Secondary Prevention of Stroke with Warfarin in Patients with

Nonvalvular Atrial Fibrillation: Subanalysis of the

J-RHYTHM Registry

Eitaro Kodani, MD, PhD,* Hirotsugu Atarashi, MD, PhD,* Hiroshi Inoue, MD, PhD,†
Ken Okumura, MD, PhD,‡ Takeshi Yamashita, MD, PhD,§ and Hideki Origasa, PhD,‖
on behalf of the J-RHYTHM Registry Investigators

Background: Prior ischemic stroke or transient ischemic attack (TIA) is a high risk
for thromboembolism in patients with nonvalvular atrial fibrillation (NVAF). To
clarify rates of thromboembolic and hemorrhagic events, and target intensities of
warfarin for secondary prevention, a subanalysis was performed using data from
the J-RHYTHM Registry. Methods: Of 7937 outpatients with atrial fibrillation, 7406
with NVAF (men 70.8%, 69.8 ± 10.0 years) were followed for 2 years or until an
event occurred. Event rates and effect of warfarin were compared between pa-
tients with (secondary prevention) and without (primary prevention) prior stroke/TIA.
Results: Prevalence of male sex, diabetes mellitus, and mean age were higher in
the secondary prevention group, showing a higher CHADS2 (congestive heart failure,
hypertension, age 75 years or older, diabetes mellitus, and history of stroke or
TIA) score than the primary prevention group (3.5 ± 1.0 versus 1.4 ± 1.0, P < .001).
In the secondary prevention group, 93.4% of patients received warfarin and their
time in therapeutic range was 62.8%. During follow-up, thromboembolism oc-
curred more frequently in the secondary than in the primary prevention
group (2.8% versus 1.5%, P = .004), especially in patients without warfarin. Major
hemorrhage also occurred more frequently in the secondary prevention group
(3.0% versus 1.7%, P = .006). Compared with patients not taking warfarin, com-
bined rates of both events were lower at an international normalized ratio
(INR) of 1.6-2.59 in patients taking warfarin in the secondary as well as in the
primary prevention groups. Conclusions: Both thromboembolism and
major hemorrhage occurred more frequently in NVAF patients with prior ischemic

From the *Department of Internal Medicine and Cardiology, Nippon Medical School, Tama-Nagayama Hospital, Tokyo, Japan; †Second
Department of Internal Medicine, Graduate School of Medicine, University of Toyama, Toyama, Japan; ‡Department of Cardiology, Respira-
tory Medicine and Nephrology, Hirosaki University Graduate School of Medicine, Aomori, Japan; §The Cardiovascular Institute, Tokyo, Japan;
and ‖Division of Biostatistics and Clinical Epidemiology, University of Toyama, Toyama, Japan.
Received June 18, 2015; revision received November 7, 2015; accepted November 14, 2015.
Financial Disclosure: The J-RHYTHM registry was supported by a grant from the Japan Heart Foundation (No. 12080025). The coauthors
have a potential conflict of interest: Dr. Atarashi has received research funding from Boehringer Ingelheim, and remuneration from Bayer
Healthcare, Boehringer Ingelheim, and Daiichi-Sankyo; Dr. Inoue has received research funding from Boehringer Ingelheim and Daiichi-
Sankyo, and remuneration from Daiichi-Sankyo, Bayer Healthcare, and Boehringer Ingelheim; Dr. Okumura has received research funding
from Boehringer Ingelheim and Daiichi-Sankyo, and remuneration from Boehringer Ingelheim, Bayer Healthcare, Daiichi-Sankyo, and Pfizer;
Dr. Yamashita has received research funding from Boehringer Ingelheim and Daiichi-Sankyo, and remuneration from Boehringer Ingelheim,
Daiichi-Sankyo, Bayer Healthcare, Pfizer, Bristol-Myers Squibb, and Eisai; and Dr. Origasa has received lecture fees from Daiichi-Sankyo.
Address correspondence to Eitaro Kodani, Department of Internal Medicine and Cardiology, Nippon Medical School, Tama-Nagayama Hospital,
1-7-1 Nagayama, Tama-shi, Tokyo 206-8512, Japan. E-mail:[email protected].
1052-3057/$ - see front matter
© 2016 National Stroke Association. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.11.020

Journal of Stroke and Cerebrovascular Diseases, Vol. 25, No. 3 (March), 2016: pp 585–599 585
586 E. KODANI ET AL.
stroke/TIA. Target INR should be 1.6-2.59 for secondary as well as primary pre-
vention of thromboembolism in Japanese NVAF patients. Key Words: Atrial
fibrillation—anticoagulation—warfarin—stroke—secondary prevention.
© 2016 National Stroke Association. Published by Elsevier Inc. All rights reserved.

