OUP user menu

★ editor's choice ★

Relationship between admission serum sodium concentration and clinical outcomes in patients hospitalized for heart failure: an analysis from the OPTIMIZE-HF registry

Mihai Gheorghiade, William T. Abraham, Nancy M. Albert, Wendy Gattis Stough, Barry H. Greenberg, Christopher M. O'Connor, Lilin She, Clyde W. Yancy, James Young, Gregg C. Fonarow
DOI: http://dx.doi.org/10.1093/eurheartj/ehl542 980-988 First published online: 19 February 2007

Abstract

Aims Hyponatraemia has been shown to be an independent predictor of mortality in selected patients with heart failure enrolled in clinical trials. The predictive value of hyponatraemia has not been evaluated in unselected patients hospitalized with heart failure.

Methods and results OPTIMIZE-HF is a registry and performance-improvement programme for patients hospitalized with heart failure and includes a subgroup with 60–90 day follow-up data. The relationship between admission serum sodium concentration and clinical outcomes was analysed in 48 612 patients from 259 hospitals. Admission serum sodium levels were analysed both as a continuous variable and by grouping patients with admission Na < 135 and Na ≥ 135 mmol/L. Patients with hyponatraemia (Na <135 mmol/L) at the time of hospital admission had modest differences in baseline clinical characteristics and management during hospitalization compared with patients who had serum sodium ≥135 mmol/L. Patients with hyponatraemia were more likely to be Caucasian, have lower admission systolic blood pressure, and receive intravenous inotropes during hospitalization. Patients with hyponatraemia had significantly higher rates of in-hospital and follow-up mortality and longer hospital stays, although no difference in re-admission rates was observed. After adjusting for differences with multivariable analysis, the risk of in-hospital death increased by 19.5%, the risk of follow-up mortality by 10%, and the risk of death or rehospitalization by 8% for each 3 mmol/L decrease in admission serum sodium below 140 mmol/L.

Conclusion Hyponatraemia in hospitalized patients with heart failure is relatively common and is associated with longer hospital stays and higher in-hospital and early post-discharge mortality. Re-admission rates were equally high in patients with or without hyponatraemia.

  • Serum sodium
  • Heart failure
  • Registry
  • Risk factors
  • Hospitalization
See page 920 for the editorial comment on this article (doi:10.1093/eurheartj/ehm046)

Introduction

Hyponatraemia, defined as a serum sodium concentration <135 mmol/L, is a relatively common finding in patients admitted to the hospital with heart failure.14 Hyponatraemia was present in 27% of patients in the Outcomes of a Prospective Trial of Intravenous Milrinone for Exacerbations of Chronic Heart Failure (OPTIME-CHF) study.5 A similar rate of hyponatraemia was observed in the Acute and Chronic Therapeutic Impact of a Vasopressin Antagonist in Congestive Heart Failure (ACTIV in CHF) trial, in which 21% of patients hospitalized for acute decompensated heart failure were hyponatraemic at baseline.6

Serum sodium is a known predictor of outcome in patients with chronic heart failure.7,8 It has also been shown to predict mortality in patients hospitalized for acute heart failure who were enrolled in the OPTIME-CHF trial. In-hospital and 60 day mortality rates were highest for the patients with the lowest admission serum sodium.5 Each 5 mmol/L increase in serum sodium was associated with a 25% lower risk of 60 day mortality.9

Randomized, controlled trials have been the setting for most studies of the relationship between serum sodium and clinical outcomes in patients hospitalized for acute heart failure. However, patients enrolled in highly controlled trials may not be truly representative of the general heart failure population.10 To understand the influence of serum sodium in a large, unselected patient population, we analysed the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF) registry. The purpose of this analysis was to evaluate the relationship among admission serum sodium, patient characteristics, treatment patterns, and clinical outcomes in a nationally representative group of patients hospitalized for acute heart failure.

