Journal of Cardiac Failure Vol. 13 No.

3 2007

Clinical Investigations

The Effect of Spironolactone Use on Heart Failure

Mortality: A Population-Based Study
MARAL OUZOUNIAN, MD,1 ANSAR HASSAN, MD,1 JAFNA L. COX, MD, FRCPC, FACC,2
DAVID E. JOHNSTONE, MD, FRCPC, FACC,2 AND JONATHAN HOWLETT, MD, FRCPC, FACC, FSCAI,2
FOR THE IMPROVING CARDIOVASCULAR OUTCOMES IN NOVA SCOTIA STUDY INVESTIGATORS
Halifax, Nova Scotia, Canada

ABSTRACT
Background: Spironolactone use for heart failure (HF) has increased dramatically after the publication of
the Randomized Aldactone Evaluation Study trial; yet, few studies have examined its real-world impact.
We aimed to determine the population effect of spironolactone use on mortality in HF patients discharged
from hospital.
Methods and Results: All patients discharged alive between October 1997 and December 2001 in Nova
Scotia, Canada, with a primary diagnosis of HF were enrolled in the Improving Cardiovascular Outcomes
Study. Two year, all-cause mortality was the primary end point. A total of 7816 patients were identified, of
whom 644 (8%) were discharged home on spironolactone. After adjusting for differences in clinical covariates, spironolactone use did not emerge as an independent predictor of long-term survival (OR 0.97,
P 5 .80). When only the subgroup of patients enrolled in a HF clinic were included (n 5 990), spironolactone use was associated with reduced rates of all-cause mortality at 2 years (OR 0.52, P 5 .003).
Conclusions: Although spironolactone use was not associated with improved long-term survival in the
general HF population, it was associated with improved long-term survival in patients enrolled in HF
clinics. These data highlight the challenges of knowledge translation from a clinical trial into practice.
(J Cardiac Fail 2007;13:165e169)
Key Words: Aldosterone antagonism, Mortality, Congestive heart failure.

Heart failure (HF) is a major public health problem, with

more than 500,000 incident cases diagnosed each year in
the United States alone.1 It is associated with significant
morbidity and mortality, thus placing a considerable burden
upon the health care system. In recent years, the medical
management of HF has been revolutionized by the publication of several landmark randomized controlled trials,
which have, in turn, led to the inclusion of angiotensin converting enzyme (ACE) inhibitors, b-blockers, and spironolactone within current HF treatment guidelines.2,3
The Randomized Aldactone Evaluation Study (RALES)
demonstrated a significant reduction in the relative risk of
mortality among low-dose spironolactone-treated patients
with New York Heart Association (NYHA) Class III or
IV heart failure.4 After the publication of RALES, there
was a widespread increase in the use of spironolactone
for the treatment of HF.5 However, little actually is known
regarding the appropriateness of spironolactone use and its

From the 1Division of Cardiac Surgery, Department of Surgery and

Division of Cardiology, Department of Medicine, Dalhousie University,
and Queen Elizabeth II Health Sciences Centre, Halifax, Nova Scotia,
Canada.
Manuscript received May 23, 2006; revised manuscript received November 23, 2006; revised manuscript accepted November 28, 2006.
Reprints requests: Jonathan Howlett MD, FRCPC, FACC, FSCAI, Room
6896, 1796 Summer Street, Halifax, Nova Scotia, Canada B3H 3A7.
Dr. Cox receives salary support from a Canadian Institutes of Health Research/Regional Partnership Program Investigator Award and through
a Clinical Research Scholarship from the Faculty of Medicine, Dalhousie
University. The QE II Heart Function Clinic was supported through an unrestricted grant from AstraZeneca Canada Inc. The ICONS study was supported through a nondirected educational grant from Merck Frosst Canada
Inc., and through in-kind support from the Queen Elizabeth II Health Sciences Centre, Halifax, Nova Scotia, Canada, and the Nova Scotia Department of Health. Since January 2003, it has been entirely funded through
the Nova Scotia Department of Health.
Originally presented as an oral abstract at the 2005 American College of
Cardiology Annual Meeting in Orlando, Florida.
1071-9164/$ - see front matter
2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.cardfail.2006.11.016
2

165

166 Journal of Cardiac Failure Vol. 13 No. 3 April 2007

impact on long-term outcomes in HF patients outside a clinical trial setting. The purpose of this study was to determine
the real-world effect of spironolactone use on long-term
mortality in a population of patients discharged from hospital with HF.

coronary syndrome. Long-term mortality data were censored at

2 years.

