Download Clinical events leading to the progression of heart failure: insights

European Heart Journal (2001) 22, 153–164
doi:10.1053/euhj.2000.2175, available online at http://www.idealibrary.com on
Clinical events leading to the progression of heart
failure: insights from a national database of hospital
discharges
A. U. Khand1, I. Gemmell2, A. C. Rankin1 and J. G. F. Cleland3
1
Department of Cardiology, Glasgow Royal Infirmary; 2Social and Public Health Sciences Unit, University of
Glasgow; 3Department of Cardiology, Castle Hill Hospital, University of Hull, Kingston-upon-Hull, U.K.
Aims To describe the sequence of clinically apparent events
causing readmission and antedating death, subsequent to a
first-time hospital admission for heart failure, in order to
give insights into the natural history and mechanisms of
progression of heart failure.
Methods A national database of linked hospital discharge
and mortality data for Scotland (population 5·1 million)
was used. Patients with a first-time admission to hospital
with heart failure in 1992 (index population) were identified
and, using a record linkage system, hospital readmissions
and their cause according to the hospital physician and
deaths were recorded over the subsequent 3 years. A
flowchart showing the sequence of events leading to death
or recurrent admission was constructed.
Results 12 640 patients had first-time admissions with
heart failure in 1992; their mean age was 74 years and
46·2% were men. A cohort of 2922 (23%) patients died on
their first admission. Among the remaining 9718 patients
there were 22 747 readmissions and 4877 deaths over the
subsequent 3 years; only 15% had neither event reported.
Nine per cent of patients died without any readmission and
a further 6% without a further readmission for cardiovascular reasons. A cohort of 5992 (61% of patients at risk)
had at least one cardiovascular readmission and half of
Introduction
Hospitalizations for heart failure are rising in
number[1–3] despite recent advances in treatment[4–7].
Ageing of the population, improved primary and secondary preventive measures and an improvement in the
Revision submitted 21 March 2000, and accepted 21 March 2000.
Correspondence: Professor John G.F. Cleland, MD, FRCP,
FACC, FESC, Department of Cardiology, Castle Hill Hospital,
University of Hull, Kingston-upon-Hull, HU16 5JQ, U.K.
0195-668X/01/2200153+12 $35.00/0
these had occurred within 6 months. Heart failure without
a report of any cardiovascular precipitating event was
responsible for 37% (2188 patients) of first cardiovascular
readmissions and of these patients approximately 12% had
evidence of renal failure or acute respiratory infection as
possible triggers for readmission. Acute ischaemic events
including myocardial infarction (19%), myocardial infarction alone (8%) and atrial fibrillation (11%) were associated
with a substantial number of first readmissions. First
readmission precipitated by acute myocardial infarction
was associated with a particularly poor prognosis (40%
inpatient mortality).
Conclusions Recurrent ischaemic events and atrial fibrillation may be the predominant mechanisms leading to
exacerbation of and progression of heart failure and death.
A substantial proportion of readmissions appear related to
heart failure alone. Whether this reflects progressive ventricular remodelling leading to worsening heart failure or
other unidentified mechanisms cannot be discerned from
this data.
(Eur Heart J 2001; 22: 153–164, doi: 10.1053/euhj.2000.2175)
2001 The European Society of Cardiology
Key Words: Heart failure, atrial fibrillation, left ventricular remodelling, ischaemic heart disease, epidemiology.
prognosis of heart failure itself[8] will conspire to drive
prevalence and therefore rates of readmission yet higher.
Because of the enormous social, economic and private
burden heart failure imposes, it is quite appropriate that
a considerable research effort is being undertaken, from
the perspectives of both basic and clinical science, in an
attempt to understand the mechanisms underlying the
progression of heart failure.
Over the last decade the concept that left ventricular
remodelling, a term used to describe the progressive
structural, molecular and functional adaptations of the
2001 The European Society of Cardiology
154
A. U. Khand et al.
heart in response to damage, is the key mechanism
underlying the progression of heart failure has become
popular. This view has developed on the basis of animal
experiments[9] as well as observations on several small
and a few large cohorts of patients after myocardial
infarction[10,11], or with chronic heart failure[12]. However, these studies have also highlighted the possibility
that recurrent ischaemic events may lead to repeated
hospital admission and sudden or worsening heart
failure deaths[13–16]. In relatively small community
studies[17,18] arrhythmias are commonly recorded as a
precipitating reason for hospital admission which contrasts with the relative rarity of such events in large
clinical trials of heart failure[19,20]. Ischaemic insults
and supraventricular arrhythmias may be important
pathways of heart failure progression: either directly
or by contributing to the adverse remodelling
process, by precipitating neurohormonal and cytokine
activation, the induction of apoptosis, or by other
mechanisms[21–23].
The purpose of this report is to identify the sequence
and to quantify the contribution of various clinically
apparent events to recurrent admission for heart failure
and as antecedents of death. To achieve this we used
individual patient-linked data from a national database
of first-time hospital admissions with heart failure. As
approximately 80% of first diagnoses of heart failure
in the U.K. occur at hospitalization[24] this approximates
to an incident, epidemiologically representative population rather than a clinical trial population.
Methods
We used data from the Scottish Hospitals In-Patient
Statistics (SHIPS) database[25]. An SMR 1 (standardized
mortality record) form is generated on patient discharge,
and diagnoses causing/precipitating hospital admission
are entered on the SMR1 form. International Classification of Disease (ICD) diagnostic codes corresponding
to these diagnoses are indexed by the local hospital
coding department and the information is then transcribed to the SHIPS database. Hospitalization records
for an individual patient are linked allowing a chosen
index event to be tracked to previous and subsequent
episodes. The data are also linked to statutory death
records. The records provide information about age,
gender, principal diagnosis and up to five other conditions coded to ICD 9 format for an admission and the
principal cause and up to three other causes for death.
The accuracy of ICD 9 coding for the principal diagnosis is 90%, and 77% for other diagnostic codes (1994
results)[26]. The system is maintained and controlled by
the Information and Statistics Division (ISD) of the
Scottish Health Service Common Services Agency.
The database was interrogated to identify patients
with a first-time admission with heart failure to Scottish
hospitals in 1992; this was the last year that allowed
linkage for three prospective years at the time of
Eur Heart J, Vol. 22, issue 2, January 2001
analysis. The codes used for heart failure were ICD 9
codes 428·0, 428·1, 428·9 and 402·9, which correspond to
congestive heart failure, left heart failure, unspecified
heart failure and hypertensive heart failure, respectively.