Introduction were selected at the discretion of the treating cardiolo-

gists. Because no nonvitamin K antagonist oral
Atrial fibrillation (AF) is the most common sustained
anticoagulant (NOAC) was available when this registry
arrhythmia and is a major risk factor for cardiogenic
was carried out in 2009, all anticoagulation therapies
embolism.1,2 CHADS23 (congestive heart failure, hyper-
were performed with warfarin in this study. Patients
tension, age 75 years or older, diabetes mellitus, and history
with valvular AF, including mitral stenosis and mechan-
of stroke or transient ischemic attack (TIA)) or CHA2DS2-
ical prosthetic valves,15 were excluded from this subanalysis.
VASc scores4 are widely used for risk stratification of
ischemic stroke and systemic embolism in patients with
Follow-Up and Definition of Endpoints
nonvalvular atrial fibrillation (NVAF). A history of isch-
emic stroke or TIA carries a stronger risk for ischemic The patients were followed for 2 years, or until an
stroke than other components of these scores.3,4 Once pa- endpoint, whichever occurred first. The thromboem-
tients suffer from cardiogenic embolism, their condition bolic endpoints consisted of symptomatic ischemic stroke,
is mostly severe, and mortality can be higher as com- TIA, and systemic embolism. Major hemorrhage as the
pared with other types of ischemic stroke.5 In addition, safety endpoint included intracranial hemorrhage, gas-
cardiogenic embolism had a higher 10-year recurrence rate trointestinal hemorrhage, and others requiring
(75.2%) after the first event in the Hisayama study.6 There- hospitalization. All-cause mortality and cardiovascular mor-
fore, prophylaxis with anticoagulation therapy is crucial tality were also determined. If any event occurred during
for reducing AF-related ischemic stroke in patients with the follow-up period, the final clinical data, including INR
a history of stroke or TIA (stroke/TIA). Oral anticoagu- at the time closest to the event, were collected.11 The di-
lation therapy with vitamin K antagonists, mainly warfarin, agnostic criteria for each event have been described
can reduce the risk of AF-related ischemic stroke by elsewhere.11,12
60%-70%.7,8 However, information regarding the inci-
dence of thromboembolic and hemorrhagic events in Classification of Patients
Japanese NVAF patients with a history of ischemic stroke/
Patients were divided into 2 groups, that is, the
TIA is still limited.9,10 Therefore, we investigated the
“primary” and “secondary” prevention groups, accord-
incidence of thromboembolic and hemorrhagic events in
ing to the history of ischemic stroke/TIA, and were divided
patients with and without a history of ischemic stroke/
further into 2 age groups of younger than 70 and 70 years
TIA, using a post hoc analysis of the J-RHYTHM
or older. In addition, patients receiving warfarin were
Registry.11-13 Additionally, optimal anticoagulation inten-
divided into 5 subgroups according to their INR (<1.6,
sities of warfarin for patients with a history of ischemic
1.6-1.99, 2.0-2.59, 2.6-2.99, and ≥3.0). Correlation of end-
stroke/TIA were determined, although these had previously
points with anticoagulation intensities was analyzed with
been determined using a small number of Japanese
INR at the time of events or at the end of follow-up. The
patients.10 Because a target international normalized ratio
time in therapeutic range (TTR) was determined with the
(INR) of prothrombin time is different between NAVF pa-
method of Rosendaal et al.16 For this determination,
tients aged younger than 70 and 70 years or older in the
the target INR level was set at 1.6-2.6 for patients aged
current Japanese guidelines,14 we also investigated event
70 years or older, and at 2.0-3.0 for patients aged younger
rates in both age groups of younger than 70 and 70 years
than 70 years, following the Japanese guidelines.14
or older.
Statistical Analysis
Methods Data are presented as mean ± standard deviation or
median, if necessary. The statistical significance of dif-
Study Design of the J-RHYTHM Registry
ferences in the mean values was analyzed using Student’s
The J-RHYTHM Registry was a prospective, observa- t-test or analysis of variance, in the median values using
tional nationwide study. Study design and baseline the Mann–Whitney U-test or Kruskal–Wallis test, as ap-
characteristics have been reported in detail elsewhere.11,12 propriate. Frequencies of parameters or events were
Briefly, the subjects were a consecutive series of outpa- compared using the chi-square test or Fisher’s exact test,
tients with AF of any type, regardless of the use of as appropriate. Event-free rates of thromboembolism and
antithrombotic drugs. Antithrombotic drugs and dosages major hemorrhage between the primary and secondary
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 587
prevention groups were compared using the Kaplan– Results
Meier method and log-rank test. A Cox proportional hazard
Of the 7937 patients with AF who had been enrolled
model was used to examine whether a history of stroke/
in the J-RHYTHM Registry, 421 patients were excluded
TIA was an independent risk factor for the events in a
because they had valvular AF. Of the remaining 7516
crude model (Model 1) and after adjustment for con-
patients with NVAF, 110 (1.5%) were lost to follow-up.
founding factors, including the other components of
Therefore, a total of 7406 patients with NVAF, that is,
CHADS2 (Model 2) and CHA2DS2-VASc scores (addition-
6384 patients without and 1022 patients with a history
ally, vascular disease [coronary artery disease], age 65-
of ischemic stroke/TIA, were eligible for the present
74 years, and female sex: Model 3), and warfarin and
analyses.
antiplatelet use (Model 4). To clarify the optimal INR level,
a multivariate logistic regression analysis was per-
Baseline Characteristics and the Status of
formed. Two-sided P values less than .05 were considered
Antithrombotic Therapy
to be statistically significant. All statistical analyses were
performed with the IBM SPSS Statistics version 23.0 (IBM Baseline clinical characteristics and antithrombotic therapy
Corporation, Somers, NY). at the time of enrollment are summarized in Tables 1 and

Table 1. Baseline characteristics of patients

Characteristics Overall Primary prevention group Secondary prevention group P value*

Number of patients 7406 6384 (86.2) 1022 (13.8)

Age (years) 69.8 ± 10.0 69.3 ± 10.1 72.3 ± 8.8 <.001
Sex, male 5241 (70.8) 4479 (70.2) 762 (74.6) .005
Type of atrial fibrillation
Paroxysmal 2835 (38.3) 2497 (39.1) 338 (33.1)
Persistent 1081 (14.6) 944 (14.8) 137 (13.4) <.001
Permanent 3490 (47.1) 2943 (46.1) 547 (53.5)
Comorbidities
Coronary artery disease 781 (10.5) 647 (10.1) 134 (13.1) .005
Cardiomyopathy 634 (8.6) 556 (8.7) 78 (7.6) .279
HCM 264 (3.6) 217 (3.4) 47 (4.6) .065
DCM 370 (5.0) 339 (5.3) 31 (3.0) .002
Congenital heart disease 96 (1.3) 88 (1.4) 8 (.8) .157
COPD 131 (1.8) 107 (1.7) 24 (2.3) .166
Hyperthyroidism 131 (1.8) 119 (1.9) 12 (1.2) .154
Risk factors for stroke
Heart failure 2055 (27.7) 1779 (27.9) 276 (27.0) .594
Hypertension 4477 (60.5) 3835 (60.1) 642 (62.8) .103
Age (≥75 years) 2565 (34.6) 2125 (33.3) 440 (43.1) <.001
Diabetes mellitus 1359 (18.3) 1140 (17.9) 219 (21.4) .007
Stroke/TIA 1022 (13.8) 0 (.0) 1022 (100.0) <.001
CHADS2 score
0 1157 (15.6) 1157 (18.1) 0 (.0)
1 2512 (33.9) 2512 (39.3) 0 (.0)
2 2056 (27.8) 1891 (29.6) 165 (16.1)
3 1059 (14.3) 711 (11.1) 348 (34.1) <.001
4 436 (5.9) 113 (1.8) 323 (31.6)
5 161 (2.2) 0 (.0) 161 (15.8)
6 25 (.3) 0 (.0) 25 (2.4)
Mean 1.7 ± 1.2 1.4 ± 1.0 3.5 ± 1.0 <.001
Median 2 [1-2] 1 [1-2] 3 [3-4] <.001

Abbreviations: CHADS2, congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, and history of stroke or TIA;
COPD, chronic obstructive pulmonary disease; DCM, dilated cardiomyopathy; HCM, hypertrophic cardiomyopathy; SD, standard deviation;
TIA, transient ischemic attack.
Data are number of patients (%), mean ± SD, or median [interquartile range].
*Comparison between primary and secondary prevention groups.
588 E. KODANI ET AL.
Table 2. Antithrombotic therapy at the time of enrollment

Antithrombotic therapy Overall Primary prevention group Secondary prevention group P value*

Warfarin 6404 (86.5) 5449 (85.4) 955 (93.4) <.001

Dosage (mg/day) 2.9 ± 1.2 2.9 ± 1.2 2.8 ± 1.1 .016
INR
<1.6 1670 (26.1) 1477 (27.1) 193 (20.2)
1.6-1.99 2348 (36.7) 1985 (36.4) 363 (38.0)
2.0-2.59 1854 (29.0) 1552 (28.5) 302 (31.6) <.001
2.6-2.99 363 (5.7) 292 (5.4) 71 (7.4)
≥3.0 169 (2.6) 143 (2.6) 26 (2.7)
Mean 1.91 ± .49 1.90 ± .49 1.97 ± .48 <.001
TTR** (%) 59.3 ± 29.2 58.7 ± 29.3 62.8 ± 28.1 <.001
(n = 6064) (n = 5166) (n = 898)
Any antiplatelet 1937 (26.2) 1550 (24.3) 387 (37.9) <.001
Aspirin 1675 (22.6) 1362 (21.3) 313 (30.6) <.001
Others 433 (5.8) 321 (5.0) 112 (11.0) <.001
Warfarin + antiplatelet 1358 (18.3) 1027 (16.1) 331 (32.4) <.001

Abbreviations: INR, international normalized ratio of prothrombin time; SD, standard deviation; TTR, time in therapeutic range.
Data are number of patients (%) or mean ± SD.
*Comparison between primary and secondary prevention groups.
**Target INR, 2.0-3.0 (<70 years) or 1.6-2.6 (≥70 years).

