A case-based approach to implementing guidelines for stroke prevention in patients with atrial fibrillation: balancing the risks and benefits

Atrial fibrillation (AF) puts patients at risk of complications, including stroke. Warfarin therapy has been the mainstay of antithrombotic treatment for reducing the risk of stroke in AF. However, warfarin has limitations that have motivated development of several novel oral anticoagulants (NOACs), including dabigatran, rivaroxaban, apixaban, and edoxaban. Clinical trials demonstrate that the NOACs offer efficacy and safety that are equivalent to, or better than, those of warfarin for reducing the risk of stroke in patients with nonvalvular AF. This review examines stroke risk reduction in patients with AF from the perspective of the clinician balancing the risks and benefits of treatment options, evaluates the most recent guidelines, and discusses 2 hypothetical patient cases to better illustrate how clinicians may apply available data in the clinical setting. We reviewed guidelines for the reduction of stroke risk in AF and data from clinical trials on the NOACs. Choosing antithrombotic treatment involves assessing the benefits of therapy versus its risks. Risk indexes, including CHADS2, CHA2DS2-VASc, and HAS-BLED can help determine how to treat patients with AF. Current guidelines suggest using these risk indexes to customize treatment to individual patients. Many current treatment guidelines also incorporate recommendations for the use of NOACs as an alternative to warfarin. As additional data emerge and guidelines are updated, these recommendations will likely evolve. In the interim, clinicians may consider published guidelines and clinical trial results on NOACs. Real-world experience will provide clinicians with additional insight into their treatment decisions.


Introduction
Successful healthcare interventions improve patient outcomes and reduce costs associated with disease management. Guidelines provide a standardized, evidence-based approach to the diagnosis and treatment of disease, with a goal of optimizing health outcomes. When available, preventive measures to reduce the risk of disease complications are an important element of treatment. Atrial fibrillation (AF), which affected an estimated 5.2 million adults in the United States in 2010, is a disease state in which patients are at risk of complications, including stroke [1]. AF is a highly prevalent cardiac arrhythmia that imparts a 2-to 7-times greater risk of embolic or transient stroke [2,3]. AF-related stroke is more severe than stroke without AF, and is associated with greater disability and subsequent medical needs [4,5]. The high AF-related morbidity is reflected in a substantial economic burden. For 2010, the projected incremental cost of AF in the United States ranged from $6.0 billion to $26.0 billion [6].
Antithrombotic therapy can reduce the risk of stroke in patients with AF [7][8][9][10][11]; however, many at-risk patients are untreated or undertreated [12][13][14][15]. Antithrombotic treatment for stroke risk reduction in patients with AF has traditionally focused on anticoagulation with a vitamin K antagonist (VKA), primarily warfarin [2]. Warfarin is highly effective for reducing the risk of stroke. However, warfarin has a narrow therapeutic window and many patients have difficulty remaining within the therapeutic range. Under-anticoagulation may lead to stroke, whereas over-anticoagulation may result in bleeding. Warfarin can be challenging to use in the clinical setting due to dietary and drug interactions, and variability in dose response due to genetic and environmental factors [13]. Warfarin necessitates regular monitoring and dose adjustments to enable individual patients to receive safe and effective treatment [13,16]. Many physicians overestimate bleeding risk and underestimate the benefit of stroke prevention [17]. Bleeding risk and concerns about bleeding risk, as well as other clinical features of warfarin, may lead clinicians to rule it out as a treatment option; hence, many patients remain untreated [18][19][20]. Until recently, aspirin was the only treatment alternative for patients unsuitable for VKA therapy. Although bleeding risk is lower and administration is simpler with aspirin, the efficacy of aspirin in stroke prevention is inferior to that of warfarin [8,21]. Thus, the conventional treatment options of warfarin and aspirin have left a segment of the patient population untreated and/or undertreated and at increased risk of stroke [22].
