Introduction 

Stroke is the leading cause of adult disability and the third leading cause of death in Canada.[1] Each year, over 50 000 Canadians, including 10,000 British Columbians, suffer a stroke.[1,2] There are 300,000 stroke survivors living in Canada, of whom almost half experience some limitation to their daily activities. It is estimated that the disease costs Canadians a staggering $3 billion annually.[3] The population in Canada is aging as baby boomers reach retirement. Over the next 15 years, a 12% increase in the annual number of strokes is expected.[3] Therefore, a stroke epidemic is inevitable unless appropriate steps are taken to tackle treatable stroke risk factors. 

Atrial fibrillation (AF) is the most common chronic cardiac arrhythmia and is a significant independent risk factor for stroke. About 1% of the population older than 65 has AF, and the prevalence increases to 10% by age 85.[4] Overall, AF increases the risk of stroke fivefold compared to those without AF.[4] When associated with AF, other stroke risk factors (such as hypertension) increase the risk further.[5] More than 90% of thromboembolic events in AF patients are strokes.[6-10]   

In population studies, AF accounts for approximately 15% of all strokes, making it one of the most common causes. Therefore, instituting effective stroke prevention strategies in patients with AF will significantly reduce the burden that this disease places on society. In the last 15 years, a wealth of information has been obtained on stroke prevention in atrial fibrillation. Despite our knowledge of very effective preventive therapy, more than 60% of patients with AF are inadequately managed in both community and academic settings.[11,12] The purpose of this review is to highlight risk factors for stroke in patients with AF, provide an overview of the rationale for the optimal prevention of stroke based upon lessons we have learned from recent randomized trials, discuss why we continue to significantly undermanage these patients, and suggest steps that can be taken to improve the situation. 

Risk factors for stroke in atrial fibrillation 

In 1994, an analysis was performed involving more than 4000 patients from pooled data obtained from five randomized primary stroke-prevention trials in AF patients.[5] The presence of any of these clinical features conferred additional stroke risk to an AF patient. Multivariant analysis identified four independent clinical risk factors. These were:  

•  Any prior stroke or transient ischemic attack (TIA) (relative risk [RR] 2.5) 

•  History of hypertension—controlled or uncontrolled (RR 1.6) 

•  Increasing age (RR 1.4 for each decade) 

•  Diabetes (RR 1.7) 

Angina, prior myocardial infarction, and congestive heart failure were identified as risk factors in the univariant analysis but not in the multivariant analysis. Paroxysmal and chronic AF conferred the same degree of risk. In a separate secondary prevention study, recent (<3 months) TIA or stroke resulted in the greatest stroke risk (12% per year, placebo arm).[13] 

Recent analysis of risk factors from three of the American AF trials identified prior TIA/stroke, increasing age, and hypertension as risks, confirming the original pooled data results.[14] Diabetes increased the risk, but to a lesser degree than the other clinical characteristics. Pooled analysis of transthoracic echocardiography demonstrated independent risk in patients with moderate to severe left ventricular dysfunction. (RR 2.5)[15] Enlargement of the left atrium was not an additional risk for stroke, although it does reduce the likelihood of maintaining sinus rhythm following cardioversion. Recent provocative transesophageal echocardiogram studies from patients in the high-risk cohort from the third Stroke Prevention in Atrial Fibrillation (SPAF III) study suggested that dense spontaneous echo contrast (fibrin clumps seen on transesophageal echocardiogram, referred to as smoke), left atrial clot, reduced left atrial ejection velocity (<20 cm/s), and complex aortic arch atheroma were additional risks.[16] Absence of all four transesophageal echocardiogram findings resulted in a low risk of stroke despite the presence of high-risk clinical features. Finally, based on cohort studies, AF patients with rheumatic valvular stenosis or prosthetic valves also have increased risk of stroke. The important additional stroke risk factors in AF patients are summarized in Table 1.

Overall, the risk of stroke in patients with paroxysmal AF may be lower than in those with chronic AF. This is likely explained by the reduced frequency of associated high-risk clinical features in paroxysmal AF patients. However, if a patient with paroxysmal AF has high-risk features, the risk of stroke is similar to a patient in chronic AF with similar high-risk features.[5,17] The risk of stroke in atrial flutter is perhaps somewhat less than in patients with AF (~3% per year).[18] Since many patients with atrial flutter develop AF, patients in flutter should be risk-stratified in a similar manner. 

