In 1985, patients with nonvalvular atrial fibrillation (AF)

Comments, Opinions, and Reviews Atrial Fibrillation and Stroke Concepts and Controversies Robert G. Hart, MD; Jonathan L. Halperin, MD In 1985, patients with nonvalvular atrial fibrillation (AF) first entered randomized clinical trials that tested antithrombotic therapies for stroke prevention. 1 3 Since then, 12 000 AF patients have been included in 25 trials that involved 40 randomized treatment comparisons (Table 1). During this interval, we reviewed the pathogenesis and prevention of stroke in AF patients in 2 previous editorials in Stroke. 4,5 Here, we offer commentary on selected concepts and controversies. The prevalence of AF increases with age, affecting 5% of people at age 70 years. Although AF-associated stroke can occur at any age, it is predominantly a problem of the elderly. The median age of AF patients with stroke in populationbased studies is 75 years; more than half are women. In people over age 75, AF is the most important single cause of ischemic stroke. This epidemiology is relevant when considering stroke prevention, because the risks of and ability to sustain preventive therapies are special problems for the very elderly. Left Atrial Appendage Thrombi The left atrial appendage is a unique substrate for stroke. It is an elongated cul-de-sac lined with endothelium, a remnant of the embryonic atrium, trabeculated by pectinate muscles (Figure 1). The contractility of the appendage is reduced in AF, but the degree varies widely and is an important determinant of stasis and thrombus formation. In AF patients, atrial thrombi almost always originate in the appendage, rather than in the smooth-walled atrium proper, and are not reliably detected with transthoracic echocardiography. Transesophageal echocardiography is much more sensitive for the detection of appendage thrombi in AF patients, but the complex structure is usually multilobed, projecting in unpredictable planes. This, coupled with the minute size of clinically important thrombi (2 to 3 mm), makes the exclusion of small thrombi problematic. 6 Transesophageal echocardiography reveals left atrial appendage thrombi in 10% of patients with nonvalvular AF and in 20% to 40% of AF patients with recent thromboembolism. Stasis is fundamental to the formation of atrial appendage thrombi in AF. 7 Endothelial lesions in the appendage have not been found, and systemic prothrombotic diatheses that contribute to thrombus formation have been suggested but not convincingly identified. 8 Hormone replacement therapy was an independent predictor of stroke risk in AF patients in 1 study, supporting a potential role for a prothrombotic state. 9 Appendage thrombi resemble those formed in the venous system; it is unclear whether they usually begin as freefloating clots in the appendage cavities and then attach to the walls or originate on the endothelial surface. It seems likely that some emboli from the appendage never adhere, and the lack of visualization of appendage thrombi on transesophageal echocardiography after stroke does not exclude the appendage as the embolic source. Thrombi in AF patients have been strongly and consistently linked to reduced blood flow velocities in the left atrial appendage, confirming the role of stasis in their genesis. 7 Appendage flow velocities range over a wide continuum in AF and are determined by the strength of the uncoordinated appendage contractions, passive emptying during left ventricular diastole (influenced by ventricular relaxation/diastolic properties), and atrial pressure. Flow velocities are only 1 component and hence an incomplete gauge of stasis, to which duration and volume of flow also contribute; the presence of spontaneous echodensities in the appendage may be a better overall indicator of stasis. Clinical risk factors for cardioembolic stroke in AF patients can generally be related to increased stasis in the left atrial appendage. Advancing age is associated with decreased appendage contractility and reduced flow velocity. Hypertension, the most prevalent stroke risk factor in AF patients, leads to atrial enlargement and appendage stasis, 10 probably mediated by ventricular diastolic dysfunction. Many