Does walking explain associations between access to greenspace and lower mortality?

Despite emerging evidence that access to greenspace is associated with longer life expectancy, little is understood about what causal mechanisms may explain this relationship. Based on social-ecological theories of health, greenspace has multifaceted potential to influence mortality but the potential alternative mediating pathways have not been empirically tested. This study evaluates relationships between access to greenspace, walking and mortality. Firstly, we test for an association between access to greenspace and self-reported levels of walking using a survey of 165,424 adults across England collected during 2007 and 2008. Negative binomial regression multilevel models were used to examine associations between greenspace access and self reported number of days walked in the last month, in total and for recreational and health purposes, after controlling for relevant confounders. Secondly we use an area level analysis of 6781 middle super output areas across England to examine if recreational walking mediates relationships between greenspace access and reduced premature mortality from circulatory disease. Results show clear evidence of better greenspace access being associated with higher reported recreational walking. There were between 13% and 18% more days of recreational walking in the greenest quintile compared with the least green after adjustment for confounders. Tests for mediation found no evidence that recreational walking explain the associations between greenspace and mortality. Futhermore, whilst the relationship between greenspace access and walking was observed for all areas, the relationship between greenspace access and reduced mortality was only apparent in the most deprived areas. These findings indicate that the association between greenspace and mortality, if causal, may be explained by mediators other than walking, such as psychosocial factors. Future research should concentrate on understanding the causal mechanisms underlying observed associations.