#cardiac-arrhythmias #acquired-arrhythmias #atrial-fibrillation #anticoagulation #stroke-prevention

Atrial Fibrillation (AF)

Contents. #Definition · #Key Facts · #Epidemiology and outcomes · #Pathophysiology · #Risk factors · #Diagnosis and classification · #Stroke prevention · #Rate and rhythm control · #Catheter ablation · #Surgery and LAA closure · #Special populations · #Genetics and biological therapy · #Contradictions / Open Questions · #Sources

Detailed pages. concepts/AF-Special-Populations · concepts/CHA2DS2-VA · concepts/Antiarrhythmic-Drugs · concepts/Catheter-Ablation-AF · concepts/AF-CARE · concepts/Atrial-Cardiomyopathy

Definition

Atrial fibrillation is the most prevalent sustained cardiac arrhythmia, characterised by disorganised atrial electrical activity and an elevated risk of thromboembolic events, heart failure and all-cause mortality. It arises from a substrate of electrical and structural remodelling, focal ectopic activity and re-entrant circuits. The 2025 ESC/HFA consensus reframes it further: AF is often a manifestation of underlying atrial cardiomyopathy rather than the primary driver of stroke and heart failure, and atrial structural disease predicts adverse outcomes more strongly than AF episodes themselves (sources/atrial-cmp-esc-2025, rating: high).

Key Facts

Scale. Global prevalence ~59.7 million in 2019; lifetime risk about 1 in 3; prevalence expected to double by 2060. (sources/AF-ESC-2024, rating: very high)
Four management pillars. Comorbidity and risk-factor management, stroke avoidance, rate and rhythm control, and evaluation — the AF-CARE framework, with comorbidity management first and foundational. (sources/AF-ESC-2024) See concepts/AF-CARE.
Staging. The AHA 2023 four-stage classification (at risk → pre-AF → AF → permanent) has replaced duration-based labels. See concepts/AF-Staging. (sources/AF-AHA-2023, rating: very high)
Anticoagulation is decided by risk score, not rhythm. AF pattern, type and burden do not determine the OAC decision. See concepts/AF-Burden and concepts/CHA2DS2-VA.
Ablation is first-line in paroxysmal AF. ESC 2024 assigns Class I/A, upgraded from IIa/B in 2020. See concepts/Catheter-Ablation-AF.
Risk-factor modification works. Weight loss ≥10% carries a Class I recommendation in both guidelines and is the one lifestyle intervention with randomised support for reducing AF burden.
Two live controversies frame the field: whether LAAO can replace indefinite anticoagulation, and whether AF or atrial cardiomyopathy should be the primary target. Both are under Contradictions.

Details

Epidemiology and outcomes

Measure	Magnitude
Global prevalence	~59.7 million (2019); doubling expected by 2060
Lifetime risk	1 in 3
Incident heart failure	4–5×
Ischaemic stroke	2.3×
All-cause mortality	2×
Cognitive impairment	HR 1.39
Vascular dementia	HR 1.68
Global deaths attributed	>250,000 (2017 data); age-standardised mortality 4.0/100,000

(Table: sources/AF-ESC-2024, rating: very high)

Pathophysiology

Electrical remodelling. Reduced I_Ca,L, increased I_K1 and constitutively active I_KACh shorten the atrial action potential and its plateau, producing a triangulated potential and a short refractory period; reduced I_Na slows conduction. The re-entry wavelength shortens until multiple wavefronts coexist — the ionic basis of "AF begets AF". (sources/membrane-potential-physrev-2021, rating: very high) See concepts/Electrical-Remodeling.
Pulmonary vein triggers. PV sleeve myocardium retains fetal-type channels with more I_f and I_Ca,T and less I_K1, so adrenergic stimulation produces spontaneous diastolic depolarisations and triggered beats that initiate episodes. (sources/membrane-potential-physrev-2021)
Structural remodelling. TGF-β-mediated atrial fibrosis creates conduction heterogeneity that sustains AF. (sources/gene-therapy-arrhythmia-2025, rating: high)
Atrial cardiomyopathy. LA enlargement, low LASr and excess fibrosis predict stroke and dementia better than AF episodes do, and subclinical AF >6 min is present in only 15% of patients in the month before an embolic event. (sources/atrial-cmp-esc-2025) See concepts/Atrial-Cardiomyopathy and concepts/Left-Atrial-Strain.

