Wat is het optimale antitrombotisch beleid voor patiënten met atriumfibrilleren (met uitzondering van patiënten met mechanische hartkleppen of matige tot ernstige mitralisstenose)?

Deze uitgangsvraag bevat de volgende deelvragen:

• Bij welke patiënten met atriumfibrilleren zijn orale anticoagulantia geïndiceerd?
• Hoe dienen patiënten met atriumfibrilleren vervolgd te worden?
• Met welke anticoagulantia moeten patiënten met atriumfibrilleren behandeld worden?

Stroke risk assessment

Overall, AF increases the risk of stroke five-fold, but this risk is not homogeneous, depending on the presence of specific stroke risk factors/ modifiers. Main clinical stroke risk factors have been identified from non-anticoagulated arms of the historical RCTs conducted >20 years ago, notwithstanding that these trials only randomized < 10% of patients screened, whereas many common risk factors were not recorded or consistently defined (Pisters, 2012). These data have been supplemented by evidence from large observational cohorts also studying patients who would not have been included in the RCTs. Subsequently, various imaging, blood, and urine biological markers (biomarkers) have been associated with stroke risk (Pisters, 2012; Szymanski, 2015). In addition, non-paroxysmal AF is associated with an increase in thrombo-embolism (multivariable adjusted HR 1.38; 95% CI 1.19 to 1.61) compared with paroxysmal AF (Ganesan, 2016). Notably, many of the risk factors for AF-related complications are also risk factors for incident AF (Allan, 2017).

Common stroke risk factors are summarized in the clinical risk-factor-based CHA₂DS₂-VASc (Congestive heart failure, Hypertension, Age ≥ 75 years, Diabetes mellitus, Stroke, Vascular disease, Age 65 to 74 years, Sex category (female)) score (Table 1) (Lip, 2010).

Table 1 CHA₂DS₂-VASc score (source: Hindricks, 2020)

Stroke risk scores have to balance simplicity and practicality against precision (Killu, 2019; Rivera-Carvaca, 2017; Alkhouli, 2019). As any clinical risk-factor-based score, CHA₂DS₂-VASc performs only modestly in predicting high-risk patients who will sustain thrombo-embolic events, but those identified as low-risk (CHA₂DS₂-VASc 0 (males), or score of 1 (females)) consistently have low ischaemic stroke or mortality rates (< 1%/ year) and do not need any stroke prevention treatment.

Female sex is an age-dependent stroke risk modifier rather than a risk factor per se (Wu, 2020; Tomasdottir, 2019). Observational studies showed that women with no other risk factors (CHA₂DS₂-VASc score of 1) have a low stroke risk, similar to men with a CHA₂DS₂-VASc score of 0 (Friberg, 2012). The simplified CHA₂DS₂-VA score could guide the initial decision about OAC in AF patients, but not considering the sex component would underestimate stroke risk in women with AF (Overvad, 2019; Nielsen, 2020). In the presence of >1 non-sex stroke risk factor, women with AF consistently have significantly higher stroke risk than men (Nielsen, 2018; Marzona, 2018).

Many clinical stroke risk factors (for example renal impairment, OSA, LA dilatation (Vinereanu, 2017, Atrial Fibrillation Investigators, 1998; Friberg, 2015; Poli, 2017; Bassand, 2018)) are closely related to the CHA₂DS₂-VASc components, and their consideration does not improve its predictive value (the relationship of smoking or obesity to stroke risk in AF is also contentious) (Overvad, 2013). Various biomarkers (for example troponin, natriuretic peptides, growth differentiation factor (GDF)-15, von Willebrand factor) have shown improved performance of biomarker-based over clinical scores in the assessment of residual stroke risk among anticoagulated AF patients (Hijazi, 2017; Lip, 2006); notwithstanding, many of these biomarkers (as well as some clinical risk factors) are predictive of both stroke and bleeding (Hijazi, 2017) or non-AF and non-cardiovascular conditions, often (non-specifically) reflecting simply a sick heart or patient.

