ARTICLE IN PRESS

Suboptimal Anticoagulant Management in Japanese Patients

with Nonvalvular Atrial Fibrillation Receiving Warfarin for
Stroke Prevention

Teruyuki Hirano, MD, PhD,* Hirokazu Kaneko,† Sari Mishina, MS,†

Feng Wang, MD, PhD,‡ and Satoshi Morita, PhD§

Background: Atrial fibrillation (AF) is the most common cardiac arrhythmia, with
increasing prevalence in Japan. Although prothrombin time–international normal-
ized ratio (PT-INR) targets for monitoring warfarin therapy in patients with nonvalvular
AF (NVAF) are well defined, real-world patient characteristics and PT-INR levels
remain unknown among Japanese patients with NVAF who initiate and continue
warfarin (warfarin maintainers) versus those who switch from warfarin to direct oral
anticoagulants (DOACs; warfarin switchers). Methods: Patients with NVAF receiv-
ing oral anticoagulants between February 2013 and June 2015 were identified using
a nationwide electronic medical record (EMR) database from 69 hospitals in Japan.
Demographics and characteristics of patients, PT-INR, time in therapeutic range (TTR),
and frequency in range (FIR) of PT-INR between warfarin maintainers and warfa-
rin switchers were assessed. Results: A total of 1705 patients met inclusion criteria
and were examined (1501 warfarin maintainers versus 204 warfarin switchers). CHADS2,
CHA2DS2-VASc, and HAS-BLED scores were comparable between groups. However,
these scores were significantly higher among warfarin switchers at the time of switch-
ing than at the time of warfarin initiation. Furthermore, TTR and FIR of PT-INR
were lower in warfarin switchers than in maintainers. Nevertheless, TTR and FIR
were below 50% (PT-INR, 1.6-2.6) in both patient groups. Conclusions: In this EMR-
based clinical study, patients who switched to DOACs had both poor or inadequate
PT-INR control and higher risk factors of stroke. Many patients receiving warfa-
rin did not achieve sufficient PT-INR therapeutic range. DOACs could be
recommended in Japanese patients with NVAF with inadequate PT-INR control
and increased risk of stroke. Key Words: Prothrombin time–international normalized
ratio—oral anticoagulant—atrial fibrillation—warfarin—time in therapeutic
range—DOAC.
© 2017 The Authors. Published by Elsevier Inc. on behalf of National Stroke
Association. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

From the *Department of Stroke and Cerebrovascular Medicine, Kyorin University School of Medicine, Tokyo, Japan; †Bristol-Myers Squibb
K.K., Tokyo, Japan; ‡Bristol-Myers Squibb Company, Princeton, New Jersey, USA; and §Department of Biomedical Statistics and Bioinfor-
matics, Kyoto University Graduate School of Medicine, Kyoto, Japan.
Received March 1, 2017; revision received April 18, 2017; accepted April 24, 2017.
Grant support: This study was funded by Bristol-Myers Squibb K.K. (CV185-492).
Conflict of interest: T.H. reports receiving honoraria from Nippon Boehringer Ingelheim, Pfizer Japan, Bristol-Myers Squibb K.K., Daiichi
Sankyo K.K., Bayer, and Otsuka Pharmaceutical Co Ltd. S.M. reports receiving honoraria from AstraZeneca, Bristol-Myers Squibb K.K., Chugai
Pharmaceutical Co, Ltd., and Eisai Co Ltd. F.W. is an employee of Bristol-Myers Squibb. H.K. and S.M. are employees of Bristol-Myers Squibb
K.K.
Address correspondence to Teruyuki Hirano, MD, PhD, 6-20-2, Shinkawa, Mitaka, Tokyo 181-8611, Japan. E-mail: terry@ks.kyorin-u.ac.jp.
1052-3057/$ - see front matter
© 2017 The Authors. Published by Elsevier Inc. on behalf of National Stroke Association. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2017.04.030

Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2017: pp ■■–■■ 1
 ARTICLE IN PRESS
2 T. HIRANO ET AL.

