Here, serum [Na]1 represents the current serum [Na], and serum [Na]2 represents the future serum [Na]. All data sources pertaining to the input and output of Na, K, and water (including infusion, urine, food, drink, drainage [Na, K, and water], electrolyte repletion [Na, K], and insensible evaporation [water]) were collected to optimize the accuracy of serum [Na]2 predictions via Eq. (3) (Supplementary Fig.
After the diagnosis of hyponatremia, the output of Na, K, and water was calculated by examining spot urine and estimating the loss volume of urine and insensible excretion. We then calculated how much serum [Na] would change by taking Na, K, and water orally or intravenously and initiated treatment to achieve the target correction rate. To predict the future urine output of Na, K, and water, we used data from the most recent time point for urine [Na], urine [K], and the amount of urine flow, as we also concurrently performed spot urine tests for each blood test. In the case of diuretic therapy for hypervolemic hyponatremia, such as in heart failure, we estimated urine [Na], urine [K], and the amount of urine flow after diuretic administration and included them in the calculation. As it was challenging to predict urine volume at the beginning of the correction, we roughly estimated it. We considered the amount of water and sodium losses via perspiration negligible in typical cases; therefore, we did not take perspiration into account for the predictive correction38 ,39 (link). In cases with a sudden and substantial dilution of urine or rapid increase in serum [Na], desmopressin or hypotonic solution was administered at the discretion of the physicians to prevent overcorrection15 (link),16 (link),29 (link). The incidence of ODS, as well as the length of hospital stay and in-hospital mortality rate, were also evaluated.