Methods

The OPTIMIZE-HF programme consisted of a registry and a quality-of-care intervention, as previously described.11,12 Detailed clinical data were collected regarding patient characteristics, medical history, physical examination, laboratories, diagnostic testing, procedures, medication use, clinical outcomes, and adherence to performance indicators. Eligible patients were enrolled if they were at least 18 years of age and were hospitalized with new-onset or worsening heart failure and had a primary discharge diagnosis of heart failure. Patients were included irrespective of ventricular function. Patients with a quantitative left ventricular ejection fraction (LVEF) <40% or moderate-to-severe or severe systolic dysfunction by qualitative assessment were considered to have left ventricular systolic dysfunction (LVSD). A pre-specified subset (10%) of the total OPTIMIZE-HF patient population was followed for 60–90 days to document clinical events post-discharge. The quality-of-care improvement programme provided participating hospitals with process-of-care tools including evidence-based treatment algorithms, standing orders, and discharge checklists. Hospital sites could view and analyse their adherence to benchmarked performance measures in real time. The registry coordinating centre was Outcome Sciences, Inc. (Cambridge, MA, USA). This study was conducted in accordance with the Declaration of Helsinki. The protocol was approved by each participating centre's institutional review board or through the use of a central institutional review board. Written informed consent was obtained from patients who participated in the follow-up data collection prior to their enrollment.

All statistical analyses were performed independently by Duke Clinical Research Institute (Durham, NC, USA). Data were reported as mean ± standard deviation for continuous variables or percentages of non-missing values for categorical variables. Performance measures and medication use are reported among eligible patients without contraindications or documented intolerance. Admission serum sodium concentrations were analysed as a continuous variable as well as categorized into two discrete groups: Na <135 mmol/L and Na ≥135 mmol/L. Serum sodium was assessed as a continuous measure in all statistical tests. Patient characteristics, evidence-based treatments, and clinical outcomes were compared. Spearman correlation, Kruskal–Wallis test, and Wilcoxon rank-sum test were used for comparisons of serum sodium with continuous measures, multilevel nominal measures, and dichotomous measures, respectively. Multiple studies have suggested that heart failure patients with LVSD have different characteristics and outcomes compared with patients with preserved systolic function and heart failure. As such, a pre-specified analysis was performed in these subgroups.

Multivariable models of in-hospital death, length of hospital stay, and post-discharge mortality or rehospitalization were developed to allow for consistent covariate adjustment across all studies. The types of models were logistic for in-hospital mortality and for the combination of post-discharge mortality or rehospitalization, general linear modelling for length of stay (LOS), and Cox proportional hazards for post-discharge mortality. The model-development process was similar for all four outcomes. The set of important factors was identified first, including variables that were predictive of outcome in univariate analysis or had been identified in previously published studies as risk factors. The linearity assumption for continuous measures was evaluated using restricted cubic spline transformations and comparing the −2 log likelihood χ2 of models with the linear vs. transformed variable. When needed, appropriate transformations such as piecewise linear splines were applied. Both backward and forward stepwise variable-selection techniques were applied to the selected variables of interest. A P-value of 0.05 was used for both entry and remaining in the model. Each model was validated using bootstrapping with 200 replications to evaluate the degree of optimism of the model C-statistic. To account for inflation of type I error because of multiple testing, we considered only P-values <0.001 to be statistically significant, except for testing for heterogeneity, where P-values <0.01 were considered significant. SAS® (Cary, NC, USA) version 8.2 was used for all statistical analyses.

Results

Enrollment in OPTIMIZE-HF began in March 2003 and ended in December 2004. A total of 259 hospitals across the USA participated in the programme, representing both academic and community-based centres of all sizes and from all regions of the country. During this period, 48 612 patients were enrolled. The population was elderly, with a mean age of 73 years. The majority of them were women (52%) and Caucasian (74%). Slightly less than half the patients had LVSD (49%). Follow-up data were obtained in 5791 patients.

A total of 47 647 patients had admission serum sodium collected and were included in this analysis. Overall, patients enrolled in OPTIMIZE-HF displayed a wide distribution of admission sodium values (Figure 1). The mean admission serum sodium in the total cohort was 138 ± 5 mmol/L, and 19.7% of the patients had values below 135 mmol/L. Patients admitted with hyponatraemia were clinically similar to patients with normonatraemia in terms of age, gender, HF aetiology, diabetes, heart rate, ejection fraction, and symptoms of congestion (Table 1). However, patients admitted with hyponatraemia were more likely to be Caucasian and to have lower admission systolic blood pressure and an atrial arrhythmia. No differences were observed in the baseline (admission) use of angiotensin-converting enzyme (ACE) inhibitors or β-blockers (Table 1).