Introduction [J-RHYTHM] Registry) have validated these targets23-25 and

provided other data on the status of anticoagulation therapy
Globally, the age-adjusted prevalence of atrial fibrilla-
in Japan. However, real-world data focusing on patient
tion (AF) was approximately 33 million in 2010, entailing
characteristics and PT-INR levels in Japanese patients with
a major public health burden.1 As in Western countries,
NVAF who started and continued on warfarin (warfa-
AF prevalence has been increasing in Japan as the pop-
rin maintainers) versus those who switched from warfarin
ulation ages; prevalence was 0.56% (720,000) in 2003 and
to DOACs (warfarin switchers) in the era of DOAC avail-
is expected to rise to 1.09% (>1 million) by 2050.2 The
ability are limited. Therefore, the objective of this
risk of stroke is approximately 5 times higher in pa-
retrospective observational study was to assess patient
tients with AF.3 Although anticoagulation therapy reduces
characteristics and anticoagulation management in pa-
stroke risk in patients with nonvalvular AF (NVAF),4-10
tients, including variation across time of PT-INR levels.
substantial undertreatment with anticoagulants such as
Additionally, time in therapeutic range (TTR) and fre-
warfarin persists,11 possibly because of the need to balance
quency in range (FIR) time were compared between
maintenance of high anticoagulation intensity to prevent
warfarin maintainers and warfarin switchers.
stroke with the importance of low-intensity anticoagu-
lant treatment to minimize the risk of major bleeding,
especially in patients at high risk of bleeding.
Methods
Warfarin, a vitamin K antagonist, is routinely used for Data Sources
anticoagulation and is effective in reducing stroke and death
The nationwide Japanese electronic medical record (EMR)
risk in patients with NVAF.7 Despite possessing a number
database was established in 2001. It now contains case-
of limitations,12 including food and drug interactions, the
based information from regional hospitals in Japan with
need for close laboratory monitoring to ensure the pro-
approximately 3 million unique patient records. It has
thrombin time–international normalized ratio (PT-INR) is
been used for previous retrospective observational studies
maintained within a narrow therapeutic range, and a high
to investigate patient demographic and clinical charac-
risk of bleeding when the high end of this range is ex-
teristics, medication utilization patterns, and treatment
ceeded, warfarin has been used extensively for patients
outcomes.26,27 De-identified data (i.e., patient demograph-
with NVAF in clinical practice in Japan.13 Recently, non–
ics, drug prescriptions, comorbid conditions, and laboratory
vitamin K-dependent, direct oral anticoagulants (DOACs;
test data from in-patient and out-patient settings) were
dabigatran, rivaroxaban, apixaban, and edoxaban) have
available from 69 hospitals throughout Japan, specifical-
become available for the prevention of stroke and sys-
ly, 12 facilities with 1-99 beds, 27 facilities with 100-299
temic embolism in patients with NVAF. Although the risk
beds, 11 facilities with 300-499 beds, 16 facilities with 500-
of major bleeding remains, DOACs offer an improved risk-
899 beds, and 3 facilities with 900 beds or more. This
benefit profile (e.g., less risk of intracranial bleeding)
study was conducted as a database study, informed consent
compared with warfarin. Of note, though, risk profiles also
was opt-out style in each institution, and the study design
differ among DOACs.14-16 Compared with warfarin, DOACs
did not require institutional review board approval.
have a more rapid onset17; shorter half-lives, which can
aid timing of surgery17,18; fewer drug and food interactions19;
Study Population
no need for routine laboratory monitoring17; and poten-
tial for fixed dosing.19 Further, the efficacy and safety of For inclusion, eligible patients were 18 years and older
DOACs have been demonstrated in large randomized con- as of the index date, which was defined as the date of
trolled trials4-6,9 and under real-world conditions8,10,20 in initiation of any oral anticoagulant (OAC) such as war-
Western populations. However, results may not be entire- farin, dabigatran, rivaroxaban, apixaban, or edoxaban.
ly applicable to Japanese populations in light of known Eligible patients had at least 1 confirmed diagnosis of
differences in patient characteristics and clinical out- AF between January 1, 2001 (database inception) and the
comes between Asians and Westerners. For example, Asians index date, and no use of any anticoagulation for at least
taking warfarin for AF have higher rates for stroke and 4 months before the index date. Patients also had re-
bleeding than non-Asians.21 ceived at least 1 OAC prescription from February 1, 2013
According to the latest (2013) Japanese Circulation Society to June 30, 2015 (or the last date of the last data cutoff
(JCS) guidelines, DOACs should be considered, when- available at the time of execution of the study; Fig 1).
ever indicated, as first-line therapy in high- or intermediate- The AF diagnosis was confirmed by any medical record
risk patients with NVAF. 22 JCS guidelines also associated with an International Classification of Dis-
recommended PT-INR targets of 2.0-3.0 in warfarin- eases (ICD-10) code of I48 (AF and flutter), including
treated patients younger than 70 years and 1.6-2.6 for those persistent AF, chronic AF, typical atrial flutter, and atyp-
aged 70 years and older.22 Previous, large-scale, prospec- ical atrial flutter.
tive, observational studies in Japan (such as the Japanese Patients were excluded if they had 1 or more diagno-
Rhythm Management Trial for Atrial Fibrillation study sis of valvular heart disease (ICD-10 code of I05, I06, I34,
 ARTICLE IN PRESS
PT-INR MANAGEMENT IN PATIENTS ON WARFARIN
Figure 1. Study design. Abbreviations: DOAC,
direct oral anticoagulant (dabigatran, rivaroxaban,
apixaban, or edoxaban); EMR, electronic medical
record; NVAF, nonvalvular atrial fibrillation; OAC,
oral anticoagulant (warfarin, dabigatran, rivaroxaban,
apixaban, or edoxaban).
or I35) or venous thromboembolism (deep vein throm-
bosis: ICD-10 codes I26, I80 except I80.0, I81, I82, O03.2,
O04.2, O22.3, O87.1; or pulmonary embolism: ICD-10 codes
I26, 088.2) during the study period before the index date
or 1 or more perinatal-related diagnosis (ICD-10 code of
O00-O99) during the baseline or follow-up periods.
Eligible patients receiving OACs were classified into
2 treatment groups in the era of DOAC availability: war-
farin maintainers and warfarin switchers (Fig 1). Warfarin
maintainers included patients receiving warfarin as the
index OAC, had at least 120 days with no use of OACs
before the index date (baseline period), and continued
to use warfarin until the end of the observation period.
Warfarin switchers included those patients who had ini-
tiated warfarin use at least 120 days before the index date
who, however, switched to using a DOAC on the index
date, had no use of DOACs at least 120 days before the
index date, and had no use of any OACs at least 120
days before the first date of warfarin continuation (base-
line period). To ascertain the factors that influenced the
switch from warfarin to a DOAC, we compared patient
characteristics at the time of warfarin initiation and at
the time of switching in warfarin switchers.
Patients were followed from the index date to June 30,
2015, or the last date of the last data cutoff available at
the time of execution of the study. OAC initiation after
discontinuation was counted as a newly initiated patient.
To account for possible nonpersistence, we defined the date
of discontinuation as 120 days with no evidence of index
OACs for 120 days after the last prescription. The follow-
up was censored at the discontinuation of the index drug.
Patient Characteristics and PT-INR Measures
For each treatment group, demographic and baseline
clinical characteristics (e.g., comorbidities including the
Charlson comorbidity index28; CHADS2 score [conges-
tive heart failure, hypertension, age ≥75 years, diabetes
mellitus, and history of stroke or transient ischemic attack
{TIA}]; CHA2DS2-VASc score; and HAS-BLED score) were
collected. TTR and FIR were also captured for each group.
3
Patient-level TTR was calculated from PT-INR levels using
the Rosendaal et al interpolation method.29 For each patient,
all PT-INR tests were conducted from the date of first
warfarin exposure until the end of the follow-up period.
Patients with only 1 PT-INR test were excluded. For the
target therapeutic range, a PT-INR of 2.0-3.0 and 1.6-2.6
was applied for patients younger than 70 years and those
70 years and older, respectively (TTR 1), according to the
JCS guidelines.22 FIR was calculated as the proportion of
PT-INR levels within the target range based on the same
target therapeutic range (FIR 1). TTR and FIR based on
a PT-INR target range of 1.6-2.6 (TTR 2 and FIR 2) were
also calculated.
To evaluate how often PT-INR levels were monitored
and how the levels varied, the frequency and mean values
of PT-INR tests were calculated for each patient and com-
pared between treatment groups. The variations across
time in the PT-INR for each day following 30 days from
the start of the drug administration were observed in all
measurable patients.
Statistical Analyses
Demographic and baseline clinical characteristics were
compared between treatment groups. Categorical and con-
tinuous variables were tested using Pearson’s χ2 test and
Student’s t-test, respectively. For warfarin switchers, cate-
gorical and continuous variables were compared between
the measurements at warfarin initiation and at the time of
the switch to any DOAC using McNemar’s test and paired
t-tests, respectively. The differences in TTR, FIR, mean PT-INR
levels, and the number and the duration of PT-INR mea-
surements were tested using the Wilcoxon rank sum test.
All statistical analyses were performed using R Foun-
dation for Statistical Computing, Vienna, Austria (http://
www.r-project.org/) version 3.1.0.
Results
OACs were prescribed at 56 of 69 hospitals in the EMR
database. A total of 24,301 patients initiated warfarin, and
 ARTICLE IN PRESS
4 T. HIRANO ET AL.

