Anticoagulation to Prevent Stroke in Atrial Fibrillation and Its Implications for Managed Care

Anticoagulation to Prevent Stroke in Atrial Fibrillation and Its Implications for Managed Care Daniel E. Singer,

MD

Nonrheumatic atrial fibrillation (AFib) is the most potent common risk factor for stroke, raising the risk of stroke 5-fold. Six randomized trials of anticoagulation in AFib consistently demonstrated a reduction in the risk of stroke by about two-thirds. In these trials, anticoagulation in AFib was quite safe. In contrast, randomized trials indicate that aspirin confers only a small reduction in risk of stroke, at best. Pooled data from the first set of randomized trials indicate that prior stroke, hypertension, diabetes, and increasing age are independent risk factors for future stroke with AFib. Individuals <65 years old with none of the other risk factors might safely avoid anticoagulation; for all others, anticoagulation seems indicated. Studies of hemorrhagic risk highlight the importance of keeping the international normalized ratio (INR) <4.0. Recent analyses also reveal that risk of ischemic stroke in AFib increases greatly at INR levels <2.0. Efficacy and safety of anticoagulation in AFib

depend on maintaining the INR between 2.0 –3.0. Costeffectiveness studies indicate that anticoagulation for AFib is among the most efficient preventive interventions in adults. Importantly, the benefits of anticoagulation in AFib accrue immediately. The implications for managed care organizations are that anticoagulation for AFib should be encouraged in their covered populations, and that dedicated anticoagulation services should be developed to promote system-wide control of anticoagulation intensity. Quality measures would include the proportion of patients with AFib who are anticoagulated, and the percentage of time patients’ INR levels are between 2.0 –3.0. Managed care organizations can benefit from recent research on anticoagulation for AFib; they have a responsibility to support future research and development efforts. Q1998 by Excerpta Medica, Inc. Am J Cardiol 1998;81(5A):35C– 40C

ver the past 2 decades, clinical insights extended by high quality epidemiologic studies O have established that atrial fibrillation (AFib) is an

tinction is the presence of mitral stenosis, which confers a greatly increased risk of stroke with AFib.2 Indeed, without any randomized trials, common practice has dictated lifelong anticoagulation for patients with mitral stenosis and AFib. Until relatively recently, the relation between nonrheumatic AFib (which is far more prevalent than rheumatic AFib) and stroke, and the likely benefit of anticoagulation in patients with AFib, were controversial. AFib occurs as people get older and as they develop other cardiac diseases. AFib might simply be a marker of these other risk factors for stroke and not a true cause of stroke itself. Data from the Framingham Heart Study, as well as other studies, have established that AFib is the most potent common risk factor for stroke.3 The prevalence of AFib increases strikingly with age, affecting 4% of those age .60 and 10% of those .80. In those without AFib, the incidence of stroke, rises from 4.5 per 1,000 person-years in people in their 60s, to 9 per 1,000 person-years for people in their 70s, to 14.3 per 1,000 person-years for people in their 80s. In the presence of AFib the rate of stroke is increased about 5-fold across the entire range of older decades. This persistence of effect into the oldest age groups is distinctive among common risk factors for stroke.4 Because AFib is such a strong determinant of stroke and because it is so common in older and thereby stroke-prone individuals, it accounts for a significant fraction of all ischemic strokes—roughly 14% of all strokes above the age of 60, or 75,000 strokes per year in the United States.

1

extraordinarily potent cause of ischemic stroke, and that long-term, well-controlled anticoagulation can safely reverse this risk. Done well, anticoagulation for the predominantly older population with AFib can prevent one of the most feared disabling events, i.e., stroke, and can do so efficiently. Done poorly, anticoagulation can result in great harm, via disastrous hemorrhagic complications. The clear implication for managed care organizations that enroll large numbers of older individuals is that anticoagulation has to be carefully organized and monitored. Dedicated anticoagulation services, or anticoagulation units, seem a logical system-wide approach to maximize control over anticoagulation intensity. Such control should allow real-world medical practice to achieve much of the substantial net benefit of anticoagulation seen in the randomized trials of AFib.

