Context Patients with nonvalvular atrial fibrillation (AF) have an increased
risk of stroke and other vascular events.
Objective To compare the risk of vascular and bleeding events in patients with
nonvalvular AF treated with vitamin K –inhibiting oral anticoagulants
or acetylsalicylic acid (aspirin).
Design Pooled analysis of patient-level data from 6 published, randomized clinical
trials.
Patients A total of 4052 patients with AF randomly assigned to receive therapeutic
doses of oral anticoagulant or aspirin with or without low-dose oral anticoagulants.
Main Outcome Measures Ischemic and hemorrhagic stroke, other cardiovascular events, all-cause
death, and major bleeding events. Person-year incidence rates were calculated
to provide crude comparisons. Relative efficacy was assessed using proportional
hazards modeling stratified by study. The variation of the oral anticoagulant's
relative effect by pertinent patient factors was explored with interaction
terms. All analyses were conducted using the intention-to-treat principle.
Results Patients receiving oral anticoagulant and aspirin were balanced for
important prognostic factors. There was no significant heterogeneity between
trials in the relative efficacy of oral anticoagulant vs aspirin for any outcome.
Patients receiving oral anticoagulant were significantly less likely to experience
any stroke (2.4 vs 4.5 events per 100 patient-years; hazard ratio [HR], 0.55;
95% confidence interval [CI], 0.43-0.71), ischemic stroke (HR, 0.48; 95% CI,
0.37-0.63), or cardiovascular events (HR, 0.71; 95% CI, 0.59-0.85) but were
more likely to experience major bleeding (2.2 vs 1.3 events per 100 patient-years;
HR, 1.71; 95% CI, 1.21-2.41). The reduction in ischemic stroke risk was similar
in patients with paroxysmal AF (1.5 vs 4.7 events per 100 patient-years; HR,
0.32; 95% CI, 0.16-0.61; P<.001). Treating 1000
patients with AF for 1 year with oral anticoagulant rather than aspirin would
prevent 23 ischemic strokes while causing 9 additional major bleeds. Overall
all-cause survival did not differ but appeared to improve for oral anticoagulant
patients 3 years after therapy was started.
Conclusions Compared with aspirin, oral anticoagulant significantly decreases the
risk of all strokes, ischemic strokes, and cardiovascular events for patients
with nonvalvular chronic or paroxysmal AF but modestly increases the absolute
risk of major bleeding. The balance of benefits and risks varies by patient
subgroup.
Nonvalvular atrial fibrillation (AF) more than quadruples the risk of
stroke.1 Stroke prophylaxis in patients with
atrial fibrillation has been examined extensively in randomized trials,2-16 analyses
of individual patient-level data,17-19 traditional
meta-analyses,20-31 and
review articles.32-42 The
clinical trials, and their quantitative reviews, show that treatment with
vitamin K–dependent oral anticoagulants significantly decrease the risk
of stroke in AF patients by more than 50%. Acetylsalicylic acid (aspirin)
also appears to decrease stroke risk but to a lesser extent.20 Because
oral anticoagulant use is more troublesome43 and
has a greater likelihood of complications,44 it
is important to determine the relative efficacy and safety of oral anticoagulant
vs aspirin overall and in relevant patient subgroups.
Six published clinical trials have randomized AF patients to oral anticoagulant
or a regimen containing aspirin.2-7,9 Three
of these studies found that oral anticoagulant was significantly more efficacious
than aspirin at reducing thromboembolic outcomes.2,5,9 However,
the conclusions from quantitative reviews of randomized trials comparing oral
anticoagulant and aspirin-containing regimens have varied extensively from
oral anticoagulants being "substantially more efficacious than aspirin"20 to finding "moderate evidence for fewer strokes among
patients on oral anticoagulants than on aspirin"31 to
"considerable uncertainty about the value of long-term anticoagulation compared
to antiplatelet treatment."45 This uncertainty
of the relative benefit of oral anticoagulant vs aspirin in AF might explain
the relatively low use of oral anticoagulant in usual practice.46
All prior meta-analyses have been limited to using summary data available
in published trial reports. We analyzed the pooled individual patient data
from all published randomized trials comparing oral anticoagulant and aspirin
for AF. These data are distinctive by permitting true time-to-event analyses,
exploration of subgroup effects, and comparison of the efficacy of oral anticoagulant
vs aspirin on a number of important cardiovascular outcomes in addition to
ischemic stroke.
