Independent ethics committee and hospital-based institutional review board approvals were obtained, as necessary, for the registry protocol. (See Ethics List S1) The registry is being conducted in accordance with the principles of the Declaration of Helsinki, local regulatory requirements, and the International Conference on Harmonisation–Good Pharmacoepidemiological and Clinical Practice guidelines. All patients provided written informed consent to participate.

The GARFIELD Registry is an ongoing, observational, multicenter, worldwide study of adults (≥18 years) with non-valvular AF diagnosed according to standard local procedures within the past 6 weeks (electrocardiogram confirmation was not mandated) and ≥1 additional factor judged by the clinician to increase the patient’s risk of stroke; such factors were not prespecified in the protocol, nor were they limited to the factors in risk-stratification schemes such as CHADS₂ (cardiac failure, hypertension, age, diabetes, stroke [doubled]) [14] or CHA₂DS₂-VASc (cardiac failure, hypertension, age ≥75 [doubled], diabetes, stroke [doubled]-vascular disease, age 65–74 and sex category [female]) [15].

Enrollment will take place in five independent, sequential cohorts [16]. Patient enrollment into Cohort 1 took place between 21 December 2009 and 26 October 2011. In parallel with prospectively enrolled patients, a validation group (part retrospective and part prospective) was enrolled, comprising patients with established AF (i.e., AF first diagnosed ≥6 months and ≤24 months before enrollment) and ≥1 additional risk factor for stroke, regardless of therapy; in these patients, data were collected retrospectively to the time of first AF diagnosis, and prospectively up to 2 years after diagnosis. The rationale for the inclusion of the retrospective cohort was to evaluate, by comparing retrospective with prospective data, whether initiation of the GARFIELD Registry influenced AF management patterns on a site level; if the effect was zero or minimal, the data from both cohorts would to be combined.

Patients for whom follow-up up to 2 years was unlikely and those with a transient reversible cause of AF were excluded. Data were collected using an electronic case report form (eCRF) [16].

Investigator sites are representative of the distribution of AF treating care settings in each country. Sufficient sites were identified from the spectrum of care settings (office-based practice, hospital departments [neurology, cardiology, geriatrics, internal medicine, emergency], anticoagulant clinics, and general or family practice) to ensure proportional representation in all countries, and the lists and ratios were validated by national coordinators [16]. Sites were selected randomly and recruited following a qualification call. Before site initiation, investigators were required to complete a training program that provided guidance on patient screening, enrollment, and follow-up. Patients were enrolled consecutively, as stipulated in the protocol.

Data collected at baseline included patient and clinical characteristics at diagnosis, medical history (including cardiovascular and bleeding history), care setting at diagnosis, type of AF, date and method of diagnosis, symptoms, antithrombotic treatment at diagnosis (VKAs, factor Xa inhibitors, thrombin inhibitors, and heparins), and reasons for not providing VKAs (when applicable). Ethnicity was classified by the investigator, in agreement with the patient, to investigate ethnic differences in the prevalence of AF [17].

Heart failure, hypertension (blood pressure >140/90 mmHg or treated hypertension), age ≥75 years, diabetes mellitus, and prior stroke or transient ischemic attack were used to calculate, retrospectively, stroke risk according to the CHADS₂ risk index [14]. Additionally, left ventricular ejection fraction <40%, prior thromboembolism, vascular disease (acute coronary syndrome, peripheral artery disease), age 65–74 years, and female gender were used to determine stroke risk using the CHA₂DS₂-VASc score [15].

Registry data were captured in electronic CRFs (designed by Dendrite Clinical Systems Ltd, Henley-on-Thames, UK, who are also responsible for ongoing database programme management). Data collection and entry are managed by Quintiles (Durham, NC, USA), who oversee all operational aspects of the programme, apart from in the UK where these aspects are undertaken by The University of Birmingham Department of Primary Care Clinical Sciences. Submitted data are examined by the coordinating center (Thrombosis Research Institute, London) to ascertain their completeness and accuracy, and data queries are sent to participating sites. Data are extracted for each analysis and analyzed by an independent statistician (PW). Confidentiality and anonymity of all patients enrolled into this registry was maintained at all times.

Continuous variables are expressed as mean±standard deviation (SD). Categorical variables are expressed as frequencies and percentages. Differences between cohorts were tested for statistical significance using the Chi-squared test for categorical variables and the unpaired t-test for continuous variables. Statistical analysis was performed using SAS® software version 9.1.3 (SAS Institute Inc., Cary, NC, USA).

One of the largest multinational studies in this field is the Euro Heart Survey on Atrial Fibrillation, conducted between 2003 and 2004 in more than 5,000 ambulatory and hospitalized patients [20]. The Euro Heart Survey provided a snapshot of the management of AF across 35 European Society of Cardiology member countries and revealed discordance between guidelines and everyday clinical practice. Oral anticoagulation was prescribed in 67% of patients considered eligible for VKA treatment according to guidelines [4], but also in 49% of ineligible patients. The Euro Heart Survey did indicate an increase in use of oral anticoagulant therapy with increasing stroke risk [21], in contrast to the results of a large nationwide retrospective US medical claims database study involving over 171,000 patients with AF (51,907 of whom had newly diagnosed AF), which indicated a low use of warfarin across all risk categories (overall rate 42.6%; 49.5% in patients newly diagnosed with AF) [22]. Both studies were consistent, however, in reporting underuse of oral anticoagulation in patients at elevated risk and overuse in those at low risk. Our present data show no apparent improvement in adherence to evidence-based guidelines [5], [6], [18] for AF in recent years.