Patients who generated more than one SMR1 record
during a single hospital admission (for example, because
of transfers between departments) were counted as a
single admission as this corresponds more closely to the
clinical understanding of what constitutes an admission.
First-time admissions with heart failure were defined as
admissions coded for heart failure without a similar
episode coded in the 3 years before the index admission.
Previous analyses indicated that <0·1% of patients have
admissions coded for heart failure >3 years apart[8].
Cases were censored for readmission or death on the
third anniversary of their index event to ensure that all
patients were exposed to the same period of risk for
events.
To describe the likely aetiological causes of heart
failure we captured hospital admissions in the 3 years
prior to the index admission (1989–1992) (Table 1). We
then identified readmissions and deaths, classified as
in- or out-of-hospital, over the subsequent 3 years
(1992–1995). Previous reports have shown that the
record linkage system is accurate in identifying these
subsequent events[27].
A flowchart was constructed showing the sequence
leading to the next two admissions or death. A hierarchical system was used to classify clinical reasons for
readmissions on the flow chart. Possible reasons for
readmissions leading to an exacerbation of heart failure
were identified from previous publications[17,18,28–31].
Atrial fibrillation/paroxysmal supraventrcular tachycardia, acute stroke, acute ischaemic event, acute
myocardial infarction, hypertensive heart disease, renal
failure or acute respiratory infection (ICD 9 coding for
acute respiratory infection: 460–466 plus 480–90, codes
for other diagnoses in Table 1) are common precipitating events for heart failure exacerbations. These diagnostic codes were defined as ‘potential triggering events’.
When these diagnoses were coded together with heart
failure then it was assumed that the ‘potential triggering
events’ had precipitated the readmission for heart
failure. All ‘potential triggering events’ were considered
co-dominant and therefore a single patient could be
counted in a number of diagnostic categories. Readmission was attributed to heart failure in the flow chart only
if the former codes were not present. Other diseases of
the circulatory system (ICD 9: 390–459, excluding above
codes) and readmission for other non-cardiovascular
diagnoses (all other codes) represented a third and
fourth tier, respectively, in this hierarchy.
To capture a ‘fresh’ acute ischaemic event rather than
diagnostic codes relating to background ischaemic heart
disease, codes for angina pectoris and chest pain (ICD
9 413 and 786·5 codes) were used in the acute ischaemic
event box (angina pectoris/chest pain) of the flowchart
and codes ICD 9 411 and 414 codes encompassing
various mostly non-acute definitions of ischaemic heart
disease were excluded. Diagnoses consistent with the
Events leading to the progression of HF
155
30
25
Rate per 1000 population
2433
4341
20
15
1953
10
1615
5
996
607
28
0
31
M
F
< 25
425
211
M
F
25–54
M
F
55–64
M
F
65–74
M
F
75+
Figure 1 Incidence rates of first time heart failure hospitalization 1992. Population of Scotland
5·1 million. Numbers above bars represent absolute numbers of patients in any ICD 9 coding
position. M=male; F=female; =other position; =first diagnostic position.
latter codes are frequently inserted on a heart failure
related admission to document the aetiology of heart
failure or copied from previous admissions as a background disease, irrespective of the evidence of an acute
ischaemic insult.
was among women aged d75 years, although rates of
admission adjusted for the population at risk were fairly
consistently higher in men.
The mean age of the population was 74 years and
46·2% were men. 2922 patients (23%) died during
admission leaving 9718 at risk of readmission or out-ofhospital death.
Results
Demography
There were 12 640 patients reported with a first admission for heart failure in 1992 (index population) coded in
any of the six diagnostic positions (of ICD 9 coding, see
methods), of which 5763 (46%) and 4378 (35%) were in
the first and second positions, respectively. Of the ICD 9
heart failure codes used, 44% of the index population
had code 428·1(left heart failure), 36% had 428·0 (congestive heart failure), 21% had 428·9 (unspecified heart
failure) and 1% had 402·9 (hypertensive heart failure),
and 2% had two of the above codes. Figure 1 describes
the rates for first admissions with heart failure of the
index population divided by age, gender and position of
ICD 9 heart failure code. The population characteristics
for Scotland were obtained from official government
publications[32]. Hospitalization rates rose markedly
with age. Numerically the largest number of admissions
Underlying causes of heart failure and
concomitant diagnoses
Table 1 shows the range of diagnostic codes prior to,
at the time of index admission and on subsequent
readmissions. 9970 of the index cohort had at least one
admission in the prior 3 years. A code for any form of
ischaemic heart disease was recorded in 3551 (36%) of
these patients, (>50% of which were acute myocardial
infarction codes) and 4662 patients had the IHD code
recorded during concomitant admission, (6400 patients
in total either previous to or on index admission, see
Table 1a). Diseases of the digestive system were the next
most frequently reported before index admission. There
were 1969 (15·6%) patients with a code for respiratory
disease prior to index admission and 2833 patients (22%)
had a respiratory disease code as a concomitant diagnosis, one-quarter of the latter were acute respiratory
Eur Heart J, Vol. 22, issue 2, January 2001
156
A. U. Khand et al.
Table 1
The burden of disease in heart failure
Diagnosis
(any diagnostic position)
Population at risk
Total number of events (any cause)‡
Acute myocardial infarction (AMI)§
Angina/chest pain
Ischaemic heart disease (IHD)
Any IHD (any of above three codes)
Atrial fibrillation/paroxysmal SVT
Ventricular fibrillation/cardiac arrest
Other arrhythmia
Hypertensive heart disease
Mitral valve disorder
Aortic valve disorder
Diseases of pulmonary circulation
Acute stroke
Heart failure
Heart failure alone0
Other diseases of circulatory system¶
Diabetes mellitus
Renal failure
Disease of urinary system
Respiratory disease
Digestive system
Cancer
Alcohol dependence/drunkenness
Other (excluding above)
ICD-9 codes
Subsequent admissions*
Previous*
admission
Concomitant
diagnosis
Subsequent
admission*
12 640
9970
1958
1584
1642
3551
730
56
375
12 640
12 640
2039
593
2402
4662
1548
193
229
9718
7381
780
1068
2500
3286
1025
217
327
—
22 747
900
1954
4335
6550
1555
223
397
3·1
1·2
1·8
1·7
2·0
1·5
1·0
1·2
814
443
348
243
502
0
0
943
833
251
1227
1969
3230
1417
207
25
737
325
182
243
297
12 640
5715
801
1022
611
463
2883
928
538
77
0
571
290
212
338
523
3719
2154
789
846
711
809
2276
1708
864
107
462
862
437
310
441
584
6660
3010
1119
2049
1353
1145
4252
2557
1932
179
645
1·5
1·5
1·5
1·3
1·1
1·8
1·4
1·4
2·4
1·9
1·4
1·9
1·5
2·2
1·7
1·4
—
—
410
413, 786·5
411, 414
427·0, 427·3
427·4, 427·5
All 427 except
above four codes
401–405
394, 424·0
395, 424·1
415–417
436
428, 402·9
428, 402·9
390–459 excl above
250
584–586
580–583, 587–599
480–519
520–579
140–239
303, 305·0
Total
number
Mean†
*Previous and subsequent admission column refers to number of patients admitted at least once whereas subsequent admissions (italics)
is number of admissions generated.