The limitations of the traditional options for stroke prevention in patients with AF have motivated the development of several novel oral anticoagulants (NOACs). Recent clinical trials have found that the NOACs offer efficacy and safety equivalent to, or better than, those of warfarin for reducing the risk of stroke in patients with nonvalvular AF (NVAF) [23][24][25][26]. As more clinical experience is gained with the NOACs, changes in treatment paradigms and guidelines will undoubtedly ensue. This review examines stroke prevention in patients with AF from the perspective of the clinician trying to balance the risks and benefits of treatment options. Use of treatment guidelines and the application of findings from recent clinical trials will help facilitate high-quality and cost-effective medical care; however, effective care must be tailored to the individual patient. In addition to reviewing the most recent guidelines for stroke prevention in AF, we will discuss 2 hypothetical patient cases to better understand how clinicians may apply these measures and the available data in the clinical setting.

Stroke risk reduction in NVAF with NOACs
Several NOACs are available in the United States for reducing the risk of stroke in patients with NVAF, including the direct thrombin inhibitor dabigatran and the factor Xa inhibitors rivaroxaban, apixaban and edoxaban [26][27][28][29][30]. A brief summary of the results of key phase 3 clinical trials comparing the NOACs with either warfarin or aspirin for reducing the risk of stroke in NVAF is provided in Table 1 [23-26, 31, 32]. These results provide support for the integration of NOACs into treatment paradigms for stroke risk reduction in AF as potential first-line options.

Guidelines and quality measures for stroke risk reduction in AF
Several guidelines discuss stroke risk reduction in patients with AF (Table 2), including recommendations from the American College of Cardiology (ACC), the American College of Chest Physicians (ACCP), the American Heart Association (AHA), the American Stroke Association (ASA), the Heart Rhythm Society (HRS), the European Society of Cardiology (ESC), and the American Academy of Neurology (AAN) [7][8][9][10][11]. In general, treatment decisions are based on an assessment of the individual patient's stroke risk, with a concurrent assessment of bleeding risk and other patient-related factors, such as his or her ability to adhere to monitoring requirements and personal preferences [7,10]. However, specific recommendations vary as to how to assess these risks, when to treat with antithrombotic therapy, and which treatments to use.
The CHADS 2 and CHA 2 DS 2 -VASc risk indexes predict the risk of stroke in patients with AF. Both the ACC/AHA/ HRS guidelines and the ESC guidelines mention the ATRIA and HEMORR 2 HAGES scores for bleeding risk, but acknowledge that the HAS-BLED risk index is most predictive of bleeding events. Together, these scoring systems can be used to evaluate an approach to treatment in patients with AF. The CHADS 2 score is based on a patient's history of Congestive heart failure (1 point), Hypertension (1 point), Age ≥75 years (1 point), Diabetes (1 point), and history of Stroke or transient ischemic attack (TIA, 2 points; Table 3) [33]. Scores of 0, 1, or ≥2 indicate low, moderate, or high stroke risk, respectively [8,34]. The CHADS 2 risk scoring index has been thoroughly validated [33][34][35]; is simple, inexpensive, and broadly applicable; and most importantly, it allows identification of patients at low risk of stroke who may not receive significant benefit from anticoagulant therapy. However, the CHADS 2 risk score has limitations, primarily that patients deemed at low risk with CHADS 2 still have a stroke risk of 2.2 % per year, in addition, there is a lack of differentiation between stroke risk factors and risk factors that do not consistently predict stroke (i.e., congestive heart failure). As a result, some patients may be identified as moderate-rather than low-risk stroke candidates, or vice versa [33][34][35]. Also, in keeping the CHADS 2 risk score simple, some common stroke risk factors are not included [35,36].
The ACC/AHA/HRS and ESC advocate using the CHA 2 DS 2 -VASc score instead of the CHADS 2 score to assess stroke risk. The ESC states that the CHA 2 DS 2 -VASc score more accurately identifies "truly low-risk" patients with AF who are not likely to benefit from oral anticoagulant therapy (Table 3) and, conversely, those patients who may benefit from anticoagulation because they are at risk for stroke or systemic thromboembolism [7]. The CHA 2 DS 2 -VASc score has been shown to identify some patients considered low risk by CHADS 2 (score of 0) to     [34]. In one study of the predictive value of risk classification schemes, when patients were categorized by CHA 2 DS 2 -VASc score, 0 % of low-risk patients, 0.6 % of moderate-risk patients, and 3 % of high-risk patients experienced a thromboembolic event.
