Multivariable Mendelian randomization to disentangle the alcohol 1 harm paradox.

Abstract

The alcohol harm paradox, whereby low socioeconomic position (SEP) groups experience greater alcohol-related harms despite reporting lower alcohol consumption, is yet to be fully understood through observational studies because key drivers are correlated and share similar confounding structures. Multivariable Mendelian randomization (MVMR) were conducted to estimate the direct causal effect of number of drinks per week (DPW) and years of schooling (YOS) on multiple health outcomes. Previously published genome-wide association summary (GWAS) statistics for DPW and YOS were utilised, and summary statistics were generated from individual-level data from UK Biobank (N = 462,818) for all health outcomes. Inverse variance weighted analyses demonstrated evidence for direct effects of DPW and YOS on liver diseases, mental and behavioural disorders due to alcohol, and stroke, indicating that increasing alcohol consumption increased the likelihood of outcomes whereas increasing years of education decreased their likelihood. There was also evidence for a direct effect of DPW on depression, anxiety, influenza/pneumonia, and heart disease. In contrast, there was evidence of a total, but not direct, effect of DPW on depression, influenza/pneumonia, epilepsy, and injuries when accounting for YOS. Although caution is required when interpreting these results due to weak instruments for alcohol, these results provide some evidence that the alcohol harm paradox is partially due to the protective effect of additional years of education, resulting in a reduced likelihood of higher SEP groups developing many alcohol-related outcomes. Replication with strong instruments would be necessary to draw causal inferences.