† mean=column 6/column 5, the median number of readmissions for all diagnostic codes was 1 except for any cause and diabetes mellitus
where it was 2.
‡ This row pertains to numbers of events (patients, readmissions or deaths) whereas the rows below this describe the number of codes for
respective diagnoses generated in any position.
§ In previous admissions only the code 412 ‘Old myocardial infarction’ is also included.
0 Heart failure alone was defined as a heart failure code without a concomitant code for the following; AMI, any IHD, stroke, AF,
hypertension, aortic and mitral valve disease and diabetes mellitus.
¶ The codes in this category principally relate to other peripheral vascular disease (ICD 9 433), and the rest are a range of conditions such
as other cardiovascular disease and transient cerebral ischaemia.
infection codes (25%). Prior admissions with cancer and
urinary tract problems occurred in 11% and 10% of the
index population, respectively.
Atrial fibrillation was recorded in 16% of patients
either prior to or on the index admission, the equivalent
figure for hypertension was 11%. Seven per cent of
patients had codes relating to aortic and mitral
valve disorders either on the previous or concomitant
admission. Of the index population 10·5% had a code
for diabetes mellitus before index admission or as a
concomitant diagnosis (Table 1a).
Overall, 5715 patients had no concomitant code for
known aetiological conditions for heart failure on index
admission (any code for ischaemic heart disease, hypertension, atrial fibrillation/paroxysmal supraventricular
tachycardia, valvular heart disease, diabetes mellitus)
and 4071 of these patients did not generate any of the
above codes in the previous 3 years either, suggesting
Eur Heart J, Vol. 22, issue 2, January 2001
approximately 32% of the index population had an
unknown aetiology for heart failure.
Cumulative diagnostic coding
By looking at the cumulative frequency of diagnostic
codes we sought to differentiate between conditions
leading up to heart failure from complications developing as a consequence of heart failure and potentially
contributing to its progression. Table 1a shows the
cumulative frequency of the diagnosis so that the right
hand column reflects a diagnosis recorded at any time
between 1989–95 for that number of patients. The
percentage in the table is a percentage of 12 640 (index
admission population) and, therefore, the magnitude of
each event as well as the principal time period in which
it occurred is given. In Fig. 2 the total number of
Events leading to the progression of HF
Table 1a
157
The burden of disease in heart failure (cumulative percentages)
Diagnosis
(any diagnostic position)
Population at risk
Acute myocardial infarction (AMI)
Angina/chest pain
Ischaemic heart disease (IHD)
Any IHD (any of above three codes)
Atrial fibrillation
Ventricular fibrillation
Other arrhythmia
Hypertensive heart disease
Mitral valve disorder
Aortic valve disorder
Diseases of pulmonary circulation
Acute stroke
Heart failure
Heart failure alone
Other diseases of circulatory system
Diabetes mellitus
Renal failure
Disease of urinary system
Respiratory disease
Digestive system
Cancer
Alcohol dependence/drunkenness
Previous
admission
% of
index
population
Concomitant
or previous
admission
Cumulative
%
Subsequent,
previous or
concomitant admission
Cumulative
%
9970
2037
1584
1642
3551
730
56
283
814
443
348
243
502
0
—
926
833
251
1227
1969
3230
1417
207
79·0
16·1
12·5
13·0
28·0
5·8
0·2
2·2
6·4
3·5
2·7
1·9
4·0
0
—
7·3
6·4
2·0
9·7
15·6
25·5
11·2
1·6
12 640
3725
1972
3458
6400
2039
193
533
1390
528
394
454
744
12 640
5715
2977
1331
765
1591
3860
3735
1657
268
100·0
29·5
15·6
27·4
50·6
16·1
1·5
4·2
11·0
4·1
3·1
3·6
5·9
100·0
45·2
23·5
10·5
6·1
12·6
30·5
29·5
13·1
2·1
12640
4096
2590
4845
7350
2552
399
802
1683
681
516
727
1192
12 640
6646
3839
1555
1305
2153
4915
4686
2167
331
—
32·4
20·5
38·3
58·1
20·2
3·1
6·3
13·3
5·4
4·1
5·7
9·4
100·0
52·6
30·4
12·3
10·3
17·0
38·9
37·1
17·1
2·6
See Table 1 for codes.
100
80
%
60
40
20
0
Diabetes Hyper
AMI
IHD
Resp
AF
Stroke
Renal
Diagnoses
Figure 2 Temporal trends in first mention of cardiovascular codes in relation
to first heart failure hospitalization 1992. Hyper=hypertension; AMI=acute
myocardial infarction; IHD=ischaemic heart disease; Resp=respiratory
disease; AF=atrial fibrillation; =subsequent; =concomitant; =previous.
Eur Heart J, Vol. 22, issue 2, January 2001
158
A. U. Khand et al.
Table 1b
Cardiovascular (CVS) readmissions and 3-year mortality according to concomitant or previous diagnosis
Diagnosis
(any diagnostic position)
Previous or
concomitant
admission
Survivors of
index admission
% with CVS
readmission*
Total number of
CVS readmission
Days in hospital
with CVS
readmission†
3-year
mortality‡ (%)
12 640
3725
4334
2309
1390
744
1331
765
3860
9718
2923
3713
1812
1250
481
1227
449
3004
60·5
65·1
68·1
63·4
66·5
65·7
70·9
65·9
61·5
12 316·0
4654
5926
2451
1978
714
1943
661
3987
18
15
17
20
18
22
21
22
20
50·2
48·4
49·9
54·1
47·1
62·4
56·3
68·1
59·3
All categories
Acute myocardial infarction
Angina/chest pain/IHD§
Atrial fibrillation
Hypertensive heart disease
Acute stroke
Diabetes mellitus
Renal failure
Respiratory disease
* % of index admission survivors admitted at least once with a cardiovascular readmission.† Median number of days spent in hospital per
patient (i.e. combining all episodes for each patient).
