8. Stroke Prevention

8.1 Stroke risk assessment

Observations from the Framingham cohort and subsequent clinical trials revealed that NVAF is an independent risk factor for stroke (annual incidence of approximately 4.1%-4.5%) and combined stroke/systemic embolism (annual incidence of 50%).^[70],[169] The risk was further refined by the delineation of various baseline characteristics that might affect the risk of the stroke.^{[169]–[173]} The first widely adopted tool for stroke risk assessment was the Congestive Heart Failure, Hypertension, Age, Diabetes, Stroke/Transient Ischemic Attack (CHADS2) score, which assigns a single point for HF, hypertension, age 75 years or older, and diabetes, and 2 points for previous stroke/systemic embolism.^[26],[172] Unfortunately, CHADS2 was unable to adequately differentiate very low risk individuals (ie, in whom OAC is associated with a greater risk than benefit) from those at low but still clinically important stroke risk. This led to the development of the expanded Congestive Heart Failure, Hypertension, Age (75 years), Diabetes, Stroke/Transient Ischemic Attack, Vascular Disease, Age (65-74 years), Sex (Female) (CHA2DS2-VASc) score. This score includes the additional risk factors of vascular disease (1 point), female sex (1 point), and age between 65 and 74 years (1 point), and also increases the risk weight to 2 points for age 75 years or older.^[26],[171] The 2010 CCS guidelines recommended stroke prevention therapies^[4] on the basis of the CHADS2 whereas the 2012 update^[5] differentiated among patients at low CHADS2 risk on the basis of risk factors derived from the CHA2DS2-VASc.^[174] The algorithm for antithrombotic prescription was further revised in 2014,^[175] on the basis of data from a Danish national cohort study, which showed that the annual risk of “stroke” (defined as a thromboembolic event precipitating hospitalization or death) was 2.1% for patients aged 65-74 years and 4.4% for those 75 years of age or older.^[174] Because age could be reliably determined in all patients, it was agreed that age 65 years or older should be the starting point for a revised algorithm.^[175] Likewise, OAC was justified for younger patients with any CHADS2 risk factors (annual risk of stroke: 2.4% with HF, 1.6% with hypertension, 2.3% with diabetes, and 7.9% with previous stroke/systemic embolism). The CCS Guidelines Committee judged that antiplatelet therapy has no role in AF-related stroke prevention and, therefore, no antithrombotic therapy was recommended for patients younger than 65 years with none of the CHADS2 risk factors. However, in the presence of vascular disease daily antiplatelet therapy is indicated to prevent ischemic vascular events independent of the presence of AF.^[176] As such, antiplatelet therapy is only indicated in the presence of established vascular disease in NVAF patients aged younger than 65 years with no CHADS2 risk factors (see section 8.3.2.1). The CCS algorithm (CHADS-65) is presented in (Fig. 8). Among several organizations that publish guidelines for antithrombotic therapies for patients with AF, there are variations in the definitions of the component risk factors, and in the selection and interpretation of data on the stroke risk associated with the individual risk factors according to their preferred risk schemes.^{[10],[26],[177]} These differences in definitions are unlikely to affect the categorization of patient stroke risk, however, they might provide different point estimates of the annual risk of stroke. It has also become evident that the risk of stroke varies among cohorts reported from different countries.^[178] For similar baseline characteristics, the Taiwanese^[179] and Danish cohorts^[174] appear to have particularly high stroke risk, whereas cohorts from Sweden^[180] and the United States^[181] appear to be lower risk. The variation might be attributable to the dissimilarities in the definitions of stroke, treatment of comorbidities, and the protocols for data collection.

8.1.1 Atrial flutter

Atrial flutter (AFL) carries a significant risk of stroke/systemic embolism but, as opposed to AF, the relationship is somewhat less precisely established due to limited number of small and retrospective cohort studies, as well as significant heterogeneity of data.^[182] Because many patients with AFL also experience episodes of AF it becomes difficult to know the exact risk of stroke/systemic embolism related to AFL alone.^[183] Moreover, there are no randomized trials of the value of OAC specifically in the AFL population. However, there is direct and indirect evidence from mechanistic, observational, and prospective studies that AFL confers a significant thromboembolic risk. TEE evidence of atrial thrombi has been documented in patients with sustained AFL not receiving long-term anticoagulation (LA thrombus in 1.6% with at least moderate spontaneous LA echo contrast in 13%; median duration of 33 days).^[184] OAC has been shown to reduce the incidence of stroke/systemic embolism after cardioversion, with an incidence similar to that of AF.^[185],[186] In a metaregression analysis, the annual risk of stroke/systemic embolism was approximately 3% in patients with AFL.^[187] Although the rate of stroke is higher than that in control participants (HR, approximately 1.4), the risk of stroke appears to be lower than for patients with AF (HR, approximately 0.70).^[183],[188] For AFL patients, the risk of stroke is higher with increasing CHA2DS2-VASc scores^{[189]–[192]} and also increased in patients who develop AF after their initial diagnosis of AFL.^[192] As such, it is recommended that patients with AFL be stratified and treated in the same manner as patients with AF.^[193]

8.2 Oral anticoagulation

The benefit of OAC must be weighed against the risk of hemorrhage. The relative importance of a stroke prevented, and a major bleed caused is a subjective judgement. There is considerable scope for physician-patient discussion (ie, shared decision-making) to ensure that patient values are concordant with the decision to prescribe OAC, particularly when the annual risk of stroke is < 2% per year.^[194] The CCS rationale for recommending OAC for most patients with age 65 years or older or CHADS2 score 1 is on the basis of the effects of OAC on the absolute risk reduction of stroke compared with the increase in major hemorrhage. In patients aged 65 years or older and without other risk factors for stroke, use of VKAs decreased the annual risk of stroke from 2.1% to 0.7% while it increased the risk of major bleeding by approximately 0.5% per year to 1.0%.^[51],[195] Although the risk of major bleeding increases with increasing CHADS2 scores, the rate of rise is not as steep as that for stroke; therefore, the benefit to risk ratio for OAC increases as stroke risk factors accumulate. Furthermore, 70% of strokes result in death or major disability, whereas most patients survive major hemorrhage without long-term effects.^[196] Thus, these results favour use of OAC in patients with age 65 years or older or CHADS2 score 1. OACs with efficacy and safety evidence in AF include VKAs and DOACs.

8.2.1 Warfarin and vitamin K antagonists

Six studies (2900 patients) have compared VKAs with placebo in the setting of NVAF; 5 primary prevention trials (2 doubleblinded) and 1 secondary prevention trial. The mean achieved international normalized ratio (INR) ranged from 2.0 to 2.6 among patients who were assigned to VKAs in the primary prevention trials and was 2.9 in the secondary prevention trial. A meta-analysis of these 6 studies showed a RR reduction of 64% (95% CI, 49%-74%) in all stroke (ischemic or hemorrhagic)^[51] with an absolute risk reduction of 2.7% per year for primary prevention trials and 8.4% per year for secondary prevention. The absolute risk increase in major extracranial hemorrhage was 0.3% per year with use of VKAs. However, there was an absolute risk reduction of 1.6% per year in overall mortality with use of VKAs. In a meta-analysis of 8 RCTs (3647 patients), the RR reduction with use of VKAs over aspirin was 39% (95% CI, 19%-53%) for all strokes, equivalent to an absolute risk reduction of approximately 0.7% per year for primary prevention and 7% per year for secondary prevention.^[51] There was an excess of bleeding in the VKA-treated patients leading to an absolute risk increase of 0.2% per year in major extracranial and intracranial hemorrhage (ICH). However, the absolute risk reduction for all-cause mortality remained in favour of VKAs at 0.5% per year. In an update of this meta-analysis to include the largest study of elderly patients^[197] the absolute risk reduction increased to 0.9%, with the risk of major extracranial hemorrhage being no different between use of VKA and aspirin.^[51] A further meta-analysis showed that adjusted-dose VKA was associated with half the rate of stroke/systemic embolism events with no significant difference in the rate of major bleeding relative to adjusted-dose antiplatelet therapy.^[198] The benefit of VKAs for stroke prevention in patients with NVAF is optimized at a target INR of 2-3, with the time spent in this therapeutic range (TTR) directly correlated with clinical outcomes. The relationship between TTR and clinical outcomes was shown in a post hoc analysis of the AF Clopidogrel Trial With Irbesartan for Prevention of Vascular Events (ACTIVE W) trial.^[199] In ACTIVE W there was no difference in stroke reduction between VKA and clopidogrel with aspirin therapy (RR, 0.93; 95% CI, 0.70-1.24; P ¼ 0.61) at centres with a TTR below the median TTR (65%). However, VKAs had a marked benefit for patients at centres with a TTR above the median, reducing vascular events by more than twofold (RR, 2.14; 95% CI, 1.61- 2.85; P < 0.0001). A subsequent systematic review reinforced this concept, showing a significant correlation between TTR and adverse outcomes (major hemorrhage and thromboembolic rate) in an analysis of 33,976 AF patients included in 38 studies.^[200] More recently, a large worldwide observational study of 9934 VKA-treated patients enrolled from 35 countries between 2010 and 2015 showed an adjusted 2.6-fold increased risk of stroke/ systemic embolism, a 1.5-fold increased risk of major bleeding, and a 2.4-fold increased risk of all-cause mortality with TTR < 65% vs 65%.^[201] Further, the study highlighted the difficulty in managing VKAs because the mean TTR for the total study group was 55%, with more than half of the North American cohort having a TTR < 65%.^[201] VKAs such as warfarin remain the preferred OAC in the presence of any mechanical prosthetic heart valve and in patients with moderate to severe mitral stenosis. With respect to the former, in the Randomized, Phase II Study to Evaluate the Safety and Pharmacokinetics of Oral Dabigatran Etexilate in Patients After Heart Valve Replacement (RE-ALIGN) study patients with mechanical heart valves were randomized to VKA or dabigatran (150 mg, 220 mg, or 300 mg twice daily [BID], on the basis of renal function).^[202] The trial was terminated early because of an excess of thromboembolic and bleeding events with dabigatran. Although RE-ALIGN is the only trial to evaluate DOACs in the setting of mechanical valve prostheses, VKA remains the treatment of choice in this population. With respect to patients with moderate to severe mitral stenosis, there are no randomized trials of OAC for prevention of thromboembolic events. Retrospective studies have suggested a 4- to 15-fold decrease in the incidence of embolic events with OAC therapy.^[203] Because patients with severe mitral valve disease were excluded from the pivotal randomized DOAC trials,^[21]–[25] VKAs remain the standard of care in this patient population until further evidence emerges. Ongoing trials of DOACs in patients with rheumatic valvular disease (Investigation of Rheumatic AF Treatment Using Vitamin K Antagonists, Rivaroxaban or Aspirin Studies [INVICTUS] studies; NCT02832531 and NCT02832544) will provide guidance.

8.2.2 Non-vitamin K direct oral anticoagulants

Four DOACs are currently approved for use in Canada: apixaban, dabigatran, edoxaban, and rivaroxaban. These four agents were developed to overcome the major limitations associated with VKAs and have been evaluated in large RCTs involving more than 70,000 patients.^[21]–[25] Individually, DOACs were shown to be at least as effective as VKAs in decreasing the risk of NVAF-associated stroke/systemic embolism, with similar or less major bleeding. A meta-analysis of these four RCTs showed that the use of the higher approved DOAC dose resulted in a statistically significant reduction in stroke/systemic embolism (RR, 0.81; 95% CI, 0.73-0.91; P < 0.0001), ICH (RR, 0.48; 95% CI, 0.39-0.59; P < 0.0001), and all-cause mortality (RR, 0.90; 95% CI, 0.85-0.95; P ¼ 0.0003) with less major bleeding (RR, 0.85; 95% CI, 0.73-1.00; P ¼ 0.06), compared with VKAs.^[52] Use of the lower-dose DOAC regimens resulted in similar rates of stroke/systemic embolism (RR, 1.03; 95% CI, 0.84-1.27; P ¼ 0.74), with less major (RR, 0.65; 95% CI, 0.43-1.00; P ¼ 0.05) and intracranial bleeding (RR, 0.31; 95% CI, 0.24-0.41; P < 0.0001), and lower mortality (RR, 0.89; 95% CI, 0.83-0.96; P ¼ 0.003), but significantly more ischemic strokes (RR, 1.28; 95% CI, 1.02-1.60; P ¼ 0.045) compared with VKAs. On the basis of these observations the CCS AF Guidelines Committee continues to recommend DOACs over VKAs for patients with NVAF. There have been no published RCTs directly comparing the four DOACs. Differences in the designs of the RCTs, including the population studied (and by extension their baseline stroke risk), preclude definitive comparisons between the agents. Although so-called “real world” comparisons are plentiful in the published literature, it is not possible to make reliable therapeutic inferences from observational associations because these types of studies are subject to significant biases and limitations.^[204] Nevertheless, there are relevant differences in the pharmacological characteristics of the DOACs, which might influence their selection in individual patients such as: (1) bioavailability, with dabigatran being the least bioavailable, thereby requiring a special formulation for optimal absorption, and rivaroxaban requiring coadministration with food to optimize absorption of the 15- and 20-mg doses; (2) renal clearance, with dabigatran being predominantly (80%) renally cleared; (3) drug-drug interaction potential, with all of the DOACs influenced by P-glycoprotein interactions but only rivaroxaban and apixaban also being influenced by cytochrome P-450 isoenzyme function, specifically 3A4; (4) the elimination half-life and dosing schedule; and, (5) the presence and availability of an antidote. Regardless of the specific DOAC agent selected, it is important to ensure that the prescribed dose is consistent with Health Canada labelling and the product monograph (Table 7). Specifically, it has been observed that overtreatment (or the prescription of a standard dose in patients with an indication for dose reduction) occurs in approximately 4% of patients and is associated with an increased risk of major bleeding, hospitalization, and death, without a significant additional reduction in stroke.^{[205]–[207]} Likewise, undertreatment (or prescription of a reduced dose without an indication to do so) occurs in 12%-15% of patients and is associated with a higher risk of thromboembolic events, hospitalization, and death, without a significant reduction in major bleeding.^{[205]–[207]}