2, respectively. Mean age and the prevalence of male sex, was comparable between the groups (Fig 1, D). As a whole
diabetes mellitus, and coronary artery disease were higher group, hazard ratios (HRs) of a prior ischemic stroke/
in the secondary than in the primary prevention group. TIA for thromboembolism remained significantly high even
It was a matter of course that the CHADS2 score was 2 in the most critical adjusting model (HR 1.82, P = .006
or higher in all patients of the secondary prevention group, in Model 4). Although the HR for major hemorrhage in
giving a higher mean CHADS2 score of 3.5 ± 1.0, as com- Model 4 did not reach statistical significance (HR 1.49,
pared with the primary prevention group (Table 1). The P = .055), HRs for combined events of thromboembo-
frequency of warfarin use as well as antiplatelet drugs lism plus major hemorrhage were consistently high even
was higher in the secondary than in the primary pre- after adjustment for multiple confounding factors (HR
vention group (Table 2). Mean INR and TTR were slightly 1.64, P = .001 in Model 4; Table 3). The HRs for all-
but significantly higher in the secondary than in the cause death were significantly high in only the crude model
primary prevention group (Table 2). (Model 1) and after adjustment for CHADS2 score (Model
2), whereas those for cardiovascular death were similar
Event Rates and a History of Stroke/TIA in any model. These results indicated that a history of
ischemic stroke/TIA was an independent risk factor for
During the 2-year follow-up period, thromboembolic
thromboembolism and for combined events of thrombo-
events occurred in 126 patients (1.7%), major hemor-
embolism and major hemorrhage (Table 3).
rhage in 140 (1.9%), and all-cause death in 195 (2.6%),
including cardiovascular death in 68 (.9%). The overall
Event Rates and Warfarin Use
incidence of thromboembolism was significantly higher
in the secondary than in the primary prevention group The characteristics of patients in the nonwarfarin and
(2.8% versus 1.5%, P = .004). The incidence rate of major warfarin groups and 5 INR subgroups are summarized
hemorrhage was also higher in the secondary than in the in Table 4. Twenty patients, including 18 without infor-
primary prevention group (3.0% versus 1.7%, P = .006). mation on warfarin use and 2 lacking INR values at the
All-cause mortality was higher in the secondary than in end of follow-up, were excluded. Consequently, the re-
the primary prevention group (3.9% versus 2.4%, P = .008), maining 7386 patients were used for post hoc analyses
whereas cardiovascular mortality was comparable between requiring INR values at the end of follow-up. In the primary
the groups (1.0% versus .9%, P = .967). The Kaplan– prevention group, prevalence of all components of CHADS2
Meier curves for endpoints are shown in Figure 1. Event- score except for prior stroke/TIA in patients on warfarin
free rates of thromboembolism, major hemorrhage, and was higher than that in the nonwarfarin group, resulting
all-cause death in the secondary prevention group were in a higher CHADS2 score in patients on warfarin. By
significantly lower than those in the primary preven- contrast, in the secondary prevention group, prevalence
tion group (P = .002, P = .003, and P = .004 by log-tank of patients aged 75 years or older and those with dia-
test, respectively; Fig 1, A-C). Cardiovascular mortality betes mellitus in the nonwarfarin group was higher than
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 589

Figure 1. Event-free curves for thromboembolism (A), major hemorrhage (B), all-cause death (C), and cardiovascular death (D) in the primary and sec-
ondary prevention groups using the Kaplan–Meier method. Additional boxes show the magnified event-free curves with 10-fold scale. P values: comparison
between the primary and secondary groups by log-rank test.

Table 3. Risk of a history of ischemic stroke/TIA for events (Cox proportional hazards analysis)

Model 1 Model 2 Model 3 Model 4

Events HR 95% CI P HR 95% CI P HR 95% CI P HR 95% CI P

Thromboembolism 1.92 1.27-2.91 .002 1.75 1.16-2.66 .008 1.72 1.13-2.61 .011 1.82 1.19-2.79 .006
Major hemorrhage 1.82 1.22-2.72 .003 1.67 1.12-2.49 .012 1.58 1.06-2.37 .025 1.49 .99-2.23 .055
Thromboembolism + major 1.87 1.40-2.49 <.001 1.71 1.28-2.28 <.001 1.65 1.23-2.20 .001 1.64 1.23-2.20 .001
hemorrhage
All-cause death 1.66 1.17-2.36 .004 1.49 1.05-2.11 .026 1.39 .98-1.97 .065 1.41 .99-2.01 .055
Cardiovascular death 1.11 .57-2.12 .755 1.03 .52-2.02 .939 .99 .50-1.94 .966 .95 .48-1.88 .888

Abbreviations: CI, confidence interval; HR, hazard ratio; TIA, transient ischemic attack.
Model 1: unadjusted (crude).
Model 2: adjusted for the other components of CHADS2 score (congestive heart failure, hypertension, age 75 years or older, and diabetes
mellitus).
Model 3: adjusted for the other components of CHA2DS2-VASc score (congestive heart failure, hypertension, age 75 years or older, dia-
betes mellitus, vascular disease [coronary artery disease], age 65-74 years, and female sex).
Model 4: adjusted for the other components of CHA2DS2-VASc score, warfarin use, and antiplatelet use.
 590
Table 4. Characteristics of nonwarfarin and warfarin groups, and 5 INR subgroups

