JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 67, NO.

25, 2016

2016 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00

PUBLISHED BY ELSEVIER http://dx.doi.org/10.1016/j.jacc.2016.03.581

ORIGINAL INVESTIGATIONS

Aspirin Instead of Oral Anticoagulant

Prescription in Atrial Fibrillation
Patients at Risk for Stroke
Jonathan C. Hsu, MD, MAS,a Thomas M. Maddox, MD, MSC,b Kevin Kennedy, MS,c David F. Katz, MD,d
Lucas N. Marzec, MD,d Steven A. Lubitz, MD, MPH,e Anil K. Gehi, MD,f Mintu P. Turakhia, MD, MAS,g
Gregory M. Marcus, MD, MASh

ABSTRACT

BACKGROUND Oral anticoagulation (OAC), rather than aspirin, is recommended in patients with atrial brillation (AF)
at moderate to high risk of stroke.

OBJECTIVES This study sought to examine patient and practice-level factors associated with prescription of aspirin
alone compared with OAC in AF patients at intermediate to high stroke risk in real-world cardiology practices.

METHODS The authors identied 2 cohorts of outpatients with AF and intermediate to high thromboembolic risk
(CHADS2 score $2 and CHA2DS2-VASc $2) enrolled in the American College of Cardiology PINNACLE (Practice Innovation
and Clinical Excellence) registry between 2008 and 2012. Using hierarchical modied Poisson regression models adjusted
for patient and practice characteristics, the authors examined the prevalence and predictors of aspirin alone versus OAC
prescription in AF patients at risk for stroke.

RESULTS Of 210,380 identied patients with CHADS2 score $2 on antithrombotic therapy, 80,371 (38.2%) were
treated with aspirin alone, and 130,009 (61.8%) were treated with warfarin or non-vitamin K antagonist OACs. In the
cohort of 294,642 patients with CHA2DS2-VASc $2, 118,398 (40.2%) were treated with aspirin alone, and 176,244
(59.8%) were treated with warfarin or non-vitamin K antagonist OACs. After multivariable adjustment, hypertension,
dyslipidemia, coronary artery disease, prior myocardial infarction, unstable and stable angina, recent coronary artery
bypass graft, and peripheral arterial disease were associated with prescription of aspirin only, whereas male sex, higher
body mass index, prior stroke/transient ischemic attack, prior systemic embolism, and congestive heart failure were
associated with more frequent prescription of OAC.

CONCLUSIONS In a large, real-world cardiac outpatient population of AF patients with a moderate to high risk of
stroke, more than 1 in 3 were treated with aspirin alone without OAC. Specic patient characteristics predicted
prescription of aspirin therapy over OAC. (J Am Coll Cardiol 2016;67:291323) 2016 by the American College of
Cardiology Foundation.

Listen to this manuscripts

From the aCardiac Electrophysiology Section, Division of Cardiology, Department of Medicine, University of California, San Diego,
audio summary by
La Jolla, California; bVA Eastern Colorado Health Care System/University of Colorado School of Medicine, Denver, Colorado; cMid
JACC Editor-in-Chief
d e
America Heart Institute, Kansas City, Missouri; University of Colorado School of Medicine, Denver, Colorado; Cardiac
Dr. Valentin Fuster.
Arrhythmia Service and Cardiovascular Research Center, Massachusetts General Hospital, Boston, Massachusetts; fUniversity of
North Carolina, Chapel Hill, North Carolina; gVA Palo Alto Health Care System/Stanford University School of Medicine, Palo Alto,
h
California; and the Section of Cardiac Electrophysiology, Division of Cardiology, Department of Medicine, University of
California, San Francisco, San Francisco, California. This research was supported by the American College of Cardiology Foun-
dations National Cardiovascular Data Registry (NCDR). The views expressed in this paper represent those of the authors, and
do not necessarily represent the ofcial views of the NCDR or its associated professional societies identied at www.ncdr.com.
2914 Hsu et al. JACC VOL. 67, NO. 25, 2016

Aspirin Use in Anticoagulation-Eligible AF Patients JUNE 28, 2016:291323

A trial brillation (AF) is the most METHODS

ABBREVIATIONS
AND ACRONYMS common cardiac arrhythmia world-
wide and imparts signicant stroke
AF = atrial brillation
risk (1). Risk stratication schemes, including
BMI = body mass index
the CHADS 2 score (2) and the more recent
CABG = coronary artery bypass
CHA 2DS2-VASc score (3), have been devel-
graft
oped to estimate the risk of thromboembo-
CI = condence interval
lism in AF patients on the basis of specic
NCDR = National
risk factors (2,3). In patients with AF deter-
Cardiovascular Data Registry
mined to be at intermediate to high risk for
OAC = oral anticoagulation/
anticoagulant thromboembolism, anticoagulation
TIA = transient ischemic attack
warfarin (a vitamin K antagonist) or the
newer non-vitamin K antagonist oral antico-
agulant agents (OACs) reduces morbidity and mor-
tality (4,5). Previous studies have demonstrated that
aspirin therapy is not as benecial as OAC for reduc-
tion of thromboembolism (6), yet it is well known
that appropriate OAC prescription in AF patients at
risk for stroke outside clinical trial settings falls short
of guideline-based expectations (7). The extent to
which AF patients at risk for stroke are prescribed
only aspirin and the factors that determine such a
practice are not well known. Additionally, the preva-
lence of concomitant antiplatelet agent prescription
in contemporary patients with AF who are and are
not prescribed OAC has not been well studied.
SEE PAGE 2924
We sought to investigate the prevalence and pre-
dictors of treatment with aspirin only compared with
OAC therapy by cardiovascular specialists in a cohort
of outpatients with AF at a moderate to high risk for
thromboembolism using data from the National Car-
diovascular Data Registry (NCDR)s Practice Innova-
tion and Clinical Excellence (PINNACLE). Use of this
prospective national registry of cardiovascular care in
the United States provides a unique opportunity to
examine patterns of both aspirin and OAC prescrip-
tion in routine practice among outpatients with AF, as
well as the use of other antiplatelet agents (i.e.,
thienopyridine prescription).
DATA SOURCE. The NCDR PINNACLE registry was
created in 2008 by the American College of Cardiol-
ogy as the rst national, prospective, ofce-based
cardiac quality-improvement registry in the United
States (8). Participating academic and private
practices collect longitudinal, point-of-care data that
includes patient demographics, symptoms, comor-
bidities, vital signs, medications, laboratory values,
with and recent hospitalizations with either paper forms,
or modication of a practices electronic medical re-
cord using a standardized collection tool to compre-
hensively obtain and transmit uniform data. NCDR
registry data quality assurance is maintained through
standardized data collection and transmission pro-
tocols, rigorous data denitions, and periodic data
quality audits, which have shown much >90% raw
accuracy of data abstraction (9). Quality checks and
analyses of the data have been performed at St.
Lukes Mid America Heart Institute (Kansas City,
Missouri), the primary analytical center for the
PINNACLE registry.
STUDY POPULATION. There were a total of 1,147,227
patients enrolled into the PINNACLE registry between
January 1, 2008, and December 30, 2012. We excluded
patients without AF (n 689,722 excluded), patients
deemed not able to be prescribed aspirin or OAC
therapy (n 17,627 excluded), as assessed by the
treating clinician and specied on data collection
forms (due to medical, patient, or system reason), and
those not prescribed any antithrombotic therapy at all
(n 112,222 excluded). We further restricted the
cohort to patients known to be at moderate to high
risk for thromboembolism (CHADS 2 score $2, with
1 point for congestive heart failure, hypertension,
age $75 years, and diabetes, and 2 points for stroke/
transient ischemic attack [TIA] [2]; n 117,276
patients excluded). Exclusion criteria for our primary
cohort (referred to as the CHADS 2 score $2 cohort)
were on the basis of the American College of

