Supplementary appendix

Transcription

1 Supplementary appendix This appendix formed part of the original submission and has been peer reviewed. We post it as supplied by the authors. Supplement to: Freedman B, Potpara TS, Lip GYH. Stroke prevention in atrial fibrillation. Lancet 2016; 388:

2 Supplemental file: WebOnly Tables and Figures REVIEW STROKE PREVENTION IN ATRIAL FIBRILLATION Ben Freedman MBBS PhD 1,2 Tatjana S Potpara MD, PhD 3 Gregory Y H Lip MD 4,5 Affiliations 1. Heart Research Institute, Charles Perkins Centre, University of Sydney, Sydney NSW Australia 2. Dept of Cardiology and Anzac Research Institute, Concord Hospital, Concord, NSW Australia 3. Cardiology Clinic, Clinical Center of Serbia, School of Medicine, University of Belgrade, Serbia 4. University of Birmingham Institute of Cardiovascular Science, City Hospital, Birmingham, United Kingdom 5. Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Aalborg University, Aalborg, Denmark. Corresponding Author Professor Ben Freedman Heart Research Institute, Level 3E D17, Charles Perkins Centre, University of Sydney, NSW 2006, Australia Ben.freedman@sydney.edu.au Tel: Fax:

3 Web-only Figure 1: Persistence of warfarin and NOACs after initiation Reproduced with permission from Martinez et al, Thomb Haemost Persistence with oral anticoagulant therapy in 27,514 patients commencing oral anticoagulant within 3 months of incident AF, in the United Kingdom. NOAC: Non-vitamin K antagonist oral anticoagulant; VKA: Vitamin K antagonist.

4 Web-only Figure 2: AF-related ischaemic strokes breakdown into proportion with known AF including antithrombotic treatment, and AF unknown at the time of stroke. Figure illustrates the high ischaemic stroke burden from unknown AF(red), and from known but inadequately treated (antiplatelet only or subtherapeutic VKA) or untreated known AF. Combined, these amount to 85-88% of all AF-related ischaemic strokes, and are potentially preventable. In the Adelaide study, 10 all with known AF not on treatment with OAC had CHADS 2 score 2. In both stroke studies, 10

5 12 AF was associated with approximately 1/3 of all ischaemic strokes. INR: International Normalized Ratio. Data in Adelaide study derived from Leyden et al. 2 Data in Swedish study derived from Friberg et al. 3 Proportions in Adelaide study refer to definite AF stoke causation by TOAST criteria in 92 /258 ischaemic strokes. In the Swedish study, timing of AF in relation to stroke (previously known AF vs unknown new AF) was available in 28,240/31,428 strokes: the proportion was assumed to be the same in the remainder. The percentages for antiplatelet only, and for no therapy, are approximate for those with known AF in the Swedish study. Additional information on the proportion of patients not on anticoagulant at time of stroke, but on an antiplatelet drug, received as personal communication from both first authors, Leyden and Friberg. INR: International Normalized Ratio. Implications for stroke prevention. In the Swedish study, looking instead at all 94,083 patients with ischaemic stroke in the registry as the denominator: 33% of strokes were associated with AF ~24% associated with AF known at the time of stroke 4% taking oral anticoagulant ~12% on oral antiplatelet alone (mostly aspirin) ~8% on no antithrombotic drug Thus approximately 20% of all ischaemic strokes could be prevented if patients with known AF were prescribed OAC instead of antiplatelet drugs or no antithrombotic at all ~ 9% associated with AF first diagnosed at the time of stroke To prevent an additional 9% of ischaemic strokes would require detection of unknown AF before stroke by screening, and treatment with OAC

6 Web-only Table 1. The proportion of atrial fibrillation-associated strokes (data from selected population-based studies/registries). Author (Study) Yiin, et al. 4,5 (The Oxford Vascular Study, OXVASC) Jannou, et al. 6 (The Brest Stroke Registry) Year of publication 2015, 2014 Study period Study population AF-associated strokes/total strokes (n) Incident stroke, TIA or peripheral embolic vascular events (general practice, Oxfordshire, UK) Jan Dec 2008 Friberg, et al July July 2010 Palm F, et al. 7 (The Ludwigshafen Stroke Study) Jan Dec 2007 Stroke (Brest, France) Ischemic stroke (the nationwide Swedish Health Registries) Incident stroke or TIA (the city of Ludwigshafen, West Germany) Bjorck, et al Incident stroke (the Vastra Gotaland Region, Sweden) Leyden, et al Acute stroke (Adelaide, (The Adelaide Stroke Incidence Australia) Study) Hannon, et al. 9 (The North Dublin Population Stroke Study) Dec Nov 2006 Stroke or TIA (the North Dublin City, Ireland) 383/1248 (ischemic) 264/835 (all) 231/733 (ischemic) Proportion of AFrelated strokes (%) Previously unknown AF, first-diagnosed at the stroke presentation, n (%) 30.7% 109/383 (28.5%) 31.7% 31.5% 48/231 (20.8%) 31428/ % 7623/28420 (26.8%) (partial data) 187/626 all 29.9% 57/187 (30.5%) 187/ % cardioembolic 1299/4565 * 28.5% * NR 1733/ /318 (all) 115/258 (ischemic) 92/109 (cardioembolic) 177/568 (incident) 139/485 (firstever) 38.0% 36.1% 44.6% 84.4% 31.2% 28.7% 37/115 (32.2%) 81/177 (45.8%) Bejot, et al Ischemic stroke (Dijon, 572/ % NR France) Marini, et al. 11 (the L Aquilla Stroke Registry) Incident ischemic stroke (L Aquilla, Italy) 869/ % NR * AF listed among hospital discharge diagnoses. AF ever diagnosed, as evident from the previous medical records or hospital discharge diagnoses. AF: Atrial Fibrillation; TIA: Transient ischemic attack; NR: Not reported.

7 Web-only Table 2. The pre-stroke use of oral anticoagulant therapy in patients with known atrial fibrillation and a high risk of stroke (a CHADS 2 of 2). Author (Study) Year of publicatio n 2015, 2014 Pre-stroke antithrombotic therapy in high-risk patients with known AF (CHADS2 2) No therapy Antiplatelet drug OAC Therapeutic INR N (%) N (%) N (%) N (%) Yiin, et al. 4,5 (The Oxford Vascular Study, OXVASC) 86 (25.6%) 194 (57.7%) 56 (16.6%) 22 (6.5% of total, or 39.3% of 56) Jannou, et al (19.1%) 69 (45.4%) 54 (35.5%) 18 (34.6%) (The Brest Stroke Registry) Friberg, et al NR NR 16.0% and 22.0% within the past NR 3 and 6 months, respectively * Palm F, et al (26.4%) 40 (37.7%) 38 (35.8%) 16 (42.1%)] (The Ludwigshafen Stroke Study) Bjorck, et al NR NR 46% (primary prevention) NR 29.9% (secondary prevention) Leyden, et al NR NR 25/57 (43.9%) 14/25 (56.0%) (The Adelaide Stroke Incidence Study) Hannon, et al /87 (17.2%) 48/87 (55.2%) 24/87 (27.6%) NR (The North Dublin Population Stroke Study) 32.0% (secondary prevention) Bejot, et al NR A 50% increase in A 2-fold increase in the use of NR the use of OAC antiplatelet drugs Marini, et al. 11 (the L Aquilla Stroke Registry) % 74.2% 11.3% NR * In general, the use of OAC in patients with known AF and a CHADS2 of 2 was inversely related to the CHADS2 score value (that is, stroke risk), age and the HAS-BLED value (that is, bleeding risk); Non-anticoagulated patients n=32, 56.1% (no therapy or antiplatelet drug not specified). OAC: Oral anticoagulant therapy; INR: International Normalized Ratio; NR: Not reported.