8.3 Anticoagulation in special populations

8.3.1 Anticoagulation in patients with chronic kidney disease and end-stage renal disease

AF patients with chronic kidney disease (CKD) represent a particularly high-risk subgroup because stroke, mortality, and major bleeding all increase as renal function deteriorates.^[208],[209] RCTs demonstrate that OAC, including DOACs, are safe and effective for AF patients with mild to moderate CKD (stage 1-3 CKD or eGFR > 30 mL/min).^[210] There are few randomized data for patients with severe renal impairment (stage 4-5 CKD or eGFR < 25 to 30 mL/min) as these patients were excluded from the large phase III RCTs. Nonrandomized and very limited randomized data support the use of OAC in patients with stage 4 CKD (eGFR 15-30 mL/min),^[211],[329] however, the benefit of OAC for AF patients with severe CKD (eGFR < 15 mL/min) or end-stage renal disease (ESRD) who require dialysis remains unclear. There are no prospective randomized trials that have evaluated OACs vs placebo for dialysis-dependent AF patients, and results of the observational studies are conflicting.^{[209],[212]–[215]} The ongoing randomized Study of the Benefit/Risk Ratio of Oral Anticoagulation in Hemodialysis Patients with Atrial Fibrillation (AVKDIAL) is studying VKA (target INR 2-3) compared with no OAC in hemodialysis patients with AF (NCT02886962). Until these results are available there is insufficient evidence to support or deny routine OAC use in the AF population with ESRD who require dialysis. When OAC is prescribed for AF patients with stage 1-4 CKD, the CCS AF guidelines panel recommends that a DOAC is preferred to a VKA. For those with moderate CKD in the landmark phase III trials, DOACs significantly reduced the risk of stroke/systemic embolism (RR, 0.79; 95% CI, 0.66-0.94) and major bleeding (RR, 0.80; 95% CI, 0.70-0.91), compared with VKAs.^[216] Moreover, in patients with mild-moderate CKD the use of DOACs was associated with a lessened rate of adverse renal outcomes, including renal function decline, doubling in serum creatinine level, or acute kidney injury.^[217] Although apixaban and rivaroxaban are approved for patients with stage 4 CKD (CrCl of > 15 mL/min), the evidence supporting DOAC use in preference to VKAs is limited.^[218] In those with more severe stage 5 CKD or ESRD requiring dialysis the evidence supporting DOACs is incomplete, with the available observational data being subject to confounding and selection bias and the randomized studies being underpowered.^{[219]–[222],[885]} The randomized Renal Hemodialysis Patients Allocated Apixaban versus VKA in AF trial (RENAL-AF; NCT02942407) was prematurely terminated following the enrollment of 154 patients with hemodialysisdependent ESRD (targeted enrollment, 760 patients). At study conclusion, the rates of bleeding and stroke/systemic embolism were similar between those randomized to apixaban 5 mg BID or VKAs. Ongoing studies include the Compare Apixaban and Vitamin-K Antagonists in Patients With Atrial Fibrillation (AF) and End-Stage Kidney Disease (ESKD) (AXADIA; NCT02933697), in which phenprocoumon vs is being compared to apixaban 2.5 mg twice daily (targeted enrollment 222 patients); and the Strategies for the Management of Atrial Fibrillation in Patients Receiving Dialysis (SAFE HD; NCT03987711) trial, in which VKA and apixaban 5 mg BID, and no anticoagulation are being compared (targeted enrollment 150 patients). Patients with clinically significant CKD require a reduction in DOAC dose in accordance with the approved prescribing information to avoid adverse clinical outcomes (Table 7).7,^{[205]–[207],[223]} Although undertreatment (or prescription of a reduced dose without an indication to do so) is associated with a higher risk of thromboembolic events, hospitalization, and death,^[205]-[207] appropriate dose reduction in those with moderate CKD can achieve clinical outcomes comparable with that in patients with preserved renal function (CrCl > 50 mL/min) who are receiving the higher DOAC dose.^[224],[225] When medication dose adjustment is performed, CockcroftGault CrCl should be used. The rationale has been extensively outlined in the 2014 AF guidelines companion.^[26] In brief, the Cockcroft-Gault CrCl was the formula used to assess eligibility in the landmark DOAC trials, and it is recommended to guide medication dose adjustment in the product monographs and approved prescribing information.^{[21]–[23],[25]} Although the Modified Diet in Renal Disease (MDRD) or the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) eGFR equations are commonly reported by laboratories, use of these formulae fail to identify a significant proportion of patients with contraindications to DOACs or those who require dose adjustment.^[226]

8.3.2 Coronary artery disease

Up to 20%-30% of AF patients also have concomitant coronary artery disease (CAD), with a significant proportion who require percutaneous coronary intervention (PCI).^[227],[228] OAC is indicated for the prevention of AF-related stroke/systemic embolism, which also provides benefit in preventing ischemic coronary events. Antiplatelet therapy is indicated for the prevention of coronary events after acute coronary syndromes (ACS) and/or PCI; however, it is inferior to OAC for the prevention of stroke/systemic embolism in an AF population at increased risk of AF-related stroke.^[229] As such, management requires a careful and balanced assessment of the individual risks of bleeding vs the anticipated effect on thrombotic outcomes. To clarify potentially confusing terminology in this area, single-agent antiplatelet therapy (SAPT) refers to the use of a single antiplatelet drug (eg, acetylsalicylic acid [ASA]), dual antiplatelet therapy (DAPT) refers to the concomitant use of 2 antiplatelet agents (eg, ASA with P2Y12 inhibitor), dual pathway therapy refers to the concomitant use of a SAPT with an OAC agent (eg, VKA with P2Y12 inhibitor), and triple antithrombotic therapy (TT), the combination of DAPT with an OAC (eg, VKA with ASA and P2Y12 inhibitor). The comprehensive recommendations regarding antithrombotic treatment in AF patients indicated for OAC with concomitant coronary/arterial vascular disease are summarized in Figure 9.

8.3.2.1 Stable vascular disease and AF in patients at low risk of stroke/systemic embolism

SAPT (eg, ASA 81 mg/d) is recommended for patients with AF who are at low risk of stroke/systemic embolism (age younger than 65 years and CHADS2 score of 0) if coronary or peripheral arterial vascular disease is present (CAD, peripheral vascular disease, or aortic plaque). The SAPT recommendation is on the basis of the efficacy of ASA therapy for the prevention of coronary events among patients with stable CAD (1.5% per year absolute risk reduction for secondary prevention).^[51],[230] Although there is extensive evidence for the efficacy of OAC for prevention of ischemic coronary events in patients with stable CAD,^[231] the CCS AF guidelines recommend SAPT in preference to OAC in those at low risk of stroke because of the favourable safety profile and ease of use associated with antiplatelet therapy. In a non-AF population, the Cardiovascular Outcomes for People Using Anticoagulation Strategies (COMPASS) trial showed that dual pathway therapy with ASA and “vascular dose” rivaroxaban (2.5 mg BID) was associated with significant reduction in cardiovascular mortality and ischemic stroke, albeit with a significantly increased risk of major bleeding.^[232] As such, the combination of ASA and rivaroxaban 2.5 mg BID may be considered a reasonable alternative to ASA alone for NVAF patients at low risk of stroke (age younger than 65 years and CHADS2 score of 0) who also have coronary or arterial vascular disease.

8.3.2.2 Stable vascular disease and AF in patients at high risk of stroke/systemic embolism

OAC is indicated for stroke prevention in patients with AF who are aged 65 years or older or with a CHADS2 score 1. When such a patient also has stable CAD (defined by the absence of ACS or PCI in the preceding 12 months), OAC provides protection against ischemic coronary events in addition to stroke and systemic embolism.^{[233]–[239]} The CCS AF Guidelines Committee does not recommend that antiplatelet agents be routinely prescribed in combination with OAC for AF patients at high risk of stroke with stable CAD. The Warfarin-Aspirin Reinfarction Study (WARIS)-II showed that the additional use of antiplatelet therapy with OAC did not confer a beneficial effect (eg, combined end point of death, MI, and stroke; 16.7% with OAC alone vs 15.0% with combination OAC and ASA; P ¼ 0.18), but did increase the risk of adverse bleeding outcomes (overall bleeding 2.82% per year with OAC alone vs 3.27% per year with combination OAC and ASA).^[236],[237] More recently, the Atrial Fibrillation and Ischemic Events With Rivaroxaban in Patients With Stable Coronary Artery Disease (AFIRE) trial^[239] randomized 2236 patients with AF with stable CAD to receive rivaroxaban monotherapy or combination therapy with rivaroxaban and SAPT. The trial was prematurely terminated because of increased all-cause mortality in the combination therapy group (1.9% per patient-year with rivaroxaban vs 3.4% per patient-year with combination therapy; HR, 0.55; 95% CI, 0.38-0.81; P < 0.05). Rivaroxaban monotherapy was noninferior to combination therapy for the primary efficacy end point (stroke/systemic embolism, MI, unstable angina requiring revascularization, or death)with event rates of 4.14% and 5.75% per patient-year, respectively (HR, 0.72; 95% CI, 0.55-0.95; P < 0.001 for noninferiority). Rivaroxaban monotherapy was superior to combination therapy for the primary safety end point of International Society on Thrombosis and Haemostasis major bleeding (1.62% vs 2.76% per patient-year; HR, 0.59; 95% CI, 0.39-0.89; P ¼ 0.01 for superiority).

8.3.2.3 PCI or ACS in patients with Af at low risk of stroke/systemic embolism

OAC is not recommended for stroke prevention for patients with AF who are at low risk of stroke/systemic embolism (age younger than 65 years and CHADS2 score of 0). DAPT is generally prescribed for patients after elective PCI for a period of 6 months and for 12 months after ACS (with or without PCI), as outlined in the 2018 CCS/CAIC antiplatelet guidelines for non-AF patients.^[240] Some patients with low thrombotic or high bleeding risk may appropriately receive shorter durations of DAPT to decrease the risk of major bleeding. Conversely, longer durations of DAPT might be appropriate for those at higher thrombotic but lower bleeding risk, because premature DAPT discontinuation might increase the risk of stent thrombosis and MI.