In the absence of antithrombotic treatment, patients can be accurately stratified to estimate their yearly risks of stroke or systemic embolism (Table 2). The annual risk in a “typical” AF patient is 5%. For patients <65 years old without concomitant clinical parameters (that is, lone atrial fibrillation), the absolute yearly embolic rate is <1%. In patients >75 with one high-risk clinical feature, the annual rate is about 8%.[5] With recent TIA or stroke, the annual stroke rate is 12%![13] 

Optimal stroke prevention in atrial fibrillation 

Many randomized trials have been performed in patients with AF. Most have been primary prevention studies (i.e., excluding patients with recent TIA/stroke). Intention-to-treat analysis of pooled data from the first five studies demonstrated that adjusted-dose warfarin reduced the risk of stroke and systemic embolism by 68% when compared to control (placebo in four studies, ASA or placebo in one study) and approximately 45% when compared to ASA.[5] However, a majority of strokes occurred in the warfarin-assigned patients at times when the warfarin was stopped (i.e., prior to surgery) or when the patient’s degree of anticoagulation was below the target level. On-treatment analysis suggested an 83% reduction in stroke risk.[19,20] Warfarin generally reduced the annual stroke risk below 2%, even compared to similar untreated patients with yearly risks approaching 8%.[5] In one secondary prevention study that randomized AF patients within 3 months of TIA or stroke, those assigned to ASA had a stroke rate of 10% per year versus 4% per year in the adjusted-dose warfarin group.[13] 

Initial studies found ASA reduced the relative risk of stroke by 21% over placebo.[5] A recent meta-analysis, which included more recent trials, found a 22% relative risk reduction.[21] Other currently available antiplatelet agents have not been adequately assessed in AF patients. 

Five trials have compared adjusted-dose warfarin to ASA and two trials have compared adjusted-dose warfarin to a combination of ASA and fixed-dose warfarin.[6,13,22-25] The doses of warfarin used in the fixed-dose studies did not significantly raise the international normalized ratio (INR).[23,24] All studies demonstrated results in favor of adjusted-dose warfarin and most were statistically significant. In one study (SPAF II), there was an increase in the number of intracranial hemorrhages that partially offset the benefits of warfarin, but in this study the INR was higher than current recommendations, and overall, the cohorts studied carried only a moderate risk of ischemic stroke.[22] The benefits of warfarin were particularly clear when a group of high-risk patients were evaluated in SPAF III.[23] This trial compared adjusted-dose warfarin (INR 2.0 to 3.0) to combination ASA 325 mg/d plus fixed-dose warfarin (mean INR 1.2). The trial was stopped early due to an excess number of ischemic events in the combination group (7.9% per year) versus the adjusted-dose warfarin group (1.9% per year). Major hemorrhages were slightly higher (2.4% per year versus 2.1% per year) in the combination cohort. The most recent meta-analysis, which summarized all randomized trials to date, found a significant 46% relative risk reduction in ischemic stroke from adjusted-dose warfarin over ASA and a 36% relative risk reduction of all stroke (ischemic and hemorrhagic).[21] Since the meta-analysis included patients entered in all randomized AF trials, it underestimated the benefit of warfarin over ASA in high-risk patients. Adjusted-dose warfarin appeared to be superior to fixed low-dose warfarin in one study.[26] 

Many AF patients have co-existent symptomatic coronary artery disease. Several studies in non-AF patients demonstrate the efficacy of adjusted-dose warfarin over placebo in preventing death and recurrent coronary events following myocardial infarction.[27] Moreover, warfarin and ASA have demonstrated similar efficacy in the prevention of coronary events in cardiac patients. However, the addition of low-dose ASA to warfarin increases the risk of bleeding.[28] Therefore, in AF patients with co-existent symptomatic coronary artery disease in whom warfarin is indicated based on clinical criteria, adjusted-dose warfarin alone is recommended for anti-thrombotic therapy. 

In assessing trials to date, ASA is only modestly effective over placebo in reducing the risk of stroke in AF patients. Compared to placebo, ASA reduces stroke by about one-fifth. Adjusted-dose warfarin is superior to placebo and reduces the risk by two-thirds. Warfarin reduces the risk of ischemic stroke over ASA by almost half. However, on-treatment analysis is even more favorable toward warfarin. The benefits of warfarin are most apparent in high-risk AF patients. 

The rationale of maintaining AF patients in sinus rhythm to prevent stroke is appealing but remains unproven. No completed trial has assessed this strategy. Further, only 50% to 60% of patients who are cardioverted and then placed on antiarrhythmics remain in sinus rhythm at 1 year.[29] Many antiarrhythmics have significant toxicity. 

Which AF patients should be placed on warfarin? The risk of stroke varies substantially in individual AF patients. In those without any of the additional clinical risks (Table 1), the annual rate of stroke is less than 1% per year compared to others whose risk is greater than 10%.[5,13] Since the risk of bleeding is greater with anticoagulation than ASA, patients selected for warfarin should be deemed at high risk for ischemic stroke. Based on the combination of data from the Atrial Fibrillation Investigator’s (AFI) and SPAF data, the American College of Chest Physicians recommends adjusted-dose warfarin for patients with one high-risk or two moderate-risk features. If one moderate risk is present, either ASA or warfarin can be used. In those without risks or in patients with contraindications to warfarin, ASA 325 mg/d should be used (Table 2).[30] 

Optimal intensity of anticoagulation 

The intensity of anticoagulation in the various adjusted-dose studies ranged from INR 1.4 to 4.5.[5,19] Most used prothrombin ratios but the INR has been extrapolated. In two case-control studies, the risk of stroke sharply increased when the INR was <2.0 (i.e., 3 times stroke risk with INR 1.5).[31,32] Ischemic events are substantially reduced at an INR of 2.0, and INRs above this level do not confer additional benefit.[31] Since it appears the ischemic stroke risk increases markedly below an INR of 2.0 and the risk of hemorrhage (in particular intracranial hemorrhage) increases substantially above 4.0, the current recommendation is to target the INR to 2.5 (range 2.0 to 3.0).[30] By targeting 2.5, it is less likely that values will fall outside the recommended 2.0 to 3.0 range. 

