[Comment] The long and winding road to precision diabetology

There is increasing interest in delineating the heterogeneity of diabetes to develop precision diabetology.1 In 2018, Ahlqvist and colleagues2 introduced unbiased phenotype-based clustering to subtype diabetes into five clusters using the Swedish All New Diabetics in Scania 1 (ANDIS1) cohort, which was validated in the German Diabetes Study3 and replicated in many cohorts worldwide. Now, in The Lancet Diabetes & Endocrinology, Olaf Asplund and colleagues4 report on the follow-up of ANDIS1 (median follow-up 9·63 years [IQR 4·05]) as well as on an additional 10 019 participants with newly diagnosed diabetes in ANDIS2 (median follow-up 2·83 years [2·76]).