Amarin ICER CVD Final Evidence Report 101719

Additive Therapies for Cardiovascular Disease:
Effectiveness and Value

Final Evidence Report
October 17, 2019
Prepared for
©Institute for Clinical and Economic Review, 2019

University of Colorado Skaggs School of Pharmacy and
ICER Staff and Consultants
Pharmaceutical Sciences Modeling Group*
Daniel A. Ollendorf, PhD R. Brett McQueen, PhD
Director of Value Measurement and Global Health Assistant Professor
Initiatives Skaggs School of Pharmacy and Pharmaceutical Sciences
Center for the Evaluation of Value and Risk in Health University of Colorado Anschutz Medical Campus
Tufts Medical Center
Jonathan D. Campbell, PhD
Serina Herron-Smith, BA Associate Professor, Department of Clinical Pharmacy,
Research Assistant Pharmaceutical Outcomes Research
Institute for Clinical and Economic Review Skaggs School of Pharmacy and Pharmaceutical Sciences
University of Colorado Anschutz Medical Campus
Katherine Fazioli, BS
Research Lead Taryn A. Quinlan, MS
Institute for Clinical and Economic Review Professional Research Assistant
Skaggs School of Pharmacy and Pharmaceutical Sciences
Patricia G. Synnott, MALD, MS University of Colorado Anschutz Medical Campus
Director, Evidence Review
Institute for Clinical and Economic Review
Remziye Zaim, PhD

(Former) Research Fellow
Rick Chapman, PhD, MS

Director of Health Economics
Ellie Adair, MPA

Program Manager
David Rind, MD, MSc

Chief Medical Officer
Institute for Clinical and Economic Review *The role of the University of Colorado Anschutz Medical
Campus modeling group is limited to the development of
Steven D. Pearson, MD, MSc the cost-effectiveness model, and the resulting ICER
President reports do not necessarily represent the views of the
Institute for Clinical and Economic Review University of Colorado Anschutz Medical Campus.
None of the above authors disclosed any conflicts of interest.
DATE OF PUBLICATION: October 17, 2019

How to cite this document: Ollendorf D, McQueen R, Fazioli K, Synnott P, Campbell J, Quinlan T, Zaim R,
Pearson S, Rind D. Additive Therapies for Cardiovascular Disease: Effectiveness and Value. Institute for Clinical and
Economic Review, October 17, 2019. http://icer-review.org/material/cvd-final-evidence-report/
©Institute for Clinical and Economic Review, 2019 Page ii

Additive CVD Therapies – Final Evidence Report
Daniel A. Ollendorf served as the lead author for the report. Patricia Synnott, Serina Herron-Smith,
and Katherine Fazioli led the systematic review and authorship of the comparative clinical
effectiveness section. R. Brett McQueen and Jonathan D. Campbell developed the cost-
effectiveness model, and along with Taryn Quinlan, authored the corresponding sections of the
report. Rick Chapman and Remziye Zaim were responsible for oversight of the cost-effectiveness
analyses and developed the budget impact model. Ellie Adair authored the section on coverage
policies and clinical guidelines, and managed the timeline and public process, and performed quality
controls. David Rind and Steve Pearson provided methodologic guidance on the clinical and
economic evaluations. We would also like to thank Laura Cianciolo for her contributions to this
report.
The Institute for Clinical and Economic Review (ICER) is an independent non-profit research
organization that evaluates medical evidence and convenes public deliberative bodies to help
stakeholders interpret and apply evidence to improve patient outcomes and control costs. Through
all its work, ICER seeks to help create a future in which collaborative efforts to move evidence into
action provide the foundation for a more effective, efficient, and just health care system. More
information about ICER is available at http://www.icer-review.org.
The funding for this report comes from government grants and non-profit foundations, with the
largest single funder being the Laura and John Arnold Foundation. No funding for this work comes
from health insurers, pharmacy benefit managers, or life science companies. ICER receives
approximately 19% of its overall revenue from these health industry organizations to run a separate
Policy Summit program, with funding approximately equally split between insurers/PBMs and life
science companies. Life science companies relevant to this review who participate in this program
include: Janssen Pharmaceuticals, Inc. For a complete list of funders and for more information on
ICER's support, please visit http://www.icer-review.org/about/support/.
About Midwest CEPAC
The Midwest Comparative Effectiveness Public Advisory Council (Midwest CEPAC) – a core program
of ICER – provides a public venue in which the evidence on the effectiveness and value of health
care services can be discussed with the input of all stakeholders. The Midwest CEPAC seeks to help
patients, clinicians, insurers, and policymakers interpret and use evidence to improve the quality
and value of health care.
The Midwest CEPAC Council is an independent committee of medical evidence experts from across
the Midwest, with a mix of practicing clinicians, methodologists, and leaders in patient engagement
and advocacy. All Council members meet strict conflict of interest guidelines and are convened to
discuss the evidence summarized in ICER reports and vote on the comparative clinical effectiveness
and value of medical interventions. More information about the Midwest CEPAC is available at
https://icer-review.org/programs/midwest-cepac/.
©Institute for Clinical and Economic Review, 2019 Page iii

Additive CVD Therapies – Final Evidence Report Return to Table of Contents
The findings contained within this report are current as of the date of publication. Readers should be aware that
new evidence may emerge following the publication of this report that could potentially influence the results.
ICER may revisit its analyses in a formal update to this report in the future.
The economic models used in ICER reports are intended to compare the clinical outcomes, expected costs, and
cost effectiveness of different care pathways for broad groups of patients. Model results therefore represent
average findings across patients and should not be presumed to represent the clinical or cost outcomes for any
specific patient. In addition, data inputs to ICER models often come from clinical trials; patients in these trials and
provider prescribing patterns may differ in real-world practice settings.
©Institute for Clinical and Economic Review, 2019 Page iv

In the development of this report, ICER’s researchers consulted with several clinical experts, patients,
manufacturers, and other stakeholders. The following clinical experts provided input that helped
guide the ICER team as we shaped our scope and report. None of these individuals is responsible for
the final contents of this report or should be assumed to support any part of this report, which is
solely the work of the ICER team and its affiliated researchers.
For a complete list of stakeholders from whom we requested input, please visit:
https://icer-review.org/material/cvd-stakeholder-list/.
Expert Reviewers
Robert A. Harrington, MD
Arthur L. Bloomfield Professor of Medicine
Stanford University
No relevant conflicts of interest to disclose, defined as more than $10,000 in health care company
stock or more than $5,000 in honoraria or consultancies during the previous year from health care
manufacturers or insurers.
Patrick T. O’Gara, MD
Watkins Family Distinguished Chair in Cardiology
Brigham and Women’s Hospital
Professor of Medicine
Harvard Medical School
Dr. Patrick T. O’Gara serves as a member of the Apollo (TMVR) Executive Committee at Medtronic
and as a member of the Early TAVR Executive Committee at Edwards Life Sciences.
©Institute for Clinical and Economic Review, 2019 Page v

Table of Contents
Executive Summary ............................................................................................................................ ES1
Background .................................................................................................................................... ES1
Interventions .................................................................................................................................. ES2
Insights Gained from Discussions with Patients and Patient Groups ............................................ ES2
Comparative Clinical Effectiveness ................................................................................................ ES3
Controversies and Uncertainties ................................................................................................... ES7
Summary and Comment ................................................................................................................ ES9
Long-Term Cost Effectiveness ...................................................................................................... ES11
Potential Other Benefits and Contextual Considerations............................................................ ES16
Value-Based Benchmark Prices.................................................................................................... ES17
Potential Budget Impact .............................................................................................................. ES18
Midwest CEPAC Votes .................................................................................................................. ES20
Key Policy Implications................................................................................................................. ES22
1. Introduction ....................................................................................................................................... 1
1.1 Background .................................................................................................................................. 1
1.2 Scope of the Assessment ............................................................................................................. 3
1.3 Definitions .................................................................................................................................... 6
1.4 Insights Gained from Discussions with Patients and Patient Groups .......................................... 6
1.5 Potential Cost-Saving Measures in the Management of CVD ..................................................... 7
2. Summary of Coverage Policies and Clinical Guidelines ..................................................................... 8
2.1 Coverage Policies ......................................................................................................................... 8
2.2 Clinical Guidelines ........................................................................................................................ 9
3. Comparative Clinical Effectiveness .................................................................................................. 11
3.1 Overview .................................................................................................................................... 11
3.2 Methods ..................................................................................................................................... 11
3.3 Results ........................................................................................................................................ 13
4. Long-Term Cost Effectiveness .......................................................................................................... 35
4.1 Overview .................................................................................................................................... 35
4.2 Methods ..................................................................................................................................... 35
©Institute for Clinical and Economic Review, 2019 Page vi

4.3 Results ........................................................................................................................................ 47
4.4 Summary and Comment ............................................................................................................ 57
5. Potential Other Benefits and Contextual Considerations ................................................................ 59
5.1 Potential Other Benefits ............................................................................................................ 60
5.2 Contextual Considerations ......................................................................................................... 61
6. Value-Based Price Benchmarks ........................................................................................................ 62
7. Potential Budget Impact .................................................................................................................. 63
7.1 Overview .................................................................................................................................... 63
7.2 Methods ..................................................................................................................................... 63
7.3 Results ........................................................................................................................................ 64
7.4 Access and Affordability Alert .................................................................................................... 66
8. Summary of the Votes and Considerations for Policy ..................................................................... 67
8.1 About the Midwest CEPAC Process ........................................................................................... 67
8.2 Voting Results ............................................................................................................................ 69
8.3 Roundtable Discussion and Key Policy Implications .................................................................. 73
References ........................................................................................................................................... 79
Appendix A. Search Strategies and Results.......................................................................................... 87
Appendix B. Previous Systematic Reviews and Technology Assessments .......................................... 91
Appendix C. Ongoing Studies ............................................................................................................... 93
Appendix D. Comparative Clinical Effectiveness Supplemental Information ...................................... 98
Appendix E. Comparative Value Supplemental Information ............................................................. 120
Appendix F. Public Comment ............................................................................................................. 125
Appendix G. Conflict of Interest Disclosures ..................................................................................... 129
©Institute for Clinical and Economic Review, 2019 Page vii

List of Acronyms Used in this Report
ACS Acute coronary syndrome
ACCF American College of Cardiology Foundation
ADA American Diabetes Association
AE Adverse event
AHA American Heart Association
AHRQ Agency for Healthcare Research and Quality
ALI Acute limb ischemia
ASA Aspirin or acetylsalicylic acid
BCBSKC Blue Cross Blue Shield of Kansas City
BI Budget impact
CAD Coronary artery disease
CE Cost effectiveness
CI Confidence interval
CMS Centers for Medicare and Medicaid Services
COPD Chronic obstructive pulmonary disease
CV Cardiovascular
CVD Cardiovascular disease
DAPT Dual antiplatelet therapy
eGFR Estimated glomerular filtration rate
EPA Eicosapentaenoic acid
EQ-5D European Quality of Life-5 Dimensions
ESC European Society of Cardiology
FDA United States Food and Drug Administration
GDP Gross domestic product
GI Gastrointestinal
HR Hazard ratio
hsCRP High-sensitivity C-reactive protein
ISTH International Society on Thrombosis and Haemostasis
LCD Local coverage determination
LDL Low-density lipoprotein
LY Life year
MI Myocardial infarction
mRS Modified Rankin Scale
NCD National coverage determination
NMA Network meta-analysis
PAD Peripheral artery disease
PCI Percutaneous coronary intervention
PICOTS Population, Intervention, Comparator, Outcome, Timing, Setting
PPI Proton pump inhibitor
PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses
QALY Quality-adjusted life year
RCT Randomized controlled trial
SAE Serious adverse event
TEAE Treatment-emergent adverse event
TG Triglyceride
TIMI Thrombolysis in myocardial infarction
US United States
USPSTF United States Preventive Services Task Force
WAC Wholesale acquisition cost
WTP Willingness to pay
©Institute for Clinical and Economic Review, 2019 Page viii

Executive Summary
Background
The term cardiovascular disease (CVD) defines a complex, burdensome, and highly prevalent set
of conditions. Three of the major types of CVD, coronary artery disease (CAD), peripheral artery
disease (PAD), and cerebrovascular disease, result most frequently from atherosclerosis, a
chronic degenerative process involving increasing buildup of plaque formed by fat- and
cholesterol-based deposits. Over time, these deposits result in arterial narrowing and wall
hardening, which in turn can result in angina, claudication, myocardial infarction (MI), stroke,
heart failure, and death, among other problems. In total, CVD is estimated to affect one-half of
adults in the United States (US), and is the leading cause of death across all races and
ethnicities, with approximately 850,000 deaths annually.2 CVD also imposes a substantial
financial burden, with annual direct and indirect costs estimated to total $351 billion; projected
annual costs are expected to exceed $1 trillion by 2035.2 Hypertension, hypercholesterolemia,
diabetes, and smoking are well-established risk factors for atherosclerotic CVD, and
overweight/obesity, poor diet, physical inactivity, and excessive alcohol use may also contribute
to its development.3
In addition to mortality risks and financial burden associated with CVD, major adverse
cardiovascular events can result in long-term disability and complicate care for other conditions.
For example, an analysis of linked data from the US Health and Retirement Study and Medicare
claims found significant increases in the number of functional limitations on activities of daily
living following hospitalization for MI or stroke; in addition, those hospitalized for stroke had a
fourfold increase in the odds of moderate-to-severe cognitive impairment, even after
controlling for pre-hospitalization cognition.4
The management of CVD has commonly consisted of behavioral and lifestyle changes (i.e., diet,
weight reduction, physical activity, smoking cessation) to interrupt atherosclerotic processes, as
well as risk factor management, including blood pressure control, treatment with lipid-lowering
agents such as statin medications and PCSK9 inhibitors, antiplatelet therapy, and when
necessary, management of diabetes as well as surgical or percutaneous revascularization.
Although low-dose acetylsalicylic acid (aspirin, or ASA) and statins have become cornerstone
therapies with proven benefit for patients with established CVD, this population remains at high
residual risk of cardiovascular events.5 In addition, those without documented CVD but with
established risk factors such as diabetes and comorbid hypertension or hypercholesterolemia
are also at elevated risk of major cardiovascular events. For these patients, there is clinical
interest in exploring other types of medical management in addition to the strategies.
©Institute for Clinical and Economic Review, 2019 Page ES1

Interventions
Rivaroxaban
Rivaroxaban (Xarelto®, Janssen Pharmaceuticals, Inc.) is an oral direct and selective inhibitor of
factor Xa in the blood coagulation pathway. Rivaroxaban was first approved for the prevention
of deep vein thrombosis in patients undergoing major orthopedic surgery and is commonly
used in the management of atrial fibrillation and venous thromboembolic disease. In October
2018, rivaroxaban 2.5 mg taken twice daily with aspirin received an indication for the
prevention of major cardiovascular events in patients with CAD or PAD.6
Icosapent Ethyl
Icosapent ethyl (Vascepa®, Amarin Pharma, Inc.) is a purified ethyl ester of the omega-3 fatty
acid known as eicosapentaenoic acid (EPA), which was initially approved in 2012 as an adjunct
to diet to treat severe hypertriglyceridemia (triglyceride levels ≥500 mg/dL). The manufacturer
has filed for an expanded indication for management of patients with established CVD or at high
risk of CV events. Initially, a decision from the Food and Drug Administration (FDA) was expected
by September 28, 2019,7 however an FDA advisory committee meeting has been scheduled for
November 14, 2019.8 Icosapent ethyl’s mechanism of action in cardioprotection is not fully
known.9 It is taken twice daily with food (total daily dose: 4 g).
Insights Gained from Discussions with Patients and Patient Groups

ICER engaged with patients, patient groups, and clinical experts to understand the specific
challenges associated with ongoing management of CVD from the patient perspective. There
was acknowledgment that the high rate of recurrent events, even in CVD patients whose risk
factors are optimally managed, continues to concern clinicians. Still, caution was urged in
considering further additions to the current armamentarium, given the need to balance the
potential for additional clinical benefit against the risk of major bleeding and other harms, as
well as the inconsistent track record of previous antithrombotic regimens and omega-3
preparations respectively in reducing the rate of recurrent cardiovascular events. Despite these
concerns, there was enthusiasm expressed around the potential for new treatments to further
reduce event risks in these high-risk populations.
We also heard that medication adherence might be a challenge in this population, given already
high rates of polypharmacy and comorbidity in older patients likely to be candidates for add-on
therapy. Indeed, patients expressed trepidation with an increased therapeutic burden, citing
concerns with both the daily complexity of treatment and increased financial responsibility for
ongoing treatment. Patients also mentioned that the value and risk of adding new treatments
to an already complex treatment regimen is not necessarily clearly and consistently
communicated. Indeed, prior research in this clinical area suggests that patients tend to
significantly overestimate both their event and bleeding risks relative to their quantified risk
scores.10 Other feedback included the need to tailor the physician-patient conversation to

reflect the patient’s specific situation—for example, a family history of CVD, management of
comorbid conditions, or the benefits of lifestyle and behavioral changes in addition to medical
management.
Comparative Clinical Effectiveness

To inform our analysis of the comparative clinical effectiveness of additive therapies for the
management of CVD, we sought evidence related to low-dose rivaroxaban + ASA compared to
ASA alone or in combination with another antiplatelet agent (i.e., dual antiplatelet therapy
[DAPT]). Separately, we also evaluated the clinical evidence for icosapent ethyl compared to
optimal medical management alone. Our review focused on clinical benefits (i.e., reduction in
cardiovascular events, mortality, and quality of life), as well as potential harms (i.e., bleeding
and other drug-related AEs). We did not attempt to indirectly compare rivaroxaban to
icosapent ethyl, as the two key Phase III trials that were the focus of our review differed in
terms of target population and definitions of key outcomes. Key findings are summarized by
drug in the sections that follow.
Rivaroxaban
Compared to treatment with ASA alone, rivaroxaban + ASA reduced the composite risk of
cardiovascular death, stroke, or MI in patients with stable CVD. Patients treated with
rivaroxaban + ASA experienced significantly fewer strokes (including disabling or fatal
strokes), less cardiovascular death, and fewer cardiovascular-related hospitalizations. No
significant effect of rivaroxaban on hemorrhagic stroke or MI was observed. Patients treated
with rivaroxaban + ASA experienced a significant increase in major bleeding events, which led
to permanent discontinuation of therapy in approximately 3% of patients. Most major
bleeding events occurred in the GI tract.
Our review of rivaroxaban was primarily informed by the Phase III COMPASS trial,11 which
randomized approximately 27,000 patients to receive rivaroxaban 2.5 mg twice daily and 100
mg once daily of ASA, 5 mg twice daily of rivaroxaban alone, or 100 mg once daily of ASA alone.
The FDA-approved indication is for rivaroxaban only in combination with ASA, however; as such,
evidence from the rivaroxaban alone arm is not summarized here. Based on a planned interim
analysis, the COMPASS trial was stopped early (after a mean of 23 months of follow-up) due to
evidence of significant clinical benefit.
Clinical Benefits
Major Adverse Cardiovascular Events
The primary outcome of the COMPASS trial was a composite endpoint consisting of the first
occurrence of cardiovascular death, stroke, or MI.11 In the time to event analysis, the hazard
ratio for the primary outcome was 0.76 (95% CI: 0.66, 0.86; p<0.001). Patients treated with
rivaroxaban + ASA had statistically significantly fewer primary outcome events (4.1%) compared

to patients in the ASA alone group (5.4%); for the ASA alone group, this translates into an annual
event rate of approximately 3%, suggesting a relatively high-risk population.
Individual Events
Individual components of the primary and secondary composite outcomes are presented in
Table 3.3 in Section 3. Patients treated with rivaroxaban + ASA experienced significantly fewer
strokes, and less cardiovascular death, death from coronary heart disease, and death from any
cause. In an exploratory analysis, rivaroxaban + ASA reduced the risk of disabling or fatal
strokes (i.e., strokes defined as a score between 3 and 6 on the modified Rankin Scale) by 42%
(HR 0.58; 95% CI: 0.37 to 0.89; p=0.01).12-14 Hemorrhagic strokes occurred in more patients in
the rivaroxaban + ASA group but differences did not reach statistical significance.
Other Benefits
Hospitalization for cardiovascular causes (Appendix Table D4) occurred less in patients
randomized to rivaroxaban + ASA versus ASA alone (14.2% vs. 15.3%; HR 0.92; 95% CI: 0.86 to
1.00; p=0.04). The non-cardiovascular-related hospitalization rate was not statistically different
between arms. Although the European Quality of Life-5 Dimensions (EQ-5D) scale was
measured in the COMPASS trial, no results have been published or presented as of the
publication of this Evidence Report.
Harms
Major bleeding events occurred in significantly more patients treated with rivaroxaban + ASA
compared to ASA alone (3.1% vs. 1.9%; HR 1.70; 95% CI: 1.40 to 2.05; p<0.001); 2.7% of patients
in the rivaroxaban + ASA group permanently discontinued treatment due to bleeding, compared
to 1.2% in the ASA alone group.6,11 The most common bleeding site was the GI tract (1.5% vs.
0.7%; HR 2.15; 95% CI: 1.60 to 2.89; p<0.0001). Selected bleeding outcomes are presented in
Table ES1 and all bleeding outcomes are reported in Appendix Table D5.
Table ES1. Selected Bleeding Outcomes in COMPASS

Rivaroxaban +
Outcome ASA Hazard Ratio (95% CI) P-Value
ASA
Major Bleeding 288 (3.1) 170 (1.9) 1.70 (1.40-2.05) <0.001
Fatal Bleeding 15 (0.2) 10 (0.1) 1.49 (0.67-3.33) 0.32
Non-Fatal Symptomatic 21 (0.2) 19 (0.2) 1.10 (0.59-2.04) 0.77
Intracranial Bleeding
Other Major Bleeding 210 (2.3) 112 (1.2) 1.88 (1.49-2.36) <0.001
Minor Bleeding 838 (9.2) 503 (5.5) 1.70 (1.52-1.90) <0.001
ASA: aspirin, CI: confidence interval
Serious adverse events (SAEs) occurred in 7.9% of patients in the rivaroxaban + ASA arm versus
7.3% of patients on ASA alone; discontinuation due to non-bleeding AEs was not reported.

Clinical Benefits and Safety of Dual Antiplatelet Therapy (DAPT)
Our literature search did not identify any studies directly comparing rivaroxaban + ASA to DAPT
in the population of focus. Although we did not systematically review DAPT versus ASA alone,
we searched for RCTs that evaluated new initiation of DAPT (as opposed to continuation of
current DAPT therapy) in patients with stable CVD. We identified two RCTs of ticagrelor + ASA
and clopidogrel + ASA, respectively.15-18 These trials are summarized in Appendix D for context.
We also indirectly compared DAPT to rivaroxaban + ASA through an NMA of major adverse
cardiovascular events in patients with a recent MI (see below).
NMA
We performed an NMA in the subgroup of patients with a recent MI (i.e., in the two years prior
to randomization for the studies of rivaroxaban and ticagrelor, and at a median of two years
prior to randomization for the study of clopidogrel) to compare ticagrelor + ASA and clopidogrel
+ ASA with rivaroxaban + ASA. The analysis estimated the comparative risk of a composite
endpoint of cardiovascular death, stroke, or MI between each of the regimens of focus. The
results of our NMA, presented in Table ES2, do not reveal statistical differences between
therapies. However, given the elevated risk of major bleeding that is associated with each of
the regimens, any analysis of comparative effectiveness is incomplete without an accompanying
analysis of comparative safety. We endeavored to also compare the incidence of major
bleeding between therapies but were unable to quantitatively synthesize the data due to the
use of important differences in definitions of major bleeding.
Table ES2. NMA Results Comparing the Risk of Cardiovascular Death, Stroke, or MI in Patients
Treated with Antithrombotic Therapy for Stable CVD
Rivaroxaban + ASA
0.91
Ticagrelor + ASA
(0.61 to 1.36)
0.91 1.00
Clopidogrel + ASA
(0.58 to 1.40) (0.75 to 1.32)
0.70 0.77 0.77
ASA
(0.48 to 1.02) (0.66 to 0.90) (0.61 to 0.98)
Each box represents the estimated hazard ratio and 95% credible interval for the combined direct and
indirect comparisons between two drugs. Estimates in bold signify that the 95% credible interval does
not contain one.

Icosapent Ethyl
Compared to optimal medical management alone (i.e., placebo), icosapent ethyl reduced the
risk of a composite outcome of cardiovascular death, stroke, MI, coronary revascularization, or
unstable angina in patients with established CVD or diabetes mellitus and additional risk
factors, as well as each individual component and the composite secondary outcome of
cardiovascular death, MI, or stroke. A treatment benefit was also observed in analyses of the
first, subsequent, and total major adverse cardiovascular events. Rates of serious adverse
events and events leading to treatment discontinuation were similar between arms. A greater
proportion of patients treated with icosapent ethyl experienced serious bleeding-related
disorders, as well as peripheral edema, constipation, and atrial fibrillation.
Evidence on icosapent ethyl was primarily derived from the Phase III REDUCE-IT trial,19 which
randomized patients at increased risk of ischemic events (either established CVD or primary
prevention in patients age ≥50 with diabetes and at least one additional risk factor) to 2 g twice
daily of icosapent ethyl (n=4089) or a placebo (n=4090) that contained mineral oil to resemble
the color and consistency of icosapent ethyl. Patients were required to have elevated fasting
triglyceride levels (≥135 and <500 mg/dL) and well-controlled LDL cholesterol levels (>40 and
≤100 mg/dL) while on a stable dose of statins for at least four weeks. At baseline, 71% of
patients had established CVD and 29% made up the high-risk primary prevention cohort.19
Patients were followed for a median of 4.9 years.
Clinical Benefits
As noted above, the REDUCE-IT trial evaluated a composite of cardiovascular death, MI, stroke,
coronary revascularization, or unstable angina as its primary endpoint. In the time-to-event
analysis, icosapent ethyl reduced the risk of a primary endpoint event by 25% (HR 0.75; 95% CI:
0.68 to 0.83; p<0.001).19 At a median follow-up of 4.9 years (maximum 6.2 years), 17.2% of
patients treated with icosapent ethyl and 22.0% treated with placebo had a first primary
endpoint event. The annual event rate in the placebo arm was approximately 4.4%, suggesting
a very high-risk population.
The REDUCE-IT trial’s key secondary endpoint (cardiovascular death, MI, or stroke) also occurred
in fewer patients treated with icosapent ethyl compared to those receiving placebo (11.2% vs.
14.8%, respectively; HR 0.74; 95% CI: 0.65-0.83; p<0.001).19
Individual and Total Events
Icosapent ethyl reduced the risk of cardiovascular death by 20%, nonfatal MI by 30%, nonfatal
stroke by 29%, coronary revascularization by 34%, and hospitalization for unstable angina by
32%; however, a 13% reduction in all-cause mortality was not statistically significant. The effect
of icosapent ethyl on total events (first and subsequent) was examined in a pre-specified
analysis using a negative binomial regression model.20 The risk of total primary endpoint events,

including cardiovascular death, MI, stroke, revascularization, and unstable angina, was reduced
by 30% with icosapent ethyl compared to placebo (rate ratio (RR): 0.70; 95% CI: 0.62, 0.78).
Treatment with icosapent ethyl resulted in a 28% risk reduction compared to placebo (RR: 0.72;
95% CI: 0.63-0.82) on the REDUCE-IT trial’s key secondary endpoint of cardiovascular death,
stroke, or MI.
Harms
The incidence of serious treatment-emergent adverse events (TEAEs) was similar in the
icosapent ethyl and placebo arms of the REDUCE-IT trial (30.6% vs. 30.7%, respectively).19
Serious TEAEs leading to death occurred in 2.3% of patients treated with icosapent ethyl and
2.5% of patients who received placebo. Serious bleeding-related disorders, identified using the
Medical Dictionary for Regulatory Activities (MedDRA), occurred in a greater proportion of
patients treated with icosapent ethyl, although the difference was not statistically significant
(2.7% vs. 2.1%, p=0.06). No fatal bleeding events occurred in either group and rates of
hemorrhagic stroke, central nervous system bleeding, and GI bleeding did not statistically differ.
TEAEs that occurred in proportionately more patients treated with icosapent ethyl included
peripheral edema (6.5% vs. 5.0%, p=0.002), constipation (5.4% vs. 3.6%, p<0.001), and atrial
fibrillation (5.3% vs. 3.9%, p=0.003).19 Hospitalization for atrial fibrillation or flutter was
significantly higher in the icosapent ethyl arm compared to placebo (3.1% vs. 2.1%; p=0.004).
Approximately 11% of patients randomized to placebo and 10% randomized to icosapent ethyl
discontinued the study early.9 The rate of TEAEs leading to discontinuation of the study drug
was similar for patients treated with icosapent ethyl and placebo (7.9% vs 8.2%, respectively) as
was the rate of drug discontinuation due to serious TEAEs (2.2% vs 2.3%, respectively).
Controversies and Uncertainties

Rivaroxaban
The generalizability of the COMPASS trial population is subject to a number of uncertainties. For
one, study entry criteria of stable CAD and PAD as well as documentation of atherosclerosis in at
least two vascular beds among patients age <65 years ensured a population at high risk of
recurrent cardiovascular events, but exclusion of patients at high bleeding risk and further
exclusion of 8% patients not tolerating or adherent to run-in ASA therapy likely resulted in a
sample at reduced bleeding risk relative to the potential candidate population for
rivaroxaban.21,22 In addition, we cannot exclude the possibility that the clinical benefits
observed in COMPASS are somewhat overstated due to the trial having been stopped early for
benefit after a mean of 23 months of follow-up. 23 In addition, the decision to separately
randomize patients to receive the proton-pump inhibitor (PPI) pantoprazole or placebo within
the rivaroxaban + ASA, rivaroxaban alone, and ASA alone groups is a puzzling one, given that
clinical guidelines recommend routine use of PPIs for gastroprotection in patients receiving
combination anticoagulation + ASA therapy24 but not for anticoagulants or ASA alone.

Finally, while the indications for combination treatment with rivaroxaban and DAPT with a P2Y 12
inhibitor do not completely overlap, there is a large subset of patients with a recent MI event
who could conceivably be candidates for either treatment approach. Indeed, some clinicians
have called for further research comparing DAPT to combination therapy with ASA and a factor
Xa inhibitor (e.g., rivaroxaban).25,26 In the absence of head-to-head trials at the time of this
report’s publication, we attempted to compare the regimens indirectly through a network meta-
analysis. However, while only small differences exist in the definitions of clinical events across
the major trials of these regimens, the same cannot be said for definitions of major bleeding,
which differed substantially across trials; in some cases, a common definition was used but
modified to enough of an extent that we could not attempt quantitative comparisons with any
confidence. While this is a source of frustration for producers of comparative effectiveness
research, the real harm done is to the patient-clinician shared decision.
Icosapent Ethyl
As noted previously, the placebo vehicle used in the REDUCE-IT trial (as well as earlier trials of
icosapent ethyl) contained mineral oil to mimic the viscosity of the active agent. Biomarker
changes observed in the trial raise the possibility that the mineral oil used was not biologically
inert, however; patients in the placebo arm experienced a threefold-higher percentage increase
in LDL-C at year one (10.2% vs. 3.1% for icosapent ethyl, p<0.001 for between-group difference)
and a substantial increase in the inflammation marker hsCRP at year two (32.9% vs. -13.9%,
p<0.001), adding to documented concerns regarding the mineral oil’s potential interference
with statin absorption.9 The manufacturer conducted a post-hoc analysis, the results of which
suggested a consistent risk reduction with icosapent ethyl irrespective of whether LDL-C
increased in the placebo arm.9 However, it is difficult to interpret analyses of effects utilizing
on-treatment response subgroups, and so residual concerns remain about a potentially
biologically active “placebo” and the possibility that the true effect of icosapent ethyl may be
attenuated from that observed in the REDUCE-IT trial.
We also note that the results of this trial stand apart from many prior studies of omega-3
preparations that showed little to no cardiovascular benefit.27 Indeed, when a Bayesian
approach is taken to the overall evidence base, the interpretation of REDUCE-IT’s findings will
differ depending on whether prior expectations for these results are pessimistic, realistic, or
optimistic.28 It is worth noting that reductions in cardiovascular events of approximately 20%
were observed in a prior EPA-only trial (JELIS), which randomized approximately 19,000
Japanese patients to 1.8 g of EPA in addition to statin therapy versus statins alone over a mean
of 4.6 years of follow-up.29 However, the JELIS trial was open-label and showed no reductions in
cardiovascular death, so its relevance to the results of REDUCE-IT is unclear. The JELIS trial was
also conducted in a Japanese population with a much higher baseline consumption of fish than
is typically seen in the US; very few patients in REDUCE-IT were from the Asia-Pacific region
making comparisons across trials difficult.30,31
Regardless of issues of trial design or interpretation, the greatest uncertainty may be in how
generalizable the REDUCE-IT results are and therefore what the most appropriate target

population will be. As with COMPASS, the patients enrolled in REDUCE-IT were at very high risk
of cardiovascular events, as illustrated by a placebo event rate of approximately 4.4% per year
over the 4.9-year median duration of follow-up.9 Patients also were on statin therapy, and it is
unclear whether icosapent ethyl would be effective in patients not treated with statins. How
the benefits of icosapent ethyl translate to an eligible population that is certain to be both
broader and at lower risk than the trial population remains to be seen.
Summary and Comment

Rivaroxaban + ASA versus ASA Alone
Compared to ASA alone, rivaroxaban + ASA significantly reduced the risk of cardiovascular
death, stroke, or MI in patients with stable CVD. Patients treated with rivaroxaban + ASA
experienced significantly fewer strokes (including disabling or fatal strokes), less cardiovascular
death, less all-cause mortality, fewer major adverse limb events, and fewer cardiovascular-
related hospitalizations. Bleeding events of greatest severity—i.e., fatal bleeding, symptomatic
bleeding into a critical organ, and nonfatal symptomatic intracranial hemorrhages—were not
significantly increased by adding rivaroxaban to ASA. We therefore have high certainty that
rivaroxaban + ASA provides a small-to-substantial net health benefit in patients with CAD, PAD,
or both conditions (“B+”).
Rivaroxaban + ASA versus DAPT
We did not identify any head-to-head studies that directly compared rivaroxaban + ASA to DAPT
in patients with stable CVD. Although an indirect comparison of the risk of major adverse
cardiovascular events in patients with a recent MI suggested that DAPT may provide a similar
cardioprotective benefit to rivaroxaban + ASA, clinically significant differences in the way major
bleeding was defined in the clinical trials of focus precluded a companion analysis of relative
bleeding risks. We also note that those with a recent MI represented a relatively small subset of
patients in the COMPASS trial, so the comparative benefits and risks of these two strategies in
the remaining CAD and PAD population are unknown. We therefore have low certainty of
whether rivaroxaban + ASA provides a negative, comparable, or positive net health benefit
compared to DAPT in patients with CAD or PAD (“I”).
Icosapent Ethyl versus Optimal Medical Management
Icosapent ethyl reduced the risk of major adverse cardiovascular events in patients with
established CVD or diabetes mellitus and additional risk factors compared to optimal medical
management alone (i.e., placebo). The therapy was generally well-tolerated, despite a slight
increase in the incidence of major bleeding disorders. However, over 4.9 years of follow-up, no
fatal bleeding events occurred, and rates of TEAEs were comparable between the icosapent
ethyl and placebo arms. Although we are uncertain whether the use of mineral oil may have
caused some harm to the placebo group, we do not believe that this theory can account for the
entire benefit observed in the REDUCE-IT trial. We believe that the results of REDUCE-IT likely

apply across a range of baseline triglyceride levels but are uncertain whether the results
generalize to patients not treated with statins. For adults with established CVD or at high risk of
cardiovascular events who are being treated with statins, we have high certainty that icosapent
ethyl provides a small-to-substantial net health benefit (“B+”).

Long-Term Cost Effectiveness
Overview and Methods
The primary aim of this analysis was to estimate the cost effectiveness of rivaroxaban and
icosapent ethyl as additive therapies to optimal medical management in patients with
established CVD, and in the case of icosapent ethyl, also in patients without evidence of CVD but
with diabetes and at least one additional risk factor. A Markov cohort model was constructed to
compare the addition of rivaroxaban to ASA therapy to ASA alone and to compare the addition
of icosapent ethyl to optimal medical management (including statins) to optimal medical
management (including statins) alone. Rivaroxaban and icosapent ethyl were modeled
separately but shared a similar overall model structure. Patient survival, quality-adjusted
survival, and health care costs from a health care sector perspective were estimated over a
lifetime time horizon for each intervention and comparator. Costs and outcomes were
discounted at 3% per year. While the base-case analysis took a health care sector perspective,
productivity losses to the patient and caregiver were considered in a scenario analysis using a
modified societal perspective.
Individuals in the CVD cohort began on treatment and could stay in that state, pass into event
states of MI or stroke, or could die (Figure ES1). Patients who experienced a cardiovascular
event moved into post-event health states, where they may have had higher likelihood for
death as compared to the general CVD prevention population. Patients remained in the model
until they died. All patients could transition to death from all-causes from any of the alive
health states. Death could have occurred from all-cause or cardiovascular event/post-event
related morality.
Figure ES1. Model Framework

Post-event states:
post-MI or post-
stroke
Established CVD or high CV event states:

CVD risk MI or stroke*
All-cause and CV-

specific death
CV: cardiovascular, CVD: cardiovascular disease, MI: myocardial infarction

Other treatment-specific modeled events include major adverse limb events and other SAEs.
*Other CV events such as revascularization and unstable angina included in scenario analysis.

For both drugs, we obtained net pricing estimates from SSR Health, LLC,32 which combine data
on unit sales with publicly-disclosed US sales figures that are net of discounts, rebates,
concessions to wholesalers and distributors, and patient assistance programs to derive a net
price (Table ES3).
Table ES3. Drug Cost Inputs

WAC Net Price
Drug per per Discount from WAC Net Price per Year
Tablet/Capsule Tablet/Capsule
Rivaroxaban $7.47 per 2.5
$3.03 59.41% $2,215
(Xarelto®, Janssen) mg tablet
Icosapent Ethyl
$2.53 per 1 g
(Vascepa®, Amarin $1.11 56.04% $1,625
capsule
Pharma)
WAC: wholesale acquisition cost
WAC per Redbook®; net pricing estimates from SSR Health.32,33
Model Validation
We used several approaches to validate the model. First, we provided preliminary methods and
results to manufacturers, patient groups, and clinical experts. Based on feedback from these
groups, we refined data inputs used in the model. Second, we varied model input parameters
to evaluate face validity of changes in results. We performed model verification for model
calculations using internal reviewers. As part of ICER’s efforts to increase modeling
transparency, we also shared the model with relevant manufacturers for external verification
and feedback shortly after publishing the draft report for this review. Finally, we compared
results to other cost-effectiveness models in this therapy area.
Base-Case Results
Base-case discounted costs and outcomes from the model are listed in Table ES4 for rivaroxaban
and in Table ES5 for icosapent ethyl. Rivaroxaban was associated with approximately $17,000 in
discounted lifetime intervention costs, whereas icosapent ethyl was associated with $15,000 in
lifetime intervention costs. Average discounted life years, equal value life years gained (evLYG)
(a measure that evenly values any gains in length of life, regardless of the impact on patients’
quality of life), and QALYs were higher for both interventions as compared to optimal medical
management alone.

Table ES4. Base-Case Discounted Results for Rivaroxaban Compared to Optimal Medical
Management including ASA
Non-
Intervention
Treatment Intervention Total Costs Life Years evLYG QALYs
Costs
Costs
Rivaroxaban $17,000 $20,000 $38,000 10.86 9.07 9.06
Medical
$200 $24,000 $24,000 10.45 8.69 8.69
Management
evLYG: equal value of life years gained, QALY: quality-adjusted life year
Table ES5. Base-Case Discounted Results for Icosapent Ethyl Compared to Optimal Medical
Management including Statins
Non-
Intervention
Costs
Costs
Icosapent
$15,000 $25,000 $40,000 12.26 10.21 10.19
Ethyl
Medical
$800 $30,000 $31,000 11.73 9.69 9.69
Management
Base-case discounted incremental results are shown in Table ES6, with rivaroxaban versus
optimal medical management yielding $36,000 per QALY gained. Icosapent ethyl versus optimal
medical management yields $18,000 per QALY gained. Discounted incremental life year results
were slightly lower than the incremental cost-per-QALY findings. Results for the incremental
evLYG were slightly more favorable than the cost-per-QALY findings given there is a life
extension to each therapy over medical management. The incremental cost per major adverse
cardiovascular event avoided should be interpreted with caution, given that this metric does not
have a known threshold for an understanding of value and does not include the differential
timing or the differential importance of major adverse cardiovascular events. Note that the
intervention-specific incremental findings are modeled using intervention-specific populations
and therefore should not be directly compared across treatments.