questions remain about the formation and embolism of left atrial appendage thrombi and, consequently, about the pathogenesis of AF-associated stroke. Aortic arch plaque, widened pulse pressure, and elevated systolic blood pressure are associated with reduced appendage flow velocity 7 ; this surprising association of aortic disease and its manifestations with appendage stasis has not been adequately explained and is a crucial missing link in the understanding of stroke mechanisms in AF. Elderly women with AF appear to have Received September 26, 2000; final revision received November 17, 2000; accepted December 22, 2000. From the Department of Medicine (Neurology) (R.G.H.), University of Texas Health Sciences Center (San Antonio); and The Zena and Michael A. Wiener Cardiovascular Institute (J.L.H.), Mt Sinai Medical Center, New York, NY. Correspondence to Robert G. Hart, MD, Department of Medicine (Neurology), University of Texas Health Science Center, MSC 7883, 7703 Floyd Curl Dr, San Antonio, TX 78229-3900. E-mail hartr@uthscsa.edu Stroke. 2001;32:803-808.) 2001 American Heart Association, Inc. Stroke is available at http://www.strokeaha.org 803

804 Stroke March 2001 TABLE 1. Randomized Trials of Antithrombotic Therapy in Atrial Fibrillation Year Published n Interventions Large published trials Copenhagen AF, Aspirin, Anticoagulation I (AFASAK I) 1989 1007 AC, ASA, PLA Copenhagen AF, Aspirin, Anticoagulation II (AFASAK II) 1998 677 AC (2 levels), ASA, AC ASA Stroke Prevention in AF I (SPAF I) 1991 1330 AC, ASA, PLA Stroke Prevention in AF II (SPAF II) 1994 1100 AC, ASA Stroke Prevention in AF III (SPAF III) 1996 1044 AC, AC ASA Boston Area Anticoagulation Trial for AF (BAATAF) 1990 420 AC, no Rx Canadian AF Anticoagulation (CAFA) 1991 378 AC, PLA Stroke Prevention In Nonrheumatic AF (SPINAF) 1992 571 AC, PLA European AF Trial (EAFT) 1993 1007 AC, ASA, PLA Studio Italiano Fibrillazione Atriale (SIFA) 1997 916 AC, indobufen Minidose Warfarin in Nonrheumatic AF 1998 303 AC (2 levels) Prevention of Arterial Thromboembolism in AF (PATAF) 1999 729 AC (2 levels), ASA Small or pilot trials Harenberg et al. 1993 75 LMW-heparin, no Rx Low-dose Aspirin, Stroke, AF (LASAF) 1999 285 ASA (2 doses), PLA Japanese Nonvalvular AF Embolism Secondary Prevention 2000 115 AC (2 levels) Fluindione Fibrillation Auriculaire Aspirin Contraste Spontane (FFAACS) 2000 157 AC, AC ASA Subgroups with AF in other trials European Stroke Prevention Study II (ESPS II) 1997 429 ASA dipyridamole, PLA United Kingdom TIA (UK-TIA) 1999 49 ASA (2 doses), PLA Ongoing or unpublished trials Swedish AF Trial (SAFT) 500 AC ASA, PLA Birmingham AF Treatment in the Aged (BAFTA) (1240) AC, ASA Athens 45 AC (2 levels), PLA Stroke Prevention Using Oral Thrombin Inhibition in AF (SPORTIF) II 257 AC, thrombin inhibitor Stroke Prevention Using Oral Thrombin Inhibition in AF (SPORTIF) III/V (6000) AC, thrombin inhibitor Japan AF Stroke Prevention Trial (JAST) (1000) ASA, no Rx Stroke Prevention in Atrial Fibrillation among Chinese (200) AC, clopidogrel AF indicates atrial fibrillation; n, number of participants; AC, oral vitamin K inhibitor anticoagulant; ASA, aspirin; LMW, low-molecular-weight; PLA, placebo; Rx, antithrombotic therapy; parentheses, anticipated recruitment in ongoing trials. more strokes than elderly men, when matched for other risk factors, and this is not accounted for by transesophageal echocardiographic markers of stasis. It is unclear whether sustained control of hypertension reverses the left atrial appendage stasis that leads to thrombus formation; in the elderly, ventricular diastolic abnormalities from hypertension do not always resolve with antihypertensive treatment. Embolism even in patients with chronic, sustained AF is intermittent, separated by days or years, suggesting that the formation of appendage thrombi occurs intermittently. In short, the formation of left atrial appendage thrombi in AF is influenced by several hemodynamic factors that promote stasis, each of which can fluctuate over time, and perhaps also by mild prothrombotic hematological perturbations that favor thrombosis, as yet incompletely defined. Although embolism of left atrial appendage thrombi accounts for most