2264 INOUE H et al.

Circulation Journal
Official Journal of the Japanese Circulation Society
ORIGINAL ARTICLE
http://www. j-circ.or.jp Arrhythmia/Electrophysiology

Target International Normalized Ratio Values for

Preventing Thromboembolic and Hemorrhagic Events in
Japanese Patients With Non-Valvular Atrial Fibrillation
– Results of the J-RHYTHM Registry –
Hiroshi Inoue, MD, PhD; Ken Okumura, MD, PhD; Hirotsugu Atarashi, MD, PhD;
Takeshi Yamashita, MD, PhD; Hideki Origasa, PhD; Naoko Kumagai, BSc;
Masayuki Sakurai, MD, PhD; Yuichiro Kawamura, MD, PhD; Isao Kubota, MD, PhD;
Kazuo Matsumoto, MD, PhD; Yoshiaki Kaneko, MD, PhD; Satoshi Ogawa, MD, PhD;
Yoshifusa Aizawa, MD, PhD; Masaomi Chinushi, MD, PhD; Itsuo Kodama, MD, PhD;
Eiichi Watanabe, MD, PhD; Yukihiro Koretsune, MD, PhD; Yuji Okuyama, MD, PhD;
Akihiko Shimizu, MD, PhD; Osamu Igawa, MD, PhD; Shigenobu Bando, MD, PhD;
Masahiko Fukatani, MD, PhD; Tetsunori Saikawa, MD, PhD;
Akiko Chishaki, MD, PhD on behalf of the J-RHYTHM Registry Investigators

Background: Target anticoagulation levels for warfarin in Japanese patients with non-valvular atrial fibrillation
(NVAF) are unclear.

Methods and Results:  Of 7,406 patients with NVAF, 1,002 did not receive warfarin (non-warfarin group), and the
remaining patients receiving warfarin were divided into 5 groups based on their baseline international normalized
ratio (INR) of prothrombin time (≤1.59, 1.6–1.99, 2.0–2.59, 2.6–2.99, and ≥3.0). Patients were followed-up prospec-
tively for 2 years. Primary endpoints were thromboembolic events (cerebral infarction, transient ischemic attack, and
systemic embolism), and major hemorrhage requiring hospital admission. During the follow-up period, thromboem-
bolic events occurred in 3.0% of non-warfarin group, but at lower frequencies in the warfarin groups (2.0, 1.3, 1.5,
0.6, and 1.8%/2 years for INR values of ≤1.59, 1.6–1.99, 2.0–2.59, 2.6–2.99, and ≥3.0, respectively; P=0.0059). Major
hemorrhage occurred more frequently in warfarin groups (1.5, 1.8, 2.4, 3.3, and 4.1% for INR values ≤1.59, 1.6–1.99,
2.0–2.59, 2.6–2.99, and ≥3.0, respectively; P=0.0041) than in non-warfarin group (0.8%/2 years). These trends were
maintained when the analyses were confined to patients aged ≥70 years.

Conclusions:  An INR of 1.6–2.6 is safe and effective at preventing thromboembolic events in patients with NVAF,
particularly patients aged ≥70 years. An INR of 2.6–2.99 is also effective, but associated with a slightly increased
risk in major hemorrhage. (UMIN Clinical Trials Registry UMIN000001569)   (Circ J 2013; 77: 2264 – 2270)

Key Words: Atrial fibrillation; Hemorrhage; International normalized ratio; Thromboembolism; Warfarin

Received March 11, 2013; revised manuscript received April 9, 2013; accepted April 19, 2013; released online May 25, 2013   Time for
primary review: 16 days
Second Department of Internal Medicine, Toyama University Hospital, Toyama (H.I.); Department of Cardiology, Hirosaki University
Graduate School of Medicine, Aomori (K.O.); Nippon Medical School, Tama-Nagayama Hospital, Tokyo (H.A.); The Cardiovascular
Institute, Tokyo (T.Y.); Division of Biostatistics and Clinical Epidemiology, University of Toyama, Toyama (H.O.); Kochi Medical School
Clinical Research Center, Kochi University, Kochi (N.K.); Department of Cardiology, Hokko Memorial Hospital Hokkaido, Sapporo
(M.S.); Health Administration Center, Asahikawa Medical University, Asahikawa (Y. Kawamura); Internal Medicine 1, Yamagata Uni-
versity School of Medicine, Yamagata (I. Kubota); Cardiology Department, International Medical Center, Saitama Medical University,
Saitama (K.M.); Department of Medicine and Biological Science, Gunma University Graduate School of Medicine, Maebashi (Y. Kaneko);
International University of Health and Welfare, Mita Hospital, Tokyo (S.O.); Division of Cardiology, Niigata University Graduate School
of Medicine and Dental Science, Niigata (Y.A.); Graduate School of Health Science, Niigata University School of Medicine, Niigata (M.C.);
Nagoya University, Nagoya (I. Kodama); Department of Cardiology, Fujita Health University School of Medicine, Toyoake (E.W.); Insti-
tute for Clinical Research, National Hospital Organization, Osaka National Hospital, Osaka (Y. Koretsune); Department of Cardiology,
Osaka General Medical Center, Osaka (Y.O.); Faculty of Health Sciences, Yamaguchi Graduate School of Medicine, Ube (A.S.); Depart-
ment of Cardiovascular Medicine, Tottori University Hospital, Tottori (O.I.); Kagawa Prefectural Shiratori Hospital, Kagawa (S.B.); Depart-
ment of Cardiology, Chikamori Hospital, Kochi (M.F.); Department of Clinical Examination and Diagnostics, Oita University, Oita (T.S.);
and Department of Health Sciences, Faculty of Medical Sciences, Kyushu University, Fukuoka (A.C.), Japan
This study was planned by the Japanese Society of Electrocardiology and supported by a grant from the Japan Heart Foundation.
Mailing address:  Hiroshi Inoue, MD, The Second Department of Internal Medicine, Toyama University Hospital, 2630 Sugitani, Toyama
930-0194, Japan.   E-mail: hiroshi@med.u-toyama.ac.jp
ISSN-1346-9843  doi: 10.1253/circj.CJ-13-0290
All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: cj@j-circ.or.jp