Table ES6. Base-Case Incremental Results
Cost per
Incr. Incr. Incr. Incr. Cost Cost per Cost per
Intervention* MACE
Costs LYs evLYG QALYs per LY evLYG QALY
Avoided
Rivaroxaban $32,000 $35,000 $36,000 $120,000

vs. Medical $13,000 0.41 0.38 0.37 per LY per evLYG per QALY per MACE
Management gained gained gained avoided
Icosapent
$17,000 $17,000 $18,000 $53,000
Ethyl vs.
$9,000 0.54 0.52 0.50 per LY per evLYG per QALY per MACE
Medical
gained gained gained avoided
Management
ICER: incremental cost-effectiveness ratio, Incr.: Incremental LY: life year, MACE: major cardiovascular event,
QALY: quality adjusted life year
*Modeled populations differed across interventions; results for the interventions are not directly comparable.
Sensitivity Analyses
To demonstrate effects of uncertainty on both costs and health outcomes, we varied input
parameters using available measures of parameter uncertainty (i.e., standard errors) or
reasonable ranges to evaluate changes in cost per additional QALY in deterministic sensitivity
analyses. Key drivers of uncertainty for both comparisons (rivaroxaban versus optimal medical
management and icosapent ethyl versus optimal medical management) included the clinical
event hazard ratios for MI, stroke, and cardiovascular death, with smaller impacts observed
from uncertainty in utility and cost inputs. In probabilistic sensitivity analyses, in which all
important parameters were varied simultaneously, 92% of iterations suggested that rivaroxaban
met the $50,000/QALY threshold. Icosapent ethyl results suggested that nearly 100% of
iterations met the $50,000/QALY threshold. Both interventions achieved 100% of iterations
meeting the $100,000/QALY and $150,000/QALY thresholds.
Threshold Analyses
We estimated threshold treatment prices that would reflect an incremental cost-per-QALY of

$50,000, $100,000, and $150,000. The findings suggest that both treatments’ net prices were
below the price needed to achieve $50,000 per QALY (Table ES7).

Table ES7. Threshold Analysis Results
Annual Annual Annual
Net Price
Price to Price to Price to
WAC per Annual per Net Price
Achieve Achieve Achieve
Tablet/Capsule WAC Tablet/ per Year
$50,000 $100,000 $150,000
Capsule
per QALY per QALY per QALY
$7.47 per 2.5
Rivaroxaban $5,457 $3.03 $2,215 $2,849 $5,223 $7,597
mg tablet
Icosapent $2.53 per 1 g
$3,699 $1.11 $1,625 $3,433 $6,282 $9,204
Ethyl capsule
QALY: quality-adjusted life year, WAC: wholesale acquisition cost
Table ES8 presents the threshold results for each drug at thresholds of $50,000, $100,000, and
$150,000 per equal value life year gained (evLYG). An analysis of the evLYG is included to
complement the cost per QALY calculations and provide policymakers with a broader view of
cost effectiveness.
Table ES8. Cost per evLYG Threshold Analysis Results

Net Price Price to Price to Price to
WAC per Annual per Net Price Achieve Achieve Achieve
Tablet/Capsule WAC Tablet/ per Year $50,000 $100,000 $150,000
Capsule per per per
evLYG evLYG evLYG
$7.47 per 2.5
Rivaroxaban $5,457 $3.03 $2,215 $2,922 $5,369 $7,780
mg tablet
$3,699 $1.11 $1,625 $3,506 $6,501 $9,423
Ethyl capsule
evLYG: equal value life year gained, WAC: wholesale acquisition cost
Summary and Comment
Our base-case results suggest that the use of rivaroxaban (plus ASA) and icosapent ethyl (in
patients receiving statins) both provide clinical benefit in terms of gains in quality-adjusted
survival and overall survival compared to optimal medical management alone in the adult,
established CVD cohort, and in the case of icosapent ethyl also for adults without known CVD
but at high risk for cardiovascular events. This translated into incremental cost-effectiveness
estimates that fell below commonly cited cost-effectiveness thresholds under the assumptions
used in this analysis. The results were relatively robust to parameter uncertainties in the one-
way and probabilistic sensitivity analyses. Further, the results were robust to a number of
scenario analyses including the modified societal perspective and others.
Our analyses had important limitations and assumptions. We assumed three-component major
adverse cardiovascular events, MI, stroke, and cardiovascular death, to form the base-case
health states within the model structure for both rivaroxaban and for icosapent ethyl. A
scenario analysis that broadened major adverse cardiovascular events to include other events
suggested similar but lower cost-effectiveness findings for icosapent ethyl. An additional
limitation of this analysis was the model calibration to the observed clinical trial event rates for
optimal medical management. Many unknowns were a part of the model calibration exercise.
Finally, it is important to note that randomized controlled trial findings may not generalize or
translate to observed signals within the real world (i.e., efficacy does not equal effectiveness).
Given that the cost-effectiveness findings relied on randomized controlled trials for estimates of
clinical benefit and harm, the findings should be interpreted with caution when estimating
whether these interventions would achieve similar value for money in actual practice.
In conclusion, the findings of our analysis suggest that the additive CVD therapies of focus for
this review provide gains in quality-adjusted survival and overall survival over optimal medical
management. Assuming clinical signals within the trial hold for patients treated with these
interventions and current net prices, the base-case results suggest that costs for treatment with
either rivaroxaban or icosapent ethyl would fall below commonly cited thresholds for cost
effectiveness. The results were relatively robust to sensitivity and scenario analyses.
Potential Other Benefits and Contextual Considerations

Our reviews seek to provide information on potential other benefits offered by the intervention
to the individual patient, caregivers, the delivery system, other patients, or the public that
would not have been considered as part of the evidence on comparative clinical effectiveness.
These elements are listed in the table below.
Potential Other Benefits
Table ES9. Potential Other Benefits

Other Benefits Description
This intervention offers reduced complexity that will Most candidate patients are older and already
significantly improve patient outcomes. taking multiple classes of medication, so the
potential is for increased complexity.
This intervention will reduce important health N/A
disparities across racial, ethnic, gender, socio-
economic, or regional categories.
This intervention will significantly reduce caregiver or N/A
broader family burden.
This intervention offers a novel mechanism of action or Icosapent ethyl may complement other
approach that will allow successful treatment of many commonly-prescribed therapies for CVD that
patients for whom other available treatments have have different mechanism of action.
failed.
This intervention will have a significant impact on N/A
improving return to work and/or overall productivity.
Other important benefits or disadvantages that should N/A
have an important role in judgments of the value of this
intervention.

Contextual Considerations
Table ES10. Potential Contextual Considerations

Contextual Consideration Description
This intervention is intended for the care of individuals Both drugs were studied in high-risk populations
with a condition of particularly high severity in terms of suggestive of significant unmet need.
impact on length of life and/or quality of life.
This intervention is intended for the care of individuals CVD is both prevalent and associated with a high
with a condition that represents a particularly high lifetime burden of illness.
lifetime burden of illness.
This intervention is the first to offer any improvement N/A
for patients with this condition.
Compared to “the comparator,” there is significant The early termination of the COMPASS trial
uncertainty about the long-term risk of serious side introduces significant uncertainty regarding the
effects of this intervention. long-term safety of rivaroxaban.
Compared to “the comparator,” there is significant The early termination of the COMPASS trial
uncertainty about the magnitude or durability of the introduces some uncertainty regarding the long-
long-term benefits of this intervention. term benefits of rivaroxaban. Uncertainty
around icosapent ethyl revolves around the
previous track record of other omega-3
preparations.
There are additional contextual considerations that N/A
should have an important role in judgments of the
value of this intervention.
Value-Based Benchmark Prices

Annual value-based price benchmarks (VBPBs) of rivaroxaban and icosapent ethyl are presented
in Table ES11. The value-based benchmark price for a drug is defined as the price range that
would achieve incremental cost-effectiveness ratios between $100,000 and $150,000 per QALY
(or evLYG) gained.
For rivaroxaban, price changes of approximately 4% discount to 39% over the list price (WAC)
would be required to reach the $100,000 to $150,000 per QALY threshold prices, respectively.
For icosapent ethyl, prices approximately 70% to 149% above WAC would achieve $100,000 to
$150,000 per QALY threshold prices. The cost per evLYG price range is quite similar to the cost
per QALY range for both rivaroxaban and icosapent ethyl.

Table ES11. Value-Based Price Benchmarks for Rivaroxaban and Icosapent Ethyl
Change from WAC
Annual Price at Annual Price at
Annual WAC to Reach Threshold
$100,000 Threshold $150,000 Threshold
Prices
Rivaroxaban
Per QALY Gained $5,457 $5,223 $7,597 -4% to +39%
Per evLYG $5,369 $7,780 -2% to +43%
Icosapent Ethyl
Per QALY Gained $3,699 $6,282 $9,204 +70% to +149%
Per evLYG $6,501 $9,423 +76% to +155%
WAC: wholesale acquisition cost; evLYG: equal value life year gained; QALY: quality-adjusted life year
Potential Budget Impact

We used results from the model to inform our budget impact estimates. For rivaroxaban, the
candidate population was estimated to be approximately 6.7% of the US population age ≥20, or
approximately 16.9 million individuals in total (~3.4 million per year over five years). For
icosapent ethyl, the candidate population included both the CAD estimate used for rivaroxaban
as well as estimates of individuals with prior stroke and those age ≥50 with diabetes and at least
one additional risk factor. The resulting estimate was approximately 33.5 million individuals, or
6.7 million per year over five years.
As shown in Figures ES2 and ES3, despite both therapies meeting common cost-effectiveness
thresholds, only a small portion of the eligible populations could be treated with crossing the
ICER budget impact threshold of $819 million per year because so many patients are potentially
eligible. When using net prices, only approximately 6% and 4% of eligible patients could be
treated in a given year with rivaroxaban and icosapent ethyl respectively without crossing the
ICER budget impact threshold.

Figure ES2. Potential Budget Impact of Rivaroxaban at Various Prices
$8,000
$150,000 per QALY
$7,000
$6,000
WAC price
$5,000 $100,000 per QALY
Annual Price
$4,000
$3,000 $50,000 per QALY
Net price
$2,000
$1,000
$0
0% 20% 40% 60% 80% 100%
Percentage of patients treated without crossing BI Threshold each year
Figure ES3. Potential Budget Impact of Icosapent Ethyl at Various Prices
$10,000
$9,000 $150,000 per QALY
$8,000
$7,000
$6,000 $100,000 per QALY

Annual Price
$5,000
$4,000 WAC price

$50,000 per QALY
$3,000
$2,000
Net price
$1,000
$0
0% 20% 40% 60% 80% 100%

Access and Affordability Alert
ICER is issuing an access and affordability alert for both rivaroxaban and icosapent ethyl. For
rivaroxaban, at the net price of $2,215 per year, approximately 6% of eligible patients could be
treated in a given year without crossing the ICER potential budget impact threshold of $819
million, while clinical experts at the Public Meeting stated that they would consider using
rivaroxaban in approximately 30% of eligible patients. For icosapent ethyl, at the net price of
$1,625 per year, approximately 4% of eligible patients could be treated in a given year without
crossing the ICER potential budget impact threshold of $819 million. Clinical experts at the
Public Meeting stated that they believe that the majority of eligible patients would want to be
on icosapent ethyl. The purpose of an ICER affordability and access alert is to signal to
stakeholders and policy makers that the amount of added health care costs associated with a
new service may be difficult for the health care system to absorb over the short term without
displacing other needed services or contributing to rapid growth in health care insurance costs.
Thus, there is a risk to sustainable access to high-value care for all patients if managing these
issues is not appropriately planned for.
Midwest CEPAC Votes

The Midwest CEPAC Panel deliberated on key questions raised by ICER’s report at a public
meeting on September 26, 2019. The results of these votes are presented below, and additional
information on the deliberation surrounding the votes can be found in the full report.
1. Is the evidence adequate to demonstrate that the net health benefit of rivaroxaban
plus ASA is superior to that provided by ASA alone?
Yes: 10 votes No: 1 vote
2. Is the evidence adequate to demonstrate that the net health benefit of rivaroxaban
plus ASA is superior to that provided by ASA as part of dual antiplatelet therapy
(DAPT) with an oral P2Y12 inhibitor (e.g., ticagrelor or clopidogrel)?
Yes: 0 votes No: 11 votes
3. Is the evidence adequate to demonstrate that the net health benefit of icosapent ethyl
added to optimal medical management (including statin therapy) is superior to that
provided by optimal medical management (including statin therapy) alone?
4. Does treating patients with rivaroxaban plus ASA offer one or more of the following
potential “other benefits or disadvantages” compared to ASA alone? (Select all that
apply).

This intervention will significantly reduce caregiver or broader family burden. 1/11
This intervention offers a novel mechanism of action or approach that will allow 6/11
successful treatment of many patients for whom other available treatments have
failed.
This intervention will have a significant impact on improving patients’ ability to 1/11
return to work and/or their overall productivity.
There are other important benefits or disadvantages that should have an important No vote
role in judgements of the value of this intervention.
5. Does treating patients with icosapent ethyl offer one or more of the following
potential “other benefits or disadvantages” compared to optimal medical
management (including statin therapy) alone? (Select all that apply).
failed.
6. Are any of the following contextual considerations important in assessing the long-
term value for money of rivaroxaban plus ASA? (Select all that apply).
This intervention is intended for the care of individuals with a condition of 6/11
particularly high severity in terms of impact on length of life and/or quality of life.
This intervention is intended for the care of individuals with a condition that 6/11
represents a particularly high lifetime burden of illness.
There is significant uncertainty about the long-term risk of serious side effects of this 9/11
intervention.
There is significant uncertainty about the magnitude or durability of the long-term 9/11
benefits of this intervention.
There are additional contextual considerations that should have an important role in N/A
judgements of the value of this intervention.
7. Are any of the following contextual considerations important in assessing the long-
term value for money of icosapent ethyl? (Select all that apply).
intervention.

There are additional contextual considerations that should have an important role in No vote
Long-Term Value for Money
As described in ICER’s value assessment framework, questions on long-term value for money are
subject to a value vote when incremental cost-effectiveness ratios for the interventions of
interest are between $50,000 and $175,000 per QALY in the primary “base-case” analysis. The
base-case estimates of the cost per QALY for both rivaroxaban + ASA and icosapent ethyl are
below the lower end of this range, and therefore the treatment is deemed “high long-term
value for money” without a vote unless the CEPAC determines in its discussion that the Evidence
Report base-case analysis does not adequately reflect the most probable incremental cost-
effectiveness ratio for either treatment.
ICER’s value assessment framework also describes that the CEPAC does not take Long-Term
Value for Money votes on treatments for which the evidence is considered inadequate during
the Clinical Vote. Such is the case for rivaroxaban + ASA versus DAPT therapy. Thus, no Value
vote was taken for rivaroxaban + ASA versus DAPT therapy (e.g. clopidogrel).
Key Policy Implications

Following its deliberation on the evidence, the Midwest CEPAC Panel engaged in a moderated
discussion with a policy roundtable about how best to apply the evidence on rivaroxaban and
icosapent ethyl to policy and practice associated with the treatment of CVD. The policy
roundtable members included two patients, two clinical experts, two payers, and one
representative from a pharmaceutical manufacturer. The discussion reflected multiple
perspectives and opinions, and therefore, none of the statements below should be taken as a
consensus view held by all participants. The top-line policy implications are presented below,
and additional information can be found in Section 8 of the full report.
Payers
1. Evidence to compare rivaroxaban to dual-action platelet therapy (DAPT) plus aspirin is

limited by lack of head-to-head trials and differing outcome measures. Additionally,
clinical experts do not view these two treatment options as interchangeable given their
different mechanisms of action and risk profiles, therefore DAPT should not be
considered an appropriate candidate in a step therapy protocol as a first step prior to
receiving coverage for rivaroxaban.
2. Icosapent ethyl has not yet received FDA approval and therefore the specific language of
the label is unknown. While awaiting the FDA decision, payers should consider
parameters of coverage criteria related to the eligibility criteria of the pivotal trial: a)

definition of risk for coronary artery disease; b) concurrent use of statins; and c)
triglyceride level of 135-499 mg/dL.
Providers
3. Clinicians, when thinking about the apparent benefit of rivaroxaban in the clinical trial,
should remember that patients at high risk of bleeding were excluded.
4. Develop options to help patients navigate complex medication regimens.
Clinical and Specialty Societies
5. Develop a decision algorithm and/or tool for clinicians to use in determining the most
appropriate additive therapies to consider for a given patient.
6. Ensure that any clinical guideline statements regarding rivaroxaban clearly warn against
assuming a class effect for direct oral anticoagulants.
Manufacturers
7. Conduct additional studies of icosapent ethyl in patients not on statin therapy.

8. Ensure that future trial recruitment reflects the demographics of the CVD population.
Regulators
9. The FDA, manufacturers, and the clinical research community should work to solidify a
common, single, outcomes definitions for key outcomes -- such as major bleeding -- so
clinicians and patients have the information they need to make informed decisions.
Researchers
10. Researchers should develop explicit head-to-head evidence of the comparative benefits
and risks of rivaroxaban + aspirin versus dual antiplatelet therapy in patients who have
completed an initial course of DAPT (12-30 months).
11. Researchers should conduct a real-world observational study to confirm the benefits of
icosapent ethyl.

1. Introduction
1.1 Background
The term cardiovascular disease (CVD) defines a complex, burdensome, and highly prevalent set
of conditions. Three of the major types of CVD, coronary artery disease (CAD), peripheral artery
disease (PAD), and cerebrovascular disease, result most frequently from atherosclerosis, a
chronic degenerative process involving increasing buildup of plaque formed by fat- and
cholesterol-based deposits. Over time, these deposits result in arterial narrowing and wall
hardening, which in turn can result in angina, claudication, myocardial infarction (MI), stroke,
heart failure, and death, among other problems. In total, CVD is estimated to affect one-half of
adults in the United States (US), and is the leading cause of death across all races and
ethnicities, with approximately 850,000 deaths annually.2 CVD also imposes a substantial
financial burden, with annual direct and indirect costs estimated to total $351 billion; projected
annual costs are expected to exceed $1 trillion by 2035.2 Hypertension, hypercholesterolemia,
diabetes, and smoking are well-established risk factors for atherosclerotic CVD, and
overweight/obesity, poor diet, physical inactivity, and excessive alcohol use may also contribute
to its development.3
In addition to mortality risks and financial burden associated with CVD, major adverse
cardiovascular events can result in long-term disability and complicate care for other conditions.
For example, an analysis of linked data from the US Health and Retirement Study and Medicare
claims found significant increases in the number of functional limitations on activities of daily
living following hospitalization for MI or stroke; in addition, those hospitalized for stroke had a
fourfold increase in the odds of moderate-to-severe cognitive impairment, even after
controlling for pre-hospitalization cognition.4
The management of CVD has commonly consisted of behavioral and lifestyle changes (i.e., diet,
weight reduction, physical activity, smoking cessation) to interrupt atherosclerotic processes, as
well as risk factor management, including blood pressure control, treatment with lipid-lowering
agents such as statin medications and PCSK9 inhibitors, antiplatelet therapy, and when
necessary, management of diabetes as well as surgical or percutaneous revascularization.
Although low-dose acetylsalicylic acid (aspirin, or ASA) and statins have become cornerstone
therapies with proven benefit for patients with established CVD, this population remains at high
residual risk of cardiovascular events.5 In addition, those without documented CVD but with
established risk factors such as diabetes and comorbid hypertension or hypercholesterolemia
are also at elevated risk of major cardiovascular events. For these patients, there is clinical
interest in exploring other types of medical management in addition to the strategies.
©Institute for Clinical and Economic Review, 2019 Page 1

Interventions
Rivaroxaban
Rivaroxaban (Xarelto®, Janssen Pharmaceuticals, Inc.) is an oral direct and selective inhibitor of
factor Xa in the blood coagulation pathway. This process also results in inhibition of
prothrombinase,6 an enzyme essential not only to hemostasis but to complex biologic processes
such as angiogenesis, cell proliferation, and inflammation; it therefore plays a role in the
development and progression of atherosclerosis, cancer, and other chronic inflammatory
diseases.34
Rivaroxaban, first approved for the prevention of deep vein thrombosis in patients undergoing
major orthopedic surgery, is commonly used in the management of atrial fibrillation and venous
thromboembolic disease. It received an indication for the prevention of major cardiovascular
events in patients with chronic CAD or PAD in October 2018, and is the latest in a line of
antithrombotic regimens that have been tested as alternatives or additions to ASA, including
vitamin K antagonists, antiplatelet therapies, and thrombin receptor antagonists.21 The
recommended dosage for prevention of cardiovascular events is 2.5 mg twice daily with or
without food, in combination with ASA (75-100 mg) once daily.6
Icosapent Ethyl
Icosapent ethyl (Vascepa®, Amarin Pharma, Inc.) is a purified ethyl ester of the omega-3 fatty
acid known as eicosapentaenoic acid (EPA), which was initially approved in 2012 as an adjunct
to diet to treat severe hypertriglyceridemia (triglyceride levels ≥500 mg/dL). Following the
completion of a Phase III trial in patients with established CVD or at high risk of cardiovascular
events,9 the manufacturer filed for an expanded indication in March of 2019, with an expected
FDA decision date of September 28, 2019.7 On August 8, 2019, the company announced that
they had received notification from the FDA that an advisory committee meeting had been
scheduled for November 14, 2019, with extension of the deadline for an FDA decision to late
December.35 At the time of publication of this report, no details are currently available
regarding the reasons for an advisory committee or specific questions or concerns posed by the
FDA.
Icosapent ethyl’s mechanism of action in cardioprotection is not fully known; while

hypertriglyceridemia increases the risk of ischemic events, reduction in these levels with
icosapent ethyl may only partially explain the treatment effects observed in the trial. Other
mechanisms, such as antithrombotic effects and stabilization and regression of coronary plaque
have also been hypothesized.9 The current recommended dosage of 2 g twice daily with food
for hypertriglyceridemia was also the dose tested in the cardiovascular prevention trial.

1.2 Scope of the Assessment
This project assesses both the comparative clinical effectiveness and economic impacts of
rivaroxaban + ASA and icosapent ethyl for the management of CVD. Evidence was collected
from available randomized controlled trials and non-randomized clinical trials. We did not
restrict studies according to number of patients or study setting. We supplemented our review
of published studies with data from conference proceedings, regulatory documents, information
submitted by manufacturers, and other grey literature when the evidence met ICER standards
(for more information, see https://icer-review.org/methodology/icersmethods/icer-value-
assessment-framework/grey-literature-policy/).
Populations
The population of focus for the review is adults with established CVD who are currently treated
with optimal medical management. For the assessment of icosapent ethyl, we also reviewed
evidence for patients without known CVD but at high risk for cardiovascular events.
Where data were available, we examined evidence for key subgroups suggested by clinical
experts, including the following:
1. Diagnosis of diabetes mellitus
2. Diagnosis of CAD alone versus CAD and concomitant PAD (rivaroxaban only)
3. Levels of high-sensitivity C-reactive protein (hsCRP) at baseline (i.e., ≤2 mg/l or >2 mg/l)
as well as changes in hsCRP from baseline to follow-up
4. Subgroups defined by level of cardiovascular risk at baseline
5. Renal dysfunction
6. Diagnosis of heart failure
Interventions
The list of interventions was developed with input from patient organizations, clinicians,
manufacturers, and payers on which drugs to include. The full list of interventions is as follows:
1. Rivaroxaban + ASA
• Patients are assumed to also be receiving optimal medical management
2. Icosapent ethyl
• Patients are assumed to also be receiving optimal medical management
including statins
Comparators
Comparators were defined to reflect the input of clinicians and other stakeholders on treatment
strategies that would be considered relevant alternatives for the overall population of interest
or a prominent subset, as well as the comparators as defined in major clinical studies of
icosapent ethyl and rivaroxaban.

1. Rivaroxaban comparators:
• optimal medical management including ASA without an additional antiplatelet
agent
• optimal medical management including ASA as part of dual antiplatelet therapy
(DAPT) with an oral P2Y12 inhibitor (e.g., clopidogrel)
2. Icosapent ethyl comparator:
• Optimal medical management including statin therapy
Outcomes
The outcomes of interest are described in Table 1.1 below.
Table 1.1. Outcomes and Harms

Outcomes Key Harms
All-cause mortality TEAEs
Cardiovascular mortality Discontinuation due to TEAEs
MI Serious TEAEs
Stroke Major bleeding events
Coronary revascularization
Unstable angina
Heart failure
Venous thromboembolism
Health-related quality of life
Cardiovascular hospitalization
Major adverse limb events
MI: myocardial infarction, TEAE: treatment-emergent adverse event
Timing
Evidence on intervention effectiveness were derived from studies of at least one year’s duration
and evidence on harms from studies of at least three month’s duration.
Settings
All relevant settings were considered, with a focus on outpatient management in the US.

Analytic Framework
The analytic framework for this assessment is depicted in Figure 1.1.
Figure 1.1. Analytic Framework: Rivaroxaban and Icosapent Ethyl for CVD
ASA: aspirin, CV: cardiovascular, CVD: cardiovascular disease, DAPT: dual antiplatelet therapy, hsCRP: high-
sensitivity C-reactive protein, LDL-C: low-density lipoprotein cholesterol, OMM: optimal medical management,
TEAE: treatment-emergent adverse event, TG: triglyceride
*For the assessment of icosapent ethyl, we will also review evidence for patients without known CVD but at
high risk for CV events.
The diagram begins with the population of interest on the left. Actions, such as treatment, are
depicted with solid arrows which link the population to outcomes. For example, a treatment may
be associated with specific clinical or health outcomes. Outcomes are listed in the shaded boxes:
those within the rounded boxes are intermediate outcomes (e.g., change in blood pressure), and
those within the squared-off boxes are key measures of clinical benefit (e.g., health-related
quality of life). The key measures of clinical benefit are linked to intermediate outcomes via a
dashed line, as the relationship between these two types of outcomes may not always be
validated. Curved arrows lead to the AEs of an action (typically treatment), which are listed
within the blue ellipsis.1

1.3 Definitions
International Society on Thrombosis and Haemostasis (ISTH) major bleeding: Fatal bleeding,
symptomatic bleeding into a critical area or organ, bleeding that causes a decrease in
hemoglobin ≥2 g/dL, or bleeding that requires a transfusion of ≥2 units of whole blood or red
cells.21,36
Modified ISTH major bleeding (used in COMPASS trial of rivaroxaban): Fatal bleeding,
symptomatic bleeding into a critical organ, bleeding into a surgical site requiring
reoperation, and bleeding that led to a visit to an acute care facility with or without an
overnight stay.21
Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary
Arteries (GUSTO) severe bleeding: Fatal bleeding events, intracranial hemorrhages, or bleeding
that causes hemodynamic compromise requiring blood or fluid replacement, inotropic support,
or surgical intervention.17
Thrombolysis in Myocardial Infarction (TIMI) major bleeding: Intracranial bleeding, clinically

overt signs of hemorrhage (drop in hemoglobin ≥5 g/dL or fall in hematocrit ≥15%), or a
bleeding event that led to death within seven days.15
Rankin Scale: A 6-point scale used to measure disability in individuals who have suffered a
stroke or other cause of neurologic disability. Scores range from 0, indicating no symptoms or
disability, to 6, indicating death; a score of 3 represents moderate disability.12-14
Ischemic stroke: Occurs when a blood vessel supplying blood to the brain becomes
obstructed.37
Hemorrhagic stroke: Occurs when a blood vessel in the brain leaks or ruptures.37
1.4 Insights Gained from Discussions with Patients and Patient

Groups
ICER engaged with patients, patient groups, and clinical experts to understand the specific
challenges associated with ongoing management of CVD from the patient perspective. There
was acknowledgment that the high rate of recurrent events, even in CVD patients whose risk
factors are optimally managed, continues to concern clinicians. Still, caution was urged in
considering further additions to the current armamentarium, given the need to balance the
potential for additional clinical benefit against the risk of major bleeding and other harms, as
well as the inconsistent track record of previous antithrombotic regimens and omega-3
preparations respectively in reducing the rate of recurrent cardiovascular events. Despite these
concerns, there was enthusiasm expressed around the potential for new treatments to further
reduce event risks in these high-risk populations.

We also heard that medication adherence might be a challenge in this population, given already
high rates of polypharmacy and comorbidity in older patients likely to be candidates for add-on
therapy. Indeed, patients expressed trepidation with an increased therapeutic burden, citing
concerns with both the daily complexity of treatment and increased financial responsibility for
ongoing treatment. Patients also mentioned that the value and risk of adding new treatments
to an already complex treatment regimen is not necessarily clearly and consistently
communicated. Indeed, prior research in this clinical area suggests that patients tend to
significantly overestimate both their event and bleeding risks relative to their quantified risk
scores.10 Other feedback included the need to tailor the physician-patient conversation to
reflect the patient’s specific situation—for example, a family history of CVD, management of
comorbid conditions, or the benefits of lifestyle and behavioral changes in addition to medical
management.
1.5 Potential Cost-Saving Measures in the Management of CVD

As described in its Final Value Assessment Framework for 2017-2019, ICER now includes in its
reports information on wasteful or lower-value services in the same clinical area that could be
reduced or eliminated to create headroom in health care budgets for higher-value innovative
services (for more information, see https://icer-review.org/final-vaf-2017-2019/). These
services are ones that would not be directly affected by rivaroxaban + ASA or icosapent ethyl
(e.g., reduction cardiovascular events), as these services will be captured in the economic
model. Rather, we are seeking services used in the current management of CVD beyond the
potential offsets that arise from a new intervention. During stakeholder engagement and public
comment periods, ICER encourages all stakeholders to suggest services (including treatments
and mechanisms of care) currently used for patients with CVD that could be reduced,
eliminated, or made more efficient. ICER has not received any such suggestions at the time of
posting this Evidence Report.

2. Summary of Coverage Policies and Clinical
Guidelines
2.1 Coverage Policies
To understand the insurance landscape for the interventions and comparators in this report, we
reviewed National and Local Coverage Determinations (NCDs and LCDs) from the Centers for
Medicare and Medicaid Services (CMS), and publicly available coverage policies from
representative public plans (MO HealthNet and Illinois Medicaid) and national and regional
private payers (Aetna, Cigna, and Blue Cross Blue Shield of Kansas City). We also looked at
Aetna’s standard Medicare plan to see if Medicare coverage differed from its standard national
plan. We surveyed the coverage policies for the currently approved indications of icosapent
ethyl and rivaroxaban, understanding these coverage policies may change for icosapent ethyl
should this intervention receive Food and Drug Administration (FDA) approval for its new
indication being reviewed for this report. We also reviewed coverage policies for two agents
used in DAPT, clopidogrel and ticagrelor, which are included in this report as comparators to
rivaroxaban.
We were unable to identify any NCDs or LCDs relating to the use of any of these therapies.38 A
summary of our other findings is as follows:
Rivaroxaban
Rivaroxaban is listed as “preferred” on the preferred drug list of both surveyed Medicaid
plans.38,39 Minnesota based payer, HealthPartners, lists rivaroxaban as a tier three drug on its
standard private plan and BCBSKC lists it as a tier-two drug—neither require prior
authorization.39,40 Aetna and Cigna’s standard national plans both list rivaroxaban as a preferred
brand on their respective drug lists.41,42 It is also a preferred brand on Aetna’s standard
Medicare plan.43
Clopidogrel and Ticagrelor
Clopidogrel, a generic anti-platelet agent, and ticagrelor, a brand anti-platelet agent, are the
preferred generic and preferred brand drugs respectively on both surveyed Medicaid plans.38
As the generic option, clopidogrel is consistently a tier one or preferred generic drug, while
ticagrelor is a tier two or three drug, or the preferred brand drug, on all surveyed private plans,
both regional and national, and on Aetna’s standard Medicare plan. None of these plans listed
any prior authorization criteria for either drug, although members are alerted that they may pay
more for the brand drug should they forego the generic.39-43

Icosapent Ethyl
Icosapent ethyl is listed as “non-preferred” on the preferred drug list of both surveyed Medicaid
plans.38 HealthPartners lists icosapent ethyl as a “non-formulary” drug on its standard private
plan, explaining that depending on an individual’s plan this medication is either not covered, or
covered at a higher out-of-pocket cost. If it is covered, it requires a prior authorization stating:
“reserved for patients with an inadequate response to two or more preferred products, such as
generic Lovaza, gemfibrozil, and fenofibrate.”39
BCBSKC lists icosapent ethyl as a tier-two drug but requires no prior authorization.40 Cigna’s
standard national plans lists it as a tier-three, non-preferred brand drug.42 Aetna’s standard
national plan lists icosapent ethyl on its preferred brand drug list, with no additional
information41, but it’s standard Medicare plan lists it as a non-preferred brand.43
2.2 Clinical Guidelines

Treatment guidelines for secondary prevention are many and large. This report focuses on only
those guidelines pertaining to the use of anticoagulation, DAPT, and fish oil in patients with
established CVD.
American Heart Association (AHA) and the American College of Cardiology Foundation (ACCF)
The AHA/ACCF guidelines for secondary prevention in patients with coronary or other
atherosclerotic vascular disease state that all patients with CAD should take daily ASA unless
contraindicated. DAPT (which refers to ASA plus a PY212 inhibitor such as clopidogrel and
ticagrelor) is recommended for patients after acute coronary syndrome (ACS) and for patients
with symptomatic PAD. The guidelines also state that DAPT may be considered in patients with
stable CAD.44
In 2016, AHA/ACCF published a focused update on guidelines for the use and duration DAPT.
The guidelines state that for patients with ACS, regardless of whether or not they have been
treated with revascularization or fibrinolytic therapy, DAPT therapy should continue for no
fewer than 12 months and if is well tolerated without bleeding complications, it may be
reasonable to continue DAPT for longer than 12 months. For patients with non-ST elevation ACS
treated with medical therapy alone, and in patients with ACS treated with DAPT after coronary
stent implantation, the guidelines state it is reasonable to prefer ticagrelor to clopidogrel for
P2Y12 maintenance therapy.45
The AHA/ACCF guidelines for secondary prevention in patients with CAD note that “it may be
reasonable” to recommend omega-3 fatty acids from fish oil or capsules to patients whose non-
HDL cholesterol levels remain elevated despite appropriate statin therapy. They also note that
omega-3 fatty acids from fish oil or capsules may be used to reduce the risk of CVD in all
patients.44

American Heart Association (AHA)
In August of 2019, the American Heart Association released a “Science Advisory” stating that
four grams per day of prescription omega-3 fatty acids (EPA+DHA or EPA-only) is effective
therapy for reducing triglycerides, either as monotherapy or in conjunction with lipid-lowering
agents. The advisory noted that, in contrast to other cited perspectives, the EPA+DHA
formulation only raises low-density (LDL) lipoprotein in the setting of very high triglycerides
(≥500 mg/dL). The advisory also counseled against patients self-treating with fish oil
supplements not approved by the FDA.46
The American Diabetes Association (ADA)
The ADA’s Standards of Medical Care in Diabetes guidelines recommend that all patients with
diabetes and a history of atherosclerotic CVD take ASA as part of their secondary prevention
strategy. They also state that DAPT is reasonable for a year after ACS.
The guidelines were also recently updated to include the recommendation that icosapent ethyl
be considered in patients with diabetes and atherosclerotic CVD, or other cardiac risk factors,
who have controlled low-density lipoprotein (LDL) cholesterol on a statin, but whose
triglycerides remain elevated.47
The European Society of Cardiology (ESC)
The ESC’s 2013 Guidelines on the Management of Stable Coronary Artery Disease recommend
low dose ASA for all patients with established CAD. These guidelines also note that N-3
polyunsaturated fatty acids, consumed mainly through fish oil, could have potential benefit on
cardiac risk factors, but trial results have shown mixed results for reducing cardiovascular
events.48
In 2017, the ESC released a focused update on guidelines for DAPT stating that for patients with
stable CAD, there is no indication for DAPT (unless overridden by a concomitant or prior
indication). For patients with ACS who have been treated with percutaneous coronary
intervention (PCI) or who are managed with medical therapy alone, DAPT is recommended to
continue for 12 months. If the patient is at high risk for bleeding, DAPT is recommended for six
months.49
In 2019, the ESC released an updated guideline that incorporated results from the REDUCE-IT
trial. The update recommends that treatment with omega-3 fatty acids, including icosapent
ethyl 2 g twice daily, in combination with statins should be considered for high-risk patients with
triglycerides between 135 and 499 mg/dL.50

3. Comparative Clinical Effectiveness
3.1 Overview
To inform our analysis of the comparative clinical effectiveness of additive therapies for the
management of CVD, we sought evidence related to low-dose rivaroxaban + ASA compared to
ASA alone or in combination with another antiplatelet agent (i.e., DAPT). Separately, we also
evaluated the clinical evidence for icosapent ethyl compared to optimal medical management
alone. Our review focused on clinical benefits (i.e., reduction in cardiovascular events,
mortality, and quality of life), as well as potential harms (i.e., bleeding and other drug-related
AEs). We did not attempt to indirectly compare rivaroxaban to icosapent ethyl given differences
in target population and definitions of key outcomes. Methods and findings of our review of the
clinical evidence are described in the sections that follow.
3.2 Methods
Data Sources and Searches
Procedures for the systematic literature review assessing the evidence on additive therapies for
CVD followed established best research methods.51,52 We conducted the review in accordance
with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)
guidelines.53 The PRISMA guidelines include a checklist of 27 items, which are described further
in Appendix Table A1.
We searched MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, and Cochrane

Central Register of Controlled Trials for relevant studies. Each search was limited to English-
language studies of human subjects and excluded articles indexed as guidelines, letters,
editorials, narrative reviews, case reports, or news items. We included abstracts from
conference proceedings identified from the systematic literature search. All search strategies
were generated utilizing the Population, Intervention, Comparator, and Study Design elements
described above. The proposed search strategies included a combination of indexing terms
(MeSH terms in MEDLINE and EMTREE terms in EMBASE), as well as free-text terms.
To supplement the database searches, we performed manual checks of the reference lists of
included trials and systematic reviews and invited key stakeholders to share references germane
to the scope of this project. We also supplemented our review of published studies with data
from conference proceedings, regulatory documents, information submitted by manufacturers,
and other grey literature when the evidence met ICER standards (for more information, see
http://icer-review.org/methodology/icers-methods/icer-value-assessment-framework/grey-
literature-policy/).

Study Selection
Subsequent to the literature search and removal of duplicate citations, references went through
two levels of screening at both the abstract and full-text levels. Two reviewers independently
screened the titles and abstracts of all publications identified using DistillerSR (Evidence
Partners, Ottawa, Canada); a third reviewer worked with the initial two reviewers to resolve any
issues of disagreement through consensus.
Citations accepted during abstract-level screening were reviewed as full text. The review
followed the same procedures as the title/abstract screening. Reasons for exclusion were
categorized according to the PICOTS elements.
Data Extraction and Quality Assessment
Two reviewers extracted key information from the full set of accepted studies (See Appendix D).
Elements included a description of patient populations, sample size, duration of follow-up, study
design features (e.g., double-blind), interventions (agent, dosage, dosing frequency, method of
administration), results, and quality assessment for each study. Extracted data were reviewed
for logic and were validated by a third investigator for additional quality assurance.
We used criteria employed by the US Preventive Services Task Force ([USPSTF], see Appendix D)
to assess the quality of clinical trials and comparative cohort studies, using the categories
“good,” “fair,” or “poor.”54
Assessment of Level of Certainty in Evidence
We used the ICER Evidence Rating Matrix to evaluate the level of certainty in the available
evidence of a net health benefit for rivaroxaban + ASA and icosapent ethyl relative to the
comparators of focus.55
Assessment of Bias
As part of our quality assessment, we evaluated the evidence base for the presence of potential
publication bias. Given the emerging nature of the evidence base for newer treatments, we
performed an assessment of publication bias for rivaroxaban + ASA and icosapent ethyl using
the ClinicalTrials.gov database of trials. We scanned the site to identify studies completed more
than two years ago that would have met our inclusion criteria and for which no findings have
been published. Any such studies may indicate whether there is bias in the published literature.
For this review, we did not find evidence of any study completed more than two years ago that
has not subsequently been published.

