Two-year outcomes of UK patients newly diagnosed with atrial fibrillation: findings from the prospective observational cohort study GARFIELD-AF

Background The outcomes of patients newly diagnosed with atrial fibrillation (AF) following the introduction of direct-acting oral anticoagulants are not well known. Aim To determine the 2-year outcomes of patients newly diagnosed with AF, and the effectiveness of oral anticoagulants in everyday practice. Design and setting This was a prospective observational cohort study in UK primary care. Method In total, 3574 patients aged ≥18 years with a new AF diagnosis were enrolled. A propensity score was applied using an overlap weighting scheme to obtain unbiased estimates of the treatment effect of anticoagulation versus no anticoagulation on the occurrence of death, non-haemorrhagic stroke/systemic embolism, and major bleeding within 2 years of diagnosis. Results Overall, 65.8% received anticoagulant therapy, 20.8% received an antiplatelet only, and 13.4% received neither. During the study period, the overall incidence rates of all-cause mortality, non-haemorrhagic stroke/systemic embolism, and major bleeding were 4.15 (95% confidence interval [CI] = 3.69 to 4.65), 1.45 (95% CI = 1.19 to 1.77), and 1.21 (95% CI = 0.97 to 1.50) per 100 person–years, respectively. Anticoagulation treatment compared with no anticoagulation treatment was associated with significantly lower all-cause mortality adjusted hazard ratio (aHR) 0.70 (95% CI = 0.53 to 0.93), significantly lower risk of non-haemorrhagic stroke/systemic embolism (aHR 0.39, 95% CI = 0.24 to 0.62), and a non-significant higher risk of major bleeding (aHR 1.31, 95% CI = 0.77 to 2.24). Conclusion The data support a benefit of anticoagulation in reducing stroke and death, without an increased risk of a major bleed in patients with new-onset AF. Anticoagulation treatment in patients at high risk of stroke who are not receiving anticoagulation may further improve outcomes.


INTRODUCTION
Atrial fibrillation (AF) increases the risk of ischaemic stroke fivefold and the risk of death twofold. 1 AF-related strokes are more severe than strokes in people without AF and are more likely to be fatal, lead to longterm disability, extended hospital stays, and increased healthcare costs. 2 Anticoagulation therapy reduces the risk of AF-related stroke (and systemic embolism) and death, with a 68% relative risk reduction for ischaemic stroke and a 25% reduction in the relative mortality. 3 Anticoagulation, however, increases the risk of bleeding with the most serious complication being intracranial haemorrhage, which can be fatal. 4 Anticoagulant drugs recommended by AF guidelines have included vitamin K antagonists (VKAs), usually warfarin, and direct-acting oral anticoagulants (DOACs), namely dabigatran, rivaroxaban, apixaban, and edoxaban. [4][5][6] The latest National Institute for Health and Care Excellence (NICE) AF guideline (2021) recommends anticoagulation for patients with a CHA 2 D 2 -VASc score of ≥2 with a DOAC as a first-choice anticoagulant and VKA as an alternative if DOACs are contraindicated or not tolerated. 7 Since the introduction of DOACs in clinical practice several UK studies have reported the clinical management of AF and changes in prescribing patterns, indicating increases in anticoagulant use overall and good uptake of DOACs. [8][9][10][11][12] However, evidence on outcomes for UK patients newly diagnosed with AF at risk of stroke following the introduction of DOACs is limited.
This study investigates the 2-year event rates for non-haemorrhagic stroke/systemic embolism, all-cause mortality, and major bleeding in UK patients enrolled in the Global Anticoagulant in the FIELD -Atrial Fibrillation registry (GARFIELD-AF).

METHOD Study design
GARFIELD-AF is a prospective, observational, international registry of adults (aged ≥18 years) newly diagnosed with AF. 13 GARFIELD-AF was conducted in 35 countries worldwide, including the Americas, Europe, Africa, Asia-Pacific and the Middle East between 2010 and 2018. The UK study was conducted in accordance with the published study protocols. 13,14 Participants were enrolled into five prospective sequential cohorts between 2011 and 2016. Inclusion criteria comprised

Abstract Background
The outcomes of patients newly diagnosed with atrial fibrillation (AF) following the introduction of direct-acting oral anticoagulants are not well known.

