Professional Documents
Culture Documents
Abstract
Objective: To assess the perception of the risk of stroke and the risks and benefits of oral
anticoagulation (OAC) in patients with atrial fibrillation (AF).
Patients and Methods: Consecutive patients with chronic AF who presented for an outpatient
cardiology visit or were admitted to a noncritical care cardiology ward service from September 15
through December 20, 2017, were invited to participate in this survey. Participants were asked to
estimate their stroke risk without OAC and bleeding risk with OAC using a quantitative risk scale. The
reported values were compared with subjectively estimated risks derived from the CHA2DS2-VASc and
HAS-BLED scores. Similarly, we compared patient perception of the stroke risk reduction afforded
with OAC compared with what is reported in the literature.
Results: A total of 227 patients were included in the analysis. The mean SD CHA2DS2-VASc score
was 4.31.6, and HAS-BLED score was 2.31.2. Atrial fibrillation was paroxysmal in 53.3% and
persistent/permanent in 46.7%. There was a negligible correlation between patient perceived and
estimated risk of stroke (r¼0.07; P¼.32), and bleeding (r¼0.16; P¼.02). Most patients overestimated
their risks of stroke and bleeding: 120 patients (52.9%) perceived an annual stroke risk greater than
20%, and 115 (53.5%) perceived an annual bleeding risk with OAC greater than 10%. Most patients
(n¼204; 89.9%) perceived that OAC would reduce their annual stroke risk by at least 50%.
Conclusion: Perceived risks of stroke and bleeding are markedly overestimated in most patients with
AF. Further research is needed to discern the root causes and to identify effective methods of bridging
this alarming disparity.
ª 2018 Mayo Foundation for Medical Education and Research n Mayo Clin Proc. 2019;nn(n):1-9
S
troke prevention is a cornerstone in bleeding and is associated with considerable
the management of patients with cost, lifestyle changes, and strict compliance
atrial fibrillation (AF). Current guide- requirements.1 Hence, patient understanding From the Division of Car-
diology, West Virginia
lines recommend lifelong oral anticoagula- of the role of OAC in stroke prevention and University, Morgantown.
tion (OAC) to reduce the risk of ischemic its accompanied risks and benefits is key.
stroke in patients with a CHA2DS2-VASc Nonetheless, several studies have reported
(congestive heart failure/left ventricular ejec- that patients with AF have major gaps in
tion fraction 40%, hypertension, age 75 knowledge of their condition, the treatment
years, diabetes mellitus, stroke/transient they are taking, and the potential alternatives
ischemic attack/thromboembolism history, despite being treated for AF for several
vascular disease, age 65-74 years, female years.2-4 We conducted a cross-sectional
sex) score of at least 2.1 However, long- study at a tertiary academic center to assess
term OAC carries a significant risk of patient perception of the risks and benefits
of OAC for stroke prevention in the setting reduction with OAC, risk of bleeding while
of AF, and we compared patient perceived receiving OAC, whether they are taking
and estimated annual risks of stroke and OACs, reasons for taking or not taking
bleeding. OACs for stroke prevention, and their
awareness of the different options for stroke
PATIENTS AND METHODS prevention in the setting of AF.
For a more quantitative assessment we
Study Design, Setting, and Participants asked patients to estimate their annual risk
The study was conducted at Ruby Memorial of stroke without prevention and the risk
Hospital, the largest teaching hospital for reduction in stroke and the risk of bleeding
West Virginia University School of Medicine. with OAC use. For this, we provided multi-
From September 15 through December 20, ple choices of customized risk categories.