In contrast, when the same cohort of patients was classified according to the CHADS 2 scoring system, 1.4 % of low-risk, 1.9 % of moderate-risk, and 3.1 % of high-risk patients experienced an event [34]. Despite these potential advantages, the CHA 2 DS 2 -VASc score index lacks the extensive validation that CHADS 2 has received, and studies have demonstrated a similar predictive ability for the 2 indexes [11,34,40]. The AAN guidelines explicitly acknowledge the discrepancies between the risk scoring systems, stating "although multiple risk stratification tools are available for estimating the absolute stroke risk of patients with NVAF, the absolute stroke risks estimated by these tools vary widely [9]." Rather than a score-based recommendation, the AAN acknowledges that determining when the benefit of reducing stroke risk will outweigh the harm of increased bleeding risk is difficult, and emphasizes the importance of patient preference and physician judgment under these circumstances [9].
Concurrent assessment of stroke and bleeding risk presents a clinical challenge, as many of the risk factors for stroke and bleeding overlap [11]. Indeed, a subgroup analysis of the RE-LY trial demonstrated an increased risk for bleeding associated with higher CHADS 2 scores in patients receiving oral anticoagulation [41]. Among the bleeding risk scores available, HAS-BLED has been independently validated, and correlates well with intracranial hemorrhage risk (Table 4) [7,42]. HAS-BLED includes the following risk factors: Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalized ratio (INR), Elderly (age ≥65 years), and Drugs (concomitant antiplatelet/nonsteroidal anti-inflammatory drug or alcohol) [7,11]. While the ESC suggests using the HAS-BLED bleeding risk   [7], the ACC/AHA/HRS, ASA, and ACCP do not endorse a specific scoring system, but do advise that an assessment of bleeding risk is needed [8,11,41]. The ESC guidelines recommend caution along with efforts to correct potentially reversible bleeding risk factors when prescribing antithrombotic therapy for patients with a HAS-BLED score ≥3. Importantly, the ESC guidelines emphasize that: whereas the HAS-BLED score should be used to identify modifiable bleeding risks, a high HAS-BLED score alone should not exclude patients from oral anticoagulant treatment [7]. NOACs have been included in the most recent updates to guidelines for stroke prevention in patients with AF (Table 2) [7][8][9][10][11]. It should be noted that edoxaban has not been included in the guideline recommendations because it was not approved at the time of publication. The ESC recommends the NOACs over warfarin based on their better efficacy, safety, and convenience [7], whereas the AHA/ASA recommend the NOACs as an alternative to warfarin in the presence of at least 1 additional risk factor for stroke [8]; neither guideline recommends one NOAC over another. While the ACC/AHA/HRS recommend any of the 3 NOACs (i.e., dabigatran, rivaroxaban, or apixaban) as an efficacious alternative to warfarin [10], the ACCP guidelines recommend dabigatran 150 mg twice daily in place of adjusted-dose VKA [11]. (The ACCP guidelines include dabigatran but not apixaban or rivaroxaban in their recommendations, as the latter 2 NOACs were not yet approved when the guidelines were drafted [11].) The ACCP guidelines favor oral anticoagulants over aspirin alone or, for patients at intermediate to high risk of stroke for whom warfarin is unsuitable, suggest aspirin given with clopidogrel; where oral anticoagulation is indicated, dabigatran is recommended over warfarin [11]. The AAN guideline update offers recommendations for specific NOACs in lieu of or as an alternative to warfarin based on type of stroke risk (i.e., general risk [dabigatran 150 mg], high risk of cerebral or systemic embolism [SE] [rivaroxaban], moderate risk of embolism [apixaban 5 mg twice daily]) [9]. Additionally, the AAN update recommends apixaban 5 mg bid over aspirin for reducing risk of stroke or SE [9], and the AAN guidelines include a recommendation for the addition of clopidogrel to aspirin as an alternative for patients with AF in whom warfarin is considered unsuitable [9]. It should be noted that while the ACC, ACCP, AHA, ASA, HRS, and ESC use the CHADS 2 , CHA 2 DS 2 -VASc, and/or HAS-BLED risk scores to determine their treatment recommendations, these risk scoring systems are based on the results of trials during the era in which patients were receiving warfarin, aspirin, or placebo, and not the new NOACs [10,[43][44][45]. The study that validated the CHADS 2 score measured stroke risk among patients who were not receiving any form of anticoagulant therapy [33].