More complex clinical scores (for example Global Anticoagulant Registry in the FIELD - Atrial Fibrillation (GARFIELD-AF)) (Fox, 2017) and those inclusive of biomarkers (for example Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) (Zhu, 2017; Singer, 2013), Intermountain Risk Score Graves, 2018), ABC-stroke (Age, Biomarkers, Clinical history)) (Hijazi, 2016) improve stroke risk prediction modestly but statistically significantly. The ABC-stroke risk score that considers age, previous stroke/transient ischaemic attack (TIA), high sensitivity troponin T (cTnT-hs) and N-terminal (NT)-prohormone B-type natriuretic peptide has been validated in the cohorts of landmark DOAC-trials (Hijazi, 2017; Oldgren, 2016; Berg, 2019). A biomarker score-guided treatment strategy to reduce stroke and mortality in AF patients is being evaluated in an ongoing RCT (the ABC-AF Study, NCT03753490).

Whereas the routine use of biomarker-based risk scores currently would not substantially add to initial stroke prevention treatment decisions in patients already qualifying for treatment based on the CHA₂DS₂-VASc score (and a limited practicality would be accompanied by increased healthcare costs) (Rivera-Caravaca, 2017; Rivera-Caravaca, 2019; Esteve-Pastor, 2017), biomarkers could further refine stroke risk differentiation among patients initially classified as low risk and those with a single non-sex CHA₂DS₂-VASc risk factor (Shin, 2019).

Studies of the CHA₂DS₂-VASc score report a broad range of stroke rates depending on study setting (community versus hospital), methodology (for example excluding patients subsequently treated with OAC would bias stroke rates towards lower levels), ethnicity, and prevalence of specific stroke risk factors in the study population (different risk factors carry different weight, and age thresholds for initiating DOACs may even differ for patients with a different single non-sex stroke risk factor, as follows: age 35 years for HF, 50 years for hypertension or diabetes, and 55 years for vascular disease) (Chao, 2019; Nielsen, 2016). No RCT has specifically addressed the need for OAC in patients with a single non-sex CHA₂DS₂-VASc risk factor (to obtain high event rates and timely complete the study, anticoagulation trials have preferentially included high-risk patients), but an overview of subgroup analyses and observational data suggests that OAC use in such patients confers a positive net clinical benefit when balancing the reduction in stroke against the potential for harm with serious bleeding (Lip, 2015; Fauchier, 2016).

For many risk factors (for example age), stroke risk is a continuum rather than an artificial low-, moderate-, or high-risk category. Risk factors are dynamic and, given the elderly AF population with multiple (often changing) comorbidities, stroke risk needs to be re-evaluated at each clinical review. Recent studies have shown that patients with a change in their risk profile are more likely to sustain strokes (Chao, 2018; Yoon, 2018). Many initially low-risk patients (> 15%) would have ≥ 1 non-sex CHA₂DS₂-VASc risk factor at 1 year after incident AF (Chao, 2019; Potpara, 2012; Weijs, 2019), and 90% of new comorbidities were evident at 4.4 months after AF was diagnosed Chao, 2019).

A Patient-Centred Outcomes Research Institute (PCORI)-commissioned systematic review of 61 studies compared diagnostic accuracy and impact on clinical decision making of available clinical and imaging tools and associated risk factors for predicting thromboembolic and bleeding risk in AF patients (Borre, 2018). The authors concluded that the CHADS₂ (CHF history, Hypertension history, Age ≥ 75 y, Diabetes mellitus history, Stroke or TIA symptoms previously), CHA₂DS₂-VASc, and ABC risk scores have the best evidence for predicting thrombo-embolic risk (moderate strength of evidence for limited prediction ability of each score).

