Lower urine sodium predicts longer length of stay in acute heart failure patients: Insights from the ROSE AHF trial

Abstract Background In patients hospitalized with acute heart failure (AHF), low urine sodium concentration (U Na) after diuretic treatment may identify patients at risk for longer length of stay (LOS) and adverse events. We investigated the prognostic significance of 24‐hour cumulative postdiuretic urine sodium concentration in a multicenter clinical trial population. Methods The Renal Optimization Strategies Evaluation AHF (ROSE AHF) trial randomized 360 patients with AHF and renal dysfunction receiving intravenous diuretic to dopamine, nesiritide, or placebo. Sodium concentration was measured in cumulative urine sample collected during the first 24 hours after randomization in 298 patients. Based on prior studies, lower U Na was defined as ≤60 mmol/L. Results Lower U Na was present in 142 (48%) patients, who had longer LOS (7 days vs 5 days, P < .001) and less 72‐hour weight loss (5.7 lb vs 9.0 lb, P < .001). These associations persisted after controlling for baseline estimated glomerular filtration rate and outpatient furosemide dose. Lower U Na did not modify the null effects of dopamine or nesiritide on clinical outcomes. Results were similar for spot rather than cumulative 24‐hour U Na concentration. Conclusion In patients hospitalized for AHF and renal dysfunction, U Na ≤ 60 mmol/L during the first 24 hours of diuresis identifies patients at risk for prolonged hospitalization but does not provide an indication for adjunctive dopamine or nesiritide.


| INTRODUCTION
Intravenous loop diuretic to relieve congestion is the primary therapy for patients hospitalized with acute heart failure. 1 Persistent congestion, despite treatment with diuretics, is associated with increased risk of adverse outcomes. [2][3][4][5] Heart failure clinicians need new biomarkers that accurately and efficiently identify patients failing conventional treatment and new therapies to offer such patients.
Urine sodium is a logical candidate biomarker because loop diuretics inhibit sodium reabsorption in the kidney, and thus increase natriuresis. Neurohormonal systems including natriuretic peptides and the renin-angiotensin-aldosterone axis closely regulate urine sodium.

| METHODS
The study design and results of the ROSE AHF trial have been published previously. 8 The study was a double-blind, placebo-controlled, randomized trial conducted by the National Heart, Lung, and Blood Institute sponsored Heart Failure Research Network. The study was approved by the institutional review board of each participating center. All participants provided written informed consent.
A total of 360 patients hospitalized with AHF and renal dysfunction at admission, defined as an eGFR of 15 to 60 mL/min/1.73 m 2 by the Modification of Diet in Renal Disease equation, were enrolled. AHF was diagnosed by at least one symptom (edema, dyspnea, or orthopnea) and at least one sign (edema, rales, ascites, or pulmonary vascular congestion on chest X-rays). Patients with reduced and preserved ejection fraction were included. Patients were first randomized in a 1:1 ratio to the dopamine or nesiritide strategy, and within each strategy were randomized in a 2:1 ratio to active therapy or placebo for a duration of 72 hours. The dopamine dosage was 2 μg/kg/minute by continuous infusion. The nesiritide dosage was 0.005 μg/kg/ minute by continuous infusion. All patients were treated with intravenous loop diuretic at a recommended total daily dose of 2.5 times the total daily outpatient dose based on the results of the DOSE trial, 10 with a minimum of 80 mg furosemide in the first 24 hours. The total daily dose was administered as two divided bolus doses. Adjustment of the loop diuretic dose and administration of other medications were at the discretion of the treating physicians. Patients with absence of 24-hour urine sodium data (n = 34) or who received less than the protocol-specified minimum diuretic dose of 80 mg intravenous furosemide or equivalent in the first 24-hour period (n = 36) were excluded from this analysis, the latter because urine composition in these patients may not have reflected the effects of diuretic.
Urine was collected in a single container for the first 24 hours after randomization. Sodium concentration was measured from the container of pooled urine. Similar collections were made for the second and third days after randomization. Patients were divided into lower urine sodium (≤60 mmol/L) and higher urine sodium (>60 mmol/L) groups based on prior published data from our institution, which found that patients in the former group had longer LOS, more frequent renal failure, and a greater risk of clinical deterioration. 6,11 Patients for whom urine sodium measurements were not collected were excluded. Spot urine sodium concentration from a single void was measured 24 hours after randomization; this measurement was used only for secondary analysis.
The primary endpoint of this post hoc analysis was LOS. Secondary endpoints included urine output, weight loss, and serum creatinine change at 72 hours after randomization, congestion score at 7 days after randomization or hospital discharge (whichever occurred earlier), and 60-day death or rehospitalization for heart failure. Congestion score was defined based on the assessment of orthopnea (≥2 pillows = 2 points; <2 pillows = 0 points) and peripheral edema (trace/ mild = 0 points; moderate = 1 point; severe = 2 points). 3 Baseline patient characteristics and short-term outcomes are presented by lower and higher urine sodium groups as median with 25th and 75th percentiles for continuous variables and as number with percentage for categorical variables. Categorical variables were compared with chi-square test or with Fisher's exact test for expected cell counts of 5 or fewer. Continuous variables were compared with Wilcoxon's rank-sum test. Short-term outcomes are also presented for each quintile of urine sodium concentration. Linear and Cox regression models measured associations between urine sodium group and clinical outcomes, with or without adjustment for baseline creatinine and outpatient furosemide-equivalent diuretic dose. Outcomes are also presented for four groups of patients in a 2 × 2 matrix of higher or lower 24-hour urine output (greater than or less than the median of 2785 mL) and higher or lower urine sodium concentration (greater than or less than/equal to 60 mmol/L); associations between urine sodium and outcomes were measured by linear regression adjusting for urine volume. Interaction tests were performed to assess whether the effects of treatment group on clinical outcomes were modified by the 24-hour urinary sodium. The linearity and proportional assumptions were satisfied for all the variables in the models.