Data Synthesis and Statistical Analyses
Data on relevant outcomes were abstracted into evidence tables (see Appendix Tables D1-D15)
and are described in the text below. Data informing the comparison of rivaroxaban + ASA to
DAPT were also synthesized quantitatively in a network meta-analysis (NMA) with a focus on
prevention of cardiovascular events. The NMA included data from the subgroup of patients
with a MI within two years of randomization. An NMA extends pairwise meta-analyses by
simultaneously combining both the direct estimates (i.e., estimates obtained from head-to-head
comparisons) and indirect estimates (i.e., estimates obtained from common comparator[s]).
The NMA was conducted in a Bayesian framework with fixed effects on the treatment
parameter using the gemtc package in R.56 A fixed effects approach was taken given the small
size of the evidence network (i.e., single-study connections throughout). The log hazard ratios
for the composite outcome of cardiovascular death, stroke, or MI were analyzed using a normal
likelihood and identity link. Inputs used for the analysis are reported in Appendix Table D9.
Tabular results are presented for the treatment effects (hazard ratio) of each intervention
versus ASA along with 95% credible intervals (95% CrI) in Section 3.3. Note that we attempted
an NMA specification for major bleeding events, but differences in definitions of this outcome
across relevant clinical trials precluded such an analysis.
3.3 Results
Study Selection
Our literature search identified 808 potentially relevant references, of which 10 met the full
PICOTS criteria (Appendix A, Figure A1). The primary reasons for exclusion included study
population outside of our scope (e.g., acute coronary syndromes), dosing or combination
therapy outside of the FDA-labeled indication (e.g., >2.5 mg BID of rivaroxaban, rivaroxaban +
DAPT), and lack of outcomes of interest (e.g., studies that only reported on changes in
laboratory parameters).
Of the 10 included references, eight publications11,14,57-62 related to a single randomized

controlled trial (RCT) of rivaroxaban and two references corresponded to a single RCT of
icosapent ethyl.19,20
Although we did not systematically review the available literature on DAPT, we searched for
RCTs that evaluated the initiation of DAPT with ticagrelor or clopidogrel in combination with
ASA. We selected two DAPT initiation trials in patients with established CVD for inclusion in an
NMA of cardiovascular outcomes. Results of the NMA are presented in the sections that follow;
evidence from four references related to the two DAPT RCTS are additionally summarized for
context in Appendix D.

Quality of Selected Studies
We rated the two key studies of rivaroxaban and icosapent ethyl, respectively, to be of good
quality using criteria from the USPSTF (Appendix D). The trials had adequate blinding of
patients, investigators, and outcome assessors. The groups were comparable at baseline and
there was non-differential follow-up.
Key Studies of Rivaroxaban
Our review of rivaroxaban was primarily informed by the Phase III COMPASS trial.11 Patients
were eligible to participate in the trial if they had CAD, PAD, or both. Patients with CAD under
the age of 65 were also required to have documented atherosclerosis in at least two vascular
beds or to have at least two additional risk factors (e.g., diabetes mellitus, heart failure). Key
exclusion criteria included a high bleeding risk, recent stroke, severe heart failure, advanced
kidney disease, and the use of other antithrombotic therapies. Additional inclusion and
exclusion criteria can be found in Appendix Table D2.
Eligible patients (n=28,275) first entered a 30-day run-in period during which they received 100
mg of ASA once daily in combination with placebo twice daily; 8.2% (2,320) of patients were
excluded after the run-in phase, with 729 withdrawing consent and 1,645 citing adherence
concerns. Patients who recently underwent coronary artery bypass graft (CABG) (n=1,448),
were exempt from the run-in phase and randomized within four to 14 days of the procedure.
Following the run-in, patients who adhered to therapy and who did not have any AEs were
randomized 1:1:1 to combination therapy with rivaroxaban 2.5 mg twice daily and 100 mg once
daily of ASA (n=9,152), 100 mg once daily of ASA alone (n=9,126), or 5 mg twice daily of
rivaroxaban alone (n=9,117); in a second randomization, the COMPASS trial also compared
pantoprazole, a proton-pump inhibitor (PPI), to placebo to assess upper gastrointestinal (GI)
complications.61 Pantoprazole randomization occurred equally across the rivaroxaban + ASA,
rivaroxaban alone, and ASA alone treatment groups. As rivaroxaban was only approved in
combination with ASA for patients with CAD or PAD, evidence pertaining to the rivaroxaban
alone arm of COMPASS was not summarized in this review.
At baseline, 91% of patients had documented CAD and 27% had a history of PAD. 11
Approximately 62% of patients had a prior MI, 4% had a previous stroke, 38% had diabetes
mellitus, and 22% had heart failure. Patients were on a number of other background
medications, including angiotensin-converting enzymes and angiotensin-receptor blockers
(71%), beta-blockers (70%), and lipid-lowering agents (90%). Based on a planned interim
analysis, the COMPASS trial was stopped early (after a mean of 23 months of follow-up) due to
evidence of significant clinical benefit. Key characteristics of the COMPASS trial are summarized
in Table 3.1 below.

Table 3.1. Summary of the COMPASS Trial11
Treatment Primary
Patient Characteristics Key Safety Outcome
Groups Outcome
COMPASS 1. Rivaroxaban n=27,395 A composite Major bleeding:* fatal
Phase III (2.5 mg BID) + Age: 68.2±7.9 of CV death, bleeding, symptomatic
Double- ASA (100 mg Previous stroke: 343 stroke, or bleeding into a critical organ,
Blind QD) (3.8) MI bleeding into a surgical site
Previous MI: 5,687.5 requiring reoperation, and
Mean 2. ASA (100 (62.2) bleeding that led to
Follow-Up: mg QD) CAD: 8,287 (90.6) hospitalization or presentation
23 Months PAD: 2,498 (27.3) to an acute care facility
ASA: aspirin, CAD: coronary artery disease, CV: cardiovascular, MI: myocardial infarction, PAD: peripheral
artery disease
*COMPASS also evaluated 5 mg twice daily of rivaroxaban alone, however only the FDA approved dose is
represented here. Modified criteria from ISTH.
The COMPASS trial’s primary endpoint was a composite of cardiovascular death, stroke, or MI;
major bleeding, which was defined using modified criteria from ISTH, was a key safety outcome.
Additional secondary and tertiary outcomes included individual components of the primary
composite endpoint, acute limb ischemia, hospitalization, revascularization, and limb
amputation.
Clinical Benefits of Rivaroxaban
Summary: Compared to treatment with ASA alone, rivaroxaban + ASA reduced the risk of
cardiovascular death, stroke, or MI in patients with stable CVD. Patients treated with
rivaroxaban + ASA experienced significantly fewer strokes (including disabling or fatal
strokes), less cardiovascular death, and fewer cardiovascular-related hospitalizations. No
significant effect of rivaroxaban on hemorrhagic stroke or MI was observed.
Time-to-First Event Composite Endpoints
As noted above, the primary outcome of the COMPASS trial was a composite endpoint
consisting of the first occurrence of cardiovascular death, stroke, or MI.11 In the time to event
analysis, the hazard ratio for the primary outcome was 0.76 (95% CI: 0.66, 0.86; p<0.001;
number needed to treat [NNT]: 77). Patients treated with rivaroxaban + ASA had statistically
significantly fewer primary outcome events (4.1%) compared to patients in the ASA alone group
(5.4%); for the ASA alone group, this translates into an annual event rate of approximately 3%,
suggesting a relatively high-risk population. As mentioned above, these results led the
independent data and safety monitoring board to recommend early termination of the trial
after the first formal interim analysis (50% of planned events) for efficacy.

Two secondary composite outcomes, comprised of ischemic stroke, MI, acute limb ischemia,
and either death from coronary heart disease or cardiovascular death, also occurred in fewer
patients treated with rivaroxaban + ASA versus ASA alone (Table 3.2).
Table 3.2. Primary and Secondary Composite Outcomes in COMPASS11

Hazard
Rivaroxaban + ASA ASA
Ratio P-Value
n (%) n (%)
(95% CI)
Primary 0.76
CV Death, Stroke, or MI 379 (4.1) 496 (5.4) <0.001
Endpoint (0.66-0.86)
Ischemic Stroke, MI, ALI, or 0.72
329 (3.6) 450 (4.9) <0.001
Secondary Death from CHD (0.63-0.83)
Endpoints Ischemic Stroke, MI, ALI, or CV 0.74
389 (4.3) 516 (5.7) <0.001
Death (0.65-0.85)
ALI: acute limb ischemia, ASA: aspirin, CHD: coronary heart disease, CI: confidence interval, CV:
cardiovascular, MI: myocardial infarction
Individual Events
Individual components of the primary and secondary composite outcomes are presented in
Table 3.3. Patients treated with rivaroxaban + ASA experienced significantly fewer strokes, and
less cardiovascular death, death from coronary heart disease, and death from any cause. In an
exploratory analysis, rivaroxaban + ASA reduced the risk of disabling or fatal strokes (i.e., strokes
defined as a score between 3 and 6 on the modified Rankin Scale) by 42% (HR 0.58; 95% CI: 0.37
to 0.89; p=0.01).12-14 Hemorrhagic strokes occurred in more patients in the rivaroxaban + ASA
group but differences did not reach statistical significance.
Table 3.3. Individual Event Rates in COMPASS11,14

Rivaroxaban + ASA ASA Hazard Ratio
P-Value
n (%) n (%) (95% CI)
Stroke 83 (0.9) 142 (1.6) 0.58 (0.44-0.76) <0.001
Ischemic Stroke 64 (0.7) 125 (1.4) 0.51 (0.38-0.69) <0.0001
Hemorrhagic Stroke 15 (0.2) 10 (0.1) 1.49 (0.67-3.31) 0.33
MI 178 (1.9) 205 (2.2) 0.86 (0.70-1.05) 0.14
CV Death 160 (1.7) 203 (2.2) 0.78 (0.64-0.96) 0.02
Non-CV Death 153 (1.7) 175 (1.9) 0.87 (0.70-1.08) 0.20
Death from Coronary Heart Disease 86 (0.9) 117 (1.3) 0.73 (0.55-0.96) 0.03
All-Cause Death 313 (3.4) 378 (4.1) 0.82 (0.71-0.96) 0.01
Revascularization* 530 (6%) 553 (7%) 0.95 (0.84-1.07) 0.39
ASA: aspirin, CI: confidence interval, CV: cardiovascular, MI: myocardial infarction
*Reported in CAD subgroup.

Other Benefits of Rivaroxaban
Hospitalization
Hospitalization for cardiovascular causes (Appendix Table D4) occurred less in patients
randomized to rivaroxaban + ASA versus ASA alone (14.2% vs. 15.3%; HR 0.92; 95% CI: 0.86 to
1.00; p=0.04). The non-cardiovascular-related hospitalization rate was not statistically different
between arms.
Quality of Life
We did not identify any evidence related to quality of life for rivaroxaban + ASA, although the
European Quality of Life-5 Dimensions (EQ-5D) was implemented as a tertiary outcome in the
COMPASS trial. As of the time of this report, these data have not been published or presented
publicly.
Harms of Rivaroxaban
Summary: Patients treated with rivaroxaban + ASA experienced a significant increase in major
bleeding events, which led to permanent discontinuation of therapy in approximately 3% of
patients. Most major bleeding events occurred in the GI tract; proton pump inhibitor therapy
(PPI) had a protective effect on gastroduodenal bleeding (although not upper GI bleeding),
which was not statistically different between patients randomized to rivaroxaban + ASA and
ASA alone.
The COMPASS trial’s primary safety endpoint was major bleeding, which was assessed using a
modified definition from ISTH. The modified ISTH criteria included fatal bleeding, symptomatic
bleeding into a critical organ, bleeding into a surgical site requiring reoperation, or bleeding
leading to hospitalization (or an acute care visit that did not require an overnight stay); bleeding
events that did not meet the ISTH criteria were counted as minor. The most severe bleeding
event was recorded among patients with more than one event; the rate of total bleeding events
was not reported.
Major bleeding events occurred in significantly more patients treated with rivaroxaban + ASA
compared to ASA alone (3.1% vs. 1.9%; HR 1.70; 95% CI: 1.40 to 2.05; p<0.001); 2.7% of patients
in the rivaroxaban + ASA group permanently discontinued treatment due to bleeding, compared
to 1.2% in the ASA alone group.6,11 Selected bleeding outcomes are presented in Table 3.4 and
all bleeding outcomes are reported in Appendix Table D5.

Table 3.4. Selected Bleeding Outcomes in COMPASS
Rivaroxaban +
Outcome ASA Hazard Ratio (95% CI) P-Value
ASA
Major Bleeding 288 (3.1) 170 (1.9) 1.70 (1.40-2.05) <0.001
Fatal Bleeding 15 (0.2) 10 (0.1) 1.49 (0.67-3.33) 0.32
Non-Fatal Symptomatic
21 (0.2) 19 (0.2) 1.10 (0.59-2.04) 0.77
Intracranial Bleeding
Other Major Bleeding 210 (2.3) 112 (1.2) 1.88 (1.49-2.36) <0.001
Minor Bleeding 838 (9.2) 503 (5.5) 1.70 (1.52-1.90) <0.001
The most common bleeding site was the GI tract (1.5% vs. 0.7%; HR 2.15; 95% CI: 1.60 to 2.89;
p<0.0001). As previously noted, the COMPASS trial evaluated whether the addition of a PPI,
pantoprazole (40 mg once daily), could reduce the risk of upper GI bleeding.61 Clinically
significant upper GI bleeding was defined as a composite of overt bleeding (i.e., hematemesis
and/or melena) with a gastroduodenal lesion (peptic ulcer or neoplasia), overt upper GI
bleeding of unknown origin, occult bleeding (drop in hemoglobin of 2 g per deciliter or more),
symptomatic gastroduodenal ulcer with at least three days of GI pain, or at least five
gastroduodenal erosions with at least three days of GI pain, and upper GI obstruction or
perforation. Statistical differences in the occurrence of clinically significant upper GI bleeding
were not observed between the pantoprazole and placebo arms on the composite bleeding
endpoint, although pantoprazole did reduce the risk of gastroduodenal bleeding events (0.2%
vs. 0.4% for the pantoprazole and placebo groups, respectively; HR 0.52; 95% CI: 0.28 to 0.94).
There was no statistically significant interaction between pantoprazole and randomization to
rivaroxaban + ASA or ASA alone. Additional results from this study are reported in Appendix
Table D6.
Serious adverse events (SAEs) occurred in 7.9% of patients in the rivaroxaban + ASA arm versus
7.3% of patients on ASA alone; discontinuation due to non-bleeding AEs was not reported. The
FDA label for rivaroxaban carries a black box warning for premature discontinuation and
spinal/epidural hematoma.6 The warning states that discontinuing any oral anticoagulant,
including rivaroxaban, increases the risk of thrombotic events. Patients who are receiving
neuraxial anesthesia or are undergoing spinal puncture are at increased risk of epidural or spinal
hematomas, which may result in long-term paralysis. Data related to these warnings were not
reported in the COMPASS trial. The FDA label also includes a warning for serious and fatal
bleeding, and advises that an agent to reverse the anti-factor Xa activity of rivaroxaban is
available.
Net Clinical Benefit
To evaluate the balance or benefits and bleeding risk, COMPASS trial investigators assessed a
net-clinical-benefit outcome, which they defined as cardiovascular death, stroke, MI, fatal
bleeding, or symptomatic bleeding into a critical organ. The risk of this composite outcome was
lower with rivaroxaban + ASA than with ASA alone (HR 0.80; 95% CI: 0.70 to 0.91; p<0.001).11

However, these results should be interpreted with discretion, as they did not account for the full
primary safety endpoint of major bleeding.
Subgroup Analyses
Summary: Subgroup analyses in patients with CAD, PAD, renal dysfunction, mild-to-moderate
heart failure, recent CABG surgery, and in patients with high-risk features demonstrated a
consistent benefit for rivaroxaban + ASA as well as a consistently elevated risk of major
bleeding. In patients with PAD, rivaroxaban + ASA reduced the risk of major adverse limb
events, including major amputations.
CAD and PAD
Patients with CAD comprised 91% of the COMPASS trial and patients with PAD represented 27%
of the trial population; rivaroxaban + ASA reduced the risk of major adverse cardiac events and
increased the risk of bleeding in both subgroups.
In patients with PAD, rivaroxaban + ASA significantly lowered the risk of major adverse limb
events, defined as the development of acute or chronic limb ischemia during trial follow-up (HR
0.54; 95% CI: 0.35 to 0.84; p=0.0054).57,59 Rivaroxaban + ASA also reduced the risk of major
amputations by approximately 70% (HR 0.30; 95% CI: 0.11 to 0.80; p=0.011).

Table 3.5. Clinical Benefit and Safety of Rivaroxaban + ASA in CAD and PAD Subgroups11,57,58,63
Concomitant
Overall
CAD CAD Alone PAD PAD Alone CAD and
COMPASS
Subgroup Subgroup Subgroup Subgroup PAD
Population
Subgroup
HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)
Primary Endpoint
CV Death, Stroke, or 0.76 0.74 0.77 0.72 0.89 0.67
MI (0.66-0.86) (0.65-0.86) (0.66-0.91) (0.57-0.90) (0.55-1.44) (0.52-0.87)
Secondary Endpoint
Ischemic Stroke, MI, 0.72 0.72 0.75 0.68 0.78 NR
ALI, or Death from (0.63-0.83) (0.62-0.83) (0.63-0.90) (0.53-0.86) (0.48-1.25)
CHD
Ischemic Stroke, MI, 0.74 0.73 0.77 0.71 0.88 NR
ALI, or CV Death (0.65-0.85) (0.64-0.84) (0.65-0.90) (0.57-0.88) (0.57-1.34)
Major Bleeding
Fatal Bleeding,
Symptomatic
Bleeding into a
Critical Organ,
Bleeding into a 1.70 1.66 1.61
NR NR NR
Surgical Site (1.40-2.05) (1.37-2.03) (1.12-2.31)
Requiring
Reoperation, and
Bleeding that Led to
a Hospital Visit
ALI: acute limb ischemia, CAD: coronary artery disease, CHD: coronary heart disease, CI: confidence interval, CV:
cardiovascular disease, HR: hazard ratio, MI: myocardial infarction, PAD: peripheral artery disease
Renal Function
A statistically-significant reduction in the primary composite efficacy endpoint from the

COMPASS trial (i.e., cardiovascular death, stroke, or MI) was observed in patients with and
without moderate renal dysfunction (defined by an estimated glomerular filtration rate [eGFR]
of <60 ml/min or ≥60 ml/min, respectively; Table 3.6); results were consistent in patients with
greater renal dysfunction (eGFR <30 ml/min, HR 0.73; 95% CI: 0.28 to 1.91).60 The occurrence of
major bleeding events was similar in groups stratified by level of renal function.

Table 3.6. Clinical Benefit and Safety of Rivaroxaban Subgroups Defined by Renal Function60
Normal Renal
Moderate Renal
Overall COMPASS Function
Dysfunction
Population (eGFR≥60
(eGFR<60 ml/min)
ml/min)
HR (95% CI) HR (95% CI) HR (95% CI)
Primary 0.76 0.76 0.75
CV death, stroke, or MI
Endpoint (0.66-0.86) (0.64-0.90) (0.60-0.94)
Fatal bleeding, symptomatic
bleeding into a critical
Major organ, bleeding into a 1.70 1.81 1.47
Bleeding surgical site requiring (1.40-2.05) (1.44-2.28) (1.05-2.07)
reoperation, and bleeding
that led to a hospital visit
ALI: acute limb ischemia, ASA: aspirin, CHD: coronary heart disease, CI: confidence interval, CV: cardiovascular,
HR: hazard ratio, MI: myocardial infarction
Mild-to-Moderate Heart Failure
Patients with mild or moderate heart failure represented 22% of the COMPASS trial’s
population; patients with severe heart failure (i.e. left ventricular ejection fraction <30% or New
York Heart Association Class III or IV symptoms) were not eligible to participate. In prespecified
subgroup analyses in patients with and without a diagnosis of heart failure, rivaroxaban + ASA
significantly reduced the risk of the composite primary endpoint of cardiovascular death, stroke,
or MI (Table 3.7; p=0.28 for interaction).62 Rivaroxaban + ASA reduced the risk of death from
any cause compared to ASA alone (HR 0.66; 95% CI: 0.50 to 0.86) in patients with heart failure
but statistical differences were not observed in patients who did not have a history of heart
failure (p-value for interaction=0.05). The risk of major bleeding was increased in both
subgroups treated with rivaroxaban + ASA.
Table 3.7. Clinical Benefit and Safety of Rivaroxaban Subgroups Defined by Diagnosis of Heart
Failure62
Overall COMPASS Mild to Moderate
No Heart Failure
Population Heart Failure
HR (95% CI) HR (95% CI) HR (95% CI)
Primary 0.76
CV death, stroke, or MI 0.79 (0.68-0.93) 0.68 (0.53-0.86)
Endpoint (0.66-0.86)
Fatal bleeding,
symptomatic bleeding into
a critical organ, bleeding
Major 1.70
into a surgical site 1.79 (1.45-2.21) 1.36 (0.88-2.09)
Bleeding (1.40-2.05)
requiring reoperation, and
bleeding that led to a
hospital visit
ALI: acute limb ischemia, ASA: aspirin, CHD: coronary heart disease, CI: confidence interval, CV: cardiovascular,
HR: hazard ratio, MI: myocardial infarction

Additional Subgroups
Additional subgroup analyses focusing on patients randomized following CABG surgery or those
with one or more high-risk features such as diabetes, ≥2 vascular beds affected, and renal
insufficiency, suggest treatment effects that are similar to or greater than those in the overall
population.64,65
Clinical Benefits and Safety of DAPT
Our literature search did not identify any studies directly comparing rivaroxaban + ASA to DAPT
in the population of focus. Although we did not systematically review DAPT versus ASA alone,
we searched for RCTs that evaluated new initiation of DAPT (as opposed to continuation of
current DAPT therapy) in patients with stable CVD. We identified two RCTs of ticagrelor + ASA
and clopidogrel + ASA, respectively.15-18 These trials are summarized in Appendix D for context.
We also indirectly compared DAPT to rivaroxaban + ASA through an NMA of major adverse
cardiovascular events in patients with a recent MI (see below).
NMA
We performed an NMA in the subgroup of patients with a recent MI (i.e., in the two years prior
to randomization for the studies of rivaroxaban and ticagrelor, and at a median of two years
prior to randomization for the study of clopidogrel) to compare ticagrelor + ASA and clopidogrel
+ ASA with rivaroxaban + ASA. The analysis estimated the comparative risk of a composite
endpoint of cardiovascular death, stroke, or MI between each of the regimens of focus. The
results of our NMA, presented in Table 3.8, do not reveal statistical differences between
therapies. However, given the elevated risk of major bleeding that is associated with each of
the regimens, any analysis of comparative effectiveness is incomplete without an accompanying
analysis of comparative safety. We endeavored to also compare the incidence of major
bleeding between therapies but were unable to quantitatively synthesize the data due to the
use of important differences in definitions of major bleeding. Data informing the NMA as well
as a network diagram are reported in Appendix D.

Table 3.8. NMA Results Comparing the Risk of Cardiovascular Death, Stroke, or MI in Patients
Treated with Antithrombotic Therapy for Stable CVD
Rivaroxaban + ASA
0.91
Ticagrelor + ASA
(0.61 to 1.36)
0.91 1.00
Clopidogrel + ASA
(0.58 to 1.40) (0.75 to 1.32)
0.70 0.77 0.77
ASA
(0.48 to 1.02) (0.66 to 0.90) (0.61 to 0.98)
Each box represents the estimated hazard ratio and 95% credible interval for the combined direct and
indirect comparisons between two drugs. Estimates in bold signify that the 95% credible interval does
not contain one.
Key Studies of Icosapent Ethyl
Evidence on icosapent ethyl was primarily derived from the REDUCE-IT trial.19 REDUCE-IT was a
multinational, double-blind, Phase III trial that randomized patients at increased risk of ischemic
events to 2 g twice daily of icosapent ethyl (n=4089) or a placebo (n=4090) that contained
mineral oil to resemble the color and consistency of icosapent ethyl. Patients were eligible to
enroll in the trial if they were at least 45 years of age with established CVD (secondary
prevention cohort) or at least 50 years of age with diabetes mellitus and at least one additional
risk factor for CVD (primary prevention cohort). Patients were required to have elevated fasting
triglyceride levels (≥135 and <500 mg/dL) and well-controlled LDL cholesterol levels (>40 and
≤100 mg/dL) while on a stable dose of statins for at least four weeks. Key exclusion criteria
included severe heart failure, severe liver disease, planned coronary intervention, glycated
hemoglobin >10%, acute or chronic pancreatitis, or known hypersensitivity to fish or shellfish.
Additional inclusion and exclusion criteria can be found in Appendix Table D11.
At baseline, 71% of patients had established CVD and 29% made up the high-risk primary
prevention cohort.19 Approximately 58% of included patients had type 2 diabetes mellitus, 87%
had hypertension, and 48% had a prior MI. Most patients (93%) were receiving moderate-to-
high intensity statin therapy. Patients were followed for a median of 4.9 years. Additional
baseline characteristics are reported in Appendix Table D12.
The primary endpoint of the REDUCE-IT trial was a composite of cardiovascular death, nonfatal
MI, nonfatal stroke, coronary revascularization, and unstable angina in a time-to-event analysis.
Per suggestions from the FDA, a protocol amendment in 2016 designated a composite of
cardiovascular death, nonfatal MI, and nonfatal stroke as a key secondary endpoint. Additional
endpoints included time-to-event analyses of the individual components of the composite
endpoints as well as all-cause mortality. The effect of icosapent ethyl on total events (first plus
subsequent) was examined in prespecified analyses for both the primary and key secondary
composite endpoints.20

Table 3.9. Summary of the Reduction of Cardiovascular Events with Icosapent Ethyl—
Intervention Trial (REDUCE-IT)
Key
Treatment Primary Efficacy Secondary
Study Design Patient Characteristics
Groups* Endpoint Efficacy
Endpoint
Phase III, double- 1. Icosapent N=8179 A composite of CV A composite
blind RCT ethyl Median age: 64.0 death, stroke, MI, of CV death,
4 g/day Secondary prevention, %: 70.7 hospitalization for stroke, or MI
Median follow- Primary prevention, %: 29.3 unstable angina, or
up: 4.9 years 2. Placebo Median TG: 216.0 mg/dL coronary
Median LDL-C: 75.0 mg/dL revascularization
CV: cardiovascular, LDL-C: low density lipoprotein-cholesterol, mg/dL: milligram per deciliter, MI: myocardial
infarction, RCT: randomized controlled trial, TG: triglyceride
*Patients were also receiving a stable dose of statins in each treatment group.
Clinical Benefits of Icosapent Ethyl
Summary: Compared to optimal medical management alone (i.e., placebo), icosapent ethyl
reduced the risk of a composite outcome of cardiovascular death, stroke, MI, coronary
revascularization, or unstable angina in patients with established CVD or diabetes mellitus
and additional risk factors. Icosapent ethyl significantly reduced the risk of all individual
components of the primary composite endpoint as well as the composite secondary outcome
of cardiovascular death, MI, or stroke. A treatment benefit was also observed in analyses of
the first, subsequent, and total major adverse cardiovascular events.
Time-to-First Event Composite Endpoints
As noted above, the REDUCE-IT trial evaluated a composite of cardiovascular death, MI, stroke,
coronary revascularization, or unstable angina as its primary endpoint. In the time-to-event
analysis, icosapent ethyl reduced the risk of a primary endpoint event by 25% (HR 0.75; 95% CI:
0.68 to 0.83; p<0.001).19 At a median follow-up of 4.9 years (maximum 6.2 years), 17.2% of
patients treated with icosapent ethyl and 22.0% treated with placebo had a first primary
endpoint event (number needed to treat [NNT]: 21; 95% CI: 15 to 33). The annual event rate in
the placebo arm was approximately 4.4%, suggesting a very high-risk population.
The REDUCE-IT trial’s key secondary endpoint (cardiovascular death, MI, or stroke) also occurred
in fewer patients treated with icosapent ethyl compared to those receiving placebo (11.2% vs.
14.8%, respectively; HR 0.74; 95% CI: 0.65-0.83; p<0.001; NNT: 28; 95% CI: 20 to 47).19
Individual Events
Individual components of the primary composite endpoint are presented in Table 3.10.
Icosapent ethyl significantly reduced the risk of cardiovascular death by 20%, nonfatal MI by

30%, nonfatal stroke by 29%, coronary revascularization by 34%, and hospitalization for
unstable angina by 32%. Icosapent ethyl treatment did not result in a statistically significant
reduction in the risk of all-cause mortality.
Table 3.10. Key Efficacy Endpoints in REDUCE-IT19

Endpoint Icosapent Ethyl, n (%) Placebo, n (%) HR (95% CI) P-Value
Primary Composite
CV Death, Nonfatal MI,
Nonfatal Stroke, 705 (17.2) 901 (22.0) 0.75 (0.68–0.83) <0.001
Revascularization, and
Unstable Angina
Key Secondary Composite
CV Death, Nonfatal MI, 459 (11.2) 606 (14.8) 0.74 (0.65–0.83) <0.001
Nonfatal Stroke
CV Death 174 (4.3) 213 (5.2) 0.80 (0.66–0.98) 0.03
Nonfatal MI 237 (5.8) 332 (8.1) 0.70 (0.59-0.82) NR
Nonfatal Stroke 85 (2.1) 118 (2.9) 0.71 (0.54-0.94) NR
Coronary
376 (9.2) 544 (13.3) 0.66 (0.58-0.76) NR
Revascularization
Hospitalization for
108 (2.6) 157 (3.8) 0.68 (0.53-0.87) 0.002
Unstable Angina
All-Cause Mortality 274 (6.7) 310 (7.6) 0.87 (0.74–1.02) NS
CI: confidence interval, CV: cardiovascular, HR: hazard ratio, MI: myocardial infarction, NR: not reported, NS:
not specified
Total Events
The effect of icosapent ethyl on total events (first and subsequent) was examined in a pre-
specified analysis using a negative binomial regression model.20 A hierarchical approach was
used for event identification, in which a cardiovascular death superseded nonfatal events
occurring on the same day, and multiple nonfatal events occurring on the same day were
counted as one event.
The risk of total primary endpoint events, including cardiovascular death, MI, stroke,
revascularization, and unstable angina, was reduced by 30% with icosapent ethyl compared to
placebo (rate ratio (RR): 0.70; 95% CI: 0.62, 0.78). Treatment with icosapent ethyl resulted in a
28% risk reduction compared to placebo (RR: 0.72; 95% CI: 0.63-0.82) on the REDUCE-IT trial’s
key secondary endpoint of cardiovascular death, stroke, or MI.
In a prespecified supportive analysis, hazard ratios for the time to first, second, and third events
were calculated using the Wei-Lin-Weissfeld method, and the rate ratio for fourth or
subsequent events was calculated using the negative binomial regression model. Icosapent
ethyl reduced the risk of first primary endpoint events by 25%, second primary endpoint events
by 32%, third primary endpoint events by 31%, and fourth or subsequent primary endpoint
events by 48%; in contrast, the hazard ratios for secondary endpoint events remained relatively
consistent across the event sequence (Table 3.11).20

Table 3.11. Hazard/Rate Ratios for Total and Subsequent Ischemic Events20
Total Events First Event Second Event Third Event ≥Fourth Event
Endpoint
RR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) RR (95% CI)
Primary Composite
CV Death, Nonfatal
0.70 0.75 0.68 0.69 0.52
MI, Nonfatal Stroke,
(0.62-0.78) (0.68-0.83) (0.60-0.78) (0.59-0.82) (0.38-0.70)
Revascularization,
and Unstable Angina
Key Secondary
Composite
0.72 0.74 0.75 0.79
CV Death, Nonfatal Not reported
(0.63-0.82) (0.65-0.83) (0.63-0.89) (0.65-0.96)
MI, and Nonfatal
Stroke
CI: confidence interval, CV: cardiovascular, HR: hazard ratio, RR: rate ratio
Harms of Icosapent Ethyl
Summary: Rates of serious TEAEs were similar in patients treated with icosapent ethyl and
placebo. A greater proportion of patients treated with icosapent ethyl experienced serious
bleeding-related disorders, as well as peripheral edema, constipation, and atrial fibrillation.
The incidence of serious and non-serious TEAEs leading to treatment discontinuation was
similar for both treatment arms.
The incidence of serious TEAEs was similar in the icosapent ethyl and placebo arms of the
REDUCE-IT trial (30.6% vs. 30.7%, respectively).19 Serious TEAEs leading to death occurred in
2.3% of patients treated with icosapent ethyl and 2.5% of patients who received placebo.
Serious bleeding-related disorders, identified using the Medical Dictionary for Regulatory
Activities (MedDRA), occurred in a greater proportion of patients treated with icosapent ethyl,
although the difference was not statistically significant (2.7% vs. 2.1%, p=0.06). No fatal
bleeding events occurred in either group and rates of hemorrhagic stroke, central nervous
system bleeding, and GI bleeding did not statistically differ. TEAEs that occurred in
proportionately more patients treated with icosapent ethyl included peripheral edema (6.5% vs.
5.0%, p=0.002), constipation (5.4% vs. 3.6%, p<0.001), and atrial fibrillation (5.3% vs. 3.9%,
p=0.003).19 Hospitalization for atrial fibrillation or flutter was significantly higher in the
icosapent ethyl arm compared to placebo (3.1% vs. 2.1%; p=0.004).
Approximately 11% of patients randomized to placebo and 10% randomized to icosapent ethyl
discontinued the study early.9 The rate of TEAEs leading to discontinuation of the study drug
was similar for patients treated with icosapent ethyl and placebo (7.9% vs 8.2%, respectively) as
was the rate of drug discontinuation due to serious TEAEs (2.2% vs 2.3%, respectively). The rate
of all-cause study drug discontinuation was not reported for the entire trial population but was
reported for patients who had a primary endpoint event. At the time of a first primary endpoint
event, 18.7% of patients randomized to icosapent ethyl and 18.2% of patients randomized to
placebo had discontinued the study drug.20

Although we did not consider earlier trials of icosapent ethyl in patients with elevated
triglyceride levels (i.e., MARINE66 and ANCHOR67) as part of our core study set, we also reviewed
safety data from these studies as they evaluated the same dose of icosapent ethyl as the
REDUCE-IT trial. Over 12 weeks of treatment in both trials, 35-45% of patients treated with 4 g
daily of icosapent ethyl experienced a TEAE. Most TEAEs were determined to be of mild or
moderate severity and were not considered related to the study drug. The only TEAEs occurring
in numerically more patients treated with icosapent ethyl 4 g/d in either trial was arthralgia in
the ANCHOR trial (1.7% vs 0.4%; see Appendix Table D15).
Subgroup Analyses
Summary: Analyses of the primary and key secondary endpoints from the REDUCE-IT trial did
not reach statistical significance in the primary prevention subgroup. A consistent treatment
benefit was observed for icosapent ethyl in subgroups of patients with and without diabetes
mellitus, with and without renal dysfunction, with and without elevated triglyceride levels,
and with and without elevated levels of the inflammation marker hsCRP.
Subgroups Defined by Baseline Characteristics
The effect of icosapent ethyl on the risk of primary and key secondary composite endpoint
events were reported for various subgroups of interest (Table 3.12). For patients with
established CVD at baseline (secondary prevention cohort), icosapent ethyl statistically-
significantly reduced the risk of the composite primary endpoint of cardiovascular death, MI,
stroke, revascularization, or unstable angina by 27%. The risk of the key secondary composite
endpoint, cardiovascular death, MI, or stroke, was reduced 28% with icosapent ethyl in the
secondary prevention cohort (Table 3.12).9 For patients in the primary prevention cohort, there
were no statistically significant risk reductions for either the primary or key secondary
composite endpoints (Table 3.12). At a prespecified alpha level of 0.15, there was evidence of a
significant differential effect between the secondary and primary prevention subgroups on the
primary composite endpoint (p-value for interaction=0.14) but not the secondary composite
endpoint.
A consistent treatment benefit was observed for both the primary and key secondary composite
endpoints in subgroups with and without diabetes mellitus, with and without renal dysfunction
(eGFR<60 vs. ≥60 mL/min/1.732), with and without elevated triglyceride levels (≥150 vs. <150
mg/dL and ≥200 vs. <200 mg/dL), and in subgroups stratified by hsCRP level at baseline (≤2 vs >2
mg/L) (Table 3.12).9

Table 3.12. Primary and Key Secondary Composite Endpoints for Key Subgroups Defined by
Baseline Characteristics9
Primary Composite Key Secondary
CV Death, MI, Stroke, Composite
Subgroup
Revascularization, and CV Death, MI, and
Unstable Angina Stroke
HR (95% CI) HR (95% CI)
Secondary Prevention 0.73 (0.65–0.81) 0.72 (0.63–0.82)
CV Risk
Primary Prevention 0.88 (0.70–1.10) 0.81 (0.62–1.06)
Diabetes Yes 0.77 (0.68–0.87) 0.70 (0.60–0.81)
Mellitus No 0.73 (0.62–0.85) 0.80 (0.65–0.98)
<60 mL/min/1.73 m2 0.71 (0.59–0.85) 0.71 (0.57–0.88)
≥60 to <90 mL/min/1.73 0.80 (0.70–0.92) 0.77 (0.64–0.91)
eGFR
m2
≥90 mL/min/1.73 m2 0.70 (0.56–0.89) 0.70 (0.52–0.94)
≤2 mg/L 0.68 (0.58–0.79) 0.73 (0.61–0.89)
hsCRP
>2 mg/L 0.81 (0.71–0.93) 0.73 (0.63–0.86)
≥150 mg/dL 0.75 (0.68-0.83) 0.74 (0.65-0.84)
<150 mg/dL 0.79 (0.57-1.09) 0.66 (0.44-0.99)
Triglycerides
≥200 mg/dL 0.73 (0.64–0.83) 0.75 (0.65–0.88)
<200 mg/dL 0.79 (0.67–0.93) 0.71 (0.58–0.86)
CI: confidence interval, CV: cardiovascular, eGFR: estimated glomerular filtration rate, HR: hazard ratio,
hsCRP: highly sensitive C-reactive protein, MI: myocardial infarction
Subgroups Defined by On-Treatment Changes
As stated earlier, the placebo used in REDUCE-IT contained mineral oil to resemble the color and
consistency of icosapent ethyl. Patients receiving placebo had marked increases in LDL
cholesterol (LDL-C) levels at year one compared to icosapent ethyl (median percent change:
10.2% vs. 3.1%, p<0.001) and in hsCRP levels at year two (32.3% vs. -13.9%, p<0.001). These
unexpected increases may indicate that the mineral oil used was not biologically inert and may
have resulted in an overstated treatment effect of icosapent ethyl relative to placebo. In
response to these concerns, the manufacturer posted the results of a post-hoc analysis on their
website, which stratified patients with and without on-trial increases in LDL-C in the placebo
arm.68 The post-hoc analysis suggested that treatment with icosapent ethyl resulted in
significant reductions in the risk for both the primary and key secondary composite endpoints,
irrespective of whether patients in the placebo arm experienced an increase in LDL-C (Table
3.13). To the best of our knowledge, however, similar analyses have not been performed in
relation to changes in hsCRP from baseline.

