Ketoacidosis on SGLT2 inhibitors never stops being a risk

The edition in brief

Today's lead is a three-country registry study of 322,597 SGLT2 inhibitor treatment episodes in type 2 diabetes. Ketoacidosis occurred at 2.43 per 1,000 person-years, and while risk was highest soon after starting, it did not go away with time on treatment. The risk sits overwhelmingly in identifiable people: HbA1c 83 mmol/mol or above, malnutrition, BMI under 20, previous ketoacidosis, recent hypoglycaemia. Infection was the commonest precipitant. The practical message is that sick-day rules belong at every review, not only at initiation. HumAn-1 randomised 400 children and young people with type 1 diabetes in Bangladesh and Tanzania to insulin glargine or usual human insulin. At six months there was no difference in time in very low range or time in target range on blinded CGM. Both arms sat around 40% time in range, which says the limiting factor in these settings is not the insulin molecule. A Bayesian meta-analysis of 1,776 people who stopped semaglutide or tirzepatide estimates regain of about 1 kg a month, with half the lost weight back by seven and a half months. A patient-level meta-analysis of 2,297 pregnancies found metformin does not prevent gestational diabetes, though it modestly prolonged gestation and reduced preterm birth. No new regulatory action for the desk today. The most useful recent guidance remains the EASO, EFAD and ECPO consensus on nutritional, functional and psychological care during incretin therapy. Today's pearl: once-weekly capillary ketone testing on well days predicts near-term DKA about as well as twice-weekly, using strips patients already hold.