JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO.

5, 2018

ª 2018 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

PUBLISHED BY ELSEVIER

NEW RESEARCH PAPERS

Burden of Atrial Fibrillation–Associated

Ischemic Stroke in the United States
Mohamad Alkhouli, MD,a,b Fahad Alqahtani, MD,a Sami Aljohani, MD,a Muhammad Alvi, MD,c David R. Holmes, MDb

ABSTRACT

OBJECTIVES This study aimed to investigate whether the excess morbidity and mortality of atrial fibrillation (AF)–
related stroke persists in the contemporary era.

BACKGROUND Acute ischemic stroke (AIS) in patients with AF is associated with worse outcomes than in patients
without AF. Stroke prevention strategies in patients with AF have improved over the last decade and AIS-related
mortality overall has also declined.

METHODS Patients $18 years of age who were admitted with AIS between 2003 and 2014 were identified in the
National Inpatient Sample. The study compared crude and propensity score–matched in-hospital morbidity and mortality,
cost, length of stay, and discharge dispositions between patients with and without AF.

RESULTS A total of 930,010 patients were admitted with AIS, and 18.2% of these patients had AF. The prevalence of AF in
these patients increased from 16.4% in 2003 to 20.4% in 2014, with the greatest increase observed in white and older
patients. Propensity score matching attained 2 pairs of 125,203 patients with AIS with and without AF. In these matched
cohorts, the mortality rate was higher in patients with AF (9.9% vs. 6.1%; p < 0.001). Ischemic stroke in patients with AF was
also associated with higher incidences of acute kidney injury, bleeding and infectious complications, and severe disability.
Hospital length of stay was significantly longer, and cost of care was 20% higher in patients with AF.

CONCLUSIONS The prevalence of AF in AIS patients continued to rise, particularly in white and older patients.
Despite the improvement in AIS-related morality overall, the differential negative impact of AF on the morbidity,
mortality, and cost of AIS was steady over the study’s 12-year period. (J Am Coll Cardiol EP 2018;4:618–25)
© 2018 by the American College of Cardiology Foundation.

A trial fibrillation (AF) is the most common

arrhythmia, with a prevalence of 1% to 2%
in the general population (1). The prevalence
of AF continues to rise markedly because of the aging
are more than 80 years of age compared with 10% to
15% across all age groups (3). Ischemic strokes in
patients with AF have been associated with worse
outcomes and greater cost than ischemic strokes in
of the population (2,3). It is estimated that w16 patients without AF (5). However, contemporary
million Americans will have AF by the year 2050 large-scale studies evaluating the impact of AF on
(2,4). AF is a major cause of stroke; it accounts for outcomes of patients with AIS in the United States
25% of acute ischemic strokes (AISs) in patients who are sparse. In addition, both stroke preventive

From the aDivision of Cardiology, Department of Medicine, West Virginia University School of Medicine, Morgantown, West
Virginia, USA; b Department of Cardiology, Mayo Clinic School of Medicine, Rochester, Minnesota, USA; and the cDepartment of Neurology,
West Virginia University School of Medicine, Morgantown, West Virginia, USA. The authors have reported that they have no re-
lationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the JACC: Clinical Electrophysiology author instructions page.

Manuscript received January 9, 2018; revised manuscript received February 22, 2018, accepted February 22, 2018.

ISSN 2405-500X/$36.00 https://doi.org/10.1016/j.jacep.2018.02.021

JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018 Alkhouli et al. 619
MAY 2018:618–25 AF-Associated Ischemic Stroke

strategies in patients with AF and outcomes of AIS patients admitted with AIS with or without ABBREVIATIONS

overall have improved over the last decade (6–10). AF were entered into a nearest neighbor 1:1 AND ACRONYMS

