Prevalence and potential risk factors of hypokalemia in pediatric patients with diabetic ketoacidosis

Abstract

Aims To examine the local prevalence of hypokalemia in patients with diabetic ketoacidosis (DKA), both at pre-sentation and during treatment, and to investigate the potential risk factors leading to significant hypokalemia during treatment of DKA. Methods Retrospective review of 114 consecutive patient-epi-sodes. Univariate analyses were performed to study any difference in mean between the group with nadir of potassium (Kn) >= 3.0mmol/L from group with Kn < 3.0mmol/L for predictors concerning patients' demo-graphics, the baseline characteristics, the therapies for DKA (including average insulin infusion rate before Kn), and the pace of recovery from DKA. Predictors deemed statistical significant in univariate analyses were sub-jected to multivariate analysis. Results The period prevalence of hypokalemia at presentation and during treatment of DKA were 13.8% and 92.5% respectively. Univariate analysis showed patients who were younger, with lower mean body weight, lower mean plasma bicarbonate at presentation, lower mean serum potassium level at presentation, higher urine output per unit body weight (in the first 24 hours of admission), higher amount of potassium supplement given before Kn, shorter time lag of starting potassium supplements (as reference to time of start of insulin) and longer dura-tion of metabolic acidosis were independently associated with risk of developing Kn < 3.0mmol/L. Multivariate analysis showed that duration of metabolic acidosis was the sole risk factor for having Kn < 3.0mmol/L.