Articles

Efficacy of β blockers in patients with heart failure plus atrial

fibrillation: an individual-patient data meta-analysis
Dipak Kotecha, Jane Holmes, Henry Krum, Douglas G Altman, Luis Manzano, John G F Cleland, Gregory Y H Lip, Andrew J S Coats, Bert Andersson,
Paulus Kirchhof, Thomas G von Lueder, Hans Wedel, Giuseppe Rosano, Marcelo C Shibata, Alan Rigby, Marcus D Flather, on behalf of the
Beta-Blockers in Heart Failure Collaborative Group

Summary
Background Atrial fibrillation and heart failure often coexist, causing substantial cardiovascular morbidity and Published Online
mortality. β blockers are indicated in patients with symptomatic heart failure with reduced ejection fraction; however, September 2, 2014
http://dx.doi.org/10.1016/
the efficacy of these drugs in patients with concomitant atrial fibrillation is uncertain. We therefore meta-analysed S0140-6736(14)61373-8
individual-patient data to assess the efficacy of β blockers in patients with heart failure and sinus rhythm compared
University of Birmingham
with atrial fibrillation. Centre for Cardiovascular
Sciences, Birmingham, UK
Methods We extracted individual-patient data from ten randomised controlled trials of the comparison of β blockers (D Kotecha PhD,
Prof G Y H Lip MD,
versus placebo in heart failure. The presence of sinus rhythm or atrial fibrillation was ascertained from the baseline Prof P Kirchhof MD); Clinical
electrocardiograph. The primary outcome was all-cause mortality. Analysis was by intention to treat. Outcome data Trials and Evaluation Unit,
were meta-analysed with an adjusted Cox proportional hazards regression. The study is registered with Clinicaltrials. Royal Brompton and Harefield
gov, number NCT0083244, and PROSPERO, number CRD42014010012. NHS Trust, London, UK
(D Kotecha); Centre of
Cardiovascular Research and
Findings 18 254 patients were assessed, and of these 13 946 (76%) had sinus rhythm and 3066 (17%) had atrial Education in Therapeutics,
fibrillation at baseline. Crude death rates over a mean follow-up of 1·5 years (SD 1·1) were 16% (2237 of 13 945) in Monash University, Melbourne,
patients with sinus rhythm and 21% (633 of 3064) in patients with atrial fibrillation. β-blocker therapy led to a VIC, Australia (D Kotecha,
Prof H Krum PhD,
significant reduction in all-cause mortality in patients with sinus rhythm (hazard ratio 0·73, 0·67–0·80; p<0·001), T G von Lueder PhD); Centre for
but not in patients with atrial fibrillation (0·97, 0·83–1·14; p=0·73), with a significant p value for interaction of Statistics in Medicine,
baseline rhythm (p=0·002). The lack of efficacy for the primary outcome was noted in all subgroups of atrial University of Oxford, Oxford,
fibrillation, including age, sex, left ventricular ejection fraction, New York Heart Association class, heart rate, and UK (J Holmes PhD,
Prof D G Altman DSc); Internal
baseline medical therapy. Medicine Department, Hospital
Universitario Ramón y Cajal,
Interpretation Based on our findings, β blockers should not be used preferentially over other rate-control medications Universidad de Alcalá, Madrid,
and not regarded as standard therapy to improve prognosis in patients with concomitant heart failure and atrial Spain (Prof L Manzano MD);
National Heart and Lung
fibrillation. Institute, Royal Brompton and
Harefield Hospitals, Imperial
Funding Menarini Farmaceutica Internazionale (administrative support grant). College, London, UK
(Prof J G F Cleland MD); Hull York
Medical School, University of
Introduction optimising the use of these drugs and providing clear Hull, Kingston upon Hull, UK
β-blocker therapy for patients with chronic heart failure guidance about efficacy and safety of treatment.4 (Prof J G F Cleland); City
with reduced ejection fraction was instituted after a Chronic heart failure and atrial fibrillation are Hospital, Sandwell and West
Birmingham NHS Trust,
series of mechanistic studies and large randomised two common illnesses that are associated with substantial
Birmingham, UK (Prof G Y H Lip,
controlled trials showed a significant reduction in the morbidity and risk of death.1,5 Importantly, both are Prof P Kirchhof); Monash
rates of morbidity and mortality. According to both predicted to continue increasing in prevalence,6,7 with the University, Melbourne, VIC,
European and American guidelines, the use of incidence of atrial fibrillation expected to double in the Australia (Prof A J S Coats DSc);
Warwick University, Warwick,
β blockers in symptomatic patients with heart failure next 20 years.8 More than 50% of patients with heart
UK (Prof A J S Coats);
has a class 1A recommendation.1,2 Nonetheless, uptake failure are readmitted to hospital within 6 months9 and Department of Cardiology,
of therapy in clinical practice remains suboptimum, nearly 40% of patients with atrial fibrillation within Sahlgrenska University
with patients at greatest risk of death least likely to 12 months.10 Despite improved medical therapy, heart Hospital, Gothenburg, Sweden
(Prof B Andersson PhD);
receive evidence-based therapy.3 There have also been failure remains an important driver of health-care cost.11
Department of Cardiovascular
concerns about treatment efficacy in some groups, Patients with concomitant atrial fibrillation have even Medicine, University Hospital
notably patients with atrial fibrillation, women, and higher mortality and hospital admission rates, Münster, Münster, Germany
elderly people. Previous analyses in these patient irrespective of which illness arises first.12,13 Additionally, (Prof P Kirchhof); Department
of Cardiology, Oslo University
subsets have lacked statistical power but further large the prevalence of atrial fibrillation is determined by the Hospital, Oslo, Norway
randomised studies are now unlikely because these severity of heart failure, as defined by the New York Heart (T G von Lueder); Nordic School
β blockers are no longer patented. The Beta-Blockers in Association (NYHA) functional class.14 of Public Health, Gothenburg,
Heart Failure Collaborative Group was formed to We assessed the efficacy and safety of β blockers in Sweden (Prof H Wedel PhD);
Department of Medical
provide definitive answers to open questions about patients with heart failure and concomitant atrial Sciences, Instituto di Ricovero
heart failure and β-blocker therapy, with the aim of fibrillation by meta-analysing individual-patient data. e Cura a Carattere Scientifico

