Author + information
- Received June 24, 2003
- Revision received October 31, 2003
- Accepted November 13, 2003
- Published online March 17, 2004.
- ↵*Reprint requests and correspondence:
Dr. Stanley G. Rockson, Division of Cardiovascular Medicine, Falk Cardiovascular Research Center, Stanford University, 300 Pasteur Drive, Stanford, California 94305, USA.
Atrial fibrillation (AF) is an important risk factor for stroke. According to a pooled analysis of controlled clinical trials with warfarin, anticoagulation therapy reduces stroke risk by 62%. However, clinicians must decide whether the benefit of long-term anticoagulation therapy with available agents outweighs the risk of bleeding for individual patients. Guidelines issued by the American College of Chest Physicians and by the joint American College of Cardiology, American Heart Association, and the European Society of Cardiology task force recommend antithrombotic therapy to protect AF patients from stroke based on risk-stratification algorithms. Risk factors for stroke AF patients include age ≥75 years; hypertension; thyrotoxicosis; diabetes; cardiovascular disease; congestive heart failure; and history of stroke, transient ischemic attack, or thromboembolism. Patients at high risk for stroke experience greater absolute benefit from anticoagulation therapy than patients at low risk. The guidelines are consistent in recommendations for high-risk patients (warfarin therapy, international normalized ratio 2.0 to 3.0) and low-risk patients (aspirin 325 mg), but differ for intermediate-risk patients with diabetes or heart disease. The guidelines continue to evolve, and future guidelines are likely to incorporate new clinical data, including the CHADS2algorithm for determining risk and the results of the Atrial Fibrillation Follow-up Investigation of Rhythm Management trial, the Rate Control versus Electrical Cardioversion for Persistent Atrial Fibrillation study, and the Stroke Prevention Using an Oral Thrombin Inhibitor in Atrial Fibrillation II to V trials.
The loss of coordinated atrial contractions in atrial fibrillation (AF) creates a predisposition for thrombus formation within the atria and thereby contributes to the risk of embolic stroke in these patients (1). As many as 15% of all strokes in the U.S. may be attributable to AF (2–4). Nonvalvular AF, the most common form of this arrhythmia, increases the risk of stroke approximately fivefold. Valvular AF increases the risk up to 17-fold (5). Approximately 2.4 million people in the U.S. currently have AF. This number, expected to increase to more than 5.6 million by the year 2050 (4), poses a major public health problem.
One of the primary therapeutic end points in AF is the reduction of stroke risk through prevention of thromboembolism. Controlled clinical trials have shown that anticoagulation with warfarin reduces the risk of ischemic stroke by 68% (3)and both ischemic and hemorrhagic stroke by 62% in patients with AF (6). The antithrombotic efficacy of warfarin in AF has been observed in patients with a variety of risk factors (Fig. 1). However, the benefits of warfarin use in routine practice are offset by the possibility of increased bleeding, including intracerebral hemorrhage (ICH) and the need for frequent monitoring and dose adjustment to achieve the optimum anticoagulant effect (7,8). In response to these concerns, warfarin use has been limited to patients at moderate or high risk of stroke, for whom the benefits clearly outweigh the risks.
Guidelines from both the American College of Chest Physicians (ACCP) (9)and the joint American College of Cardiology/American Heart Association/European Society of Cardiology (ACC/AHA/ESC) task force (10)recommend anticoagulation therapy for the prevention of stroke in patients with AF based upon the presence of risk factors for stroke. Although the guidelines generally agree on therapy for high-risk patients, their assessments of risk:benefit ratios differ for patients at moderate risk of stroke.
This review will consider the issues surrounding identification of AF patients who merit oral anticoagulation therapy, on the basis of a variety of risk-stratification approaches and guidelines. Optimal management strategies for anticoagulation will also be discussed in light of recently published controlled clinical trials in AF patients.
The creation of risk classification schemata is intended to guide treatment decisions in AF, with attention to the evidence that defined risk factors increase the likelihood of clinical events. In AF, risk factors for stroke include age, gender, hypertension, ischemic or rheumatic heart disease, prosthetic heart valves, congestive heart failure, history of stroke or transient ischemic attack (TIA), prior thromboembolism, evidence of intracardiac thrombus (demonstrable by transesophageal echocardiography), and diabetes mellitus. Thyrotoxicosis, hormone replacement therapy, smoking status, and alcohol consumption are less well-defined risk factors (3,9,11).
