Economic Evaluation in Genomic Medicine

Economic Evaluation in
GENOMIC MEDICINE
Economic Evaluation in
GENOMIC MEDICINE
VASILIOS FRAGOULAKIS
University of Patras School of Health Sciences,
Department of Pharmacy, Patras, Greece;
National School of Public Health,
Athens, Greece
CHRISTINA MITROPOULOU
Erasmus University Medical Center,
Department of Clinical Chemistry,
Rotterdam, The Netherlands
MARC S. WILLIAMS
Geisinger Health System, Danville,
Pennsylvania, USA
GEORGE P. PATRINOS
University of Patras School of Health Sciences,
Department of Pharmacy, Patras, Greece
AMSTERDAM • BOSTON • HEIDELBERG • LONDON

NEW YORK • OXFORD • PARIS • SAN DIEGO
SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO
Academic Press is an imprint of Elsevier
Academic Press is an imprint of Elsevier
32 Jamestown Road, London NW1 7BY, UK
525 B Street, Suite 1800, San Diego, CA 92101-4495, USA
225 Wyman Street, Waltham, MA 02451, USA
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
Copyright r 2015 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our arrangements with
organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be
found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the
Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research
and experience broaden our understanding, changes in research methods, professional
practices, or medical treatment may become necessary.
Practitioners and researchers may always rely on their own experience and knowledge
in evaluating and using any information, methods, compounds, or experiments described herein. In
using such information or methods they should be mindful of their own safety and the safety of
others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume
any liability for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas
contained in the material herein.
ISBN: 978-0-12-801497-4
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
For information on all Academic Press publications

visit our website at http://store.elsevier.com/
Typeset by MPS Limited, Chennai, India

www.adi-mps.com
Publisher: Christine Minihane

Acquisition Editor: Catherine Van Der Laan
Editorial Project Manager: Lisa Eppich
Production Project Manager: Lucı́a Pérez
Designer: Maria Inês Cruz
PREFACE
The subject of this textbook is the emerging scientific field of applied

economic evaluation of genomic medicine with a focus on cost-
effectiveness analysis, which constitutes the most important part of health
economics from a research standpoint. Given that this field is relatively
new, we thought that its concepts should be presented in a more intuitive
and less formalized manner to familiarize the reader with the field’s tools
and methods.
Because this is an emerging field bridging two distinct and unrelated
disciplines, those of economics and genomic medicine, we have tried to
make our text understandable to a broad audience. In particular, this
book is addressed to the competent clinical and biomedical scientist, stu-
dent, or health care professional who possesses a very basic level of
knowledge about economic evaluation and would like to delve deeper to
be able to evaluate work by others or to apply that knowledge to his or
her own studies. This book is also addressed to the economist and eco-
nomics students who envision acquiring training in health economics
with an emphasis on genomic medicine. In this work, we have focused
on what we thought was necessary with a thorough presentation of basic
concepts and approaches.
The readers of this book should bear in mind that economic evalua-
tion is a highly technical subject that requires a significant investment of
time and effort to be fully understood. We believe that this will become
apparent as one reads the book. At present, economic evaluation is still at
the stage of developing new approaches, whether these are entirely new
or borrowed from other related fields such as statistics, mathematics, or
econometrics. It is our firm belief that only by understanding its method-
ologies, including their weaknesses and capabilities, can we make a true
and valid judgment of the capabilities of the field itself at the theoretical
level. Naturally, we should not forget that, despite the attractive veneer of
objectivity given by the concise and elegant mathematical nomenclature,
the actual subject of financial resource management is in practice funda-
mentally a political problem. Unfortunately, quantitative methods are
inadequate for such problems and, in many cases, the translation of
“knowledge” into “political decision” involves other factors that are
mostly outside the province of the academic community. In this book,
ix
x Preface
however, we have kept the researcher’s point of view and have only
superficially dealt with certain political issues to contribute impartially, we
hope, to the reader’s critical thinking.
With regard to the text itself, we offer this suggestion: while reading,
do not skip any words, symbols, or gene and biomarker names you do
not understand. The reason why most people abandon the study of any
subject is because they skip over a word or a symbol they cannot decipher
and thus become unable to apply or understand what they are reading.1 If
you find that you are having trouble understanding the text, go back, find
a point where you had complete conceptual understanding, and look for
a word whose definition you do not know. As soon as you locate it, use a
suitable dictionary (either an economics dictionary, if the word is a term
of the nomenclature; a biologic or medical dictionary,2 if the term relates
to genetics; or even a conventional dictionary for an “ordinary” word)
and clarify it. Both economic evaluation and genomic medicine use a few
specialized terms that need to be genuinely understood and not just
memorized. You should also keep in mind the fact that evaluating the
data presented here is just as important, if not more, than simply learning
it. As you study this text, we hope that you will remain critical, that you
will remember all that you consider to be true or useful, and that
you will compare the information you read with your own personal
observations.
We hope that the time you spend reading the text will be worth it,
and that it will also give you some ideas for further study or application
of the subject. If you achieve this, then we might say your effort had high
personal utility and was a cost-effective choice for you and for us as well.
The Authors
January 2015
1
http://www.appliedscholastics.org. Applied Scholastics International, a nonprofit educa-
tional organization based in Missouri, was founded by a consortium of American educa-
tors in 1972. Administered by the Association for Better Living and Education, it is
dedicated to the broad implementation of learning tools researched and developed by
American author and educator L. Ron Hubbard.
2
Genetics Home Reference (http://www.ghr.nlm.nih.gov/glossary) and National Human
Genome Research Institute (http://www.genome.gov/glossary).
CHAPTER 1
Economic Evaluation in Health

