doi: 10.1111/j.1751-7133.2010.00206.

ORIGINAL PAPER

Burden of Sodium Abnormalities in Patients Hospitalized

for Heart Failure

H eart failure (HF) is an important

public health problem that results
in frequent emergency department visits
Hyponatremia presumably is associated with adverse clinical outcomes in patients with
congestive heart failure (CHF), but risk thresholds and economic burden are less studied.
The authors analyzed 115,969 patients hospitalized for CHF and grouped them by
and hospitalizations.1–5 Consequently, serum sodium levels (severe hyponatremia, 130 mEq ⁄ L; hyponatremia,
the associated economic burden contin- 131–135 mEq ⁄ L; normonatremia, 136–145 mEq ⁄ L; hypernatremia, >145 mEq ⁄ L).
ues to escalate. In 2009, the combined Univariable and multivariable analyses on the associated clinical and economic
direct and indirect costs were estimated outcomes were performed. The most common abnormality was hyponatremia (15.9%),
to be $37 billion in the United States.2,5 followed by severe hyponatremia (5.3%) and hypernatremia (3.2%). Hospital mortality
Similar to renal insufficiency, intense was highest for severe hyponatremia (7.6%), followed by hypernatremia (6.7%) and
diuretic therapy, diabetes, and atrial hyponatremia (4.9%) ( P<.0001). Compared with normonatremia, risk-adjusted mortality
fibrillation, both incident and persistent was highest for severe hyponatremia (odds ratio [OR], 1.78; 95% confidence interval
hyponatremia have been shown to [CI], 1.59–1.99), followed by hypernatremia (OR, 1.55; 95% CI, 1.34–1.80) and
hyponatremia (OR, 1.29; 95% CI, 1.19–1.40; all P<.0001). Risk-adjusted hospital
increase the risk of hospital and subse-
prolongation was greater for each level of sodium abnormality than for normonatremia,
quent mortality and re-hospitalization
ranging from 0.42 (CI, 0.26–0.60) days for hypernatremia to 1.28 (CI, 1.11–1.47)
rates in both observational studies and days for severe hyponatremia. Risk-adjusted attributable hospital cost increase was
controlled clinical trials.6–9 Recognized highest for severe hyponatremia ($1132; CI, $856–$1425; all ( P<.0001). Sodium
as the most common electrolyte abnor- abnormalities were common in patients hospitalized for CHF. Adverse outcomes resulted
mality in clinical practice,9,10 hyponat- not only from severe hyponatremia, but also from mild hyponatremia and hypernatremia.
remia occurs in approximately 20% of Congest Heart Fail. 2011;17:1–7. 2011 Wiley Periodicals, Inc.
patients admitted for HF.8,9 The high
prevalence of hyponatremia in HF Andrew F. Shorr, MD, MPH;1 Ying P. Tabak, PhD;2
is multifactorial. The most common Richard S. Johannes, MD, MS;2,3 Vikas Gupta, PharmD;2
causes are volume expansion due to Mitchell T. Saltzberg, MD;4 Maria R. Costanzo, MD5
vasopressin-induced free water retention From the Washington Hospital Center, Washington, DC;1 Clinical Research,
and diuretic-mediated renal electrolyte CareFusion, Marlborough, MA;2 Brigham and Women’s Hospital and Harvard
Medical School, Boston, MA;3 Christiana Healthcare, Newark, DE;4 and
losses.11 Regardless of whether a marker
Midwest Heart, Naperville, IL5
or cause of pathology, hyponatremia is
associated with increased mortality rates
Address for correspondence:
during and after hospitalization, even Maria R. Costanzo, MD, Medical Director, Midwest Heart Specialists Heart
after controlling for potentially con- Failure and Pulmonary Hypertension Programs, Edward Hospital Center
founding factors.6,8,9 Hyponatremia is for Advanced Heart Failure, Edward Heart Hospital, Fourth Floor,
also associated with prolonged length of 801 South Washington Street, PO Box 3226, Naperville, IL 60566
hospital stay (LOS),9 but the thresholds E-mail: mcostanzo@midwestheart.com
of sodium abnormalities that increase Manuscript received July 22, 2010; revised November 9, 2010;
the risk of adverse outcomes and the accepted November 30, 2010
excess cost attributable to disorders of
this electrolyte have not been evaluated. in this high-risk patient group, we per- diagnosis of HF from January 2004
Recognition of the impact of sodium formed a retrospective analysis of a large through December 2005 contributing to
abnormalities on the outcomes of acute contemporary population of patients one of the Clinical Research Databases
decompensated HF patients may lead to hospitalized for HF. from CareFusion (formerly Cardinal
earlier use of therapies shown in con- Health, Marlborough, MA). Used for
trolled clinical trials to effectively cor- Methods research since the late 1980s,12–18 this
rect hyponatremia. Study Design. A retrospective analysis database comprises acute-care admis-
To assess the clinical and economic was conducted in patients admitted to sions at participating hospitals, including
impact of serum sodium abnormalities 185 acute-care hospitals with a principal electronically imported or manually