Table 3.13. Subgroup Analysis by Change in LDL-C at Year One
Primary Composite
CV Death, MI, Stroke, Key Secondary Composite
Post-Hoc Analysis Revascularization, and CV Death, MI, and Stroke
Unstable Angina HR (95% CI)
HR (95% CI)
Icosapent ethyl vs.
0.79 0.80
placebo LDL-C
(0.70 to 0.88) (0.70 to 0.93)
Increase
Icosapent ethyl vs.
0.79 0.74
placebo LDL-C
(0.69 to 0.91) (0.63 to 0.88)
No change/decrease
Placebo LDL- increase vs.
1.01 0.92
placebo LDL-C
(0.87 to 1.17) (0.77 to 1.10)
No change/decrease
CI: confidence interval, CV: cardiovascular, HR: hazard ratio, LDL: low-density lipoprotein, MI: myocardial
infarction
Controversies and Uncertainties
While the available evidence for both rivaroxaban and icosapent ethyl is indicative of a potential
net clinical benefit, there are several concerns with the design and results of these trials that
should be considered along with data on clinical outcomes and potential harms of these
treatments. Concerns are organized by intervention of interest in the sections that follow.
Rivaroxaban
The generalizability of the COMPASS trial population is subject to a number of uncertainties. For
one, study entry criteria of stable CAD and PAD as well as documentation of atherosclerosis in at
least two vascular beds among patients age <65 years ensured a population at high risk of
recurrent cardiovascular events, but exclusion of patients at high bleeding risk and further
exclusion of 8% patients not tolerating or adherent to run-in ASA therapy likely resulted in a
sample at reduced bleeding risk relative to the potential candidate population for
rivaroxaban.21,22
In addition, we cannot exclude the possibility that the clinical benefits observed in COMPASS are
somewhat overstated due to the trial having been stopped early for benefit after a mean of 23
months of follow-up.23 The Kaplan-Meier estimates provide some reassurance that benefits
observed after approximately one year of follow-up continued until the trial was stopped21, but
the balance of event reduction and bleeding risks beyond this point is currently unknown.
Relatedly, while the trial considered a measure of “net benefit” that included both
cardiovascular and bleeding events and found a statistically-significant 20% reduction in risk,
this definition only included fatal bleeding and symptomatic bleeding into a critical organ, and
no other definitions of major bleeding that comprised the primary safety outcome. Our own
calculations of risk-benefit using the full event definitions indicate an NNT of 77 for

cardiovascular events and a number needed to harm (NNH) of 83 for bleeding events,
suggesting a much smaller net benefit on average than that reported in COMPASS. To be sure,
the individual calculus may be clear for avoiding a severe MI or disabling stroke relative to a
treatable bleed at an acute care facility without an overnight stay, but the calculus for avoiding a
moderately symptomatic MI with increased troponin levels relative to a critical organ bleed
would be very different.22
The decision to separately randomize patients to receive the PPI pantoprazole or placebo within
the rivaroxaban + ASA, rivaroxaban alone, and ASA alone groups is a puzzling one, given that
clinical guidelines recommend routine use of PPIs for gastroprotection in patients receiving
combination anticoagulation + ASA therapy24 but not for anticoagulants or ASA alone. Recently-
reported results from COMPASS suggest that pantoprazole does not reduce the risk of upper GI
bleeding but does reduce bleeding from gastroduodenal lesions relative to placebo, and that
this finding is consistent regardless of the anticoagulation strategy used.69 It is unclear how this
finding would change clinical practice, however, given the well-established benefit-risk profile
and generic availability of PPIs.70-72
Finally, while the indications for combination treatment with rivaroxaban and DAPT with a P2Y12
inhibitor do not completely overlap, there is a large subset of patients with a recent MI event
who could conceivably be candidates for either treatment approach. Indeed, some clinicians
have called for further research comparing DAPT to combination therapy with ASA and a factor
Xa inhibitor (e.g., rivaroxaban).25,26 In the absence of head-to-head trials at the time of this
report’s publication, we attempted to compare the regimens indirectly through a network meta-
analysis. However, while only small differences exist in the definitions of clinical events across
the major trials of these regimens, the same cannot be said for definitions of major bleeding,
which differed substantially across trials; in some cases, a common definition was used but
modified to enough of an extent that we could not attempt quantitative comparisons with any
confidence. While this is a source of frustration for producers of comparative effectiveness
research like us, the real harm done is to the patient-clinician shared decision. Patients, in
particular, deserve to know how the major benefits and risks compare for treatments they
expect to receive over a lifetime. Standard and well-accepted definitions of both cardiovascular
events and bleeding risks exist, and the fact that they have not been used consistently and
without modification in as important a clinical area as this is a disservice to patients and the
clinicians who care for them.
Icosapent Ethyl
As noted previously, the placebo vehicle used in the REDUCE-IT trial (as well as earlier trials of
icosapent ethyl) contained mineral oil to mimic the viscosity of the active agent. Biomarker
changes observed in the trial raise the possibility that the mineral oil used was not biologically
inert, however; patients in the placebo arm experienced a threefold-higher percentage increase
in LDL-C at year one (10.2% vs. 3.1% for icosapent ethyl, p<0.001 for between-group difference)
and a substantial increase in the inflammation marker hsCRP at year two (32.9% vs. -13.9%,
p<0.001), adding to documented concerns regarding the mineral oil’s potential interference

with statin absorption.9 As described above, the manufacturer conducted a post-hoc analysis,
which stratified patients with and without on-trial increases in LDL-C in the placebo arm.68 The
results of this analysis suggested a consistent risk reduction with icosapent ethyl irrespective of
whether LDL-C increased in the placebo arm.9 However, it is difficult to interpret analyses of
effects utilizing on-treatment response subgroups, and so residual concerns remain about a
potentially biologically active “placebo” and the possibility that the true effect of icosapent ethyl
may be attenuated from that observed in the REDUCE-IT trial. To the best of our knowledge,
similar analyses have not been performed in relation to changes in hsCRP from baseline.
Other findings from REDUCE-IT give rise to additional uncertainties. For example, a separate
publication described a larger effect size for icosapent ethyl when total ischemic events (rather
than time to first event) are considered, as well as improved levels of risk reduction with each
subsequent event.20 This type of analysis is controversial, however, given the relation that often
exists between event types (e.g., nonfatal MI followed by revascularization or death) and the
consequent inflation of event rates.73 The authors addressed this by bundling multiple events
occurring on the same day into one and specifying multiple statistical models. Subsequent
events were evaluated using the Wei-Lin-Weissfeld model, however, which has been previously
criticized for overstating the population at risk of subsequent events, which may lead to
overestimates of reductions in the risk of these events.74,75 Other model specifications not
subject to this form of bias, such as the Prentice-Williams-Petersen and kinetic modeling
techniques,73,75 were not employed in this analysis.
We also note that the results of this trial stand apart from many prior studies of omega-3
preparations that showed little to no cardiovascular benefit.27 Indeed, when a Bayesian
approach is taken to the overall evidence base, the interpretation of REDUCE-IT’s findings will
differ depending on whether prior expectations for these results are pessimistic, realistic, or
optimistic.28
Several possible explanations for differences between REDUCE-IT’s conclusions and those of
previous studies have been posited, including use of an EPA-only formulation. Docosahexaenoic
acid (DHA), another common component of omega-3 preparations, has been found to increase
LDL-C levels when used alone or in combination with EPA,76 although findings from a recent AHA
science advisory indicate this is only the case in patients with very high triglycerides, and
recommend prescription forms of EPA+DHA and EPA alone for triglyceride reduction.46 Other
possible reasons include a higher daily dose than previously studied, and the possibility of
metabolic effects of EPA other than triglyceride-lowering alone.9 Indeed, patients in the
REDUCE-IT trial had baseline elevations in triglyceride levels, but subgroup analyses suggested
that the effect of icosapent ethyl may be similar across triglyceride categories.
It is worth noting that reductions in cardiovascular events of approximately 20% were observed
in a prior EPA-only trial (JELIS), which randomized approximately 19,000 Japanese patients to
1.8 g of EPA in addition to statin therapy versus statins alone over a mean of 4.6 years of follow-
up.29 However, the JELIS trial was open-label and showed no reductions in cardiovascular death,
so its relevance to the results of REDUCE-IT is unclear. The JELIS trial was also conducted in a

Japanese population with a much higher baseline consumption of fish than is typically seen in
the US; very few patients in REDUCE-IT were from the Asia-Pacific region making comparisons
across trials difficult.30,31
Regardless of issues of trial design or interpretation, the greatest uncertainty may be in how
generalizable the REDUCE-IT results are and therefore what the most appropriate target
population will be. As with COMPASS, the patients enrolled in REDUCE-IT were at very high risk
of cardiovascular events, as illustrated by a placebo event rate of approximately 4.4% per year
over the 4.9-year median duration of follow-up.9 Patients also were on statin therapy, and it is
unclear whether icosapent ethyl would be effective in patients not treated with statins. How
the benefits of icosapent ethyl translate to an eligible population that is certain to be both
broader and at lower risk than the trial population remains to be seen.

3.4 Summary and Comment
Using the ICER Evidence Matrix (Figure 3.1), we assigned evidence ratings to rivaroxaban + ASA
and icosapent ethyl relative to the comparators of interest for this review (Table 3.14).
Figure 3.1. ICER Evidence Rating Matrix
Table 3.14. ICER Evidence Ratings

Intervention Comparator ICER Evidence Rating
Rivaroxaban + ASA ASA B+
Rivaroxaban + ASA DAPT I
Icosapent Ethyl Optimal medical management B+
ASA: aspirin, DAPT: dual antiplatelet therapy

Rivaroxaban + ASA versus ASA Alone
Compared to ASA alone, rivaroxaban + ASA significantly reduced the risk of cardiovascular
death, stroke, or MI in patients with stable CVD. Patients treated with rivaroxaban + ASA
experienced significantly fewer strokes (including disabling or fatal strokes), less cardiovascular
death, less all-cause mortality, fewer major adverse limb events, and fewer cardiovascular-
related hospitalizations. Although rivaroxaban + ASA significantly increased the risk of major
bleeding events, the COMPASS trial’s inclusion of bleeds leading to presentation at an acute
care facility in its primary safety analysis of major bleeding may have led to the inclusion of both
consequential and somewhat inconsequential bleeding events in this analysis. Bleeding events
of greatest severity—i.e., fatal bleeding, symptomatic bleeding into a critical organ, and nonfatal
symptomatic intracranial hemorrhages—were not significantly increased by adding rivaroxaban
to ASA. We therefore have high certainty that rivaroxaban + ASA provides a small-to-substantial
net health benefit in patients with CAD, PAD, or both conditions (“B+”).
Rivaroxaban + ASA versus DAPT
We did not identify any head-to-head studies that directly compared rivaroxaban + ASA to DAPT
in patients with stable CVD. Although an indirect comparison of the risk of major adverse
cardiovascular events in patients with a recent MI suggested that DAPT may provide a similar
cardioprotective benefit as rivaroxaban + ASA, clinically significant differences in the way major
bleeding was defined in the clinical trials of focus precluded a companion analysis of relative
bleeding risks. We also note that those with a recent MI represented a relatively small subset of
patients in the COMPASS trial, so the comparative benefits and risks of these two strategies in
the remaining CAD and PAD population are unknown. We therefore have low certainty of
whether rivaroxaban + ASA provides a negative, comparable, or positive net health benefit
compared to DAPT in patients with CAD or PAD (“I”).
Icosapent Ethyl versus Optimal Medical Management
Icosapent ethyl reduced the risk of major adverse cardiovascular events in patients with
established CVD or diabetes mellitus and additional risk factors compared to optimal medical
management alone (i.e., placebo). The therapy was generally well-tolerated, despite a slight
increase in the incidence of major bleeding disorders. However, over 4.9 years of follow-up, no
fatal bleeding events occurred, and rates of TEAEs were comparable between the icosapent
ethyl and placebo arms. Although we are uncertain whether the use of mineral oil may have
caused some harm to the placebo group, we do not believe that this theory can account for the
entire benefit observed in the REDUCE-IT trial. We believe that the results of REDUCE-IT likely
apply across a range of baseline triglyceride levels but are uncertain whether the results
generalize to patients not treated with statins. For adults with established CVD or at high risk of
cardiovascular events who are being treated with statins, we have high certainty that icosapent
ethyl provides a small-to-substantial net health benefit (“B+”).

4. Long-Term Cost Effectiveness
4.1 Overview
The primary aim of this analysis was to estimate the cost effectiveness of rivaroxaban and
icosapent ethyl as additive therapies to optimal medical management in patients with
established CVD, and in the case of icosapent ethyl, also in patients without evidence of CVD but
with diabetes and at least one additional risk factor. A Markov cohort model was constructed to
compare the addition of rivaroxaban to ASA therapy to ASA alone and to compare the addition
of icosapent ethyl to optimal medical management (including statins) to optimal medical
management (including statins) alone. Rivaroxaban and icosapent ethyl were modeled
separately but shared a similar overall model structure. Patient survival, quality-adjusted
survival, and health care costs from a health care sector perspective were estimated over a
lifetime time horizon for each intervention and comparator. Costs and outcomes were
discounted at 3% per year. While the base-case analysis took a health care sector perspective,
productivity losses to the patient and caregiver were considered in a scenario analysis using a
modified societal perspective.
4.2 Methods
Model Structure
The Markov model focused on an intention to treat analysis, with a hypothetical cohort of adult
patients with established CVD (or at high CVD risk) being treated with optimal medical
management entering the model. The model included health states that define the pathways of
CVD and that have been used in previous modeling efforts.77-81 The base-case health states
included major cardiovascular events of MI and stroke, as well as post-event health states and
death (from cardiovascular and other causes). A scenario analysis included other cardiovascular
events beyond MI and stroke (e.g., revascularization or unstable angina) in the event and post-
event health states. Additional consequences such as major adverse limb events for
rivaroxaban, as well as SAEs, were tracked in the model. For these additional consequences, we
assumed event probabilities were equal for all living health states and therefore did not require
additional health states in the model structure.
Specifically, the CVD cohort began on treatment and could stay in that state, pass into event
states of MI or stroke, or death (Figure 4.1). Patients who experienced a cardiovascular event
moved into post-event health states, where they may have had higher likelihood for death as
compared to the general CVD prevention population. Patients remained in the model until they
died. All patients could transition to death from all-causes from any of the alive health states.
Death could have occurred from all-cause or cardiovascular event/post-event related morality.
As patients moved through the model over the course of their lifetime, they collected costs and
health utility weights related to the management and treatment of specific cardiovascular

conditions. The cumulative sum of costs, survival time and utility weights produced model
outputs such as lifetime costs, life years (LYs), quality-adjusted life years (QALYs), and equal
value of life years gained (evLYG). An analysis of the incremental cost per evLYG is included in
this report to complement the cost per QALY calculations and provide policymakers with a
broader view of cost effectiveness. A description of the methodology used to derive the evLYG
can be found in Appendix E. Other outputs of the model included cumulative incidence of
cardiovascular events and cardiovascular events avoided. The model was developed in
Microsoft Excel 2016 (Redmond, WA).
Figure 4.1. Model Framework
Post-event states:
post-MI or post-
stroke
Established CVD or high CV event states: MI

CVD risk or stroke*
All-cause and CV-

specific death
CV: cardiovascular, CVD: cardiovascular disease, MI: myocardial infarction

Other treatment-specific modeled events include major adverse limb events and other SAEs.
*Other CV events such as revascularization and unstable angina included in scenario analysis.
Target Population
The population of focus was adults with established CVD being treated with optimal medical
management. For the assessment of icosapent ethyl, patients without known CVD but at high
risk for cardiovascular events were also considered. The modeled populations’ characteristics
were consistent with the average across trial arms in the pivotal trials (Tables 4.1 and 4.2). The
population of study for rivaroxaban was, on average, 68 years old, 78% male, 21% with smoking
history, 38% with diabetes, 62% with prior MI, 4% with prior stroke, 22% with heart failure, and
with a number of commonly prescribed therapies as part of their medical management. The
population of study for icosapent ethyl was, on average, 64 years old, 71% male, 15% with
smoking history, 58% with diabetes, and 71% with prior CVD events. These model
characteristics have limited impact within the model, by influencing the time-varying clinical
event rate estimates (see Transition Probabilities).

Table 4.1. Base-Case Model Patient Characteristics for Rivaroxaban Evaluation
Characteristic Overall Source
Age, Years, Mean (SD) 68.2 (7.9)
Male (%) 78.0 82,83
Total Cholesterol, mmol/liter 4.3 (3.5)

Systolic Blood Pressure, mmHg 135.5 (17.57)
82
Smoking (% Yes) 21.4
Diabetes (% Yes) 37.9
Prior MI (% Yes) 62.1
Prior Stroke (% Yes) 3.8
Heart Failure (% Yes) 21.5
CAD (% Yes) 90.6
82,83
PAD (% Yes) 27.3
ACE Inhibitor or ARBs (% Yes) 70.8
Calcium-Channel Blocker (% Yes) 26.8
Diuretic (% Yes) 28.5
Beta-Blocker (% Yes) 70.0
Lipid-Lowering Agent (% Yes) 89.7
ACE: angiotensin-converting-enzyme, ARB: angiotensin II receptor blocker, CAD: coronary artery disease, MI:
myocardial infarction, PAD: peripheral artery disease, SD: standard deviation
Table 4.2. Base-Case Model Patient Characteristics for Icosapent Ethyl Evaluation
Characteristic Overall Source
19
Age, Years, Median (IQR) 64.0
Male (%) 71.2
High-Density Lipoprotein, mg/dL, Median (IQR) 40.0
Low-Density Lipoprotein, mg/dL, Median (IQR) 75.0
Triglycerides, mg/dL, Median (IQR) 216
84
Smoking (% Yes) 15.2
19,84
Diabetes – Type 2 (% Yes) 57.8
19
Prior CVD Events (% Yes) 70.7
CVD: cardiovascular disease, IQR: interquartile range
*Data not available in publicly disclosed sources.
Treatment Strategies
Interventions
The list of interventions was developed with input from patient organizations, clinicians,
manufacturers, and payers. The full list of interventions is as follows:
1. Rivaroxaban + ASA
• Patients are assumed to also be receiving optimal medical management.
2. Icosapent ethyl
• Patients are assumed to also be receiving optimal medical management,
including statins.

Comparators
Comparators were defined to reflect the input of clinicians and other stakeholders on treatment
strategies that would be considered relevant alternatives for the overall population of interest
or a prominent subset, as well as the comparators as defined in major clinical studies of
rivaroxaban and icosapent ethyl.
1. Rivaroxaban comparators:
• Optimal medical management including ASA without an additional antiplatelet
agent.
• DAPT with clopidogrel (scenario analysis only).
2. Icosapent ethyl comparator:
• Optimal medical management including statin therapy.
Key Model Characteristics and Assumptions
Model cycle length was one year, based on what was observed in prior published economic
models and clinical data. The base-case analysis assumed a lifetime horizon, consistent with the
ICER Value Framework. The base-case analysis took a health care sector perspective and thus
focused on direct medical care and drug costs only. Costs and outcomes were discounted at 3%
per year. Key model assumptions are described in Table 4.3.
Table 4.3. Key Model Assumptions

Assumption Rationale
Given differences across the severity of endpoints in
Individual hazard ratios were used for each
terms of cost, disutility, or likelihood of CV death,
subcomponent of composite endpoints observed in
endpoint subcomponents were tracked in the
the clinical trials. Endpoint subcomponents
model. Relative reductions in other CV events such
included the common major adverse CV events: MI,
as revascularization and unstable angina were not
stroke, and CV death. Other CV events (e.g.,
included in the base case due to potential
revascularization and unstable angina) were
associations and double counting issues with MI and
evaluated in a scenario analysis.
stroke events.
Based on the clinical review critique, time to first
event analyses were the statistical analyses of
primary focus in trials, and there were statistical
Subsequent CV events (second, third, and fourth
concerns regarding correlations between
events) had the same overall HR as the first CVD
subsequent event types. The model was calibrated
event.
to the overall CV event rate observed in the control
arm (MI, stroke, and CV death), with the HR applied
to events associated with the modeled intervention.
This cohort-level model allowed for multiple events
Patients could have more than one event in the within a model cycle by assuming that the costs and
same cycle, with costs and disutilities for multiple disutility of one event (stroke or MI) could be added
events assumed to be additive. to that of the costs and disutility of another event
(MI or stroke).

Assumption Rationale
There was a higher risk of CV death for patients in a
Literature-based evidence.85-90
CV event or post-event health state.
Because the trial evidence suggested high-risk
Model event rate was consistent with control arm
populations, the model was calibrated to the trial
first event rates (for MI, stroke, and CV death) from
populations in terms of first MI, stroke, and CV
clinical trials.
death events.
Patients continued on treatment after first event in Patients continuing on therapy after an event was
the model. consistent with the trial evidence.
Patients who discontinued treatment were not re- Patients discontinuing therapy did not re-initiate in
treated with the same initiating therapy. the trials for both therapies. After the average trial
Discontinuation rates mirrored trial evidence and duration, the model assumed an annualized
were forecasted based on annualized discontinuation rate consistent with discontinuation
discontinuation due to SAEs. due to SAEs from the trials.
CV: cardiovascular, CVD: cardiovascular disease, HR: hazard ratio, MI: myocardial infraction, SAE: serious
adverse event
Model Inputs
Model inputs were estimated from the evidence review, published literature, and information
provided by stakeholders. The inputs that informed the models for rivaroxaban and icosapent
ethyl are described below.
Clinical Inputs
Key clinical inputs for the model included validated CVD risk prediction models, baseline trial-
based clinical markers (e.g., cholesterol, triglycerides), baseline comorbid conditions (e.g.,
diabetes), and other baseline factors (e.g., smoking, event history, etc.).
Transition Probabilities
Cardiovascular events included in the base-case model were: MI, stroke, and cardiovascular-
related mortality. Validated cardiovascular risk calculators91 were used to estimate time-varying
annualized event rates within the control arm (Table 4.4). The control arm’s 10-year risk of
cardiovascular events was calibrated such that the model produced consistent first
cardiovascular events observed over the same period as within the trial.
For rivaroxaban, the model calibration varied baseline risk while holding constant the
proportion of MI (59% of non-fatal first events restricted to MI and stroke only), stroke (41% of
non-fatal first events restricted to MI and stroke only), and death derived from first event trial-
based results (37% of non-fatal MI, non-fatal stroke, and cardiovascular death). For rivaroxaban,
the mean treatment follow-up duration was 23 months. At the beginning of the model, the
control arm’s annualized cardiovascular event rate was determined by comparing the observed
first cardiovascular event and cardiovascular death over an average of 23 months to that of the
model’s estimates at the end of two years (24 months). These annualized rates in the model
varied with time based on the validated cardiovascular risk calculators.91

For icosapent ethyl, the model calibration varied baseline risk while holding constant the
distribution of MI (73% of non-fatal first events restricted to MI and stroke only), stroke (26% of
non-fatal first events restricted to MI and stroke only), and death derived from first event trial-
based results (32% of non-fatal MI, non-fatal stroke, and cardiovascular death). For icosapent
ethyl, the median follow-up duration was 58.8 months. At the beginning of the model, the
control arm’s annualized cardiovascular event rate was determined by comparing the observed
first cardiovascular event and cardiovascular death over an average of 58.8 months to that of
the model’s estimates at the end of five years (60 months). These annualized rates in the model
varied with time based on the validated cardiovascular risk calculators.91
Subsequent events were included in the model as well as first events. We assumed that the risk
of subsequent events would be the same as that of first events even if the risk calculator91
suggested lower likelihood of non-fatal cardiovascular events, given the relatively high severity
of the populations in the COMPASS and REDUCE-IT trials and the challenges inherent in
evaluating subsequent event risks in situations with event types that are not independent from
one another. Once calibrated to the trial’s control arm first observed events, these same risk
calculator parameters were also used in the model’s treatment arm in combination with the
treatment- and event-specific hazard ratios.
Table 4.4. Sources for Baseline Risk Equations for First Future Events, Subsequent Events, and
Mortality
Baseline Risk Equations Values Source
First Future Event (MI or Stroke) Varies by age and risk factors
91
Subsequent Future Events (MI or
Varies by age and risk factors
Stroke)
Calibrated to death from first
Event-Specific Mortality event using trial-based results
*Multiple sources20,21,85-
from standard of care arms 90,92
Increased mortality relative risk of
Mortality Post-MI or Stroke
2.5
Mortality, All-Cause Varies by age US Life Tables93
* An earlier version of this report incorrectly detailed these sources. It has been corrected here.
MI: myocardial infarction
The treatment- and event-specific hazard ratios for endpoints from the treatment-specific
clinical trials were applied to baseline risk estimates to model the transition from the
established CVD to the cardiovascular event health states at the end of each model cycle (each
year). Efficacy estimates for each intervention are detailed in Tables 4.5 and 4.6. The base case
assumed efficacy estimates from time to first event. Alternatives such as the relative risk of
total events and the inclusion of other cardiovascular events (revascularization and unstable
angina) were evaluated in a scenario analysis.

Table 4.5. Efficacy Estimates for Rivaroxaban
Rivaroxaban + ASA ASA Alone
Parameter HR (95% CI) P-Value Source
n (%) n (%)
Composite Primary
0.76 (0.66-
Outcome: Stroke, CV 379 (4.1) 496 (5.4) <0.001
0.86)
Death, MI*
0.58 (0.44-
Stroke† 83 (0.9) 142 (1.6) <0.001 83
0.76)
0.78 (0.64-
CV Death† 160 (1.7) 203 (2.2) 0.02
0.96)
0.86 (0.70-
MI† 178 (1.9) 205 (2.2) 0.14
1.05)
0.54 (0.35- 57
MALE‡ 30 (1) 56 (2) 0.0054
0.84)
ASA: aspirin, CI: confidence interval, CV: cardiovascular, HR: hazard ratio, MALE: major adverse limb event, MI:
myocardial infarction
*Only p-values for the primary outcome are confirmatory.
†As the statistical analysis plan for the trial did not specify modifications to the pre-specified control of
multiple testing of other efficacy outcomes in the case of early termination of the study, any HRs,
corresponding CIs, and P-values reported for other efficacy outcomes cannot be interpreted as statistically
significant.
‡MALE was defined as acute or chronic limb ischemia and included all major amputations. MALE was a pre-
specified outcome for patients with PAD in the COMPASS trial.
Table 4.6. Efficacy Estimates for Icosapent Ethyl

Comparator/
Icosapent Ethyl
Parameter Placebo HR (95% CI) P-Value Source
n (%)
n (%)
Composite Outcome:
0.74 (0.65-
CV Death, Nonfatal 459 (11.2) 606 (14.8) <0.001
0.83)
Stroke, Nonfatal MI
0.73 (0.65-
Secondary Prevention 559 (19.3) 738 (25.5)
0.81)
0.88 (0.70-
Primary Prevention 146 (12.2) 163 (13.6) 19
1.10)
0.71 (0.54-
Non-Fatal Stroke 85 (2.1) 118 (2.9) 0.01
0.94)
0.80 (0.66-
CV Death 174 (4.3) 213 (5.2) 0.03
0.98)
0.70 (0.59-
Non-Fatal MI 237 (5.8) 332 (8.1) <0.001
0.82)
Total Events (Primary 0.70 (0.62- 20
1076 1546 <0.0001
Composite Endpoint) 0.78)
CI: confidence interval, CV: cardiovascular, HR: hazard ratio, MI: myocardial infarction

Discontinuation
Treatment discontinuation rates were based on trial-specific data for each comparison. For
rivaroxaban, 16.5% of patients in the rivaroxaban + ASA arm had permanently discontinued
treatment at the final study visit (mean follow-up duration of 23 months).83
For icosapent ethyl, after an average follow-up of approximately two years, 18.7% of patients in
the icosapent ethyl arm had discontinued treatment at the time of a first event.20
Beyond two years duration in the model for rivaroxaban and five years duration for icosapent
ethyl, we assumed an annualized discontinuation based on SAE-related discontinuation of 2.7%
for rivaroxaban and 2.2% for icosapent ethyl. The observed trial-based hazard ratios were
assigned for all patients in the first two or five years of the model (no matter the discontinuation
status, i.e., consistent with an intention to treat analysis) for rivaroxaban and icosapent ethyl,
respectively. For model cycles beyond two years (rivaroxaban) and five years (icosapent ethyl),
the proportion remaining on active treatment were assigned the observed trial-based hazard
ratios whereas the proportion who discontinued treatment were assigned the costs and
outcomes consistent with optimal medical management.
Utilities
To adjust for quality of life, health state utilities were derived from publicly available literature
and applied to health states. Utility values were primarily from a study on preference-based EQ-
5D index scores for chronic conditions, based on survey results for a nationally representative
sample of the US adult population.94 We used consistent health state utility values across both
comparisons. We assigned higher disutilities to MI and stroke events compared to the post-
event states, consistent with prior studies (Table 4.7). Disutilities for AEs were applied to the
proportion of the cohort with an event within each cycle.
Since the severity of stroke can differ with patients experiencing a wide range of symptoms and
disability, we compared the disutility values for stroke listed in Table 4.7 to a weighted average
stroke utility value that incorporates stroke severity. The Modified Rankin Scale (mRS) is a
commonly used clinical outcome measure to classify strokes based on symptom severity, with
severity ranging from 0 (no symptoms) to 6 (dead). The COMPASS trial classified the severity of
all stroke events using the mRS, and the associated utilities for each stroke category have been
published in the literature95 (Appendix Table E5). We weighted the utility values for mRS
categories by the proportion of patients in the COMPASS control arm who experienced a stroke
in each mRS category to estimate a weighted average stroke utility (Appendix Table E5). This
weighted average utility value is 0.6615, which is higher than the utility value applied to a
patient in our model cohort who experiences an event cycle stroke (0.5976). While the severity
differed for COMPASS patients who experienced a stroke, the risk reduction for stroke with
rivaroxaban + ASA versus ASA alone was consistent across all mRS categories.14

Table 4.7. Utility and Disutility Values for Health States
Parameter Value Source
96,97
Treated Population without Observed Events 0.854*
Post-Event MI (Disutility Applied to State) -0.150
Post-Event Stroke (Disutility Applied to State) -0.204 94
Event Cycle MI (Disutilities Applied to Event) -0.0409 + -0.150
Event Cycle Stroke (Disutilities Applied to Event) -0.0524 + -0.204
98
Severe Atrial Fibrillation (Disutility Applied to Event) -0.164
99
Major Bleeding (Disutility Applied to Event) -0.181
78
Acute Non-Fatal MALE (Disutility Applied to Event) -0.220
MALE: major adverse limb event, MI: myocardial infarction
*Based on average utilities of coronary heart disease patients who had undergone coronary artery bypass
grafting (CABG) and PCI and later stabilized. (CABG=0.847, PCI=0.861).
Adverse Events
The model included all reported treatment-related SAEs and bleeding events for each of the two
comparisons. Each SAE had an associated cost and disutility (if available) that was applied for
each occurrence of the event. Inputs related to SAEs for each intervention are detailed in Tables
4.8 and 4.9.
For the rivaroxaban comparison, major bleeding events occurred more frequently in patients in
the rivaroxaban + ASA group than in the ASA alone group (3.1% vs. 1.9%). The annualized
probability for each treatment arm was used within the model.83
Table 4.8. Bleeding Event Parameter for Rivaroxaban Evaluation

Rivaroxaban +
Parameter ASA Alone HR (95% CI) P-Value Source
ASA
Modified ISTH Major 1.70 (1.40- 83
288 (3.1) 170 (1.9) <0.001
Bleeding* 2.05)
ASA: aspirin, CI: confidence interval, HR: hazard ratio, ISTH: International Society on Thrombosis and
Haemostasis
*Modified ISTH bleeding was defined as a composite of fatal bleeding, symptomatic bleeding into a critical
organ, bleeding into a surgical site requiring reoperation, and bleeding that led to hospitalization (including
presentation to an acute care facility without an overnight stay).
For the icosapent ethyl comparison, overall AE rates were low in both treatment arms and none
of the AEs were fatal. There was an observed trend toward increased serious bleeding in the
icosapent ethyl arm. The evidence on icosapent ethyl did not suggest significant increases in
serious central nervous system bleeding, GI bleeding, or adjudicated hemorrhagic stroke and
therefore no difference in major bleeding was assumed within the model. There was a small,
statistically significant increase in hospitalization for atrial fibrillation or flutter endpoints that
was included in the model.20

Table 4.9. Adverse Event Parameters for Icosapent Ethyl Evaluation
Parameter Icosapent Ethyl Comparator/Placebo P-Value Source
Serious TEAE (%) 30.6 30.7 0.98
Hospitalization for Atrial
3.1 2.1 0.004 20
Fibrillation or Flutter (%)
Bleeding-Related Disorders
2.7 2.1 0.06
(%)
TEAE: treatment-emergent adverse event
Drug Utilization
The following inputs were used to model drug utilization and associated costs for each
intervention:
• Duration of treatment
• Schedule of doses for each drug
Table 4.10. Treatment Regimens

Characteristic Rivaroxaban + ASA ASA Alone Icosapent Ethyl
Rivaroxaban 2.5 mg twice
ASA 100 mg once
Recommended Dose daily + ASA 100 mg once 2 g twice daily
daily
daily
Route of
Oral Oral Oral, with food
Administration
ASA: aspirin
Economic Inputs
Drug Acquisition Costs
For both drugs, we obtained net pricing estimates from SSR Health, LLC,32 which combine data
on unit sales with publicly-disclosed US sales figures that are net of discounts, rebates,
concessions to wholesalers and distributors, and patient assistance programs to derive a net
price. We estimated net prices by comparing the four-quarter averages (i.e., 2nd quarter of
2018 through 1st quarter of 2019) of both net prices and wholesale acquisition cost (WAC) per
unit to calculate a mean discount from WAC for the drug. Finally, we applied this average
discount to the most recently available WAC (July 1, 2019) to arrive at an estimated net price
per unit.
Rivaroxaban’s WAC is $7.47 per 2.5 mg tablet. The average discount from WAC was 59.41% for
rivaroxaban, leading to an estimated net price of $3.03 per dose.
Icosapent ethyl’s WAC for a one-month supply of 4 g/day is $303.65 (each bottle contains 120
each of 1 g capsules). The average discount from WAC was 56.04% for icosapent ethyl, leading
to an estimated net price of $1.11 per dose.

Table 4.11. Drug Cost Inputs
WAC Net Price
Drug per per Discount from WAC Net Price per Year
Tablet/Capsule Tablet/Capsule
Rivaroxaban $7.47 per
$3.03 59.41% $2,215
(Xarelto®, Janssen) 2.5mg tablet
Icosapent Ethyl
$2.53 per 1g
(Vascepa®, Amarin $1.11 56.04% $1,625
capsule
Pharma)
WAC: wholesale acquisition cost
WAC per Redbook®; net pricing estimates from SSR Health.32,33
Please refer to the ICER Reference Case for more details on drug pricing.
Non-Drug Costs
Health state costs were derived from literature-based estimates. Indirect costs were not
included in the base-case analysis but were included in a scenario analysis. All costs were
inflated to year 2019 levels using the health care component of the personal consumption
expenditure index,100 in accordance with the ICER Reference Case.101
Table 4.12. Non-Drug Cost Inputs

Input 2019 USD Mean Value* Source
MI Treatment and Event Year Cost $55,316
Stroke Treatment and Event Year Cost $58,932 102
and supporting
Post-MI Annual Cost (Assumed Same as Subsequent
$2,728 references
Years of Coronary Heart Disease)
Post-Stroke Annual Cost $5,742
103
CV Death Cost $18,341
Major Bleeding Cost (Applied to Event Year) $3,367 78
Acute Non-Fatal MALE Cost (Applied to Event) $17,979
Hospitalization for Atrial Fibrillation 104
$9,957
CV: cardiovascular, MALE: major adverse limb event, MI: myocardial infarction, USD: United States
dollar
*Estimates varied in sensitivity analyses using the 2.5th and 97.5th percentiles of evidence-based
probability distributions where available.
Sensitivity Analyses
We ran one-way sensitivity analyses to identify the key drivers of model outcomes, using
available measures of parameter uncertainty (i.e., standard errors) or reasonable ranges for
each input described in the model inputs section above. Probabilistic sensitivity analyses were
also performed by jointly varying all model parameters over 5,000 simulations, then calculating
95% credible range estimates for each model outcome based on the results. Additionally, we
performed a threshold analysis by systematically altering the price of rivaroxaban and icosapent

ethyl to estimate the maximum prices that would correspond to willingness to pay (WTP)
thresholds of $50,000/QALY, $100,000/QALY, and $150,000/QALY.
Scenario Analyses
Multiple scenario analyses were conducted to evaluate the impact of key model choices and
assumptions on the robustness of the results and conclusions. First, the perspective was
expanded to a modified/restricted societal one that included productivity losses. Evidence
suggests that workers have workplace absenteeism and short-term disability equal to 13.6
hours per month within the first year after a cardiovascular event, but no differences beyond
the first year.105 We assumed these first-year annualized hours, 163.2 hours, would apply to all
individuals in the model who experienced a cardiovascular event.106 The average hourly wage of
$27.71 was assumed to apply to all hours no matter the working status of the individual.
Second, the addition of other cardiovascular events, such as revascularization and unstable
angina, were included in the cardiovascular event health state within the model. Coronary
revascularization and unstable angina requiring hospitalization were pre-specified
subcomponents of the primary composite endpoint in the icosapent ethyl trial. Third, the
composite primary endpoints and corresponding hazard ratios from the intervention-specific
trials were modeled instead of the individual subcomponents of the primary endpoints. Fourth,
DAPT with clopidogrel was modeled as a comparator to rivaroxaban + ASA (Appendix Table E2-
E4). Due to differences in severe bleeds, the clinical review was not able to produce a hazard
ratio that indirectly compared the rates of major bleeding with DAPT with clopidogrel to
rivaroxaban + ASA. The annualized bleeding rate for the ASA alone arm of the “CAPRIE-like
subgroup” (documented prior MI, ischemic stroke, or symptomatic PAD) from CHARISMA was
0.0065.18 Rate ratios between 1 and 5 (for rivaroxaban + ASA versus DAPT with clopidogrel)
were modeled and compared versus commonly cited thresholds. Finally, we estimated the
potential cost effectiveness of the interventions versus their respective optimal medical
management comparator by assuming the same baseline cardiovascular risk through averaging
the baseline risk across the two interventions’ trial populations, but assuming the same
intervention-specific hazard ratios.
Model Validation
We used several approaches to validate the model. First, we provided preliminary methods and
results to manufacturers, patient groups, and clinical experts. Based on feedback from these
groups, we refined data inputs used in the model. Second, we varied model input parameters
to evaluate face validity of changes in results. We performed model verification for model
calculations using internal reviewers. As part of ICER’s efforts to increase modeling
transparency, we will also share the model with relevant manufacturers for external verification
and feedback shortly after publishing the draft report for this review. Finally, we compared
results to other cost-effectiveness models in this therapy area.

4.3 Results
Base-Case Results
The base-case lifetime undiscounted clinical events are found in Tables 4.13 and 4.14 for
rivaroxaban and icosapent ethyl, respectively. Over a lifetime horizon, we present the
percentage of the modeled cohort who experienced their first modeled event as an MI or as a
stroke. Separately, we present the proportion of the modeled cohort who experienced death
due to CVD or due to all other causes. Further, we present the cumulative incidence of
cardiovascular events (MI, stroke, and cardiovascular death). The column percentage sum of
death due to CVD and all other causes approximates to 100 but may differ slightly due to
rounding. For both rivaroxaban and icosapent ethyl, lower lifetime first MI, first stroke, and
death due to CVD were observed. The major adverse cardiovascular event cumulative incidence
including MI, stroke, and cardiovascular-related death was lower for both interventions as
compared to optimal medical management alone. Note that subsequent events were also
tracked within the model alongside associated costs and disutilities but are not reported here.
Table 4.13. Rivaroxaban Long-Run Clinical Outcomes (Lifetime Horizon, Undiscounted)

Lifetime Events
Absolute
Event Intervention Medical Management
Difference
First Event MI 20% 21% -1%
First Event Stroke 10% 14% -4%
Death (CV) 30% 35% -5%
Cumulative CV events (MI, Stroke, and CV Death) 61% 72% -11%
Death (All Other Cause) 70% 64% N/A
CV: cardiovascular, MI: myocardial infarction
Table 4.14. Icosapent Ethyl Long-Run Clinical Outcomes (Lifetime Horizon, Undiscounted)
Lifetime Events
Absolute
Event Intervention Medical Management
Difference
First Event MI 29% 35% -6%
First Event Stroke 9% 11% -2%
Death (CV) 38% 46% -8%
Cumulative CV events (MI, Stroke, and CV Death) 81% 98% -17%
Death (All Other Cause) 61% 54% N/A
Base-case discounted costs and outcomes from the model are listed in Table 4.15 for
rivaroxaban and in Table 4.16 for icosapent ethyl. Rivaroxaban was associated with
approximately $17,000 in discounted lifetime intervention costs, whereas icosapent ethyl was
associated with $15,000 in lifetime intervention costs. Although additional medical
management intervention and non-intervention costs are likely within the modeled cohorts,

such costs were assumed to be equal across treatment arms, and therefore not included within
this analysis. Average discounted life years and QALYs were higher for both interventions as
compared to optimal medical management alone. The items included within the health care
sector base-case results as well as those included within the modified societal perspective are
listed in the impact inventory (Appendix Table E1).
Table 4.15. Base-Case Discounted Results for Rivaroxaban Compared to Optimal Medical
Management including ASA
Non-
Intervention
Costs
Costs
Rivaroxaban $17,000 $20,000 $38,000 10.86 9.07 9.06
Medical
$200 $24,000 $24,000 10.45 8.69 8.69
Management
Table 4.16. Base-Case Discounted Results for Icosapent Ethyl Compared to Optimal Medical
Management including Statins
Non-
Intervention
Costs
Costs
Icosapent
$15,000 $25,000 $40,000 12.26 10.21 10.19
Ethyl
Medical
$800 $30,000 $31,000 11.73 9.69 9.69
Management
Base-case discounted incremental results are shown in Table 4.17, with rivaroxaban versus
optimal medical management yielding $36,000 per QALY gained. Icosapent ethyl versus optimal
medical management yields $18,000 per QALY gained. Discounted incremental life year results
were slightly lower than the incremental cost-per-QALY findings. Discounted incremental equal
value life years gained (evLYG) evenly measures any gains in length of life, regardless of the
impact on patients’ quality of life. Results for the incremental evLYG were slightly lower than
the cost-per-QALY findings given there is a life extension to each therapy over medical
management. Incremental cost per major adverse cardiovascular event avoided could be
interpreted as the expected costs to achieve one less major adverse cardiovascular event (MI,
stroke, or cardiovascular death) when treating a population with rivaroxaban versus optimal
medical management or separately, when treating a population with icosapent ethyl versus
optimal medical management. The incremental cost per major adverse cardiovascular event
avoided should be interpreted with caution, given that this metric does not have a known
threshold for an understanding of value and does not include the differential timing or the
differential importance of major adverse cardiovascular events. Note that the intervention-
specific incremental findings are modeled using intervention-specific populations and therefore
should not be directly compared across treatments.