8.3.2.4 PCI or ACS in patients with AF at high risk of stroke/systemic embolism

Combination OAC and antiplatelet therapy is required for patients with AF who are aged 65 years or older or with CHADS2 score 1, who are undergoing PCI or treatment of ACS. In these patients the optimal therapeutic regimen should be individualized on the basis of a balanced assessment of their risk of AF-related stroke/systemic embolism (“CCS algorithm”, Fig. 8), ischemic coronary events, and clinically relevant bleeding (Fig. 10). The risk of ischemic coronary events is modulated by the clinical presentation (eg, ACS being higher risk than elective PCI), clinical characteristics (higher risk with comorbid diabetes mellitus or CKD, current tobacco use, or previous stent thrombosis), as well as PCI-related factors (higher risk with multivessel disease, multiple stent implantation, total stent length > 60 mm, bifurcation lesion, chronic total occlusion intervention, and stent type).^[240],[241] The risk of bleeding can be estimated from clinical risk scores such as Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly (> 65 Years), Drugs/Alcohol Concomitantly (HASBLED), Predicting Bleeding Complications in Patients Undergoing Stent Implantation and Subsequent Dual Anti Platelet Therapy (PRECISE-DAPT), and Cardiovascular Disease Research Using Linked Bespoke Studies and Electronic Health Records (CALIBER), with the former validated in a VKA-treated population, and the latter 2 in a population with CAD treated with PCI and DAPT.^[240]

8.3.2.5 Key trials of dual pathway therapy vs triple therapy in patients with AF and ACS/PCI

The key trials that have compared dual pathway therapy with VKA-based TT include: the What Is the Optimal Antiplatelet and Anticoagulation Therapy in Patients With Oral Anticoagulation and Coronary Stenting (WOEST) study, the Prevention Of Bleeding In Patients With AF Undergoing PCI (PIONEER AF-PCI) study, the Randomized Evaluation of Dual Antithrombotic Therapy With Dabigatran versus Triple Therapy With Warfarin in Patients With Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention (RE-DUAL PCI) trial, the Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Percutaneous Coronary Intervention (ENTRUST-AF PCI) study, and the AUGUSTUS trial.^{[243]–[247]} Relevant details of these trials are presented in Table 8. Collectively these studies showed that the use of DOACs with a P2Y12 inhibitor was associated with a significant reduction in major bleeding (HR, 0.62; 95% CI, 0.47-0.81), without a statistically significant excess in the occurrence of MI (HR, 1.18; 95% CI, 0.93-1.52), definite stent thrombosis (HR, 1.55; 95% CI, 0.99-2.41), and stroke (HR, 0.89; 95% CI, 0.58-1.36).^[246] When considering these studies, it is important to recognize that a large proportion of patients were undergoing elective PCI for stable CAD (38%-72%), possibly affecting the absolute risk of thromboembolic complications relative to patients with ACS. Second, measures to decrease bleeding risk were underutilized, suggesting that the rates of bleeding in the TT arms might have been greater than in contemporary practice. Third, most of the patients who participated in these trials received clopidogrel as their P2Y12 inhibitor, meaning no conclusions can be drawn about the effects of prasugrel or ticagrelor.

8.3.2.6 Duration of triple therapy

The benefit of TT (reduction of recurrent MI and stent thrombosis) must be balanced against the increased bleeding risk with this therapeutic regimen. In the AF population, it is likely that shorter durations of DAPT after ACS or PCI are reasonable when concomitant OAC will be used for the prevention of stroke/systemic embolism. This concept was shown in the Intracoronary Stenting and Antithrombotic Regimen: Triple Therapy in Patients on Oral Anticoagulation After Drug Eluting Stent Implantation (ISAR-TRIPLE) study, which showed no significant difference in the primary end point of “net clinical benefit” (combination of death, MI, stent thrombosis, stroke, or Thrombolysis in Myocardial Infarction [TIMI] major bleeding) between those who received 6 weeks vs 6 months of TT with a significant reduction in any bleeding in the group that received 6 weeks of TT.^[248] A post hoc analysis of the AUGUSTUS trial, which compared the risk of bleeding and ischemic outcomes from randomization to 30 days and from 30 days to 6 months showed that aspirin use (ie, TT) was associated with more severe bleeding (absolute risk difference 0.97%) but fewer ischemic events (absolute risk difference -0.91%) in the first 30 days following randomization, when compared to placebo (ie, dual pathway therapy). From 30 days to 6 months, the risk of severe bleeding was higher with aspirin than placebo (absolute risk difference 1.25%) whereas the risk of ischemic events was similar (absolute risk difference -0.17%). As such, it appears that the benefit of triple therapy on ischemic coronary outcomes is limited to the first 30 days of therapy, with continuation of triple therapy beyond 30 days resulting in increased rates of bleeding without reduction in ischemic outcomes. Likewise, the PIONEER AFPCI and RE-DUAL PCI trials showed no relationship between major adverse cardiovascular events and TT duration.^[244],[245] On balance, these findings suggest that limiting the course of TT to 1 month, and thereafter continuing dual pathway therapy (OAC and a P2Y12) might achieve the optimal balance of ischemic events prevented and bleeding caused.

8.3.3 Liver disease

The optimal management of patients with AF and advanced liver disease is complex. Patients with advanced liver disease are at increased risk of bleeding because of perturbations in coagulation factor synthesis,^[249],[250] which was affirmed in a large Swedish registry-based study of 182,678 AF patients.^[249] In this study the presence of liver disease was associated with an increased risk of major bleeding (adjusted HR, 1.8; 95% CI, 1.45-2.23) even in the absence of concomitant OAC use.^[249] In addition, there is a concern that advanced liver disease results in a prothrombotic state, which might increase the risk of stroke/systemic embolism.^{[249],[251],[252]} Finally, liver disease results in altered metabolism of VKAs and DOACs, which further complicates OAC choice.^{[253]–[257]} Although OACs should be considered in most AF patients with liver disease, the optimal treatment is more challenging because such patients were excluded from the landmark anticoagulation trials. Specifically, patients with significant active liver disease and those with persistent elevation of liver enzymes or bilirubin were excluded from the large RCTs of a DOAC vs VKA.^{[21]–[23],[25]} Although randomized data are limited, observational studies have shown that VKA use in patients with AF and advanced liver disease resulted in reduction of ischemic stroke risk with a higher risk of bleeding complications.^{[250],[252],[256]–[258]} Moreover, a post hoc analysis of a prospective observational multicentre study of AF patients showed that the presence of advanced liver fibrosis increased the risk of major bleeding in those who received a VKA, but not in those who received a DOAC.^[259] More recently, a nationwide retrospective Taiwanese cohort study reported outcomes in 2428 cirrhotic patients with NVAF who were taking apixaban (n ¼ 171), dabigatran (n ¼ 535), rivaroxaban (n ¼ 732), or a VKA (n ¼ 990).^[258] In this study a similar risk of ischemic stroke/systemic embolism between DOAC- and VKA-treated patients was observed, however, the rates of major bleeding (2.9% vs 5.4% per year; P ¼ 0.0003) and gastrointestinal (GI) bleeding (1.9% vs 3.6% per year; P ¼ 0.0030) were significantly lower with a DOAC. For patients with advanced cirrhosis the DOAC group had a lower risk of ICH (HR, 0.17; 95% CI, 0.03-0.96; P ¼ 0.04). The Child-Pugh score can be used to guide OAC choice in patients with AF and advanced liver disease.^[260] This score is commonly used to express the severity of chronic liver disease on the basis of clinical (ascites and encephalopathy) and laboratory (INR, bilirubin, albumin) parameters. When OAC is being prescribed for AF patients with Child-Pugh grade C cirrhosis it is recommended that a VKA be the preferred agent, because such patients were excluded from the landmark DOAC vs VKA trials.^{[21]–[23],[25]} For those with Child-Pugh grade B cirrhosis it is recommended that rivaroxaban be avoided because it undergoes substantial hepatic metabolism,^[253],[254] whereas apixaban, dabigatran, and edoxaban may be used with caution in such patients.^{[256]–[258],[261]} Large studies appear to confirm a very low risk of hepatotoxicity with the approved DOACs.^[262] The use of the appropriate antithrombotic agent in AF patients with advanced liver disease requires individualized therapy and might benefit from the guidance of an expert multidisciplinary team including a hepatologist and hematologist.

8.3.4 Cancer

AF and malignancy are relatively common conditions, with the prevalence of both increasing with age. Patients with cancer have an elevated risk of AF, possibly because of the presence of comorbid conditions, a local/direct tumour effect, systemic inflammation, altered sympathovagal balance, or as a complication of surgical or targeted therapies (eg, tyrosine kinase inhibitors such as ibrutinib).^[263],[264] The management of coexisting AF and cancer is challenging, because of the increased rates of thrombosis and thromboembolism observed in association with malignancy, as well as the increased risk of bleeding associated with coagulation defects, blood vessel erosion, tumour vascularity, radiation injury, and the antiplatelet effects of chemotherapies.^[265],[266] These factors increase the complexity of OAC decision-making. The evidence for efficacy of anticoagulation among AF patients with cancer comes from several sources. These include 1 population-based retrospective cohort study, 2 retrospective cohort studies, 1 case-control study, and 3 post hoc analyses of the landmark DOAC vs VKA RCTs.^{[267]–[273]} Among patients where were receiving a DOAC, the annual incidence of thromboembolic events varied from 0.0% to 4.9% with cancer and from 1.3% to 5.1% without; the annual incidence of a major bleed during DOAC treatment varied from 1.2% to 4.4% with cancer and 1.2% to 3.1% without cancer.^[274] A large registry using prescription-based analysis for AF patients receiving VKAs or DOACs, with and without cancer reported equivalence for bleeding and thromboembolic risk, although the rates of both were lower in the DOAC population.^[270] Within the landmark randomized trials of DOACs vs VKAs, there did not appear to be any significant difference in relative efficacy and safety of DOACs compared with VKAs in patients with and without a history of cancer.^{[267],[271],[273]} Likewise, RCTs evaluating the cancer associated venous thromboembolism population have shown that DOACs were noninferior to low molecular-weight heparin (LMWH) for efficacy as well as safety.^[275] The use of the appropriate antithrombotic agent in a patient with a concomitant malignancy requires individualized therapy under the guidance of an expert multidisciplinary team. When myelosuppressive chemotherapy or radiation therapy is undertaken or intensified, further steps to reduce bleeding risk, such as dose reductions or temporary cessation of OAC might be necessary.

8.3.5 Congenital heart disease

8.3.5.1 Stratification for anticoagulation

Although the link between atrial arrhythmias and thromboembolic complications is less well established in patients with CHD compared with the general AF population, the existing data appear consistent. Absence of sinus rhythm was found to be the factor associated with the highest prevalence of stroke in a cohort of > 23,000 adults with CHD.^[276] Moreover, a strong association was observed between atrial arrhythmias and stroke in a Quebec administrative database study of > 38,000 patients with CHD.^[277] As such, questions regarding long-term anticoagulation for atrial arrhythmias frequently arise in the care of adults with CHD.^[278] Two retrospective studies have assessed anticoagulation practices, thromboembolic event rates, and bleeding complications in adults with CHD and atrial arrhythmias. A single centre retrospective study derived from the Congenital Corvitia (CONCOR) registry in the Netherlands followed 229 adults with CHD and intra-atrial reentrant tachycardia (IART), AF, or atrial tachycardia for a median of 6 years.^[279] Overall, 67% of patients received a VKA and 7% antiplatelet therapy. The thromboembolic event rate in patients without a mechanical valve was 1.4% per year, with a major bleeding rate of 4.4% per year during treatment with a VKA. In univariable analyses, a CHA2DS2-VASc score 2 was associated with a higher thromboembolic event rate (HR, 3.7; P ¼ 0.021). Similarly, a HAS-BLED score 2 was associated with a higher major bleeding rate in univariable analyses (HR, 2.9; P ¼ 0.017). The Anticoagulation Therapy in Congenital Heart Disease (TACTIC) study enrolled 482 patients (age 32.0  18.0 years) from 12 North American centres.^[280] Although the study was retrospective, it adhered to clinical trial data management standards including blinded adjudication of arrhythmias and outcomes and multiple layers of data quality control. Patients were classified as having simple (18.5%), moderate (34.4%), or severe (47.1%) forms of CHD on the basis of an established classification scheme.^[281] OAC, predominantly VKAs, were administered to 54% of participants and antiplatelet agents to 38% of participants. Freedom from thromboembolic events was 89% at 10 years and 85% at 15 years. Rates did not differ significantly according to whether patients had AF vs IART. In multivariable analyses, complexity of CHD was the only factor independently associated with thromboembolism. Corresponding thromboembolic event rates in patients with simple, moderate, and severe forms of CHD were 0.00%, 0.93%, and 1.95% per year, respectively (P < 0.001). Notably, no thromboembolic event occurred in patients with simple forms of CHD despite 44% not receiving OAC. In patients with moderate and severe forms of CHD, the thromboembolic event rate exceeded the major bleeding rate associated with a VKA (0.77% per year). Although 93% of the population had a HAS-BLED score of 0 or 1, a higher score was associated with a greater risk of bleeding in multivariable analyses (HR, 3.15; P ¼ 0.047). These results were consistent with the concept of incorporating complexity of CHD in stratification decisions for anticoagulation, as proposed by the Pediatric and Congenital Electrophysiology Society (PACES)/Heart Rhythm Society (HRS) expert consensus statement on the recognition and management of arrhythmias in adults with CHD.^[282]

