Real-world EVIdence on

Stroke prevention In
patients with aTrial
Fibrillation in the United
States
Craig I. Coleman, Matthias Antz, Edgar Simard, Thomas Evers, Kevin Bowrin,
Hendrik Bonnemeier, Riccardo Cappato
University of Connecticut School of Pharmacy, Storrs, CT, USA; Hospital Oldenburg, Department of Cardiology, Oldenburg,
Germany; Aetion, Inc., New York, NY, USA,; Bayer Pharma AG, Wuppertal, Germany; Bayer Pharma AG, Berlin, Germany; University
Medical Center of Schleswig-Holstein, Department of Electrophysiology and Rhythmology, Kiel, Germany; Humanitas Clinical and
Research Centre, Rozzano, Italy

Disclosures
I

have received grant funding and/or

consultancy fees from:
Bayer Pharma AG, Berlin, Germany
Janssen Scientific Affairs, LLC, Raritan, NJ, USA
Boehringer-Ingelheim Pharmaceuticals Inc.,

Ridgefield, CT, USA

Portola Pharmaceuticals, South San Francisco,
CA, USA
Pfizer, Groton, CT, USA

Background
Phase

III clinical trial data (ROCKET AF1

and ARISTOTLE2) demonstrated the
favorable efficacy and safety profiles of
rivaroxaban and apixaban vs. vitamin K
antagonists (VKAs)
Real-world evidence is needed to
confirm the magnitude of benefits and
risks in routine clinical practice

1. N Engl J Med 2011; 365:883-891; 2. N Engl J Med 2011; 365:981

Objective
To

assess the real-world

effectiveness and safety of newlyinitiated rivaroxaban or apixaban
compared to warfarin in NVAF
patients

Methods

Retrospective analysis performed using US MarketScan

claims data from 1/1/2012-10/31/2014

Inclusion:
Adult patients newly initiated on rivaroxaban, apixaban

or warfarin
Baseline CHA2DS2-VASC score2,
2 ICD-9 diagnosis codes for NVAF, and
180 days of continuous medical and prescription
coverage
Exclusion:

Methods
Utilized

validated FDA Mini-Sentinel1 coding for

ischemic stroke and ICH endpoints (primary ICD-9
code position only)
Patients were followed until the occurrence of a
primary event, discontinuation/switch of therapy,
disenrollment, or end of the study
Propensity-score matched each FXaI patient to a
warfarin patient (during the same time period) to
reduce differences between baseline
characteristics
1. http://www.mini-sentinel.org/work_products/Assessments/Mini-Sentinel_Protocol-for-

Considerations When Performing

NOAC/NVAF Claims Database Studies
Claims

databases have little, if any, data on vital

signs or laboratory results
Not possible to assess whether prescribing is consistent

with labeling in MarketScan

Time

lag in data availability

Data through 10/2014 = limited apixaban usage and

decreased power
NOACs

were approved for NVAF at different times

Experience with NOACs likely grows over time, changing

benefit/risk
All

claims databases subject to potential

Our Approach
Design

study to optimize internal validity

Show consistency between HRs in Phase III trials and real-

world analyses, not head-to-head NOAC comparisons

Selected endpoints most likely to be coded accurately (and
with less variability) in claims data and of equal importance
to allow for benefit/risk assessment (ischemic stroke and
ICH)
Used validated ICD-9 coding algorithms and restricted codes to the
primary diagnosis code position (may miss cases, but greater faith in
those identified)
Exclusion of patients with baseline events

Included rivaroxaban and apixaban patients starting on their

38,831 Patients Met Selection Criteria

(Rivaroxaban vs. Warfarin)
Newly initiated on
rivaroxaban or warfarin
meeting selection
criteria
N=38,831

Rivaroxaba
warfarin
n
n=26,083
n=12,748
Following propensityscore matching
stratified by
exposure status.
11,411 rivaroxaban and
11,411 warfarin users
were identified
*PY=Patient-years