Table 4.17. Base-Case Incremental Results
Cost per
Incr. Incr. Incr. Incr. Cost Cost per Cost per
Intervention* MACE
Costs LYs evLYG QALYs per LY evLYG QALY
Avoided
Rivaroxaban $32,000 $35,000 $36,000 $120,000

vs. Medical $13,000 0.41 0.38 0.37 per LY per evLYG per QALY per MACE
Management gained gained gained avoided
Icosapent
$17,000 $17,000 $18,000 $53,000
Ethyl vs.
$9,000 0.54 0.52 0.50 per LY per evLYG per QALY per MACE
Medical
gained gained gained avoided
Management
ICER: incremental cost-effectiveness ratio, Incr.: Incremental, LY: life year, MACE: major cardiovascular event,
QALY: quality adjusted life year
Sensitivity Analysis Results
To demonstrate effects of uncertainty on both costs and health outcomes, we varied input
parameters using available measures of parameter uncertainty (i.e., standard errors) or
reasonable ranges to evaluate changes in cost per additional QALY. Key drivers of uncertainty
for rivaroxaban versus optimal medical management included the clinical event hazard ratios for
MI, stroke, and cardiovascular death, with smaller impacts observed from uncertainty in utility
and cost inputs (Figure 4.2). Similarly, key drivers of uncertainty for icosapent ethyl versus
optimal medical management included the clinical event hazard ratios for MI, stroke, and
cardiovascular death, with smaller impacts observed from uncertainty in utility and cost inputs
(Figure 4.3).

Figure 4.2. Tornado Diagram for One-Way Sensitivity Analyses of Rivaroxaban versus Medical
Management (Accompanied Values Below)
Input Name Lower Input ICER Upper Input ICER Lower Input Upper Input
Relative Risk – Non-Fatal MI $23,914 $52,746 0.70 1.05
Relative Risk – Non-Fatal
$26,651 $45,463 0.44 0.76
Stroke
Relative Risk – CV Death $29,304 $40,500 0.64 0.96
Health Utility – Treated
Population without $34,981 $31,854 0.81 0.89
Observed Events
Health State Cost – Stroke $34,698 $31,582 $42,894 $77,477
Health State Cost – CV Death $33,937 $32,461 $13,350 $24,113
Health State Cost – MI $33,647 $32,797 $40,262 $72,724
Health State Cost – Post MI $33,131 $33,394 $1,986 $3,586
Health State Cost – Post
$33,349 $33,142 $4,179 $7,549
Stroke
Disutility – Post Event Stroke $33,302 $33,202 0.19 0.21
CV: cardiovascular, MI: myocardial infarction, ICER: incremental cost-effectiveness ratio

Figure 4.3. Tornado Diagram for One-Way Sensitivity Analyses of Icosapent Ethyl versus
Medical Management (Accompanied Values Below)
Lower Input Upper Input

Input Name Lower Input Upper Input
Incremental Result Incremental Result
Hazard Ratio – Non-Fatal MI $12,183 $27,305 0.59 0.82
Hazard Ratio – Non-Fatal
$14,223 $24,083 0.54 0.94
Stroke
Hazard Ratio – CV Death $16,232 $20,434 0.66 0.98
Health State Cost – MI $19,383 $15,918 $40,262 $72,723
Health Utility – Treated
Population without $18,674 $17,048 0.81 0.89
Observed Events
Health State Cost – Stroke $18,314 $17,155 $42,894 $77,477
Health State Cost – CV Death $18,292 $17,181 $13,350 $24,113
Health State Cost – Post MI $17,878 $17,658 $1,986 $3,586
Health State Cost – Post
$17,839 $17,704 $4,179 $7,549
Stroke
Disutility – Post Event MI $17,808 $17,744 0.14 0.16
Ninety two percent of iterations suggested that rivaroxaban met the $50,000/QALY threshold.
Icosapent ethyl results suggested that nearly 100% of iterations met the $50,000/QALY
threshold (Table 4.18). Both interventions achieved 100% of iterations that met the
$100,000/QALY and $150,000/QALY thresholds.

Table 4.18. Probabilistic Sensitivity Analysis Results
Cost-Effective at Cost-Effective at Cost-Effective at
Comparison*
$50,000 per QALY $100,000 per QALY $150,000 per QALY
Rivaroxaban vs. Medical
92% 100% 100%
Management
Icosapent Ethyl vs.
100% 100% 100%
Medical Management
QALY: quality-adjusted life year
Scenario Analyses Results
Modified Societal Perspective
We conducted a scenario analysis by assigning an annualized productivity-related cost of $4,522

to each cardiovascular event in the model. This scenario did not discriminate against those who
were not working as it assigned the same cost to all individuals who experienced an event. The
rounded incremental results did not materially differ from the base-case findings (Table 4.19).
Table 4.19. Incremental Results for Modified Societal Perspective

Incremental Incremental Incremental CE Ratio per
Comparison* CE Ratio per QALY
Costs LYs QALYs LY
Rivaroxaban vs.
Medical $13,000 0.41 0.37 $32,000 per LY $35,000 per QALY
Management
Icosapent Ethyl
vs. Medical $9,000 0.54 0.50 $16,000 per LY $17,000 per QALY
Management
CE: cost effectiveness, LY: life year, QALY: quality-adjusted life year
Other Cardiovascular Events Health State
A scenario analysis was conducted with the addition of other cardiovascular events such as
revascularization and unstable angina included in the cardiovascular event health state within
the icosapent ethyl evaluation. This broader inclusion of other cardiovascular events shifted the
incremental cost per QALY from $18,000/QALY (base-case) to $16,000/QALY (Table 4.20).

Table 4.20. Incremental Results for Other Cardiovascular Events Health State
Incremental Incremental Incremental CE Ratio per CE Ratio per
Comparison
Costs LYs QALYs LY QALY
Icosapent Ethyl vs.
$14,000 per $16,000 per
Medical $8,000 0.56 0.50
LY QALY
Management
Composite Primary Endpoints
A scenario analysis was conducted using the composite primary endpoints instead of individual
primary endpoints from the intervention-specific trials. This scenario yielded higher but similar
incremental cost-per-QALY findings (Table 4.21).
Table 4.21. Incremental Results for Composite Primary Endpoints

Comparison*
Rivaroxaban vs.
$34,000 per $38,000 per
Medical $14,000 0.41 0.36
LY QALY
Management
Icosapent Ethyl vs.
$19,000 per $21,000 per
Medical $10,000 0.52 0.47
LY QALY
Management
DAPT with Clopidogrel Comparator for Rivaroxaban
A scenario analysis was conducted comparing rivaroxaban + ASA to DAPT with clopidogrel +
ASA. We present the incremental findings assuming the ASA annualized major bleed rate of
0.0065 (Table 4.22) for both rivaroxaban + ASA as well as for DAPT with clopidogrel (i.e.,
assuming this rate did not differ between treatment strategies). Further, we present the
incremental findings assuming the ASA annualized major bleed rate of 0.0065 for DAPT with
clopidogrel and solved for the rate ratio that would generate an incremental cost-effectiveness
ratio per QALY of $100,000 (rivaroxaban + ASA vs. DAPT with clopidogrel). The rate ratio for
major bleed was 2.83 (rivaroxaban + ASA vs. DAPT with clopidogrel) to generate an incremental
cost-effectiveness ratio of $100,000 per QALY (results not shown).
Table 4.22. Incremental Results for DAPT with Clopidogrel Comparator

Comparison
Rivaroxaban vs.
$89,000 per $86,000 per
DAPT with $14,000 0.16 0.16
LY QALY
Clopidogrel
CE: cost effectiveness, DAPT: dual antiplatelet therapy, LY: life year, QALY: quality-adjusted life year

Same Baseline Cardiovascular Risk Assumption
We estimated the potential cost effectiveness of the interventions versus their respective
optimal medical management comparator by assuming the same baseline cardiovascular risk by
averaging the baseline risk across the two interventions’ trial populations but assuming the
base-case intervention-specific hazard ratios. The incremental results for both treatments move
toward one another with rivaroxaban’s incremental cost-per-QALY of $33,000/QALY and
icosapent ethyl’s incremental cost-per-QALY of $19,000/QALY. However, given the lack of
formal direct or indirect treatment comparisons, the results for the interventions may not be
directly comparable.
Table 4.23. Incremental Results for Same Baseline Cardiovascular Risk

Comparison
Rivaroxaban vs.
$30,000 per $33,000 per
Medical $13,000 0.43 0.40
LY QALY
Management
Icosapent Ethyl vs.
$18,000 per $19,000 per
Medical $9,000 0.51 0.48
LY QALY
Management
Threshold Analyses Results
We estimated threshold treatment prices that would reflect an incremental cost-per-QALY of

$50,000, $100,000, and $150,000. The findings suggest that both treatments’ net prices were
below the price needed to achieve $50,000 per QALY (Table 4.24). Interestingly, the annual
price to achieve $100,000/QALY was slightly lower but approximated the WAC annualized price
for rivaroxaban. The annual price to achieve $50,000/QALY was slightly lower but approximated
the WAC annualized price for icosapent ethyl.
Table 4.24. Cost per QALY Threshold Analysis Results

Net Price
Price to Price to Price to
WAC per Annual per Net Price
Achieve Achieve Achieve
Tablet/Capsule WAC Tablet/ per Year
$50,000 $100,000 $150,000
Capsule
per QALY per QALY per QALY
$7.47 per 2.5
Rivaroxaban $5,457 $3.03 $2,215 $2,849 $5,223 $7,597
mg tablet
$3,699 $1.11 $1,625 $3,433 $6,282 $9,204
Ethyl capsule
QALY: quality-adjusted life year, WAC: wholesale acquisition cost
Table 4.25 presents the threshold results for each drug at thresholds of $50,000, $100,00 and
$150,000 per equal value life year gained (evLYG). An analysis of the evLYG is included to

complement the cost per QALY calculations and provide policymakers with a broader view of
cost effectiveness. The cost per evLYG considers any extension of life at the same “weight” no
matter what treatment is being evaluated.
Table 4.25. Cost per evLYG Threshold Analysis Results

Net Price Price to Price to Price to
WAC per Annual per Net Price Achieve Achieve Achieve
Tablet/Capsule WAC Tablet/ per Year $50,000 $100,000 $150,000
Capsule per per per
evLYG evLYG evLYG
$7.47 per 2.5
Rivaroxaban $5,457 $3.03 $2,215 $2,922 $5,369 $7,780
mg tablet
$3,699 $1.11 $1,625 $3,506 $6,501 $9,423
Ethyl capsule
evLYG: equal value life year gained, WAC: wholesale acquisition cost
Model Validation
Model validation followed standard practices in the field. We tested all mathematical functions
in the model to ensure they were consistent with the report (and supplemental Appendix
materials). We also conducted sensitivity analyses with null input values to ensure the model
was producing findings consistent with expectations. Further, independent modelers tested the
mathematical functions in the model as well as the specific inputs and corresponding outputs.
We reported on the model calibration findings to suggest that over a time horizon consistent
with the randomized controlled trials, the model produced first event MI, first event stroke, and
cardiovascular death rates within small margins of error of the randomized controlled trial
findings (Appendix Table E7-Table E8).
Model validation was also conducted in terms of comparisons to other model findings. We
searched the literature to identify models that were similar to our analysis, with comparable
populations, settings, perspective, and treatments. Fonarow et al. evaluated the cost
effectiveness of evolocumab in patients with atherosclerotic CVD when added to standard
background therapy.107 They used a Markov cohort state-transition model from a US societal
perspective assuming a lifetime time horizon to capture the progression of atherosclerotic CVD
in adults. We used inputs from this study in our model to validate against the Fonarow model
findings and achieved agreeable estimates (Appendix Table E6).
Prior Economic Models
In our review of prior economic models, Fonarow et al. evaluated the cost effectiveness of the
proprotein convertase subtilisin/kexin type 9 inhibitor evolocumab in patients with
atherosclerotic cardiovascular disease when added to standard background therapy.107 A
Markov cohort state-transition model was used, integrating US population-specific
demographics, risk factors, background therapy, and event rates, along with trial-based event
risk reduction. In the base case, using patients with atherosclerotic cardiovascular disease,
evolocumab had an ICER of approximately $269,000 per QALY gained (approximately $166,000
using a discounted price based on an average rebate of 29% for branded pharmaceuticals).107
Sensitivity and scenario analyses demonstrated ICERs ranging from approximately $100,000 to
$489,000 per QALY, with an ICER of approximately $414,000 per QALY when using trial patient
characteristics. At its list price of $14,523, the addition of evolocumab to standard background
therapy in patients with atherosclerotic cardiovascular disease exceeded generally accepted
cost-effectiveness thresholds.107
A study by Kazi et al. estimated the cost effectiveness of PCSK9 inhibitors in patients with
heterozygous familial hypercholesterolemia (FH) or atherosclerotic cardiovascular disease
(ASCVD).102 A simulation model of US adults aged 35 to 94 years, called the Cardiovascular
Disease Policy Model, was used to evaluate the cost effectiveness of PCSK9 inhibitors or
ezetimibe in heterozygous FH or ASCVD.102 Adding PCSK9 inhibitors to statins in heterozygous
FH was estimated to prevent 316,300 MACE at a cost of $503,000 per QALY gained compared
with adding ezetimibe to statins (80% uncertainty interval [UI], $493,000-$1,737,000).102 In
ASCVD, adding PCSK9 inhibitors to statins was estimated to prevent 4.3 million MACE compared
with adding ezetimibe, at $414,000 per QALY (80% UI, $277,000-$1,539,000).102 In this study,
PCSK9 inhibitor use in patients with heterozygous FH or ASCVD did not meet generally
acceptable incremental cost-effectiveness thresholds.102
Two studies assessed the cost effectiveness of rivaroxaban and aspirin compared to aspirin
alone in patients with stable cardiovascular disease and in patients with peripheral or carotid
artery disease from the Australian perspective.78,79 Ademi et al. developed a Markov model
using input data from the COMPASS trial. Compared to aspirin alone, rivaroxaban plus aspirin
was estimated to have an ICER of AUD$31,436/QALY gained.79 The authors concluded that
compared to aspirin, rivaroxaban in combination with aspirin is likely to be cost-effective in
preventing recurrent cardiovascular events in patients with stable atherosclerotic vascular
disease.79 Similarly, Zomer et al. estimated that rivaroxaban plus aspirin therapy prevented 143
non-fatal cardiovascular disease events, 118 major adverse limb events and 10 deaths
compared to aspirin therapy alone, resulting in an ICER of AUD$26,769 per QALY. 78 The authors
concluded that in patients with peripheral artery disease or carotid artery disease, rivaroxaban
plus aspirin therapy was cost-effective in the prevention of recurrent cardiovascular disease
compared to aspirin therapy alone.78
Other published economic evaluations included the assessment of eicosapentaenoic acid (EPA)
in CVD prevention in the settings of the US and Japan.80,81 The results of the US model showed
that combining EPA with statin therapy for secondary prevention of cardiovascular disease may
be a cost-saving option compared to statin monotherapy.81 In the Japanese assessment, EPA
plus statin combination therapy showed acceptable cost effectiveness for secondary prevention,
but not primary prevention, of CVD in patients with hypercholesterolemia.80

Lastly, Jiang et al. assessed the cost effectiveness of 30-month vs 12-month dual antiplatelet
therapy (DAPT) with clopidogrel and aspirin after drug-eluting stents in patients with acute
coronary syndrome.77 Continuation of DAPT appears to be cost-effective in acute coronary
syndrome patients who were event-free for 12-month DAPT after drug-eluting stents. However,
the cost effectiveness of DAPT for 30 months was highly subject to the odds ratios of nonfatal
stroke and death.
4.4 Summary and Comment

Our base-case results suggest that the use of rivaroxaban (plus ASA) and icosapent ethyl (in
patients receiving statins) both provide clinical benefit in terms of gains in quality-adjusted
survival and overall survival compared to optimal medical management alone in the adult,
established CVD cohort, and in the case of icosapent ethyl also for adults without known CVD
but at high risk for cardiovascular events. This translated into incremental cost-effectiveness
estimates that fell below commonly cited cost-effectiveness thresholds under the assumptions
used in this analysis. The results were relatively robust to parameter uncertainties in the one-
way and probabilistic sensitivity analyses. Further, the results were robust to a number of
scenario analyses including the modified societal perspective and others. When comparing
rivaroxaban + ASA to generic clopidogrel + ASA (DAPT), the incremental results were between
$85,000 and $100,000 per QALY assuming plausible (but unknown) relative risks of major
bleeding.
Limitations
Our analyses had important limitations and assumptions. We assumed three-component major
adverse cardiovascular events, MI, stroke, and cardiovascular death, to form the base-case
health states within the model structure for both rivaroxaban and for icosapent ethyl.
Acknowledging that other potential cardiovascular events are problematic to model due to
potential double-counting and competing events, we did not cost out or apply disutilities to any
other potential differences in cardiovascular events such as revascularization. This model
structure was consistent with prior ICER evaluations within established CVD.108 A scenario
analysis that broadened major adverse cardiovascular events to include other events suggested
similar but lower cost-effectiveness findings for icosapent ethyl. For all evaluated treatments,
the evidence was more uncertain when quantifying the cumulative incidence of subsequent
events as well as the relative impacts that the treatments had on subsequent events. Base-case
assumptions held treatment-specific first-event relative reductions constant for subsequent
events. Icosapent ethyl trial evidence suggested this assumption is appropriate or slightly
conservative for the purposes of estimating cost effectiveness. Patient-level projections of trial
evidence could aid in supporting or refuting our model assumptions surrounding ASA
cumulative incidence of subsequent events as well as the relative impact of treatment.
An additional limitation of this analysis was the model calibration to the observed clinical trial
event rates for optimal medical management. Many unknowns were a part of the model
calibration exercise. We opted to calibrate the model to first MI, first stroke and cardiovascular

death. Alternative calibration options were tested with similar incremental lifetime findings
(results not reported). Finally, it is important to note that randomized controlled trial findings
may not generalize or translate to observed signals within the real world (i.e., efficacy does not
equal effectiveness). Given that the cost-effectiveness findings relied on randomized controlled
trials for estimates of clinical benefit and harm, the findings should be interpreted with caution
when estimating whether these interventions would achieve similar value for money in actual
practice.
Conclusions
In conclusion, the findings of our analysis suggest that the additive CVD therapies of focus for
this review provide gains in quality-adjusted survival and overall survival over optimal medical
management. Assuming clinical signals within the trial hold for patients treated with these
interventions and current net prices, the base-case results suggest that costs for treatment with
either rivaroxaban or icosapent ethyl would fall below commonly cited thresholds for cost
effectiveness. The results were relatively robust to sensitivity and scenario analyses. When
comparing rivaroxaban + ASA to clopidogrel + ASA, the incremental results are more uncertain,
but generally fall below $100,000/QALY.

5. Potential Other Benefits and Contextual
Considerations
Our reviews seek to provide information on potential other benefits offered by the intervention
to the individual patient, caregivers, the delivery system, other patients, or the public that
would not have been considered as part of the evidence on comparative clinical effectiveness.
We also recognize that there may be broader contextual issues related to the severity of the
condition, whether other treatments are available, and ethical, legal, or other societal priorities
that influence the relative value of illnesses and interventions. These general elements are
listed in the table below, and the subsequent text provides detail about the elements that are
applicable to the comparisons of icosapent ethyl and rivaroxaban to optimal medical
management. We sought input from stakeholders, including individual patients, patient
advocacy organizations, clinicians, and manufacturers, to inform the contents of this section.
Each ICER review culminates in a public meeting of an independent voting Council of clinicians,
patients, and health services researchers. As part of their deliberations, Council members will
judge whether a treatment may substantially impact the considerations listed in Table 5.1. The
presence of substantial other benefits or contextual considerations may shift a council
member’s vote on an intervention’s long-term value for money to a different category than
would be indicated by the clinical evidence and cost-effectiveness analyses alone. For example,
a council member may initially consider a therapy with an incremental cost-effectiveness ratio
of $150,000 per QALY to represent low long-term value for money. However, the Council
member may vote for a higher value category if they consider the treatment to bring substantial
other benefits or contextual considerations. Conversely, disadvantages associated with a
treatment may lead a Council member to vote for a lower value category. A Council member
may also determine that there are no other benefits or contextual considerations substantial
enough to shift their vote. All factors that are considered in the voting process are outlined in
ICER’s value assessment framework. The content of these deliberations is described in the last
chapter of ICER’s Final Evidence Report, which is released after the public meeting.
This section, as well as the Council’s deliberation, provides stakeholders with information to
inform their decisions on a range of issues, including shared decision-making between patients
and clinicians, coverage policy development, and pricing negotiations.

Table 5.1. Potential Other Benefits or Contextual Considerations (Not Specific to Any Disease
or Therapy)
Potential Other Benefits
This intervention offers reduced complexity that will significantly improve patient outcomes.
This intervention will reduce important health disparities across racial, ethnic, gender, socio-economic, or
regional categories.
This intervention will significantly reduce caregiver or broader family burden.
This intervention offers a novel mechanism of action or approach that will allow successful treatment of
many patients for whom other available treatments have failed.
This intervention will have a significant impact on improving return to work and/or overall productivity.
Other important benefits or disadvantages that should have an important role in judgments of the value of
this intervention.
Potential Other Contextual Considerations
This intervention is intended for the care of individuals with a condition of particularly high severity in terms
of impact on length of life and/or quality of life.
This intervention is intended for the care of individuals with a condition that represents a particularly high
lifetime burden of illness.
This intervention is the first to offer any improvement for patients with this condition.
Compared to optimal medical management alone, there is significant uncertainty about the long-term risk
of serious side effects of this intervention.
Compared to optimal medical management alone, there is significant uncertainty about the magnitude or
durability of the long-term benefits of this intervention.
There are additional contextual considerations that should have an important role in judgments of the value
of this intervention.
5.1 Potential Other Benefits

The majority of patients in the target populations for icosapent ethyl and rivaroxaban (i.e.,
established CVD or patients with multiple risk factors for CVD) are older individuals already
taking multiple classes of medication for risk factor control. Accordingly, we see the potential
for increased treatment complexity and patient/family burden rather than reductions.
Because both agents represent new mechanisms of action, they represent important treatment
options that may be complementary to existing treatment mechanisms (e.g., PCSK9 inhibition),
and may offer benefit if adherence to existing treatment is sub-optimal. They also appear to
offer further risk reductions in patients already optimally managed on other medications,
however, an important consideration given the high residual risk of cardiovascular events in
those with established disease.
There are no expected benefits with regard to reduction in disparities, return to work, or other
aspects of disease management and treatment.

5.2 Contextual Considerations
As noted previously, CVD is both prevalent and associated with a high lifetime burden of illness.
In addition, both icosapent ethyl and rivaroxaban were studied in CVD populations at
particularly high risk of recurrent events, suggesting a significant level of unmet need. That
being said, utilization of these therapies may ultimately include lower-risk individuals, with risk-
benefit tradeoffs that are more uncertain.
While icosapent ethyl and rivaroxaban represent new treatment mechanisms for CVD, there are
multiple other options available, including antithrombotics and multiple classes of lipid-lowering
agents and antihypertensives, many of which are available in generic form. The incremental
benefit of adding either of these two treatments to current medical management relative to
adding another relatively new treatment such as a PCSK9 inhibitor is unclear, as are the
potential benefits of all of these agents in combination.
The Phase III COMPASS trial of rivaroxaban was stopped early due to evidence of clinical benefit,
so the evidence base is limited to a median of approximately two years of follow-up. While this
introduces a fair degree of uncertainty to the assessment of long-term clinical benefits,
uncertainty around long-term risks is likely greater. For example, guideline statements for long-
term use of dual antiplatelet therapy use cautionary language specifically because of the
relatively short-term nature of clinical trials and concerns regarding bleeding risks over the long-
term.109 In contrast, the Phase III REDUCE-IT trial of icosapent ethyl included a median of nearly
five years of follow-up. Residual uncertainty among the clinical community is primarily focused
on the poor previous track record of other omega-3-based products in reducing cardiovascular
risks.
The benefits of rivaroxaban were robust among most patient subgroups, except for the very
elderly (≥75 years) as well as those with PAD only and without baseline dyslipidemia, although
these were small subsets of the COMPASS trial.21 Similarly, icosapent ethyl’s effects were
relatively consistent across subgroups, although the 95% confidence interval around the hazard
ratio estimate did include 1.0 in primary-prevention patients, those receiving ezetimibe, and
patients with baseline triglycerides <150 mg/dL; again, the latter two subsets were quite small.9
As mentioned above, however, the major challenge in interpretation of the trial data is in
generalizability, as patients were at very high risk of recurrent CVD events and optimally
managed on current therapy, conditions unlikely to be replicated with real-world use.

6. Value-Based Price Benchmarks
Annual value-based price benchmarks (VBPBs) of rivaroxaban and icosapent ethyl are presented
in Table 6.1. The value-based benchmark price for a drug is defined as the price range that
would achieve incremental cost-effectiveness ratios between $100,000 and $150,000 per QALY
(or evLYG) gained.
For rivaroxaban, price changes of approximately 4% discount to 39% over the list price (WAC)
would be required to reach the $100,000 to $150,000 per QALY threshold prices, respectively.
For icosapent ethyl, prices approximately 70% to 149% above WAC would achieve $100,000 to
$150,000 per QALY threshold prices (Table 6.1).
Table 6.1. Value-Based Price Benchmarks for Rivaroxaban and Icosapent Ethyl
Change from WAC
Annual Price at Annual Price at
Annual WAC to Reach Threshold
$100,000 Threshold $150,000 Threshold
Prices
Rivaroxaban
Per QALY Gained $5,457 $5,223 $7,597 -4% to +39%
Per evLYG $5,369 $7,780 -2% to +43%
Icosapent Ethyl
Per QALY Gained $3,699 $6,282 $9,204 +70% to +149%
Per evLYG $6,501 $9,423 +76% to +155%
WAC: wholesale acquisition cost; evLYG: equal value life year gained; QALY: quality-adjusted life year
We are including results for price per evLYG to ensure that policymakers are aware of the
complementary information these results can provide to the cost per QALY findings. The annual
price at which rivaroxaban meets the $100,000 to $150,000 per evLYG range for use in these
patients is $5,369 to $7,780. For icosapent ethyl, the relevant cost per evLYG price range is
$6,501 to $9,423 for the $100,000 to $150,000 per evLYG thresholds. The cost per evLYG price
range is quite similar to the cost per QALY range for both rivaroxaban and icosapent ethyl.

7. Potential Budget Impact
7.1 Overview
We used the cost-effectiveness model to estimate the potential total budgetary impacts of
rivaroxaban and icosapent ethyl as additive therapies to optimal medical management in
patients (age ≥20) with established CVD, and in the case of icosapent ethyl, also in patients
without evidence of CVD but with diabetes (age ≥50) and at least one additional risk factor.
7.2 Methods
We used results from the same model employed for the cost-effectiveness analyses to estimate
total potential budget impact. Potential budget impact was defined as the total differential cost
of using each new technology for the eligible population in this indication, calculated as
differential health care costs (including drug costs) minus any offsets in these costs from
avoided treatments and averted health care events. We assumed that patients matching those
in the respective trials would be eligible for rivaroxaban and icosapent ethyl. All costs were
undiscounted and estimated over a five-year time horizon, given the potential for cost offsets to
accrue over time and to allow a more realistic impact on the number of patients treated with
these new therapies.
For rivaroxaban, to estimate the size of the eligible prevalent population, we used a baseline
estimate for CAD of 6.7% of the US population (age≥20) based on a recent update from the AHA
in conjunction with the National Institutes of Health and other agencies.2 We then applied this
estimate to the average of the projected 2019 to 2023 US population (age≥20) to derive the
eligible population over the next five years. This resulted in an eligible population size for
rivaroxaban of approximately 16,908,000 patients over five years, or an estimated 3,385,000
patients each year.
For icosapent ethyl, to estimate the secondary prevention group, we used the same baseline
estimate for CAD of 6.7% of the US population (age≥20).2 In addition, we accounted for the
prevalence of patients with stroke (2.5%).2 To estimate the primary prevention population with
diabetes and one additional risk factor, we used an estimate of diabetes mellitus prevalence in
the US population of 9.8%2 and assumed 87% of these patients would have an additional risk
factor, based on the estimated proportion of diabetes patients with metabolic syndrome. 110 We
applied these estimated proportions for patients in the secondary prevention group and for
patients of age ≥50 years in the primary prevention group to the projected average of the 2019
to 2023 US population to derive the estimated eligible population over the next five years. This
resulted in an eligible population size for icosapent ethyl of approximately 33,522,000 patients
over five years, or an estimated 6,704,000 patients each year.
ICER’s methods for estimating potential budget impact are described in detail elsewhere111 and
have been recently updated. ICER recalculates the potential budget impact threshold each

calendar year, using the most recent inputs available. In the recalculation of ICER’s potential
budget impact threshold for calendar year 2019, we extended the time period over which we
average the annual number of drugs approved by the FDA from two to five years to reduce
fluctuations in the threshold due to this variable. The intent of our revised approach to
budgetary impact is to document the percentage of patients who could be treated at selected
prices without crossing a budget impact threshold that is aligned with overall growth in the US
economy. For 2019-2020, the five-year annualized potential budget impact threshold that
should trigger policy actions to manage access and affordability is calculated to be
approximately $819 million per year for new drugs.
To estimate potential budget impact, we evaluate a new therapy that would take market share
from one or more existing therapies and calculate the blended budget impact associated with
displacing use of existing therapies with the new intervention. For this analysis, we evaluated
the potential budget impact of rivaroxaban and icosapent ethyl as additive therapies to optimal
medical management in patients with established CVD, and in the case of icosapent ethyl, also
in patients without evidence of CVD but with diabetes and at least one additional risk factor.
For each treatment, we assumed equal uptake over five years, with treatment duration ranging
from one year (for the year-five cohort) to five years (for the year-one cohort). In other words,
patients initiating therapy in year one would accrue all drug costs and cost offsets over the full
five years, but those initiating in other years would only accrue a proportional amount of the
five-year costs. Note that the purpose of these analyses is to estimate the potential budget
impact, not to predict actual uptake or expected budget impact.
7.3 Results
For rivaroxaban, per-patient budget impact calculations are based on the WAC price of $5,457
per year, the net price of $2,215 per year, and the prices to reach $150,000, $100,000, and
$50,000 per QALY thresholds (at $7,597, $5,223, and $2,849 per year, respectively).
The average five-year annualized per patient potential budgetary impact when using the WAC
price and the net price were approximately $3,800 and $1,450, respectively. The average five-
year annualized potential budgetary impact at the three cost-effectiveness threshold prices
ranged from approximately $5,900 per patient using the annual price to achieve $150,000 per
QALY to approximately $2,000 using the annual price to achieve a $50,000 per QALY cost-
effectiveness threshold.
As shown in Figure 7.1, the ICER budget impact threshold of $819 million would be crossed if
greater than approximately 2% of eligible patients were treated in a given year with rivaroxaban
at the WAC price. When using the net price, the threshold would be reached when
approximately 6% of eligible patients were treated in a given year with rivaroxaban.
Approximately 1%, 2% and 4% of patients could be treated in a given year without crossing the
budget impact threshold at the $150,000, $100,000, and $50,000 per QALY threshold prices.
Although the per patient budget impact is relatively small, due to the very large population of

patients potentially eligible for treatment, the threshold is reached even at prices that would
meet standard cost-effectiveness thresholds.
Figure 7.1. Potential Budget Impact of Rivaroxaban at Various Prices
$8,000
$150,000 per QALY
$7,000
$6,000
WAC price
$5,000 $100,000 per QALY
Annual Price
$4,000
$3,000 $50,000 per QALY
Net price
$2,000
$1,000
$0
0% 20% 40% 60% 80% 100%
For icosapent ethyl, per-patient budget impact calculations are based on the WAC price of
$3,699 per year, the net price of $1,625 per year, and the prices to reach $150,000, $100,000,
and $50,000 per QALY thresholds (at $9,204, $6,282, and $3,433 per year, respectively).
The average five-year annualized per patient potential budgetary impact when using the WAC
price and the net price were approximately $2,500 and $960, respectively. The average five-
year annualized potential budgetary impact at the three cost-effectiveness threshold prices
ranged from approximately $7,200 per patient using the annual price to achieve $150,000 per
QALY to approximately $2,450 using the annual price to achieve a $50,000 per QALY cost-
effectiveness threshold.
As shown in Figure 7.2, the ICER budget impact threshold of $819 million would be crossed if
greater than approximately 2% of eligible patients were treated in a given year with icosapent
ethyl at the WAC price. When using the net price, the threshold would be reached if
approximately 4% of eligible patients were treated in a given year with icosapent ethyl.
Between 1% and 2% of patients could be treated in a given year without crossing the budget
impact threshold at the $150,000, $100,000, and $50,000 per QALY threshold prices. As with
rivaroxaban, the relatively small budget impact per patient, when combined with the very large
population of patients potentially eligible for treatment, has the potential to exceed the budget
impact threshold.

Figure 7.2. Potential Budget Impact of Icosapent Ethyl at Various Prices
$10,000
$9,000 $150,000 per QALY
$8,000
$7,000
$6,000 $100,000 per QALY

Annual Price
$5,000
$4,000 WAC price

$50,000 per QALY
$3,000
$2,000
Net price
$1,000
$0
0% 20% 40% 60% 80% 100%
7.4 Access and Affordability Alert

ICER is issuing an access and affordability alert for both rivaroxaban and icosapent ethyl. For
rivaroxaban, at the net price of $2,215 per year, approximately 6% of eligible patients could be
treated in a given year without crossing the ICER potential budget impact threshold of $819
million, while clinical experts at the Public Meeting stated that they would consider using
rivaroxaban in approximately 30% of eligible patients. For icosapent ethyl, at the net price of
$1,625 per year, approximately 4% of eligible patients could be treated in a given year without
crossing the ICER potential budget impact threshold of $819 million. Clinical experts stated at
the Public Meeting that they believe that the majority of eligible patients would want to be on
icosapent ethyl. The purpose of an ICER affordability and access alert is to signal to stakeholders
and policy makers that the amount of added health care costs associated with a new service
may be difficult for the health care system to absorb over the short term without displacing
other needed services or contributing to rapid growth in health care insurance costs. Thus,
there is a risk to sustainable access to high-value care for all patients if managing these issues is
not appropriately planned for.

8. Summary of the Votes and Considerations
for Policy
8.1 About the Midwest CEPAC Process
During Midwest CEPAC public meetings, the Midwest CEPAC Panel deliberates and votes on key
questions related to the systematic review of the clinical evidence, an economic analysis of the
applications of treatments under examination, and the supplementary information presented.
Panel members are not pre-selected based on the topic being addressed and are intentionally
selected to represent a range of expertise and diverse perspectives.
Acknowledging that any judgment of evidence is strengthened by real-life clinical and patient
perspectives, subject matter experts are recruited for each meeting topic and provide input to
Midwest CEPAC Panel members before the meeting to help clarify their understanding of the
different interventions being analyzed in the evidence review. The same clinical experts serve
as a resource to the Midwest CEPAC Panel during their deliberation and help to shape
recommendations on ways the evidence can apply to policy and practice.
After the Midwest CEPAC Panel votes, a policy roundtable discussion is held with the Midwest
CEPAC Panel, clinical experts, patient advocates, payers, and when feasible, manufacturers. The
goal of this discussion is to bring stakeholders together to apply the evidence to guide patient
education, clinical practice, and coverage and public policies. Participants on policy roundtables
are selected for their expertise on the specific meeting topic, are different for each meeting, and
do not vote on any questions.
At the September 26, 2019 meeting, the Midwest CEPAC Panel discussed issues regarding the
application of the available evidence to help patients, clinicians, and payers address important
questions related to the use of rivaroxaban and icosapent ethyl. Following the evidence
presentation and public comments (public comments from the meeting can be accessed
[https://www.youtube.com/watch?v=2LFCTqW_hSc&feature=youtu.be], starting at minute
[1:28:55]), the Midwest CEPAC Panel voted (votes from the meeting can be accessed
[https://www.youtube.com/watch?v=ATMvwRQ_RPk&feature=youtu.be], starting at minute
[0:02:23]), on key questions concerning the comparative clinical effectiveness, and potential
other benefits and contextual considerations related to rivaroxaban and icosapent ethyl. These
questions are developed by the ICER research team for each assessment to ensure that the
questions are framed to address the issues that are most important in applying the evidence to
support clinical practice, medical policy decisions, and patient decision-making. The voting
results are presented below, along with specific considerations mentioned by Midwest CEPAC
Panel members during the voting process.

In its deliberations and votes related to value, the Midwest CEPAC Panel considered the
individual patient benefits, and incremental costs to achieve such benefits, from a given
intervention over the long term.
There are four elements to consider when deliberating on long-term value for money (see
Figure 8.1 below):
• Comparative clinical effectiveness is a judgment of the overall difference in clinical

outcomes between two interventions (or between an intervention and placebo),
tempered by the level of certainty possible given the strengths and weaknesses of the
body of evidence. Midwest CEPAC uses the ICER Evidence Rating Matrix as its
conceptual framework for considering comparative clinical effectiveness.
• Estimated incremental cost effectiveness is the average incremental cost per patient of
one intervention compared to another to achieve a desired “health gain,” such as an
additional stroke prevented, case of cancer diagnosed, or gain of a year of life.
Alternative interventions are compared in terms of cost per unit of effectiveness, and
the resulting comparison is presented as a cost-effectiveness ratio. Relative certainty in
the cost and outcome estimates continues to be a consideration. As a measure of cost
effectiveness, the Midwest CEPAC voting panel follows common academic and health
technology assessment standards by using cost per quality-adjusted life year (QALY),
with formal voting on “long-term value for money” when the base-case incremental
cost-effectiveness ratio is between $50,000 per QALY and $175,000 per QALY.
• Potential other benefits refer to any significant benefits or disadvantages offered by the
intervention to the individual patient, caregivers, the delivery system, other patients, or
the public that would not have been considered as part of the evidence on comparative
clinical effectiveness. Examples of potential other benefits include better access to
treatment centers, mechanisms of treatment delivery that require fewer visits to the
clinician’s office, treatments that reduce disparities across various patient groups, and
new potential mechanisms of action for treating clinical conditions that have
demonstrated low rates of response to currently available therapies. Other
disadvantages could include increased burden of treatment on patients or their
caregivers. For each intervention evaluated, it will be open to discussion whether
potential other benefits or disadvantages such as these are important enough to factor
into the overall judgment of long-term value for money. There is no quantitative
measure for potential other benefits or disadvantages.
• Contextual considerations include ethical, legal, or other issues (but not cost) that
influence the relative priority of illnesses and interventions. Examples of contextual
considerations include whether there are currently any existing treatments for the
condition, whether the condition severely affects quality of life or not, and whether

there is significant uncertainty about the magnitude of benefit or risk of an intervention
over the long term. There is no quantitative measure for contextual considerations.
Figure 8.1. Conceptual Structure of Long-term Value for Money
8.2 Voting Results
Midwest CEPAC Votes
Clinical Evidence
a) Is the evidence adequate to demonstrate that the net health benefit of rivaroxaban
plus ASA is superior to that provided by ASA alone?
Yes: 10 votes No: 1 vote
A majority of the Council determined that the evidence was adequate to demonstrate
that the net health benefit of rivaroxaban plus ASA is superior to that provided by ASA
alone.
The Council member who voted in the negative was concerned that the absolute
difference between the two options was quite small, and there is no certainty that the
benefit would be maintained over a lifetime. In addition, given the inclusion criteria in
the COMPASS trial, the bleeding risk of patients in the trial is likely substantially lower
than that of patients in the real world.

b) Is the evidence adequate to demonstrate that the net health benefit of rivaroxaban
plus ASA is superior to that provided by ASA as part of dual antiplatelet therapy
(DAPT) with an oral P2Y12 inhibitor (e.g., ticagrelor or clopidogrel)?
The Council unanimously judged that the evidence was inadequate to demonstrate a
superior net health benefit of rivaroxaban plus ASA compared to that provided by ASA
as part of dual antiplatelet therapy (DAPT) with an oral P2Y12 inhibitor (e.g., ticagrelor or
clopidogrel) in patients with cardiovascular disease. Given the lack of head-to-head
trials and the different ways that bleeding was measured across trials, panelists
concluded there was not yet evidence to make a determination on rivaroxaban plus ASA
compared to DAPT.
c) Is the evidence adequate to demonstrate that the net health benefit of icosapent ethyl
added to optimal medical management (including statin therapy) is superior to that
provided by optimal medical management (including statin therapy) alone?
A majority of the Council determined that the evidence was adequate to demonstrate a
net health benefit of icosapent ethyl added to optimal medical management (including
statin therapy).
The Council members who voted in the negative expressed concerns regarding the
unexpected results in the REDUCE-IT trial compared to the many previous studies
investigating fish oil that showed no benefit to cardiovascular health, and questioned
whether these results could be replicated. One council member argued that the
potential benefit of icosapent ethyl is relatively small when considering absolute effects.
The other dissenting voter was concerned about the confounding factor of mineral oil
being used in the placebo arm and potentially inflating the results. Council members did
discuss the specific DHA and EPA composition of icosapent ethyl as it compares with
other fish oil interventions currently on the market, as this composition may be
responsible for the divergent results produced by the REDUCE-IT trial. However, they
agreed that additional study is required to gain any certainty on this question.