The Atrial Fibrillation Investigators (AFI) pooled five large, randomized controlled trials of anticoagulation therapy in patients with AF and performed multivariate analysis to define predictors of stroke. This analysis demonstrated that a variety of risk factors increase stroke risk to more than 5% per year; anticoagulation therapy consistently reduced this risk by approximately three-fold (Fig. 1). Age was also shown to increase stroke risk by a factor of 1.4 per decade. A simple table was devised to express the relationship of age with the risk factors selected as having the most significance, namely, a history of hypertension, a history of diabetes, and a history of stroke or TIA (Table 1) (3).
In parallel, the Stroke Prevention in Atrial Fibrillation (SPAF) investigators assessed the risk for stroke in patients with AF given aspirin therapy to prevent stroke in three separate studies (Table 2). Although risk factors were assessed in each trial individually, a meta-analysis was also performed (12). Risk increased by 1.8-fold per decade of life (p < 0.001), 1.6-fold for female gender (p = 0.01), 2.0-fold for a history of hypertension (p < 0.001), and 2.3-fold for systolic blood pressure (SBP) >160 mm Hg (p < 0.001). The highest risk was found in patients who had already experienced a stroke (adjusted relative risk of 2.9-fold increase, p < 0.001) (12). As in the AFI study, patients with multiple risk factors consistently demonstrated an elevated yearly stroke rate (12).
The differences between the AFI criteria and the SPAF III criteria were assessed in 13,559 patients from the Anticoagulation and Risk Factors in Atrial Fibrillation study. According to the AFI criteria, low-risk patients include those <65 years of age who have no history of stroke, hypertension, diabetes, or left ventricular dysfunction. According to the SPAF III criteria, patients at low risk include men, women <75 years with no history of prior stroke or recent heart failure, patients with SBP ≤160 mm Hg, and patients with fractional shortening ≥25% on echocardiography (13). When patients were classified in parallel applying both sets of criteria, 10.5% were classified as low risk using the AFI criteria, and 29.3% were defined to be at low risk based on the SPAF III criteria (13). More than 90% of the differences between the two groups reflected inclusion of older patients with no other risk factors (men and women 65 to 75 years old, and men >75 years old) in the higher risk AFI group. If one assumes that patients defined to be at low risk would not receive anticoagulation therapy, these classification distinctions might significantly affect therapeutic decisions for a substantial cohort of patients.
A recently proposed risk classification scheme, CHADS2, incorporates features from both AFI and SPAF (Table 3) (3,11,14). In CHADS2, risk factors are assigned a value of 1 or 2, thus achieving a numerical stratification of stroke risk within each patient. The CHADS2score ignores risk factors other than congestive heart failure, hypertension, age ≥75 years, diabetes, and prior stroke. However, when scrutinized, the CHADS2scheme had a greater predictive value for stroke (c-statistic 0.82, 95% confidence interval [CI], 0.80 to 0.84) than either the AFI (c = 0.68, 95% CI, 0.65 to 0.71) or SPAF scheme (c = 0.74, 95% CI, 0.71 to 0.76) (14). Although the c-statistic is >0.5, which would indicate a random relationship, it is not 1.0, which would indicate perfect agreement. Thus, some patients at risk are not identified.
The stroke risk associated with an increasing CHADS2score was demonstrated using the national registry of AF participants (Fig. 2). The incremental difference in risk between sequential CHADS2scores illustrates the difficulty in establishing cutoff values for antithrombotic intervention. A CHADS2score of 0 defines the AF population that would be offered aspirin therapy, and those at highest risk would be given warfarin therapy (international normalized ratio [INR], 2.0 to 3.0), but therapeutic guidelines for intermediate CHADS2values have not been established. For example, individual physicians may disagree regarding whether a patient with a predicted yearly stroke rate of 2.8% (CHADS2score = 1) should receive aspirin or warfarin.
