Researchers linked national registers in Sweden, Denmark and Norway covering 322,597 SGLT2 inhibitor treatment episodes among 282,282 adults with type 2 diabetes, mean age 63. Over a median 1.3 years, 1,452 ketoacidosis events occurred — an incidence of 2.43 per 1000 person-years. A nested case-control analysis then asked who those patients were.
The risk factors are not subtle. Compared with an HbA1c of 52 mmol/mol or below, an HbA1c of 83 mmol/mol or above carried an odds ratio of 15.37 (95% CI 11.50–20.53). Malnutrition gave an OR of 10.54, previous ketoacidosis 10.40, and a BMI below 20 kg/m2 an OR of 9.98 (5.68–17.53) against a BMI of 20 to under 25. Recent hypoglycaemia carried an OR of 5.22. Infection was the commonest precipitant, present in 31.8% of cases versus 6.1% of controls (OR 7.60). Alcohol intoxication, acute kidney injury, acute abdomen, stroke and major surgery all clustered around events.
Two things change practice here. First, the risk did not concentrate only in the first weeks after starting — it was highest early but persisted throughout follow-up, so a patient stable at six months is not past it. Second, the profile of the high-risk patient is thin, poorly controlled and malnourished, which is not the patient most prescribers picture when they think of an SGLT2 inhibitor.
That profile is common in Indian practice. A lean patient with an HbA1c above 9% and poor intake is exactly the person who may be started on an SGLT2 inhibitor for its cardiorenal benefit and is also the person most likely to present in ketoacidosis with a near-normal glucose. Sick-day rules are the intervention, and they need repeating at every visit, not only at initiation.
- Reassess ketoacidosis risk at every review, not just at initiation — the risk persisted well beyond the first months
- Treat low BMI, malnutrition, very high HbA1c and any prior ketoacidosis episode as strong reasons to be cautious about starting or continuing
- Give written sick-day rules: pause the SGLT2 inhibitor during acute illness, fasting, alcohol excess, or before major surgery
- In a patient on an SGLT2 inhibitor who is unwell, check ketones even when glucose is normal — euglycaemic ketoacidosis is the trap
- After an episode, plan the restart deliberately: only a quarter were still on the drug at one year, and insulin use rose from 32% to 73%
The statistics, in plain English
An absolute incidence of 2.43 per 1000 person-years is low — about 1 in 400 patient-years — so this is not an argument against a drug class with proven cardiorenal benefit. What the odds ratios do is redistribute that low average risk very unevenly. An OR above 10 for malnutrition or prior ketoacidosis means the risk in those patients is an order of magnitude above the average, which is what makes selection worthwhile. Note the authors' own caution: some associations with transient conditions may be inflated because milder illnesses go unrecorded in controls, and their exploratory work suggested several risk factors are general to type 2 diabetes rather than specific to the drug.
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