Whether these changes have resulted in a reduction variable ratio, parallel, balanced propensity-
AF = atrial fibrillation
in the excess morbidity and mortality and cost of matching model using a caliper of 0.01
AIS = acute ischemic stroke
AIS among AF patients is not known. without replacement to derive 2 propensity-
NIS = National Inpatient
This study used a large, nationwide representative matched groups of patients for comparative
Sample
sample and aimed to assess the following: 1) the analyses. Variables included in the
temporal change in the prevalence and of AF among propensity-match model are listed in Online Table 1.
patients admitted with AIS; 2) characteristics and
STATISTICAL ANALYSIS. Outcomes analysis was
clinical risk profiles of patients with AF admitted
performed using the actual 20% sample available in
with AIS; and 3) temporal trends in in-hospital
NIS, whereas the trend analysis was performed using
morbidity and mortality, cost, and resource use
the national estimate. This is a standard methodology
among propensity-matched cohorts of patients with
in other research involving the NIS. Descriptive sta-
AIS with and without AF.
tistics were presented as frequencies with percent-
ages for categorical variables. Mean, standard
METHODS deviation, median, and interquartile range were re-
ported for continuous measures. Baseline character-
STUDY DATA. The National Inpatient Sample (NIS)
istics were compared using a Pearson chi-square test
was used to derive patient-relevant information
and Fisher exact test for categorical variables and an
between January 2003 and December 2014. The NIS is
independent-samples Student’s t-test for continuous
the largest publicly available all-payer administrative
variables. Trend weights accounting for changes in
claims-based database and contains information about
the NIS sampling design are available only for data
patients’ discharges from 1,000 hospitals in 45 states.
between 1998 and 2011. For 2012 and 2013, trend
It contains clinical and resource use information on
weights were not available, and the standard survey
5 to 8 million discharges annually, with safeguards to
weights were used. Matched categorical variables
protect the privacy of individual patients, physicians,
were presented as frequencies with percentages and
and hospitals. These data are stratified to represent
compared using McNemar’s test. Matched continuous
approximately 20% of U.S. inpatient hospitalizations
variables were presented as means with standard
across different hospital and geographic regions
deviations and compared using a paired-samples
(random sample). National estimates of the entire U.S.
Student’s t-test. A type-I error of <0.50 was consid-
hospitalized population were calculated using the
ered statistically significant. All statistical analyses
Agency for Healthcare Research and Quality sampling
were performed with SPSS software version 24 (IBM
and weighting method.
Corp., Armonk, New York) and R software version
3.3.1 (R Foundation, Vienna, Austria).
STUDY POPULATION. Patients with a principal
discharge diagnosis of AIS (International Classifica- RESULTS
tion of Diseases-9th Revision-Clinical Modification
[ICD-9-CM] codes 433 to 437.1) during the study A total of 930,010 patients (representing a national
period were identified. The study population was estimate of 4,579,486 patients) were admitted with
then divided into 2 groups on the basis of the pres- AIS between 2003 and 2014, and 168,806 (18.2%) of
ence of AF (ICD-9-CM code 427.3). these patients had AF. The prevalence of AF among
patients with AIS increased by 24% from 16.4% in
STUDY ENDPOINTS. The primary endpoints of the
2003 to 20.4% in 2014 (Figures 1A to 1D). Although the
study were as follows: 1) in-hospital mortality; 2) in-
prevalence of AF among patients with AIS increased
hospital morbidities (bleeding complications, renal
among all age groups, sexes, and ethnicities, the
failure, and infectious complications); 3) surrogates
largest increase was in white patients and patients
of severe disability (nonhome discharges, gastro-
>85 years of age (Figures 1A to 1D).
stomy, mechanical ventilation, and tracheostomy);
Patients with AF were older (age 82
10 years vs.
and 4) cost of hospitalization and length of stay.
70
15 years), and they had a higher percentages of
These endpoints were compared between 2 patients
female patients (59.3% vs. 51.8%) and Caucasian pa-
with AIS with and without AF before and after pro-
tients (80.6% vs. 67.3%) compared with patients
pensity score matching.
without AF (p < 0.001). With the exception of dia-
PROPENSITY SCORE MATCHING. To reduce the het- betes and smoking, major comorbidities were more
erogeneity between the 2 cohorts in our study, prevalent in the AF group (Table 1). Thrombolytic
620 Alkhouli et al. JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018

AF-Associated Ischemic Stroke MAY 2018:618–25

F I G U R E 1 Prevalence of AF Among Patients Admitted with AIS in the United States Between 2003 and 2014

