OUP user menu

Atrial fibrillation and long-term prognosis in patients hospitalized for heart failure: results from heart failure survey in Israel (HFSIS)

Avraham Shotan, Moshe Garty, David S. Blondhein, Simcha R. Meisel, Basil S. Lewis, Michael Shochat, Ehud Grossman, Avi Porath, Valentina Boyko, Reuven Zimlichman, Abraham Caspi, Shmuel Gottlieb
DOI: http://dx.doi.org/10.1093/eurheartj/ehp422 309-317 First published online: 15 October 2009


Aims Atrial fibrillation (AF) and heart failure (HF) commonly coexist, and each adversely affects the other. The aim of the study was to prospectively evaluate the impact of AF and its subtypes on management, and early and long-term outcome of hospitalized HF patients.

Methods and results Data were prospectively collected on HF patients hospitalized in all public hospitals in Israel as part of a national survey (HFSIS). Atrial fibrillation patients were subdivided into intermittent and chronic AF subgroups. During March–April 2003, we enrolled 4102 HF patients, of whom 1360 (33.2%) had AF [600 (44.1%) intermittent, 562 (41.3%) chronic]. Patients with AF were older (76.9 ± 10.5 vs. 71.7 ± 12.6 years, P = 0.0001), males, with preserved LV systolic function. Crude mortality rates for AF patients were progressively and consistently higher during hospitalization and during the 4-year follow-up period, especially in the chronic AF group (P = 0.0001). After covariate adjustment, AF was associated with increased 1-year mortality [HR 1.19, 95% CI (1.03–1.36)].

Conclusion AF was present in a third of hospitalized HF patients, and identified a population with increased mortality risk, largely due to co-morbidities.

  • Atrial fibrillation
  • Heart failure
  • Prognosis


Atrial fibrillation (AF) and heart failure (HF) commonly coexist, and adversely affect each other. The prevalence of AF varies widely in HF studies, somewhat greater among patients with preserved left-ventricular (LV) systolic function.15 HF results in haemodynamic and neurohormonal alterations, cellular and extracellular remodelling, that may promote arrhythmogenesis.4,5 Adverse effects of AF include reduction in stroke volume and peak oxygen consumption, tachycardia-induced cardiomyopathy, and drug toxicity of antiarrhythmics. Co-existence of HF and AF increases the risk of thromboembolism.25

The characteristics and long-term outcome of patients with HF complicated by AF and the effect of medications have not been fully investigated. Moreover, few studies have examined possible differences in patients with chronic as opposed to intermittent AF.

The purpose of this nationwide study was to characterize and evaluate the impact of AF and its subtypes on HF patients, focusing primarily on 1 year outcome, and secondarily on hospital management and hospital and 4 year outcome.


Patient population

The study population consisted of patients included in the HFSIS (Heart Failure Survey in ISrael) 2003 heart failure survey—a prospective nationwide survey of hospitalized HF patients, conducted in all 25 public hospitals in Israel. The study design and its main results have been published.6 The survey included hospitalized HF patients ACC/AHA stages B–D. Detailed data on patients’ characteristics, in-hospital course, management during their hospitalization period, pre-hospital and discharge medications and diagnoses were collected and recorded on pre-specified structured forms. Diagnoses of HF, hypertension, valvular heart disease (at least moderate stenosis or insufficiency), diabetes mellitus, and dyslipidaemia were determined by the local survey team. Killip classification, which clinically describes HF signs severity, was used by us to describe not only acute MI on admission, but all HF patients during the entire hospitalization. We did not follow-up our patients clinically post-discharge, and therefore have no data about further hospitalizations, compliance with medications, and additional patients developing AF.

Mortality was assessed for 99% of patients, by matching their identification numbers with the Israeli National Population Registry. We could not retrieve the cause of death. The protocol was approved by the Ethics Committee at each of the participating hospitals.