8 Web-only Table 3A-3D. Stroke risk assessment scores in atrial fibrillation, with event rates per 100 person-years. 3A) The CHA 2DS 2-VASc stroke risk score 12 Clinical risk factor Points (max. 9) Score value Stroke rate C Congestive heart failure/lv systolic dysfunction a H Hypertension A 2 Age 75 years D Diabetes mellitus S 2 Stroke/TIA V Vascular disease b A Age years Sc Sex category female gender a Moderate or severe systolic LV dysfunction, defined arbitrarily as LVEF of 40%, or recent decompensated HF requiring hospitalization; b Prior myocardial infarction, peripheral artery disease, aortic plaque. 3B) The CHADS 2 stroke risk score 13 Clinical risk factor Points (max. 6) Score value Stroke rate C Congestive heart failure H Hypertension A Age > 75 years D Diabetes mellitus S 2 Stroke/TIA TIA: Transient ischemic attack.

9 3C) The ATRIA stroke risk score 14 Risk factor Points without Points with Score Stroke rate prior stroke prior stroke value Age < Female sex Diabetes mellitus Congestive heart failure Hypertension Proteinuria egfr<45ml/min or ESRD ESRD: End-stage renal disease. 3D) The R 2CHADS 2 stroke risk score 15 Risk factor Points (max. 8) Score value Renal dysfunction (i.e., CrCl<60mL/min Congestive heart failure (recent) Hypertension Age 75 years Diabetes mellitus Stroke/TIA CrCl: Creatinine clearance; TIA: Transient ischemic attack. Stroke rate (at 3 years)

10 Web-only Table 4A-4E. Bleeding risk assessment scores in atrial fibrillation (4A-4C); The SAMe-TT2R2 score (4D). 1 or 2 Reduced haemoglobin, 4A) The HAS-BLED bleeding risk score 16 4B) The ORBIT bleeding risk score 17 Clinical parameter Points (max. 9) Clinical parameter Points (max. 7) H Hypertension (SBP>160mmHg) 1 Older age (>74 years) 1 A Abnormal renal and liver function 2 (one point each) a Haematocrit/Anemia e S Stroke 1 Bleeding history 2 B Bleeding b 1 Insufficient kidney function f 1 L Labile INRs c 1 Treatment with antiplatelet drug(s) 1 E Elderly (e.g., age > 65 years, frail 1 condition) D Drugs or alcohol concomitant use (one point each) d 1 or 2 a Abnormal renal function classified as the presence of long-term dialysis, renal transplant, or serum creatinine of 2.3mg/dL (200mmol/L). Abnormal liver function defined as chronic hepatic disease (e.g., cirrhosis) or biochemical evidence of significant hepatic derangement (e.g., bilirubin 2-3x the upper limit of normal, in association with aspartate aminotransferase, alanine aminotransferase, or alkaline phosphatase 3x the upper limit normal); b History of bleeding or predisposition (e.g., anemia); c Labile INR (i.e., TTR<60%, applies only if the patient is taking warfarin); d Concomitant antiplatelet drugs or nonsteroidal anti-inflammatory drugs, or alcohol excess. e Haemoglobin <13mg/dL (males) or <12mg/dL (females), or Haematocrit <40% (males) or <36% (females); f <60mL/min/1.73m 2. 4C) The HEMORR2HAGE bleeding risk score 18 4D) The ATRIA bleeding risk score 19 Parameter Points (max.12) Parameter Points (max.10) H Hepatic or renal disease 1 Anemia 3 E Ethanol abuse 1 Severe renal disease 3 (egfr<30ml/min or on dialysis) M Malignancy 1 Age 75 years 2 O Older age ( 75 years) 1 Prior haemorrhage 1 R Reduced platelet count (or function) 1 Diagnosed hypertension 1 R 2 Re-bleeding risk 2 H Hypertension (uncontrolled) 1 A Anemia 1 G Genetic factors a 1 E Excessive fall risk 1 S Stroke 1 a Cytochrome P 4502C9 single nucleotide polymorphism; egfr: Estimated glomerular filtration rate.

11 4E) The SAMe-TT2R2 score Component Points S Sex (female) 1 A Age (<60 years) 1 Me Medical history* 1 T Treatment (interacting drugs, e.g., amiodarone) 1 T2 Tobacco use (within 2 years) 2 R2 Race (non-caucasian) 2 Max. points 8 *More than two of the following: hypertension, diabetes mellitus, coronary artery disease/myocardial infarction, peripheral arterial disease, congestive heart failure, previous stroke, pulmonary disease, and hepatic or renal disease.

12 Web-onlyTable 5. Pharmacological characteristics of VKAs and NOACs (A) and main results of the landmark NOAC trials (B) A) Pharmacological characteristics of VKAs and NOACs 20 Mechanism of action Warfarin Dabigatran Rivaroxaban Apixaban Edoxaban Direct thrombin inhibitor Direct activated Direct activated factor X factor X inhibitor inhibitor Vitamin K antagonist (inhibition of vitamin K epoxide reductase) Direct activated factor X inhibitor Prodrug No Yes No No No Bioavailability 100% 3-7% 66% (fasting), 50% 62% 100% with food Absorption with food No effect No effect Increased by 39% No effect Increased by 6-22%, but minimal effect on exposure Intake with food Not required Not required Recommended Not required Not required Elimination Renal 80% Renal 80% Renal 35% Renal 27% Renal 50% Liver metabolism Half-life Plasma peak level after ingestion Plasma trough level after ingestion Dosing Yes, cytochrome P450 involved No Yes, cytochrome P3A4 involved (moderate contribution) 12-17h 5-9h (young), 11-13h (elderly Yes, cytochrome P3A4 involved (moderate contribution) Minimal (<4% of elimination) R-warfarin: 35-38h; 12h 10-14h S-warfarin: 24-33h 2-8h 2h 2-4h 1-4h 1-2h 48-72h 12h 24h 12h 24h INR-guided dose adjustments Twice daily, 150mg or 110mg (consider using 110mg in patients >75-80 years old, particularly in the presence of additional bleeding risk factors) Once daily, 20mg (15mg if CrCl<50mL/min) Twice daily, 5mg (2.5mg if 2 out of 3 following criteria are fulfilled: age 80years, body weight 60kg, serum creatinine 1.5mg/dL or 133mmol/L) Once daily, 30mg or 60mg (both doses halved if any of the following criteria are fulfilled: a CrCl of 30-50mL/min, body weight 60kg or the concomitant use of verapamil, quinidine or dronedarone)

