NOAC Management in Atrial Fibrillation
Definition
Non-vitamin K antagonist oral anticoagulants (NOACs — also called DOACs) are four agents for stroke prevention in AF: dabigatran, a direct thrombin inhibitor, and the factor Xa inhibitors apixaban, rivaroxaban and edoxaban. They are preferred to vitamin K antagonists when anticoagulation is newly started (ESC Class I/A), except in EHRA Type 1 valvular disease — a mechanical prosthesis or moderate-to-severe mitral stenosis. This page follows the structure of the EHRA Practical Guide and updates it where later trials and guidelines have moved.
Abbreviations: AF, atrial fibrillation; BID, twice daily; CHA₂DS₂-VASc and CHA₂DS₂-VA, stroke risk scores; CrCl, creatinine clearance; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; EHRA, European Heart Rhythm Association; GI, gastrointestinal; INR, international normalised ratio; LAA, left atrial appendage; LMWH, low-molecular-weight heparin; NOAC, non-vitamin K antagonist oral anticoagulant; OD, once daily; P-gp, P-glycoprotein; TOE, transoesophageal echocardiography; TAVI, transcatheter aortic valve implantation; UFH, unfractionated heparin; VKA, vitamin K antagonist; VTE, venous thromboembolism.
Key Concepts
Eligibility — EHRA Valvular Classification
| Type | What it contains | Anticoagulation |
|---|---|---|
| Type 1 (VKA-mandatory) | Mechanical prosthetic valve; moderate-to-severe mitral stenosis, usually rheumatic | NOACs contraindicated |
| Type 2 (NOAC-eligible) | All other native valve disease; bioprosthesis beyond 3 months and not for rheumatic mitral stenosis; mitral repair; TAVI with AF; HCM with AF | NOACs appropriate |
(sources/noac-ehra-2018, rating: high)
- HCM with AF has no mechanistic reason for NOAC inferiority to VKA, so these patients are eligible — which aligns with the separate rule that anticoagulation in HCM is mandatory regardless of stroke score. (sources/noac-ehra-2018)
- TAVI is Type 2 on the 2018 classification, but see the GALILEO contradiction below: the trial published afterwards, and it points the other way.
Pharmacokinetics and Clearance
Four differences between the agents drive every choice in kidney disease and every interaction decision.
| NOAC | Renal clearance (unchanged drug) | CYP3A4 metabolism | Protein binding | Half-life |
|---|---|---|---|---|
| Dabigatran | 80% | No | 35% | 12–17 h |
| Edoxaban | 50% | Minimal | 55% | 10–14 h |
| Rivaroxaban | 35% | Yes (18%) | 95% | 5–13 h |
| Apixaban | 27% | Yes (25%) | 87% | 12 h |
(sources/noac-ehra-2018, rating: high)
Standard Dosing and Dose Reduction
| NOAC | Standard | Reduced | Criteria for reduction |
|---|---|---|---|
| Apixaban | 5 mg BID | 2.5 mg BID | ≥2 of: age ≥80, weight ≤60 kg, creatinine ≥133 µmol/L |
| Dabigatran | 150 mg BID | 110 mg BID | Age >75, high bleeding risk, CrCl 30–49 |
| Edoxaban | 60 mg OD | 30 mg OD | Weight ≤60 kg, CrCl ≤50, strong P-gp inhibitor |
| Rivaroxaban | 20 mg OD | 15 mg OD | CrCl ≤50 |
(sources/noac-ehra-2018, rating: high)
- Never use a lower dose outside a tested reduction algorithm. Underdosing raises stroke rates without any safety benefit — it is the commonest prescribing error in this class.