www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8 1

 Articles

committee approval was judged to be unnecessary by the

Sinus rhythm Atrial fibrillation
(n=13 946) (n=3066) National Research Ethics Service, London, UK.
We identified published and unpublished randomised
Age (years) 64 (54–71) 69 (60–74)
controlled trials through computer-aided searches of
Women 3498 (25%) 594 (19%)
Medline and Current Contents using MeSH terms
Diabetes mellitus 3221 (25%) 674 (23%)
relating to heart failure with free text search for β blockers
Years with heart failure diagnosis 3·0 (1·0–6·0) 3·0 (1·0–7·0)
since the inception of Medline (about 1960), reference lists
LVEF 0·27 (0·21–0·33) 0·27 (0·22–0·33)
of trials, trials registries, meeting abstracts, review articles,
NYHA class III or IV 7782 (63%) 1901 (72%)
and discussion with group members and pharmaceutical
Systolic blood pressure (mm Hg) 123 (110–140) 127 (113–140)
manufacturers. There were no language restrictions.
Diastolic blood pressure (mm Hg) 78 (70–82) 80 (70–85)
Randomised controlled trials in which mortality was a
Heart rate (bpm) 80 (72–88) 81 (72–92) primary or composite outcome of the comparison of
Body-mass index (kg/m²) 27 (24–31) 27 (25–31) β blockers versus placebo in patients with heart failure
Estimated glomerular filtration rate (mL/min) 64 (52–78) 61 (49–74) were included in the meta-analysis. Only unconfounded
Any diuretic therapy 11 888 (85%) 2866 (93%) head-to-head trials with recruitment of more than
Angiotensin-converting-enzyme inhibitor or 13 213 (95%) 2898 (95%) 300 patients and a planned follow-up of more than
angiotensin-receptor blocker
6 months were eligible to make the project technically
Aldosterone antagonists 1093 (8%) 500 (17%)
feasible and clinically relevant. The search results,
Digoxin 7380 (53%) 2560 (83%)
individual study demographics, and a standardised data
Amiodarone 797 (6%) 319 (10%)
request form to obtain individual-patient data from each
Oral anticoagulation 3652 (26%) 1772 (58%)
trial have been reported previously.4
Data are median (IQR) or percentage, unless otherwise indicated. Within-group characteristics according to treatment Eleven studies accounted for 95·7% of the eligible
allocation are presented in the appendix. Data were missing for diabetes mellitus (844 patients in the sinus rhythm recruited participants.18–28 All included studies had low risk
group and 149 in the atrial fibrillation group), years with heart failure (2464 and 320), LVEF (60 and 16), NYHA class of bias as assessed with the Cochrane Collaborations Risk
(1506 and 431), systolic blood pressure (ten and none), diastolic blood pressure (15 and two), heart rate (seven and
one), body-mass index (128 and 22), and glomerular filtration rate (634 and 101). LVEF=left-ventricular ejection of Bias Tool.29 The CHRISTMAS trial21 had atrial fibrillation
fraction. NYHA=New York Heart Association functional class. as an exclusion criterion, so although the data could not be
used in this meta-analysis they were extracted for future
Table 1: Baseline characteristics of pooled patients with heart failure and sinus rhythm or atrial
fibrillation analysis in other subgroups. Data were extracted from
original source files and additional follow-up outcomes
were available in seven studies.18–20,24,25,27,28 The primary
San Raffaele Pisana, Rome, Patients with atrial fibrillation are often prescribed outcome was all-cause mortality, including an analysis of
Italy (Prof G Rosano PhD); β blockers for both prognostic benefit in heart failure and total mortality where deaths occurred after early study
Division of Cardiology,
University of Alberta,
heart-rate control, although there is little and underpowered termination or following a fixed censor point. The mean
Edmonton, Canada evidence for efficacy in terms of clinical outcomes.15 follow-up until death or censoring was 1·5 years (SD 1·1)
(M C Shibata MD); Academic Regarded as the gold-standard of meta-analysis, individual- for all studies; range 0·9–5·3 years in the individual trials.
Cardiology, Castle Hill Hospital, patient data enable assessment of subgroups, accurate Major secondary outcomes were cardiovascular death, the
Kingston upon Hull, UK
(A Rigby MSc); and Norwich
combination of original data (thereby improving data composite of all-cause mortality and cardiovascular
Medical School, University of quality), full time-to-event analyses, and generation of hospitalisation, and non-fatal stroke. Hospital admission
East Anglia, Norwich, UK hazard ratios (HRs) adjusted for baseline covariates.16 outcomes included the time to hospital admission
(Prof M D Flather MBBS) Analysis of data from more than 18 000 patients randomly (any cause), cardiovascular hospital admission and
Correspondence to: assigned to β blockers or placebo allows a robust and heart-failure-related hospital admission, and the number
Dr Dipak Kotecha, University of
Birmingham Centre for
adequately powered analysis of the clinical benefits of and duration of cardiovascular or heart-failure hospital
Cardiovascular Sciences, Medical β-blocker therapy in patients with heart failure and atrial admissions. An additional post-hoc defined outcome was
School, Vincent Drive, fibrillation compared with those in sinus rhythm. the composite of cardiovascular death and heart-failure-
Edgbaston, Birmingham related hospital admission. Drug safety outcomes were
B15 2TT, UK
d.kotecha@bham.ac.uk
Methods focused on discontinuation of study drug therapy due to
Search strategy and data gathering hypotension, bradycardia, renal impairment, heart failure
The detailed rationale and design have been reported exacerbation, or any adverse event.
previously.4 Briefly, the Beta-Blockers in Heart Failure The diagnosis of sinus rhythm and atrial fibrillation
Collaborative Group is a multinational effort to combine or atrial flutter was based on the baseline
individual-patient data from the major randomised electrocardiograph (ECG). The two atrial arrhythmias
controlled trials of β blockers in patients with heart could only be distinguished in two trials.19,27 Consistent
failure. This report was prepared according to the with clinical expectation, flutter accounted for only 4%
PRISMA guidelines.17 of the combined group. For the purposes of this report,
All the original trials were done under the supervision reference to atrial fibrillation therefore includes atrial
of an appropriate human ethics committee. The current flutter. Incident atrial fibrillation was defined as atrial
analysis involves anonymised data only; hence, ethics fibrillation during follow-up in patients with sinus