Prevalence of atrial fibrillation (AF) in (A) the overall population with acute ischemic stroke (AIS) and stratified by (B) age, (C) sex, (D) and race.

therapy and mechanical thrombectomy were more vs. 5
6; p < 0.001) and mean hospital cost was
commonly used in the AF group (5.1% vs. 3.5% and higher ($13 091
15,233 vs. $10,874
12,456; p <
0.9% vs. 0.4%, respectively; p < 0.001). After pro- 0.001) in the AF group.
pensity matching for age, sex, comorbidities, and The disparity in the rates of in-hospital mortality,
hospital characteristics (Online Table 1), baseline rates of nonhome discharges, cost of hospitalization,
characteristics became well balanced between the 2 and length of stay remained constant during the
matched groups (Table 1). study period (Figures 2A to 2D).
In-hospital mortality of AIS was significantly
higher among patients with AF compared with pa- DISCUSSION
tients without AF (10% vs. 3.9% and 9.9% vs. 6.1% in
the unmatched and propensity-matched cohorts, This contemporary nationwide analysis of 930,010
respectively; p < 0.001 for all). After propensity patients who were admitted with AIS between 2003
matching, patients with AF had higher rates of acute and 2014 features several key findings: 1) the prev-
kidney injury (8% vs. 7.4%; p < 0.001), dialysis- alence of AF among patients admitted with AIS
requiring acute kidney injury (1.8% vs. 1.6%; p ¼ increased from 16.4% in 2003 to 20.4% in 2014, with
0.001), intracranial hemorrhage (2.4% vs. 1.2%; the greatest increase occurring in white patients and
p < 0.001), gastrointestinal bleeding (1% vs. 0.6%; older (>85 years of age) patients; 2) among patients
p < 0.001), and blood transfusion (3.5% vs. 2.8%; p < admitted with AIS, rates of in-hospital death, renal
0.001). Infectious complications were also more failure, bleeding events, infectious complications
common in the AF group (Table 2). and severe disability were significantly higher in
Surrogates of severe disability were more frequent patients with AF; and 3) the excess mortality, higher
among patients with AF: nonhome discharges (62.1% cost and longer length of stay, and higher rates of
vs. 56.7%; p < 0.001), gastrostomy (7% vs. 4.5%; nonhome discharges among patients with AF
p < 0.001), and mechanical ventilation (2.4% vs. 1.2%; remained constant during the study’s 12-year
p < 0.001). Hospital length of stay was longer (6
6 duration.
JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018 Alkhouli et al. 621
MAY 2018:618–25 AF-Associated Ischemic Stroke

T A B L E 1 Characteristics of Acute Ischemic Stroke Patients Stratified by the Presence of AF

Unmatched Cohorts Matched Cohorts

Non AF (N ¼ 761,204; AF (N ¼ 168,806; Non AF (N ¼ 125,203; AF (N ¼ 125,203;

