DOI: 10.1161/CIRCULATIONAHA.113.

002601

A National Assessment of Warfarin Anticoagulation Therapy for

Stroke Prevention in Atrial Fibrillation

Running title: Dlott et al.; Anticoagulant Management in the United States

Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Jeffrey S. Dlott, MD1; Roberta A. George, MS2; Xiaohua Huang, MS2; Mouneer Odeh, MS3;

Harvey W. Kaufman, MD3; Jack Ansell, MD4; Elaine M. Hylek, M

MDD5

1
Quest
Ques
Qu esst Di
Diagnostics
Diag
a noost
ag stic
ics Nichols Institute, Chantilly
Chantilly,
ly V ; 2Quest Diagno
y, VA;
VA Diagnostics,
ost
stic
i s, West Norriton, PA;
ic
3
Quest
Q uest Diagnostics,
Diag
Di agno
nost
no s iccs, Madison,
st Mad
adis
ison N ; 4De
on,, NJ;
on NJ Dept
D pt ooff Me
Medicine,
edicine
ne,, Lenox
ne Leno
Len x Hill
no Hil
illl Ho
Hos
Hospital,
sp tal
spital,, New
New York,
York
Yo rk,, NY;
rk NY;
5
Deept of Medicine,
Dept Meediccine,, Section
Secctio
on of
of General
Gen
ner
eral
a Internal
al Inteernnal Medicine,
Med
edic
iccin
ne, Boston
Bostton Me
Medical
edi
dica
call Cent
C
Center,
en err, Bo
Bost
Boston,
ton,, M
MA
A

Address
Addr
Ad dres
esss for
for Correspondence:
Corr
Co rres
espo
pond
nden
ence
ce::
Jeffrey S. Dlott, MD
Quest Diagnostics Nichols Institute
Coagulation and Hematology
14225 Newbrook Drive
Chantilly, VA20151
Tel: 703-802-7259
Fax: 703-802-7099
E-mail: Jeffrey.S.Dlott@QuestDiagnostics.com

Journal Subject Codes: Anticoagulants:[184] Coumarins

1
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Abstract

Background—Anticoagulation control with warfarin, as assessed by the international

normalized ratio (INR), is challenging. Time in the therapeutic range (TTR) has been inversely

correlated with major hemorrhage, thrombosis, and mortality. Quest Diagnostics offers

standardized INR laboratory testing services to approximately half of U.S. physician practices.

To inform national stroke prevention strategies, we evaluated anticoagulation control in office-

based community practices.

Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Methods and Results—We selected individuals with •2 months of INR data, INR results of

>1.2, and an INR diagnosis code of atrial fibrillation. Frequency of INR testing aand TTR
nd T were
TR w ee
er

analyzed by age, sex, length of testing period, number of referred patients per provider, and

median
me
medi
dian
di an hhousehold
ouse
ou seehoold income
income (based on home ZIP code.
odee. We identified
cod identifiied 138,319
e 138
38,3
38 ,319 individuals referred

bby
y 337,939 physicians,
7 939 phys
7, ysic
iciaans
ic yielding
ns,, yi
iel ding a ttotal
eldding otaal ooff 2,6
ot 2,683,674
6833,67
74 IN
INR results.
R re
ressult
l s. P
lt Patients
atient
atie nts ha mean
hadd a me n aage
ean ge ooff 74
ge

years;
ye
year s;; most
ars;
ar ostt (81%)
mos (81%
(8 %) were
1% •65
were • 65 years
yeaars old
old and
and 55%
% were
55% •75.
werre • 75.. Th
75 Thee mean TTR
mean T TR
Rwwas
ass 553.7%
7% ooverall
3.7%
3.7 veralll aand
vera n
nd

improved
mproved with
ithh time
wit t me on
ti on treatment,
t eaatm
tr nt, increasing
tmen inc
ncreeas ingg from
asin
in f om 47.6%
fr 47.6
47 % for
.6% for patients
pati
pa tien
ti t with
nts ithh <6
wit < months
mon thss of testing
onth
th

to 57.5% for those with •6 months (p < 0.0001). The number of patients tested per physician

practice was positively associated with TTR. Younger age, female sex, and lower income were

also independently associated with poorer anticoagulant control.

Conclusions—This study demonstrates widespread suboptimal anticoagulation control,

suggesting an urgent need to improve oral anticoagulation care for most patient segments in the

United States.

Key words: anticoagulant, coagulation, stroke, atrial fibrillation, INR

2
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Introduction

Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting approximately 2.2

million Americans 1. AF increases the risk of stroke approximately five-fold, and these strokes

confer a 30-day mortality of 24% 2,3. The vitamin K antagonist warfarin greatly reduces the risk

of stroke and is the most commonly used therapy to prevent stroke and stroke recurrence in

patients with AF 2,4,3,5. Because of its variable dose-response relationship and narrow therapeutic

window, warfarin requires frequent monitoring of its anticoagulant effect as reflected by the

International Normalized Ratio (INR) 6. Percent time in the therapeutic range (TTR; INR of 2.0
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

- 3.0) has been shown to correlate with mortality, stroke, and major hemorrhage5,7,8.

To date, few studies have assessed the quality of warfarin management iin
n ro
routine
out
utin
inee cl
in clin
clinical
inic
in ical

care outside of anticoagulation clinics. Quantification and evaluation of this metric is timely, as

he
health
eal
alth
th care
carre providers
provvid
pr ideers
er and organizations are increasingly
increa
easiingly faced with
ea wit
itth decisions
deeci
cissions
si regarding

transitioning
ran s tioning to more
nsi moree costly
cost
co ly alternatives
stlly
st alt
lter
erna tivees to warfarin
nattive
na wa rin that
warfar thaat may
maay provide
proovid
pr bbetter
ovidee be tter aanticoagulation
nticooagu
ntic oagu
g la
l tiion
on

control,
co
ont ol, ssuch
n ro ucch as ddabigatran.
abig
abig
gat
atrran I a rrecent
ran. In ecen
ec en study,
nt st udy, tthe
stud
ud level
eveel of IINR
he le NR ccontrol
onntr oll bby
trol ywwarfarin
a fa
ar f riin wa sshown
wass show
how
wn to
t

be a factor inn determining

det
e er
ermi
m ni
mi n the
n ng the relative
rel
elat
el a iv
at ve cost
cost effectiveness
effe
effect
fe c iv
ctiven
enesss of ddabigatran
nes ab
big
igat
atra versus
rann ve
ra vers
rsus
rs u w
us riin 9.
warfarin
arfa
arfari
fa

Maximal benefit of dabigatran was derived by centers in the lowest quartile of INR control (i.e.,

TTR <57.1%) 9. Indeed, the quality of warfarin control may factor into the consideration to

switch established patients prescribed warfarin to a novel anticoagulant such as dabigatran10,11

The goal of our study was to provide a comprehensive and unbiased assessment of

anticoagulation care in the United States by taking advantage of a unique data resource. Quest

Diagnostics provides laboratory testing services to approximately half of the physician practices

in the United States, with more than 145 million patient encounters annually. The national scope

of this database, coupled with the uniformity of the INR assay, provides an unprecedented

3
 DOI: 10.1161/CIRCULATIONAHA.113.002601

opportunity to assess anticoagulation care in routine practice outside of hospitals (medical

centers). We also examined the relationship of TTR to demographic features, physician account

size, geographic region, duration of INR monitoring, and economic status.