2 www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8

 Articles

rhythm at baseline. Follow-up ECGs or adverse event

Deaths in sinus Deaths in atrial
data reflecting new-onset or recurrence of atrial rhythm (n=2237) fibrillation (n=633)
fibrillation were available in all studies, with data
Acute myocardial 126 (6%) 13 (2%)
missing for 862 (6%) of 13 946 individual patients. infarction
Sudden death 927 (41%) 231 (36%)
Statistical analysis Heart failure 539 (24%) 184 (29%)
Data are presented as median and IQR, or percentages. Cardiac (not heart failure) 59 (3%) 11 (2%)
Estimated glomerular filtration rate was calculated with Stroke 43 (2%) 27 (4%)
the Modification of Diet in Renal Disease formula, Vascular (not stroke) 99 (4%) 38 (6%)
normalised to a body surface area of 1·73 m². Three patients
Non-cardiovascular 180 (8%) 45 (7%)
(two with atrial fibrillation and one with sinus rhythm) had
Unknown 264 (12%) 84 (13%)
missing event dates and were excluded from outcome
analyses. Hospital admission was not recorded for the Data are number (%), unless otherwise indicated. *Including deaths reported after
MDC trial and NYHA class was not explicitly obtained in the closure or early termination of the study.

the COPERNICUS study.24,25 Because the amount of Table 2: Causes of deaths* in patients with heart failure and sinus
missing data for other major variables was low, there was rhythm or atrial fibrillation
no need to impute missing values.
All analyses were by intention to treat. The primary and
major secondary outcomes were analysed with a stratified Pooled data for sinus rhythm Pooled data for atrial fibrillation
Cox proportional hazards regression model.30 This is a (n=13 645) (n=3002)
one-stage fixed-effects approach and the assumption is All-cause hospital admissions
that all trials are estimating a common treatment effect Patients with one or more admissions 5150 (38%) 1205 (40%)
with baseline hazards that vary across studies. HRs and Per patient (mean; range) 0·78 (0–26) 0·79 (0–26)
95% CIs are presented with corresponding p values, with Per patient per year 0·86 0·94
adjustment for age, sex, and baseline left-ventricular Cardiovascular hospital admissions
ejection fraction (LVEF), heart rate, and use of Patients with one or more admissions 3508 (26%) 866/3002 (29%)
angiotensin-converting-enzyme inhibitors or angiotensin- Per patient (mean; range) 0·45 (0–16) 0·49 (0–14)
receptor blockers. Kaplan-Meier plots are used to present Per patient per year 0·52 0·60
the data (pooled from all trials). Because the follow-up in
Length of stay (mean, median; range)* 9·7 days, 6 days (1–368) 11·9 days, 8 days (1–179)
individual studies varied, data were censored at 1200 days
Heart-failure hospital admissions
(3·3 years). Heterogeneity for pooled outcomes was
Patients with one or more admission 2241 (16%) 631/3002 (21%)
assessed with the χ² test and I² statistic, with the
Per patient (mean; range) 0·30 (0–16) 0·36 (0–14)
estimate of heterogeneity taken from the inverse-variance
Per patient per year 0·36 0·41
fixed-effects two-stage model.31
Length of stay (mean, median; range)* 9·8 days, 6·5 days (1–148) 12·0 days, 8·0 days (1–179)
Sensitivity analyses included alternative censor
points, separate exclusion of the BEST19 and Data are number (%), unless otherwise indicated. Hospital admission data were not available for the Metoprolol in
CAPRICORN20 studies, additional baseline adjustment, Dilated Cardiomyopathy Trial.25 Data do not account for differences in baseline demographics between groups.
*Average of first five hospital admissions for a cardiovascular or heart-failure cause in patients with at least one
and random-effects modelling.32 Exploratory analyses hospital admission.
included a per-protocol analysis of patients who
remained on study therapy throughout the trial and Table 3: Hospital admissions for patients with heart failure and sinus rhythm or atrial fibrillation
factors associated with incident atrial fibrillation (with
an adjusted logistic regression model because time to Medicine, Oxford, UK, had full access to all the data and
diagnosis of atrial fibrillation was not available). had joint responsibility for the decision to submit for
A two-tailed p value of 0·05 was judged to be significant. publication after discussion with all the named authors.
Analyses were done with Stata (version 11.2) and
R (version 3.0.2). Results
This study is registered with Clinicaltrials.gov, Of 18 254 patients with heart failure, 13 946 (76%) had
number NCT0083244, and with PROSPERO, number sinus rhythm at baseline, 3066 (17%) had atrial
CRD42014010012.33 fibrillation, 1124 (6%) had other rhythms (predominantly
paced rhythm or heart block), and 118 (<1%) had a
Role of the funding source missing or uninterpretable baseline ECG. Minimal
Menarini Farmaceutica Internazionale provided an differences were noted in baseline characteristics
unrestricted research grant for administrative costs and between patients who were randomly assigned to
GlaxoSmithKline provided data extraction support. None β blockers or placebo in any group (appendix). Table 1 See Online for appendix
of the pharmaceutical groups had any role in data shows the baseline characteristics of patients with heart
analysis or manuscript preparation. The steering failure and sinus rhythm or atrial fibrillation. The
committee lead (DK) and the Centre for Statistics in median duration of heart failure before enrolment in the