Figure 2. Patient flow. Abbreviations: AF, atrial fibrillation; DOAC, direct oral anticoagulant; OAC, oral anticoagulant; VTE: venous thromboembolism.
*Many patients who were not diagnosed with AF but who had the disease and were prescribed warfarin for the treatment and prevention of thromboem-
bolism (venous thrombosis, myocardial infarction, pulmonary embolism, cerebral embolism, slowly progressing cerebral thrombosis, etc.) were excluded.

1479 patients initiated any DOAC after using warfarin
during the prior 120 days (Fig 2). Overall, 4272 patients
with AF and without valvular disease met the inclusion
criteria. Patients who were prescribed warfarin for treat-
ment and prevention of diseases other than AF such as
thromboembolism (venous thrombosis, myocardial in-
farction, pulmonary embolism, cerebral embolism, and
slowly progressing cerebral thrombosis, etc.) were ex-
cluded from this study. After excluding pregnant patients
and those with venous thromboembolism history during
the baseline and follow-up periods, 3430 patients re-
mained. Of these, 1501 patients were considered warfarin
maintainers, and 204 were deemed warfarin switchers.
Among the warfarin switchers, 107 switched to rivaroxaban,
53 switched to apixaban, 40 switched to dabigatran, and
4 switched to edoxaban. More than 1 INR measurement
was available for 754 patients in the warfarin maintainers
group and 89 patients in the warfarin switchers group.
TTR and FIR data were evaluated in 653 and 75 pa-
tients in the warfarin maintainers and the warfarin
switchers groups, respectively.
Patient Demographics and Baseline Characteristics
Patients who switched from warfarin were signifi-
cantly younger than warfarin maintainers (mean: 72.0
versus 75.7 years; P < .001). In addition, significantly more
warfarin switchers than warfarin maintainers were pre-
viously treated with aspirin, heparin, anti-epilepsy, or
antiplatelet drugs (P ≤ .0220); had a history of stroke, TIA,
renal disease, and alcohol-related disease (P ≤ .0294); and
were hospitalized for any coagulation disorder (pre-
scribed any drugs with anticoagulant effects in the hospital)
within 120 days before the index date (62.7% versus 39.6%;
χ2 = 38.314, df = 1, P < .001; data not shown). In contrast,
the CHADS2 (3.0 ± 1.5), CHA2DS2-VASc (4.4 ± 1.8), and HAS-
BLED (3.1 ± 1.4) scores of the warfarin switchers were
similar to the CHADS2 (2.9 ± 1.5), CHA2DS2-VASc (4.3 ± 1.9),
and HAS-BLED (3.0 ± 1.4) scores of the warfarin maintainers
(Table 1). Both groups had similar proportions of pa-
tients with CHADS2 score of zero; warfarin switchers (6.9%)
versus warfarin maintainers (5.7%), P = .6259.
Among warfarin switchers, CHADS2, CHA2DS2-VASc,
and HAS-BLED scores were significantly lower at war-
farin initiation than at the time of switching to a DOAC
(CHADS2, 2.8 ± 1.5 versus 3.0 ± 1.5; CHA2DS2-VASc, 4.1 ± 1.8
versus 4.4 ± 1.8; and HAS-BLED, 2.7 ± 1.4 versus 3.1 ± 1.4;
P < .001; Table 2). Further, significantly fewer patients had
a history of hypertension, congestive heart failure, or vas-
cular disease at warfarin initiation than at the time of
switching (P ≤ .0233). Significantly more patients had a
CHADS2 score of 0 (P = .0412) and were treated with
 ARTICLE IN PRESS
PT-INR MANAGEMENT IN PATIENTS ON WARFARIN 5
Table 1. Patient demographics and baseline characteristics