13 Not recommended in CKD if --- CrCl<30mL/min (75mg twice daily approved in US if CrCl 15-30mL/min) CrCl<15mL/min CrCl<15mL/min CrCl<15mL/min Food interactions +++ No No No No Drug-drug interactions +++ +* Dronedarone, Rifampicinavoid concomitant use; Verapamil-use lower dose of Dabigatran and use concomitantly B) Main results of the landmark NOAC trials in non-valvular AF 21 RE-LY trial 22 +* Rifampicin-avoid concomitant use +* Rifampicin-avoid concomitant use +* Dronedarone, Clarithromycin, Erythromycin-reduce Edoxaban dose by 50% Stroke or SE Major bleeding GI bleeding ICH All-cause mortality Dabigatran 150mg 0.66 ( ) 1.11%/year 0.93 ( ) 3.11%/year 1.50 ( ) 1.51%/year 0.40 ( ) 0.30%/year 0.88 ( ) 3.64%/year Dabigatran 110mg 0.91 ( ) 1.53%/year 0.80 ( ) 2.71%/year 1.10 ( ) 1.12%/year 0.31 ( ) 0.23%/year 0.91 ( ) 3.75%/year Warfarin 1.69%/year 3.36%/year 1.02%/year 0.74%/year 4.13%/year ROCKET-AF trial 23 ARISTOTLE trial 25 Rivaroxaban 0.88 ( ) 1.04 ( ) 1.61 ( ) 0.67 ( ) 0.92 ( ) %/year 3.6%/year 2.0%/year 0.5%/year 4.5%/year Warfarin 2.4%/year 3.4%/year 1.24&/year 0.7%/year 4.9%/year ENGAGE AF trial 26 Apixaban 0.79 ( ) 0.69 ( ) 0.89 ( ) 0.42 ( ) 0.89 ( ) 1.27%/year 2.13%/year 0.76%/year 0.33%/year 3.59%/year Warfarin 1.60%/year 3.09%/year 0.86%/year 0.80%/year 3.94%/year Edoxaban 60mg 0.87 ( ) 1.57%/year 0.80 ( ) 2.75%/year 1.23( ) 1.51%/year 0.47 ( ) 0.39%/year 0.92 ( ) 3.99%/year Edoxaban 30mg 1.13 ( ) 2.04%/year 0.47 ( ) ( ) 0.67%/year 0.30 ( ) 0.26%/year 0.87 ( ) 3.80%/year Warfarin 1.80%/year 3.43%/year 1.23%/year 0.85%/year 4.35%/year AVERROES trial 27 Apixaban 0.45 ( ) 1.6%/year 1.13 ( ) 1.4%/year 0.86 ( ) 0.4%/year 0.85 ( ) 0.4%/year 0.79 ( ) 3.5%/year

14 Aspirin 3.7%/year 1.2%/year 0.4%/year 0.4%/year 4.4%/year *Strong inducers or inhibitors of p-glycoprotein or cytochrome P3A4. VKA: Vitamin K antagonist; NOAC: Non-vitamin K oral anticoagulant; INR: International Normalized Ratio; CrCl: Creatinine clearance; AF: Atrial fibrillation.

15 Web-only Table 6. Observational real-world data on the use of oral anticoagulant therapy in patients with non-valvular atrial fibrillation Real-world data on the use of Dabigatran in patients with non-valvular AF Author Study period AF cohort type; Dataset(s) Dabigatran bid dose Main findings Subgroup analyses (Dabigatran compared to warfarin) Vaughan Sarrazin MS, et al Larsen T, et al Graham DJ, et al Hernandez I, et al Lauffenburger JC, et al Abraham NS, et al Avgil-Tsadok M, et al (2011) VKA experienced; Veterans Affairs administrative data OAC naïve/ VKA experienced; Danish health registries OAC naive, Age 65y; Medicare OAC naïve; Medicare (a 5% random sample) OAC naïve; Truven Health MarketScan Commercial Claims; Encounters and Medicare Supplement Databases OAC naïve; Optum Labs Data Warenhouse OAC naïve/vka experienced; Quebec hospital discharge database 150mg 110mg or 150mg 150mg or 75mg (16% of the cohort) NR 150mg or 75mg 150mg 110mg or 150mg Increased rates of major bleeding and GI bleeding, significantly lower ICH rates and similar mortality with Dabigatran vs. warfarin Lower rates of major bleeding and ICH, and similar rates of GI bleeding with Dabigatran as compared to warfarin Lower ischemic and hemorrhagic stroke rates, similar major bleeding rates (with higher rates of GI bleeding and lower rates of ICH) and lower mortality with Dabigatran vs. warfarin Higher major bleeding and GI bleeding rates, but lower rates of ICH with Dabigatran as compared to warfarin Dabigatran was associated with similar rates of ischemic stroke, lower rates of hemorrhagic stroke and increased rates of GI bleeding compared with warfarin. Myocardial infarction was less common in patients taking Dabigatran than in those taking warfarin Similar rates of any stroke and GI bleeding in patients taking Dabigatran and those taking warfarin Similar rates of any stroke or major bleeding, increased risk of GI bleeding and lower ICH rate in patients taking Dabigatran None -The highest bleeding rates were noted in VKA naive patients starting warfarin; -VKA experienced patients switched to Dabigatran 110mg had higher rates of major bleeding compared with those who remained on warfarin. -Increased GI bleeding risk in female AF patients 75years old and male patients 85years old; -Comparable bleeding rates in younger patients. -Increased risk of bleeding especially in blacks (HR 2.12; 95%CI, ) and CKD patients (HR 2.07; 95%CI, ). -Male patients, those 55-64years old and patients at low-to-intermediate bleeding risk may have lower risk of harm with Dabigatran. -Renal impairment was associated with significantly increased the risk of bleeding with Dabigatran. -Similar risk of both upper and lower GI bleeding with Dabigatran; -Higher risk of GI bleeding in Dabigatran patients >75years old compared to those on warfarin. -Among patients taking Dabigatran, those aged 75 years were mostly prescribed the110mg dose, whilst patients <75 years old more often received the 150mg dose;