Aim
To determine the 2-year outcomes of patients newly diagnosed with AF, and the effectiveness of oral anticoagulants in everyday practice.

Design and setting
This was a prospective observational cohort study in UK primary care.

Method
In total, 3574 patients aged ≥18 years with a new AF diagnosis were enrolled. A propensity score was applied using an overlap weighting scheme to obtain unbiased estimates of the treatment effect of anticoagulation versus no anticoagulation on the occurrence of death, non-haemorrhagic stroke/systemic embolism, and major bleeding within 2 years of diagnosis.

Conclusion
The data support a benefit of anticoagulation in reducing stroke and death, without an increased risk of a major bleed in patients with new-onset AF. Anticoagulation treatment in patients at high risk of stroke who are not receiving anticoagulation may further improve outcomes. males and females aged ≥18 years with a new diagnosis of non-valvular AF of up to 6 weeks before entry into the registry and an investigator-determined risk factor for stroke, meaning the risk factors for stroke were not pre-specified in the protocol and left to the clinical judgement of the site investigator.
Eligible patients were recruited consecutively at participating sites to prevent selection bias. All participants provided informed consent. Patients were followed up for a minimum of 2 years (study end) or the occurrence of the event of interest, or loss to follow-up, whichever came first. Patients with transient AF, secondary to a reversible cause, and patients for whom follow-up was not possible were excluded.

Setting
The UK specific study recruited from primary care, with 185 sites (GP practices) across the country. Participants were enrolled between June 2011 and August 2016. Data were collected from participants' primary care records at baseline and at 4-month intervals up to 24 months post-diagnosis using an electronic case report file by trained local site staff.

Variables
Data collected at baseline included patient characteristics, medical history, and antithrombotic therapy initiated at diagnosis. The main outcomes were non-haemorrhagic stroke and systemic embolism, major bleeding, and death. Major bleeding was defined as clinically overt bleeding associated with: • a fall in haemoglobin of ≥2 g/dL; or • a transfusion of ≥2 units of packed red blood cells or whole blood; or • a critical site (intracranial, intraspinal, intraocular, pericardial, intra-articular, intramuscular with compartment syndrome, and retroperitoneal); or • a fatal outcome.
The full classification of bleeding events can be found in Supplementary Box S1. Anticoagulant use was measured as anticoagulation prescribed at diagnosis. Anticoagulant use includes patients receiving anticoagulants with an antiplatelet (that is with or without an antiplatelet).

Statistical analysis
Continuous variables are expressed as mean (standard deviation [SD]) and categorical variables as frequency and percentage. Use of antithrombotic therapy at baseline was analysed by CHA 2 DS 2 -VASc and HAS-BLED scores; HAS-BLED was modified to exclude fluctuations in the international normalised ratio as these data were not available. Females with no other risk factors were assigned a CHA 2 DS 2 -VASc score of 0. The occurrence of the major outcomes nonhaemorrhagic stroke/systemic embolism, major bleeding, and mortality are presented using number of events and person-time event rate per 100 person-years and 95% confidence intervals (CIs). Only the first occurrence of each event was taken into account.
A propensity score was applied using an overlap weighting scheme to reduce biased estimates of the treatment effect. Weights were applied to Cox proportional hazards models to estimate the effects of the anticoagulant versus no anticoagulant comparison on the occurrence of death, non-haemorrhagic stroke/systemic embolism, and major bleeding within 2 years of enrolment. This newly developed method of overlap propensity weighting avoids excluding patients (as with matching) and gives the most weight to propensities where there is equipoise. This applied method overlaps weights and optimises the efficiency of comparisons by defining the population with the most overlap in the covariates between treatment groups. This scheme eliminates the potential for outlier weights by avoiding a weight based on a ratio calculation using values bounded by 0 and 1. Thus, when using overlap weights, many of the concerns regarding the assessment and the trimming of the weights are eliminated. 15 Covariates evaluated in the weighting scheme included demographic characteristics (sex, age, and ethnicity), lifestyle factors (current smoking and alcohol consumption), clinical measurements at