AFib AS A RISK FACTOR FOR STROKE Historically, AFib has been categorized as rheumatic or nonrheumatic in etiology. The central disFrom the Clinical Epidemiology Unit, General Medicine Division, Medical Services, Massachusetts General Hospital, and the Departments of Medicine and of Health Care Policy, Harvard Medical School, and the Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts. Address for reprints: Daniel E. Singer, MD, Clinical Epidemiology Unit, S50-9, Massachusetts General Hospital, Boston, Massachusetts 02114. ©1998 by Excerpta Medica, Inc. All rights reserved.

0002-9149/98/$19.00 PII S0002-9149(98)00185-4

35C

THE RANDOMIZED TRIALS OF ANTITHROMBOTIC THERAPY IN AFib Despite the evidence that AFib was a risk factor for stroke, there remained uncertainty whether warfarin would actually prevent stroke in AFib, and, even if it did, whether anticoagulation in such predominantly older patients would be adequately safe. In response to the mounting epidemiologic evidence that AFib led to embolic stroke, a series of randomized trials of antithrombotic therapy was begun in the mid-1980s.5–10 The core findings of these randomized trials were dramatic and consistent (Table I). Each of the 5 primary prevention trials stopped early because of the marked efficacy of warfarin (the Canadian Atrial Fibrillation Anticoagulation [CAFA] trial stopped early because of the results of other trials). The observed efficacy, measured as relative risk reduction, was 52– 86%. Because the trials were stopped early, the number of outcome events observed in each trial was small, resulting in wide confidence intervals and limiting the power of subgroup analyses. These problems were addressed by pooling the trials’ data (discussed below). The secondary prevention European Atrial Fibrillation Trial (EAFT) studied patients with AFib who had had a prior minor stroke or transient ischemic attack.10 These patients were clearly at increased risk, with the control group rate of stroke being .12% per year. Still, the relative risk reduction due to anticoagulants seen in EAFT was essentially the same as that for the primary prevention trials. Two additional points about these trials’ results are worth noting. First, the efficacy values were estimated according to the intention-to-treat principle. In fact, more than one-quarter of the strokes counted in the warfarin arms occurred among patients randomized to anticoagulants who were not taking their assigned medicine at the time of their stroke. Second, the lowest target intensity of anticoagulation tested was as effective as higher intensities. The Boston Area Anticoagulation Trial for Atrial Fibrillation (BAATAF) and Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Study (SPINAF) trials used a prothrombin time ratio of 1.2–1.5, roughly corresponding to an INR target of 2–3, yet their estimates of efficacy were actually somewhat higher than those of the other trials. The first 5 trials pooled their results to provide more precise estimates of the effect of anticoagulation and of risk factors for stroke in AFib.11 The pooled relative risk reduction for the 5 primary prevention trials was 68% (analysis according to intention to treat), with a fairly narrow confidence interval of 50 –79%. The EAFT results demonstrating a relative risk reduction of 66% were remarkably consistent with this estimate from the first set of trials. Since AFib raises the risk of stroke 5-fold, a complete reversal of the effect of AFib would result in a relative risk reduction of 80%. The observed relative risk reduction due to anticoagulants in AFib is close to such a complete reversal. For the primary prevention trials, the average absolute risk reduction was 3.1% per year. This is a large 36C THE AMERICAN JOURNAL OF CARDIOLOGYT

effect for primary prevention of cardiovascular disease, corresponding to about 5 times the impact of treating hypertension in the elderly.12 For the secondary prevention EAFT study the absolute risk reduction of 8% per year is clearly sizable. In these trials, anticoagulation was also very safe. The only hemorrhagic complication of warfarin comparable in impact to ischemic stroke is intracranial hemorrhage. The increase in intracranial hemorrhage in the 5 primary prevention trials was only 0.2% per year; no intracranial hemorrhages were reported in the EAFT study. Aggregate major hemorrhage (e.g., intracranial hemorrhage plus hemorrhage leading to hospitalization and/or transfusion) was higher, but the highest rate observed (in the EAFT study) was only 1.8% per year. The results of the trials were dramatically favorable. Anticoagulation virtually removed the risk of stroke due to AFib, at intensities of anticoagulation that conferred only a small increase in bleeding complications. As with all such randomized trials, the generalizability of these findings to usual care is not certain. In particular, it is likely that the participants in the trials were selected, in part, because they were good candidates for safe anticoagulation. Bleeding rates may be higher among patients with AFib in general practice. Indeed, one subsequent trial in AFib observed a rate of intracranial hemorrhage on warfarin of almost 1.8% per year in patients with AFib over age 75.13 However, this represented only 7 events, and subsequent studies of anticoagulation in clinical practice report rates of major bleeding similar to those in the trials.14

