Urban nature and physical activity: Investigating associations using self-reported and accelerometer data and the role of household income

Background: Physical inactivity is a major public health concern. Natural, or semi-natural, environments may encourage physical activity, but the influences of socio-economic factors have been under-researched. Methods: We explored the associations between meeting physical activity (PA) guidelines and both neighbourhood green (area coverage) and blue (freshwater coverage and coastal proximity) environments for urban adults using data from the Health Survey for England [HSE] (2008/2012). We considered different domains of selfreported PA: walking (n = 18,391), sports and other exercise (n = 18,438), non-recreational (domestic/gardening/occupational; n = 18,446) and all three domains combined (n = 18,447); as well as accelerometerderived PA data using a subsample (n = 1,774). Relationships were stratified by equivalised household income as an indicator of socio-economic status. Results: After adjusting for covariates, living <5 km from the coast was associated with significantly higher odds of meeting UK 2010 guidelines through self-reported total, walking and non-recreational PA (e.g. total PA, <5 km vs. >20 km, adjusted odds ratio (ORadj) = 1.26; 95% confidence interval (CI) = 1.15-1.39) but unrelated to sports and exercise. Greater neighbourhood greenspace, however, was only associated with significantly higher odds of meeting guidelines through non-recreational PA alone (e.g. 80-100% vs. <20% ORadj = 1.32; 95% CI = 1.12-1.56). Although associations were most consistent in the lowest income quintile, income-related results were mixed. Relationships were not replicated in the smaller accelerometry subsample. Conclusion: Our self-report findings for the differing domains of PA as a function of neighbourhood green and blue space broadly replicated previous research, yet the reasons for the observed differences between PA domains and environments remain unclear. We did not observe any associations between environmental variables and accelerometer-measured PA; further research with larger samples is needed.