Care: Evidence-Based Medicine
and Evidence-Based Health
Economics
THE SCOPE OF ECONOMIC EVALUATION OF

HEALTH SERVICES: THE SCIENCE OF ECONOMICS
Economic evaluation of health services is a branch of economics that
deals with “the systematic evaluation of the benefits and costs arising
from the comparison of different health technologies.” Economic eval-
uation is a way of thinking and problem-solving rather than a sterile
set of terms or methods used by health economists. Before moving on
to its subject matter, we should define its general framework, which is
economics itself.
It is a widely held opinion that economics deals exclusively with the
processes of production and distribution of wealth, as well as the properties
and origin of that well-known means of transaction for a society: money.
This approach could be made even broader. The object of economics is pri-
marily the combination of options we should adopt to maximize our wel-
fare under conditions of limited resources. We do not analyze the term
“welfare” from an ethical or social point of view, but the term is used liber-
ally by economists to describe the degree of euphoria, exultation, and
pleasure produced specifically by the consumption of commodities and
services available in the market.
Households, employees, and businesses all face similar “survival”
problems and are forced to make certain choices regarding the use of the
available resources, regardless of whether the society they live in follows a
tender-based or an exchange-based economy. According to economics,
Economic Evaluation in Genomic Medicine © 2015 Elsevier Inc.

All rights reserved. 1
2 Economic Evaluation in Genomic Medicine
the behavior of all these actors should be determined by certain simple

tenets, including:
1. That the economic actor wishes to survive and achieve an infinite
level of welfare (maximize welfare). This does not include people who
do not wish to survive or who do not think rationally. The concept
of “welfare” usually considers the person in an “individualistic” sense,
with no regard to altruism or social responsibility.
2. In the standard case, economics assumes that “greater consumption of
commodities leads to greater welfare,” with no consideration of issues
of saturation.
3. That the combined consumption of commodities and services (that
can be purchased with the available resources) can increase or decrease
welfare depending on the ratio, and that there is a specific mix of
choices that is ideal for each case (because it maximizes welfare),
depending on income and preferences.
4. That resources are limited and economic commodities have a certain cost.
5. That economic stakeholders are willing to sacrifice part of the con-
sumption of a commodity they possess in large quantities to obtain
a small quantity of another desirable commodity that they consume at a
smaller degree. In economics, this is called “the principle of convexity”
because the diagrammatic description of this principle uses a convex
mathematical curve.
Tenet 5 is based on the economics concept that the consumption
of any desirable commodity provides welfare that is great at first, but
gradually decreases (decreasing utility); therefore, it would be wise for
one to refrain from this additional consumption (and the associated
cost) to consume something else, which would give them greater wel-
fare. This principle indicates that, generally speaking, economic stake-
holders avoid “unilateral consumption” and prefer to consume “a
little of each” commodity, but not necessarily in the same ratio.
6. There are specific preferences that can be clearly defined.
From these simple “polished” tenets, we see that achieving infinite
welfare is an unobtainable goal, whereas the maximization of welfare
with the available means is achievable. For example, a worker follows his
or her own maximization pattern and spends any available money to buy
a “basket” of commodities (food, clothes, entertainment, health care,
etc.), balancing market prices against his or her own preferences.
Without going further into the principles of economics, we should also
note that a certain level of welfare can be achieved by different combinations
Economic Evaluation in Health Care 3
of commodities associated with different costs, and that there is a specific