sodium abnormalities in heart failure january february 2011

• 1
 abstracted demographic, clinical (eg, LOS, and cost associated with sodium mic, hyponatremic, and hypernatremic
comorbidities, vital signs, laboratory val- abnormalities, adjusting for severity on patients, those with severe hyponatre-
ues, other clinical findings), and admin- admission. We used an aggregated sever- mia were more likely to be white and to
istrative (eg, diagnosis, LOS, total costs) ity score generated from a multivariable have a systolic blood pressure 100
data. The current study was conducted logistic regression model reported previ- mm Hg, altered mental status, and
in compliance with the New England ously.18 The original model was devel- Medicare health insurance coverage
Institutional Review Board ⁄ Human oped and validated using data for (P<.005). Patients with hypernatremia
Subjects Research Committee (Welles- patients with CHF from more than 200 were more likely to be black, with mal-
ley, MA), federal regulations, and hospitals in 2000 to 2003. That model nutrition and altered mental status at
the Health Insurance Portability and included demographics, acute physio- admission.
Accountability Act. All data were ren- logic presentation on admission (vital The unadjusted hospital mortality
dered anonymous so that patient-specific signs, altered mental status, laboratory rate was 3.6% for the entire population.
information could not be identified findings, and renal function variables), Hospital mortality rate was highest in
directly or indirectly through linked and comorbidities (eg, metastatic can- patients with severe hyponatremia
identifiers. cer). For the current study, we recali- (7.6%), followed by that of patients with
Principal diagnosis of HF was defined brated the mortality risk adjustment hypernatremia (6.7%) and hyponatre-
as the presence of a primary diagnostic model for the current study population mia (4.9%) (P<.0001) (Table II).
code (International Classification of Dis- to estimate the independent contribu- In the multiple regression analysis,
eases, Ninth Revision, Clinical Modification tion of each level of sodium abnormality each sodium abnormality was indepen-
[ICD-9-CM]) for that condition. Each to mortality. For hospital LOS and cost, dently associated with mortality risk
patient was assigned to one of the fol- we used log-transformed LOS and cost higher than that in normonatremic
lowing 4 groups based on serum sodium as outcome. We used the median, 2.5th, patients (Figure 1). Compared with the
values at admission: severe hyponatre- and 97.5th percentile from 1000 boot- normonatremic group and after control-
mia, 130 mEq ⁄ L; hyponatremia, 131 strap iterations of the multivariable ling for admission severity, hospital mor-
mEq ⁄ L to 135 mEq ⁄ L; normonatremia, models for estimates of the parameter tality risk was highest for patients with
136 mEq ⁄ L to 145 mEq ⁄ L; and hyper- and 95% confidence intervals (CIs).20,21 severe hyponatremia (odds ratio [OR],
natremia, >145 mEq ⁄ L. (Serum sodium We controlled for severity of illness on 1.78; 95% CI, 1.59–1.99), followed by
values can be converted to mmol ⁄ L by admission, mortality status, interaction that of patients with hypernatremia
multiplying the value in mEq ⁄ L by 1.0). of mortality and severity of illness, and (OR, 1.55; 95% CI, 1.34–1.80), and
Patients without admission serum teaching status of the hospital. We hyponatremia (OR, 1.29; 95% CI, 1.19–
sodium values were excluded from the included additional variables other than 1.40).
analysis. The variables of patients with severity to account for potential differ- The unadjusted mean hospital LOS
serum sodium abnormalities were com- ences in resource consumption for was longest in patients with severe hyp-
pared with those of normonatremic patients in different types of hospitals onatremia (6.9 days) and shortest in
patients, who were used as the reference and for those who died in the hospital those with normonatremia (5.4 days)
group. vs those who survived hospitalization. (P<.0001). The risk-adjusted attribut-
The end points were hospital mortal- All statistical analyses were performed able hospital LOS was longer for
ity, hospital LOS, and cost of hospital- using Statistical Analysis Software (SAS patients with each type of sodium
ization. Cost per patient was calculated version 9.01; SAS Institute Inc, Cary, abnormality than for normonatremic
as total cost ⁄ charge ratio multiplied by NC). patients, ranging from 0.42 (95% CI,
total charge. Cost-charge ratios were 0.26–0.60) days in patients with hyper-
based on data from the Centers for natremia to 1.28 (95% CI, 1.11–1.47)
Medicare & Medicaid Services for each Results days in those with severe hyponatremia
hospital for the calendar years 2004 and Serum sodium abnormalities were pres- (Figure 2).
2005.19 ent on admission in 28,287 of 115,969 The unadjusted mean hospitalization
(24.4%) patients hospitalized for acute cost was highest for patients with severe
