Article Text

Original article
Symptoms, functional status and quality of life in patients with controlled and uncontrolled atrial fibrillation: data from the RealiseAF cross-sectional international registry
  1. P Gabriel Steg1,2,
  2. Samir Alam3,
  3. Chern-En Chiang4,
  4. Habib Gamra5,
  5. Marnix Goethals6,
  6. Hiroshi Inoue7,
  7. Laura Krapf1,2,
  8. Thorsten Lewalter8,
  9. Ihsen Merioua9,
  10. Jan Murin10,
  11. Lisa Naditch-Brûlé9,
  12. Piotr Ponikowski11,
  13. Mårten Rosenqvist12,
  14. José Silva-Cardoso13,
  15. Oleg Zharinov14,
  16. Sandrine Brette15,
  17. James O Neill16 on behalf of the RealiseAF investigators*
  1. 1INSERM U-698, Paris, France, Université Paris-Diderot, Paris, France
  2. 2Assistance Publique - Hôpitaux de Paris, Centre Hospitalier Bichat-Claude Bernard, Paris, France
  3. 3Division of Cardiology, American University of Beirut Medical Centre, Beirut, Lebanon
  4. 4General Clinical Research Center, Taipei Veterans General Hospital and National Yang-Ming University, Taipei, Taiwan
  5. 5Cardiology A Department, Cardiothrombosis Research Unit, Fattouma Bourguiba University Hospital, Monastir, Tunisia
  6. 6Department of Cardiology-Electrophysiology, H.-Hartziekenhuis Roeselare-Menen, Roeselare, Belgium
  7. 7Second Department of Internal Medicine, University of Toyama, Toyama, Japan
  8. 8St Marienhospital, Academic Hospital, University of Bonn, Bonn, Germany
  9. 9Global Anti-arrhythmics, Sanofi, Paris, France
  10. 10Department of Internal Medicine and Cardiology, Comenius University, Bratislava, Slovakia
  11. 11Department of Heart Disease, Medical University, Wroclaw, Poland
  12. 12Karolinska Institutet, Södersjukhuset, Stockholm, Sweden
  13. 13Department of Cardiology, Porto Medical School, Hospital São João, Porto, Portugal
  14. 14National Medical Academy of Postgraduate Education, Kiev, Ukraine
  15. 15Statistician, Lincoln, Boulogne-Billancourt, France
  16. 16Connolly/Mater Hospitals/RCSI, Dublin, Ireland
  1. Correspondence to Professeur Ph. Gabriel Steg, INSERM U-698 “Recherche Clinique en Athérothrombose” Université Paris VII - Denis Diderot, Assistance Publique - Hôpitaux de Paris, Centre Hospitalier Bichat-Claude Bernard, 46 rue Henri Huchard, 75877 PARIS Cedex 18, FRANCE; gabriel.steg{at}


Background Rate control and rhythm control are accepted management strategies for atrial fibrillation (AF).

Objective RealiseAF aimed to describe the success of either strategy and the impact of control on symptomatic status of patients with AF.

Methods This international, observational, cross-sectional survey of patients with any history of AF in the previous year, recorded AF characteristics, management and frequency of control (defined as sinus rhythm or AF with resting heart rate ≤80 bpm).

Results Overall, 9665 patients were evaluable for AF control, with 59.0% controlled (sinus rhythm 26.5%, AF ≤80 bpm 32.5%) and 41.0% uncontrolled. Symptom prevalence in the previous week was lower in controlled than uncontrolled AF (55.7% vs 68.4%; p<0.001) and similar for patients in sinus rhythm versus AF ≤80 bpm (54.8% vs 56.4%; p=0.23). At the visit, AF-related functional impairment (EHRA class >I) was seen in 67.4% of patients with controlled AF and 82.1% of patients with uncontrolled AF (p<0.001). Quality-of-life (QoL, measured using EQ-5D) was better for patients with controlled versus uncontrolled AF using the Visual Analogue Scale (mean (SD) score 67.1 (18.4) vs 63.2 (18.9); p<0.001), single index utility score (median 0.78 vs 0.73; p<0.001), or five dimensions of well-being (all p<0.001). Irrespective of AF control, cardiovascular events had led to hospitalisation in the past year in 28.1%.

Conclusion AF control is not optimal. Control appears to be associated with fewer symptoms and better QoL, but even patients with controlled AF have frequent symptoms, functional impairment, altered QoL and cardiovascular events. New treatments are needed to improve control and minimise the functional and QoL burden of AF.