PINNACLE Registry is an initiative of the American College of Cardiology Foundation. Bristol-Myers Squibb and Pzer Inc.
are Founding Sponsors of the PINNACLE Registry. Dr. Hsu has received honoraria from St. Jude Medical, Medtronic,
Biotronik, Janssen Pharmaceutical, and Bristol-Myers Squibb; has received research support from Biotronik and Biosense
Webster. Dr. Lubitz has received grants from the National Institutes of Health (NIH) and the Doris Duke Charitable
Foundation. Dr. Gehi has received speaker honoraria from Medtronic, Zoll Medical, St. Jude Medical, and Biotronik. Dr.
Turakhia has received research support from the Veterans Affairs, Gilead Sciences, iRhythm, Medtronic, Janssen Phar-
maceuticals, and SentreHeart; is a consultant to Janssen Pharmaceuticals, Medtronic, St. Jude Medical, and Precision
Health Economics; and has equity ownership in thryva and Zipline. Dr. Marcus has received research support from the NIH,
PCORI, Medtronic, Pzer, Rhythm Diagnostic Systems, and SentreHeart; and is a consultant for and has equity ownership
in InCarda. All other authors have reported that they have no relationships relevant to the contents of this paper to
disclose.

Manuscript received December 26, 2015; revised manuscript received March 22, 2016, accepted March 29, 2016.
JACC VOL. 67, NO. 25, 2016 Hsu et al. 2915
JUNE 28, 2016:291323 Aspirin Use in Anticoagulation-Eligible AF Patients

F I G U R E 1 Flowchart of PINNACLE Registry Patients Included in the Analyses

Total Patients in the NCDR

PINNACLE Registry
(N=1,147,227)

Excluded (N=689,722):
Patients without AF

Patients with AF
(N=457,505)

Excluded (N=17,627):
Contraindication to aspirin or oral
anticoagulant therapy

Excluded (N=112,222):
Not prescribed aspirin or oral
anticoagulant therapy

Analytic Cohort of AF
Patients
(N=327,656)

Excluded (N=117,276): Excluded (N=33,014):

CHADS2 Score 1 CHA2DS2-VASc Score 1

Final Analytic Cohort of Final Analytic Cohort of

CHADS2 Score 2 CHA2DS2-VASc Score 2
AF Patients AF Patients
(N=210,380) (N=294,642)

The owchart depicts the total cohort of patients in the PINNACLE registry from which exclusion criteria were administered to arrive at both the CHADS2 $2
and the CHA2DS2-VASc $2 nal AF cohorts for analyses. AF atrial brillation; NCDR National Cardiovascular Data Registry; PINNACLE Practice
Innovation and Clinical Excellence.