‡ Excluding deaths on index admission.
§411, 413, 786·5: all codes denoting any diagnosis of ischaemic heart disease (IHD) except AMI.
700
600
500
400
300
200
100
ok
e
st
r
cu
te
A
er
t
yp
H
en
a
lf
ai
en
si
o
n
lu
re
A
M
I
D
ia
R
m
el
an
be
te
s
e
ig
D
lit
us
ce
r
F
C
sy
A
st
em
e
se
as
es
tiv
y
at
or
ir
es
p
R
H
ea
r
A
tf
ai
ny
lu
IH
re
di
al
co
D
fa
rt
ea
H
on
e
de
ilu
re
0
Figure 3 Subsequent admission rates per thousand population at risk. Most frequent admissions in
descending order. See text for abbreviations and Table 1 for diagnostic codes; AMI is a subset of any IHD.
Population at risk 9718 (index admission survivors). =number of admissions; =patient admissions.
patients having been coded for the listed events anytime
between 1989–95 is defined as 100%. The 100% is
subdivided in the multi-column bar chart according to
the proportions of patients at each specified time period.
A diagnosis of ischaemic heart disease was recorded at
some time (1989–95) in 7350 (58%) patients (excluding
codes for death) and in the great majority of these cases
(50·6%) this was prior to or on the index admission. The
pattern of recording of these events is as might be
expected with conditions such as hypertension, diabetes
and ischaemic heart disease generally antedating the
onset of heart failure, atrial fibrillation and respiratory
Eur Heart J, Vol. 22, issue 2, January 2001
disease occurring prior to or concomitant with a diagnosis of heart failure while stroke and renal failure
occurred after the onset of heart failure in a large
percentage of patients.
Readmissions 1992–1995
Figure 3 shows the number of patients readmitted and
the total number of readmissions according to diagnostic discharge codes, per thousand population at risk (i.e.
excluding deaths on the index admission). Table 1b, by
Events leading to the progression of HF
contrast, categorizes the index population according to
concomitant or previous codes for heart failure associated diagnoses and describes the percentage of index
admission survivors with cardiovascular admissions
(1992–95), median length of stay per admission and the
respective 3-year mortality for these cohorts, excluding
deaths on index admission.
The mean number of readmissions for survivors of the
index admission was 3·1 (range 0–23). 1485 patients
survived without any readmission and 2231 without a
cardiovascular readmission over 3 years. Heart failure
was the most common coding for readmission. 3719
patients had 6660 readmissions for heart failure (mean
of 1·8 admissions/patient) over 3 years and these constituted 29% of the total number of admissions. Readmission for heart failure was particularly high in the early
post-discharge period, 16% of index admission survivors
were readmitted with heart failure within 90 days of
discharge. Approximately 21% of heart failure readmissions were associated with a concomitant code for acute
myocardial infarction, an acute ischaemic event (chest
pain/angina pectoris) or atrial fibrillation/paroxysmal
supraventricular tachycardia.
Ischaemic heart disease codes, as a group, were the
second most frequent coded reason for readmission
(6550 admissions in 3286 patients). Acute ischaemic
event codes (acute myocardial infarction and chest pain/
angina pectoris) constituted almost 50% of the codes.
Non-cardiovascular diagnostic codes, such as respiratory disease and diseases of the digestive system figured
frequently as a cause of readmission (Fig. 3).
Survivors of the index admission who had diabetes
mellitus, any ischaemic heart disease code, atrial
fibrillation, hypertensive heart disease or renal failure
coded either in the index or a previous admission
had a significantly higher rate of cardiovascular
readmission compared with the index admission
survivors as a whole. Those with previous or concomitant renal failure, diabetes mellitus, acute stroke,
respiratory disease or atrial fibrillation codes suffered a
significantly greater mortality. Acute stroke patient
subsets had significantly longer lengths of stay in
hospital (Table 1b).
Death
Three-year mortality for the index population (Table 2)
was 62% (7799 deaths), 52% for those <75 years and
70% for those d75 years (P<0·001 for age difference).
Of these, 2922 deaths occurred on the index admission,
867 patients died after the index admission but before
readmission and 535 patients died subsequent to only
non-cardiovascular readmissions. Of the index admission survivors, 4877 (50%) subsequently died over 3
years and 2087 (43%) of these deaths occurred out of
hospital. The median (IQR) time to death during index
admission was 9 days (2–25) and that for discharge to
out of hospital death without subsequent readmission
was 100 days (18–345), 46% of these deaths occurring
within 90 days of discharge.
159
Figure 4 illustrates time from discharge from last
hospital admission to death for all the 2087 deaths
occurring out of hospital. The risk of death was highest
early after hospital discharge, 25% of all out of hospital
deaths had occurred by 16 days after last hospital
discharge. Age, gender or cause of last admission did not
explain the skewing the incidence of death towards an
early post-discharge period.
Sequence of events
Figure 5 describes the coded reasons for next hospitalization after index admission (according to the criteria in
the methods section). Second hospitalizations are also
shown for larger cohorts.
Of the total at risk, 6093 patients (63%) had at least
one cardiovascular readmission or admission with acute
respiratory infection or renal failure. Of these patients
an acute ischaemic event code (defined as code for chest
pain or angina pectoris) was reported in 656 patients
(7% of patients at risk), 61 (9%) of whom subsequently
died on or after discharge without readmission. A further acute ischaemic event was also the commonest
reason for readmission in this subset of patients. Heart
failure readmission in the absence of a clinical trigger
(concomitant first rank codes excluded) accounted for
only 15·7% of second readmissions in this subgroup. The
number of patients with acute myocardial infarction
causing first cardiovascular readmission is also shown,
nearly 50% patients in this cohort died during admission
or subsequent to discharge without a second readmission.
Atrial fibrillation was reported at the time of first
cardiovascular readmission in 680 patients, in 307 (45%)
cases this was the first report of atrial fibrillation. The
vast majority of this subgroup had code 427·3, atrial
fibrillation (97·5%) compared to only 2·5% for paroxysmal supraventricular tachycardia. These patients had
a better prognosis during and after readmission; only
20% died during or subsequent to a first cardiovascular
readmission compared with 30% during or subsequent
to all first cardiovascular readmissions (P<0.001). However, the next admission in this cohort was associated
with stroke in 5% of the patients at risk of an event
(survivors of first readmission with atrial fibrillation
code).
