55

Abstract

Introduction: Insulin infusion therapy (IIT) is commonly used in the hospital setting to manage diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic state (HHS). Clinical evidence suggests that both hypoglycemia and glycemic variability experienced while on insulin infusions negatively impacts patient outcomes. To address these concerns our institutional DKA protocol was revised to include evidence-based targets for glycemic control and to emphasize wide safety thresholds via use of an online infusion rate calculator embedded within the electronic medical record. Extensive, targeted outreach efforts were also employed to increase staff education about the new protocol. The aim of this study was to characterize the impact of a conservative-correction DKA protocol on the prevalence of hypoglycemic events, glycemic variability, and hospital length of stay. Methods: This single-center, retrospective cohort study incorporated a before-after design and included all adult patients admitted with a DKA or HHS diagnosis and initiated on a continuous insulin infusion. Outcome data was collected on 100 subjects in the pre-protocol group (October 2010 – October 2012) and 38 subjects in the post-protocol group (March 2013 – July 2013). The primary endpoint was hospital length of stay (LOS) in hours. Other endpoints included mean absolute glucose change (mg/dL/hr), hypoglycemia (blood glucose < 70 mg/dL), severe hypoglycemia (blood glucose < 40 mg/dL), and ICU LOS. Sample size was determined assuming a 25% reduction in hospital LOS with an alpha of 0.05 and power of 80%. Descriptive statistics, Fischer’s exact test, and the paired student t-test were used. Results: A total of 138 subjects with 3187 blood glucose readings were evaluated. Seventy-five (54.3%) of these subjects required admission to the ICU. There was a significant reduction in the risk of hypoglycemia (35% vs. 2.6%) and severe hypoglycemia (4% vs. 0%) when compared to the pre-protocol group. The average number of extreme glucose variability episodes, defined as mean absolute glucose change > 100 mg/dL/hr, were also notably decreased after protocol implementation (3.2 per patient vs. 1.7 per patient). Hospital and ICU length of stay were reduced by 20% and 27% respectively, suggesting that a conservative-correction approach does not delay ketosis resolution or time to discharge. Conclusions: This preliminary data support the hypothesis that a conservative-correction approach to DKA management is safe and effective in minimizing the risk of hypoglycemia and glycemic variability without impacting hospital length of stay.