CLINICAL RESEARCH STUDY

Absence of Oral Anticoagulation and Subsequent

Outcomes Among Outpatients with Atrial
Fibrillation
Paul L. Hess, MD, MHS,a,b Sunghee Kim, PhD,c Gregg C. Fonarow, MD,d Laine Thomas, PhD,c Daniel E. Singer, MD,e
James V. Freeman, MD, MPH,f Bernard J. Gersh, MB, ChB, DPhil,g Jack Ansell, MD,h Peter R. Kowey, MD,i
Kenneth W. Mahaffey, MD,j Paul S. Chan, MD, MSc,k,l Benjamin A. Steinberg, MD, MHS,c,m Eric D. Peterson, MD, MPH,c
Jonathan P. Piccini, MD, MHS,c on behalf of the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation
(ORBIT-AF) Patients and Investigators
a
Cardiology Section, VA Eastern Colorado and Health Care System, Denver; bDepartment of Medicine, University of Colorado Anschutz
Medical Campus, School of Medicine, Aurora; cDuke Clinical Research Institute, Durham, NC; dDepartment of Medicine, University of
California, Los Angeles; eHarvard Medical School and Massachusetts General Hospital, Boston; fDepartment of Medicine, Yale University
School of Medicine, New Haven, Conn; gDepartment of Medicine, Mayo Clinic College of Medicine, Rochester, Minn; hDepartment of
Medicine, New York School of Medicine, Lenox Hill Hospital; iLankenau Institute for Medical Research, Wynnewood, Penn; jDepartment of
Medicine, Stanford University School of Medicine, Palo Alto, Calif; kDepartment of Cardiovascular Research, St. Luke’s Mid America Heart
Institute, Kansas City, Mo; lDepartment of Medicine, University of Missouri-Kansas City; mUniversity of Utah, Salt Lake City.

ABSTRACT

BACKGROUND: Prior studies have shown a treatment gap in oral anticoagulation (OAC) use among patients
with atrial fibrillation yet have incompletely characterized factors associated with failure to treat and
subsequent outcomes in contemporary practice.
METHODS: Using data collected between June 2010 and August 2011 from 174 ambulatory care sites in the
Outcomes Registry for Better Informed Treatment of Atrial Fibrillation, we identified factors associated
with absence of OAC via stratified logistic regression. Using weighted Cox regression, we assessed the
association between OAC non-use and subsequent outcomes over 2.5 years.
RESULTS: Among 9553 patients, 2202 (23.0%) were not on OAC. Among OAC nonrecipients, 1846 (83.8%)
had a CHA2DS2-VASc score
2. Factors independently associated with OAC non-use included atrial
fibrillation type (paroxysmal odds ratio [OR] 0.73, 95% confidence interval [CI] 0.54-0.99; persistent OR
0.14, 95% CI 0.10-0.21; permanent OR 0.35, 95% CI 0.25-0.49; reference ¼ new-onset), left atrial diameter
enlargement (mild OR 0.80, 95% CI 0.66-0.97; moderate 0.58, 95% CI 0.47-0.73; severe 0.53, 95% CI
0.42-0.68; reference ¼ normal diameter), and age >80 years (OR 1.04, 95% CI 1.02-1.08). Untreated
patients had a higher risk of death (adjusted hazard ratio [HR] 1.22, 95% CI 1.05-1.41), a lower bleeding risk
(adjusted HR 0.35, 95% CI 0.15-0.81), and a nonsignificant trend toward higher risk of stroke/non-central
nervous system embolism/transient ischemic attack than those treated (adjusted HR 1.18, 95% CI 0.91-1.54).
CONCLUSIONS: A majority of atrial fibrillation patients not treated with an OAC in current community
practice meet guideline indications for treatment. Atrial fibrillation burden, chronicity, and comorbidity are
associated with nontreatment. Untreated patients are at increased risk for adverse outcomes.
Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).  The American Journal of Medicine (2017) 130, 449-456

KEYWORDS: Atrial fibrillation; Oral anticoagulation; Outcomes; Quality of care

Funding: See last page of article. Requests for reprints should be addressed to Paul L. Hess, MD, MHS,
Conflict of Interest: See last page of article. Denver VA Medical Center, Cardiology Section (111B), 1055 Clermont
Authorship: See last page of article. Street, Denver, CO 80220.
E-mail address: paul.hess@ucdenver.edu

0002-9343/Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.amjmed.2016.11.001
450 The American Journal of Medicine, Vol 130, No 4, April 2017