Warfarin

INR level*

Characteristics Nonwarfarin All <1.6 1.6-1.99 2.0-2.59 2.6-2.99 ≥3.0 P value**

Primary prevention group

Number of patients 1023 5343 1193 1898 1694 331 227
Age (years) 66.9 ± 12.0 69.8 ± 9.6 70.6 ± 9.9 69.7 ± 9.4 69.3 ± 9.7 69.8 ± 9.6 70.9 ± 9.0 <.001
Sex, male 723 (70.7) 3745 (70.1) 798 (66.9) 1334 (70.3) 1222 (72.1) 236 (71.3) 155 (68.3) <.001
Risk factors for stroke
Heart failure 186 (18.2) 1587 (29.7) 395 (33.1) 501 (26.4) 510 (30.1) 103 (31.1) 78 (34.4) <.001
Hypertension 549 (53.7) 3275 (61.3) 717 (60.1) 1151 (60.6) 1038 (61.3) 222 (67.1) 147 (64.8) <.001
Age (≥75 years) 280 (27.4) 1841 (34.5) 466 (39.1) 637 (33.6) 555 (32.8) 100 (30.2) 83 (36.6) <.001
Diabetes mellitus 128 (12.5) 1009 (18.9) 245 (20.5) 348 (18.3) 310 (18.3) 63 (19.0) 43 (18.9) <.001
Stroke/TIA 0 (.0) 0 (.0) 0 (.0) 0 (.0) 0 (.0) 0 (.0) 0 (.0) –
CHADS2 score
Mean 1.1 ± 1.0 1.4 ± 1.0 1.5 ± 1.0 1.4 ± .9 1.4 ± .9 1.5 ± .9 1.6 ± 1.0 <.001
Median 1 [0-2] 1 [1-2] 1 [1-2] 1 [1-2] 1 [1-2] 1 [1-2] 2 [1-2] <.001
Secondary prevention group
Number of patients 103 917 196 333 293 54 41
Age (years) 74.7 ± 9.0 72.1 ± 8.7 72.2 ± 9.9 72.2 ± 8.9 71.6 ± 8.3 72.1 ± 8.7 73.2 ± 8.6 .083
Sex, male 77 (74.8) 684 (74.6) 132 (67.3) 251 (75.4) 236 (80.5) 36 (66.7) 29 (70.7) <.001
Risk factors for stroke
Heart failure 32 (31.1) 243 (26.5) 55 (28.1) 87 (26.1) 70 (23.9) 18 (33.3) 13 (31.7) .534
Hypertension 68 (66.0) 573 (62.5) 111 (56.6) 222 (66.7) 182 (62.1) 34 (63.0) 24 (58.5) .291
Age (≥75 years) 60 (58.3) 379 (41.3) 83 (42.3) 142 (42.6) 111 (37.9) 11 (20.4) 20 (48.8) <.001
Diabetes mellitus 26 (25.2) 192 (20.9) 36 (18.4) 79 (23.7) 57 (19.5) 23 (42.6) 9 (22.0) .005
Stroke/TIA 103 (100.0) 917 (100.0) 196 (100.0) 333 (100.0) 293 (100.0) 54 (100.0) 41 (100.0) –
CHADS2 score
Mean 3.8 ± 1.0 3.5 ± 1.0 3.5 ± 1.0 3.6 ± 1.0 3.4 ± 1.0 3.6 ± 1.1 3.6 ± 1.0 .014
Median 4 [3-5] 3 [3-4] 3 [3-4] 4 [3-4] 3 [3-4] 4 [3-4] 4 [3-4] .025

E. KODANI ET AL.
Abbreviations: CHADS2, congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, and history of stroke or TIA; INR, international normalized ratio of prothrombin time;
SD, standard deviation; TIA, transient ischemic attack.
Data are number of patients (%), mean ± SD, or median [interquartile range].
*At the time of events or at the end of follow-up.
**Comparison among 6 groups (nonwarfarin group and 5 INR subgroups) in patients receiving warfarin.
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 591

Figure 2. Two-year incidence of thromboembo-

lism (A), major hemorrhage (B), all-cause death
(C), and cardiovascular death (D). P values: com-
parison between patients with and without warfarin
in each group.

that in patients on warfarin, resulting in a higher CHADS2 patients in the primary and secondary prevention groups,
score in the nonwarfarin group than in patients on war- respectively (Table 5), including typical cardiogenic em-
farin (Table 4). Rates of thromboembolism and all-cause bolism in 18 and 7, atherothrombotic stroke in 5 and 2,
death were significantly higher in the secondary than in lacunar infarction in 3 and 1, and undetermined in others.
the primary prevention group (8.7% versus 2.5%, P = .002;
and 17.5% versus 5.8%, P < .001, respectively), when war-
Event Rates in INR Subgroups
farin was not given (Table 5). However, the rate of
thromboembolic events was significantly lower in the war- Incidence rates of thromboembolism for both the primary
farin group than in the nonwarfarin group, especially in and secondary prevention groups showed an obvious de-
the secondary prevention group (Fig 2, A). By contrast, creasing trend among the nonwarfarin group and 5 INR
rates of major hemorrhage in patients with and without subgroups of patients receiving warfarin (P < .001 for trend
warfarin were not different in either the primary or sec- in both groups, Table 5). Rates of major hemorrhage for
ondary prevention group (Fig 2, B). All-cause and both prevention groups also showed an apparent in-
cardiovascular mortality showed similar trends to that creasing trend among the nonwarfarin group and 5 INR
of thromboembolism (Fig 2, C,D). subgroups in patients receiving warfarin (P < .001 and
In patients who received warfarin, cerebral infarction oc- P = .009 for trend, respectively; Table 5). Consequently,
curred eventually in 54 (1.0%/2 years) and 19 (2.1%/2 years) the combined rates of thromboembolic and major
 592
Table 5. Incidence rates of events during the 2-year follow-up period

Warfarin

INR level*

Events Nonwarfarin All <1.6 1.6-1.99 2.0-2.59 2.6-2.99 ≥3.0 P value**

Primary prevention group

Number of patients 1023 5343 1193 1898 1694 331 227
Thromboembolism 26 (2.5%) 71 (1.3%) 37 (3.1%) 17 (.9%) 10 (.6%) 4 (1.2%) 3 (1.3%) <.001†
Cerebral infarction 22 54 26 15 6 4 3
TIA 2 6 2 1 3 0 0
Systemic embolism 2 11 9 1 1 0 0
Major hemorrhage 12 (1.2%) 97 (1.8%) 11 (.9%) 17 (.9%) 26 (1.5%) 8 (2.4%) 35 (15.4%) <.001†
Intracranial 4 28 5 6 9 2 6
Gastrointestinal 3 36 3 6 9 3 15
Others 5 33 3 5 8 3 14
Thromboembolism + major 38 (3.7%) 168 (3.1%) 48 (4.0%) 34 (1.8%) 36 (2.1%) 12 (3.6%) 38 (16.7%) <.001
hemorrhage
All-cause death 59 (5.8%) 96 (1.8%) 24 (2.0%) 34 (1.8%) 19 (1.1%) 7 (2.1%) 12 (5.3%) <.001
Cardiovascular death 17 (1.7%) 41 (.8%) 21 (.9%) 14 (.7%) 10 (.6%) 2 (.6%) 4 (1.8%) .049
Secondary prevention group
Number of patients 103 917 196 333 293 54 41
Thromboembolism 9 (8.7%) 20 (2.2%) 11 (5.6%) 6 (1.8%) 2 (.7%) 1 (1.9%) 0 (.0%) <.001†
Cerebral infarction 9 19 11 5 2 1 0
TIA 0 1 0 1 0 0 0
Systemic embolism 0 0 0 0 0 0 0
Major hemorrhage 2 (1.9%) 29 (3.2%) 6 (3.1%) 5 (1.5%) 9 (3.1%) 5 (9.3%) 4 (9.8%) .009†
Intracranial 0 18 2 4 7 3 2
Gastrointestinal 1 7 3 1 0 2 1
Others 1 4 1 0 2 0 1
Thromboembolism + major 11 (10.7%) 49 (5.3%) 17 (8.7%) 11 (3.3%) 11 (3.8%) 6 (11.1%) 4 (9.8%) <.001
hemorrhage
All-cause death 18 (17.5%) 21 (2.3%) 5 (2.6%) 3 (.9%) 7 (2.4%) 1 (1.9%) 5 (12.2%) <.001
Cardiovascular death 5 (4.9%) 5 (.5%) 0 (.0%) 2 (.6%) 1 (.3%) 0 (.0%) 2 (4.9%) <.001