Methods

Study Design and Data Collection

Test results from patients with AF were collected from the Quest Diagnostics Informatics Data

Warehouse. The study period extended two years, from January 1, 2007, through December 31,
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

2008. To be eligible for inclusion, patients had to be at least 18 years of age and have a diagnosis

of AF recorded (ICD-9-CM code 427.31). Individuals with a diagnosis code representing

rep
prese ingg heart
s ntin
se ntin hear
he art

valve replacement (V43.3) were excluded. Patients with two or more consecutive months of

INR
NR te st rresults
test esul
es u ts coupled
ul oup ed with more than one INR
coupl R value
vaalue of >1.2 during
r ngg a 12-month
duuri 12-month period were

deemed
m to be oon
deemed
deem nwwarfarin
arf
rfaarin
rf in aand
nd eeligible
ligi
li gibble
gi for
bl fo the
or th study.
he stud To eensure
dy. To nsure llongitudinal
ure lon
ongit
ngitud warfarin
u inall w arfa
ar fari
fa riin

management
ma
ana
nage
geme
ge meent aand
nd
dmminimize
inim
in imiz
imizee bi ass iinherent
bias nher
nh ent to eepisodic
ereent piso
pi s dic
so dic ccare, aanalyses
aree, anal
nal
alys were
yssess w restricted
erre re
rest
stri
st riict d tto
c ed o ppatients
ati
tiien
entts

with continuous
ouss uninterrupted
continuuou unin
un inte
in teerr tedd INR
r uppte INR testing.
t sttin
te ing.
g In
g. Individuals
ndi
d vi
v du
dual
a s with aany
with gap
ap in ttesting
ny ga t ngg ggreater
esti
es ti reat
re ater
at e than 60

days were excluded, as were individuals with extreme testing frequency (over 100 annualized

tests). Finally, INR data from hospitals (medical centers) were excluded to avoid patients who

may have been hospitalized or managed by hospital-based anticoagulation clinics.

INR test results were extracted from the Quest Diagnostics Data Warehouse, along with

patient demographics and insurance status. Age was categorized into seven groups (see

Appendix). Economic status was assigned using the median income from the 2007 United States

Census from the patient’s home ZIP Code. Median income was categorized into five groups.

This study was determined to be exempt from institutional review by the Western

4
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Institutional Review Board (Olympia, WA).

Laboratory

All Quest Diagnostics laboratories (>100) in the U.S., use a standard thromboplastin reagent

from the same lot with an International Sensitivity Index of 0.97 [Siemens- Innovin® Marburg,

Germany]. A recombinant human source of tissue factor is used. All assays were performed

using Siemens automated coagulation analyzers. Quality control procedures include both

validation and surveillance correlation analyses for each instrument/reagent combination across

the reportable INR range compared to company mean values. These correlation coefficients are
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

routinely 0.99 or greater for every instrument combination. Standardization of the INR assay

minimizes the variability in measurement inducedd by differences in reagents, ins

strum
umen
uments
ents,, an
ts
instruments, andd

assay methods and ensures comparability across sites.

Sttat
atis
isti
is tica
ti
Statistical ca
al Analysis
Anallys
An ysis
is

Statistical
St tis
Stat i tical analyses
anal
alys
yssess of
of summary
suumm
mmar
aryy st
ar statistics
tat
atis
isttics we
were
eree per
performed
rform
med
ed uusing
siing S
SAS
AS vversion
AS ersi
er sion
on
n 99.2
.2 ((SAS
SAS
SA S In
Institute
nst
stit
itut
it utte

Inc.,
c , Cary,
nc. C ry
Ca y, NC,
NC, U.S.).
U.S.).
U.S. ). Means
eanss and
Mea nd standard
sta
tand ardd deviations
nddar devi
viat
vi ions were
atiion eree used
wer e too describe
ussed ribee tthe
d scrribe
de he ppatient
atiient
at ientt

characteristics
characteristiccs for
f r age
fo age and
and number
numb
nu mber
mbe of
er of INR
IN
NR tests
test
te stss per
st p r year.
pe year
ye arr. T
TTR
TR w
was
ass ccalculated
alcu
al cula
culate
la t d ac
te acco
according
cord
co rdin
rd i g to the

Rosendaal linear interpolation method using a time interval of 60 days 12. The mean patient TTR

was used for analysis and comparison. Frequency of INR testing and TTR were analyzed by

age, sex, median income, geographic region (U.S. Health and Human Services; HHS), and

physician account size. Because the initiation phase of warfarin is associated with wide

variability in INR control, TTRwas also analyzed according to the length of the INR testing

period: <6 months or •6 months.

To assess the impact of risk factors on TTR, we applied a beta regression model with a

probit link function using the ‘betareg’ open source statistical analysis package in R3.0.213. The

5
 DOI: 10.1161/CIRCULATIONAHA.113.002601

beta regression model reflects the bounded property of our dependent variable (TTR is bounded

by 0 and 1), and is particularly well suited for proportions data such as TTR. The explanatory

variables included sex, age group, median income level, physician account size, HHS region,

INR testing frequency, and length of testing. Overall model significance was assessed using the

pseudo R-square and the log-likelihood ratio test. Coefficient estimates were evaluated using

partial Wald statistics. Slope estimates measured in TTR units were imputed from the beta

regression coefficient estimates using the probability density function of the standard normal

distribution evaluated at the mean of the fitted values.

Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Results

Identified
dentified were 138,319 individuals with AF referred for testing by 37,939 unique physicians,

providing
pr
rovvid
idin
ingg a to
in totall of analysis.
of 2,683,674 INR results for ana
naly
na mean
ysis. The mea
an agee of the patient population

was
w ass 74
7 years, an were
andd 55 % w ere age
ere age 75 yyears
eaars oorr oolder
ldeer ((Table
ble 11).
Taable ). The total
The to
tota observation
taal ob
obsservat
ser
rvattio
ionn time was
tiime w as

130,124
13
30,
0,12
1244 pe
12 person-years.
pers
rson
rs on-y
-yeears
-y The
ears.. The aaverage
v ra
ve ge llength
rage engthh ooff follow-up
ngt fol
o low up was
low-uup was 8.8
8.8 person-months
per
erso
s n-m
so mont
months (IQR:
n hs (IQR: 4.2
IQR: 4..2

months, 19.44 months);

mon
onth s);; 62%
ths)
th s) 6 % of
62 o the cohort
the co ort hhad
coho ad sstudy
dy llength
tudy
tu engt
engt •66 months.
gthh of • mo
ont
nths
hs..
hs

Time in the Therapeutic Range

Patients underwent INR testing approximately every 19 days. About half (50.6%) of INR results

were in the therapeutic range (INR = 2.0 to 3.0); 16.9% were >3.0: and 32.5% were <2.0. The

overall mean patient TTR was 53.7%. The length of the INR testing period (i.e., duration of

warfarin exposure in the study) was the most statistically significant factor associated with TTR:

TTR was 47.6% for patients with <6 months of INR data versus 57.5% for those with •6 months

(p < 0.0001). Younger individuals exhibited the poorest control and had a higher frequency of

testing (Figure 1). The overall mean TTR for individuals <45 years of age was 45.5% compared

6
 DOI: 10.1161/CIRCULATIONAHA.113.002601

to 53.9% for those age >75 years (p<0.001).

The relationship between TTR and age persisted beyond the first 6 months of INR

measurement. Important differences in TTR were also evident by sex: within every age

category, women had lower TTR (~2-3%) compared to men (p<0.001) and a higher frequency of

INR testing (Figure 1). This relationship persisted beyond the initial 6-month period.

Geographic region

The influence of geographic residence on TTR is presented according to the ten regions as

defined by the U.S. Department of Heath and Human Services (Table 1). In most regions, TTR
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

increased by approximately 10 percentage points when analyses were stratified by duration of

INR
NR testing. When we restricted our comparisons to regions with statistically robust
rob
obbusst

representation
epresentation (i.e., excluding Regions 8 and 10), the Northeast and Midwest regions (Regions 11,

2,, 33,, 5, aand

nd 77)) ha
had
ad hhigher
igher mean TTR (53.3% – 556.8%)
6.8%) than the su
6.8 sunbelt
unb states
n ellt st
stat
a es of the South, the

Southwest,
S ouuthw nd tthe
uth est, and West
hee W Coast
estt Co
es Coas (Region
astt (R
as Reg on 4, 66,, aand
egio
io mean
nd 9; meann TT
ea TTR 449.1%
R 49.1
9.1% 52.8%).
% – 52
52.8
.8%
.8 % ). R
%). Region
e io
eg i n1

(Connecticut,
Con
o ne
nect
ctic
iccut Massachusetts,
ut,, Massac
achu
huuse
sett New
t s, N Hampshire,
ew H am
mps
pshi
hiree, Rh
hi Rhode
hode Island,
ode Is
sla
lannd,
nd, an Vermont)
andd Ve
Verm
rm ont)) hhad
mont) a tthe
ad highest
he hi
ighest
igh
hest

mean TTR ((57% overall,

577% ov
over
e al
er l 660.0%
all,
l, 0% ffor
0.0%
0. or tthose
hoose iindividuals
ndiv
nd ivid
iv id
dua
uals with
ls w it •66 mo
ithh • months
ntths ttime
mont i tthe
imee in
im he sstudy),
tudy
tu d ), and thee

highest mean frequency of INR testing (25.2 tests per year). Regions 1, 3, and 7 achieved a

58% threshold when analyses were restricted to patients with at least 6 months of INR data 5.

Patient Volume

As shown in Figure 2, there was a direct relationship between the number of patients referred

(physician account size) and TTR that was evident in both the inception and established

warfarin periods. Of the 37,939 ordering physicians, nearly 95% ordered warfarin testing on

fewer than 10 patients. A larger referral panel was associated with increased TTR which may

partly represent participation of non-hospital based anticoagulation clinics.

7
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Other Factors

An association between income and TTR was found, as previously reported by Rose et al14.

Patients living in ZIP codes with the highest median incomes had higher mean TTR (Figure 3;

p<0.001). This relationship persisted beyond the first 6 months of INR measurement.

Multivariate Regression Model

Independent factors associated with TTR included length of the INR testing period, age, sex,

physician account size, median income level, and HHS region (Table 2). The pseudo R-square for

the model is 0.09353 and the p-value for the log-likelihood ratio test is <0.0001). Patients with a
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

shorter INR testing period (<6 months) had lower TTR than those with 6-24 months of testing (-

4.3%, 95% CI: -4.6% to -4.0%). Compared to patients in the •85 age group, thos
se in
those n tthe
he 555-64
5 64 aage
5- g
ge

group had higher TTR (2.4%, 95% CI: 1.9% to 2.9%), and those in the 35-44 age group had lower

TT
TR (-
TTR (-3.
3.8%
8%, 995%
8%
(-3.8%, 5%
%C I: -5.1% to -2.5%). Similarly,
CI: y ffemales
y, emales had low
wer
e T
lower TR than males (-1.3%, 95%
TTR

CI: -1.5%
CI: -1.5% to -1.0%).
-1.
1 0%
0%). Compared
Com
mpa
parred
red to accounts
acco
ccoun
ounts comprised
comprrissed of a large
com larg
largee nnumber
umb
ber
er ooff ppatients
atiennts ((>72),
atie > 2)
>72), th
tthose
ose

wiith oonly
with n y 2-
nl 2-44 pa
pat
tients
tie
ents ttested
patients esste
ted ha
hhad
d lo
lowe
werr TT
lower TR (-8.
TTR 8.6%
6%
%, 95
(-8.6%, 5% C
95% I: --9.1%
CI: 9 1% tto
9. o --8.1%).
8..1%)
%). Pa
P tiien
ents w
Patients ho llived
ho
who iveed
ed

n the lowestt me
in medi
diian inc
median ncom
ncom
me ar
income aarea,
ea,, de
ea efi
f ne
nedd ass <
defined $ 0,
$30,00
000,
000 hhad
<$30,000, ad llower
oweer TT
ow TTR
R th
than
an tthose
hoose iin
n th
tthee highest

income areas (•$80,000) (-5.6%, 95% CI: -6.3% to -4.9%). Finally, compared to HHS Region 1

(Connecticut, Massachusetts, New Hampshire, Rhode Island, and Vermont ), Region 6 (Arkansas,

Louisiana, New Mexico, Oklahoma, and Texas) had lower TTR (-7.5%, 95% CI: -8.2%, -6.8%)

In addition, we found a nonlinear association between INR testing frequency and TTR:

INR testing frequency was positively associated with TTR among patients with <14 INR tests per

year, but inversely associated with TTR among those with more frequent testing. This pattern

likely reflects increased testing among individuals with poor INR control (data not shown).