Warfarin Warfarin t-Statistics or

Characteristics maintainer switcher χ2 values P value

N 1501 204
Sex (female), n (%) 537 (35.8) 67 (32.8) .553 .4570
Mean (SD) age 75.7 10.6 72.0 11.3 −4.491 <.0001
Mean (SD) weight n 1172 144
kg 57.2 16.3 60.1 15.9 2.033 .0435
Mean (SD) CCr n 872 130
mL/min 53.2 28.9 70.1 51.0 5.684 <.0001
Mean (SD) CHADS2 score 2.9 1.5 3.0 1.5 1.363 .1741
Mean (SD) CHA2DS2-VASc score 4.3 1.9 4.4 1.8 .736 .4623
Mean (SD) HAS-BLED score 3.0 1.4 3.1 1.4 .427 .6698
Mean (SD) CCI 4.1 3.4 3.8 3.3 −1.175 .2412
Lowest CHADS2 score (CHADS2 = 0), n (%) 86 5.7% 14 6.9% .238 .6259
Labile PT-INR (TTR <60%), n (%) 457 30.4% 67 32.8% .379 .5384
Disease history, n (%)
Stroke or TIA 484 32.2% 89 43.6% 9.924 .0016
Diabetes mellitus 722 48.1% 97 47.5% .005 .9414
Hypertension 925 61.6% 133 65.2% .827 .3633
Congestive heart failure 820 54.6% 120 58.8% 1.113 .2915
Vascular disease 287 19.1% 50 24.5% 2.958 .0855
Severe or moderate renal disease 254 16.9% 12 5.9% 15.795 <.0001
Liver disease 212 14.1% 26 12.7% .181 .6705
Alcohol-related disease 18 1.2% 7 3.4% 4.745 .0294
Type of prescription drug(s) at baseline, n (%)
Anti-epilepsy 40 2.7% 12 5.9% 5.247 .0220
Antimycotic 4 .3% 0 0% .000 1.0000
Aspirin 117 7.8% 37 18.1% 22.138 <.0001
Heparin 499 33.2% 90 44.1% 8.915 .0028
LMWH 6 .4% 2 1.0% .351 .5534
Other antiplatelet* 88 5.9% 39 19.1% 43.867 <.0001

Abbreviations: CCI, Charlson comorbidity index; CCr, creatinine clearance; LMWH, low molecular weight heparin; PT-INR, prothrom-
bin time–international normalized ratio; SD, standard deviation; TIA, transient ischemic attack; TTR, time in therapeutic range.
Categorical and continuous variables were tested using Pearson’s χ2 test and Student’s t-test, respectively.
*Aspirin and platelet aggregation inhibitors were divided in this study.

aspirin, heparin, or antiplatelet drugs (P ≤ .0221) at war-
farin initiation than at the time of switching to a DOAC.
PT-INR Control
As described in Table 3, significantly fewer warfarin
switchers than maintainers achieved target TTR 1
(P < .0001), TTR 2 (P < .0001), FIR 1 (P = .0002), and FIR
2 ranges (P = .0016). Median TTR 1, TTR 2, FIR 1, and
FIR 2 were 36.4%, 45.4%, 33.3%, and 40.0%, respective-
ly, in warfarin maintainers, but only 4.8%, 18.5%, 16.7%,
and 31.8% in warfarin switchers.
Mean PT-INR values were not significantly different
between warfarin maintainers and warfarin switchers (1.732
and 1.713, respectively); however, the median number of
PT-INR tests conducted on each patient within 1 year after
warfarin initiation was significantly lower in warfarin
switchers (3.0) than in warfarin maintainers (5.0) (P = .0070;
Table 3). The PT-INR variation across time within 30 days
after warfarin initiation was consistently lower than the
therapeutic range, and the median was always below 2.0
(Fig 3). The PT-INR variation across time in warfarin
maintainers and warfarin switchers within the 30 days
after warfarin initiation is provided in Supplementary
Table S1.
Discussion
To the best of our knowledge, this is the first real-
world evidence from the EMR database in Japan showing
patient characteristics and subtherapeutic PT-INR range in
patients with NVAF taking warfarin. This study included
patients who were exclusively classified as having AF (ICD-
10 code of I48). Even though this study has high sensitivity,
the number of eligible patients was limited. Both TTR and
FIR were significantly lower in warfarin switchers, com-
pared with warfarin maintainers, suggesting that PT-INR
control needs to be more closely monitored in clinical practice
 ARTICLE IN PRESS
6 T. HIRANO ET AL.
Table 2. Changes in clinical characteristics between warfarin initiation and switch in warfarin switchers

