The SGLT-2 inhibitor ketoacidosis signal was maintained, not diluted, across expanding cardiorenal indications, with euDKA concentrated within this class and early onset, warranting ketone-based assessment when ketoacidosis is suspected particularly soon after initiation.
The DKA signal for SGLT-2 inhibitors met four-method consensus within every indication stratum, including heart failure and chronic kidney disease, and was not diluted as indications expanded.
Overall disproportionality reporting ratios for DKA with SGLT-2 inhibitors were large and statistically significant in both databases.
SGLT-2 inhibitors accounted for the large majority of all euglycaemic DKA reports in both databases.
Time-to-onset modelling indicated that ketoacidosis events associated with SGLT-2 inhibitors occurred relatively early after initiation, with a median onset of 60 days.
Both pre-specified negative controls produced null signals in both databases, supporting the specificity of the SGLT-2 inhibitor DKA finding.
The study used a cross-database disproportionality design that cannot establish incidence, relative risk, or causality, and findings are hypothesis-generating.
This research suggests that a known safety concern with SGLT-2 inhibitors — a dangerous condition called ketoacidosis, including a form where blood sugar appears normal (euglycaemic DKA or euDKA) — remains prominent even as these drugs are increasingly prescribed to patients with heart failure and chronic kidney disease, not just type 2 diabetes. The researchers analyzed two large databases of adverse drug event reports, one from the United States and one from Japan, and found that the signal for ketoacidosis with SGLT-2 inhibitors was strong and consistent across all patient groups studied. SGLT-2 inhibitors accounted for over 85% of all euDKA reports in both databases, even though euDKA made up only a small fraction of all reports linked to these drugs. One particularly important finding is that the risk appeared to be highest soon after patients started taking these medications, with a median time to onset of about 60 days and a statistical pattern suggesting the hazard decreases over time. The signal was also strongest in patients using these drugs off-label for type 1 diabetes. The fact that the same pattern emerged independently in both the U.S. and Japanese reporting systems adds confidence that this is a real pharmacological signal rather than a data artifact, especially since two pre-specified 'negative control' outcomes showed no signal, supporting the reliability of the methods used. This research suggests that clinicians and patients should be aware that euDKA can occur even when blood sugar levels look normal, particularly in the weeks following initiation of SGLT-2 inhibitors — and this applies not just to diabetes patients but also to those taking these drugs for heart failure or kidney disease, who may not have routine glucose monitoring. The authors recommend ketone-based testing (rather than relying solely on blood glucose) when ketoacidosis is suspected. Because this was a reporting database study, it cannot prove causation or determine how often these events occur in the broader population, and the findings are described as hypothesis-generating.
Chen J, Wang D, Duan X, Wang J, Zhao Z, Xing X. (2026). Disproportionate reporting of euglycaemic and diabetic ketoacidosis with SGLT-2 inhibitors across expanding cardiorenal indications: a cross-database study of FAERS and JADER.. Frontiers in endocrinology. https://doi.org/10.3389/fendo.2026.1957734