Page - 110 - in Biodiversity and Health in the Face of Climate Change

110 confounders mostly address variables of social status that include education of sur- vey participants or households, followed by income or employment status. For income, it was shown that a higher household income was found to be less likely for residents with psychiatric morbidity, and that this was less prevalent in greener neighbourhoods (Astell-Burt et  al. 2014). Significant associations between greenness and physical activity were identified in all income groups (McMorris et  al. 2015). Mukherjee et  al. (2017) showed that park size was significantly nega- tively associated with depression even when models were adjusted for confounders such as income, education or employment status, but also for age, gender, marital status and household composition. Similarly, Calogiuri et  al. (2016) and Nichani et  al. (2017) found that socio-economic inequality in mental well-being was sig- nificantly lower among respondents reporting good access to a green space com- pared with those who had less access. Other studies, however, could not find any relationship between health outcome and socio-economic or socio-demographic confounders (Calogiuri et  al. 2016; Nichani et  al. 2017) or showed that significant associations between green space and health outcome became non-significant after models were adjusted for confounders (Cusack et  al. 2017; Richardson et  al. 2017a). For example, Padilla et  al. (2016) showed that the significant association between green space and stress disappeared when models were adjusted for socio- economic confounders. A case study city in France showed a significant relation- ship between infant and neonatal mortality risk and level of deprivation, but could not clearly explain the link to urban green space (Kihal-Talantikite et  al. 2013) (Fig.  5.4). age sex/gender race/ethnicity marital status household composition population density 0 5 10 15 20 No. of articles Fig. 5.3 Socio-demographic confounders N. Kabisch