NE ¼ 3,746,733) NE ¼ 832,753) p Value NE ¼ 617,110) NE ¼ 617,110) p Value

Age, yrs 70
15 82
10 <0.001 79
11 79
10 0.002

<65 38.7 10.0 <0.001 10.7 11.0 0.002
66–84 47.9 56.6 56.9 56.6
>85 13.3 33.3 32.3 32.5
Female 51.8 59.3 <0.001 58.7 58.7 0.908
Race <0.001 0.418
Caucasian 67.3 80.6 79.5 79.8
African American 18.8 9.1 9.8 9.6
Hispanic 8.2 5.5 5.7 5.8
Medical comorbidities
Hypertension 78.5 78.1 <0.001 79.5 79.3 0.36
Diabetes 35.2 28.9 <0.001 30.3 30.1 0.248
Prior sternotomy 6.8 10.2 <0.001 10.3 10.2 0.302
Chronic lung disease 14.1 16.8 <0.001 16.9 16.8 0.474
Anemia 10.8 13.3 <0.001 14.2 14.0 0.36
Conduction abnormality 1.7 2.4 <0.001 2.6 2.5 0.076
Heart failure 9.4 28.7 <0.001 26.6 26.8 0.111
Smoking 17.0 5.3 <0.001 5.7 5.9 0.03
Vascular disease 8.3 8.8 <0.001 9.2 9.2 0.741
Chronic renal disease 10.1 13.7 <0.001 14.3 14.1 0.235
Coronary disease 13.7 19.7 <0.001 20.1 20.0 0.534
Liver disease 1.0 0.8 <0.001 0.9 0.9 0.847
Thrombolytic therapy 3.5 5.1 <0.001 5.2 5.3 0.921
Thrombectomy 0.4 0.9 <0.001 0.8 0.8 0.062
Patient and hospital demographic
Teaching hospital 44.3 42.9 <0.001 42.7 43.0 0.166
Hospital bed size <0.001 0.938
Small 12.7 13.1 13.1 13.1
Medium 25.5 25.6 26.2 26.3
Large 61.8 61.2 760.6 60.7
Rural location 14.2 15.0 <0.001 13.4 13.3 0.383
Primary payer <0.001 0.361
Medicare/Medicaid 70.1 87.3 87.4 87.1
Private including HMO 21.4 9.8 9.6 9.9
Self-pay 5.4 1.3 1.4 1.4

Values are mean
SD or %.

AF ¼ atrial fibrillation; HMO ¼ health maintenance organization; NE ¼ national estimate.

Stroke preventive strategies in AF patients have principal diagnosis in our study. These rising
improved significantly in the last decade because of numbers likely reflect the continuous aging of the
the expanding use of direct thrombin inhibitors and population, with octogenarians projected to comprise
the increasing awareness of role of stroke prevention 8% of the total U.S. population in 2050 (11). However,
in AF (7–10). Yet the burden of disease is still sub- our data also show that although the increase in AF
stantial. Our data showed that an estimated 832,753 prevalence was the greatest among patients who were
patients with AF had AIS in the United States between >85 years of age (40% of patients with AIS in this age
2003 and 2014, and the prevalence of AF among pa- group had AF in 2014), a significant but a more
tients with AIS increased by 24% during this period modest increase was noted in younger patients
(Figure 1A). During this period, the actual prevalence (Figure 1B). Whether these age-specific trends mirror
of AF in patients with AIS could have been even the trends in the prevalence and incidence of AF in
higher, however, because of: 1) the limited ascer- the general population remains uncertain.
tainment of the absence of AF during the index Remarkable sex and racial disparities in AF prev-
admission of AIS; and 2) the exclusion of patients who alence among patients with AIS were observed in our
had AIS during hospitalizations for a different study; AF was more prevalent in women than in men
622 Alkhouli et al. JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018

AF-Associated Ischemic Stroke MAY 2018:618–25

T A B L E 2 In-Hospital Morbidity and Mortality and Resource Use Among Acute Ischemic Stroke Patients With and Without AF

Unmatched Cohorts Matched Cohorts

Non AF (N ¼ 761,204; AF (N ¼ 168,806; Non AF (N ¼ 125,203; AF (N ¼ 125,203;

NE ¼ 3,746,733) NE ¼ 832,753) p Value NE ¼ 617,110) NE ¼ 617,110) p Value

In-hospital death 3.9 10.0 <0.001 6.1 9.9 <0.001

Acute kidney injury 5.7 7.5 <0.001 7.4 8.0 <0.001
New dialysis 1.6 1.6 0.70 1.6 1.8 0.001
Bleeding complications
Intracranial hemorrhage 1.0 2.3 <0.001 1.2 2.4 <0.001
Craniotomy 0.2 0.2 0.431 0.1 0.2 <0.001
Blood transfusion 2.1 3.4 <0.001 2.8 3.5 <0.001
Gastrointestinal bleeding 0.5 0.9 <0.001 0.6 1.0 <0.001
Infectious complications
Urinary tract infection 11.0 16.6 <0.001 15.3 16.7 <0.001
Acquired pneumonia 2.5 5.0 <0.001 3.6 4.9 <0.001
Sepsis 0.9 1.7 <0.001 1.2 1.9 <0.001
Severe disability surrogate
Nonhome discharge 46.0 62.4 <0.001 56.7 62.1 <0.001
Gastrostomy 3.7 6.8 <0.001 4.5 7.0 <0.001
Mechanical ventilation 1.3 2.3 <0.001 1.2 2.4 <0.001
Tracheostomy 0.06 0.11 <0.001 0.07 0.12 <0.001
LOS, days 5.93
6.12 6.04
6.12 <0.001 5.10
5.62 6.35
6.34 <0.001
Cost, $ 7,546 (5,119, 11,789) 8,683 (5,626, 14,372) <0.001 8,127 (5,319, 11,661) 9,475 (5,937, 14,216) <0.001

