You are on page 1of 13

Clinical Review & Education

Review

Stroke Prevention in Atrial Fibrillation


A Systematic Review
Gregory Y. H. Lip, MD; Deirdre A. Lane, PhD

Author Audio Interview at


IMPORTANCE Atrial fibrillation (AF) is associated with an increase in mortality and morbidity, jama.com
with a substantial increase in stroke and systemic thromboembolism. Strokes related to AF CME Quiz at
are associated with higher mortality, greater disability, longer hospital stays, and lower jamanetworkcme.com and
chance of being discharged home than strokes unrelated to AF. CME Questions page 1970

OBJECTIVE To provide an overview of current concepts and recent developments in stroke


prevention in AF, with suggestions for practical management.

EVIDENCE REVIEW A comprehensive structured literature search was performed using


MEDLINE for studies published through March 11, 2015, that reported on AF and stroke,
bleeding risk factors, and stroke prevention.

FINDINGS The risk of stroke in AF is reduced by anticoagulant therapy. Thromboprophylaxis


can be obtained with vitamin K antagonists (VKA, eg, warfarin) or a non-VKA oral
anticoagulant (NOAC). Major guidelines emphasize the important role of oral anticoagulation
(OAC) for effective stroke prevention in AF. Initially, clinicians should identify low-risk AF
patients who do not require antithrombotic therapy (ie, CHA2DS2-VASc score, 0 for men; 1 for
women). Subsequently, patients with at least 1 stroke risk factor (except when the only risk is Author Affiliations: University
of Birmingham Centre for
being a woman) should be offered OAC. A patients individual risk of bleeding from Cardiovascular Sciences,
antithrombotic therapy should be assessed, and modifiable risk factors for bleeding should City Hospital, Birmingham,
be addressed (blood pressure control, discontinuing unnecessary medications such as aspirin United Kingdom (Lip, Lane);
Aalborg Thrombosis Research Unit,
or nonsteroidal anti-inflammatory drugs). The international normalized ratio should be tightly
Department of Clinical Medicine,
controlled for patients receiving VKAs. Aalborg University, Aalborg,
Denmark (Lip).
CONCLUSIONS AND RELEVANCE Stroke prevention is central to the management of AF, Corresponding Author: Gregory Y.
irrespective of a rate or rhythm control strategy. Following the initial focus on identifying H. Lip, MD, University of Birmingham
Centre for Cardiovascular Sciences,
low-risk patients, all others with 1 or more stroke risk factors should be offered OAC.
City Hospital, Birmingham,
West Midlands B18 7QH, England
JAMA. 2015;313(19):1950-1962. doi:10.1001/jama.2015.4369 (g.y.h.lip@bham.ac.uk).
Corrected on July 15, 2015. Section Editor: Mary McGrae
McDermott, MD, Senior Editor.

A
trial fibrillation (AF) increases the risk of morbidity and The objective of this article is to provide an overview of cur-
mortality because of a substantially increased risk of rent concepts and recent developments in stroke prevention in pa-
stroke and systemic thromboembolism. Overall, AF tients with nonvalvular AF.
increases the stroke risk 5-fold; the attributable risk of ischemic
stroke related to AF increases from 4.6% from the ages of 50
through 59 years to 20.2% from the ages of 80 through 89 years.1
Search Strategy
Compared with non-AFrelated strokes, strokes related to AF are
associated with higher mortality, greater disability, longer hospital A comprehensive structured literature search was performed
stays, poorer functional outcome, and lower chance of being dis- using MEDLINE for studies published through March 11, 2015, that
charged home.2 Given the increasing prevalence and incidence of reported on AF and stroke or bleeding risk factors, as well as on
AF, this arrhythmia confers a major public health and economic stroke prevention. The search terms included each of the follow-
burden on any health care system. ing terms individually and in combination: atrial fibrillation,
The risk of stroke in AF is reduced by antithrombotic therapy.3 thromboembolism, stroke risk, risk stratification, oral anticoagula-
Thromboprophylaxis can be obtained with vitamin K antagonists tion, warfarin, aspirin, and thromboprophylaxis. We focused on
(VKA, eg, warfarin) or a non-VKA oral anticoagulant (NOAC). All the primary published research articles and systematic reviews, as
major guidelines emphasize the role of oral anticoagulation (OAC) well as on clinical trials complemented by large observational
use for stroke prevention in AF.4,5 cohorts.

1950 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

Risk Factors for Stroke At a Glance

The risk of stroke and mortality in AF is reduced by anticoagulant


Atrial fibrillation can cause stroke by several mechanisms consis- therapy, whether with vitamin K antagonists (VKA, eg, warfarin)
tent with the Virchow triad for thrombogenesis (thrombus forma- or a non-VKA oral anticoagulant (NOAC).
tion): (1) stasis in the left atrium causing flow abnormalities; (2) struc- As an initial step, clinicians should identify low-risk AF patients
tural heart and vascular disease (eg, mitral stenosis; fulfilling the who do not require antithrombotic therapy. Patients with at least
abnormal vessel wall Virchow criterion); and (3) abnormal coagula- 1 stroke risk factor (except when the only risk factor is being a
woman) should be offered OAC.
tion and fibrinolysis (eg, abnormal procoagulant hemostatic and
Clinicians also need to assess each patients individual risk of
platelet factors; fulfilling the abnormal blood constituents criterion).6
bleeding with antithrombotic therapy and correct modifiable
The prothrombotic state in AF has recently been comprehensively bleeding risk factors: control blood pressure; remove unneces-
reviewed.6 sary concomitant medication, such as aspirin, or NSAIDs; control
The most common risk factors associated with stroke (eg, heart INR and avoid labile INRs; counsel patient to reduce alcohol
failure, hypertension, diabetes, age, prior stroke) were initially iden- intake if excessive. Ultimately, the main priority is stroke
tified from the non-VKA cohorts of randomized trials conducted 2 prevention.
decades ago.2 Although these trials provided information on pa-
tients treated with placebo,
AF atrial fibrillation
they have been criticized for In a recent systematic review, peripheral arterial disease was
CHADS2 congestive heart failure, randomizing fewer than 10% significantly associated with AF-related stroke risk in all 10 obser-
hypertension, age 75 years,
diabetes mellitus, stroke
of patients screened and for vational studies, with a reported risk range of 1.3- to 2.5-fold.16
not recording all the risk fac- Complex aortic plaque on the descending aorta, as identified by
CHA2DS2-VASc CHADS2 criteria plus
vascular disease, age 65-74 years, sex tors nor consistently defin- transesophageal echocardiography, was also a significant risk
category ing them in the studies.7 factor.17 Prior myocardial infarction was validated as a significant
HAS-BLED hypertension, abnormal Numerous systematic predictor of the primary end point in 5 of the 6 studies. There
renal/liver function, stroke, bleeding reviews have examined AF- may be ethnic differences evident for peripheral arterial disease
history or predisposition, labile related stroke risk factors.8,9 as a risk factor, with an RR of stroke approximately 1.22-fold
international normalized ratio, elderly,
drugs/alcohol
A study by the Stroke in AF higher in European populations,18 but 1.80-fold higher in Far East-
Working Group 8 reported ern populations.19
HEMORR2HAGES hepatic or renal
disease, ethanol abuse, malignancy, that the strongest, most con-
older age, reduced platelet count or sistent independent risk fac-
function, rebleeding, hypertension, tors for stroke were prior
anemia, genetic factors, excessive fall
stroke or transient ischemic
Risk Stratification Schemes
risk, and stroke
attack (TIA) (relative risk The various stroke risk factors in AF have been used to formulate sev-
INR international normalized ratio
[RR], 2.5; 95% CI, 1.8-3.5), in- eral stroke risk prediction tools,20-28 to help risk stratify patients with
LAA left atrial appendage
creasing age (RR, 1.5 per AF (Table 1). The most common in use are the CHADS2 (congestive
NOAC non-VKA oral anticoagulant decade; 95% CI, 1.3-1.7), heart failure; hypertension; age 75 years; diabetes mellitus,1 point
OAC oral anticoagulation a history of hypertension for each; stroke, 2 points) and CHA2DS2-VASc scores (see Table 3 for
TIA transient ischemic attack (RR, 2.0; 95% CI, 1.6-2.5), definition). All the risk stratification schemes based on clinical risk
TTR time in therapeutic range and diabetes mellitus (RR, factors have weak or modest predictive value for identifying high-
VKA vitamin K antagonists 1.7; 95% CI, 1.4-2.0). In this risk patients.2 However, the CHA2DS2-VASc score has particular value
systematic review, female in identifying low-risk patients who do not need antithrombotic
sex was inconsistently associated with stroke risk, and the evi- therapy.31,32 Hence, more recent guidelines have moved toward ini-
dence was inconclusive for heart failure or coronary artery disease tial identification of low-risk patients first (because these patients
as independent predictors of AF-related stroke. do not need any antithrombotic therapy), rather than focus on iden-
Other systematic reviews that included large observational tifying high-risk patients.4
cohort studies also found evidence for vascular disease and
female sex 9 as risk factors for stroke. In a recent systematic CHADS2
review and meta-analysis of 17 studies, there was a 1.31-fold (95% The CHADS2 score was formulated and initially validated in a regis-
CI, 1.18-1.46) elevated risk of stroke in women with AF.10 Even in try of hospitalized patients with AF.22 The CHADS2 score is rela-
populations that were entirely anticoagulated, stroke rates were tively simple for identifying high-risk patients (score >2), whereas
higher in women, varying from 1.2% to 1.44% per patient-year for moderate risk was initially defined as a score of 1 to 2 and low risk as
men and 2.08% to 2.43% per patient-year for women. The risk of a score of 0.22 The CHADS2 had various limitations, and does not in-
stroke in women with AF is age dependent, with women younger clude many of the commonly accepted stroke risk factors (eg, age
than 65 and having no other risk factors being at low risk.11,12 In 65-74 years, vascular disease, female sex, asymptomatic left ven-
one European study involving AF patients younger than 65 years, tricular diastolic dysfunction, etc). Furthermore, according to the
female sex did not emerge as an independent risk factor13; how- original validation study, those with a prior stroke and no other risk
ever, female sex appears to still confer some stroke risk, especially factor (which represented the highest risk category) would only score
in Asian cohorts.14,15 2 and would be classified as moderate risk.33

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1951

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

Table 1. Risk Factors Included in Various Clinical Stroke Risk Stratification Schema and Guidelines