8.3.5.2 Choice of anticoagulant

Evidence on the use of DOACs in adults with CHD and atrial arrhythmias has recently emerged in the form of observational studies, as featured in a recent review.^[283] In the first series of 75 adults with CHD receiving DOACs, 57 (76%) were anticoagulated for atrial arrhythmias.^[284] No thrombotic or major hemorrhagic event occurred over a mean follow-up of 12 months. Nevertheless, minor bleeds were observed in 47%. An international registry reported a 1-year experience on the use of DOACs in 530 adults with CHD, 482 of whom had atrial arrhythmias.285 The DOACs consisted of rivaroxaban in 43%, apixaban in 39%, dabigatran in 12%, and edoxaban in 7%. Overall, complexity of CHD was simple in 15%, moderate in 45%, and severe in 40%. Nearly half the population (46%) was considered to have a significant valve lesion. Bioprosthetic valves were present in 11%, with no patient having a mechanical valve. At 1-year of follow-up, the thromboembolic event rate was 1.1%, major bleeding rate 1.3%, and minor bleeding rate 6.3%. In light of this reassuring data and considering that valve disease is common in adults with CHD, it appears reasonable to select a DOAC instead of a VKA in CHD patients with IART or AF and forms of valve disease that are similar to those with reassuring efficacy and safety data from large DOAC trials.^[283] However, thromboembolic and bleeding rates in the large international registry were disproportionately high in the 14% of patients with Fontan palliation, who experienced a remarkable 50% rate of thrombotic and bleeding complications.^[285] In a separate series 10 minor bleeding events occurred in 21 patients with Fontan surgery who received DOACs (12 for atrial arrhythmias).^[286] One patient with a right-to-left shunt through a fenestration developed deep vein thrombosis during treatment with dabigatran. Another patient receiving apixaban had progression of thrombosis within the Fontan circuit. A separate report described worsening thrombus in a patient with an intracardiac lateral tunnel Fontan and IART during treatment with apixaban.^[287] Thus, rigourous comparative safety and efficacy data of DOACs vs VKAs in certain subgroups of patients with severe forms of, including Fontan palliation, are required before DOACs can be endorsed for routine use in this setting.^[282]

8.3.6 Secondary AF

Although secondary AF has been associated with an increased risk for stroke and mortality irrespective of the precipitant, it is unclear whether the benefits of OAC apply equally to secondary AF patients compared with those with primary AF.^[288],[289] As such, it remains uncertain whether such patients should be treated with long-term OAC on the basis of the standard stroke risk stratification schema. Until randomized trials assessing OAC in the secondary AF population are available an individualized approach to OAC is warranted. One such strategy would evaluate the likelihood that the secondary AF is “reversible” (ie, AF that occurs solely secondary to an acute illness, with little to no abnormal underlying substrate and therefore limited risk of recurrence) or “provoked” (ie, AF that is unmasked by the acute illness, occurring in patients with significant abnormal underlying substrate and therefore significant risk for recurrence). In both instances OAC might be warranted in selected patients in the short-term (ie, until resolution of the acute AF episode) however the long-term need for OAC should take into consideration a given patient’s underlying risk of stroke (eg, CHADS-65), the specific secondary precipitant for AF, and the likelihood of recurrence. In some cases, such as sepsis, the acute administration of intravenous (I.V.) anticoagulation increases the risk of bleeding but does not appear to reduce the risk of ischemic events.^{[289]–[291]} Conversely, OAC is required for most patients with thyrotoxicosis-related AF because hyperthyroidism shifts the balance of the coagulation fibrinolytic system toward a hypercoagulable state, acting as an independent risk factor for thromboembolic events.^{[292]–[294]} Of note, frequent monitoring of INR is advisable when a VKA is used in patients with hyperthyroid-associated AF, because the potency of VKAs will change as the degree of hyperthyroidism changes.^[295]

8.3.7 Hypertrophic cardiomyopathy

The risk of stroke/systemic embolism in patients with AF and hypertrophic cardiomyopathy (HCM) is substantial and is significantly greater than in the non-HCM population with AF.^{[296]–[301]} Stroke/systemic embolism in the HCM population appears to be unrelated to the type of AF (paroxysmal vs persistent) or AF burden.^[297] Moreover, traditional risk prediction scores (eg, CHADS2 or CHA2DS2-VASc) have not been validated in the HCM population, and do not reliably predict thromboembolic outcomes in the HCM population.^{[296],[297],[302]} All subjects with HCM who develop AF should started treatment with OAC because there is currently insufficient evidence to support the use of stroke risk prediction tools in the HCM population.^{[296],[297],[301],[302]} VKAs have been traditionally used as the preferential OAC in subjects with HCM and AF.^[303] Observational data subject to selection and ascertainment bias suggest that DOACs are at least as effective as VKAs in preventing thromboembolic events with similar or lower bleeding and safety events, with some studies suggesting improved survival and greater treatment satisfaction with DOACs.^{[304]–[306]} As such, it is suggested that either DOACs or VKAs can be used to reduce thromboembolic events in subjects with HCM and AF.

8.3.8 Cardiac amyloidosis

Patients with cardiac amyloidosis are at a particularly high risk for thromboembolism due to progressive atrial myopathy, electromechanical dissociation, and reduced LAA emptying velocity. LA thrombus has been reported in up to one-third of patients with cardiac amyloidosis in sinus rhythm, as well as in patients with AF who are receiving therapeutic anticoagulation.^{[307]–[309]} Among patients with cardiac amyloidosis who undergo cardioversion, 22.4% had intracardiac thrombus identified on TEE, which was significantly more common than in matched controls. Of patients with cardiac amyloidosis with intracardiac thrombus, 4 of 13 had received therapeutic OAC for a period of 3 weeks and 2 of 13 had an AF episode duration of < 48 hours. Patients with amyloid light-chain (AL) amyloidosis appear to be at particularly high risk of intracardiac thrombosis.^[310] There is limited information to inform a specific OAC strategy. Data regarding the safety and efficacy of warfarin and DOACs is limited, however, their use appears to be reasonably safe in patients with cardiac amyloidosis the absence of conventional contraindications.^[311]

8.3.9 Frail elderly patients

Improvements in life expectancy have resulted in an increasing number of AF patients being cared for into advanced age.^[36],[312] Advanced age is a well established risk factor for ischemic stroke and hemorrhagic events in subjects with AF.^{[313]–[316]} Unfortunately, elderly patients with AF might present with multiple health conditions that might affect QOL (eg, anemia or cognitive impairment/dementia), and/or increase the risk of adverse drug events (eg, impaired renal or hepatic function). In addition, polypharmacy is common in the elderly, which increases the risk of adverse drug reactions due to drug-drug interactions. Further, these patients are often excluded from RCTs, leading to a relative paucity of data to guide treatment decisions. Unfortunately, despite clear benefit,^{[21]–[23],[25],[270],[317]} OAC remains underutilized in this population.^[317],[318] A recent study of older patients with AF noted 35% of OAC-eligible patients did not receive anticoagulation.^[317] The major factors responsible for underutilization of anticoagulation in subjects with AF include older age, female sex, abnormal liver enzyme levels, history of falls, excessive alcohol consumption, and concomitant prescription of antiplatelet agents.^{[318]–[327]} Despite these concerns, there is evidence that the net clinical benefit favours prescription of OAC in this population.^[328],[329] Although there are limited RCTs specifically in the elderly population with AF, secondary analyses of the landmark DOAC vs VKA trials did not show a difference in efficacy among subjects aged 75 years or older and those aged younger than 75 years.^{[330]–[333]} Specifically, a meta-analysis of 4 RCTs (20,165 AF participants 75 years of age or older) showed a significantly lower rate of stroke/systemic embolism in DOAC-treated elderly patients (3.3% vs 4.7%; OR, 0.65; 95% CI, 0.48-0.87), with no significant difference in the rates of major or clinically relevant nonmajor (CRNM) bleeding (6.2% vs 6.6%; OR, 0.82; 95% CI, 0.58-1.16).^[334] A pooled analysis from 2 European registries showed that DOAC use was associated with net clinical benefit in AF patients 75 years of age or older; with significant reduction in the composite end point of major bleeding and ischemic cardiovascular events (6.6% per year with DOACs vs 9.1% per year with VKAs; OR, 0.71; 95% CI, 0.51-0.99; P ¼ 0.042), which was predominantly due to a lower rate of major bleeding (OR, 0.58; 95% CI, 0.38-0.90; P ¼ 0.013) and a trend to reduction in ischemic events (OR, 0.71; 95% CI, 0.51 1.00; P ¼ 0.05) with DOACs.^[328] In those at risk of falls, consideration should be made to: provide walking aids and proper footwear; promote cardiovascular, resistance, and balance activities; and, avoid medications that result in hypotension.

8.3.10 Increased BMI, obesity, and morbid obesity

Obesity is an established risk factor for the development of AF.^[120] Clinicians face challenges regarding dosing of OAC in obese patients, because increasing BMI is independently associated with a higher risk of bleeding but lower risk of stroke/systemic embolism.^[335],[336] Recent publications regarding outcomes in obese patients enrolled in the landmark DOAC trials are reassuring. A post hoc analysis of the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial revealed that the treatment effect of apixaban was consistent with the overall trial results across all of the weight categories (P for interaction ¼ 0.64).^[337] Specifically, in 982 patients > 120 kg, there was a significant reduction in major and CRNM bleeding with apixaban (HR, 0.58; 95% CI, 0.35-0.95) with comparable efficacy for stroke/systemic embolism (HR, 0.39; 95% CI, 0.12-1.22), relative to warfarin. The Effective Anticoagulation With Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48(ENGAGE AF-TIMI 48) trial included 2099 patients (10%) with a BMI of 35-39, 1149 with a BMI > 40 (5.5%), and 148 with a BMI > 50 (0.7%).^[335] There was no significant difference in trough edoxaban plasma concentrations and anti-Factor Xa activity across BMI categories, and there was no significant interaction between BMI and clinical outcomes (stroke/systemic embolism, all-cause mortality, major bleeding, major or CRNM bleeding, or the net clinical outcome). The Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonist for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trial enrolled 5206 patients with a BMI > 30.^[336] There was no significant treatment interaction by BMI category (< 25, 25-35, or > 35) for either the primary end point of stroke/systemic embolism, or major or CRNM bleeding.^[22] Finally, the Randomized Evaluation of Longterm Anticoagulation Therapy (RE-LY) trial included 3099 patients> 100 kg. There was no significant treatment interaction by weight category on the rate of stroke/systemic embolism or major bleeding, with dabigatran 110 mg BID and 150 mg BID being at least as effective as warfarin in obese patients.^[21],[338] Meta-analyses of these phase III trials suggest that DOACs are more effective than warfarin for overweight patients (BMI 25-30: OR, 0.87; 95% CI, 0.76-0.99 for stroke/systemic embolism; OR, 0.83; 95% CI, 0.71-0.96 for major bleeding), and similarly effective in obese patients (BMI > 30: OR, 0.87; 95% CI, 0.76-1.00 for stroke/systemic embolism; OR, 0.91; 95% CI, 0.81-1.03 for major bleeding).^[338] Although it is important to note that severely obese patients were not well represented, these meta-analyses provide reassurance that DOACs are at least as effective as warfarin in overweight and obese patients.

8.4 Anticoagulation in special circumstances

8.4.1 Cardioversion

8.4.1.1 Increased thromboembolism risk at the time of cardioversion

The incidence of stroke/systemic embolism is increased after cardioversion. A pooled analysis of 16 studies published between 1960 and 1969 showed a 30-day postcardioversion incidence of thromboembolic events of 1.76% (2665 cardioversions) in patients not receiving OAC with AF of > 48 hours in duration.^[339] Similarly, a more contemporary pooled analysis of 10 studies describing outcomes after 3564 cardioversions of AF of > 48 hours duration published between 1986 and 2003 showed a 30-day incidence of thromboembolic events of 2.39% in patients not receiving OAC.^[339] These rates contrast with the monthly 0.5% background thromboembolic risk of patients with AF in the absence of OAC when cardioversion has not been performed.^[51] The increased thromboembolic risk associated with cardioversion might be related to 3 phenomena. The first is generation of thrombi during the persistent AF episode, with subsequent embolization after restoration of organized atrial contraction.^[340] The second relates to a period of transient atrial mechanical dysfunction after the restoration of sinus rhythm.^{[341]–[343]} This “atrial stunning” might be responsible for development of new intracardiac thrombi post cardioversion despite the restoration of sinus rhythm.^[344],[345] Such atrial mechanical dysfunction has been reported with pharmacologic, electrical, and spontaneous cardioversion and is maximal in the period immediately after cardioversion.^{[344]–[347]} The duration and severity of atrial stunning varies depending on the duration of the atrial arrhythmia, atrial size, and presence of underlying structural heart disease.^{[344]–[347]} The third possibility is that cardioversion does not directly cause stroke/systemic embolism but rather patients who require cardioversion have transient as well as permanent characteristics associated with a higher stroke/systemic embolism risk.^[348]

8.4.1.2 Anticoagulation for cardioversion of AF of > 48 hours

In patients with continuous AF for > 48 hours, observational studies reported a lower 30-day incidence of post cardioversion thromboembolic events in patients who receive OAC than in patients who do not receive OAC (0.45% vs 1.76% in studies from 1960-1969; 0.20% vs 2.39% in studies from 1986-2003).^[339] Although none of these studies were randomized, the results are compelling in that patients who received OAC were those at highest risk of stroke/systemic embolism (eg, those with rheumatic mitral stenosis and/or previous thromboembolic events). Accordingly, there is widespread agreement that OAC is required for 3 weeks before and 4 weeks after planned cardioversion in patients who present with either “valvular AF,” or NVAF of > 48 hours.