Values are %, mean
SD, or median (25th, 75th).

LOS ¼ length of stay; other abbreviations as in Table 1.

and in white than in black or Hispanic patients 12% in patients with and without AF, respectively
admitted with AIS (Figures 1C and 1D). The higher (p < 0.001) (18). Similar disparity in 28-day mortality
prevalence in women is consistent with the known rates was found in Japanese patients with AIS (11.3%
differential impact of sex on stroke risk in patients vs. 3.4%; p < 0.001) (19). In a large Canadian pro-
with AF (12). The lower prevalence of AF among black spective registry of 12,686 patients with AIS, patients
and Hispanic patients is, however, interesting and with AF had a 22.3% death rate at 30 days compared
deserves more investigation. Several studies had with 10.2% among patients without AF (20). However,
previously suggested a lower incidence of AF among only 3 small studies in the 1980s to 1990s assessed the
black than among white patients after adjustment for impact of AF on AIS outcomes in the United States
conventional risk factors (13–17). Some of these Wolf et al. (21) in 1983 found no difference in 30-day
studies also ruled out a major role for underdetection mortality rates in AF versus patients without AF
or less ascertainment of AF diagnosis in this (17% vs. 19%; p ¼ NS), whereas a higher mortality rate
disparity, by confirming higher rates of AF among in patients with AF was found in the other 2 studies,
nonwhite patients with implantable pacemakers and by Broderick et al. (22) (23% vs. 8%; p < 0.001) and
defibrillators compared with white patients (16,17). Lin et al. (23) (25% vs. 14%; p < 0.05). Our large study
Underlying predisposing genetic factors have been affirms the higher in-hospital mortality of AIS in pa-
proposed to explain the racial disparity in AF inci- tients with AF compared with propensity-matched
dence: Marcus et al. (15) used ancestry informative patients without AF and shows the persistence of
markers in patients enrolled in 2 large cohort studies this excess mortality over time.
and found that for every 10% increase in European Our study also confirms prior observations that
ancestry, the risk of developing AF increased by 17%, ischemic strokes in patients with AF are associated
after adjusting for demographic and clinical risk fac- with more severe disability compared with AIS in
tors. Further studies are needed to identify specific patients without AF (5,24,25). Ischemic strokes in
genes or mutations that could explain this racial patients with AF are thought to be more disabling
variance. than in patients without AF for the following reasons:
AF-related AISs have been shown to be associated 1) patients with AF are older and have a higher
with worse short and long-term mortality. In a pro- prevalence of major morbidity; and 2) AF-related
spective multi-European centers registry of 4,462 strokes may result from large and organized thrombi
patients, the 28-day mortality rate was 19.1% versus leading to larger AISs compared with AISs from other
JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018 Alkhouli et al. 623
MAY 2018:618–25 AF-Associated Ischemic Stroke

F I G U R E 2 Temporal Trends of Stroke Mortality, Resource Use, and Cost in Patients With and Without Atrial Fibrillation Between 2003 and 2014

(A) In-hospital mortality. (B) Hospital length of stay. (C) Discharge disposition. (D) Cost.