Risk Factors Other Factors


Female Prior TE Hyper- Heart Diabetes Vascular
Age, y Sex Event tension Failure Mellitus Disease
Scheme
SPAF,20 1999 >75a a
AFI, 21
1994 65-75 >75
CHADS2,22 2001 75
Framingham,23
2003
van Walraven,24
2003
Rietbrock,25 2008
CHA2DS2-VASc,26 65-74 75
2009
QStroke, 27
2013 Many others
ATRIA,28 2013 Proteinuria and eGFR; different weighting
for primary and secondary prevention
Guidelines
ACCP,29 2012 65-74 >75 CHADS2 score, 0, non-CHADS2 risk
factors (similar to CHA2DS2-VASc)
should be considered
ESC,4 2012 65-74 75 Based on CHA2DS2-VASc
CCS,30 2014 65
AHA/ACC/HRS,5 65-74 75 Based on CHA2DS2-VASc
2014
NICE,4 2014 65-74 75 Based on CHA2DS2-VASc

Abbreviations: ACCP, American College of Chest Physicians; AFI, Atrial D, diabetes, and S2, previous stroke, transient ischemic attack (TIA), or systemic
Fibrillation Investigators; AHA/ACC/HRS, American Heart Association, American embolism; CHA2DS2-VASc: C, congestive heart failure, H, hypertension, A2, age
College of Cardiology, and Heart Rhythm Society; CCS, Canadian Cardiovascular at least 75 years, D, diabetes, S2, previous stroke, TIA, or systemic embolism,
Society; eGFR, estimated glomerular filtration rate; ESC, European Society of V, vascular disease, Sc, sex category female sex.
Cardiology; NICE, National Institute for Health and Care Excellence; SPAF, a
Age and female sex combined are a single risk factor.
Stroke Prevention in Atrial Fibrillation; TE, thromboembolism. CHADS2:
C, congestive heart failure, H, hypertension, A, age 65 through 74 years,

CHA2DS2-VASc Other Risk Stratification Schemes


The CHA2DS2-VASc score (Table 2) was initially validated in the Clinical stroke risk prediction scores such as CHADS 2 and
Euro-Heart survey cohort.26 This score included various non- CHA 2 DS 2 -VASc are based on common risk factors in patients
CHADS2 risk factors such as age from 65 to 74 years, female sex, with AF whom clinicians encounter in clinical practice. The
and vascular disease. For example, in contrast to CHADS2 the C in CHA2DS2-VASc score does not include other rare factors that have
CHA2DS2-VASc refers to congestive heart failure, including moder- been associated with stroke or systemic embolism in AF such as amy-
ate to severe left ventricular dysfunction (ejection fraction, <40%) loid heart disease and end-stage renal failure. Large observational
and recent decompensated heart failure irrespective of ejection cohorts have clarified the effect of severe renal impairment on stroke
fraction, thus including both heart failure with reduced ejection and bleeding risks in AF, but renal impairment does not add predic-
fraction or heart failure with preserved ejection fraction.4 In the lat- tive value to established scores, such as CHA2DS2-VASc.37
ter group, patients with cardiomyopathy (eg, restrictive or hyper- The other risk factor score schemes that have been devel-
trophic) could also be included under the C criterion, although oped, including the CHADS2 score,38 the Anticoagulation and
robust data are limited. Risk Factors in Atrial Fibrillation (ATRIA) stroke score, 28 the
The CHA2DS2-VASc score has been validated in numerous CHA2DS2-VASc score,39,40 and the Qstroke27 score, all have limi-
independent cohorts, including non-Western populations. In 2 Far tations. All the major guidelines have continued to emphasize the
Eastern studies, the CHA2DS2-VASc but not the CHADS2 score use of the CHA2DS2-VASc score.
was significantly predictive of stroke.34,35 Most studies demon- Most risk stratification schemes based on clinical criteria have
strate that CHA2DS2-VASc is best at identifying truly low-risk weak predictive value for high-risk patients who sustain events
patients for whom the absolute risks of stroke or systemic embo- (C statistic, approximately 0.6); thus, there has been much inter-
lism were less than 1% per year and was as good aspossibly est in the use of biomarkers to enhance the predictive value of
better thanthe older CHADS 2 score for predicting high-risk clinical risk scores. Such biomarkers can include blood markers
patients. 31,32,36 The CHA 2 DS 2 -VASc score is now the recom- (eg, von Willebrand factor, D-dimer, natriuretic peptides),41-43
mended risk score by European,4 US,5 and National Institute for urine (eg, proteinuria, estimated glomerular filtration rate, or cre-
Health and Care Excellence (NICE)4 guidelines. atinine clearance),44 cardiac imaging (echocardiography, whether

1952 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

Table 2. Summary of Guideline Recommendationsa

Stroke Risk Stratification Treatment Recommendation Comments


ACCP,29 2012
CHADS2 score
2 OAC Warfarin or dabigatran
1 OAC Warfarin or dabigatran
0 plus additional non-CHADS2 risk OAC Warfarin or dabigatran
factors (eg, age 65-74 y, woman,
and vascular disease)
No risk factors No antithrombotic therapy
ESC,4 2012
Initial step: identify low risk patients No antithrombotic therapy
(CHA2DS2-VASc 0 in males,
1 in females)
Subsequent step: for patients OAC is recommended for OAC refers to a VKA (eg, warfarin)
with 1 additional stroke risk factors CHA2DS2-VASc score 2 with TTR>70%, or a NOAC
or should be considered for (preferred); antiplatelet therapy
CHA2DS2-VASc score of 1 in men with aspirin-clopidogrel combination
therapy orless effectivelyaspirin
monotherapy is recommended only
when patients refuse any form of OAC
CCS,30 2014
Algorithm: identify those aged 65 y OAC Warfarin or a NOAC, preferred
and CHADS2 score risk factors
Algorithm: vascular disease Aspirin Abbreviations: ACCP, American
College of Chest Physicians; AF, atrial
Algorithm: no risk factors, No antithrombotic therapy
ie, age <65 y with no CHADS2 risk fibrillation; AHA/ACC/HRS, American
factors nor vascular disease Heart Association, American College
AHA/ACC/HRS,5 2014 of Cardiology, Heart Rhythm Society;
CCS, Canadian Cardiovascular
CHA2DS2-VASc score Society; European Society of
2 OAC OAC refers to warfarin or a NOAC Cardiology; NICE, National Institute
as an alternative for Health and Care Excellence;
1 Nothing, aspirin, or OAC NOAC, nonvitamin K antagonist oral
0 No antithrombotic therapy anticoagulant; OAC, oral
anticoagulant; TTR, time in
NICE,4 2014 therapeutic range; VKA, vitamin K
Evaluative steps antagonist.
a
Initial: identify low-risk patientsb No antithrombotic therapy See Table 1 footnotes for CHADS2
and CHA2DS2-VASc expansions.
Subsequent: for AF patients Offer OAC (CHA2DS2-VASc 2) OAC refers to warfarin or a NOAC
with 1 additional stroke risk or consider OAC (CHA2DS2-VASc score as an alternative. b
Low-risk patients include men who
factors of 1 in men) Aspirin monotherapy should not be have a CHA2DS2-VASc score of 0
used for stroke prevention in AF.
and women 1.

transthoracic or transesophageal), or cerebral imaging (eg, com- risk patients who do not need antithrombotic therapy. All other AF
puted tomography or magnetic resonance imaging), which would patients who have 1 or more stroke risk factors can be offered stroke
provide additional refinement to clinical risk stratification for the prevention therapy.
identification of high-risk individuals. The real practical value of
measuring various biomarkers for stroke-risk prediction would
require testing large cohorts of individuals who have not received
Risk Factors for Bleeding
anticoagulation treatment rather than testing highly selected
cohorts of patients who have. Also, stroke risk scoring schemes In addition to stroke risk assessment, it is important to evaluate
must have simplicity and ease of use, for rapid application in bleeding risk in AF patients, especially cases in which thrombopro-
everyday clinical practice. Although scoring systems incorporat- phylaxis is being considered.46 Various bleeding risk stratification
ing complex multivariate analyses and modeling or addition of schemes have been proposed but until recently have not been widely
biomarkers may offer some improved predictive value, the resul- used due to complexity and overlaps with stroke risk.47 Of the vari-
tant scores are less practical for everyday clinical use because of ous bleeding risk scores,48-53 only 3 have been derived and vali-
their complexity. dated in AF populations: HEMORR2HAGES (hepatic or renal dis-
Any single stroke risk factor confers risk, and if left untreated, ease, ethanol abuse, malignancy, older age, reduced platelet count
patients with AF may be exposed to a fatal or disabling stroke.45 or function, re-bleeding, hypertension, anemia, genetic factors, ex-
Stroke risk stratification schemes have been formulated with as- cessive fall risk, and stroke),50 HAS-BLED (hypertension, abnormal
sumptions that risk factors carry equal weight, but they do not.15 With renal/liver function, stroke, bleeding history or predisposition, la-
the clinical practice change advocated by contemporary guide- bile international normalized ratio [INR], elderly, drugs/alcohol),52
lines, the focus should be on the initial identification of truly low- and ATRIA.53