Statistical Analysis. We conducted decompensated HF. The most common hyponatremia ($11,109) and lowest for
univariable analysis on the prevalence sodium abnormality on admission those with normonatremia ($9192)
of sodium abnormality and its relation was hyponatremia (131–135 mEq ⁄ L, (P<.0001). The risk-adjusted attribut-
to mortality, LOS, and cost. We used 15.9%), followed by severe hyponatre- able hospital cost was higher for
the chi-square test for dichotomous vari- mia (130 mEq ⁄ L, 5.3%) and hyper- patients with severe hyponatremia
ables such as mortality and the Kruskal– natremia (>145 mEq ⁄ L, 3.2%). ($1132; 95% CI, $856–$1425) and
Wallis test for continuous variables such Patient baseline clinical character- hyponatremia ($509 95% CI, $361–
as LOS and costs. istics differed according to type of $653) but not significant for those with
We conducted multivariable analyses admission serum sodium abnormality hypernatremia ($99; 95% CI, $140–
to estimate the attributable mortality, (Table I). Compared with normonatre- $374) (Figure 3).
2 sodium abnormalities in heart failure january february 2011
•
Table I. Patient Characteristics at Admission by Sodium Group
SEVERE HYPONATREMIA HYPONATREMIA NORMONATREMIA HYPERNATREMIA TOTAL
VARIABLE (N=6117) (N=18,445) (N=87,682) (N=3725) (N=115,969) P VALUEa
Demographics
Male 2634 (43.1) 8536 (46.3) 40,774 (46.5) 1686 (45.3) 53,630 (46.3) <.0001
Age, mean, median (IQR) 75.4, 79 (17.0) 74.3, 77 (18.0) 74.6, 77 (17.0) 76.9, 79 (15.0) 74.7, 77 (17.0) <.0001
Race ⁄ ethnicity
White 4993 (81.6) 14,722 (79.8) 66,933 (76.3) 2620 (70.3) 89,268 (77.0) <.0001
Black 433 (7.1) 1830 (9.9) 12,133 (13.8) 751 (20.2) 15,147 (13.1) <.0001
Other 441 (7.2) 1231 (6.7) 5189 (5.9) 185 (5.0) 7046 (6.1) <.0001
Unknown 250 (4.1) 662 (3.6) 3427 (3.9) 169 (4.5) 4508 (3.9) <.0001
Admitted from health 1064 (17.4) 2996 (16.2) 12,766 (14.6) 476 (12.8) 17,302 (14.9) <.0001
care facility
Payers
Medicare 4052 (66.2) 11,944 (64.8) 54,176 (61.8) 2361 (63.4) 72,533 (62.6) <.0001
Medicaid 241 (3.9) 853 (4.6) 3624 (4.1) 110 (3) 4828 (4.2) <.0001
Health maintenance 993 (16.2) 3065 (16.6) 16,244 (18.5) 710 (19.1) 21,012 (18.1) <.0001
organization
Commercial insurance 90 (1.5) 371 (2.0) 1714 (2.0) 70 (1.9) 2245 (1.9) <.0001
Self-pay 119 (1.9) 380 (2.1) 2104 (2.4) 88 (2.4) 2691 (2.3) <.0001
Other ⁄ unknown 622 (10.2) 1832 (9.9) 9820 (11.2) 386 (10.4) 12,660 (10.9) <.0001
Comorbidities and medical history
COPD 2015 (32.9) 6360 (34.5) 30,368 (34.6) 1362 (36.6) 40,105 (34.6) <.003
Metastatic cancer 225 (3.7) 581 (3.1) 2094 (2.4) 69 (1.9) 2969 (2.6) <.0001
Severe malnutrition 204 (3.3) 482 (2.6) 1635 (1.9) 131 (3.5) 2452 (2.1) <.0001
Respiratory effusion 2062 (33.7) 5763 (31.2) 24,414 (27.8) 1160 (31.1) 33,399 (28.8) <.0001
Chronic liver disease 163 (2.7) 363 (2.0) 955 (1.1) 41 (1.1) 1522 (1.3) <.0001
Chronic pulmonary heart 823 (13.5) 2416 (13.1) 10,449 (11.9) 420 (11.3) 14,108 (12.2) <.0001
disease
Chronic renal failure ⁄ urine 271 (4.4) 868 (4.7) 3094 (3.5) 140 (3.8) 4373 (3.8) <.0001
calc
Cardiomyopathy 1145 (18.7) 3231 (17.5) 14,864 (17.0) 634 (17.0) 19,875 (17.1) .002
Immunocompromised 424 (6.9) 1321 (7.2) 5511 (6.3) 227 (6.1) 7483 (6.5) <.0001
Laboratory and physiologic findings
Albumin 2.7 g ⁄ dL 555 (9.0) 1429 (7.7) 3913 (4.5) 210 (5.6) 6107 (5.2) <.0001
Arterial pCO2 35 or 595 (9.7) 1491 (8.1) 5545 (6.3) 365 (9.8) 7996 (6.9) <.0001
>60 mm Hg
Serum urea nitrogen 1209 (19.8) 3306 (17.9) 9961 (11.4) 597 (16.0) 15,073 (13.0) <.0001
>50 mg ⁄ dL
Creatinine 2.0 mg ⁄ dL 1307 (21.4) 4127 (22.4) 15,325 (17.5) 833 (22.4) 21,592 (18.6) <.0001
Troponin I > 1 ng ⁄ mL or 312 (5.1) 879 (4.8) 3403 (3.9) 230 (6.2) 4824 (4.2) <.0001
CKMB > 9 ng ⁄ mL
Temperature 95 or >100F 549 (9.0) 1408 (7.6) 5454 (6.2) 279 (7.5) 7690 (6.6) <.0001
Pulse 99 beats per min 1946 (31.8) 6108 (33.2) 27,035 (30.8) 1228 (33.0) 36,317 (31.3) <.0001
Systolic blood pressure 1664 (27.2) 4399 (23.9) 13,109 (14.9) 494 (13.3) 19,666 (17.0) <.0001
100 mm Hg
Altered mental statusb
Mild 844 (13.8) 2034 (11.0) 8202 (9.4) 636 (17.1) 11,716 (10.1) <.0001
Moderate or severe 140 (2.3) 312 (1.7) 1273 (1.5) 169 (4.5) 1894 (1.6) <.0001
Hospital characteristics
Teaching 3434 (56.1) 10,232 (55.5) 50,796 (57.9) 2316 (62.2) 66,778 (57.6) <.0001
Urban 5051 (82.6) 15,058 (81.6) 75,438 (86.0) 3450 (92.6) 98,997 (85.4) <.0001
Size
<100 beds 667 (10.9) 2107 (11.4) 7919 (9.0) 216 (5.8) 10,909 (9.4) <.0001
100–299 beds 3191 (52.2) 9481 (51.4) 44,369 (50.6) 2119 (56.9) 59,160 (51.0)
>299 beds 2259 (36.9) 6857 (37.2) 35,394 (40.4) 1390 (37.3) 45,900 (39.6)
Abbreviations: CKMB, creatinine kinase Mb; COPD, chronic obstructive pulmonary disease; IQR, interquartile range; pCO2, partial
pressure of carbon dioxide. aP value based on global chi-square test. bAltered mental status based on Glasgow Coma Scale (GCS) as
mild: GCS score of 10 to 14 or disoriented ⁄ lethargy; moderate: GCS score of 5 to 9; or severe: GCS score of 5 or comatose. Values
are expressed as number (percentage) unless otherwise indicated.

