(869) A Single-Center Randomized Controlled Trial of IV Fluid Content in Children With DKA in the PICU

Introduction: Diabetic ketoacidosis (DKA) is a life-threatening condition resulting from insulin deficiency that is treated with aggressive intravenous fluids (IVF) to correct the dehydration from hyperglycemia. Normal saline (NS) is most often used despite having supraphysiologic chloride levels which can cause hyperchloremia and metabolic acidosis. This study investigated if children with DKA treated with IVF containing lactated ringers (LR) will have a shorter duration of metabolic acidosis after hospital presentation than those treated with IVF containing NS.

Methods: A single center, double blinded, randomized controlled trial with two treatment arms of IVF containing NS versus LR during DKA management per the institutional protocol. Subjects were children ages 0-18 years with Type I DM admitted to the pediatric intensive care unit in DKA. Primary outcome is the duration of metabolic acidosis after hospital presentation defined as: serum ketone level less than 1 mmol/L, venous pH over 7.3 and serum bicarbonate level over 15 mmol/L.

Results: A total of 49 patients were enrolled: 25 in the NS and 24 in the LR arm. The study ended early due to national shortages of IVF additives. Treatment groups at baseline were similar in age, Hemoglobin A1C level, and proportion who were new onset diabetics (56% in NS, 58% in LR, p=0.869). Children in the LR arm received 1.97 liters per unit of body surface area of IVF and those treated with NS received 1.65 liters, p=0.390, imparting a median chloride load of 305,364 and 378,224 mmol, respectively. With median PICU lengths of stay of 0.97-1.1 days, there was no difference in days in the PICU or hospital nor hours of acidosis between arms, even after adjusting for age, new onset and HgbA1c. Subjects with new onset Type 1 diabetes were 9.5 hours faster (95%CI 2.1 to 16.9, p=0.014) in achieving a bicarbonate over 15mEq/L. In the LR arm, every mmol of chloride given extended the time on an insulin infusion by 0.15 hours (95%CI 0.074-0.232, p=0.001)

Conclusions: This study was limited by early termination due to additive shortage. However, the data suggest that LR for intravenous rehydration provides a lower chloride load and less dependence on IV insulin.