www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8 3

 Articles

Table 3 shows the pooled data for all-cause,

Number of Sinus rhythm, Atrial fibrillation, pinteraction, atrial
events/ β blockers versus β blockers versus fibrillation versus cardiovascular disease, and heart-failure-related hospital
sample size placebo placebo sinus rhythm admissions. The total number of hospital admissions
Hazard ratio p value Hazard ratio p value and the yearly rate per patient were higher and the mean
(95% CI)* (95% CI)* length of stay was longer in patients with atrial fibrillation
All-cause mortality 2870/17 009 0·73 <0·001 0·97 0·73 0·002 than in those with sinus rhythm (table 3).
(including all (0·67–0·80) (0·83–1·14) A consistent benefit of β blockers versus placebo was
reported deaths) noted for all death or hospital admission outcomes in
All-cause mortality 2577/17 009 0·73 <0·001 0·93 0·43 0·01 patients with sinus rhythm, but the differences were not
(deaths during study) (0·67–0·80) (0·79–1·10)
significant in patients with atrial fibrillation (table 4). The
Cardiovascular 2297/17 009 0·72 <0·001 0·92 0·35 0·02
deaths (including all (0·65–0·79) (0·77–1·10) pinteraction values for treatment efficacy and baseline heart
reported deaths) rhythm were significant for each of these outcomes
First cardiovascular 4374/16 644 0·78 <0·001 0·91 0·15 0·05 (table 4). Including all reported deaths, the adjusted HR
hospital admission (0·73–0·83) (0·79–1·04) was 0·73 for all-cause mortality in patients with sinus
Death or 5670/16 644 0·76 <0·001 0·89 0·06 0·01 rhythm and 0·97 in patients with atrial fibrillation
cardiovascular (0·72–0·81) (0·80–1·01)
(pinteraction=0·002; table 4; figure 1). These results were
hospital admission
similar to those for cardiovascular deaths or when the
First heart-failure- 2872/16644 0·71 <0·001 0·91 0·26 0·005
related hospital (0·65–0·77) (0·78–1·07) analysis was restricted to deaths during the study (table 4).
admission For time-to-first cardiovascular hospital admission, the
Cardiovascular death 4151/16 644 0·70 <0·001 0·90 0·13 0·001 adjusted HR for patients with sinus rhythm was 0·78 and
(during study) or (0·65–0·75) (0·79–1·03) for those with atrial fibrillation was 0·91 (pinteraction=0·05;
heart-failure-related
hospital admission†
table 4; figure 2). Results were similar for heart-failure-
Non-fatal stroke 296/16 644 1·02 0·91 1·04 0·87 0·94
related hospital admission and the composite clinical
(0·78–1·32) (0·66–1·63) outcomes (death or cardiovascular-hospital admission
and cardiovascular-death or heart-failure-related hospital
Metoprolol in Dilated Cardiomyopathy Trial25 did not contribute to analyses of hospital admissions or incident stroke.
admission). β-blocker therapy had no effect on incident
*Hazard ratios derived from the one-stage Cox regression model were stratified by study and adjusted for age, sex, baseline
left ventricular ejection fraction (data missing for 76 patients), baseline heart rate (data missing for eight patients), and use non-fatal stroke in either sinus rhythm or atrial fibrillation
of angiotensin-converting-enzyme inhibitor or angiotensin-receptor blocker. †Outcome was not prespecified. (table 4).
Sensitivity analyses for the primary outcome are
Table 4: Primary and secondary outcomes
presented in the appendix. There were no effects with
additional baseline adjustment, exclusion of specific trials,
studies was 3 years (table 1). Compared with patients in or use of different censor points (appendix). The HRs in
sinus rhythm, those with atrial fibrillation were 5 years the two-stage meta-analysis for all-cause mortality (figure 3)
older, with a higher percentage being men (table 1). were almost identical to those obtained with the one-stage
Small differences were noted in systolic blood pressure approach, using both fixed and random effects modelling
and glomerular filtration rate, but LVEF and heart rate (data not shown). Importantly, we identified no hetero-
were similar (table 1). Patients with atrial fibrillation geneity between the individual studies for all-cause
were more symptomatic (NYHA class III or IV) than mortality in patients with atrial fibrillation (figure 3).
were those with sinus rhythm (table 1). More patients Heterogeneity was significant for sinus rhythm (figure 3).
with atrial fibrillation used diuretics, aldosterone In an exploratory subgroup analysis of the atrial
antagonists, digoxin, or oral anticoagulants; however, fibrillation cohort for all-cause mortality, no significant
95% in both groups were taking an angiotensin- interactions were noted at clinical cutoffs for variables
converting-enzyme inhibitor or angiotensin-receptor including age, sex, LVEF, NYHA class, blood pressure or
blocker at baseline (table 1). heart rate, and medical therapy (figure 4). We also did a
Inclusion of deaths reported after early study per-protocol assessment of all reported deaths in the
termination or study closure resulted in a crude patients with atrial fibrillation. No difference was noted
mortality of 21% in patients with atrial fibrillation in the efficacy of β-blocker therapy in patients with atrial
(633 of 3064) and 16% in those with sinus rhythm fibrillation who remained on therapy (HR 0·84, 95% CI
(2237 of 13 945). Based on the number of deaths during 0·68–1·04), with no significant interaction for
the individual study periods only, crude mortality was discontinuation of study treatment in the atrial fibrillation
18% in patients with atrial fibrillation (556 of 3064) and group (p=0·09).
14% in those with sinus rhythm (2021 of 13 945). The Of 13 084 patients with sinus rhythm at baseline and
most common causes of death in both groups were data for follow-up heart rhythm, 610 (5%) developed
sudden death and heart failure (table 2). Fatal stroke atrial fibrillation. The factors independently associated
was uncommon, but as expected more frequent in with incident atrial fibrillation in an exploratory analysis
patients with atrial fibrillation than in those with sinus were advanced age, male sex, BMI of at least 30 kg/m²,
rhythm (table 2). and NYHA class III or IV at baseline (appendix).

4 www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8

 Articles

A B
100 β-blocker group
Placebo group

90
Survivors (%)

80

70

60

HR 0·73 (95% CI 0·67–0·80); p<0·001 HR 0·97 (95% CI 0·83–1·14); p=0·73

50
0 1 2 3 0 1 2 3
Time (years) Time (years)
Number at risk
β-blocker group 7123 5014 1798 722 1521 997 331 113
Placebo group 6819 4604 1530 561 1542 1020 346 115

Figure 1: Kaplan-Meier survival curve for patients with sinus rhythm (A) and atrial fibrillation (B) in the β-blocker and placebo groups
Data are unadjusted survival curves for all reported deaths. HRs are derived from the adjusted one-stage Cox regression model and stratified by study. HR=hazard ratio.

A B
100 β-blocker group
Placebo group

90
Event-free (%)

80

70

60

HR 0·78 (95% CI 0·73–0·83); p<0·001 HR 0·91 (95% CI 0·79–1·04); p=0·15

50
0 1 2 3 0 1 2 3
Time (years) Time (years)
Number at risk
β-blocker group 6969 4089 1335 482 1485 791 238 71
Placebo group 6665 3540 1013 305 1512 768 230 58

Figure 2: Kaplan-Meier event curve for cardiovascular hospital admission in patients with sinus rhythm (A) and atrial fibrillation (B) in the β-blocker and
placebo groups
Data are unadjusted event curves for cardiovascular hospital admission during the study. HRs are derived from the adjusted one-stage Cox regression model,
stratified by study. HR=hazard ratio.

