M116 - SMOKING AND NEUROTICISM: USING MENDELIAN RANDOMIZATION TO INVESTIGATE CAUSALITY

Abstract

Background Smoking is one of the leading modifiable risk factors for disability, disease and death. There is a well-documented association between smoking and neuroticism, with smokers reporting increased levels of neuroticism. However, much of this data comes from observational studies and we are unable to make any causal inference regarding this relationship. Methods Mendelian Randomization (MR) is a method of assessing causality using observational data through the use of genetic instrumental variables for modifiable risk factors. Recent GWAS have identified variants robustly associated with both smoking phenotypes and neuroticism. This enabled us to use a range of MR methods to attempt to unpick this relationship. We use publicly available summary statistics from these studies in addition to data from UK Biobank to investigate whether there appears to be a causal link between smoking and neuroticism. Results We found evidence of a modest genetic correlation between smoking initiation and neuroticism (rG=0.124, p=0.0078). However, we found no strong evidence of a causal relationship in either direction when using 2-sample MR or when looking at individual level data in UK Biobank. Within UK Biobank it was possible to stratify on smoking status, which meant that in addition to smoking initiation, we were able to look at the association between smoking heaviness and neuroticism. We found some evidence of a causal relationship from neuroticism to increased smoking heaviness (β=0.198, p=0.014). Discussion In conclusion, although much of the observed association between smoking and neuroticism appears to be non-causal, we found evidence of a modest genetic correlation, suggesting some shared genetic aetiology. We also found some evidence of a causal relationship from neuroticism to smoking heaviness, which would lend support to the self-medication hypothesis.