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1953

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

Table 3. Assessment of Stroke and Bleeding Risk in Atrial Fibrillation


INRs. Use of a decision support tool that integrates patient-specific
Patients by Schema stroke and bleeding risk would result in significant gains in quality-
adjusted life expectancy.60
Score
Falls are not an independent predictor of bleeding for those
CHA2DS2-VASca
who are taking OAC medication.61 A modeling exercise has sug-
Congestive heart failure 1
gested that a patient would need to fall 295 times a year for the
Hypertension 1
benefit of stroke prevention with OAC to outweigh the risk of
Age 75 y 2
serious bleeding.62 Although a history of falls should not be a con-
Diabetes mellitus 1
traindication to OAC, AF patients who have actually sustained a
Stroke/TIA/TE 2 fall remain at high risk of thromboembolism and morbidity.63 Fur-
Vascular disease (prior MI, PAD, or aortic plaque) 1 ther investigation to ascertain the reasons for the fall and target-
Age 65-74 y 1 ing interventions to reduce risk of fallsfor example, treatment
Sex category (ie, female sex) 1 for correctable causes of falls (such as lowering blood pressure,
Maximum score 9 vertigo, etc), polypharmacy, and the need for walking aids
HAS-BLEDb should be paramount. Nevertheless, treatment decisions should
Hypertension (systolic blood pressure >160 mm Hg) 1 be individualized.
Abnormal renal and liver function (1 point each)c 1 or 2 Stroke and bleeding risk are closely related. Recent compari-
Stroke 1 sons of HAS-BLED with the CHADS2 and CHA2DS2-VASc scores
Bleeding tendency/predispositionc 1 showed an increase in bleeding rates among patients who have
Labile INRs (eg, TTR<60%, applies only if the patient 1 higher HAS-BLED, CHADS2, and CHA2DS2-VASc scores, but this was
is taking warfarin)c generally only significant for the HAS-BLED score; the latter score
Elderly (eg, age >65 y, frail condition) 1 was also superior to CHADS2 and CHA2DS2-VASc for bleeding risk
Drugs or alcohol excess (1 point each)c 1 or 2 prediction.54,55 Other efforts have led to derivation and validation
Maximum score 9 of a composite stroke and bleeding risk score.64,65 Although there
Abbreviations: INR, international normalized ratio; MI, myocardial infarction; was a marginal improvement in the C statistic for predicting high-
PAD, peripheral artery disease; TE, thromboembolism; TIA, transient ischemic risk patients (C for composite score, 0.728 [95% CI, 0.659-0.798]
attack; TTR, time in therapeutic range.
compared with 0.706 [95% CI, 0.638-0.774] for HAS-BLED and
a
CHA2DS2-VASc is used to initially identify low-risk patients (CHA2DS2-VASc,
0.654 [95% CI, 0.574-0.733] for CHA2DS2-VASc scores), the dis-
0 in males, 1 in females) who do not need any antithrombotic therapy);
subsequently, stroke prevention is offered to those with at least 1 additional crimination compared with individual stroke and bleeding risk scores
stroke risk factor. was nonsignificant. Thus, stroke risk should be assessed using a vali-
b
HAS-BLED categorizes those into low (score 0-2) or high (3) risk, where the dated stroke score scheme and bleeding risk with a specific bleed-
latter can be identified for more regular review and follow-up. ing risk score scheme.
c
Abnormal renal function classified as the presence of long-term dialysis, renal
transplant, or serum creatinine of 2.3 mg/dL (200 mmol/L) or higher.
Abnormal liver function defined as chronic hepatic disease (eg, cirrhosis) or
biochemical evidence of significant hepatic derangement (eg, bilirubin 2 to 3
the upper limit of normal, in association with aspartate aminotransferase,
Net Clinical Benefit of Oral Anticoagulants
alanine aminotransferase, or alkaline phosphatase 3 the upper limit normal); Effective stroke prevention for patients with AF requires the use of
history of bleeding or predisposition (anemia); labile INR (ie, TTR<60%);
OACs, whether a VKA or a NOAC. NOACs fall into 2 broad classes,
concomitant antiplatelet drugs or nonsteroidal anti-inflammatory drugs, or
alcohol excess. the oral direct thrombin inhibitors (eg, dabigatran) and oral factor
Xa inhibitors (eg, rivaroxaban, apixaban, and edoxaban) (Table 4).
Stroke or systemic embolism is significantly reduced by VKAs
Of the different scores, the simple HAS-BLED score (Table 3) in patients with AF (RR reduction, 64%; 95% CI, 49%-74%).66
has been validated in multiple independent cohorts and outper- Vitamin K antagonists are associated with an absolute RR of 2.7%
forms other bleeding risk scores in the prediction of serious per year (number needed to treat, 37) for embolism in patients
bleeding.54-56 HAS-BLED has been validated in NOACs and in un- with no history of prior stroke (primary prevention) and 8.4% per
treated and aspirin-treated patients,52,57 HAS-BLED is the only score year (number needed to treat, 12) for patients with a history of
predictive of intracranial hemorrhage risk and also has been shown prior stroke (secondary prevention). Compared with the control
to predict bleeding during bridging therapy58 and during percuta- or placebo groups, the relative RR attributable to VKA is 26%
neous coronary intervention.59 (95% CI, 3%-43%; absolute RR 1.6% per year) for all-cause
A high HAS-BLED score is not a reason to withhold OAC, but it mortality.66
can be used to highlight patients potential risk of bleeding and alert A meta-analysis examining the efficacy of aspirin compared
physicians to schedule more regular follow-up visits and provide with either a control treatment or placebo for stroke or systemic
straightforward warning about the necessity of avoiding falls and not embolism or mortality in AF found no benefit.66 A 19% reduction
engaging in other high-risk activities. A high HAS-BLED score sug- in stroke was observed with aspirin but that finding was attribut-
gests the need to focus on correcting the potentially reversible fac- able to the only positive study, the Stroke Prevention on Atrial
tors that contribute to bleeding, which include uncontrolled hyper- Fibrillation (SPAF-I) trial, which reported a 42% reduction in
tension, concomitant use of aspirin or nonsteroidal anti- stroke or systemic embolism with 325 mg/d of aspirin compared
inflammatory drugs with OAC, excessive alcohol intake, and labile with a control treatment. However, the SPAF-I trial had major

1954 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

Table 4. Optimal Selection of Oral Anticoagulation for Stroke Prevention in Atrial Fibrillation
Factor X
VKA, Warfarin Direct Thrombin Inhibitors Inhibitors Rivaroxaban Apixaban Edoxaban
Recurrent stroke or TIA despite
150 mg of dabigatran, 2/d
treatment VKAa
Moderate or severe renal impairmentb
GI tract symptoms or dyspepsiac
75 mg dabigatran, 2/d (US);
High risk of bleedingd 110 mg dabigatran, e
2/d (rest of world)
Preference for 1 dose per day
c
Abbreviations: GI, gastrointestinal; TIA, transient ischemic attack; VKA, vitamin K Also has an increased risk of bleeding.
antagonist. d
High risk of bleeding is defined by a HAS-BLED score of 3 or higher. Use agent
a
Superior efficacy for preventing both stroke and hemorrhage. with the lowest incidence of bleeding, especially for GI tract bleeding.
b e
For creatinine clearance of <15 mL/min use VKA. Awaiting approval.

internal heterogeneity for the aspirin effect. The study had 2 tion Compared With Vitamin K Antagonism for Prevention
major groups: one group could not, or would not, take anticoagu- of Stroke and Embolism Trial in Atrial Fibrillation), ARISTOTLE
lation medicine. Of these, one subgroup of AF patients received (Apixaban for Reduction in Stroke and Other Thromboembolic
placebo and the other aspirin. For the group that could take anti- Events in Atrial Fibrillation), and ENGAGE-AF TIMI 48 (Effective
coagulation medication, one subgroup received placebo; another, Anticoagulation with Factor Xa Next Generation in Atrial
aspirin; and a third group, warfarin. There was a significant, 94% FibrillationThrombolysis In Myocardial Infarction study 48) trials,
reduction in stroke risk attributable to aspirin compared with pla- 42 411 participants received a NOAC and 29 272 participants
cebo in the anticoagulation-eligible group. In the group that could received warfarin. NOACs significantly reduced stroke or systemic
not receive anticoagulation medication, there was no significant embolism events by 19% compared with warfarin (RR, 0.81; 95%
difference in stroke rates between aspirin and placebo.67 Indeed, CI, 0.73-0.91; P < .001) mainly related to a reduction in hemor-
aspirin is often considered for those deemed inappropriate for rhagic stroke (RR, 0.49; 0.38-0.64; P < .001). NOACs also signifi-
anticoagulation. In the SPAF-I trial, aspirin did not reduce stroke cantly reduced all-cause mortality (RR, 0.90; 0.85-0.95; P < .001)
among those older than 75 years, nor did it prevent severe and intracranial hemorrhage (RR, 0.48; 0.39-0.59; P < .001) but
strokes. In a meta-analysis of AF studies, the reduction of stroke increased gastrointestinal tract bleeding (RR, 1.25; 1.01-1.55;
or vascular events using aspirin declined with age, whereas bleed- P = .04). Low-dose NOACs had similar overall reductions in stroke
ing risk between it and warfarin was not markedly different.68 or systemic embolism events to warfarin (RR, 1.03; 0.84-1.27;
This was also evident from recent randomized trials and cohort P = .74) and had a more favorable bleeding profile (RR, 0.65; 0.43-
studies reporting that major bleeding and intracranial hemor- 1.00; P = .05), with significantly more ischemic strokes (RR, 1.28;
rhage did not differ between aspirin and OAC use and that 1.02-1.60; P = .045).
adverse events occurred more frequently among very elderly With the availability of 4 different NOACs as well as the VKAs,
patients who were taking aspirin. physicians now have a choice and can fit the drug to the patient.
With VKAs, the quality of anticoagulation control is critical (as Various patient characteristics may influence the initial choice of
reflected by the time in therapeutic range [TTR], with a target INR one drug over another (Table 4). In the absence of head-to-head
of 2.0-3.0). Time in therapeutic ranges are closely related to effi- clinical trials, clinicians cannot directly compare one NOAC
cacy and safety, with low rates of stroke and bleeding with high against another. Nevertheless, regulatory bodies and health eco-
TTRs. 69,70 With INRs persistently higher than 3.0, there is an nomic analyses have performed indirect comparisons of one
increase in bleeding, especially in elderly patients, whereas the agent against another using warfarin as the comparator; such
rate of thromboembolisms increase when INRs are less than 2.0. indirect comparisons (or network meta-analyses) should be inter-
Even at an average INR of 1.7, the risk of stroke already increases preted with some caution given heterogeneity in trial inclusion-
2-fold. Thus, VKAs require regular anticoagulation monitoring and exclusion criteria, quality of INR control. 73 Overall, there are
are open to significant interpatient and intrapatient variability few substantial differences in efficacy, although 150 mg of dabi-
due to interactions with diet, drugs, and alcohol, among other gatran twice daily may offer the greatest potency by reducing
factors.71 both ischemic and hemorrhagic stroke, and it has a similar risk of
In more recent years, the availability of the NOACs has major bleeding as warfarin. For safety, the best bleeding profile
changed the landscape of stroke prevention in AF. The NOACs is seen with 110 mg of dabigatran twice daily, apixaban, and
offer efficacy, safety, and convenience for stroke prevention com- edoxaban.74-76
pared with the VKAs. The various phase 3 trials comparing NOACs In some health care systems, there is a preference for a 3- to
against warfarin have been the subject of numerous reviews and 6-month warfarin trial to determine if high TTRs (eg, >65%) can
meta-analyses. be achieved. NOACs are only used if the TTRs remain poor
In a recent meta-analysis72 that included the RE-LY (Random- (<65%). This strategy may put patients at risk of stroke or sys-
ized Evaluation of Long-term Anticoagulation Therapy), temic embolism because in the initial treatment stages, inad-
ROCKET-AF (Rivaroxaban Once-daily, Oral, Direct Factor Xa Inhibi- equate anticoagulation may result in an excess of thromboem-