Allocation to β blockers was associated with a 33%
reduction in the adjusted odds ratio of incident atrial
fibrillation (253 [4%] of 6722 patients randomly assigned
to β blockers developed atrial fibrillation vs 357 [6%] of
6362 allocated to placebo; appendix).
There were minimal differences in the dose of study
drug achieved according to baseline heart rhythm, with
overall 84% of patients achieving maximal study dose in
the placebo group and 73% in the β-blocker group
(appendix). The attained heart rate and change from
baseline heart rate were similar in patients with sinus
rhythm and those with atrial fibrillation, although the
interpretation was confounded by lack of measurements
in patients who died before the interim study visit
(appendix). 15% of patients discontinued the study drug
due to adverse effects in those randomised to β blockers
or placebo. No differences were noted in β-blocker
discontinuation in the sinus rhythm and atrial
fibrillation groups (14% vs 15%; appendix). The
incidences of withdrawal of therapy due to specific
adverse effects were low (eg, hypotension or bradycardia
in 1–2%).
Discussion
Our principal findings in this individual-patient data
analysis are that patients with heart failure and atrial
fibrillation given β blockers had no significant reduction
in all-cause mortality, cardiovascular hospital admission,
or composite clinical outcomes compared with those
receiving placebo. The atrial fibrillation group
comprised 3066 participants with 633 deaths, and
although we cannot exclude limited power, the results

www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8 5

 Articles

A fibrillation in 14–50% of patients with symptomatic heart

HR (95% CI) Weight failure.14 Atrial remodelling often occurs in heart failure
because of sustained increases in pressure, volume, and
MDC 25
0·67 (0·38–1·18) 2·3%
neurohormonal stress, making the development of atrial
CIBIS I22 0·70 (0·47–1·06) 4·4%
fibrillation more likely. Similarly, atrial fibrillation can
US-HF28 0·74 (0·50–1·10) 4·9%
lead to heart failure both as a direct cause (eg, in
ANZ18 0·91 (0·49–1·70) 1·9%
tachycardia-induced cardiomyopathy) and due to loss of
CIBIS II23 0·60 (0·47–0·76) 12·3%
atrioventricular synchrony and impairment in diastolic
MERIT-HF26 0·58 (0·45–0·74) 12·6%
filling. Heart failure with preserved ejection fraction is as
COPERNICUS24 0·53 (0·39–0·72) 8·0%
common as the syndrome with reduced LVEF;2 however,
CAPRICORN20 0·83 (0·64–1·09) 10·3%
the effect of pharmacotherapy on diastolic function in
BEST19 0·91 (0·78–1·06) 31·8%
patients with atrial fibrillation has not been investigated
SENIORS27 0·78 (0·61–1·01) 11·5%
previously. Irrespective of the type of heart failure, the
Overall (I2=56%, p=0·016) 0·74 (0·68–0·81)
combination with atrial fibrillation is known to adversely
0·25 0·5 1·0 2·0 4·0
affect prognosis37 and overall represents a massive
burden on affected patients, health-care systems, and
Favours β blocker Favours placebo societies, including substantial health-care costs.38
B
Hospital admission in heart failure is the greatest cost
HR (95% CI) Weight contributor11 and the results of our analysis suggest that
atrial fibrillation increases both the risk of hospital
MDC 25
1·00 (0·34–2·95) 2·3% admission and the length of stay, reinforcing the
CIBIS I22 1·14 (0·46–2·83) 3·2% importance of finding efficacious therapies in
US-HF28 1·14 (0·56–2·32) 5·3% this population.
ANZ18 0·28 (0·05–1·63) 0·8% A cause for concern is the previous lack of focus on
CIBIS II23 0·98 (0·64–1·51) 14·4% optimal management of the combination of heart failure
MERIT-HF26 1·03 (0·65–1·64) 12·4% and atrial fibrillation. Indeed, all current guideline
COPERNICUS24 0·91 (0·54–1·54) 9·8% recommendations for treatment of heart failure with
CAPRICORN20 0·90 (0·46–1·75) 6·0% reduced LVEF stem from trials predominantly with
BEST19 0·76 (0·54–1·06) 23·4% patients who have sinus rhythm. Guidelines from the
SENIORS27 1·14 (0·81–1·62) 22·4% European Society of Cardiology1 and the American
Overall (I2=0%, p=0·65) 0·95 (0·81–1·12) College of Cardiology Foundation and American Heart
Association2 recommend β-blocker therapy in patients
0·25 0·5 1·0 2·0 4·0
with heart failure and atrial fibrillation, based on the
Favours β blocker Favours placebo efficacy shown in the trials assessed in this report. The
Figure 3: Two-stage adjusted meta-analysis for all-cause mortality in patients with sinus rhythm (A) and
results of our analysis suggest that the substantial benefit
atrial fibrillation (B) in the studies included in the meta-analysis identified in patients with sinus rhythm should not be
Cox regression models, adjusted for age, sex, left ventricular ejection fraction, heart rate, and use of angiotensin- extrapolated to patients with atrial fibrillation. The reason
converting-enzyme inhibitors or angiotensin-receptor blockers, were meta-analysed with a fixed-effects approach. for the lack of efficacy of β blockers in patients with atrial
Analysis included all reported deaths, censored at 1200 days (3·3 years). HR=hazard ratio. ANZ=Australia/New
Zealand Heart Failure Study. BEST=Beta-Blocker Evaluation Survival Trial. CAPRICORN=Carvedilol Post-Infarct
fibrillation might be due to several physiological
Survival Control in Left Ventricular Dysfunction Study. CIBIS I=Cardiac Insufficiency Bisoprolol Study. CIBIS differences.15 Unlike sinus rhythm, slower heart rates are
II=Cardiac Insufficiency Bisoprolol Study II. COPERNICUS=Carvedilol Prospective Randomized Cumulative Survival not associated with improved survival in atrial
Study. MDC=Metoprolol in Dilated Cardiomyopathy Trial. MERIT-HF=Metoprolol CR/XL Randomised Intervention fibrillation,39 although this association remains to be
Trial in Congestive Heart Failure. SENIORS=Study of the Effects of Nebivolol Intervention on Outcomes and
Rehospitalisation in Seniors with Heart Failure Study. US-HF=US Carvedilol Heart Failure Study.
adequately tested in prospective studies. The irregular
rhythm in atrial fibrillation is also associated with a
of this analysis suggest that clinical benefit from detrimental effect on systolic and diastolic cardiac
β blockers is unlikely in patients with combined heart function that is independent of heart rate.40,41 There are
failure and atrial fibrillation. Patients with atrial also structural42 and cellular43 consequences of atrial
fibrillation had higher crude rates of death, more fibrillation that might affect treatment efficacy. These
frequent hospital admission, and longer stay in hospital observations, however, do not fully explain why the
than did those with sinus rhythm. positive effects of β blockers, particularly on myocardial
Heart failure and atrial fibrillation are two common metabolism, do not correspond to prognostic benefit in
disorders that are increasing in prevalence. Although the patients with atrial fibrillation, an anomaly that requires
incidence of heart failure has remained static over the further investigation.
past 25 years,34 the incidence of atrial fibrillation is It is also important to consider the substantial
increasing35 and not simply as a function of the ageing reduction in the rates of hospital admission and death in
population.36 These two disorders often coexist, with patients with sinus rhythm. Rates of β-blocker uptake
observational data suggesting the presence of atrial among patients with heart failure have been consistently