- How the bottom of the indication range behaves in practice (SINGLE-AF, 2026): among 902 patients with AF at CHA₂DS₂-VASc 1 (men) or 2 (women) dosed at physician discretion, 81.6% received a full dose and 18.4% a reduced dose, most often apixaban 5 mg (67.4%), then rivaroxaban 20 mg (14.1%) and 15 mg (13.9%). Mean HAS-BLED was 0.5 and major bleeding occurred in 0.3% at 24 months, about 0.2% per year. Roughly one patient in five met a reduction criterion at this risk level, and no excess bleeding was seen — but no outcomes were reported by dose, so the trial cannot say whether the reduced-dose group was appropriately selected. (sources/single-af-nejm-2026, rating: high)
Chronic Kidney Disease
- Use Cockcroft-Gault, not CKD-EPI, for every NOAC dosing decision, matching the method used in the trials. Reassess renal function every CrCl/10 months — a CrCl of 30 mL/min means every 3 months. (sources/noac-ehra-2018, rating: high)
- Renal dependence differs enough to decide the drug. Dabigatran is 80% renally cleared and is not approved below CrCl 30; a factor Xa inhibitor is preferred at 15–29. Apixaban is the least renally dependent at 27%, and its bleeding advantage over warfarin widens as CrCl falls. Edoxaban is paradoxically less effective than warfarin above CrCl 95 mL/min (FDA warning, 2015), so an alternative agent is preferable there. Below CrCl 15, the 2018 guide indicates no NOAC.
- ESC 2026 separates the bands more decisively. AF is about 50% more common at an eGFR of 30–59 and 2.4-fold more common below 30, with 11.6% prevalence on kidney replacement therapy. Existing ischaemic and bleeding risk scores are not recommended below eGFR 30 (Class III B and III C) because their c-statistics fall between 0.49 and 0.68 — these patients should be presumed high risk for both and are systematically undertreated. (sources/CVD-CKD-ESC-2026, rating: very high)
| Renal function | Anticoagulant | Class |
|---|---|---|
| eGFR ≥30 | DOAC over VKA — 14% fewer thromboembolic and 19% fewer bleeding events in a network meta-analysis of 170,059 patients | I A |
| eGFR 15–29 | Factor Xa inhibitor preferred, because only the factor Xa trials enrolled patients at this level | IIa B1 |
| eGFR <15, including dialysis | Individualised decision | IIa C |
- Below eGFR 15 the dose ladder becomes counter-intuitive. If anticoagulating: VKA or apixaban 2.5 mg twice daily below eGFR 15 off dialysis (IIb C), or a low-dose factor Xa inhibitor on dialysis (IIb C). The guideline also advises 5 mg twice daily on dialysis unless weight ≤60 kg or age ≥80 — a higher dose than it advises off dialysis at the same degree of kidney failure, and one that drops the serum-creatinine limb of the standard reduction criteria.
- In venous thromboembolism a DOAC is preferred over LMWH plus VKA at eGFR ≥30 (I B2; EINSTEIN major bleeding HR 0.54), with a factor Xa inhibitor at 15–29 (IIb C) and on dialysis (IIb C).
- Antiarrhythmic choice must account for QT and electrolytes (I C). Digoxin is about 75% renally cleared and needs major dose reduction below eGFR 30; amiodarone needs no adjustment and is not dialysable; dronedarone raises creatinine by roughly 9 µmol/L reversibly. See concepts/Chronic-Kidney-Disease.
Liver Disease
| Child-Pugh | Dabigatran | Apixaban | Edoxaban | Rivaroxaban |
|---|---|---|---|---|
| A (5–6 points) | Standard | Standard | Standard | Standard |
| B (7–9 points) | Caution | Caution | Caution | Avoid |
| C (10–15 points) | Avoid | Avoid | Avoid | Avoid |
(sources/noac-ehra-2018, rating: high)
All four are contraindicated in Child-Pugh C. Rivaroxaban is also contraindicated in Child-Pugh B, where exposure more than doubles.
Extremes of Body Weight
- Patients under 50 kg or over 120 kg are underrepresented in every trial, so the dose-response relationship in both directions is extrapolated rather than measured. (sources/noac-ehra-2018, rating: high)
- Trough-level monitoring may be warranted in this group; it is one of the few accepted indications for measurement outside an emergency.