combination that achieves a target level of welfare with minimum cost. For
example, if we were interested in an individual’s welfare, we could identify
several combinations of the commodities “consumption of health care
services” and “available income for other services” that would provide a spe-
cific target level of welfare. Naturally, we could always achieve this by
increasing the consumption of one commodity at the expense of the other,
and vice versa, for each combination of choices. To make such an estimate,
however, the stakeholders’ preferences and the commodities’ prices must be
expressly specified in each case.
By expanding the application of this example, we can see how a society
may decide that it can achieve maximum welfare through a public health
system that provides free health care at the expense of “sacrificing” part of
the citizens’ income, because insurance and taxation costs will take a part
of the income available for consumption to invest it in public health institu-
tions. Other societies with more privately oriented health systems provide
“less public health care” but compensate for this by taking less income,
leaving the rest available for consumption and savings. In accordance with
tenet 5, most societies avoid the extreme choices and do not accept a “fully
private” or “fully public” character for their health care, instead selecting a
mixture of the two. This analysis is simplistic and does not take into
account special interest groups or the political scenery; however, it is obvi-
ous that each country, financial organization, or individual needs to identify
all available choices and rank them by the level of welfare they provide.
This process is called “evaluation.”
Therefore, “evaluation” refers to a process of comparing various
choices to rank them (based on a certain criterion) by order of attrac-
tiveness. At a much narrower level, the same evaluation process is per-
formed within a health system or (even more narrowly) within
a segment of a health system (e.g., hospitals), or even within the meth-
ods of management of a single disease (available budget, treatment
options, maximization of patient/taxpayer welfare, etc.). In this book
we deal extensively with the criteria we use to make decisions in the
health care industry.
The definition of economic evaluation we first provided includes the word
“systematic.” It is obvious that analyses based on simplistic criteria of cost
comparison between treatments (such as the price of one product compared
with another), which include neither the entire financial burden nor the
associated benefit for each treatment, are not systematic. When performing
economic evaluations, we seek a complete, thorough, and accurate compari-

son of the alternatives to judge their attractiveness and then rank them.
In conclusion, we might say that economic evaluation combines objective
data (prices of production factors, medical technology, etc.) with preference
data to rule on which of the available options maximize the welfare of the
society in general or of specific patient groups.
EVIDENCE-BASED MEDICINE AND EVIDENCE-BASED

HEALTH ECONOMICS
A common misconception is that economic evaluation deals exclusively
with the (albeit systematic) determination of the cost or the difference in
cost between two therapeutic interventions, with no regard for their
effectiveness. If that were true, then economic evaluation would be
demoted to a simple “accounting” assessment of burden, which would
overlook the social dimension of this science.
For a correct, ambitious, and complete analysis, this would be inaccu-
rate. Economic evaluation of health services combines cost and benefit
data to help the decision-makers evaluate them. In this context, eco-
nomic evaluation is useful not only to economists but also to clinical
scientists and policymakers, who can so be informed about the available
options (and the consequences of adopting them for the population’s
health and the country’s budget). In the United States and elsewhere, this
combination is referred to as value. Value can be thought of as
a relationship between outcomes and cost, and this relationship can be
graphically represented (Figure 1.1) (Williams, 2014). Value-based health
care is emerging as a central dogma for health care reform. As such, intel-
ligent application of systematic economic evaluation becomes a necessary
element.
This section requires more elucidation. Generally speaking, we are all
more or less familiar with the concept of evidence-based medicine (EBM).
EBM is usually synonymous with the search for knowledge (regarding
effectiveness) through systematic review of the literature to identify and
propose optimal practices for a health system. Its purpose is to inform the
people doing clinical work about these practices and to change current
practices if they are suboptimal. In formalistic terms, we might say that
EBM is the foundation of medical decisions on the process of systematic
search, evaluation, and implementation of the findings of current research.
This “movement,” which is closely linked to the principles and the
Medical and/or Cost of care Cost of care Cost of care

