Unravelling the complex causal effects of substance use behaviours on common diseases

Abstract

Abstract Substance use behaviours (SUB) including smoking, alcohol consumption, and coffee intake are associated with many health outcomes. However, whether the health effects of SUB are causal remains controversial, especially for alcohol consumption and coffee intake. In this study, we assess 11 commonly used Mendelian Randomization (MR) methods by simulation and apply them to investigate the causal relationship between 7 SUB traits and health outcomes. Smoking initiation shows widespread risk effects on common diseases such as asthma, type 2 diabetes, and peripheral vascular disease. Alcohol consumption shows risk effects specifically on cardiovascular diseases, dyslipidemia, and hypertensive diseases. By integrating stratified regression, genetic correlation, and MR analyses, we find evidence of dosage-dependent effects of coffee and tea intake on common diseases (e.g., cardiovascular disease and osteoarthritis). We observe that the minor allele effect of rs4410790 (the top signal for tea intake level) is negative on heavy tea intake (ƀGWAS = -0.091, s.e. = 0.007, P = 4.90 × 10-35) but positive on moderate tea intake (ƀGWAS = 0.034, s.e. = 0.006, P = 3.40 × 10-8), compared to the non-tea-drinkers. Our study provides novel insights into the complexity of the health effects of SUB and informs design for future analysis to dissect the causal relationship between behavioural traits and complex diseases.