Investigators linked nationwide registers in Sweden, Denmark and Norway to follow 322,597 SGLT2 inhibitor treatment episodes among 282,282 adults with type 2 diabetes, mean age 63 years. Over a median 1.3 years there were 1,452 ketoacidosis events, an incidence of 2.43 per 1,000 person-years. A nested case-control analysis, matched on age, sex, region of birth, calendar time and time since starting the drug, looked at who those patients were and what happened around the event.
The headline is that risk did not settle down. It was highest shortly after initiation, as expected, but persisted throughout follow-up. And it was concentrated in people who can be picked out in clinic. Compared with HbA1c 52 mmol/mol or below, an HbA1c of 83 mmol/mol or above carried an odds ratio of 15.37 (95% CI 11.50 to 20.53). Malnutrition gave an OR of 10.54 (7.32 to 15.18), previous ketoacidosis 10.40 (7.09 to 15.24), BMI under 20 kg/m2 compared with 20 to under 25 an OR of 9.98 (5.68 to 17.53), and recent hypoglycaemia 5.22 (2.60 to 10.49). Infection was the commonest precipitating or co-occurring event, present in 31.8% of cases against 6.1% of controls (OR 7.60, 6.62 to 8.71). Alcohol intoxication, acute renal events, acute abdomen, stroke and major surgery all clustered around events.
This matters in Indian practice more than the Scandinavian setting suggests. A BMI under 20 kg/m2 is not unusual in Indian adults with type 2 diabetes, undernutrition coexists with diabetes far more often here, and the infection burden that precipitates these events is higher. The phenotype the study flags as high risk is closer to the average Indian patient than to the average Swedish one.
The practical change is to make ketoacidosis risk a recurring question rather than a starting question. Ask about sick-day rules at every review of anyone on an SGLT2 inhibitor, and be explicit that the drug is paused during acute illness, before surgery, and when eating stops. What happened afterwards is also worth knowing: a year after ketoacidosis only 25.1% of patients were still on an SGLT2 inhibitor, and insulin use had risen from 31.9% to 73.0%.
- Re-check the risk profile at every visit: HbA1c, BMI, nutritional state, prior DKA, recent hypoglycaemia.
- Give written sick-day rules and repeat them — stop the drug with fever, vomiting, diarrhoea, or any reason not to eat.
- Hold the SGLT2 inhibitor before elective surgery and during any admission for acute illness.
- Keep ketone strips available for patients on SGLT2 inhibitors, and check ketones when unwell even if glucose looks acceptable.
- Be cautious about starting an SGLT2 inhibitor in a patient with BMI under 20 kg/m2 or documented undernutrition.
The statistics, in plain English
These are odds ratios from a case-control comparison, so they describe how much more common a factor was among people who developed ketoacidosis, not the chance any one patient will. An odds ratio of 15 with a confidence interval of 11.50 to 20.53 is a large effect measured precisely — the interval sits nowhere near 1.0, so chance is not a plausible explanation. The absolute risk stays low at 2.43 per 1,000 person-years, so these multipliers act on a small baseline. The authors caution that associations with short-lived events such as milder infections may be overstated, because those episodes are recorded less reliably in people who did not present unwell.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free