Bleeding risk assessment

When initiating antithrombotic therapy, potential risk for bleeding also needs to be assessed. Non-modifiable and partially modifiable bleeding risks (Table 2) are important drivers of bleeding events in synergy with modifiable factors (Chao, 2018). Notably, a history of falls is not an independent predictor of bleeding on OAC (a modelling study estimated that a patient would need to fall 295 times per year for the benefits of ischaemic stroke reduction with OAC to be outweighed by the potential for serious bleeding) (Man-Son-Hing, 1999).

Table 2 Risk factors for bleeding with OAC and antiplatelet therapy (source: Hindricks, 2020)

CKD-EPI= Chronic Kidney Disease Epidemiology Collaboration; CrCl = creatinine clearance; cTnT-hs = high-sensitivity troponin T; CYP = cytochrome P; GDF-15 = growth differentiation factor-15; INR = international normalized ratio; NSAID = non-steroidal anti-inflammatory drug; OAC = oral anticoagulant; SBP = systolic blood pressure; TTR = time in therapeutic range; VKA = vitamin K antagonist.

^aWalking aids; appropriate footwear; home review to remove trip hazards; neurological assessment where appropriate

^bIncreased INR monitoring, dedicated OAC clinicals, self-monitoring/self-management, education/behavioural interventions

^cFor patients receiving VKA treatment

^dDose adaptation based on patient’s age, body weight, and serum creatinine level

Modifiable and non-modifiable bleeding risk factors have been used to formulate various bleeding risk scores (Fox, 2017; Gage, 2006; Fang, 2011; O’Brien, 2015; Rohla, 2019; Pisters, 2010) generally with a modest predictive ability for bleeding events (Mori, 2019; Yao, 2017). Studies comparing specific bleeding risk scores provided conflicting findings (O’Brien, 2015; Rohla, 2019; Rutherford, 2018). Various biomarkers have been proposed as bleeding risk predictors, but many have been studied in anticoagulated trial cohorts (while bleeding risk assessment is needed at all parts of the patient pathway - when initially not using OAC, if taking aspirin, and, subsequently, on OAC). Additionally, biomarkers are non-specifically predictive of stroke, death, HF, et cetera (Thomas, 2017; Khan, 2019). or even non-cardiovascular conditions (for example glaucoma) (Ban, 2017), and the availability of some biomarkers is limited in routine clinical practice.

The biomarker-based ABC-bleeding risk score (Age, Biomarkers (GDF-15, cTnT-hs, haemoglobin) and Clinical history (prior bleeding)) (Berg, 2019; Hijazi, 2016) reportedly outperformed clinical scores, but in another study there was no long-term advantage of ABC-bleeding over HAS-BLED score (Table 3), whereas HAS-BLED was better in identifying patients at low risk of bleeding (HAS-BLED 0 - 2) (Esteve-Pastor, 2017). In the PCORI-commissioned systematic review (Borre, 2018), encompassing 38 studies of bleeding risk prediction, the HAS-BLED score had the best evidence for predicting bleeding risk (moderate strength of evidence), consistent with other systematic reviews and meta-analyses comparing bleeding risk prediction approaches (Caldeira, 2014; Zhu, 2015; Chang, 2020).

Table 3 Clinical risk factors in the HAS-BLED score (source: Hindricks, 2020)

ALP = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartate aminotransferase; SBP = systolic blood pressure; INR = international normalized ratio; NSAID = Non-steroidal anti-inflammatory drug; TTR = time in therapeutic range; VKA = vitamin K antagonist.

^aHaemorrhagic stroke would also score 1 point under the ‘B’ criterion

^bAlcohol excess or abuse refers to a high intake (for example > 14 units per week), where the clinician assesses there would be an impact on health or bleeding risk

A high bleeding risk score should not lead to withholding OAC, as the net clinical benefit of OAC is even greater amongst such patients. However, the formal assessment of bleeding risk informs management of patients taking OAC, focusing attention on modifiable bleeding risk factors that should be managed and (re)assessed at every patient contact, and identifying high-risk patients with non-modifiable bleeding risk factors who should be reviewed earlier (for instance in 4 weeks rather than 4 to 6 months) and more frequently (Chao, 2018; Lip, 2016). Identification of ‘high bleeding risk’ patients is also needed when determining the antithrombotic strategy in specific AF patient groups, such as those undergoing percutaneous coronary intervention (PCI).