Risk factors

Factor	Association with AF	What modification buys
Obesity and epicardial fat	~1/5 of cases; 29% higher risk per 5 BMI units; 54% and 87% higher progression risk in class 1 and class 2 obesity; epicardial fat stronger than general adiposity	≥10% weight loss → ~6-fold higher freedom from AF at 5 years; Class I
Sleep-disordered breathing	Present in 21–74% of AF patients; ~5-fold odds after adjustment (HR 2.18 vs 1.49 for obesity); nocturnal hypoxia, not AHI, predicts incident AF (HR 2.47)	CPAP restores recurrence rates towards non-OSA levels observationally; treatment Class I (AHA)
MASLD / metabolic syndrome	×1.2–1.5 long-term risk	Modifiable marker; GLP-1 RAs and SGLT2 inhibitors act through weight and metabolism
Hypertension	Highest attributable risk of any modifiable factor; present in >80% of AF patients	BP <130/80 reduces incident AF, especially in HF
Alcohol	Pooled OR/RR 1.51, dose-dependent	Reduction advised; no protective low-dose effect
Fish oil supplements	Possible increased AF risk, including icosapent ethyl (~1% absolute)	Dietary fish does not carry the signal
Drugs	Anticancer agents, antipsychotics, catecholamines, bronchodilators, class IC agents, NSAIDs	Withdraw the agent; many self-convert

(Table compiled from sources/obesity-cv-aha-2021, rating: very high; sources/sdb-arrhythmia-aha-2022, rating: very high; sources/osa-af-jama-2018, rating: high; sources/masld-nejm-2025, rating: high; sources/HT-AHA-2025, rating: very high; sources/diet-aha-2026, rating: very high; sources/drug-arrhythmia-aha-2020, rating: very high)

Where the detail lives. entities/Obesity and concepts/Epicardial-Adipose-Tissue-Arrhythmogenesis for adiposity; concepts/OSA-Arrhythmogenic-Substrate and entities/Obstructive-Sleep-Apnea for sleep apnoea (including the 31% higher ablation recurrence and the VARIOSA-AF directionality data); entities/MASLD for metabolic disease; entities/Hypertension for BP targets; concepts/Drug-Induced-Arrhythmia for the full drug catalogue.
Screening. Symptom-based screening for sleep apnoea is unreliable — most AF patients with OSA do not report daytime sleepiness — so sleep study evaluation is reasonable in anyone being considered for rhythm control.

Diagnosis and classification

ECG confirmation with ≥30 s of recording and physician review (Class I/A); PPG or smartwatch recordings alone are not diagnostic. (sources/AF-ESC-2024)
Opportunistic screening — pulse-taking or rhythm strip at every visit — is recommended during BTK inhibitor therapy (Class I/C). (sources/Cardio-Oncology-ESC-2022, rating: very high)
Device-detected subclinical AF. False-positive atrial high-rate episodes occur in ~17% (ASSERT), so electrogram review is required; SCAF appears in 10–50% of CIED populations, carries 2.4× stroke risk and predicts clinical AF with a 5.6–5.9× hazard. (sources/subclinical-af-aha-2019, rating: high) See concepts/Subclinical-AF.
Staging. Stage 3 is subdivided into paroxysmal, persistent, long-standing and post-ablation sinus rhythm. See concepts/AF-Staging.