THE RANDOMIZED TRIALS OF ASPIRIN IN AFib Aspirin would be an attractive alternative to warfarin if it were comparably effective. However, substantial evidence indicates that aspirin’s efficacy is small, at best. There have been 4 separately randomized trials of aspirin versus placebo in AFib: Atrial Fibrillation, Aspirin, Anticoagulation Study (AFASAK) which used 75 mg per day,6 2 separately randomized trials within the Stroke Prevention in Atrial Fibrillation Study (SPAF I) using 325 mg per day,15 and EAFT using 300 mg per day.10 The SPAF I Group 1 trial included patients who could be randomized to warfarin and compared groups treated with aspirin, aspirin-placebo, as well as warfarin. Group 2 of the SPAF I trial included patients with AFib who would not, or could not take warfarin, and simply compared groups assigned to aspirin versus placebo. In AFASAK, the rate of outcome events was 16% lower in the aspirin group, a statistically nonsignificant result. In SPAF I Group 2, there were 28 events in control versus 25 in aspirin for a nonsignificant reduction of 8%. There was, as well, no significant effect of aspirin seen in the EAFT study, which observed 88 events in the aspirin category versus 90 in control. Interest in aspirin as an antithrombotic in AFib stems virtually entirely from the SPAF I Group

VOL. 81 (5A)

MARCH 12, 1998

TABLE I Randomized Trials of Anticoagulation for Atrial Fibrillation1

Anticoagulation target Subjects, n Emboli, n Annual rate Control subjects, n Emboli, n Annual rate Preventive efficacy 95% CI

AFASAK

BAATAF

SPAF

CAFA

SPINAF

EAFT

INR 2.8–4.2

PTR 1.2–1.5

PTR 1.3–1.8

INR 2–3

PTR 1.2–1.5

INR 2.5–4

335 4 1.6% 336 21 5.5% 71% 3–90%

212 2 0.41% 208 13 3.0% 86% 51–96%

210 6 2.3% 211 18 7.4% 69% 27–85%

187 5 2.5% 191 11 5.2% 52% (236)–87%

260 4 0.88% 265 19 4.3% 79% 52–90%

225 20 3.9% 214 50 12.3% 66% 43–80%

1 Abbreviated titles of trials: AFASAK 5 Atrial Fibrillation, Aspirin, Anticoagulation Study; BAATAF 5 Boston Area Anticoagulation Trial for Atrial Fibrillation; CAFA 5 Canadian Atrial Fibrillation Anticoagulation Study; EAFT 5 the secondary prevention European AF Trial; SPAF 5 Stroke Prevention in Atrial Fibrillation Study; SPINAF 5 Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Study. InR 5 international normalized ratio; PTR 5 prothrombin time ratio; CI 5 confidence interval. Preventive efficacy is the relative risk reduction calculated as (1 2 RR) 3 100, where RR is the annual rate in the anticoagulation group divided by the annual rate in the control group. Adapted with permission from Ann Epidemiol.30

TABLE II Pooled Analysis of First Five AFib Trials: Efficacy of Warfarin by Risk Category* Control 1

Risk Category

Age ,65 years: No risk factor $1 risk factor Age 65–75 years: No risk factor $1 risk factor Age .75: No risk factor $1 risk factor

# Strokes

Warfarin Rate (95% CI)

# Strokes

Rate (95% CI)

3 16

1.0% (0.3–3.1) 4.9% (3.0–8.1)

3 6

1.0% (0.3–3.0) 1.7% (0.8–3.9)

16 27

4.3% (2.7–7.1) 5.7% (3.9–8.3)

4 7

1.1% (0.4–2.8) 1.7% (0.9–3.4)

6 13

3.5% (1.6–7.7) 8.1% (4.7–13.9)

3 2

1.7% (0.5–5.2) 1.2% (0.3–5.0)

