Background Neighbourhood environments may influence risk of cardiovascular disease (CVD), via diet and physical activity (PA) behaviours. However, if the effects of CVD-related neighbourhood risks vary by socioeconomic position, efforts to improve population health by improving neighbourhood built environments may widen health inequalities. We examined whether associations between two neighbourhood characteristics – availability of PA facilities and fast-food store proximity – and CVD-related hospital admissions, were modified by income and area deprivation.

Methods 336,156 UK Biobank participants aged 40–70 years, linked to the UK Biobank Urban Morphometric Platform, were followed up through linked Hospital Episodes Statistics (mean follow-up=6.8 years). We examined whether associations between neighbourhood density of formal PA facilities and proximity of home address to a fast-food/takeaway store (at baseline), and hospital admissions with a primary diagnosis of CVD, were modified by household income or area deprivation (Townsend). We used Cox proportional hazards models, adjusted for likely confounding, and calculated relative excess risks due to interaction (RERI) to assess effect modification on the additive scale. We also examined the combined modifying role of income and deprivation.

Results Household income and area deprivation modified associations between neighbourhood exposures and CVD-related hospital admissions. Greater density of PA facilities may have a larger public health impact in more deprived areas (RERI=0.088), but high-income households benefit more than low-income households (RERI=-0.075). The estimated benefit was restricted to high-income households in deprived areas, where we observed 21% lower hazard of being admitted to hospital with CVD for people living <1km from at least four PA facilities than among people with no local PA facilities (HR=0.79, 95%CI:0.65–0.95). For fast-food proximity, reduced access to fast-food stores might have the biggest impact for low-income households (RERI=0.075), but mostly in less deprived areas (RERI=-0.104). A beneficial association was only observed among low-income households in affluent areas, where the hazard of CVD-related admission was 12% lower among people living ≥2km from a fast-food store than among people living <500m from one (HR=0.88, 95%CI:0.80–0.97).

Conclusion Among mid-life adults in the UK, associations between neighbourhood food and PA environments and hospital admissions for CVD varied according to household income and area deprivation. Results suggest that formal PA facilities may reduce CVD risk in deprived areas, but not among low-income households, raising important implications for health inequalities. Reducing access to fast-food stores may have greatest impact for low-income households but mostly in affluent areas. This may imply a greater range of alternatives to fast food in those areas.