Recent data shows significant cardiovascular and renal benefit due to SGLT2 inhibitor (SGLT2i) therapy. However, use in the acute setting is limited and deferred due to concern of ketoacidosis especially in the setting of decreased nutrition intake and illness. Between 5/1/19 and 11/25/19 in one hospital, the inpatient formulary allowed the use of SGLT2i in patients with diabetes on the cardiology service as part of a supervised pilot program. As a result, 94 hospitalized patients were exposed to at least one dose of SGLT2i, 24 (25%) directed by the pilot program (PILOT), and 70 (75%) undirected (UND). In the UND group, 12 (13%) were newly started on SGLT2i and 58 (62%) were continued from home. Among all patients, 29.8 % were women, mean age 65, A1c 8.3% and GFR 67 mL/min. 17 patients (18%) had GFR ≤ 45, who were 5 years older on average (p= 0.04). In hospital there was no incident metabolic acidosis or clinically significant hypoglycemia. 79.8% of patients remained on SGLT2i at 1 month. 55 patients had data within 3 months post discharge. Of those still taking SGLT2i at 1 month, there was no significant change in systolic BP (SBP) at 1 or 3 months across all groups. In the UND group, there was no significant change in GFR. By comparison, in the PILOT group there was both a significant decrease in estimated GFR at 1 month (-6.95 mL/min, p= <0.01) and in SBP of -4.62 mmHg (p=0.04). Of all those newly started on SGLT2i, 6 had ≥ 25% reduction in GFR and moved up one CKD stage by 3 months; one moved from CKD stage 4 to 5. Among the patients with 3 month data, 40% were readmitted within 90 days, 0.3% for CHF, 36% other cardiac and 6% infection. Baseline GFR did not affect readmission risk.

In conclusion, permitting use of SGLT2i in the acute setting to select populations is likely to increase overall exposure of the drug to the general inpatient diabetes population. While we did not identify ketoacidosis, which is rare, reduction in GFR is common. Substantial education and direction is needed for inpatient clinicians in both the inadvertent and intended use of SGLT2is.


N.E. Palermo: None. M. Goodberlet: None. G. Cromwell: None. C.M. Smith: None. A. Bilodeau: None. P.M. Szumita: None. M.E. McDonnell: None.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at