Statistical Analysis

The province of Nova Scotia has a population of approximately

940,000 persons. It has among the highest provincial rates of
cardiovascular disease-associated mortality and cardiovascular
disease-related risk-factors in Canada.6 Our primary data source
was the Improving Cardiovascular Outcomes in Nova Scotia (ie,
ICONS) project, a large, prospective cohort disease management
study established in October 1997, the rationale and methods of
which are described elsewhere.7 The study was approved by the
Queen Elizabeth II Health Sciences Centre Research Ethics
Board. Since its inception, it has captured all admissions to hospital for the diagnosis of acute myocardial infarction (MI), unstable
angina, HF, and atrial fibrillation in Nova Scotia. Detailed information pertaining to clinical, demographic, and socioeconomic
and laboratory measures were collected prospectively on all patients. Included in this study were all patients discharged alive between October 1997 and December 2001 with a primary discharge
diagnosis of HF and with at least 2 years of follow-up or confirmed death within those 2 years. We excluded patients who
were hospitalized with a primary diagnosis of acute MI or acute

Patients who were discharged from hospital on spironolactone

were compared with those who were not, in terms of baseline
demographic and clinical characteristics, using analysis of variance for continuous variables and Pearsons chi-square analysis
for categoric variables. The primary outcome was all-cause
mortality at 2 years. Multivariable analysis included multiple logistic regression modeling techniques. To determine which variables were entered into the final model, univariate comparisons
were carried out comparing patients who were alive at 2 years
and those who were not (Table 2). Those variables that differed
between the 2 groups at a P value of less than .2 were included
in the model-building process. To determine whether the impact
of discharge spironolactone had remained constant over time,
we also analyzed survival effects with a Kaplan Meier analysis.
A subgroup of the patients discharged from hospital also was
followed by 1 of our 4 specialized HF clinics in the province of
Nova Scotia. We sought to determine the effect of spironolactone use on survival in this cohort of patients with more structured follow-up. Finally, to better understand the mechanisms
behind any survival effect of spironolactone, we repeated this
analysis on the subgroup of hospitalized patients who best fulfilled RALES criteria by our available data: creatinine ! 2.5
mg/dL, K ! 5.0 mM, and an ejection fraction (EF) less
than 35%.

Table 1. Baseline Variables According to Spironolactone

Prescription at Hospital Discharge

Table 2. Baseline Variables According to 2-Year Mortality

Methods
Study Cohort and Protocol

Expired at 2 Years

Discharged on Spironolactone
Variable
Mean age, yrs (SD)
Male gender (%)
EF performed
within 60 days (%)
EF ! 35 % (%)
Diabetes (%)
Hyperlipidemia (%)
Smoking (% ever)
Peripheral vascular
disease (%)
Creatinine O
2.5 mg/dL (%)
Admission K O
5 mmol/L (%)
Loop diuretics (%)
ACE inhibitors (%)
Angiotensin receptor
blockers (%)
Digoxin (%)
b-blockers (%)
Statins (%)
NSAIDs (%)

No
(n 5 7172)

Yes
(n 5 644)

P Value

Variable

75.4 (11.4)
3377 (47.1)
3123 (43.5)

73.0 (12.8)
344 (53.4)
289 (44.9)

!.0001
.002
.51

795
2515
1668
3148
737

128
263
162
297
66

!.0001
.003
0.28
.28
.98

Age O 77 years (%)