16 Seeger J, et al Villines T, et al and Quebec physician and prescription claims database OAC naïve; MarketScan, Truven and Cliniformatic, Optum OAC naïve; The US Department of Defense 150mg or 75mg (4% of the cohort) 150mg or 75mg (12% of the cohort) Real-world data on the use of Rivaroxaban in patients with non-valvular AF Study (or Author) The XANTUS study 37 Study period (prospectively) AF cohort 6784 consecutive AF patients, newly started on Rivaroxaban Geographic region Europe, Israel, Canada Dabigatran use was associated with similar rates of all stroke and ischemic stroke, lower rates of hemorrhagic stroke and major bleeding, similar rates of GI bleeding, lower rates of ICH and similar rates of myocardial infarction compared with warfarin Lower rates of any stroke, similar rates of ischemic stroke, lower rates of hemorrhagic stroke, similar rates of major or GI bleeding and lower rates of ICH, myocardial infarction or death in patients on Dabigatran compared to those taking warfarin Baseline characteristics Mean age 71.5±10.0 years (37.2% >75 years old); Female 41.0%; Mean CHADS 2 score: 2.0±1.3; Mean CHA 2DS 2-VASc score: 3.4±1.7; First diagnosed AF: 18.5%; Paroxysmal AF: 40.6%; Persistent AF: 13.6%; Permanent AF: 27.0%; VKA naïve: 54.5%; Rivaroxaban dosing: -20mg once daily (78.7%), -15mg once daily (20.8%) -Younger patients on Dabigatran experienced less ICH and GI bleeding than patients taking warfarin; -Elderly patients on Dabigatran had less ICH and more GI bleeding than patients taking warfarin. -Decreased upper GI bleeding risk; -Similar lower GI bleeding risk; -Greater stroke reduction in patients 75years old, and -Greater major bleeding reduction if <55years old or at low risk of stroke. -Similar upper GI bleeding risk; -Significantly increased lower GI bleeding risk: A separate analysis including only patients taking Dabigatran 150mg bid: HR 0.73 (95%CI, ) for stroke HR 0.82 (95%CI, ) for major bleeding Main findings -Most patients (96.1%) did not experience any of the study outcomes (treatment-emergent major bleeding, stroke/se or all-cause death); -The incidence of major bleeding: 2.1%/year; -ICH: 0.4%/year; fatal bleeding: 0.7%/year; major GI bleeding: 0.9%/year; stroke:0.7%/year; SE: 0.1%/year; hemorrhagic stroke:0.2%/year; ischemic stroke: 0.5%/year; all-cause death: 1.9%/year; -The incidence of thromboembolic and bleeding events increased with age, with higher risk scores and with declining renal function; -The 15mg dose was mostly used in patients with a CrCl of <50mL/min, but 36% of such patients were given the 20mg dose; -Study outcomes were numerically higher in the 15mg dose group;

17 The RELIEF study (retrospectively) A propensity score-matched AF patients newly started on Rivaroxaban (n=1039) or VKA (n=1039) Germany (outpatients from the Primary Care Physician panel) Mean age: 74.0±10.7 years and 74.4±9.9 years; Female 48.2% (both arms); Mean CHADS 2 score: 1.7±1.0 and 1.8±0.9; Mean CHA 2DS 2-VASc score: 3.9±1.5 and 3.9±4; Mean ATRIA bleeding score: 2.1±1.7 and 2.1±1.7, in the Rivaroxaban and warfarin arm, respectively -The rate of Rivaroxaban discontinuation was low (20.1%), and 75.1% of patients reported satisfaction with their treatment. -The incidence of composite end-point of thromboembolism, bleeding or mortality was significantly lower in the Rivaroxaban arm compared to warfarin (1.97%/year vs. 3.68%/year), HR 0.54 (95%CI, ). Tamayo S, et al (a retrospective pharmacovigilance study) AF patients taking Rivaroxaban US Department of Defence electronic health care records Mixed real-world observational studies addressing the use of NOACs in patients with AF Study (or Author) Olesen JB, et al. 40 Gorst- Rasmussen A, et al. 41 Study period AF cohort OAC naive patients: N=9902 on warfarin; N=7128 on dabigatran; N=1303 on rivaroxaban; N=278 on apixaban OAC new users: N=776 on rivaroxaban 15mg; Patients with major bleeding were older (78.4±7.7 years vs. 75.7±9.7 years), with higher mean CHADS 2 and CHA 2DS 2-VASc score values (3.0±1.2 vs. 2.2±1.3 and 4.8±1.5 vs. 3.7±1.7, respectively) and more co-morbidity Geographic region Denmark (Danish nationwide administrative registries) Denmark (Danish nationwide Main findings -The incidence of major bleeding: 2.86%/year (95%CI, ); -Of all major bleeding events, 88.5% were GI bleedings and 7.5% were ICHs; -Of patients suffering a major bleeding event, 63.2% were taking Rivaroxaban 20mg, 32.2% were taking the 15mg dose, and 4.6% of patients were given the 10mg dose; -The incidence of fatal bleeding was 0.08%/year (95%CI, ). -Gradual decrease in the use of warfarin over the study period, with increasing uptake of NOACs; -Patients initiated on a NOAC were more likely to be older and have certain co-morbidities (e.g., stroke, prior bleeding and alcohol abuse); -A history of chronic kidney disease was the factor most strongly associated with warfarin use; -Patients initiated on warfarin vs. those initiated on a NOAC had comparable predicted stroke or bleeding risk (as measured by the CHA2DS2-VASc and HAS-BLED score, respectively) -Rivaroxaban users were older and with more co-morbidities as compared to dabigatran and warfarin users;

18 N=1629 on rivaroxaban 20mg; N=3588 on dabigatran 110mg; N=5320 on dabigatran 150mg; N=11045 on warfarin administrative registries) - After propensity adjustment, rivaroxaban had lower stroke rates vs. warfarin (HR 0.46, 95%CI: for rivaroxaban 15mg; HR: 0.72, 95%CI: for rivaroxaban 20mg, and similar stroke rates vs. dabigatran; -The bleeding rate was similar to warfarin and moderately higher vs. dabigatran (rivaroxaban 15 vs. dabigatran 110mg HR: 1.28, 95%CI: ; rivaroxaban 20mg vs. dabigatran 150mg HR: 1.81, 95%CI: ); -The mortality rate was higher vs. dabigatran (rivaroxaban 15mg vs. dabigatran110mg HR: 1.43, 95%CI: ; rivaroxaban 20mg vs. dabigatran 150mg HR: 1.52, 95%CI: ); Larsen T, et al OAC naive patients: N=35436 on warfarin; N=12701 on dabigatran 150mg; N=7192 on rivaroxaban 20mg; N=6349 on rivaroxaban 20mg Denmark (Danish nationwide administrative registries) - NOACs were not significantly different from warfarin in the analysis restricted to ischaemic stroke; -Rivaroxaban was associated with lower annual rates of ischaemic stroke or SE compared with warfarin (3.0% vs. 3.3%, respectively, HR 0.83; 95%CI: 0.69 to 0.99). The HRs for dabigatran (2.8% annually) and apixaban (4.9% annually) were non-significant compared with warfarin; -The annual risk of death was significantly lower with apixaban (5.2%) and dabigatran (2.7%) compared with warfarin (8.5%), but not with rivaroxaban (7.7%); -For the combined endpoint of any bleeding, annual rates for apixaban (3.3%) and dabigatran (2.4%) were significantly lower than for warfarin (5.0%), whilst warfarin and rivaroxaban had comparable annual bleeding rates (5.3%). AF: atrial fibrillation; GI: gastrointestinal; ICH: intracranial haemorrhage; VKA: vitamin K antagonist; OAC: oral anticoagulant; bid: twice daily; NR: not reported; SE: Systemic embolism; US: United States; HR: Hazard Ratio; CI: Confidence Interval.