Included Studies: Design and Treatment
This study included patient-level data from every published clinical
trial in which patients with AF were randomly assigned to full-dose oral anticoagulant
or aspirin. These trials included the Atrial Fibrillation, Aspirin, Anticoagulation
(AFASAK) Studies 12 and 2,3 Primary
Prevention of Arterial Thromboembolism in patients with Nonrheumatic Atrial
Fibrillation in Primary Care (PATAF),4 the
European Atrial Fibrillation Trial (EAFT),5 and
the Stroke Prevention in Atrial Fibrillation (SPAF) Studies 1,6 2,7 and 3.9 Since SPAF
1 patients randomized to receive oral anticoagulant or aspirin all continued
into SPAF 2, SPAF 1 data are combined with SPAF 2 as in the original publication.7 To our knowledge, there are no other published studies
in which AF patients were randomized to receive a full-dose of oral anticoagulant
or aspirin.
All patients were adults with nonvalvular AF. The exclusion criteria
varied only slightly between the studies. In general, patients were excluded
if they had clinical indications for, or contraindications to, oral anticoagulation
or aspirin. These contraindications included pregnancy, alcoholism, renal
or hepatic failure, thrombocytopenia, or bleeding disorders. With the exception
of the AFASAK studies,2,3 patients
were also excluded if they had a recent acute coronary event or cardiac revascularization.
All studies also excluded patients with thyrotoxicosis except for AFASAK 1,
which included only 28 such patients.
Patients were randomly assigned to receive a full-dose of oral anticoagulant
or aspirin with or without low-dose oral anticoagulant. Patients in the oral
anticoagulant group were treated primarily with coumarin derivatives including
warfarin sodium and 4-hydroxycoumarin. The target international normalized
ratio (INR) varied slightly in each trial (Table 1). The daily aspirin dose ranged between 75 and 325 mg (Table 1).
In our analyses, we combined patients treated with aspirin alone with
patients treated with aspirin plus low-dose warfarin because there were no
significant differences in outcomes between such patients.3 We
also reasoned that, if low-dose warfarin has a previously unmeasured effect
on cardiovascular outcomes, including such patients would likely decrease
the differences between the oral anticoagulant and aspirin groups for cardiovascular
outcomes. The AFASAK 2 and SPAF 3 trials were the only ones in which patients
receiving aspirin also received low-dose warfarin in dosages of 1.25 mg/d
and a median of 2.0 mg/d, respectively, and with negligible prolongation of
the INR. To ensure that adding low-dose warfarin to aspirin did not bias our
study, we repeated the analyses after excluding patients taking aspirin plus
low-dose warfarin.
Patient clinical features were collected by research coordinators and
physicians before the initiation of therapy
(see eTable 1). These included previous stroke or transient ischemic attack,
hypertension, congestive heart failure, diabetes, and coronary artery disease.
All studies classified patients as hypertensive if they were taking medications
given to lower blood pressure. In the AFASAK studies, patients were considered
to have congestive heart failure if it was graded as moderate or severe. We
used the Atrial Fibrillation Investigators risk stratification to classify
patient stroke risk.18 Using these data, we
labeled the patients with hypertension, diabetes, or prior cerebral ischemia
as high risk. Patients without these risk factors
were considered low risk if they were younger than
65 years. All others were considered moderate risk.
We compared the relative efficacy of oral anticoagulant and aspirin
for 6 outcomes: ischemic or hemorrhagic stroke; ischemic stroke alone; hemorrhagic
stroke alone (including subarachnoid and subdural hemorrhage); aggregate cardiovascular
events (including ischemic stroke, myocardial infarction, systemic embolism,
or cardiovascular death); major bleeding (including intracranial and systemic
bleeding); and all-cause death. Cardiovascular deaths included those due to
stroke, myocardial infarction, congestive heart failure, pulmonary embolism,
or systemic embolism. Ischemic stroke and major bleeds were classified as
lethal if death occurred within 30 days of the event. (For outcome criteria
for each trial, see eTable 2.)