Stroke prevention

Decision point	ESC 2024	AHA 2023
Risk score	CHA₂DS₂-VA (sex removed)	CHA₂DS₂-VASc (sex retained)
OAC threshold	≥2 recommended; = 1 considered	≥2%/year (Class I/A); 1–2%/year reasonable (IIa/A)
HCM and cardiac amyloidosis	Treat regardless; score ≥1 threshold	Treat HCM as though score 3
DOAC versus VKA	DOAC preferred (I/A) except mechanical valves and moderate-to-severe mitral stenosis	Same
Device-detected AHREs	Single Class IIb/B, no duration cut-off	≥24 h + score ≥2 (IIa); 5 min–24 h + score ≥3 (IIb); <5 min none (III)
LAAO	—	Class IIa/B-NR when OAC is contraindicated

(Table: sources/AF-ESC-2024, rating: very high; sources/AF-AHA-2023, rating: very high)

Intermediate risk now has RCT evidence. SINGLE-AF (n=1,803, DOAC versus none at CHA₂DS₂-VA = 1) reduced the composite of stroke, systemic embolism, major bleeding or cardiovascular death from 1.5% to 0.5% (HR 0.31, 0.10–0.94), on 17 events with a confidence interval touching null. (sources/single-af-nejm-2026, rating: high) See concepts/CHA2DS2-VA.
Burden is not a decision variable. Persistent AF carries higher stroke risk than paroxysmal in trial and meta-analysis data, but the effect disappears at high risk and there is no monotonic burden–stroke gradient across a >250-fold range. See concepts/AF-Burden. (sources/afburden-aha-2018, rating: high)
LAAO is a three-trial controversy, not a settled alternative. OPTION showed non-inferiority with less bleeding after ablation; CHAMPION-AF showed non-inferiority plus bleeding superiority in NOAC-eligible patients; CLOSURE-AF failed non-inferiority in elderly patients unsuitable for anticoagulation. See entities/LAAO for the full landscape.
Factor XI inhibition has not displaced DOACs. The class reliably reduces bleeding — abelacimab by 62–69% versus rivaroxaban — but asundexian produced 3.8-fold more strokes or systemic emboli than apixaban in a 14,810-patient phase 3 trial that was stopped for that reason, with the composite of stroke, systemic embolism or major bleeding also favouring apixaban. No agent is approved. See concepts/Factor-XI-Inhibition. (sources/asundexian-oceanicaf-nejm-2025, rating: very high; sources/abelacimab-azaleatimi71-nejm-2025, rating: high)

Rate and rhythm control

Drug	Class	Role	Key caveat
Flecainide	IC	Pharmacological cardioversion (up to 95% within 1 h), pill-in-the-pocket (75–85% by 6–8 h), sinus-rhythm maintenance	Structurally normal hearts only; AV nodal blocker mandatory because of 1:1 flutter risk
Amiodarone	III	Acute cardioversion (35–65%); maintenance — 84% freedom at 1 year, 45% at 5 years	Only safe AAD in HFrEF; no mortality benefit there (SCD-HeFT); thyroid, pulmonary and neurological toxicity
Dronedarone	III	Paroxysmal or persistent AF	Contraindicated in severe systolic HF (ANDROMEDA HR 2.13); not for permanent AF (PALLAS)

(Table: sources/flecainide-af-europace-2011, rating: medium; sources/amiodarone-cvdrug-2020, rating: high; sources/dronedarone-circ-2009, rating: medium)

Early rhythm control. Initiating it within 12 months of diagnosis reduces cardiovascular death, stroke and HF hospitalisation — EAST-AFNET 4 HR 0.79 (96% CI 0.66–0.94), stopped early for efficacy at a median 5.1 years, with benefit independent of symptoms. (sources/EAST-AFNET4-NEJM-2020, rating: very high)
Rate control. Resting target <100–110 bpm (RACE II); diltiazem and verapamil are contraindicated at LVEF <40%. (sources/AF-AHA-2023)
Full detail: concepts/Antiarrhythmic-Drugs for the class comparison, and entities/Amiodarone, entities/Dronedarone, entities/Flecainide for pharmacology, dosing and toxicity.