Definition of atrial fibrillation

Atrial fibrillation, determined as a main topic in our HF survey, was diagnosed by the local survey team. In spite of our detailed questionnaire we could not get sufficient data to subdivide AF accurately into subgroups. Therefore we predefined AF subtypes, grouping first-detected, paroxysmal, and persistent AF together as intermittent AF, and compared it with permanent AF, which we termed chronic AF.2

Statistical analysis

First we compared HF patients with AF vs. patients with no-AF, and thereafter AF subgroups: intermittent vs. chronic AF. All analyses were performed using SAS software (version 8.2, SAS Institute, Inc., Cary, NC). The χ2 and unpaired t-test were used to determine the significance of differences between proportions and means, respectively. Results of continuous variables are reported as means ± SD.

A two-sided P < 0.05 was considered as statistically significant. No adjustments for multiple comparisons were made to the significant level.

Multivariable analysis of hospital, 1 year and 4 year mortality were done, using the Cox proportional hazards model, and its results are reported as hazards ratio (HR) and 95% CI. We included in the models a broad range of characteristics, that we considered clinically important as they were statistically significant (P < 0.10) by univariable analysis (Table 5). A test based on a different time-dependent covariate was used for assessing the proportionality of hazards. Missing values like New York Heart Association (NYHA) class [103 (2.5%)], type of HF [95 (2.3%)] and LV ejection fraction (LVEF) [1257 (30.6%)] were included in the models as separate categories.

Survival curves were constructed by the Kaplan–Meier method. The log-rank test was used for comparing the curves.


The HFSIS survey, conducted during the 2-month period March–April 2003 included 4102 consecutive unselected HF patients. Atrial fibrillation was reported in 1360 (33.2%) patients. Among AF patients, 600 (44.1%) had intermittent AF, 562 (41.3%) had chronic AF, and in 198 (14.6%) type undefined (AF type unknown).

Demographic data

The baseline characteristics of the patients are given in Table 1. Atrial fibrillation patients were hospitalized more often in internal medicine departments, were older and more frequently males. Chronic AF patients had more frequently a diagnosis of valvular heart disease, but less often coronary artery disease (CAD), acute coronary syndrome (ACS), and other co-morbidities.

View this table:
Table 1

Baseline characteristics (%)

No AF (n = 2742) (66.8%)AF—all (n = 1360) (33.2%)P-value (AF vs. no AF)Chronic AF (n = 562) (13.7%)Intermittent AF (n = 600) (14.6%)AF type unknown (n = 198) (4.8%)P-value (intermittent vs. chronic AF)
Site of hospitalization internal medicine76.186.50.000190.482.082.60.0001
Age (years), mean ± SD71.7 ± 12.676.9 ± 10.50.000177.3 ± 10.176.6 ± 10.576.4 ± 11.30.27
 Male69.8 ± 12.675.6 ± 11.00.000176.5 ± 10.175.1 ± 11.674.8 ± 11.30.12
 Female74.6 ± 12.078.2 ± 9.80.000178.2 ± 10.178.3 ± 8.977.9 ± 11.00.87
Diabetes mellitus53.443.60.000139.547.044.90.01
 Insulin treated14.48.80.00016.810.59.10.03
Current smoking12.65.30.00013.
Coronary artery disease84.976.70.000174.081.569.70.0002
Acute coronary syndrome41.527.10.000123.533.019.20.0004
Valvular heart disease32.443.00.000146.840.240.90.03
Renal failure (creatinine ≥1.5 mg/dL)39.542.8<0.0540.245.541.90.07
Anaemia (haemoglobin ≤12 g/dL)
Chronic obstructive lung disease19.619.50.9219.620.017.70.86
Peripheral vascular disease9.29.00.826.911.76.60.006
Stroke/transient ischaemic attack11.813.80.0814.
  • AF, atrial fibrillation.

Atrial fibrillation patients were more symptomatic (higher NYHA class) prior to hospitalization (Table 2), and had more often signs of right HF on physical examination. Killip classification tended to be higher among AF patients. Chronic AF patients were more symptomatic (NYHA class III/IV) than intermittent AF patients, and presented more often with signs of right HF. They tended to develop more frequently signs of mild left HF (Killip class II), but less Killip class III/IV.