E. KODANI ET AL.
Abbreviations: INR, international normalized ratio of prothrombin time; TIA, transient ischemic attack.
Data are number of patients (%).
*At the time of events or at the end of follow-up.
**Comparison among 6 groups (nonwarfarin group and 5 INR subgroups) in patients receiving warfarin (chi-square test).
†For trend.
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 593
hemorrhagic events were lower at INR levels of 1.6-2.59 adjustment for confounding factors, in both the primary
in both prevention groups (Table 5). All-cause mortality and secondary prevention groups (Table 6). ORs for all-
was lower in patients on warfarin than in those without cause death were significantly lower in INR subgroups
warfarin for both prevention groups (Table 5). This trend of lower than 3.0 in both prevention groups (Table 6).
was negated at INR levels of 3.0 or higher in both pre-
vention groups. Odds ratios (ORs) for thromboembolism
Event Rates in Age Groups
were significantly lower in INR subgroups of 1.6-1.99 and
2.0-2.59 than in the nonwarfarin group, even after ad- When the same analyses were performed in 2 age groups
justment for confounding factors, in both the primary and of younger than 70 and 70 years or older, rates of throm-
secondary prevention groups (Table 6). By contrast, the boembolism were significantly lower in patients taking
significance of ORs for major hemorrhage in INR sub- warfarin than in the nonwarfarin group among patients
groups disappeared after adjustment for confounding aged 70 years or older in both the primary and second-
factors in the secondary prevention group (Table 6). Con- ary groups (Fig 3, A). By contrast, the differences did not
sequently, ORs for thromboembolism plus major reach a significant level even in the secondary group, al-
hemorrhage were significantly lower in INR subgroups though rates of major hemorrhage appeared to be slightly
of 1.6-2.59 than in the nonwarfarin group, even after higher in patients on warfarin in all groups (Fig 3, B).

Figure 3. Two-year incidence of thromboembo-

lism (A), major hemorrhage (B), all-cause death
(C), and cardiovascular death (D) in each age
group. P values: comparison between patients with
and without warfarin in each group.
 594
Table 6. Odds ratios for events in each INR subgroup

Warfarin

INR level*
Nonwarfarin
Events (reference) <1.6 1.6-1.99 2.0-2.59 2.6-2.99 ≥3.0 P value

Primary prevention group

Thromboembolism 1.00 1.04 (.61-1.78) .31 (.16-.58) .21 (.10-.43) .43 (.15-1.25) .44 (.13-1.49) <.001
Major hemorrhage 1.00 .69 (.30-1.60) .70 (.33-1.50) 1.21 (.60-2.44) 1.90 (.76-4.76) 13.90 (6.95-27.78) <.001
Thromboembolism + major hemorrhage 1.00 .91 (.58-1.43) .42 (.26-.68) .50 (.31-.80) .87 (.44-1.69) 4.47 (2.73-7.33) <.001
All-cause death 1.00 .21 (.12-.35) .22 (.14-.35) .13 (.07-.22) .26 (.11-.59) .55 (.28-1.08) <.001
Cardiovascular death 1.00 .35 (.16-.78) .35 (.17-.73) .26 (.11-.58) .27 (.06-1.22) .65 (.21-2.03) <.001
Secondary prevention group
Thromboembolism 1.00 .75 (.29-1.96) .23 (.08-.70) .08 (.02-.41) .23 (.03-1.96) 0 .005
Major hemorrhage 1.00 1.65 (.34-8.02) .76 (.14-4.12) 1.65 (.34-8.02) 5.30 (.96-29.28) 5.64 (.96-33.08) .157
Thromboembolism + major hemorrhage 1.00 .90 (.39-2.01) .32 (.13-.78) .36 (.15-.89) 1.17 (.40-3.47) 1.02 (.30-3.49) .044
All-cause death 1.00 .11 (.04-.35) .04 (.01-.14) .11 (.04-.29) .10 (.01-.79) .66 (.20-2.14) <.001
Cardiovascular death 1.00 0 .18 (.03-1.09) .11 (.01-1.02) 0 1.27 (.21-7.79) .001

Abbreviations: INR, international normalized ratio of prothrombin time.

Data are odds ratios (95% confidence interval) adjusted for the components of CHA2DS2-VASc score (congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, vascular
disease [coronary artery disease], age 65-74 years, and female sex) and antiplatelet use, using the nonwarfarin group as a reference.
*At the time of events or at the end of follow-up.