Figure 1

Distribution of admission serum sodium in patients hospitalized with a primary discharge diagnosis of heart failure.

View this table:
Table 1

Patient clinical characteristics and treatments by admission serum sodium groups

VariableNa <135 mmol/L (n = 9368)Na ≥135 mmol/L (n = 38 279)P-value
Mean age, years (%)74.1 (14.0)73.0 (14.0)0.262
Female, n (%)5025 (53.6)19 585 (51.2)0.008
Caucasian, n (%)7547 (80.6)27 765 (72.5)<0.0001
African American, n (%)1012 (10.8)7435 (19.4)<0.0001
LVSD (LVEF <40% or moderate/severe LVD), n (%)4954 (52.9)19 854 (51.9)<0.0001
Mean LVEF, % (SD)38.5 (18.3)39.1 (17.5)<0.0001
Ischaemic aetiology, n (%)4310 (46.0)17 455 (45.6)<0.0001
Hypertensive aetiology, n (%)1924 (20.5)9019 (23.6)<0.0001
No known prior HF, n (%)939 (10.0)4600 (12.0)0.0003
HF hospitalizations within 6 months, n (%)<0.0001
 04258 (45.5)19 007 (49.7)
 11775 (18.9)6085 (15.9)
 2484 (5.2)1519 (4.0)
 ≥3358 (3.8)1183 (3.1)
 Unknown2493 (26.6)10 485 (27.4)
Atrial arrhythmia, n (%)3190 (34.1)11 469 (30.0)<0.0001
ICD, n (%)552 (5.9)1887 (4.9)<0.0001
Insulin-treated diabetes, n (%)1618 (17.3)6302 (16.5)<0.0001
Non-insulin-treated diabetes, n (%)2363 (25.2)9533 (24.9)0.0003
Mean serum sodium, mmol/L (SD)130.6 (3.9)139.5 (2.9)
Mean BNP, pg/mL (SD)1386.7 (1385.2)1250.0 (1307.1)<0.0001
Mean serum creatinine at admission, mg/dL (SD)1.86 (1.6)1.74 (1.5)0.0003
Mean serum creatinine at discharge, mg/dL (SD)1.74 (1.43)1.75 (1.39)<0.0001
Mean weight change from admission to discharge, kg (SD)−2.5 (4.9)−2.6 (4.8)0.198
Mean admission SBP, mmHg (SD)135.7 (32.6)144.4 (32.7)<0.0001
Mean discharge SBP, mmHg (SD)122.1 (22.9)125.4 (22.3)<0.0001
Mean admission heart rate, b.p.m. (SD)86.7 (21.6)86.6 (21.4)0.139
Mean discharge heart rate, b.p.m. (SD)77.5 (14.4)75.6 (14.0)<0.0001
JVD at admission, n (%)2706 (33.1)10 792 (32.7)0.147
JVD at discharge, n (%)260 (5.0)943 (4.4)0.160
Oedema at admission, n (%)5937 (64.6)24 209 (64.7)<0.0001
Oedema at discharge, n (%)2025 (27.6)8145 (26.7)0.349
Rales at admission, n (%)5822 (63.2)24 235 (64.5)0.004
Rales at discharge, n (%)1276 (16.6)4898 (15.1)0.043
Dyspnoea at rest at admission, n (%)4103 (43.8)16 727 (43.7)0.074
Dyspnoea on exertion at admission, n (%)5584 (59.6)23 710 (61.9)<0.0001
Orthopnoea at admission, n (%)2331 (24.9)10 730 (28.0)<0.0001
PND at admission, n (%)1235 (13.2)5995 (15.7)<0.0001
ACE-inhibitor at admission, n (%)3610 (38.5)15 257 (39.9)0.252
ARB at admission, n (%)1116 (11.9)4471 (11.7)0.638
ACE-inhibitor and/or ARB at admission, n (%)4626 (49.4)19 182 (50.1)0.580
β-blocker at admission, n (%)4938 (52.7)20 353 (53.2)0.801
Statin at admission, n (%)2731 (29.2)12 386 (32.4)<0.0001
Aldosterone antagonist at admission, n (%)968 (10.3)2417 (6.3)<0.0001
Warfarin at admission, n (%)2302 (24.6)8483 (22.2)<0.0001
Diuretic at admission, n (%)6338 (67.7)24 954 (65.2)<0.0001
Thiazide/thiazide-like diuretic at admission, n (%)1027 (11.0)2996 (7.8)<0.0001
Loop diuretic at admission, n (%)5842 (62.4)23 264 (60.8)<0.0001
Digoxin at admission, n (%)2405 (25.7)8714 (22.8)<0.0001
Dobutamine, n (%)554 (5.9)1299 (3.4)<0.0001
Dopamine, n (%)579 (6.2)1148 (3.0)<0.0001
Milrinone, n (%)201 (2.1)444 (1.2)<0.0001
Nesiritide, n (%)1092 (11.7)4043 (10.6)<0.0001
Other IV vasodilator, n (%)274 (2.9)1357 (3.5)0.010
Dialysis, n (%)576 (6.1)1828 (4.8)<0.0001
LV assist device, n (%)24 (0.3)33 (<0.1)<0.0001
Mechanical ventilation, n (%)359 (3.8)1137 (3.0)<0.0001
  • BNP, B-type natriuretic peptide; b.p.m., beats per minute; HF, heart failure; ICD, implantable cardioverter-defibrillator; IV, intravenous; JVD, jugular venous distention; LVD, LV dysfunction; PND, paroxysmal nocturnal dyspnoea; SBP, systolic blood pressure.