Performance and quality measures from the Joint Commission, a US healthcare accrediting organization formerly known as the Joint Commission on Accreditation of Healthcare Organizations, and from the ACC/AHA also provide guidance on the treatment of patients with NVAF [46,47]. Published in 2008, the Joint Commission's Stroke Performance Implementation Guide recommends warfarin (unless contraindicated) for patients with AF-related stroke [47]. ACC/AHA performance measures, also published in 2008, recommend antithrombotic therapy with aspirin or warfarin for patients with NVAF based on the patient's stroke risk category [46].
More recently, the Stroke and Stroke Rehabilitation Work Group provided guidance in 2012 with quality measures to improve outcomes in stroke, TIA, and stroke rehabilitation. Composed of the AAN, American College of Radiology, National Committee for Quality Assurance, and the American Medical Association-convened Physician Consortium for Performance Improvement® (PCPI®), the Stroke and Stroke Rehabilitation Work Group advocates anticoagulant therapy with dabigatran, rivaroxaban, warfarin, or low-molecular-weight heparin at discharge for patients with AF [48].

Application of treatment guidelines to clinical practice
With multiple oral anticoagulants now available to reduce the risk of stroke in patients with NVAF, guidelines and treatment paradigms will continue to evolve to reflect findings from clinical trials and emerging clinical experience. The hypothetical cases that follow explore how clinicians might apply guidelines and clinical trial data to treatment decisions for the patient throughout the clinical course of NVAF.
Hypothetical patient case 1: Balancing the benefit of thromboprophylaxis with the risk of bleeding BL is a 64-year-old white woman who presents to the emergency room with cellulitis of the left forearm that has not responded to outpatient antibiotic treatment. BL takes metoprolol for hypertension, but no other medications. She has no history of diabetes. She is a nonsmoker who exercises regularly and occasionally has a glass of wine with dinner. Physical examination reveals the following: body mass index (BMI), 22 kg/m 2 (height, 5′7′′; weight, 139 lbs); blood pressure, 144/82 mm Hg; pulse, 78 bpm, irregular; temperature, 101°F. On electrocardiogram (ECG), an irregularly irregular rhythm is noted with no ST-elevation and a normal axis. Apart from moderate cellulitis of the left forearm and associated signs of infection, no other abnormal findings are noted during the physical examination.
BL is admitted to the hospital and the cellulitis responds to intravenous antibiotics. She asks if her irregular heart rhythm will require additional treatment or a repeat of treatment she had taken for it previously. Upon further inquiry, her hospitalist notes that 2 years ago BL had been examined for complaints of breathlessness and a rapid heart rate. After a complete cardiac workup at that time, she had been diagnosed with paroxysmal NVAF and was prescribed warfarin. BL states she didn't like the frequent blood tests that were associated with warfarin and often found it difficult to get to the anticoagulant clinic due to her work schedule. During a follow-up visit approximately 2 months after BL had started taking warfarin, her physician found that she was still in AF, although her heart rate had returned to normal, and although he recommended that she continue taking warfarin, she chose to stop it at that time.

Case discussion
Due to her history of hypertension, BL has a HAS-BLED score of 1, a CHADS 2 score of 1, and a CHA 2 DS 2 -VASc score of 2. Her HAS-BLED score indicates a low risk of bleeding. Her CHADS 2 score indicates a moderate risk of stroke, which is associated with a stroke rate of 2.8 % per year [33]. In contrast, her CHA 2 DS 2 -VASc score places her in the high-risk category, although the rate of stroke associated with a CHA 2 DS 2 -VASc score of 2 is 2.2 % per year, similar to that associated with her CHADS 2 score [10]. Her moderate to high stroke risk (depending on the scoring system used) suggests the importance of stroke prevention in this patient.
Treatment guidelines provide various recommendations for patients with a CHADS 2 score of 1 (or a CHA 2 DS 2 -VASc score of ≥2), ranging from daily aspirin to oral anticoagulation with warfarin or a NOAC (Table 2) [7][8][9][10][11]. In the present case, however, an effective treatment choice for BL must also consider her low risk of bleeding and her past noncompliance with warfarin therapy. Based on BL's history of difficulty maintaining a therapeutic INR while on warfarin and adhering to monitoring requirements, warfarin would not be the ideal first choice for this patient. Instead, one of the NOACs may provide a safe, efficacious, and convenient alternative.