Circulation Journal  Vol.77, September 2013

Target INR in NVAF 2265

Table 1.  Clinical Characteristics of the Japanese Patients With Non-Valvular AF in the J-RHYTHM Registry
Warfarin
Non-
INR level P value
warfarin
≤1.59 1.6–1.99 2.0–2.59 2.6–2.99 ≥3
n 1,002 1,670 2,348 1,854 363 169
Age (years) 67.8±11.8 70.1±10.0 70.4±9.4 69.6±9.5 69.6±9.3 71.5±9.6 <0.0001
Men (%) 70.4 69.7 71.7 70.3 73.3 71.0 　0.66　　　　
Paroxysmal AF (%) 60.6 37.7 33.5 33.6 37.7 30.2 <0.0001
Heart failure (%) 15.5 29.9 28.4 29.8 30.3 41.4 <0.0001
Diabetes (%) 13.6 21.0 19.0 18.1 16.5 18.9 　0.0002
Hypertension (%) 53.3 59.6 62.6 62.0 60.6 63.9 <0.0001
Prior CI or TIA (%) 8.7 11.6 15.5 16.3 19.6 15.4 <0.0001
CHADS2 score <0.0001
  0 29.1 15.7 13.5 12.8 10.7 10.7
  1 37.2 34.0 33.0 35.7 34.4 27.2
  2 20.5 27.4 30.1 26.8 32.2 33.1
  ≥3 13.2 22.9 23.4 24.7 22.7 29.0
Mean 1.2±1.2 1.7±1.2 1.7±1.2 1.8±1.2 1.8±1.1 1.9±1.2 <0.0001
Antiplatelet (%) 57.8 22.9 20.9 20.1 21.8 20.1 <0.0001
Data are mean ± SD. AF, atrial fibrillation; CI, cerebral infarction; INR, international normalized ratio; TIA, transient isch-
emic attack.

A
nticoagulation treatment with warfarin is effective at
preventing ischemic stroke and peripheral embolism
in patients with atrial fibrillation (AF).1,2 The guide-
lines used in Western countries for the management of AF
recommend target anticoagulation levels, as measured by the
international normalized ratio (INR) of prothrombin time, of
2–3.3 The investigators in the ATRIA study reported that a
target INR of between 2 and 3 is optimal for preventing throm-
boembolic and hemorrhagic events in patients with non-valvu-
lar AF (NVAF), irrespective of their age and CHADS2 score.4
However, in Japan, an INR of between 1.6 and 2.6 has been
recommended for patients ≥70 years of age with NVAF,5 based
on results obtained by combining 2 studies that had prospec-
tively determined the optimal INR values for the secondary
prevention of thromboembolism.6 However, the sample size
of the combined study was not large enough to determine the
optimal INR values for preventing thromboembolic and hem-
orrhagic events.6 A recent retrospective study7 supported the
assertion that lower INR values, such as those suggested in the
study by Yasaka et al,6 are optimal for elderly patients with
NVAF in Japan. Lower INR values were also effective at pre-
venting thromboembolic events among Chinese patients with
AF.8,9 However, for Japanese NVAF patients <70 years old, an
INR of between 2 and 3 is recommended for preventing throm-
boembolism,5 which agrees with the guidelines used in West-
ern countries.3
Editorial p 2242
Recently, novel oral anticoagulants have been demonstrated
to be effective at preventing thromboembolism in patients with
NVAF.10,11 The RE-LY10 and J-ROCKET-AF11 studies used
somewhat lower target INR values for Japanese AF patients
who had been allocated to the warfarin arm. Therefore, the
target INR values for preventing thromboembolic and hemor-
rhagic events among Japanese patients with NVAF need be
reappraised in a large-scale prospective study. The J-RHYTHM
Registry12–14 is an observational cohort study that aims to de-
termine the current status of anticoagulation therapy with war-
farin and the optimal anticoagulation levels for preventing
thromboembolic events in Japanese patients with valvular AF
or NVAF. In the present analysis of the J-RHYTHM Registry,
target INR values for preventing both thromboembolic and
hemorrhagic events in patients with NVAF were determined.
Methods
The details of the study design and the subjects’ baseline char-
acteristics have been reported elsewhere.12,13 Briefly, consecu-
tive AF patients were recruited at the outpatient clinic of each
participating institution. All participating physicians were car-
diologists who had been selected by the local members of
the Steering Committee of the J-RHYTHM Registry. To avoid
particular geographic regions being over- or underrepresented
among the study population, enrollment targets that reflected
the population densities of 10 geographical regions of Japan
were set.12,13 The antithrombotic drugs and their dosages were
selected by the participating physicians. A total of 7,937 AF
patients (mean age, 69.7 years) were enrolled in the J-
RHYTHM Registry.13 Of these, 421 patients had mitral steno-
sis or had undergone mechanical valve replacement, and so the
remaining 7,516 patients with NVAF comprised the present
study group.
The patients were followed for 2 years or until an endpoint
occurred. The endpoints consisted of thromboembolic events,
including symptomatic cerebral infarction (CI), transient isch-
emic attack (TIA), and systemic embolism; major hemorrhage
requiring hospital admission; and death. The diagnostic crite-
ria for CI and TIA have been reported previously.12,13
Statistical Analysis
Data are presented as percentage or mean ± SD values. The pa-
tients in the warfarin group were divided into subgroups ac-
cording to their baseline antithrombotic therapy status and
baseline INR values (≤1.59, 1.6–1.99, 2.0–2.59, 2.6–2.99, and
≥3.0). As the Japanese guidelines recommend different INR
values according to the patient’s age,5 the patients were also
divided into 2 age groups (ie, <70 years and ≥70 years). Vari-