| RESULTS
Of the 360 patients randomized in the ROSE AHF trial, 298 met criteria for this analysis. The mean urine sodium concentration was 62.9 mmol/L (SD 25.9 mmol/L); the range was 20 to 135 mmol/L.
Urine sodium concentration for cumulative urine from the first 24 hours was ≤60 mmol/L in 142 patients (48%) and >60 mmol/L in 156 patients (52%). Baseline characteristics of the higher and lower urine sodium concentration groups are shown in Table 1 Urine sodium concentration at 24 hours was a more powerful predictor of LOS than urine volume at 24 hours. Patients in the lower urine sodium group had a longer LOS (7d vs 5d) regardless of whether 24-hour urine output was above or below the median value ( Figure 2).
After controlling for dichotomized 24-hour urine sodium concentration, 24-hour urine volume was not associated with LOS (P < .001 for urine sodium concentration, and P = .69 for urine volume in a linear regression model including both variables).
Outcomes across 5 quintiles of urine sodium concentration are shown in Figure 3. The relationships between lower urine sodium and longer LOS, greater weight loss, and higher rates of death or HF rehospitalization were consistent across the range of urine sodium.
Urine sodium concentration at 24 hours did not modify the null effect of nesiritide and dopamine on clinical outcomes (P interaction of >.05 for all outcomes). Among patients with urine sodium ≤60 mmol/L, there was no significant difference between dopamine and placebo or nesiritide and placebo with respect to LOS, urine output, weight loss, change in creatinine, or 60-day survival free of rehospitalization (Table S1). There was a trend toward lower 60-day survival free of rehospitalization in patients with lower urine sodium randomized to nesiritide compared with placebo that did not reach statistical significance (P interaction of .06 unadjusted, .053 adjusted for eGFR, and furosemide-equivalent dose). As previously reported, among the patients with left ventricular ejection fraction ≤40%, patients randomized to dopamine had lower rates of death or heart failure rehospitalization compared with placebo. 12 This relationship was not modified by urine sodium concentration.