Discussion decompensated HF confirms the results higher hospital mortality rates and
This analysis of a large contemporary of previous studies6–9 that show the asso- longer hospitalizations in this high-risk
population of patients with acute ciations between hyponatremia and population. In addition, this study yields
sodium abnormalities in heart failure january february 2011
• 3
 Table II. Univariable Analysis on Mortality, Length of Stay, and Cost by Serum Sodium Level at Admissiona
SEVERE HYPONATREMIA HYPONATREMIA NORMONATREMIA HYPERNATREMIA TOTAL
VARIABLE (N=6117) (N=18,445) (N=87,682) (N=3725) (N=115,969)
In-hospital mortality
No. (%) 466 (7.6) 906 (4.9) 2513 (2.9) 250 (6.7) 4135 (3.6)
P value <.0001 <.0001 Reference <.0001
Length of hospital stay, d
Mean  standard 6.9  7.3 6.1  5.6 5.4  5.2 6.0  5.2 5.6  5.4
deviation
Median (1st and 5 (3, 8) 5 (3, 7) 4 (3, 7) 5 (3, 7) 4 (3, 7)
3rd quartile)
P value <.0001 <.0001 Reference <.0001
Hospital cost, $
Mean  standard 11,109  19,229 10,033  15,265 9192  13,523 9418  13,074 9433  14,160
deviation
Median (1st and 6346 (4005, 11,067) 6006 (3748, 10,400) 5596 (3566, 9458) 5994 (3808, 10,118) 5708 (3624, 9727)
3rd quartile)
P value <.0001 <.0001 Reference <.0001
a
All P values represent comparisons of patients in abnormal sodium groups with those in the normonatremia group.