Circulation Journal  Vol.77, September 2013

2266 INOUE H et al.

Table 2.  Clinical Characteristics of Japanese Patients With Non-Valvular AF Aged ≥70 Years in the
J-RHYTHM Registry
Warfarin
Non-
INR level P value
warfarin
≤1.59 1.6–1.99 2.0–2.59 2.6–2.99 ≥3
n 459 924 1,362 1,004 196 96
Age (years) 78.0±5.7 77.2±5.0 76.8±4.9 76.7±4.6 76.4±5.1 77.8±6.5 <0.0001
Men (%) 62.3 63.7 65.4 62.1 70.4 68.8 　0.16　　　　
Paroxysmal AF (%) 55.3 33.9 31.6 32.7 33.7 30.2 <0.0001
Heart failure (%) 22.4 35.3 30.8 32.0 30.6 47.9 <0.0001
Diabetes (%) 16.3 21.6 18.6 19.7 17.3 14.6 　0.15　　　　
Hypertension (%) 62.5 65.8 66.1 66.2 61.2 64.6 　0.57　　　　
Prior CI or TIA (%) 11.8 13.9 17.3 18.1 20.9 15.6 　0.0037
CHADS2 score 　0.096　　
  0 10.5 6.1 7.3 5.2 5.6 5.2
  1 29.8 27.1 26.2 27.9 27.6 18.8
  2 34.9 32.8 34.7 31.9 36.7 40.6
  ≥3 24.8 34.0 31.8 35.0 30.1 35.4
Mean 1.9±1.2 2.2±1.2 2.1±1.2 2.2±1.2 2.1±1.2 2.2±1.1 　0.0031
Antiplatelet (%) 64.3 27.2 23.5 23.6 26.5 21.9 <0.0001
Data are mean ± SD. Abbreviations as in Table 1.

Table 3.  Clinical Characteristics of Japanese Patients With Non-Valvular AF Aged <70 Years in the
J-RHYTHM Registry
Warfarin
Non-
INR level P value
warfarin
≤1.59 1.6–1.99 2.0–2.59 2.6–2.99 ≥3
n 543 746 986 850 167 73
Age (years) 59.2±8.2 61.2±7.1 61.4±6.5 61.3±6.7 61.6±6.2 63.2±6.0 <0.0001
Men (%) 77.2 77.1 80.3 80.0 76.6 74.0 　0.33　　　　
Paroxysmal AF (%) 65.0 42.5 36.1 34.7 42.5 30.1 <0.0001
Heart failure (%) 9.6 23.3 25.1 27.3 29.9 32.9 <0.0001
Diabetes (%) 11.2 20.1 19.4 16.2 15.6 24.7 　0.0002
Hypertension (%) 45.5 52.0 57.7 57.1 59.9 63.0 <0.0001
Prior CI or TIA (%) 2.4 8.7 13.0 14.1 18.0 15.1 <0.0001
CHADS2 score <0.0001
  0 44.9 27.7 22.1 21.8 16.8 17.8
  1 43.5 42.5 42.4 44.8 42.5 38.4
  2 8.3 20.8 23.7 20.8 26.9 23.3
  ≥3 0.3 9.0 11.8 12.6 13.8 20.5
Mean 0.7±0.8 1.1±1.0 1.3±1.0 1.3±1.1 1.4±1.0 1.5±1.1 <0.0001
Antiplatelet (%) 52.3 17.6 17.3 15.9 16.2 17.8 <0.0001
Data are mean ± SD. Abbreviations as in Table 1.

