Development and Validation of Equations to Estimate 24-H Urinary Sodium Excretion from Urine Samples of Patients with Chronic Kidney Disease

Abstract

Introduction: The assessment of sodium intake is difficult due to low accuracy of dietary records and to the inconvenience of 24-h urine collections. Therefore, equations based on spot urine samples have been proposed to estimate sodium intake. In this study, we aimed to develop and to validate equations to estimate 24-h urinary sodium excretion (24hUNa) from several urine samples in chronic kidney disease (CKD) patients. Methods: Cross-sectional study with 76 CKD patients (males 55.3%; age: 64.5 [56.0–69.0] years; glomerular filtration rate 27.8 [24.7–32.1] mL/min). Sodium excretion was measured in 12-h daytime and 12-h nighttime collections; spot 1 (first urine of the day) and spot 2 (second urine of the day). By multivariable linear regression analysis, 4 equations were developed. The equations’ accuracy was evaluated by P30 test. Association between estimated and measured 24hUNa was assessed by intraclass correlation coefficient (ICC); mean differences and limits of agreement by Bland-Altman plot. Data from 51 CKD patients of other CKD outpatient clinic were used to validate the equation developed from spot 2. Results: The 4 equations showed significant (p < 0.001) ICC and relatively good accuracy when compared to 24hUNa (Daytime: ICC = 0.89; P30 = 84%; Nighttime: ICC = 0.90; P30 = 83%; spot 1: ICC = 0.85; P30 = 78%; and spot 2: ICC = 0.70; P30 = 76%). In validation set, the equation from spot 2 was moderately accurate (P30 = 67%). Mean bias and ICC were 19.9 mmol/day and 0.58 (p = 0.001), respectively. A high sensitivity (97%) and specificity (89%) were found for a cutoff of 3.6 g of sodium/day. Conclusion: Equations derived from 12 h collections better performed than spot urine when compared to gold standard 24hUNa. The equation from spot 2 showed good sensitivity to identify excessive sodium intake.