8
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Discussion

Quest Diagnostics provides laboratory testing services to approximately half of the physician

practices in the United States. The quality control and uniformity of INR measurement coupled

with the national breadth of the AF population represented provides a unique opportunity to

assess the quality of anticoagulation management in office-based community practices. We

purposefully excluded hospital-based accounts to enhance the representation of patients managed

outside of hospital-based anticoagulation clinics. Of the 37,939 ordering physician practices in

our study, nearly 95% ordered INR testing on fewer than 10 patients. Among the 138,319
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

individuals with AF evaluated, we found an overall mean TTR of 53.7%. The overall mean TTR

for 9 out of 10 HHS regions in the U.S. fell below the TTR threshold (58% to 65
65%) reported
5%) rep
epor
ep orte
or ted
te d to

be necessary to achieve the efficacy of warfarin versus dual antiplatelet therapy 5. Our TTR

esuult
ltss ar
results aree lo
low
werr th
lower thaan the center-based TTR valu
than ues
e rreported
values eported from
m rrecent
ecen
en
nt rrandomized
andomized trials:

ddabigatran
dab
abiga
g tran versus
verrsuus warfarin
warf
wa r arrin (mean
rf (me
mean
an TTR
TTR 64%),
64%
%), rivaroxaban
rivaaroxaabaan versus
vers
ve rsuus warfarin
war
arfa
f rin (mean
fa (mea
(meann TTR
TTR 55%),
555%)
%), and
a d
an

apix
ap
apixaban
ixab
ix aban
ab a vversus
an ersu
er suss w
warfarin
arf
rfaarin
in
n ((mean
meean
nTTTR 2%))115,16,17
TR 662%)
2% 5,16,
16,17
17
. When
When we
we restricted
r stri
re rict
ri c ed
ct d our
our analyses
ana
naly
lyse
ly sess too the
se the
h

established warfarin
waarffarrin subgroup,
sub
ubgr
grou
oup,
ou p TTR
TTR improved
imp
mpro
rove
ro v d to 58%
ve 58%
% ((vs
v 48%
vs 48% ffor
or individuals
ind
ndiv
ivid
ividua
idu lss w
ua ithh <6
it
with < months ooff

continuous INR data).

The stability of INR control has been shown to differ considerably between patients

starting warfarin and established users, in both clinical trials and observational studies 18. The

levels of anticoagulation control for the inception and established warfarin groups in our study

are similar to those reported from a large Veterans Health Administration population. In that

study of 100 anticoagulation clinics, the mean TTR values across all standard warfarin

indications (not limited to AF) were 48% for the inception warfarin period and 61% for the

established warfarin period 14. The overall mean TTR in our study (53.7%) is also similar to the

9
 DOI: 10.1161/CIRCULATIONAHA.113.002601

54.5% reported for 8,992 individuals with AF from Kaiser Permanente Southern California 19,

but lower than the 62.5% reported from Kaiser Permanente Northern California 20. Differences

in the proportion of patients newly starting warfarin, the proportion of patients managed through

anticoagulation clinics, and the health status of the AF populations studied may partly account

for the disparity.

We also found younger age, female sex, and lower income to be independently associated

with poorer anticoagulation control, similar to findings from the Veterans Health Administration
21
. The mechanisms underlying these associations are unknown but may be related to differences
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

in treatment adherence, polypharmacy, health status, or underlying pathobiology 21. The strong

association between improved TTR and physician account size may reflect volum
volume-related
um
me--re
rela
laated
teed dr
drug
ug

familiarity or better-organized warfarin management in the office setting or through a non-

hospital-based
ho
osp
spit
ital
it al-bas
bas
ased
e aanticoagulation
ed ntiicoagulation
nt ic clinic.

The nov
novelty
oveelty ooff ou
our study
ur st
stud lies
iess inn tthe
udyy li
ud aavailability
he ava
vailabbil y aand
bility analysis
nd an
naly
naly the
he nnational
ysiis off th atio
ationa Quest
naal Qu
Ques
estt

Diagnostics
Di
iag
agno
nost
no stic ddatabase,
iccs daata
taba
baase which
se,, whic
w pprovides
hichh pr ovid
ov es iinsight
ides
id ighht iinto
nssig o tthe
nto he aanticoagulation
ntic
nt icoa
ic oaagu latiion ccare
g lati provided
arre pr
prov
ovid
ov id outside
ded ou
utsside off

the
he hospital or
or medical
medi
me dica
di call center
ca c nt
ce er setting.
nter set
etti
et t ng
ti ng. The
The average
aver
av erag
eragee length
ag nggth of
leng
le of follow-up
foll
fo llow
ll ow
w-u
up wa 8.8
wass 8. person-months.
.8 pe
pers
rson
rs o -months.
on

To most objectively evaluate warfarin management, we required continuous serial INR

measurements for inclusion in the study. Because individuals with gaps in care may represent a

less healthy or less adherent population, it is possible that our results are an overestimate of TTR

in practice.

A limitation of our study is the lack of prescription drug data. However, we believe the

requirements for two or more months of serial INR measurements, two or more INR results of

>1.2, and a diagnosis code of AF collectively minimized misclassification of warfarin exposure.

Our analysis of 138,319 individuals with AF represents the most comprehensive

10
 DOI: 10.1161/CIRCULATIONAHA.113.002601

assessment of current anticoagulation care in routine practice in the United States to date, and

complements prior studies that focused on care rendered in organized health systems or

anticoagulation clinics. Our results demonstrate suboptimal warfarin treatment for

approximately half of individuals with AF. Strategies to optimize stroke prevention for this

population are urgently needed. Potential approaches include implementation of processes

shown to improve TTR and consideration of transitioning appropriate candidates to newer oral