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1955

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

Table 5. Definition of the SAMe-TT2R2 Score, Used to Aid Initial Decision bolic events (RR, 1.71; 95% CI, 1.39-2.12).77 Various clinical and
Making Between Vitamin K Antagonist (With Good Quality Anticoagulation pharmacogenetic factors have been proposed as determinants of
Control) and a NonVitamin K Antagonist Oral Anticoagulanta warfarin dose or TTR.71,78 Recent randomized trials based on a
pharmacogenetic testing as a means of predicting the response
Definitions Points
Sex (female) 1
to warfarin have been disappointing.
Simple clinical risk factors have been incorporated into a
Age (<60 y) 1
new score, the SAMe-TT2R2 score78 (Table 5), which has been
Medical historyb 1
proposed as a means of identifying those newly diagnosed
Treatment (interacting drugs, eg, amiodarone for 1
rhythm control) patients with AF who have not been treated with anticoagulation
Tobacco use (within 2 y) 2 agents and who have a probability of doing well taking a VKA
Race (not white) 2 (SAMe-TT2R2 score, 0-2) and achieve a TTR greater than 65% or
Maximum points 8 70%78,79 (Figure). In contrast, a SAMe-TT2R2 score higher than 2
a
suggests that such patients are unlikely to achieve a good TTR by
The SAMe-TT2R2 score is proposed as a means to help with decision making,
to identify those newly diagnosed nonanticoagulated AF patients who have a taking a VKA, so a NOAC should be used initially, without subject-
probability of doing well while taking a vitamin K antagonist (VKA) (with ing the patient to a trial of warfarin period. The SAMe-TT2R2
SAMe-TT2R2 score, 0-2) and achieve a time in therapeutic range (TTR) of at score has since been validated in independent studies to show it
least 65% or 70%. In contrast, a SAMe-TT2R2 score of more than 2 suggests
can discern those likely to achieve poor TTRs. A high SAMe-TT2R2
that such patients are unlikely to achieve a good TTR while taking a VKA, and a
non-VKA oral anticoagulant should be used upfront, without a trial of score translates to a greater chance of labile INRs, thromboembo-
warfarin period. lism, death, and bleeding.80-83 This approach has been recom-
b
Two of the following: hypertension, diabetes mellitus, coronary artery disease mended in a position document from the European Society of
or myocardial infarctions, peripheral artery disease, congestive heart failure , Cardiology (ESC) Working Group on Thrombosis Anticoagulation
previous stroke, pulmonary disease, or hepatic or renal disease.
Task Force.71

Figure. Algorithm for Risk Stratification and Selection of Anticoagulation Therapy for Stroke Prevention in Atrial Fibrillation

Patient with newly diagnosed


atrial fibrillation

Calculate CHA2DS2-VASc score


to determine stroke risk

High stroke risk?


No Yes a
Male, CHA2DS2-VASc score 1
(Low stroke Female, CHA2DS2-VASc score 2 (High stroke
risk) risk)

No antithrombotic therapy Calculate SAMe TT2R2 score to determine


initial anticoagulation treatmenta

No Yes
SAMe TT2R2 score >2?
(Score 0-2) (Score >2)

Vitamin K antagonist (VKA) therapy


(eg, warfarin)

Monitor anticoagulation control (goal:


time in therapeutic range [TTR] >70%)

Inadequate anticoagulation control?


TTR <65%
OR
within past 6 mo
No INR >5 twice Yes
or
INR >8 once
or
INR <2 twice

Continue VKA therapy with regular Non-VKA-oral anticoagulant therapy


monitoring (oral direct thrombin inhibitors or oral
factor Xa inhibitors)

a
See Table 1 footnotes for abbreviation expansions; SAMe-TT2R2 is explained in Consider NOAC or VKA. If considering VKA, calculate SAMe-TT2R2 first before
Table 5. deciding on optimal OAC therapy.

1956 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

patients with a CHADS2 score of 0, additional non-CHADS2 risk fac-


Net Clinical Benefit tors should be considered, including age from 65 through 74 years,
female sex, and vascular disease, whereby the presence of multiple
When making decisions about thromboprophylaxis, an important non-CHADS2 risk factors would merit OAC treatment.
consideration is to balance ischemic stroke prevention against the In 2014, the new AHA/ACC/HRS guidelines5 for AF advocated
potential risk of serious bleeding. The net clinical benefit of anti- use of the CHA2DS2-VASc score for stroke risk stratification, but in
thrombotic therapy can be determined in many ways, eg, by bal- a categorical approach whereby OAC was recommended for a
ancing the reduction in ischemic stroke against the risk of serious CHA2DS2-VASc score of 2 or higher, and no therapy for a score of
bleeding (ie, intracranial hemorrhage) weighted 1.5-fold.84 Other 0; however, for CHA2DS2-VASc score of 1, the recommendation
complex weighted formulae for net clinical benefit have been pro- was no therapy, aspirin or OAC. Oral anticoagulants could be
posed, but none are perfect. either NOACs or VKAs. The value of the US guidelines approach
Recent studies have found a positive net clinical benefit (bal- for CHA2DS2-VASc score of 1 was recently tested in a nationwide
ancing ischemic stroke reduction against the risk of intracranial cohort study of 12 935 men with AF and a CHA2DS2-VASc score of
hemorrhage weighted 1.5-fold) for OAC in patients with AF and at 1 but were not receiving antithrombotic therapy, where the
least 1 additional stroke risk factor.85,86 With the NOACs, the net annual stroke rate overall was 2.75%. 92 The risk of ischemic
clinical benefit was also generally positive for dabigatran and stroke ranged from 1.96% per year for AF patients with vascular
apixaban with CHA2DS2-VASc score of 1, and for all NOACs with disease to 3.50% per year for those aged 65 through 74 years.
CHA2DS2-VASc score of at least 2,87 and in a modeling analysis, For women with AF and a CHA2DS2-VASc score of 2 (ie, 1 addi-
could translate to an additional 65 000 cardiovascular events tional risk factor), the annual stroke rate was 2.55%, with the risk
and deaths prevented in Europe.88 The threshold for initiating of ischemic stroke increasing from 1.91% per year for patients
OAC treatment has been calculated as a stroke rate of 0.9% per with hypertension to 3.34% per year for those aged 65 through
year, based on the balance of ischemic stroke reduction vs intra- 74 years. Similar findings were observed from the Danish nation-
cranial hemorrhage with the availability of new safer OAC wide cohort study, where event rates in nontreated AF patients
agents.89 with 1 additional stroke risk factor increased stroke by 3.01-fold
and death by 3.12-fold, at 1 year follow-up.93 Thus, OAC should
still be considered for AF patients with 1 additional stroke risk fac-
tor given their high risk of ischemic stroke and death.
Guidelines for AF Treatment
In 2014, NICE published its new AF guidelines.4 In contrast to
When the only available oral anticoagulants were the VKAs, the some guidelines, the NICE guidelines are based on systematic re-
focus was to identify high-risk patients to be targeted for this views, evidence appraisal and cost-effectiveness. NICE guidelines
inconvenient drug, VKA (eg, warfarin), given the need for antico- advocate use of the CHA2DS2-VASc and HAS-BLED scores for stroke
agulation monitoring and other restrictions necessary with VKA and bleeding risk assessment, respectively. The recommendations
use. Also, older guidelines, such as the 2006 ACC/AHA/ESC joint are to initially identify low-risk patients who do not need any anti-
guidelines had divided patients using the CHADS2 score into low-, thrombotic therapy. Patients with a CHA2DS2-VASc score 2 or higher
moderate-, and high-risk strata, for which low-risk patients were should be offered OAC, while men with a score of 1 should be con-
recommended aspirin; moderate-risk patients, aspirin or warfarin; sidered for OAC. If VKAs are used, a TTR of higher than 65% was rec-
and high-risk patients, warfarin.90 Such an approach did not work ommended. The guidelines clearly state that aspirin should not be
well, and numerous studies have shown that high-risk patients used for stroke prevention in AF, due to its minimal efficacy, poor
were undertreated, with OAC being used in approximately 50% safety, and lack of cost-effectiveness.
of patients and with regional differences evident.91 Also, stroke The Canadian Cardiovascular Society (CCS) published its
risk is a continuum, and any 1 risk factor can confer risk if left focused update in 2014, proposing a simplified algorithm-based
untreated, exposing the patient with AF to a fatal or disabling approach to stroke risk stratification.30 The first step in the algo-
stroke.45 rithm is to identify those aged 65 years or older who should be
In 2010, and in a focused update in 2012, the ESC guidelines ad- offered OAC. The second step is to identify those younger than
vocated a change in clinical practice highlighting the need to iden- 65 years with CHADS2 risk factors who should have OAC. Next,
tify the low-risk patients (age <65 years and lone AF, including wom- those younger than 65 years who have a CHADS2 score of 0 with
en) who do not need any antithrombotic therapy. Patients with 1 or arterial disease, ie, coronary, aortic or peripheral are recom-
more stroke risk factors should be offered OAC, with a preference mended aspirin alone. Finally, those patients younger than 65
for NOACs. If VKAs are used, a TTR higher than 70% was recom- years with no CHADS2 risk factors nor vascular disease are recom-
mended. Only if patients refuse any form of OAC, then antiplatelet mended no antithrombotic therapy.30 This approach was tested
therapy with aspirin-clopidogrel combination therapy (if not at an in a nationwide cohort study, in which the overall rate of the com-
unduly high bleeding risk) orless effectivelyaspirin mono- bined end point of ischemic stroke, systemic embolism , or TIA
therapy, may be considered. The HAS-BLED score is recom- was 4.32 per 100 person-years at 1 year, among the patients who
mended for bleeding risk assessment. would have had an indication for OAC therapy according to ESC
In 2012, the American College of Chest Physicians (ACCP) pub- guidelines but were not recommended for OAC according to the
lished its consensus guidelines on antithrombotic therapy in 2014 CCS algorithm; also, the subgroup of patients with prior vas-
AF patients.29 This was based on the CHADS2 score, where OAC cular disease and a CHADS2 score of 0 (ie, only recommended
was recommended for those with CHADS2 score of 1 or higher. In aspirin treatment according to CCS algorithm) had an event rate