E. KODANI ET AL.
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 595
18
All-cause death and cardiovascular mortality in both age practice and indicated for NVAF patients who have renal
groups showed similar trends to that of thromboembo- insufficiency and other comorbidities. Therefore, the present
lism (Fig 3, C,D). Event rates among the nonwarfarin group results would be practically important even in the era
and 5 INR subgroups are summarized in Table 7. There of NOACs.
were significant trends in rates of thromboembolism and In the present study, efficacy of warfarin for prevent-
major hemorrhage in all age groups, except for major hem- ing thromboembolism was apparent in both the secondary
orrhage in patients aged younger than 70 years in the and primary prevention groups (Fig 2). This salutary effect
secondary group (Table 7). of warfarin was evident particularly in patients aged 70
ORs for events in each age group and INR subgroup years or older for both the primary and the secondary
are summarized in Table 8. In patients aged younger prevention of stroke (Fig 3). Although rates of thrombo-
than 70 years in the primary prevention group, ORs for embolism in patients aged younger than 70 years on
thromboembolism, combined events of thromboembolism warfarin appeared to be lower than in those without war-
plus major hemorrhage, and all-cause death were sig- farin in both the primary and secondary groups, these
nificantly lower in INR subgroups of 1.6-1.99 and 2.0-2.59, differences were not statistically significant (Fig 3, A). As
whereas those for major hemorrhage were significantly the number of patients and events in each group was
higher in the INR subgroups of 2.6-3.0 and 3.0 or small, it could be explained that statistical power was
higher, as compared with the nonwarfarin group (Table 8, insufficient when patients were divided into 2 preven-
A). In patients aged younger than 70 years in the tion groups and an additional 2 age groups. In our previous
secondary prevention group, OR for all-cause death subanalysis on event rates in age groups in patients with
was significantly lower in only the INR subgroup of NVAF,19 there was a significant difference in rates of throm-
2.0-2.59 because the numbers of patients and events boembolism between patients with and without warfarin
were small (Table 8, B). In patients aged 70 years or even in patients aged younger than 70 years. Therefore,
older in both the primary and secondary prevention warfarin therapy would be beneficial in preventing throm-
groups, ORs for thromboembolism were significantly boembolism even in patients aged younger than 70 years.
lower in the INR subgroups of 1.6-1.99 and 2.0-2.59, However, it remains controversial whether warfarin therapy
and those for combined events of thromboembolism is really beneficial in younger patients aged younger than
plus major hemorrhage were significantly lower in the 65 years because the difference disappeared in younger
INR subgroup of 1.6-1.99 (Table 8, A,B). patients when the cutoff age was 65 years.19 A meta-
analysis of 6 randomized controlled studies on the effects
of warfarin in patients with NVAF showed comparable
Discussion
risk reduction of thromboembolism for primary preven-
There were major findings in the present study. First, tion studies (including several % of patients with prior
as expected, patients with a history of ischemic stroke/ ischemic stroke) and a secondary prevention study, al-
TIA were characterized as a high-risk population, with though target INR levels differed among the studies.20
a CHADS2 score of 3.5 ± 1.0. Second, warfarin was pre- The European Atrial Fibrillation Trial 21 was a second-
scribed in approximately 93%, and the TTR, based on the ary prevention study with 1007 NVAF patients to determine
Japanese guidelines,14 was 62.8% in the secondary pre- the efficacy of warfarin and aspirin. The target INR level
vention group. This indicates that anticoagulation therapy was set at 2.5-4.0, and aspirin was given at a dose of
was performed frequently, and the quality of warfarin 300 mg. Warfarin was effective at preventing thrombo-
therapy seemed acceptable among NVAF patients with embolic events, as compared to aspirin (HR .38, P < .001)
a history of ischemic stroke/TIA in Japan.17 However, rates and placebo (HR .34, P < .001). However, warfarin was
of both thromboembolism and major hemorrhage were associated with major hemorrhage more frequently
higher in the secondary than in the primary prevention (2.8%/year), as compared with aspirin (.9%/year) and
group. Prior ischemic stroke/TIA emerged as an inde- placebo (.7%/year). In our secondary prevention group,
pendent risk for both events after adjustment for multiple somewhat different results were observed; overall, war-
confounding factors. Third, for prevention of thrombo- farin was associated with a significantly lower incidence
embolism, an INR level of 1.6-2.59 would be optimal of thromboembolic events (2.2%/2 years), but not with
among Japanese NVAF patients in the secondary as well a higher incidence of major hemorrhage (3.2%/2 years).
as in the primary prevention groups. This could be attributed to lower INR levels in the sec-
ondary prevention group in the present study (1.97 ± .48).

Efficacy and Safety of Warfarin

Target Anticoagulation Intensity for Secondary
Although the use of NOACs is increasing worldwide,18
Prevention
the current Japanese guidelines14 recommend both war-
farin and NOACs for the prevention of thromboembolism Current guidelines in Western countries for manage-
for patients with NVAF. Warfarin is still used in clinical ment of patients with AF recommend the same INR
 596
Table 7. Incidence rates of events in each age group and each INR level

Warfarin

INR level*

Events Nonwarfarin All <1.6 1.6-1.99 2.0-2.59 2.6-2.99 ≥3.0 P value**

Primary prevention group

<70 years
Number of patients 592 2398 499 855 810 140 94
Thromboembolism 10 (1.7%) 18 (.8%) 9 (1.8%) 5 (.6%) 1 (.1%) 2 (1.4%) 1 (1.1%) <.001†
Major hemorrhage 6 (1.0%) 28 (1.2%) 4 (.8%) 6 (.7%) 7 (.9%) 6 (4.3%) 5 (5.3%) <.001†
Thromboembolism + major hemorrhage 16 (2.7%) 46 (1.9%) 13 (2.6%) 11 (1.3%) 8 (1.0%) 8 (5.7%) 6 (6.4%) <.001
All-cause death 9 (1.5%) 24 (1.0%) 6 (1.2%) 6 (.7%) 7 (.9%) 3 (2.1%) 2 (2.1%) .420
Cardiovascular death 3 (.5%) 12 (.5%) 2 (.4%) 3 (.4%) 5 (.6%) 2 (1.4%) 0 (.0%) .610
≥70 years
Number of patients 431 2945 694 1043 884 191 133
Thromboembolism 16 (3.7%) 53 (1.8%) 28 (4.0%) 12 (1.2%) 9 (1.0%) 2 (1.0%) 2 (1.5%) .008†
Major hemorrhage 6 (1.4%) 69 (2.3%) 7 (1.0%) 11 (1.1%) 19 (2.1%) 2 (1.0%) 30 (22.6%) .004†
Thromboembolism + major hemorrhage 22 (5.1%) 122 (4.1%) 35 (5.0%) 23 (2.2%) 28 (3.2%) 4 (2.1%) 32 (24.1%) <.001
All-cause death 50 (11.6%) 72 (2.4%) 18 (2.6%) 28 (2.7%) 12 (1.4%) 4 (2.2%) 10 (7.5%) <.001
Cardiovascular death 14 (3.2%) 29 (1.0%) 9 (1.3%) 11 (1.1%) 5 (.6%) 0 (.0%) 4 (3.0%) <.001
Secondary prevention group
<70 years
Number of patients 27 339 72 119 116 21 11
Thromboembolism 2 (7.4%) 5 (1.5%) 3 (4.2%) 1 (.8%) 1 (.9%) 0 (.0%) 0 (.0%) .014†
Major hemorrhage 0 (.0%) 8 (2.4%) 4 (5.6%) 1 (.8%) 2 (1.7%) 1 (4.8%) 0 (.0%) .653†
Thromboembolism + major hemorrhage 2 (7.4%) 13 (3.8%) 7 (9.7%) 2 (1.7%) 3 (2.6%) 1 (4.8%) 0 (.0%) .013
All-cause death 3 (11.1%) 2 (.6%) 1 (1.4%) 0 (.0%) 1 (.9%) 0 (.0%) 0 (.0%) <.001
Cardiovascular death 1 (3.7%) 0 (.0%) 0 (.0%) 0 (.0%) 0 (.0%) 0 (.0%) 0 (.0%) .028
≥70 years
Number of patients 76 578 124 214 177 33 30
Thromboembolism 7 (9.2%) 15 (2.6%) 8 (6.5%) 5 (2.3%) 1 (.6%) 1 (3.0%) 0 (.0%) <.001†
Major hemorrhage 2 (2.6%) 21 (3.6%) 2 (1.6%) 4 (1.9%) 7 (4.0%) 4 (12.1%) 4 (13.3%) .001†
Thromboembolism + major hemorrhage 9 (11.8%) 36 (6.2%) 10 (8.1%) 9 (4.2%) 8 (4.5%) 5 (15.2%) 4 (13.3%) .028
All-cause death 15 (19.7%) 19 (3.3%) 4 (3.2%) 3 (1.4%) 6 (3.4%) 1 (3.0%) 5 (16.7%) <.001

