Mendelian randomization analysis supports the causal role of dysglycaemia and diabetes in the risk of coronary artery disease.