Fig. 4

Two-sample Mendelian randomisation results of the potential causal effect of drinks per week on glycaemic traits. Results from the main IVW two-sample Mendelian randomisation (MR) analysis of drinks per week on fasting glucose, fasting insulin and HbA1c and the sensitivity analysis results from MR Egger, SIMEX, weighted median, PRESSO and GSMR analyses. Results are the difference in mean fasting glucose (mmol/l), fasting insulin (pmol/l) and HbA1c (NGSP percent or equivalent) per 1 SD higher log-transformed drinks per week, with 95% confidence intervals (CI). Multivariable Mendelian randomisation analyses including body mass index for the outcomes of fasting glucose and fasting insulin are not presented here but effects for fasting glucose were attenuated with this approach, whilst results for fasting insulin were in line with the main IVW results. SNP single nucleotide polymorphism, IVW inverse-variance weighted, SIMEX simulation extrapolation, PRESSO pleiotropy RESidual sum and outlier, GSMR generalised summary-data-based Mendelian randomization.