E. KODANI ET AL.
Cardiovascular death 4 (5.3%) 5 (.9%) 0 (.0%) 2 (.9%) 1 (.6%) 0 (.0%) 2 (6.7%) .003

Abbreviations: INR, international normalized ratio of prothrombin time; TIA, transient ischemic attack.
Data are number of patients (%).
*At the time of events or at the end of follow-up.
**Comparison between nonwarfarin group and 5 INR subgroups in patients receiving warfarin (chi-square test).
†For trend.
 SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF
Table 8. Odds ratios for events in each age group and INR subgroup

Warfarin

INR level*

Events Nonwarfarin (reference) <1.6 1.6-1.99 2.0-2.59 2.6-2.99 ≥3.0 P value

A. Primary prevention group

<70 years
Thromboembolism 1.00 .83 (.32-2.16) .26 (.09-.80) .05 (.01-.43) .60 (.13-2.87) .49 (.06-3.97) .032
Major hemorrhage 1.00 .60 (.16-2.19) .54 (.17-1.72) .65 (.21-2.02) 3.50 (1.07-11.43) 4.04 (1.17-13.96) .002
Thromboembolism + major hemorrhage 1.00 .74 (.34-1.60) .36 (.16-.81) .27 (.11-.66) 1.65 (.67-4.05) 1.86 (.69-4.99) <.001
All-cause death 1.00 .43 (.15-1.26) .28 (.09-.80) .31 (.11-.88) .83 (.21-3.22) .78 (.16-3.91) <.001
Cardiovascular death 1.00 .40 (.06-2.51) .42 (.08-2.19) .67 (.15-2.94) 1.55 (.24-9.95) 0 .070
≥70 years
Thromboembolism 1.00 1.11 (.58-2.14) .31 (.14-.68) .28 (.12-.65) .31 (.07-1.37) .42 (.09-1.88) <.001
Major hemorrhage 1.00 .74 (.24-2.25) .84 (.31-2.33) 1.62 (.63-4.16) .81 (.16-4.09) 21.96 (8.66-35.67) <.001
Thromboembolism + major hemorrhage 1.00 1.02 (.58-1.79) .45 (.25-.83) .64 (.36-1.14) .44 (.15-1.30) 6.24 (3.40-11.45) <.001
All-cause death 1.00 .17 (.09-.30) .21 (.12-.34) .09 (.05-.18) .16 (.06-.47) .51 (.24-1.07) <.001
Cardiovascular death 1.00 .34 (.14-.83) .15 (.15-.77) .16 (.06-.46) 0 .77 (.24-2.49) <.001
B. Secondary prevention group
<70 years
Thromboembolism 1.00 .63 (.09-4.37) .12 (.01-1.51) .12 (.01-1.54) 0 0 .629
Major hemorrhage 1.00 – – – – – –
Thromboembolism + major hemorrhage 1.00 .58 (.29-8.73) .24 (.03-1.85) .36 (.05-2.39) .73 (.06-9.04) 0 .446
All-cause death 1.00 .04 (.00-1.12) 0 .02 (.00-.46) 0 0 .007
Cardiovascular death 1.00 0 0 0 0 0 .313
≥70 years
Thromboembolism 1.00 .78 (.25-2.37) .21 (.08-.91) .06 (.01-.52) .33 (.04-2.92) 0 .033
Major hemorrhage 1.00 .59 (.18-4.50) .64 (.11-3.71) 1.47 (.28-7.61) 4.64 (.77-28.07) 5.79 (.94-33.70) .123
Thromboembolism + major hemorrhage 1.00 .70 (.26-1.87) .34 (.12-.91) .36 (.13-1.02) 1.34 (.40-4.51) 1.29 (.35-4.73) .155
All-cause death 1.00 .12 (.03-.41) .05 (.01-.21) .13 (.04-.38) .15 (.02-1.28) .93 (.27-3.17) <.001
Cardiovascular death 1.00 0 .27 (.04-1.78) .16 (.02-1.60) 0 1.72 (.26-11.50) .015

Abbreviations: INR, international normalized ratio of prothrombin time.

Data are odds ratios (95% confidence interval) adjusted for the components of CHA2DS2-VASc score (congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, vascular
disease [coronary artery disease], age 65-74 years, and female sex) and antiplatelet use, using the nonwarfarin group as a reference.
*At the time of events or at the end of follow-up.