  • Acute coronary syndrome
  • trials
  • randomised
  • reperfusion
  • angina—unstable
  • atrial fibrillation
  • atrial flutter
  • cardiac function
  • interventional cardiology
  • coronary stenting
  • percutaneous valve therapy
  • valvuloplasty
  • cardiac function
  • artificial heart
  • heart transplant
  • heart failure
  • cardiogenic shock
  • sinus node function
  • heart failure
  • systolic heart failure
  • heart failure treatment
  • atrial fibrillation

Statistics from


Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia and its prevalence is expected to increase substantially in future years owing to population ageing.1 2 There are two strategies for the management of AF: restoration and maintenance of sinus rhythm (rhythm-control strategy) and control of the ventricular response rate (rate-control strategy). In addition, prevention of stroke is a cornerstone of management and may require long-term antithrombotic treatment.3 4 Randomised trials have compared rate-control and rhythm-control strategies, but neither has demonstrated better clinical outcomes in patients who had minimally symptomatic or asymptomatic AF.5–10 Therefore, a reasonable therapeutic goal in AF for both strategies is to achieve either rhythm control—that is, restoration of sinus rhythm, or rate control, with slowing of AF heart rate. However, it is not known how often such ‘control’ of AF is achieved in routine clinical practice.

While evidence-based recommendations are available from randomised trials about optimal therapeutic strategies for rhythm management and prevention of stroke in AF, there are limited data on the clinical presentation of AF in outpatients, the degree to which rate/rhythm control is actually achieved and whether this affects clinical symptoms and quality of life (QoL) in patients with AF.11–14

The RealiseAF study was established to provide reliable information about patient characteristics, cardiovascular risk, AF type, symptoms, medical history, impact on QoL and management practices in AF.



RealiseAF was an international, cross-sectional observational survey of >10 000 patients with all types of AF seen at >800 sites in 26 countries.

Patient population

Inclusion criteria were: male/female patients with a history of AF (treated or not and regardless of rhythm at the time of enrolment), with ≥1 AF episode documented by standard ECG/Holter-ECG monitoring in the previous 12 months, or documented current AF, providing written informed consent. Exclusion criteria included cognitive impairment, inability to provide written informed consent, patients with AF within 3 months of cardiac surgery and participation in clinical trials in the AF or antithrombotic field in the previous month.

Site selection

Participating doctors were randomly selected from a global list of cardiologists and internists (hospital- and office-based), in a ratio predetermined to reflect the practice of each country. The list and ratio were validated by national coordinators. To maximise recruitment of consecutive patients, the enrolment period was short (<6 weeks). Each site had an enrolment target ranging from 10 to 30 patients.

Data collection

Data were collected from the participating doctors on patient demographics, risk factors, medical history, symptoms, management strategy, use of antithrombotic agents and antiarrhythmic drugs and prior use of procedures and devices. Symptoms were categorised according to the New York Heart Association (NYHA) and European Heart Rhythm Association (EHRA) (for arrhythmia-related symptoms) classifications by the investigator. A high symptom burden was defined as ≥1 symptom in the previous 7 days. Independently of the investigator assessment of AF by ECG, patients were asked to assess their QoL using the EQ-5D instrument, including the Visual Analogue Scale.15–19 As this was an observational study only, some patients did not have an ECG performed on the day of the visit and were non-evaluable for the primary outcome of the analysis.

Data quality

In each country, 10% of sites were randomly selected for data quality control and the source of key items was verified for all patients at the site.


Primary aims were to determine the frequency of AF ‘control,’ defined as being either in sinus rhythm or having AF with a resting ventricular rate ≤80 beats per minute (bpm) at the time of the visit on resting ECG; and to describe the cardiovascular risk profile.

Secondary objectives were to describe the characteristics and management of AF; and to determine whether practice was in keeping with the practice guidelines available at the time of recruitment.3

Additional goals were to determine the frequency and severity of symptoms in the preceding 12 months and the past 7 days and the severity of symptoms on the day of the visit using the EHRA classification of cardiac symptoms in AF20: the impact of cardiovascular risk factors and comorbidities on management; the cardiovascular events and interventions leading to hospitalisation within the past 12 months; the impact of previous testing on antiarrhythmic treatment changes; and to assess AF-associated health-related QoL.