Cardiology/American Heart Association/European
Society of Cardiology guidelines in place during the
study timeframe, which dened as a Class I indication
(i.e., recommendation that the treatment is useful/
effective) the use of OAC instead of aspirin alone for
the primary prevention of stroke in at-risk AF pa-
tients with a CHADS2 score $2 (10). Therefore, our
nal primary study cohort was composed of 210,380
patients with AF from 123 practices in 38 states across
the United States (Figure 1). Of the 123 practices
included in the PINNACLE registry, 97 (78.9%) prac-
tices were located in an urban/suburban setting.
The regional breakdown of practice locations
included the following: 27 (22%) in the West; 17
(13.8%) in the Northeast; 33 (26.8%) in the Midwest;
and 46 (37.4%) in the South. We also conducted a
secondary analysis, using the CHA2DS2 -VASc score, a
more sensitive tool to risk-stratify AF patients who
may be at risk for stroke and benet from anticoag-
ulant therapy (3); use of this risk score may have
inuenced cardiovascular specialist prescription of
OAC during the study timeframe, as reected in
updated European Society of Cardiology guidelines
from 2012 (11) and subsequently published updated
guidelines after the study timeframe (12). These
updated guidelines, published in 2014, advise the use
of the CHA2 DS2-VASc score for the assessment of
stroke risk, and state as a Class I recommendation the
use of OAC therapy for patients with nonvalvular AF
and a CHA2DS 2-VASc score $2 (12). In addition to
carrying over all of the same exclusion criteria listed
previously, this cohort was restricted to AF patients
with a CHA2DS2 -VASc score $2, with 1 point for
congestive heart failure, hypertension, age $65 years
[2 points if age $75 years], diabetes, female sex, and
coronary or peripheral arterial disease, and 2 points
for stroke/TIA. Selection of this CHA 2DS2-VASc
score $2 cohort led to inclusion of an additional
84,262 patients, for a total cohort of 294,642 patients.
Because antiplatelet therapy may be indicated for
reasons other than AF among those with cardiovas-
cular disease, we also stratied analyses in both the
CHADS2 score $2 and CHA 2 DS2-VASc score $2 cohorts
by those patients with and without a coronary heart
disease risk equivalent condition, which was consid-
ered a diagnosis of any of the following: coronary
2916 Hsu et al.
Aspirin Use in Anticoagulation-Eligible AF Patients
artery disease; unstable angina; stable angina; prior
myocardial infarction; prior coronary artery bypass
graft (CABG) surgery; or peripheral arterial disease.
To minimize over-representation by patients with
multiple visits, only data from each patients index
visit during the study period were used. The index
visit was considered the rst encounter where a
diagnosis of AF was specied. To ensure that
misclassication of OAC prescription was not over-
looked by examining only the index visit, we per-
formed a sensitivity analysis to determine the
number of patients that would be reclassied as being
prescribed an OAC by increasing the window from
baseline to within 1 year following the index visit.
STUDY OUTCOMES. Our primary study outcome was
treatment with aspirin therapy (alone or in combina-
tion with another antiplatelet agent) versus any U.S.
Food and Drug Administrationapproved OAC for
stroke prevention in patients with AF, which included
warfarin, dabigatran or rivaroxaban (apixaban and
edoxaban had not yet been approved by the U.S. Food
and Drug Administration during the study timeframe).
Aspirin therapy was dened as prescription of aspirin,
either alone or in combination with clopidogrel,
ticlopidine, prasugrel, and/or dipyridamole. Fre-
quencies of OAC prescription with aspirin and/or other
antiplatelet agents (i.e., thienopyridine or dipyr-
idamole), as well as aspirin plus antiplatelet agent
prescription were analyzed. Treatment with a thie-
nopyridine was dened as prescription of clopidogrel,
ticlopidine, or prasugrel. Patients treated
aspirin/dipyridamole combination only were included
in the aspirin group.
STATISTICAL ANALYSIS. Normally distributed contin-
uous variables are expressed as mean  SD, whereas
categorical variables are expressed as proportions.
Unadjusted differences were compared using the
chi-square test for categorical variables, and Student t
tests for continuous variables, as appropriate.
To investigate the independent associations of
various characteristics with the outcome of aspirin
alone versus OAC prescription, we constructed hier-
archical modied Poisson regression models adjusted
for patient and practice-level characteristics. These
models included site as a random effect to account for
patient clustering within sites. All covariates consid-
ered to be potential predictors were entered in the
multivariable model and included patient-level
characteristics (age, body mass index [BMI], sex,
congestive heart failure, diabetes mellitus, stroke or
TIA, systemic embolism, hypertension, dyslipidemia,
coronary artery disease, unstable angina, stable
angina, prior myocardial infarction, prior CABG
JACC VOL. 67, NO. 25, 2016
JUNE 28, 2016:291323
surgery, and peripheral arterial disease) and practice-
level characteristics (U.S. geographical region, urban
location, and clinic volume). Candidate covariates
selected for the multivariate analyses were chosen a
priori on the basis of plausibility that they could be
associated with differential prescription of aspirin
alone versus OAC. Additional analyses adjusted for
any thienopyridine use were performed to evaluate
associations of these predictors with prescription of
aspirin alone versus OAC, independent of the need
for thienopyridine prescription. The highest missing
rate for variables was BMI (28.7%); therefore, a
missing indicator was included in models that con-
tained this variable. Additionally, missing data were
assumed to be missing at random and were imputed
with 10 imputation sets, in which all patient variables
were used to inform the imputation model (13).
Because the rate of OAC prescription exceeded 10%,
we used modied Poisson regression models at all
steps to estimate relative prescription rates with 95%
condence intervals (CIs) directly (instead of odds
ratios obtained from logistic regression, which may
overestimate effect differences) (14). Statistical tests
were 2-sided and considered signicant if they yiel-
ded a p value <0.05. Analyses were performed using
the SAS statistical package, version 9.3 (SAS Institute,
Cary, North Carolina), R version 2.15.3 (Foundation
for Statistical Computing, Vienna, Austria), and IVE-
Ware (Institute for Social Research, University of
Michigan, Ann Arbor, Michigan).
with
RESULTS
Among the 210,380 patients with a CHADS2 score $2,
there were 80,371 (38.2%) treated with aspirin alone
and 130,009 (61.8%) treated with warfarin or non-
vitamin K antagonist OACs. Of the 294,642 patients
with CHA 2DS2-VASc $2, 118,398 (40.2%) were treated
with aspirin alone, and 176,244 (59.8%) were treated
with warfarin or non-vitamin K antagonist OACs.
Patient and practice characteristics among treated
AF patients in each cohort stratied by prescription of
aspirin alone versus OAC are shown in Table 1. In both
the CHADS2 score $2 cohort and the CHA 2DS2-VASc
score $2 cohort, AF patients prescribed aspirin alone
instead of OAC were younger, had lower BMI, were
more often female, and were more likely to have
diabetes mellitus, hypertension, dyslipidemia, coro-
nary artery disease, unstable angina, stable angina,
prior myocardial infarction, prior CABG surgery, and
peripheral artery disease. Compared with those who
were prescribed OAC, patients in both cohorts who
were prescribed aspirin alone were more often
treated in the South and West, in nonurban practices,
JACC VOL. 67, NO. 25, 2016 Hsu et al. 2917
JUNE 28, 2016:291323 Aspirin Use in Anticoagulation-Eligible AF Patients

T A B L E 1 Baseline Characteristics of AF Patients at Intermediate to High Risk of Stroke as Categorized by CHADS 2 Score $2 and
CHA 2 DS 2 -VASc Score $2, Stratied by Prescription of Aspirin Versus OAC

CHADS 2 Score $2 Cohort CHA2DS2-VASc Score $2 Cohort

(N 210,380) (N 294,642)