At warfarin At warfarin
initiation switch Test value P value

N 204 204
Mean (SD) number of days before switch 96.4 145.4
Mean (SD) CCr n 78 130
mL/min 72.0 152.2 70.1 51.0 .033† .9737
Mean (SD) CHADS2 score 2.8 1.5 3.0 1.5 5.643† <.0001
Mean (SD) CHA2DS2-VASc score 4.1 1.8 4.4 1.8 5.846† <.0001
Mean (SD) HAS-BLED score 2.7 1.4 3.1 1.4 7.712† <.0001
Mean (SD) CCI 3.5 3.3 3.8 3.3 7.065† <.0001
Lowest CHADS2 score (CHADS2 = 0), n (%) 20 9.8% 14 6.9% 4.167‡ .0412
Disease history, n (%)
Stroke or TIA 84 41.2% 89 43.6% 3.200‡ .0736
Type 2 diabetes mellitus 93 45.6% 97 47.5% 2.250‡ .1336
Hypertension 119 58.3% 133 65.2% 12.071‡ .0005
Congestive heart failure 107 52.5% 120 58.8% 11.077‡ .0009
Vascular disease 43 21.1% 50 24.5% 5.143‡ .0233
Renal disease 12 5.9% 12 5.9%
Liver disease 24 11.8% 26 12.7% .500‡ .4795
Labile PT-INR (TTR <60%) 67 32.8% 67 32.8%
Alcohol-related disease 6 2.9% 7 3.4% .000‡ 1.0000
Type of prescription drug(s) at baseline, n (%)
Anti-epilepsy 10 4.9% 12 5.9% .100‡ .7518
Antimycotic 0 0% 0 0% — —
Aspirin 19 9.3% 37 18.1% 11.115‡ .0009
Heparin 47 23.0% 90 44.1% 32.073‡ <.0001
LMWH 6 2.9% 2 1.0% 2.250‡ .1336
Other antiplatelet* 22 10.8% 39 19.1% 9.482‡ .0021

Abbreviations: CCI, Charlson Comorbidity Index; CCr, creatinine clearance; LMWH, low molecular weight heparin; PT-INR, prothrom-
bin time–international normalized ratio; SD, standard deviation; TIA, transient ischemic attack; TTR, time in therapeutic range.
*Aspirin and platelet aggregation inhibitors were divided in this study.
†Paired t-test.
‡McNemar’s test.

in Japan. A higher proportion of warfarin switchers also
had a history of stroke or TIA and concomitant use of aspirin,
heparin, and other antiplatelet agents at baseline, in com-
parison with warfarin maintainers. The switch to DOACs
could be attributed to the incidence of stroke or TIA coupled
with inadequate PT-INR control. Baseline CHADS2 or
CHA2DS2-VASc and HAS-BLED scores were similar between
warfarin maintainers and warfarin switchers.
Although analysis of the J-RHYTHM Registry sug-
gested no benefit for warfarin in patients with a CHA2DS2-
VASc score of 0,30 we noted a greater proportion of patients
with CHADS2 score of 0 at warfarin initiation than at the
time of switching from warfarin. Taken together, the results
suggest that although the use of warfarin is high, DOACs
may also be considered for use in Japanese patients with
low event risks. Additionally, at the time of switch to
DOACs, patients were treated with aspirin, heparin, and
antiplatelet drugs, and their Charlson comorbidity index,
CHADS2, CHA2DS2-VASc, and HAS-BLED scores were
higher than at the time of warfarin initiation. This sug-
gests that an incident or event or inadequate PT-INR control
in high-risk patients might have occurred, justifying the
switch to DOACs. DOACs are preferentially recom-
mended for stroke prevention in Asian patients with AF
because of their improved bleeding profile and similar
or better stroke prevention ability.21 DOACs are also rec-
ommended upfront by the JCS guidelines.22 Although
Japanese physicians may choose to switch any patient
to DOACs to take advantage of these benefits, the results
suggest that vascular events might have affected their de-
cision to switch from warfarin to DOACs because a
significantly higher baseline stroke or TIA rate was ob-
served in warfarin switchers than in warfarin maintainers.
Warfarin use is also associated with an increased risk
of stroke within the first 30 days of initiation.31 Addi-
tionally, Japanese patients with NVAF may require narrower
PT-INR therapeutic ranges than Westerners.32 For these
patients, DOACs could offer an alternative to warfarin
as initial OAC therapy. The prospective, observational
Stroke Acute Management with Urgent Risk-factor As-
sessment and Improvement-Non Valvular Atrial Fibrillation
(SAMURAI-NVAF) study showed that the index stroke
 ARTICLE IN PRESS
PT-INR MANAGEMENT IN PATIENTS ON WARFARIN 7
Table 3. Variations across time and number of PT-INR measurements between groups

