait speed, falls, knee pain, depressive symptoms, cognitive status, perceived energy), social (driving status, social engagement), and residential/neighborhood (residence type, neighborhood socioeconomic status [SES], population density) characteristics were assessed by interaction terms in logistic regression models, adjusted for gender, race, gait speed, prior falls, high depressive symptoms, currently driving, and cognitive status. When effect modification was suggested (p for interaction <.1), adjusted analyses of walkability with walking stratified on the effect modifier were conducted.

Results: In this sample, 59% walked in the past week and greater walkability was associated with greater odds of walking (odds ratio [OR] = 1.16, 95% confidence interval [CI]: 1.01-1.34). This association (p for interaction range 0.002-0.07) was present for those who were not currently married (OR = 1.35, 95% CI: 1.17-1.56), who reported knee pain (OR = 1.40, 95% CI: 1.14-1.72) or high depressive symptoms (OR = 1.30, 95% CI: 1.06-1.60), or who had declining cognitive function (OR = 1.30, 95% CI: 1.09-1.55).

Discussion and implications: High walkability may influence physical activity, particularly for those with vulnerabilities related to social, pain, and brain health characteristics. These results should inform neighborhood planning and targeted interventions for vulnerable older adults.