Characteristics of included patients (matched

cohorts)
Rivaroxaba Warfarin
(n=11,411
n
)
Parameter
(n=11,411)
6,271
7,715 PYs*
PYs*
70.7
Age in years, mean (SD)
70.7 (10.99)
(11.35)
Male, %
53.6
53.9
CHADS2 score, 180 day;
1.92 (1.08) 1.94 (1.08)
mean (SD)
CHA2DS2-VASc score, 180
3.46 (1.37) 3.48 (1.35)
day;
mean (SD)

18,591 Patients Met Selection Criteria

(Apixaban vs. Warfarin)
Newly initiated on
apixaban or warfarin
meeting selection
criteria
N=18,591

Apixaban
n=4,332

Warfarin
n=14,259

Following propensityscore matching,

stratified by exposure
status
4,083 apixaban and
4,083 warfarin users
were identified
*PY=Patient-years

Characteristics of included patients (matched

cohorts)
Warfarin
Apixaban
(n=4,083)
Parameter
(n=4,083)
1,951
2,125 PYs*
PYs*
71.15
71.00
Age in years, mean (SD)
(11.32)
(11.25)
Male, %
53.2
53.6
CHADS2 score, 180 day;
1.93 (1.07) 1.92 (1.07)
mean (SD)
CHA2DS2-VASc score, 180
3.47 (1.38) 3.47 (1.35)
day; mean (SD)
HAS-BLED Score, 180 day;
1.65 (0.69) 1.66 (0.72)
mean (SD)

Rivaroxaban vs. Warfarin

Rivaroxaban was associated with a significant* 47%
reduction in ICH vs.warfarin
29% non-significant decrease in ischaemic stroke vs.
warfarin
Significant* 39% reduction in the combined endpoint of
ICH and
ischemic
stroke vs.
warfarin
Rivaroxaban
Warfarin
HR (95%
CI)
HR (95% CI)

Rate (%/year) Rate (%/year)

rivaroxaban vs.
warfarin

ICH

0.49

0.96

0.53 (0.350.79)

Ischemic
stroke

0.54

0.83

0.71 (0.471.07)

Combined
*p<0.05

0.95

1.6

0.61 (0.450.82)

rivaroxaban vs. warfarin

0.13

0.25

0.5

Favors
rivaroxaban

2
Favors
warfarin

Apixaban vs. Warfarin

Apixaban was associated with a significant* 62%
reduction in ICH vs. warfarin
13% non-significant increase in ischemic stroke vs.
warfarin
Non-significant 37% reduction in the combined endpoint
of ICH and
ischemic
strokeHRvs.
Apixaban
Warfarin
(95% warfarin
CI)
HR (95% CI)

Rate (%/year) Rate (%/year)

apixaban vs.
warfarin

ICH

0.38

0.97

0.38 (0.170.88)

Ischemic
stroke

0.56

0.51

1.13 (0.492.63)

Combined
*p<0.05

0.89

1.44

0.63 (0.351.12)

apixaban vs. warfarin

0.13

0.25

0.5

Favors
apixaban

2
Favors
warfarin

Conclusion
This study was designed to optimize internal validity (obtain
the most unbiased HR estimates), however caution is
warranted when comparing these results to Phase III trials
It provides reassurance that both oral agents have lower ICH
rates vs. warfarin in routine practice
Rivaroxaban significantly reduced the combined endpoint of
ischemic stroke and ICH

These results are generally consistent with that of ROCKET AF1 and

other real-world studies (XANTUS2, PMSS3)

Observed apixaban HRs were less consistent with those seen

in ARISTOTLE4 (particularly 13% increased hazard of ischemic
stroke, p=NS)

1. N Engl J Med 2011; 365:883-891; 2. Eur Heart J. 2015 Sep 1. pii: ehv466. [Epub ahead of print]; 3. Clin Cardiol. 2015;38:63-8; 4. N Engl J Med 20

Thank You!