6 www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8

 Articles

suboptimum44,45 and might suggest concern about Deaths/number logHR (95% CI) pinteraction
symptom deterioration after initiation of treatment or Age
the poor generalisability of data from randomised <70 years 279/1617 0·92 (0·73–1·16)
0·643
controlled trials to the real-world setting.46 One of the ≥70 years 337/1446 0·98 (0·79–1·22)
major aims of the Beta-Blockers in Heart Failure Sex
Collaboration was to improve rates of appropriate Female 114/593 0·82 (0·56–1·19)
0·402
β-blocker use by identifying key patient groups that Male 502/2470 0·99 (0·83–1·17)
benefit most from therapy. In this respect, use of Diabetes
β blockers in patients with sinus rhythm is strongly No 413/2242 0·88 (0·72–1·06)
0·305
recommended and further subgroup analyses in relation Yes 161/674 1·09 (0·79–1·49)
NYHA class
to age, sex, and other patient characteristics are planned
I or II 105/734 0·87 (0·59–1·28)
(to explain the remaining heterogeneity). We also noted a 0·513
III or IV 447/1898 0·98 (0·82–1·19)
reduction in incident atrial fibrillation in patients with
Blood pressure
sinus rhythm treated with β blockers. This finding <140/90 mm Hg 566/2634 0·94 (0·80–1·11)
0·867
confirms a previous tabular analysis of β-blocker trials in ≥140/90 mm Hg 50/429 1·04 (0·59–1·83)
heart failure47 and might be an important component of Estimated GFR
the benefit noted in patients with sinus rhythm. ≥60 mL/min 249/1517 0·97 (0·76–1·25)
0·940
Although we found no evidence that β-blocker therapy <60 mL/min 342/1447 0·94 (0·76–1·17)
prevents adverse clinical events in patients with heart BMI
failure and atrial fibrillation, it did seem to be safe with no ≤30 kg/m2 422/2077 0·95 (0·78–1·15)
0·828
increase in mortality or hospital admission rates. This >30 kg/m2 186/964 0·93 (0·69–1·24)
finding should reassure clinicians, particularly for patients LVEF
>0·35 83/510 1·17 (0·76–1·80)
with another indication for β blockers—eg, myocardial 0·403
≤0·35 528/2537 0·92 (0·77–1·09)
infarction or the need for rate control of rapid atrial
Heart rate
fibrillation with ongoing symptoms. However, for the
≤90 bpm 453/2231 1·00 (0·83–1·20)
primary reason of preventing major adverse cardiovascular 0·444
>90 bpm 163/831 0·88 (0·65–1·21)
outcomes in patients with chronic heart failure and Digoxin
reduced LVEF, β blockers do not seem to be effective in No 102/505 1·11 (0·75–1·65)
0·165
patients with atrial fibrillation and should no longer be Yes 514/2558 0·91 (0·76–1·08)
regarded as standard therapy to improve prognosis. Angiotensin-converting-enzyme inhibitor
The strength of our analysis was the use of or angiotensin-receptor blocker
individual-patient data from large, high-quality No 37/167 0·85 (0·43–1·67)
0·580
Yes 579/2896 0·97 (0·82–1·14)
randomised controlled trials, with nearly all available
Oral anticoagulant
randomised data. Our careful and methodical data
No 488/2480 0·93 (0·73–1·18)
extraction from original datasets4 resulted in improved 0·835
Yes 105/500 0·96 (0·78–1·18)
quality of baseline and outcome data across trials. Overall, crude 616/3063 0·97 (0·83–1·14)
Although the process of individual-patient data 0·97 (0·83–1·14)
Overall, adjusted 611/3046
meta-analysis is arduous, there are substantial benefits
including the ability to adjust for covariates and produce –1·0 –0·5 0 0·5 1·0
time-to-event analyses.16 We were also able to include
post-publication data for mortality, explaining the small Favours β blocker Favours placebo
differences from previously published results in the Figure 4: Subgroup analysis of all-cause mortality in patients with atrial fibrillation for the comparison of
component randomised controlled trials. We confirmed β-blocker therapy versus placebo
that our conclusions apply across various subgroups of Dashed line is the overall effect of β blockers versus placebo in patients with atrial fibrillation. HRs and interaction
atrial fibrillation and irrespective of meta-analysis p values were derived from the one-stage Cox regression model, stratified by study and censored at 1200 days
(3·3 years). HR=hazard ratio. NYHA=New York Heart Association. GFR=glomerular filtration rate. BMI=body-mass
methods. To the best of our knowledge, the results of this index. LVEF=left ventricular ejection fraction. bpm=beats per min.
study provide the most powerful analysis available of the
efficacy of β blockers in atrial fibrillation, thereby patients had atrial flutter. The rate of incident atrial
addressing a key clinical question about the management fibrillation was lower than expected from clinical practice
of this important group of patients with heart failure. and might reflect under-reporting, particularly of
As with all meta-analytical techniques, we are limited paroxysmal atrial fibrillation. Our inability to characterise
by the data provided from the individual studies. Although the type, persistence, and duration of atrial fibrillation is
there were missing data for some variables, their effect also a limitation. Although the validity and reproducibility
was minimised by extracting data from source datasets of LVEF measurement has not been adequately shown in
with a published data extraction plan.4 As already noted, patients with atrial fibrillation, we did not see any
we were unable to separate patients with atrial fibrillation difference in the variance of LVEF when comparing
and atrial flutter; however, only a small percentage of patients with sinus rhythm or atrial fibrillation. Heart

