Does potassium concentration measured on blood gas analysis agree with serum potassium in patients with diabetic ketoacidosis?

OBJECTIVE: The aims of this study were to define the maximum clinically acceptable difference between potassium concentrations on different samples and to determine the degree of agreement between potassium concentration measured on blood gas analysis and serum for patients with diabetic ketoacidosis (DKA).

METHODS: This project comprised two sub studies. In the first, 15 emergency physicians, intensivists and endocrinologists were surveyed and asked to mark on a line with markings at 0.5 mmol/L intervals, the maximum clinically acceptable differences (both above and below the 'true' value) between potassium concentration measured on different samples. The maximum clinically acceptable difference was calculated as the median of responses. The second study was a retrospective agreement study. Patients with an ED diagnosis of DKA were identified from a computer database. They were eligible for inclusion if they had both blood gas analysis including potassium concentration and serum potassium concentration and pH was less than 7.3. Data collected included potassium concentration on serum and blood gas samples, pH, serum glucose concentration and time of sample collections. Data were analysed using bias plot and Spearman correlation analyses.

RESULTS: The maximum clinically acceptable difference was defined as 0.5 mmol/L for both over and underestimation of potassium concentration. Fifty patients were studied with a median pH of 7.17 and median serum glucose of 29.5 mmol/L. Difference in potassium concentration between samples ranged from -0.9-2.9 mmol/L. 80% of sample pairs had a difference within the maximum clinically acceptable difference defined previously. The magnitude of difference between samples correlated with serum glucose (P = 0.0033, coefficient 0.41) but not with pH.

CONCLUSION: This study suggests that potassium concentration derived from blood gas analysis may not be an acceptable substitute for serum potassium concentration in patients with diabetic ketoacidosis, particularly at higher serum glucose concentrations.