*The first 5 trials are AFASAK, BAATAF, CAFA, SPAF, and SPINAF (listed in footnote to Table I). Risk factors are history of hypertension, diabetes, or prior stroke or transient ischemic attack. Rate is annual rate; CI is confidence interval. Adapted with permission from Arch Intern Med.11

1 trial, where there were 18 events in control versus only 1 on aspirin, for the highest efficacy seen for any therapy: 94%. Detailed analyses to explain the disparate results of SPAF I Group 1 according to patient features have been unsatisfactory.15 A recent analysis of the pooled data from the AFASAK, SPAF I, and EAFT studies provided an estimate of aspirin’s efficacy of 21%, with the 95% confidence interval ranging from 0 (i.e., no effect) to 38%.16 This analysis did not adjust for the heterogeneous results of the two SPAF I studies. Accounting for such heterogeneity would have broadened the confidence limits further, making the effect of aspirin clearly statistically nonsignificant. The results of the SPAF III trial add even more evidence that aspirin is a very weak agent for preventing stroke in AFib.17 SPAF III compared warfarin targeted at INR 2–3 to a combination of aspirin at 325 mg per day plus a fixed very low dose of warfarin targeted at INR 1.2–1.5 in high-risk patients with AFib. The standard intensity of warfarin (INR 2–3) resulted in more than a two-thirds reduction in rate of embolic events, roughly the same as standard intensity anticoagulation versus no therapy in the

previous trials. A reasonable conclusion from these studies is that aspirin’s effect in AFib is very small to nil.

RISK FACTORS FOR STROKE IN AFib The pooled analysis of the first 5 trials assessed possible clinical risk factors for stroke among the control patients in the trials.11 In this analysis, prior stroke, hypertension, older age, and diabetes were independent risk factors for subsequent stroke in AFib. Patients in the first 5 trials were categorized according to age and the presence of at least one of the other risk factors— hypertension, diabetes, or prior stroke (Table II). For patients who were younger than 65 and had no risk factor, the rate of stroke was 1% per year. This rate did not appear to be decreased by anticoagulation. Such findings support the conclusion that younger patients with socalled “lone” AFib do not need to be anticoagulated. However, for all other categories of patients with AFib, the rate of stroke was substantial and the benefit of warfarin so dramatic that warfarin is strongly indicated. A SYMPOSIUM: AFIB AND MANAGED CARE

37C

THE RISK OF BLEEDING WITH ANTICOAGULATION

THE LOWEST EFFECTIVE INTENSITY OF ANTICOAGULATION IN AFib

Reluctance to use anticoagulants, particularly in older patients, stems largely from concern about inducing bleeding complications. Perhaps the most fortuitous decision in designing several of the original randomized trials in AFib was the choice of a relatively low intensity target. Such targets (prothrombin time ratio of 1.2–1.5, or INR 2.0 –3.0) were based on the efficacy of anticoagulants demonstrated in trials in thrombophlebitis reported shortly before the design of the trials in AFib.18 The rates of major bleeding observed in the trials of AFib were really quite low. As a result, even the pooled analysis provides insufficient data for powerful assessment of risk factors for bleeding complications on warfarin. Analyses from anticoagulation clinics in the United States and in the Netherlands are now providing the information needed to identify such risk factors.19 –21 These latter sources of data have the very important additional strength that they represent actual clinical practice. The general conclusions from such studies are: (1) the rates of major hemorrhage among patients managed by anticoagulation units are not much higher than the rates seen in the trials; (2) the most powerful determinant of hemorrhagic risk is the intensity of anticoagulation; and (3) increasing age appears to confer a modestly greater risk of major hemorrhage, although the data are somewhat inconsistent on this point.21–23 The category of major bleeding complications includes many types of events, each with different health and survival implications. In the context of stroke prevention in AFib, the risk of intracranial hemorrhage on warfarin is particularly germane. It is the only type of hemorrhagic complication that regularly produces health impairments as severe as ischemic stroke. However, its low frequency of occurrence makes study of intracranial hemorrhage difficult. Case-control study designs are one approach to surmount the problem of small incidence rates. We assembled 121 cases of intracranial hemorrhage on warfarin hospitalized at Massachusetts General Hospital over an 11-year period and compared each with 3 calendar-matched controls selected from the large Massachusetts General Hospital anticoagulant therapy unit.21 There were 77 cases of intracerebral hemorrhage and 44 cases of subdural hemorrhage. Intracranial hemorrhage was devastating. Of those with intracerebral hemorrhage, 46% died, as did 20% of those with subdural hemorrhage. Few in either group left the hospital without major disability. The most potent risk factor in our analysis was one that physicians can control, i.e., the intensity of anticoagulation (Figure 1). The odds ratio for intracranial hemorrhage increased dramatically for prothrombin time ratios .2.0. This corresponds to INR levels .4.0. These findings have been largely corroborated by follow-up studies at Thrombosis Centers in the Netherlands.20