www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8 7

 Articles
failure with preserved ejection fraction accounts for over
half of patients with heart failure, but there are few data
from randomised controlled trials comparing β blockers
versus placebo in this group.48 Of the studies included in
our analysis, only SENIORS27 recruited patients with
LVEF of at least 0·50, which accounted for only 1·8% of
the pooled dataset. Hence, we are unable to comment on
the efficacy of β blockers according to the rhythm status
for patients with heart failure and preserved LVEF.
By contrast with the beneficial effects noted for patients
with sinus rhythm, β-blocker therapy has no or minimal
effect on mortality or cardiovascular hospital admission
in patients with heart failure with reduced ejection
fraction and atrial fibrillation. Based on our results, we
dispute the preferential use of β blockers compared with
other rate-control medications and emphasise the need
for further trials in this common and increasingly
important group of patients.
Contributors
DK participated in the design of the study, managed the collaborative
group, and did the data management, statistical analysis, and
manuscript preparation. HK, LM, and MDF participated in the design
and coordination of the study. JH and DGA did the primary statistical
analyses. All authors read, revised, and approved the final manuscript.
Declaration of interests
DK has received grants from Menarini during the conduct of the study;
and honoraria from Menarini (>36 months ago). HK has received
honoraria from GlaxoSmithKline, Roche, and Merck (>36 months ago).
JGFC has received grants and personal fees from Roche and
GlaxoSmithKline outside of the submitted work. AJSC has received grants
and personal fees from Menarini during the conduct of the study, and
personal fees from Lone Star Heart outside of the submitted work. PK has
received grants and personal fees from several research funders including
European Union, British Heart Foundation, German Research Foundation,
Leducq Foundation, German Ministry of Education and Research, and
from medical device and pharmaceutical companies outside the submitted
work; PK also has patents pending for atrial fibrillation therapy and
markers. HW has received consulting fees from AstraZeneca as a member
of the MDC and MERIT-HF steering committees. MDF has received grants
from Menarini during the conduct of the study, and personal fees from
AstraZeneca outside the submitted work. TGvL was supported by a grant
from Southern and Eastern Norway Health Authority during the conduct of
the study. GYHL has served as a consultant for Bayer, Astellas, Merck,
AstraZeneca, Sanofi, Bristol-Myers Squibb–Pfizer, Biotronik, Portola, and
Boehringer Ingelheim, and has been on the speakers’ bureau for Bayer,
Bristol-Myers Squibb–Pfizer, Boehringer Ingelheim, and Sanofi-Aventis.
MCS has received honoraria and speakers’ fees from Forest Laboratories.
The other authors declare no competing interests.
Acknowledgments
The Beta-Blockers in Heart Failure Collaborative Group was investigator
initiated. DK is funded by the National Institute for Health Research
(NIHR) Clinical Lectureship scheme. The BEST trial was principally
sponsored by the US National Heart, Lung and Blood Institute and the
Department of Veterans Affairs Cooperative Studies Program. All other
trials were funded by pharmaceutical companies. We are indebted to the
other members of the Beta-Blockers in Heart Failure Collaborative
Group for database access and extraction support (full list available in
Kotecha and colleagues4) and the late Philip Poole-Wilson (National
Heart and Lung Institute, Imperial College, London, UK). This project
was only possible with the support of the pharmaceutical companies that
have marketed β blockers in heart failure, and the group wishes to
extend their gratitude to AstraZeneca, GlaxoSmithKline, Menarini
Farmaceutica, and Merck Serono. The views expressed in this
publication are those of the authors and not those of the NIHR or the
UK Department of Health.
8 www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8
References
1 McMurray JJ, Adamopoulos S, Anker SD, et al. ESC Guidelines for
the diagnosis and treatment of acute and chronic heart failure 2012.
Eur Heart J 2012; 33: 1787–847.
2 Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA Guideline
for the Management of Heart Failure: A Report of the American
College of Cardiology Foundation/American Heart Association Task
Force on Practice Guidelines. Circulation 2013; 128: e240–319.
3 Lee DS, Tu JV, Juurlink DN, et al. Risk-treatment mismatch in the
pharmacotherapy of heart failure. JAMA 2005; 294: 1240–47.
4 Kotecha D, Manzano L, Altman DG. Individual patient data
meta-analysis of beta-blockers in heart failure: rationale and design.
Syst Rev 2013; 2: 7.
5 Camm AJ, Kirchhof P, Lip GYH, et al. Guidelines for the
management of atrial fibrillation: The Task Force for the Management
of Atrial Fibrillation of the European Society of Cardiology (ESC).
Eur Heart J 2010; 31: 2369–429.
6 Heidenreich PA, Trogdon JG, Khavjou OA, et al. Forecasting the
Future of Cardiovascular Disease in the United States: A Policy
Statement From the American Heart Association. Circulation 2011;
123: 933–44.
7 Krijthe BP, Kunst A, Benjamin EJ, et al. Projections on the number
of individuals with atrial fibrillation in the European Union, from
2000 to 2060. Eur Heart J 2013; 34: 2746751.
8 Colilla S, Crow A, Petkun W, Singer DE, Simon T, Liu X. Estimates
of Current and Future Incidence and Prevalence of Atrial Fibrillation
in the U.S. Adult Population. Am J Cardiol 2013; 112: 1142–47.
9 Desai AS, Stevenson LW. Rehospitalization for heart failure: predict
or prevent? Circulation 2012; 126: 501–06.
10 Amin AN, Jhaveri M, Lin J. Hospital readmissions in US atrial
fibrillation patients: occurrence and costs. Am J Ther 2013;
20: 143–50.
11 Braunschweig F, Cowie MR, Auricchio A. What are the costs of
heart failure? Europace 2011; 13: ii13–ii17.
12 Dries D, Exner D, Gersh B, Domanski M, Waclawiw M, Stevenson L.
Atrial fibrillation is associated with an increased risk for mortality
and heart failure progression in patients with asymptomatic and
symptomatic left ventricular systolic dysfunction: a retrospective
analysis of the SOLVD trials. J Am Coll Cardiol 1998; 32: 695–703.
13 Wang TJ, Larson MG, Levy D, et al. Temporal relations of atrial
fibrillation and congestive heart failure and their joint influence on
mortality: the Framingham Heart Study. Circulation 2003;
107: 2920–25.
14 Maisel WH, Stevenson LW. Atrial fibrillation in heart failure:
epidemiology, pathophysiology, and rationale for therapy.
Am J Cardiol 2003; 91: 2D–8D.
15 Rienstra M, Damman K, Mulder BA, Van Gelder IC, McMurray JJV,
Van Veldhuisen DJ. Beta-blockers and outcome in heart failure and
atrial fibrillation: a meta-analysis. JACC: Heart Failure 2013; 1: 21–28.
16 Stewart LA, Tierney JF. To IPD or not to IPD? Advantages and
disadvantages of systematic reviews using individual patient data.
Eval Health Prof 2002; 25: 76–97.
17 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting
items for systematic reviews and meta-analyses: the PRISMA
statement. BMJ 2009; 339: b2535.
18 Australia/New Zealand Heart Failure Research Collaborative Group.
Randomised, placebo-controlled trial of carvedilol in patients with
congestive heart failure due to ischaemic heart disease. Lancet 1997;
349: 375–80.
19 A trial of the beta-blocker bucindolol in patients with advanced
chronic heart failure. N Engl J Med 2001; 344: 1659–67.
20 The CAPRICORN Investigators. Effect of carvedilol on outcome after
myocardial infarction in patients with left-ventricular dysfunction:
the CAPRICORN randomised trial. Lancet 2001; 357: 1385–90.
21 Cleland JG, Pennell DJ, Ray SG, et al. Myocardial viability as a
determinant of the ejection fraction response to carvedilol in
patients with heart failure (CHRISTMAS trial): randomised
controlled trial. Lancet 2003; 362: 14–21.
22 CIBIS Investigators and Committees. A randomized trial of
beta-blockade in heart failure. The Cardiac Insufficiency Bisoprolol
Study (CIBIS). Circulation 1994; 90: 1765–73.
23 CIBIS-II Investigators and Committees. The Cardiac Insufficiency
Bisoprolol Study II (CIBIS-II): a randomised trial. Lancet 1999;
353: 9–13.