anticoagulants 10,22,23.
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Conflict of Interest Disclosures: Dr. Ansell is a consultant for Bristol Myers Squibb, Janssen,
Boehringer-Ingelheim, Daiichi, Alere. He is on the Scientific Advisory Boards ooff
nstrumentation Laboratories and Perosphere, Inc (equity). Dr. Hylek receives ffunding
Instrumentation undi
un dinng
di ng ffrom
rom
ro m th
thee
NIH/NINDS (5R01NS070307). She is a researcher for NIH/NINDS, NIH/NHLBI,
/ Bristol-Myers
Squi
Sq uibb
uibb- Ex
bb
Squibb- Exec
ecutiv
ec iv
ve St
Executive S eering Committee, ARISTO
Steering OTL
TLE trial, Ortho
ARISTOTLE o-McN
cN
Neil-Executive
ei
Ortho-McNeil-Executive Steering
C om
mm
mmittee,
Committee, ORB
RBIT
IT-A
AF Registry.
ORBIT-AF Regi
Re gist
gistry
st ry.. She
ry She serves
seerv
ves
e on
on the
the advisory
ad
dvi
viso
sory
so ry boards
boa
o rd
ds of B ay
yer
er,, Bo
Bayer, Boeh
ehri
eh rinnger
ri nger
Boehringer-er--
ngeelh
l eim, Bristol-Myers
Ingelheim, Bri
r stool-
ol-Myyerrs Squibb,
Squ
quib
i b,
b Daiichi
Dai
aiic
ichhi Sankyo,
Saankyoo, and
and Johnson
John
Jo hnsson & Johnson,
hn Johnnson, Pfizerr. T
fi
Pfizer. he
The
emaain
inin
ingg auth
in
remaining aauthors
uthor
orss re
or rreport
port
po rt nno
o co
onf
nfli
lict
li cts.
conflicts.s

References:

1. Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, Halperin JL, Le
Heuzey JY, Neal KG, Lowe JE, Bertil OS, Prystowsky EN, Tamargo JL, Wann S . [Guidelines
for the management of patients with atrial fibrillation. Executive summary]. Rev Esp Cardiol.
2006;59:1329.

2. Hylek EM, Go AS, Chang Y, Jensvold NG, Henault LE, Selby JV, Singer DE . Effect of
intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J
Med. 2003;349:1019-1026.

3. Benjamin EJ, Wolf PA, Agostino RB, Silbershatz H, Kannel WB, Levy D . Impact of Atrial
Fibrillation on the Risk of Death. Circulation. 1998;98:946-952.

4. Lin HJ, Wolf PA, Kelly-Hayes M, Beiser AS, Kase CS, Benjamin EJ, D'Agostino RB . Stroke
Severity in Atrial Fibrillation. Stroke. 1996;27:1760-1764.

11
 DOI: 10.1161/CIRCULATIONAHA.113.002601

5. Connolly SJ, Pogue J, Eikelboom J, Flaker G, Commerford P, Franzosi MG, Healey JS, Yusuf
S . Benefit of oral anticoagulant over antiplatelet therapy in atrial fibrillation depends on the
quality of international normalized ratio control achieved by centers and countries as measured
by time in therapeutic range. Circulation. 2008;118:2029-2037.

6. Wysowski DK, Nourjah P, Swartz L . Bleeding complications with warfarin use: a prevalent
adverse effect resulting in regulatory action. Arch Intern Med. 2007;167:1414-1419.

7. White HD, Gruber M, Feyzi J, Kaatz S, Tse HF, Husted S, Albers GW . Comparison of
Outcomes Among Patients Randomized to Warfarin Therapy According to Anticoagulant
ControlResults From SPORTIF III and V. Arch Intern Med. 2007;167:239-245.

8. Ansell J, Hirsh J, Hylek E, Jacobson A, Crowther M, Palareti G . 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.
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

9. Shah SV, Gage BF . Cost-Effectiveness of Dabigatran for Stroke Prophylaxis in Atrial

Fibrillation / Clinical Perspective. Circulation. 2011;123:2562-2570.

10. Writing Group, Wann LS, Curtis AB, January CT, Ellenbogen KA, Lowe JE, E Estes
Est ess NAM,
stes NAM
AM,
Page RL, Ezekowitz MD, Slotwiner DJ, Jackman WM, Stevenson WG, Tracy CM . 2011
ACCF/AHA/HRS
ACCFC /AHA/HRS Focused Update on the Manag Management
agement of Patients With Atrial Fibrillation
(Updating
Upd
pdat
atin
at ingg th
in 2006
thee 20 2011;123:104-123.
006 Guideline). Circulation. 2011
11;1
11 123:104-123.

111.
1. Pengo
Pengo V, Crippa
ripppa L,
Cri L, Falanga
Fal
alan
an ga A,
nga A, Finazzi
nazzi G,
Finaz G Marongiu
Maaroonggiuu F,
F, Palareti
Palare
Pala ti G,
reti G, Poli
Poli D,
D, Testas a S,
Teest S Tiraferri
Tiraf
iraf
afer
errri
E, Tosetto
o etto A,
Tos A, Tripodi
ipoodi A, Manotti
Trip Manoanottii C . Questions
Questiionns and
nd answers
an an
nsw errs on
swer on the
the usee ofof dabigatran
dabiga
dabbig trran n and
andd
perspectives
pers
pe rspe
rs pect
pec iv
ct es on
ives on the
the use
use of o other
othherr new
th new oralr l anticoagulants
ora anttico
an coaagul
co nts in ppatients
gulannts atie
at i nt
ie n s with
w th atrial
wi atriall fibrillation.
tria
ia f brrilllati
fi tiionn. A
consensus
cons
nsen ddocument
ensuss do cume
cu mentnt ooff th Italian
thee It
tal
alia Federation
iann Fe
F dera
rati
tion Thrombosis
on off Th
T ro
omb
mbososis Centers
is C ennte rs ((FCSA).
ters FCSA
FC SA).) Th
Thromb mb Haemost.
Hae
aemo
most st.
2011;106:868-876.
2011;106:868 68
8-88766.

12. Rosendaal FR, Cannegieter SC, van der Meer FJ, Briet E . A method to determine the
optimal intensity of oral anticoagulant therapy. Thromb Haemost. 1993;69:236-239.

13. Francis Cribari-Neto, Achim Zeileis. Beta Regression in R. J Stat Softw. 2010;34:1-24.

14. Rose AJ, Hylek EM, Ozonoff A, Ash AS, Reisman JI, Berlowitz DR . Patient characteristics
associated with oral anticoagulation control: results of the Veterans AffaiRs Study to Improve
Anticoagulation (VARIA). J Thromb Haemost. 2010;8:2182-2191.

15. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A, Pogue J, Reilly
PA, Themeles E, Varrone J, Wang S, Alings M, Xavier D, Zhu J, Diaz R, Lewis BS, Darius H,
Diener HC, Joyner CD, Wallentin L . Dabigatran versus warfarin in patients with atrial
fibrillation. N Engl J Med. 2009;361:1139-1151.

16. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W, Breithardt G, Halperin JL,
Hankey GJ, Piccini JP, Becker RC, Nessel CC, Paolini JF, Berkowitz SD, Fox KA, Califf RM .

12
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883-891.

17. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, Al-Khalidi HR,
Ansell J, Atar D, Avezum A, Bahit MC, Diaz R, Easton JD, Ezekowitz JA, Flaker G, Garcia D,
Geraldes M, Gersh BJ, Golitsyn S, Goto S, Hermosillo AG, Hohnloser SH, Horowitz J, Mohan
P, Jansky P, Lewis BS, Lopez-Sendon JL, Pais P, Parkhomenko A, Verheugt FW, Zhu J,
Wallentin L . Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med.
2011;365:981-992.

18. Hylek EM, Frison L, Henault LE, Cupples A . Disparate stroke rates on warfarin among
contemporaneous cohorts with atrial fibrillation: potential insights into risk from a comparative
analysis of SPORTIF III versus SPORTIF V. Stroke. 2008;39:3009-3014.

19. Shen AY, Yao JF, Brar SS, Jorgensen MB, Wang X, Chen W . Racial/Ethnic differences in
ischemic stroke rates and the efficacy of warfarin among patients with atrial fibrillation. Stroke.
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

2008;39:2736-2743.

20. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, Capra AM, Jensvold NG NG, Selby
G, Se
Selb
lbyy JV
JV,
Singer DE . Anticoagulation therapy for stroke prevention in atrial fibrillation: ho
how well
ow weell ddo
o
randomized
andomized trials translate into clinical practice? JAMA. 2003;290:2685-2692.

Osende
21. Os
O ende JI,
JI, Badimon JJ, Fusterr V, Herson P, Rabito P, Vidhun R, Zaman A, Rodriguez OJ,
Lev
Le
ev EI Rauch
EI,, Ra
Rauc
u h U,
uc U H Heflt
eflt G, Fallon JT, Crandall JP . B Blood thrombogenicity
lood thrombo bo
ogeni
niici
citty in type 2 diabetes
mellitus
mell
me lllit patients
itus pat ieentts is associated
atie ass
ssoc
o ia
oc t d with
iate
te with glycemic
gly
yce micc control.
cemi cont
co ntro
ro Coll
ol. J Am C Cardiol.
olll Ca
C rdio
rd ioll. 2001;38:1307-1312.
l. 200
001;
1;38
1;38:1
38:130
307-
7-13
1312
12..
12

22.
22 Garcia
2. G DA,
arcia D A W
A, Witt
itt DM
DM, HHylek
ylek
le E,, WWittkowsky
itttkow
wskky AK AK, NuNutescu
Nutetesscuu EA
te EA, Jacobson
A, Jaco obson
bsonn A A,, M
Moll S,, M
o ll S Merli GJ,
errli GJ
Crowther
Crow
Cr owth
ow ther
ther M,
M, Earl
Earrl L,
Ea L, Becker
c er RC,
Beck
Beck RC, Oertel
erteel L,
Oerte L, JJaffer
afffe Ansell
ferr A,, A nssel
elll JE
E.D Delivery
eliv
el i eryy off O
iv Optimized
ptim
pt im
miz
izedd
Anticoagulant
Anti
tico
coagul Therapy:
ulaantt Th
T erap
er apy: CConsensus
onse
on sens Statement
nsuus State
temmen from
entt fr
froom tthe
he AAnticoagulation
ntic
nt icoa
o guula
lati
tion
on FForum.
o um
or um. A Ann
nnn
Pharmacother.
Pharmacother 2008;42:979-988.
err. 20
2008
08;4
08 ;4
42:
2 97
79-
9 98
988.
8.

23. Granger CB, Armaganijan LV . Newer Oral Anticoagulants Should Be Used as First-Line
Agents to Prevent Thromboembolism in Patients With Atrial Fibrillation and Risk Factors for
Stroke or Thromboembolism / Response to Granger and Armaganijan. Circulation.
2012;125:159-164.

13
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Table 1. Atrial Fibrillation Patient Characteristics, INR Frequency, and Time in the Therapeutic
Range.

Patients Study Time

All <6 Months •6 Months
N = 138,319 n = 52,142 n = 86,177
Age (years)
Mean (± SD) 74.1 (±11.2) 73.4 (±12.1) 74.6 (±10.6)
• 65, n (%) 112,252 (81.2%) 40,756 (78.2%) 74,496 (86.4%)
• 75, n (%) 76,569 (55.4%) 27,740 (53.2%) 48,829 (56.7%)
Sex, n (%)
Female 67,741 (49.0%) 25,094 (48.1%) 42,647 (49.5%)
Male 70,578 (51.0%) 27,048 (51.9%) 43,530 (50.5%)
INR Tests per Year, Median
(25th, 75th %-iles)
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Overall 19.4 (15.1, 26.6) 22.9 (16.9, 32.3) 17.9 (14.4, 23.6)
Female 19.9 (15.4, 27.3) 23.2 (17.1, 33.2) 18.5 (14.7, 24.5)
Male 18.9 (14.9, 26.1) 22.6 (16.6, 31.7) 17.
7.4
17.44 (1
(14.
4.1,
1, 222.8)
(14.1, 2.8)
2. 8
NR Tests per Month
INR
Median (25th, 75th %-iles) 1.6 (1.3, 2.2) 1.9 (1.4, 2.7) 11.5
.5
5 (1
(1.2,
1.2,
2 2.0)
2.0)
0)
NR Categories, n (%)
INR 2,683,674 416,743 2,266,931
<1.5 255,08
0 5 (9.5%)
255,085 (15.
(1 5.6%)
64,854 (15.6%) 190,231 (8.4%)
1.55 – 2.0
1.
1.5 2.0 617,71717
71
617,717 7 (23.0%)
(23.0%) 102, 2,,56
5 5 (24.6%)
102,565 (24
(2 4.
4.6%) 515,152 (22.7%)
2.0 – 3.0
2.0 1,35
1, 357,
35 7,08
1,357,08282 (50.0..6%
6%)) 1175,568
(50.6%) 75,5568
75 6 ((42.1%)
42.1%)
42 %)) 1,181,514
1,1
,18181,5
,514
14 (52.1%)
(52
52.1
. %)
.1 %
> .0 – 4.0
>3
>3.0 3328,465
28
8,4665 (12.2%)
(12.
2.2 2%)
2%) 47
7,2247 (11.3%)
47,247 (11.3% %)
%) 2281,218
8 ,2
81 , 18 ((12.4%)
12.44%)
12.4
>4.0
>4.0 12
25,32
125,32525 (4.7%)
(4.7
(4 .7%)
.7 %) 26
6,5509 ((6.4%)
26,509 6.44%)
%) 998,816
8,,81
816 ((4.4%)
4.4%)
%)
%
TTR (Mean
TTR (Mea
(M eann %,
ea % ±S D)
SD) 53.77 (±
53.7 (± 23.3)
233.3
.3)) 47.66 (±
47.6 (± 277.3
.3))
27.3) 57.55 (±
57 (± 19
19..6)
.6)
19.6)
Geog
og
gra
r ph
p ic R
Geographic eggion – St
Region States,, TT
TR
TTR
Mean %, ± S
(Mean D)
SD)
CT ME,
1 – CT, ME MA,
MA NH,
NH RI,
RI VT 56 8 (±
56.8 (± 21.7)
21 7) 50 4 (±
50.4 (± 26 0)
26.0) 60 0 ((±
60.0 ± 18 3)
18.3)
2 – NJ, NY 54.0 (± 23.2) 47.7 (± 27.2) 57.4 (± 19.8)
3 – DE, DC, MD, PA, VA, WV 55.4 (± 22.1) 48.3 (± 26.3) 58.7 (± 19.0)
4 – AL, FL, GA, KY, MS, NC, SC, TN 52.8 (± 24.5) 48.2 (± 28.4) 56.4 (± 20.2)
5 – IL, IN, MI, MN, OH, WI 53.3 (± 23.8) 47.6 (± 27.4) 57.2 (± 20.1)
6 – AR, LA, NM, OK, TX 49.1 (± 24.9) 43.4 (± 27.8) 53.6 (± 21.4)
7 – IA, KS, MO, NE 56.7 (± 22.5) 50.4 (± 27.4) 59.9 (± 18.8)
8 – CO, MT, ND, SD, UT, WY 51.8 (± 25.4) 46.0 (± 28.1) 58.1 (± 20.1)
9 – AZ, CA, HI, NV 51.9 (± 23.5) 46.1 (± 27.2) 55.8 (± 19.7)
10 – AK, ID, OR, WA 58.8 (± 22.7) 56.3 (± 26.1) 61.2 (± 18.4)
Insurance Type
Medicare 66,531 (49.8%) 25,067 (49.8%) 41,464 (49.7%)
Private Insurance 63,616 (47.6%) 23,133 (45.9%) 40,483 (48.6%)
Other 3,572 (2.7%) 2,172 (4.3%) 1,400 (1.7%)
INR indicates international normalized ratio; TTR, time in therapeutic range.