Statistical analysis

Determination of sample size

Assuming that ‘AF control’ would be achieved in 50% of patients, with ≤10% of patients non-evaluable, enrolment of 450–900 patients for each geographical area (country/region) would provide a precision ranging from 3.5% to 5% in the calculation of the 95% CI. If the rate of control were to be either higher or lower than 50%, precision would be better. To determine the relationship between ‘AF control’ and cardiovascular risk factors or comorbidities, a sample size of 10 000 patients would allow measurement of an OR of 1.2 to be ‘not controlled,’ for a risk factor prevalence ranging from 15% to 85% of patients (with α=5% and β=15%).

Statistical methods

Population characteristics were summarised as mean (SD) for continuous variables and count (percentage) for qualitative variables, unless otherwise indicated. Descriptive analyses were conducted on the total population and according to control of AF. Comparisons between subgroups were made using χ2 tests, Student t test, or Wilcoxon test.

To identify factors associated with control of AF, a multivariable stepwise logistic regression was performed, entering into the model all variables with enough available data to allow meaningful analysis and using a p value of 0.05 to retain the variable in the model. Variables tested were age by class; gender; ethnic origin; patients in-out; country; ≥1 symptom in the past week including the day of the visit; obesity (body mass index ≥30 kg/m2); arterial hypertension; diabetes; type of AF; time since AF diagnosis by class; left ventricular hypertrophy; history of heart failure by NYHA class; history of coronary artery disease; history of peripheral arterial disease; history of cardiovascular disease; history of valvular heart disease; chronic advanced renal failure; therapeutic strategy chosen before this visit; statins in the previous 7 days; ACE inhibitors/angiotensin receptor blockers/aldosterone in the previous 7 days. Discrimination between models was assessed using c-statistics and calibration was assessed using Hosmer–Lemeshow χ2 statistics. The ORs and associated 95% CIs for AF control were determined. Multivariable analysis was adjusted for country. Analyses were performed using the SAS statistical software, V. 9·1 (SAS Institute).

Data reporting

The results of the RealiseAF survey are reported in accordance with the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) statement (



From October 2009 to May 2010, 831 sites screened 11 198 patients and enrolled 10 546 patients, of whom 23 (0.2%) were found to be ineligible (figure 1). The final dataset included 10 523 eligible patients from 26 participating countries located in Western and Eastern Europe, the Middle East, Africa, Asia and Central and South America. Participant numbers from each country are listed online in appendix A.

Figure 1

Patient enrolment in the RealiseAF survey. AF, atrial fibrillation.

Patient characteristics are shown in table 1. Most patients enrolled were outpatients (69.9%). Lone AF (defined as patients aged <60 years with no coronary artery disease/heart failure/valvular heart disease/chronic pulmonary disease/venous thromboembolism/arterial hypertension)3 was present in 5.1% of the cohort. Just under half of the population had permanent AF, 24.8% paroxysmal AF, 22.3% persistent AF and 6.4% of the patients had undefined AF, where the presenting AF was the first episode.

Table 1

Clinical characteristics of patients (AF controlled or not controlled), n (%)*

Risk factors, comorbidities and history of prior events

The prevalence of risk factors and previous cardiovascular and non-cardiovascular disease was high (table 1), with 32.3% of patients with AF having a history of coronary artery disease, 4.6% peripheral arterial disease, 14.1% cerebrovascular disease and 26.7% valvular heart disease. A history of congestive heart failure was present in 45.8%. Overall, 28.1% of all patients (controlled AF 27.4%, uncontrolled AF 29.1%) reported ≥1 hospital admission related to a cardiovascular event (excluding major bleeding) in the previous 12 months, with the most common causes being acute decompensated heart failure (11.3%), acute coronary syndrome (7.8%), stroke (6.1%), transient ischaemic attack (2.8%), pulmonary embolism (1.0%) and systemic embolism (0.8%). Likewise, a history of cardiovascular procedures not specifically related to AF was common in the past 12 months: 12.4% of patients had had a previous intervention, with the most common being percutaneous coronary intervention (6.3%), valvular heart surgery (3.8%) and coronary artery bypass grafting (2.1%). Additionally, 1.9% of the patients had undergone an attempt at catheter ablation before the past 12 months, 0.5% an attempt at surgical ablation and 5.2% had undergone implant of a pacemaker or intracorporeal cardiac defibrillator; these figures during the past 12 months were 1.8%, 0.2% and 2.2%, respectively.