service outcomes decreased unchanged increased
Improved
Unchanged
Worsened
Figure 1.1 The value framework. Value is represented as a relationship between

cost and outcomes of care. High-value interventions are depicted in green, lower-
value interventions are depicted in yellow, and interventions of unacceptable value
are depicted in red. High-value interventions should be rapidly implemented.
Interventions of unacceptable value should be eliminated. The yellow boxes require
weighing the magnitude of the changes in outcomes and costs to determine the
acceptability of the intervention to the relevant stakeholder.
philosophy of evaluation, supports the notion that EBM should draw its
conclusions from outcome research (Braslow, 1999; Claridge and Fabian,
2005; Sur and Dahm, 2011).
In this way, EBM attempts to link its indications with medical practice to
improve health care quality and effectiveness at a personalized level. The imple-
mentation of its principles requires the integration of personal clinical experi-
ence with the most reliable scientific clinical data. Historically, the method of
physician self-education in practicing EBM was developed at the Canadian
McMaster University, and in Europe important work is being done at
Oxford, at the NHS Research and Development Centre for EBM, which is
its counterpart in the United Kingdom (Cohen, 1996; Hadley et al., 2007).
In contrast, the United States does not have a single center for EBM. Instead,
a number of stakeholders have assumed roles, including the government (e.g.,
Agency for Healthcare Research and Quality [AHRQ], http://www.ahrq.
gov; Health Resources and Services Administration [HRSA], http://www.
hrsa.gov), public private partnerships (Patient Centered Outcomes Research
Institute [PCORI], http://www.pcori.org; United States Preventive Services
Task Force [USPSTF], http://www.uspreventiveservicestaskforce.org), and
professional societies.
In many cases, however, the distinction between “good clinical practice”
and “available economic resources” is controversial and often provides results
opposite to those initially expected. For example, the adoption of practices

that offer very little therapeutic benefit but unbearable burden may undercut
the overall ability of the health system to treat patients in the near future
and over the next generation. In other cases, society elects—through its
health care system—to transfer resources to those less fortunate, regardless
of whether this transfer will achieve a small increase in overall social welfare.
Such social groups are usually unable to “obtain welfare” as easily as wealth-
ier population groups. In this sense, help for such groups is not a behavior
that leads to efficiency maximization; however, society might choose to
uphold the concept of equity at the cost of maximizing efficiency, a fact
that may not be taken into account by EBM.
In this framework, economic evaluation attempts to link EBM, whose
approach is more “clinically targeted,” to the wishes of society, patients,
and the state to better achieve multiple goals such as viability, societal
fairness, and improved efficiency in the health system. This approach is
called evidence-based health economics (EBHE). Today, at a theoretical level,
EBHE is a basic tool for the rational distribution of resources and a cen-
tral point of reference for health economics methodology. The arguments
presented here might indicate that, at a purely theoretical level, EBHE is
broader than EBM and contains EBM, but in practice they are equal allies
with different primary goals. EBM aims to influence basic clinical prac-
tices, whereas the evaluation of health technology is aimed at supporting
the exercise of health policies (Banta, 2003).
THE NECESSITY OF DRAFTING ECONOMIC EVALUATIONS