View this table:
Table 2

Clinical findings (%)

No AF (n = 2742) (66.8%)AF—all (n = 1360) (33.2%)P-value (AF vs. no AF)Chronic AF (n = 562) (13.7%)Intermittent AF (n = 600) (14.6%)AF type unknown (n = 198) (4.8%)P-value (intermittent vs. chronic AF)
New York Heart Association functional class prior to hospitalization, n (%)2669 (66.7)1330 (33.3)544 (13.6)592 (14.8)194 (4.8)
Physical examination findings
 Systolic blood pressure (mmHg)143 ± 33139 ± 290.0001139 ± 28138 ± 30139 ± 310.86
 Diastolic blood pressure (mmHg)78 ± 1777 ± 160.0777 ± 1776 ± 1676 ± 160.23
 Heart rate (b.p.m.)82 ± 1989 ± 250.000188 ± 2389 ± 2691 ± 270.81
 Jugular venous distension28.533.50.00140.027.732.30.0001
 Pulmonary rales53.153.30.9053.454.350.00.77
Killip classification during hospitalization2742 (66.8)1360 (33.2)562 (13.7)600 (14.6)198 (4.8)
Chest X-ray2596 (67.0)1279 (33.0)531 (13.7)566 (14.6)182 (4.7)
 Pulmonary congestion38.642.90.00842.342.346.51.0
 Pulmonary oedema15.112.90.0611.714.212.10.22
 Pleural effusion16.119.60.00621.416.822.7<0.05
Electrocardiographic findings
Echocardiographic findings2195 (67.8)1041 (32.2)416 (12.9)479 (14.8)146 (4.5)
 Aortic stenosis8.810.70.0411.69.711.60.14
 Mitral regurgitation48.249.00.6247.951.544.90.34
 Left-ventricular hypertrophy14.715.90.3013.216.820.70.04
Left-ventricular ejection fraction1944 (68.4)898 (31.6)353 (12.4)420 (14.8)125 (4.4)
 Normal (≥50%)23.933.20.00135.731.432.00.15
 Mildly reduced (40–49%)21.320.722.418.623.2
 Preserved (≥40%)45.353.958.150.055.2
 Moderately reduced (30–39%)28.719.716.722.419.2
 Severely reduced (<30%)26.026.425.227.625.6
Coronary angiography34.220.30.000117.124.516.70.0001
Percutaneus coronary intervention18.17.50.00016.
  • AF, atrial fibrillation.

Atrial fibrillation patients presented similar to no-AF patients with regard to acute HF (59.0% AF vs. 56.8% no-AF, P = NS). Chronic HF patients presented less often with acute new onset HF than intermittent AF (7.5 vs. 17.4%, P < 0.0001).

The chest X-ray disclosed more often signs of HF in AF patients, but less often florid pulmonary oedema. Atrial fibrillation patients had normal (≥50%) and preserved (≥40%) LVEF [33 and 54%, respectively vs. 24 and 45% in non-AF group (P < 0.001)].

Patient management

During hospitalization AF patients underwent fewer coronary catheterizations and percutaneous coronary interventions (PCI) (Table 2). Intravenous furosemide was given to 62.6% of AF patients vs. 59.4% of no-AF patients (P < 0.05), positive inotropic medications to 7.6 and 6.4%, respectively (P = NS), and vasodilators to 7.4 and 12.1%, respectively (P = 0.0001). Electrical cardioversion was performed during the index hospitalization in 50 (3.7%) AF patients vs. 35 (1.3%) no-AF patients (P = 0.0001).

Comparing chronic with intermittent AF, intravenous furosemide, and positive inotropic medications were given similar to both the chronic and intermittent AF subgroups (63.2 and 7.7% vs. 62.3 and 7.5%, NS). Intravenous vasodilators were administered less often to chronic than intermittent AF patients (4.6 vs. 10.5%, P = 0.0001). Electrical cardioversion was performed in 6 (1.1%) chronic vs. 37 (6.2%) intermittent AF patients (P = 0.0001).