independent predictor of increased

hospital mortality rates is similar to the
findings of OPTIME-CHF,6 OPTI-
MIZE-HF,9 and ESCAPE.8 Notably, a
recent evaluation22 of 145 patients hos-
pitalized for acute decompensated HF
showed that increased central venous
pressure, but not measures of systolic
performance, was independently associ-
ated with worsening renal function and
unfavorable outcomes. In addition, work
by Androne and colleagues23 indicates
that chronic HF patients with hypervo-
lemia by quantitative blood volume
determination have dramatically worse
outcomes than those with normovol-
emia or hypovolemia. These results sup-
port the notion that hypervolemia by
itself increases mortality.24,25 Therefore,
it is plausible that hyponatremia predicts
Figure 1. Multivariable-adjusted mortality odds ratio and 95% confidence intervals (CIs)
associated with serum sodium abnormality at admission. unfavorable outcomes because it is a
marker of hypervolemia and of worsen-
new findings. First, even mild hyponat- studies, including the Outcomes of a Pro- ing renal function due to venous conges-
remia (131–135 mEq ⁄ L) is associated spective Trial of Intravenous Milrinone tion. In healthy individuals, 85% of
with unfavorable outcomes in patients for Exacerbations of Chronic Heart total plasma volume is present in the
hospitalized for HF. Second, in patients Failure (OPTIME-CHF; 135 sodium venous circulation and 15% in the arte-
with acute decompensated HF, not only mEq ⁄ L, 27% of 949 patients),6 the Orga- rial vessels. Decreased arterial volume in
hyponatremia but also hypernatremia nized Program to Initiate Lifesaving the setting of HF inactivates high-
(>145 mEq ⁄ L) is associated with in- Treatment in Hospitalized Patients pressure arterial baroreceptors, reduces
creased mortality and prolonged hospital With Heart Failure (OPTIMIZE-HF; tonic inhibition of afferent parasympa-
LOS. Third, hyponatremia increases sodium <134 mEq ⁄ L, 19.7% of 38,279 thetic input to the central nervous
considerably the soaring costs of hospi- patients),9 and the Evaluation Study of system, and potentiates sympathetic
talization for HF. Congestive Heart Failure and Pulmonary efferent tone, which, in turn, activates
The percentage of hospitalized HF Artery Catheter Effectiveness (ESCAPE; the renin-angiotensin-aldosterone sys-
patients with hyponatremia at admission <134 mEq ⁄ L, 23.8% of 433 patients).8 tem and induces nonosmotic release of
(21.2% of almost 116,000 patients) In this study, the emergence of vasopressin.26 In the kidney, angiotensin
was within the range reported in other hyponatremia at admission as an II–mediated renal efferent arteriolar
4 sodium abnormalities in heart failure january february 2011
•
 increases renal elimination of free water,
decreases renal sympathetic tone, and
promotes natriuretic peptides secretion.
In patients with HF, these atrial-renal
reflexes are overwhelmed by neurohor-
monal activation. This explains the con-
tinued renal sodium and water retention
even with elevated atrial pressures.26
Venous congestion extends to the renal
veins and worsens impairment of glo-
merular filtration rate. In fact, in the
ESCAPE trial, right atrial pressure was
the only hemodynamic measure corre-
lated with baseline renal function,
which independently predicts mortality
and HF hospitalization.29 Loop diuretics
block sodium chloride uptake in the
macula densa, and thus promote rennin
secretion independent of any effect on
Figure 2. Multivariable-adjusted attributable length of stay associated with serum sodium sodium and water reabsorption.30 Thus,
abnormality at admission. CIs indicates confidence intervals. the hyponatremia associated with acute
HF may simultaneously reflect escala-
tion of neurohormonal activity and the
detrimental effects of intense diuretic
therapy on the kidney.
It has been known for more than 2
decades that compared with normo-
natremic HF patients, those with hypo-
natremia have higher circulating levels
of neurohormones, impaired orthostatic
responses, and decreased hepatic and
renal function. In addition, hyponatre-
mia has been correlated to increased
ventricular ectopy, sudden death, and
susceptibility to the hypotensive and
azotemic effects of angiotensin-convert-
ing enzyme inhibitors.
Worsening renal function has repeat-
edly emerged as an independent predic-
tor of poor outcomes in hospitalized HF
patients, and severity of renal dysfunc-
tion at baseline and during hospitaliza-
Figure 3. Multivariable-adjusted attributable cost associated with serum sodium abnormal- tion has been correlated to intensity of
ity at admission. CIs indicates confidence intervals. diuretic therapy.29,31–34 Importantly, the
analysis from OPTIME-CHF showed
vasoconstriction decreases renal blood nephron.26 Notably, increased proximal that lower serum sodium was correlated
flow and increases filtration fraction. In sodium reabsorption decreases distal with higher serum urea nitrogen (BUN)
concert with renal nerve stimulation, sodium and water delivery, a signal for level and lower blood pressure,6 the two
the increased peritubular capillary onco- macula densa cells to secrete renin, fur- strongest predictors of hospital mortality
tic pressure and reduced peritubular cap- ther escalating neurohormonal activa- in patients enrolled in the Acute De-
illary hydrostatic pressure enhance tion.28 The additional sodium and water compensated Heart Failure National
sodium reabsorption in the proximal reabsorbed by the kidney predominantly Registry (ADHERE).33 Decreased renal
tubule. Angiotensin II also directly stim- fills the compliant venous circulation, perfusion and neurohormonal activation
ulates proximal sodium reabsorption27 increasing central venous and atrial lead to enhanced proximal tubule reab-
and aldosterone secretion, which increa- pressures. Normally, an increase in atrial sorption of sodium, water, and urea.
ses sodium reabsorption in the distal pressure inhibits vasopressin release and Binding of vasopressin to V2 receptors
sodium abnormalities in heart failure january february 2011
• 5
 in the inner medullary–collecting ducts 20% of cases, our findings suggest that date. The analysis also showed that even
