841: THE IMPACT OF MILD-TO-MODERATE HYPOGLYCEMIA & NUMBER OF HYPOGLYCEMIC EVENTS ON ICU LENGTH OF STAY

Abstract

An association was recently reported between in-ICU mild-to-moderate hypoglycemia (MH) and increased ICU length of stay (ICULOS), but the independent predictive role of the number of hypoglycemic events (MHE) is uncertain.

MH is a predictor of ICULOS, independent of the number of MHE.

We studied a retrospective cohort of patients in a Med/Surg ICU, requiring IV insulin infusion, targeting blood glucose (BG) 90-130 mg/dl, using computer-guided insulin dosing (CG) [GlucoStabilizer™] between 3/1/08-6/30/09. Exclusions were age <18 yrs, <10 BG values, diabetic ketoacidosis/hyperglycemic hyperosmolar state, & hospital readmission. MH was defined as BG?70 mg/dl. We examined ICULOS across minimal BG (Glumin) strata of >70 mg/dl, 41-70 mg/dl, and?40 mg/dl, and determinants of ICULOS. Kruskal-Wallis test and stepwise multivariate regression modeling were used.

Four hundred forty three patients met study criteria (overall mortality 19.6%). Patients had mean (SD) age 58.6 (15.4) yrs; 56% were male; 41% Hispanic; 56% had diabetes; median (IQR) ICU LOS 4.8 (2.8-9.5) days; mean (SD) of patients’ average BG 125.4 (25.5) mg/dl. One or more MHE occurred in 52% of patients (>2 MHE in 49% of MH patients). The median (IQR) ICULOS for Glumin of >70 mg/dl, 41-70 mg/dl, &?40 mg/dl, were 3.3 (2.0-5.8), 7.3 (3.5-13), & 9.5 (4.3-14.1) days, respectively (p <0.0001). However, MH was not a predictor of ICULOS, once the number of MHE was included in a multivariate analysis (OR = 2.5 per hypoglycemic event [95% CI 1.9-3.3; p = 0.024]).

MH was associated with increased ICU LOS in the present cohort. However, this adverse association was likely mediated by recurrence of MHE. The generalizability of these findings needs further examination in other patient populations, CGs, and glycemic targets.