Warfarin Warfarin
maintainers switchers W statistic P value

Time or frequency of PT-INR in therapeutic range

N 653 75
Median MAD Median MAD
TTR 1 36.4% 44.9% 4.8% 7.1% 32,232 <.0001
TTR 2 45.4% 43.9% 18.5% 27.5% 31,274 <.0001
FIR 1 33.3% 34.6% 16.7% 24.7% 30,823 .0002
FIR 2 40.0% 29.7% 31.8% 37.6% 29,905 .0016
PT-INR per patient (mean) 1.665 (1.732) .430 (.423) 1.610 (1.713) .549 (.685) 27,982 (26,755) .0428 (.1888)
Duration and number of PT-INR tests per patient within 1 year after warfarin initiation
N 754 89
Median MAD Median MAD
Duration of PT-INR tests (d) 21.0 31.1 11.0 16.3 37,910 .0446
Number of PT-INR tests per patient 5.0 4.4 3.0 3.0 39,393 .0070
Number of PT-INR tests per patient 2.3 1.9 2.0 1.5 36,878 .1238
per month (30 d)*

Abbreviations: FIR 1, frequency in range defined as PT-INR between 2.0 and 3.0 for patients younger than 70 years, and between 1.6 and
2.6 for patients aged 70 years and older; FIR 2, frequency in range defined as PT-INR between 1.6 and 2.6 for all patients; MAD, median
absolute deviation; PT-INR, prothrombin time–international normalized ratio; TTR 1, time in therapeutic range defined as PT-INR between
2.0 and 3.0 for patients younger than 70 years and between 1.6 and 2.6 for patients aged 70 years and older; TTR 2, time in therapeutic
range defined as PT-INR between 1.6 and 2.6 for all patients.
Statistical tests were performed using Wilcoxon rank sum test.
*Calculated as the ratio between the number of PT-INR tests per patient and the duration of PT-INR tests per patient (months). The latter
was determined by dividing the number of days in the duration of PT-INR tests per patient by 30 days and rounding up to the nearest integer.

was milder in patients initiated with DOACs than in those
initiated with warfarin.33
Data from a global registry showed that whereas a greater
proportion of patients newly diagnosed with NVAF are pre-
scribed DOACs in Europe and North America, treatment
with vitamin K antagonists is more prevalent in Asia.34
Analysis of the Fushimi AF Registry showed that warfa-
rin is overused in low-risk patients in Japan.35 Although
the patients in this study were not anticoagulant naïve, the
results provide only subjective rationale for switching from
warfarin to DOACs, particularly in low-risk patients with
NVAF and those with poor PT-INR control.

Figure 3. PT-INR distribution curve of warfarin maintainers and warfarin switchers (within 30 days after warfarin initiation). The dashed purple line
indicates a PT-INR of 2.0-3.0 and the dashed and dotted blue line indicates a PT-INR of 1.6-2.6. Abbreviation: PT-INR, prothrombin time–international
normalized ratio.
 ARTICLE IN PRESS
8 T. HIRANO ET AL.

