Comparison of three formulas to estimate 24-hour urinary sodium and potassium excretion in patients hospitalized in a hypertension unit

Abstract

Measurements of 24-hour urinary sodium (24hrUNa) and potassium (24hrUK) excretion are useful in hypertensives but 24-hour urine collection may be difficult or unreliable. We compared three formulas (Tanaka, Kawasaki, Pan American Health Organization [PAHO]) proposed to estimate 24hrUNa and 24hrUK based on spot urine measurements. We studied 382 patients admitted to a hypertension unit. Sodium, potassium, and creatinine levels were measured using standard laboratory methods in a morning urine sample, followed by 24-hour urinary collection. Agreement between estimated and measured 24hrUNa and 24hrUK was evaluated using the Pearson correlation and Bland-Altman plots. Measured 24hrUNa was 158 ± 75 mmol/d and 24hrUK was 54 ± 24 mmol/d. The correlation coefficient was r = 0.53 for estimated versus measured 24hrUNa, r = 0.69–0.73 for estimated versus measured 24hrUK (all P < .001). The mean bias for 24hrUNa was significantly smaller for Tanaka (10.5 mmol/d) and PAHO (11.5 mmol/d) compared with Kawasaki formula (−29.9 mmol/d). The mean bias for 24hrUK was significantly smaller for Kawasaki (7.3 mmol/d) and PAHO (8.3 mmol/d) compared with Tanaka formula (16.5 mmol/d). Using a single morning urine sample, we found the PAHO formula to be the best for predicting mean 24hrUK and 24hrUNa in hospitalized hypertensive patients. However, precision and accuracy of all the evaluated formulas was inadequate.