Drug-Drug and Food Interactions
| Mechanism | Agents | Effect | Action |
|---|---|---|---|
| P-gp inhibition | Verapamil, dronedarone, amiodarone, quinidine, ticagrelor loading | Raises levels of all NOACs | Dronedarone with dabigatran is contraindicated; verapamil with dabigatran uses 110 mg BID |
| Combined strong CYP3A4 and P-gp inhibition | Ketoconazole, ritonavir | Raises rivaroxaban and apixaban levels | Avoid |
| CYP3A4 and P-gp induction | Rifampicin, carbamazepine, phenytoin, St John's wort | Markedly lowers rivaroxaban and apixaban levels | Avoid |
| Food | Rivaroxaban 15 mg and 20 mg | Area under the curve rises 39% with food | Must be taken with food; the others are unaffected |
(sources/noac-ehra-2018, rating: high)
Measuring Anticoagulant Effect
- PT and aPTT cannot quantify a NOAC. For dabigatran use the dilute thrombin time or ecarin clotting time (quantitative), with the thrombin time as a qualitative check — a normal thrombin time excludes dabigatran — and aPTT only qualitatively. For the factor Xa inhibitors use a calibrated anti-Xa chromogenic assay; do not convert a PT to an INR. (sources/noac-ehra-2018, rating: high)
| Drug at standard AF dose | Peak (ng/mL) | Trough (ng/mL) |
|---|---|---|
| Dabigatran | 64–443 | 31–225 |
| Apixaban | 69–321 | 34–230 |
| Edoxaban | 91–321 | 31–230 |
| Rivaroxaban | 184–343 | 12–137 |
- Routine monitoring is not recommended. Measurement is reserved for emergencies, uncertain adherence, extreme body weight and complex interaction regimens.
Follow-Up and Adherence
- Schedule: review at one month, then at least every three months. Bloods — renal and liver function, haemoglobin — yearly, six-monthly at age ≥75 or if frail, and every CrCl/10 months if CrCl is below 60. (sources/noac-ehra-2018, rating: high)
- HAS-BLED is validated in NOAC patients, and a high score never justifies withdrawing anticoagulation: stroke and bleeding risk travel together. A proton pump inhibitor is worth considering where GI bleeding risk is high — prior GI bleed or concomitant antiplatelet therapy.
- The effect wanes within 12–24 hours of the last dose, and a plasma level reflects only the preceding day or two, so it cannot establish adherence. Real-world adherence ranges from 38% to 99%, and discontinuation is lower with NOACs than with VKAs.
- Once-daily regimens generally achieve better adherence than twice-daily, though whether the difference translates into fewer thromboembolic events is unclear.
- Where adherence is the problem, treat it as the problem: education, an anticoagulation card, a nurse-led AF clinic, pharmacy monitoring, medication apps and electronic blister packs. If it remains poor despite all of these, converting to a VKA may be considered — not because VKA is superior, but because a monitored drug can be verified.
- SAMe-TT2R2 of 3 or more predicts poor INR control on VKA and is useful where a VKA must be used before NOAC access is available.
Cardioversion Anticoagulation
- AF lasting 48 hours or more, or of unknown duration: anticoagulate for at least three weeks before cardioversion, or use a TOE-guided early strategy with a single NOAC dose at least four hours beforehand. (sources/noac-ehra-2018, rating: high)
- After cardioversion, anticoagulate for at least four weeks regardless of stroke score.
- A LAA thrombus on TOE means do not cardiovert. Treat with a NOAC — rivaroxaban resolved 41.5% in X-TRA — and reassess.
- AF under 48 hours: a single NOAC dose 2–4 hours before may substitute for LMWH; use a TOE-guided strategy if the stroke score is ≥4.
- Long-term anticoagulation after cardioversion follows the stroke risk score, not the outcome of the procedure.