At hospital discharge, AF patients were treated more often with anticoagulants, digoxin, calcium channel blockers (CCBs), especially verapamil, diuretics, and spironolactone, while no-AF patients were treated more with aspirin, clopidogrel, ACE-I, beta-blockers, and statins (Table 3). Antiarrhythmic medications, mostly amiodarone were prescribed more often to AF patients. At discharge, 281 (6.9%) patients were reported as having a permanent cardiac pacemaker. Pacemakers were more frequent in patients with AF than without AF [113 (8.3%) vs. 168 (6.1%), (P = 0.04)], and more often in patients with intermittent rather than chronic AF [58 (10.3%) vs. 44 (7.3%), P = 0.004]. An implantable cardio-defibrillator (ICD) was noted in 104 (2.5%) patients, similarly in AF and no-AF patients [33 (2.4%) vs. 71 (2.6%), NS] and intermittent [15 (2.3%)] vs. chronic [15 (2.7%), NS] AF.

View this table:
Table 3

Discharge medications (%)

No AF (n = 2619) (67.2%)AF—all (n = 1278) (32.8%)P-value (AF vs. no AF)Chronic AF (n = 531) (13.6%)Intermittent AF (n = 566) (14.5%)AF type unknown (n = 181) (4.6%)P-value (intermittent vs. chronic AF)
Other antiarrhythmics1.03.60.00010.
Anticoagulants: all9.948.00.000157.639.247.50.0001
Anticoagulants (p.o.)
Anticoagulants (i.v./s.c.)
ARB and/or ACE-I69.365.30.0267.463.464.60.16
Calcium channel blockers23.927.80.00827.126.932.60.92
Alpha blockers10.110.10.9910.410.19.40.87
Antidiabetics: oral26.418.60.000117.
  • AF, atrial fibrillation.

Early and late mortality

Crude mortality rates were higher in AF patients during index hospitalization and during follow-up (Table 4 and Figure 1). Mortality, both at 1 and 4 years was higher in patients with chronic vs. intermittent AF.

Figure 1

Unadjusted survival curves. (A) Atrial fibrillation and no-atrial fibrillation patients. (B) Atrial fibrillation subtypes and no-atrial fibrillation patients.

View this table:
Table 4

All-cause mortality (unadjusted)

PeriodNo AF (n = 2734) (66.8%)AF—all (n = 1359) (33.2%)P-value (AF vs. No AF)Chronic AF (n = 566) (13.6%)Intermittent AF (n = 599) (14.5%)AF type unknown (n = 198) (4.6%)P-value (Intermittent vs. chronic AF)
Hospital113 (4.1)80 (5.9)0.0230 (5.3)33 (5.5)17 (8.6)0.90
30 day189 (6.9)121 (8.9)0.0246 (8.2)50 (8.3)25 (12.6)0.92
6 month467 (17.1)300 (22.1)0.0001132 (23.5)117 (19.5)51 (25.8)0.10
1 year705 (25.8)447 (32.9)0.0001206 (36.7)174 (29.0)67 (33.8)0.006
4 year1480 (54.3)882 (64.9)0.0001397 (70.6)359 (59.9)126 (64.3)0.0001
  • AF, atrial fibrillation.

After covariate adjustment, hospital all-cause mortality of AF in comparison to no-AF patients was HR (±95% CI) 1.26 (0.89–1.83), 1 year mortality 1.19 (1.03–1.36), and 4 year mortality 1.09 (0.99–1.21), reaching statistical significance only at 1 year, and particularly so for patients with unknown type of AF (hospital and 1 year) and chronic AF only at 4 years (Table 5).

Variables associated with increased mortality for the entire study population are presented in Table 5. Subtypes of AF were included as variables in the model.