Two of the original five primary prevention AF trials used prothrombin time targets equivalent to an INR of 1.5 to 2.7 and 1.4 to 2.8.[8,9] The benefit of warfarin over placebo was similar to other primary prevention studies. Recently a Japanese trial compared an INR target range of 1.5 to 2.1 versus 2.2 to 3.5.[33] The ischemic event rate was too low (1.1% to 1.7 % per year) to draw conclusions of efficacy. However the trial was stopped early due to an excessive number of major bleeds (one intracranial, five peripheral, all non-life threatening) in the high-intensity group. Further prospective studies are required to determine the optimal intensity of anticoagulation and whether specific racial groups may respond differently. 

Risk of hemorrhage with warfarin 

In the AFI analysis, the annual major bleed rates were similar in the placebo (1%), ASA (1%), and warfarin (1.3%) groups. The risk of intracranial hemorrhage attributable to warfarin was 0.3% per year. However, these were randomized trials that closely monitored patients and excluded patients at high risk of bleeding. In SPAF II, the rate of intracranial hemorrhage was 1.3% per year, but in that study, the INR intensity was above the current recommendation (INR 2.0 to 4.5). The risk of bleeding increases in the elderly (who are at highest ischemic stroke risk), as the INR climbs above 4.0, and in patients with uncontrolled hypertension.[34,35] Since the latter two factors can be controlled with appropriate patient monitoring, anticoagulation can be safely achieved. Safety of anticoagulation has been demonstrated in clinical practice.[36,37] Anticoagulation clinics improve INR control and reduce bleeding rates over individual practitioners.[38] Efforts should be further directed toward establishing anticoagulation clinics in Canada. Recent studies have also demonstrated that selected patients are able to correctly adjust their own INR.[39] Suggestions to improve individual patient selection and monitoring are presented in Table 3.

Underutilization of anticoagulation in clinical practice 

Despite the results of many randomized trials that clarify optimal patient selection for anticoagulation, there remains a significant underuse of warfarin in clinical practice. In one American hospital-based study, 56% of AF patients with one or more risk factors for stroke and no contraindication to anticoagulation were not on warfarin.[12] Similar results were seen in a recent Canadian study.[40] 

Inadequate practitioner education, physician concern regarding bleeding risk, and tediousness of INR monitoring are potential reasons for poor utilization of warfarin in appropriate patients. A frequent reason used to exclude patients from warfarin is concern regarding falls. However, the risk of bleeding due to falls is minimal. It is estimated that 295 falls per year are required to negate the benefit of warfarin in elderly patients with AF.[41] Further, falls can be predicted and prevented.[42] 

Use of warfarin in elderly AF patients poses additional challenges. This group is at increased risk for ischemic and hemorrhagic events.[5,22] However, trials clearly demonstrate the substantial benefit and safety of warfarin in patients >75 years of age.[5] Patient compliance is sometimes more problematic, but difficulties can be overcome by use of medication bubble packs, spouse/family member education and participation in care, increased INR monitoring/use of mobile labs, scrutiny of concomitant medications that interact with warfarin, and inclusion of patients in anticoagulation clinics where available. 

When properly used, warfarin prevents 20 strokes for every major bleeding complication it causes.[43] Comparing warfarin to placebo, in an AF patient over the age of 75 with one additional clinical risk factor for stroke (Table 1), the number needed to treat to prevent one embolic event in 1 year is 15.[5] If a patient has suffered a TIA or stroke in the preceding 3 months and is placed on warfarin, to prevent one stroke in a year, compared to placebo, the number needed to treat drops to 13 (number needed to treat for warfarin over ASA is 17).[13] When warfarin is compared to ASA, the number needed to harm is 333.[5] 

Summary 

AF is the most common single cause of stroke. Adjusted-dose warfarin (target INR 2.5, range 2.0 to 3.0) is superior to ASA for stroke prevention and should be prescribed for patients with additional clinical risk factors for stroke. Further efforts are required to increase warfarin use in appropriate high-risk patients. 

Competing interests 

None declared. 

  Table 1. Clinical stroke risks.[30]

Table 2. Optimal antithrombotic therapy for AF. 

Table 3. Improving the safety of anticoagulation. 

Contraindications

Andrew R. Woolfenden, MD, FRCPC and Gregory W. Albers, MD 

Dr Woolfenden is the associate director of the British Columbia Centre for Stroke and Cerebrovascular Diseases. Dr Albers is the director of the Stanford Stroke Center.