2017, we invited consecutive patients with These categories were carefully developed
AF presenting to a cardiology clinic (80%) using the following methods: (1) Annual
or admitted to a low-acuity nonecritical risk of stroke: The CHA2DS2-VASc score is
care inpatient medicine/cardiology service the most-used risk score for stroke risk pre-
(20%) to participate in an interview assess- diction in patients with AF. The maximum
ing their understanding of stroke prevention annual stroke risk estimated by the
in AF and the risks and benefits associated CHA2DS2-VASc score is 12.2% for a score
with OAC. After informed consent was ob- of 9. We expected that some patients will
tained, trained research personnel conducted overestimate their stroke risk. Hence, we
brief structured interviews with the patients used the following stroke risk categories in
at the end of their clinic visit in the outpa- the questionnaire: 1% to 5%, 6% to 10%,
tient setting and once they were clinically 11% to 20%, 21% to 50%, and greater than
stable in the inpatient setting. Demographic 50%. The former 3 categories corresponded
and clinical data were extracted from elec- with CHA2DS2-VASc scores of 1 to 4, 5 and
tronic medical records after the interview. 6, and 7 or greater, respectively
The person conducting the interview was (Supplemental Table 1, available online at
blinded to the clinical information at the http://www.mayoclinicproceedings.org). The
time of the interview. The institutional re- latter 2 categories do not correspond with a
view board of Ruby Memorial Hospital CHA2DS2-VASc score but were created to pro-
approved this study. vide patients with a full spectrum of options to
indicate their own perception of stroke risk
Study End Points and Questionnaire (from 1% to 100%). (2) Annual risk of
Structure bleeding while taking OACs: The following
We hypothesized that there is a significant risk categories were used in the survey: 1%
gap in patient understanding of the stroke to 3%, 4% to 6%, 7% to 10%, and greater
risk associated with AF and the risks and than 10%, corresponding to HAS-BLED (hy-
benefits of OAC for stroke prevention. How- pertension, abnormal renal/liver function,
ever, there are no validated questionnaires to stroke, bleeding history or predisposition,
specifically assess this issue. Therefore, we labile international normalized ratio, elderly
developed a questionnaire in collaboration [age 65 years], drugs/alcohol concomi-
with a behavioral scientist with experience tantly) scores of 1, 2 and 3, 4 and 5, and 6
in qualitative research methods and cross- or greater, respectively (Supplemental
sectional study design. The questionnaire Table 2, available online at http://www.
was intended to be simple and intuitive for mayoclinicproceedings.org). (3) Effective-
patients at a sixth-grade level of education ness of OAC for stroke prevention: OAC
and included 7 questions inquiring about with warfarin reduces the risk of ischemic
the patient’s knowledge of the increased stroke by 50% to 65%.5 The following choices
stroke risk with AF, their estimation of the were then provided to patients for them to
annual stroke risk without prevention, risk estimate the effectiveness of OAC in reducing
n n
2 Mayo Clin Proc. XXX 2019;nn(n):1-9 https://doi.org/10.1016/j.mayocp.2018.08.043
www.mayoclinicproceedings.org
RISK PERCEPTION IN PATIENTS WITH AF
stroke: 10%, 30%, 50%, 70%, and 90%. Clin- Validation of the Survey Instrument. The
ical data collected from patient medical re- study cohort (n¼227) was randomly divided
cords included demographic characteristics, into a derivation set (n¼114) and a valida-
clinical risk factors, socioeconomic status, tion set (n¼113) to examine the validity of
and all the elements required to calculate the the study’s questionnaire. The outcome of
CHA2DS2-VASc and HAS-BLED scores. the prediction models was false risk percep-
tion (inaccurate assessment) of actual risk
score (CHA2DS2-VASc and HAS-BLED).
Statistical Analyses Prediction models were developed using
Continuous data are presented as mean multivariable logistic regression analyses.
SD, and categorical data are presented as Variables included in the logistic regression
count (percentage). For categorical data, analysis included age, sex, level of education,
the association between patient perception history of bleeding, and CHA2DS2-VASc and
of stroke risk and estimated risk scores for HAS-BLED scores. The model predictive
stroke and the association between patient performance (discrimination value) was
perception of annual risk of bleeding on assessed using the C-statistic (area under the
OAC and the estimated risk scores for receiver operating characteristic curve) on
bleeding were assessed using the Pearson the validation set.
c2 test or the Spearman rank correlation
coefficient. Correlation coefficients were RESULTS
interpreted as follows: negligible correlation, A total of 287 patients participated in the
0.0-0.3; low correlation, 0.3-0.5; moderate study, of whom 60 (20.9%) were not aware
correlation, 0.5-0.8; and strong correlation, of the increased stroke risk with AF and
0.8-1.0.4 Analyses were performed using were thus excluded. This yielded a study
IBM SPSS Statistics for Windows, Version cohort of 227 patients. Mean SD patient
24.0 (IBM Corp), and all the tests were age was 72.210.2 years, and 125 patients
2-sided with a significance level of P¼.05. (55.1%) were men. Providers of AF care
were cardiologists in 94.3% of participants
Predictors of False Risk Perception. To (n¼214). Mean SD CHA2DS2-VASc score
assess the possible association of certain clin- was 4.31.6 and HAS-BLED score was
ical and demographic factors with patients’ 2.31.2. Most patients (n¼158; 69.6%) had
inaccurate (false) perceptions of their stroke a high school or less level of education
and bleeding risks, patients were grouped (Table 1). Atrial fibrillation was paroxysmal
into (1) those who accurately estimated their in 121 patients (53.3%) and persistent/perma-
risks, (2) those who had inaccurate estima- nent in 106 (46.7%). Most patients (40.5%)
tion of their risks (false perception), and had a history of bleeding, with gastrointes-
(3) those who had severe inaccurate estima- tinal bleeding being the most common
tion of their risks (severe false perception). bleeding location (53.0%). Other bleeding lo-
False risk perception was defined as cations included musculoskeletal in 8.7%,
choosing a risk category that is different genitourinary in 7.6%, intracranial in 6.5%,
than what is predicted by CHA2DS2-VASc pulmonary in 4.3%, and others in 38%.