597
598 E. KODANI ET AL.
levels for the primary and secondary prevention of underreporting of endpoints. Finally, the present study
thromboembolism in NVAF patients with and without had been performed before NOACs were approved for
prior ischemic stroke/TIA.22,23 In the Japanese guidelines,14 clinical use in Japan. Therefore, the present results should
target INR levels of 1.6-2.6 are recommended for pa- be interpreted carefully in the era of NOACs.
tients aged 70 years or older; this is based on results of
the secondary prevention study by Yasaka et al10 and the Conclusions
J-RHYTHM Registry.13 Concentration of D-dimer as an
index of fibrinolytic activity and prothrombin fragment Both thromboembolism and major hemorrhage oc-
F1+2 as an index of coagulation activity was not differ- curred more frequently in NVAF patients with prior stroke/
ent between INR levels of 1.5-1.9 and 2.0 or higher in TIA even when TTR was higher as compared with those
Japanese patients with NVAF.24 This indicated that slightly without prior stroke/TIA. A history of ischemic stroke/
lower anticoagulation intensities of warfarin could be ef- TIA is an independent risk factor for both events.
fective at inhibiting prothrombotic activity in Japanese Anticoagulation therapy with warfarin was performed fre-
patients with NVAF. quently for the secondary prevention of stroke in Japan
NVAF patients with prior ischemic stroke/TIA are at and could have a reducing effect for thromboembolism.
higher risk for thromboembolism than those without prior An INR level of 1.6-2.59 would be optimal for prevent-
ischemic events, and therefore anticoagulation therapy is ing thromboembolism and for avoiding major hemorrhage
indicated. One might ask whether higher INR values would in Japanese NVAF patients for the secondary as well as
be required for prevention of thromboembolism in the primary prevention of thromboembolism.
secondary prevention group as compared to the primary
prevention group, but this was not the case. An INR level Acknowledgments: The present study was presented in part
of 1.6-2.59 would be optimal for prevention of throm- at the 62nd Annual Scientific Session of the Japanese College
boembolism in the secondary as well as in the primary of Cardiology (in Sendai, Japan, September 27, 2014). In-
prevention group among Japanese NVAF patients. vestigators in the J-RHYTHM Registry are listed in References
The frequency of intracranial hemorrhage on warfa- 11, 12, and 27.
rin therapy with a target INR level of 2.0-3.0 was 4-fold
higher in Asian than in Caucasian patients.25 Suzuki et al26 References
also showed that INR levels of 2.27 or higher were as-
1. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as
sociated with an increased risk of major hemorrhage in
an independent risk factor for stroke: the Framingham
Japanese patients with NVAF. The present results showing study. Stroke 1991;22:983-988.
that INR levels of 2.6 or higher were associated with higher 2. Feinberg WM, Blackshear JL, Laupacis A, et al. Prevalence,
rates of major hemorrhage are consistent with prior age distribution, and gender of patients with atrial
results.25,26 fibrillation. Analysis and implications. Arch Intern Med
1995;155:469-473.
3. Gage BF, Waterman AD, Shannon W, et al. Validation
of clinical classification schemes for predicting stroke:
Study Limitations
results from the national registry of atrial fibrillation.
The present study had several limitations. First, it was JAMA 2001;285:2864-2870.
performed in a single country and the subjects were en- 4. Lip GY, Nieuwlaat R, Pisters R, et al. Refining clinical
risk stratification for predicting stroke and
rolled in only 158 institutions in Japan. Because most thromboembolism in atrial fibrillation using a novel risk
participating physicians were cardiologists, the pa- factor-based approach: the euro heart survey on atrial
tients’ clinical backgrounds might not be extrapolated to fibrillation. Chest 2010;137:263-272.
a general Japanese patient population with NVAF. There- 5. Kubo M, Kiyohara Y, Ninomiya T, et al. Decreasing
fore, generalizing from the present results of this specific incidence of lacunar vs other types of cerebral infarction
in a Japanese population. Neurology 2006;66:1539-1544.
population to individual clinical situations should be done 6. Hata J, Tanizaki Y, Kiyohara Y, et al. Ten year recurrence
carefully. Second, although this registry was relatively large, after first ever stroke in a Japanese community: the
analyzing a total of 7406 patients with NVAF, the numbers Hisayama study. J Neurol Neurosurg Psychiatry
of patients and events in INR subgroups were small es- 2005;76:368-372.
pecially in the secondary prevention group. This might 7. Atrial Fibrillation Investigators. Risk factors for stroke
and efficacy of antithrombotic therapy in atrial fibrillation.
have reduced the statistical power of the present study. Analysis of pooled data from five randomized controlled
Third, because the study design was prospective but ob- trials. Arch Intern Med 1994;154:1449-1457.
servational, antithrombotic treatments with warfarin or 8. Hart RG, Pearce LA, Aguilar MI. Meta-analysis:
antiplatelet agents were not randomized. Dosage of war- antithrombotic therapy to prevent stroke in patients who
farin and clinical targets of INR levels for individual have nonvalvular atrial fibrillation. Ann Intern Med
2007;146:857-867.
patients were selected at the discretion of treating phy- 9. Yamaguchi T. Optimal intensity of warfarin therapy
sicians. Fourth, 1.5% of patients were lost to follow-up for secondary prevention of stroke in patients with
in the present study, which could have led to the nonvalvular atrial fibrillation: a multicenter, prospective,
SECONDARY STROKE PREVENTION WITH WARFARIN IN NONVALVULAR AF 599
randomized trial. Japanese nonvalvular atrial fibrillation—subanalysis of the J-RHYTHM Registry. Circ
fibrillation-embolism secondary prevention cooperative J 2015;79:2345-2352.
study group. Stroke 2000;31:817-821. 20. Hart RG, Benavente O, McBride R, et al. Antithrombotic
10. Yasaka M, Minematsu K, Yamaguchi T. Optimal intensity therapy to prevent stroke in patients with atrial fibrillation:
of international normalized ratio in warfarin therapy for a meta-analysis. Ann Intern Med 1999;131:492-501.
secondary prevention of stroke in patients with non- 21. EAFT (European Atrial Fibrillation Trial) Study Group.
valvular atrial fibrillation. Intern Med 2001;40:1183-1188. Secondary prevention in non-rheumatic atrial fibrillation
11. Atarashi H, Inoue H, Okumura K, et al. Investigation after transient ischaemic attack or minor stroke. Lancet
of optimal anticoagulation strategy for stroke prevention 1993;342:1255-1262.
in Japanese patients with atrial fibrillation -The 22. Fuster V, Ryden LE, Cannom DS, et al. ACC/AHA/ESC
J-RHYTHM Registry study design. J Cardiol 2011;57:95-99. 2006 guidelines for the management of patients with atrial
12. Atarashi H, Inoue H, Okumura K, et al. Present status fibrillation: a report of the American College of
of anticoagulation treatment in Japanese patients with Cardiology/American Heart Association Task Force on
atrial fibrillation—a report from the J-RHYTHM Registry. Practice Guidelines and the European Society of
Circ J 2011;75:1328-1333. Cardiology Committee for Practice Guidelines (writing
13. Inoue H, Okumura K, Atarashi H, et al. Target committee to revise the 2001 Guidelines for the
international normalized ratio values for preventing Management of Patients with Atrial Fibrillation):
thromboembolic and hemorrhagic events in Japanese developed in collaboration with the European Heart
patients with non-valvular atrial fibrillation: results of Rhythm Association and the Heart Rhythm Society.
the J-RHYTHM Registry. Circ J 2013;77:2264-2270. Circulation 2006;114:e257-e354.
14. JCS Joint Working Group. Guidelines for pharmacotherapy 23. Camm AJ, Lip GY, De Caterina R, et al. 2012 Focused
of atrial fibrillation (JCS 2013). Circ J 2014;78:1997-2021. update of the ESC Guidelines for the management of
15. Kodani E, Atarashi H, Inoue H, et al. Target intensity atrial fibrillation: an update of the 2010 ESC Guidelines
of anticoagulation with warfarin in Japanese patients with for the Management of Atrial Fibrillation. Developed with
valvular atrial fibrillation: subanalysis of the J-RHYTHM the special contribution of the European Heart Rhythm
Registry. Circ J 2015;79:325-330. Association. Eur Heart J 2012;33:2719-2747.
16. Rosendaal FR, Cannegieter SC, van der Meer FJ, et al. 24. Nozawa T, Inoue H, Iwasa A, et al. Effects of
A method to determine the optimal intensity of oral anticoagulation intensity on hemostatic markers in patients
anticoagulant therapy. Thromb Haemost 1993;69:236-239. with non-valvular atrial fibrillation. Circ J 2004;68:29-34.
17. Connolly SJ, Pogue J, Eikelboom J, et al. Benefit of oral 25. Shen AY, Yao JF, Brar SS, et al. Racial/ethnic differences
anticoagulant over antiplatelet therapy in atrial fibrillation in the risk of intracranial hemorrhage among patients
depends on the quality of international normalized ratio with atrial fibrillation. J Am Coll Cardiol 2007;50:309-
control achieved by centers and countries as measured 315.
by time in therapeutic range. Circulation 2008;118:2029- 26. Suzuki S, Yamashita T, Kato T, et al. Incidence of major
2037. bleeding complication of warfarin therapy in Japanese
18. Barnes GD, Lucas E, Alexander GC, et al. National trends patients with atrial fibrillation. Circ J 2007;71:761-765.
in ambulatory oral anticoagulant use. Am J Med 27. J-RHYTHM Registry Investigators. Determinants of
2015;doi:10.1016/j.amjmed.2015.05.044. warfarin use and international normalized ratio levels
19. Kodani E, Atarashi H, Inoue H, et al. Use of warfarin in atrial fibrillation patients in Japan—subanalysis of the
in elderly patients with non-valvular atrial J-RHYTHM Registry. Circ J 2011;75:2357-2362.