To assess whether 24-h urinary sodium excretion (24uNa) can be estimated from spot samples in adult patients who attend hospital clinics. Methods: A cross-sectional study with a development (284 patients) and a validation cohort (229 patients) was conducted at our hospital. A multivariate linear regression model was built which was compared with former models. Concordance analyses and comparison of the ability to correctly classify each patient against a prespecified uNa cut-off value of 130 mmol/24h were performed, assessed by the C-statistic. The model was well calibrated (slope [95%CI] in internal validation: 0.965 [0.947-0.987], showing good discrimination, and performed robustly in an external validation cohort (slope: 0.811 [0.675-0.946]). The mean bias between the measured and the estimated 24uNa by NaRYC was 24.85 mmol/24h [17.06-32.63]. The NaRYC had the highest values of Pearson coefficient (0.613 p<0.0001), accuracy (P30): 56.8%, and AUC-ROC: 0.822 [0.766-0.869] as compared to other seven equations. Although the mean bias of the results is quite acceptable, the variability observed in the 95%CI makes not recommend the general use of a spot as a substitute of the 24-hour urine in order to estimate the total urine excretion of Na in a single subject basis.
Read full abstract