Catheter ablation

Guideline position. ESC 2024: Class I/A first-line in paroxysmal AF, I/B in persistent AF after failed AADs, and I where tachycardiomyopathy is likely. AHA 2023: I/A in younger patients with few comorbidities, IIa/B-R more broadly, I/A in HFrEF on GDMT. (sources/AF-ESC-2024, sources/AF-AHA-2023)
End-stage HFrEF. CASTLE-HTx: primary composite 8% versus 30% (HR 0.24), all-cause mortality 6% versus 20% (HR 0.29) — single-centre, open-label and stopped early, so the effect size needs replication. (sources/CA-HF-CASTLEHTx-NEJM-2023, rating: very high)
Persistent AF, first-line. AVANT GUARD: PFA versus AADs, 56% versus 30% treatment success at 12 months (HR 0.46). (sources/PFA-LPeAF-AVANT-GUARD-NEJM-2026, rating: very high)
Progression and cognition. First-line cryoablation reduced progression to persistent AF from 7.4% to 1.9% (EARLY-AF), and ablation is associated with HR 0.60 for incident dementia. (sources/ca-af-ehj-2024, rating: high)
Full detail: concepts/Catheter-Ablation-AF for indications, energy sources and lesion sets; concepts/PFA-Atrial-Arrhythmias for the pulsed-field platform.

Surgery and LAA closure

Concomitant surgical LAA closure is Class I/B (ESC 2024) and I/A (AHA 2023) as an adjunct to OAC; LAAOS III showed HR 0.67 for ischaemic stroke or systemic embolism. (sources/AF-ESC-2024, sources/AF-AHA-2023)
Valve surgery (ESC 2025). Surgical LAAO is Class I/B in any valve surgery with AF; surgical AF ablation is I/A with mitral valve surgery and IIa/A with non-mitral valve surgery. (sources/vhd-esc-2025, rating: very high)
AF drives atrial secondary MR and TR through annular dilatation without LV dilatation; rhythm control may reduce both and reverse atrial dilatation. See concepts/Secondary-Mitral-Regurgitation and concepts/Tricuspid-Regurgitation.
DOACs in valve disease. Preferred with AS, AR or MR (I/A); contraindicated in rheumatic mitral stenosis with MVA ≤2.0 cm² (III/B), where VKA is mandatory.

Special populations

Population	The AF-specific issue	Full detail
Heart failure	AF in 25–50%; tachycardiomyopathy has a better course than AF in established HF; CASTLE-AF HR 0.62 for death or HF hospitalisation; amiodarone is the only safe AAD	concepts/AF-Special-Populations
Cancer	2–16% during active therapy; bidirectional association; 2× stroke and 6× HF risk; CHA₂DS₂-VASc underestimates and HAS-BLED is unreliable	concepts/AF-Special-Populations, concepts/Cancer-Associated-Arrhythmia
Noncardiac surgery (POAF)	0.4–40% incidence; HR 2.00 for 30-day stroke; 39% recurrence at 5 years; anticoagulate if it persists beyond 24–48 h	concepts/AF-Special-Populations, concepts/Perioperative-Cardiovascular-Assessment
ACS with PCI	Aspirin withdrawn at 1–4 weeks and clopidogrel plus OAC continued (I/B-R); ~80% of stent thromboses occur within 30 days	concepts/AF-Special-Populations, concepts/DAPT-Strategies
Stable CAD beyond 6–12 months	DOAC monotherapy Class IIb/C-LD; EPIC-CAD net adverse events HR 0.44; ADAPT AF-DES (DES-specific) 9.6% vs 17.2%, HR 0.54	concepts/AF-Special-Populations, concepts/AF-Stable-CAD-Antithrombotic
Acute ischaemic stroke	Early OAC within 4–14 days is reasonable (COR 2a); ELAN 30-day composite 2.9% vs 4.1%, sICH 0.2% both arms; heparin bridging is harmful (COR 3)	concepts/AF-Special-Populations, entities/Ischemic-Stroke