Consideration of pharmacologic properties and patientspecific characteristics, such as comorbidities, may help in the identification of a specific NOAC for stroke prevention (Table 5) [23][24][25][26][27][28][29][30][31][49][50][51][52][53][54][55][56][57][58]. However, in this case, BL's medical history and concomitant medication (metoprolol) do not rule out any of the NOACs. Given a CHADS 2 score of 1, one might consider aspirin versus a NOAC. In contrast, her CHA 2 DS 2 -VASc score of 2 would suggest that an anticoagulant is a more appropriate choice. As noted previously, aspirin is less effective than warfarin in preventing stroke in patients with NVAF [21]. Of the NOACs, only apixaban has been studied head-to-head versus aspirin (AVERROES trial) [32]. The AVERROES trial demonstrated greater efficacy with apixaban in significantly reducing the risk of stroke and SE compared with aspirin (hazard ratio 0.45; 95 % confidence interval, 0.32, 0.62; p < 0.001; Table 1) [32]. For example, the annual rate of stroke or SE was 1.6 % (51/2808) with apixaban versus 3.7 % (113/2791) with aspirin, with no significant difference between treatments in rates of major (p = 0.57), gastrointestinal (GI; p = 0.71), or intracranial (p = 0.69) bleeding [31]. Based on the results of the AVERROES trial, and given the low risk of bleeding anticipated for BL based on her HAS-BLED score of 1, using an oral anticoagulant instead of aspirin on balance appears to be the better choice, considering that complications of stroke can be life-debilitating.
Hypothetical patient case 2: Identifying and managing stroke risk in the midst of comorbidities and a high risk of bleeding AS is a 73-year-old black man with type 2 diabetes, hypertension, and persistent NVAF who presents to the emergency room with complaints of occasional palpitations, lightheadedness, and a productive cough. He takes metformin for diabetes, valsartan with chlorthalidone for hypertension, and a daily aspirin tablet for his AF. He is a nonsmoker and leads a generally sedentary lifestyle. On physical examination, AS has a BMI of 34 kg/m 2 (height 5′10′′; weight, 237 lbs); blood pressure, 154/90 mm Hg; pulse, 100 bpm; and temperature, 102°F. His lab and test results were consistent with a diagnosis of acute pneumonia.
AS is admitted to the hospital and treated for pneumonia. Past history reveals that he was diagnosed 2 years ago with AF and treated at that time with warfarin, which was stopped 1 year ago due to a GI bleed. Subsequently, he was put on an aspirin regimen of 75 mg per day. During the current visit AS expresses concern to the hospitalist about his persistent AF, diabetes, and high blood pressure, all of which increase his risk of stroke. He says he wants to resume oral anticoagulant treatment, which he thinks may be better for him than aspirin.

Case discussion
This patient's age and history of hypertension and diabetes give him a CHADS 2 score of 2 (and CHA 2 DS 2 -VASc score of 3), which indicates an increased risk of stroke and is associated with a stroke rate of 4.0 % per year based on his CHADS 2 score, or 3.2 % per year based on his CHA 2 DS 2 -VASc score [10,33]. His age (≥65 years), past history of GI bleeding, and hypertension add up to a HAS-BLED score of 3, which is linked to an increased risk of bleeding, and data show it to be associated with a bleeding rate of 3.74 % per year [42]. The high risk of both stroke and bleeding call for immediate stroke prevention. Guideline recommendations for patients with a CHADS 2 score of 2   [50] rFVIIa [51] rFVIIa [56] aPCC (FEIBA®) [55] aPCC (FEIBA®) [53] aPCC (FEIBA®) [51] aPCC (FEIBA®) [56] PER977 (when available) [76] Andexanet alfa (when available) [52] Activated charcoal [71] PPSB-HT [56] PER977 (when available) [76] Andexanet alfa (when available) [49,52] PER977 (when available) [76] aPCC activated prothrombin complex concentrate; bid twice dailym, CrCl creatinine clearance, CYP3A4 cytochrome P450 3A4, ESRD end-stage renal disease, FEIBA® factor VIII inhibitor bypass activity, NA not available, NSAID nonsteroidal anti-inflammatory drug, PCC prothrombin complex concentrate, PD pharmacodynamics, P-gp P-glycoprotein, PPSB-HT prothrombin complex concentrate, rFVIIa recombinant activated factor VIIa, SNRI serotonin/norepinephrine reuptake