Other Benefits and Disadvantages, and Contextual Considerations
d) Does treating patients with rivaroxaban plus ASA offer one or more of the following
potential “other benefits or disadvantages” compared to ASA alone? (Select all that
apply).
failed.
Approximately half of the council judged that treating patients with rivaroxaban plus
ASA offers a novel mechanism of action or approach that will allow successful treatment
of many patients for whom other available treatments have failed. One Council member
felt that treating patients with rivaroxaban plus ASA will significantly reduce caregiver or
broader family burden, and one member voted that it will have a significant impact on
improving patients’ ability to return to work and/or their overall productivity. Prior to
voting, one Council member reiterated that, given the inclusion criteria in the COMPASS
trial, the bleeding risk of patients in the trial is likely substantially lower than that of
patients in the real world. The Council decided that this issue pertains to generalizability
and clinical effectiveness and does not represent an important disadvantage that should
have an important role in judgements of the value of this intervention. Another council
member mentioned that rivaroxaban’s introduction to the market could increase the
decision-making burden for patients and caregivers, as it is a potential addition to
patients’ already long and complex list of drug therapies. However, the Council agreed
that this is not a disadvantage specific to rivaroxaban plus ASA. No other important
benefits or disadvantages were raised, and the council did not vote on the final
question.
e) Does treating patients with icosapent ethyl offer one or more of the following
potential “other benefits or disadvantages” compared to optimal medical
management (including statin therapy) alone? (Select all that apply).
failed.

Approximately half of the Council judged that treating patients with icosapent ethyl
offers a novel mechanism of action or approach that will allow successful treatment of
many patients for whom other available treatments have failed. One Council member
stated his opinion that icosapent ethyl may offer a novel mechanism of action or
approach that will allow successful treatment of many patients for whom other available
treatments have failed; however, due to the nature of cardiovascular disease, it likely
will not significantly reduce caregiver burden or have a significant impact on improving
patients’ ability to return to work.
f) Are any of the following contextual considerations important in assessing the long-
term value for money of rivaroxaban plus ASA? (Select all that apply).
intervention.
There are additional contextual considerations that should have an important role in N/A
The Council considered cardiovascular disease to represent a condition of high severity

with a high lifetime burden of illness. One Council member noted the particularly large
proportion of Americans diagnosed with cardiovascular disease compared with those
affected by other, less prevalent diseases, as a possible contextual consideration of this
intervention. Most Council members agreed that there is significant uncertainty about
the long-term risk of serious side effects and the magnitude or durability of the long-
term benefits of rivaroxaban plus ASA. Bleeding risk is one long-term serious side effect
that can change with age and comorbidity status. Although the Council did not vote on
additional contextual considerations, members noted that less than 1% of participants in
the COMPASS trial self-identified as black Americans, a population that is at greater risk
for cardiovascular disease and stroke than white Americans. They also discussed
rivaroxaban plus ASA’s potential for good in large populations, as well as its potential for
harm.

g) Are any of the following contextual considerations important in assessing the long-
term value for money of icosapent ethyl? (Select all that apply).
intervention.
There are additional contextual considerations that should have an important role in No vote
T
The Council considered cardiovascular disease to represent a condition of high severity

with a high lifetime burden of illness. Five Council members voted that there is
significant uncertainty about the long-term risk of serious side effects and the
magnitude or durability of the long-term benefits of icosapent ethyl .
Long-Term Value for Money
As described in ICER’s value assessment framework, questions on long-term value for money are
subject to a value vote when incremental cost-effectiveness ratios for the interventions of
interest are between $50,000 and $175,000 per QALY in the primary “base-case” analysis. The
base-case estimates of the cost per QALY for both rivaroxaban + ASA and icosapent ethyl are
below the lower end of this range, and therefore the treatment is deemed “high long-term
value for money” without a vote unless the CEPAC determines in its discussion that the Evidence
Report base-case analysis does not adequately reflect the most probable incremental cost-
effectiveness ratio for either treatment.
ICER’s value assessment framework also describes that the CEPAC does not take Long-Term
Value For Money votes on treatments for which the evidence is considered inadequate during
the Clinical Vote. Such is the case for rivaroxaban + ASA versus DAPT therapy. Thus, no Value
vote was taken for rivaroxaban + ASA versus DAPT therapy (e.g. clopidogrel).
8.3 Roundtable Discussion and Key Policy Implications

Following its deliberation on the evidence, the Midwest CEPAC Panel engaged in a moderated
discussion with a policy roundtable about how best to apply the evidence on rivaroxaban and
icosapent ethyl to policy and practice associated with the treatment of CVD. The policy
roundtable members included two patients, two clinical experts, two payers, and one
representative from a pharmaceutical manufacturer. discussion reflected multiple perspectives
and opinions, and therefore, none of the statements below should be taken as a consensus view
held by all participants. The names of the Policy Roundtable participants are shown below, and
conflict of interest disclosures for all meeting participants can be found in Appendix G.

Table 8.3 Policy Roundtable Members
Name Title and Affiliation
Executive Director, Mended Hearts
Andrea Baer, MS, BCPA
Chester “Bernie” Good, MD,

Senior Medical Director, UPMC Health Plan
MPH
Craig Granowitz, MD, PhD Senior Vice President, Chief Medical Officer, Amarin Corporation
Kayla Leeser, PharmD Clinical Pharmacist, IngenioRx
Jeremy Sussman, MD, MS Associate Professor, University of Michigan
Marie Warshauer, MS Support Network
Director of Quality and Analytics, Massachusetts General Hospital Heart
Jason Wasfy, MD
Center
The roundtable discussion was facilitated by Dr. Steven Pearson, MD, MSc, President of ICER.
The main themes and recommendations from the discussion are organized by audience and
summarized below.
Payers
1. Evidence to compare rivaroxaban to dual-action platelet therapy (DAPT) plus aspirin is

limited by lack of head-to-head trials and differing outcome measures. Additionally, clinical
experts do not view these two treatment options as interchangeable given their different
mechanisms of action and risk profiles, therefore DAPT should not be considered an
appropriate candidate in a step therapy protocol as a first step prior to receiving coverage for
rivaroxaban.
Although subgroups in trials of rivaroxaban were comparable to some trial populations in which
DAPT was evaluated, looking at these subgroups limited the population size and increased
imprecision in results. Additionally, bleeding outcomes were reported differently and were not
comparable. ICER concluded the evidence was insufficient to compare the net benefits of
rivaroxaban and DAPT.
Other considerations for key prior authorization criteria for rivaroxaban are shown below:
Patient Eligibility Criteria:
a. Severity of disease: The FDA indication for rivaroxaban allows for treatment in a
broader population than the high-risk patients enrolled in COMPASS, yet there is no
strong evidence-based approach to defining a narrower set of eligibility criteria for
coverage.
b. Bleeding risk: Patients at high risk of major bleeding were excluded from the COMPASS
trial of rivaroxaban. Payers might consider instituting coverage criteria requiring
clinicians to attest that patients have not had a prior major bleed and/or are not
currently at high risk for future bleeds. However, balancing bleeding risk with CV event

risk needs to be an individualized decision, and payers may wish to frame coverage
language without any determination of bleeding risk.
Provider Criteria:
1) Specialist prescribing should not be viewed as a necessary part of a coverage policy.

While payers might consider limiting prescribing to cardiologists or in consultation with a
cardiologist, many non-specialists have experience prescribing rivaroxaban for
conditions such as atrial fibrillation, deep venous thrombosis, and pulmonary embolism.
The dose of rivaroxaban for CV risk reduction is much lower than that used for these
other conditions, and so non-specialists may be comfortable prescribing rivaroxaban for
this indication as well.
2. Icosapent ethyl has not yet received FDA approval and therefore the specific language of
the label is unknown. While awaiting the FDA decision, payers should consider parameters of
coverage criteria related to the eligibility criteria of the pivotal trial: a) definition of risk for
coronary artery disease; b) concurrent use of statins; and c) triglyceride level of 135-499
mg/dL.
a. Definition of risk of CAD: Patients in the REDUCE-IT trial were required to either have
established CVD or be 50 or older with diabetes and at least one additional risk factor.
Many primary prevention patients would have a constellation of risk factors creating a
similar risk to that in the primary prevention cohort of the trial, and the evidence does
not strongly support the use of the specific trial criteria around risk. Payers may wish to
consider coverage criteria based on total risk rather than based on the trial criteria.
b. Concurrent use of statins: The REDUCE-IT trial required patients to be on a stable dose
of a statin and to have an LDL-cholesterol level of 41-100 mg/dL. Given uncertainties
around the mechanism of benefit of icosapent ethyl and whether it would be effective in
patients not receiving statins, payers may consider requiring that patients be taking
statin therapy when prescribed icosapent ethyl. Payers face a challenging situation for
patients who are statin intolerant—they could limit eligibility, or they could use this as
an opportunity to get patients on a statin, since many patients felt to be statin
“intolerant” are able to take statins with appropriate clinical support.
c. Triglyceride level greater than 135: The REDUCE-IT trial did not suggest that the
benefits of icosapent ethyl were related to baseline triglyceride level, and other
evidence has also suggested that therapies that reduce triglycerides do not necessarily
reduce CV risk. As such, there is no strong reason to believe that icosapent ethyl is more
effective at reducing CV risk for patients with triglyceride levels meeting the entry
criteria for the trial. While payers could decide to limit icosapent ethyl coverage to
match the trial eligibility criteria, if the FDA label does not include a triglyceride level
requirement, then plans may choose also not to have a criterion related to triglyceride
level.

Providers
3. Clinicians, when thinking about the apparent benefit of rivaroxaban in the clinical trial,
should remember that patients at high risk of bleeding were excluded.
Clinicians should individualize decisions about adding treatments that decrease CV risk but
increase risk of bleeding based on patients’ individual risks for these events. Patients at high
risk of bleeding are likely to have a different net benefit from what was seen in the COMPASS
trial of rivaroxaban.
4. Develop options to help patients navigate complex medication regimens.
Patients with CVD are already treated with many medications each with their own regimens and
side effects. Medications include statins, aspirin or DAPT, angiotensin-converting-enzyme (ACE)
inhibitors, beta blockers, and potentially many more medications for comorbidities such as
diabetes. In this milieu, adding one or two additional therapies may be overwhelming for
patients. Shared decision making and patient support through education, caregivers, and
potentially the use of health coaches may assist in mitigating the cognitive and emotional
burden on patients from having so many therapies to manage.
Clinical and Specialty Societies
5. Develop a decision algorithm and/or tool for clinicians to use in determining the most
appropriate additive therapies to consider for a given patient.
In addition to the therapies evaluated in this report, treatments that can be considered for
residual CV risk include PCSK9 inhibitors, GLP-1 receptor agonists, and SGLT-2 inhibitors, among
others. There is overlap in the populations eligible for many of these therapies, and evidence-
based guidance around how to order considerations of these treatments for different patients
could standardize decision making and improve outcomes.
6. Ensure that any clinical guideline statements regarding rivaroxaban clearly warn against
assuming a class effect for direct oral anticoagulants.
While there are other direct oral anticoagulants (DOACs) available on the market, none
currently share rivaroxaban’s indication for prevention of cardiovascular events. The DOACs
also vary significantly in terms of dosing, and an observational study conducted in a Medicare
population suggests that the benefit-risk profile of these agents also differs when used for
stroke prevention in atrial fibrillation.112 Clinical societies should therefore direct clinicians to
consider only low-dose rivaroxaban for cardioprotection until similar evidence is available for
other DOACs.

Manufacturers
7. Conduct additional studies of icosapent ethyl in patients not on statin therapy.
Given the nature and findings of REDUCE-IT, payers may require patients to be on statin therapy
concurrently with icosapent ethyl. Yet many patients, if not strictly statin-intolerant, are
unwilling to take statins, and such a requirement might exacerbate this problem and prevent
some patients from receiving what could be a promising intervention. The manufacturer should
conduct additional clinical trials to assess the clinical performance of icosapent ethyl in patients
not receiving statins.
8. Ensure that future trial recruitment reflects the demographics of the CVD population.
COMPASS and REDUCE-IT were large, multicenter clinical trials, yet trial patients were not
reflective of the CVD population at large in important ways. For example, African Americans are
35% more likely to develop CVD than their white counterparts113 and have rates of stroke that
are twice as high.114 Yet only 1% of COMPASS participants were black. Enrollment of more
representative CVD populations will not only increase generalizability but can also potentially
provide reassurance that treatment effects persist across appropriately sized and pre-specified
subgroups.
Regulators
9. The FDA, manufacturers, and the clinical research community should work to solidify a
common, single, outcomes definitions for key outcomes -- such as major bleeding -- so
clinicians and patients have the information they need to make informed decisions.
As noted in the report and discussed during the roundtable, standard definitions of key
outcomes such as major bleeding have been developed, yet they have been modified to such an
extent that formal indirect comparisons of results from clinical trials of competing treatment
options were not possible. This represents a disservice to patients and clinicians, who need to
understand how to best weigh the risks and benefits of the alternatives available to them.
Given that these standard definitions exist, regulators, manufacturers, and the research
community should come together to identify a core outcome definition that can be used across
trials and treatments.
Researchers
10. Researchers should develop explicit head-to-head evidence of the comparative benefits
and risks of rivaroxaban + ASA versus dual antiplatelet therapy in patients who have
completed an initial course of DAPT (12-30 months).
There is an important need for head-to head comparative analyses of DAPT and rivaroxaban +
ASA. Prolonged use (i.e., 12-30 months) of dual antiplatelet therapy (DAPT) after an MI is
currently standard clinical practice. After an initial course of DAPT, some patients may benefit

from indefinite treatment. However, low-dose rivaroxaban + ASA is also now a potential option
rather than ongoing DAPT. How the balance of clinical benefits and bleeding risks compares
between these two alternatives is an open question in the absence of direct comparative
evidence. Subtle but important factors in populations presenting for treatment might introduce
selection bias in an observational setting, therefore an observational analysis is unlikely to
provide persuasive findings and a randomized trial is needed. The Patient-Centered Outcomes
Research Institute (PCORI) or the National Institute of Health (NIH) are well-suited to sponsor
this kind of trial to provide critical information for clinicians and patients that will remain
unanswered otherwise.
11. Researchers should conduct a real-world observational study to confirm the benefits of
icosapent ethyl.
The results of REDUCE-IT, while impressive, have not eliminated uncertainty regarding the
potential effects of the placebo composition in the trial, as well as the largely unimpressive
evidence from prior omega-3 studies. In addition, the FDA is unlikely to require a second
confirmatory clinical trial for this indication. A rigorous prospective observational study of
icosapent ethyl that had results consistent with those seen in the REDUCE-IT trial would help
alleviate concerns, while a conflicting result would suggest the need for a second randomized
trial.
****
This is the first ICER review of rivaroxaban and icosapent ethyl for CVD.

References
1. Woolf S. An organized analytic framework for practice guideline development: using the
analytic logic as a guide for reviewing evidence, developing recommendations, and
explaining the rationale. Clinical Practice Guideline Development: Methodology Perspectives
AHCPR Pub. 1994;95(0009):105-113.
2. Benjamin EJ, Muntner P, Alonso A, et al. Heart Disease and Stroke Statistics-2019 Update: A
Report From the American Heart Association. Circulation. 2019;139(10):e56-e66.
3. CD F. Prevalence of Uncontrolled Risk Factors for Cardiovascular Disease: United States,
1999–2010. NCHS Data Briefs. 2012(103).
4. Levine DA, Davydow DS, Hough CL, Langa KM, Rogers MA, Iwashyna TJ. Functional disability
and cognitive impairment after hospitalization for myocardial infarction and stroke.
Circulation Cardiovascular quality and outcomes. 2014;7(6):863-871.
5. Patel R, Tragante V, Schmidt AF, et al. Subsequent Event Risk in Individuals with Established
Coronary Heart Disease: Design and Rationale of the GENIUS-CHD Consortium. Circulation
Genomic and precision medicine. 2019.
6. Administration USFaD. Xarelto (R) [package insert]. Janssen Pharmaceuticals, Inc. 2011;2019.
7. Amarin Pharma I. Priority Review for Icosapent Ethyl. 2019.
8. Feuerstein A. Amarin now says FDA plans outside panel review of its Vascepa heart drug,
delaying approval decision. Stat news. August 8, 2019, 2019.
9. Bhatt DL, Steg PG, Miller M, et al. Cardiovascular Risk Reduction with Icosapent Ethyl for
Hypertriglyceridemia. The New England journal of medicine. 2019;380(1):11-22.
10. Hijazi M, Aljohani S, Alqahtani F, et al. Perception of the Risk of Stroke and the Risks and
Benefits of Oral Anticoagulation for Stroke Prevention in Patients With Atrial Fibrillation: A
Cross-Sectional Study. Mayo Clinic proceedings. 2019;94(6):1015-1023.
11. Eikelboom JW, Connolly SJ, Bosch J, et al. Rivaroxaban with or without Aspirin in Stable
Cardiovascular Disease. The New England journal of medicine. 2017;377(14):1319-1330.
12. van Swieten J, Koudstaal P, Visser M, Schouten H, van Gijn J. Interobserver agreement for
the assessment of handicap in stroke patients. Stroke. 1988;19(5):604-607.
13. Bonita R, Beaglehole R. Recovery of motor function after stroke. Stroke. 1988;19(12):1497-
1500.
14. Sharma M, Hart RG, Connolly SJ, et al. Stroke Outcomes in the COMPASS Trial. Circulation.
2019;139(9):1134-1145.
15. Bonaca MP, Bhatt DL, Cohen M, et al. Long-Term Use of Ticagrelor in Patients with Prior
Myocardial Infarction. New England Journal of Medicine. 2015;372(19):1791-1800.
16. Bonaca MP, Storey RF, Theroux P, et al. Efficacy and Safety of Ticagrelor Over Time in
Patients With Prior MI in PEGASUS-TIMI 54. Journal of the American College of Cardiology.
2017;70(11):1368-1375.
17. Bhatt DL, Fox KAA, Hacke W, et al. Clopidogrel and Aspirin versus Aspirin Alone for the
Prevention of Atherothrombotic Events. New England Journal of Medicine.
2006;354(16):1706-1717.
18. Bhatt DL, Flather MD, Hacke W, et al. Patients With Prior Myocardial Infarction, Stroke, or
Symptomatic Peripheral Arterial Disease in the CHARISMA Trial. Journal of the American
College of Cardiology. 2007;49(19):1982-1988.
19. Bhatt DL, Steg PG, Miller M, et al. Cardiovascular Risk Reduction with Icosapent Ethyl for
Hypertriglyceridemia. The New England journal of medicine. 2019;380(1):11-22.
20. Bhatt DL, Steg PG, Miller M, et al. Effects of Icosapent Ethyl on Total Ischemic Events: From
REDUCE-IT. Journal of the American College of Cardiology. 2019.
21. Eikelboom JW, Connolly SJ, Bosch J, et al. Rivaroxaban with or without Aspirin in Stable
Cardiovascular Disease. The New England journal of medicine. 2017;377(14):1319-1330.

22. Niessner A, Agewall S. Critical appraisal of the COMPASS trial. Eur Heart J Cardiovasc
Pharmacother. 2018;4(4):191-192.
23. Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and
estimation of treatment effects: systematic review and meta-regression analysis. JAMA.
2010;303(12):1180-1187.
24. Jeppsson A, Petricevic M, Kolh P, Valgimigli M. 2017 European Society of Cardiology (ESC)
focused update on dual antiplatelet therapy in collaboration with the European Association
for Cardio-Thoracic Surgery (EACTS). European journal of cardio-thoracic surgery : official
journal of the European Association for Cardio-thoracic Surgery. 2018;53(1):3-4.
25. Braunwald E. An Important Step for Thrombocardiology. The New England journal of
medicine. 2017;377(14):1387-1388.
26. Cho SW, Franchi F, Angiolillo DJ. Role of oral anticoagulant therapy for secondary prevention
in patients with stable atherothrombotic disease manifestations. Therapeutic advances in
hematology. 2019;10:2040620719861475.
27. Abdelhamid AS, Brown TJ, Brainard JS, et al. Omega-3 fatty acids for the primary and
secondary prevention of cardiovascular disease. The Cochrane database of systematic
reviews. 2018;11:Cd003177.
28. Pocock SJ, Collier TJ. Statistical Appraisal of 6 Recent Clinical Trials in Cardiology: JACC State-
of-the-Art Review. J Am Coll Cardiol. 2019;73(21):2740-2755.
29. Matsuzaki M, Yokoyama M, Saito Y, et al. Incremental effects of eicosapentaenoic acid on
cardiovascular events in statin-treated patients with coronary artery disease. Circulation
journal : official journal of the Japanese Circulation Society. 2009;73(7):1283-1290.
30. Zhang J, Sasaki S, Amano K, Kesteloot H. Fish consumption and mortality from all causes,
ischemic heart disease, and stroke: an ecological study. Preventive medicine.
1999;28(5):520-529.
31. Brinton EA, Mason RP. Prescription omega-3 fatty acid products containing highly purified
eicosapentaenoic acid (EPA). Lipids in health and disease. 2017;16(1):23.
32. SSRHealth. US Brand Rx Net Price. 2019.
33. Redbook. 2019. Accessed June 1, 2019.
34. Ten Cate H, Hemker HC. Thrombin Generation and Atherothrombosis: What Does the
Evidence Indicate? Journal of the American Heart Association. 2016;5(8).
35. Feuerstein A. Amarin now says FDA plans outside panel review of its Vascepa heart drug,
delaying approval decision. 2019. https://www.statnews.com/2019/08/08/amarin-now-
says-fda-plans-outside-panel-review-of-its-vascepa-heart-drug-delaying-approval-decision/.
Accessed 09/11/19.
36. Schulman S, Kearon C. Definition of major bleeding in clinical investigations of
antihemostatic medicinal products in non-surgical patients. Journal of thrombosis and
haemostasis : JTH. 2005;3(4):692-694.
37. Centers for Disease Control and Prevention. Types of Stroke.
https://www.cdc.gov/stroke/types_of_stroke.htm. Accessed July 9, 2019.
38. 2019. https://ilpriorauth.com/. Accessed June 27, 2019.
39. .
https://healthpartners.adaptiverx.com/webSearch/index?key=8F02B26A288102C27BAC82D
14C006C6FC54D480F80409B6897145FB47DE4F581. Accessed June 27, 2019.
40. City BCBSoK. Member Perscription Drug List 2019. 2019;
https://www.bluekc.com/consumer/pdfs/MemberPDL-2019.pdf. Accessed June 27 2019.
41. Management AP. Plan for your Best Health: 2019 Aetna Pharmacy Drug Guide, Aetna
Standard Plan. 2019; https://www.aetna.com/individuals-families/find-a-medication/2019-
aetna-standard-plans.html. Accessed June 27, 2019.
42. Cigna. 2019 Cigna Drug Lists. Standard 3 Tier List. Available at:
https://www.cigna.com/individuals-families/member-resources/prescription/drug-

list.html?consumerID=cigna&indicator=nonIFP&pdlYearType=CD&effectiveDate=2019-01-
01#/pdlv. Accessed June 26, 2019.
43. Medicare A. 2019 Comprehensive Formulary: Aetna Medicare List of Covered Drugs B2.
2019.
44. Foundation AHAaACoC. AHA/ACCF Secondary Prevention and Risk Reduction Therapy for
Patients with Coronary and Other Atherosclerotic Vascular Disease: 2011 Update. Journal of
American College of Cardiology. 2011;58(23):2432-2435.
45. Association ACoCFatAh. 2016 ACC/AHA Guideline Focused Update on Duration of Dual
Antiplatelet Therapy in patients with Coronary Artery Disease. Journal of American College
of Cardiology. 2016;68(10):1082-1099.
46. Skulas-Ray AC WP, Harris WS, Brinton EA, Kris-Etherton PM, Richter CK, Jacobson TA, Engler
MB, Miller M, Robinson JG, Blum CB, Rodriguez-Leyva D, de Ferranti SD, Welty FK; on behalf
of the American Heart Association Council on Arteriosclerosis, Thrombosis and Vascular
Biology; Council on Lifestyle and Cardiometabolic Health; Council on Cardiovascular Disease
in the Young; Council on Cardiovascular and Stroke Nursing; and Council on Clinical
Cardiology. Omega-3 fatty acids for the management of hypertriglyceridemia: a science
advisory from the American Heart Association. Circulation. 2019;140.
47. Association AD. 10. Cardiovascular Disease and Risk management: Standards of Medical Care
in Diabetes--2019. Diabetes Care. 2019;42(Supplement 1).
48. Cardiology TTFotmoscadotESo. 2013 ESC guidelines on the management of stable coronary
artery disease. European Heart Journal. 2013;34:2949-3003.
49. Valgimigli M. 2017 ESC focused update on dual antiplatelet therapy in coronary artery
disease developed in collaboration with EACTS: The Task Force for dual antiplatelet therapy
in coronary artery disease of the European Society of Cardiology (ESC) and of the European
Association for Cardio-Thoracic Surgery (EACTS) European Heart Journal. 2018;39(3):213-
260.
50. Mach F, Baigent C, Catapano AL, et al. 2019 ESC/EAS Guidelines for the management of
dyslipidaemias: lipid modification to reduce cardiovascular risk: The Task Force for the
management of dyslipidaemias of the European Society of Cardiology (ESC) and European
Atherosclerosis Society (EAS). European Heart Journal. 2019.
51. Cook DJ, Mulrow CD, Haynes RB. Systematic reviews: synthesis of best evidence for clinical
decisions. Ann Intern Med. 1997;126(5):376-380.
52. Higgins JPT GSe. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0
[updated March 2011]. The Cochrane Collaboration. Available from
http://handbook.cochrane.org.; 2011.
53. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. Int J Surg. 2010;8(5):336-341.
54. Agency for Healthcare Research and Quality. U.S. Preventive Services Task Force Procedure
Manual. 2008.
55. Ollendorf DA, Pearson SD. An integrated evidence rating to frame comparative effectiveness
assessments for decision makers. Med Care. 2010;48(6 Suppl):S145-152.
56. van Valkenhoef G, Kuiper J. gemtc: Network Meta-Analysis Using Bayesian Methods. R
package version 0.8-2. 2016.
57. Anand SS, Bosch J, Eikelboom JW, et al. Rivaroxaban with or without aspirin in patients with
stable peripheral or carotid artery disease: an international, randomised, double-blind,
placebo-controlled trial. Lancet (London, England). 2018;391(10117):219-229.
58. Connolly SJ, Eikelboom JW, Bosch J, et al. Rivaroxaban with or without aspirin in patients
with stable coronary artery disease: an international, randomised, double-blind, placebo-
controlled trial. Lancet (London, England). 2018;391(10117):205-218.

59. Anand SS, Caron F, Eikelboom JW, et al. Major Adverse Limb Events and Mortality in Patients
With Peripheral Artery Disease: The COMPASS Trial. Journal of the American College of
Cardiology. 2018;71(20):2306-2315.
60. Fox KAA, Eikelboom JW, Shestakovska O, Connolly SJ, Metsarinne KP, Yusuf S. Rivaroxaban
Plus Aspirin in Patients With Vascular Disease and Renal Dysfunction. Journal of the
American College of Cardiology. 2019;73(18):2243.
61. Moayyedi P, Eikelboom JW, Bosch J, et al. Pantoprazole to Prevents Gastroduodenal Events
in Patients Receiving Rivaroxaban and/or Aspirin in a Randomized, Double-blind, Placebo-
controlled Trial. Gastroenterology. 2019.
62. Branch KR, Probstfield JL, Eikelboom JW, et al. Rivaroxaban With or Without Aspirin in
Patients with Heart Failure and Chronic Coronary or Peripheral Artery Disease: The
COMPASS Trial. Circulation. 2019.
63. CADTH. Clinical Review Report -Rivaroxaban (Xarelto). Canadian Agency for Drugs and
Technologies in Health (CADTH);2019.
64. Lamy A, Eikelboom J, Sheth T, et al. Rivaroxaban, Aspirin, or Both to Prevent Early Coronary
Bypass Graft Occlusion: The COMPASS-CABG Study. J Am Coll Cardiol. 2019;73(2):121-130.
65. Anand SS, Eikelboom JW, Dyal L, et al. Rivaroxaban Plus Aspirin Versus Aspirin in Relation to
Vascular Risk in the COMPASS Trial. J Am Coll Cardiol. 2019;73(25):3271-3280.
66. Bays HE, Ballantyne CM, Kastelein JJ, Isaacsohn JL, Braeckman RA, Soni PN. Eicosapentaenoic
Acid Ethyl Ester (AMR101) Therapy in Patients With Very High Triglyceride Levels (from the
Multi-center, plAcebo-controlled, Randomized, double-blINd, 12-week study with an open-
label Extension [MARINE] Trial). The American journal of cardiology. 2011;108(5):682-690.
67. Ballantyne CM, Bays HE, Kastelein JJ, et al. Efficacy and Safety of Eicosapentaenoic Acid Ethyl
Ester (AMR101) Therapy in Statin-Treated Patients With Persistent High Triglycerides (from
the ANCHOR Study). The American journal of cardiology. 2012;110(7):984-992.
68. What is Amarin’s perspective on the use of mineral oil in its clinical trials and the variability
commonly observed in blood-based lipid values in clinical trials of statin-stabilized patients
(updated November 23, 2018)? [press release]. 2018.
69. Moayyedi P, Eikelboom JW, Bosch J, et al. Pantoprazole to Prevents Gastroduodenal Events
in Patients Receiving Rivaroxaban and/or Aspirin in a Randomized, Double-blind, Placebo-
controlled Trial. Gastroenterology. 2019.
70. Arora G, Singh G, Triadafilopoulos G. Proton pump inhibitors for gastroduodenal damage
related to nonsteroidal anti-inflammatory drugs or aspirin: twelve important questions for
clinical practice. Clinical gastroenterology and hepatology : the official clinical practice
journal of the American Gastroenterological Association. 2009;7(7):725-735.
71. Kinoshita Y, Ishimura N, Ishihara S. Advantages and Disadvantages of Long-term Proton
Pump Inhibitor Use. Journal of neurogastroenterology and motility. 2018;24(2):182-196.
72. Leontiadis GI, Sreedharan A, Dorward S, et al. Systematic reviews of the clinical effectiveness
and cost-effectiveness of proton pump inhibitors in acute upper gastrointestinal bleeding.
Health technology assessment (Winchester, England). 2007;11(51):iii-iv, 1-164.
73. Diamond GA, Kaul S. Forbidden fruit: on the analysis of recurrent events in randomized
clinical trials. The American journal of cardiology. 2013;111(10):1530-1536.
74. Metcalfe C, Thompson SG. Wei, Lin and Weissfeld's marginal analysis of multivariate failure
time data: should it be applied to a recurrent events outcome? Stat Methods Med Res.
2007;16(2):103-122.
75. Ozga AK, Kieser M, Rauch G. A systematic comparison of recurrent event models for
application to composite endpoints. BMC medical research methodology. 2018;18(1):2.
76. Singh S, Arora RR, Singh M, Khosla S. Eicosapentaenoic Acid Versus Docosahexaenoic Acid as
Options for Vascular Risk Prevention: A Fish Story. American journal of therapeutics.
2016;23(3):e905-910.

77. Jiang M, You JHS. Cost-effectiveness analysis of 30-month vs 12-month dual antiplatelet
therapy with clopidogrel and aspirin after drug-eluting stents in patients with acute coronary
syndrome. Clinical cardiology. 2017;40(10):789-796.
78. Zomer E, Si S, Hird TR, et al. Cost-effectiveness of low-dose rivaroxaban and aspirin versus
aspirin alone in people with peripheral or carotid artery disease: An Australian healthcare
perspective. European journal of preventive cardiology. 2018:2047487318817910.
79. Ademi Z, Zomer E, Tonkin A, Liew D. Cost-effectiveness of rivaroxaban and aspirin compared
to aspirin alone in patients with stable cardiovascular disease: An Australian perspective.
International journal of cardiology. 2018;270:54-59.
80. Kodera S, Morita H, Kiyosue A, Ando J, Komuro I. Cost-Effectiveness of Statin Plus
Eicosapentaenoic Acid Combination Therapy for Cardiovascular Disease Prevention in
Japanese Patients With Hypercholesterolemia- An Analysis Based on the Japan
Eicosapentaenoic Acid Lipid Intervention Study (JELIS). Circulation journal : official journal of
the Japanese Circulation Society. 2018;82(4):1076-1082.
81. Philip S, Chowdhury S, Nelson JR, Benjamin Everett P, Hulme-Lowe CK, Schmier JK. A novel
cost-effectiveness model of prescription eicosapentaenoic acid extrapolated to secondary
prevention of cardiovascular diseases in the United States. Journal of medical economics.
2016;19(10):1003-1010.
82. Bosch J, Eikelboom JW, Connolly SJ, et al. Rationale, Design and Baseline Characteristics of
Participants in the Cardiovascular Outcomes for People Using Anticoagulation Strategies
(COMPASS) Trial. Can J Cardiol. 2017;33(8):1027-1035.
83. Eikelboom JW, Connolly SJ, Bosch J. Rivaroxaban with or without aspirin in stable
cardiovascular disease [Supplementary Appendix]. The New England journal of medicine.
2017;377(14):S1-S37.
84. Amarin Pharma I. Important New Information: REDUCE-IT Cardiovascular Outcomes Study of
Vascepa. 2019; https://www.vascepahcp.com/vascepa-efficacy/reduce-it/#important-new-
info. Accessed 5/28/2019.
85. Alahmar AE, Nelson CP, Snell KI, et al. Resuscitated cardiac arrest and prognosis following
myocardial infarction. Heart. 2014;100(14):1125-1132.
86. Bronnum-Hansen H, Davidsen M, Thorvaldsen P, Danish MSG. Long-term survival and causes
of death after stroke. Stroke. 2001;32(9):2131-2136.
87. Owen-Smith V, Hannaford PC, Elliott AM. Increased mortality among women with Rose
angina who have not presented with ischaemic heart disease. Br J Gen Pract.
2003;53(495):784-789.
88. Shahar E, Lee S, Kim J, Duval S, Barber C, Luepker RV. Hospitalized heart failure: rates and
long-term mortality. J Card Fail. 2004;10(5):374-379.
89. Tancredi M, Rosengren A, Svensson AM, et al. Excess Mortality among Persons with Type 2
Diabetes. The New England journal of medicine. 2015;373(18):1720-1732.
90. Ultee KH, Rouwet EV, Hoeks SE, et al. Coronary revascularization induces a shift from cardiac
toward noncardiac mortality without improving survival in vascular surgery patients. J Vasc
Surg. 2015;61(6):1543-1549 e1541.
91. D’Agostino RB, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in
primary care: the Framingham Heart Study. Circulation. 2008;117(6):743-753.
92. Bhatt DL, Steg PG, Miller M, Brinton EA, Jacobson TA, Ketchum SB. Cardiovascular risk
reduction with icosapent ethyl for hypertriglyceridemia [Supplementary Appendix]. The New
England journal of medicine. 2019;380(1):S1-S50.
93. Arias E, Heron M, Xu J. United States Life Tables, 2012. National vital statistics reports : from
the Centers for Disease Control and Prevention, National Center for Health Statistics,
National Vital Statistics System. 2016;65(8):1-65.
94. Sullivan PW, Ghushchyan V. Preference-Based EQ-5D index scores for chronic conditions in
the United States. Med Decis Making. 2006;26(4):410-420.

95. Ali M, MacIsaac R, Quinn TJ, et al. Dependency and health utilities in stroke: Data to inform
cost-effectiveness analyses. Eur Stroke J. 2017;2(1):70-76.
96. Cohen DJ, Van Hout B, Serruys PW, et al. Quality of life after PCI with drug-eluting stents or
coronary-artery bypass surgery. The New England journal of medicine. 2011;364(11):1016-
1026.
97. Stevanovic J, Pechlivanoglou P, Kampinga MA, Krabbe PF, Postma MJ. Multivariate Meta-
Analysis of Preference-Based Quality of Life Values in Coronary Heart Disease. PLoS One.
2016;11(3):e0152030.
98. Wynn GJ, Todd DM, Webber M, et al. The European Heart Rhythm Association symptom
classification for atrial fibrillation: validation and improvement through a simple
modification. Europace. 2014;16(7):965-972.
99. Sullivan PW, Arant TW, Ellis SL, Ulrich H. The cost effectiveness of anticoagulation
management services for patients with atrial fibrillation and at high risk of stroke in the US.
Pharmacoeconomics. 2006;24(10):1021-1033.
100. Statistics BoL. Consumer Price Index Historical Table, U.S. City Average, All items, 1982-
84=100. 2019;
https://www.bls.gov/regions/midwest/data/consumerpriceindexhistorical_us_table.pdf.
Accessed 5/28/2019.
101. ICER. ICER’s Reference Case for Economic Evaluations: Principles and Rationale. 7/16/2018
2018.
102. Kazi DS, Moran AE, Coxson PG, et al. Cost-effectiveness of PCSK9 Inhibitor Therapy in
Patients With Heterozygous Familial Hypercholesterolemia or Atherosclerotic Cardiovascular
Disease. JAMA. 2016;316(7):743-753.
103. O'Sullivan AK, Rubin J, Nyambose J, Kuznik A, Cohen DJ, Thompson D. Cost estimation of
cardiovascular disease events in the US. Pharmacoeconomics. 2011;29(8):693-704.
104. Quality AfHRa. HCUPnet Healthcare Cost and Utilization Project. 2019;
https://hcupnet.ahrq.gov/#setup. Accessed June 1, 2019.
105. Song X, Quek RG, Gandra SR, Cappell KA, Fowler R, Cong Z. Productivity loss and indirect
costs associated with cardiovascular events and related clinical procedures. BMC Health Serv
Res. 2015;15:245.
106. Statistics BoL. Average hourly and weekly earnings of all employees on private nonfarm
payrolls by industry sector, seasonally adjusted. 2019;
https://www.bls.gov/news.release/empsit.t19.htm#ces_table3.f.p. Accessed July 1, 2019.
107. Fonarow GC, Keech AC, Pedersen TR, et al. Cost-effectiveness of Evolocumab Therapy for
Reducing Cardiovascular Events in Patients With Atherosclerotic Cardiovascular Disease.
JAMA Cardiol. 2017;2(10):1069-1078.
108. Jeffrey A. Tice M. PCSK9 Inhibitors for Treatment of High Cholesterol: Effectiveness, Value,
and Value-Based Price Benchmarks. www.icer-review.org: Institute for Clinical and Economic
Review; November 24, 2015 2015.
109. Levine GN, Bates ER, Bittl JA, et al. 2016 ACC/AHA guideline focused update on duration of
dual antiplatelet therapy in patients with coronary artery disease: A report of the American
College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines.
The Journal of thoracic and cardiovascular surgery. 2016;152(5):1243-1275.
110. Bertoluci MC. Cardiovascular risk assessment in patients with diabetes. Diabetology &
Metabolic Syndrome. 2017.
111. Pearson SD. The ICER Value Framework: Integrating Cost Effectiveness and Affordability in
the Assessment of Health Care Value. Value in health : the journal of the International
Society for Pharmacoeconomics and Outcomes Research. 2018;21(3):258-265.
112. Graham D. BE, Zhang R., Liao J., Wernecke M., Reichman M.E., Hu M., Illoh O., Wei Y.,
Goulding M.R., Chillarige Y., Southworth M.R., MaCurdy T.E., Kelman J.A.. Comparative

Stroke, Bleeding, and Mortality Risks in Older Medicare Patients Treated with Oral
Anticoagulants for Nonvalvular Atrial Fibrillation. The American Journal of Medicine.
113. Cardiology ACo. One Size Does Not Fit All: The Role of Sex, Gender, Race and Ethnicity in
Cardiovascular Medicine. Cardiology Magazine. 2018.
114. Association AH. Facts- Bridging the Gap: CVD Health Disparities.
115. Udell JA, Bonaca MP, Collet J-P, et al. Long-term dual antiplatelet therapy for secondary
prevention of cardiovascular events in the subgroup of patients with previous myocardial
infarction: a collaborative meta-analysis of randomized trials. European Heart Journal.
2015;37(4):390-399.
116. Zannad F, Anker SD, Byra WM, et al. Rivaroxaban in Patients with Heart Failure, Sinus
Rhythm, and Coronary Disease. The New England journal of medicine. 2018;379(14):1332-
1342.
117. Sanders GD, Neumann PJ, Basu A, et al. Recommendations for Conduct, Methodological
Practices, and Reporting of Cost-effectiveness Analyses: Second Panel on Cost-Effectiveness
in Health and Medicine. JAMA. 2016;316(10):1093-1103.
118. Pickard AS, Law EH, Jiang R, et al. United States Valuation of EQ-5D-5L Health States Using an
International Protocol. Value in health : the journal of the International Society for
Pharmacoeconomics and Outcomes Research. 2019;22(8):931-941.
119. Sabatine MS, Giugliano RP, Keech AC, et al. Evolocumab and Clinical Outcomes in Patients
with Cardiovascular Disease. The New England journal of medicine. 2017;376(18):1713-1722.