Overall, bleeding risk assessment based solely on modifiable bleeding risk factors is an inferior strategy compared with formal bleeding risk assessment using a bleeding risk score (Chao, 2018; Guo, 2018; Esteve-Pastor,2017), thus also considering the interaction between modifiable and non-modifiable bleeding risk factors. Bleeding risk is dynamic, and attention to the change in bleeding risk profile is a stronger predictor of major bleeding events compared with simply relying on baseline bleeding risk. In a recent study, there was a 3.5-fold higher risk of major bleeding in the first 3 months amongst patients who had a change in their bleeding risk profile (Chao, 2018).

In the mAFA-II trial, prospective dynamic monitoring and reassessment using the HAS-BLED score (together with holistic App-based management) was associated with fewer major bleeding events, mitigated modifiable bleeding risk factors, and increased OAC uptake; in contrast, bleeding rates were higher and OAC use overall decreased by 25% in the ‘usual care’ arm when comparing baseline with 12 months (Guo, 2020).

Absolute contraindications to oral anticoagulants

The few absolute contraindications to OAC include active serious bleeding (where the source should be identified and treated), associated comorbidities (for example severe thrombocytopenia < 50 platelets/µL, severe anaemia under investigation, et cetera), or a recent high-risk bleeding event such as intracranial haemorrhage (ICH). Non-drug options may be considered in such cases (module 8.3).

Stroke prevention therapies

Vitamin K antagonists

Compared with control or placebo, vitamin K antagonist (VKA) therapy (mostly warfarin) reduces stroke risk by 64% and mortality by 26% (Hart, 2007), and is still used in many AF patients worldwide. VKAs are currently the only treatment with established safety in AF patients with rheumatic mitral valve disease and/or an artificial heart valve.

The use of VKAs is limited by the narrow therapeutic interval, necessitating frequent international normalized ratio (INR)monitoring and dose adjustments (De Caterina, 2013). At adequate time in therapeutic range ((TTR) > 70%), VKAs are effective and relatively safe drugs. Quality of VKA management (quantified using the TTR based on the Rosendaal method, or the percentage of INRs in range) correlates with haemorrhagic and thrombo-embolic rates (Wan, 2008). At high TTR values, the efficacy of VKAs in stroke prevention may be similar to DOACs, whereas the relative safety benefit with DOACs is less affected by TTR, with consistently lower serious bleeding rates (for example ICH) seen with DOACs compared with warfarin, notwithstanding that the absolute difference is small (Sjalander, 2018; Amin, 2014).

Numerous factors (including genetics, concomitant drugs, et cetera) influence the intensity of VKA anticoagulant effect; the more common ones have been used to derive and validate the SAMe-TT2R2 {Sex (female), Age (60 years), Medical history of ≥ 2 comorbidities (hypertension, diabetes mellitus, CAD/myocardial infarction, peripheral artery disease (PAD), HF, previous stroke, pulmonary disease, and hepatic or renal disease), Treatment (interacting drugs, for example amiodarone), Tobacco use, Race (non-Caucasian)) score (Apostolakis, 2013), which can help to identify patients who are less likely to achieve a good TTR on VKA therapy (score > 2) and would do better with a DOAC. If such patients with SAMe-TT₂R₂> 2 are prescribed a VKA, greater efforts to improve TTR, such as more intense regular reviews, education/counselling, and frequent INR monitoring are needed or, more conveniently, the use of a DOAC should be reconsidered (Proietti, 2015).