inhibitor, SSRI selective serotonin reuptake inhibitor (or a CHA 2 DS 2 -VASc score ≥2) include oral anticoagulation with warfarin or a NOAC (Table 2) [7,8,10,11]. AS has a history of GI bleeding with warfarin, and has several characteristics associated with poor INR control including his age, non-white race, and diabetes [59][60][61]. A NOAC may provide a safer alternative to warfarin. However, in the RE-LY and ROCKET-AF trials, dabigatran and rivaroxaban, respectively, were associated with similar rates of major bleeding and increased rates of GI bleeding versus warfarin [23,25] [23]. Multiple GI bleeding events with dabigatran have also been reported in recent postmarketing surveillance [62]. In the ROCKET-AF trial, the rates of major bleeding were 3.6 % per year and 3.4 % per year for rivaroxaban and warfarin, respectively (p = 0.58), and GI bleeding occurred in 3.2 % of patients with rivaroxaban versus 2.2 % of patients with warfarin (p < 0.001) [25]. In the ARISTOTLE trial comparing apixaban with warfarin, apixaban demonstrated a 31 % reduction in major bleeding and no increased risk of GI bleeding compared with warfarin [24]. Non-bleeding GI adverse events (AEs) were more common with dabigatran than with rivaroxaban in the RE-LY and ROCKET-AF trials, respectively, versus warfarin (16.9 % vs. 9.4 %; 5.33 % vs. 5.57 %, respectively) [25,63]. In the RE-LY trial, dyspepsia was the most common AE among patients receiving dabigatran, occurring significantly more often with either dose of dabigatran than with warfarin (p < 0.001 for both doses) [25]. An analysis of data from the RE-LY trial found that non-bleeding upper GI AEs were generally mild or moderate with dabigatran and warfarin [63]. However, patients receiving dabigatran compared with warfarin were more likely to discontinue the study drug due to non-bleeding upper GI AEs or dyspepsia-like symptoms (4.0 % vs. 1.7 %; p < 0.001, respectively) [63]. Similar findings were not observed for apixaban or rivaroxaban in the ARISTOTLE and ROCKET-AF trials, respectively [24,25]. In the ROCKET-AF trial, the only difference in incidence of AEs between treatment groups was epistaxis, which occurred more frequently in patients receiving rivaroxaban than patients receiving warfarin (10.14 % vs. 8.55 %, p < 0.05) [25].

Discussion and conclusions
Conventional treatment options for reducing the risk of stroke in patients with NVAF, including VKAs and aspirin, have limitations that currently leave many patients undertreated or untreated, and thus, suboptimally protected from stroke. The NOACs provide an additional treatment pathway that may help address this trend by offering treatment options with equivalent or improved efficacy and safety compared with warfarin. Current treatment guidelines have begun to incorporate recommendations for use of the available NOACs, and as additional data emerge and guidelines are further updated, these recommendations will likely evolve. In the interim, clinicians may need to consider the recommendations of published guidelines while also weighing results of clinical trials on NOACs. Furthermore, real-world experience will provide clinicians with additional insight into their treatment decisions. The challenge, however, will be for clinicians to assess the benefits and risks of treatment within the framework of the individual characteristics that define each patient case scenario. Use of scoring systems to assess stroke and bleeding risk; consideration of patient characteristics such as concomitant medications, age, and renal failure; and an understanding of relevant pharmacologic characteristics of the NOACs will all assist in these treatment decisions (Table 5).
With the availability of the NOACs, success in reducing the risk of stroke among patients with NVAF no longer has to pivot on the suitability of warfarin for patients. Until treatment guidelines are updated to incorporate all available data on the NOACs, clinicians would be well advised to consider recent clinical trial data for these agents alongside current treatment guidelines for stroke prevention in AF. Such attentiveness may help render treatment decisions that accurately address the constellation of characteristics composing each individual case and result in safe, high-quality, and cost-effective patient care.