APPENDICES

Appendix A. Search Strategies and Results
Table A1. PRISMA 2009 Checklist
# Checklist Item
TITLE
Title 1 Identify the report as a systematic review, meta-analysis, or both.
ABSTRACT
Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria,
Structured Summary 2 participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and
implications of key findings; systematic review registration number.
INTRODUCTION
Rationale 3 Describe the rationale for the review in the context of what is already known.
Provide an explicit statement of questions being addressed with reference to participants, interventions,
Objectives 4
comparisons, outcomes, and study design (PICOS).
METHODS
Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide
Protocol and Registration 5
registration information including registration number.
Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered,
Eligibility Criteria 6
language, publication status) used as criteria for eligibility, giving rationale.
Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify
Information Sources 7
additional studies) in the search and date last searched.
Present full electronic search strategy for at least one database, including any limits used, such that it could be
Search 8
repeated.
State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable,
Study Selection 9
included in the meta-analysis).
Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes
Data Collection Process 10
for obtaining and confirming data from investigators.
List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and
Data Items 11
simplifications made.
Risk of Bias in Individual Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done
12
Studies at the study or outcome level), and how this information is to be used in any data synthesis.

# Checklist Item
Summary Measures 13 State the principal summary measures (e.g., risk ratio, difference in means).
Describe the methods of handling data and combining results of studies, if done, including measures of consistency
Synthesis of Results 14
(e.g., I2) for each meta-analysis.
Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective
Risk of Bias across Studies 15
reporting within studies).
Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating
Additional Analyses 16
which were pre-specified.
RESULTS
Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at
Study Selection 17
each stage, ideally with a flow diagram.
For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and
Study Characteristics 18
provide the citations.
Risk of Bias within Studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12).
For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each
Results of Individual Studies 20
intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.
Synthesis of Results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency.
Risk of Bias across Studies 22 Present results of any assessment of risk of bias across studies (see Item 15).
Additional Analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regression [see Item 16]).
DISCUSSION
Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to
Summary of Evidence 24
key groups (e.g., healthcare providers, users, and policy makers).
Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of
Limitations 25
identified research, reporting bias).
Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research.
FUNDING
Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the
Funding 27
systematic review.
From: Moher D, Liberati A, Tetzlaff J, Altman DG. The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The
PRISMA Statement. PLoS Med 6(6): e1000097. doi:10.1371/journal.pmed1000097

Table A2. Search Strategy of MEDLINE and Cochrane Central Register of Controlled Trials (via
Ovid)*
1 ((cardiovascular or heart or coronary or atherosclero$ or isch?emic or peripheral or arter$
or cerebrovascular) adj2 (diseas$ or disorder$ or syndrome$ or atherosclero$)).ti,ab.
2 ('icosapent ethyl' or vascepa or amr101).ti,ab.
3 exp rivaroxaban/ or (rivaroxaban or Xarelto).ti,ab.
4 2 or 3
5 1 and 4
6 (animals not (humans and animals)).sh.
7 5 not 6
8 (addresses or autobiography or bibliography or biography or case reports or classical article
or clinical conference or clinical trial, phase i or comment or congresses or consensus
development conference or consensus development conference, nih or dataset or dictionary
or directory or duplicate publication or editorial or equivalence trial or "expression of
concern" or guideline or historical article or interactive tutorial or interview or introductory
journal article or lectures or legal cases or legislation or letter or news or newspaper article
or overall or patient education handout or periodical index or personal narratives or portraits
or practice guideline or "published erratum review" or "scientific integrity review" or
technical report or twin study or validation studies or video-audio media or webcasts).pt.
9 7 not 8
10 limit 9 to english language
*Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R)
Daily, Ovid MEDLINE and Versions(R) 1946 to Present
Table A3. Search Strategy of EMBASE

1 ((cardiovascular OR heart OR coronary OR atherosclero* OR isch*emic OR peripheral OR
arter* OR cerebrovascular) NEAR/2 (diseas* OR disorder* OR syndrome* OR
atherosclero*)):ti,ab
2 'icosapentaenoic acid ethyl ester'/exp OR 'icosapent ethyl':ti,ab OR vascepa:ti,ab OR
amr101:ti,ab
3 'rivaroxaban'/exp OR rivaroxaban:ti,ab OR xarelto:ti,ab
4 #2 OR #3
5 #1 AND #4
6 ('animal'/exp OR 'nonhuman'/exp OR 'animal experiment'/exp) NOT 'human'/exp
7 #5 NOT #6
8 #7 AND ('case report'/de OR 'in vitro study'/de OR 'chapter'/it OR 'editorial'/it OR 'letter'/it
OR 'note'/it OR 'review'/it)
9 #7 NOT #8
10 #9 NOT [medline]/lim
11 #10 AND [english]/lim

Figure A1. PRISMA Flow Chart Showing Results of Literature Search for Rivaroxaban and
Icosapent Ethyl
808 references identified 4 references identified

through literature search through other sources
779 references after

duplicate removal
779 references screened 695 citations excluded
70 citations excluded
84 references assessed for 7 Population
eligibility in full text 7 Intervention
35 Outcome
21 Study Type
14 total references
related to 4 RCTs
1 RCT of Rivaroxaban
1 RCT of Icosapent Ethyl
2 RCTs of DAPT*
3 references included in
quantitative synthesis
1 Rivaroxaban
2 DAPT*
*DAPT: dual antiplatelet therapy

Appendix B. Previous Systematic Reviews and
Technology Assessments
We identified two completed technology assessment of rivaroxaban for CVD, which is summarized
below. We also identified one ongoing, but temporarily suspended technology assessment of
icosapent ethyl, and a peer-reviewed systematic review of DAPT for secondary prevention.
Canadian Agency for Drugs and Technologies in Health (CADTH) Common Drug Review:
Rivaroxaban (Xarelto) 2019
Although Health Canada approved rivaroxaban + ASA for CAD with or without PAD, CADTH
recommended that rivaroxaban be reimbursed only for patients with concomitant CAD and PAD.
CADTH’s reimbursement recommendation was informed by a review of the COMPASS trial,21 a
manufacturer-provided NMA comparing rivaroxaban to DAPT, and a manufacturer-submitted cost-
utility analysis. Based on this analysis, CADTH concluded that rivaroxaban + ASA is cost-effective in
patients with concomitant CAD and PAD compared with ASA alone, at an incremental cost-
effectiveness ratio of Can$17,764 per QALY gained. Reviewers noted a few gaps in the available
evidence, including uncertainties about the generalizability of the COMPASS trial results, the long-
term efficacy and safety of rivaroxaban + ASA, the optimal duration of therapy, whether
rivaroxaban + ASA improves health-related quality of life or daily function, and the comparative
efficacy and safety of rivaroxaban + ASA versus DAPT.
NICE: Rivaroxaban for Preventing Major Cardiovascular Events in People with Coronary or
Peripheral Artery Disease [ID1397], Expected publication date August 28, 2019
NICE recommends the use of rivaroxaban plus aspirin within its marketing authroitization, as an
option for preventing atherothrombotic events in adults with coronary artery disease or
symptomatic peripheral artery disease who are at high risk for ischemic events. Additionally, NICE
recommends assessing a person’s risk for bleeding before considering rivaroxaban.
Canadian Agency for Drugs and Technologies in Health (CADTH) Common Drug Review: Icosapent
Ethyl [SR0619-000], Review temporarily suspended.
CADTH is currently evaluating the clinical and cost effectiveness of icosapent ethyl for the
prevention of ischemic events in statin-treated patients with elevated triglyceride levels with CVD
or at high risk for CVD, however the review has been temporarily suspended pending receipt and
review of information.

Udell JA, et al. Long-term dual antiplatelet therapy for secondary prevention of cardiovascular
events in the subgroup of patients with previous myocardial infarction: a collaborative meta-
analysis of randomized trials. European Heart Journal. 2015;37(4):390-399.115
We identified a meta-analysis evaluating DAPT for the secondary prevention of cardiovascular

events in patients with a prior MI.115 Six trials that randomized patients to DAPT (n=20,203) or ASA
alone (n=13,232) for at least one year were included. One trial was conducted exclusively in
patients with a prior MI, and the other five trials were conducted in high-risk populations and
reported on subgroups of patients with a prior MI. Two trials compared the initiation of DAPT to
treatment with ASA alone, and the other four trials compared continuation of DAPT versus
discontinuation of the P2Y12 inhibitor (i.e., treatment with ASA alone). The primary outcome in all
six trials was a composite of cardiovascular death, MI, or stroke. The mean duration of follow-up
was 31 months.
At baseline, the mean age across studies was 64.0 years, 23.6% were female, 83.9% underwent or
had a history of PCI, and 29.6% had diabetes.115 Results from the meta-analysis showed treatment
with DAPT reduced the risk of the primary composite endpoint (cardiovascular death, MI, or stroke)
by 22% compared to ASA alone (6.4% vs. 7.5%, respectively, risk ratio [RR] 0.78; 95% CI: 0.67 to
0.90; p=0.0001). Patients treated with DAPT also experienced a significant reduction in
cardiovascular death (RR 0.85; 95% CI: 0.74 to 0.98; p=0.03), MI (RR 0.70; 95% CI: 0.55 to 0.88;
p=0.003), and stroke (RR 0.81; 95% CI 0.68 to 0.97; p=0.02).
The six trials measured the incidence of major bleeding using a variety of definitions; three trials
used the TIMI definition, two trials used the GUSTO definition, and one trial used the STEEPLE
definition. Nevertheless, the investigators conducted a meta-analysis of major bleeding endpoints
and found treatment with DAPT significantly increased the risk of major bleeding compared to ASA
alone (1.9% vs. 1.1%; RR: 1.73; 95% CI: 1.19 to 2.50, p=0.004), but not fatal bleeding (0.14% vs
0.17%; RR 0.91; 95% CI: 0.53 to 1.58; p=0.75) or intracranial hemorrhage (0.41% vs. 0.31%; RR 1.34;
95% CI: 0.89 to 2.02; p=0.17).

Appendix C. Ongoing Studies
Title/Trial Sponsor Study Design Arms Patient Population Primary Outcome Estimated Completion Date
Xarelto® (Rivaroxaban)
Efficacy and Safety of Phase III, Arm 1: Inclusion Time from randomization October 2019
Rivaroxaban in Reducing randomized, Rivaroxaban 2.5 • Age ≥50 to the first occurrence of
the Risk of Major double-blind, mg twice daily • Moderate to severe any of the following
Thrombotic Vascular parallel symptomatic lower major thrombotic
Events in Subjects with assignment Arm 2: Placebo extremity PAD vascular events: MI,
Symptomatic Peripheral • Peripheral revascularization ischemic stroke, CV
Artery Disease Enrollment: distal to the external iliac death, acute limb
Undergoing Peripheral 6564 artery for symptomatic ischemia, and major
Revascularization PAD within the last 10 days amputation
Procedures of the Lower
Extremities (VOYAGER Exclusion
PAD) • Patients undergoing
revascularization for
Bayer and Janssen asymptomatic PAD
Research & • Prior revascularization on the
Development, LLC index leg within 10 days of
the qualifying
NCT02504216 revascularization
• Planned use of additional
antiplatelet agent other
than clopidogrel and ASA

Switching for DAPT to Phase IV, Arm 1: ASA + Inclusion: Maximal platelet December 2020
Dual Pathway Inhibition randomized, clopidogrel → • Age ≥18 aggregation by light
With Low-Dose open label, continued DAPT • Known CAD and have transmittance
Rivaroxaban in Adjunct parallel completed their required aggregometry at 30 days
to Aspirin in Patients assignment Arm 2: ASA + duration of care DAPT and
With Coronary Artery clopidogrel → ASA still be on treatment: ≥6
Disease (SWAP-AC) Enrollment: 90 + rivaroxaban months after an elective
PCI, ≥12 months after
University of Florida & Arm 3: ASA + experiencing an ACS
Janssen, LP prasugrel → Exclusion:
continued DAPT • Deemed to be at high risk of
NCT04006288 bleeding, active bleeding,
Arm 4: ASA + or history of major
prasugrel → ASA + bleeding
rivaroxaban • Estimated GFR <15 ml/min by
MDRD equation
Arm 5: ASA + • Known non-CVD that is
ticagrelor → associated with poor
continued DAPT prognosis
Arm 6: ASA +
ticagrelor → ASA +
rivaroxaban

Vascepa® (Icosapent Ethyl)
Effect of Vascepa on Phase IV, Arm 1: Inclusion Change in baseline in low September 2019
Improving Coronary randomized, Icosapent ethyl • Age ≥30 attenuation plaque
Atherosclerosis in double blind, (4 g/day) • Elevated triglycerides (fasting at 18 months
People with High parallel value between 200-499
Triglycerides Taking assignment Arm 2: Placebo mg/dL)
Statin Therapy • LDL-C >40 mg/dL and ≤115
(EVAPORATE) Enrollment: 80 mg/dL
• Stable diet + exercise
Los Angeles Biomedical • Stable treatment with statin
Research Institute for at least four weeks
NCT02926027 Exclusion
• A contraindication to fish or
fish oils
• Use of non-study lipid
altering medications or
supplements
• Bleeding disorder
• Uncontrolled hypertension
• MI, stroke, life-threatening
arrhythmia within prior six
months

Other Ongoing Trials of Omega-3 Fatty Acids
Randomized Trial for Phase IV, Arm 1: Inclusion: Time from randomization October 2021
Evaluation in Secondary randomized, Eicosapentaenoic • Age ≥20 to the first occurrence of
Prevention Efficacy of open-label, acid (1.8 g/day) • CAD, defined by having either any of the following
Combination Therapy - parallel 1) history of acute coronary major CV events: CV
Statin and assignment Arm 2: Placebo system; 2) history of death, non-fatal MI, non-
Eicosapentaenoic Acid coronary revascularization; fatal cerebral infarction,
(RESPECT-EPA) or 3): diagnosis of ischemic unstable angina requiring
Target heart disease and severe emergent hospitalization
UMIN000012069 enrollment: coronary artery stenosis and coronary
3900 (≥75%) revascularization, and
Japan Heart Foundation • Stable treatment with statin coronary
for at least four weeks revascularization
Exclusion:
• Acute coronary syndrome or
coronary revascularization
within prior three months
• Planned coronary
revascularization or
angiography
• Inadequately controlled
diabetes mellitus
• Active bleeding or bleeding
tendency

Outcomes Study to Phase III, Arm 1: Omega-3- Inclusion: Time from randomization September 2020
Assess Statin Residual randomized, carboxylic acids • Age ≥18 to the first occurrence of
Risk Reduction With double-blind, • High risk for CV events, any of the following
Epanova in High CV Risk parallel Arm 2: Placebo defined as 1) CVD; 2) men major CV events: CV
Patients With assignment age ≥40 or women age ≥50 death, nonfatal MI,
Hypertriglyceridemia with diabetes mellitus and nonfatal stroke, coronary
(STRENGTH) Enrollment: at least one additional risk revascularization, or
13086 factor; or 3) men age>50 or hospitalization for
NCT02104817 women age>60 with at least unstable angina
one additional risk factor
AstraZeneca • Stable treatment with statin
for at least four weeks
• LDL-C <100 mg/dL
• TG level ≥180 and <500 mg/dL
Exclusion:
• Use of fibrates, bile acid
sequestrants, or niacin
within four weeks
Sources: www.ClinicalTrials.gov (NOTE: studies listed on site include both clinical trials and observational studies); UMIN Clinical Trials Registry (UMIN-CTR)
https://www.umin.ac.jp/ctr/index.htm

Appendix D. Comparative Clinical Effectiveness
Supplemental Information
US Preventive Services Task Force (USPSTF) Criteria for the Quality Assessment
of Clinical Trials and Comparative Cohort Studies
Good: Meets all criteria: Comparable groups are assembled initially and maintained throughout the
study; reliable and valid measurement instruments are used and applied equally to the groups;
interventions are spelled out clearly; all important outcomes are considered; and appropriate
attention is paid to confounders in analysis. In addition, intention to treat analysis is used for RCTs.
Fair: Studies were graded "fair" if any or all of the following problems occur, without the fatal flaws
noted in the "poor" category below: Generally comparable groups are assembled initially but some
question remains whether some (although not major) differences occurred with follow-up;
measurement instruments are acceptable (although not the best) and generally applied equally;
some but not all important outcomes are considered; and some but not all potential confounders
are addressed. Intention to treat analysis is done for RCTs.
Poor: Studies were graded "poor" if any of the following fatal flaws exists: Groups assembled
initially are not close to being comparable or maintained throughout the study; unreliable or invalid
measurement instruments are used or not applied equally among groups (including not masking
outcome assessment); and key confounders are given little or no attention. For RCTs, intention to
treat analysis is lacking.
Note that case series are not considered under this rating system – because of the lack of
comparator, these are generally considered to be of poor quality.
ICER Evidence Rating
We used the ICER Evidence Rating Matrix (see Figure D1) to evaluate the evidence for a variety of
outcomes. The evidence rating reflects a joint judgment of two critical components:
The magnitude of the difference between a therapeutic agent and its comparator in “net health
benefit” – the balance between clinical benefits and risks and/or adverse effects AND
The level of certainty in the best point estimate of net health benefit.55

Figure D1. ICER Evidence Rating Matrix

Evidence Tables for the Review of Rivaroxaban
Table D1. Study Quality of COMPASS

Outcome Appropriate
Comparable Adequate Patient Physician Non-Differential ITT Overall
Study Adjudication Handling of
Groups Randomization Blinding Blinding Follow-Up Analysis Quality
Blinding Missing Data
COMPASS Yes Yes Yes Yes Yes Yes Yes Yes Good
ITT: intent-to-treat

Table D2. Study Design of COMPASS
Study COMPASS (Cardiovascular Outcomes for People Using Anticoagulation Strategies)
Multicenter, double-blind, placebo-controlled, double-dummy, randomized trial. Patients were randomized 1:1:1 to received rivaroxaban (2.5 mg twice
Design daily) + ASA (100 mg daily), rivaroxaban (5 mg twice daily), or ASA (100 mg once daily) and stratified according to center and use of PPI at the time of
randomization. Eligible patients were also randomly assigned 1:1 to receive pantoprazole (40 mg daily) or matched placebo.
o CAD or PAD
Patients with CAD must also meet at least one of the following
Inclusion Criteria
o Age ≥65, or
o Age <65 and documented atherosclerosis or revascularization involving at least two vascular beds, or at least two additional risk factors
o High risk of bleeding
o Need for DAPT, other non-ASA antiplatelet therapy or oral anticoagulant therapy
Exclusion Criteria o Stroke within one month or any history of hemorrhagic or lacunar stroke
o Severe heart failure with known ejection fraction <30% or New York heart Association (NYHA) class III or IV symptoms
o Estimated glomerular filtration rate (eGFR) <15 mL/min
N 27,395
• 2.5 mg twice daily rivaroxaban + ASA (n=9,152)
Interventions • 5 mg twice daily rivaroxaban (n=9,117)
• 100 mg twice daily ASA (n=9,126)
Follow-Up Mean follow-up of 23 months
• Primary Endpoint: A composite of CV death, stroke, or MI
• Secondary Endpoints:
o A composite of ischemic stroke, MI, acute limb ischemia, or death from CAD
o A composite of ischemic stroke, MI, acute limb ischemia, or CV death
o Death from any cause
• Tertiary Endpoints:
o Individual components of primary and secondary endpoints
Outcomes o Hospitalization for CV causes
o Revascularization
o Limb amputation
o Stent thrombosis
o Angina
o Heart failure
o Resuscitated cardiac arrest
o Venous thromboembolism
o New diagnosis of cancer
ASA: aspirin, CAD: coronary artery disease, CV: cardiovascular, DAPT: dual antiplatelet therapy, MI: myocardial infarction, PPI: a proton-pump inhibitor, PAD: peripheral
artery disease

Table D3. Key Baseline Characteristics of COMPASS*11,57,58
Study Overall COMPASS CAD Subgroup PAD Subgroup
Intervention RV + ASA ASA RV + ASA ASA RV + ASA ASA
N 9152 9216 8313 8261 2492 2504
Age, Mean (SD) 68.3 (7.9) 68.2 (8.0) 69 (65-73)† 69 (65-73)† 67.9 (8.45) 67.8 (8.47)
Body Mass Index,
28.3 (4.8) 28.4 (4.7) 28.4 (4.7) 28.5 (4.7) 28.3 (5.0) 28.4 (5.0)
Mean (SD)
Blood Pressure, Mean 136 (17) systolic; 136 (18) systolic; 135 (17) systolic; 135 (17) systolic; 138.9 (18.5) systolic; 138.6 (18.2) systolic;
(SD) 77 (10) diastolic 78 (10) diastolic 77 (10) diastolic 78 (10) diastolic 77.7 (10.1) diastolic 77.8 (10.3) diastolic
Prior Stroke, No. (%) 351 (3.8) 335 (3.7) 279 (3) 268 (3) 171 (6.9) 154 (6.2)
Prior MI, No. (%) 5654 (61.8) 5721 (62.7) 5654 (68) 5721 (69) NR NR
Heart Failure, No. (%) 1963 (21.4) 1979 (21.7) 1909 (23) 1912 (23) NR NR
CAD, No. (%) 8313 (90.8) 8261 (90.5) -- -- 1656 (66.5) 1641 (65.5)
Symptomatic: 2026 Symptomatic: 2039
PAD, No. (%) 2492 (27.2) 2504 (27.4) 1656 (20) 1641 (20)
(81.3)‡ (81.4)‡
Diabetes, No. (%) 3448 (37.7) 3474 (38.1) 3043 (37) 3040 (37) 1100 (44.1) 1104 (44.1)
1679 (20) current 1687 (20) current 682 (27.4) current; 1147 685 (27.4) current; 1143
Smoking Status, No. 1944 (21.2) 1972 (21.6)
smoker; 3944 (47) smoker; 3908 (47) (46) former; 663 (26.6) (45.6) former; 676 (27)
(%) tobacco use tobacco use
former smoker former smoker never never
Concomitant Medications
ACE Inhibitor or ARB,
6475 (70.7) 6462 (70.8) 5970 (72) 5939 (72) 1715 (68.8) 1765 (70.5)
No. (%)
Beta-Blocker, No. (%) 6389 (69.8) 6394 (70.1) 6124 (74) 6154 (75) 1477 (59.3) 1485 (59.3)
Lipid-Lowering Agent,
8239 (90.0) 8158 (89.4) 7667 (92) 7573 (92) 2088 (83.8) 2074 (82.8)
No. (%)
NSAID, No. (%) 531 (5.8) 473 (5.2) NR NR NR NR
ACE: angiotensin-converting-enzyme, ARB: angiotensin II receptor blocker, ASA: aspirin, CAD: coronary artery disease, MI: myocardial infarction, No.: number of patients,
NSAID: nonsteroidal anti-inflammatory drug, PAD: peripheral artery disease, RV: rivaroxaban; SD: standard deviation
*Only reporting the results of the FDA approved 2.5 mg dose of rivaroxaban.
†Median (IQR).
‡Symptomatic PAD (the sum of systematic PAD of the lower extremities and carotid artery disease).

Table D4. Key Efficacy Outcomes in COMPASS at a Mean of 23 Months of Follow-Up*11,14,57-60,116
Study Overall COMPASS CAD Subgroup PAD Subgroup Renal Dysfunction Heart Failure
RV + ASA RV + ASA RV + ASA RV eGFR ASA RV eGFR ASA ≥60 RV + ASA ASA RV + ASA ASA
Intervention
ASA ASA ASA <60 ml <60 ≥60 ml No HF No HF HF HF
N 9152 9126 8313 8261 2492 2504 2054 2114 7094 7012 7189 7147 1963 1979
Primary Outcome
CV Death,
379 496 177
Stroke, or MI, 347 (4) 460 (6) 126 (5) 174 (7) 132 (6.4) 247 (3.5) 319 (4.5) 271 (3.8) 339 (4.7) 108 (5.5) 157 (7.9)
(4.1) (5.4) (8.4)
No. (%)
HR (95% CI) 0.76 (0.66-0.86) 0.74 (0.65-0.86) 0.72 (0.57-0.90) 0.75 (0.60-0.94) 0.76 (0.64-0.90) 0.79 (0.68-0.93) 0.68 (0.53-0.86)
P-Value p<0.001 p<0.0001 p=0.0047 p=0.01 p=0.001 -- --
Secondary Outcomes
Ischemic
Stroke, MI, ALI, 329 450
299 (4) 411 (5) 115 (5) 169 (7) NR NR NR NR NR NR NR NR
Death from (3.6) (4.9)
CHD, No. (%)
HR (95%CI) 0.72 (0.63-0.83) 0.72 (0.62-0.83) 0.68 (0.53-0.86)
NR NR NR NR NR NR NR NR
P-Value p<0.001 p<0.0001 p=0.0011
Ischemic
Stroke, MI, ALI 389 516
349 (4) 470 (6) 142 (6) 198 (8) NR NR NR NR NR NR NR NR
or CV Death, (4.3) (5.7)
No. (%)
HR (95%CI) 0.74 (0.65-0.85) 0.73 (0.64-0.84) 0.71 (0.57-0.88)
NR NR NR NR NR NR NR NR
P-Value p<0.001 p<0.0001 p=0.0019
Other Outcomes
CV Death, No. 203
160(1.7) 139 (2) 184 (2) 64 (3) 78 (3) 71 (3.5) 83 (3.9) 89 (1.3) 120 (1.7) NR NR NR NR
(%) (2.2)
HR (95%CI) 0.78 (0.64-0.96) 0.75 (0.60-0.93) 0.82 (0.59-1.14) 0.88 (0.64-1.20) 0.73 (0.56-0.96)
NR NR NR NR
P-Value p=0.02 p=0.010 -- p=0.41 p=0.02
Non-CV Death, 153 175
123 (2) 155 (2) NR NR NR NR NR NR NR NR NR NR
No. (%) (1.7) (1.9)
HR (95%CI) 0.87 (0.70-1.08) 0.79 (0.62-1.00)
NR NR NR NR NR NR NR NR NR
P-Value p=0.20 p=0.048

Death from 117 107
86 (0.9) 8 (1.0) NR NR NR NR NR NR NR NR NR NR
CHD, No. (%) (1.3) (1.3)
HR (95%CI) 0.73 (0.55-0.96) 0.74 (0.55-0.99)
NR NR NR NR NR NR NR NR NR NR
P-Value p=0.03 p=0.04
178 205
MI, No. (%) 169 (2) 195 (2) 51 (2) 67 (3) 53 (2.6) 73 (3.5) 125 (1.8) 132 (1.9) NR NR NR NR
(1.9) (2.2)
HR (95%CI) 0.86 (0.70-1.05) 0.86 (0.70-1.05) 0.76 (0.53-1.09) 0.73 (0.51-1.04) 0.93 (0.73-1.19)
NR NR NR NR
P-Value p=0.14 p=0.15 -- p=0.08 p=0.57
142
Stroke, No. (%) 83 (0.9) 74 (1) 130 (2) 25 (1) 47 (2) 20 (1.0) 49 (2.3) 63 (0.9) 93 (1.3) NR NR NR NR
(1.6)
HR (95% CI) 0.58 (0.44-0.76) 0.56 (0.42-0.75) 0.54 (0.33-0.87) 0.42 (0.25-0.70) 0.67 (0.48-0.92)
NR NR NR NR
P-Value p<0.001 p<0.0001 -- p=0.0007 p=0.01
Ischemic 132
64 (0.7) 60 (1) 120 (2) NR NR 14 (0.7) 46 (2.2) 54 (0.8) 86 (1.2) NR NR NR NR
Stroke, No. (%) (1.4)
HR (95% CI) 0.51 (0.38-0.68) 0.50 (0.36-0.67) 0.31 (0.17-0.57) 0.62 (0.44-0.87)
NR NR NR NR NR NR
P-Value p<0.001 p<0.0001 p<0.0001 p=0.005
Hemorrhagic
15 (0.2) 10 (0.1) 14 (<1) 10 (<1) NR NR 6 (0.3) 3 (0.1) 9 (0.1) 7 (<0.1) NR NR NR NR
Stroke, No. (%)
HR (95% CI) 1.49 (0.67-3.31) 1.39 (0.62-3.32) 2.01 (0.50-8.06) 1.26 (0.47-3.39)
NR NR NR NR NR NR
P-Value p=0.33 p=0.43 p=0.31 p=0.64
Heart Failure, 197 192
178 (2) 182 (2) NR NR NR NR NR NR NR NR NR NR
No. (%) (2.2) (2.1)
HR (95% CI) 1.02 (0.84-1.24) 0.97 (0.79-1.19)
P-Value p=0.84 p=0.78
Hospitalization
1303 1394 1189 1270
for CV Cause, NR NR NR NR NR NR NR NR NR NR
(14.2) (15.3) (14) (15)
No. (%)
HR (95% CI) 0.92 (0.86-1.00) 0.92 (0.85-1.00)
P-Value p=0.04 p=0.046
Hospitalization
1701 1624 1552 1481
for non-CV NR NR NR NR NR NR NR NR NR NR
(18.6) (17.8) (19) (18)
Cause, No. (%)

HR (95% CI) 1.05 (0.98-1.13) 1.05 (0.98-1.13)
P-Value p=0.14 p=0.18
Hospitalization
NR NR NR NR NR NR NR NR NR NR 98 (1.4) 76 (1.1) 57 (2.9) 69 (3.5)
for HF, No. (%)
HR (95% CI) 1.29 (0.95-1.74) 0.82 (0.58-1.16)
P-Value -- --
MALE, No. (%) NR NR NR NR 30 (1) 56 (2) NR NR NR NR NR NR NR NR
HR (95%CI) 0.54 (0.35-0.84)
NR NR NR NR NR NR NR NR NR NR NR NR
P-Value p=0.0054
Major
Amputation, NR NR NR NR 5 (<1) 17 (1) NR NR NR NR NR NR NR NR
No. (%)
HR (95%CI) 0.30 (0.11-0.80)
P-Value p=0.011
Vascular
Amputation, NR NR NR NR 11 (<1) 28 (1) NR NR NR NR NR NR NR NR
No. (%)
HR (95%CI) 0.40 (0.20-0.79)
P-Value p=0.0069
ALI: acute limb ischemia, ASA: aspirin, CAD: coronary artery disease, CHD: coronary heart disease, CI: confidence interval, CV: cardiovascular, eGFR: estimated
glomerular filtration rate, HF: heart failure, HR: hazard ratio, MALE: major adverse limb event, MI: myocardial infarction, No.: number of patients, NR: not reported,
PAD: peripheral artery disease, RV: rivaroxaban

Table D5. Key Safety Events in COMPASS at a Mean of 23 Months of Follow-Up*11,57,58,116
RV + ASA ASA RV + ASA RV + ASA RV eGFR ASA RV eGFR ASA RV + ASA RV + ASA
Intervention ASA ASA <60 ml <60 ≥60 ml ≥60 ASA No HF ASA HF
No HF HF
N 9152 9126 8313 8261 2492 2504 2054 2114 7094 7012 7189 7147 1963 1979
Any SAE, No. (%) 721 (7.9) 662 (7.3) NR NR NR NR NR NR NR NR NR NR NR NR
288 (3.1) 170 (1.9) 263 (3) 158 77 (3) 48 (2) 81 (3.9) 57 206 (2.9) 113 239 134 49 36
Major Bleeding, No. (%)
(2) (2.7) (1.6) (3.3) (1.9) (2.5) (1.8)
P-Value p<0.001 p<0.001 p=0.0089 p=0.02 p<0.0001 -- --
15 (02) 10 (0.1) 14 (<1) 9 (<1) 4 (<1) 3 (<1) 5 (0.2) 4 (0.2) 10 (0.1) 6 NR NR NR NR
Fatal Bleeding, No. (%)
(<0.1)
P-Value p=0.32 p=0.30 -- p=0.74 p=0.34 NR NR NR NR
21 (0.2) 19 (0.2) 19 (<1) 19 4 (<1) 8 (<1) NR NR NR NR NR NR NR NR
Nonfatal ICH, No. (%)
(<1)
P-Value p=0.77 p=0.98 -- NR NR NR NR NR NR NR NR
42 (0.5) 29 (0.3) 36 (<1) 25 13 (1) 8 (<1) NR NR NR NR NR NR NR NR
Nonfatal Non-ICH, No. (%)
(<1)
Site of Major Bleeding
GI, No. (%) 140 (1.5) 65 (0.7) 130 (2) 61 (1) 41 (2) 18 (1) NR NR NR NR NR NR NR NR
P-Value p<0.001 p<0.0001 p=0.0027 NR NR NR NR NR NR NR NR
28 (0.3) 24 (0.3) 26 (<1) 23 5 (<1) 9 11 (0.5) 7 (0.3) 17 (0.2) 17 19 (0.3) 18 9 6 (0.3)
Intracranial, No. (%)
(<1) (< 1) (0.2) (0.3) (0.5)
P-Value p=0.60 p=0.69 -- p=0.33 p=0.95 -- --
Skin or Injection Site, No. 28 (0.3) 12 (0.1) 25 (<1) 10 5 (<1) 8 (<1) NR NR NR NR NR NR NR NR
(%) (<1)
13 (0.1) 21 (0.2) 13 (<1) 21 NR NR NR NR NR NR NR NR NR NR
Urinary, No. (%)
(<1)
P-Value p=0.16 p=0.16 NR NR NR NR NR NR NR NR NR NR
Other

838 (9.2) 503 (5.5) 775 (9) 454 198 (8) 141 NR NR NR NR NR NR NR NR
Minor Bleeding, No. (%)
(6) (6)
P-Value p<0.001 p<0.0001 p=0.0011 NR NR NR NR NR NR NR NR
Net-Clinical-Benefit Outcome, n (%)
CV Death, Stroke, MI, Fatal 431 (4.7) 534 (5.9) 392 (5) 494 140 (6) 185 147 (7.2) 188 284 (4.0) 346 315 369 116 165
Bleeding, or Symptomatic (6) (7) (8.9) (4.9) (4.4) (5.2) (5.9) (8.3)
Bleeding into Critical Organ
HR (95% CI) 0.80 (0.70-0.91) 0.78 (0.69-0.90) 0.75 (0.60-0.94) 0.79 (0.64-0.98) 0.81 (0.69-0.94) 0.85 (0.73-0.99) - 0.69 (0.55-0.88)
P-Value p<0.001 p=0.0003 p=0.011 p=0.03 p=0.007 -- --
AE: Adverse event, ASA: aspirin, eGFR: estimated glomerular filtration rate, HF: heart failure, ICH: intracranial hemorrhage, MI: myocardial infarction, RV: rivaroxaban, SAE:
serious adverse event

Table D6. Summary of Pantoprazole to Prevent Gastroduodenal Events61
Study and Placebo, No.
Patient Characteristics Pantoprazole, No. (%) Between Group Differences
Population (%)
Primary outcome: clinically
HR (95% CI): 0.88 (0.67-1.15)
significant upper GI events*: 116 (1.3)
Pantoprazole (n=8791) Placebo (n=8807) p=0.35
102 (1.2)
Age, median: 68 Age, median: 68
Pantoprazole to Overt bleeding of
Female sex, no. (%): 1937 (22) Female sex, no. (%): 1869 (21) HR (95% CI): 0.52 (0.28-0.94)
Prevent gastroduodenal organ: 31 (0.4)
Previous MI, no. (%): 5403 (62) Previous MI, no. (%): 5404 (61) p=0.03
Gastroduodenal 16†(0.2)
Previous stroke, no. (%): 350 (4) Previous stroke, no. (%): 366 (4)
Events in Patients Overt upper GI bleeding of HR (95% CI): 1.09 (0.73-1.63)
Previous cancer, no. (%): 450 (5) Previous cancer, no. (%): 491 (6) 45 (0.5)
Receiving unknown origin: 50 (0.6) p=0.68
Diabetes, no. (%): 3363 (38) Diabetes no. (%): 3369 (38)
Rivaroxaban and/or Bleeding of presumed occult
Heart failure, no. (%): 2181 (25) Heart failure no. (%): 2138 (24)
Aspirin in a upper GI tract w/ documented HR (95% CI): 1.00 (0.42-2.40)
Randomized, 10 (0.1)
Medication Medication decrease in Hb ≥2 g/dL: 10 p=0.99
Double-Blind, (0.1)
NSAID, no. (%): 425 (5) NSAID, no. (%): 447 (5)
Placebo-Controlled
SSRIs, no. (%): 257 (3) SSRI, no. (%): 258 (3) Symptomatic gastroduodenal HR (95% CI): 0.47 (0.20-1.09)
Trial 17 (0.2)
ACE inhibitors/arbs, no. (%): ACE inhibitors/ARBs, no. (%): ulcer: 8 (<0.1) p=0.07
6269 (71) 6286 (71)
COMPASS GI pain with underlying
Beta blockers, no. (%): 6137 (70) Beta blockers, no. (%): 6122 (70) HR (95% CI): 0.57 (0.17-1.95)
Moayyedi 2019 multiple gastroduodenal 7 (<0.1)
Lipid-lowering agents, no. (%): Lipid-lowering agents, no. (%): p=0.37
erosions: 4 (<0.1)
7775 (88) 7823 (89)
Upper GI obstruction or HR (95% CI): 1.32 (0.69-2.52)
16 (0.2)
peroration: 21 (0.2) p=0.41
ACE: angiotensin-converting-enzyme, ARB: angiotensin II receptor blocker, GI: gastrointestinal, Hb: hemoglobin, HR: hazard ratio, MI: myocardial infarction, NSAID: nonsteroidal
anti-inflammatory drug, SSRI: selective serotonin reuptake inhibitor
*Composite of overt bleeding of gastroduodenal organ confirmed by endoscopy or radiography, overt upper GI bleeding of unknown origin, bleeding of presumed occult upper
GI tract origin with documented decrease in Hb of 2 g/dL, symptomatic gastroduodenal ulcer, GI pain with underlying multiple gastroduodenal erosions, and upper GI
obstruction/perforation.
†Includes one gastric cancer in pantoprazole group, no upper GI cancers in the placebo group.