able frequencies were compared using Pearson’s chi-square
test, and mean values were compared with ANOVA. Hazard
ratios were calculated with the Cox proportional hazard model,
using the non-warfarin group as a reference. P<0.05 were re-
garded as statistically significant.
Results
Of the 7,516 patients with NVAF, 110 (1.5%) were lost to fol-
low-up. Therefore, the data for remaining 7,406 patients were
used in the subsequent analyses.
Baseline Characteristics of the Study Patients
Entire NVAF Patient Population (Table 1)   Of the 7,406 pa-
tients, 1,002 were not taking warfarin at the time of enrollment
(non-warfarin group). The patients in this group were younger
and had lower CHADS2 scores than the warfarin group; that
is, the non-warfarin group was at low risk of thromboembo-
lism; more than half of them had paroxysmal AF and were
taking antiplatelet drugs instead of warfarin. Of the warfarin
group, 36.7% had baseline INR values of 1.6–1.99, and 29.0%
displayed baseline INR values of 2.0–2.59. Only 2.6% of the
warfarin group had an INR ≥3.0 at the time of enrollment.
NVAF Patients Aged ≥70 Years (Table 2)   There were 4,041
patients who were aged ≥70 years (mean age: 77.0 years), and
approximately two-thirds of them were men. Of these patients
459 did not receive anticoagulation therapy. Compared with
the warfarin group, the non-warfarin group had a higher prev-

Circulation Journal  Vol.77, September 2013

Target INR in NVAF 2267

Table 4.  Incidence of Thromboembolic and Major Hemorrhagic Events During the 2-Year Follow-up of
Japanese Patients With Non-Valvular AF in the J-RHYTHM Registry
Warfarin
Non-
INR level P value
warfarin
≤1.59 1.6–1.99 2.0–2.59 2.6–2.99 ≥3
Whole group
  Thromboembolism 30 (3.0%) 33 (2.0%) 31 (1.3%) 27 (1.5%) 2 (0.6%) 3 (1.8%) 0.0059
    Cerebral infarction 27 26 25 23 1 2
    TIA 2 3 2 0 1 1
    Systemic embolism 1 4 4 4 0 0
  Major hemorrhage 8 (0.8%) 25 (1.5%) 43 (1.8%) 45 (2.4%) 12 (3.3%) 7 (4.1%) 0.0041
    Intracranial 3 8 15 16 5 3
    Gastrointestinal 3 11 13 15 3 3
    Others 2 6 15 14 5 1
  All-cause mortality 34 (3.4%) 50 (3.0%) 63 (2.7%) 33 (1.8%) 6 (1.7%) 9 (5.3%) 0.0095
  Cardiovascular mortality 9 (0.9%) 14 (0.8%) 26 (1.1%) 12 (0.6%) 2 (0.6%) 5 (3.0%) 0.0697
<70 years old
  Thromboembolism 11 (2.0%) 8 (1.1%) 6 (0.6%) 10 (1.2%) 0 0 0.2894
    Cerebral infarction 9 8 5 7 0 0
    TIA 2 0 1 0 0 0
    Systemic embolism 0 0 0 3 0 0
  Major hemorrhage 3 (0.6%) 6 (0.8%) 18 (1.8%) 12 (1.4%) 3 (1.8%) 0 0.3513
    Intracranial 2 4 8 2 1 0
    Gastrointestinal 1 2 5 4 0 0
    Others 0 0 5 6 2 0
  All-cause mortality 5 (0.9%) 7 (0.9%) 11 (1.1%) 10 (1.2%) 3 (1.8%) 2 (2.7%) 0.7955
  Cardiovascular mortality 3 (0.6%) 3 (0.4%) 3 (0.3%) 5 (0.6%) 1 (0.6%) 1 (1.4%) 0.8443
≥70 years old
  Thromboembolism 19 (4.1%) 25 (2.7%) 25 (1.8%) 17 (1.7%) 2 (1.0%) 3 (3.1%) 0.0241
    Cerebral infarction 18 18 20 16 1 2
    TIA 0 3 1 0 1 1
    Systemic embolism 1 4 4 1 0 0
  Major hemorrhage 5 (1.1%) 19 (2.1%) 25 (1.8%) 33 (3.3%) 9 (4.6%) 7 (7.3%) 0.0008
    Intracranial 1 4 7 14 4 3
    Gastrointestinal 2 9 8 11 3 3
    Others 2 7 10 8 3 1
  All-cause mortality 29 (6.3%) 43 (4.7%) 52 (3.8%) 23 (2.3%) 3 (1.5%) 7 (7.3%) 0.0004
  Cardiovascular mortality 6 (1.3%) 11 (1.2%) 23 (1.7%) 7 (0.7%) 1 (0.5%) 4 (4.2%) 0.0456
Abbreviations as in Table 1.

alence of both paroxysmal AF and antiplatelet therapy, but
lower prevalence of each of heart failure and prior CI and
TIA. The mean CHADS2 score of the non-warfarin group was
slightly lower than that of the warfarin group.
NVAF Patients Aged <70 Years (Table 3)   There were 3,365
patients aged <70 years (mean age: 61.0 years), and approxi-
mately three-quarters of them were men. Of these patients, 543
did not receive anticoagulation therapy. The mean CHADS2
score of the non-warfarin group was 0.7, and that of the war-
farin group ranged from 1.1 to 1.5. This indicated that patients
aged <70 years were not at high risk of thromboembolic
events.3,5 The patients in the non-warfarin group were at par-
ticularly low risk of thromboembolism, which might explain
the higher frequency of antiplatelet treatment in this group.
Thromboembolic and Hemorrhagic Events, and Mortality
(Tables 4,5)
During the 2-year follow-up, 126 patients suffered thrombo-
embolic events (CI, TIA, or systemic embolism), 140 patients
suffered major hemorrhagic events requiring hospitalization,
and 195 patients died.
In the warfarin group, the incidence of thromboembolic
events decreased as the INR value increased (P=0.0059 for
trend) and was significantly lower in each warfarin group than
in the non-warfarin group. In contrast, the frequency of hem-
orrhagic events increased significantly as the INR value rose
(P=0.0041 for trend). In addition, all-cause mortality was low-
est at an INR of between 2.0 and 2.99. Analyses based on the
Cox proportional hazard model showed that thromboembolic
risk was significantly lower at an INR of between 1.6 and 2.99
than that in the non-warfarin group, whereas hemorrhagic risk
increased along with the rise in the INR value (Table 5). This
suggested that an INR of 1.6–2.6 was safe and effective at
preventing thromboembolic events in Japanese patients with
NVAF, as Yasaka et al reported previously.6 An INR of 2.6 to
2.99 was also effective but associated with a slightly increased
risk of major hemorrhage.
The elderly patients group (≥70 years old) displayed simi-
lar results to those for the whole patient group (Tables 4,5).
Thromboembolic risk was lowest at an INR of 1.6 to 2.99, and