8.4.1.3 TEE instead of 3 weeks of OAC before cardioversion

When cardioversion is desired without 3 weeks of therapeutic OAC another approach is to establish immediate anticoagulation, then perform a TEE to exclude LA thrombi before immediate cardioversion. This practice is supported by the Assessment of Cardioversion Using Transesophageal Echocardiography (ACUTE) trial, which randomized 1222 patients with nonacute AF to TEE-guided cardioversion or to the conventional 3 weeks of OAC before cardioversion.^[349] Although there was no significant difference in the rate of periprocedural thromboembolic events (0.8% in the TEE group vs 0.5% in the conventional group; P ¼ 0.50), patients in the TEE group had fewer hemorrhagic events (2.9% vs 5.5%; P ¼ 0.03), a shorter time to cardioversion (3.0 5.6 vs 30.6 10.6 days; P < 0.001), and more frequently had sinus rhythm restored (71.1% vs 65.2%; P ¼ 0.03). Although other imaging modalities such as cardiac computed tomography (CT), cardiac magnetic resonance (MR) imaging (MRI), and intracardiac ultrasound might have sensitivities and specificities for the detection of LA clot that are comparable with those of TEE,^[350] only the use of TEE for screening patients for candidacy for cardioversion has been subjected to an RCT. Accordingly, TEE remains the goldstandard procedure for this purpose; nevertheless, the other procedures listed may be used for this purpose when TEE is contraindicated or is not available.

8.4.1.4 Anticoagulation for cardioversion of AF of version of AF of ≤ 48 hours

Patients who present with AF of ≤ 48 hours have long been considered, on a theoretical basis, to have a low risk of stroke/systemic embolism after cardioversion. This practice has been supported by observational reports of outcomes after cardioversion in patients with AF ≤ 48 hours, suggesting a low risk of stroke/systemic embolism in the 30 days after cardioversion (0.27% after 4836 cardioversions in 4380 patients).^[339] Although this risk is comparable with that of planned cardioversion in patients with AF of > 48 hours who receive OAC, these reports were limited by selection bias; describing low-risk patients who presented in a stable state early in the 48-hour window, with a significant proportion receiving OAC.^[339] Moreover, despite the significant selection bias the rate of stroke/systemic embolism substantially exceeded the established threshold for OAC initiation (1.5% per year or 0.12% per 30 days).^[351] More recent data specifically focused on patients who underwent cardioversion for AF of < 48 hours in the absence of OAC provide a more disquieting viewpoint.^{[352]–[359]} In the Cleveland Clinic Study^[359] consecutive patients having cardioversion of AF of ≤ 48 hours duration were examined, and a significantly higher 30-day post cardioversion rate of stroke/systemic embolism was reported in patients with no or subtherapeutic OAC (0.88%; 6 events after 683 cardioversions) compared with those with therapeutic OAC (0.22%; 2 events after 898 cardioversions; OR, 4.8; P ¼ 0.03). In the ANTIKogulation Registry^[352] 366 consecutive patients with AF of ≤ 48 hours were examined, and a significantly higher rate of LA thrombosis (4% vs 0%; P ¼ 0.02) and a nonsignificantly higher rate of 30-day post cardioversion stroke/TIA (1.3% vs 0.5%; P ¼ 0.58) was observed in those who did not receive OAC. Although the latter was not statistically significant, it is noteworthy that each of the patients with a stroke/TIA had a CHADS2 score of 0. The FinCV studies were retrospective, observational studies that determined outcomes after cardioversion of AF in catchment area patients of 8 hospitals in Finland between 2003 and 2016.^{[353]–[358]} The program enrolled consecutive patients in whom electrical or pharmacological cardioversion was attempted for AF of ≤ 48 hours (FinCV: 3143 patients; 7660 cardioversions), consecutive patients in whom elective electrical cardioversion was attempted for AF of > 48 hours (FinCV2: 1271 patients; 1894 cardioversions), and patients with AF who received DOACs in whom electrical or pharmacological cardioversion was attempted (FinCV3: 1028 patients; 1298 cardioversions). Together these 6 reports, on the short-term outcomes after 10,852 pharmacologic or electrical cardioversions in 5441 patients, demonstrated the following: (1) in the absence of OAC, time to cardioversion is a very strong predictor of 30- day risk of stroke/systemic embolism^[354]; (2) in the absence of OAC the incidence of stroke/systemic embolism after cardioversion is significantly higher in patients with AF duration of 12-48 hours than in patients with < 12 hours of AF (1.1% [30/ 2767 patients] vs 0.33% [8/2440 patients], respectively)^[354]; (3) for patients who receive OAC, AF duration is not associated with the incidence of stroke/systemic embolism (0.1% for < 24 hours; 0% for 24-48 hours; 0% for 48 hours to 30 days; and 0.2% for > 30 days)^[358]; (4) in the absence of periprocedural OAC, independent predictors of stroke/systemic embolism postcardioversion are older age (OR, 1.05 per year; 95% CI, 1.02-1.08; P < 0.001), female sex (OR, 2.1; 95% CI, 1.1-4.0; P ¼ 0.03), HF (OR, 2.9; 95% CI, 1.1-7.2; P ¼ 0.03), and diabetes mellitus (OR, 2.3; 95% CI, 1.1-4.9; P ¼ 0.03)^[353],[355]; and (5) therapeutic OAC significantly decreases the risk of postcardioversion stroke/systemic embolism.^{[353]–[357],[360]} Specifically, for patients with AF of≤ 48 hours the 30 day incidence of stroke/systemic embolism was significantly lower in the presence of OAC (0.13% [3 events in 2298 encounters] vs 0.71% [38 events in 5362 encounters]; P ¼ 0.001).^[355] When parsed according to CHA2DS2-VASc score, OAC use significantly reduced the 30 day rate of definite stroke/systemic embolism in those with a CHA2DS2-VASc score of 2 (0.2% [3/1; 708] vs 1.1% [28/2590]; P ¼ 0.001), but the benefit of OAC was not statistically significant in patients with a CHA2DS2-VASc score of 0-1 (0.0% [0/590] vs 0.40% [10/2772]; P ¼ 0.23). Of note, 10 of the 38 (26%) definite thromboembolic events that occurred after successful cardioversion in patients without OAC occurred in patients with a CHA2DS2-VASc score of 0-1. In 2 other recent reports the thromboembolic risk associated with cardioversion of recent-onset AF in the absence of OAC were indirectly assessed: a report from the Danish National Patient Registry^[361] and a report from the Swedish National Patient Registry.^[362] Each used administrative databases to evaluate the 30-day rate of stroke/systemic embolism after cardioversion of AF as a function of the presence or absence of precardioversion OAC. Although these databases did not include the duration of AF, the authors proposed that it would be reasonable to assume that cardioversion would only have been performed without previous OAC in patients with recentonset AF of≤ 48 hours. In the Danish analysis,^[361] the 30-day incidence of stroke/systemic embolism after cardioversion was 1.06% (54 events in 5084 patients) without precardioversion OAC and 0.29% (32 events in 11,190 patients) with precardioversion OAC. Stroke/systemic embolism did occur in patients with CHADS2 or CHA2DS2-VASc scores of 0 or 1 but the rates were not stated. In the Swedish analysis,^[362] the crude 30-day incidence of stroke/systemic embolism in the absence of precardioversion OAC was significantly higher than in the presence of precardioversion OAC (0.86% [104 events in 12,152 patients] vs 0.33% [35 events in 10,722 patients]). After adjustment for unequal distributions of the CHA2DS2-VASc factors, the OR for stroke/systemic embolism in the 30 days after cardioversion was 2.54 (95% CI, 1.70-3.79; P < 0.001) without precardioversion OAC relative to patients with precardioversion OAC. Propensity analysis of 9500 patients who did not receive OAC matched with 9500 patients who did receive OAC showed that patients who did not receive OAC were more likely to have stroke/systemic embolism (OR, 2.51; 95% CI, 1.69-3.75; P < 0.001) with similar rates of major bleeding (OR, 1.00; 95% CI, 0.48-2.10) relative to those who received OAC.^[362] Taken together, the authors concluded that cardioversion in patients with AF of < 48 hours without OAC conferred a greater risk of stroke/systemic embolism than cardioversion for patients with AF of > 48 hours who received OAC. In summary, the risk of stroke/systemic embolism after cardioversion is elevated, even in patients who present within 48 hours of AF onset. This risk does not differ on the basis of the method of cardioversion (30 day risk of cardioversion of 0.26% with electrical vs 0.39% with pharmacological cardioversion).^[363] Immediate cardioversion without 3 weeks of therapeutic OAC appears to be associated with a low risk of stroke/systemic embolism only in patients who presenting within 12 hours of AF onset and in patients who presenting 12-48 hours after AF onset who have a low risk of stroke (eg, patients aged younger than 65 years with a CHADS2 score of 0-1). Other patients deemed to be at higher risk of stroke should receive at least 3 weeks of therapeutic OAC before cardioversion (or undergo a TEE to exclude LA thrombus). After cardioversion, all patients should receive at least 4 weeks of OAC in the absence of a strong contraindication, and such therapy should be initiated as soon as possible and preferably before the cardioversion.^[364] Thereafter, the need for OAC should be on the basis of the risk of stroke/systemic embolism as determined by the “CCS Algorithm” (CHADS-65). The approach to anticoagulation at time of cardioversion is presented in Figure 11.

8.4.1.5 DOAC or VKA for pericardioversion OAC

Three randomized trials have compared a DOAC with adjusted-dose VKAs in the setting of planned cardioversion of AF.^{[365]–[367]} A meta-analyses of these 3 RCTs^{[365]–[367]} showed that DOAC compared with adjusted-dose VKA was associated with significant reduction in stroke/systemic embolism (0.2% vs 0.6%; RR, 0.33; 95% CI, 0.12-0.91; P ¼ 0.03),^[368] but no significant differences in any bleeding (1.8% vs 2.5%; RR, 0.85; 95% CI, 0.158-1.23; P ¼ 0.38),^[369] major bleeding (0.4% vs 0.7%; RR, 0.61; 95% CI, 0.28-1.34; P ¼ 0.22),^[368] or mortality (0.3% vs 0.4%; RR, 0.70; 95% CI, 0.23-2.10; P ¼ 0.52).369 In a meta-analysis that combined these RCT^{[365]–[367]} data with that from patients having cardioversion in the 4 pivotal trials that compared DOAC with VKA therapy,^{[370]–[373]} it was reported that no significant differences were observed in the singular outcomes of ischemic stroke, hemorrhagic stroke, mortality, or major bleeding, or in the composite outcome of stroke/systemic embolism with DOACs compared with adjusted-dose VKAs.^[374],[375] It should be noted that these trials, individually and in their aggregate, were not sufficiently powered to exclude a clinically meaningful difference in either safety or efficacy.

8.4.2 Catheter ablation of AF

AF ablation is an invasive procedure associated with a risk of stroke/systemic embolism as well as a risk of bleeding. There has been a tremendous amount of research in the past few years regarding periablation OAC management. The use of uninterrupted OAC with a VKA was associated with a lower risk of bleeding and a decreased risk of thromboembolic complications compared with interrupted VKAs with LMWH bridging.^[376] More recently, several randomized trials have shown superior safety and efficacy of uninterrupted DOAC (apixaban, dabigatran, edoxaban, and rivaroxaban) compared with uninterrupted VKA.^{[377]–[381]} For these reasons, uninterrupted OAC has been considered the standard of care for AF ablation. Uninterrupted OAC typically means initiating OAC 3-4 weeks before the procedure, continuing OAC right up to the time of procedure, and reinitiating OAC 6-8 hours after the procedure. More recently, 2 trials have assessed the safety of “minimal” DOAC interruption.^[382],[383] On the basis of these trials, it might be reasonable to hold a DOAC for 1-2 doses before ablation with early reinitiation 6-12 hours post ablation. Intraprocedural anticoagulation is critical to avoid the risk of clinical and subclinical cerebral thromboembolism.^[384],[385] Anticoagulation during ablation is typically given as I.V. boluses and/or infusion of heparin with frequent measurement of the activated clotting time. More recent guidelines have suggested more aggressive activated clotting time targets of 350-400 seconds to prevent cerebral microembolism.^[384],[386] Post ablation management of OAC can be divided into 2 phases: the initial 2 months after the procedure and the period thereafter. Catheter ablation of AF transiently damages the LA endothelium, creating an early prothrombotic state, which might lead to thromboembolism. Such a prothrombotic state can occur even in patients considered to have a low risk of stroke/systemic embolism according to traditional risk schema. For this reason, OAC is recommended for at least 2 months after AF ablation. Thereafter, the need for ongoing OAC should be on the basis of the CCS Algorithm (CHADS-65), rather than the apparent success of the ablation procedure. Although limited retrospective data series have suggested a low risk of stroke after successful AF ablation in a wide variety of patient risk profiles, other studies, including the large randomized Catheter Ablation vs Anti-Arrhythmic Drug Therapy for Atrial Fibrillation (CABANA) trial, have not shown a significant reduction of stroke post ablation.^{[387]–[394]} Furthermore, discontinuation of OAC is limited by asymptomatic episodes of AF (meaning patients would be unaware of AF recurrence), late AF recurrence (meaning short-term freedom from recurrent AF might not predict long-term success), as well as a lack of clear temporal association between AF recurrence and stroke/systemic embolism.^{[75],[95],[395]} Because of the lack of large RCTs to provide a definitive answer on how to manage OAC late after ablation, at this point in time, AF ablation should not be considered as an alternative to OAC. If a patient has a high thromboembolic risk profile, then the patient should continue OAC even after successful AF ablation. If discontinuation of anticoagulation is being considered on the basis of patient values and preferences, regular and prolonged monitoring for AF screening is suggested. The Optimal Anticoagulation for Enhanced Risk Patients Post-Catheter Ablation for Atrial Fibrillation (OCEAN) trial will provide guidance in the future (NCT02168829).^[396]