Direct oral anticoagulants

In four pivotal RCTs, apixaban, dabigatran, edoxaban, and rivaroxaban have shown non-inferiority to warfarin in the prevention of stroke/systemic embolism (Connoly, 2009; Patel, 2011; Granger, 2011; Giugliano, 2013). In a meta-analysis of these RCTs, DOACs were associated with a 19% significant stroke/systemic embolism risk reduction, a 51% reduction in haemorrhagic stroke (Ruff, 2014), and similar ischaemic stroke risk reduction compared with VKAs, but DOACs were associated with a significant 10% reduction in all-cause mortality (supplementary table 8). There was a non-significant 14% reduction in major bleeding risk, significant 52% reduction in ICH, and 25% increase in gastrointestinal bleeding with DOACs versus warfarin (Ruff, 2014).

The major bleeding relative risk reduction with DOACs was significantly greater when INR control was poor (i.e. centre-based TTR< 66%). A meta-analysis of the five DOAC-trials (RE-LY (Randomized Evaluation of Long Term Anticoagulant Therapy), 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), J-ROCKET AF, ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation), and ENGAGE AF TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation - Thrombolysis in Myocardial Infarction 48)) showed that, compared with warfarin, standard-dose DOACs were more effective and safer in Asians than in non-Asians (Wang, 2015). In the AVERROES (Apixaban Versus Acetylsalicylic Acid (ASA) to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment) trial of AF patients who refused or were deemed ineligible for VKA therapy, apixaban 5 mg b.i.d. (twice a day) significantly reduced the risk of stroke/systemic embolism with no significant difference in major bleeding or ICH compared with aspirin (Connolly, 2011).

Post-marketing observational data on the effectiveness and safety of dabigatran (Carmo, 2016; Huisman, 2018), rivaroxaban (Camm, 2016; Martinez, 2018), apixaban (Li, 2017), and edoxaban (Lee, 2018) versus warfarin show general consistency with the respective RCT. Given the compelling evidence about DOACs, AF patients should be informed of this treatment option.

Persistence to DOAC therapy is generally higher than to VKAs, being facilitated by a better pharmacokinetic profile of DOACs (Ingrasciotta, 2018) (Supplementary Table 9) and favourable safety and efficacy, especially amongst vulnerable patients including the elderly, those with renal dysfunction or previous stroke, and so on (Chao, 2018). Whereas patients with end-stage renal dysfunction were excluded from the pivotal RCTs, reduced dose regimens of rivaroxaban, edoxaban, and apixaban are feasible options for severe CKD (creatinine clearance (CrCl) 15 to 30 mL/min using the Cockcroft-Gault equation) (Stanton, 2017; Siontis, 2018). Considering that inappropriate dose reductions are frequent in clinical practice (Steinberg, 2016), thus increasing the risks of stroke/systemic embolism, hospitalization, and death, but without decreasing bleeding risk (Yao, 2017), DOAC therapy should be optimized based on the efficacy and safety profile of each DOAC in different patient subgroups.

Other antithrombotic drugs

In the ACTIVE W (Atrial Fibrillation Clopidogrel Trial with Irbesartan for Prevention of Vascular Events) trial, dual antiplatelet therapy (DAPT) with aspirin and clopidogrel was less effective than warfarin for prevention of stroke, systemic embolism, myocardial infarction, and vascular death (the annual risk of events was 5.6% versus 3.9%, P =0.0003), with a similar rate of major bleeding (Connolly, 2006). In the ACTIVE-A trial, patients unsuitable for anticoagulation had a lower rate of thrombo-embolic complications when clopidogrel was added to aspirin compared with aspirin alone, but with a significant increase in major bleeding (Connolly, 2009). Aspirin monotherapy was ineffective for stroke prevention compared with no antithrombotic treatment and was associated with a higher risk of ischaemic stroke in elderly patients (Sjalander, 2014).

Overall, antiplatelet monotherapy is ineffective for stroke prevention and is potentially harmful, (especially amongst elderly AF patients) (Mant, 2007; Lip, 2011), whereas DAPT is associated with a bleeding risk similar to OAC therapy. Hence, antiplatelet therapy should not be used for stroke prevention in AF patients.