strokes in AF patients, and particularly the larger and more disabling strokes, some brain infarcts in these typically elderly, hypertensive patients are due to other mechanisms. Based on the best available clinical estimates, about two thirds of strokes in AF patients are due to atrial thrombi. This fraction varies according to the distribution of additional risk factors and antithrombotic therapy; high-risk AF patients have a particularly high proportion of cardioembolic strokes. Overall, 10% of all ischemic strokes in population-based cohorts are due to embolism of left atrial appendage thrombi in AF patients (Figure 2). Clinical Trials of Antithrombotic Therapies The results of 18 randomized trials have revolutionized the management of AF patients. 11 These trials have focused on comparisons of oral vitamin K antagonists and aspirin in various doses with placebo, with each other, and with their combination (Table 1). Some trials were restricted to highrisk AF patients, 12 14 others involved those at a low inherent risk, 15,16 and still others were limited to participants aged 75 or older. 17,18 Results from 10 146 participants tell a reasonably consistent story 19 : adjusted-dose warfarin is highly

Hart and Halperin Atrial Fibrillation and Stroke 805 Figure 2. Pathophysiology of AF-associated ischemic stroke. About 16% of ischemic strokes are associated with AF, and 10% are likely due to embolism of atrial appendage thrombus, with the remainder caused by other stroke mechanisms. Figure 1. Pathological specimens of the left atrial appendage. A, Bilobed appendage, seen in about half of autopsies (arrows indicate junction of appendage with atrium proper). B, Multilobed appendage, found in 25% of autopsies (arrows indicate junction of appendage with atrium proper. Adapted from Agnon et al 6 with permission. efficacious (60% stroke reduction), aspirin is modestly efficacious (20% reduction), warfarin is much more efficacious than aspirin (40% reduction), and low-intensity warfarin (INRs 1.5) alone or combined with aspirin offers minimal protection (Table 2). Other randomized comparisons of antithrombotic agents (including dipyridamole) have involved too few patients to be conclusive. Recent surveys of warfarin use in AF patients from clinical practice reveal that the striking efficacy and the modest increases in bleeding are comparable to those observed in the clinical trials. Nevertheless, antithrombotic regimens that are more efficacious than aspirin and easier to administer than adjusted-dose warfarin are needed. The effect of antithrombotic therapy varies according to the ischemic stroke mechanism in AF patients. Aspirin reduces noncardioembolic strokes more than cardioembolic strokes in AF patients, whereas adjusted-dose warfarin is much more efficacious than aspirin for the prevention of cardioembolic strokes. 21,22 Cardioembolic strokes in AF patients are particularly disabling and frequent among AF patients at high risk for stroke. Warfarin reduces cardioembolic stroke in AF patients by 85% during therapeutic anticoagulation, whereas the effect of aspirin on this stroke subtype is probably closer to 15%. 21 Not surprisingly, then, the reduction in disabling stroke (most cardioembolic) with aspirin in a meta-analysis of 3 clinical trials was only 13%. 11 The differential effect of antithrombotic therapies according to stroke mechanism in AF explains discrepant results from clinical trials and is important to consider when choosing antithrombotic prophylaxis for individual patients. For example, lower-risk AF patients have a lesser proportion of cardioembolic stroke, and the reduction in stroke with warfarin compared with aspirin is smaller, in both relative and absolute terms (Figure 3). Hence, although warfarin is clearly more efficacious than aspirin for the prevention of stroke in all AF patients, the magnitude of benefit varies according to the inherent stroke risk, explained by differing proportions of stroke subtypes. AF patients with the highest risk of stroke have the largest proportion of cardioembolic strokes and the