Heart failure without any reported triggering event
(no concomitant first rank code) was coded as the reason
for first cardiovascular readmission in 1931 patients
(20% of patients at risk), 575 of whom died during
admission or prior to a further readmission. 205 of the
surviving patients (13% of at risk population) were
subsequently readmitted with an acute ischaemic event,
acute myocardial infarction or atrial fibrillation and 522
(33% of at risk population) with heart failure in the
absence of a potential triggering event.
Renal failure was reported in 6·2% (380 patients) of all
first readmissions (cardiovascular, renal failure or respiratory infection codes) and of these admissions heart
Eur Heart J, Vol. 22, issue 2, January 2001
160
A. U. Khand et al.
Table 2
Death: out of hospital and during cardiovascular readmission
Population
Patients
at risk
Numbers with
event (n)
Median times to event*
(IQ range) days
12 640
—
9718
9718
6048
4915
4915
3057
2922
9718
867
6048
1131
631
3057
526
9 (2, 25)
—
100 (18, 345)
150 (53, 400)
8 (2, 23)
123 (35, 332)
92 (29, 252)
10 (3, 25)
9718
9718
2790
2087
325 (124, 631)
296 (77, 635)
Deaths during index admission
Survivors of index admission
Deaths subsequent to discharge but without readmission
First cardiovascular readmission
Deaths during 1st cardiovascular readmissions
Deaths after 1st cardiovascular readmission but without readmission
Second Cardiovascular readmission
Deaths during second cardiovascular readmission
Deaths during readmission over 3 years
Deaths out of hospital over 3 years
* Median times for deaths during readmission are calculated from admission date, whereas for out-of-hospital deaths median time refer
to time to death from discharge of preceding admission. Deaths over 3 years (last two rows) are from index admission discharge.
2500
Number of deaths
2000
1500
1000
500
0
1
3
2
Time to death from discharge from last admission (years)
Figure 4
Time to death from last admission for out of hospital deaths.
failure was the sole cardiovascular code in 149 patients
(39%). The equivalent figures for acute respiratory infection was 6·5% (399 patients) of first readmissions and
112 (28%) of these patients had heart failure as a sole
cardiovascular code. The most frequent diagnoses for
second readmissions were heart failure in the absence of
a concomitant potential triggering event code (933
patients), an acute ischaemic event or acute myocardial
infarction (540) and atrial fibrillation (361).
Discussion
The present study identifies key clinical events associated
with readmission or antedating death from a large
database of first-time admissions with heart failure. The
Eur Heart J, Vol. 22, issue 2, January 2001
relative magnitude and sequence of these events provides
an insight into the clinical progression of heart failure.
Among the 9718 patients who survived the index
admission there were 22 747 readmissions and 4877
(50%) deaths over the following 3 years. Ischaemic heart
disease was the dominant cause of heart failure and 16%
(n=2039) of first admissions with heart failure were
associated with acute myocardial infarction and a further 13% had had an admission for acute myocardial
infarction within the past 3 years (Table 1a). The single
most common triggering event for first and subsequent
readmissions was an acute ischaemic event or a myocardial infarct (Fig. 5). Once a patient had been admitted
with an acute ischaemic event this remained the dominant reason for readmission. This is despite restricting
codes for acute ischaemic events to chest pain/angina
AMI
496
D
25
Angina/
chest pain*
656
D
77
D
61
AF
680
HF
74
AMI CVA Ang
14
33 24
Hypertension
341
AMI CVA Ang H AF Resp Renal
36
164 28 23 14
24
12
Resp Renal HF
8
10
68
Acute
stroke
301
Renal
380
D
123
AMI CVA Ang
45
31 75
H AF Resp Renal HF
18 167 14
24
80
Resp
399
Heart failure admission
12 640
Death
535
D
164
HF
522
Ang H AF Resp Renal HF
66 40 51 44
26 142
Other
cardiovascular
disease
1327
H AF Resp Renal
30 85 65
68
AMI CVA
24 29
D
123
D
345
Heart
failure
1931
Other diagnosis
1273
No admission and
no death in 3 years
1485
Figure 5 The sequence of events after first hospital admission with Heart Failure 1992. See text for description, Table 1 for codes of diagnostic
categories (acute respiratory infection: ICD 9 codes: 450–466 plus 480–90, subset of respiratory disease). The diagnosis and below this the number
of codes with this event are illustrated in each box. The solid bars in first readmission separate codes with equal dominance from those with lesser
dominance, the same hierarchy exists for 2nd readmissions. The hierarchy of codes is from left to right. For 1st readmission, deaths during
readmission are slightly below event boxes and deaths after discharge but without readmission are slightly above 2nd event rows. *denotes acute
ischaemic event. AMI: acute myocardial infarction, CVA: Acute Cerebrovascular Accident, Ang: angina/chest pain (acute ischaemic event), H:
Hypertension, AF: Atrial Fibrillation/paroxysmal SVT, Resp: Acute Respiratory infection, Renl: Renal Failure, HF: Heart Failure, D: Death.
D
36
AMI CVA Ang H AF
43
8 27 11 29
D
35
2nd Readmission
D
197
Potential
triggering factors
1st Readmission
Death
867
Death on index
2922
Events leading to the progression of HF
161
Eur Heart J, Vol. 22, issue 2, January 2001
162
A. U. Khand et al.
pectoris. Whilst this may underestimate the contribution
of acute ischaemic episodes to first readmissions it is
likely to represent a truer picture compared to including
ICD 9 codes 411 and 414 codes as well as other acute
and subacute forms of ischaemic heart disease and other
forms of chronic ischaemic heart disease. The latter
codes would largely represent background ischaemic
heart disease rather than a fresh event. (When incorporating all four ischaemic heart disease codes, first
readmission increased to 2597 patients, as defined in the
flowchart). Although we cannot similarly differentiate
acute hypertensive crisis precipitating admission from
background or ‘burnt out’ hypertension in the flowchart
the potential contribution of hypertension to precipitating readmission is far less in magnitude.
Those patients who were hospitalized with acute
myocardial infarction, subsequent to first heart failure
admission, had a particularly poor prognosis, as has
been noted by others[13,14].