Starting Anticoagulation After Ischaemic Stroke
- Timing is set by infarct size, not by the calendar. ELAN (n=2,013) used imaging-defined severity: a DOAC within 48 hours for minor or moderate stroke and on day 6–7 for major stroke, against day 3–4, 6–7 and 12–14 in the later arm. The 30-day composite was 2.9% versus 4.1% (risk difference −1.18 percentage points, 95% CI −2.84 to 0.47), with symptomatic intracranial haemorrhage 0.2% in both arms and recurrent ischaemic stroke 1.4% versus 2.5%. (sources/elan-doac-timing-nejm-2023, rating: very high)
- ELAN tested no formal hypothesis — it was powered for precision, so the Class 2a recommendation rests on a confidence interval that excludes more than half a point of harm rather than on a demonstrated benefit. The trial excluded patients already therapeutically anticoagulated and those with parenchymal haematoma, and enrolled a population with a median NIHSS of 3. (sources/ais-aha-2026, rating: very high)
- Do not bridge with heparin while waiting — no benefit and more intracranial haemorrhage (COR 3). The severity-based table and the full trial detail are on concepts/AF-Special-Populations and entities/Ischemic-Stroke.
Missed Doses, Doubled Doses and Overdose
| Situation | Twice-daily agent | Once-daily agent |
|---|---|---|
| A dose is missed | May be taken up to 6 hours late; skip it if later, though at high stroke risk the window may be extended to the next dose | May be taken up to 12 hours late; skip it if later, with the same caveat |
| A double dose has been taken | Skip the next scheduled dose and restart 24 hours after the double dose | Continue the normal schedule |
| Unsure whether a dose was taken | Do not take an extra dose; continue the schedule | Take another dose if CHA₂DS₂-VASc ≥3; otherwise wait for the next |
(sources/noac-ehra-2018, rating: high)
- Overdose without bleeding: activated charcoal within 2–3 hours of ingestion, maintained diuresis, and dialysis if the drug is dabigatran — 50–60% is removable, while the factor Xa inhibitors are under 14% dialysable because of their protein binding. Specific reversal agents are reserved for confirmed high levels or bleeding.
Bleeding Management
| Severity | Action |
|---|---|
| Nuisance or minor | Delay up to one dose; treat the cause; local antifibrinolytics such as tranexamic acid |
| Major, not immediately life-threatening | Causal haemostasis; tranexamic acid 1 g IV every 6 hours; desmopressin 0.3 µg/kg IV; maintain diuresis; dialysis for dabigatran only |
| Life-threatening | Idarucizumab 5 g IV for dabigatran (RE-VERSE AD); andexanet alfa for the factor Xa inhibitors; activated or four-factor prothrombin complex concentrate if specific reversal is unavailable; fresh frozen plasma is not useful |
(sources/noac-ehra-2018, rating: high)
Switching Between Anticoagulants
| Transition | When to start the new agent |
|---|---|
| VKA → NOAC | At an INR below 2.0; the same or next day at 2.0–2.5; wait above 2.5 |
| NOAC → VKA | Co-administer until the INR is in range — for edoxaban, a half-dose bridge until three consecutive in-range readings |
| NOAC → intravenous UFH | When the next NOAC dose is due |
| NOAC → LMWH | When the next NOAC dose is due |
| Intravenous UFH → NOAC | 2–4 hours after stopping the infusion |
| LMWH → NOAC | At the next scheduled LMWH dose time |
| NOAC → NOAC | At the next dose time, allowing a longer interval where clearance is impaired |
(sources/noac-ehra-2018, rating: high)
Beyond DOACs — Factor XI Inhibition
- Not an option yet. The class was built to solve the bleeding problem this page is largely about — undertreatment driven by gastrointestinal bleeding — and it does reduce bleeding. What it does not yet do is prevent stroke comparably. Full detail: concepts/Factor-XI-Inhibition.
- The two trial results point opposite ways. Abelacimab reduced major or clinically relevant non-major bleeding by 62–69% against rivaroxaban in a 1,287-patient phase 2b trial that was never powered for stroke prevention (sources/abelacimab-azaleatimi71-nejm-2025, rating: high); asundexian reduced major bleeding by 68% against apixaban but produced 3.8 times as many strokes or systemic emboli in 14,810 patients, and the composite favoured apixaban (sources/asundexian-oceanicaf-nejm-2025, rating: very high).