View this table:
Table 5

Predictors of mortality

Hospital, HR (±95% CI)1 year, HR (±95% CI)4 year, HR (±95% CI)
Intermittent AF1.30 (0.85–1.99)1.15 (0.96–1.38)1.02 (0.90–1.15)
Chronic AF1.02 (0.63–1.66)1.18 (0.98–1.42)1.18 (1.031.35)
AF unknown1.74 (1.003.03)1.31 (1.011.71)1.17 (0.97–1.42)
Age1.02 (1.011.04)1.02 (1.021.03)1.03 (1.031.04)
Male gender0.56 (0.410.77)0.89 (0.78–1.01)0.99 (0.90–1.08)
Hypertension0.53 (0.380.74)0.79 (0.690.91)0.81 (0.730.89)
Diabetes1.28 (0.94–1.75)1.09 (0.96–1.23)1.24 (1.131.35)
CAD0.84 (0.58–1.22)1.03 (0.88–1.19)0.96 (0.86–1.07)
ACS1.47 (1.052.07)1.01 (0.88–1.16)0.98 (0.89–1.08)
Valvular1.18 (0.87–1.60)1.04 (0.92–1.18)1.06 (0.97–1.16)
Renal failure2.69 (1.933.75)1.59 (1.401.810)1.48 (1.361.62)
Anaemia1.12 (0.82–1.52)1.37 (1.211.55)1.33 (1.221.45)
PVD1.21 (0.76–1.93)1.12 (0.93–1.36)1.31 (1.151.50)
COPD0.96 (0.66–1.40)1.17 (1.021.35)1.18 (1.071.30)
Stroke1.74 (1.192.55)1.48 (1.261.73)1.26 (1.121.42)
NYHA III–IV1.81 (1.083.03)1.70 (1.372.11)1.47 (1.271.70)
Killip II1.66 (0.89–3.09)1.38 (1.141.68)1.27 (1.121.45)
Killip III–IV3.53 (1.966.34)1.57 (1.291.92)1.40 (1.231.60)
LVEF 30–39%1.47 (0.93–2.34)1.11 (0.92–1.35)1.06 (0.93–1.21)
LVEF<30%1.70 (1.082.67)1.45 (1.201.74)1.38 (1.181.52)
Primary HF diagnosis0.47 (0.340.64)0.78 (0.680.90)0.96 (0.871.06)
Acute HF: new olnset2.40 (1.523.80)1.08 (0.88–1.33)0.97 (0.84–1.12)
Acute HF: exacerbation1.62 (1.102.38)1.11 (0.96–1.29)1.08 (0.97–1.19)
Amiodarone0.57 (0.32–1.01)0.82 (0.67–1.01)0.97 (0.85–1.11)
Other antiarrhythmics0.53 (0.13–2.18)0.69 (0.42–1.14)0.76 (0.55–1.05)
Anticoagulants1.04 (0.69–1.57)0.85 (0.71–1.00)0.85 (0.750.95)
Aspirin1.00 (0.73–1.38)0.98 (0.86–1.12)0.96 (0.87–1.05)
Clopidogrel0.56 (0.23–1.39)0.91 (0.60–1.31)0.84 (0.67–1.05)
ACE-I/ARB0.79 (0.58–1.07)0.85 (0.750.97)0.88 (0.810.96)
Beta blockers0.76 (0.54–1.06)0.82 (0.720.93)0.82 (0.7510.90)
Digoxin1.53 (1.012.30)1.19 (1.001.41)1.24 (1.10140)
CCBs: non-dihydropyridines1.29 (0.75–2.19)0.78 (0.610.99)0.77 (0.650.91)
CCBs: dihydropyridines1.38 [0.94–2.02]0.94 (0.80–1.11)0.94 (0.44–1.05)
Furosemide1.23 (0.85–1.78)1.31 (1.131.53)1.36 (1.231.51)
Spironolactone1.09 (0.71–1.66)1.17 (0.99–1.37)1.19 (1.061.33)
Nitrates0.58 (0.410.83)0.92 (0.80–1.05)1.00 (0.91–1.09)
Statins0.76 (0.52–1.10)0.83 (0.720.95)0.81 (0.730.88)
  • Adjusted for AF subtypes, age, gender, hypertension, diabetes mellitus, CAD, ACS, valvular heart disease, non-ischaemic CMP, renal failure, anaemia, PVD, COPD, stroke, NYHA class III–IV, Killip class, LVEF—preserved, moderately, and severely decreased, primary HF, secondary HF, acute new onset HF, acute exacerbation of chronic HF, chronic HF, amiodarone, antiarrhythmics, anticoagulants, aspirin, clopidogrel, ACE-I, ARBs, beta-blockers, furosemide, aldosterone blockers, CCBs—dihydropyridines, CCBs—non-dihydropyridines, digoxin, nitrates, and statins.

  • AF, atrial fibrillation.

  • Results displayed in bold are statistically significant.