8.4.2.1 Catheter ablation of AFL

Patients in sustained AFL need to be anticoagulated for at least 3 weeks before the ablation procedure and at least 1 month after ablation, on the basis of the same principles underlying the rationale for OAC after cardioversion. Similarly, a preprocedural TEE can be used to rule out atrial thrombus before AFL ablation if the arrhythmia has been continuous for more than 48 hours without appropriate anticoagulation.^[397] Analogous to AF, it is reasonable to perform catheter ablation procedures for AFL with uninterrupted OAC. Continued anticoagulation after successful AFL ablation is generally warranted in light of recognition that patients with isolated typical right AFL have an increased risk of later developing AF. The risk of thromboembolic events remains high despite successful AFL ablation.^{[189],[398],[399]} The main predictors for the occurrence of stroke after AFL ablation are older age, higher CHA2DS2-VASc score, recurrence of AFL, and subsequent development of AF.^{[189],[190],[399]} Multivariate predictors of AF after ablation of AFL include AF before ablation, reduced LV ejection fraction (LVEF), hypertension, increased LA size, inducible AF during the electrophysiology study, endurance sports, and structural heart disease.^[400] In patients without a preexisting history of AF up to 20%-40% will develop AF over the 18-36 months after AFL ablation.^{[189],[398],[400]} In patients with a history of preexisting AF up to 40%-75% will continue to experience AF after AFL ablation.^[400] Although there is no RCT on the benefits of continued anticoagulation after AFL ablation, observation data suggest that patients with AFL and a CHA2DS2-VASc score 2 have a higher risk of stroke.^[189],[190]

8.4.3 Management of OAC for patients undergoing invasive procedures

It is estimated that one in six AF patients who are receiving OAC will require an elective surgery or invasive procedure annually.^[401] For these patients, the periprocedural risk of a thromboembolic event with OAC interruption must be weighed against the risk of periprocedural bleeding. An OAC interruption interval that is longer than necessary might increase the thromboembolic risk, whereas an insufficient period of interruption might increase the bleeding risk, with consequent delays in OAC resumption.^[402] As such, perioperative anticoagulation management necessitates an assessment of the patient’s thrombotic risk, procedural bleeding risk (eg, minimal, low/moderate, and high bleed risk), specific anticoagulant used, renal function, and procedural bleeding risk (Table 9 and Figs. 12 and 13). An electronic tool incorporating these parameters to provide an OAC dosing schedule is provided by Thrombosis Canada (thrombosiscanada.ca).

8.4.3.1 No bridging for interrupted VKA in low thromboembolic risk patients

When a decision to interrupt VKA therapy has been made for an invasive procedure with a low/moderate or high risk of bleeding, interruption of VKA without LMWH or unfractionated heparin (UFH) bridging is recommended for most patients at low thromboembolic risk (Fig. 13). The rationale for the avoidance of bridging has come from several sources. A meta-analysis of 33 observational studies and 1 RCT involving 7118 predominantly non-AF patients showed that VKA interruption with bridging therapy was associated with increased overall bleeding (13.1% vs 3.4%; P < 0.0001) and increased major bleeding (4.2% vs 0.9%; P ¼ 0.004), but without reduction in thromboembolic events (0.9% vs 0.6%; P ¼ 0.50) compared with VKA interruption without bridging therapy.^[403] As well, in the randomized double-blinded placebo-controlled trial, Bridging Anticoagulation in Patients Who Require Temporary Interruption of Warfarin Therapy for an Elective Invasive Procedure or Surgery (BRIDGE),^[404] 1884 AF patients (most with CHADS2 score < 4) undergoing interruption of VKA for an elective surgery/invasive procedure were evaluated. Patients were randomized to placebo or LMWH from 3 days to 1 day before the procedure, and for 5- 10 days after the procedure. No bridging was noninferior to bridging for arterial thromboembolism (0.4% vs 0.3%; P ¼ 0.01 for noninferiority), but was associated with significantly less major bleeding (1.3% vs 3.2%; P ¼ 0.005 for superiority) and significantly less minor bleeding (12.0% vs 20.9%; P < 0.001). There were no significant differences for any other outcomes (all-cause mortality, MI, deep vein thrombosis, or pulmonary embolism).

8.4.3.2 Bridging for interrupted VKA in high thromboembolic risk patients

Bridging should be considered only for patients at high risk of stroke/systemic embolism when a decision is made to interrupt VKA therapy for an invasive procedure with a low/moderate or high risk of bleeding. Such patients would include those with “valvular AF” (mechanical heart valves or moderate severe mitral valve stenosis), NVAF with a CHADS2 score of 5-6, and those with a recent stroke or TIA. Before the procedure, patients should receive 3 days of LMWH bridging, although the dose of LMWH bridging should be reduced by 50% on the day before the procedure (Fig. 13).

8.4.3.3 No bridging for interrupted DOACs

Bridging is not necessary for DOAC-treated patients who need DOAC interruption. DOACs have a rapid onset and offset of action and relatively short elimination half-lives (ie, 10-14 hours), rendering bridging unnecessary from a pharmacological perspective.^[405] In addition, the use of bridging with interrupted DOAC has been associated with more bleeding and limited efficacy. Increased bleeding with bridging was shown in the prospective observational Dresden Registry (76% rivaroxaban, 24% dabigatran) in which heparin bridging did not reduce major cardiovascular events (stroke, venous thromboembolism [VTE], or ACS; 0.8% with no bridging vs 1.6% with bridging; P ¼ 0.265), but led to significantly higher rates of major bleeding complications (2.7% vs 0.5%; P ¼ 0.010).^[406] Post hoc analyses of the phase III DOAC trials^[21]-[25] support no bridging for DOAC interruption. Interruption of dabigatran, rivaroxaban, and apixaban without bridging in these studies had comparable rates of bleeding and thromboembolic events compared with their respective VKA groups.^{[407]–[409]} The Perioperative Anticoagulant Use for Surgery Evaluation for Patients on a Direct Oral AnticoagulantWho Need an Elective Surgery or Procedure (PAUSE) study was a large prospective cohort study of 3007 DOAC-treated AF patients scheduled to undergo an elective surgery or procedure.^[410] This study demonstrated the safety with preserved efficacy of a simple standardized DOAC interruption and resumption protocol that did not involve perioperative heparin bridging or preoperative coagulation function testing. Specifically, interruption of DOAC was associated with low rates of major bleeding (1.35% in the apixaban cohort, 0.90% in the dabigatran cohort, and 1.85% in the rivaroxaban cohort) and low rates of stroke/systemic embolism (0.16% in the apixaban cohort, 0.60% in the dabigatran cohort, and 0.37% in the rivaroxaban cohort).

8.4.3.4 OAC resumption

After the invasive procedure, DOAC is reintroduced after hemostasis has been achieved (usually 24 hours after a low/moderate bleed risk procedure and 48-72 hours after a high bleed risk procedure). VKA therapy may be reintroduced on the same evening after the procedure, recognizing that it will take several days for the INR to achieve therapeutic range. When bridging LMWH is used, we suggest using a therapeutic-dose regimen for 3 days before the procedure and to resume 24 hours after a low/moderate bleed risk procedure and 48-72 hours after a high bleed risk procedure. In selected patients, for example, those undergoing a cardiac, intracranial, or spinal procedure, it is suggested to forego postprocedure LMWH bridging altogether, although low-dose heparin regimens, typically used for VTE prophylaxis, can be considered post procedure. The recommended approach to OAC interruption is presented in Figures 12 and 13 and Table 9.

8.5 Bleeding

8.5.1 Prevention

Bleeding is a known adverse effect of OAC. Despite the similar or lower risk of bleeding with DOACs vs VKAs, there remains a significant residual 2%-4% annual risk of major bleeding. Assessment of bleeding risk is important to understand the risk-benefit balance of OAC, to identify and optimize modifiable bleeding risk factors, to identify patients who might benefit from more frequent follow-up, and to provide a framework for ongoing monitoring. The use of a validated bleeding risk prediction algorithm appears to be better at predicting major bleeding than using individual modifiable bleeding risk factors.^{[411]–[413]} Despite lack of supportive RCT evidence, the identification of modifiable bleeding risk factors before OAC initiation and at each point of contact should be routinely done. GI protection for patients at high risk of GI events and require concomitant ASA or NSAID therapy has been associated with a reduction in GI bleeding and ulcer complications. In one of the earliest studies, albeit a small study with limited number of events, patients with acute GI bleeding or obstruction due to ulcer who received Helicobacter pylori eradication therapy restarted ASA 100 mg daily treatment with lansoprazole 30 mg daily or placebo.^[414] The trial was stopped early after the second of 2 planned interim analyses revealed a statistically significant reduction in recurrent bleeding in the patients who received lansoprazole (1.6% vs 15%; HR, 10.6; 95% CI, 1.3-86.1). Subsequently the benefit of using a PPI for patients with no baseline evidence of peptic ulcer disease who required daily ASA (in doses ranging from 75 to 325 mg daily) was shown in the Randomized, Double-Blind, Placebo Controlled Study to Assess the Prevention of Low Dose Acetylsalicylic Acid (ASA) Associated Gastroduodenal Lesions and Upper Gastrointestinal Symptoms in Patients Taking Esomeprazole 20 mg Once Daily for Two Weeks (ASTERIX) study.^[415] In 991 patients 60 years of age or older, the additional use of esomeprazole 20 mg daily reduced the risk of endoscopically diagnosed gastric and duodenal ulcers (1.6% with esomeprazole vs 5.4% with placebo; P ¼ 0.0007). In addition, the use of PPIs lowered the risk of upper GI bleeding in patients who required DAPT with ASA and clopidogrel in the Clopidogrel and the Optimization of Gastrointestinal Events Trial (COGENT) trial, and decreased the occurrence of peptic ulcers in patients at high concomitant risk of cardiovascular and GI events who were taking ASA in the Randomized, DoubleBlind, Parallel Group, Multicentre, Phase III Study to Assess the Effect of Esomeprazole 20 or 40 mg od Versus Placebo on the Occurrence of Peptic Ulcers During 26 Weeks in Subjects on Continuous Low Dose Acetylsalicylic Acid (OBERON) trial.^[416],[417] There is no published RCT in which PPI use was investigated in patients with AF who were receiving standard stroke prevention dose OAC alone or in combination with or. In a retrospective cohort study of > 1.6 million Medicare beneficiaries who were receiving anticoagulation, PPI co-therapy was associated with a lower rate of upper GI bleeding hospitalizations: incidence rate ratio, 0.66 (95% CI, 0.62-0.69).^[418] However, more recently in the COMPASS – Proton Pump Inhibitor (COMPASS-PPI) study^[419] the effect of PPI use in patients receiving ASA in combination with rivaroxaban 2.5 mg BID was investigated. In this study, patients with stable atherosclerotic vascular disease and no clinical need for a PPI who were receiving 1 of 3 regimens: rivaroxaban 2.5 mg BID with ASA 100 mg daily, rivaroxaban 5 mg BID, or ASA 100 mg daily without anticoagulation, were randomized to pantoprazole 40 mg daily or placebo. The main reason for exclusion from the PPI arm of this trial was physician judgement that their patients had a clinical need for a PPI; this accounted for more than 35% of eligible patients (n ¼ 9797). Of the 17,598 patients randomized, the mean age was 68 years, 78% were male, and 2.6% had baseline peptic ulcer disease. These patients were followed for a mean of 3 years. Clinically significant GI events were no different in the 2 arms: 1.2% in the PPI arm vs 1.3% in the placebo arm (HR, 0.88; 95% CI, 0.67-1.15). The event rate in the nonrandomized cohort was 0.6%. In an analysis of the individual GI events included in the primary outcome, there was a reduction in overt gastroduodenal bleeding with pantoprazole (0.4% per year) over placebo (1.2% per year; HR, 0.52; 95% CI, 0.28-0.94; P ¼ 0.03). In addition, the rate of discontinuation of pantoprazole was similar to that of placebo overall (21.4% vs 22.4%) and for serious adverse events (0.9% vs 0.75%). Of note, a Cochrane review is currently under way to assess the benefits and harms of PPIs vs histamine-2 receptor antagonists or placebo for the prevention of upper GI bleeding in adults receiving single-agent and combination antithrombotic therapy for cardiovascular conditions.^[420]

8.5.2 Management of a bleeding event

DOACs are the preferred agents for stroke prevention in NVAF patients who merit anticoagulation. Although less lifethreatening bleeding was shown with DOACs, annual rates of major bleeding were observed to be 2%-4%.^{[21]–[23],[25]} Bleeding management protocols for DOACs have included supportive therapy alone, antifibrinolytic therapy, hemostatic factors such as fresh-frozen plasma, 3- or 4-factor prothrombin complex concentrates, and recombinant activated factor VII; activated charcoal for overdose or unintentional ingestion; dialysis or continuous renal replacement therapy; and, specific reversal agents such as idarucizumab and andexanet alfa. An approach to bleeding is outlined in Figure 14.