Patients were followed up for a mean of 1.9 years (Table 1). Patients were routinely seen every 3 to 6 months by study
investigators or when an outcome event was suspected. With the exception of
patients in AFASAK 1, a central events committee that was blinded to therapeutic
group reviewed and confirmed all events. Of all strokes, 97% were assessed
by neuroimaging, almost exclusively by computed tomography.
Crude event rates were calculated as events per 100 person-years of
observation and compared using the normal approximation of the Poisson distribution
as described by Rothman and Greenland.47 Kaplan-Meier
curves were generated to compare time to each outcome, and the log-rank statistic
determined the significance of differences between the curves. We used proportional
hazards regression modeling to estimate the relative effect of oral anticoagulant
on all 6 outcomes. Each model was stratified by study, thereby allowing different
hazard functions to be estimated for each study and then a pooled hazard ratio
(HR) across all studies was estimated. Such modeling helps adjust for factors
that might be unique to a particular study.48,49 We
tested the proportional hazards assumption for all Cox models using an interaction
term between the predictor variable and time.49 If
the P value for this term was less than .05, we concluded
that the assumption of proportional hazards was not met over time.
We assessed heterogeneity between studies for each outcome by visually
comparing HRs for each study with the overall estimate and by calculating
the DerSimonian and Laird Q statistic for heterogeneity.50 To measure interactions between treatment group and
baseline factors, appropriate interaction terms were included in the model.
All analyses were conducted using SAS 8.1 software (Cary, NC).
Table 1 describes the patients
according to the study in which the patients were enrolled. Atrial fibrillation
was nonparoxysmal in 83% and was present for more than a year in 67% of patients.
Sixty-five percent of patients were classified as having high, 27% as having
moderate, and 8% as having low risk of stroke. Reflecting different patient
selection criteria between studies, the prevalence of patient characteristics
varied between trials (Table 1).
After pooling patients from all 6 trials, patients receiving oral anticoagulant
and aspirin were well matched (Table 2).
Compared with aspirin, oral anticoagulant significantly decreased the
rate of all stroke, ischemic stroke, and cardiovascular events (Table 3). The decrease in the rate of all stroke was due to a large
decrease in ischemic stroke with only a small absolute increase in hemorrhagic
stroke. Of all ischemic strokes, 12.5% were lethal. Oral anticoagulants also
decreased the lethal ischemic stroke rate (0.5 vs 0.2 events per 100 patient-years; P = .01). The significant decrease in cardiovascular events
seen in the oral anticoagulant group was due primarily to decreased rates
of ischemic stroke and myocardial infarction.
The use of oral anticoagulant significantly increased the rate of major
bleeding (Table 3), with 15.3%
of all major bleeding episodes being lethal. Hemorrhagic stroke accounted
for 21.9% of all major bleeding and 52.4% of fatal bleeding events. Patients
receiving oral anticoagulant had a nonsignificantly increased rate of lethal
hemorrhages. Overall mortality did not differ between patient groups.
Life-table analyses (Figure 1)
confirmed the crude-rate comparisons and demonstrated, with the exception
of all-cause death, a consistent relative beneficial effect of oral anticoagulant
over time. All-cause death was the only outcome for which the assumption of
proportionality was not satisfied. This is reflected in the survival plots
for death showing a difference between treatment groups starting 2½
years into observation (Figure 1).
The outcomes for each trial are presented in Figure 2. The relative effect of oral anticoagulant for each outcome
was similar across studies with the exception of major bleeds. For major bleeding,
EAFT and SPAF 3 provided particularly high and low HRs, respectively. However,
the DerSimonian and Laird Q statistic for heterogeneity50 in treatment effect was not significant for any outcome,
including major bleeding.