The main findings of this nationwide HF survey of patients hospitalized for HF were the high prevalence of AF, a detailed clinical picture including physical examination findings, a lower rate of ACS, and a worse outcome of mainly chronic AF patients, largely due to co-morbidities.

Atrial fibrillation was common and occurred in a third of the survey patients, particularly in older, males, and in those with preserved (LVEF >40%) or ‘normal’ (LVEF >50%) LV systolic function. The prevalence of AF in our study was based on previous AF reported and/or the presence of AF on the baseline ECG. As many patients were not monitored, especially in internal medicine, we could miss episodes of transient AF. Consequently, the prevalence of AF in our survey, although high may be underestimated.

Atrial fibrillation was associated with progressively increasing crude total mortality rates over a 4 year follow-up period. The presence of AF identified a high-risk HF population, even though the increased hazard was explained for the most part by co-morbidities.

Similar to other series AF patients had more valvular heart disease, renal failure, and hypertension.79 They had a lower rate of CAD, especially ACS, and most CAD risk factors. It seems that AF itself may cause symptomatic HF or cardiac dysfunction, without the insult of ischaemia, or myocardial and vascular damage attributable to diabetes mellitus. Atrial fibrillation patients, especially with chronic AF, underwent less coronary diagnostic and therapeutic procedures. To what extent the coexistence of AF contributed to HF symptoms and signs, or was merely the result of older age and higher prevalence of co-morbidities, like valvular disease or renal failure, cannot be answered by this survey.

Atrial fibrillation patients had higher rates of normal and preserved LV function, implying mainly diastolic dysfunction. Atrial fibrillation patients included, therefore, two main populations: the larger (54% of patients) with preserved LV, and the second with severely reduced LVEF (<30%). Left-ventricular dysfunction (diastolic or systolic) and AF, are mutually detrimental in terms of pathogenesis, but whichever is cause and whichever effect, is beyond the scope of this survey.

Chronic vs. intermittent atrial fibrillation

Chronic AF patients had less co-morbidities compared with intermittent AF patients, but suffered more frequently from valvular heart disease. Chronic AF patients presented with more advanced, but relatively stable HF, and less with acute HF, especially acute new onset symptoms.

The reported rate of thyroid function abnormalities among AF patients was low (7%), as 23% of them were discharged on amiodarone, especially intermittent AF patients (40%). We lack information, in whom had thyroid function tests actually been checked, although thyroid function tests were recommended in both HF and AF guidelines.2,10

Patient management

Management of AF is primarily rate control or maintaining sinus rhythm, and prevention of thromboembolism.2,5,79 At discharge antiarrhythmics were prescribed only to 27% of AF patients (amiodarone 23%, other 4%), 8% of chronic AF (amiodarone 7%, other 1%), and 47% of intermittent AF patients (amiodarone 40%, other 7%). The precise indications for antiarrhythmics were not available, and at least in some patients the indication was ventricular arrhythmias. Verapamil, which is not recommended for systolic dysfunction, and digoxin were prescribed more often to chronic than to intermittent AF patients, probably mainly for rate-control. The similar rate of ICD use in the AF and no-AF HF population, exclude the effect of ICD on long-term prognosis in terms of sudden cardiac death.

Several HF indicated medications can prevent AF, such as ACE-I and ARBs. However, ACE-I were prescribed less to AF than to no-AF patients, and ARBs at similar rates. Beta-blockers that are indicated for HF, and for rate and rhythm control, were prescribed less often to AF patients (53%) than to no-AF patients (63%). The relatively low rate is probably due to older age of AF patients, conduction disturbances, and renal failure.

The AFFIRM and RACE studies, published prior to our survey showed no advantage to rhythm over rate control, and contributed to the relatively low usage of antiarrhythmics,7,9 although pharmacological maintenance of sinus rhythm may reduce morbidity in patients with HF.11 However, the recent AF-CHF study also showed no benefit to rhythm over rate control.5

Prevention of thromboembolism

Atrial fibrillation guidelines recommend warfarin unless contraindicated for AF patients at moderate to high risk of thromboembolism (CHADS2>1).2,12,13 According to baseline characteristics, almost all our AF patients had an indication for chronic anticoagulation. However, only 48% of the AF patients were discharged on anticoagulant therapy, more so in chronic (58%), than intermittent AF (39%), and only 10% no-AF patients. This rate is similar to other ‘real world’ series, in which only about half of patients with indication for anticoagulation are treated,3,14 and leaves a great deal of improvement and implementation of guidelines-based therapy.


