Nonlinear Association Between Daily Mean Temperature and Daily Step Counts in Community-Dwelling Older Adults in Japan: The Kyotango Longevity Cohort Study

Ambient temperature has been associated with daily step counts, but evidence characterizing continuous, non-linear relationships in older adults remains limited, particularly in large cohorts with repeated daily measurements. This study examined the association between daily mean temperature and objectively measured daily step counts in the Kyotango Longevity Cohort, conducted in a region of Japan recognized for its high longevity. This cross-sectional analysis included 4344 person-days from 769 community-dwelling adults aged ≥65 years (mean age 74.0 years; 59.3% women). A generalized additive mixed model examined the non-linear association while accounting for repeated measurements and relevant covariates. A significant non-linear association was observed (effective degrees of freedom = 3.3; χ2 = 21.5; p < 0.001). Predicted step counts peaked at 14,941 steps/day (95% confidence interval, 14,567–15,350) at an estimated optimal temperature of 20.4 °C (95% confidence interval, 17.6–22.3) and declined toward colder and warmer conditions. Compared with the optimum, predicted step counts were 828 steps/day lower at −10 °C and 1448 steps/day lower at +10 °C. No significant asymmetry was observed. Daily mean temperature showed an inverted U-shaped association with daily step counts. These findings may inform climate-adaptive strategies to support physical activity and healthy longevity.