results in activation of urea transporters, sodium abnormalities increase by at least mild hyponatremia is associated with
which enhance its reabsorption. Thus, in $600 million the already enormous costs poorer hospital outcomes in patients
HF patients, enhanced neurohormonal of inpatient HF care. Furthermore, this with acute decompensated HF. Further-
activation may explain both elevated figure does not account for the costs of more, the study clearly showed that hy-
BUN levels and hyponatremia. re-hospitalization, the rate of which is pernatremia is associated with higher
Data from the ESCAPE trial8 show higher in hyponatremic than in nor- hospital mortality. Finally, this is the
that, despite similar clinical improve- monatremic HF patients.8 first study to estimate the excess cost
ments, patients with persistent hypo- specifically attributable to admission
natremia had higher post-discharge Limitations serum sodium abnormalities.
mortality and HF re-hospitalization than First, use of a retrospective design
those whose serum normalized during restricted our analysis to existing data;
hospitalization. These findings support however, a large database could provide Acknowledgments: Study concept
the concept that low serum sodium con- and design: Shorr, Tabak, Gupta,
a more accurate estimate as evident by
centration in the setting of congestion Johannes. Acquisition of data: Tabak.
the narrow CIs in this study. The num-
Analysis and interpretation of data: Shorr,
identifies patients with greater neurohor- ber of patients included in this study was Tabak, Johannes, Gupta, Saltzberg,
monal activation and severity of illness. more than twice that of the next largest Costanzo. Drafting the manuscript: Shorr,
Alternatively, the finding that persistent study.9 Second, morbidity and mortality Tabak, Gupta, Saltzberg, Costanzo. Critical
hyponatremia increases the risk of post- after discharge were not evaluated, but revision of the manuscript for important
discharge adverse events suggests that previous analyses indicated that baseline intellectual content: Shorr, Tabak, Gupta,
hyponatremia per se may contribute to sodium abnormalities predict both hos- Johannes, Saltzberg, Costanzo. Statistical
the pathogenesis of HF. Many studies pital and long-term outcomes, not only expertise: Shorr, Tabak. Obtained funding:
have conclusively shown that hyponat- in patients with HF but also in individu- Gupta, Tabak. Administrative, technical, or
remia alters structure and function of als with acute coronary syndromes.6,9 material support: Tabak, Gupta. Study
central nervous system cells.35 supervision: Shorr, Tabak. Dr Tabak has full
Third, only admission serum sodium lev-
A unique finding of this analysis is access to the data and takes responsibility
els were considered, so differential effects
for data integrity and accuracy. We thank
that serum sodium levels >145 mEq ⁄ L of persistent vs corrected hyponatremia Karen Derby and Xiaowu Sun, PhD, from
had adverse effects similar to those could not be determined. Fourth, study CareFusion Clinical Research Services for
occurring in patients with severe hypo- patients were not stratified according to their dedicated contributions in database
natremia. In fact, the risk of hospital decreased or preserved systolic function. management, statistical analysis, and
mortality was second only to that associ- However, a recent study36 prospectively technical support. We also thank Cindy W.
ated with severe hyponatremia. The assessed the impact of baseline natremia Hamilton, PharmD, ELS (Hamilton House,
patients in this analysis were elderly, and changes in sodium level during hos- Virginia Beach, VA), for editorial
with an average age of approximately pitalization in 358 patients surviving first assistance. Hamilton House received
75 years. Elderly CHF patients may be hospitalization for HF with preserved compensation from CareFusion for its
at higher risk for developing hypernatre- contributions.
systolic function. At 7 years, survival
mia resulting from an impaired ability to was higher in normonatremic than in
replace water losses due to greater physi- hyponatremic patients (63% vs 31%).36 Disclosure: The preliminary results of
cal and mental functional limitations As in patients with systolic dysfunction, this study were presented at the 27th
that restrict access to water in addition persistent hyponatremia portends a Annual Scientific Session of the American
to the greater risk of infection in this poorer prognosis than corrected hypo- College of Cardiology; March 29 to April
population. The finding in our analysis natremia in HF patients with normal 1, 2008; Chicago. Dr Shorr has received
that black patients were more likely to systolic function. Fifth, cost was based research funding and honoraria from
be in the hypernatremia group than on charges for each hospital and calcu- Astellas Pharma US, Inc and serves as a
other racial ⁄ ethnic groups calls for fur- lated using the cost ⁄ charge ratio.19 consultant to Astellas Pharma US, Inc.
ther study on the potential genetic com- Sixth, treatment data were not avail- Dr Costanzo and Dr Saltzberg served as
ponent of this clinical condition among consultants for this study and received
able, so our analysis did not include this
CHF patients. modest compensation from CareFusion, Inc
important variable. Finally, we were not
for the time spent working directly on this
This is the first evaluation of sodium able to evaluate out-of-hospital events manuscript. Dr Costanzo also served as a
abnormalities in HF patients to include or rehospitalization. consultant and speaker for CHF Solutions,
an assessment of the excess hospitaliza- St Jude, and received research grant from
tion costs specifically attributable to this Conclusions Cardiomems and Paracor. Dr Saltzberg
electrolyte disorder. Assuming that 1 This analysis is valuable because it dem- received modest consulting fees and
million hospitalizations occur yearly due onstrated the impact of admission serum honoraria from CHF Solutions. Drs Tabak,
to acute decompensated HF and that sodium abnormalities in the largest con- Gupta, and Johannes are employees of
hyponatremia occurs in approximately temporary HF population reported to CareFusion Inc Per contract with Astellas