Prescribed Aspirin Prescribed OAC Prescribed Aspirin Prescribed OAC

(n 80,371) (n 130,009) p Value (N 118,398) (N 176,244) p Value

Patient characteristics
Age, yrs 75.8  11.5 76.1  10.0 <0.001 73.2  11.6 74.6  10.0 <0.001
BMI, kg/m2 29.2  6.1 30.0  6.4 <0.001 29.2  6.1 29.9  6.4 <0.001
Male 53.2 54.6 <0.001 52.8 54.6 <0.001
Congestive heart failure 36.2 39.7 <0.001 25.6 30.6 <0.001
Diabetes mellitus 37.6 33.9 <0.001 26.6 25.7 <0.001
Stroke or TIA 22.3 21.9 0.046 15.1 16.2 <0.001
Systemic embolism 1.0 1.6 <0.001 0.8 1.4 <0.001
Hypertension 93.7 93.3 <0.001 84.6 84.2 0.001
Dyslipidemia 67.7 63.4 <0.001 64.2 59.8 <0.001
Coronary artery disease 64.8 54.5 <0.001 62.2 52.4 <0.001
Unstable angina 1.7 0.9 <0.001 1.5 0.8 <0.001
Stable angina 10.5 5.8 <0.001 9.0 5.2 <0.001
Prior myocardial infarction 23.2 20.2 <0.001 22.4 19.5 <0.001
Prior CABG surgery 13.2 10.1 <0.001 11.8 9.2 <0.001
Peripheral arterial disease 13.9 9.3 <0.001 12.1 8.4 <0.001
Practice characteristics
Region <0.001 <0.001
Northeast 14.3 14.7 13.6 14.8
Midwest 28.2 36.8 26.5 34.7
South 38.9 31.6 42.1 33.0
West 18.5 16.8 17.9 17.5
Urban practice location 88.3 88.9 <0.001 88.0 88.8 <0.001
Clinic volume (mean visits/yr) 37,658.5  27,990.8 36,662.3  27,089.6 <0.001 37,193.3  27,721.1 36,807.6  27,238.7 <0.001

Values are mean  SD or %.

AF atrial brillation; BMI body mass index; CABG coronary artery bypass graft; OAC oral anticoagulant.

and at practices with a higher volume of patients.
Patients with AF treated with aspirin or OAC therapy
differed from patients with AF excluded from the
analyses with respect to several baseline character-
istics (Online Table 1).
In the CHADS 2 score $2 cohort, the mean CHADS 2
score was 2.8  1.0. Of the 130,009 patients (61.8%)
who were treated with OACs, warfarin was the most
commonly used therapy (n 118,178 [90.9%]), fol-
lowed by dabigatran (n 9,363 [7.2%]) and rivarox-
aban (n 2,468 [1.9%]). In the CHA 2DS2-VASc
score $2 cohort, the mean CHA 2DS2-VASc score was
2.2  1.2. Similarly, of the 176,244 patients (59.8%)
who were treated with OACs, warfarin was the most
commonly used therapy (n 159,668 [90.6%]), fol-
lowed by dabigatran (n 13,122 [7.4%]), and rivar-
oxaban (n 3,454 [2.0%]). In a sensitivity analysis
evaluating reclassication of patients prescribed an
OAC within 1 year of the index visit, a small propor-
tion of patients (1.9%, n 2,507 of 135,056) not
prescribed an anticoagulant agent at baseline were
prescribed OAC in follow-up. Evaluation of practice-
level variation of OAC prescription revealed that the
median practice rate for OAC prescription in the
CHADS2 score $2 cohort was 64.9%, with a signicant
variation in OAC prescription, as evidenced by
an interquartile range of 56.4% to 69.9%. The median
practice rate for OAC prescription in the CHA 2 DS2-
VASc score $2 cohort was 63.3%, with variation
represented by an interquartile range of 55.2% to
68.7% (Figures 2A and 2B).
In both the CHADS2 score $2 cohort and the
CHA2DS2-VASc score $2 cohort, concomitant use of
any thienopyridine (mainly clopidogrel) was higher
in those patients prescribed aspirin. Concomitant
prescription of aspirin alone as well as aspirin
thienopyridine was not uncommon in patients who
were prescribed an OAC (Table 2). Combination dual
antiplatelet therapy with aspirin and any thienopyr-
idine was much less common in patients prescribed
OAC than aspirin, in the range of 5.7% to 6.0% in both
the CHADS2 score $2 and CHA2DS 2-VASc score $2
cohorts. Among patients treated with OAC, concom-
itant antiplatelet therapy was more often prescribed
in patients prescribed a non-vitamin K antagonist
OAC compared with warfarin (Online Table 2).
2918 Hsu et al. JACC VOL. 67, NO. 25, 2016

Aspirin Use in Anticoagulation-Eligible AF Patients JUNE 28, 2016:291323

use was as high as 31.2% in patients also prescribed an

F I G U R E 2 Variation in OAC Prescription Prevalence Across Practices
OAC. Thienopyridine use was mostly reserved for
patients with a coronary heart disease risk equivalent
CHADS2 Score 2 Cohort
A 116 diagnosis (Table 2).
111
After multivariable adjustment, hypertension,
106
101 dyslipidemia, coronary artery disease, prior myocar-
96
91 dial infarction, unstable and stable angina, recent
86 CABG, and peripheral arterial disease were associated
81
76 with more frequent prescription of aspirin only,
71 whereas male sex, higher BMI, prior stroke/TIA, prior
66
Practice

61 systemic embolism, and congestive heart failure were

56
51
associated with more frequent prescription of OAC in
46 both the CHADS2 score $2 and CHA 2DS2-VASc
41
36 score $2 cohorts (Central Illustration). The associa-
31 tions of all covariates included in the adjusted
26
21 multivariable models evaluating predictors of aspirin
16
versus OAC prescription are summarized in Online
11
6 Table 3. In subanalyses including any thienopyr-
1
0 10 20 30 40 50 60 70 80 90 100 idine use in the multivariable model, thienopyridine
Proportion of Patients in Practice Prescribed an Oral Anticoagulant prescription was associated with more frequent pre-
scription of aspirin compared with OAC in both the
CHA2DS2-VASc Score 2 Cohort
B 121 CHADS2 score $2 (relative prescription rate: 1.61; 95%
116 CI: 1.58 to 1.61; p < 0.001) and CHA2 DS2-VASc score $2
111
106 cohorts (relative prescription rate: 1.55; 95% CI: 1.52
101
96 to 1.57; p < 0.001) (Online Figure 1).
91
86
81 DISCUSSION
76
71
Practice