and HAS-BLED scores. Severe false risk
perception was defined as choosing a risk Correlation Between Patient Perception and
category that is at least 2 categories different Estimated Outcomes
than what is predicted by the CHA2DS2- Annual Risk of Stroke. There was a negli-
VASc and HAS-BLED scores. The following gible correlation between patient perceived
elements were then entered into univariate and estimated risk of stroke (r¼0.07;
and multivariate regression models: age, sex, P¼.32). The estimated annual stroke risk
level of education, CHA2DS2-VASc score, based on the CHA2DS2-VASc score was 1%
HAS-BLED score, duration of AF, history of to 5%, 6% to 10%, and 11% to 20% in
bleeding, and history of stroke. 53.7%, 37.0%, and 9.3% of patients,
Mayo Clin Proc. n XXX 2019;nn(n):1-9 n https://doi.org/10.1016/j.mayocp.2018.08.043 3
www.mayoclinicproceedings.org
MAYO CLINIC PROCEEDINGS
100 100
80 80
Patients (%)
Patients (%)
60 60
40 40
20 20
0 0
Actual Perceived Actual Perceived
Stroke risk based on CHA2DS2-VASc score Bleeding risk based on HAS-BLED score
FIGURE 1. Differences in perceived vs objectively estimated annual risks of stroke (A) and bleeding (B) in patients with atrial
fibrillation.
methods: in the study by Zweiker et al, pa- patients’ decisions and potentially their
tients with newly diagnosed AF were asked compliance with the recommended preven-
to estimate their risk of stroke and bleeding tion strategy. Patients seem to accept certain
as low, intermediate, high, or very high; in measures based on a false perception of a
the present study, we surveyed patients substantial annual risk of stroke if these
with chronic AF and used numeric risk esti- measures are not taken. Would patients
mate scales. We used this method not only accept lifelong OAC if they become aware
to obtain a more quantitative assessment of that they have a greatly exaggerated percep-
patient perception but also to allow the in- tion of the risk? For example, patients with
clusion of perceived risk values beyond the AF and a CHA2DS2-VASc score of 3 have
traditional values calculated by the various an estimated annual stroke risk of 3.2%,
risk scores. The present finding that more which is an indication for lifelong OAC in
than 50% of patients perceived their annual major guidelines. Would these patients
stroke risk to be greater than 20% is con- accept long-term OAC if they correctly iden-
cerning and suggests that most patients (at tified their annual stroke risk without OAC
least in this survey) may have made an unin- as 3.2%? Similarly, most patients in this
formed decision about stroke prevention study overestimated their bleeding risk while
based on an unrealistically exaggerated receiving OAC. This excess perceived
perception of their risks. bleeding risk was the main reason for
The decision to accept or decline preven- declining OAC in the 21% of patients who
tive measures for a potential event (eg, are not taking OAC despite their physician’s
ischemic stroke in the context of AF) ulti- recommendation. Would those patients have
mately belongs to the patient. The physi- accepted OAC if they had a more accurate
cian’s role is to integrate the patient’s risk perception of the associated bleeding risk?
profile with the clinical data to provide an Another intriguing finding in the present
evidence-based recommendation to facilitate study is that the magnitude of discrepancy
a well-informed decision by that patient. between perceived and estimated risks was
However, there seems to be a significant most profound regarding estimates that are
gap between physician and patient interpre- traditionally presented in absolute values
tation of risks, which may greatly affect (eg, stroke risk) compared with those
n n
6 Mayo Clin Proc. XXX 2019;nn(n):1-9 https://doi.org/10.1016/j.mayocp.2018.08.043
www.mayoclinicproceedings.org
RISK PERCEPTION IN PATIENTS WITH AF
P=.32 r=0.066
Estimated stroke risk (%)
11-20
6-10
1-5
P=.02 r=0.16
>10
Estimated bleeding risk (%)
7-10
4-6
1-3
FIGURE 2. Distribution of perceived and objectively estimated annual risks of stroke (A) and bleeding (B)
in individual patients with atrial fibrillation.
