Urban Residential Greenness and Adiposity: A Longitudinal Cohort Study in Stockholm County

Abstract

Background/Aim: This longitudinal cohort study, using objectively measured exposure and outcomes, aims to explore the association between residential green space and body fat distribution, potentially mediated via psycho-endocrine mechanisms. Methods: Based on the Stockholm Diabetes Prevention Program (SDPP), this study included 5 355 subjects from Stockholm County aged 35-56 years at recruitment in 1992-1994, and followed up 2002-2006. We defined surrounding greenness as a time-weighted average of satellite-based Normalized Difference Vegetation Index (NDVI) at 30 m×30 m resolution during 1990-2015 in buffers of 100 m, 250 m, and 500 m around participants’ place of residence. Multiple linear regressions with adjustment for potential confounders at individual and contextual level, including behavioural, socio-economic and environmental factors (traffic-related air pollution and noise), were used to estimate the change in BMI, waist circumference and waist-hip ratio. Results: Waist circumference increased on average by 0.51 cm/year in this population, and time-weighted NDVI in 500 m buffers ranged 0.2-0.7. Higher levels of surrounding greenness were associated with reduced increase in waist circumference, with 0.78 cm less increase in waist circumference per year for one unit difference in NDVI, (95% CI 0.16-1.41). This is equivalent to an average annual increase of 0.60 cm among the 10 percent least exposed to greenness, compared to 0.43 cm among the 10 percent most exposed. A modest but yet statistically significant association was observed for the waist-hip ratio, but no significant association was found between NDVI and change in BMI. Using the 100 m and 250 m buffers generated less clear associations than for 500 m. Conclusions: Our findings suggest a beneficial impact of surrounding greenness on waist circumference and waist-hip ratio. The absence of a statistically significant relationship between NDVI and BMI suggests that the effect of NDVI may be mediated by the cortisol-induced differences in body fat distribution.