During hospitalization, a higher proportion of patients in the low serum sodium group received parenteral inotropes. In addition, dialysis, mechanical ventilatory support, and LV assist devices were used more often in patients with low admission serum sodium. Patients in each group experienced a similar amount of diuresis during hospitalization, as seen by equivalent weight change from admission to discharge (Table 1).

The discharge use of aldosterone antagonists was higher in the lower serum sodium group, but patients in the lower serum sodium group were less likely to receive an ACE-inhibitor or angiotensin receptor blocker (ARB) or a statin at discharge. The discharge use of β-blockers did not differ between serum sodium groups (Table 2). The addition of a β-blocker, ACE-inhibitor, and/or ARB to patients not treated at admission was similar between the groups. Diuretics were prescribed less often at discharge in the low serum sodium group, with 74.1% of patients receiving diuretics compared with 78.1% receiving them in the high sodium group. Loop diuretics followed this pattern, with hyponatraemic patients being less likely to receive a loop diuretic at discharge. However, thiazide-type diuretics were used slightly more often in patients with hyponatraemia (8.8% in the lower serum sodium group; 7.1% in the higher serum sodium group) (Table 2). Modest differences were observed between admission serum sodium concentration and several heart failure performance indicators (Table 2).

View this table:
Table 2

Performance measures and discharge therapy by admission serum sodium group

VariableNa <135 mmol/L (n = 9368)Na ≥135 mmol/L (n = 38 279)P-value
Complete discharge instructions55.1 (53.8–56.4)53.4 (52.8–54.0)<0.0001
LVEF assessed85.7 (84.9–86.5)86.6 (86.2–87.0)0.037
Smoking cessation counselling65.2 (62.5–67.9)61.8 (60.5–63.1)0.010
ACE-inhibitor72.4 (70.6–74.1)76.0 (75.2–76.8)0.0001
ACE-inhibitor and/or ARBa79.7 (78.2–81.3)83.2 (82.5–83.9)<0.0001
β-blockerb82.0 (80.6–83.5)83.5 (82.8–84.1)0.262
Statinc33.8 (32.7–34.8)40.5 (39.9–41.1)<0.0001
Aldosterone antagonistd19.5 (18.2–20.7)17.8 (17.2–18.4)0.0001
Warfarine50.2 (48.2–52.1)52.7 (51.7–53.7)0.397
Diuretic74.1 (73.2–75.0)78.1 (77.7–78.5)<0.0001
Thiazide/thiazide-like diuretic8.8 (8.2–9.4)7.1 (6.9–7.4)<0.0001
Loop diuretic70.9 (70.0–71.8)74.9 (74.4–75.3)<0.0001
Outpatient IV vasoactive therapies2.0 (1.2–2.9)1.2 (0.9–1.5)0.001
  • All values are expressed as percentage (95% confidence interval).