To determine whether our conclusions changed without patients assigned
to receive aspirin plus low-dose warfarin, we repeated the analysis after
excluding such patients. This analysis excluded SPAF 3 and the combined aspirin-low-dose
warfarin group of AFASAK 2, resulting in 2837 patients. We noted similar results
for all stroke, 2.4 events among those taking oral anticoagulant vs 3.8 taking
aspirin per 100 patient-years (HR, 0.65; 95% CI, 0.49-0.86; P = .003); ischemic stroke, 2.0 vs 3.7 events per 100 patient years
(HR, 0.56; 95% CI, 0.41-0.76; P<.001), cardiovascular
events, 5.3 vs 6.9 events per 100 patient-years (HR, 0.76; 95% CI, 0.62-0.93; P = .02); and death, 4.7 vs 5.1 (HR, 0.92; 95% CI, 0.74-1.15; P = .41). There was a slight increase of the difference
between the 2 groups in hemorrhagic stroke, 0.5 vs 0.2 events per 100 patient-years
(HR, 2.26; 95% CI, 0.93-5.50; P = .06) and major
bleeding, 2.2 vs 1.2 events per 100 patient-years (HR, 1.93; 95% CI, 1.30-2.88; P = .001). The HRs excluding those who received combined
aspirin-low-dose warfarin are similar to HRs attained with the entire cohort
(Table 3). When the patients taking
aspirin and low-dose warfarin were excluded, cardiovascular outcome rates
did not change extensively in the oral anticoagulant group but were noticeably
lower in the aspirin group. This is likely due to the exclusion of high-risk
patients enrolled in the SPAF 3 study.
Effect of Treatment in Patient Subgroups
We explored the interaction of treatment effect with relevant patient
features on ischemic stroke and major bleeding (Figure 3). In almost all patient subgroups, oral anticoagulant was
significantly more efficacious than aspirin for reducing the risk of ischemic
stroke. Although we noted some variability in the relative benefit of oral
anticoagulant between patient subgroups, most of these interactions were not
significant with P>.10. There were 2 possible exceptions.
The relative benefit of oral anticoagulant vs aspirin in ischemic stroke prevention
appeared greater for patients younger than 75 years vs those who were 75 years
or older (P for interaction = .08) and for women
vs men (P for interaction = .04). The increased risk
of major bleeding for patients taking oral anticoagulants appeared consistent
in all patient subgroups and none of the interaction terms had a P<.10.
The absolute rate reduction in ischemic stroke by oral anticoagulant
vs aspirin varied with stroke risk (Figure
3). The absolute risk reduction in ischemic stroke achieved with
oral anticoagulant was greater in patients at highest baseline risk of stroke.
Patients with a previous history of stroke or transient ischemic attack had
an absolute risk reduction of 6.0% per year (number needed to treat [NNT],
17) while those patients without previous cerebrovascular disease had an absolute
risk reduction of 1.2% per year (NNT, 83). Patients who were classified as
being at high risk of ischemic stroke by the AF investigators' criteria had
an absolute risk reduction of 3.3% per year (NNT, 30), whereas those at low
risk of ischemic stroke only had an absolute risk reduction of 0.4% per year
(NNT not significant). In contrast to ischemic stroke, the absolute risk increase
of major bleeding with oral anticoagulant vs aspirin was more consistent between
patient subgroups.
Compared with aspirin, oral anticoagulant significantly decreased the
risk of all strokes, ischemic strokes, and cardiovascular events in patients
with nonvalvular paroxysmal or chronic AF. The absolute risk reductions of
these events were not substantially offset by small but statistically significant
increases in major hemorrhage. In these patients, the overall risk of stroke
or cardiovascular events was substantially higher than that of major bleeding.
These results support the use of oral anticoagulant as first-line preventive
therapy for high-risk patients with nonvalvular AF. Among trial participants,
the absolute rate reduction for all stroke (2.1 events per 100 patient-years),
ischemic stroke (2.3 events per 100-patient-years), and cerebrovascular events
(2.2 events per 100 patient-years) was considerable. Since the absolute rate
increase of major bleeding with oral anticoagulant is 0.9 events per 100 patient-years,
treating 1000 AF patients for 1 year with oral anticoagulant rather than aspirin
would prevent 23 ischemic strokes while causing 9 additional major bleeding
episodes.