Oral anticoagulation (OAC) prevents stroke and improves Those missing data on OAC status at baseline (n ¼ 1) or
all-cause survival in patients with atrial fibrillation. In his- without any follow-up (n ¼ 329) were also excluded. In an
torical clinical trials, vitamin K antagonists reduced the risk analysis dedicated to patients with unequivocal American
of thromboembolism by 64% and all-cause death by 26%, Heart Association/American College of Cardiology/Heart
with an acceptable increase in bleeding risk compared with Rhythm Society guideline indications for OAC,8 we excluded
1
no treatment. Trials of non-vitamin K OACs have patients with a CHA2DS2-VASc (Congestive heart failure;
demonstrated efficacy and safety Hypertension; Age
75 years;
equivalent or superior to vitamin Diabetes mellitus; prior Stroke,
K antagonism.2-5 CLINICAL SIGNIFICANCE TIA, or thromboembolism;
Clinical guidelines recommend  In ORBIT-AF, a national, ongoing regis- Vascular disease; Age 65-74 years;
that atrial fibrillation patients who try of outpatients with atrial fibrillation, Sex category) score <2 (n ¼ 930).
are at moderate to high risk of stroke To identify factors associated with
a majority of atrial fibrillation patients
without a known contraindication non-use of OAC, sites with >95%
not treated with oral anticoagulation
should be treated with OAC. 6-8
OAC use and/or <20 patients were
Despite demonstrated efficacy and meet guideline indications for excluded (n ¼ 2564). For the pur-
professional guideline recommen- treatment. pose of the outcomes assessment
dations, OAC treatment rates are  Atrial fibrillation burden, chronicity, and among patients with a CHA2DS2-
generally below 60%9 and have comorbidity are associated with VASc score <2, sites with >95%
remained low over time.10 Howev- nontreatment. OAC use and/or <20 patients were
er, factors underlying absence of included.
OAC therapy and associated out-  Untreated patients are at elevated risk
comes in the current therapeutic era for death.
are incompletely understood.  Future quality improvement initiatives Outcome Measures
Accordingly, the present analysis The condition of interest was
should emphasize appropriate risk
sought to identify factors associated OAC use at baseline. Outcomes of
with the absence of OAC therapy in
stratification and underscore the survival interest included all-cause death,
US clinical practice. Additionally, benefit of oral anticoagulation use. stroke or systemic embolism, and
we sought to describe outcomes in major bleeding. In ORBIT-AF,
patients who were not treated with patients were evaluated every 6
OAC in contemporary US outpatient practice. months, and the time and date of intervening cardiovascular
events were recorded, including but not limited to stroke,
bleeding events, and death. Stroke was defined as a new,
METHODS sudden, focal neurologic deficit that persisted beyond 24
Data Source hours and was not due to a readily identifiable, nonvascular
Outcomes Registry for Better Informed Treatment of Atrial cause (eg, seizure). All stroke and systemic embolism events
Fibrillation (ORBIT-AF) is a national registry of US out- were verified and adjudicated using source documentation.
patients with atrial fibrillation. The rationale, study design, Major bleeding was defined according to the International
data collection, and methods have been described previ- Society of Thrombosis and Haemostasis criteria.12 Interna-
ously.11 The primary dataset for the present analysis con- tional Society of Thrombosis and Haemostasis acute major
sisted of baseline data collected from 174 primary care, bleeding events were those that were fatal; occurred in a
cardiology, or electrophysiology sites between June 2010 critical area or organ such as intracranial, intraspinal,
and August 2011. Trained personnel abstracted data on intraocular, retroperitoneal, intra-articular, pericardial, or
eligible atrial fibrillation outpatients and submitted them to intramuscular with compartment syndrome; and/or led to a
the ORBIT-AF registry by means of a Web-enabled case fall in hemoglobin level of 2 g/L or more, leading to
report form. Data included demographic and clinical char- transfusion of 2 or more units of whole or red blood cells.12
acteristics, medical history, heart rhythm history, and
pharmacologic treatment, including OAC use. The Duke
Clinical Research Institute serves as the data and coordi-
Statistical Analysis
nating center for the registry. In the overall study population, we compared the baseline
characteristics of patients who did not receive OAC with
those of patients who did, using c2 tests for categorical
Study Population variables and Wilcoxon rank-sum tests for continuous var-
A total of 10,135 patients aged
18 years with electrocar- iables. Percentages for categorical variables and medians
diographically documented atrial fibrillation were enrolled in and interquartile ranges (IQRs) for continuous variables are
ORBIT-AF. For the present analysis, patients with an abso- reported. An analysis of only patients with guideline in-
lute contraindication to OAC use, including prior intracranial dications for OAC was performed among patients with a
hemorrhage, allergy, and pregnancy, were excluded (n ¼ 89). CHA2DS2-VASc score
2.
Hess et al Oral Anticoagulation and Outcomes in AF Outpatients
To identify factors associated with OAC non-use in the
total study population, we used stratified logistic regression
modeling after excluding sites with >95% OAC use and/or
sites with fewer than 20 patients. Covariates missing <14%
were imputed using multiple imputation by the Markov
chain Monte Carlo and regression methods. Estimates and
associated standard errors reflect the combined analysis over