66 In a large, quality-improvement registry of >200,000

61
56 outpatients with AF at intermediate to high
51
46 stroke risk, as measured by both CHADS2 and
41
CHA 2DS2-VASc scores $2, approximately 40% were
36
31 treated with aspirin alone instead of OAC prescrip-
26
21 tion. Coronary atherosclerosis-related comorbidities,
16
11
including hypertension, dyslipidemia, coronary ar-
6 tery disease, prior myocardial infarction, unstable
1
0 10 20 30 40 50 60 70 80 90 100 and stable angina, recent CABG, and peripheral
Proportion of Patients in Practice Prescribed an Oral Anticoagulant arterial disease were associated with more frequent
prescription of aspirin only, whereas male sex,
Variation of oral anticoagulant prescription within both the CHADS2 score $2 (A), and higher BMI, prior stroke/TIA, prior systemic embo-
CHA2DS2-VASc score $2 (B) cohorts. The proportion of patients with AF prescribed an oral lism, and congestive heart failure were associated
anticoagulant (OAC) within each practice is shown. Abbreviations as in Figure 1.
with more frequent prescription of OAC. Our ndings
have important implications for AF patients, partic-
ularly as the benet of stroke risk reduction with
The majority of patients in both the CHADS2 OAC over aspirin prescription increases as annual
score $2 cohort (n 127,188 [60.5%]) and CHA 2DS2- stroke risk increases, as measured by both the
VASc score $2 cohort (n 171,401 [58.2%]) had a CHADS2 and CHA 2DS 2-VASc scores (2,3), the latter of
coronary heart disease equivalent condition. In both which includes coronary atherosclerosis and its
the CHADS 2 score $2 and CHA 2DS2-VASc score $2 equivalent disease processes. Because many of the
cohorts, when stratied by the presence of a coronary predictors of aspirin use alone include conditions
heart disease risk equivalent diagnosis, aspirin use that may warrant aspirin therapy regardless of the
was more common in patients prescribed OAC, but presence of AF, much of the underutilization of
did not top 55.2%. Even in patients without a coro- appropriate anticoagulant agents may be driven
nary heart disease risk equivalent diagnosis, aspirin by either the perception that aspirin by itself is
JACC VOL. 67, NO. 25, 2016 Hsu et al. 2919
JUNE 28, 2016:291323 Aspirin Use in Anticoagulation-Eligible AF Patients

T A B L E 2 Prevalence of Solo and Combination Antiplatelet Prescription in AF Patients at Intermediate to High Risk of Stroke, Stratied by
Prescription of Aspirin Versus OAC in the Overall CHADS 2 Score $2 and CHA 2 DS 2 -VASc Score $2 Cohorts and Subcohorts of Patients With
and Without Any Coronary Heart Disease Risk Equivalent

CHADS 2 Score $2 Cohort: CHADS2 Score $2 Cohort:

CHADS2 Score $2 Coronary Heart Disease No Coronary Heart
Overall Cohort Risk Equivalent Disease Risk Equivalent
(N 210,380) (N 127,188) (N 83,192)

Prescribed Prescribed Prescribed Prescribed Prescribed Prescribed

Aspirin OAC Aspirin OAC Aspirin OAC
Antithrombotic Therapy (n 80,371) (n 130,009) p Value (n 53,630) (n 73,558) p Value (n 26,741) (n 56,451) p Value

Aspirin 100.0 44.8 <0.001 100.0 55.2 <0.001 100.0 31.2 <0.001
Any thienopyridine 20.6 8.4 <0.001 27.5 12.9 <0.001 6.6 2.2 <0.001
Clopidogrel 20.1 8.1 <0.001 26.9 12.7 <0.001 6.6 2.1 <0.001
Ticlopidine 0.1 0.1 <0.001 0.2 0.1 <0.001 0.0 0.0 0.539
Prasugrel 0.4 0.2 <0.001 0.5 0.3 <0.001 0.0 0.0 0.047
Aspirin any thienopyridine 20.6 6.0 <0.001 27.5 9.6 <0.001 6.6 1.3 <0.001
Dipyridamole/aspirin 1.2 0.3 <0.001 1.2 0.4 <0.001 1.2 0.2 <0.001

CHA2DS 2-VASc Score $2 Cohort: CHA2DS2-VASc Score $2 Cohort:

CHA2DS2-VASc Score $2 Coronary Heart Disease No Coronary Heart Disease
Overall Cohort Risk Equivalent Risk Equivalent
(N 294,642) (N 171,401) (N 83,192)

Prescribed Prescribed Prescribed Prescribed Prescribed Prescribed

Aspirin OAC Aspirin OAC Aspirin OAC
Antithrombotic Therapy (n 118,398) (n 176,244) p Value (n 75,647) (n 95,754) p Value (n 42,751) (n 80,490) p Value

Aspirin 100.0 43.8 <0.001 100.0 55.1 <0.001 100.0 31.2 <0.001
Any thienopyridine 19.2 7.8 <0.001 27.0 12.7 <0.001 5.4 2.2 <0.001
Clopidogrel 18.7 7.6 <0.001 26.2 12.4 <0.001 5.3 2.1 <0.001
Ticlopidine 0.1 0.1 <0.001 0.2 0.1 <0.001 0.0 0.0 0.384
Prasugrel 0.5 0.2 <0.001 0.7 0.3 <0.001 0.0 0.0 0.219
Aspirin any thienopyridine 19.2 5.7 <0.001 27.0 9.5 <0.001 5.4 1.3 <0.001
Dipyridamole/aspirin 0.9 0.2 <0.001 0.9 0.3 <0.001 0.9 0.2 <0.001

Values are %. A patient was considered to have a coronary heart disease risk equivalent with any of the following diagnoses: coronary artery disease; unstable angina; stable
angina; prior myocardial infarction; prior CABG surgery; or peripheral arterial disease. OAC therapy was dened as prescription of either warfarin, dabigatran, or rivaroxaban.
Aspirin therapy was dened as prescription of either aspirin alone or in combination with clopidogrel, ticlopidine, prasugrel, and/or dipyridamole. Treatment with a thieno-
pyridine was dened as prescription of clopidogrel, ticlopidine, or prasugrel.
Abbreviations as in Table 1.