Combination therapy with oral anticoagulant and antiplatelet drugs

The use of antiplatelet therapy remains common in clinical practice, often in patients without an indication (for example PAD, CAD, or cerebrovascular disease) beyond AF (Verheugt, 2018). There is limited evidence to support the combination therapy solely for stroke prevention in AF, with no effect on reductions in stroke, myocardial infarction, or death, but with a substantial increase in the risk of major bleeding and ICH (Mant, 2007; Lip, 2011).

Long-term oral anticoagulation per atrial fibrillation burden

Although the risk of ischaemic stroke/systemic embolism is higher with non-paroxysmal versus paroxysmal AF, and AF progression is associated with an excess of adverse outcomes (Potpara, 2012; Ogawa, 2018), the clinically determined temporal pattern of AF should not affect the decision regarding long-term OAC, which is driven by the presence of stroke risk factors (Ganesan, 2016). Management of patients with AHRE/subclinical AF is reviewed in module 10. Stroke risk in AHRE patients may be lower than in patients with diagnosed AF (Mahajan, 2018), and strokes often occur without a clear temporal relationship with AHRE/subclinical AF (Brambatti, 2014; Boriani, 2014), underscoring its role as a risk marker rather than a stroke risk factor (Freedman, 2017a; Freedman, 2017b). Whether AHRE and subclinical AF have the same therapeutic requirements as clinical AF 7 is presently unclear, and the net clinical benefit of OAC for AHRE/subclinical AF > 24 h is currently being studied in several RCTs (Freedman, 2017).

Notably, patients with subclinical AF/AHRE may develop atrial tachyarrhythmias lasting more than 24 h (Van Gelder, 2017) or clinical AF; hence careful monitoring of these patients is recommended, even considering remote monitoring, especially with longer AHRE and higher risk profile (Boriani, 2018). Given the dynamic nature of AF as well as stroke risk, a recorded duration in one monitoring period would not necessarily be the same in the next.

Long-term oral anticoagulation per symptom control strategy

Symptom control focuses on patient-centred and symptom-directed approaches to rate or rhythm control. Again, symptom control strategy should not affect the decision regarding long-term OAC, which is driven by the presence of stroke risk factors, and not the estimated success in maintaining sinus rhythm.

Management of anticoagulation-related bleeding risk

Strategies to minimize the risk of bleeding

Ensuring good quality of VKA treatment (TTR> 70%) and selecting the appropriate dose of a DOAC (as per the dose reduction criteria specified on the respective drug label) are important considerations to minimize bleeding risk. As discussed in Bleeding risk assessment, attention to modifiable bleeding risk factors should be made at every patient contact, and formal bleeding risk assessment is needed to help identify high-risk patients who should be followed up or reviewed earlier (for example 4 weeks rather than 4 to 6 months) (Lip, 2016). Concomitant regular administration of antiplatelet drugs or non-steroidal anti-inflammatory drug (NSAID) should be avoided in anticoagulated patients. Bleeding risk is dynamic, and attention to the change in bleeding risk profile is a stronger predictor of major bleeding events, especially in the first 3 months (Chao, 2018).

High-risk groups

Certain high-risk AF populations have been under-represented in RCTs, including the extreme elderly (≥ 90 years), those with cognitive impairment/dementia, recent bleeding or previous ICH, end-stage renal failure, liver impairment, cancer, and so on. Observational data suggest that such patients are at high risk for ischaemic stroke and death, and many would benefit from OAC.

Patients with liver function abnormalities may be at higher risk of bleeding on VKA, possibly less so on DOACs. Observational data in cirrhotic patients suggest that ischaemic stroke reduction may outweigh bleeding risk (Pastori, 2018; Kuo, 2017; Lee, 2019).

