Abstract Rationale Hypoglycemia and hypokalemia are common complications during diabetic ketoacidosis (DKA) treatment that are associated with increased resource utilization and mortality. Data predicting which patients with DKA are most vulnerable to these complications are limited. We aimed to identify risk factors associated with treatment-related complications among adults with DKA to optimize future personalized treatment strategies. Methods We conducted a retrospective observational analysis of adult (18 years) emergency department (ED) encounters with DKA (beta-hydroxybutyrate BOHB ≥3 mmol/L, glucose ≥200 mg/dL, pH 7.3 and/or bicarbonate 18 mmol/L) between August 2015 - April 2024 at a quaternary medical center. Visits were included if a standardized DKA treatment order set was used that included fixed-dose intravenous insulin infusion (0.1u/kg/hr, no bolus), intravenous fluid, potassium, and dextrose supplementation (dextrose added if glucose ≤ 250 mg/dL), and long-acting subcutaneous insulin at time of DKA resolution. Inclusion criteria also required at least one BOHB measured ≤3 hours after insulin infusion start time. Outcomes included incidence of hypoglycemia (glucose 70 mg/dL) or hypokalemia (potassium 3.3 mmol/L) during the duration of insulin infusion. Separate bivariate and multivariable logistic regression analyses tested for predictors of hypoglycemia and hypokalemia; covariates included age, sex, race, weight, Charlson Comorbidity Index, diabetes type, duration of insulin infusion, and initial BOHB, glucose, and potassium. Results We included 856 encounters from 527 unique patients; mean age 40.2 years; 61% female; 28% Black; 76% type I diabetes; mean initial BOHB 8.8±3.2 mmol/L. Fifty-two (6.1%) and 156 (18.2%) patients experienced hypoglycemia and hypokalemia, respectively. After controlling for covariates, type II diabetes (versus type I) was associated with a statistically significant decrease in the adjusted odds of hypoglycemia (AOR 0.11 95%CI 0.02-0.6, p = 0.008), while all other covariates were not. Statistically significant predictors of hypokalemia included younger age (AOR 0.98 0.97-0.99, p=0.04), type II diabetes (versus type I) (AOR 1.75 1.0-3.0, p=0.04), higher initial BOHB level (AOR 1.1 1.05-1.2, p0.001), lower initial potassium level (AOR 0.4 0.2-0.5, p0.001), and longer duration of insulin infusion (AOR 1.01 1.00-1.03, p=0.01). Conclusions In this single center study of treatment of adults with DKA using a standardized intravenous insulin protocol, type I diabetes was associated with an 89% higher odds of hypoglycemia and type II diabetes was associated with a 75% higher odds of hypokalemia. These findings highlight an opportunity for diabetes type-specific DKA treatment strategies that aim to proactively mitigate rates of iatrogenic complications. This abstract is funded by: None
Clark et al. (Fri,) studied this question.