Circulation Journal  Vol.77, September 2013

2268 INOUE H et al.

Table 5.  Hazard Ratios of Thromboembolic and Major Hemorrhagic Events Among Japanese Patients With Non-Valvular AF in the
J-RHYTHM Registry
Warfarin
Non- P value
INR level
warfarin for trend
≤1.59 1.6–1.99 2.0–2.59 2.6–2.99 ≥3
Whole group
  Thromboembolism 1.00 0.65 (0.40–1.07) 0.42 (0.26–0.70)　　 0.47 (0.28–0.79) 0.18 (0.04–0.73)　　 0.60 (0.18–1.95)　　 0.1351
  Major hemorrhage 1.00 1.85 (0.83–4.09) 2.21 (1.04–4.70)　　 2.94 (1.39–6.24) 3.96 (1.62–9.69)　　 5.26 (1.91–14.50) 0.0014
<70 years old
  Thromboembolism 1.00 0.53 (0.21–1.31) 0.29 (0.11–0.79)　　 0.57 (0.24–1.34) 0 0 0.3768
  Major hemorrhage 1.00 1.45 (0.36–5.82) 3.22 (0.95–10.93) 2.49 (0.70–8.84) 3.12 (0.63–15.45) 0 0.7424
≥70 years old
  Thromboembolism 1.00 0.63 (0.35–1.15) 0.42 (0.23–0.76)　　 0.39 (0.20–0.75) 0.23 (0.05–1.01)　　 0.78 (0.23–2.65)　　 0.2528
  Major hemorrhage 1.00 1.82 (0.68–4.88) 1.59 (0.61–4.16)　　 2.87 (1.12–7.35) 3.99 (1.33–11.89) 7.02 (2.23–22.13) 0.0002
Data in parentheses are 95% confidence intervals.

the risk of hemorrhagic events was significantly increased at
an INR ≥2.0. Again, these findings suggested that the optimal
target INR for Japanese elderly patients with NVAF ranges
from 1.6 to 2.6.5,6 An INR of 2.6–2.99 was associated with a
trend toward an increase in major hemorrhage.
The results for the younger patient group (<70 years old)
were somewhat different from those of the whole group and
the elderly group (Table 4). The frequencies of thromboem-
bolic and major hemorrhagic events were very low, and throm-
boembolic events did not occur at an INR ≥2.6. The incidence
of major hemorrhagic events did not differ significantly among
the non-warfarin and warfarin groups.
Discussion
Major Findings
The major findings of the present observational study were as
follows. First, in the whole warfarin group, INR ≥1.6 was ef-
fective at reducing the frequency of thromboembolic events
among Japanese patients with NVAF. However, the frequency
of major hemorrhagic events increased as the INR value in-
creased and was higher in each warfarin group than in the non-
warfarin group. Second, for elderly Japanese patients (≥70
years old) with NVAF, the risk of thromboembolic events was
lowest at an INR of 1.6–2.99. At an INR ≥2.6, the risk of hem-
orrhagic events was 4-fold higher than that of the non-warfarin
group. These findings support the Japanese guidelines5 for tar-
get anticoagulation levels among elderly patients with NVAF
(ie, target INR between 1.6 and 2.6). An INR of 2.6–2.99 was
effective at reducing thromboembolic events, but associated
with a slight increase in major hemorrhagic events as compared
with an INR of 1.6–2.59. Third, in the younger patient group
(<70 years old), the optimal target INR values for preventing
thromboembolic and major hemorrhagic events could hardly
be determined, which might be related to younger patients with
NVAF being at lower risk of thromboembolic and hemor-
rhagic events.
Anticoagulation Levels
An INR between 2 and 3 is recommended in the guidelines
used in Western countries for the management of AF.3,15 Previ-
ous studies from Western countries have revealed that the risk
of stroke increased among patients with INR values between
1.5 and 2.16,17 Even in patients ≥75 years of age, an INR of 2–3
were found to be effective at reducing the frequencies of fatal
stroke and non-fatal disabling stroke in the primary care set-
ting.18 The investigators in the ATRIA study reported that an
INR of between 2 and 3 were effective at preventing thrombo-
embolism and major hemorrhage in patients with NVAF, ir-
respective of their age and CHADS2 score.4 However, their
results suggested that increased intracranial hemorrhage would
occur in patients ≥70 years old at an INR >2.6.4 A meta-anal-
ysis involving patients with various pathological conditions
that required anticoagulation treatment with warfarin also
showed that an INR of between 2 and 3 was optimal for pre-
venting thromboembolic events without increasing the risk of
hemorrhagic events.19 Taken together, studies from Western
countries have repeatedly suggested that the optimal target
INR for managing patients with AF ranges from 2 to 3.
However, 2 major prospective randomized studies from
Western countries have demonstrated the efficacy of slightly
lower INR values in patients with NVAF.20,21 The BAATAF
study used a target INR of between 1.5 and 2.7 in 420 patients
with nonrheumatic AF (mean age, 68 years old),20 and the SPI-
NAF study used a target INR of between 1.4 and 2.8 in 571
male patients with nonrheumatic AF (mean age, 67 years old).21
These 2 studies clearly demonstrated that lower INR values
are effective at reducing the risk of thromboembolic events. In
the SPINAF study, the utility of lower INR values for prevent-
ing ischemic stroke was evident, particularly in patients ≥70
years of age.21
As for Asian NVAF patients, only a few prospective studies
have attempted to determine the optimal INR values for this
patient group. Yasaka et al6 combined the results of 2 prospec-
tive secondary prevention trials to determine the optimal INR
values for Japanese patients with NVAF. They found that major
ischemic stroke and systemic embolism only occurred in pa-
tients with an INR <1.6, and major hemorrhage only devel-
oped in patients with an INR ≥2.6 (mean, 2.81).6 Based on
these findings, INR values between 1.6 and 2.6 were recom-
mended as optimal for preventing embolic events and major
hemorrhagic events in Japanese patients with NVAF.5 In a
retrospective cohort study, Naganuma et al7 found that an INR
of between 1.5 and 2.5 was associated with low incidences of
thromboembolic and major bleeding events in 845 elderly
Japanese NVAF patients (≥70 years old; mean, 74 years old)
with CHADS2 scores ≥2. In addition, they found that an INR
≥2.5 was associated with a trend toward an increased risk of
major hemorrhage.7
Among Chinese patients receiving warfarin, lower INR
values were also suggested to be optimal for preventing em-
bolic and bleeding events. In a retrospective study involving