Since risk of hemorrhage increases with the intensity of anticoagulation, there is great interest in defining the lowest intensity of anticoagulation that is still effective in preventing ischemic stroke in AFib. Surveys indicate that physicians frequently use very low intensity targets, e.g., INR 1.5, in older patients, to minimize the risk of hemorrhage.24 The core set of completed randomized trials in AFib indicate that INR targets as low as 2.0 seem effective. Because of the efficacy of anticoagulation there were too few strokes observed in the warfarin arms of the trials to powerfully assess at what level of INR warfarin loses its efficacy. Here again, case-control approaches can be very informative. We assembled a consecutive series of 74 cases of ischemic stroke among patients taking warfarin admitted to our hospital and compared them with randomly selected control patients also taking warfarin who were managed by our anticoagulation therapy unit (3 controls per case).25 For both groups INR values were available. The findings were striking (Figure 2). The stroke-preventive efficacy of warfarin was even more sensitive to INR level than was the rate of hemorrhagic complications. At INR levels ,2.0, the risk of ischemic stroke escalated rapidly. Our data indicate that most of the risk of ischemic stroke on anticoagulants is due to periods when the INR is ,2.0. No clear additional stroke-preventive benefit is gained by INRs much .2.0.

38C THE AMERICAN JOURNAL OF CARDIOLOGYT

COST-EFFECTIVENESS OF ANTICOAGULATION FOR AFib Anticoagulation leads to clear health benefits for patients with AFib. A variety of analyses have demonstrated that these benefits can be achieved very efficiently. Indeed, analyses for both Sweden and for the United States indicate that for the many patients with AFib who are at moderate risk of stroke (i.e., have a risk factor) anticoagulation is actually costsaving.26,27 That is, anticoagulation provides dramatic health benefits at the same time that it also decreases healthcare costs. Very few medical interventions in adults are so efficient. A financial concern for managed care companies is whether they will realize the benefits of their investment in preventive therapies. With time individuals may shift into other insurance plans. As a result, insurers find less attractive preventive therapies that have a long latent period before benefits accrue. A distinct advantage of anticoagulation for AFib is that there is no latent period; patients realize an immediate reduction is risk of ischemic stroke. The distinct health benefits for enrollees, the cost advantages, and the timing of the adverse events prevented all should make widespread anticoagulation for AFib an appealing goal for managed care.

VOL. 81 (5A)

MARCH 12, 1998

agulants for AFib, from very low pretrial levels, following the published reports of the initial randomized trials.28

IMPLICATIONS FOR MANAGED CARE

FIGURE 1. The odds ratio of intracranial hemorrhage among patients receiving warfarin as a function of the prothrombin time ratio. The reference category is prothrombin time ratio of 1.0 – 1.5. 95% confidence limits are indicated. Reprinted with permission from, and further details are provided in, Ann Intern Med.21

FIGURE 2. The odds ratio for ischemic stroke among patients with atrial fibrillation receiving warfarin as a function of the international normalized ratio (INR). The reference category is INR of 2.0. Reprinted with permission from, and further details are provided in, N Engl J Med.25

TRANSLATION OF THE FINDINGS OF THE RANDOMIZED TRIALS INTO CLINICAL PRACTICE Anticoagulation is generally burdensome to physicians and to patients and is risky. Nonetheless, such therapy can prevent stroke and the loss of independence and function that stroke produces. Translation of effective preventive practices into usual clinical care is often slow. One might anticipate that it would be particularly slow for a complicated therapy such as anticoagulation. We studied this process using the National Ambulatory Medical Care Survey. For this survey, physicians record features of all medical care encounters during a designated period. Analyses of these data demonstrate that only 34% of patients with AFib were receiving anticoagulants during the most recent analyzable period (1992–1993). However, the time trends revealed a rapid increase in use of antico-

