Examining the association between genetic risk for depression, wellbeing and schizophrenia, and proximity to greenspace

Abstract Previous studies indicate that residing in areas with greater greenspace is associated with better mental health and wellbeing. It is unclear whether these associations reflect those with better mental health seeking out greener environments. To examine this we 1) test associations between depression, wellbeing and schizophrenia polygenic scores (PGS) with two greenspace measures in UK Biobank (N=238,306 and 293,922), 2) estimate multilevel-models (MLM), clustering individuals by local geography to investigate whether observed global effect estimates may be capturing between area differences and, 3) conduct one-sample Mendelian randomisation (MR) to estimate causal effects. Depression and schizophrenia PGS were associated with residing in areas with lower greenspace, whilst wellbeing PGS was associated with higher greenspace. Locally-clustered MLM demonstrated attenuation for the individual wellbeing PGS association and a reversal of effect for the schizophrenia PGS association. MR revealed evidence of a causal effect of increased depression liability on decreased greenspace. Our study provides evidence for a putative causal relationship whereby people with greater liability to depression may select into less green neighbourhoods. Our results also highlight the potential for apparently individual-level PGS effect estimates to be biased by contextual, between-area differences in outcome, which are not effectively addressed using traditional principal component adjustment.