causes (24,25). Although it is not feasible to confirm hospitalizations among these patients. Further
the latter assumption in our dataset, we adjusted for studies are needed to identify possible interventions
the differences in comorbidities between patients to reduce this excess mortality in AF-associated AIS.
with and without AF with vigorous propensity score Ischemic strokes in patients with AF were also more
matching. Yet the rate of nonhome discharge, a sur- resource intensive than AISs in patients without AF. In
rogate for severe disability, was higher in AF patients the propensity-matched cohorts, AIS in the AF group
compared with patients without AF (62.1% vs. 56.7%; was associated with longer hospitalizations and a 20%
p < 0.001), and this was steady over time. In addition, increase in in-hospital cost, and both of these param-
patients with AF were 100% and 55% more likely to eters remained constant over time. The added costs of
receive mechanical ventilation and gastric tube AF-related AIS over the costs of AIS in patients without
placement following AIS compared with patients AF are, however, likely higher because of the more
without AF. In the propensity-matched cohorts, pa- frequent use of nursing homes and intermediate care
tients with AF had 100%, 67%, and 25% increases in facilities following AIS in the latter group. These
the rates of intracranial hemorrhage, gastrointestinal findings have to be taken into consideration when
bleeding, and blood transfusion, respectively, assessing the cost effectiveness of the current and
compared with patients without AF. This finding may emerging stroke prevention strategies in patients
be related to the higher percentage of patients taking with AF.
oral anticoagulant agents in the AF cohort. Infectious
complications and renal failure were also more com- FUTURE DIRECTIONS. The increasing prevalence of
mon in patients with AF, perhaps secondary to the AF in patients admitted with AIS, especially among
higher prevalence of more severe strokes and longer certain subgroups (e.g., white, older age), and the
624 Alkhouli et al. JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018

AF-Associated Ischemic Stroke MAY 2018:618–25

persistently worse outcomes of AIS in patients with cannot completely exclude the possibility of
AF are concerning. Further research is needed to improved coding and documentation over time. Fifth,
address several relevant questions. 1) Do these trends The NIS allows detailed assessment of in-hospital
represent a mere population growth, better AF outcomes. However, baseline and post-procedural
detection, or an alarming failure of our current pre- laboratory and brain imaging data are not captured.
ventive measures? 2) Is the racial disparity in AF and Moreover, data needed to calculate traditional stroke
AF-related AIS genetically inherited, or does it stem severity scale numbers and information on anti-
from potentially modifiable factors? 3) Stroke pre- coagulation and implantable AF monitoring devices
vention strategies in patients with AF reduce but do are not available in the NIS. We used surrogates of
not eliminate the risk of AIS. Does the negative stroke severity that were previously used in admin-
impact of AF on the morbidity and mortality of AIS istrative databases (28,29). Finally, the potential for
persist among patients receiving adequate oral anti- unmeasured confounders may bias the outcomes re-
coagulation? 4) Are there other measures that may sults. However, we believe that our rigorous pro-
reduce the excess morbidity of AIS in patients with pensity matching has adequately addressed this
AF (e.g., mechanical thrombectomy, left atrial selection bias.
appendage occlusion)? 5) The number of geriatric
patients with AF is rapidly growing, and this group is
CONCLUSIONS
usually characterized by a high prevalence of
The prevalence of AF among patients with AIS is ris-
bleeding risk factors and frailty. Given the increasing
ing, especially among white and older patients.
use of non-pharmacological stroke prevention stra-
Despite the improvement in AIS-related morality
tegies in recent years, will these methods affect the
overall, AIS in patients with AF continues to be asso-
overall incidence, morbidity, and mortality of AIS in
ciated with higher morbidity and mortality, cost, and
these patients?
resource use. Further studies are needed to identify
STUDY LIMITATIONS. First, the NIS is an adminis- preventive and management strategies to reduce the
trative database that gathers data for billing purposes risk of stroke-associated mortality in AF patients.
and can be limited by erroneous coding. However, we
used ICD-9-CM codes for AIS and its complications ADDRESS FOR CORRESPONDENCE: Dr. Mohamad
that have been shown to have high specificity and Alkhouli, West Virginia University Heart & Vascular
positive predictive value (26,27). In addition, the hard Institute, 1 Medical Drive, Morgantown, West Virginia
clinical endpoints used in our analysis (in-hospital 26505, USA. E-mail: [email protected].
mortality) are difficult to miscode. Second, it is not
possible using this database to ascertain the absence
PERSPECTIVES
of AF among patients in the non-AF group. Never-
theless, if certain numbers of patients in the non-AF
COMPETENCY IN MEDICAL KNOWLEDGE: In-
group were found to have AF later, this would not
hospital outcomes of AIS have improved overall in the
change the main findings of the study of higher
last decade. However, risk-adjusted morbidity and
morbidity, mortality, and cost in patients with AF.
mortality, as well as cost, remained substantially
Third, it is not possible to exclude completely the
higher in patients with AF compared with patients
possibility of AF occurrence after the stroke from this
without AF.
database. However, even when AF is first detected
after the stroke, it does not preclude the possibility of
TRANSLATIONAL OUTLOOK: Further studies are
prior undetected AF. Fourth, the increasing preva-
needed to understand this persistent excess in
lence of AF can in theory be related to the improved
morbidity and mortality of AF-associated AIS and to
AF detection techniques in contemporary practice.
identify more effective stroke preventive and man-
However, this is likely pertinent to outpatient sur-
agement strategies for these patients.
veillance and less likely to affect AF detection in pa-
tients who are hospitalized with AIS. However, we

