Predictors of asymptomatic ketonemia in outpatients with diabetic kidney disease receiving SGLT-2 inhibitors: a cross-sectional study

Sodium–glucose cotransporter-2 (SGLT-2) inhibitors are now foundational therapy for diabetic chronic kidney disease (CKD), yet their ketogenic metabolic effects in this population remain incompletely characterised. While euglycaemic diabetic ketoacidosis is a recognised complication, the prevalence and clinical determinants of subclinical asymptomatic ketonemia in patients with diabetic CKD receiving SGLT-2 inhibitors have not been systematically evaluated. We aimed to determine the prevalence of asymptomatic ketonemia, defined by point-of-care capillary β-hydroxybutyrate (β-HB) positivity, and to identify its independent clinical predictors in this population.

This cross-sectional study enrolled 211 consecutive adult outpatients with diabetic CKD receiving SGLT-2 inhibitor therapy (empagliflozin or dapagliflozin). Capillary β-HB was measured under standardised fasting conditions using a point-of-care ketone meter; positivity was defined as β-HB ≥ 0.6 mmol/L. Univariate comparisons were performed using the Mann–Whitney U test, independent-samples t-test, χ² test, or Fisher’s exact test as appropriate. Given the low number of outcome events (n = 16), Firth’s penalised logistic regression was employed as the primary multivariable analysis to mitigate small-sample bias.

Asymptomatic ketonemia was detected in 16 of 211 patients (7.6%), all of whom were clinically asymptomatic at measurement. In Firth’s penalised logistic regression, heart failure (OR 4.46; 95% CI 1.48–13.51; p = 0.008), absence of statin use (OR 4.94; 95% CI 1.54–15.86; p = 0.007), and higher HbA1c per 1% increment (OR 1.35; 95% CI 1.01–1.81; p = 0.041) were independent predictors of ketone positivity. Systolic blood pressure showed a borderline inverse association (OR 0.97 per mmHg; 95% CI 0.93–1.00; p = 0.063). No ketone-positive patient was receiving thiazide diuretics (0% vs. 30.3%; p = 0.007). SGLT-2 inhibitor type was not independently associated with ketonemia. Findings were robust on sensitivity analysis with standard logistic regression (Hosmer–Lemeshow p = 0.217; Nagelkerke R²=0.234).

Asymptomatic ketonemia is present in approximately one in thirteen outpatients with diabetic CKD receiving SGLT-2 inhibitors. Heart failure, absence of statin therapy, and suboptimal glycemic control independently identify a higher-risk metabolic phenotype, while ketonemia appears unrelated to CKD stage. Although these findings are exploratory and require confirmation in larger prospective studies, they suggest that targeted ketone monitoring may be warranted in selected higher-risk patients.