Per current guidelines, our institution advocates using basal-bolus insulin regimens for inpatient hyperglycemia. However, it is a challenging practice in the real-world setting. Hypoglycemia is an unwanted consequence and a national quality metric for hospitals. We sought to characterize the factors around use of basal insulin and hypoglycemic events (HE) in our hospital system for quality improvement and to develop preventative strategies.

Between April-June 2018, 4130 admissions receiving basal insulin (glargine, NPH, or Mixed 70/30, 75/25) were retrospectively reviewed for HE, defined as glucose <70 mg/dl. ICU and non-ICU patients were included. Insulin infusion or insulin pump therapy was excluded. Median age was 66 years, 48% were male and median A1C was 7.8%. In the entire cohort, 22% (923) had >/= 1 HE. Of those, 76% had 1-3 HE and 24% had >/= 4 HE per admission. The frequency of basal insulins were 75% glargine, 18% NPH, and 6% Mixed. Within 24 hours of the first HE, insulin dose was decreased in 42% receiving glargine, 66% NPH, and 59% mixed insulin by an average of 23%, 23% and 11%, respectively. Oral antidiabetic agents (OADs) were used concurrently in 16% of admissions with HE, of which sulfonylureas (SU) represented 43%, metformin 25%, and DPP-4i 26%. At the time of the HE, patients were on an oral diet in 73% of admissions, TPN or tube feeds in 6%, and NPO in 12%. In 75% of the HE, basal insulin was administered without the use of guideline-based hospital order sets and 83% of HE did not have an endocrinology consultation.

Our analysis demonstrates several key issues that need to be addressed around the recommended practice of basal insulin in the hospital setting. Among them, inadequate response in dose reduction after the first HE, underutilization of standardized order sets, concurrent use of OADs, particularly SU, add risk for HE. Endocrinology consultation is low, and the complicated nature of a basal-bolus insulin regimen as well as the many variables involved remain real world challenges.


A. Pandya: None. S. Nguyen: None. V. Saldivar: None. A.R. Sadhu: 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