Generally speaking, one may say that the goal of a health system is to pro-
vide high-quality health services to their defined population on an equal
basis, to allow quick access to innovation that improves value, to produce
a large number of health services to meet the needs of the population,
and to do this efficiently by consuming as few resources as possible. These
health system activities are centrally organized through complex systems
of political oversight, planning, and financing.
Nevertheless, achieving these goals is impeded by certain factors, the
most important of which are the following:
• The demographic problem (few active workers to support the system
but many retirees in societies where health care funding is tied to
employment)
• The resolution to provide full health care coverage to the population

(especially in Western societies), which increases costs
• The modern unhealthy lifestyle (carbohydrate-rich foods, sedentary
lifestyle, use of alcohol, smoking, lack of exercise, poor diet, excessive
consumption of drugs, etc.), which causes chronic diseases and com-
plications that can only be treated at significant costs and with modest
therapeutic results
• The financial costs imposed by the technological advances in health
services (businesses bear very high research and development costs that
they wish to transfer to the end consumer or the public insurance
funds while also making some profit because they are profit-seeking
enterprises), many of which add little incremental benefit and result in
poor value
• The extended average lifespan (older people suffer from multiple
conditions and chronic diseases, with high treatment costs)
• The public’s expectations (it is the express conviction of a democratic
state that the citizens’ needs must be met with no particular consider-
ation of the cost; furthermore, the citizens’ demands have increased
because of improved educational systems and cheap and widely
available communication/information channels such as the Internet,
among others)
• Medical errors (which can be harmful or fatal for the patient): it is
estimated that medical errors cost billions of euros or other currency
and thousands of lives each year (van den Bos et al., 2011)
• Unnecessary consumption of health care resources (supply induced)
Because all of these factors constitute a direct or indirect financial
burden on modern health systems, governments believe that the money
spent for health care is excessive and that priorities must be set, or the
ratio at which the state and the patient share this expense must be altered.
Economic evaluation allows a more rational process to evaluate the factors
that impact the system. We should underline that an absolute restriction
of health care expenditures is a difficult goal from a social point of view,
whereas a reduction in the rate of expenditure growth is easier to achieve
(so-called bending of the cost curve). Therefore, the aim of economic
evaluation is not necessarily to restrict health care expenditures, but rather
to rationally distribute the available resources to those uses that will
achieve the highest possible level of population health based on certain
societal criteria. In certain cases, such criteria may lead to an increase in
expenditures when this is financially or socially acceptable (yellow box
in web version of Figure 1.1). If the ultimate goal were to reduce expen-
ditures, then the state would simply cease to provide health care services
to certain citizens, which would achieve immediate savings but would
inflame the public sense of justice and would strain social cohesion. Such
expenditure restrictions are socially justified and financially effective only
if they include a substantial restructuring of the system to save resources
without reducing benefits.
In conclusion, economic evaluation is more of a guideline for a
systematic and strict way of thinking than a dogmatic set of quantification
methods that will mechanically determine the therapeutic options to
which society should direct its resources.
REFERENCES
Banta, D., 2003. The development of health technology assessment. Health Policy 63 (2),
121 132.
Braslow, J.T., 1999. History and evidence-based medicine: lessons from the history of
somatic treatments from the 1900s to the 1950s. Ment. Health Serv. Res. 1 (4),
231 240.
Claridge, J.A., Fabian, T.C., 2005. History and development of evidence-based medicine.
World J. Surg. 29 (5), 547 553.
Cohen, L., 1996. McMaster’s pioneer in evidence-based medicine now spreading his mes-
sage in England. Can. Med. Assoc. J. 154 (3), 388 390.
Hadley, J.A., Davis, J., Khan, K.S., 2007. Teaching and learning evidence-based medicine
in complementary, allied, and alternative health care: an integrated tailor-made course.
J. Altern Complement Med. 13 (10), 1151 1155.
Sur, R.L., Dahm, P., 2011. History of evidence-based medicine. Ind. J. Urol. 27 (4),
487 489.
Van Den Bos, J., Rustagi, K., Gray, T., Halford, M., Ziemkiewicz, E., Shreve, J., 2011.
The $17.1 billion problem: the annual cost of measurable medical errors. Health Aff.
(Millwood) 30 (4), 596 603.
Williams, M.S., 2014. Genomic medicine implementation: learning by example. Am. J.
Med. Genet. C Semin. Med. Genet. 166, 8 14.
CHAPTER 2
Genomic Medicine Today:

An Introduction for Health
Economists
INTRODUCTION
The central dogma of genomic medicine is to exploit the individual’s
genomic sequence in combination with other clinical information to
enhance the clinical decision-making process. In recent years, significant
advances have enriched our knowledge regarding the molecular etiology
of a wide range of human genetic diseases, allowing for better disease
prevention, prognosis, and treatment. In parallel, genomic technology
has progressed rapidly and, as a result, genomics research has the poten-
tial to aid clinicians in their task of estimating disease risk and individu-
alizing treatment modalities. This constitutes the basis of genomic
medicine, a new discipline that promises to enhance opportunities for
the customization of patient care and the personalization of conventional
therapeutic interventions.
The concept of genomic or personalized medicine stems from as early as
400 B.C., at which time Hippocrates of Kos (460 370 B.C.E.) stated that
“. . .it is more important to know what kind of person suffers from a disease
than to know the disease a person suffers.” This ancient statement encapsulates
the essence of modern personalized genomic medicine. The first application
of individualized care based on presumed genetic information is documented
in the Talmud (Yevamot 64b). Rabbi Judah the Prince (135 217 C.E.) ruled
that if a woman’s first two children died from blood loss after circumcision,
the third son should not be circumcised. A second disagreed and ruled that
the third son may be circumcised; however, if this infant died, the fourth child
should not be circumcised. There was agreement that abnormal bleeding was
hereditary but disagreement regarding how many events were required to
establish a pattern and therefore exempt a child from circumcision

(http://download.yutorah.org/2013/1053/793670.pdf). At present, there is a

significant effort worldwide to support the translation of genomic research
into clinical practice so that genomic medicine can ultimately be used to ben-
efit the global community.
PHARMACOGENOMICS AND GENOMIC MEDICINE

For more than 50 years, it has been known that there is substantial inter-
individual variability in the clinical response to drug treatments and that
only a percentage of patients respond satisfactorily to their medications.
This implies that the balance of the patient population either is not
receiving proper medication or, worse, is suffering from either adverse
drug reactions, varying from mild to even lethal, or marked therapeutic
delays, which are caused by empiric trials of different medications until an
acceptable alternative is identified (Spear et al., 2001). Furthermore, the
side effects for the same drug vary from patient to patient. Experimental
evidence suggests that the variable phenotypic expression of drug treatment
efficacy and toxicity is determined by a complex interplay of multiple
genetic variants and environmental factors (Squassina et al., 2010).
Pharmacogenomics is referred to as “...the delivery of the right drug
to the right patient at the right dose” (Piquette-Miller and Grant, 2007),
and several pharmacogenomic testing approaches currently exist to iden-
tify the underlying pharmacogenomic biomarkers. The availability of
knowledge about the impact of pharmacogenomics variants on drug
response is the reason that pharmacogenomics is the branch of genomic
medicine that is being actively implemented in many centers.
APPLICATIONS OF PHARMACOGENOMICS
At present, the US Food and Drug Administration (FDA) and the European
Medicines Agency (EMA) provide pharmacogenomic information on the
label of more than 150 drugs. Although only a few recommend that
pharmacogenomic tests must be performed prior to prescribing these drugs,
the inclusion of the information for providers is indicative of the importance
these agencies place on this emerging field. The Clinical Pharmacogenetics
Implementation Consortium has published a series of evidence-based clinical
guidelines for use of pharmacogenetic information to guide therapy
Genomic Medicine Today: An Introduction for Health Economists 11
(http://www.pharmgkb.org/page/cpic). Currently, pharmacogenomics