We believe that our results both highlight and clarify the trade-off
between potential harms and benefits of oral anticoagulant prophylaxis for
AF patients. Compared with aspirin, oral anticoagulant decreased the risk
of ischemic stroke but increased the risk of major bleeding. We also found
that the overall absolute risk of ischemic stroke was approximately twice
that for major bleeding. Several studies show that patients with AF perceive
stroke as a much more serious outcome than bleeding. Hing et al51 also
found that patients had a strong aversion to strokes and would take warfarin
if it reduced the absolute risk of stroke by 1% annually. The AF patients
that Gage et al52 interviewed gave a median
utility value of 0.03 for moderate strokes and 0.0 for major stroke, with
death having a utility value of 0.52 In a study
by Solomon et al,53 patients equated the utility
of severe stroke with that of death. Devereaux et al54 found
that the average patient would take warfarin if it reduced the absolute risk
of stroke by at least 0.9 events per year but would only stop the drug if
it increased the risk of bleeding by 8.7 events per year. These findings show
that patients fear the consequences of having a stroke more than they fear
the consequences of being prone to severe bleeding, which supports the use
of oral anticoagulant treatment for such patients.
However, our results also highlight the importance of determining baseline
risk to identify AF patients most likely to benefit from oral anticoagulant.
There were some notable variations in the absolute difference of stroke rates
(Figure 3). For patients at higher
risk of ischemic stroke, such as the elderly or those with previous cerebrovascular
disease, the absolute risk reduction in ischemic stroke with oral anticoagulant
supercedes the associated bleeding risk. In contrast, patients at low risk
of ischemic stroke have only a small absolute risk reduction in ischemic stroke.
Therefore, proper risk stratification is essential to identify patients who
have the best chance of benefiting from oral anticoagulant. This is especially
true for patients without prior cerebrovascular disease for whom the absolute
benefit of oral anticoagulant is less (Figure
3). Several risk stratification schemes have been published18,43,55,56 and
further research is required to better refine our estimates of stroke risk
in AF patients.
Three meta-analyses have compared oral anticoagulant with antiplatelet
therapy for patients with AF.20,31,45 These
analyses have summarized reported aggregate rates of trial outcome events.
Our analysis used individual patient data, which allows a more complete and
accurate comparison of outcome risk between oral anticoagulant and aspirin
for patients with AF. Individual patient data permitted full survival analyses
of a variety of relevant outcome events, not just ischemic stroke. We were
able to provide absolute as well as relative comparisons of person-year rates
of outcomes. Finally, analysis of pooled individual patient data allowed the
exploration of treatment effects in relevant patient subgroups, such as those
with paroxysmal AF. Because we pooled individual patient data, we had the
opportunity to study 672 patients with paroxysmal AF. It is noteworthy that
these patients had a similar reduction in stroke risk with oral anticoagulant
(Figure 3).
Our results differ most from the meta-analysis by Taylor et al,30 which concluded that "the evidence for current clinical
practice in long-term anticoagulation for patients with nonrheumatic atrial
fibrillation is not strong." In addition to the methodological differences
between the studies, other factors explain why our findings were considerably
different. Taylor et al excluded the EAFT study5 and
the high-risk component of the SPAF 3 study,9 both
of which showed considerable advantage of oral anticoagulant over aspirin
for all strokes, ischemic strokes, and cardiovascular events (Figure 2). The EAFT study was excluded from the analysis by Taylor
et al because data were not presented to allow direct comparison of patients
randomized to oral anticoagulant and aspirin.57 Our
access to individual patient data avoided this problem. The SPAF 3 study was
excluded because patients taking low-dose warfarin were included with those
taking aspirin. We believe that it was preferable to include these patients
in our analysis because including patients taking low-dose warfarin in an
analysis with those taking aspirin patients would, if anything, decrease differences
in cardiovascular events. Excluding these patients from the analysis did not
alter our results. The only study that randomized patients to receive aspirin
or low-intensity oral anticoagulant with aspirin found no significant differences
in outcomes when low-dose warfarin was added to aspirin for patients with
AF.3 Finally, their meta-analysis dichotomized
outcomes into exclusive fatal and nonfatal events. This greatly decreases
their statistical power to identify differences between treatment groups.
A potential limitation of our analyses stems from their uncertain applicability
to AF patients outside of clinical trials. When compared with other patients,
those enrolled in clinical trials tend to be healthier and more compliant.