The mortality among AF patients in the general population is approximately double that of patients in normal sinus rhythm, and depends on the severity of underlying heart disease.3,15,16 In large HF trials, AF as an independent mortality predictor had conflicting results. In V-HeFT trials mortality was not increased among patients with concomitant AF,17 whereas in SOLVD mortality was 34% for AF vs. 23% for sinus rhythm patients (P < 0.001).18 In COMET there was no difference in all-cause mortality in patients with AF at entry, but mortality increased in those who developed AF during follow-up.19 In Val-HeFT development of AF was associated with significantly worse outcome.20

Overall crude mortality for the entire survey population was reasonable during hospitalization (4.7%), but very high at 1 (28.2%) and at 4 years (57.7%).6 AF patients had higher 1 and 4 year crude mortality (32.9 and 64.9%, respectively), especially chronic AF patients (36.7 and 70.6%, respectively). After adjustment AF was associated with significant 19% increased mortality only at 1 year, similar to the 1 year mortality data of the recent Euro-Heart survey on AF.21 Adjusted 4 year mortality was only 9% insignificantly higher among AF patients, but with 18% higher mortality among chronic AF patients. Intermittent AF patients had initially insignificant higher adjusted hospital mortality, similar to the higher hospital mortality among new onset AF patients in EuroHeart Failure Survey I.22 However, adjusted mortality among intermittent AF patients had declined at 1 year and became at 4 years similar to no-AF patients. The better outcome of intermittent AF patients can partially be explained by lower NYHA class prior to hospitalization, younger age, and somewhat lower rate of valvular heart disease. The group of AF patients with unknown type comprised high-risk patients who had higher early crude and adjusted mortality and mixture of intermittent AF and chronic AF patients, whose 1 year and 4 year post-discharge outcome was accordingly between intermittent and chronic AF (Tables 4 and 5, Figure 1).

Predictors associated with increased adjusted mortality for the entire HF population were sicker HF patients, like NYHA III–IV, Killip III/IV, and LVEF<30%, co-morbidities, such as renal failure and prior stroke, age and digoxin, during hospitalization ACS and acute HF, at 1 year and 4 years anaemia, COPD, Killip II, and furosemide, and at 4 year diabetes, PVD, and spironolactone (Table 5). Predictors associated with lower mortality were hypertension, during hospitalization: male gender, CAD, and nitrates, at 1 year and 4 years ACE-I/ARB, beta-blockers, non-dihydropyridines CCBs, and statins, and at 4 years antiacoagulants.

Major recent large HF surveys, like ADHERE and second Euro-HF survey, describe only acutely hospitalized HF patients. As reflected by the long-term mortality data in the survey, the long-term outcome of patients with chronic HF with or without AF is no better than the outcome of patients with acute HF.

Clinical implications

In hospitalized HF patients, AF is associated with increased morbidity and mortality. After adjustment for clinical variables and medications, chronic AF, severe forms of HF, comorbidities, and use of diuretics and digoxin were associated with higher mortality, while intermittent AF is similar to HF with no-AF. Managing AF in conjunction with HF requires greater attention to guidelines and additional trials and surveys to guide and optimize its management.


The study was conducted with the support of The Israel Center for Disease Control (ICDC), The Israeli Medical Association, Teva, Pfizer, MSD, Aventis, Medtronic, Dexxon, Guidant (Levant), Medisson, Neopharm, Novartis, and Schering-Plough companies.

Conflict of interest: none declared.


  • The survey was conducted cooperatively by the Israel Society of Internal Medicine, the Israel Heart Society, the Working Group on Heart Failure, and the Israel Center for Disease Control (ICDC), Ministry of Health, under the auspices of the Israel Medical Association. Data Processing and analysis was conducted by the Israel Society for Prevention of Heart Attacks (ISPHA).


View Abstract