In patients with a recent bleeding event, attention should be directed towards addressing the predisposing pathology (for example bleeding ulcer or polyp in a patient with gastrointestinal bleeding), and the reintroduction of OAC as soon as feasible, as part of a multidisciplinary team decision. Consideration should be made for drugs such as apixaban or dabigatran 110 mg b.i.d., which are not associated with an excess of gastrointestinal bleeding compared with warfarin. Where OAC is not reintroduced, there is a higher risk of stroke and death compared with restarting OAC, although the risk of rebleeding may be higher (Staerk, 2015). Similarly, thromboprophylaxis in cancer may require a multidisciplinary team decision balancing stroke reduction against serious bleeding, which may be dependent on cancer type, site(s), staging, anti-cancer therapy and so on.

Thromboprophylaxis in specific high-risk groups is discussed in detail throughout module 8.

Decision making to avoid stroke

In observational population cohorts, both stroke and death are relevant endpoints, as some deaths could be due to fatal strokes (given that endpoints are not adjudicated in population cohorts, and cerebral imaging or post-mortems are not mandated). As OAC significantly reduces stroke (by 64%) and all-cause mortality (by 26%) compared with control or placebo (Hart, 2007), the endpoints of stroke and/or mortality are relevant in relation to decision making for thromboprophylaxis.

The threshold for initiating OAC for stroke prevention, balancing ischaemic stroke reduction against the risk of ICH and associated QoL, has been estimated to be 1.7%/year for warfarin and 0.9%/year for a DOAC (dabigatran data were used for the modelling analysis) (Eckman, 2011). The threshold for warfarin may be even lower, if good-quality anticoagulation control is achieved, with average TTR> 70% (Proietti, 2016). Given the limitations of clinical risk scores, the dynamic nature of stroke risk, the greater risk of stroke and death among AF patients with ≥ 1 non-sex stroke risk factor, and the positive net clinical benefit of OAC among such patients, we recommend a risk-factor-based approach to stroke prevention rather than undue focus on (artificially defined) ‘high-risk’ patients. As the default is to offer stroke prevention unless the patient is low risk, the CHA₂DS₂-VASc score should be applied in a reductionist manner, to decide on OAC or not (Lip, 2016).

Thus, the first step in decision making (‘A’ Anticoagulation/Avoid stroke) is to identify low-risk patients who do not need antithrombotic therapy. Step 2 is to offer stroke prevention (i.e., OAC) to those with ≥ 1 non-sex stroke risk factors (the strength of evidence differs, with multiple clinical trials for patients with ≥ 2 stroke risk factors, and subgroups from trials/observational data on patients with 1 non-sex stroke risk factor). Step 3 is the choice of OAC-a DOAC (given their relative effectiveness, safety and convenience, these drugs are generally first choice as OAC for stroke prevention in AF) or VKA (with good TTR at > 70%). This ‘AF 3-step’ patient pathway (Freedman, 2016; Lip, 2015) for stroke risk stratification and treatment decision making is shown in Figure 1.

 Figure 1 A - Anticoagulation/Avoid stroke: The ‘AF 3-step’ pathway (source: Hindricks, 2020)

AF = atrial fibrillation; CHA2DS2-VASc = Congestive heart failure, Hypertension, Age ≥ 75 years, Diabetes mellitus, Stroke, Vascular disease, Age 65 to 74 years, Sex category (female); HAS-BLED = Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile INR, Elderly (> 65 years), Drugs/alcohol concomitantly; INR = international normalized ratio; NOAC = non-vitamin K antagonist oral anticoagulant (referred to as DOAC in this guideline); OAC = oral anticoagulant; SAMe-TT2R2 = Sex (female), Age (< 60 years), Medical history, Treatment (interacting drug(s)), Tobacco use, Race (non-Caucasian) (score); TTR = time in therapeutic range; VKA = vitamin K antagonist

^aIf a VKA being considered, calculate SAMe-TT2R2 score: if score 0 to 2, may consider VKA treatment (for example warfarin) or DOAC; if score > 2, should arrange regular review/frequent INR checks/ counselling for VKA users to help good anticoagulation control, or reconsider the use of DOAC instead; TTR ideally >70%