In his letter to the Lancet in 1986, reprinted in this issue of the International Journal of Epidemiology (IJE), Katan described the idea of using data from genetic studies to test for a relationship between a quantitative intermediate phenotype and a disease in a way that is not distorted by confounding or reverse causality.1 Following the application of these ideas by other authors2–5 interest in the concept has grown, although it is still not widely understood. This important and novel method has the potential to improve the way that the quantitative phenotypes that underlie common diseases are investigated, so better informing public health interventions that alter the level of the phenotype in order to reduce the risk of disease.4