In addition, study investigators might more effectively monitor oral anticoagulant
therapy than usual clinical caregivers. As a consequence, both bleeding and
stroke rates might be higher among patients anticoagulated in general clinical
practice. However, a recently published systematic review of anticoagulated
AF patients in actual clinical practice found stroke and bleeding rates that
were very similar to those found in randomized trials,58 supporting
the external validity of our conclusions.
In summary, oral anticoagulant is more effective than aspirin in decreasing
the risk of stroke and other cardiovascular events in patients with nonvalvular
AF. Although oral anticoagulant also increases the risk of major bleeding,
the frequency of strokes and the gravity of their consequences exceed that
of therapy-associated bleeding events in most patients with AF. As a result,
oral anticoagulant should be the preferred prophylactic treatment for patients
with AF at a significant risk of thromboembolism.
1.Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham
Study.
Stroke.1991;22:983-988.Google Scholar 2.Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention
of thromboembolic complications in chronic atrial fibrillation: the Copenhagen
AFASAK study.
Lancet.1989;1:175-179.Google Scholar 3.Gullov AL, Koefoed BG, Petersen P.
et al. Fixed minidose warfarin and aspirin alone and in combination vs adjusted-dose
warfarin for stroke prevention in atrial fibrillation: Second Copenhagen Atrial
Fibrillation, Aspirin, and Anticoagulation Study.
Arch Intern Med.1998;158:1513-1521.Google Scholar 4.Hellemons BS, Langenberg M, Lodder J.
et al. Primary prevention of arterial thromboembolism in non-rheumatic atrial
fibrillation in primary care: randomised controlled trial comparing two intensities
of coumarin with aspirin.
BMJ.1999;319:958-964.Google Scholar 5.EAFT (European Atrial Fibrillation Trial) Study Group. Secondary prevention in non-rheumatic atrial fibrillation after transient
ischaemic attack or minor stroke.
Lancet.1993;342:1255-1262.Google Scholar 6. Stroke Prevention in Atrial Fibrillation Study: final results.
Circulation.1991;84:527-539.Google Scholar 7. Warfarin vs aspirin for prevention of thromboembolism in atrial fibrillation:
Stroke Prevention in Atrial Fibrillation II Study.
Lancet.1994;343:687-691.Google Scholar 8.The SPAF III Writing Committee for the Stroke Prevention in Atrial
Fibrillation Investigators. Patients with nonvalvular atrial fibrillation at low risk of stroke
during treatment with aspirin: Stroke Prevention in Atrial Fibrillation III
Study.
JAMA.1998;279:1273-1277.Google Scholar 9. Adjusted-dose warfarin vs 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.Google Scholar 10.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.Google Scholar 11.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.Google Scholar 12.Ezekowitz MD, Bridgers SL, James KE.
et al. 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.Google Scholar 13.Diener H, Cunha L, Forbes C.
et al. European stroke prevention study 2: dipyridamole and acetylsalicylic
acid in the prevention of stroke.
J Neurol Sci.1996;143:1-13.Google Scholar 14.Morocutti C, Amabile G, Fattapposta F.
et al. for the SIFA (Studio Italiano Fibrillazione Atriale) Investigators. Indobufen vs warfarin in the secondary prevention of major vascular
events in nonrheumatic atrial fibrillation.
Stroke.1997;28:1015-1021.Google Scholar 15.Posada IS, Barriales V.for the LASAF Pilot Study Group. Alternate-day dosing of aspirin in atrial fibrillation.
Am Heart J.1999;138:137-143.Google Scholar 16.Pengo V, Zasso A, Barbero F.
et al. Effectiveness of fixed minidose warfarin in the prevention of thromboembolism
and vascular death in nonrheumatic atrial fibrillation.
Am J Cardiol.1998;82:433-437.Google Scholar 17.The Atrial Fibrillation Investigators. The efficacy of aspirin in patients with atrial fibrillation: analysis
of pooled data from 3 randomized trials.
Arch Intern Med.1997;157:1237-1240.Google Scholar 18.Atrial Fibrillation Investigators. Risk factors for stroke and efficacy of antithrombotic therapy in atrial
fibrillation: analysis of pooled data from five randomized controlled trials.