5 imputed data sets. We then explored relationships between
demographic and clinical variables and OAC non-use using
multivariable logistic regression modeling. Continuous
variables were tested for linearity, and nonlinear relation-
ships were accounted for with the use of splines. Using
logistic regression with generalized estimating equations,
candidate baseline characteristics were backward selected
using an a level of 0.15. The final regression model was
then fitted using backward selection in stratified logistic
regression with an a level >0.05. An analogous analysis
was performed among patients with a CHA2DS2-VASc
score
2. A sensitivity analysis was performed among
patients with a CHADS2 (Congestive heart failure; Hyper-
tension; Age
75 years; Diabetes mellitus; prior Stroke,
TIA, or thromboembolism) score
2.
To assess the association of baseline OAC use with subse-
quent outcomes in guideline eligible/recommended patients
without contraindications, we constructed Cox regression
models for each outcome weighted for a patient’s inverse
propensity score of getting OAC for each outcome. Markov
chain Monte Carlo and regression methods were used. The
propensity score of predicting OAC use at baseline was
calculated from the logistic regression adjusted for all clinically
relevant covariates, as well as independent predictors of each
outcome identified using backward selection. Covariates with a
P value <.05 were maintained in the model. Linearity was
checked, and splines were used as needed. Models were
adjusted for demographic and clinical covariates.
P values <.05 were considered statistically significant.
Tests were 2-sided. Analyses were performed using SAS
software version 9.4 (SAS Institute, Cary, NC). The
ORBIT-AF registry was approved by the institutional re-
view board of the Duke University Health System and the
local institutional review board at each enrolling center.
Analyses were performed in aggregate using deidentified
data. PLH and JPP had full access to all of the data in the
study and take responsibility for the integrity of the data and
the accuracy of the data analysis. All authors read the final
manuscript and agree to it as written.
RESULTS
Overall, the median age of the cohort was 75 (IQR 67-82)
years, 57.5% were male, and 89.5% were white. Table 1
shows the baseline characteristics of the overall study
cohort stratified by OAC use. Compared with patients
receiving OAC, those not receiving OAC were younger
(73 [IQR 64-82] years vs 75 [IQR 68-82] years), more
frequently had new-onset (6.7% vs 3.8%) or paroxysmal
(65.5% vs 46.3%) atrial fibrillation, and less frequently had
451
persistent (12.9% vs 17.9%) or permanent (14.9% vs
32.0%) atrial fibrillation. They had slightly higher systolic
blood pressure (126 [IQR 116-138] mm Hg vs 125 [IQR
116-138] mm Hg). They more commonly had a prior
myocardial infarction (16.9% vs 15.7%) and less frequently
had a prior cerebrovascular event (10.9% vs 16.8%). They
had a lower burden of cardiovascular comorbidities and risk
factors such as heart failure (27.0% vs 34.5%) and advanced
heart failure symptoms (5.5% vs 8.5% had New York Heart
Association class III or IV), diabetes mellitus (26.1% vs
30.5%), hypertension (77.7% vs 84.8%), or a history of
valve replacement or repair (4.4% vs 9.5%). Patients
without OAC more commonly had preserved systolic
function (74.3% vs 70.0%) and less frequently had enlarged
left atrial diameters (13.5% vs 22.8% had severely enlarged
left atrial diameters). Among those not on OAC, 1332
(60.4%) had a CHADS2 score
2 and 1846 (83.8%) had a
CHA2DS2-VASc score
2. Similar between-group
differences were observed among patients with a
CHA2DS2-VASc score
2.
Table 2 shows factors independently associated with non-
use of OAC at baseline in the overall cohort (all
CHA2DS2DVASc scores included), including age >80 years
(odds ratio [OR] 1.04, 95% confidence interval [CI] 1.02-
1.08), type of atrial fibrillation (paroxysmal OR 0.73, 95% CI
0.54-0.99; persistent OR 0.14, 95% CI 0.10-0.21; permanent
OR 0.35, 95% CI 0.25-0.49; reference ¼ new-onset) and left
atrial diameter enlargement (mild OR 0.80, 95% CI 0.66-
0.97; moderate 0.58, 95% CI 0.47-0.73; severe 0.53, 95% CI
0.42-0.68; reference ¼ normal diameter). Systolic blood
pressure >120 mm Hg was not associated with OAC non-use.
Prior stroke or transient ischemic attack (OR 0.52, 95% CI
0.41-0.65), heart failure symptoms (New York Heart Asso-
ciation class II vs none OR 0.62, 95% CI 0.48-0.80), and left
ventricular dysfunction (severe dysfunction vs normal OR
0.51, 95% CI 0.31-0.83) were negatively associated with
OAC non-use. Most of the factors observed in the overall
population were also seen among patients with CHA2DS2-
VASc scores
2. A sensitivity analysis among patients with
CHADS2 score
2 (Supplementary Table 1, available online)
yielded similar results.
Table 3 shows outcomes associated with OAC non-
receipt among patients with a CHA2DS2-VASc score
2
after a median follow-up of 2.5 years. Rates of death, stroke
or non-central nervous system embolism, and stroke or non-
central nervous system embolism or transient ischemic
attack were numerically higher among patients not receiving
OAC. Bleeding rates were numerically lower. After
adjustment for the propensity to receive treatment, absence
of OAC was associated with a higher likelihood of death
(hazard ratio [HR] 1.22, 95% CI 1.05-1.41) and a nonsig-
nificant increase in rates of the composite outcomes of
stroke and non-central nervous system embolism (HR 1.14,
95% CI 0.79-1.64) and stroke, non-central nervous system
embolism, and transient ischemic attack (HR 1.18, 95% CI
0.91-1.54), as well as lower rates of bleeding (HR 0.35, 95%
CI 0.74-0.96).
452 The American Journal of Medicine, Vol 130, No 4, April 2017