806 Stroke March 2001 TABLE 2. Antithrombotic Therapies for Stroke Prevention in Atrial Fibrillation: Meta-Analysis of Randomized Trials No. of Trials No. of Participants Risk Reduction (95% CI) P Adjusted-dose warfarin vs placebo 6 2900 62% (48 72%) 0.001 Aspirin vs placebo 6 3119 22% (2 38%) 0.03 Antiplatelet vs placebo* 6 3337 24% (7 39%) 0.01 Adjusted-dose warfarin vs aspirin 5 2837 36% (14 52%) 0.003 Adjusted-dose warfarin vs low-dose warfarin 3 893 38% ( 20 68%) NS Aspirin vs low-dose warfarin 2 934 15% ( 42 49%) NS Adjusted-dose warfarin vs low-dose warfarin aspirin 2 1385 Adapted from Hart et al. 11 Stroke includes both ischemic and hemorrhagic stroke. *Includes 218 participants administered dipyridamole alone or combined with aspirin. Estimated mean achieved INRs of 2.6 vs 1.2 from pooling of the 3 trials. 15,16,19 An additional small trial compared 2 different intensities of adjusted-dose warfarin. 20 Statistically significant heterogeneity of the results (73% risk reduction in SPAF III high-risk AF patients 13 vs 0% in AFASAK II low-risk AF patients 15 precluded meta-analysis and is likely explained by stroke subtypes; see text). largest relative and absolute reductions in stroke by warfarin compared with aspirin. Intracerebral hemorrhage is the most devastating complication of anticoagulation. Regarding intracerebral bleeding, AF patients appear to tolerate anticoagulation better than do patients with intrinsic cerebrovascular disease. 23 In the 6 most recent clinical trials that tested the use of warfarin in AF patients, 1486 warfarin-treated participants had a mean age of 72 years, a mean achieved INR estimated as 2.5 during an average follow-up of 1.4 years, and an observed rate of intracerebral hemorrhage of 0.5%/y. 13 16,19,20 The only clinical trial that involved AF patients and reported a substantially higher rate of intracerebral bleeding associated with warfarin (1.8%/y) included participants with a mean age of 80 years and an upper limit of anticoagulation intensity equivalent to an INR of 4.5, monitored with prothrombin time ratios rather than the more accurate INRs. 17 Optimal Intensity of Anticoagulation in AF Optimal protection against ischemic stroke for AF patients is probably achieved with INRs between 2 and 3, 24 with minimal protection when INRs fall below 1.6 (Figure 4). The use of lower-intensity anticoagulation makes warfarin safer and better tolerated, particularly for the elderly, who are at special risk of both major bleeding and the minor bleeding that leads to the discontinuation of anticoagulation. The optimal intensity that balances risk with benefit remains controversial and probably is not the same for all AF patients but rather depends on the estimated risks of cardioembolic stroke and of major bleeding during anticoagulation. 20,24,25 There is no safe INR prolongation that does not increase intracerebral bleeding to some degree; most warfarinassociated intracerebral hemorrhages in recent reports occurred when the INR was within the therapeutic range. 26,27 A target of 2.5 (target range 2 to 3) seems appropriate for younger AF patients and for secondary prevention at any age, whereas an INR of 2.0 (target range 1.6 to 2.5) may be sensible for primary prevention in AF patients over age 75. There may be ethnic influences. 20,28 Additional trials that focus on AF patients over age 75 are warranted to better define the risks and benefits of antithrombotic therapies and Figure 3. Benefit of anticoagulation instead of aspirin according to the inherent rate of stroke in 6 randomized trials. The y axis shows the number of strokes prevented yearly per 100 patients treated with warfarin instead of aspirin. The largest reductions are seen in AF patients at highest risk (EAFT, SPAF III) due to both a larger relative risk reduction and the higher intrinsic stroke rate. See Table 1 for trial eponyms. Figure 4. Relationship between anticoagulation intensity and protection from ischemic stroke in patients with AF. Achieving INRs equally distributed in the range of 1.6 to 2.5 is predicted to provide 90% of the protection of INRs between 2.0 and 3.0 for primary prevention of stroke in patients with nonvalvular AF. Data from Hylek et al 24 and the Stroke Prevention in Atrial Fibrillation Investigators 13 (reprinted with permission of the American College of Physicians).