Circulation Journal  Vol.77, September 2013

Target INR in NVAF
555 Chinese patients (mean age, 69.7 years) with valvular AF
or NVAF who were taking warfarin, Cheung et al8 found that
an INR range of 1.5–1.9 was associated with the lowest fre-
quencies of thromboembolic and serious bleeding events. They
concluded that an INR range of 1.5–3.0 was safe and effective
at preventing stroke in Chinese AF patients.8 Another retro-
spective study by You et al enrolled 491 Chinese patients
(mean age, 65.8 years) who had recently started taking warfa-
rin for various indications with the aim of achieving a target
INR of between 2 and 3.9 AF was the main indication for
warfarin in 72% of the patients. The incidence of hemorrhagic
or thromboembolic events was lowest at an INR of between
1.8 and 2.4. These studies from China8,9 and Japan,6,7 as well
as the present study suggest that an INR slightly less than 2.0
(ie, 1.5 or 1.6–2.) is effective at preventing embolic events in
Japanese and Chinese patients with NVAF. It should be noted
that the risk of major hemorrhagic events is increased at INR
values exceeding 2.5 or 2.6.6,7,9 Based on an observational
study, Suzuki et al reported that an INR of 2.27 or higher was
the only significant predictor of major hemorrhage in patients
receiving warfarin.22
In a previous study, the effects of the INR value on hemo-
static marker levels were analyzed in Japanese patients with
NVAF.23 It was found that the reduction in D-dimer levels
observed at an INR of between 1.5 and 1.99 was similar to that
seen at an INR of 2.00 or higher.23 Therefore, an INR of be-
tween 1.5 and 1.99 is also effective at preventing thrombus
formation in the cardiovascular system, which suggests that
lower INR values (between 1.5 and 1.99) are as effective as
INR values of between 2 and 3 at preventing embolic events
in patients with NVAF.
Study Limitations
The present study had several limitations. First, the study de-
sign was observational, and the antithrombotic drugs were se-
lected by the participating physicians, although the patients
were followed-up prospectively for 2 years. In addition, the
non-warfarin group consisted of patients who were at low
risk of thromboembolism according to their CHADS2 scores
(mean, 1.2). This could have hampered our ability to examine
the utility of warfarin for preventing thromboembolic events,
but this was not the case. The incidence of thromboembolic
events was lower in the warfarin groups with INR values ≥1.6
than in the non-warfarin group. Second, our analyses only in-
volved the subjects’ baseline antithrombotic therapy status and
baseline INR values. The present study collected outpatients
who were being treated by the participating physicians, and the
antithrombotic treatment and dosage were selected by the par-
ticipating physicians. Therefore, the baseline INR values were
used as values that represented anticoagulation intensity. The
effects of switching antithrombotic therapy were not taken into
consideration in the present analyses. These limitations could
have explained the slightly higher rate of thromboembolic
events in patients with an INR ≥3.0. However, this could not
be the case, because the hemorrhagic event rate was highest in
these patients. More sophisticated analyses involving the time
in the therapeutic range24 need be done to determine the asso-
ciations between the variation in INR values during the follow-
up period and event frequencies. Third, 110 patients were lost
to follow-up in the present study, which could lead to under-
reporting of endpoints. The incidence of CI among the elderly
patients (≥70 years old) with baseline INR values of 2.0–2.59
(mean CHADS2 score of 2.2, Table 2) in the present study was
16/1,004 or 1.59%/2 years, a slightly lower rate as compared
with the Japanese patients in the warfarin arm (mean CHADS2
Circulation Journal  Vol.77, September 2013
2269
score of 2.2; target INR 2.0–2.6 for elderly patients aged ≥70
years, and 2.0–3.0 for the remaining patients) in the RE-LY
trial (1.33%/year).10 As for intracranial hemorrhage, the inci-
dence was similar between the present study (14/1,004 or
1.39%/2 years for elderly patients with INR of 2.0–2.59) and
the RE-LY trial (0.66%/year for Japanese patients receiving