An age-independent clinical prediction rule has been developed recently to determine the patients with low stroke risk who can be treated with aspirin alone (15). According to this algorithm, aspirin may represent sufficient therapy if the patient does not have a history of previous stroke or TIA, treated hypertension or SBP ≥140 mm Hg, angina or previous myocardial infarction, or diabetes. These criteria were applied to 24% of the 2,501 patients included in six clinical trials, including 16% of patients >75 years (15). Thus considered, low-risk patients experienced 1.1 events per 100 patient-years when not given oral anticoagulation.
Other approaches for determining individual risk are in development. In a follow-up study to the SPAF III trial, Conway et al. (16)measured levels of von Willebrand factor (vWf), a marker of endothelial damage and a possible indicator of hypercoagulability, in 1,321 trial participants. The vWf increased incrementally from low-, to moderate-, to high-risk groups (p < 0.001) (16). It is possible that vWf or other indexes of hypercoagulability may eventually be used to help predict risk for thromboembolic stroke.
The ACCP and the ACC/AHA/ESC have issued similar guidelines for the use of oral anticoagulants to treat patients with AF. These major therapeutic guidelines have been established based upon thorough review of risk-stratification studies (9,10). Both of the guidelines employ an age-based stratification, comparable with the AFI schema. Furthermore, both guidelines recommend either aspirin (325 mg) or warfarin, with the treatment choice dependent upon the presence of additional risk factors or the presence of contraindications to the use of warfarin. It should be noted that a substantial portion of patients may have contraindications to warfarin that prevent use of the recommended anticoagulation therapy. In a recent review of frail, elderly Medicare beneficiaries, 46% had a contraindication to warfarin, such as previous hemorrhage, blood dyscrasia, renal disease, or hepatic disease (17). Because such patients are excluded from clinical trials, no data are available on the relative risk of bleeding versus the risk of stroke in these patients, and their management is challenging.
Description of the guidelines
The ACCP guidelines represent a risk-stratification scheme based on age and other conditions that predispose an individual to thromboembolic stroke (9). The guidelines define risk as low (an essential absence of risk factors), moderate (age 65 to 75 years, diabetes, or coronary artery disease), or high (age >75 years, multiple intermediate risk factors, history of hypertension, poor left ventricular function, valvular disease, or prior thromboembolism) (Table 4). While therapy is defined for both low-risk (325 mg aspirin) and high-risk patients (anticoagulation with warfarin to an INR of 2.0 to 3.0), the choice of antithrombotic therapy for patients at intermediate risk is left to the discretion of the physician.
The ACC/AHA/ESC guidelines present recommendations based on an integration of the ACCP consensus risk-stratification scheme with an evidence-based analysis of risk factors for stroke (Table 5) (10). The lowest risk patients may not require antithrombotic therapy. However, low-risk patients should receive aspirin (325 mg). Oral anticoagulation is recommended for high-risk patients, with a target INR in the range of 2.5 to 3.5; patients ≥75 years should receive warfarin therapy with more modest therapeutic targets (INR, ∼2.0; range, 1.6 to 2.5).
Comparison of the guidelines
The two sets of guidelines are quite similar and, in general, would result in comparable therapeutic strategies for most patients. However, the routine use of long-term anticoagulation therapy in patients with intermediate risk differs between the guidelines. In the ACC/AHA/ESC approach, anticoagulation therapy with warfarin is recommended for patients ≥60 years of age who have diabetes or coronary artery disease (10). However, either warfarin therapy or aspirin therapy is acceptable for patients <65 years with one of these risk factors, according to the ACCP guidelines (9).
The INR targeting ranges for warfarin also differ in some cases. The ACCP guidelines recommend an INR of 2.0 to 3.0 for all patients on anticoagulation therapy, which is the most effective range (9). The ACC/AHA/ESC guidelines recommend a lower therapeutic range for patients ≥75 years (those who are at increased risk for thromboembolic stroke but also for hemorrhagic stroke) (10). When compared with standard INR targets, the reduced INR range appears to be about 80% as effective at preventing thromboembolism, but the bleeding risk may be reduced (10). In contrast, according to the ACC/AHA/ESC, a higher INR range (2.5 to 3.5) may be appropriate for patients at extremely high risk of stroke because of the presence of valvular disease or a history of prior stroke. In these high-risk patients, it is important to avoid subtherapeutic anticoagulation; however, there is no data to indicate that the efficacy of anticoagulation is improved in AF patients when the INR exceeds 2.5 (10).