Andexanet alfa is a recombinant modified human factor Xa decoy protein with high affinity for the active site of factor Xa inhibitors, sequestering them within the vascular space.^[428] In patients requiring anticoagulation that involves inhibition of Factor Xa, the presence of andexanet alfa will inhibit the anticoagulant effectiveness. The pharmacodynamic half-life is approximately 1 hour. The Andexanet Alfa, A Novel Antidote to the Anticoagulation Effects of Factor Xa Inhibitors (ANNEXA-4) was a single-group cohort study designed to assess the efficacy and safety of andexanet alfa in patients with acute major bleeding occurring while taking a factor Xa inhibitor. Patients who had received 1 of 4 factor Xa inhibitors (apixaban, edoxaban, enoxaparin, or rivaroxaban) within 18 hours were treated with a bolus and 2-hour infusion of andexanet alfa.^[429] The dose of andexanet alfa was dependent on the timing of the last dose of factor Xa inhibitor. Patients were excluded if they required surgery within 12 hours, if they received other bleeding treatments such as prothrombin complex concentrate, recombinant factor VIIa, whole blood, or plasma, and if they had a Glascow Coma Scale of < 7, an estimated ICH volume of > 60 mL, or an expected survival of < 1 month. Patients enrolled (n ¼ 352) had a mean age of 77 years. Bleeding was predominantly ICH (64%) and GI (26%). Most of the patients were receiving apixaban (54%) or rivaroxaban (36%). Anti-factor Xa activity decreased by 92% for apixaban (149.7 ng/mL to 11.1 ng/mL) and rivaroxaban (211.8 ng/mL to 14.2 ng/mL). Of 249 patients who were eligible for the hemostatic efficacy analysis, 204 were adjudicated as having excellent or good hemostasis at 12 hours. Reduction in anti-factor Xa activity was not predictive of hemostatic efficacy overall but was modestly predictive in patients with ICH. At 30 days of follow-up of the overall study population 49 (14%) of the patients had died and 34 (10%) had a thrombotic event. All thrombotic events occurred before reinitiation of OAC. Andexanet alfa received accelerated Food and Drug Administration approval in the United States in 2018 for the management of life-threatening or uncontrolled bleeding in patients treated with rivaroxaban and apixaban.^[430] A condition of this accelerated approval was the requirement to complete an RCT including patients with acute ICH, the results of which might affect the modification or withdrawal of approval. The Food and Drug Administration also mandated a black box warning that andexanet alfa has been associated with thromboembolic events, ischemic events, cardiac arrest, and sudden death. There appears to be some increase in endogenous thrombin generation potential upon bolus administration of andexanet alfa although the mechanism is not yet known^[431] and did not appear to be associated with clinical thrombotic events in healthy volunteers.^[428] The thrombosis rate in a single health system retrospective cohort study of 13 patients who received andexanet alfa was 30% (4 patients) and mortality was 15% (2 patients).^[432] In another, the thrombosis rate was 0 and mortality was 40%.^[433] Practical challenges reported with andexanet alfa include access to product, lack of availability of anti-factor Xa levels, and drug preparation requiring reconstitution of multiple vials.^[433]

8.6 Secondary stroke prevention

8.6.1 Stroke during OAC treatment

Use of OAC is associated with a substantial risk reduction in the occurrence of ischemic stroke (see section 8.2), however, some patients treated with OAC will still experience ischemic stroke or TIA despite use of OAC.^[51],[52] Studies have shown that strokes that occur in patients receiving OAC tend to be less severe than strokes that occur in AF patients not taking OAC; in VKA-treated patients, stroke severity has been inversely correlated with INR level at the time of the event.^[434],[435] After confirmation of the stroke diagnosis, an on-treatment event should prompt an evaluation for potential causes (other than AF) and ensure optimization of OAC and risk factor management.

8.6.1.1 Potential causes of an ischemic stroke in a patient receiving OAC

I. Inadequate intensity of anticoagulation. In patients treated with VKAs, a subtherapeutic INR is associated with an increased risk of stroke as well as greater stroke severity.^[434],[435] In those treated with DOACs, undertreatment (eg, low plasma levels) has been associated with higher risk of thromboembolic events, and increased stroke severity (eg, large vessel occlusion).^{[205]–[207],[436]} For DOAC-treated patients, undertreatment might take the form of the prescription of a reduced dose without an indication to do so, failure to anticipate drug-drug interactions (eg, OAC levels might be decreased by interaction with anticonvulsants such as phenytoin) or, in the case of rivaroxaban, a failure to account for GI absorption. Specifically, rivaroxaban should be taken with food for maximum GI absorption.

II. Suboptimal medication adherence or persistence. Poor patient adherence is a common problem with many chronic medications, adherence and persistence to stroke prevention therapies being particularly problematic in the AF population. As outlined in section 7.2 suboptimal adherence and persistence with OAC has been associated with higher rates of all-cause mortality and stroke.^{[156],[157],[160]–[162]} Missed doses is a particular concern for DOACs because of their short half-life (average 12 hours). Strategies to improve persistence and adherence are outlined in section 7.2 and Table 6.

III. Alternate stroke etiology. Patients with AF might have other etiologies for stroke, some of which require specific management. Examples include large artery atherosclerosis of the intra- or extracranial circulation, cerebral small-vessel disease, other cardiac sources of embolism eg, patent foramen ovale) or another determined cause (carotid or vertebral artery dissection, vasculitis, etc).

IV. Suboptimal risk factor management. Untreated or suboptimally controlled stroke risk factors (eg, hypertension, dyslipidemia, etc) might contribute to the occurrence of stroke, independent of the presence of AF. Of particular importance, reduction in BP reduces the risk of recurrence of ischemic stroke as well as incident hemorrhagic stroke.^[118]

8.6.1.2 Practical clinical approach to patients with an ischemic stroke who are receiving OAC therapy

I. Ensure an accurate diagnosis of the event. Brain imaging with CT or MRI is recommended for the diagnostic evaluation of acute stroke to exclude ICH, hemorrhagic transformation of ischemic stroke, and other structural pathology (MRI has greater diagnostic sensitivity than CT for identifying acute ischemia and small lesions). TIAs are frequently overdiagnosed, with up to 60% of patients suspected to have had a TIA having nonischemic causes for their symptoms after systemic evaluation.^[437],[438] Although TIA remains a clinical diagnosis mainly on the basis of a detailed history of the event, brain imaging is useful in assessing for mimics (eg, migraine aura, seizure, peripheral vertigo, presyncope, neuropathy, multiple sclerosis) including small subarachnoid hemorrhages.^[439]

II. Investigate for the most likely etiology of the stroke event. A typical etiological stroke workup for a patient with known AF consists of brain imaging (CT or MRI); vascular imaging (carotid ultrasound at a minimum; ideally CT angiography or MR angiography from the aortic arch to vertex), echocardiography, and laboratory investigations.

III. Determine if any other treatments are indicated for stroke risk reduction. For example, symptomatic carotid artery stenosis that is moderate (50%-69%) or severe (70%-99%) might benefit from urgent revascularization, whereas nonsurgical management is usually recommended for asymptomatic carotid stenosis.^[118]

IV. Assess medication adherence. Patients and/or family members should be questioned regarding missed doses or prolonged interruptions (eg, periprocedural or in association with a bleeding event). For VKA-treated patients, the INR record for the preceding months should be reviewed. Patients should be counselled about the importance of daily medication adherence, the dangers of missed doses, and avoidance of prolonged or unnecessary treatment interruptions. Encourage adherence- and persistence-enhancing strategies as outlined in section 7.2 and Table 6.

V. Determine whether or not any changes are needed to the patient’s current OAC (agent or dose). For VKA-treated patients with suboptimal INR control, either switch to a DOAC (if eligible) or aim for improved INR control (consider a higher target INR and more frequent INR monitoring). For DOAC-treated patients, it is reasonable to continue the same DOAC (after ensuring it is at the correct dose for the patient’s age, renal function, and body weight) or switch to a different DOAC. For patients unable to comply with DOACs, then a switch to a VKA might be reasonable because of the longer half-life and ability to monitor the INR. The additional use of an antiplatelet agent is sometimes considered, however, the additional use of an antiplatelet agent with therapeutic OAC is associated with a significant increase in bleeding events without an additional benefit for stroke prevention. LAA occlusion (LAAO) might be an option for some patients but the optimal candidates for this procedure remains unclear.

VI. Identify and address any untreated vascular risk factors. Optimize BP and lipids, recommend smoking cessation, and promote general secondary stroke prevention lifestyle recommendations (diet, exercise).

For further details on the etiological stroke workup and secondary stroke prevention treatment recommendations, see the Canadian stroke best practice recommendations^[118] available at: strokebestpractices.ca.

8.6.2 Timing of OAC initiation after acute ischemic stroke in patients with AF

In the 2-week period immediately after a TIA or ischemic stroke, patients with AF have an increased risk of ischemic stroke that ranges from 0.5%-1.3% per day.^[440],[441] Initiation of OAC during this phase must balance the benefit of OAC with the risk of hemorrhagic transformation of the acute brain infarct. Hemorrhagic transformation can range from asymptomatic petechiae to symptomatic parenchymal hematoma with mass effect.^[442],[443] Hemorrhagic transformation is more frequent in large-volume infarcts and in the setting of thrombolysis or mechanical revascularization of the index stroke.^[442],[443] The optimal timing for OAC initiation post stroke has not yet been clearly defined and practice patterns are highly variable.^[444] RCTs are under way to help address this uncertainty. For now, the heterogeneity of strokes and the lack of randomized data (particularly for DOACs in the early poststroke period) preclude any specific recommendations regarding the timing of OAC initiation. In practice, treatment decisions are typically individualized on the basis of a benefit-risk assessment guided by brain imaging appearance and clinical context. Bridging with low-dose aspirin is often prescribed until OAC treatment is initiated. Serial head CT scans within the first days or weeks post stroke can be helpful to monitor the evolution of an acute infarct, especially for moderate or large infarcts, and to assess for the presence and extent of any hemorrhage, before OAC treatment initiation. Outcomes in relation to timing of OAC initiation have been studied in prospective cohort studies.^[445],[446] In the Clinical Relevance of Microbleeds in Stroke-2 study 1490 participants with AF and stroke or TIA in whom anticoagulation was indicated were enrolled.^[446] Of these, 1335 (90%) had a known date of OAC initiation post stroke that was dichotomized as early (≤ 4 days) or late ( ≥ 5 days or never). The groups differed significantly, with lower stroke severity, better premorbid functioning, and a lower probability of thrombolysis for the index event in those in the early OAC group. At 90 days of follow-up the combined end point of intracerebral hemorrhage, ischemic stroke, and death was similar between groups (2% with early and 5% with late initiation; adjusted OR, 1.17; 95% CI, 0.48-2.84; P ¼ 0.736). Recognizing that the event rate was low, the authors concluded that there did not appear to be a hazard associated with early anticoagulation in carefully selected patients. Results of a prospective cohort study consisting of 1029 individuals with ischemic stroke and AF suggested that initiation of anticoagulation between 4 and 14 days optimized the combined outcome of stroke, TIA, systemic embolus, symptomatic intracranial bleeding, and bleeding within 90 days of onset.^[445] Most participants were treated with VKAs or mixed treatment protocols including heparin, with only 12% treated with DOACs. Results of 2 small RCTs have suggested the early use of DOACs is safe for small strokes using an MRI measure of hemorrhagic transformation.^[447],[448] Participants (N ¼ 195) randomized to VKA or rivaroxaban with a median National Institutes of Health Stroke Scale (NIHSS) score of 2 (ie, very mild strokes) did not differ in the rate of hemorrhagic transformation detected using MRI, an imaging modality with high sensitivity for the detection of hemorrhage.^[447] The Dabigatran Following Acute Transient Ischemic Attack and minor stroke trial (DATAS II) randomized 301 individuals with a median NIHSS score of 1 (ie, very mild strokes) to ASA or dabigatran within 72 hours of symptom onset. No symptomatic hemorrhagic transformation was observed and the rate of minor petechial hemorrhage within the acute infarct did not differ between the treatment arms. Infarct volume predicted the probability of hemorrhagic transformation.^[448] These trials do not establish optimal time points for OAC initiation but suggest that an identifiable group of patients with very mild stroke severity and small volume infarcts can be safely anticoagulated early.