EF performed
within 60 days (%)
EF ! 35 % (%)
Diabetes (%)
Hyperlipidemia (%)
Smoking (% ever)
Peripheral vascular
disease (%)
Creatinine O
2.5 mg/dL (%)
Admission K O
5 mmol/L (%)
Loop diuretics (%)
ACE inhibitors (%)
ARBs (%)
Digoxin (%)
b-blockers (%)
Statins (%)
Calcium channel
blockers (%)
NSAIDs (%)
Spironolactone (%)

(41.7)
(35.1)
(23.2)
(43.9)
(10.3)

(62.8)
(40.8)
(25.2)
(46.1)
(10.2)

639 (9.1)

40 (6.3)

.02

735 (10.4)

60 (9.4)

.32

5277 (73.6)
4577 (63.8)
468 (6.5)

562 (87.3)
485 (75.3)
59 (9.2)

!.0001
!.0001
.01

2144
3565
1239
182

278
360
144
13

!.0001
.003
.001
.42

(29.9)
(49.7)
(17.3)
(2.5)

(43.2)
(55.9)
(22.4)
(2.0)

All medications prescribed at hospital discharge, other parameters on

admission. Ejection fraction, and creatinine were also significant
continuous predictors of mortality.
SD, standard deviation; EF, ejection fraction; ACE, angiotensinconverting enzyme; NSAID, nonsteroidal anti-inflammatory drug.

No
(n 5 4975)

Yes
(n 5 2841)

P Value

2046 (41.2)
2411 (48.5)

1698 (60.0)
1001 (35.2)

!.0001
!.0001

606
1726
1368
2676
458

317
1052
462
1690
345

(50.8)
(37.0)
(16.3)
(59.6)
(12.1)

!.0001
.04
!.0001
!.0001
!.0001

281 (5.8)

398 (14.3)

!.0001

421 (8.6)

375 (13.4)

!.0001

3612
3310
336
1404
2681
1055
1295

(40.7)
(34.7)
(27.5)
(53.8)
(9.2)

(72.5)
(66.5)
(6.8)
(28.2)
(53.9)
(21.2)
(26.0)

126 (2.5)
405 (8.1)

2227
1752
191
1018
1244
328
785

(78.4)
(61.7)
(6.7)
(35.8)
(43.8)
(11.6)
(27.6)

69 (2.4)
239 (9.2)

!.0001
!.0001
.96
!.0001
!.0001
!.0001
.12
.78
.67

All medications prescribed at hospital discharge, other parameters on

admission.
EF, ejection fraction; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; NSAID, nonsteroidal anti-inflammatory drug.

Spironolactone and Heart Failure Mortality

Statistical significance was indicated by a P value of less than
.05. All statistical analyses were performed using SAS version
8.2 (SAS Institute Inc, Cary, NC).

Results
A total of 7816 patients were discharged from hospitals
in Nova Scotia with a primary diagnosis of HF during the
observation period. Their average age was 76 years; 52%
were female; and their median EF was 40%. Of these patients, 644 (8%) had been discharged on spironolactone.
Patients discharged from hospital on spironolactone were
younger and were more likely to be male and have diabetes
(Table 1). Their mean EF was 31.5%, although less than
half of those discharged on spironolactone had had a measure of their EF within the 60 days preceding their discharge. Furthermore, 6% had a creatinine O 2.5 mg/dL,
9% had K O 5.0 mmol/L on admission, and 5% were
given concomitant nonsteroidal anti-inflammatory drug
therapy. Of the 644 patients who were prescribed spironolactone at hospital discharge, 326 (51%) either had a documented serum creatinine O2.5 mg/dL, a serum K O5.5
mmol/L, or no documentation of left ventricular EF
!35% within 60 days of hospital admission.
At 2 years, rates of all-cause mortality were similar between patients who had been discharged on spironolactone
and those who had not been (37.1% vs. 36.3%, P 5 .67).
After adjusting for differences between the 2 groups, we
found that spironolactone use did not emerge as an independent predictor of long-term survival (OR 0.97, 95% confidence interval [CI] 0.79e1.20) (Table 3). A time-to-event
analysis yielded similar results (Fig. 1).
A total of 990 patients with a discharge diagnosis of HF
were followed in 1 of 4 specialized HF clinics during the
study period (Table 4). As a group, they were slightly younger than the overall HF population. Furthermore, clinic