REFERENCES

1. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ Task Force on Practice Guidelines and the Heart implications for rhythm management and stroke
ACC/HRS guideline for the management of patients Rhythm Society. J Am Coll Cardiol 2014;64:e1–76. prevention: the AnTicoagulation and Risk Factors
with atrial fibrillation: a report of the American 2. Go AS, Hylek EM, Phillips KA, et al. Prevalence in Atrial Fibrillation (ATRIA) Study. JAMA 2001;
College of Cardiology/American Heart Association of diagnosed atrial fibrillation in adults: national 285:2370–5.
JACC: CLINICAL ELECTROPHYSIOLOGY VOL. 4, NO. 5, 2018 Alkhouli et al. 625
MAY 2018:618–25 AF-Associated Ischemic Stroke

3. Miyasaka Y, Barnes ME, Gersh BJ, et al. Secular 12. Lip GY, Nieuwlaat R, Pisters R, Lane DA, fibrillation and imminence of stroke: the Fra-
trends in incidence of atrial fibrillation in Olmsted Crijns HJ. Refining clinical risk stratification for mingham study. Stroke 1983;14:664–7.
County, Minnesota, 1980 to 2000, and implica- predicting stroke and thromboembolism in atrial
22. Broderick JP, Phillips SJ, O’Fallon WM,
tions on the projections for future prevalence. fibrillation using a novel risk factor-based
Frye RL, Whisnant JP. Relationship of cardiac dis-
Circulation 2006;114:119–25. approach: the Euro Heart Survey on Atrial
ease to stroke occurrence, recurrence, and mor-
Fibrillation. Chest 2010;137:263–72.
4. Benjamin EJ, Blaha MJ, Chiuve SE, et al. Heart tality. Stroke 1992;23:1250–6.
disease and stroke statistics—2017 update: a 13. Dewland TA, Olgin JE, Vittinghoff E,
23. Lin HJ, Wolf PA, Kelly-Hayes M, et al. Stroke
report from the American Heart Association. Cir- Marcus GM. Incident atrial fibrillation among
severity in atrial fibrillation. The Framingham
culation 2017;135:e146–603. Asians, Hispanics, blacks, and whites. Circulation
Study. Stroke 1996;27:1760–4.
2013;128:2470–7.
5. Ali AN, Abdelhafiz A. Clinical and economic 24. Steger C, Pratter A, Martinek-Bregel M, et al.
implications of AF related stroke. J Atr Fibrillation 14. Alonso A, Agarwal SK, Soliman EZ, et al.
Stroke patients with atrial fibrillation have a worse
2016;8:1279. Incidence of atrial fibrillation in whites and African-
prognosis than patients without: data from the
Americans: the Atherosclerosis Risk in Communities
6. Ma J, Ward EM, Siegel RL, Jemal A. Temporal Austrian Stroke registry. Eur Heart J 2004;25:
(ARIC) study. Am Heart J 2009;158:111–7.
trends in mortality in the United States, 1734–40.
1969-2013. JAMA 2015;314:1731–9. 15. Marcus GM, Alonso A, Peralta CA, et al. Euro-
25. Tu HT, Campbell BC, Christensen S, et al.
pean ancestry as a risk factor for atrial fibrillation in
7. Chang SS, Dong JZ, Ma CS, et al. Current status Pathophysiological determinants of worse stroke
African Americans. Circulation 2010;122:2009–15.
and time trends of oral anticoagulation use among outcome in atrial fibrillation. Cerebrovasc Dis
chinese patients with nonvalvular atrial fibrilla- 16. Lau CP, Gbadebo TD, Connolly SJ, et al. Ethnic 2010;30:389–95.