  • aACE-inhibitor and/or ARB use in patients with LVSD, excluding patients with contraindications to ACE-inhibitor and/or ARB.

  • bβ-Blocker use in patients with LVSD, excluding patients with contraindications to β-blockers.

  • cStatin use in patients with medical history of coronary artery disease, cerebrovascular accident/transient ischaemic attack, diabetes, hyperlipidaemia, or peripheral vascular disease.

  • dAldosterone antagonist use in patients with LVSD.

  • eWarfarin use in patients with chronic or paroxysmal atrial fibrillation.

Clinical outcomes

The in-hospital mortality rate was 3.8% in the overall OPTIMIZE-HF programme. When clinical outcomes were examined by admission serum sodium values, lower admission serum sodium was associated with higher in-hospital mortality, 6.0% for the lower sodium group compared with 3.2% for those patients with higher serum sodium (Table 3; Figure 2). The relationship between admission serum sodium levels and in-hospital mortality is plotted in Figure 3. The risk of mortality begins to significantly rise at serum sodium <138 mmol/L and is more than double for the patients with serum sodium levels in the 132–135 mmol/L range. Overall, the mean LOS was 5.6 ± 5.5 days, and the median LOS was 4.0 (25th and 75th percentiles 3.0 and 7.0, respectively). Lower admission serum sodium was associated with longer mean hospital LOS, ranging from 6.4 days for the lower group to 5.5 days for the higher group, as shown in Table 3 and Figure 2.

Figure 2

LOS and clinical outcomes by admission serum sodium groups. Values are expressed as mean ± SD or percentage and 95% confidence intervals.

Figure 3

Relationship between admission serum sodium level and in-hospital mortality. Restrictive cubic spline transformation plot with 95% confidence intervals is shown.

View this table:
Table 3

Clinical event rates by admission serum sodium and LVSDa

Na <135 mmol/L (n = 7882)Na ≥135 mmol/L (n = 32 572)P-value
Between groupsInteraction
In-hospital mortality: overall6.0 (5.5–6.5)3.2 (3.0–3.4)<0.0001
LVSD6.8 (6.0–7.6)3.2 (2.9–3.4)<0.00010.009
No LVSD4.2 (3.6–4.8)2.6 (2.3–2.8)<0.0001
Mean (SD)/median LOS (days)6.4 (6.4)/5.05.5 (5.2)/4.0<0.0001
LVSD6.9 (7.2)/5.05.7 (5.6)/4.0<0.00010.009
No LVSD6.4 (5.8)/5.05.6 (5.0)/4.0<0.0001
Na <135 mmol/L (n = 1024)Na ≥135 mmol/L (n = 4019)
Post-discharge mortality: overall12.4 (10.0–14.9)7.6 (6.2–8.1)<0.0001
LVSD15.5 (11.0–20.1)6.4 (5.1–7.7)<0.00010.026
No LVSD9.3 (6.1–12.4)6.9 (5.5–8.3)0.168
60–90 day rehospitalization: overall32.4 (29.7–35.1)29.0 (27.7–30.4)0.292
LVSD31.5 (27.6–35.4)29.8 (27.8–31.7)0.7870.244
No LVSD33.5 (29.3–37.7)28.1 (26.0–30.1)0.054
Death or rehospitalization post-discharge: overall42.5 (39.6–45.4)34.8 (33.4–36.2)<0.0001
LVSD42.4 (38.2–46.7)34.8 (32.8–36.9)0.0110.714
No LVSD42.1 (37.6–46.6)33.5 (31.4–35.7)0.002
  • All values are expressed as percentage (95% confidence interval).

  • aThis analysis includes only patients with LV function documented: 41 267/48 612 (84.9%) and admission serum sodium between 110 and 178 mmol/L. Post-discharge mortality and 60–90 day rehospitalization analyses include only patients with follow-up data and LV function documented (n = 5043) and admission serum sodium between 110 and 178 mmol/L.