Table 1 Patient Characteristics by Oral Anticoagulation Use*

Total Study Population Patients with CHA2DS2-VASc Score
2

No OAC OAC No OAC OAC

Characteristic (n ¼ 2202) (n ¼ 7351) P (n ¼ 1870) (n ¼ 6916) P
Age (y) 73 (64-82) 75 (68-82) <.0001 76 (68-83) 76 (69-82) .3254
Male 57.0 57.7 <.0001 51.8 55.0 .0122
Race .0838 .0374
White 89.2 89.6 89 89.7
Black 5.6 4.6 5.9 4.6
Hispanic 3.5 4.3 3.4 4.3
Other 1.5 1.4 1.4 1.3
Insurance
Private 30.8 23.5 <.0001 21.8 20.2 .0977
Medicare or Medicaid 63.9 72 73.3 75.6
Type of atrial fibrillation <.0001
New-onset 6.7 3.8 6.4 3.6 <.0001
Paroxysmal 65.5 46.3 63.9 45.9
Persistent 12.9 17.9 12.7 17.6
Permanent 14.9 32.0 17.0 33.0
Systolic blood pressure 126 (118-138) 125 (116-138) .0368 128 (118-140) 126 (116-138) .0004
(mm Hg)
Diastolic blood pressure 72 (68-80) 72 (66-80) .5207 72 (65-80) 72 (66-80) .6641
(mm Hg)
Heart rate (beats per 70 (62-78) 70 (63-80) .0027 70 (62-79) 70 (64-80) .0532
minute)
Weight (kg) 84 (71-100) 87 (73-103) <.0001 82 (69-98) 86 (72-102) <.0001
Cardiovascular comorbidities and
risk factors
Coronary artery disease 36.1 36.3 .8579 42.2 38.7 .0064
Prior myocardial 16.9 15.7 .1635 19.8 16.8 .0024
infarction
Diabetes mellitus 26.1 30.5 <.0001 30.7 32.6 .1281
Heart failure 27.0 34.5 <.0001 31.8 36.6 .0008
New York Heart <.0001 .0008
Association class
No CHF 73.0 65.4 68.2 63.4
I 8.5 10.8 10.2 11.5
II 9.5 17.2 13.9 17.0
III/IV 5.5 8.5 7.5 8.1
Hyperlipidemia 67.9 73.7 <.0001 71.7 75.2 .0020
Hypertension 77.7 84.8 <.0001 85.2 87.9 .0020
Peripheral vascular 12.7 13.4 .3622 15 14.4 .5025
disease
Valve replacement/repair 4.4 9.5 <.0001 4.7 9.5 <.0001
Prior stroke or transient 10.9 16.8 <.0001 12.2 17.5 <.0001
ischemic
attack
Other medical history
Anemia 20.9 17.7 .0006 24.4 18.6 <.0001
Alcohol abuse 5.5 3.4 <.0001 4.9 3.3 .001
Cancer 23.8 23.8 .4988 26.3 24.9 .2202
COPD 16.4 16.4 .9460 19 17.1 .062
Dialysis 1.9 1.0 .0009 2.3 1.0 <.0001
Frailty 8.0 5.3 <.0001 9.5 5.6 <.0001
GI bleed 13.2 8.0 <.0001 14.9 8.4 <.0001
Hyperthyroidism 1.4 2.2 .0183 1.3 2.2 .0225
Laboratory data
eGFR (mL/min/1.73 m2) 68.1 (53.2-85.1) 66.6 (52.6-81.0) .007 64.6 (50.9-81.5) 65.2 (51.8-79.5) .6527
Hess et al Oral Anticoagulation and Outcomes in AF Outpatients 453