Optimal PT-INR Levels in Japanese Patients for all the patients included in this study. Second, clin-
with NVAF ical diagnoses by physicians registered as ICD-10 codes
may not have been recorded uniformly among physi-
In this study, the mean and median PT-INR along with
cians, especially for comorbidities with relatively mild
its fluctuation in each patient was consistently below 2.0.
symptoms. Third, only patients who were diagnosed as
This suggests that in Japan, the perceived risk of bleed-
ICD-10 code I48 were included in this study; however,
ing with warfarin is relatively tightly controlled, while
patients who received warfarin and had AF but were dif-
adhering to the JCS guidelines. A nested case-control study
ficult to diagnose with AF as per the ICD-10 code I48
in elderly (≥70 years) Japanese patients with NVAF re-
were excluded. Fourth, the presence of a prescription record
ceiving warfarin showed that patients with a PT-INR
does not ensure treatment compliance.
greater than or equal to 3 had approximately 20 times
the risk of developing major bleeding compared with those
with a PT-INR of 2.0-2.5; patients whose PT-INR was 2.0- Conclusions
2.5 or 2.5-3.0 had a similar risk. 36 Analysis of the
Our results showed that patients who switched to
J-RHYTHM Registry supports targeting a PT-INR of 1.6-
DOACs had both poor or inadequate PT-INR control and
2.6 in Japanese patients with NVAF, even for those patients
higher risk factors of stroke, suggesting that switching
younger than 70 years (but >65 years)23 and 85 years and
to DOACs should be recommended for this patient group.
older.24 Large real-world studies of Japanese patients with
Most patients with NVAF taking warfarin in this study
NVAF or AF have shown that only approximately half
did not achieve adequate PT-INR control. Therefore, PT-
of patients attain the therapeutic PT-INR range.23,35 Another
INR needs to be more closely monitored among patients
analysis of 6404 warfarin-treated patients with NVAF from
on warfarin in Japan. Furthermore, patients with unsta-
the J-RHYTHM Registry showed a gradual increase in
ble PT-INR control may benefit from the use of DOACs.
the risk of ischemic stroke and systemic embolism for
Further studies in Japan are needed to evaluate the reasons
patients treated with a PT-INR less than 2.0.32
for switching from warfarin to DOACs based on safety,
TTR is an optimal measure of PT-INR control.37 Signifi-
effectiveness, and compliance outcomes.
cantly fewer strokes were observed in patients treated with
than in those without warfarin having a CHADS2 score
Acknowledgments: This study was funded by Bristol-
greater than or equal to 2, TTR greater than 70%, and a
Myers Squibb K.K. Data analysis was provided by Kinya
PT-INR of 2.0-3.0.37 Yet undertreatment with anticoagu-
Kokubo, Ayako Shoji, and Hironaga Kudo of Nomura Re-
lants is common in patients with NVAF with a CHADS2
search Institute, Ltd, and was funded by Bristol-Myers Squibb
score greater than or equal to 2.38 In the present study, TTR
K.K. Medical writing support was provided by Steven Tresker
1, TTR 2, FIR 1, and FIR 2 were below 50%, and much
of Cactus Communications, and was funded by Bristol-
lower in warfarin switchers. Thus, a large proportion of
Myers Squibb K.K.
the high-risk Japanese patients in our study did not achieve
optimal warfarin anticoagulation, particularly those who
switched to DOACs. To confirm the impact of a low target Appendix: Supplementary material
range, we also calculated TTR and FIR based on the con-
Supplementary data to this article can be found online
sistent target therapeutic range between 1.6 and 2.6 (TTR
at doi:10.1016/j.jstrokecerebrovasdis.2017.04.030.
2 and FIR 2) and found them to be similarly low. Our results
stand in contrast to those of the J-RHYTHM Registry, which
indicated that, even for high-risk patients 85 years and older, References
warfarin was prescribed in approximately 80% of such pa-
1. Rahman F, Kwan GF, Benjamin EJ. Global epidemiology
tients, who maintained a TTR (with a PT-INR of 1.6-2.6) of atrial fibrillation. Nat Rev Cardiol 2014;11:639-654.
of 67.1%.24 Differences in age and baseline characteristics, 2. Inoue H, Fujiki A, Origasa H, et al. Prevalence of atrial
and the fact that J-RHYTHM Registry patients were from fibrillation in the general population of Japan: an analysis
institutions that specialized in cardiology, could explain this based on periodic health examination. Int J Cardiol
2009;137:102-107.
difference. In contrast to J-RHYTHM, the strength of the
3. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as
present study is the use of an EMR database with a large an independent risk factor for stroke: the Framingham
sample size reflecting the standard of care in a real-world study. Stroke 1991;22:983-988.
scenario in Japan. 4. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran
versus warfarin in patients with atrial fibrillation. N Engl
J Med 2009;361:1139-1151.
Study Limitations 5. Giugliano RP, Ruff CT, Braunwald E, et al. Edoxaban
In our opinion, there are 4 limitations to this study. versus warfarin in patients with atrial fibrillation. N Engl
J Med 2013;369:2093-2104.
First, because of the sample size of the warfarin switcher 6. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban
group, the study was not able to explore stroke and bleed- versus warfarin in patients with atrial fibrillation. N Engl
ing events. INR measurements were not available on record J Med 2011;365:981-992.
 ARTICLE IN PRESS
PT-INR MANAGEMENT IN PATIENTS ON WARFARIN
7. Hart RG, Pearce LA, Aguilar MI. Meta-analysis:
antithrombotic therapy to prevent stroke in patients who
have nonvalvular atrial fibrillation. Ann Intern Med
2007;146:857-867.
8. Lauffenburger JC, Farley JF, Gehi AK, et al. Effectiveness
and safety of dabigatran and warfarin in real-world US
patients with non-valvular atrial fibrillation: a retrospective
cohort study. J Am Heart Assoc 2015;4.
9. Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus
warfarin in nonvalvular atrial fibrillation. N Engl J Med
2011;365:883-891.
10. Villines TC, Schnee J, Fraeman K, et al. A comparison
of the safety and effectiveness of dabigatran and warfarin
in non-valvular atrial fibrillation patients in a large
healthcare system. Thromb Haemost 2015;114:1290-1298.
11. Bjorck S, Palaszewski B, Friberg L, et al. Atrial fibrillation,
stroke risk, and warfarin therapy revisited: a population-
based study. Stroke 2013;44:3103-3108.
12. Ansell J, Hirsh J, Hylek E, et al. Pharmacology and
management of the vitamin K antagonists: American
College of Chest Physicians Evidence-Based Clinical
Practice guidelines (8th edition). Chest 2008;133:160S-198S.
13. Atarashi H, Inoue H, Okumura K, et al. Present status
of anticoagulation treatment in Japanese patients with
atrial fibrillation: a report from the J-RHYTHM Registry.
Circ J 2011;75:1328-1333.
14. Graham DJ, Reichman ME, Wernecke M, et al.
Cardiovascular, bleeding, and mortality risks in elderly
Medicare patients treated with dabigatran or warfarin
for nonvalvular atrial fibrillation. Circulation 2015;131:157-
164.
15. Lip GY, Pan X, Kamble S, et al. Major bleeding risk
among non-valvular atrial fibrillation patients initiated
on apixaban, dabigatran, rivaroxaban or warfarin: a
“real-world” observational study in the United States.
Int J Clin Pract 2016;70:752-763.
16. Sharma M, Cornelius VR, Patel JP, et al. Efficacy and
harms of direct oral anticoagulants in the elderly for stroke
prevention in atrial fibrillation and secondary prevention
of venous thromboembolism: systematic review and
meta-analysis. Circulation 2015;132:194-204.
17. Heidbuchel H, Verhamme P, Alings M, et al. Updated
European Heart Rhythm Association Practical guide on
the use of non-vitamin K antagonist anticoagulants in
patients with non-valvular atrial fibrillation. Europace
2015;17:1467-1507.
18. Pollack CV. Introduction to direct oral anticoagulants and
rationale for specific reversal agents. Am J Emerg Med
2016;34:1-2.
19. Eikelboom J, Merli G. Bleeding with direct oral
anticoagulants vs warfarin: clinical experience. Am J
Emerg Med 2016;34:3-8.
20. Yao X, Abraham NS, Sangaralingham LR, et al.
Effectiveness and safety of dabigatran, rivaroxaban, and
apixaban versus warfarin in nonvalvular atrial fibrillation.
J Am Heart Assoc 2016;5.
21. Senoo K, Lau YC, Dzeshka M, et al. Efficacy and safety
of non-vitamin K antagonist oral anticoagulants vs.
warfarin in Japanese patients with atrial fibrillation—
meta-analysis. Circ J 2015;79:339-345.
22. JCS Joint Working Group. Guidelines for pharmacotherapy
of atrial fibrillation (JCS 2013): digest version. Circ J
2014;78:1997-2021.
23. Inoue H, Okumura K, Atarashi H, et al. Target
international normalized ratio values for preventing
9
thromboembolic and hemorrhagic events in Japanese
patients with non-valvular atrial fibrillation: results of
the J-RHYTHM Registry. Circ J 2013;77:2264-2270.
24. Kodani E, Atarashi H, Inoue H, et al. Use of warfarin
in elderly patients with non-valvular atrial
fibrillation—subanalysis of the J-RHYTHM Registry. Circ
J 2015;79:2345-2352.
25. Kodani E, Atarashi H, Inoue H, et al. Secondary
prevention of stroke with warfarin in patients with
nonvalvular atrial fibrillation: subanalysis of the
J-RHYTHM Registry. J Stroke Cerebrovasc Dis
2016;25:585-599.
26. Kodani E, Atarashi H, Inoue H, et al. Beneficial effect
of non-vitamin K antagonist oral anticoagulants in patients
with nonvalvular atrial fibrillation—results of the
J-RHYTHM Registry 2. Circ J 2016;80:843-851.
27. Teramoto T, Kokubo K, Shoji A. Exploring factors affecting
achievement of cholesterol management goal by statin
treatment in 2012 edition of atherosclerotic cardiovascular
disease prevention guidelines—a study of the Japan lipid
management view (J-LiView). Jpn Pharmacol Ther
2014;42:97-106.
28. Charlson ME, Pompei P, Ales KL, et al. A new method
of classifying prognostic comorbidity in longitudinal
studies: development and validation. J Chronic Dis
1987;40:373-383.
29. Rosendaal FR, Cannegieter SC, van der Meer FJ, et al.
A method to determine the optimal intensity of oral
anticoagulant therapy. Thromb Haemost 1993;69:236-239.
30. Okumura K, Inoue H, Atarashi H, et al. Validation of
CHA(2)DS(2)-VASc and HAS-BLED scores in Japanese
patients with nonvalvular atrial fibrillation: an analysis
of the J-RHYTHM Registry. Circ J 2014;78:1593-1599.
31. Azoulay L, Dell’Aniello S, Simon TA, et al. Initiation of
warfarin in patients with atrial fibrillation: early effects
on ischaemic strokes. Eur Heart J 2014;35:1881-1887.
32. Yamashita T, Inoue H, Okumura K, et al. Warfarin
anticoagulation intensity in Japanese nonvalvular atrial
fibrillation patients: a J-RHYTHM Registry analysis. J
Cardiol 2015;65:175-177.
33. Toyoda K, Arihiro S, Todo K, et al. Trends in oral
anticoagulant choice for acute stroke patients with
nonvalvular atrial fibrillation in Japan: the SAMURAI-
NVAF study. Int J Stroke 2015;10:836-842.
34. Huisman MV, Rothman KJ, Paquette M, et al.
Antithrombotic treatment patterns in patients with newly
diagnosed nonvalvular atrial fibrillation: the GLORIA-AF
Registry, Phase II. Am J Med 2015;128:1306-1313, e1.
35. Akao M, Chun YH, Esato M, et al. Inappropriate use
of oral anticoagulants for patients with atrial fibrillation.
Circ J 2014;78:2166-2172.
36. Ohgushi A, Ohtani T, Nakayama N, et al. Risk of major
bleeding at different PT-INR ranges in elderly Japanese
patients with non-valvular atrial fibrillation receiving
warfarin: a nested case-control study. J Pharm Health
Care Sci 2016;2:2.
37. Wan Y, Heneghan C, Perera R, et al. Anticoagulation
control and prediction of adverse events in patients with
atrial fibrillation: a systematic review. Circ Cardiovasc
Qual Outcomes 2008;1:84-91.
38. Suzuki S, Yamashita T, Okumura K, et al. Incidence of
ischemic stroke in Japanese patients with atrial fibrillation
not receiving anticoagulation therapy—pooled analysis
of the Shinken Database, J-RHYTHM Registry, and
Fushimi AF Registry. Circ J 2015;79:432-438.