Hart and Halperin Atrial Fibrillation and Stroke 807 TABLE 3. Predictors of Ischemic Stroke in Nonvalvular Atrial Fibrillation Consistent independent predictors Advancing age* Hypertension Prior stroke or TIA Left ventricular dysfunction Possible independent predictors Diabetes mellitus Systolic blood pressure 160 mm Hg Women over age 75 Postmenopausal hormone replacement therapy Regular alcohol use ( 14 drinks/2 wk) (decreased risk) Coronary artery disease/prior myocardial infarction Postoperative status Mitral regurgitation (decreased risk) Transesophageal echocardiography: appendage thrombi, spontaneous dense echocontrast, reduced appendage flow velocity Not independently predictive Left atrial diameter Intermittency (ie, paroxysmal) From multivariate analyses of prospectively evaluated cohorts or casecontrol studies, adapted from Hart and Halperin. 29 *Some studies dichotomized at age 65 years while others at age 75 years to confer high risk. Clinical congestive heart failure or systolic dysfunction by echocardiography. While predictive, it remains unclear whether independently predictive when other clinical risk factors are considered. the long-term tolerability of different intensities of anticoagulation. Stratification of Stroke Risk Warfarin reduces stroke for all AF patients, but the magnitude of reduction is small for those with low inherent risks of stroke. The stroke rate varies 20-fold among AF patients, from 0.5%/y for young ( 65 years old) AF patients without organic heart disease or hypertension (ie, lone AF) to 12%/y for AF patients with prior stroke or TIA. Clinical features independently associated with high stroke rates in AF patients have been defined (Table 3) and integrated into several risk stratification schemes. 29,30 Although available schemes have much in common, there are important differences that bear on individual patient management. For example, by some schemes, all AF patients age 75 or older are predicted to be high risk for stroke, whereas other schemes classify up to one third of AF patients of this age as low risk. 30 Surprisingly, many elderly patients with recurrent intermittent (ie, paroxysmal) AF have substantial rates of stroke, predicted by similar risk factors that apply to those with sustained AF. 31,32 Although it seems likely that transesophageal echocardiographic measures of stasis (ie, appendage ejection fraction and flow velocity, dense spontaneous echo contrast) are independent predictors of embolic risk, the role of transesophageal echocardiography has yet to be adequately defined and integrated into clinical risk stratification. 33,34 AF patients with relatively low rates of stroke can be identified. 30 In a large prospective study, the Stroke Prevention in Atrial Fibrillation Investigators identified AF patients with a low risk for stroke during treatment with aspirin through the use of specific clinical criteria. 35 During a 2-year mean follow-up of 892 AF patients, the observed stroke rates were 2.0%/y for stroke of any severity and 0.8%/y for disabling stroke (Rankin level 2 or worse). A major research imperative, in our view, is additional study of stratification of stroke risk in AF patients, validating application outside of clinical trials. Reliable risk stratification to identify AF patients who benefit most and least from lifelong anticoagulation is an important precursor to the selection of antithrombotic prophylaxis. Underuse of Anticoagulation in AF Recent studies have documented the increasing use of warfarin in AF patients: about half of AF patients in North American reports are anticoagulated. 36,37 In our view, warfarin is not so much underused as it is poorly used: those at high risk are often not given warfarin in favor of younger, low-risk AF patients who are easier and safer to anticoagulate. 38 At least half of AF patients with high risks of stroke who would stand to benefit substantially from anticoagulation are not treated, and this is especially so for those over age 75. Patient values and preferences regarding anticoagulation have not received sufficient attention; patient-perceived thresholds of benefit for choosing anticoagulation vary widely and may not coincide with those in physician-generated guidelines. 39 The appropriate use of anticoagulation for the 2 million Americans with nonvalvular AF should consider the individual patient s absolute risk reduction afforded by warfarin (derived from stroke risk stratification), the estimated bleeding risk, and the patient s preference after an explanation of the benefits, risks and disutility of anticoagulation therapy. Our initial commentary in 1988 stated that while the importance of AF-associated stroke is not in doubt, preventive strategies remain empiric, citing major uncertainties about mechanisms of AF-associated stroke, stratification of stroke risk, and optimal antithrombotic prophylaxis. 4 We have come very far since then. Evidence-based effective prevention strategies tailored to the individual patient s risks, benefits, and preferences are now available for most AF patients. Results of randomized trials and high-quality observational and case-control studies have prompted a revolution in antithrombotic management, saving many tens of thousands of persons each year from having a stroke. References 1. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomized trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation: the Copenhagen AFASAK study. Lancet. 1989;1:175 179. 2. Stroke Prevention in Atrial Fibrillation Investigators. The Stroke Prevention in Atrial Fibrillation Study: final results. Circulation. 1991;84: 527 539. 3. Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. The effect of low-dose warfarin on the risk of stroke in nonrheumatic atrial fibrillation. N Engl J Med. 1990;323:1505 1511.