Genetics and biological therapy

Early-onset AF is often genetic. In 1,293 patients diagnosed before 66, 10.1% carried a pathogenic or likely pathogenic variant in a cardiomyopathy or arrhythmia gene, rising to 16.8% below age 30, mostly in cardiomyopathy genes (TTN, MYH7, MYH6, LMNA) rather than channelopathy genes. (sources/eoaf-jama-2021, rating: high) See concepts/Genetic-Testing-in-AF.
Testing thresholds. EP study is reasonable for AF under 30 (24–39% have re-entrant SVT); genetic testing is reasonable under 45 without obvious risk factors (Class IIb/B-NR). (sources/AF-AHA-2023)
Very young AF without structural disease still carries a high modifiable-risk-factor burden (89% had at least one) and an 11.6% P/LP yield on panel testing, and 4.1% of a cohort with CHA₂DS₂-VA = 0 had a stroke. (sources/eoaf-riskfactor-ehj-2026, rating: medium) See concepts/Early-Onset-Atrial-Fibrillation.
Sarcomeric HCM shows primary atrial myopathy: more LA low-voltage area than gene-negative controls at comparable LA pressures. See concepts/HCM-Atrial-Myopathy. (sources/MYBPC3-MYH7-JACCEP-2024, rating: medium)
Biological therapy remains preclinical. Pooled across 25 animal studies, gene, miRNA and cell therapies reduced AF inducibility by 85% (OR 0.15) and atrial fibrosis by 6.7%, yet none has entered clinical trials and every study delivered therapy before AF onset. See concepts/AF-Biological-Therapy and concepts/AAV-Gene-Delivery. (sources/biological-tx-af-hrs-2019, rating: medium)

Contradictions / Open Questions

LAAO: same intervention, opposite conclusions. OPTION showed non-inferiority after ablation with 56% less bleeding; CHAMPION-AF showed non-inferiority plus bleeding superiority in NOAC-eligible patients (HR 0.55); CLOSURE-AF failed non-inferiority in elderly patients unsuitable for anticoagulation (P=0.44), with more bleeding and numerically more cardiovascular death. Risk profile, post-implant antithrombotic strategy (NOAC versus DAPT) and non-inferiority margin structure explain the divergence — but the group LAAO was most advocated for is the group in which it failed. (sources/laao-closureaf-nejm-2026, sources/laao-championaf-nejm-2026, rating: very high)
Atrial cardiomyopathy versus the AF-first paradigm. If atrial structural disease predicts stroke independently of AF, then thresholds tied to AF detection and duration may systematically miss the patients most at risk — yet ARCADIA, which selected cryptogenic stroke patients by atrial cardiomyopathy markers, was stopped for futility. Either the markers were inadequate or anticoagulation without documented AF does not work. (sources/atrial-cmp-esc-2025, rating: high)
The intermediate-risk threshold has RCT support but low precision. SINGLE-AF rests on 17 events with a CI touching null, the endpoint was amended mid-trial, and antiplatelet use differed between arms. What makes net benefit favourable there is its low bleeding rate; ARTESiA and NOAH-AFNET 6 ran major bleeding at 2.1% and 1.5% per patient-year at median CHA₂DS₂-VASc 4. The unresolved question is at what bleeding risk net benefit inverts — no trial has delimited that boundary. (sources/single-af-nejm-2026, sources/apixaban-scaf-artesia-nejm-2024, rating: very high)
Sex category: ESC versus AHA. CHA₂DS₂-VA removes sex; CHA₂DS₂-VASc retains it, so a woman whose only risk factor is female sex scores 1 under AHA and 0 under ESC. SINGLE-AF cannot arbitrate — it enrolled to the sex-differentiated threshold and had only 2 control-arm events among 427 women. The disagreement reappears in the ESC 2021 HF guideline versus ESC 2024. (sources/AF-ESC-2024, sources/AF-AHA-2023)
Device-detected AF thresholds are guideline artefacts. AHA uses explicit duration bands; ESC uses one Class IIb recommendation without cut-offs. ARTESiA and NOAH-AFNET 6 gave discordant results, and no trial has defined the duration or burden that warrants anticoagulation. (sources/AF-ESC-2024, sources/AF-AHA-2023)
Anticoagulation after successful ablation. Guidelines recommend continuing OAC by stroke risk regardless of rhythm outcome, and no trial has tested stopping it in patients whose risk score falls only because of ablation success. (sources/AF-ESC-2024, sources/AF-AHA-2023)
CABANA versus CASTLE-AF and CASTLE-HTx. CABANA found no mortality benefit in all-comers while the HF trials found large benefit, so patient selection rather than the procedure explains the difference — and CASTLE-HTx's HR of 0.24 remains unreplicated. (sources/AF-ESC-2024, sources/CA-HF-CASTLEHTx-NEJM-2023)
Cancer-specific risk prediction does not exist. CHA₂DS₂-VASc and HAS-BLED both fail in cancer — the former underestimates stroke and the latter ignores thrombocytopenia and brain metastases — and no validated cancer-specific algorithm has replaced them. (sources/Cardio-Oncology-ESC-2022)
Anticoagulation for perioperative AF is empirical. The recommendation to start OAC when POAF persists beyond 24–48 h rests on registry data showing a 48% reduction in thromboembolism, not on a randomised trial. (sources/periop-aha-2024, rating: very high)
Atrial cardiomyopathy has no therapy. It can be diagnosed and it predicts outcomes, but nothing has been shown to reverse it, and the ARCADIA result questions whether treating it without documented AF helps. (sources/atrial-cmp-esc-2025)