Arch Intern Med.1994;154:1449-1457.Google Scholar 19. Echocardiographic predictors of stroke in patients with atrial fibrillation:
a prospective study of 1066 patients from 3 clinical trials.
Arch Intern Med.1998;158:1316-1320.Google Scholar 20.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.Google Scholar 21.Benavente O, Hart R, Koudstaal P, Laupacis A, McBride R. Oral anticoagulants for preventing stroke in patients with nonvalvular
atrial fibrillation and no previous history of stroke or transient ischemic
attacks.
Cochrane Database Syst Rev.2000;(2):CD001927.Google Scholar 22.Antithrombotic Trialists' Collaboration. Collaborative meta-analysis of randomised trials of antiplatelet therapy
for prevention of death, myocardial infarction, and stroke in high risk patients.
BMJ.2002;324:71-86.Google Scholar 23.Ezekowitz MD, Levine JA. Preventing stroke in patients with atrial fibrillation.
JAMA.1999;281:1830-1835.Google Scholar 24.Green CJ, Hadorn DC, Bassett K, Kazanjian A. Anticoagulation in chronic nonvalvular atrial fibrillation: a critical
appraisal and meta-analysis.
Can J Cardiol.1997;13:811-815.Google Scholar 25.Hart RG, Halperin JL. Atrial fibrillation and stroke: concepts and controversies.
Stroke.2001;32:803-808.Google Scholar 26.Howard PA, Duncan PW. Primary stroke prevention in nonvalvular atrial fibrillation: implementing
the clinical trial findings.
Ann Pharmacother.1997;31:1187-1196.Google Scholar 27.Koudstaal PJ. Anticoagulants vs antiplatelet therapy for preventing stroke in patients
with nonrheumatic atrial fibrillation and a history of stroke or transient
ischemic attacks.
Cochrane Database Syst Rev.2000;(2):CD000187.Google Scholar 28.Koudstaal PJ. Anticoagulants for preventing stroke in patients with nonrheumatic
atrial fibrillation and a history of stroke or transient ischemic attacks.
Cochrane Database Syst Rev.2000;(2):CD000185.Google Scholar 29.Singer DE. Overview of the randomized trials to prevent stroke in atrial fibrillation.
Ann Epidemiol.1993;3:563-567.Google Scholar 30.Taylor FC, Cohen H, Ebrahim S. Systematic review of long term anticoagulation or antiplatelet treatment
in patients with non-rheumatic atrial fibrillation.
BMJ.2001;322:321-326.Google Scholar 31.Segal JB, McNamara RL, Miller MR.
et al. Anticoagulants or antiplatelet therapy for non-rheumatic atrial fibrillation
and flutter.
Cochrane Database Syst Rev.2001;(1):CD001938.Google Scholar 32. Practice parameter: Stroke prevention in patients with nonvalvular
atrial fibrillation. Report of the Quality Standards Subcommittee of the American
Academy of Neurology.
Neurology.1998;51:671-673.Google Scholar 33.Albers GW, Atwood JE, Hirsh J, Sherman DG, Hughes RA, Connolly SJ. Stroke prevention in nonvalvular atrial fibrillation.
Ann Intern Med.1991;115:727-736.Google Scholar 34.Albers GW, Sherman DG, Gress DR, Paulseth JE, Petersen P. Stroke prevention in nonvalvular atrial fibrillation: a review of prospective
randomized trials.
Ann Neurol.1991;30:511-518.Google Scholar 35.Albers GW. Atrial fibrillation and stroke: three new studies, three remaining
questions.
Arch Intern Med.1994;154:1443-1448.Google Scholar 36.Albers GW. Choice of antithrombotic therapy for stroke prevention in atrial fibrillation:
warfarin, aspirin, or both?
Arch Intern Med.1998;158:1487-1491.Google Scholar 37.Dunn M, Alexander J, de Silva R, Hildner F. Antithrombotic therapy in atrial fibrillation.
Chest.1989;95(2 suppl):118S-127S.Google Scholar 38.Fuster V, Ryden LE, Asinger RW.
et al. ACC/AHA/ESC guidelines for the management of patients with atrial fibrillation:
executive summary. A Report of the American College of Cardiology/American
Heart Association Task Force on Practice Guidelines and the European Society
of Cardiology Committee for Practice Guidelines and Policy Conferences (Committee
to Develop Guidelines for the Management of Patients With Atrial Fibrillation):
developed in Collaboration With the North American Society of Pacing and Electrophysiology.