In the follow-up cohort, admission hyponatraemia was associated with higher 60–90 day mortality; 12.4% for <135 mmol/L vs. 7.1% for ≥135 mmol/L (P < 0.0001) (Table 3; Figure 2). Rehospitalization rates during follow-up were not different between admission serum sodium groups (32.4% for the lower sodium group and 29.0% for the higher sodium group; P = 0.292). The association between admission serum sodium and LOS or in-hospital mortality was present regardless of LV systolic function. For in-hospital mortality and follow-up mortality, the magnitude of effect was most pronounced in patients with LVSD (Table 3).

After adjusting for other prognostic factors, admission serum sodium concentration remained a significant independent predictor of in-hospital mortality, post-discharge mortality, and 60–90 day death or rehospitalization (Table 4). Serum sodium was also independently predictive of LOS. However, it was found that the relationships between admission sodium and these clinical outcomes are not monotonic. Admission sodium <140 mmol/L had a different effect than admission sodium >140 mmol/L. The risk of in-hospital mortality increased by 19.5% for LVSD patients and by 8.6% for non-LVSD patients for each 3 mmol/L decrease in admission sodium below 140 mmol/L. In contrast, the risk of in-hospital mortality decreased 14.0% for LVSD patients and 10.9% for non-LVSD patients for each 3 mmol/L decrease in admission sodium above 140 mmol/L. The risk of follow-up mortality increased by 10.0% and the risk of death or rehospitalization increased by 8.0% for each 3 mmol/L decrease in admission serum sodium below 140 mmol/L, and there was no significant effect of admission sodium on these two clinical outcomes if admission sodium is >140 mmol/L.

View this table:
Table 4

Multivariable logistic regression and Cox proportional hazards analysis

OR or hazard ratioa95% Confidence limitsP-value
In-hospital mortality (per 3 mmol/L decrease up to 140 mmol/L, LVSD = Yes)1.1951.134–1.259Interaction between admission sodium and LVSD: 0.024
In-hospital mortality (per 3 mmol/L decrease above 140 mmol/L, LVSD = Yes)0.8590.740–0.998
In-hospital mortality (per 3 mmol/L decrease up to 140 mmol/L, LVSD = No)1.0861.020–1.155Overall main effect of admission sodium: <0.0001
In-hospital mortality (per 3 mmol/L decrease above 140 mmol/L, LVSD = No)0.8910.759–1.045
60–90 day follow-up mortality (per 3 mmol/L decrease up to 140 mmol/L)1.1001.027–1.1780.007
60–90 day mortality and rehospitalization (per 3 mmol/L decrease up to 140 mmol/L)1.0801.026–1.1360.004
  • aFor admission serum sodium >140 mmol/L, the effects of sodium on 60–90 day follow-up mortality and 60–90 day mortality and rehospitalization are not statistically significant (P > 0.05).

Discussion

The findings of this analysis demonstrate that hyponatraemia at the time of admission is relatively common in a representative group of patients hospitalized with acute decompensated heart failure. Hyponatraemia was present in 20% of the OPTIMIZE-HF cohort, a finding that is similar to the rates observed in other studies of acute decompensated heart failure.5,6 Hyponatraemia may identify a population with a pathophysiological profile that differs from that of normonatraemic patients. It may reflect more severe activation of the rennin–angiotensin–aldosterone or the sympathetic nervous system and/or vasopressin release.2,5,13,14 In the OPTIMIZE-HF registry, low serum sodium was more common among patients with lower admission systolic blood pressure and a prior history of heart failure. These findings are supported by results of other studies in which patients with a history of heart failure were more likely to have hyponatraemia.15

Hyponatraemia can be associated with the hypervolemic, hypovolemic, and euvolemic states. However, in OPTIMIZE-HF data, patients with hyponatraemia at the time of admission were most likely to be hypervolemic since they were admitted for worsening heart failure and had evidence of fluid overload. Low serum sodium was associated with a higher rate of in-hospital and post-discharge mortality after adjustment for multiple other prognostic factors.