| DISCUSSION
The major finding of this study is that lower urine sodium concentration predicted less weight loss and longer LOS in a cohort of patients hospitalized with AHF and renal dysfunction who were managed with protocol-driven loop diuretic. The relationship between urine sodium and outcomes was consistent across the range of urine sodium values, without a clear threshold. Urine sodium concentration was more predictive of LOS than urine volume at 24 hours. These results suggest that lower urine sodium identifies diuretic-refractory patients who may require escalation of therapy. However, patients with lower urine sodium did not benefit from adjunctive dopamine or nesiritide.
This study is consistent with and extends prior reports demonstrating that urine sodium is a marker of diuretic effectiveness. Testani and colleagues found that a prediction model including urine sodium concentration for 2 hours after diuretic administration, eGFR, and serum/urine creatinine ratio accurately predicted cumulative 6-hour sodium excretion and urine output. 7 Single-center cohort studies have demonstrated that lower urine sodium after loop diuretic predicts worsening heart failure, defined in various reports as worsening renal function, delayed symptom improvement, or need for inotrope or mechanical support. 4,6,13 Our study adds to this literature in several Our findings complement a recent report by Hodson et al from the same cohort, which found that total urinary sodium excretion and net sodium excretion are associated with 6-month all-cause mortality. 9 This study adds short-term outcomes such as weight loss and LOS, which underscore that urinary sodium is a marker of the effectiveness of current diuretic dose in addition to long-term heart failure prognosis. We also demonstrate that both cumulative and spot urine sodium concentration at 24 hours are associated with LOS and weight loss. Spot and total urine sodium concentration may be complementary tests: spot urine sodium concentration yields actionable information more quickly, while cumulative urine sodium concentration integrates the rate of natriuresis during periods between diuretic doses. Notably, spot urine in this study remained useful despite collection at 24 hours rather than at a precisely specified interval after diuretic administration. F I G U R E 1 Time to death or heart failure rehospitalization in higher and lower urine sodium groups. eGFR: estimated glomerular filtration rate; HR, hazard ratio; U Na : urine sodium concentration F I G U R E 2 Prognostic significance of urine sodium concentration and urine volume at 24 hours. Lower urine sodium concentration defined as ≤60 mmol/L. Lower urine volume defined ≤2875 mL, which was the median value. Rate of 60 days death or HF rehospitalization refers to Kaplan-Meier event rate. HF, hazard ratio; LOS, length of stay We chose to focus on spot and cumulative urine sodium concentration rather than total urinary sodium content because urine output is difficult to measure and was poorly correlated with outcomes in our study. We found that 24-hour urine volume was not associated with LOS after adjustment for 24-hour urine sodium concentration. Others have reported that urine output recorded by bedside nurses is often inconsistent with gold standard measurements by research coordinators and with changes in weight. 7,14 The robust association of cumulative urine sodium concentration with short-term outcomes in our study may be due to lower sensitivity to missed voids.
A second finding of this study is that lower postdiuretic urine sodium did not identify a population of AHF patients who benefitted from addition of dopamine or nesiritide compared with placebo. Some have attributed the neutral result of the ROSE AHF trial to enrollment of subjects who were not truly refractory to diuretics, as reflected by the average 72-hour urine output of 8.3 L. 7 Our results do not support this hypothesis. Unfortunately, other potential therapies for diuretic-resistant patients such as ultrafiltration, tolvaptan, or spironolactone have not shown benefit. [15][16][17] Addition of a thiazide diuretic may be an effective strategy for these patients because increased sodium reabsorption in the distal convoluted tubule is a frequent cause of diuretic resistance. 18,19 Development of a better medical therapy for diuretic-resistant patients is an important research priority.
Our study should be interpreted in light of two limitations. First, the timing of urine collection after randomization could have introduced bias into the comparison of the treatments among the lower urine sodium subgroup. Dopamine or nesiritide might increase urine sodium concentration, thereby reassigning patients to the lower risk group and biasing assessment of drug benefit toward the null. However, the data do not support this effect; in fact, median urine sodium concentration was the highest in the placebo group. Patients receiving dopamine (n = 51) and nesiritide (n = 48) were overrepresented in the lower urine sodium group compared with patients receiving placebo (n = 43). Second, other factors not measured or controlled in this study-non-osmotic vasopressin release, acute kidney injury, glycosuria, and variations in dietary sodium intake-also influence urine sodium. Metabolic alkalosis, which may occur in this population due to diuretic-associated volume contraction, may also confound interpretation of urine sodium.
In conclusion, lower urine sodium concentration during the first 24 hours of protocol-driven intravenous loop diuretic therapy was associated with less weight loss and longer LOS among patients with acute heart failure and renal dysfunction. Patients with lower urine sodium did not benefit from adjunctive therapy with dopamine or nesiritide. Urine sodium is a promising biomarker, which may assist in the early identification of diuretic-resistant patients. Further research is needed to identify effective adjunctive therapies for this population.
F I G U R E 3 Outcomes by quintile of 24-h urine sodium concentration. Median values presented for LOS and 72-hour weight loss. The range of urine sodium concentration for patients included in each quintile is provided below the corresponding bar. LOS, length of stay; U Na , urine sodium concentration