instances are most relevant to the specialties of oncology, cardiology, and
psychiatry, and they are emerging for several other medical specialties.
Oncology Examples
Trastuzumab, a monoclonal antibody blocking v-erb-b2 erythroblastic
leukemia viral oncogene homolog 2 (HER2, also ERBB2) receptor
protein, is one of the prime examples for which pharmacogenomic test-
ing is routinely used. It has been shown that variable expression of the
HER2 receptor gene determines the likelihood that a patient will respond
to trastuzumab. HER2 overexpression is observed in approximately one-
fourth of breast cancer patients. It is correlated with poor prognosis,
increased tumor formation, and metastasis, as well as resistance to chemo-
therapeutic agents. As such, HER2 testing either through genetic testing
or through analysis of the HER2 protein product identifies patients who
will likely respond to trastuzumab. Other examples are the tyrosine kinase
inhibitors erlotinib and gefitinib, which are designed to target the epider-
mal growth factor receptor (EGFR). EGFR has been shown to play a
role in predisposition to lung cancer. EGFR variants are now used to pre-
dict improved progression-free survival with gefitinib in a comparison
with carboplatin paclitaxel (Mok et al., 2009).
Irinotecan is yet another example of a drug for which a pharmacoge-
nomic test can help to identify colorectal cancer patients who are likely
to experience adverse reactions during treatment, such as diarrhea and
severe neutropenia. The UGT1A1 28 variant is associated with increased
toxicity to irinotecan; as such, patients homozygous for the UGT1A1 28
allele are at higher risk for development of irinotecan-associated neutro-
penia and diarrhoea. The FDA recommended an addition to the irinote-
can package insert to include UGT1A1 28 genotype as a risk factor for
severe neutropenia. However, it is important to note that in the
Evaluation of Genomic Applications in Practice and Prevention (EGAPP)
working group’s recommendation regarding irinotecan, it was noted that
there is evidence that presence of this genotype results in improved
response to irinotecan, that is, improved efficacy (Berg et al., 2009). This
means that when evaluating a pharmacogenomic test, care must be taken
to identify impact on both efficacy and adverse events. The following
example also illustrates this principle.
Azathioprine, 6-mercaptopurine, and 6-thioguanine are drugs that are

metabolized by the thiopurine methyltransferase (TPMT) enzyme. They are
primarily used for the treatment of acute lymphoblastic leukemia (ALL), but
they are also used for inflammatory conditions such as Crohn’s disease, ulcer-
ative colitis, and psoriasis. Patients with inherited TPMT deficiency have
been shown to be likely to experience severe (potentially fatal) hematopoi-
etic toxicity when exposed to standard doses of thiopurine drugs. As such, a
pharmacogenomic test could classify patients according to normal, interme-
diate, and deficient levels of TPMT activity, which enables physicians to
prescibe the correct dose to these patients based on their TPMT activity
levels. In other words, patients classified as having normal activity are treated
with conventional doses, whereas lower doses (from 50% [intermediate
metabolizers] to as low as 10% [poor metabolizers] of the normal dose) are
administrated to avoid toxicity in deficient and intermediate patients who
are likely to suffer exaggerated, potentially life-threatening toxic responses to
normal doses of azathioprine and thiopurine drugs (Relling et al., 1999).
Again, there is evidence that individuals with intermediate or poor metabo-
lizers may have improved clinical response (efficacy) to these agents (Stanulla
et al., 2005). This is less of an issue in ALL because the endpoint of treatment
is measured by carefully monitoring end organ response (white count,
minimal residual disease, etc.). This demonstrates another important principle:
if you can measure a relevant phenotype, that is preferable to using a surrogate
endpoint such as a genotype. So in the case of ALL, although the starting dose
of the medication could be chosen based on genotype, the doses will be
adjusted based on the phenotypic response of the bone marrow. This is likely
quite different for other treatment indications such as inflammatory bowel
disease (IBD) or psoriasis. In these conditions, white blood count is not
measured as an indicator of treatment response; therefore, there is higher risk
of missing an incipient bone marrow catastrophe in individuals with interme-
diate or poor metabolizer phenotypes. As a consequence, the importance of
the pharmacogenomic information could be more important in these
conditions. This has led to development of a therapeutic algorithm for use of
thiopurines in IBD (Chouchana et al., 2012).
Cardiology Examples
In recent years, cardiology became the other medical specialty in which
pharmacogenomics applications are emerging into practice, particularly
related to warfarin, acenocoumarol, and clopidogrel. For more than
half a century, coumarinic oral anticoagulants (COAs), namely warfa-

