Hypertens Res. 2026 May 1. doi: 10.1038/s41440-026-02656-y. Online ahead of print.
ABSTRACT
To prevent hypertension and cardiovascular diseases, many studies have aimed to develop a convenient formula for estimating 24-hour sodium excretion. However, previous formulas that use regression analysis based on spot urine samples have been criticized for several evaluation metric problems including adjusted coefficient of determination, correlation coefficient, mean bias, and systematic bias, and being not suitable for individual-level predictions. This study aimed to address these limitations by developing a new formula with higher predictive accuracy in a population. We hypothesized that more accurate predictions could be achieved by incorporating additional individual-specific information into the regression model. Specifically, we proposed that four refinements would improve prediction accuracy: accounting for sex differences, including voided volume, considering the time elapsed since the last meal at the time of spot urine collection, and categorizing participants into two groups based on predicted sodium salt excretion. To test these hypotheses, we provided low- and high-salt meals to 104 young healthy volunteers over 3 days. We collected all urine samples from the participants and measured sodium and creatinine levels. Using these data, we refined the Tanaka formula to provide a more accurate estimate of 24-hour sodium excretion in young people. The four refinements significantly improved the performance of previous formulas, leading to better correlation between predicted and observed sodium excretion values, reduced systematic bias, and overall enhanced accuracy. Although our refined formula has improved accuracy in predicting the average sodium excretion of a population, there may be still room for improvement in the future.
PMID:42067680 | DOI:10.1038/s41440-026-02656-y