Atrial fibrillation contributed to a smaller but still
significant proportion (11%) of first readmissions
(potential triggering event codes, heart failure and other
cardiovascular disease). It is unclear what role atrial
fibrillation played in precipitating readmission, but 45%
of these patients had no previous report of atrial fibrillation. Preliminary reports of patients from clinical trials
suggest that atrial fibrillation is associated with <5%
of first readmissions[20]. However, in some case series of
hospital admissions the proportion of patients with
atrial fibrillation acting as a trigger for heart failure
exacerbation is in the order of 20%[17,28]. The true figure
is likely to be closer to the latter studies, as atrial
fibrillation is undercoded (unpublished data) and therefore likely to be underestimated in this study, perhaps
because it may be seen, by some, as a condition of
secondary importance to a major diagnosis such as heart
failure. However, atrial fibrillation may be the principal
reason for heart failure deterioration either by causing a
tachycardiomyopathy[33] or because the stress of tachycardia and loss of atrioventricular synchrony reveals
underlying disease. The prognosis of first cardiovascular
readmissions with atrial fibrillation was better than the
overall population suggesting, perhaps, that many
patients did not have major underlying left ventricular
systolic dysfunction and that heart failure may have
resolved with control of the arrhythmia. Others have
noted that atrial fibrillation confers a worse prognosis
among patients with heart failure[34,35] although this has
not been consistent across studies. We cannot be certain
from our data-set what proportion of patients had acute
onset or paroxysmal atrial fibrillation compared with
those with chronic atrial fibrillation.
However, atrial fibrillation was associated with an
increased risk of stroke: 10% of patients with either a
previous or concomitant code for atrial fibrillation
suffered an acute stroke over 3 years compared with 6%
for the whole index population.
Concomitant codes for heart failure were listed in first
readmissions in 20–50% of all patients with a potential
triggering event code, mainly in the first or second
Eur Heart J, Vol. 22, issue 2, January 2001
coding position, suggesting a high rate of clinically overt
heart failure deterioration.
Although acute ischaemic events and atrial fibrillation
may be the most important clinical events contributing
to the progression of heart failure, in a substantial
proportion of first readmissions (31%) (Fig. 5) heart
failure was coded in the absence of potential triggering
events. These cases may truly reflect the contribution of
progressive ventricular remodelling. However, even in
this subset, almost a quarter of second readmissions
were reported to have an acute ischaemic or arrhythmic
triggering event (Fig. 5).
Death
The mortality in this cohort of patients is considerably
worse than that reported from clinical trials in patients
with chronic, stable heart failure but very similar to that
reported in epidemiological studies of incident heart
failure[5,6,8,36,37].
Although available to us (both inpatient and out of
hospital) we did not incorporate codes for death in this
analysis. There is considerable evidence that the cause of
death recorded on death certificates is unreliable (in
comparison, hospital discharge coding appears relatively
reliable)[38,39]. Furthermore, in Scotland, as in other
countries, reporting death as sudden or due to heart
failure on death certificates is discouraged and such
deaths are frequently reported, without supporting
evidence, such as myocardial infarction[40,41]. This is
consistent with this dataset; 41% of outpatient deaths
had a primary code for acute myocardial infarction
compared with 21% for inpatient deaths.
A substantial proportion of deaths occurred out of
hospital (27% of all deaths) with an especially high-risk
of death shortly after last hospital discharge. Most of
these are likely to fulfil common definitions of sudden
death[16,42]. However, death in these patients may not be
primarily arrhythmic[43]. Preliminary results from the
ATLAS trial[44] reveal that myocardial ischaemia is
indeed a frequent trigger for sudden death in heart
failure both in the presence and absence of documented
ischaemic heart disease.
The high risk of out of hospital death so soon after
discharge suggests that better risk stratification for
sudden death is needed, as is further research into
triggers and mechanisms of death in heart failure.
Non-cardiovascular disease
Another insight from the present study is the contribution that non-cardiovascular morbidity makes to
admissions or death. Gastro-intestinal and respiratory
disease and cancer made a significant overall contribution to readmissions (Fig. 3, Table 1). Renal failure
and acute respiratory infection triggered a substantial minority of first readmissions (Fig. 5) suggesting
Events leading to the progression of HF
non-cardiovascular disease may also drive the progression of heart failure. Improved treatment of acute heart
failure may prevent the development of secondary organ
dysfunction such as renal failure[45] (Fig. 2), which is
associated with a particularly poor prognosis. There
is also some evidence to suggest that angiotensinconverting enzyme inhibitors can reduce the risk of
respiratory infection[46]. Diabetes mellitus could be
responsible, in part, for rising hospital admissions with
heart failure in the future[47]. Treating heart failure well
may make these other conditions more tolerable, but
these problems will often require treatment in their own
right.
163
that we may need new strategies to reduce readmissions
in patients with heart failure. Some Class III antiarrhythmic agents[54–57] help to maintain patients with
heart failure in sinus rhythm; much more research into
effective strategies for the prevention and management
of atrial fibrillation are required. Unfortunately patients
with heart failure are usually excluded from trials of
agents investigating coronary events. Only now are trials
of antithrombotic agents[58] and statins being set up in
patients with heart failure to evaluate their risks and
benefits.
We thank the staff of the Information and Statistics Division of
the Scottish Health Services Agency for providing the data. We
would also like to thank Mrs Elaine Clark for her help with the text
and Morag Shaw for her help with the illustration.
Previous literature
Preliminary reports from cohorts of patients with heart
failure selected for inclusion in randomized controlled
trials suggest that ischaemic events account for about
10% of readmissions and that arrhythmias make a much
smaller contribution[19]. However these populations
have generally been younger, with chronic (>90 days)
stable heart failure and with left ventricular systolic
dysfunction and may have under-reported triggering
events for worsening heart failure or death.
Observational cohorts analysing precipitating factors
for heart failure readmission have described inconsistent
results. The variability between case series is probably
due to a narrow patient population; single secondary[17,48] or tertiary centre hospital settings[28], analysis
of certain age groups[49] or NYHA classes[28], and exclusion of first-time admissions with heart failure[17,48] and
also the effect of racial group is likely to have influenced
results in some studies[17].
