A large number of studies have focused on the associations between single built environment (BE) characteristics and physical activity (PA). Combinations of BE characteristics offer a more comprehensive approach to identify the BE-PA associations. We aimed to examine the BE-PA associations in a cohort of elderly Hong Kong Chinese. Between 2001 and 2003, 3944 participants (65-98 years of age) were recruited and followed for a mean of 7.8 years. BE characteristics were assessed via geographic information system. PA levels were obtained using the Physical Activity Scale for the Elderly questionnaire at baseline and three follow-ups. Latent profile analysis was first conducted to classify the BE characteristics, and linear mixed-effects models were then used to explore the longitudinal associations between the BE classes and changes in the PA levels. Three classes of BE were identified. Class 3 (characterized by greater green space and sky view factor) demonstrated a significant decline in household PA (β = -1.26, 95% confidence interval: -2.20, -0.33) during the study period, and a slower decline in walking PA (1.19 (0.42, 1.95)) compared with Class 2 (characterized by a greater proportion of residential land use). Our results indicate that BE patterns characterized by high green space and a sky view factor may help promote the walking PA level.
Keywords: built environment; built environment pattern; cohort study; latent profile analysis; physical activity.