sufcient or that the risk of aspirin plus anti-
coagulation is not worth the benet.
Despite a well-established association of AF with
stroke, signicant lack of OAC prescription to reduce
thromboembolism in at-risk candidates has been
demonstrated in several large-scale studies, with
aspirin prescription prevalence as high as 28% to 38%
in patients who were prescribed an antithrombotic
agent at all (15,16). These previous studies described
primarily U.S. patients in the era of warfarin therapy,
and were before the promulgation of the importance
of stroke risk scores to aid in risk stratication. Clin-
ical risk scores, including the CHADS 2 and CHA2DS2 -
VASc scores, have been developed to elucidate and
quantify stroke risk in AF patients to aid in the deci-
sion to prescribe antithrombotic therapies (2,3). In
our study, we used the CHADS 2 score, because this
risk scheme was the predominant scoring system
contemporary with the study period. In order to
expand upon the robustness of our ndings, we also
studied the same cohort of AF patients as assessed by
the CHA2DS2 -VASc score, because this risk scheme
may improve discrimination of AF patients at risk for
stroke and thromboembolism and is supported by
updated guidelines published near the end of and
after the study timeframe (11,12). In the GARFIELD
registry (Global Anticoagulant Registry in the FIELD),
a multinational, observational study of 10,614 AF
patients enrolled between 2009 and 2011 at 540 sites
in 19 non-U.S. countries and cared for by general
practitioners, cardiologists, and neurologists, 25.3%
of patients with a CHADS 2 score $2 were prescribed
aspirin instead of OAC, but the study did not evaluate
predictors of aspirin prescription (17). Because 38.2%
of AF patients with a CHADS2 score $2 were pre-
scribed aspirin alone in our study, there is clearly a
continued lack of a guideline-adhering prescription
of OAC in U.S. cardiovascular specialist practices. The
assertion that aspirin does not adequately protect
against thromboembolism in AF patients has been
extensively studied, highlighted by a meta-analysis
of 4,052 patients with AF who experienced nearly
2920 Hsu et al. JACC VOL. 67, NO. 25, 2016

Aspirin Use in Anticoagulation-Eligible AF Patients JUNE 28, 2016:291323

C E NT R A L IL L U ST R A T I O N Predictors of Aspirin Versus OAC Therapy Prescription in AF Patients at Intermediate to

High Stroke Risk Among Both CHADS 2 Score $2 and CHA 2 DS 2 -VASc Score $2 Cohorts

Hsu, J.C. et al. J Am Coll Cardiol. 2016;67(25):291323.

Shown are characteristics associated with aspirin treatment after multivariable adjustment in AF patients with intermediate to high stroke risk, as dened by CHADS2
score $2 (A), and CHA2DS2-VASc score $2 (B). Error bars denote 95% condence intervals. AF atrial brillation; BMI body mass index; CABG coronary artery
bypass graft; CI condence interval; OAC oral anticoagulant; RR relative prescription rate; TIA transient ischemic attack.
JACC VOL. 67, NO. 25, 2016
JUNE 28, 2016:291323
one-half the stroke risk when treated with warfarin
rather than aspirin (hazard ratio: 0.55; 95% CI: 0.43 to
0.71) (18). Although inadequate knowledge regarding
these data may explain the guideline nonadherence
among cardiology specialists, other reasons,
including justiable clinical circumstances, concerns
for bleeding in specic patients, or patient refusal of
this therapy, may be responsible.
Patients with risk factors for coronary athero-
sclerosis and myocardial infarction are often treated
with aspirin as primary prevention therapy, and
those with previous acute coronary syndrome are
treated with aspirin as secondary prevention ther-
apy, as per clinical guidelines (19). In our large
cohort of AF patients at risk for stroke, risk factors
for coronary atherosclerosis, including hypertension
and dyslipidemia, predicted aspirin over OAC ther-
apy prescription. Additionally, characteristics asso-
ciated with coronary atherosclerosis, including
coronary artery disease itself, prior myocardial
infarction, unstable and stable angina, recent CABG,
and peripheral arterial disease, also were more
often predictive of aspirin prescription. It is possible
that cardiovascular specialists may be more inclined
to treatment with aspirin therapy in such patients,
despite concomitant indications for anticoagulation
due to AF. This may reect the misperception that
aspirin exhibits adequate efcacy compared with
OAC, or may be driven by concerns for bleeding risk
when using combination therapy (20). That prior
stroke/TIA, prior systemic embolism, and congestive
heart failure were associated with more frequent
prescription of OAC suggests that these specic
components of both the CHADS2 and CHA2 DS2-VASc
scores may be weighed more by the prescribing
clinician in a decision to choose appropriate OAC
over aspirin-only therapy. Interestingly, after
multivariable adjustment, men with AF in our study
had a 6% greater odds of OAC prescription versus
aspirin therapy, despite previous studies that
demonstrate an increased stroke risk in women (3).
This paradoxical practice pattern suggests that
female sex as an important stroke risk factor in AF
remains largely unappreciated. Previous data from
the PINNACLE registry suggest that a sex bias may
exist on the basis of lower prescription rates of OAC
in women treated in real-world U.S. cardiology
practices (21). Signicant variability of OAC pre-
scription across individual practices was observed
(as shown in Figure 2); this suggests that focusing
on factors at the practice level, and not simply at
the individual physician level, may prove important
in future efforts to rectify proper prescription of
OAC in AF patients at risk for stroke.
Hsu et al. 2921
Aspirin Use in Anticoagulation-Eligible AF Patients
A signicant number of patients prescribed OAC for
AF were also prescribed antiplatelet therapy. Dual
antiplatelet therapy was prescribed in 19.2% to 20.6%
of the patients who were prescribed aspirin alone.
Among those receiving OAC, aspirin was prescribed in
just over one-half of all patients with coronary heart
disease equivalent conditions and one-third of pa-
tients without any coronary heart disease equivalent
conditions. It is also important to emphasize that a
substantial proportion of AF patients without a car-
diovascular risk equivalent (approximately one-third)
were prescribed both an OAC and aspirin, placing
those patients at higher risk for bleeding, without any
evidence of benet. This category of patients likely
represents low-hanging fruit for changing practice
(presumably via prescriber education) in a fashion
that would mitigate risk for bleeding, without
reducing the efcacy of thromboembolism preven-
tion (22). Triple therapy, including an anticoagulant
agent and 2 antiplatelet drugs, was prescribed in
approximately 6%. Although guidelines recommend
aspirin therapy in all acute coronary syndrome pa-
tients who are able to take it (23), consensus guide-
lines for anticoagulant-indicated AF patients with
stable coronary heart disease and its equivalents are
lacking, with either little mention (24), or recom-
mendations to only continue antiplatelet therapy
for 1 year after an acute coronary syndrome. Our
nding that a large proportion of patients prescribed
OAC are also prescribed aspirin therapy suggests that
further studies evaluating cardiovascular outcomes in
AF patients with stable coronary heart disease pre-
scribed various antithrombotic strategies are needed.
Additionally, clarication and subsequent promulga-
tion of consensus guidelines on concomitant anti-
platelet and anticoagulation use appears warranted,
particularly because coronary heart disease risk
equivalents have now been incorporated into the
CHA 2DS2-VASc score as part of consensus AF guide-
lines (12).
STUDY LIMITATIONS. First, although the PINNACLE
registry does ascertain whether an anticoagulant
agent is contraindicated, the data are not sufciently
granular to calculate the HAS-BLED score (i.e., the
PINNACLE registry does not capture data on medica-
tion use predisposing to bleeding, labile international
normalized ratios, or alcohol or drug use) (25), nor
does it contain data on renal insufciency or aspirin
dose prescribed. This additional information may
have been helpful to quantify the bleeding risk among
those who received aspirin versus OAC prescription to
determine whether bleeding risk was associated with
prescription preference. However, because the net
clinical benet of warfarin actually increases with the
2922 Hsu et al. JACC VOL. 67, NO. 25, 2016

