Spot urine sodium-to-creatinine ratio surpasses sodium in identifying poor diuretic response in acute heart failure.

AIMS

We aim to identify the most accurate marker for early prediction of poor diuretic response in acute heart failure (AHF) patients with signs of congestion requiring intravenous diuretic treatment.

METHODS

In this single-centre, prospective observational study, AHF patients with signs of congestion received a standardized intravenous furosemide dose (1 mg/kg of body weight; 40 mg in bolus and remaining dose in 2 h continuous infusion). Subsequently, we assessed spot urine composition at 2 h post-administration, comparing it with total urine output at 6 h. Various potential urine markers were analysed for predicting urine output using receiver operating characteristic (ROC) curves and logistic regression models. We investigated guideline-recommended markers, including spot urine sodium (UNa) and its cut-off, and introduced the UNa/UCr (urine creatinine concentration) ratio adjusting UNa for urine dilution.

RESULTS

Out of 111 patients (85% males, 66.4 ± 13.9 years old, NTproBNP 7290 [4493-14 582] pg/ml), there were 18 (16%) with a poor diuretic response (cumulative urine output <600 ml during the first 6 h). The mean 6 h cumulative diuresis in patients with poor and good diuretic response was 406 ± 142 and 2114 ± 1164 ml, respectively, P < 0.005. After an initial evaluation of several potential biomarkers, only UNa, UCr and UNa/UCr were selected as candidates with the highest predictive value. The cut-off for UNa adjusted for urine dilution: UNa/UCr ratio <0.167 mmol/mg × 10 was determined by ROC analysis with the highest area under the curve (95% confidence interval): 0.956 (0.915-0.997), P < 0.001. When compared with the guideline-recommended cut-off (UNa <50 mmol/L as a reference, specificity-0.97; sensitivity-0.83), the odds ratio (OR) for UNa/UCreat to identify a poor diuretic response was 2.5 times greater, regardless of kidney function (OR for estimated glomerular filtration rate in the logistic regression model was 0.978 [0.945-1.013, P = 0.222]).

CONCLUSIONS

The UNa/UCr ratio in a spot urine sample 2 h after intravenous diuretic administration is a simple, highly predictive marker for the identification of AHF patients with poor diuretic response, surpassing guidelines-recommended markers like UNa.