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1957

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

of 4.84 per 100-person-years at 1-year follow-up.94 Thus, based Acute Coronary Syndrome and Percutaneous Coronary
on the 2014 CCS algorithm, the subgroup not recommended for Intervention
OAC were not low risk and use of the ESC guideline approach Many patients with AF have associated coronary artery disease.
offered additional refinement of stroke risk stratification in such Some may present with an acute coronary syndrome (ACS), be un-
patients. dergoing percutaneous coronary intervention, or both. Such pa-
In summary, stroke risk stratification and treatment decisions tients are at high risk, so management is complex, having to use OAC
can now be simplified, notwithstanding the various approaches ad- plus antiplatelet therapy to balance between stroke prevention, re-
vocated in guidelines. The initial step is to identify the low-risk pa- current cardiac ischemia, stent thrombosis, and risk of serious bleed-
tients who do not need any antithrombotic therapy, given their low ing (especially intracranial hemorrhage).105,106 Observational and trial
absolute risk of stroke. Low-risk patients are those aged younger than data have recently been reviewed by a European consensus group,107
65 years with no stroke risk factors (a CHA2DS2-VASc score of 0 for which recommends an initial period of triple therapy (OAC, aspirin,
males or 1 for females). Subsequently, patients with at least 1 addi- clopidogrel) followed by OAC plus a single antiplatelet drug
tional stroke risk factor can be offered effective stroke prevention, (preferably clopidogrel). After 1 year, the patients can receive OAC
which is OAC, either NOAC or VKA (Figure). Thus, this includes male alone. In some patients at low thrombotic risk or high-bleeding risk,
AF patients with a CHA2DS2-VASc score of 1 and a score of 2 or higher the initial triple therapy period could be substituted with OAC plus
for everyone else, irrespective of other risk stratification enhance- clopidogrel. In guidelines in which OAC is mentioned, this could re-
ments. In terms of clinical application and practicality, such an ap- fer to a NOAC or well-managed VKA (with TTR >70%). Third-
proach makes clinical decision making very simple and practical, with- generation drug eluting stents are also the preferred option, and with
out the necessity for complex investigations that may delay improvements in stent technology, the period during which addi-
treatment decisions. tional dual antiplatelet drugs are needed can be shortened.
Does use of risk scores in guidelines to aid treatment deci-
sions improve outcomes? As mentioned above, use of the Bridging
CHA2DS2-VASc score enables refinement of stroke risk stratifica- Bridging refers to the use of a short-acting anticoagulants, eg,
tion and enables identification of those AF patients still at appre- low-molecular-weight heparin, during the interruption of a
ciable stroke risk who were previously considered at low risk longer-acting oral agent. Bridging anticoagulation is often used
when using other risk schemes.31,45,94-96 The net clinical benefit is during anticoagulation interruptions for operative procedures
also positive for patients with a CHA 2 DS 2 -VASc score of 1 or and is associated with a higher risk of bleeding and adverse
higher, whether from real-world cohorts or modeling analyses events. Indeed, observational and trial data do not support the
with the NOACs.85-87 Indeed, modeling analyses show that use of use of routine bridging given the higher bleeding risks and no dif-
the ESC guidelines would translate to a reduction in additional ference in thromboembolism when compared with an uninter-
thromboembolic, mortality, and bleeding events in Europe rupted OAC strategy; thus, the latter approach is used for most
and Asia, compared with older guideline strategies.88,97 Finally, procedures.108 However, bridging may perhaps be considered for
guideline adherent treatment is associated with improved effi- some very high-risk subgroups (eg, prosthetic mechanical heart
cacy and safety outcomes, whether from observational98,99 or valves); thus, risk stratification is important.109 NOACs may be
trial-based100 cohorts. beneficial based on promising data from post hoc analysis from
randomized trials.110 Nevertheless, the use of NOACs requires
understanding of their pharmacology, given their fast onset and
offset of action, as well as the effect of renal function.
Specific Considerations
Chronic Renal Failure Left Atrial Appendage Occlusion
Atrial fibrillation patients with renal impairment represent a high- The concept of left atrial appendage (LAA) occlusion arises be-
risk group, demonstrating a greater risk of stroke, myocardial infarc- cause most atrial thrombi arise from the LAA, in patients with AF.
tion, and major bleeding.101 Approximately 20% of AF patients show Thus, over the last decade, the LAA has been removed during car-
a significant reduction in renal function over 2 years, and even nor- diac surgery, or more recently subject to percutaneous occlusion with
mal or mild renal impairment at baseline did not preclude some pa- a variety of devices. For example, the WATCHMAN device has been
tients from progressing to severe renal impairment.102 tested in a randomized trial (PROTECT-AF),111 which showed that the
Patients with severe renal impairment were largely excluded LAA occlusion device was noninferior for safety and efficacy to stan-
from randomized trials, and all NOACs do have some degree of re- dard therapy (warfarin). There was an initial excess of complica-
nal excretion (the highest being with dabigatran, whereby the drug tions (eg, pericardial tamponade) in the device intervention group
is also removed by hemodialysis). Initial observational studies were than in the control group (7.4 vs 4.4 per 100 patient-years; RR, 1.69,
conflicting suggesting that stroke prevention with warfarin was off- 95% CI, 1.01-3.19), but as operators became more experienced this
set by a high rate of major bleeding, especially in patients requiring diminished, as has been seen in subsequent registries. These find-
dialysis.103 The most recent data from the Swedish AF cohort study37 ings were reinforced in the PREVAIL trial,112 which showed that LAA
suggests that warfarin is still associated with a positive net clinical occlusion was noninferior to warfarin for ischemic stroke preven-
benefit, balancing stroke reduction against bleeding risk, but em- tion or systemic embolism at more than 7 days after the procedure.
phasizes the importance of a high TTR to minimize adverse effects. Other methods of LAA occlusion are also available, but random-
Similar data have been published from the Danish nationwide co- ized trials are lacking. A recent position document from the European
hort study.104 Heart Rhythm Association comprehensively reviews the literature

1958 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

and suggests practical points for its use.113 For now, LAA occlusion well as treatment adherence.117 Particularly with NOACs (with their
may be considered in patients with a high stroke risk and contrain- short half-life), poor adherence can be associated with a greater risk
dications for long- term OAC.4 of stroke and mortality.118 Simply to avoid 1 stroke, patients are even
prepared to sustain 4 major bleeding events.119 In contrast, physi-
Cardioversion cians are more concerned with avoiding major bleeding, even at the
Cardioversion of AF requires OAC for a minimum of 3 weeks expense of patients sustaining strokes.120,121
before cardioversion, and OAC continued during and after cardio-
version. Recent data on the NOACs may support their use to facili-
tate cardioversion and may be associated with less delay to the
Conclusions
procedure, in contrast to using VKAs for which delays or cancella-
tions were common pending a therapeutic INR.114-116 Irrespective Substantial recent advances are evident in the area of stroke pre-
of apparent successful rhythm control (whether by cardioversion vention in AF. Treatment has changed greatly, recognizing that stroke
or ablation), patients should continue taking OAC in the presence prevention is the cornerstone of AF management. Indeed, follow-
of stroke risk factors. Rhythm control should not be performed as ing the initial identification of low-risk patients using the CHA2DS2-
a basis to stop OAC. VASc score, physicians can then offer effective thromboprophy-
laxis (which is OAC) to patients with at least 1 additional stroke risk
Patient Values and Preferences factor. Clinicians should not offer aspirin for stroke prevention in AF.
Increasing awareness is evident on the importance of patient atti- With this approach, clinicians can improve AF patient outcomes and
tudes and preferences in deciding on managing strategies for AF, as reduce the huge burden of AF-related stroke.