In OPTIMIZE-HF, the in-hospital mortality risk for LVSD and non-LVSD patients and follow-up mortality risk increased 19.5, 8.6, and 10.0%, respectively, for each 3 mmol/L decrease in serum sodium up to 140 mmol/L after adjustment for other prognostic variables. The rate of in-hospital and follow-up mortality for patients with hyponatraemia was approximately twice that of patients in the normal-to-high sodium group. This finding is similar to the results of other multivariable analyses, such as the OPTIME-CHF study, in which retrospective data revealed an 18% higher risk of 60 day mortality for each 3 mmol/L decrease in serum sodium.5 The magnitude of the association between serum sodium and mortality was greater in the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) study.16 This retrospective study included 2624 newly admitted patients with a primary diagnosis of heart failure from 34 community and teaching hospitals in Ontario, Canada from 1999 to 2001. The mean age of heart failure patients included in this study was 76 years and approximately half of patients were women. In an analysis of predictors of mortality, hyponatraemia predicted both 30 day and 1 year mortality. The odds ratio (OR) was 1.53 and 1.46 for each 1 mmol/L decrease in serum sodium below 136 mmol/L at the 30 day and 1 year time-points, respectively.16

The unadjusted relationship between serum sodium and in-hospital mortality and LOS was present regardless of LV function. In the past, the ability to evaluate interactions between LVSD and serum sodium has been limited because the clinical trials used for these analyses excluded patients with ejection fractions >40%.5,6 However, the EFFECT study did include patients with normal ejection fraction, and a total of 362 and 764 patients who had LV function assessed had preserved systolic function in the validation and derivation cohorts, respectively. The authors stated that the coefficients of model covariates were similar after adjustment for LVSD, suggesting that the relationship between serum sodium and mortality was present regardless of LVSD.16 The analysis from OPTIMIZE-HF provides a much larger data set of patients with preserved systolic function (n = 21 149), which can be used to assess the association among serum sodium, mortality, and potential interactions with systolic function. On the basis of these data, serum sodium is of prognostic importance in patients with LVSD as well as in those with preserved systolic function.

Although the differences were small in the use of evidence-based therapies between serum sodium groups, underuse of drugs such as ACE-inhibitors, β-blockers, or aldosterone antagonists could have important implications for hyponatraemic patients. Previous work by Lee and Packer14 has shown that patients with hyponatraemia treated with ACE-inhibitors have lower mortality than hyponatraemic patients who did not receive ACE-inhibitors. In OPTIMIZE-HF, ACE-inhibitors or ARBs were used in 80% of eligible patients in the low serum sodium group and 83% of eligible patients in the higher group, leaving 17–20% of eligible patients untreated. These data suggest that every effort should be made to ensure that all eligible patients are treated with ACE-inhibitors or ARB and other evidence-based therapies. The results of the present study also indicate that β-blocker use did not differ between groups, and a β-blocker was prescribed in the majority of eligible patients (83.1% overall). Although this rate of use is encouraging, efforts should continue to encourage β-blocker use in all eligible patients. Aldosterone antagonist use was higher in the low sodium group compared with the higher group (19.5 vs. 17.8%), but overall use was low. Patients with hyponatraemia are a high-risk population that may benefit from aldosterone blockade. Future strategies should focus on improving the use of aldosterone antagonists among eligible patients with LVSD who are hospitalized with heart failure.

The therapeutic approach to the treatment of hyponatraemia in heart failure has traditionally relied on attempts to improve cardiac function and at the same time limit fluid intake. However, this approach is not likely to improve or normalize serum sodium concentration. Recently, agents that selectively block the type 2 vasopressin receptor and increase free water excretion without any of the adverse consequences of other therapies, such as loop diuretics, have been shown to improve or normalize serum sodium in patients with mild, moderate, or severe heart failure without affecting heart rate, blood pressure, or renal function.6,1720 Although oral vasopressin antagonists are known to improve or normalize serum sodium in heart failure patients, their effect on clinical outcomes are not yet known. The Efficacy of Vasopressin Antagonism in Heart Failure Outcome Study with Tolvaptan (EVEREST) trial recently completed evaluating the effects of tolvaptan, an oral vasopressin-2 antagonist, when added to standard therapy in more than 4000 patients hospitalized with heart failure and low ejection fraction, with respect to post-discharge mortality and hospitalization. This study also examined the correlation between tolvaptan use and normalization or improvement in serum sodium levels in the nearly 20% of study patients with hyponatraemia.21