AF becomes a significant issue for managed care organizations when they enroll Medicare populations. The past 2 decades’ research on preventing stroke in AFib has clear implications for managed care organizations. Most patients with chronic AFib should receive anticoagulants unless there are clear contraindications. Even more importantly, systems should be established to provide organized, obsessive monitoring and management of anticoagulation. The success of anticoagulation in AFib depends on tight control of the INR within the range of 2.0 –3.0. This fact allows an extremely focused program of quality maintenance. The advantages of organized anticoagulation units seem obvious. Dedicated personnel who track patients and INR test results and who are supported by computer-based management systems will certainly achieve better control of INR levels than the average isolated practitioner. Such organized units may be able to more quickly apply new technology that may improve management of anticoagulation. An example would be the use of patient self-testing of INRs to improve patient compliance and anticoagulation control.29 The structure of such units will need to be adapted to local circumstances. Managed care organizations are always searching for meaningful measures of quality that can be easily obtained. For anticoagulation for AFib the most obvious such measures are: (1) the proportion of patients with chronic AFib, either sustained or paroxysmal, who are anticoagulated; and (2) the proportion of time patients with AFib have INR levels of 2.0 –3.0. The former measure is really a proxy for the true measure of quality, i.e., the proportion of patients with AFib who should be receiving anticoagulants who are anticoagulated. The latter measure is a direct measure of the quality of anticoagulation management. To implement these measures of quality, managed care organizations need valid and comprehensive registries of patients with AFib. The diagnostic code for AFib has to be available and periodic validation of such coding needs to be implemented. The ideal proportion of patients with AFib who are receiving anticoagulants is unclear, but at least one staff model health maintenance organization has reported that about 60% of their patients with AFib were anticoagulated with good results.14 The extensive and largely successful research on preventing stroke in AFib is both good for patients and for managed care companies. In turn, managed care’s emphasis on system-wide approaches to providing health care and monitoring its quality should allow more widespread and higher quality anticoagulation for AFib. One final point should be emphasized. Managed care companies will reap the benefits of decades of research on AFib. There are numerous fertile areas for future research in this field. These include defining more sophisticated markers of stroke risk in AFib and A SYMPOSIUM: AFIB AND MANAGED CARE

39C

of bleeding risk with anticoagulation, and of developing more highly engineered anticoagulation dosing regimens. Managed care should sponsor such clinical research, both because it is a responsibility of important healthcare organizations to provide for future advances and because it is a good business investment. 1. Fisher CM. Treatment of chronic atrial fibrillation. (Letter.) Lancet 1972;1: 1284. 2. Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: The Framingham Study. Neurology 1978;28:973–977. 3. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation: a major contributor to stroke in the elderly. Arch Intern Med 1987;147:1561–1564. 4. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: The Framingham Study. Stroke 1991;22:983–988. 5. The Boston Area Anticoagulation trial for Atrial Fibrillation Investigators. The effect of low-dose warfarin on the risk of stroke in patients with nonrheumatic atrial fibrillation. N Engl J Med 1990;323:1505–1511. 6. Petersen P, Godtfredsen J, Boysen G, Andersen ED, Andersen B. Placebocontrolled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation: the Copenhagen AFASAK study. Lancet 1989;1:175–179. 7. Stroke Prevention in Atrial Fibrillation Investigators. Stroke prevention in atrial fibrillation study: final results. Circulation 1991;84:527–539. 8. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian atrial fibrillation anticoagulation (CAFA) study. J Am Coll Cardiol 1991;18:349 – 355. 9. Ezekowitz MD, Bridgers SL, James KE, Carliner NH, Colling CL, Gornick CC, Krause-Steinrauf H, Kurtzke JF, Nazarian SM, Radford MJ, Rickles FR, Shabetal R, Deykin D, for the Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Investigators. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. N Engl J Med 1992;327:1406 – 1412. 10. EAFT (European Atrial Fibrillation Trial) Study Group. Secondary prevention in non-rheumatic atrial fibrillation after transient ischaemic attack or minor stroke. Lancet 1993;342:1256 –1262. 11. Atrial Fibrillation Investigators. Atrial fibrillation: Risk factors for embolization and efficacy of anti-thrombotic therapy. Arch Intern Med 1994;154:1449 – 1457. 12. Mulrow CD, Cornell JA, Herrera CR, Kadri A, Farnett L, Aguilar C. Hypertension in the elderly: implications and generalizability of randomized trials. JAMA 1994;272:1932–1938. 13. Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atrial fibrillation: The Stroke Prevention in Atrial Fibrillation II Study. Lancet 1994;343:687– 691.