ARTICLE INFORMATION patients with atrial fibrillation: a report of the atrial fibrillation and CHA2DS2-VASc score of 1. Pacing
Conflict of Interest Disclosures: Both authors American College of Cardiology/American Heart Clin Electrophysiol. 2014;37(11):1442-1447.
have completed and submitted the ICMJE Form for Association Task Force on Practice Guidelines and 16. Anandasundaram B, Lane DA, Apostolakis S,
Disclosure of Potential Conflicts of Interest. Dr Lip the Heart Rhythm Society. J Am Coll Cardiol. 2014; Lip GY. The impact of atherosclerotic vascular disease
reported that he has served as a consultant for 64(21):e1-e76. in predicting a stroke, thromboembolism and
Bayer Healthcare, Merck, AstraZeneca, sanofi- 6. Watson T, Shantsila E, Lip GY. Mechanisms of mortality in atrial fibrillation patients: a systematic
aventis, Bristol-Myers Squibb/Pfizer, and thrombogenesis in atrial fibrillation: Virchows triad review. J Thromb Haemost. 2013;11(5):975-987.
Boehringer Ingelheim, and has been on the speaker revisited. Lancet. 2009;373(9658):155-166. 17. The Stroke Prevention in Atrial Fibrillation
bureaus for Bayer, Bristol-Myers Squibb/Pfizer, 7. Sweeney KG, Gray DP, Steele R, Evans P. Use of Investigators Committee on Echocardiography.
Boehringer Ingelheim, and sanofi-aventis. Dr Lane warfarin in non-rheumatic atrial fibrillation: Transesophageal echocardiographic correlates of
reported that she has received an investigator a commentary from general practice. Br J Gen Pract. thromboembolism in high-risk patients with
initiated educational grants from Bayer Healthcare 1995;45(392):153-158. nonvalvular atrial fibrillation. Ann Intern Med. 1998;
and Boehringer Ingelheim and has been on the 128(8):639-647.
speakers bureau for Boehringer Ingelheim, Bayer, 8. Stroke Risk in Atrial Fibrillation Working Group.
and Bristol-Myers Squibb/Pfizer. Independent predictors of stroke in patients with 18. Friberg L, Rosenqvist M, Lip GY. Evaluation of
atrial fibrillation: a systematic review. Neurology. risk stratification schemes for ischaemic stroke and
Correction: This article was corrected July 15, 2015, 2007;69(6):546-554. bleeding in 182 678 patients with atrial fibrillation:
to correct a study name mentioned in the text. the Swedish Atrial Fibrillation cohort study. Eur
9. Pisters R, Lane DA, Marin F, Camm AJ, Lip GY.
Submissions:We encourage authors to submit Stroke and thromboembolism in atrial fibrillation. Heart J. 2012;33(12):1500-1510.
papers for consideration as a Review. Please Circ J. 2012;76(10):2289-2304. 19. Lin LY, Lee CH, Yu CC, et al. Risk factors and
contact Mary McGrae McDermott, MD, at incidence of ischemic stroke in Taiwanese with
mdm608@northwestern.edu. 10. Wagstaff AJ, Overvad TF, Lip GY, Lane DA.
Is female sex a risk factor for stroke and nonvalvular atrial fibrillationa nationwide
thromboembolism in patients with atrial database analysis. Atherosclerosis. 2011;217(1):292-
REFERENCES 295.
fibrillation? a systematic review and meta-analysis.
1. Bjrck S, Palaszewski B, Friberg L, Bergfeldt L. QJM. 2014;107(12):955-967. 20. Hart RG, Pearce LA, McBride R, Rothbart RM,
Atrial fibrillation, stroke risk, and warfarin therapy Asinger RW; The Stroke Prevention in Atrial
revisited: a population-based study. Stroke. 2013; 11. Friberg L, Benson L, Rosenqvist M, Lip GY.
Assessment of female sex as a risk factor in atrial Fibrillation (SPAF) Investigators. Factors associated
44(11):3103-3108. with ischemic stroke during aspirin therapy in atrial
fibrillation in Sweden: nationwide retrospective
2. Lip GY. Stroke and bleeding risk assessment in cohort study. BMJ. 2012;344:e3522. fibrillation: analysis of 2012 participants in the SPAF
atrial fibrillation: when, how, and why? Eur Heart J. I-III clinical trials. Stroke. 1999;30(6):1223-1229.
2013;34(14):1041-1049. 12. Mikkelsen AP, Lindhardsen J, Lip GY, Gislason
GH, Torp-Pedersen C, Olesen JB. Female sex as a 21. Risk factors for stroke and efficacy of
3. Camm AJ, Lip GY, De Caterina R, et al; ESC risk factor for stroke in atrial fibrillation: antithrombotic therapy in atrial fibrillation. Analysis
Committee for Practice Guidelines-CPG; Document a nationwide cohort study. J Thromb Haemost. of pooled data from five randomized controlled
Reviewers. 2012 focused update of the ESC 2012;10(9):1745-1751. trials. Arch Intern Med. 1994;154(13):1449-1457.
Guidelines for the management of atrial fibrillation: 22. Gage BF, Waterman AD, Shannon W, Boechler
an update of the 2010 ESC Guidelines for the 13. Olesen JB, Fauchier L, Lane DA, Taillandier S,
Lip GY. Risk factors for stroke and M, Rich MW, Radford MJ. Validation of clinical
management of atrial fibrillation. Europace. 2012;14 classification schemes for predicting stroke: results
(10):1385-1413. thromboembolism in relation to age among
patients with atrial fibrillation: the Loire Valley Atrial from the National Registry of Atrial Fibrillation. JAMA.
4. National Institute for Health and Care Fibrillation Project. Chest. 2012;141(1):147-153. 2001;285(22):2864-2870.
Excellence. Atrial fibrillation: the management of 23. Wang TJ, Massaro JM, Levy D, et al. A risk score
atrial fibrillation. http://guidance.nice.org.uk/CG180. 14. Chao TF, Liu CJ, Chen SJ, et al. Atrial fibrillation
and the risk of ischemic stroke: does it still matter in for predicting stroke or death in individuals with
Published June 2014. Accessed June 18, 2014. new-onset atrial fibrillation in the community: the
patients with a CHA2DS2-VASc score of 0 or 1? Stroke.
5. January CT, Wann LS, Alpert JS, et al; American 2012;43(10):2551-2555. Framingham Heart Study. JAMA. 2003;290(8):
College of Cardiology/American Heart Association 1049-1056.
Task Force on Practice Guidelines. 2014 15. Huang D, Anguo L, Yue WS, Yin L, Tse HF, Siu CW.
Refinement of ischemic stroke risk in patients with 24. van Walraven C, Hart RG, Wells GA, et al.
AHA/ACC/HRS guideline for the management of A clinical prediction rule to identify patients with

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1959

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

atrial fibrillation and a low risk for stroke while 37. Friberg L, Benson L, Lip GY. Balancing stroke major bleeding in outpatients treated with warfarin.
taking aspirin. Arch Intern Med. 2003;163(8):936- and bleeding risks in patients with atrial fibrillation Am J Med. 1998;105(2):91-99.
943. and renal failure: the Swedish Atrial Fibrillation 49. Kuijer PM, Hutten BA, Prins MH, Bller HR.
25. Rietbrock S, Heeley E, Plumb J, van Staa T. Cohort study. Eur Heart J. 2015;36(5):297-306. Prediction of the risk of bleeding during
Chronic atrial fibrillation: Incidence, prevalence, and 38. Piccini JP, Stevens SR, Chang Y, et al; ROCKET anticoagulant treatment for venous
prediction of stroke using the congestive heart AF Steering Committee and Investigators. Renal thromboembolism. Arch Intern Med. 1999;159(5):
failure, hypertension, age >75, diabetes mellitus, dysfunction as a predictor of stroke and systemic 457-460.
and prior stroke or transient ischemic attack embolism in patients with nonvalvular atrial 50. Gage BF, Yan Y, Milligan PE, et al. Clinical
(CHADS2) risk stratification scheme. Am Heart J. fibrillation: validation of the R(2)CHADS(2) index in classification schemes for predicting hemorrhage:
2008;156(1):57-64. the ROCKET AF (Rivaroxaban Once-daily, oral, results from the National Registry of Atrial
26. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns Direct Factor Xa Inhibition Compared With Vitamin Fibrillation (NRAF). Am Heart J. 2006;151(3):713-719.
HJ. Refining clinical risk stratification for predicting K Antagonism for Prevention of Stroke and
Embolism Trial in Atrial Fibrillation) and ATRIA 51. Shireman TI, Mahnken JD, Howard PA,
stroke and thromboembolism in atrial fibrillation Kresowik TF, Hou Q, Ellerbeck EF. Development of a
using a novel risk factor-based approach: the Euro (AnTicoagulation and Risk Factors In Atrial
Fibrillation) study cohorts. Circulation. 2013;127(2): contemporary bleeding risk model for elderly
Heart Survey on Atrial Fibrillation. Chest. 2010; warfarin recipients. Chest. 2006;130(5):1390-1396.
137(2):263-272. 224-232.
39. Chao TF, Liu CJ, Wang KL, et al. Using the 52. Pisters R, Lane DA, Nieuwlaat R, de Vos CB,
27. Hippisley-Cox J, Coupland C, Brindle P. Crijns HJ, Lip GY. A novel user-friendly score
Derivation and validation of QStroke score for CHA2DS2-VASc score for refining stroke risk
stratification in low-risk Asian patients with atrial (HAS-BLED) to assess 1-year risk of major bleeding
predicting risk of ischaemic stroke in primary care in patients with atrial fibrillation: the Euro Heart
and comparison with other risk scores: fibrillation. J Am Coll Cardiol. 2014;64(16):1658-1665.
Survey. Chest. 2010;138(5):1093-1100.
a prospective open cohort study. BMJ. 2013;346: 40. Lip GY, Nielsen PB, Skjoth F, Lane DA,
f2573. Rasmussen LH, Larsen TB. The value of the 53. Fang MC, Go AS, Chang Y, et al. A new risk
European Society of Cardiology guidelines for scheme to predict warfarin-associated hemorrhage:
28. Singer DE, Chang Y, Borowsky LH, et al. A new The ATRIA (Anticoagulation and Risk Factors in
risk scheme to predict ischemic stroke and other refining stroke risk stratification in patients with
atrial fibrillation categorized as low risk using the Atrial Fibrillation) Study. J Am Coll Cardiol. 2011;58
thromboembolism in atrial fibrillation: the ATRIA (4):395-401.
study stroke risk score. J Am Heart Assoc. 2013;2(3): anticoagulation and risk factors in atrial fibrillation
e000250. stroke score: a nationwide cohort study. Chest. 54. Roldn V, Marn F, Manzano-Fernndez S, et al.
2014;146(5):1337-1346. The HAS-BLED score has better prediction accuracy
29. You JJ, Singer DE, Howard PA, et al; American for major bleeding than CHADS2 or CHA2DS2-VASc
College of Chest Physicians. Antithrombotic 41. Hijazi Z, Oldgren J, Andersson U, et al. Cardiac
biomarkers are associated with an increased risk of scores in anticoagulated patients with atrial
therapy for atrial fibrillation: Antithrombotic fibrillation. J Am Coll Cardiol. 2013;62(23):2199-2204.
Therapy and Prevention of Thrombosis, 9th ed: stroke and death in patients with atrial fibrillation:
American College of Chest Physicians a Randomized Evaluation of Long-term 55. Apostolakis S, Lane DA, Buller H, Lip GY.
Evidence-Based Clinical Practice Guidelines. Chest. Anticoagulation Therapy (RE-LY) substudy. Comparison of the CHADS2, CHA2DS2-VASc and
2012;141(2)(suppl):e531S-e575S. Circulation. 2012;125(13):1605-1616. HAS-BLED scores for the prediction of clinically
42. Conway DS, Pearce LA, Chin BS, Hart RG, relevant bleeding in anticoagulated patients with
30. Verma A, Cairns JA, Mitchell LB, et al; CCS atrial fibrillation: the AMADEUS trial. Thromb
Atrial Fibrillation Guidelines Committee. 2014 Lip GY. Prognostic value of plasma von Willebrand
factor and soluble P-selectin as indices of Haemost. 2013;110(5):1074-1079.
focused update of the Canadian Cardiovascular
Society Guidelines for the management of atrial endothelial damage and platelet activation in 994 56. Lip GY, Banerjee A, Lagrenade I, Lane DA,
fibrillation. Can J Cardiol. 2014;30(10):1114-1130. patients with nonvalvular atrial fibrillation. Circulation. Taillandier S, Fauchier L. Assessing the risk of
2003;107(25):3141-3145. bleeding in patients with atrial fibrillation: the Loire
31. Olesen JB, Torp-Pedersen C, Hansen ML, Valley Atrial Fibrillation project. Circ Arrhythm
Lip GY. The value of the CHA2DS2-VASc score for 43. Lip GY, Lane D, Van Walraven C, Hart RG.
Additive role of plasma von Willebrand factor levels Electrophysiol. 2012;5(5):941-948.
refining stroke risk stratification in patients with
atrial fibrillation with a CHADS2 score 0-1: to clinical factors for risk stratification of patients 57. Apostolakis S, Lane DA, Guo Y, Buller H, Lip GY.
a nationwide cohort study. Thromb Haemost. 2012; with atrial fibrillation. Stroke. 2006;37(9):2294- Performance of the HEMORR 2 HAGES, ATRIA, and
107(6):1172-1179. 2300. HAS-BLED bleeding risk-prediction scores in
44. Go AS, Fang MC, Udaltsova N, et al; ATRIA nonwarfarin anticoagulated atrial fibrillation
32. Zhu W-G, Xiong Q-M, Hong K. Meta-Analysis of patients. J Am Coll Cardiol. 2013;61(3):386-387.
CHADS2 versus CHA2DS2-VASc for predicting stroke Study Investigators. Impact of proteinuria and
and thromboembolism in atrial fibrillation patients glomerular filtration rate on risk of 58. Omran H, Bauersachs R, Rbenacker S, Goss F,
independent of anticoagulation. Tex Heart Inst J. thromboembolism in atrial fibrillation: the Hammerstingl C. The HAS-BLED score predicts
2015;42(1):6-15. anticoagulation and risk factors in atrial fibrillation bleedings during bridging of chronic oral
(ATRIA) study. Circulation. 2009;119(10):1363-1369. anticoagulation: results from the national
33. Karthikeyan G, Eikelboom JW. The CHADS2 multicentre BNK Online Bridging Registry
score for stroke risk stratification in atrial 45. Olesen JB, Torp-Pedersen C. Stroke risk in atrial
fibrillation: do we anticoagulate CHADS2 or (BORDER). Thromb Haemost. 2012;108(1):65-73.
fibrillationfriend or foe? Thromb Haemost. 2010;
104(1):45-48. CHA2DS2-VASc 1, or higher? [published online 59. Smith JG, Wieloch M, Koul S, et al. Triple
March 6, 2015]. Thromb Haemost. 2015;113(2). antithrombotic therapy following an acute coronary
34. Siu CW, Lip GY, Lam KF, Tse HF. Risk of stroke syndrome: prevalence, outcomes and prognostic
and intracranial hemorrhage in 9727 Chinese with 46. Lip GY, Andreotti F, Fauchier L, et al; European
Heart Rhythm Association. Bleeding risk utility of the HAS-BLED score. EuroIntervention.
atrial fibrillation in Hong Kong. Heart Rhythm. 2014; 2012;8(6):672-678.
11(8):1401-1408. assessment and management in atrial fibrillation
patients. Executive Summary of a Position 60. Eckman MH, Wise RE, Speer B, et al.
35. Guo Y, Apostolakis S, Blann AD, et al. Validation Document from the European Heart Rhythm Integrating real-time clinical information to provide
of contemporary stroke and bleeding risk Association [EHRA], endorsed by the European estimates of net clinical benefit of antithrombotic
stratification scores in non-anticoagulated Chinese Society of Cardiology [ESC] Working Group on therapy for patients with atrial fibrillation. Circ
patients with atrial fibrillation. Int J Cardiol. 2013; Thrombosis. Thromb Haemost. 2011;106(6):997-1011. Cardiovasc Qual Outcomes. 2014;7(5):680-686.
168(2):904-909.
47. Marcucci M, Lip GY, Nieuwlaat R, Pisters R, 61. Donz J, Clair C, Hug B, et al. Risk of falls and
36. Taillandier S, Olesen JB, Clmenty N, et al. Crijns HJ, Iorio A. Stroke and bleeding risk major bleeds in patients on oral anticoagulation
Prognosis in patients with atrial fibrillation and co-distribution in real-world patients with atrial therapy. Am J Med. 2012;125(8):773-778.
CHA2DS2-VASc Score = 0 in a community-based fibrillation: the Euro Heart Survey. Am J Med.
cohort study. J Cardiovasc Electrophysiol. 2012;23 62. Man-Son-Hing M, Nichol G, Lau A, Laupacis A.
2014;127(10):979-986.e2, e972. Choosing antithrombotic therapy for elderly
(7):708-713.
48. Beyth RJ, Quinn LM, Landefeld CS. Prospective patients with atrial fibrillation who are at risk for
evaluation of an index for predicting the risk of falls. Arch Intern Med. 1999;159(7):677-685.