Table 1 Continued
Total Study Population Patients with CHA2DS2-VASc Score
2

No OAC OAC No OAC OAC

Characteristic (n ¼ 2202) (n ¼ 7351) P (n ¼ 1870) (n ¼ 6916) P
Hematocrit 40.0 (36.3-43.0) 40.3 (37.0-43.4) <.0001 39.3 (35.7-42.3) 40.1 (36.8-43.2) <.0001
Left ventricular ejection <.0001
fraction
Normal (
50%) 74.3 70.0 72.6 69.2
Mild dysfunction 5.0 6.7 5.5 6.8
(>40%, <50%)
Moderate dysfunction 7.4 9.3 8.5 9.6
(
30% to 40%)
Severe dysfunction 2.5 4.8 2.9 5.0
(<30%)
Left atrial diameter <.0001 <.0001
Normal 33.6 21.0 30.8 20.8
Mild enlargement 23.8 22.4 24.3 22.0
Moderate 14.5 20.0 15.6 20.2
enlargement
Severe enlargement 13.5 22.8 15.1 23.5
CHADS2 score 2 (1-3) 2 (2-3) <.0001 2 (1-3) 2 (2-3) <.0001
0 12.7 4.4 3.7 1.3
1 26.8 20.1 24.3 17.6
2 28.0 34.6 33.4 37.0
3 19.4 23.9 23.2 25.8
4 8.6 10.7 10.1 11.5
5 3.5 5.0 4.2 5.3
6 1.0 1.4 1.1 1.5
CHA2DS2-VASc score 2 (1-3) 2 (2-3) <.0001 4 (3-5) 4 (3-5) <.0001
0 5.1 1.3 <.0001 e e <.0001
1 11.0 5.5 e e
2 14.7 11.3 17.8 12.1
3 17.6 18.9 21.0 20.3
4 19.7 24.8 23.5 26.6
5 16.0 19.4 19.0 20.9
6 9.1 11.2 10.9 12.0
7 5.2 5.1 6.0 5.5
8 1.3 2.1 1.5 2.2
9 0.3 0.5 0.4 0.5
ATRIA score 3 (1-4) 3 (1-4) .0454 3 (1-6) 3 (1-4) .0003
0-3 71.7 74.9 .0062 58.6 64.7
4-10 28.3 25.1 41.4 35.3
Warfarin
History of treatment 48 92.5 <.0001 48.3 92.7 <.0001
Current treatment 0.0 93.7 <.0001 e 94.0 <.0001
Antithrombotic therapy
Aspirin 73.2 35.5 <.0001 72.2 35.9 <.0001
Clopidogrel 15.5 4.6 <.0001 17.8 4.8 <.0001
Prasugrel 0.5 0.0 <.0001 0.5 0.03 <.0001
Dabigatran 0.0 6.4 <.0001 0 6.1 <.0001
Other antithrombotic 1.0 0.4 1.2 0.4 <.0001
Values are percentages or median (interquartile range).
ATRIA ¼ anticoagulation and risk factors in atrial fibrillation; CHF ¼ congestive heart failure; COPD ¼ chronic obstructive pulmonary disease; eGFR ¼
estimated glomerular filtration rate; GI ¼ gastrointestinal.
*Data are based on patients with available data for each characteristic.

DISCUSSION non-receipt as well as long-term outcomes. There were 3 main

In a nationwide cohort of approximately 10,000 patients with findings. First, the majority of patients who were not receiving
atrial fibrillation, we described current patterns of OAC OAC had a class I guideline recommendation for
454 The American Journal of Medicine, Vol 130, No 4, April 2017

Table 2 Independent Factors Associated with Non-Receipt of Oral Anticoagulation

Total Study Population Patients with CHA2DS2-VASc Score
2

Factor OR (95% CI) P OR (95% CI) P

Type of AF
Paroxysmal vs new-onset 0.73 (0.54-0.99) .0411 0.75 (0.56-1.00) .0483
Persistent vs new-onset 0.35 (0.25-0.49) <.0001 0.35 (0.25-0.48) <.0001
Permanent vs new-onset 0.14 (0.10-0.21) <.0001 0.15 (0.10-0.21) <.0001
Age (y)
80, per 1 increase 0.96 (0.95-0.97) <.0001 0.96 (0.95-0.97) <.0001
>80, per 1 increase 1.05 (1.02-1.08) .0017 1.06 (1.03-1.09) <.0001
Prior stroke or transient ischemic attack 0.52 (0.41-0.66) <.0001 0.49 (0.39-0.62) <.0001
Left atrial diameter
Mild enlargement vs normal 0.80 (0.66-0.97) .9440 0.79 (0.65-0.95) .0136
Moderate enlargement vs normal 0.59 (0.47-0.73) <.0001 0.58 (0.47-0.72) <.0001
Severe enlargement vs normal 0.53 (0.42-0.68) <.0001 0.54 (0.42-0.68) <.0001
History of coronary artery disease 1.36 (1.12-1.65) .0017 1.45 (1.24-1.70) <.0001
Prior valve replacement/repair 0.48 (0.34-0.68) <.0001 0.45 (0.32-0.63) <.0001
New York Heart Association class
I vs none 0.81 (0.62-1.06) .1318 0.82 (0.63-1.07) .1464
II vs none 0.62 (0.48-0.80) .0002 0.61 (0.48-0.78) <.0001
III/IV vs none 0.83 (0.59-1.17) .2890 0.79 (0.57-1.11) .1740
Alcohol abuse 1.84 (1.29-2.64) .0009 1.74 (1.24-2.44) .0015
Systolic blood pressure 120 mm Hg, 1.02 (1.01-1.03) .0040 1.02 (1.01-1.03) .0045
per 1 increase
Hyperthyroidism 0.30 (0.15-0.61) .0009 0.42 (0.23-0.77) .0051
Left ventricular ejection fraction
Mild dysfunction vs normal 0.99 (0.74-1.32) .0259 1.01 (0.78-1.32) .9291
Moderate dysfunction vs normal 0.73 (0.55-0.97) .0285 0.79 (0.60-1.03) .0770
Severe dysfunction vs normal 0.51 (0.31-0.83) .0077 0.53 (0.33-0.85) .0085
Weight (kg), per 1 increase 1.00 (0.99-1.00) .0388 1.00 (0.99-1.00) .0094
Hematocrit (%), per 1 increase 0.98 (0.97-1.00) .0386 0.98 (0.97-1.00) .0412
Cancer 1.21 (1.02-1.45) .0325 1.19 (1.00-1.42) .0471
Diabetes mellitus 0.82 (0.69-0.98) .0253 e e
Prior myocardial infarction 1.29 (1.01-1.65) .0400 e e
CI ¼ confidence interval; OR ¼ odds ratio.