Table 3. Logistic Regression Model for 2-Year Mortality

Effect
Spironolactone
Age (per year)
Female gender
Smoker (ever)
Diabetes
Hyperlipidemia
Peripheral
vascular disease
Serum creatinine
(per mg/dL)
Loop diuretics
Angiotensin-converting
enzyme inhibitor
b-blocker
Statin

Point
Estimate

95%
Confidence
Limits
1.24)
1.05)
0.92)
1.34)
1.46)
0.96)
1.92)

P
Value

0.97
1.04
0.79
1.16
1.16
0.77
1.57

(0.76,
(1.03,
(0.69,
(1.00,
(0.93,
(0.61,
(1.28,

.80
!.0001
.002
.05
.19
.02
!.0001

1.20

(1.11, 1.30)

.001

1.20
0.80

(1.02, 1.41)
(0.69, 0.94)

.03
.005

0.79
0.73

(0.68, 0.91)
(0.57, 0.95)

.001
.02

All medications prescribed at hospital discharge, laboratory and parameters on admission.

Ouzounian et al

167

patients were noted to have higher rates of concurrent treatment with other HF medications, including loop diuretics,
ACE inhibitors, and b-blockers. Of these patients, 11%
had been discharged from hospital on spironolactone. Of
the 107 patients who were prescribed spironolactone at hospital discharge, 18 (17%) either had a documented serum
creatinine O 2.5 mg/dL, a serum K O 5.5 mmol/L, or
no documentation of left ventricular EF !35% within 60
days of hospital admission. After fully adjusting for differences in baseline variables, spironolactone emerged as an
independent predictor of reduced long-term mortality (OR
0.52, 95% CI 0.34e0.79) (Table 5).
In an attempt to dissect out the mechanism for the incongruent effects of spironolactone use on mortality in our
population, we repeated our logistic regression model for
only those patients who met RALES criteria, with an EF
of less than 35%, a serum creatinine less than 2.5 mg/dL,
and a serum K less than 5 mM (n 5 628).
Spironolactone use in this subgroup was not associated
with improved long-term survival (OR 1.20, 95% CI
0.68e2.11).
Discussion
We found that treatment with spironolactone among all
patients admitted for HF was not associated with improved
survival. In contrast, spironolactone use in patients followed in HF clinics was associated with a significant
decrease in the relative risk of all-cause mortality.
In RALES, 1663 patients with NYHA III-IV HF and an
ejection fraction ! 35% were randomized to receive either
a low dose of spironolactone (25 mg/day) or placebo. Exclusion criteria included a plasma potassium concentration
O 5.0 mmol/L and a serum creatinine concentration O 2.5
mg/dL. Importantly, only 11% of the treatment group had
been receiving a b-blocker, and 25% of the patients were
diabetic.8 The RALES protocol called for monitoring of serum potassium levels 1 week after initiating therapy, and
then again at 1 month and every 3 months thereafter. An
increase in serum potassium O 5.5 mmol/L at any time
prompted a review of concomitant medications associated
with hyperkalemia, such as potassium supplements or nonsteroidal anti-inflammatory drugs, and a reduction in the
dose of study medication to 25 mg every other day. Serious
hyperkalemia occurred in 1% of the placebo group and 2%
of the spironolactone group (P 5 .42).
There are several possible reasons for our studys failure
to detect the positive effect of spironolactone use on survival observed in the RALES trial. To begin with, the differences in outcome likely reflect differences between the
2 studies in terms of patient selection and extent of
follow-up after discharge. Our patient population represents
a broader group of HF patients, with less severe HF than in
RALES (mean EF 31% vs. 25%). Also, the closeness of follow-up dictated in the RALES trial may have minimized
the risk of hyperkalemia and other sources of mortality associated with spironolactone use in an unselected HF