1960 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Stroke Prevention in Atrial Fibrillation Review Clinical Review & Education

63. Banerjee A, Clementy N, Haguenoer K, fibrillation: early effects on ischaemic strokes. Eur 91. Ogilvie IM, Newton N, Welner SA, Cowell W,
Fauchier L, Lip GY. Prior history of falls and risk of Heart J. 2014;35(28):1881-1887. Lip GY. Underuse of oral anticoagulants in atrial
outcomes in atrial fibrillation: the Loire Valley Atrial 78. Apostolakis S, Sullivan RM, Olshansky B, Lip GY. fibrillation: a systematic review. Am J Med. 2010;123
Fibrillation Project. Am J Med. 2014;127(10):972-978. Factors affecting quality of anticoagulation control (7):638-645.e4.
64. Banerjee A, Fauchier L, Bernard-Brunet A, among patients with atrial fibrillation on warfarin: 92. Chao TF, Liu CJ, Wang KL, et al. Should atrial
Clementy N, Lip GY. Composite risk scores and the SAMe-TT2R2 score. Chest. 2013;144(5):1555-1563. fibrillation patients with 1 additional risk factor of
composite endpoints in the risk prediction of 79. Proietti M, Lip GY. Simple decision making the CHA2DS2-VASc score (beyond sex) receive oral
outcomes in anticoagulated patients with atrial between a vitamin K antagonist and non-vitamin K anticoagulation? J Am Coll Cardiol. 2015;65(7):
fibrillation: the Loire Valley Atrial Fibrillation antagonist oral anticoagulant: using the 635-642.
Project. Thromb Haemost. 2014;111(3):549-556. SAMe-TT2R2 Score [published online March 4, 93. Lip GY, Skjth F, Rasmussen LH, Larsen TB.
65. Lip GY, Lane DA, Buller H, Apostolakis S. 2015]. Eur Heart J. doi.org/10.1093/ehjcvp/pvv012. Oral anticoagulation, aspirin or no therapy in
Development of a novel composite stroke and 80. Poli D, Antonucci E, Testa S, Lip GY. A patients with nonvalvular AF with 0 or 1 stroke risk
bleeding risk score in patients with atrial fibrillation: prospective validation of the SAME-TT2R2 score: factor based on the CHA2DS2-VASc score. J Am Coll
the AMADEUS Study. Chest. 2013;144(6):1839-1847. how to identify atrial fibrillation patients who will Cardiol. 2015;65(14):1385-1394.
66. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: have good anticoagulation control on warfarin. 94. Lip GY, Nielsen PB, Skjth F, Rasmussen LH,
antithrombotic therapy to prevent stroke in Intern Emerg Med. 2014;9(4):443-447. Larsen TB. Atrial fibrillation patients categorized as
patients who have nonvalvular atrial fibrillation. 81. Lip GY, Haguenoer K, Saint-Etienne C, Fauchier not for anticoagulation according to the 2014
Ann Intern Med. 2007;146(12):857-867. L. Relationship of the SAMe-TT2R2 score to Canadian Cardiovascular Society algorithm are not
67. The-Stroke-Prevention-in-Atrial-Fibrillation- poor-quality anticoagulation, stroke, clinically low risk. Can J Cardiol. 2015;31(1):24-28.
Investigators. A differential effect of aspirin in relevant bleeding, and mortality in patients with 95. Lip GY, Nielsen PB, Skjth F, Lane DA,
prevention of stroke on atrial fibrillation. J Stroke atrial fibrillation. Chest. 2014;146(3):719-726. Rasmussen LH, Larsen TB. The value of the
Cerebrovasc Dis. 1993;3:181-188. 82. Gallego P, Roldn V, Marin F, et al. SAMe-TT2R2 European Society of Cardiology guidelines for
68. van Walraven C, Hart RG, Connolly S, et al. score, time in therapeutic range, and outcomes in refining stroke risk stratification in patients with
Effect of age on stroke prevention therapy in anticoagulated patients with atrial fibrillation. Am J atrial fibrillation categorized as low risk using the
patients with atrial fibrillation: the Atrial Fibrillation Med. 2014;127(11):1083-1088. anticoagulation and risk factors in atrial fibrillation
Investigators. Stroke. 2009;40(4):1410-1416. stroke score: a nationwide cohort study. Chest.
83. Abumuaileq RR, Abu-Assi E, Raposeiras-Roubin 2014;146(5):1337-1346.
69. Wan Y, Heneghan C, Perera R, et al. S, et al. Evaluation of SAMe-TT2R2 risk score for
Anticoagulation control and prediction of adverse predicting the quality of anticoagulation control in a 96. Nielsen PB, Chao TF. The risks of risk scores for
events in patients with atrial fibrillation: real-world cohort of patients with non-valvular stroke risk assessment in atrial fibrillation. Thromb
a systematic review. Circ Cardiovasc Qual Outcomes. atrial fibrillation on vitamin-K antagonists Haemost. doi:10.1160/TH15-03-0210.
2008;1(2):84-91. [published online February 6, 2015]. Europace. 97. Senoo K, Lane DA, Lip GY. Stroke risk reduction
70. Gallego P, Roldan V, Marn F, et al. Cessation of doi:10.1093/europace/euu353. with oral anticoagulation using CHA2DS-VASc in a
oral anticoagulation in relation to mortality and the 84. Singer DE, Chang Y, Fang MC, et al. The net Japanese AF population: a modeling analysis. Int J
risk of thrombotic events in patients with atrial clinical benefit of warfarin anticoagulation in atrial Cardiol. 2015;181:247-254.
fibrillation. Thromb Haemost. 2013;110(6):1189-1198. fibrillation. Ann Intern Med. 2009;151(5):297-305. 98. Gorin L, Fauchier L, Nonin E, Charbonnier B,
71. De Caterina R, Husted S, Wallentin L, et al. 85. Olesen JB, Lip GY, Lindhardsen J, et al. Risks of Babuty D, Lip GY. Prognosis and guideline-adherent
Vitamin K antagonists in heart disease: current thromboembolism and bleeding with antithrombotic treatment in patients with atrial
status and perspectives (Section III). Position paper thromboprophylaxis in patients with atrial fibrillation and atrial flutter: implications of
of the ESC Working Group on ThrombosisTask fibrillation: a net clinical benefit analysis using a real undertreatment and overtreatment in real-life
Force on Anticoagulants in Heart Disease. Thromb world nationwide cohort study. Thromb Haemost. clinical practice: the Loire Valley Atrial Fibrillation
Haemost. 2013;110(6):1087-1107. 2011;106(4):739-749. Project. Chest. 2011;140(4):911-917.

