To investigate whether the development of hypokalemia in patients with diabetic ketoacidosis (DKA) treated in the pediatric critical care unit (PCCU) could be caused by increased potassium (K⁺) excretion and its association with insulin treatment.

In this prospective observational study of patients with DKA admitted to the PCCU, blood and timed urine samples were collected for measurement of sodium (Na⁺), K⁺, and creatinine concentrations and for calculations of Na⁺ and K⁺ balances. K⁺ excretion rate was expressed as urine K⁺-to-creatinine ratio and fractional excretion of K⁺.

Of 31 patients, 25 (81%) developed hypokalemia (plasma K⁺ concentration <3.5 mmol/L) in the PCCU at a median time of 24 hours after therapy began. At nadir plasma K⁺ concentration, urine K⁺-to-creatinine ratio and fractional excretion of K⁺ were greater in patients who developed hypokalemia compared with those without hypokalemia (19.8 vs 6.7, P = .04; and 31.3% vs 9.4%, P = .004, respectively). Patients in the hypokalemia group received a continuous infusion of intravenous insulin for a longer time (36.5 vs 20 hours, P = .015) and greater amount of Na⁺ (19.4 vs 12.8 mmol/kg, P = .02). At peak kaliuresis, insulin dose was higher in the hypokalemia group (median 0.07, range 0-0.24 vs median 0.025, range 0-0.05 IU/kg; P = .01), and there was a significant correlation between K⁺ and Na⁺ excretion (r = 0.67, P < .0001).

Hypokalemia was a delayed complication of DKA treatment in the PCCU, associated with high K⁺ and Na⁺ excretion rates and a prolonged infusion of high doses of insulin.