168 Journal of Cardiac Failure Vol. 13 No. 3 April 2007

Fig. 1. Kaplan-Meier survival analysis.

population.9e12 However, we also must not ignore the role

that chance may have played in obtaining these results, as
well as the potential effect of unmeasured variables for
which we could not adjust.
Table 4. Baseline Characteristics of HF Clinic Patients
Discharged on Spironolactone
Variable
Age O 77 years
Male gender (%)
EF performed
within 60 days (%)
EF ! 35% (%)
Diabetes (%)
Hyperlipidemia (%)
Smoking (% ever)
Peripheral vascular
disease (%)
Creatinine O
2.5 mg/dL (%)
ACE inhibitors (%)
ARB (%)
b-blockers (%)
Digoxin
Statins
NSAIDs

No
(n 5 883)

Yes
(n 5 107)

P Value

213 (24.1)
552 (62.5)
729 (82.6)

15 (14.0)
68 (63.6)
91 (85.0)

.02
.83
.52

338
257
417
487
107

65
43
47
65
11

(53.8)
(29.1)
(47.3)
(55.2)
(12.1)

41 (5.8)
522
41
465
323
256
11

(59.1)
(4.6)
(52.7)
(36.6)
(29.0)
(1.2)

(77.4)
(40.2)
(43.9)
(60.4)
(9.9)

5 (4.7)
90
8
73
59
37
1

(84.1)
(7.5)
(68.2)
(55.4)
(34.6)
(1.0)

!.0001
.02
.51
.34
.54
.65
!.0001
.20
.002
.0004
.26
.00

All medications prescribed at hospital discharge, laboratory and parameters on admission.

EF, ejection fraction; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; NSAID, nonsteroidal anti-inflammatory drug.

Several smaller studies have warned of a higher risk of hyperkalemia associated with spironolactone in the community
than what was observed in RALES.9e12 Several hypotheses
have been proposed: inappropriate monitoring of potassium
levels; neglect of predisposing patient attributes, such as
diabetes; neglect of conditions that develop during therapy,
including renal insufficiency; inappropriate doses of
spironolactone; deliberate increases in dietary potassium
intake by patients; and the inappropriate extrapolation of
RALES findings to a broader patient population.13 Although
we do not have information regarding rates of hyperkalemia
or the cause of death in our patients, inappropriate prescription of spironolactone may have resulted in higher rates of
Table 5. Logistic Regression Model for Mortality
in Clinic Patients

Variable
Spironolactone on
discharge
Age (per year)
Peripheral vascular disease
Creatinine (per mg/dL)
b-blocker
Statin
Loop diuretic

Point
Estimate

95%
Confidence
Limits

P Value

0.52

(0.34, 0.79)

.003

1.03
1.46
1.35
0.60
0.44
1.54

(0.99,
(0.78,
(0.94,
(0.39,
(0.27,
(0.83,

.08
.24
.39
.01
.0007
.17

1.05)
2.73)
1.95)
0.93)
0.70)
2.87)

All medications prescribed at hospital discharge, laboratory and parameters on admission.