tion: the Chinese Atrial Fibrillation Registry Study. differences in atrial fibrillation identified using
26. Kokotailo RA, Hill MD. Coding of stroke and
Stroke 2016;47:1803–10. implanted cardiac devices. J Cardiovasc Electro-
stroke risk factors using international classification
physiol 2013;24:381–7.
8. Patel PA, Zhao X, Fonarow GC, et al. Novel oral of diseases, revisions 9 and 10. Stroke 2005;36:
anticoagulant use among patients with atrial 17. Kamel H, Kleindorfer DO, Bhave PD, et al. 1776–81.
fibrillation hospitalized with ischemic stroke or Rates of atrial fibrillation in black versus white
27. Ramirez L, Kim-Tenser MA, Sanossian N, et al.
transient ischemic attack. Circ Cardiovasc Qual patients with pacemakers. J Am Heart Assoc 2016;
Trends in acute ischemic stroke hospitalizations in
Outcomes 2015;8:383–92. 5:e002492.
the United States. J Am Heart Assoc 2016;5:
9. Hsu JC, Maddox TM, Kennedy K, et al. Aspirin 18. Lamassa M, Di Carlo A, Pracucci G, et al. e003233.
instead of oral anticoagulant prescription in atrial Characteristics, outcome, and care of stroke
28. Moradiya Y, Levine SR. Comparison of short-
fibrillation patients at risk for stroke. J Am Coll associated with atrial fibrillation in Europe: data
term outcomes of thrombolysis for in-hospital
Cardiol 2016;67:2913–23. from a multicenter multinational hospital-based
stroke and out-of-hospital stroke in United
registry (the European Community Stroke Proj-
States. Stroke 2013;44:1903–8.
10. Katz DF, Maddox TM, Turakhia M, et al. ect). Stroke 2001;32:392–8.
Contemporary trends in oral anticoagulant pre- 29. Qureshi AI, Chaudhry SA, Hassan AE, et al.
scription in atrial fibrillation patients at low to 19. Kimura K, Minematsu K, Yamaguchi T, Japan
Thrombolytic treatment of patients with acute
moderate risk of stroke after guideline- Multicenter Stroke Investigators; Collaboration
ischemic stroke related to underlying arterial
recommended change in use of the CHADS2 to (J-MUSIC). Atrial fibrillation as a predictive factor
dissection in the United States. Arch Neurol 2011;
the CHA2DS2-VASc score for thromboembolic risk for severe stroke and early death in 15,831
68:1536–42.
assessment: analysis from the National Cardio- patients with acute ischaemic stroke. J Neurol
vascular Data Registry’s Outpatient Practice Neurosurg Psychiatry 2005;76:679–83.
Innovation and Clinical Excellence Atrial Fibrilla- 20. Saposnik G, Gladstone D, Raptis R, et al. Atrial KEY WORDS atrial fibrillation, comparative
tion Registry. Circ Cardiovasc Qual Outcomes fibrillation in ischemic stroke: predicting response outcomes, ischemic stroke
2017;10:e003476. to thrombolysis and clinical outcomes. Stroke
2013;44:99–104.
11. Colby SL, Ortman JM. Projections of the size
and composition of the U.S. population: 2014 to 21. Wolf PA, Kannel WB, McGee DL, Meeks SL, A PP END IX For a supplemental table, please
2060. Curr Popul Rep P-25 2015:P25–1143. Bharucha NE, McNamara PM. Duration of atrial see the online version of this article.