warfarin). Therefore, the effects of underreporting of endpoints
because of a lack of follow-up data in a small proportion
(1.5%) of the patients should be limited in the present study.
Finally, the patients in the <70-year-old group were at low risk
of thromboembolism (ie, the mean CHADS2 score was 0.7 in
the non-warfarin group and ranged from 1.1 to 1.5 in the war-
farin groups). Therefore, the incidence rates of thromboembo-
lism and major hemorrhage were quite low in this age group
and did not differ significantly among the non-warfarin group
and the warfarin group (Table 4).
Conclusion
Although our study was affected by the stated limitations, it
clearly demonstrated that an INR of 1.6–2.6 is a safe and ef-
fective target for the prevention of thromboembolic events in
Japanese patients with NVAF, particularly those aged ≥70
years. An INR between 2.6 and 2.99 is also effective, but as-
sociated with a trend toward an increased risk of major hemor-
rhage. The management of Japanese patients with NVAF
should be based on data obtained from Japanese patients.
Disclosures
Conflict of Interest: Dr Inoue received research funding from Boehringer
Ingelheim and Daiichi-Sankyo, and remuneration from Daiichi-Sankyo,
Bayer Healthcare and Boehringer Ingelheim; Dr Okumura received re-
search funding from Boehringer Ingelheim and Daiichi-Sankyo, and remu-
neration from Boehringer Ingelheim, Bayer Healthcare, Daiichi-Sankyo,
and Pfizer; Dr Atarashi received research funding from Daiichi-Sankyo
and Boehringer Ingelheim, and lecture fees from Bayer Healthcare and
Boehringer Ingelheim; Dr Yamashita received research funding from
Boehringer Ingelheim and Daiichi-Sankyo, and remuneration from Boeh-
ringer Ingelheim, Daiichi-Sankyo, Bayer Healthcare, Pfizer, Bristol-Myers
Squibb, and Eisai; Dr Origasa received lecture fees from Daiichi-Sankyo;
Dr Kawamura received lecture fees from Boehringer Ingelheim; Dr Kubota
received research funding from Daiichi-Sankyo; Dr Watanabe received
lecture fees from Bayer Healthcare and Boehringer Ingelheim; Dr Koretsune
received remuneration from Boehringer Ingelheim, Bayer Healthcare, and
Daiichi-Sankyo; Dr Okuyama received lecture fees from Boehringer In-
gelheim; Dr Shimizu received lecture fees from Boehringer Ingelheim and
Bayer Healthcare; Dr Bando received lecture fees from Bayer Healthcare
and Boehringer Ingelheim; Dr Saikawa received research funding from
Boehringer Ingelheim; and Dr Chishaki received lecture fees from Bayer
Healthcare. No other potential conflict of interest relating to this article are
reported.
References
  1. Atrial Fibrillation Investigators. Risk factors for stroke and efficacy
of antithrombotic therapy in atrial fibrillation: Analysis of pooled
data from five randomized controlled trials. Arch Intern Med 1994;
154: 1449 – 1457.
  2. Connolly SJ, Pogue J, Eikelboom J, Flaker G, Commerford P, Franzosi
MG, et al; on behalf of the ACTIVE W investigators. Benefit of oral
anticoagulant over antiplatelet therapy in atrial fibrillation depends on
the quality of international normalized ratio control achieved by cen-
ters and countries as measured by time in therapeutic range. Circula-
tion 2008; 118: 2029 – 2037.
  3. Fuster V, Reiden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen
KA, et al. ACC/AHA/ESC guidelines for the management of patients
with atrial fibrillation: Executive summary. Circulation 2006; 114:
700 – 752.
 4. Singer DE, Chang Y, Fang MC, Borowsky LH, Pomernacki NK,
Udaltsova N, et al. Should patient characteristics influence target an-
ticoagulation intensity for stroke prevention in nonvalvular atrial fi-
brillation? The ATRIA study. Circ Cardiovasc Qual Outcomes 2009;
2: 297 – 304.
2270
  5. JCS Joint Working Group. Guidelines for pharmacotherapy of atrial
fibrillation (JCS 2008): Digest version. Circ J 2010; 74: 2479 – 2500.
  6. Yasaka M, Minematsu K, Yamaguchi T. Optimal intensity of interna-
tional normalized ratio in warfarin therapy for secondary prevention
of stroke in patients with non-valvular atrial fibrillation. Intern Med
2001; 40: 1183 – 1188.
  7. Naganuma M, Shiga T, Sato K, Murasaki K, Hashiguchi M, Mochizuki
M, et al. Clinical outcome in Japanese elderly patients with non-val-
vular atrial fibrillation taking warfarin: A single-center observational
study. Thromb Res 2012; 130: 21 – 26.