808 Stroke March 2001 4. Halperin JL, Hart RG. Atrial fibrillation and stroke: new ideas, persisting dilemmas. Stroke. 1988;19:937 941. 5. Hart RG, Halperin JL. Atrial fibrillation and stroke: revisiting the dilemmas. Stroke. 1994;25:1337 1341. 6. Agmon Y, Khandheria BK, Gentile F, Seward JB. Echocardiographic assessment of the left atrial appendage. J Am Coll Cardiol. 1999;34: 1867 1877. 7. Goldman ME, Pearce LA, Hart RG, Zabalgoitia M, Asinger RW, Safford R, Halperin JL. Pathophysiologic correlates of thromboembolism in nonvalvular atrial fibrillation: reduced flow velocity in the left atrial appendage. J Am Soc Echocardiogr. 1999;12:1080 1087. 8. Feinberg WM, Pearce LA, Hart RG, Cushman M, Cornell ES, Lip GYH, Bovill EG. Markers of thrombin and platelet activity in atrial fibrillation. Stroke. 1999;30:2547 2553. 9. Hart RG, Pearce LA, McBride R, Rothbart RM, Asinger RW. Factors associated with ischemic stroke during aspirin therapy in atrial fibrillation: analysis of 2012 participants in the SPAF I III clinical trials. Stroke. 1999;30:1223 1229. 10. Zabalgoitia M, Pearce LA, Halperin JL, Asinger RW, Blackshear JL, Hart RG. Transesophageal echocardiographic correlates of clinical risk of thromboembolism in nonvalvular atrial fibrillation. J Am Coll Cardiol. 1998;31:1622 1626. 11. Hart RG, Benavente O, McBride R, Pearce LA. Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Ann Intern Med. 1999;131:492 501. 12. European Atrial Fibrillation Trial Study Group. Secondary prevention of vascular events in patients with nonrheumatic atrial fibrillation and recent transient ischemic attack or minor ischemic stroke. Lancet. 1993;342: 1255 1262. 13. Stroke Prevention in Atrial Fibrillation Investigators. Adjusted-dose warfarin versus low-intensity, fixed-dose warfarin plus aspirin for high-risk patients with atrial fibrillation. The Stroke Prevention in Atrial Fibrillation III randomised clinical trial. Lancet. 1996;348:633 638. 14. Morocutti C, Amablie G, Fattapposta F, Nicolosi A, Matteoli S, Tappolini M, Cataldo G, Milanesi G, Lavezzari M, Pamparana F, Coccheri S, for the SIFA (Studio Italiano Fibrillazione Atrial) Investigators. Indobufen versus warfarin in the secondary prevention of major vascular events in nonrheumatic atrial fibrillation. Stroke. 1997;28:1015 1021. 15. Gullov AL, Koefoed BG, Petersen P, Pedersen TS, Andersen ED, Godtfredsen J, Boysen G. Mini-dose warfarin and aspirin in atrial fibrillation. Second Copenhagen Atrial Fibrillation, Aspirin, and Anticoagulation Study (AFASAK 2). Arch Int Med 1998:158:1513 1521. 16. Hellemons BSP, Langenberg M, Lodder J, Vermeer F, Schouten HJA, Lemmens T, van Ree JW, Knotterus JA. Primary prevention of arterial thromboembolism in nonrheumatic atrial fibrillation in primary care: randomised controlled trial comparing two intensities of coumarin with aspirin. BMJ. 1999;319:958 964. 17. Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atrial fibrillation. Stroke Prevention in Atrial Fibrillation II Study. Lancet. 1994;343:687 691. 18. Mant J, Fitzmaurice D, Murray E, Hobbs R. Study does not have the power to show aspirin is as good as anticoagulation. BMJ. 2000; 320:1009. Letter. 19. Pengo V, Zasso A, Barbero F, Banzato A, Nante G, Parissenti L, John N, Noventa F, Dalla Volta S. Effectiveness of fixed minidose warfarin in the prevention of thromboembolism and vascular death in nonrheumatic atrial fibrillation. Am J Cardiol. 1998;82:433 437. 20. Yamaguchi T, for Japanese Nonvalvular Atrial Fibrillation Embolism Secondary Prevention Cooperative Study Group. Optimal intensity of warfarin therapy for secondary prevention of stroke in patients with nonvalvular atrial fibrillation. Stroke. 2000;31:817 821. 