I. It is reasonable to initiate a DOAC 1 day post event for patients with a TIA, 3 days for patients with a small infarct/mild stroke severity (NIHSS score < 8), 6 days in patients with a moderate-sized infarct/moderate severity stroke (NIHSS score 8-15), and 12-14 days in patients with a large infarct/severe stroke (NIHSS score >15). For a brief-duration TIA with no residual symptoms or deficits and no acute infarct or hemorrhage on head CT scan, it is reasonable to consider DOAC initiation on the day of the event provided timely follow-up is available (Figure 15).

II. It is reasonable to delay OAC initiation for more than 2 weeks post stroke if, in the judgement of the clinician the risk of bleeding is believed to be high (eg, for some patients with large infarcts and those with hemorrhagic transformation).

III. If OAC initiation is recommended after hospital discharge, coordination with the treating physicians and close followup is suggested because postdischarge recommendations are implemented in only two-thirds of patients.^[449]

IV. Hypertension is a significant risk factor for intracerebral hemorrhage, and it is reasonable to delay OAC in the setting of uncontrolled hypertension.

V. In most instances, brain imaging should be repeated before initiating OAC to detect asymptomatic hemorrhagic transformation that might influence the timing for OAC initiation. Consideration should be given to delaying OAC in patients with hemorrhage other than minor petechial changes, because OAC in the presence of existing hemorrhage might increase bleeding. Delayed initiation in patients with hemorrhagic transformation was not associated with increased risk of ischemic stroke in an observational study of > 2000 patients with stroke and AF.^[450]

VI. Neurological instability might suggest recurrent ischemia or hemorrhage and in either case, might require a repeat CT examination or reassessment of the timing of OAC initiation.

8.6.3 OAC initiation after hemorrhagic stroke

For patients with AF who have had an acute primary intracerebral hemorrhage, subdural, or subarachnoid hemorrhage, OAC is typically avoided in the acute subacute period, and decisions regarding OAC initiation should be made in consultation with a stroke/neurology specialist on the basis of careful risk stratification guided by brain MRI appearance, the presumed etiology for the hemorrhage, and the estimated risk of recurrence. Patients with a lobar (superficial) ICH attributed to cerebral amyloid angiopathy have a higher risk of recurrent ICH than patients with a deep hypertensive ICH. If OAC is initiated post-ICH, a DOAC is preferred over VKAs because of the lower rate of incidence of ICH with these agents.^[52] Observational studies support the cautious use of OAC for selected patients post ICH, usually initiated weeks to months post ICH, but clinical equipoise exists.^[451],[452] RCTs are currently under way to evaluate the safety and efficacy of DOAC vs aspirin in AF patients who have had an ICH.

8.7 Left atrial appendage occlusion

Imaging and postmortem studies suggest that most AFassociated ischemic strokes are cardioembolic, and that the majority arise from the LAA.^[453],[454] In sinus rhythm the LAA has pulsatile flow, however in AF appendage emptying is reduced leading to stasis and clot formation. Physical elimination of the LAA (removal or occlusion) from the circulation is postulated to prevent thrombus formation and subsequent embolization, without suffering from the limitations of pharmacological therapy. Specifically, although OAC is effective at preventing AF-associated stroke/systemic embolism, its use is limited by short- and long term nonadherence, nonpersistence, and side effects such as the risk of major bleeding. Conversely, when performed, the effects of the physical elimination of the LAA are considered permanent and are not reliant on patient compliance.

8.7.1 Percutaneous LAAO

Two RCTs and a number of single and multicentre registries that have evaluated percutaneous device closure have been performed. The WATCHMAN LAA Closure Technology for Embolic Protection in Patients With Atrial Fibrillation (PROTECT AF) study enrolled 707 patients with AF and a CHADS2 score of ≥ 1, and the Prospective Randomized Evaluation of the WATCHMAN LAA Closure Device in Patients With Atrial Fibrillation vs Long-Term Warfarin Therapy (PREVAIL) study enrolled 407 patients with a CHADS2 score of 2.^[455],[456] Both studies randomized patients to LAAO or VKA. The 5-year outcome data from these 2 studies were combined in a meta-analysis.^[457] Although LAAO was deemed noninferior to VKAs for the combined end points, much of the benefit of LAAO is due to a reduction in intracranial bleeding (0.2% per year with LAAO vs 0.9% per year with VKA; HR, 0.20; 95% CI, 0.07-0.56; P ¼ 0.002).^[457] However, the effect of LAAO on ischemic stroke remains to be determined, and is suggested to be inferior (1.6% per year with LAAO vs 0.95% per year with VKA; HR, 1.71; 95% CI, 0.94-3.11; P ¼ 0.08).^[457] Higher risk of ischemic stroke might relate to device-related thrombosis, which occurs in 2%-4% of cases and is a known independent predictor of stroke (HR, 4.4; 95% CI, 1.05-18.43).^[458] In the recently published randomized multicentre Left Atrial Appendage Closure Versus Novel Anticoagulation Agents in Atrial Fibrillation (PRAGUE-17) study the safety and efficacy of LAAO was compared with DOAC in high-risk AF patients with a history of significant bleeding or thromboembolic event while receiving OAC (CHA2DS2-VASc score 4.7 ±1.5; HAS-BLED score 3.1 ± 0.9). This study showed no significant difference between LAAO and DOAC, with similar rates of all-cause stroke/TIA (2.20% with LAAO vs 2.68% with DOACs), ischemic stroke/TIA (2.20% with LAAO vs 2.38% with DOACs), and major and CRNM bleeding (5.5% per year with LAAO vs 7.42% per year with DOACs). There was no difference in the end points of cardiovascular death, noncardiovascular death, or all-cause mortality. Major LAAO implant-related complications occurred in 4.5%. LAAO implantation was unsuccessful in 10.0% of the patients, however, the results of the study did not differ when analyzed according to treatment received (P ¼ 0.31) or per protocol (P ¼ 0.40).

8.7.2 Surgical LAA

The device-based trials provide proof of concept that LAAO might provide benefit, however they do not address the surgical AF population, do not inform on the combination of LAAO and OAC, and do not evaluate a surgical intervention that can be offered with little increase in risk. Results of existing observational studies suggest that standalone surgical LAAO can be safely and effectively performed, however, there are limited high-quality data to suggest that a stand-alone surgical LAAO approach is reasonable in patients eligible for a percutaneous approach.^[459] In patients with AF who are already undergoing a thoracotomy, it is possible that concomitant surgical LAAO might offer significant benefit with minimal incremental risk. However, it is important to recognize that patients who undergo surgical LAAO often have higher CHA2DS2-VASc scores and more often have coexisting valvular heart disease (ie, are more likely to develop AF-associated thrombosis outside of the LAA).^[454] To date the observational data of concomitant surgical LAAO are somewhat conflicting, with some observational series suggesting that surgical LAAO is associated with significantly greater freedom from stroke/systemic embolism, and others suggesting that LAAO did not reduce the incidence of stroke/systemic embolism in patients with AF (P ¼ 0.69), with the apparent stroke reduction being related to continued VKA use and not the LAAO procedure itself.^[460],[461] However, these observational studies do not have sufficient power or freedom from bias to provide the level of evidence needed to clearly answer this important question. A large RCT of surgical exclusion of the LAA vs VKA therapy has completed recruitment of > 4800 patients and is in the follow-up phase.^[462] The primary end point is stroke/systemic embolism over 4 years. Total mortality and safety end points will also be compared. Until this trial is completed, the only basis for recommendations regarding surgical LAA removal is consensus. As such, we have made a weak recommendation on the basis of the existing low-quality evidence.

8.8 Cognitive function/dementia

An association between AF and cognitive impairment has been frequently reported over the past 30 years.^[463],[464] Although the association has not been consistently observed,^[465],[466] recent meta-analyses have shown a significant correlation between AF and cognitive impairment, irrespective of a history of previous stroke.^{[467]–[469]} It has been estimated that the risk of dementia in individuals with AF and no overt cerebrovascular events is increased 30%-60% compared with patients without AF, even after adjustment for age, sex, multiple comorbidities, and cardiovascular medications.^{[470]–[472]} An American study of community-dwelling individuals showed that AF was associated with a 50% increase in risk of developing Alzheimer disease compared with individuals without AF.^[473] The findings are not just unique to North America and Europe, with a Taiwanese study involving 332,665 patients,^[470] and a South Korean study of 262,611 patients^[472] demonstrating that patients with AF had a 42% and a 63% higher risk of dementia, respectively, even after adjustment for age, sex, baseline differences, and medication. More recent data suggest that AF increases the risk of vascular and degenerative (eg, Alzheimer disease) dementia,^[474],[475] and also that the relationship between AF and cognitive decline appears to be strongest when AF develops in middle age,^[476],[477] when the AF is persistent and of long duration,^[477],[478] and when the ventricular rate in AF is fast.^[479],[480] Although AF is related to cognitive impairment and dementia, the etiological mechanisms remain a matter for conjecture. Shared risk factors have been proposed as part of the explanation. The CHADS2 and CHA2DS2-VASc scores have been shown to predict cognitive impairment as well as stroke in patients with AF. Nevertheless, the association between AF and cognitive decline persists even after controlling for age and these comorbidities.^{[467],[470],[481]} Proposed mechanisms include silent cerebral infarcts due to microemboli, cerebral microbleeds, disruptions of the blood-brain barrier, cerebral vascular disease, cerebral hypoperfusion due to variability in the cardiac cycle and reduced cardiac output, oxidative stress, inflammation, and endothelial dysfunction.^{[471],[482]–[485]} Of these, subclinical embolic cerebrovascular ischemia is thought to be the most likely mechanism.^[485] Although white matter disease had been raised as another potential driver, recent work appears to refute this.^{[482],[486],[487]} Treatments that have been proposed to reduce the risk of cognitive impairment include risk factor management,^[488] sinus rhythm maintenance with ablation,^[489] and anticoagulation.^{[490]–[492]} A modest effect has been observed with specific risk factor treatments, such as statins or angiotensin conversion enzyme inhibitor and angiotensin receptor blocker therapy in some studies,^[481],[488] but not in others.^[493] There also remains controversy concerning the ability of anticoagulants to decrease the incidence of cognitive impairment and dementia in patients with AF,^[494] with reviews of the available literature failing to find convincing evidence that they do so.^[495],[496] Although positive data continue to accumulate, including from large population based studies,^[497],[498] there have been no clinical trials to eliminate potential confounders. VKA use appears to be most effective when TTR is kept consistently very high (eg, > 75%).^[499],[500] Data relating to the DOACs and dementia are sparse, with some studies showing relative benefit compared with VKAs,^{[491],[501],[502]} whereas a nationwide Danish cohort study did not.503 Several clinical trials are being undertaken to allow greater clarity. In 2 of these studies, VKAs are being compared with dabigatran (Cognitive Impairment Related to Atrial Fibrillation Prevention Trial [GIRAF; NCT01994265] and Impact of Anticoagulation Therapy on the Cognitive Decline and Dementia in Patients With Non-Valvular Atrial Fibrillation [CAF; NCT03061006]) and in a third study, Blinded Randomized Trial of Anticoagulation to Prevent Ischemic Stroke and Neurocognitive Impairment in Atrial Fibrillation (BRAIN-AF; NCT02387229), rivaroxaban is being compared with standard of care. There are currently only 2 observational studies on the association between catheter ablation and cognitive dysfunction. The largest was a retrospective case control study from Intermountain Health in Utah, in which outcomes in 4212 patients who underwent AF ablation were compared with 16,848 age- and sex-matched patients with AF who did not undergo ablation. At 3 years, the incidence of dementia was low in both groups, but it was significantly less in the AF patients who had undergone ablation.^[489] A small prospective case-control study was recently published, which involved 308 AF patients treated with ablation and 50 medically managed control participants.^[504] Cognitive function was assessed at baseline, 3 months, and 1 year. Significant cognitive improvement was observed at 3 and 12 months in the ablation group compared with the control group, and no major adverse events were reported at either point in time. A larger prospective case-control study (Cognitive Impairment in Atrial Fibrillation [DIAL-F]; NCT01816308) is specifically addressing the effect of AF catheter ablation on dementia and is currently under way. Finally, LAAO might be expected to have a potential role on decreasing the incidence of cognitive impairment on the basis of subclinical strokes. However, whether the available devices can prevent microemboli remains to be seen.