Calibration of 24-h urinary sodium excretion estimates from fasting morning urine in Chinese populations: a comparative analysis of 2 calibration models and a locally developed model

Abstract

BackgroundSpot urine collection offers a convenient alternative to the more cumbersome 24-h urine collection. However, the widely recognized estimation models, such as Tanaka and International Cooperative Study on Salt, Other Factors, and Blood Pressure (INTERSALT), have not been adequately adapted for widespread use in the general Chinese population. ObjectivesThis study was designed to evaluate the precision of the Tanaka and INTERSALT calibration models, alongside a locally Zhejiang Province-formulated model, in predicting 24-h urinary sodium (24-hUNa) excretion among the Chinese population. MethodsThe study comprised 1424 participants, aged 18–69 y, who provided both comprehensive 24-h urine and fasting morning urine samples. The researchers assessed the accuracy of the measured 24-hUNa against the estimates obtained from the Tanaka, INTERSALT, and Zhejiang models. This evaluation was conducted at both population and individual levels, employing a range of statistical techniques, including bias analysis, correlation coefficients, intraclass correlation coefficients, receiver operating characteristic curves, Bland–Altman plots, as well as relative and absolute difference calculations, and misclassification rates. ResultsThe measured average 24-hUNa excretion was found to be 165.7 ± 71.5 mmol/24-h. Notably, there was a significant deviation between the estimated and measured values for the Tanaka-adjusted model [−11.7 mmol, 95% confidence interval (CI): −16.7, −6.7 mmol/24-h], indicating a statistically significant difference. In contrast, the deviations for the INTERSALT-adjusted model (0.6 mmol, 95% CI: −4.2, 5.4 mmol/24-h) and the Zhejiang model (0.2 mmol, 95% CI: −4.6, 5.0 mmol/24-h) were nonsignificant. The correlation coefficients for the models were 0.303, 0.398, and 0.391, respectively, with the INTERSALT-adjusted and Zhejiang models showing superior performance at the population level. ConclusionsThe 3 evaluation models may serve as effective, low-burden alternatives for assessing urinary sodium levels in the population. However, to enhance the accuracy and reliability of predictions at the individual level, further repeated measurements are necessary to minimize measurement errors and augment the validity of the estimations.