Aspirin Use in Anticoagulation-Eligible AF Patients JUNE 28, 2016:291323

HAS-BLED score (given that the score generally mir-
rors the stroke-risk scores), a higher bleeding risk
alone should not justify aspirin, rather than an anti-
coagulant agent. In fact, data support a net clinical
benet for OAC over aspirin therapy in AF patients
at risk for stroke, despite increased bleeding risks
(26). Second, the PINNACLE program enrolled pa-
tients from motivated cardiology practices dedicated
to quality improvement. Therefore, antithrombotic
therapy prescription patterns in other U.S. practices
may differ from those reported in this study, poten-
tially reducing the generalizability of our results.
Third, specic data are unavailable regarding previ-
ous bleeding complications or exact reasons for con-
traindications to anticoagulant therapy, and therefore
we cannot determine the validity of a reported
contraindication. Fourth, the main data analyzed
consisted of OAC prescription at the index AF visit.
Although the index visit may not capture OAC pre-
scription at any time in patient follow-up, a sensitivity
analysis that included follow-up visits out to 1 year
after the index visit demonstrated that only a small
proportion of additional patients (<2%) were pre-
scribed OAC. Fifth, the PINNACLE registry does not
contain follow-up data to ascertain outcomes, such as
subsequent stroke or bleeding, and does not contain
follow-up laboratory data to assess therapeutic anti-
coagulation status with warfarin. Future studies
should focus on these outcomes as they relate to
aspirin versus OAC prescription, as well as effects of
prescription of additional antiplatelet agents. Finally,
some may argue that the PINNACLE data collection
form may not reect actual prescription of the drug,
much less what patients actual receive or consume;
however, that distinction is arguably minimally rele-
vant for the purposes of this study, because the
data recorded on the form likely more purely reects
the intent or perceived correct prescription of
medications. Nevertheless, we cannot rule out the
possibility that, despite best efforts for accurate data
collection, underreporting of
occurred by those recording this information.
CONCLUSIONS
Despite a lack of evidence for aspirins use for
thromboembolic prophylaxis in patients with AF at
moderate to high risk of stroke, and clear benet with
the use of OACs, nearly 1 in 3 of these patients in this
large, quality-improvement registry of cardiology
outpatients with AF at intermediate to high risk of
stroke were prescribed aspirin alone. Several specic
patient characteristics predicted prescription of
aspirin therapy over OAC therapy, particularly
comorbidities related to coronary atherosclerosis and
its risk equivalent diseases. These data indicate a gap
in care, most prominent in patients with or at risk for
coronary artery disease, and should draw attention to
a high rate of prescription of aspirin therapy in AF
patients at risk for stroke, despite previous data that
show aspirin to be inferior to OAC in this population.
The specic patient characteristics associated with
this practice, including those related to coronary ar-
tery disease, highlight opportunities to improve
appropriate prescription of OAC in AF, including
identifying knowledge gaps that might be informed
by future studies.
REPRINT REQUESTS AND CORRESPONDENCE: Dr.
Jonathan C. Hsu, Cardiac Electrophysiology Section,
Division of Cardiology, Department of Medicine,
University of California-San Diego, 9444 Medical
Center Drive, MC7411, La Jolla, California 92037.
E-mail: Jonathan.Hsu@ucsd.edu.
PERSPECTIVES
COMPETENCY IN PATIENT CARE: In a cardiac
outpatient population of patients with AF at
moderate to high risk of stroke, a relatively high
proportion of patients were treated with aspirin
rather than anticoagulant agents, and this was
associated with conditions related to coronary
heart disease.
TRANSLATIONAL OUTLOOK: Future studies
OAC prescription should investigate the impact of the availability of
non-vitamin K antagonist OACs on the choice of
antithrombotic agent for patients with AF and the
role of performance improvement programs on
adherence to guideline-directed management.