Control of AF

Control of AF was evaluable in 9665 patients (91.8% of the population). On the day of the visit, AF was controlled in 59.0% of the patients (95% CI 58.0% to 60.0%) and not controlled in 41.0% (95% CI 40.0% to 42.0%). Among the patients controlled, 26.5% were in sinus rhythm and 32.5% were in AF with heart rate ≤80 bpm (figure 2). Among patients in sinus rhythm, this had been achieved by spontaneous conversion in 34.3%, drug-induced cardioversion in 51.3%, electrical cardioversion in 13.7% and both drug-induced and electrical cardioversion in 0.6%. The characteristics of patients with uncontrolled AF as a function of the heart rate achieved are described in detail online (appendix B).

Figure 2

Distribution of atrial fibrillation (AF) control on the day of the visit. HR, heart rate; SR, sinus rhythm.

Multivariate logistic regression analysis found that the main predictors of AF control were age ≥75 years, lack of obesity, lack of recent symptoms, paroxysmal/persistent AF (as opposed to permanent), longer time since diagnosis, history of coronary artery disease, valvular heart disease, lack of advanced symptoms of heart failure, lack of hypertension and left ventricular hypertrophy, selection of rhythm-control strategy and use of drugs interfering with the renin–angiotensin–aldosterone system (table 2).

Table 2

Predictors of AF control


On the day of the visit, rate control was the most commonly selected treatment strategy (57.5% of patients overall; controlled AF 48.1%, uncontrolled 71.3%). Overall, β blockers were used in 50.2% of patients (including 17.8% for an indication other than AF), class Ia antiarrhythmic drugs in 0.4%, class Ic in 6.5%, class III in 25.8%, calcium channel blockers in 13.8% and cardiac glycosides in 27.3% (table 3).

Table 3

Treatment on the day of visit, n (%)

Symptoms, functional status and QoL

AF was associated with a high symptom burden. Symptom burden was lower among patients with controlled AF (55.7%) than with uncontrolled AF (68.4%; p<0.001). Among patients with controlled AF, symptom burden was similar for those in sinus rhythm and those in AF ≤80 bpm (54.8% vs 56.4%; p=0.226).

The proportion of patients with AF-related symptoms at the time of the visit (EHRA class >I) among patients with controlled versus uncontrolled AF was 67.4% and 82.1% (p<0.001). Among patients with controlled AF, AF-related symptom frequency was 61.1% among those in sinus rhythm and 72.6% among those in AF ≤80 bpm (p<0.001).

AF control was associated with better QoL reported by patients compared with uncontrolled AF, as measured by the EQ-5D Visual Analogue Scale (mean (SD), 67.1 (18.4) vs 63.2 (18.9); p<0.001), the single index utility score (median score of 0.78 (range −0.59–1.00) vs 0.73 (−0.59–1.00); p<0.001), or the five dimensions of well-being measured by the EQ-5D (figure 3, p<0.001 for each dimension). Of note, QoL was highest for patients in sinus rhythm, intermediate for patients in AF ≤80 bpm and lowest for patients with AF with heart rate >80 bpm (figure 4).

Figure 3

Impact of atrial fibrillation (AF) control on the five dimensions of well-being from the EQ-5D questionnaire.

Figure 4

Mean and SD of EQ-5D scores (for Visual Analogue Scale and for single index utility) by group. AF, atrial fibrillation; HR, heart rate.

Role of heart rate among patients with uncontrolled AF

As depicted online (appendix B), the frequency of previous cardiovascular events and cardiovascular- and AF-related symptoms increased steadily and consistently with heart rate. When a less stringent definition of control was used, defined as sinus rhythm or AF rate ≤110 bpm (instead of ≤80 bpm), the proportion of patients not controlled was 11% (figure 2). As heart rate increased, the demographics changed, comprising a younger population, an increasing proportion of women, higher blood pressure, more patients with paroxysmal/persistent AF and fewer with permanent AF and a shorter time since diagnosis (online appendix B). Importantly, the increase in heart rate was associated with a clear increase in AF-related symptoms (greater prevalence of EHRA >II), more symptoms in the previous 7 days and a greater frequency of cardiovascular events (excluding major bleeding) leading to hospitalisation in the past 12 months (26.8% for patients with AF and heart rate 81–90 bpm vs 31.7% for patients with AF and heart rate >110 bpm).