Implementation of the guidelines
Compliance with either of the guidelines is likely to improve patient outcomes. Patients who are initially judged to be at low risk should be evaluated frequently for changes in their risk profiles, particularly because late-onset hypertension, diabetes, cardiovascular disease, or advancing age will increase the dynamically evolving risk for stroke. With time, a given patient's stroke risk is likely to increase.
When applying the guidelines, it may be helpful to test the tolerability of warfarin therapy in intermediate-risk patients. Some patients maintain a relatively stable INR with little difficulty, have a low incidence of bleeding problems, and tolerate the need for frequent testing with equanimity. In this situation, the extra protection offered by warfarin therapy is offset by minimal undesirable attributes. However, patients with highly variable INRs, bleeding, or poor compliance may not be good candidates for anticoagulation therapy. The subjective input of the patient must be considered in arriving at this therapeutic decision.
When using warfarin, individualization of therapy is important. Elderly patients are particularly sensitive to warfarin's anticoagulant effect and may require lower maintenance doses than younger patients. One study showed that, among patients with an INR of 2.0 to 3.0, only 25% of patients older than 80 years required a weekly maintenance dose of warfarin exceeding 30 mg, compared with nearly 70% of those younger than 65 years (18).
Anticoagulation intensity, or the magnitude of the INR, is the primary risk factor for bleeding. The optimal INRs achieved in clinical trials substantially reduced the risk for any form of bleeding, including ICH. However, bleeding risk increases significantly when the INR rises above 3.0. Patient factors that predict the risk of ICH include advanced age, history of stroke, and hypertension (19). Because age and history of stroke are nonmodifiable risk factors, the best strategy to minimize bleeding risk is to carefully control the patient's INR and blood pressure. Particular care should be taken when adding new medications, because warfarin has the potential to interact with numerous other drugs frequently prescribed in aging patients, including antacids, antiarrhythmics, antidepressants, aspirin, nonsteroidal anti-inflammatory agents, and HMG-CoA reductase inhibitors (statins) (20,21). Patients should also be educated about the importance of adhering to recommended doses and the importance of consistent diet and activity to maintain a stable therapeutic INR.
Because the guidelines are evidence-based, they evolve as new clinical data become available. Data that have been published since the guidelines were written may influence future versions of the guidelines and, hence, influence clinical practice.
The AFFIRM trial and the RACE study
The Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial and the Rate Control versus Electrical Cardioversion for Persistent Atrial Fibrillation (RACE) study were large multi-year trials that examined the benefits of rate control versus rhythm control for patients with AF (22–24). Risk for thromboembolic stroke was also examined.
Patients in the AFFIRM trial were at moderate or high risk of stroke. The study enrolled 4,060 patients of at least 65 years of age who had AF and at least one risk factor without contraindication for anticoagulant therapy. A lower number of deaths occurred in patients treated with the rate-control strategy (310 of 2,027) versus rhythm control (356 of 2,033; p = 0.08) (24). The results of the RACE trial, in a population of 522 patients, were similar to those of AFFIRM. The composite primary end point, including death from cardiovascular causes, heart failure, thromboembolic complications, bleeding, the need for a pacemaker, or severe adverse effects of antiarrhythmic drugs, did not differ significantly between rate-control and rhythm-control strategies (44 of 256, 17.2% vs. 60 of 66, 22.6%; p = NS) (22).
Perhaps the most significant insight with respect to stroke prevention was the finding that patients with ischemic stroke either had a subtherapeutic INR or discontinued anticoagulation. All together, 23 of 35 (66%) of patients who experienced thromboembolic events in the RACE trial had an INR <2.0, and most (73%) had AF at the time of the event. In AFFIRM, 113 of 157 (72%) of patients who experienced ischemic stroke had either discontinued anticoagulation therapy or had an INR <2.0 (24). Most importantly, 75% of the patients in the rhythm-control group who experienced a thromboembolic event were believed to be in sinus rhythm. These data suggest that chronic anticoagulation to an INR of 2.0 to 3.0 should be continued without regard to whether rate or rhythm control is achieved (23).