anticoagulation. Second, factors unrelated to recommended non-use was associated with a significantly higher risk of death
stroke risk stratification, including atrial fibrillation duration, among eligible patients with a CHA2S2DVASc score
2.
chronicity, and comorbidity, were all independently associated Most patients not receiving OAC had multiple risk
with OAC non-use. Finally, compared with OAC use, OAC factors and a class I guideline indication for OAC. Fewer

Table 3 Outcomes According to Oral Anticoagulation Use and Associations with Non-Use Among Patients with a CHA2DS2-VASc Score
2
Outcome No OAC* OAC* Unadjusted HR (95% CI) P Adjusted HR (95% CI) P
Death 287 (7.42) 895 (5.78) 1.29 (1.14-1.49) .0002 1.22 (1.05-1.41) .0060
Stroke/non-central nervous 43 (1.12) 158 (1.03) 1.10 (0.79-1.52) .5917 1.14 (0.79-1.64) .4755
system embolism
Stroke/non-central nervous 72 (1.89) 252 (1.65) 1.15 (0.90-1.47) .2550 1.18 (0.91-1.54) .2191
system embolism/transient
ischemic attack
International Society of Thrombosis 9 (0.23) 95 (0.62) 0.38 (0.18-0.79) .0102 0.35 (0.15-0.81) .0147
and Haemostasis bleeding
Composite 338 (8.90) 1103 (7.25) 1.23 (1.09-1.41) .0010 1.19 (1.04-1.35) .0098
CI ¼ confidence interval; HR ¼ hazard ratio; OAC ¼ oral anticoagulation.
*Data are expressed as number of events (number of events per 100 patient-years).
Hess et al Oral Anticoagulation and Outcomes in AF Outpatients
than 1 in 5 patients who were not receiving OAC were low
risk (CHA2DS2VASc score of 0-1). That elevated blood
pressure above 120 mm Hg was not a significant factor
suggests hypertension may be overlooked in favor of
weighting of other CHADS2 score components. Prior work
has in fact shown that prescribers often underestimate
stroke risk.13 Patients and/or providers may also prioritize
bleeding risk over stroke risk, as indicated by the lower
use of OAC in those older than 80 years in the present
analysis. Further study of OAC in patients with frailty,
which is common in octogenarians and nonagenarians,
may help address equipoise for treatment in this special
and growing population.
Initially, investigators hypothesized that patients with
more sustained forms of atrial fibrillation have higher risks
of thromboembolism due to longer duration of atrial stasis
and potential for thrombus formation. This hypothesis was
supported by early reports assessing stroke risk according to
atrial fibrillation duration.14,15 However, an analysis of
pooled data from the Stroke Prevention in Atrial Fibrillation
trials indicated stroke rates were comparable between
groups stratified by atrial fibrillation duration on aspirin.16
An analysis of the Euro Heart Survey yielded similar re-
sults.17 By contrast, more recent studies with more granular
data than those preceding them show that the stroke risk
associated with paroxysmal atrial fibrillation is indeed lower
than that of persistent or permanent atrial fibrillation.18,19
Although patients with paroxysmal atrial fibrillation and a
CHA2DS2VASc score
2 likely have lower rates of
thromboembolism, they nonetheless have sufficient stroke
risk to derive net clinical benefit from OAC.8 Despite clear
professional guideline recommendations, prior reports
indicate that atrial fibrillation burden has been an important
determinant of whether patients receive OAC.20,21 The
present analysis confirms this trend persists in the current
therapeutic era among US outpatients. Our study also shows
that left atrial diameter, a marker of atrial fibrillation chro-
nicity,22 also influences treatment decisions. These data
suggest that efforts to improve OAC use should emphasize
the importance of therapy in patients with paroxysmal atrial
fibrillation and those earlier in the course of the disease.
Potential explanations for the observed association be-
tween the absence of OAC and death among guideline-
eligible patients are several. First, OAC may have been
withheld in the setting of more advanced illness, as sug-
gested by a higher burden of a number of several comor-
bidities among patients not on OAC. Although the present
analysis accounted for multiple comorbidities in multi-
variable adjustment, residual confounding may exist.
Second, patients on OAC may have more regular exposure
to healthcare providers, as required by anticoagulation
monitoring, or receive higher quality of care not related to
OAC status. Alternatively, the finding may reflect a true
survival benefit of OAC, as seen in many clinical trials.1,23
Importantly, adjusted rates of stroke and systemic embo-
lism were nonsignificantly increased among patients not
receiving OAC compared with those receiving it. Potential
455
explanations include limited power or patient selection
bias. Although absence of OAC is associated with less
bleeding, prior studies have demonstrated net clinical
benefit from OAC therapy even in those with high
bleeding risk.24
Compared with other registries, the use of OAC in
ORBIT-AF is relatively high.25 This is in contrast to prior
studies, reporting a rate generally below 60% among
eligible patients.9 Participation in ORBIT-AF is voluntary
and may reflect an interest in atrial fibrillation quality of
care. Stated another way, the data in ORBIT-AF probably
represent an “optimistic” estimation of the problem of un-
derutilization. In this setting, that factors not related to
recommended stroke risk stratification are operative in OAC
non-use, the high proportion of patients not receiving OAC
with a class I indication for it, and the observed mortality
signal are all the more striking.
The present analysis should inform future quality
improvement initiatives to improve low OAC treatment
rates among guideline-eligible patients. Future efforts
should include an educational component regarding appro-
priate risk stratification as well as the survival benefit of
OAC use. Specifically, the relative importance of older age
and hypertension versus atrial fibrillation duration and
chronicity should be underscored. In addition, the signifi-
cance of the mortality benefit vis-à-vis the bleeding risk of
OAC use should be discussed among physicians and pa-
tients within a shared decision-making framework.
This study has some limitations. Participation in ORBIT-
AF is voluntary, and enrolled patients may not fully repre-
sent atrial fibrillation outpatients in the United States.
Specifically, our findings may not be generalizable to mi-
nority and female populations, because they were under-
represented in our study sample. Further, the inclusion of
specialty practices in ORBIT-AF may have led to higher
rates of OAC use than that seen in primary care practices.
Contraindications to OAC may have been present but not
adequately documented in the medical record or abstracted.
Our analysis was observational and thus subject to residual
or unmeasured confounding.
CONCLUSIONS
In the largest dedicated registry of US ambulatory patients
with atrial fibrillation, the majority of patients untreated with
OAC have a guideline recommendation to receive it. Factors
unrelated to recommended stroke risk stratification,
including atrial fibrillation duration, chronicity, and co-
morbidity, were associated with absence of OAC. Finally,
consistent with prior studies of net clinical benefit, absence
of OAC among CHA2DS2 VASc score
2 patients was
associated with an increased risk of death. These data
highlight the need to improve atrial fibrillation patients’
outcomes by eliminating the OAC treatment gap. Quality
improvement efforts should focus attention on patients less
likely to receive OAC, including those with paroxysmal
atrial fibrillation, older age, and multiple comorbidities.
456 The American Journal of Medicine, Vol 130, No 4, April 2017