The main findings from this survey are that for approximately 40% of patients seen in routine practice their AF is not controlled (ie, they are in AF with heart rate >80 bpm) and all patients with AF (controlled or uncontrolled) have a high symptom burden and their QoL is affected. Patients with controlled AF had a lower prevalence of symptoms and better QoL indices than those with uncontrolled AF. Among patients with controlled AF, patients in sinus rhythm had less functional impairment and better QoL than patients with AF <80 bpm. Conversely, among patients with uncontrolled AF, increasing heart rate was associated with more frequent symptoms and functional impairment.

In addition to the symptom burden, patients with either controlled or uncontrolled AF had frequent hospital admissions related to cardiovascular events excluding major bleeding in the previous 12 months. Finally, regardless of AF control, 25–30% of patients with AF experienced cardiovascular events leading to hospitalisation and >10% needed cardiovascular procedures. The enormous burden of AF shown by symptoms, QoL, cardiovascular events, hospitalisations, procedures and their attendant costs, emphasises the need for improved treatments for this increasingly common condition.

Although randomised trials in AF provide information on optimal management strategies, there are limited data on the clinical epidemiology of AF in routine practice, particularly outside North America and Western Europe.21 Epidemiological information often comes from randomised clinical trials, which recruit highly selected populations that may not be representative of routine clinical practice, or patients from specific geographical or practice settings.22 23 Data may be gathered predominantly from hospital sources24 or on the occasion of specific clinical events (such as cardioversion or stroke) and therefore may not reflect the status of stable outpatients with AF. Information sometimes pertains to specific types of AF, such as paroxysmal or persistent AF.25 26 Finally, epidemiology and practice patterns change rapidly. Therefore, a contemporary, international, representative source of information on patients with the whole spectrum of AF is needed.

Several studies have compared rate- and rhythm-control strategies in AF5–9: results have not indicated a clear superiority of rhythm-control compared with rate control strategies, even in patients with heart failure.27 Therefore, both strategies are acceptable for the management of patients with AF. It is not clear, however, how frequently restoration and maintenance of sinus rhythm or adequate control of AF rate are achieved. This study provides information about the frequency of control in a large international sample of patients with AF, representative of routine practice. Possibly even more important is the observation that regardless of whether AF control was achieved, cardiovascular events, symptoms and functional impairment were common and QoL was impaired. A high symptom frequency has already been seen in other registries: in the EuroHeart Survey on AF,24 69% of patients were symptomatic at the time of the survey. Likewise, in a Swiss office-based survey, 73.8% of patients with AF had symptoms,28 a figure consistent with the 73.9% observed in our report. Many previous studies on QoL in AF have come from studies of rate- or rhythm-control interventions and are, therefore, skewed towards selection of highly symptomatic patients.12 In contrast, this cross-sectional study enrolled patients regardless of their status and treatment and therefore has good external validity. However, our results on the impact of AF on QoL, both in patients with controlled and uncontrolled AF, are consistent with previous observations,29 30 even though they stem from a much larger and geographically diverse cohort.

Furthermore, one recent trial (RACE II)31 found that in patients with permanent AF, ‘lenient’ rate control (targeting a resting heart rate <110 bpm) achieved similar clinical outcomes, symptom control and adverse events as a strategy of ‘strict’ control (targeting a heart rate <80 bpm). In this study, functional impairment and prevalence of symptoms increased steadily with increasing heart rate above 80 bpm, with the worse rates being seen in patients with AF >110 bpm.


This report should be interpreted with caution given its observational and cross-sectional nature, as it cannot include patients with fatal complications or in-hospital patients on wards other than cardiology or internal medicine and precludes linking clinical features with outcomes. However, if anything, this would tend to minimise rather than overestimate the clinical impact of AF. Specifically, the data in this observational analysis should not be interpreted to indicate that either strict rate control or rhythm control should be the goals of treatment in AF, given the lack of improvement in clinical outcomes seen in large randomised trials.5–9 27 31 They do, however, indicate that patients with AF often have poor functional status, frequent symptoms and poor QoL and that all of these outcomes tend to be worse among the subset which fails to be ‘controlled’, using a broad definition of control, encompassing both rate and rhythm control. In addition, the adequacy of heart rate control was only assessed at rest, not at exercise.

Despite the wide geographical scope of this study, it does not include large geographical regions such as the USA or Central Africa. There are probably major differences in patient characteristics and management in these countries; indeed, preliminary data from sub-Saharan Africa suggests that the clinical presentation of AF may be more severe than in developed market economies.17 However, RealiseAF does have unprecedented geographical relevance by including many developing low- and middle-income countries.