19 Web-only Table 7. Real-world registries on thromboprophylaxis in patients with non-valvular atrial fibrillation. Registry Main registry features Demographics, stroke risk profile and AF data (where reported) Start year 2009 Mean age: 70.2±11.2 years; Status Ongoing GARFIELD 42 G L O Registry cohort Geographic region(s) Recruitment settings Purpose N of patients Cohort I ( ): Consecutive adult patients newly diagnosed with non-valvular AF and 1 additional investigator-defined stroke risk factors Europe, Asia-Pacific, Central/South America and Canada Randomly selected sites from officebased practice, hospital departments (neurology, cardiology, geriatrics, internal medicine, emergency), anticoagulation clinics and general or family practice To describe everyday antithrombotic treatment patterns and to understand the burden of thromboembolic and bleeding complications in patients with newly-diagnosed AF at risk of stroke Female: 43.2%; Mean CHA 2DS 2-VASc: 3.2±1.6; CHA 2DS 2-VASc 2: 84.4%; Mean CHADS2: 1.9±1.2; CHADS 2c 2: 57.2%; Paroxysmal AF: 27.5%; Persistent AF: 17.9%; Permanent AF: 24.9%; Unclassified AF: 29.7%; Cardiologists or internal medicine specialists enrolled 80% of patients. Antithrombotic therapies (as reported) Overall treatment OAC: 60.3% (VKAs 55.8%, NOACs 4.5%); APLT: 25.3%; No therapy: 14.4%. OAC per CHA2DS2-VASc category Score=0: 40.0% Score=1: 51.0%; Score 2: 63.0%. APLT per CHA2DS2-VASc category Score=0: 32.8% Score=1: 28.4%; Score 2: 24.6%. No therapy per CHA2DS2- VASc category Score=0: 27.2% Score=1: 20.7%; Score 2: 12.4%. Other findings The reasons for OAC non-use: 42 -OAC non-use was the physician choice in 48.3% of cases, and the main reasons were the risk of bleeding (7.4%), fall risk (6.5%) or concerns over patient compliance (5.3%). Patient s refusal to use OAC was sated in 7.2% of cases. Sex-related differences in OAC use: 43 (Patients prospectively enrolled during (n=17 184), 43.8% female) -Similar overall rates of OAC use in both genders (60.9% of male and 60.8% of female patients); -OAC underuse in patients at moderateto-high risk of stroke (<65% in both genders) and OAC overuse in low-risk patients (around 42% in both genders); -APLT alone was used in 28% of male and 27.2% of female patients; -No therapy was given to 12% of both male and female patients; -A high proportion of patients (36.7%) with missing data for the HAS-BLED score calculation; -OAC was less used in patients with a HAS-BLED of 3, regardless of the gender; -The physician choice was the most common reason for OAC non-use both in male and female patients (>45%). Start year 2011 Overall treatment Regional differences: 44

20 PREFER in AF 45 O R B Status Registry cohort Ongoing Consecutive adult patients with newly diagnosed non-valvular AF Median age: 71.0 years ( ); Geographic and a CHA2DS2-VASc of 1 Asia, Europe, North America, Latin Female: 45.5%; region(s) Recruitment settings Purpose America, Africa/Middle East Outpatients from university hospitals, specialist offices, community hospitals Mean CHA 2DS 2-VASc: 3.2±1.5; CHA 2DS 2-VASc=1: 14.5%; CHA 2DS 2-VASc 2: 85.5%; Mean HAS-BLED: 1.4±0.9; N of patients To prospectively collect information on patients with newly diagnosed non-valvular AF at risk of stroke, to identify antithrombotic treatment patterns and to address the issues of effectiveness and safety of antithrombotic therapies Paroxysmal AF: 54.5%; Persistent AF: 34.9%; Permanent AF: 10.6%; Asymptomatic AF: 31.4%; The main prescribing physicians were cardiologists: 91.7% Start year 2012 Mean age: 71.5±11.0 years; Status Ongoing Age>65 years: 75.0% Registry cohort Geographic region(s) Recruitment settings Purpose N of patients 7243 Consecutive adult patients with a history of AF documented by an ECG or implanted cardiac electronic device in the preceding 12 months 7 European countries (Austria, France, Germany, Italy, Spain, Switzerland and the UK) Office-based outpatient centers, hospital-based centers To gain detailed insight on the characteristics and management of AF patients with focus on the prevention of thromboembolic events, in particular stroke Female: 39.9%; Mean CHA 2DS 2-VASc: 3.4±1.8; CHA 2DS 2-VASc=0: 4.8%; CHA 2DS 2-VASc=1: 10.1%; CHA 2DS 2-VASc 2: 84.1%; Mean HAS-BLED: 2.0±NR Cardiologists enrolled 89% of patients. OAC: 80.0% (VKAs 32.3%, NOACs 47.7%); APLT: 12.3%; No therapy: 7.6%. OAC per CHA2DS2-VASc category Score=1: 65.1%; Score 2: 82.4%. APLT per CHA2DS2-VASc category Score=1: 20.3%; Score 2: 10.0%. No therapy per CHA2DS2- VASc category Score =1: 13.2%; Score 2: 6.7%. Overall treatment OAC: 82.3% (VKAs 78.0%, NOACs 6.0%); APLT: 22.1%; No therapy: 6.5% OAC per CHA2DS2-VASc category Score=0: 62.5%; Score=1: 70.1%; Score 2: 85.6%. Start year 2010 Median age: 75 years Overall treatment -The highest rate of OAC use was noted in Europe (90.2%), and the lowest in Asia (57.4%), -NOAC were used in 52% of patients in Europe and North America, and in 25.5% of patients in Asia, -The use of APLT in Asia was 25.8%, in North America 14.1%, in Europe 5.7%. General treatment patterns: -OAC was underused in high-risk patients, -APLT or no therapy were overused, and OAC was underused in paroxysmal AF patients, -No therapy was more commonly chosen in asymptomatic patients. -The proportion of patients with paroxysmal AF progressively decreased with increasing CHA 2DS 2-VASc score; -NOACs were used in younger patients at either high or low stroke risk; A small proportion of patients taking VKAs (13.5%) had labile INR values. -The combination therapy of OAC plus single or dual APLT is not uncommon among AF patients (791 of 7243 patients [10.9%]). Of 660 patients treated with OAC plus single APLT, 629 (95.3%) did not have an accepted indication for such therapy, and of patients taking OAC plus dual APLT 67 (63.8%) did not have an accepted indication for such treatment (e.g., coronary or peripheral vascular disease) 46.