Summary of Key DAPT Trials Incorporated into NMA
Ticagrelor + ASA
PEGASUS-TIMI 54 was a Phase III randomized controlled trial that evaluated the efficacy and safety
of DAPT with ticagrelor (90 mg or 60 mg) + ASA compared to ASA alone.15 Patients were eligible to
participate if they had had a MI one to three years before enrollment, were at least 50 years of age,
and had at least one additional risk factor, including ≥65 years of age, diabetes mellitus, a second
prior MI, multivessel CAD, or chronic renal dysfunction. Eligible patients were randomized 1:1:1 to
receive 90 mg of ticagrelor orally (n=7050), 60 mg of ticagrelor orally (n=7045), or placebo
(n=7067); all patients received 75-150 mg of ASA daily. At baseline, patients had a mean age of 65,
32% had diabetes, 60% had multivessel CAD, and the median time since the qualifying MI was 1.7
years. Background medications at baseline included statins (93%), beta-blockers (82%), and ACE
inhibitors or ARBs (81%). Patients were followed for a median of 33 months (IQR 28-37).15
The two ticagrelor doses (90 mg and 60 mg) significantly reduced the risk of the composite primary
endpoint of cardiovascular death, MI, or stroke. In a pooled analysis of both ticagrelor doses versus
placebo, ticagrelor reduced the risk of the composite endpoint by 16% (HR: 0.84; 95% CI (0.76 to
0.94); p=0.001); results for each dose arm compared to ticagrelor are reported in Table D6. 15 When
stratified by time from MI (<2 years vs. ≥2 years), patients whose MI occurred <2 years before
enrollment appeared to derive a greater benefit from ticagrelor (60 mg) than those who had a MI
≥2 years prior to enrollment (p-value for interaction=0.09; Table D6); results for the 90 mg dose of
ticagrelor were similar in these subgroups.15,16

Table D7. PEGASUS-TIMI 54 Summary Results15
Ticagrelor Ticagrelor Ticagrelor (90 Ticagrelor (60
(90 mg) + (60 mg) + ASA mg) + ASA vs. mg) + ASA vs.
ASA ASA ASA ASA
3-Yr 3-Yr
3-Yr Kaplan-
Kaplan- Kaplan- HR (95% CI) HR (95% CI)
Meier Event
Meier Meier P-Value P-Value
Rate
Event Rate Event Rate
Composite of CV Death, MI, or Stroke
0.85 (0.75-0.96) 0.84 (0.74-0.95)
All Patients 7.9% 7.8% 9.0%
p=0.008 p=0.004
Subgroup: MI <2 Years 0.83 (0.71-0.96) 0.77 (0.66-0.90)
8.2% 7.8% 9.7%
before Enrollment p=NR p=NR
Subgroup: MI ≥2 Years 0.89 (0.73-1.08) 0.96 (0.79-1.17)
7.3% 7.8% 7.9%
before Enrollment p=NR p=NR
TIMI Major Bleeding
2.69 (1.96-3.70) 2.32 (1.68-3.21)
All Patients 2.6% 2.3% 1.1%
p<0.001 p<0.001
Subgroup: MI <2 Years
2.4% 2.4% 1.2% 2.18 (1.48-3.23) 2.05 (1.38-3.03)
before Enrollment
Subgroup: MI ≥2 Years
2.9% 2.2% 0.7% 4.15 (2.34-7.36) 3.17 (1.76-5.70)
before Enrollment
ASA: aspirin, HR: hazard ratio, CI: confidence interval, CV: cardiovascular, MI: myocardial infarction, NR: not
reported, TIMI: Thrombolysis in Myocardial Infarction
The PEGASUS trial’s primary safety endpoint was Thrombolysis in Myocardial Infarction (TIMI)
major bleeding, defined as intracranial bleeding, clinically overt signs of hemorrhage (drop in
hemoglobin ≥5 g/dL or fall in hematocrit ≥15%), or a bleeding event that led to death within seven
days.15 Both doses of ticagrelor significantly increased the rate of major bleeding (90mg: 2.6%;
60mg: 2.3%) versus ASA (1.1%; p<0.001 for both comparisons, see Table D6). Bleeding led to the
discontinuation of study drug in significantly more patients treated with ticagrelor (7.8% and 6.2%
for the 90 mg and 60 mg arms, respectively) versus ASA (1.5%; p<0.001 for both comparisons).
Clopidogrel + ASA
CHARISMA was a Phase III, multicenter, double-blind study evaluating the efficacy and safety of 75
mg of clopidogrel + ASA (75-162 mg/day; n=7802) versus ASA alone (n=7801).17 Patients were
eligible for the study if they were at least 45 years of age and had multiple atherothrombotic risk
factors, documented coronary disease, cerebrovascular disease, or symptomatic PAD. At baseline,
the median age of patients was 64, 42% had diabetes, 78% had documented vascular disease, and
34% had a prior MI. Background medications at baseline included statins (77%), beta-blockers
(55%), ACE inhibitors (46%), and ARBs (26%). Patients were followed for a median of 28 months.

Clopidogrel + ASA did not significantly reduce the risk of the primary endpoint, a composite of MI,
stroke, or cardiovascular death, in the overall trial population (Table D7). However, in a post hoc
subgroup analysis of higher-risk patients (i.e., patients with a prior MI, stroke, or symptomatic PAD)
the rate of MI, stroke or cardiovascular death was statistically significantly lower with clopidogrel +
ASA (7.8%) versus ASA alone (8.8%; HR: 0.83; 95% CI 0.72 to 0.96; p=0.010); results were consistent
in the subgroup with a prior MI (Table D7).18
The CHARISMA trial’s primary safety endpoint was Global Utilization of Streptokinase and Tissue
Plasminogen Activator for Occluded Coronary Arteries (GUSTO)-defined severe bleeding; fatal
bleeding events, intracranial hemorrhages, or bleeding that caused hemodynamic compromise
requiring blood or fluid replacement, inotropic support, or surgical intervention were considered
severe.17 The rate of severe bleeding was 1.7% for clopidogrel + ASA versus 1.3% for ASA alone
(relative risk [RR] 1.25 (95% CI 0.97 to 1.61; p=0.09); moderate bleeding, i.e., bleeding that led to a
transfusion and a statistically significant increase in moderate bleeding but did not meet the severe
definition, occurred in 2.1% of the clopidogrel + ASA arm and 1.3 % of the ASA alone arm (RR: 1.62;
95% CI 1.25 – 2.08; p<0.001).
Table D8. CHARISMA Trial Summary Results17,18

Clopidogrel + ASA ASA Clopidogrel + ASA vs. ASA
HR (95% CI)
% %
P-Value
Composite of CV Death, MI, or Stroke
[RR] 0.93 (0.83-1.05)
All Patients 6.8% 7.3%
p=0.22
Subgroup: Prior MI, Stroke, or 0.83 (0.72-0.96)
7.8% 8.8%
Symptomatic PAD p=0.010
0.77 (0.61-0.98)
Subgroup: Prior MI 6.6% 8.3%
p=0.031
Severe of Life-Threatening Bleeding
[RR] 1.25 (0.91-1.61)
All Patients 1.7% 1.3%
p=0.09
Subgroup: Prior MI, Stroke, or 1.11 (0.81-1.54)
1.7% 1.5%
Symptomatic PAD p=0.509
ASA: aspirin, CI: confidence interval, CV: cardiovascular, HR: hazard ratio, MI: myocardial infarction, PAD:
peripheral artery disease, RR: relative risk

NMA Supplemental Information
Table D9. Data Inputs for NMA of Composite Cardiovascular Death, MI, or Stroke in Patients with
a Recent MI
Trial Regimen Hazard Ratio 95% CI
58
COMPASS Rivaroxaban + ASA vs. ASA 0.70 0.48-1.01
Ticagrelor (60 mg) + ASA vs. 0.77 0.66-0.90
PEGASUS15
ASA
CHARISMA18 Clopidogrel + ASA vs. ASA 0.77 0.61-0.98
Figure D1. Network Diagram for NMA in Patients with a Recent MI
Riv +
Clop +
ASA
ASA
ASA
Tic +
ASA

Evidence Tables for the Review of Icosapent Ethyl
Table D10. Study Quality of REDUCE-IT

Outcome Non- Appropriate
Comparable Adequate Patient Physician ITT Overall
Study Adjudication Differential Handling of
Groups Randomization Blinding Blinding Analysis Quality
Blinding Follow-Up Missing Data
REDUCE-IT Yes Yes Yes Yes Yes Yes Yes Yes Good
ITT: intention-to-treat

Table D11. Study Design of REDUCE- IT
Study REDUCE-IT (Reduction of Cardiovascular Events with Icosapent Ethyl—Intervention Trial)
Design Phase IIIb, multicenter, double-blind, placebo-controlled, randomized trial with parallel assignment
General Criteria
• ≥45 years with established CVD (Secondary Prevention Cohort) OR
≥50 years with diabetes mellitus and ≥1 additional risk factor for CVD (Primary Prevention Cohort)
• Fasting triglyceride levels ≥135 mg/dL and <500 mg/dL
• LDL-C >40 mg/dL and ≤100 mg/dL
• On stable statin therapy (with or without ezetimibe) for ≥4 weeks
Secondary Prevention Cohort (≥45 years with ≥1 of the following):
• Documented CAD (≥1 of the following):
o Documented multivessel CAD (≥50% stenosis in ≥2 major arteries, with or without revascularization)
o Documented prior MI
o Hospitalization for high risk NSTE-ACS
• Documented cerebrovascular or carotid disease (≥1 of the following):
o Prior ischemic stroke
o Symptomatic carotid artery disease with ≥50% carotid arterial stenosis
Inclusion Criteria o Asymptomatic carotid artery disease with ≥70% carotid arterial stenosis
o History of carotid revascularization
• Documented PAD (≥1 of the following):
o Ankle-brachial index <0.9 with symptoms of intermittent claudication or
o History of aortoiliac or peripheral arterial intervention
Primary Prevention Cohort (≥50 years with diabetes mellitus and ≥1 of the following additional risk factors for CVD):
• Men ≥55 years and women ≥65 years
• Cigarette smoker within three months
• Hypertension (SBP ≥140 mm Hg OR DBP ≥90 mm Hg)
• HDL-C ≤40 mg/dL for men or ≤50 mg/dL for women
• hsCRP >3 mg/L
• Renal dysfunction
• Retinopathy
• Micro- or macroalbuminuria
• Ankle-brachial index <0.9 without symptoms of intermittent claudication
• Severe (class IV) heart failure
• Any life-threatening disease expected to result in death within the next two years (other than CVD)
Exclusion Criteria
• Active severe liver disease
• History of pancreatitis

• HbA1c >10%
• Poorly controlled hypertension (SBP ≥200 mm Hg OR DBP ≥100 mm Hg) despite antihypertensive treatment
• Planned coronary intervention or any other major surgical procedure
• Known hypersensitivity to fish or shellfish, or ingredients of the study treatments
• Use of non-statin lipid-altering medications, dietary supplements, or foods (e.g., niacin, PCSK9 inhibitors, products containing
omega-3 fatty acids) unless washout
N 8,179
• Icosapent ethyl 4 g/day (n=4,089)
Interventions
• Placebo (n=4,090)
Follow-Up Median 4.9 years (maximum 6.2 years)
• Primary Endpoint: A composite of CV death, nonfatal MI, nonfatal stroke, coronary revascularization, or hospitalization for
unstable angina
• Key Secondary Endpoint: A composite of CV death, nonfatal MI, or nonfatal stroke
• Other Secondary Efficacy Endpoints:
o A composite of CV death or nonfatal MI
o A composite of death from any cause, nonfatal MI, or nonfatal stroke
o Death from any cause
o CV death
o Fatal or nonfatal MI
o Fatal or nonfatal stroke
Outcomes o Urgent or emergency revascularization
o Hospitalization for unstable angina
• Tertiary/Exploratory Efficacy Endpoints (selected):
o Total (first and subsequent) CV events
o New coronary heart failure
o Transient ischemic attack
o Amputation for peripheral vascular disease
o Carotid revascularization
o New onset type 2 diabetes or hypertension
o Cardiac arrhythmias requiring hospitalization ≥24 h
o Cardiac arrest
CAD: coronary artery disease, CVD: cardiovascular disease, DBP: diastolic blood pressure, HbA1c: glycated hemoglobin. HDL-C: high-density lipoprotein-
cholesterol, hsCRP: high-sensitivity C-reactive protein, LDL-C: low-density lipoprotein- cholesterol; MI: myocardial infarction, NSTE-ACS: non–ST-segment
elevation acute coronary syndrome, PAD: peripheral artery disease, SPB: systolic blood pressure

Table D12. Key Baseline Characteristics of REDUCE-IT19,20
Characteristic Icosapent Ethyl 4 g/d (n=4089) Placebo (n=4090)
Demographics and Cardiovascular Risk
Age, Median (IQR) 64.0 (57.0-69.0) 64.0 (57.0-69.0)
BMI, Median (IQR) 30.8 (27.8-34.5) 30.8 (27.9-34.7)
Secondary Prevention, No. (%) 2892 (70.7) 2893 (70.7)
Primary Prevention, No. (%) 1197 (29.3) 1197 (29.3)
Type 2 Diabetes, No. (%) 2367 (57.9) 2363 (57.8)
Prior MI, No. (%) 1938 (47.4) 1881 (46.0)
Prior Unstable Angina, No. (%) 1017 (24.9) 1015 (24.8)
Hypertension, No. (%) 3541 (86.6) 3543 (86.6)
Concomitant Medications
High Statin Intensity, No. (%) 1290 (31.5) 1226 (30.0)
Moderate Statin Intensity, No. (%) 2533 (61.9) 2575 (63.0)
Low Statin Intensity, No. (%) 254 (6.2) 267 (6.5)
Ezetimibe, No. (%) 262 (6.4) 262 (6.4)
Anti-Diabetic, No. (%) 2190 (53.6) 2196 (53.7)
Anti-Hypertensive, No. (%) 3895 (95.3) 3895 (95.2)
Anti-Platelet, No. (%) 3257 (79.7) 3236 (79.1)
ACE Inhibitor, No. (%) 2112 (51.7) 2131 (52.1)
ARB, No. (%) 1108 (27.1) 1096 (26.8)
Beta-Blocker, No. (%) 2902 (71.0) 2880 (70.4)
Laboratory Parameters
hsCRP, Median (IQR) 2.2 (1.1-4.5) 2.2 (1.1-4.5)
TG, Median (IQR) 216.5 (176.5-272.0) 216.0 (175.5-274.0)
HDL-C, Median (IQR) 40.0 (34.5-46.0) 40.0 (35.0-46.0)
LDL-C, Median (IQR) 74.0 (61.5-88.0) 76.0 (63.0-89.0)
EPA (μg/mL), Median (IQR) 26.1 (17.1-40.1) 26.1 (17.1-39.9)
ARB: angiotensin II receptor blockers; BMI: body mass index, EPA: eicosapentaenoic acid, HDL-C: high-density
lipoprotein- cholesterol, hsCRP: high-sensitivity c-reactive protein, IQR: interquartile range, LDL-C: low-density
lipoprotein- cholesterol, No.: number of patients, TG: triglyceride

Table D13. Key Efficacy Outcomes in REDUCE-IT19 at a Median of 4.9 Years of Follow-Up
Icosapent Ethyl 4
Outcome Placebo (n=4090) HR (95% CI) P-Value
g/d (n=4089)
Primary Outcome
CV Death, Nonfatal MI, Nonfatal Stroke,
705 (17.2) 901 (22.0) 0.75 (0.68-0.83) <0.001
Revascularization, or Unstable Angina, No. (%)
Individual Components of Primary Outcome
CV Death, No. (%) 174 (4.3) 213 (5.2) 0.80 (0.66-0.98) 0.03
Nonfatal MI, No. (%) 237 (5.8) 332 (8.1) 0.70 (0.59-0.82) NR
Nonfatal Stroke, No. (%) 85 (2.1) 118 (2.9) 0.71 (0.54-0.94) NR
Revascularization, No. (%) 376 (9.2) 544 (13.3) 0.66 (0.58-0.76) NR
Hospitalization for Unstable Angina, No. (%) 108 (2.6) 157 (3.8) 0.68 (0.53-0.87) 0.002
Key Secondary Outcome
CV Death, Nonfatal MI, or Nonfatal Stroke, No. (%) 459 (11.2) 606 (14.8) 0.74 (0.65-0.83) <0.001
Other Secondary Outcomes
CV Death or Nonfatal MI, No. (%) 392 (9.6) 507 (12.4) 0.75 (0.66-0.86) <0.001
Death from Any Cause, Nonfatal MI, or Nonfatal
549 (13.4) 690 (16.9) 0.77 (0.69-0.86) <0.001
Stroke, No. (%)
Death from Any Cause, No. (%) 274 (6.7) 310 (7.6) 0.87 (0.74-1.02) NS
Fatal or Nonfatal MI, No. (%) 250 (6.1) 355 (8.7) 0.69 (0.58-0.81) <0.001
Fatal or Nonfatal Stroke , No. (%) 98 (2.4) 134 (3.3) 0.72 (0.55-0.93) 0.01
Emergency or Urgent Revascularization, No. (%) 216 (5.3) 321 (7.8) 0.65 (0.55-0.78) <0.001
CV: cardiovascular, CI: confidence interval, HR: hazard ratio, No. number of patients, NS: not significant, MI: myocardial infarction

Table D14. Key Safety Events in REDUCE-IT19 at a Median of 4.9 Years of Follow-Up
Icosapent Ethyl 4 g/d
Event Placebo (n=4090) P-Value
(n=4089)
Any TEAE, No. (%) 3343 (81.8) 3326 (81.3) 0.63
Serious TEAE, No. (%) 1252 (30.6) 1254 (30.7) 0.98
TEAE Leading to Withdrawal of Study Drug, No. (%) 321 (7.9) 335 (8.2) 0.60
Serious TEAE Leading to Withdrawal of Study Drug, No. (%) 88 (2.2) 88 (2.2) 1.00
Serious TEAE Leading to Death, No. (%) 94 (2.3) 102 (2.5) 0.61
Bleeding TEAEs
Any Bleeding-Related Disorder , No. (%) 111 (2.7) 85 (2.1) 0.06
GI Bleeding, No. (%) 62 (1.5) 47 (1.1) 0.15
Central Nervous System Bleeding, No. (%) 14 (0.3) 10 (0.2) 0.42
Other Bleeding, No. (%) 41 (1.0) 30 (0.7) 0.19
AEs occurring ≥5%
Peripheral Edema, No. (%) 267 (6.5) 203 (5.0) 0.002
Atrial Fibrillation, No. (%) 215 (5.3) 159 (3.9) 0.003
Constipation, No. (%) 221 (5.4 ) 149 (3.6) <0.001
Diarrhea, No. (%) 367 (9.0) 453 (11.1) 0.002
Anemia, No. (%) 191 (4.7) 236 (5.8) 0.03
Back Pain, No. (%) 335 (8.2) 309 (7.6) 0.29
Hypertension, No. (%) 320 (7.8) 344 (8.4) 0.35
Nasopharyngitis, No. (%) 314 (7.7) 300 (7.3) 0.56
Arthralgia, No. (%) 313 (7.7) 310 (7.6) 0.9
Upper Respiratory Tract Infection, No. (%) 312 (7.6) 320 (7.8) 0.77
Bronchitis, No. (%) 306 (7.5) 300 (7.3) 0.8
Chest pain, No. (%) 273 (6.7) 290 (7.1) 0.48
Pneumonia, No. (%) 263 (6.4) 277 (6.8) 0.56
Influenza, No. (%) 263 (6.4) 271 (6.6) 0.75
Dyspnea, No. (%) 254 (6.2) 240 (5.9) 0.52
Urinary Tract Infection, No. (%) 253 (6.2) 261 (6.4) 0.75
Cough, No. (%) 241 (5.9) 241 (5.9) 1.00
Osteoarthritis, No. (%) 241 (5.9) 218 (5.3) 0.27
Dizziness, No. (%) 235 (5.7) 246 (6.0) 0.64
Pain in Extremity, No. (%) 235 (5.7) 241 (5.9) 0.81
Cataract, No. (%) 233 (5.7) 208 (5.1) 0.22
Fatigue, No. (%) 228 (5.6) 196 (4.8) 0.11
AE: adverse event, GI: gastrointestinal, No.: number of patients, TEAE: treatment-emergent adverse event

Table D15. Key Safety Events in MARINE66 and ANCHOR67 at 12 Weeks
MARINE ANCHOR
Icosapent Ethyl 4 g/d (n=77) Placebo (n=76) Icosapent Ethyl 4 g/d (n=233) Placebo (n=233)
Any TEAE, No. (%) 27 (35.1) 28 (36.8) 106 (45.5) 112 (48.1)
Serious Adverse Event, No. (%) 1 (1.3) 0 7 (3.0) 5 (2.1)
TEAE Leading to Withdrawal of
0 3 (3.9) 5 (2.1) 12 (5.2)
Study Drug, No. (%)
Musculoskeletal and Connective
NR NR 18 (7.7) 10 (4.3)
Tissue Disorders, No. (%)
Arthralgia, No. (%) NR NR 4 (1.7) 1 (0.4)
GI Disorders, No. (%) NR NR 27 (11.6) 40 (17.2)
Diarrhea, No. (%) 1 (1.3) 5 (6.6) 8 (3.4) 10 (4.3)
Nausea, No. (%) 1 (1.3) 4 (5.3) 5 (2.1) 7 (3.0)
Eructation, No. (%) 0 3 (3.9) 2 (0.9) 4 (1.7)
Infections and Infestations, No. (%) NR NR 31 (13.3) 38 (16.3)
Nasopharyngitis, No. (%) NR NR 1 (0.4) 7 (3.0)
GI: gastrointestinal, No.: number of patients, TEAE: treatment-emergent adverse event

Appendix E. Comparative Value Supplemental
Information
Table E1. Impact Inventory
Included in This Analysis Notes on
from… Perspective? Sources (if
Type of Impact Quantified),
Sector
(Add Additional Domains, as Relevant) Health Care Likely
Societal
Sector Magnitude &
Impact (if Not)
Formal Health Care Sector
Longevity effects X X
Health
Health-related quality of life effects X X
Outcomes
Adverse events X X
Paid by third-party payers X X
Paid by patients out-of-pocket  
Medical Costs
Future related medical costs  
Future unrelated medical costs  
Informal Health Care Sector
Patient time costs NA 
Health-
Unpaid caregiver time costs NA 
Related Costs
Transportation costs NA 
Non-Health Care Sectors
Labor market earnings lost NA X
Cost of unpaid lost productivity due to
NA X
Productivity illness
Cost of uncompensated household
NA 
production
Consumption Future consumption unrelated to health NA 
Cost of social services as part of
Social Services NA 
intervention
Legal/Criminal Number of crimes related to intervention NA 
Justice Cost of crimes related to intervention NA 
Impact of intervention on educational
Education NA 
achievement of population
Housing Cost of home improvements, remediation NA 
Production of toxic waste pollution by
Environment NA 
intervention
Other Other impacts (if relevant) NA 
NA: not applicable
Adapted from Sanders et al.117

Description of the evLYG Calculations
The cost per evLYG considers any extension of life at the same “weight” no matter what
treatment is being evaluated. Below are the stepwise calculations used to derive the evLYG.
i) First, we attribute a utility of 0.851, the age- and gender-adjusted utility of the
general population in the U.S. that are considered healthy.118
ii) For each cycle (cycle i) in the model where using the intervention results in
additional years of life gained, we multiply this general population utility with the
additional life years gained (ΔLYG).
iii) We sum the product of the life years and average utility (cumulative LYs/cumulative
QALYs) for cycle i in the comparator arm with the value derived in step 2 to derive
the equal value of life years (evLY) for that cycle.
iv) If no life years were gained using the intervention versus the comparator, we use the
conventional utility estimate for that cycle i.
v) The total evLY is then calculated as the cumulative sum of QALYs gained using the
above calculations for each arm.
vi) We use the same calculations in the comparator arm to derive its evLY.
vii) Finally, the evLYG is the incremental difference in evLY between the intervention
and the comparator arms.
Table E2. Model Cohort Characteristics for DAPT Scenario Analysis

Characteristic Clopidogrel + ASA ASA Alone Source
Age, Yrs., Median 64 64
Male (%) 72.7 73.1
Smoking (% Yes) 21.6 22.2
Diabetes (% Yes) 30.8 31.3
Prior MI 40.2 41.0
Prior Ischemic Stroke 34.5 34.0
Congestive Heart Failure 6.3 6.5
Atrial Fibrillation 3.6 3.4 18
PAD 32.3 32.3
Hypertension 68.3 69.9
Hypercholesterolemia 69.8 70.5
Angiotensin II Receptor 22.1
21.9
Blockers
Calcium-Channel Blocker 33.8 34.7
Diuretic 44.4 44.0
Beta-Blocker 55.7 56.8
From CHARISMA post-hoc analysis in patients with prior MI, documented prior ischemic stroke, or
symptomatic PAD.

Table E3. Clinical Inputs and Costs for DAPT Scenario Analysis
Rivaroxaban + ASA Clopidogrel + ASA Rivaroxaban + ASA vs.
Input
vs. ASA vs. ASA Clopidogrel + ASA
Estimated HR for
Composite Primary 0.70 (0.48-1.02) 0.77 (0.61-0.98) 0.91 (0.58-1.40)
Endpoint*
WAC per Dose Price per Year
$0.18 per 75mg
Clopidogrel* $65.70
tablet
*A network meta-analysis was conducted as part of the clinical review to estimate the comparative risk of a
composite endpoint of cardiovascular death, stroke, or MI between each of the regimens of focus.
ꝉDosage 75mg/day. Wholesale acquisition cost (WAC) per Redbook®.33
Table E4. Efficacy Estimates for DAPT Scenario Analysis

Clopidogrel +
ASA Alone P-
Parameter ASA HR (95% CI) Source
n (%) Value
n (%)
Composite Primary
Endpoint: Stroke, CV 347 (7.3) 416 (8.8) 0.829 (0.719-0.956) 0.010
Death, MI
Stroke* 144 (3.0) 179 (3.8) 0.802 (0.644–0.998) 0.048 18
Ischemic Stroke* 126 (2.7) 152 (3.2) 0.828 (0.654-1.048) 0.115
CV Death 142 (3.0) 163 (3.4) 0.870 (0.695-1.090) 0.224
MI* 117 (2.5) 145 (3.1) 0.805 (0.631-1.027) 0.080
Severe Bleeding 79 (1.7) 71 (1.5) 1.114 (0.808-1.535) 0.509
*Fatal plus nonfatal events.
The CHARISMA trial had a median follow-up of 28 months.

Table E5. Weighted Average Stroke Utility Estimate using COMPASS Control Arm
Stroke Severity, Control
Stroke Control
Modified Ranking mRS Score Description Arm Source
Utility* Arm (n)
Scale (mRS) Weight
mRS score 0 No symptoms at all 0.92 29 0.2057
No significant disability
despite symptoms; able
mRS score 1 0.85 32 0.2270
to carry out all usual
duties and activities
Slight disability; unable to
carry out all previous
mRS score 2 activities, but able to look 0.77 27 0.1915
after own affairs without
assistance
Moderate disability;
requiring some help, but
mRS score 3 0.64 16 0.1135
able to walk without 14,95
assistance
Moderately severe
disability; unable to walk
without assistance and
mRS score 4 0.41 18 0.1277
unable to attend to own
bodily needs without
assistance
Severe disability;
bedridden, incontinent,
mRS score 5 and requiring constant 0.14 7 0.0496
nursing care and
attention
mRS score 6 Dead 12 0.0851
Weighted Stroke
Utility Value 0.6615
*These utility values were elicited from the general US population using the time trade-off method and EQ-5D-
3L.

Table E6. Model Comparison Results
Cumulative
Incidence of Total Incremental Cost Incremental
Model ICER (per QALY)
Events at 36 (Discounted) QALYs
Months‡
Model with Fonarow et
14.6% $120,307 0.28 $435,169
al. Inputsꝉ
Fonarow et al. Model
14.9% $139,817 0.34 $413,579
Findings*
*Table 3, FOURIER trial participants, full annual list price of $14,500. 107
†Inputs changed include baseline risk to FOURIER trial participants, health state costs, intervention effects,
baseline demographics and lipid panels, discontinuation. Indirect costs were not included, and health state
utility input calculations differed between models.
‡Sabatine et al. Figure 2A.119
Table E7. Rivaroxaban Model Calibration

Sum of First MI, First
First First
Model Stroke, and CV CV Death*
MI Stroke
Death
Model Estimate at 2
Years (3 Cycles in the
6% 2.2% 1.5% 2.2%
Model for Death; 2 Cycles
for MI, Stroke)
Trial-Based Findings at 23
6% 2.2% 1.6% 2.2%
Months
*From Eikelboom NEJM 2017.11
Table E8. Icosapent Ethyl Model Calibration

Sum of First MI,
First First
Model First Stroke, and CV CV Death*
MI Stroke
Death*
Model Estimate at 5
Years (6 Cycles in the
16.1% 8.2% 2.7% 5.2%
Model for Death; 5 Cycles
for MI, Stroke)
Trial-Based Cumulative
Incidence Sum of
16.2% 8.1% 2.9% 5.2%
Composite Endpoints at 5
Years
*From Bhatt NEJM 2019 (approximate).92
†Bhatt NEJM 2019 Supplementary Figure 3.92

Appendix F. Public Comment
This section includes summaries of the public comments prepared for the Midwest CEPAC Public
Meeting on September 26, 2019 in St. Louis, MO. These summaries were prepared by those who
delivered the public comments at the meeting and are presented in order of delivery.
A video recording of all comments can be found here, beginning at minute 01:28:55. Conflict of
interest disclosures are included at the bottom of each statement for each speaker who is not
employed by a pharmaceutical manufacturer.
Alina Kolomeyer, Pharm.D.

Associate Director, Corporate Alliances at Amarin
On behalf of my colleagues at Amarin Pharma, Inc, thank you for the opportunity to submit
comments for the record following the September 26, 2019 Midwest Comparative Effectiveness
Public Advisory Council (CEPAC) Institute for Clinical and Economic Review’s (ICER) Public
Meeting on additive therapies for cardiovascular (CV) disease. We thank ICER and the Midwest
CEPAC for the opportunity to meet with experts in the field, the public, and fellow
manufacturers, and to share Amarin’s position on ICER’s Draft Evidence Report.
Amarin takes very seriously its mission of developing innovative, life-saving, and quality-of-life
enhancing medicines for patients and in conducting rigorous, high-quality science to clearly
evaluate their benefits.
While statin use and optimal medical management have reduced cardiovascular risk and
improved patient health outcomes over the last few decades, substantial residual cardiovascular
risk remains. This residual risk represents an important unmet medical need for patients and
accounts for significant health care and economic burden.
With regard to ICER’s economic analysis, Amarin believes that it’s essential to use all of the
evidence from REDUCE-IT in the economic analysis, including (1) using all five components of
the prespecified primary endpoint in the Markov model base-case analysis and (2) using total
event data. We believe it is beneficial to use a patient-level, microsimulation approach to
modeling as well as incorporating more recent US national cost data for cardiovascular events.
In REDUCE-IT, 16 hospitalizations for unstable angina and 76 coronary revascularization

procedures were prevented per 1000 patients treated over a five-year period. Despite ICER’s
analysis reaching favorable conclusions based on select events, these data represent clinically
meaningful and statistically significant benefits for patients, as well as cost offsets for payers,
and should be included in the base-case cost-effectiveness and budget impact analyses due to
the high quality of this scientific evidence.

We also believe that recurrent long-term, downstream cardiovascular events after the first
cardiovascular event involve real costs to patients, payers, and society, and need to be included
in the model. Each component of the primary endpoint was reduced by icosapent ethyl therapy
in a statistically significant, clinically meaningful, and economically relevant manner.
On behalf of Amarin, we again thank ICER and the Midwest CEPAC for the opportunity to submit
comments and to be part of an open dialogue around the clinical and economic benefits of
icosapent ethyl to patients, payers, and society. Amarin is dedicated to working with relevant
stakeholders to ensure access for eligible and appropriate patients to innovative, life-saving, and
quality-of-life-enhancing therapies such as icosapent ethyl.
Marie Warshauer, M. Ed.

Former Program Director of WomenHeart: the National Coalition for Women with Heart
Disease
Why am I here? Eighteen years ago my life changed. In my early 40’s I was busy raising three
boys, exercising, eating healthy, just living a busy life, however my heart had other ideas. After
several trips to my GP, complaining of being exhausted, having numbness in my left arm and
feeling short of breath, I was diagnosed with Coronary Artery Disease. I had 99% blockage in my
LAD. Angioplasty was preformed unexpectedly the day of my stress test. Five months later
stents had closed, bypass surgery followed by several months of rehab all while adding on to our
home. My life, my family’s life had changed.
Why am I here? As one of the first women to participate in WomenHeart’s Science and
Leadership Symposium, I learned about the science of my disease as well as understanding that
it was also a disease that touches your confidence, emotional balance, and depth of
understanding for what I am now living with. Understanding that I would forever be on
medications that might cause side effects, cause damage to other organs, make a difference in
so many ways other than how it helps my heart was just the start. Through WomenHeart I
learned the value of participating in drug trials and research studies. With patient help, the
research might be of value to so many and not just me.
Why am I here? In 2005 just as we are getting ourselves settled after my diagnosis, my husband
is diagnosed with a meningioma. His easy surgery was anything but. After a month in the
hospital and weeks in a rehab center learning to walk, swallow, and talk again, our home
became a full time health care center. I was no longer the patient I had become the caregiver.
The navigation of the medical world was truly an eye opening experience. My husband endured
10 years of medical chaos and disconnected care. We quickly learned the ropes and had the
where with all to be able to seek additional advice. Our nonstop research and inquiring about
treatments or to try new drugs and explore how we could keep David with us longer kept us
going. We sought out holistic treatments, acupuncture, therapeutic massages, exercise and
meditation all in the name of trying to make a difference.

Why am I here? Because I have experienced first hand the side effects and challenges of taking
various drugs. Emotional impact of taking medicines include being afraid to change drugs or
brands of
drugs, dose adjustments, lack of sleep or being constantly tired. Stomach problems, headaches,
and weight gain, long-term effects. The drugs caused emotional ups and down’s. Remembering
to take them at the right time. The expense.
The effects of the variety of drugs my husband was given changed his personality, made him
grumpy, made a difference on what he could and wanted to eat. In so many ways it controlled
his life. Every time he went to the doctor there were new prescriptions. What do we do with the
old medicine we have hardly used? How does someone who is ill keep track of when they are to
take what drug? How do patients know if drug A or B causing the side effect? How does a drug
interact with other meds? How will it help me?
I am here today to make a small contribution to something I would hope helps others navigate
the world of medical diagnosis and the drugs we take to make us feel better.
Marie Warshauer has no conflicts of interest to disclose.
Taylor Kelly
Law Clerk at DCBA Law & Policy, Policy Advisor at Aimed Alliance
Aimed Alliance is pleased that the Institute for Clinical and Economic Review (ICER) has deemed
rivaroxaban and icosapent ethyl to be cost effective. Yet, we remain concerned that its analysis
and recommendations will impact patient access by encouraging the use of restrictive policies
on plan formularies, such as prior authorization and step therapy, to control costs. Such policies
can be unethical and inconsistent with standards of care, interfere with the patient-practitioner
relationship, and result in significant delays to prescribed treatments.
Moreover, ICER’s analysis does not appear to incorporate several important considerations into
its potential budget impact analysis and value-based benchmark prices, such as data that
represents the patient perspective; data from two of the REDUCE-IT trial’s primary endpoints –
coronary revascularization and hospitalization; and adequate data about the risk of subsequent
adverse events. We believe that the exclusion of such pertinent data does not adequately
capture the full value that these therapies provide to patients.
Finally, Aimed Alliance reiterates its long-standing opposition to the use of QALYs. QALYs can
result in inappropriate health care rationing and discrimination based on age and health status,
unfairly favoring younger and healthier populations. QALYs put a price tag on the value of a
human life that merely reflects the individual’s diagnosis and deems those with chronic,
debilitating, and rare conditions, such as cardiovascular disease, as being worth less than the
rest of us.

Taylor Kelly is an employee of DCBA Law & Policy, a law firm whose practice is focused on
corporate, drug and device, and health law. The majority of DCBA’s clients are health care
companies, health care professionals, biopharmaceutical companies, and not-for-profit
organizations, and it consequently receives more than 25% of its funding from these
organizations and individuals. Aimed Alliance is a client of DCBA.

Appendix G. Conflict of Interest Disclosures
Tables G1 through G3 contain conflict of interest (COI) disclosures for all participants at the
September 26, 2019 Public meeting of the Midwest CEPAC.
Table G1. ICER Staff and Consultant COI Disclosures

Name Organization Disclosures
Ellie Adair, MPA Institute for Clinical and Economic Review *
Received funding from ICER for
Daniel A. Ollendorf, PhD Tufts Medical Center
this evaluation and report
Jonathan Campbell, PhD University of Colorado *
Katherine Fazioli Institute for Clinical and Economic Review *
Serina Herron-Smith Institute for Clinical and Economic Review *
R. Brett McQueen University of Colorado *
Steve Pearson, MD, MSc Institute for Clinical and Economic Review *
Michelle Poulin Institute for Clinical and Economic Review *
David Rind, MD, MSc Institute for Clinical and Economic Review *
Patty Synnott, MALD, MS Institute for Clinical and Economic Review *
Table G2. Midwest CEPAC Panel Member COI Disclosures

Saint Louis University Center for Health Outcomes *
Eric Armbrecht, PhD Research, School of Medicine and College for Public
Health & Social Justice
Bijan Borah, PhD Mayo Clinic College of Medicine and Science *
Aaron Carroll, MD, MS Indiana University School of Medicine *
Donald Casey, MD, MPH, MBA Medicision *
Gregory Curfman, MD JAMA *
Jill Johnson, PharmD University of Arkansas for Medical Sciences *
Center for Health Economics and Policy, Washington *
Timothy McBride, PhD
University in St. Louis
Rachel Sachs, JD, MPH Washington University in St. Louis
Kurt Vanden Bosch, PharmD St. Luke’s Health System *
Minnesota Evidence-Based Synthesis Program, *
Timothy Wilt, MD, MPH Minneapolis VA Center for Chronic Disease Outcomes
Research
Integrated Health Associates, St Joseph Mercy Health *
Stuart Winston
System
* No relevant conflicts of interest to disclose, defined as more than $10,000 in healthcare company stock or
more than $5,000 in honoraria or consultancies during the previous year from health care manufacturers or
insurers.

Table G3. Policy Roundtable Participant COI Disclosures
Mended Hearts has received more
Andrea Baer, MS, MCPA Mended Hearts than $5,000 in funding from Janssen
and AstraZeneca
Full-time employee of UMPC Health
Chester B. Good, MD, MPH UPMC Health Plan
Plan.
Full-time employee of Amarin
Craig Granowitz Amarin Corporation
Corporation
Kayla Lesser IngenioRx Full-time employee of IngenioRx.
University of Michigan Medical *
Jeremy Sussman, MD, MS
School
WomenHeart, The National Coalition *
Marie Warshauer
for Women with Heart Disease
Harvard Medical School Received speaking fees (<$5,000) for
Jason Wasfy, MD, MPhil Quality and Outcomes Research participation at the iHEAR
Massachusetts General Hospital conference sponsored by Biotronik
* No relevant conflicts of interest to disclose, defined as more than $10,000 in healthcare company stock or
more than $5,000 in honoraria or consultancies during the previous year from health care manufacturers or
insurers.


Amarin ICER CVD Final Evidence Report 101719

Uploaded by

Copyright:

Available Formats

Amarin ICER CVD Final Evidence Report 101719

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Amarin ICER CVD Final Evidence Report 101719

Uploaded by

Copyright:

Available Formats

Additive Therapies for Cardiovascular Disease:

Effectiveness and Value

October 17, 2019

©Institute for Clinical and Economic Review, 2019

Remziye Zaim, PhD

Rick Chapman, PhD, MS

Ellie Adair, MPA

David Rind, MD, MSc

DATE OF PUBLICATION: October 17, 2019

©Institute for Clinical and Economic Review, 2019 Page ii

About Midwest CEPAC

©Institute for Clinical and Economic Review, 2019 Page iii

©Institute for Clinical and Economic Review, 2019 Page iv

©Institute for Clinical and Economic Review, 2019 Page v

©Institute for Clinical and Economic Review, 2019 Page vi

©Institute for Clinical and Economic Review, 2019 Page vii

©Institute for Clinical and Economic Review, 2019 Page viii

©Institute for Clinical and Economic Review, 2019 Page ES1

Insights Gained from Discussions with Patients and Patient Groups

©Institute for Clinical and Economic Review, 2019 Page ES2

Comparative Clinical Effectiveness

Major Adverse Cardiovascular Events

©Institute for Clinical and Economic Review, 2019 Page ES3

Table ES1. Selected Bleeding Outcomes in COMPASS

©Institute for Clinical and Economic Review, 2019 Page ES4

©Institute for Clinical and Economic Review, 2019 Page ES5

Major Adverse Cardiovascular Events

Individual and Total Events

©Institute for Clinical and Economic Review, 2019 Page ES6

Controversies and Uncertainties

©Institute for Clinical and Economic Review, 2019 Page ES7

©Institute for Clinical and Economic Review, 2019 Page ES8

Summary and Comment

Rivaroxaban + ASA versus DAPT

Icosapent Ethyl versus Optimal Medical Management

©Institute for Clinical and Economic Review, 2019 Page ES9

©Institute for Clinical and Economic Review, 2019 Page ES10

Overview and Methods

Figure ES1. Model Framework

Established CVD or high CV event states:

All-cause and CV-

CV: cardiovascular, CVD: cardiovascular disease, MI: myocardial infarction

©Institute for Clinical and Economic Review, 2019 Page ES11

Table ES3. Drug Cost Inputs

©Institute for Clinical and Economic Review, 2019 Page ES12

©Institute for Clinical and Economic Review, 2019 Page ES13

Rivaroxaban $32,000 $35,000 $36,000 $120,000

We estimated threshold treatment prices that would reflect an incremental cost-per-QALY of

©Institute for Clinical and Economic Review, 2019 Page ES14

Table ES8. Cost per evLYG Threshold Analysis Results

Summary and Comment

Potential Other Benefits and Contextual Considerations

Potential Other Benefits

Table ES9. Potential Other Benefits

©Institute for Clinical and Economic Review, 2019 Page ES16

Table ES10. Potential Contextual Considerations

Value-Based Benchmark Prices

©Institute for Clinical and Economic Review, 2019 Page ES17

Potential Budget Impact

©Institute for Clinical and Economic Review, 2019 Page ES18