Implications for treatment
ACE inhibitors, beta-blockers and spironolactone have
been shown to reduce recurrent hospitalization and
mortality in patients with heart failure, yet the mechanism of their benefit remains elusive[4,5]. Each can probably affect ventricular remodelling favourably, but ACE
inhibitors[13,15,50], beta-blockers[6,51,52] and spironolactone[53] probably also reduce reinfarction and/or
arrhythmias and these could be their dominant mechanisms of benefit. Whilst these data cast doubt on
ventricular remodelling as the sole factor driving the
progression of heart failure this does not necessarily
diminish its significance. Ventricular remodelling,
myocardial ischaemia and arrhythmias may represent
different aspects of a ‘vicious matrix’ in heart failure.
Progressive ventricular remodelling could precipitate
arrhythmias, including atrial fibrillation, and myocardial ischaemic syndromes could contribute to
remodelling and tachyarrhythmias.
If ischaemic events and supraventricular arrhythmias
are important pathways for progression, this suggests
References
[1] Mcmurray J, Mcdonagh T, Morrison CE, Dargie HJ. Trends
in hospitalisation for heart failure in Scotland 1980–1990. Eur
Heart J 1993; 14: 1158–62.
[2] Haldemand GA, Croft JB, Giles WH, Rashidee A. Hospitalization of patients with heart failure: National Hospital Discharge Survey, 1985 to 1995. Am Heart J 1999; 137: 352–60.
[3] Reitsma JB, Dalstra JAA, Bonsel GJ et al. Cardiovascular
disease in the Netherlands, 1975 to 1995: decline in mortality,
but increasing numbers of patients with chronic conditions.
Heart 1999; 82: 52–6.
[4] Swedberg K, Idanpaan Heikkila U, Remes J and CONSENSUS trial study group. Effects of enalapril on mortality in
severe congestive heart failure. Results of the Coperative
North Scandinavian Enalapril Survival Study (CONSENSUS). New Eng J Med 1987; 316: 1429–35.
[5] Packer M, Bristow MR, Cohn JN et al. The effect of
carvedilol on morbidity and mortality in patients with chronic
heart failure. U.S. Carvedilol Heart Failure Study Group.
New Eng J Med 1996; 334: 1349–55.
[6] CIBIS-II investigators and Committees. The cardiac insufficiency Bisoprolol Study II (CIBIS-II): a randomised trial.
Lancet 1999; 353: 9–13.
[7] Ball SG, Hall AS, Mackintosh AF et al. Effect of ramipril and
morbidity of surviviors of acute myocardial infarction with
clinical evidence of heart failure. Lancet 1993; 342: 821–8.
[8] Cleland JGF, Gemmel I, Khand AU, Boddy A. Is the
prognosis of heart failure improving? Eur J Heart Failure
1999; 1: 229–41.
[9] Teerlink JR, Pfeffer JM, Pfeffer MA. Effect of left ventricular
sphericity on the evolution of ventricular dysfunction in rats
with diffuse isoproteronol-induced myocardial necrosis. J
Cardiac Failure 1998; 4: 45–56.
[10] Torp-Pederson LKC, Kober L, Carlsen J. Angio-converting
enzyme inhibition after myocardial infarction: the Trandolapril Cardiac Evaluation Study. Am Heart J 1996; 132:
235–43.
[11] St. John Sutton M, Pfeffer MA, Moye L et al. Cardiovascular
death and left ventricular remodeling two years after myocardial infarction: baseline predictors and impact of long-term
use of captopril: information from the Survival and Ventricular Enlargement (SAVE) trial. Circulation 1997; 96: 3294–9.
[12] Cohn JN, Johnson GR, Ziesche S et al. A comparison of
enalapril with hydralazine-isosorbide dinitrate in the treatment of chronic congestive heart failure. N Eng J Med 1991;
325: 303–10.
[13] Yusuf S, Pepine CJ, Garces C et al. Effect of enalapril on
myocardial infarction and unstable angina in patients with
low ejection fractions. Lancet 1992; 340: 1173–8.
Eur Heart J, Vol. 22, issue 2, January 2001
164
A. U. Khand et al.
[14] Rutherford JD, Pfeffer MA, Moye LA et al. Effects of
captopril on ischaemic events after myocardial infarction.
Results of the Survival and Ventricular Enlargement trial
SAVE Investigators. Circulation 1994; 90: 1731–8.
[15] Lonn EM, Yusuf S, Montague TJ et al. Emerging role of
angiotensin-converting enzyme inhibitors in cardiac and vascular protection. Circulation 1994; 90: 2056–69.
[16] Cleland JGF, Erhardt L, Murray G, Hall AS, Ball SG and
AIRE study investigators. Effect of ramipril on morbidity and
mode of death among survivors of acute myocardial infarction
with clinical evidence of heart failure. Eur Heart J 1997; 18:
41–51.
[17] Ghali JK, Kadakia S, Cooper R, Ferlinz J. Precipitating
factors leading to decompensation of heart failure. Traits
among urban blacks. Arch Internal Medicine 1988; 148:
2013–6.
[18] Krumholz HM, Parent EM, Tu N et al. Early readmission of
elderly patients with congestive heart failure. J Am Geriatr
Soc 1990; 38: 1290–5.
[19] Cleland JGF, Poole-Wilson PA, Packer M et al. Critical event
pathways for progression of heart failure and mortality:
Insights from the ATLAS study. [Abstr]. J Am Coll Cardiol
1999; 33 Suppl A 212A: 1228–8.
[20] Anonymous. Effect of enalapril on survival in patients with
reduced left ventricular ejection fractions and congestive heart
failure. The SOLVD investigators. N Eng J Med 1991; 325:
293–302.
[21] Remme WJ. Prevention of worsening heart failure: future
focus. Eur Heart J 1998; 19B: 47–53.
[22] Patten RD, Udelson JE, Konstam MA. Ventricular remodeling and its prevention in the treatment of heart failure. Curr
Opin Cardiol 1998; 13: 162–7.
[23] Lechat P. Prevention of heart failure progression: current
approaches. Eur Heart J 1998; 19: B12–8.
[24] Cowie MR, Wood DA, Coats AJS et al. Incidence and
aetiology of heart failure, A population-based study. Eur
Heart J 1999; 20: 421–8.
[25] Kendrick S, Clarke J. The Scottish Record Linkage System.
Health Bull 1993; 51: 72–9.
[26] Harley K, Jones C. Quality of Scottish Morbidity Record
(SMR) Data. Health Bull 1996; 54: 410–7.
[27] Anonymous. The West of Scotland Coronary Prevention
Group: Computerised record linkage compared with traditional patients follow-up methods in clinical trials and
illustrated in a prospective epidemiological study. J Clin
Epidemiology 1995; 48: 1441–52.