REFERENCES

1. Go AS, Hylek EM, Phillips KA, et al. Prevalence of
diagnosed atrial brillation in adults: national impli-
cations for rhythm management and stroke preven-
tion: the AnTicoagulation and Risk Factors in Atrial
Fibrillation (ATRIA) study. JAMA 2001;285:23705.
2. Gage BF, Waterman AD, Shannon W, et al.
Validation of clinical classication schemes for
predicting stroke: results from the National Reg-
istry of Atrial Fibrillation. JAMA 2001;285:
286470.
3. Lip GYH, Nieuwlaat R, Pisters R, et al. Rening
clinical risk stratication for predicting stroke and
thromboembolism in atrial brillation using a novel
risk factor-based approach: the Euro Heart Survey
on Atrial Fibrillation. Chest 2010;137:26372.
JACC VOL. 67, NO. 25, 2016
JUNE 28, 2016:291323
4. Hart RG, Pearce LA, Aguilar MI. Meta-analysis:
antithrombotic therapy to prevent stroke in pa-
tients who have nonvalvular atrial brillation. Ann
Intern Med 2007;146:85767.
5. Connolly SJ, Ezekowitz MD, Yusuf S, et al.,
RE-LY Steering Committee and Investigators.
Dabigatran versus warfarin in patients with
atrial brillation. N Engl J Med 2009;361:113951.
6. Aguilar MI, Hart R, Pearce LA. Oral anticoagu-
lants versus antiplatelet therapy for preventing
stroke in patients with non-valvular atrial brilla-
tion and no history of stroke or transient ischemic
attacks. Cochrane Database Syst Rev 2007;(3).
CD006186.
7. Chan PS, Maddox TM, Tang F, et al. Practice-
level variation in warfarin use among outpatients
with atrial brillation (from the NCDR PINNACLE
program). Am J Cardiol 2011;108:113640.
8. Chan PS, Oetgen WJ, Buchanan D, et al. Cardiac
performance measure compliance in outpatients:
the American College of Cardiology and National
Cardiovascular Data Registrys PINNACLE (Practice
Innovation And Clinical Excellence) program. J Am
Coll Cardiol 2010;56:814.
9. Messenger JC, Ho KKL, Young CH, et al., NCDR
Science and Quality Oversight Committee Data
Quality Workgroup. The National Cardiovascular
Data Registry (NCDR) Data Quality Brief: the NCDR
Data Quality Program in 2012. J Am Coll Cardiol
2012;60:14848.
10. Fuster V, Rydn LE, Cannom DS, et al. ACC/
AHA/ESC 2006 guidelines for the management of
patients with atrial brillationexecutive sum-
mary: a report of the American College of Cardi-
ology/American Heart Association Task Force on
Practice Guidelines and the European Society of
Cardiology Committee for Practice Guidelines
(Writing Committee to Revise the 2001 Guidelines
for the Management of Patients With Atrial
Fibrillation). J Am Coll Cardiol 2006;48:854906.
11. Camm AJ, Lip GYH, De Caterina R, et al., ESC
Committee for Practice Guidelines (CPG). 2012
focused update of the ESC guidelines for the
management of atrial brillation. Eur Heart J
2012;33:271947.
12. January CT, Wann LS, Alpert JS, et al. 2014
AHA/ACC/HRS guideline for the management
of patients with atrial brillation: executive
summary: a report of the American College of
Cardiology/American Heart Association Task Force
on Practice Guidelines and the Heart Rhythm So-
ciety. J Am Coll Cardiol 2014;64:224680.
13. Raghunathan TE, Solenberger PW, Van
Hoewyk J. IVEware: Imputation and Variance Esti-
mation Software. Ann Arbor, MI: Survey Method-
ology Program, Survey Research Center, Institute
for Social Research, University of Michigan, 2002.
14. Zou G. A modied Poisson regression approach
to prospective studies with binary data. Am J
Epidemiol 2004;159:7026.
15. Go AS, Hylek EM, Borowsky LH, et al. Warfarin
use among ambulatory patients with nonvalvular
atrial brillation: the anticoagulation and risk
factors in atrial brillation (ATRIA) study. Ann
Intern Med 1999;131:92734.
16. Glazer NL, Dublin S, Smith NL, et al. Newly
detected atrial brillation and compliance with
antithrombotic guidelines. Arch Intern Med 2007;
167:24652.
17. Kakkar AK, Mueller I, Bassand JP, et al., GARFIELD
Registry Investigators. Risk proles and antith-
rombotic treatment of patients newly diagnosed
with atrial brillation at risk of stroke: perspectives
from the international, observational, prospective
GARFIELD registry. PLoS One 2013;8. e63479.
18. van Walraven C, Hart RG, Singer DE, et al. Oral
anticoagulants vs aspirin in nonvalvular atrial
brillation: an individual patient meta-analysis.
JAMA 2002;288:24418.
19. Vandvik PO, Lincoff AM, Gore JM, et al. Pri-
mary and secondary prevention of cardiovascular
disease: Antithrombotic Therapy and Prevention
of Thrombosis, 9th ed: American College of Chest
Physicians Evidence-Based Clinical Practice
Guidelines. Chest 2012;141 Suppl:e637S68S.
20. Hansen ML, Srensen R, Clausen MT, et al.
Risk of bleeding with single, dual, or triple therapy
with warfarin, aspirin, and clopidogrel in patients
with atrial brillation. Arch Intern Med 2010;170:
143341.
Hsu et al. 2923
Aspirin Use in Anticoagulation-Eligible AF Patients
21. Thompson LE, Maddox TM, Bradley SM, et al.
Abstract 14594: gender differences in use of anti-
coagulant for atrial brillation: a report from the
NCDR (abstr). Circulation 2015;132:A14594.
22. Steinberg BA, Kim S, Piccini JP, et al., ORBIT-
AF Investigators and Patients. Use and associ-
ated risks of concomitant aspirin therapy with
oral anticoagulation in patients with atrial
brillation: insights from the Outcomes Registry
for Better Informed Treatment of Atrial Fibrilla-
tion (ORBIT-AF) registry. Circulation 2013;128:
7218.
23. OGara PT, Kushner FG, Ascheim DD, et al.
2013 ACCF/AHA guideline for the management of
ST-elevation myocardial infarction: a report of
the American College of Cardiology Foundation/
American Heart Association Task Force on Prac-
tice Guidelines. J Am Coll Cardiol 2013;61:
e78140.
24. You JJ, Singer DE, Howard PA, et al. Antith-
rombotic therapy for atrial brillation: Antith-
rombotic Therapy and Prevention of Thrombosis,
9th ed: American College of Chest Physicians
Evidence-Based Clinical Practice Guidelines. Chest
2012;141 Suppl:e531S75S.
25. Pisters R, Lane DA, Nieuwlaat R, et al. A novel
user-friendly score (HAS-BLED) to assess 1-year
risk of major bleeding in patients with atrial
brillation: the Euro Heart Survey. Chest 2010;
138:1093100.
26. Olesen JB, Lip GYH, Lindhardsen J, et al. Risks
of thromboembolism and bleeding with throm-
boprophylaxis in patients with atrial brillation: a
net clinical benet analysis using a real world
nationwide cohort study. Thromb Haemost 2011;
106:73949.
KEY WORDS aspirin, CHADS 2 score,
CHA 2 DS 2 -VASc score, thromboembolism
A PP END IX For a supplemental gure and
tables, please see the online version of this
article.