Questions?

Back Up Slides

Over-Use of the Reduce Doses of FXaIs

in the US?1
PERCENTAGE OF APIXABAN AND RIVAROXABAN
PRESCRIPTIONS FILLED FOR A REDUCED DOSE
VS. THE PROPORTION OF PATIENTS REQUIRING
A REDUCED DOSE IN CORRESPONDING
REGISTRATION TRIALS

PERCENTAGE OF APIXABAN AND RIVAROXABAN

PRESCRIPTIONS WRITTEN FOR THE REDUCED DOSE IN
ROUTINE CARDIOLOGY PRACTICE

IMS LifeLink data for apixaban and rivaroxaban from 9/19/2014

1. Curr Med Res Opin. 2016 24:1-13. [Epub
9/11/2015

Apixaban and Rivaroxban

Prescription Data
Pivotal RCTs and labeling provide clear recommendation for the use of the reduced
dose each agent
In routine practice, a high proportion ofApixaban
FXaI prescriptions are written
for the
Rivaroxaban
reduced
NOACdose
Prescription Data
Q2 2015
Q2 2015
Country
2.5 MG
5 MG
10 MG
15 MG
20 MG
UNITED STATES
25%
75%
5%
20%
75%
JAPAN
60%
40%
55%
45%
0%
GERMANY
45%
55%
4%
34%
62%
CANADA
38%
62%
5%
26%
69%
AUSTRALIA
40%
60%
2%
30%
68%
UNITED KINGDOM
39%
61%
5%
22%
74%
SPAIN
38%
62%
5%
31%
64%
FRANCE
45%
55%
7%
36%
57%
BELGIUM
31%
69%
2%
42%
57%
ITALY
41%
59%
3%
37%
60%
IMS LifeLink data for apixaban and

Impact of Lower Intensity

Anticoagulation
High
Edox1
%/year

Low
Edox1
%/year

Warfarin
Warfarin
INR 2.6-3.02 INR 1.5-1.92
%/year
%/year

ICH

0.39

0.26

0.5

0.3

Ischemic
stroke

1.25

1.77

0.9

1.9

1. N Engl J Med 2013;369:2093-104; 2. N Engl J Med

2003;349:1019-26

od dosing is associated with higher adherence than bid dosing:

adherence to chronic cardiovascular disease medication 1

Adherence (%)

6.9 (95% CI 11.2 to 2.6)

p<0.01
14.0 (95% CI 19.9 to 8.1)
22.9 (95% CI 33.1 to 12.7)
p<0.01
100
p<0.01
93.1

90
80
70
60
50
40
30
20
10
0

86.2

73.8

76.3

Dosing
frequency
50.4

Taking adherence

Number of bottle
cap openings
divided by the
Less stringent prescribed number
of doses
1. Curr Med Res Opin.
2012;28:669-80.

84.9

Regimen adherence
Percentage of
dayswith the
appropriate
number of doses
taken

o
d

Timing adherence
Percentage of
nearoptimal
interadministration
intervals

More stringent

Are differences in NOAC adherence

clinically-relevant?
1.15

1.13
95%CI, 1.08
1.19

1.13
95%CI, 0.97
1.33

Non-adherence is
associated with worse
outcomes during NVAF
NOAC treatment

Hazard
ratio/
1.10
10% decrease in PDC
1.05

1.04
95%CI, 0.94
1.14

1.00

0.95NVAF (71.39.7 years; 98.3% were men and mean CHADS score was 2.41.2; mean PDC
5376 US veterans with
2
Mortality/stroke
Stroke
Non-fatal
bleeding
event
84%22%; 27.8% with a PDC <80%; median follow-up of 244 days) initiated on dabigatran
from October
2010 to
September 2012

Am Heart J. 2014;167:810-7.