24 Packer M, Coats AJ, Fowler MB, et al. Effect of carvedilol on survival
in severe chronic heart failure. N Engl J Med 2001; 344: 1651–58.
25 Waagstein F, Bristow MR, Swedberg K, et al, for the Metoprolol in
Dilated Cardiomyopathy (MDC) Trial Study Group. Beneficial effects
of metoprolol in idiopathic dilated cardiomyopathy. Lancet 1993;
342: 1441–46.
26 Effect of metoprolol CR/XL in chronic heart failure: Metoprolol
CR/XL Randomised Intervention Trial in Congestive Heart Failure
(MERIT-HF). Lancet 1999; 353: 2001–07.
27 Flather MD, Shibata MC, Coats AJ, et al. Randomized trial to
determine the effect of nebivolol on mortality and cardiovascular
hospital admission in elderly patients with heart failure (SENIORS).
Eur Heart J 2005; 26: 215–25.
28 Packer M, Bristow MR, Cohn JN, et al. The effect of carvedilol on
morbidity and mortality in patients with chronic heart failure. US
Carvedilol Heart Failure Study Group. N Engl J Med 1996;
334: 1349–55.
29 Higgins JPT, Altman DG, Sterne JAC. Chapter 8: assessing risk of
bias in included studies. In: Higgins JPT, Green S, eds. Cochrane
Handbook for systematic reviews of interventions version 5.1.0.
Cochrane Collaboration, 2011.
30 Tudur Smith C, Williamson PR. A comparison of methods for fixed
effects meta-analysis of individual patient data with time to event
outcomes. Clin Trials 2007; 4: 621–30.
31 Koopman L, van der Heijden GJMG, Hoes AW, Grobbee DE,
Rovers MM. Empirical comparison of subgroup effects in
conventional and individual patient data meta-analyses.
Int J Tech Assess Health Care 2008; 24: 358–61.
32 Smith CT, Williamson PR, Marson AG. Investigating heterogeneity
in an individual patient data meta-analysis of time to event outcomes.
Stat Med 2005; 24: 1307–19.
33 Kotecha D, Manzano L, Krum H, Altman DG, Holmes J, Flather M.
The Beta-Blockers in Heart Failure Collaborative Group: individual
patient data meta-analysis. PROSPERO register. 2014: http://www.
crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014010012
(accessed July 27, 2014).
34 Roger VL, Weston SA, Redfield MM, et al. Trends in heart failure
incidence and survival in a community-based population. JAMA
2004; 292: 344–50.
35 Chugh SS, Havmoeller R, Narayanan K, et al. Worldwide
epidemiology of atrial fibrillation: a Global Burden of Disease 2010
Study. Circulation 2014; 129: 837–47.
36 Miyasaka Y, Barnes ME, Gersh BJ, et al. Secular trends in incidence
of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000,
and implications on the projections for future prevalence.
Circulation 2006; 114: 119–25.
www.thelancet.com Published online September 2, 2014 http://dx.doi.org/10.1016/S0140-6736(14)61373-8
Articles
37 Mamas MA, Caldwell JC, Chacko S, Garratt CJ, Fath-Ordoubadi F,
Neyses L. A meta-analysis of the prognostic significance of atrial
fibrillation in chronic heart failure. Eur J Heart Fail 2009; 11: 676–83.
38 Stewart S, Murphy N, Walker A, McGuire A, McMurray JJV. Cost of
an emerging epidemic: an economic analysis of atrial fibrillation in
the UK. Heart 2004; 90: 286–92.
39 Cullington D, Goode KM, Zhang J, Cleland JG, Clark AL. Is heart
rate important for patients with heart failure in atrial fibrillation?
JACC Heart Fail 2014; 2: 213–20.
40 Daoud EG, Weiss R, Bahu M, et al. Effect of an irregular ventricular
rhythm on cardiac output. Am J Cardiol 1996; 78: 1433–36.
41 Clark DM, Plumb VJ, Epstein AE, Kay GN. Hemodynamic effects of
an irregular sequence of ventricular cycle lengths during atrial
fibrillation. J Am Coll Cardiol 1997; 30: 1039–45.
42 Neilan TG, Shah RV, Abbasi SA, et al. The incidence, pattern, and
prognostic value of left ventricular myocardial scar by late
gadolinium enhancement in patients with atrial fibrillation.
J Am Coll Cardiol 2013; 62: 2205–14.
43 Rudolph V, Andrie RP, Rudolph TK, et al. Myeloperoxidase acts as a
profibrotic mediator of atrial fibrillation. Nat Med 2010; 16: 470–74.
44 Komajda M, Follath F, Swedberg K, et al. The EuroHeart Failure
Survey programme—a survey on the quality of care among patients
with heart failure in Europe: part 2: treatment. Eur Heart J 2003;
24: 464–74.
45 Komajda M, Hanon O, Hochadel M, et al. Contemporary
management of octogenarians hospitalized for heart failure in
Europe: Euro Heart Failure Survey II. Eur Heart J 2009; 30: 478–86.
46 Lenzen MJ, Boersma E, Reimer WJ, et al. Under-utilization of
evidence-based drug treatment in patients with heart failure is only
partially explained by dissimilarity to patients enrolled in landmark
trials: a report from the Euro Heart Survey on Heart Failure.
Eur Heart J 2005; 26: 2706–13.
47 Abi Nasr I, Bouzamondo A, Hulot J-S, Dubourg O, Le Heuzey J-Y,
Lechat P. Prevention of atrial fibrillation onset by beta-blocker
treatment in heart failure: a meta-analysis. Eur Heart J 2007;
28: 457–62.
48 Liu F, Chen Y, Feng X, Teng Z, Yuan Y, Bin J. Effects of beta-blockers
on heart failure with preserved ejection fraction: a meta-analysis.
PLoS One 2014; 9: e90555.
9