The EQ-5D tool used to measure QoL has shortcomings as it is a health utility scale rather than a ‘true’ QoL questionnaire, but it is validated and available in the many languages used within RealiseAF. Other validated generic questionnaires which assess QoL (eg, SF-36, living with heart failure questionnaire), are not specific for AF-related symptoms.17 Disease-specific questionnaires (eg, AFSS,32 AF-QoL,33 34 AF6,35 or AFEQT36) may have given a better profile of health-related QoL in this population but most were not yet available at the time of patient enrolment, were not well validated12 or were not available in the many languages from our catchment regions. In addition, the EHRA classification, which we used to describe symptoms, has been specifically designed to describe AF-related symptoms.17


In this large, contemporary, international, observational, cross-sectional survey, AF control was not optimal. Patients with AF experienced frequent symptoms, functional impairment and altered QoL, even when AF was controlled. The burden of disease also included frequent hospital admissions for cardiovascular events and a high requirement for procedures, regardless of AF control. This highlights the need for improved treatments in patients with AF in order to minimise the proportion of patients with uncontrolled AF, reduce symptoms, minimise functional impairment and improve QoL.


The authors acknowledge the assistance of Sally-Anne Mitchell, PhD (PPSI, UK) with editing of the manuscript, references and figures, supported by Sanofi. RealiseAF was sponsored by Sanofi, who helped with data collection. The statistical analysis was performed by a contract organisation (Registrat-MAPI), funded by the sponsor, under the supervision of a statistician from Lincoln, mandated by Sanofi.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • * A full list of investigators can be found on the RealiseAF website at:

  • Presented in part at the European Society of Cardiology Congress, Stockholm, Sweden, September 2010.

  • Funding RealiseAF was sponsored by Sanofi, who provided assistance with data collection. The study was designed, conducted and the manuscript written by an academic international steering committee, with non-voting participation of the sponsor representatives. The statistical analysis was performed by a contract organisation (Registrat-MAPI), funded by the sponsor, under the supervision of a statistician from Lincoln, mandated by Sanofi. The sponsor had no role in submitting the paper for publication.

  • Competing interests All authors have completed the unified competing interest form and make the following declarations: PGS has received research grants from Servier; consultancy fees/honoraria from Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo/Eli Lilly alliance, GlaxoSmithKline, Medtronic, Merck Sharpe and Dohme, Roche, Sanofi, Servier and The Medicines Company; and has equity ownership in Aterovax. C-EC has received honoraria from AstraZeneca, Boehringer Ingelheim, Daiichi-Sankyo, MSD, Novartis, Pfizer and Sanofi. HG has received honoraria from Sanofi and Novartis; and acts as a non-paid advisor to Sanofi and AstraZeneca. MG has received consultancy fees/honoraria from Boehringer-Ingelheim and Sanofi. HI has received consultancy fees/honoraria from Bayer, Boehringer-Ingelheim, Bristol-Myers Squibb, Daiichi-Sankyo and Sanofi. TL has received research grants and consultancy fees/honoraria from Boehringer Ingelheim, Biotronik, Medtronic, Novartis and Sanofi. IM and LN-B are employees of Sanofi. JM has received consultancy fees from Sanofi. PP has received consultancy fees/honoraria from Merck and Sanofi. MR has received research grants from Boehringer Ingelheim, Bristol-Myers Squibb, Medtronic and Sanofi and consultancy fees/honoraria from Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Medtronic, Nycomed and Sanofi. JSC reports board membership for Abbott, AstraZeneca, Pfizer and Sanofi; has received grants from Abbott; consultancy fees/honoraria from Abbott, AstraZeneca, Menarini, Merck, Merck Sharpe and Dohme, Novartis, Pfizer and Sanofi. OZ has received consultancy fees/honoraria from Sanofi. SB is an employee of Lincoln, mandated by Sanofi to oversee and review statistical analyses performed by Registrat-MAPI. JO'N has received consultancy fees/honoraria from Sanofi. SA and LK declare no competing interests.

  • Ethics approval This study was conducted with the approval of the appropriate boards in each country.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement As the corresponding author, I, Gabriel Steg, have the right to grant on behalf of all authors and do grant on behalf of all authors, an exclusive licence (or non-exclusive for UK Crown Employees) on a worldwide basis to the BMJ Publishing Group Ltd and its Licensees to permit this article (if accepted) to be published in Heart and any other BMJPGL products and to exploit all subsidiary rights, as set out in our licence.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.