Connections

concepts/AF-CARE — the four-pillar ESC framework this page is organised around
concepts/AF-Staging — the AHA four-stage classification replacing duration labels
concepts/AF-Burden — why burden is not used to decide anticoagulation
concepts/CHA2DS2-VA — the score that governs the OAC decision, and its evidence base
concepts/Subclinical-AF — device-detected AF, its stroke risk and its treatment thresholds
concepts/NOAC-AF-Management — dosing, renal and interaction detail for the DOACs
concepts/Factor-XI-Inhibition — the investigational class, and why it failed its first phase 3 test
entities/LAAO — the three-trial LAAO landscape in full
concepts/Antiarrhythmic-Drugs — the class comparison behind the drug table
concepts/Catheter-Ablation-AF — indications, energy sources and lesion sets
concepts/PFA-Atrial-Arrhythmias — the pulsed-field platform
concepts/AF-Special-Populations — HF, cancer, perioperative, ACS/CAD and post-stroke AF in detail
concepts/Atrial-Cardiomyopathy — the substrate-first reframing of AF
concepts/Atrial-Failure — advanced atrial cardiomyopathy
concepts/Left-Atrial-Strain — the imaging marker of atrial mechanical dysfunction
concepts/OSA-Arrhythmogenic-Substrate — the sleep-apnoea substrate
concepts/Epicardial-Adipose-Tissue-Arrhythmogenesis — how epicardial fat creates the substrate
concepts/Drug-Induced-Arrhythmia — the drug catalogue behind the risk-factor table
concepts/Genetic-Testing-in-AF — who to test and what the yield is
concepts/Early-Onset-Atrial-Fibrillation — AF under 40 without structural disease
concepts/HCM-Atrial-Myopathy — sarcomeric atrial myopathy
concepts/AF-Biological-Therapy — gene, miRNA and cell therapy in preclinical development
entities/Atrial-Flutter — the related organised atrial arrhythmia
entities/Ischemic-Stroke — post-stroke anticoagulation timing
entities/HCM — HCM, where anticoagulation follows regardless of score
concepts/Cancer-Associated-Arrhythmia — cancer therapy and AF, extending the special-populations material

Sources

Guidelines and frameworks

Mechanisms and atrial cardiomyopathy

Epidemiology and risk factors

Anticoagulation and stroke prevention

Rhythm control and ablation

Genetics and biological therapy

Perioperative and coronary disease