Agreement between methods to assess potassium intake in patients with chronic kidney disease

In the general population, 24-h urine potassium excretion is considered the reference standard for estimating potassium intake. However, its agreement with food records and spot urine collections in adults living with chronic kidney disease (CKD) is not well-established. Given the risk of hyperkalemia related to changes in renal potassium handling, understanding if this reference standard is appropriate for the CKD population is important. This study aimed to compare agreement between methods for estimating potassium intake, using food records, 24-h urine measures, and spot urine samples.

A cross-sectional study was conducted among adults recruited from a kidney care clinic. Participants were instructed to complete a detailed food record and a 24-h urine collection, both performed on the same day. The following day, participants provided a spot urine sample from the second void, at the research site. Potassium excretion was estimated from the spot urine sample using Tanaka's and Kawasaki's equations. Agreement between methods was assessed using the highest p-value for paired t-test and the lowest Bland-Altman bias combined with the narrowest upper and lower limits of agreements (LoA).

60 adults with Stage 3 and 4 CKD completed the study (48 % male, 62.9 ± 14.6 years; eGFR = 34.9 ± 12.7 mL/min). The food records showed the highest agreement with 24-h urine potassium (1823.9 ± 746.7 vs 1918.2 ± 809.3, p = 0.584). This was followed by Kawasaki's equation for spot urine (1994.8 ± 441.9, p = 0.231) and Tanaka's equation (1630.0 ± 325.9, p = 0.174). Food records and Kawasaki's equation had slightly higher mean values compared to 24-h urine with mean differences <100-200 mg/day (bias; 95%CI: bias = -94.4 mg/day; -438.3 to 249.6 mg/day and -170.9 mg/day; -454.1 to 112.2 mg/day, respectively). Tanaka's equation had a lower mean value compared to 24-h urine with a mean difference of 193.9 mg/day; -88.5 to 476.3 mg/day). The limits of agreement were as follows: in the Kawasaki's equation from -2082.2 to 1740.3 mg/day, in the Tanaka's equation from -1712.2 to 2100.0 mg/day and in the food records from -2416.1 to 2227.4 mg/day.

Mean potassium intake estimates were similar across methods. Food records demonstrated the highest agreement with 24-h urine potassium, followed by Kawasaki's equation. The Tanaka's equation showed the highest bias compared to 24-h urine and was significantly different from food records. Combining food records with potassium excretion estimated using the Kawasaki's equation from spot urine samples may be a clinically useful tool for assessing potassium intake in adults with CKD.