The established guidelines already recommend well-controlled anticoagulation therapy for patients with risk factors and atrial flutter, paroxysmal AF, or persistent AF. In view of the results of AFFIRM and RACE, it is likely that ongoing anticoagulation therapy will be recommended for at-risk patients believed to be in sinus rhythm after rhythm-control therapy.
The CHADS2algorithm was published after release of the guidelines (14). Prospective testing of this algorithm is ongoing. More data on the predictive value of this algorithm will indicate whether this single index is sufficient for risk stratification, or whether inclusion of other risk factors (as in the existing guidelines) continues to be necessary.
New anticoagulants, which may have more predictable dosing or require less frequent administration than warfarin, are in development and being tested in large-scale trials. For example, the Stroke Prevention using an Oral Thrombin Inhibitor in atrial Fibrillation (SPORTIF) trial was recently published (25). Once new drugs are approved, their study results will be assessed for possible inclusion in the guidelines.
Patients with AF are at significant risk for thromboembolic strokes. Treatment with anticoagulants substantially reduces this risk, but the adverse effects common with currently available anticoagulation therapy may outweigh the benefits for specific populations, particularly in those at low risk for stroke. Guidelines have been established by the ACCP and the ACC/AHA/ESC to help identify which patient populations are most likely to benefit from anticoagulation. Antithrombotic therapy for patients between 60 and 65 years with diabetes or coronary artery disease entails the practitioner's choice of either aspirin 325 mg or adjusted-dose warfarin (ACCP) or, alternatively, warfarin therapy (ACC/AHA/ESC). In contrast, the therapeutic anticoagulation goals for the intensity of anticoagulation are clearly defined as an INR of 2.0 to 3.0 with a target of 2.5 in the ACCP guidelines, but are slightly increased (for highest risk patients) or decreased (for patients ≥75 years) according to the ACC/AHA/ESC guidelines. The guidelines continue to evolve with the evidence, and may be influenced by emerging studies, such as AFFIRM, RACE, and SPORTIF II to V. The emergence of a new oral anticoagulant with decreased bleeding risk would improve the risk:benefit ratio for anticoagulation therapy, reducing the need for risk stratification.
☆ Drs. Rockson and Albers have served as consultants for AstraZeneca and have served as investigators in AstraZeneca trials. Dr. Albers has been a member of the Executive Committee for the SPORTIF trials.
- American College of Cardiology/American Heart Association/European Society of Cardiology
- American College of Chest Physicians
- atrial fibrillation
- Atrial Fibrillation Follow-up Investigation of Rhythm Management trial
- Atrial Fibrillation Investigators
- confidence interval
- intracerebral hemorrhage
- international normalized ratio
- Rate Control versus Electrical Cardioversion for Persistent Atrial Fibrillation study
- systolic blood pressure
- Stroke Prevention in Atrial Fibrillation
- Stroke Prevention Using an Oral Thrombin Inhibitor in Atrial Fibrillation trial
- transient ischemic attack
- von Willebrand factor
- Received June 24, 2003.
- Revision received October 31, 2003.
- Accepted November 13, 2003.
- American College of Cardiology Foundation
- Fuster V,
- Ryden L.E,
- Asinger R.W,
- et al.
- Hart R.G,
- Pearce L.A,
- McBride R,
- Rothbart R.M,
- Asinger R.W
- Go A.S,
- Hylek E.M,
- Phillips K.A,
- et al.
- Conway D.S,
- Pearce L.A,
- Chin B.S,
- Hart R.G,
- Lip G.Y
- Gage B.F,
- Boechler M,
- Doggette A.L,
- et al.
- Hart R.G,
- Boop B.S,
- Anderson D.C
- Warfarin. RxList LLC. Available at: www.rxlist.com. Accessed February 13, 2003.
- ↵Halperin J, Nevinson M. Stroke prevention using oral direct thrombin inhibitor, ximelagatran, compared with warfarin in patients with nonvascular atrial fibrillation: the SPORTIF III trial. Lancet 2004. In Press.