How this fits in
Atrial fibrillation (AF) increases the risk of stroke and death; anticoagulation reduces these risks at the cost of an increased risk of bleeding. There has been an increase in the proportion of patients with AF receiving anticoagulants, with patients receiving either vitamin K antagonists or directacting oral anticoagulants. Evidence on outcomes following the increase in the use of anticoagulant therapy in the UK is limited. In this study, the benefit of anticoagulation in this real-world cohort of patients with AF affirms recommendations in AF management guidelines. Addressing gaps in anticoagulation treatment for patients with AF may reduce AF-related stroke and death. diagnosis (body mass index, heart rate, and blood pressure), medical history (congestive heart failure, acute coronary syndromes, vascular disease, carotid occlusive disease, prior stroke/transient ischaemic attack/ systemic embolism, prior bleeding, venous thromboembolism, hypertension, hypercholesterolemia, diabetes, cirrhosis, moderate-to-severe chronic kidney disease, dementia, hyperthyroidism, and hypothyroidism), and baseline antiplatelet use.
Treatment was defined as the first treatment received at the time of enrolment, approximating 'intention to treat'. Patients with missing values were not removed from the study; multiple imputation combining estimates from five imputed datasets was applied for the comparative effectiveness analysis.
Data analysis was performed centrally by study statisticians using SAS (version 9.4).
Anticoagulant therapy was prescribed in 67.1% (n = 2108/3142) of patients with CHA 2 DS 2 -VASc score of ≥2, 56.1% (n = 161/287) of patients with CHA 2 DS 2 -VASc score of 1 (that is males with a score of 1) and 47.2% (n = 42/89) of patients with CHA 2 DS 2 -VASc score of 0 (that is females with a score of 1 and males with a score of 0). Of the participants, 51.6% (n = 247/479) with a HAS-BLED score ≥3 received anticoagulation (data not shown).
The proportion of patients receiving anticoagulant therapy increased progressively with CHA 2 DS 2 -VASc score (Figure 1a). The proportion of patients receiving anticoagulant therapy peaked in patients with a HAS-BLED score of 1, then decreased with increasing HAS-BLED score (Figure 1b). The proportion of patients receiving antiplatelet treatment only according to their CHA 2 DS 2 -VASc score ranged from 16.9% to 22.5% (Figure 1a). The proportion of patients receiving antiplatelet treatment only increased progressively with their HAS-BLED score from 0 in patients with a HAS-BLED score of 0 to 47.7% in patients with a HAS-BLED score from 4 to 6 ( Figure 1b). Most baseline characteristics were similar in patients who received anticoagulant therapy and patients who did not receive anticoagulant therapy; however, a higher proportion of patients receiving anticoagulation had hypertension, diabetes, prior stroke, and venous thromboembolism, whereas a higher proportion of patients with heavy alcohol consumption and history of bleeding did not receive anticoagulants (Table 2). This is of course due to hypertension, diabetes, and prior stroke being risk factors for AF-related stroke. Heavy alcohol consumption and history of bleeding are risk factors for bleeding.
Treatment changes are described in Supplementary Box S2.

DISCUSSION Summary
In this recent cohort of UK patients newly diagnosed with AF, death was the most frequent clinical outcome at 2 years occurring at 2.9 times the rate of nonhaemorrhagic stroke/systemic embolism and 3.4 times the rate of major bleeding. Death remained the most frequent outcome regardless of whether patients were receiving anticoagulation or not. Anticoagulation treatment compared with no anticoagulation treatment was associated with significantly lower all-cause mortality, significantly lower risk of non-haemorrhagic stroke/systemic embolism, and a nonsignificant higher risk of major bleeding.      Strengths and limitations GARFIELD-AF was conducted to highquality standards and data for 20% of the UK cohort were monitored against source documentation. Enrolling patients within 6 weeks of diagnosis ensured the sample included patients who may not survive long after an AF diagnosis by capturing disease burden early on.
The main limitation of the study is that the analysis is intention to treat, based on therapy initiated at diagnosis, and does not account for treatment changes during the 2-year follow-up. Also, the study did not collect data on deprivation and therefore it was not possible to adjust for deprivation in the analysis. Despite having applied appropriate propensity score methodology to balance confounding factors across treatment groups, the authors cannot exclude the presence of unobserved confounding.