Despite mandating the presence of one or more additional stroke risk factors in an effort to exclude patients with lone AF, the GARFIELD population fares as relatively low risk compared with other observational cohorts, and is substantially lower risk than the populations included in randomized trials of novel oral anticoagulant drugs (Table 4) [23], [24], [25], [26]. “Additional risk factors” were investigator defined in order to identify patients that physicians themselves perceived – during the course of their usual practice – to be at risk of stroke. This approach contrasts with other studies that mandated the presence of one or more specific risk factors, such as prior stroke or transient ischemic attack, hypertension, and heart failure, resulting in much higher-risk populations than typically seen in everyday clinical practice. Further, other data sets have been based on patients identified in emergency departments or hospitalized for another condition, who may be at high risk of a poor outcome [27]. In the nationwide US Outcomes Registry for Better Treatment of Atrial Fibrillation (ORBIT-AF) [28], for example, the mean±SD CHADS₂ score was 2.3±1.3 [28], the mean age was 76 years, and approximately 30% of patients had diabetes and 30% heart failure. In contrast, in GARFIELD, the mean CHADS₂ score was 1.9±1.2, mean age was 70 years, and only 22% patients had diabetes and 21% had heart failure. The differences in these two study populations may be due the fact that GARFIELD enrolled patients newly diagnosed with AF whereas ORBIT-AF enrolled patients with prevalent or incident AF.

RealiseAF was a large, international, contemporary, cross-sectional study conducted in 10,523 outpatients in 26 countries between 2009 and 2010 [29], with the aim of investigating the success of rhythm versus rate control, and the impact of control on patients’ clinical symptoms and quality of life. The patients in this registry were slightly younger than those in GARFIELD (67 vs. 70 years), and they had higher rates of heart failure (46% vs. 21%), coronary artery disease (32% vs. 19%) and hypercholesterolemia (46% vs. 39%); these differences in baseline characteristics likely reflect differences in the treatment settings. In RealiseAF, for example, all of the patients were enrolled by cardiologists or internists (hospital- and office-based), whereas in GARFIELD the aim was to reflect the spectrum of care settings in each country; consequently 80% of the patients were enrolled by cardiologists or internists, 18% were enrolled by primary care/general practice physicians, and 2.1% were enrolled by neurologists. Furthermore, 62% of the RealiseAF population was diagnosed with AF >12 months previously, whereas patients in GARFIELD were newly diagnosed, a finding supported by the higher rate of permanent AF in RealiseAF (46% vs. 25%).

Several other large registries in AF have been launched recently, including the international Global Registry on Long-Term Oral Antithrombotic Treatment in Patients with Atrial Fibrillation (GLORIA-AF), and the nationwide US Practice INNovation And Clinical Excellence (PINNACLE-AF) registry – part of the American College of Cardiology National Cardiovascular Data Registry. Physician participation in PINNACLE-AF is voluntary, and the study will report on outcomes in 121,000 patients with AF [30]. Like GARFIELD, PINNACLE-AF will rigorously assess current and evolving practice patterns; it will also help US providers evaluate and improve adherence to guidelines and performance measures through the provision of checklists of guideline-recommended care and provision of quarterly reports. These national US data will be complementary to GARFIELD, which will be conducted in up to 35 countries throughout the world.

The second GARFIELD cohort was initiated in October 2011, and an additional 11 countries (Argentina, Belgium, Chile, Czech Republic, Hungary, India, Russia, Singapore, South Africa, Thailand, Ukraine) joined the registry. Owing to its unique methodology, the GARFIELD Registry will continue to provide prospective and rigorous global and national data on “real-world” risk stratification, AF management, and clinical outcomes in patients newly diagnosed with AF and at risk of stroke.

As an observational study, GARFIELD is subject to certain limitations inherent in all such studies, such as the collection of non-randomized data and missing or incomplete information. Given the wide spectrum of care settings, the presence of missing or incomplete data is not surprising, particularly as site selection was random, encompassed the spectrum of care settings, and included sites with no clinical research experience. The study does, however, provide broader insights into the management of AF patients, with the inclusion of care environments not normally included in such studies. The data collected will provide a “snapshot” of anticoagulant use at the point of evaluation, which may change over time. GARFIELD is currently the largest ongoing international academic registry in patients with non-valvular AF; it represents a novel approach to outcomes research through the recruitment of unselected patients in five consecutive cohorts. In addition, through random selection of nationally representative sites, consecutive patient enrolment, and inclusion of patients perceived by their physicians to be at risk of stroke, the population will reflect those treated in everyday clinical practice. The contemporary data reported here provide a benchmark against which subsequent data, incorporating new therapies for AF, can be compared.

These data highlight a substantial gap between evidence-based risk stratification, management recommendations, and their application in everyday clinical practice. The long-term implications of these management decisions will become apparent in the follow-up data at 1 and 2 years. With the introduction of new anticoagulants for AF, GARFIELD will describe how management strategies, patient outcomes, and use of healthcare resources evolve over time on a global level and in participating countries.