21 Status Registry cohort Active A prospective registry of adult (67-82); Female: 42.0%; outpatients patients with AF Geographic US region(s) Recruitment settings Purpose N of patients Heterogeneous community-based centers from anticoagulation, cardiology and electrophysiology clinics To identify patterns in the characteristics and management of real-world community-based outpatients with AF (9957 non-valvular AF) CHADS 2=0-1: 28.8%; CHADS 2 2: 71.2%; ATRIA=0-3: 73.7%; ATRIA 4: 26.2%; New onset AF: 4.7%; Paroxysmal AF: 50.7%; Persistent or permanent AF: 44.5%; 81.5% of patients were included by electrophysiologists or cardiologists OAC: 76.0% (VKAs 71.0%, NOAC [Dabigatran] 4.9%); APLT: 47.3%; OAC per CHADS2 category CHADS 2=0: 52.5%; CHADS 2 2: 80.0%; OAC per CHA2DS2-VASc category Score=0: 44.8%; Score=1: 56.7%; Score 2: 80.0%; APLT: 75.2% of patients with a CHADS 2 of 0 -The rate of OAC use was decreasing with increasing bleeding risk (as measured by the ATRIA bleeding score). There was a significant interaction between OAC use and the level of bleeding risk across all stroke risk strata (nonetheless, the use of OAC increased with increasing stroke risk; -The rate of OAC use was higher among electrophysiologists or cardiologists compared with internal medicine specialists or internal medicine providers (p=0.02), across all stroke risk strata 49 ; -In 5210 patients taking warfarin, the overall mean TTR was 65%±20%, and median TTR was 68% (53-79%); 59% of all measured INRs were within the range of The frequency of INR measurement, patient s comorbidities, increasing CHA 2DS 2-VASc or ATRIA score values and treatment setting (e.g., outside the anticoagulation clinic) were significantly associated with lower TTR values 50 ; -Physician-reported contraindications to OAC were present in 13% of outpatients with non-valvular AF. Whilst patient refusal was the most common in younger patients at low risk of stroke (as measured by the CHA 2DS 2-VASc score), prior bleeding, high bleeding risk, frequent falls or frailty were more common in elderly patients at high risk of stroke. Around 30% of patients with a reported contraindication were taking OAC 51 ; -Over the 1-year follow-up, permanent warfarin discontinuation occurred in 10.1% of patients who were taking warfarin at baseline. The rate of warfarin discontinuation was the highest (17.1%) in patients newly started on warfarin during the last 12 months and in younger patients. The most commonly reported reasons for discontinuation were physician preference (47.7%), patient refusal/preference (21.1%), bleeding event (20.2%), frequent falls/frailty (10.8%), high bleeding risk (9.8%) and patient inability to adhere to/monitor therapy (4.7%) 52 ; -Of 7347 patients taking OAC, 2543 (35.0%) concomitantly used aspirin. These patients were slightly younger, more likely to be male and had more comorbidities. Multivariable factors associated with combination therapy were a history of coronary artery disease (OR 2.23; 95%CI, ), previous Maze procedure (OR 1.56; 95%CI, ), previous drug-eluting stent (OR 1.53; 95%CI, ) and previous stroke or TIA (OR 1.45; 95%CI, ). However, 39% of patients receiving OAC+aspirin had no history of atherosclerotic disease and, conversely, 37%of patients taking OAC alone had a history of atherosclerotic disease. Compared to OAC alone, combination therapy of OAC+aspirin was associated with increased risk of major bleeding (HR 1.53; 95%CI, ) and bleeding hospitalizations (HR 1.52; 95%CI, ), whilst rates of ischemic events were low in both groups 53. E O year 2012 Mean age: 68.8 years; RStart Status Active Age 75 years: 33.6%; Overall treatment -OAC use has increased, but NOAC use was still low. Compliance with the

22 Registry cohort Geographic region(s) Recruitment settings Purpose N of patients 3119 Consecutive adult in- and outpatients with documented nonvalvular AF treated by cardiologists 9 European countries (Belgium, Denmark, Netherlands, Norway, Poland, Romania, Greece, Italy, Portugal) Tertiary, secondary and general hospitals with and without interventional cardiology, electrophysiology or cardiac surgery services To describe the implementation of the current guidance for AF management, to evaluate the mortality and morbidity in relation to therapeutic decisions including adherence to guidelines in the EORP AF cohort at 1 year and subsequent follow-up Age years: 46.8%; Female: 40.4%; Mean CHADS 2: 1.93±1.27; Median CHADS 2: 2.0 ( ); Mean CHA 2DS 2-VASc: 3.24±1.80; Median CHA 2DS 2-VASc: 3.0 ( ); Mean HAS-BLED: 1.37±1.06; Median HAS-BLED: 1.00 ( ) First-detected AF: 36.5%; Paroxysmal AF: 19.0%; Persistent AF: 14.9%; Permanent AF: 29.5%; 100% of patients treated by cardiologists OAC: 80.0% (VKA: 71.6%, NOACs: 8.4%); APLT: 41.0%; No therapy: 4.8%; OAC per CHA2DS2-VASc category Score=0: 56.4% Score=1: 72.2%; Score 2: 79.5%. APLT per CHA2DS2-VASc category Score=0: 16.8% Score=1: 17.4%; Score 2: 18.3%. No therapy per CHA2DS2- VASc category Score=0: 26.3%; Score=1: 9.7%; Score 2: 1.8%. treatment guidelines for patients with the lowest and higher stroke risk scores remained suboptimal 54 ; -In elderly AF patients OAC were prescribed less often, and APLT alone was prescribed more commonly compared to younger AF patients; -APLT was more commonly used in patients with a HAS-BLED score of 2; -Female patients, those with paroxysmal AF and patients with coronary or peripheral vascular disease were more likely to receive APLT only, and the latter were also more likely to receive OAC in combination with APLT 55 ; -At 1-year follow-up, OAC use remained high (84% of patients taking VKAs at baseline, and 86% of patients taking a NOAC), with a minority of patients switching to other therapy (11.8% had changed to VKAs, and 1.1% to APLT). Despite good adherence to OAC, 1-year mortality was 5.7%, mostly due to cardiovascular death (particularly from heart failure) 56 ; -Of 2643 patients from the follow-up dataset, treatment was guideline adherent in 60.5%, whilst under-treatment was noted in 17.3% and over-treatment in 21.7% of patients. Guideline-adherent treatment was associated with significantly better outcomes compared with under- or over-treatment (for example, the outcome of all-cause death or any thromboembolism was more likely in undertreated patients (HR 1.68; 95%CI, ), whilst over-treatment was associated with increased risk of cardiovascular death, any thromboembolism or bleeding (HR 1.72; 95%CI, ) 57 ; -At 1-year follow-up OAC use was similar among patients with paroxysmal AF and those with non-paroxysmal AF (>80% in both groups). At 1 year, the rates of stroke or other thromboembolic events were low in both groups, but non-paroxysmal AF patients had a worse outcome (that is, all-cause mortality) due to a more severe clinical profile 58 ; -Female patients were older than male patients, with more heart failure and hypertension and a greater proportion of patients with a CHA 2DS 2-VASc score of 2 (94.7% vs. 74.6%), and the use of OAC was similar in both female and male patients 59 ;