6 sodium abnormalities in heart failure january february 2011

•
Pharma US, Inc, CareFusion Inc generated analysis) in collaboration with principal Funding: This study was supported by
the research data set from one of its Clinical investigators, All authors participated the Astellas Pharma US, Inc.
Research Databases and provided research study and manuscript drafting and ⁄ or
services (database management, statistical critical revision.

REFERENCES
1 Curtis LH, Whellan DJ, Hammill BG, et al. 12 Iezzoni LI, Moskowitz MA. A clinical assess- 24 Drazner MH, Rame JE, Stevenson LW, Dries
Incidence and prevalence of heart failure in ment of MedisGroups. JAMA. 1988;260(21): DL. Prognostic importance of elevated jugular
elderly persons, 1994–2003. Arch Intern 3159–3163. venous pressure and a third heart sound in
Med. 2008;168(4):418–424. 13 Fine MJ, Auble TE, Yealy DM, et al. A patients with heart failure. N Engl J Med.
2 Fang J, Mensah GA, Croft JB, Keenan NL. prediction rule to identify low-risk patients 2001;345(8):574–581.
Heart failure-related hospitalization in the with community-acquired pneumonia. N Engl 25 Gheorghiade M, Gattis WA, O’Connor CM,
U.S., 1979 to 2004. J Am Coll Cardiol. J Med. 1997;336(4):243–250. et al. Effects of tolvaptan, a vasopressin
2008;52(6):428–434. 14 Kollef MH, Shorr A, Tabak YP, et al. Epide- antagonist, in patients hospitalized with wors-
3 Hugli O, Braun JE, Kim S, et al. United States miology and outcomes of health-care-associ- ening heart failure: a randomized controlled
emergency department visits for acute decom- ated pneumonia: results from a large US trial. JAMA. 2004;291(16):1963–1971.
pensated heart failure, 1992 to 2001. Am J database of culture-positive pneumonia. 26 Schrier RW. Role of diminished renal function
Cardiol. 2005;96(11):1537–1542. Chest. 2005;128(6):3854–3862. in cardiovascular mortality: marker or patho-
4 Levy D, Kenchaiah S, Larson MG, et al. Long- 15 Shorr AF, Tabak YP, Killian AD, et al. Health- genetic factor? J Am Coll Cardiol. 2006;
term trends in the incidence of and survival care-associated bloodstream infection: a 47(1):1–8.
with heart failure. N Engl J Med. 2002; distinct entity? Insights from a large U.S. data- 27 Schrier RW. Body fluid volume regulation in
347(18):1397–1402. base. Crit Care Med. 2006;34(10):2588– health and disease: a unifying hypothesis.
5 Lloyd-Jones D, Adams R, Carnethon M, et al. 2595. Ann Intern Med. 1990;113(2):155–159.
Heart disease and stroke statistics – 2009 16 Shorr AF, Tabak YP, Gupta V, et al. Morbidity 28 Castrop H, Schweda F, Mizel D, et al.
update: a report from the American Heart and cost burden of methicillin-resistant Staph- Permissive role of nitric oxide in macula
Association Statistics Committee and Stroke ylococcus aureus in early onset ventilator-asso- densa control of renin secretion. Am J Physiol
Statistics Subcommittee. Circulation. 2009; ciated pneumonia. Crit Care. 2006;10(3): Renal Physiol. 2004;286(5):F848–F857.
119(3):e21–e181. R97. 29 Nohria A, Hasselblad V, Stebbins A, et al.
6 Klein L, O’Connor CM, Leimberger JD, et al. 17 Silber JH, Rosenbaum PR, Schwartz JS, et al. Cardiorenal interactions: insights from the
Lower serum sodium is associated with Evaluation of the complication rate as a ESCAPE trial. J Am Coll Cardiol. 2008;
increased short-term mortality in hospitalized measure of quality of care in coronary artery 51(13):1268–1274.
patients with worsening heart failure: results bypass graft surgery. JAMA. 1995;274(4): 30 Ellison DH, Berl T. Clinical practice. The