References 19. Steinberg BA, Hellkamp AS, Lokhnygina Y, et al. Higher risk of
death and stroke in patients with persistent vs. paroxysmal atrial
1. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic ther-
fibrillation: results from the ROCKET-AF Trial. Eur Heart J.
apy to prevent stroke in patients who have nonvalvular atrial fibrilla-
2015;36:288-296.
tion. Ann Intern Med. 2007;146:857-867.
20. Waldo AL, Becker RC, Tapson VF, Colgan KJ. Hospitalized
2. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus
patients with atrial fibrillation and a high risk of stroke are not being
warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:
provided with adequate anticoagulation. J Am Coll Cardiol.
1139-1151.
2005;46:1729-1736.
3. Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in
21. Nieuwlaat R, Capucci A, Lip GY, et al. Antithrombotic treatment in
nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883-891.
real-life atrial fibrillation patients: a report from the Euro Heart Survey
4. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban versus
on Atrial Fibrillation. Eur Heart J. 2006;27:3018-3026.
warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:
22. Sanfilippo AJ, Abascal VM, Sheehan M, et al. Atrial enlargement as a
981-992.
consequence of atrial fibrillation. A prospective echocardiographic
5. Giugliano RP, Ruff CT, Braunwald E, et al. Edoxaban versus warfarin
study. Circulation. 1990;82:792-797.
in patients with atrial fibrillation. N Engl J Med. 2013;369:2093-2104.
23. Mant J, Hobbs FD, Fletcher K, et al. Warfarin versus aspirin for
6. Fuster V, Rydén LE, Cannom DS, et al. ACC/AHA/ESC 2006
stroke prevention in an elderly community population with atrial
guidelines for the management of patients with atrial fibrillation: a
fibrillation (the Birmingham Atrial Fibrillation Treatment of the
report of the American College of Cardiology/American Heart Asso-
Aged Study, BAFTA): a randomised controlled trial. Lancet.
ciation Task Force on Practice Guidelines and the European Society of
2007;370:493-503.
Cardiology Committee for Practice Guidelines (Writing Committee to
24. Friberg L, Rosenqvist M, Lip GY. Net clinical benefit of warfarin in
Revise the 2001 Guidelines for the Management of Patients With Atrial
patients with atrial fibrillation: a report from the Swedish atrial fibril-
Fibrillation) developed in collaboration with the European Heart
lation cohort study. Circulation. 2012;125:2298-2307.
Rhythm Association and the Heart Rhythm Society. Circulation.
25. Cullen MW, Kim S, Piccini JP Sr, et al. Risks and benefits of anti-
2006;114:e257-e354.
coagulation in atrial fibrillation: insights from the Outcomes Registry
7. Fuster V, Rydén LE, Cannom DS, et al. 2011 ACCF/AHA/HRS focused
for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF) reg-
updates incorporated into the ACC/AHA/ESC 2006 guidelines for the
istry. Circ Cardiovasc Qual Outcomes. 2013;6:461-469.
management of patients with atrial fibrillation: a report of the American
College of Cardiology Foundation/American Heart Association Task
Funding: The ORBIT-AF registry is sponsored by Janssen Scientific
Force on Practice Guidelines. Circulation. 2011;123:e269-e367.
Affairs, LLC (Raritan, NJ). The sponsor did not have a role in the design
8. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guide-
and conduct of the study; collection, management, analysis, and interpre-
line for the management of patients with atrial fibrillation: a report of
tation of the data; or preparation, review, or approval of the manuscript.
the American College of Cardiology/American Heart Association Task
Conflict of Interest: GCF reports consulting for Janssen (modest) and
Force on Practice Guidelines and the Heart Rhythm Society. J Am Coll
Medtronic (modest). DES reports consulting for Boehringer Ingelheim,
Cardiol. 2014;64:e1-e76.
Bristol-Myers Squibb, Merck, Johnson & Johnson, and Pfizer; and research
9. Ogilvie IM, Newton N, Welner SA, Cowell W, Lip GY. Underuse of