21. Hart RG, Pearce LA, Miller VT, Anderson DC, Rothrock JF, Albers GW, Nasco E. Cardioembolic vs noncardioembolic strokes in atrial fibrillation: frequency and effect of antithrombotic agents in the Stroke Prevention in Atrial Fibrillation studies. Cerebrovasc Dis. 2000;10:39 43. 22. Miller VT, Rothrock JF, Pearce LA, Feinberg WM, Hart RG, Anderson DC. Ischemic stroke in patients with atrial fibrillation: effect of aspirin according to stroke mechanism. Neurology. 1993;43:32 36. 23. Gorter JW. Major bleeding during anticoagulation after cerebral ischemia: patterns and risk factors. Stroke Prevention in Reversible Ischemia Trial (SPIRIT) and European Atrial Fibrillation Trial (EAFT) study groups. Neurology. 1999;53:1319 1327. 24. Hylek EM, Skates SJ, Sheehan MA, Singer DE. An analysis of the lowest effective intensity of prophylactic anticoagulation for patients with nonrheumatic atrial fibrillation. N Engl J Med. 1996;335:540 546. 25. Hart RG. Intensity of anticoagulation to prevent stroke in patients with atrial fibrillation. Ann Intern Med. 1998;128:408. Letter. 26. Hylek EM, Singer DE. Risk factors for intracranial hemorrhage in outpatients taking warfarin. Ann Intern Med. 1994;120:897 902. 27. Rosand J, Hylek EM, O Donnell HC, Greenberg SM. Warfarinassociated hemorrhage and cerebral amyloid angiopathy: a genetic pathological study. Neurology. 2000;55:947 951. 28. Nagao T, Hamamoto M, Nagayama H, Nito C, Miyazaki T, Terashi A. Low-intensity anticoagulant therapy in elderly cardioembolic stroke patients with nonvalvular atrial fibrillation. Cerebrovasc Dis. 1996; 6(suppl 2):147. Abstract. 29. Hart RG, Halperin JL. Atrial fibrillation and thromboembolism: a decade of progress in stroke prevention. Ann Intern Med. 1999;131:688 695. 30. Pearce LA, Hart RG, Halperin JL. Assessment of three schemes for stratifying stroke risk in patients with nonvalvular atrial fibrillation. Am J Med. 2000;109:45 51. 31. Hart RG, Pearce LA, Rothbart RM, McAnulty JH, Asinger RW, Halperin JL. Stroke with intermittent atrial fibrillation: incidence and predictors during aspirin therapy. J Am Coll Cardiol. 2000;35:183 187. 32. Muramatsu H, Shimura T, Teramoto A. Incidence of cerebral thromboembolic infarction associated with paroxysmal or persistent atrial fibrillation. Stroke. 2000;31:321. Abstract. 33. Stroke Prevention in Atrial Fibrillation Investigators Committee on Echocardiography. Transesophageal echocardiographic correlates of thromboembolism in high-risk patients with atrial fibrillation. Ann Intern Med. 1998;128:639 647. 34. Stollberger C, Chnupa P, Kronik G, Brainin M, Finsterer J, Schneider B, Slany J. Transesophageal echocardiography to assess embolic risk in patients with atrial fibrillation. Ann Intern Med. 1998;128:630 638. 35. Stroke Prevention in Atrial Fibrillation Investigators. Prospective identification of patients with nonvalvular atrial fibrillation at low-risk of stroke during treatment with aspirin. JAMA. 1998;279:1273 1277. 36. Go AS, Hylek EM, Borowsky LH, Phillips KA, Selby JV, Singer DE. Warfarin use among ambulatory patients with nonvalvular atrial fibrillation. Ann Intern Med. 1999;131:927 934. 37. Smith NL, Psaty BM, Furberg CD, White R, Lima JA, Newman AB, Manolio TA. Temporal trends in the use of anticoagulants among older adults with atrial fibrillation. Arch Intern Med. 1999;159:1574 1578. 38. Sudlow M, Thomson R, Thwaites B, Rodgers H, Kenny RA. Prevalence of atrial fibrillation and eligibility for anticoagulants in the community. Lancet. 1998;352:1167 1171. 39. Howitt A, Armstrong D. Implementing evidence based medicine in general practice: audit and qualitative study of antithrombotic treatment for atrial fibrillation. BMJ. 1999;318:1324 1327. KEY WORDS: anticoagulants aspirin atrial fibrillation cerebral embolism cerebrovascular disorders stroke