- Practical consequence: a patient who cannot take a DOAC because of bleeding cannot be offered a factor XI inhibitor outside a trial. The evidence-backed alternatives remain a dose-reduced or alternative DOAC, warfarin, and left atrial appendage occlusion.
Contradictions / Open Questions
- No head-to-head NOAC trial exists in AF. Every comparison between agents is indirect or from network meta-analysis. COBRRA (2026) compares apixaban with rivaroxaban directly, but in venous thromboembolism, so AF choice remains inference from trials that each used a different comparator and population.
- Plasma-level monitoring has no outcome data. Level-guided dose adjustment is recommended on the reasoning that levels predict bleeding and thrombosis, not on a trial showing that adjusting to them prevents either. The expected ranges are the class's least evidenced number.
- Post-cardioversion timing for AF under 48 hours is undefined. The 48-hour threshold is not binary — AF lasting 12–48 hours still carries a thromboembolic risk at cardioversion — and no trial establishes a minimum period of anticoagulation for that band.
- NOAC after TAVI points the other way from the classification. The 2018 guide places TAVI with AF in Type 2, but GALILEO, published afterwards, found harm with rivaroxaban after TAVI, and an antiplatelet-based strategy may be preferred.
- Cancer and AF remains contested. The 2018 preference for LMWH is challenged by CARAVAGGIO and Hokusai-Cancer, while GI bleeding risk in GI tumours keeps NOACs the less certain option. (sources/noac-ehra-2018, rating: high; sources/Cancer-Associated-VTE)
- The apixaban ladder below eGFR 15 does not follow its own reduction rule. Standard reduction requires two of age ≥80, weight ≤60 kg and creatinine ≥133 µmol/L. ESC 2026 advises 2.5 mg twice daily for everyone below eGFR 15 off dialysis, and 5 mg twice daily on dialysis unless weight ≤60 kg or age ≥80 — a higher dose on dialysis than off it, and one that discards the creatinine limb. (sources/CVD-CKD-ESC-2026, rating: very high)
- Two stroke-risk scores are in circulation on this page. The EHRA guide and every trial quoted here use CHA₂DS₂-VASc; ESC 2024 decisions use CHA₂DS₂-VA, with the sex category removed. A score of 1 therefore means different things depending on which half of the page a reader is in. See concepts/CHA2DS2-VA.
- Factor XI inhibition has no successor role yet. The class was expected to uncouple thrombosis from haemostasis and did not survive its first phase 3 efficacy test — see concepts/Factor-XI-Inhibition for why.
Connections
- concepts/AF-CARE — the "avoid stroke" pillar this page implements
- concepts/CHA2DS2-VA — the score that sets the threshold for starting
- concepts/NOAC-Perioperative — interruption and resumption around procedures
- concepts/AF-Stable-CAD-Antithrombotic — combining with antiplatelet therapy after PCI
- concepts/AF-Special-Populations — cancer, ACS, post-operative AF and post-stroke initiation
- concepts/Subclinical-AF — device-detected AF and the ARTESIA and NOAH evidence
- concepts/Venous-Thromboembolism-Anticoagulation — the same drugs on different dose schedules
- concepts/Cancer-Associated-VTE — malignancy and anticoagulant choice
- concepts/Chronic-Kidney-Disease — the renal bands and the ESC 2026 recommendations
- entities/Ischemic-Stroke — post-stroke timing in full
- concepts/Factor-XI-Inhibition — the investigational class that has not displaced DOACs
- concepts/TAVI — the GALILEO signal after transcatheter replacement
Sources
Core guide
Guidelines
Trials
- sources/single-af-nejm-2026
- sources/elan-doac-timing-nejm-2023
- sources/abelacimab-azaleatimi71-nejm-2025
- sources/asundexian-oceanicaf-nejm-2025
Other