14
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Table 2. Coefficient and Slope Estimates of Regression Model.

Beta Regression Model* TTR Units**

(Probit link function) (original scale units)
Explanatory Variable*** ȕ 95% LL 95% UL Slope 95% LL 95% UL
Intercept 0.82 0.80 0.85 n/a n/a n/a
Test Frequency freq -0.03 -0.03 -0.03 -1.1% -1.1% -1.0%
freq^2 2.33E-4 2.19E-4 2.47E-4 9.3E-5 8.7E-5 9.8E-5
Study Length <6 months -0.11 -0.11 -0.10 -4.28% -4.56% -4.00%
Case Load 1 afib patient -0.25 -0.26 -0.23 -9.8% -10.3% -9.2%
2 to 4 afib patients -0.22 -0.23 -0.20 -8.6% -9.1% -8.1%
5 to 9 afib patients -0.19 -0.21 -0.18 -7.7% -8.2% -7.3%
10 to 18 afib patients -0.13 -0.15 -0.12 -5.3% -5.8% -4.9%
19 to 36 afib patients -0.10 -0.11 -0.09 -4.0% -4.4% -3.5%
37 to 72 afib patients -0.08 -0.09 -0.07 -3.2% -3.7% -2.6%
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Age <35years old -0.14 -0.20 -0.09 -5.7% -8.0% -3.4%

35 to 44 years old -0.10 -0.13 -0.06 -3.8% -5.1% -2.5%
45 to 54 years old -0.02 -0.04 0.00 -0.9% -1.6% -0.2%
55 to 64 years old 0.06 0.05 0.07 2.4% 1.9%
1.9% 22.9%
.9
9%
65 to 74 years old 0.06 0.05 0.07 2.4% 2.
.0%
2.0% 22.8%
.8%
.8 %
75 to 84 years old 0.04 0.03 0.05 1.6% 1.3
3%
3%
1.3% 22.0%
.0
0%
Gender Female -0.03 -0.04 -0.02 -1.3% -1.5% -1.0%
an Income
Median Inc
ncom
omee ($)) 0 to 29,999 -0.1
.144
-0.14 -0.16 -0.12 -5.6% -6.3% -4.9%
30,,000 to 39,999
30
30,000 -0
0.08
.0
08
-0.08 -0.10 -0
0.0
. 7
-0.07 --3.3%
-3 .3% -3.9% -2.8%
4 ,0
40 ,000
00 tto
40,000 o 49
49,9
,999
99
49,999 9 -0.055
-0 -0
0.06
06
-0.06 -0.0
-0 .04
04
-0.04 --2.0%
2.00% -2.6
-2 .6%
-2.6%% -1.5
-11.5
. %
-1.5%
50,0000
00 tto
50,000 o 79
79,9
,9
999
99
79,999 -0.044 -0
0.0
.055
-0.05 -0
0.022
-0.02 -11.55%
-1.5% -2.0
-2 .0%
0%
-2.0% -11.0
-1.0%0%
HHS Region
HHS R egio
Regionon 2 (NNJ, N
(NJ, Y)
NY) -0.06 6 -0 .08
.0
08
-0.08 -0
0.05
-0.05 -22.6%
-2.6% --3.0%
-3.0
3 0% -22.1%
-2.1% %
R
Regi
egioon 3 (DE,
Region (DE
E, DC
DC, MD,
M , PA,
MD PA, -0
0.0
.044
-0.04 -0
0.0
.066
-0.06 -0.0
-0 .03
03
-0.03 -1.8
-1 .8
8%
-1.8% -2.2
-2 .2%
-2.2% % -11.3
-1.3%3%
VA, WV)
VA, WV)
Reg
Re gi 4 (AL,
gion
Region (AL
AL,, FL
FL,, GA
GA,, KY
KY,, -0.0
0.099
-0.09 -0
-0.1
.100
-0.10 -0.0
0.088
-0.08 -3.6
3.6%
-3.6%% -4.1
4.1%
-4.1%% -3
3.0%
0%
-3.0%
MS, NC
MS, NC,, SCC, T
SC, N))
TN)
Region 5 (IL, IN, MI, MN, -0.07 -0.09 -0.06 -2.9% -3.5% -2.3%
OH, WI)
Region 6 (AR, LA, NM, OK, -0.19 -0.21 -0.17 -7.5% -8.2% -6.8%
TX)
Region 7 (IA, KS, MO, NE) -0.04 -0.06 -0.02 -1.5% -2.3% -0.8%
Region 8 (CO, MT, ND, SD, -0.09 -0.14 -0.05 -3.8% -5.7% -1.8%
UT, WY)
Region 9 (AZ, CA, HI, NV) -0.13 -0.14 -0.12 -5.3% -5.8% -4.8%
Region 10 (AK, ID, OR, WA) 0.08 0.01 0.14 3.0% 0.5% 5.6%
* The precision parameter (I for the beta regression model is 3.34 (95% CI 3.32 to 3.36). The Pseudo R2for the overall model
fit is 0.09353, and the log-likelihood ratio statistic is 12,407 (p<0.0001). Statistical significance of the coefficients is evaluated
using the partial Wald statistics. Two variables (Age Group 45 to 54 years old and Region 10) were significant at the p<0.05
level. Al other variables were significant at the p<0.001 level.
** The slope and approximate 95% confidence intervals are calculated in TTR units. They are derived from the coefficients of
the beta regression with a probit link function with the probability density function of the standard normal distribution evaluated
at the mean of the fitted values ( f(ȕx) = 0.3976).
*** The comparators for the categorical variables are Study Length •6 months; Case Load •72 afib patients; Age •85 years old;
Gender = Male; Median Income •$80,000; HHS Region = Region 1 (CT, MA, NH, RI, and VT).

