Interrelationship Between Baseline HbA1c, SGLT-2 Inhibitor Use and Risk of Diabetic Ketoacidosis in Adults With Type...

There is ongoing uncertainty about whether initiating sodium-glucose co-transporter 2 inhibitor (SGLT-2i) therapy in individuals with type 2 diabetes (T2D) who have elevated baseline HbA1c levels may lead to an additive or potentially synergistic increase in the risk of diabetic ketoacidosis (DKA). This systematic review and meta-analysis aimed to investigate the interrelationship between baseline HbA1c, SGLT-2i use and the risk of DKA in adults with T2D.

Observational cohort studies and randomized controlled trials (RCTs) comparing SGLT-2is with placebo or active comparators in adults with T2D that reported baseline HbA1c and DKA events were identified through searches in MEDLINE, Embase, CENTRAL, ClinicalTrials.gov and bibliographies up to January 2026. Key characteristics of the study design, patients, interventions and outcomes were extracted. Risk of bias was evaluated. Risk ratios (RRs) were pooled and stratified by HbA1c (high vs. < low). Effect modification was assessed using random-effects meta-regression.

Twenty-two studies (15 cohorts, 7 RCTs) were included. SGLT-2is were associated with increased DKA risk overall. In observational studies, risk was higher in those with elevated HbA1c (RR 1.63, 95% CI 1.46-1.81) but not in those with lower HbA1c (RR 1.10, 0.80-1.51), with significant effect modification (p = 0.018). In RCTs, pooled RRs were 2.37 (1.44-3.90) and 2.01 (0.84-4.79) in high and low HbA1c groups, respectively, without significant interaction (p = 0.73). Elevated HbA1c was independently associated with DKA among SGLT-2i users (RR 1.50, 1.17-1.92). Evidence certainty ranged from high to low.

SGLT-2i use is associated with an increased risk of DKA in adults with T2D, with a higher risk observed in those with elevated baseline HbA1c, particularly in real-world settings. However, the narrow range of HbA1c values across studies limits definitive conclusions, and further research is warranted.

https://www.crd.york.ac.uk/PROSPERO/view/CRD420251149386: CRD420251149386.