40C THE AMERICAN JOURNAL OF CARDIOLOGYT

14. Gottlieb LK, Salem-Schatz S. Anticoagulation in atrial fibrillation: does

efficacy in clinical trials translate into effectiveness in practice. Arch Intern Med 1994;154:1945–1953. 15. The Stroke Prevention in Atrial Fibrillation Investigators. A differential effect of aspirin on prevention of stroke in atrial fibrillation. J Stroke Cerebrovasc Dis 1993;3:181–188. 16. Atrial Fibrillation Investigators. The efficacy of aspirin in patients with atrial fibrillation: Analysis of pooled data from three randomized trials. Arch Intern Med 1997;157:1237–1240. 17. The Stroke Prevention in Atrial Fibrillation Investigators. Adjusted dose warfarin versus low-intensity, fixed dose warfarin plus aspirin for high-risk patients with atrial fibrillation: Stroke Prevention in Atrial Fibrillation III randomised clinical trial. Lancet 1996;348:633– 638. 18. Hull R, Hirsh J, Jay R, Carter C, England C, Gent M, Turpie AGG, McLaughlin D, Dodd P, Thomas M, Raskob G, Ockelford P. Different intensities of oral anticoagulant therapy in the treatment of proximal-vein thrombosis. N Engl J Med 1982;307:1676 –1681. 19. Fihn SD, McDonell M, Martin D, Henikoff J, Vermes D, Kent D, White RH, for the Warfarin Optimized Outpatient Follow-up Study Group. Risk factors for complications of chronic anticoagulation. Ann Intern Med 1993;118:511–520. 20. Cannegieter SC, Rosendaal FR, Wintzen AR, van der Meer FJM, Vandenbroucke JP, Briet E. Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl J Med 1995;333:11–17. 21. Hylek EM, Singer DE. Risk factors for intracranial hemorrhage in outpatients taking warfarin. Ann Intern Med 1994;120:897–902. 22. Gurwitz JH, Avorn J, Ross-Degnan D, Choodnovsky I, Ansell J. Aging and the anticoagulant response to warfarin therapy. Ann Intern Med 1992;116:901– 904. 23. Fihn SD, Callahan CM, Martin DC, McDonell MB, Henikoff JG, White RH. The risk for and severity of bleeding complications in elderly patients with warfarin. Ann Intern Med 1996;124:970 –979. 24. McCrory DC, Matchar DB, Samsa G, Sanders LL, Pritchett ELC. Physician attitudes about anticoagulation for nonvalvular atrial fibrillation in the elderly. Arch Intern Med 1995;155:277–281. 25. 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. 26. Gustafsson C, Asplund K, Britton M, Norrving B, Olsson B, Marke L-A. Cost effectiveness of primary stroke prevention in atrial fibrillation: Swedish national perspective. Br Med J 1992;305:1457–1460. 27. Gage BF, Cardinalli AB, Albers GW, Owens DK. Cost-effectiveness of warfarin and aspirin for prophylaxis of stroke in patients with nonvalvular atrial fibrillation. JAMA 1995;274:1839 –1845. 28. Stafford RS, Singer DE. National patterns of warfarin use in atrial fibrillation. Arch Intern Med 1996;156:2537–2541. 29. Ansell JE, Patel N, Ostrovsky D, Nozzolillo E, Petersen AM, Fish L. Long-term patient self-management of oral anticoagulation. Arch Intern Med 1995;155:2185–2189. 30. Singer DE. Overview of the randomized trials to prevent stroke in atrial fibrillation. Ann Epidemiol 1993;3:563–567.

VOL. 81 (5A)

MARCH 12, 1998