from the Outcomes of a Prospective Trial of 317–323. syndrome of inappropriate antidiuresis. N
Intravenous Milrinone for Exacerbations of 18 Tabak YP, Johannes RS, Silber JH. Using Engl J Med. 2007;356(20):2064–2072.
Chronic Heart Failure (OPTIME-CHF) study. automated clinical data for risk adjustment: 31 Gottlieb SS, Abraham W, Butler J, et al. The
Circulation. 2005;111(19):2454–2460. development and validation of six disease- prognostic importance of different definitions
7 Rossi J, Bayram M, Udelson JE, et al. Improve- specific mortality predictive models for pay- of worsening renal function in congestive heart
ment in hyponatremia during hospitalization for-performance. Med Care. 2007;45(8): failure. J Card Fail. 2002;8(3):136–141.
for worsening heart failure is associated with 789–805. 32 Krumholz HM, Chen YT, Vaccarino V, et al.
improved outcomes: insights from the Acute 19 Killian AD, Gupta V, Goetz AE. Cost analysis Correlates and impact on outcomes of wors-
and Chronic Therapeutic Impact of a Vaso- of erythropoietic-stimulating therapy dosing in ening renal function in patients 65 years of
pressin Antagonist in Chronic Heart Failure oncology inpatients. Ann Pharmacother. 2006; age with heart failure. Am J Cardiol.
(ACTIV in CHF) trial. Acute Card Care. 40(3):421–426. 2000;85(9):1110–1113.
2007;9(2):82–86. 20 Basu A, Rathouz PJ. Estimating marginal and 33 Fonarow GC, Adams KF Jr, Abraham WT,
8 Gheorghiade M, Rossi JS, Cotts W, et al. incremental effects on health outcomes using et al. Risk stratification for in-hospital mortality
Characterization and prognostic value of flexible link and variance function models. in acutely decompensated heart failure: clas-
persistent hyponatremia in patients with se- Biostatistics. 2005;6(1):93–109. sification and regression tree analysis. JAMA.
vere heart failure in the ESCAPE Trial. Arch 21 Efron B, Tibshirani R. An Introduction to 2005;293(5):572–580.
Intern Med. 2007;167(18):1998–2005. the Bootstrap. London, UK: Chapman & Hall; 34 Butler J, Forman DE, Abraham WT, et al.
9 Gheorghiade M, Abraham WT, Albert NM, 1993. Relationship between heart failure treatment
et al. Relationship between admission serum 22 Mullens W, Abrahams Z, Francis GS, et al. and development of worsening renal function
sodium concentration and clinical outcomes Importance of venous congestion for worsen- among hospitalized patients. Am Heart J.
in patients hospitalized for heart failure: an ing of renal function in advanced decompen- 2004;147(2):331–338.
analysis from the OPTIMIZE-HF registry. Eur sated heart failure. J Am Coll Cardiol. 2009; 35 Verbalis JG. Hyponatremia: epidemiology,
Heart J. 2007;28(8):980–988. 53(7):589–596. pathophysiology, and therapy. Curr Opin
10 Upadhyay A, Jaber BL, Madias NE. Incidence 23 Androne AS, Hryniewicz K, Hudaihed A, Nephrol Hypertens. 1993;2(4):636–652.
and prevalence of hyponatremia. Am J Med. et al. Relation of unrecognized hyper- 36 Rusinaru D, Buiciuc O, Leborgne L, et al.
2006;119(7 suppl 1):S30–S35. volemia in chronic heart failure to clinical Relation of serum sodium level to long-term
11 Ghali JK. Hyponatraemia in heart failure: a status, hemodynamics, and patient out- outcome after a first hospitalization for heart
call for redefinition. Eur Heart J. 2007; comes. Am J Cardiol. 2004;93(10):1254– failure with preserved ejection fraction. Am J
28(8):920–921. 1259. Cardiol. 2009;103(3):405–410.

sodium abnormalities in heart failure january february 2011

• 7