72. Ruff CT, Giugliano RP, Braunwald E, et al. 86. Friberg L, Rosenqvist M, Lip GY. Net clinical 99. Nieuwlaat R, Olsson SB, Lip GY, et al; Euro
Comparison of the efficacy and safety of new oral benefit of warfarin in patients with atrial fibrillation: Heart Survey Investigators; The Euro Heart Survey
anticoagulants with warfarin in patients with atrial a report from the Swedish atrial fibrillation cohort on Atrial Fibrillation. Guideline-adherent
fibrillation: a meta-analysis of randomised trials. study. Circulation. 2012;125(19):2298-2307. antithrombotic treatment is associated with
Lancet. 2014;383(9921):955-962. improved outcomes compared with
87. Banerjee A, Lane DA, Torp-Pedersen C, Lip GY. undertreatment in high-risk patients with atrial
73. Skjth F, Larsen TB, Rasmussen LH. Indirect Net clinical benefit of new oral anticoagulants fibrillation. Am Heart J. 2007;153(6):1006-1012.
comparison studiesare they useful? insights from (dabigatran, rivaroxaban, apixaban) versus no
the novel oral anticoagulants for stroke prevention treatment in a real world atrial fibrillation 100. Lip GY, Clemens A, Noack H, Ferreira J,
in atrial fibrillation. Thromb Haemost. 2012;108 population: a modelling analysis based on a Connolly SJ, Yusuf S. Patient outcomes using the
(3):405-406. nationwide cohort study. Thromb Haemost. 2012; European label for dabigatran. A post-hoc analysis
107(3):584-589. from the RE-LY database. Thromb Haemost.
74. Skjth F, Larsen TB, Rasmussen LH, Lip GY. 2014;111(5):933-942.
Efficacy and safety of edoxaban in comparison with 88. Pisters R, Nieuwlaat R, Lane DA, Crijns HJ,
dabigatran, rivaroxaban and apixaban for stroke Lip GY. Potential net clinical benefit of 101. Olesen JB, Lip GY, Kamper AL, et al. Stroke and
prevention in atrial fibrillation. Thromb Haemost. population-wide implementation of apixaban and bleeding in atrial fibrillation with chronic kidney
2014;111(5):981-988. dabigatran among European patients with atrial disease. N Engl J Med. 2012;367(7):625-635.

75. Verdecchia P, Angeli F, Lip GY, Reboldi G. fibrillation. Thromb Haemost. 2013;109(2):328-336. 102. Roldn V, Marn F, Fernndez H, et al. Renal
Edoxaban in the evolving scenario of non vitamin K 89. Eckman MH, Singer DE, Rosand J, Greenberg impairment in a real-life cohort of anticoagulated
antagonist oral anticoagulants imputed placebo SM. Moving the tipping point: the decision to patients with atrial fibrillation (implications for
analysis and multiple treatment comparisons. PLoS anticoagulate patients with atrial fibrillation. Circ thromboembolism and bleeding). Am J Cardiol.
One. 2014;9(6):e100478. Cardiovasc Qual Outcomes. 2011;4(1):14-21. 2013;111(8):1159-1164.

76. Verdecchia P, Angeli F, Bartolini C, et al. Safety 90. Fuster V, Rydn LE, Cannom DS, et al; 103. Ng KP, Edwards NC, Lip GY, Townend JN,
and efficacy of nonvitamin K oral anticoagulants in American College of Cardiology; American Heart Ferro CJ. Atrial fibrillation in CKD: balancing the
nonvalvular atrial fibrillation: a Bayesian Association Task Force; European Society of risks and benefits of anticoagulation. Am J Kidney Dis.
meta-analysis approach. Expert Opin Drug Saf. Cardiology Committee for Practice Guidelines; 2013;62(3):615-632.
2015;14:(1)7-20. European Heart Rhythm Association; Heart Rhythm 104. Bonde AN, Lip GY, Kamper AL, et al.
77. Azoulay L, DellAniello S, Simon TA, Renoux C, Society. ACC/AHA/ESC 2006 guidelines for the Net clinical benefit of antithrombotic therapy in
Suissa S. Initiation of warfarin in patients with atrial management of patients with atrial fibrillation. patients with atrial fibrillation and chronic kidney
Europace. 2006;8(9):651-745.

jama.com (Reprinted) JAMA May 19, 2015 Volume 313, Number 19 1961

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015


Clinical Review & Education Review Stroke Prevention in Atrial Fibrillation

disease: a nationwide observational cohort study. Evidence-Based Clinical Practice Guidelines. Chest. 116. Flaker G, Lopes RD, Al-Khatib SM, et al;
J Am Coll Cardiol. 2014;64(23):2471-2482. 2012;141(2 suppl):e326S-e350S. ARISTOTLE Committees and Investigators. Efficacy
105. Bernard A, Fauchier L, Pellegrin C, et al. 110. Douketis JD, Healey JS, Brueckmann M, et al. and safety of apixaban in patients after
Anticoagulation in patients with atrial fibrillation Perioperative bridging anticoagulation during cardioversion for atrial fibrillation: insights from the
undergoing coronary stent implantation. Thromb dabigatran or warfarin interruption among patients ARISTOTLE Trial (Rivaroxaban Once-daily, oral,
Haemost. 2013;110(3):560-568. who had an elective surgery or procedure. Direct Factor Xa Inhibition Compared With Vitamin
Substudy of the RE-LY trial. Thromb Haemost. 2015; K Antagonism for Prevention of Stroke and
106. Rubboli A, Faxon DP, Juhani Airaksinen KE, Embolism Trial in Atrial Fibrillation). J Am Coll Cardiol.
et al. The optimal management of patients on oral 113(3):625-632.
2014;63(11):1082-1087.
anticoagulation undergoing coronary artery 111. Holmes DR, Reddy VY, Turi ZG, et al; PROTECT
stenting. Thromb Haemost. 2014;112(6):1080-1087. AF Investigators. Percutaneous closure of the left 117. Lane DA, Lip GY. Patients values and
atrial appendage versus warfarin therapy for preferences for stroke prevention in atrial
107. Lip GY, Windecker S, Huber K, et al; Document fibrillation: balancing stroke and bleeding risk with
Reviewers. Management of antithrombotic therapy prevention of stroke in patients with atrial
fibrillation: a randomised non-inferiority trial. Lancet. oral anticoagulation. Thromb Haemost. 2014;111(3):
in atrial fibrillation patients presenting with acute 381-383.
coronary syndrome and/or undergoing 2009;374(9689):534-542.
percutaneous coronary or valve interventions: 112. Holmes DR Jr, Kar S, Price MJ, et al. 118. Shore S, Carey EP, Turakhia MP, et al.
a joint consensus document of the European Prospective randomized evaluation of the Adherence to dabigatran therapy and longitudinal
Society of Cardiology Working Group on Watchman left atrial appendage closure device in patient outcomes: insights from the veterans health
Thrombosis, European Heart Rhythm Association patients with atrial fibrillation versus long-term administration. Am Heart J. 2014;167(6):810-817.
(EHRA), European Association of Percutaneous warfarin therapy: the PREVAIL trial. J Am Coll Cardiol. 119. Lahaye S, Regpala S, Lacombe S, et al.
Cardiovascular Interventions (EAPCI) and European 2014;64(1):1-12. Evaluation of patients attitudes towards stroke
Association of Acute Cardiac Care (ACCA) endorsed 113. Meier B, Blaauw Y, Khattab AA, et al; prevention and bleeding risk in atrial fibrillation.
by the Heart Rhythm Society (HRS) and Asia-Pacific Document Reviewers. EHRA/EAPCI expert Thromb Haemost. 2014;111(3):465-473.
Heart Rhythm Society (APHRS). Eur Heart J. 2014; consensus statement on catheter-based left atrial 120. Devereaux PJ, Anderson DR, Gardner MJ,
35(45):3155-3179. appendage occlusion. Europace. 2014;16(10):1397- et al. Differences between perspectives of
108. Steinberg BA, Peterson ED, Kim S, et al. Use 1416. physicians and patients on anticoagulation in
and outcomes associated with bridging during 114. Cappato R, Ezekowitz MD, Klein AL, et al; patients with atrial fibrillation: observational study.
anticoagulation interruptions in patients with atrial X-VeRT Investigators. Rivaroxaban vs vitamin K BMJ. 2001;323(7323):1218-1222.
fibrillation: findings from the Outcomes Registry for antagonists for cardioversion in atrial fibrillation. 121. Rosendaal FR, Cannegieter SC, van der Meer
Better Informed Treatment of Atrial Fibrillation Eur Heart J. 2014;35(47):3346-3355. FJ, Brit E. A method to determine the optimal
(ORBIT-AF). Circulation. 2015;131(5):488-494. intensity of oral anticoagulant therapy. Thromb
115. Nagarakanti R, Ezekowitz MD, Oldgren J, et al.
109. Douketis JD, Spyropoulos AC, Spencer FA, Dabigatran versus warfarin in patients with atrial Haemost. 1993;69(3):236-239.
et al. Perioperative management of antithrombotic fibrillation: an analysis of patients undergoing
therapy and prevention of Thrombosis, 9th ed: cardioversion. Circulation. 2011;123(2):131-136.
American College of Chest Physicians

1962 JAMA May 19, 2015 Volume 313, Number 19 (Reprinted) jama.com

Copyright 2015 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ by a STANFORD Univ Med Center User on 11/04/2015

You might also like