grants from Boehringer Ingelheim and Bristol-Myers Squibb. BJG reports
oral anticoagulants in atrial fibrillation: a systematic review. Am J Med.
serving on Data and Safety Monitoring boards for studies sponsored by
2010;123:638-645.e4.
Mt. Sinai St. Lukes, Boston Scientific Corporation, Teva Pharmaceutical
10. Kirley K, Qato DM, Kornfield R, Stafford RS, Alexander GC. National
Industries, St. Jude Medical Inc, Janssen Research & Development, Baxter
trends in oral anticoagulant use in the United States, 2007 to 2011. Circ
Healthcare Corporation, Cardiovascular Research Foundation, and
Cardiovasc Qual Outcomes. 2012;5:615-621.
Thrombosis Research Institute; serving on an advisory board for Medtronic,
11. Piccini JP, Fraulo ES, Ansell JE, et al. Outcomes registry for better
Inc; and consulting for Janssen Scientific Affairs and Cipla Limited. JA
informed treatment of atrial fibrillation: rationale and design of ORBIT-
reports consulting for Boehringer Ingelheim, Daiichi Sankyo, Bristol-Myers
AF. Am Heart J. 2011;162:606-612.e1.
Squibb, Pfizer, and Janssen. PRK reports consulting for Johnson and
12. Schulman S, Kearon C; Subcommittee on Control of Anticoagulation
Johnson and Janssen. KWM reports consulting for the American College of
of the Scientific and Standardization Committee of the International
Cardiology, AstraZeneca, BAROnova, Bayer, Bio2 Medical, Boehringer
Society on Thrombosis and Haemostasis. Definition of major bleeding
Ingelheim, Bristol-Myers Squibb, Cubist, Eli Lilly, Elsevier, Epson, Forest,
in clinical investigations of antihemostatic medicinal products in non-
Glaxo Smith Kline, Johnson & Johnson, Medtronic, Merck, Mt. Sinai,
surgical patients. J Thromb Haemost. 2005;3:692-694.
Myokardia, Omthera, Portola, Purdue Pharma, Springer Publishing, The
13. Steinberg BA, Kim S, Thomas L, et al. Lack of concordance between
Medicines Company, Theravance, Vindico, and WebMD; and receiving
empirical scores and physician assessments of stroke and bleeding risk
research grants from Amgen, Daiichi Sankyo, Medtronic, Merck, St. Jude,
in atrial fibrillation: results from the Outcomes Registry for Better
and Tenax. PSC reports significant consulting for Optum Rx and significant
Informed Treatment of Atrial Fibrillation (ORBIT-AF) registry. Cir-
funding from the National Institutes of Health. BAS reports consulting for
culation. 2014;129:2005-2012.
Janssen (modest), Bristol-Myers Squibb (modest), and Pfizer (modest).
14. Petersen P, Godtfredsen J. Embolic complications in paroxysmal atrial
EDP reports having consulted for and participated in research sponsored by
fibrillation. Stroke. 1986;17:622-626.
Janssen Pharmaceuticals and Bayer Corporation. JPP reports consulting for
15. Treseder AS, Sastry BS, Thomas TP, Yates MA, Pathy MS. Atrial
Gilead (modest), GlaxoSmithKline (modest), Janssen (modest), Medtronic,
fibrillation and stroke in elderly hospitalized patients. Age Ageing.
Inc (modest), Pfizer (modest), SpectraNetics (modest); and receiving
1986;15:89-92.
research grants from Boston Scientific, Inc, Johnson & Johnson, ResMed,
16. Hart RG, Pearce LA, Rothbart RM, McAnulty JH, Asinger RW,
St. Jude, and SpectraNetics.
Halperin JL. Stroke with intermittent atrial fibrillation: incidence and
Authorship: All authors had access to the data and a role in writing the
predictors during aspirin therapy. Stroke Prevention in Atrial Fibrilla-
manuscript.
tion Investigators. J Am Coll Cardiol. 2000;35:183-187.
17. Nieuwlaat R, Dinh T, Olsson SB, et al. Should we abandon the com-
mon practice of withholding oral anticoagulation in paroxysmal atrial SUPPLEMENTARY DATA
fibrillation? Eur Heart J. 2008;29:915-922.
18. Al-Khatib SM, Thomas L, Wallentin L, et al. Outcomes of apixaban vs. Supplementary table accompanying this article can be found
warfarin by type and duration of atrial fibrillation: results from the in the online version at http://dx.doi.org/10.1016/j.amjmed.
ARISTOTLE trial. Eur Heart J. 2013;34:2464-2471. 2016.11.001.