Spironolactone and Heart Failure Mortality

complications, which negated any survival-benefit the drug

otherwise might have exerted in our population.
In contrast to what we observed in the general HF population, spironolactone use among those patients followed
in 1 of our HF clinics was associated with a significant
decrease in the relative risk of all-cause mortality. A recent
systematic review demonstrated that the use of multidisciplinary teams providing specialized follow-up for HF patients was associated with significant reductions in both
mortality and hospitalization rates.13 We hypothesize that
there may be at least 2 reasons for these results, as suggested by our own findings. First, entry criteria for clinical
trials are more likely to be rigorously applied in such specialty clinics than in routine practice. Second, for all
patients, but especially for those who fall outside strict
clinical trial populations, the close follow-up and specialized care provided in the subspecialty clinic setting more
closely mimic that provided in the efficacy studies in which
outcome benefit has been demonstrated. The findings therefore may reflect not the effect of the HF clinic on mortality,
but rather the importance of prescribing spironolactone in
accordance with published indications and in the presence
of closer follow-up.
The primary strength of the current study is that it was
population-based, with full capture of consecutive hospitalizations and outcomes across an entire health care system.
However, this study is not without its limitations. First, details regarding spironolactone use are unknown, including
average dose, duration of treatment, and medication compliance after hospital discharge. The proportion of patients
not discharged on spironolactone but subsequently prescribed the drug is unknown, as is the proportion of patients
who never filled their prescription for spironolactone after
hospital discharge. Second, data concerning New York
Heart Association class and ejection fraction are lacking
for a large percentage of patients, in particular those not
seen in a specialized HF clinic. Finally, in contrast to a
randomized trial wherein differences between groups are
eliminated through the process of randomization, the
retrospective nature of this study did not permit us to completely eliminate confounding, especially that which might
have resulted from unmeasured variables.
Conclusion
In a general population of patients admitted for HF, prescribing spironolactone at the time of discharge from hospital was not associated with improved survival. Among
those patients followed in HF clinics, however, spironolactone use was associated with enhanced survival. These conflicting results highlight the challenges of knowledge

Ouzounian et al

169

translation from a structured clinical trial into actual

clinical practice.
Acknowledgments
The authors wish to acknowledge the work of the ICONS
Staff and Steering Committee. The authors also wish to
acknowledge the statistical expertise of Heather Merry.

References
1. OConnell JB, Bristow M. Economic impact of heart failure in the
United States: time for a different approach. J Heart Lung Transplant
1993;13:S107e12.
2. Hunt SA, Baker DW, Chin MH, et al. ACC/AHA guidelines for the
evaluation and management of chronic heart failure in the adult: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll
Cardiol 2001;38:2101e13.
3. Liu P, Arnold JM, Belenkie I, et al. The 2002/3 Canadian Cardiovascular Society consensus guideline update for the diagnosis and
management of heart failure. Can J Cardiol 2003;19:347e56.
4. Pitt B, Zannad F, Remme WJ, et al. The effect of spironolactone on
morbidity and mortality in patients with severe heart failure. N Engl
J Med 1999;341:709e17.
5. Cox JL, Ramer SA, Lee DS, Humphries K, Pilote L, Svenson L,
Tu JV, on behalf of the Canadian Cardiovascular Outcomes Research
Team Investigators. Pharmacological treatment of congestive heart
failure in Canada: a description of care in five provinces. Can J Cardiol 2005;21:337e43.
6. Canadian Cardiovascular Outcomes Research Team Investigators.
Burden of cardiovascular disease in Canada. Can J Cardiol 2003;19:
997e1004.
7. Cox JL, ICONS Investigators. Optimizing disease management at
a health care system level: the rationale and methods of the Improving
Cardiovascular Outcomes in Nova Scotia (ICONS) study. Can J
Cardiol 1999;15:787e96.
8. Pitt B, Perez A. Spironolactone in patients with heart failure. N Engl J
Med 2000;342:133e4.
9. Berry C, McMurray JJ. Serious adverse events experienced by patients
with chronic heart failure taking spironolactone. Heart 2001;85:E8.
10. Schepkens H, Vanholder R, Billiouw JM, Lameire N. Life-threatening
hyperkalemia during combined therapy with angiotensin-converting
enzyme inhibitors and spironolactone: an analysis of 25 cases. Am J
Med 2001;110:438e41.
11. Svensson M, Gustafsson F, Galatius S, Hildebrandt PR, Atar D. Hyperkalaemia and impaired renal function in patients taking spironolactone for congestive heart failure: retrospective study. BMJ 2003;327:
1141e2.
12. Juurlink DN, Mamdani MM, Lee DS, et al. Rates of hyperkalemia
after publication of the Randomized Aldactone Evaluation Study.
N Engl J Med 2004;351:543e51.
13. McAlister FA, Stewart S, Ferrua S, McMurray JJ. Multidisciplinary
strategies for the management of heart failure patients at high risk
for admission: a systematic review of randomized trials. J Am Coll
Cardiol 2004;44:810e9.