15
 DOI: 10.1161/CIRCULATIONAHA.113.002601

Figure Legends:

Figure 1. Association of age and sex with time in therapeutic range(TTR) and frequency of

international normalization ratio (INR)measurements among 138,319US patients with an INR

diagnosis code indicating atrial fibrillation. Panels A and B show the mean percent TTR for

women (dark gray bars) and men (light gray bars) by age group; panel A represents the inception

period of warfarin therapy (patients with <6 months of INR data) and panel B represents the

established period (•6 months of INR data). Panels C and D show the frequency of INR
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

measurements for women (black diamonds) and men (white diamonds) by age group; panel C

represents established
epresents the inception period of warfarin therapy and panel D represents the est
sttab
bli
lish
shed
sh ed pperiod.
e io
er iod

Within each age category, women had a significantly lower TTR (p<0.001, two sample t-test)

and
an higher
nd hi
high err ffrequency
gher
gh reequuen
ency two
wo sample t-test)) tthan
c of INR testing (p<0.001, tw hann di
didd men.

Figure
Figu
Fi re 22.. Re
gure
gu Relationship
Rellation
lati onsh
on ip between
ship bet
etwe n physician
weeen phy
hysi an aaccount
siccian o nt ssize
cccou izze ((number
numb
nu mber
mb e ooff pa
er ppatients
nts rreferred
tien
tien effer
erre for
redd fo
re or INR
INR

testing/physician)
esting/physic
iccia n aand
ian)
n) mean
nd m eaan percent
p rc
pe e t time
rcen
en t me
ti m in
in therapeutic
th
her
e ap
apeu
eu
uti range
t c ra
rang
ngee (T
ng (TTR).A,
(TTR
TR
R). A, IInception
).A, ncep
nc ept on pperiod
pti e iod of
er

warfarin treatment (patients with <6 months of INR data). B, Established warfarin period

(patients with (•6 months of INR data).Bubble size is proportional to the number of physicians

referring the indicated number of patients; of the 37,939 ordering physicians, nearly 95% ordered

warfarin testing on fewer than 10 patients. Physician account size was significantly correlated

with TTR during the inception and established periods of warfarin therapy.

Figure 3. Association of annual income in patient’s area of residence (ZIP code) with time in

therapeutic range (TTR). Dark gray bars represent patients in the inception period of warfarin

16
 DOI: 10.1161/CIRCULATIONAHA.113.002601

therapy (<6 months of data) and light gray bars represent those in the established period (•6

months of data). Income level was significantly associated with TTR in both the inception and

established periods of therapy (P < 0.001, two sample t-test).

Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

17
 65 30
Study Time <6 Months

60

55

INR Tests per Year (Median)

25
TTR (Mean %)

50

45

20
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

40

35

30 15
18-34 35-44 45-54 55-64 65-74 75-84 85-
5 100
00
85-100
(N=300) (N=805) (N=2,897) (N=7,384) (N=13,016) (N=18,654) (N=9,0
(N=
(N 9,086)
9,086
86)
(N=9,086)
Age Range (Years) (Patient)

65 30
Study
St
tud T me •
dy Time
Ti •66 Mo
Mont
Months
nths
hs

60

55

INR Tests per Year (Median)

25
5
TTR (Mean %)

50

45

20
40

35

30 15
18-34 35-44 45-54 55-64 65-74 75-84 85-100
(N=160) (N=718) (N=3,135) (N=10,668) (N=22,667) (N=34,601) (N=14,228)
Age Range (Years) (Patients)

TTR (Mean %), Females TTR (Mean %), Males

INR Tests per Year (Median), Females INR Tests per Year (Median), Males

Figure 1
 65
Study Time <6 Months

60

55 16
TTR (Mean %)

50 40
172
524
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

45
1,605

5,710
40
8,894

35
5

30
0
0 1
1 2 2- 4 5 3- 9 10 4- 18 19 5- 36 37 6- 72 72+
7 8
Patients in a Ph
Physician's
P Account
ysician's Accoun
nt

5
65
tudy Time •6
Study
S 6MMonths
ontths

69
0
60
146
46
508
508
55 1,067

2,313
TTR (Mean %)

50
5,616
7,869

45

40

35

30
0 1
1 2 2- 4 5 3- 9 10 4- 18 19 5- 36 37 6- 72 72+
7 8
Patients in a Physician's Account

Number of Physician Accounts

Figure 2
 65

60

55
Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

TTR (Mean %)

50

45

0
40

5
35

30
0
29
$0 - 29 $3 - 39
$30 $4 - 49
$40 79
$ 50 - 79 80+
$ 80+
(N=10,2888)
(N=10,288) (N
N=2
= 7,98
982)
(N=27,982)2 (N
N=32,75
757)
7)
(N=32,757) (N=
=53
3,511 4)
(N=53,514) (N
N=100,4
414)
(N=10,414)
Med
Me di n Income
dian
Median Inc
come ($000s
($
$00
000
0s U SD
D)
USD)

<6
< Months
Mo ths
Mon •6
•6 Months
Mont
Mont
on hs

Figure 3
 A National Assessment of Warfarin Anticoagulation Therapy for Stroke Prevention in Atrial
Fibrillation
Jeffrey S. Dlott, Roberta A. George, Xiaohua Huang, Mouneer Odeh, Harvey W. Kaufman, Jack
Ansell and Elaine M. Hylek

Circulation. published online February 3, 2014;

Downloaded from http://circ.ahajournals.org/ by guest on November 5, 2017

Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2014 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/early/2014/01/30/CIRCULATIONAHA.113.002601

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click Request
Permissions in the middle column of the Web page under Services. Further information about this process is
available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:

http://circ.ahajournals.org//subscriptions/