386-P: Inpatient Hypoglycemic Rate Reduction through the Implementation of Prescriber Targeted Decision Support Tools

Abstract

Introduction: Hypoglycemia has been shown to increase mortality and length of hospital stay in patients with and without diabetes. The purpose of this study is to determine if the implementation of prescriber targeted decision support tools reduces the incidence of inpatient hypoglycemic events. Methods: This pre-post intervention study included adult patients who experienced an inpatient hypoglycemic event within 24 hours of insulin administration from October 2018 to December 2019. Pregnant women and patients receiving insulin infusions were excluded. Best Practice Alerts (BPAs) and banners were implemented in January 2019 to alert providers of patients at risk for hypoglycemia. Triggers for the BPAs and banners include use of sulfonylureas with reduced renal function, high weight-based basal doses, a previous episode of hypoglycemia during the admission, or concomitant use of fluoroquinolones and scheduled insulin. The primary outcome was to measure the incidence of hypoglycemia per 1000 patient days in pre- vs. post-intervention groups. Results: In total, 1186 hypoglycemic events in the pre-intervention group were compared to 5260 events in the post-intervention group. The implementation of EMR alerts significantly decreased hypoglycemia (glucose ≤ 70 mg/dL) from 22.91 events/1000 patient days to 18.27 events/1000 patient days (p<0.001). Severe hypoglycemia (glucose < 54 mg/dL) was not significantly reduced (1.65 events/1000 patient days vs. 1.79 events/1000 patient days (p=0.453)). The use of high weight-based basal insulin doses was reduced in the post-intervention patients with normal renal function, 23.0% vs. 16.7% (p<0.001), and patients with impaired renal function, 7.1% vs. 4.9% (p=0.017). Conclusion: The implementation of prescriber targeted decision support tools was successful in reducing the overall incidence of inpatient hypoglycemia. Disclosure R.Goldstein: None. J.M.Odom: None.