presented in relative values (eg, benefit of Similar to what has been reported in
OAC). These provocative data imply that other studies, patients also overestimated
the current methods of sharing risk and their risk of bleeding.11 Interestingly, despite
risk reduction values with patients might that and despite the high incidence of previ-
need to be revisited. For example, a high- ous bleeding in this cohort of patients
risk CHA2DS2-VASc score of 5 has an esti- (40%), only 26% of patients were not
mated annual stroke risk of 6.7%. Compared receiving OAC. This may be partially
with aspirin, OAC with warfarin in this pa- explained by the finding that most patients
tient would reduce the risk of stroke to in this study (n¼214; 94.3%) received AF-
3.7%, a substantial risk reduction when pre- related care from a cardiology specialist. In
sented to the patient in relative terms (45%) the TREAT-AF (Retrospective Evaluation
but less remarkable when presented in abso- and Assessment of Therapies in AF) study,
lute terms (3%). Whether the way in which cardiology care was associated with
the benefits of OAC are presented (relative improved outcomes in patients with newly
vs absolute risk reduction) will affect pa- diagnosed AF, mostly mediated by the early
tients’ attitudes toward management of their prescription of OAC therapy.12
stroke risk remains to be assessed in further This study has several limitations
studies. inherent to cross-sectional cohort studies.
Mayo Clin Proc. n XXX 2019;nn(n):1-9 n https://doi.org/10.1016/j.mayocp.2018.08.043 7
www.mayoclinicproceedings.org
MAYO CLINIC PROCEEDINGS
CONCLUSION
0.2
In contemporary practice, perceived risks of
stroke and bleeding are markedly overesti-
mated in most patients with AF. Further
0.0
0.0 0.2 0.4 0.6 0.8 1.0 studies are needed to identify effective strate-
1-Specificity gies to address this profound gap between
patient perceptions and objectively esti-
FIGURE 3. Validation of the survey instrument in the validation set cohort. mated risks.
n n
8 Mayo Clin Proc. XXX 2019;nn(n):1-9 https://doi.org/10.1016/j.mayocp.2018.08.043
www.mayoclinicproceedings.org
RISK PERCEPTION IN PATIENTS WITH AF
5. van Walraven C, Hart RG, Singer DE, et al. Oral anticoagulants 11. Andrade JG, Krahn AD, Skanes AC, Purdham D, Ciaccia A,
vs aspirin in nonvalvular atrial fibrillation: an individual patient Connors S. Values and preferences of physicians and patients
meta-analysis. JAMA. 2002;288(19):2441-2448. with nonvalvular atrial fibrillation who receive oral anticoagulation
6. Dentali F, Riva N, Crowther M, Turpie AG, Lip GY, Ageno W. therapy for stroke prevention. Can J Cardiol. 2016;32(6):747-753.
Efficacy and safety of the novel oral anticoagulants in atrial fibril- 12. Perino AC, Fan J, Schmitt SK, et al. Treating specialty and out-
lation: a systematic review and meta-analysis of the literature. comes in newly Diagnosed atrial fibrillation: from the TREAT-
Circulation. 2012;126(20):2381-2391. AF Study. J Am Coll Cardiol. 2017;70(1):78-86.
7. Hsu JC, Maddox TM, Kennedy KF, et al. Oral anticoagulant ther- 13. Rothberg MB, Scherer L, Kashef MA, et al. The effect of infor-
apy prescription in patients with atrial fibrillation across the mation presentation on beliefs about the benefits of elective
spectrum of stroke risk: insights from the NCDR PINNACLE percutaneous coronary intervention. JAMA Intern Med. 2014;
Registry. JAMA Cardiol. 2016;1(1):55-62. 174(10):1623-1629.
8. Yao X, Abraham NS, Alexander GC, et al. Effect of adherence 14. Kureshi F, Jones PG, Buchanan DM, Abdallah MS, Spertus JA.
to oral anticoagulants on risk of stroke and major bleeding Variation in patients’ perceptions of elective percutaneous cor-
among patients with atrial fibrillation. J Am Heart Assoc. 2016; onary intervention in stable coronary artery disease: cross
5(2):e003074. sectional study. BMJ. 2014;349:g5309.
9. Kaufman BG, Kim S, Pieper K, et al. Disease understanding in 15. Badin A, Parr AR, Banga S, Wigant RR, Baman TS. Patients’ and
patients newly diagnosed with atrial fibrillation. Heart. 2018; physicians’ perceptions regarding the benefits of atrial fibrillation
104(6):494-501. ablation. Pacing Clin Electrophysiol. 2017;40(4):362-371.
10. Frankel DS, Parker SE, Rosenfeld LE, Gorelick PB. HRS/NSA 16. Balci KG, Balci MM, Akboga MK, et al. Perceived benefits of
2014 survey of atrial fibrillation and stroke: gaps in knowledge implantable cardioverter defibrillator implantation among heart
and perspective, opportunities for improvement. J Stroke Cere- failure patients and its relation to quality of life: a cross-sectional
brovasc Dis. 2015;24(8):1691-1700. study. Cardiol Ther. 2015;4(2):155-165.