J Am Coll Cardiol.2001;38:1231-1266.Google Scholar 39.Gage BF, Fihn SD, White RH. Warfarin therapy for an octogenarian who has atrial fibrillation.
Ann Intern Med.2001;134:465-474.Google Scholar 40.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.Google Scholar 41.Hart RG, Sherman DG, Easton JD, Cairns JA. Prevention of stroke in patients with nonvalvular atrial fibrillation.
Neurology.1998;51:674-681.Google Scholar 42.Kitchens JM, Flegel KM. Atrial fibrillation, stroke and anticoagulation: what is to be done?
J Gen Intern Med.1986;1:126-129.Google Scholar 43.Albers GW, Dalen JE, Laupacis A, Manning WJ, Petersen P, Singer DE. Antithrombotic therapy in atrial fibrillation.
Chest.2001;119(suppl 1):194S-206S.Google Scholar 44.The Stroke Prevention in Atrial Fibrillation Investigators. Bleeding during antithrombotic therapy in patients with atrial fibrillation.
Arch Intern Med.1996;156:409-416.Google Scholar 45.Taylor FC, Ramsay ME, Renton A, Cohen H. Methods for managing the increased workload in anticoagulant clinics.
BMJ.1996;312:286.Google Scholar 46.Sudlow M, Thomson R, Thwaites B, Rodgers H, Kenny RA. Prevalence of atrial fibrillation and eligibility for anticoagulation
in the community.
Lancet.1998;352:1167-1171.Google Scholar 47.Greenland S, Rothman KJ. Introduction to categorical statistics. In: Rothman KJ, Greenland S, eds. Modern Epidemiology. Philadelphia, Pa: Lippincott-Raven Publishers; 1998:253-280.
48.Harrell FE. Cox proportional hazards regression model. In: Regression Modelling Strategies. New
York, NY: Springer; 2001:465-507.
49.Allison PD. Estimating Cox-regression models with PROC PHREG. In: Survival Analysis Using the SAS System.
Cary, NC: SAS Institute Inc; 2000:111-184.
50.Takkouche B, Cadarso-Suarez C, Spiegelman D. Evaluation of old and new tests of heterogeneity in epidemiologic meta-analysis.
Am J Epidemiol.1999;150:206-215.Google Scholar 51.Man-Son-Hing M, Laupacis A, O'Connor A.
et al. Warfarin for atrial fibrillation: the patient's perspective.
Arch Intern Med.1996;156:1841-1848.Google Scholar 52.Gage BF, Cardinalli AB, Owens DK. The effect of stroke and stroke prophylaxis with aspirin or warfarin
on quality of life.
Arch Intern Med.1996;156:1829-1836.Google Scholar 53.Solomon NA, Glick HA, Russo CJ, Lee J, Schulman KA. Patient preferences for stroke outcomes.
Stroke.1994;25:1721-1725.Google Scholar 54.Devereaux PJ, Anderson DR, Gardner MJ.
et al. Differences between perspectives of physicians and patients on anticoagulation
in patients with atrial fibrillation: observational study.
BMJ.2001;323:1218-1222.Google Scholar 55.Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke:
results from the national registry of atrial fibrillation.
JAMA.2001;285:2864-2870.Google Scholar 56.Stroke Prevention in Atrial Fibrillation Investigators. Risk factors for thromboembolism during aspirin therapy in patients
with atrial fibrillation: the stroke prevention in atrial fibrillation study.
J Stroke Cerebrovasc Dis.1995;5:147-157.Google Scholar 57.Ebrahim S, Taylor F. Review: anticoagulants may be better than antiplatelet agents for nonfatal
stroke but not other vascular or fatal events in nonrheumatic AF.
ACP J Club.2001;135:A17.Google Scholar 58.Evans A, Kalra L. Are the results of randomized controlled trials on anticoagulation
in patients with atrial fibrillation generalizable to clinical practice?
Arch Intern Med.2001;161:1443-1447.Google Scholar