[28] Opasich C, Febo O, Riccardi PG et al. Concomitant factors of
decompensation in chronic heart failure. American Journal of
Cardiology 1996; 78: 354–7.
[29] Krumholz HM, Parent EM, Tu N et al. Readmission after
hospitalization for congestive heart failure among medicare
beneficiaries. Arch Intern Med 1997; 157: 99–104.
[30] Gooding J, Hette AM. Hospital readmissions among the
elderly. J Am Geriatr Soc 1990; 33: 595–601.
[31] Feenstra J, Grobbee DE, Jonkman FAM, Hoes AW, Stricker
BHC. Prevention of relapse in patients with congestive heart
failure: the role of precipitating factors. Heart 1998; 80: 432–6.
[32] Registrar General. Annual Report of the Registrar General
For Scotland 1992. The Scottish Office, Edinburgh, Scotland.
[33] Van Den Berg MP, Tuinenberg AE, Crijns HJGM et al. Heart
failure and atrial fibrillation: current concepts and controversies. Heart 1997; 77: 309–13.
[34] Cowburn PJ, Cleland JGF, Coats AJS, Komajda M. Risk
stratification in chronic heart failure. Eur Heart J 1998; 19:
696–710.
[35] Khand AU, Rankin AC, Kaye G, Cleland JGF. Systematic
review of the management of atrial fibrillation in patients with
heart failure. Eur Heart J 2000; 21: 614–32.
[36] Ho KK, Pinsky JL, Kannel WB, Levy D. The epidemiology of
heart failure: the Framingham Study. J Am Coll Cardiol 1993;
22: 6A–13A.
[37] Cowie MR, Mosterd A, Wood DA. The epidemiology of
heart failure. Eur Heart J 1997; 18: 208–15.
Eur Heart J, Vol. 22, issue 2, January 2001
[38] Kohli HS, Knil-Jones RP. How accurate are SMR 1 (Scottish
Morbidity Record 1). Data. Health Bull 1992; 50: 14–23.
[39] Patel AR, Gray G, Lang GD, Baillie FGH, Fleming L,
Wilson GM. Errors in diagnostic returns. Health Bull 1976;
34: 215–20.
[40] Murdoch DR, Love MP, Robb SD et al. Importance of heart
failure as a cause of death. Changing contribution to overall
mortality and coronary heart disease mortality in Scotland
1979–1992. Eur Heart J 1998; 19: 1829–35.
[41] Lauer MS, Blackstone EH, Young J, Topol EJ. Cause of
Death in Clinical Research. J Am Coll Cardiol 1999; 34:
618–20.
[42] Narang R, Cleland JGF, Erhardt L et al. Mode of death in
chronic heart failure: a request for more accurate classification. Eur Heart J 1996; 17: 1390–403.
[43] Cleland JGF, Massie BM, Packer M. Sudden Death in Heart
Failure: vascular or electrical? Eur J Heart Fail 1999; 1: 41–5.
[44] Uretsky B, Armstrong PW, Cleland JGF et al. Ischaemia as a
’trigger’ for sudden death in patients with cardiomyopathy
and heart failure secondary to coronary artery disease: results
from the ATLAS trial [Abstr]. Circulation 1998; (suppl)I199.
[45] Morooka S, Hayashi T, Takayanagi K, Inoue T, Sakai Y,
Takabatake Y. Heart Failure progression due to secondary
organ dysfunction in acute heart failure. Japan Heart J 1995;
36: 29–36.
[46] Okaishi K, Morimoto S, Fukuo K et al. Reduction of Risk of
Pneumonia Associated with use of Angiotensin I Converting
Enzyme Inhibitors in elderly Inpatients. Am J Hypertension
1999; 99(12): 778–83.
[47] Wikstrom G, Malmberg K, Ryden L. Improved knowledge of
antidiabetic treatment — a necessity for the modern cardiologist. Eur Heart J 1999; 20: 403–5.
[48] Michalsen A, Konig G, Thimme W. Preventable causative
factors leading to hospital admission with decompensated
heart failure. Heart 1998; 80: 437–41.
[49] Vinson JM, Rich MW, Sperry JC et al. Early readmission of
elderly patients with congestive heart failure. J Am Geriatr
Soc 1990; 38: 1290–5.
[50] Pedersen OD, Bagger H, Torp-Pederson LKC. Trandolapril
reduces incidence of atrial fibrillation following acute myocardial infarction. Circulation 1999; 100: 376–80.
[51] Anonymous. Effects of carvedilol, a vasodilator beta-blocker,
in patients with congestive heart failure due to ischaemic heart
disease. Australia-New Zealand Heart Failure Research Collaborative Group. Circulation 1995; 92: 212–8.
[52] Anonymous. Randomised trial of intravenous atenolol among
16 027 cases of suspected acute myocardial infarction: ISIS-1.
First international study of Infarct survival collaborative
group. Lancet 1986; ii: 57–66.
[53] Pitt B, Zannad F, Remme WJ et al. The effects of spironolactone on morbidity and mortality in patients with severe
heart failure. N Eng J Med 1999; 341: 709–17.
[54] Pedersen OD, The DIAMOND study group. Dofetilide in the
treatment of atrial fibrillation in patients with impaired left
ventricular function: Atrial Fibrillation in the DIAMOND
study [Abstract]. Circulation 1998; 98: I-632.
[55] Cappuci A, Aschieri GQ, Villani F, Passerini F, Groppi A.
Oral amiodarone increases efficacy of DC cardioversion in
patients with chronic atrial fibrillation and organic heart
disease. J Am Coll Cardiol 1997; 29: 443A.
[56] Chapman MJ, Moran JL, O’Fathartaigh MS, Peisach AR,
Cunningham DN. Management of atrial tachyarrhythmias in
the critically ill: a comparison of intravenous procainamide
and amiodarone. Intensive Care Medicine 1993; 19: 48–52.
[57] Deedwania PC, Singh BN, Ellenbogen KA, Fisher S, Fletcher
R, Singh SN. Spontaneous conversion and maintenance of
sinus rhythm by amiodarone in patients with heart failure and
atrial fibrillation. Circulation 1998; 98: 2574–9.
[58] The Wash Study Steering Committee and Investigators. The
WASH study (Warfarin/Aspirin Study in Heart Failure)
rationale, design and end-points. Eur J Heart Fail 1999;
1:95–9.