Comparison with existing literature
The findings of the present study regarding the benefit of anticoagulation fits with previous meta-analyses of randomised controlled trials in the VKA-only era as well as the DOAC trials. 16,[17][18][19][20] The estimated lower risk of 30% and 61% for all-cause mortality and non-haemorrhagic stroke/systemic embolism in UK patients on anticoagulants in this study are similar to the results from these studies. The estimated 31% higher risk of major bleeding did not reach statistical significance, which might be because of the relatively small number of these events in this cohort of patients.
As in the UK cohort, death was the most frequent outcome in the global cohort, 21 occurring at over three times the rate of nonhaemorrhagic stroke/systemic embolism and almost five times the rate of major  22 Overall though, the findings on the relative effects of anticoagulation in the global study are reproduced in this UK-only cohort. Marginal differences might be because of the wider uncertainty around the obtained estimates in the UK data. In addition, the global analysis on effectiveness of anticoagulants was based on a different group of patients comprising patients enrolled to cohort 3 to 5 and patients with a CHA 2 DS 2 -VASc score of ≥2.

Implications for practice
Findings regarding the benefit of anticoagulation for reduction of all-cause mortality and non-haemorrhagic stroke/ systemic embolism without a significant increase in the risk of bleeding suggests that for most patients the benefits of anticoagulation outweigh the risks, and affirms recommendations relating to anticoagulation in AF management guidelines.
The findings regarding a non-significant increase in the risk of bleeding are reassuring, particularly as 47.0% of patients with a HAS-BLED score of 4-6 received anticoagulation. Nevertheless, this finding should be interpreted with caution, and it must be emphasised that despite the benefit, anticoagulation is recommended for patients at increased risk of stroke with a CHA 2 D 2 -

Data
Requests for patient-level data can be made to the head of statistics at the Thrombosis Research Institute (Karen S Pieper, kpieper@ tri-london.ac.uk). These requests should include a protocol summary and a summary of the statistical analysis plan. The request will be reviewed by the data-sharing committee for approval and next steps will be discussed with the requestor.
VASc score of ≥2 and should be considered for those patients with a CHA 2 D 2 VASc score of 1.
There is scope for improvement in the management of patients newly diagnosed with AF to align with NICE guidelines. 7 Overall, 67.1% of participants at high risk of stroke (CHA 2 D 2 -VASc score ≥2) received anticoagulation at diagnosis. The authors have previously reported a progressive increase in the proportion of patients in the UK cohort receiving anticoagulation, with 75.6% of the final cohort of patients (diagnosed June 2015 to July 2016) receiving anticoagulation. 23 Further increment in the proportion of patients at high risk of stroke receiving anticoagulation will optimise anticoagulation in patients with AF and improve outcomes. On the other hand, 47.2% of patients defined as very low risk in the NICE guidelines (CHA 2 D 2 -VASc score of 0 for males or 1 for females) received anticoagulation, which is contrary to the guidelines.
In addition, the practice of prescribing antiplatelet treatment alone (20.8%) is contrary to AF guidelines; the guidelines indicate patients at risk of stroke must receive anticoagulants or no antithrombotic therapy. The practice of prescribing anticoagulant with antiplatelet treatment (12.5%) is also not recommended in the guidelines.
The older profile of the patient population and the prevalence of comorbidities are likely to be contributory factors to the allcause mortality. The prominence of allcause mortality as an outcome indicates more attention should be given to mortality risk in the management of patients newly diagnosed with AF.

Provenance
Freely submitted; externally peer reviewed.

Competing interests
FD Richard Hobbs has received occasional consult or speaker fees from Pfizer/BMS, Boehringer Ingelheim, and Bayer outside the submitted work. A John Camm has received institutional grants and personal fees from Bayer, Boehringer Ingelheim, Pfizer/BMS, and Daiichi Sankyo. Keith AA Fox has received grants and personal fees from Bayer/Janssen and AstraZeneca, and personal fees from Sanofi/Regeneron and Verseon outside the submitted work. All other authors have declared no competing interests.