  8. Cheung CM, Tsoi TH, Huang CY. The lowest effective intensity of
prophylactic anticoagulation for patients with atrial fibrillation. Cere-
brovasc Dis 2005; 20: 114 – 119.
  9. You JHS, Chan FWH, Wong RSM, Cheng G. Is INR between 2.0
and 3.0 the optimal level for Chinese patients on warfarin therapy for
moderate-intensity anticoagulation? Br J Clin Pharmacol 2005; 59:
582 – 587.
10. Hori M, Connolly SJ, Ezekowitz MD, Reilly PA, Yusuf S, Wallentin
L. Efficacy and safety of dabigatran vs. warfarin in patients with atrial
fibrillation: Sub-analysis in Japanese population in RE-LY trial. Circ J
2011; 75: 800 – 805.
11. Hori M, Matsumoto M, Tanahashi N, Momomura S, Uchiyama S,
Goto S, et al; on behalf of the J-ROCKET AF study investigators.
Rivaroxaban vs. warfarin in Japanese patients with atrial fibrillation:
The J-ROCKET AF study. Circ J 2012; 76: 2104 – 2111.
12. Atarashi H, Inoue H, Okumura K, Yamashita T, Origasa H; for the J-
RHYTHM Registry investigators. Investigation of optimal anticoagu-
lation strategy for stroke prevention in Japanese patients with atrial
fibrillation: The J-RHYTHM Registry study design. J Cardiol 2011;
57: 95 – 99.
13. Atarashi H, Inoue H, Okumura K, Yamashita T, Kumagai N, Origasa
H; for the J-RHYTHM Registry investigators. Present status of anti-
coagulation treatment in Japanese patients with atrial fibrillation: A
report from the J-RHYHTHM Registry. Circ J 2011; 75: 1328 – 1333.
14. J-RHYTHM Registry Investigators. Determinants of warfarin use
and international normalized ratio levels in atrial fibrillation patients
in Japan: Subanalysis of the J-RHYTHM Registry. Circ J 2011; 75:
2357 – 2362.
15. Camm AJ, Lip GYH, De Caterina R, Savelieva I, Atar D, Hohnloser
SH, et al. 2012 focused update of the ESC guidelines for the manage-
ment of atrial fibrillation: An update of the 2010 ESC guidelines for
the management of atrial fibrillation developed with the special con-
Circulation Journal  Vol.77, September 2013
INOUE H et al.
tribution of the European Heart Rhythm Association. Eur Heart J
2012; 33: 2719 – 2747.
16. Reynolds MW, Fahrbach K, Hauch O, Wygant G, Estok R, Cella C, et
al. Warfarin anticoagulation and outcomes in patients with atrial fibril-
lation: A systematic review and meta-analysis. Chest 2004; 126: 1938 – 
1945.
17. Oden A, Fahlen M, Hart RG. Optimal INR for prevention of stroke
and death in atrial fibrillation: A critical appraisal. Thromb Res 2006;
117: 493 – 499.
18. Mant J, Hobbs R, Fletcher K, Roalfe A, Fitzmaurice D, Lip GYH, et
al on behalf of the BAFTA investigators and the Midland Research
Practice Network (MidReC). Warfarin versus aspirin for stroke pre-
vention in an elderly community population with atrial fibrillation
(the Birmingham Atrial Fibrillation Treatment of the Aged Study,
BAFTA): A randomised controlled study. Lancet 2007; 370: 493 – 
503.
19. Oake N, Jennings A, Forster AJ, Fergusson D, Doucette S, van
Walraven C. Anticoagulation intensity and outcomes among patients
prescribed oral anticoagulant therapy: A systematic review and meta-
analysis. Can Med Assoc J 2008; 179: 235 – 244.
20. The Boston Area Anticoagulation Trial for Atrial Fibrillation Inves-
tigators. The effect of low-dose warfarin on the risk of stroke in pa-
tients with nonrheumatic atrial fibrillation. N Engl J Med 1990; 323:
1505 – 1511.
21. Ezekowitz MD, Bridgers SL, James KE, Carliner NH, Dolling CL,
Gornick CC, et al; for the Veterans Affairs Stroke Prevention in Non-
rheumatic Atrial Fibrillation Investigators. Warfarin in the prevention
of stroke associated with nonrheumatic atrial fibrillation. N Engl J
Med 1992; 327: 1406 – 1412.
22. Suzuki S, Yamashita T, Kato T, Fujino T, Sagara K, Sawada H, et
al. Incidence of major bleeding complication of warfarin therapy in
Japanese patients with atrial fibrillation. Circ J 2007; 71: 761 – 765.
23. Nozawa T, Inoue H, Iwasa A, Okumura K, Lee JD, Shimizu A, et al.
Effects of anticoagulation intensity on hemostatic markers in patients
with non-valvular atrial fibrillation. Circ J 2004; 68: 29 – 34.
24. Rosendaal FR, Cannegieter SC, van der Mer FJ, Briet E. A method to
determine the optimal intensity of oral anticoagulant therapy. Thromb
Haemost 1993; 69: 236 – 239.
Appendix
The participating physicians are listed in references 12–14.