23 BALKAN-AF Registry 61,62 -Although the overall use of OAC was high, some regional differences in patient risk profiles and patterns of management were noted. For example, patients from East and South European countries had higher stroke and bleeding risk compared with patients from West Europe, but only 50% of patients undergoing planned cardioversion were given OAC before the procedure 60. Start year Status Registry cohort 2014 Completed Consecutive adult patients with documented non-valvular AF Mean age: 69.1±10.9 years; Age years: 33.0%; Age 75 years: 35.4%; Geographic Female: 44.6%; region(s) Recruitment settings Purpose N of patients 2712 Balkan Region (Albania, Bosnia&Herzegovina, Bulgaria, Croatia, Montenegro, Romania and Serbia) Adult in- and outpatients from university or non-university centers located in capital or non-capital cities or rural regions To collect contemporary real-world data on demographics, patient risk profiles and AF management in daily clinical practice in European countries which are underrepresented in ongoing AF registries Mean CHA 2DS 2-VASc: 3.48±1.78; Median CHA 2DS 2-VASc: 3.0 ( ) Mean HAS-BLED: 1.97±1.23; First-diagnosed AF: 23.5%; Paroxysmal AF: 27.2%; Persistent AF: 18.8%; Permanent AF: 53.1%; 79.2% of patients were enrolled by cardiologists Overall treatment OAC: 73.6% (VKA: 61.5%, NOACs: 12.1%); APLT: 28.6%; No therapy: 9.9%; OAC per CHA2DS2-VASc category Score=0: 56.5% Score=1: 73.5%; Score 2: 74.4%. APLT per CHA2DS2-VASc category Score=0: 13.7% Score=1: 17.4%; Score 2: 30.4%. No therapy per CHA2DS2- VASc category Score=0: 33.6%; Score=1: 13.4%; Score 2: 8.5%. -Most of the CHA 2DS 2-VASc components were not significantly associated with the use of OAC, whilst increasing age and coronary artery disease were inversely related with OAC use); -Patients with paroxysmal AF were less likely to receive OAC; -Patients treated by a cardiologist or in an university center were more likely to receive OAC; -Elderly patients, those with coronary artery disease or paroxysmal AF were more likely to receive APLT (that is, aspirin) -INR was within the therapeutic range ( ) in only 55% of patients taking VKAs at baseline; -Quality of anticoagulation was rarely monitored by the calculation of TTR. Data were available for only 18.7% of patients, and the mean TTR was 49.5±22.3%. NCD R year 2008 Mean age: 71.3±12.9 years; Status Ongoing Registry cohort Adult outpatients with documented Female: 44.2%; PINNStart AF Overall treatment OAC: 44.9% (VKA: 90.3%, NOACs: 9.7%); APLT: 35.4%; -The risk of bleeding, as measured by the HAS-BLED score, was not a significant determinant of OAC use. -The prescription of OAC was significantly associated with the increasing stroke risk;

24 FUSHIMI AF Registry 66 Geographic region(s) Recruitment US Academic and private practice, Mean CHADS 2: 2.0±1.1; Mean CHA 2DS 2-VASc: 3.7±1.8; settings cardiovascular specialists Purpose To identify the use of antithrombotic therapy N of patients No therapy: 23.8%; -Patients with paroxysmal AF were less likely to be prescribed OAC compared to patients with paroxysmal AF (RR 0.74; 95%CI, ); -OAC was prescribed to 26.6% of patients with a CHA 2DS 2-VASc score of 0 64 ; -Significant practice-level variations in OAC prescription, with suboptimal OAC use 65. Start year Status 2011 Ongoing Mean age: 74.2±11.0 years; Age >80 years: 33.2%; Registry cohort Adult patients with documented AF Geographic Fushimi-ku area of Kyoto, Japan Female: 40.7%; region(s) Recruitment General practice clinics Mean CHADS 2: 2.09±1.35; settings Mean CHA 2DS 2-VASc: Purpose To identify the use of antithrombotic 3.43±1.71; therapy N of patients 3183 Paroxysmal AF: 46.0%; Persistent AF: 7.3%; Permanent AF: 46.7%; Asymptomatic AF: 50.3% Overall treatment OAC: 50.5% (VKA: 48.5%, NOACs: 4.2%); APLT: 31.1%; No therapy: NR%; -For each 1-point increase in the CHADS 2 score there were a 16.7% greater odds of OAC prescription vs. no antithrombotic therapy (OR 1.67; 95%CI, ) and a 15.8% greater odds of OAC prescription vs. aspirin only (OR 1.16; 95%CI, ). A similar association was noted for the CHA 2DS 2-VASc score (OR 1.19; 95%CI, and OR 1.16; 95%CI, , respectively); -However, OAC prescription capped at 50%, even in the high-risk AF patients. -The use of OAC generally increased with increasing stroke risk, but plateaued at 50% of patients even in the high-risk strata; OAC was overused in low-risk patients and underused in highrisk AF patients; -Of 2554 patients available at 1-year follow-up, 54.6% were taking OAC, and the use of NOAC did not increase substantially (6.3%) 67 ; -INR was within the recommended range in 54.4% of patients taking warfarin; -At 1-year follow-up, mortality was higher in patients not taking OAC (10.1% vs. 6.0%), but there was no difference in the rates of stroke or bleeding between the groups 67 ; -Paroxysmal AF was associated with a lower risk of stroke (HR 0.45; 95%CI, ), although the use of OAC was less common in patients with paroxysmal AF 68 ;