Several limitations should be considered when interpreting the results from the present analysis. First, OPTIMIZE-HF was not a prospective, randomized trial. Therefore, unmeasured variables that could have influenced the findings may have been present. OPTIMIZE HF hospitals are self-selected and thus may not be entirely representative of national care patterns and clinical outcomes. Admission diuretic dose was not collected in OPTIMIZE-HF. Consequently, potential associations between diuretic dose and hyponatraemia could not be examined.2 Discharge serum sodium was not collected, so we cannot evaluate the influence of serum sodium change on clinical outcomes or ascertain the mortality effect of interventions to increase serum sodium, which are important and relevant concerns given the development of vasopressin antagonists for the treatment of heart failure. Some of the observed differences may not be clinically relevant, although they were statistically significant because of the large number of patients studied overall.

OPTIMIZE-HF has confirmed that hyponatraemia is common in a representative population of patients hospitalized with acute heart failure. Despite receiving evidence-based therapies to an extent similar to that of normonatraemic patients, hyponatraemic patients have a longer LOS and significantly greater in-hospital and post-discharge mortality. This was true for patients with systolic or diastolic heart failure. This observation suggests that additional treatment strategies are needed in these patients. Future studies should evaluate the mortality effects of interventions that increase serum sodium to within the normal range in patients hospitalized with acute decompensated heart failure.

Acknowledgement

The OPTIMIZE-HF registry and this study were supported by GlaxoSmithKline, Philadelphia, PA, USA.

Conflict of interest: M.G. reported that he is a consultant and has received honoraria from Otsuka, Protein Design Lab, Sigma Tau, Medtronic, Pfizer, and GSK. W.G.S. reported that she has received a research grant, is a consultant, and is on the speaker's bureau for GSK. C.M.O. reported that he is a consultant for Amgen, GSK, Guidant, Medtronic, Merck, Novartis, Otsuka, Pfizer, and Scios, Inc. W.T.A. has reported that he has received a research grant from Amgen, Biotronik, CHF Solutions, GSK, HFSA, NIH, Medtronic, Myogen, Orqis Medical, Otsuka Maryland Research Institute, Paracor Inc, and Scios, Inc. He is a consultant/on the speakers bureau for Amgen, AstraZeneca, Boehringer-Ingelheim, CHF Solutions, GSK, Guidant, Medtronic, Merck, Pfizer, ResMed, Respironics, Scios, Inc., and St. Jude Medical. He is on the Advisory Board of CardioKinetix, Inc., CHF Solutions, Department of Veterans Affairs Cooperative Studies Program, NIH, and Savacor, Inc. He has received honoraria from AstraZeneca, Boehringer-Ingelheim, GSK, Guidant, Medtronic, Merck, Pfizer, ResMed, Respironics, Scios, Inc., and St. Jude Medical. B.H.G. reported that he is on the speaker's bureau for GSK, AstraZeneca, Pfizer, Merck, and Novartis. He is a consultant for GSK and CHF Solutions. J.B.Y. reported that he has received research grants from and is a consultant for AstraZeneca and GSK. N.M.A. reported that she is a consultant for Medtronic and GSK. She is also on the speaker's bureau for Medtronic and GSK. She is a consultant for Medtronic, GSK, NitroMed, and Scios, Inc. C.W.Y. reported that he has received research grants from GSK, Scios, Inc., Medtronic, and NitroMed. He is also a consultant for Scios, Inc., GSK, Medtronic, NitroMed, and CHF Solutions. L.S. is an employee of DCRI. G.C.F. reported that he has received research grants from Medtronic, GSK, Pfizer, Merck, BMS/Sanofi, and Guidant. He is also on the speaker's bureau or has received honoraria from Medtronic, GSK, Pfizer, Merck, BMS/Sanofi, and Guidant. He is a consultant for Medtronic, GSK, Pfizer, Merck, BMS/Sanofi, and Guidant.

Footnotes

  • The Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF) registry is registered: www.clinicaltrials.gov, study number NCT00344513.

References

View Abstract