312

Abstract

Introduction: Hospitals vary in their use of intensive care units (ICUs) for patients with diabetic ketoacidosis (DKA). At Beth Israel Medical Center, insulin infusions for DKA are only permitted in the emergency department (ED) or in an ICU. DKA patients must, therefore, remain in the ED until anion gap normalization (followed by admission to a regular medical ward) or go to an ICU. We sought to assess the impact on length of stay (LOS) - of triaging DKA patients with anion gaps to the ICU versus triaging of patients with anion gap normalization in the ED to medical floors. Hypothesis: Patients admitted to the medical ICU for DKA have longer hospital LOS without appreciable changes in mortality. Methods: We conducted a retrospective analysis of all adults presenting to the ED with a primary diagnosis of DKA from November 1st 2006-October 31st 2007, and November 1st 2010-October 31st 2011. Demographics, co-morbidities, concomitant acute diagnoses, laboratory values and DKA severity were compared between patients triaged to the ICU and those sent to the floor after anion gap closure in the ED using t-test and chi square analyses as appropriate. DKA severity (0/1 (least)-IV (most) severe) was determined using initial HCO3, calculated osmolarity and anion gap. Hospital LOS (primary outcome) and ED LOS were compared for each grade of DKA severity using Spearman rank correlation and after adjustment for DKA severity using linear regression. Results: Our cohort included 112 patients; 61 (54.5%) admitted to the ICU and 51 to the floor. There were no statistically significant differences in gender, age and co-morbidities between groups. Patients who went to the ICU had higher finger sticks, potassium, leukocytes and anion gaps, with lower HCO3 (P<0.05). ICU use increased with DKA severity (grade 0/I 10.7% up to grade IV 84.6%, (p<0.05). Patients who went to the ICU had shorter ED LOS but longer hospital LOS (P<0.05). When stratified by DKA severity, ED LOS was shorter for patients who went to the ICU with DKA grades II, III and IV (P<0.05), but not for grade 0/I. Hospital LOS was longer for patients who went to the ICU with grades 0/I and IV DKA (P<0.05), but not for grades II and III. After adjustment for DKA severity, all patients who went to the ICU had longer hospital LOS (median (IQR): 5.5 (3.2,4.6) vs. 4.6 (2.8,3.2) days, p<0.001) and shorter ED LOS (8.5 (6.8,7.2) vs. 10.0 (7.2,8.5) hours, p<0.001). Hospital survival was 100% in both groups. Conclusions: Patients with DKA who go to the ICU in our institution have shorter ED LOS, but longer hospital LOS even after adjustment for DKA severity.