25 -Among 1541 patients not taking OAC at baseline, advanced age (HR 1.68; 95%CI, ), underweight (HR 1.71; 95%CI, ), previous stroke, systemic embolism or TIA (HR 1.70; 95%CI, ), heart failure (HR 1.59; 95%CI, ), chronic kidney disease (HR 1.53; 95%CI, ) and anemia (HR 2.41; 95%CI, ) were independent predictors of the composite outcome of death or stroke/systemic embolism 69 ; J-RHYTHM AF Registry 71 -Patients with low body weight had higher incidence of stroke or systemic embolism, but similar incidence of major bleeding compared with other AF patients (HR 2.19; 95%CI, and HR 1.05; 95%CI, , respectively) 70. Start year 2009 Mean age: 69.7±9.9 years; Overall treatment -INR was in 66.0% of patients Status Ongoing Age 80 years: 16.0%; OAC: 87.3% (all VKA); taking VKAs, and 36.4% of patients had Registry cohort Adult consecutive patients with APLT: 25.4%; an INR of ; documented AF Female: 31.1%; No therapy: 5.3%; Geographic Japan (all 10 regions) -The optimal INR range may be region(s) CHADS 2=0: 15.6%; OAC per CHADS category narrower in Japanese patients compared Recruitment Outpatients from a range of CHADS 2=1: 34.0%; Score=0: 76.5% with Caucasians: the relationship of INR settings specialized cardiovascular centers CHADS 2 2: 50.4%; Score=1: 86.1%; value and stroke incidence remained Score 2: 91.8%; virtually identical, but the risk of ICH Purpose To identify the use of antithrombotic Paroxysmal AF: 37.1%; sharply increased above the INR values therapies in AF patients in Japan Persistent AF: 14.4%; of > ; N of patients 7937 Permanent AF: 48.5%; APLT per CHADS category Score=0: 10.7% Score=1: 7.8%; Score 2: 5.9%; No therapy per CHADS category Score=0: 12.8% Score=1: 6.1%; Score 2: 2.5%; -The rates of stroke and bleeding were higher in patients 85 years old compared to younger AF patients. In the oldest group of AF patients 79.7% were taking warfarin, and 67.1% of those had a target INR of The rates of stroke and all-cause mortality were lower in this subgroup compared with the elderly AF patients not taking warfarin, without increase in the rates of major bleeding 73 ; -In a pooled analysis of 3588 AF patients not taking OAC (the Shinken Database, J-RHYTHM Registry and Fushimi AF Registry) the rates of ischemic stroke were 5.3, 5.5 and 18.4 per 1000 person-years in patients with a CHA 2DS 2-VASc score of 0, 1 and 2, respectively. History of stroke or TIA (HR 3.25; 95%CI, ), age 75 years (HR 2.31; 95%CI, ) and

26 hypertension (HR 1.69; 95%CI, ) were independent risk factors for stroke 74 ; -The use of warfarin was associated with lower risk of all-cause mortality (OR 0.30; 95%CI, ), whilst the use of NOACs was associated with lower rates of stroke (OR 0.42; 95%CI, ), major bleeding (OR 0.53; 95%CI, ) and all-cause death (OR 0.10; 95%CI, ), as compared to patients not taking OAC 75. AF: Atrial fibrillation; OAC: Oral anticoagulant therapy; VKA: Vitamin K antagonist; NOAC: Non-vitamin K antagonist oral anticoagulant; APLT: Antiplatelet therapy; ECG: Electrocardiography; NR: Not reported; GARFIELD: The Global Anticoagulant Registry in the Field. GLORIA-AF: The Global Registry on Long-Term Oral Antithrombotic Treatment in Patients with Atrial Fibrillation. PREFER in AF: Prevention of thromboembolic events European Registry in Atrial Fibrillation. ORBIT-AF: The Outcomes Registry for Better Informed Treatment of Atrial Fibrillation. EORP-AF Pilot General Registry: EurObservational Research Programme Atrial Fibrillation Pilot General Registry. BALKAN-AF: The Snapshot Survey on the management of atrial fibrillation in the Balkan countries. NCDR PINNACLE: The National Cardiovascular Data Registry Practice Innovation and Clinical Excellence Registry. FUSHIMI-AF Registry: Japan, general practice; J-RHYTHM Registry: Japan, specialized cardiovascular centres.

27 References 1. Martinez C, Katholing A, Wallenhorst C, Freedman SB. Therapy persistence in newly diagnosed non-valvular atrial fibrillation treated with warfarin or NOAC. A cohort study. Thromb. Haemost. 2016;115(1): Leyden JM, Kleinig TJ, Newbury J, et al. Adelaide stroke incidence study: declining stroke rates but many preventable cardioembolic strokes. Stroke. 2013;44(5): Friberg L, Rosenqvist M, Lindgren A, Terent A, Norrving B, Asplund K. High prevalence of atrial fibrillation among patients with ischemic stroke. Stroke. 2014;45(9): Yiin GS, Howard DP, Paul NL, et al. Recent time trends in incidence, outcome and premorbid treatment of atrial fibrillation-related stroke and other embolic vascular events: a populationbased study. J Neurol Neurosurg Psychiatry Yiin GS, Howard DP, Paul NL, et al. Age-specific incidence, outcome, cost, and projected future burden of atrial fibrillation-related embolic vascular events: a population-based study. Circulation. 2014;130(15): Jannou V, Timsit S, Nowak E, et al. Stroke with atrial fibrillation or atrial flutter: a descriptive population-based study from the Brest stroke registry. BMC Geriatr. 2015;15: Palm F, Kleemann T, Dos Santos M, et al. Stroke due to atrial fibrillation in a population-based stroke registry (Ludwigshafen Stroke Study) CHADS(2), CHA(2) DS(2) -VASc score, underuse of oral anticoagulation, and implications for preventive measures. European journal of neurology : the official journal of the European Federation of Neurological Societies. 2013;20(1): Bjorck S, Palaszewski B, Friberg L, Bergfeldt L. Atrial fibrillation, stroke risk, and warfarin therapy revisited: a population-based study. Stroke. 2013;44(11): Hannon N, Sheehan O, Kelly L, et al. Stroke associated with atrial fibrillation--incidence and early outcomes in the north Dublin population stroke study. Cerebrovascular diseases. 2010;29(1): Bejot Y, Ben Salem D, Osseby GV, et al. Epidemiology of ischemic stroke from atrial fibrillation in Dijon, France, from 1985 to Neurology. 2009;72(4): Marini C, De Santis F, Sacco S, et al. Contribution of atrial fibrillation to incidence and outcome of ischemic stroke: results from a population-based study. Stroke; a journal of cerebral circulation. 2005;36(6): Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2): Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA : the journal of the American Medical Association. 2001;285(22): Singer DE, Chang Y, Borowsky LH, et al. A new risk scheme to predict ischemic stroke and other thromboembolism in atrial fibrillation: the ATRIA study stroke risk score. Journal of the American Heart Association. 2013;2(3):e Piccini JP, Stevens SR, Chang Y, et al. Renal dysfunction as a predictor of stroke and systemic embolism in patients with nonvalvular atrial fibrillation: validation of the R(2)CHADS(2) index in the ROCKET AF (Rivaroxaban Once-daily, oral, direct factor Xa inhibition Compared with vitamin K antagonism for prevention of stroke and Embolism Trial in Atrial Fibrillation) and ATRIA (AnTicoagulation and Risk factors In Atrial fibrillation) study cohorts. Circulation. 2013;127(2): Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138(5): O'Brien EC, Simon DN, Thomas LE, et al. The ORBIT bleeding score: a simple bedside score to assess bleeding risk in atrial fibrillation. European heart journal. 2015;36(46):