rin, acenocoumarol, and phenprocoumon, have been the standard oral
anticoagulants for thromboembolic disorders. These COAs, however,
have a narrow therapeutic window and are associated with high risk of
major bleeding, especially during the initial phase of treatment, making
them one of the leading causes of emergency hospitalizations
worldwide. Experimental evidence suggests that there is a substantial
interindividual variation in COA treatment response, which requires
regular monitoring and dosage adjustment (Pirmohamed, 2006).
It has been shown for many years that CYP2C9 and VKORC1 geno-
mic variations influence interindividual the variable response of COAs.
CYP2C9 metabolizes COA, and vitamin K epoxide reductase (VKORC1)
is the pharmacologic target enzyme of these drugs (Bodin et al., 2005;
Schalekamp et al., 2007). The CYP2C9 2 and 3 variants lead to decreased
CYP2C9 activity, whereas the VKORC1 2 1639G . A variant influences
pharmacodynamic response to coumarins. Based on these findings, there are
several dosing algorithms incorporating CYP2C9 and VKORC1 genotype
information to predict the patient’s response to warfarin (Gage, et al., 2008;
Klein et al., 2009). Warfarin is among the drugs that have been relabeled, as
early as in 2007, to include CYP2C9 and VKORC1 genotyping as a means
to optimize warfarin dosing.
Clopidogrel is another drug that is the standard for the care of acute
coronary syndromes for which nonresponsiveness, corresponding to approxi-
mately 25% of patients receiving clopidogrel, is related to recurrent ischemic
events (Gladding et al., 2008). Response to clopidogrel has been shown to
be determined by the CYP2C19 genotype (Geisler et al., 2008) because
clopidogrel is a pro-drug that requires enzymatic conversion to an active
ingredient. CYP2C19 is the primary cytochrome P450 enzyme for this
conversion. The CYP2C19 2 allele has been shown to impair CYP2C19
function, and this is associated with a marked decrease in patient respon-
siveness to clopidogrel (Mega et al., 2009).
Psychiatric Examples
Unlike oncology and cardiology, pharmacogenomics and personalized medi-
cine are still far from being achieved in psychiatry (Alda, 2013).
Pharmacokinetic pathways in which cytochrome P450 isoenzymes, such as
CYP2D6, CYP1A1, CYP3A4, CYP2C9, and CYP2C19, are involved may
predict serum levels of antidepressants and antipsychotics. Numerous studies
implicated CYP2D6 genomic variants as a predictor of risperidone-induced

adverse reactions, but not those of clozapine (reviewed in Tsermpini et al.,
2014). A recommendation by the EGAPP working group found insufficient
evidence to support a recommendation for or against use of CYP450 genetic
testing in adults beginning selective serotonin reuptake inhibitor (SSRI)
treatment for nonpsychotic depression (EGAPP, 2007). As far as atypical
antipsychotics are concerned, pharmacogenomic studies have mostly focused
on the HTR2A and HTR2C serotonin receptor genes (Arranz and de Leon,
2007) and their role in SSRIs. Finally, there are few genetic determinants
that have been recently reported to predict response to lithium chloride
treatment of bipolar disorder (Perlis et al., 2009a; Squassina et al., 2011).
Infectious Disease Examples

Last but not least, pharmacogenomics for infectious diseases is an expanding
area that is gradually assuming an important role. One example is predic-
tion of adverse effects caused by antiretroviral drug therapies that are used
in the treatment of HIV/AIDS and chronic hepatitis C (Picard and
Bergeron, 2002). Mallal et al. (2008) demonstrated that the HLA-B 5701
allele is indicative of a hypersensitive reaction to abacavir. The latter variant
displays varying frequencies in patients from different populations and
reaches high frequencies in Asian populations, making HLA-B 5701 an
important pharmacogenomics marker to predict abacavir-induced toxicity.
Genetic variation can also inform differences in response to treatment.
An example of this is variation in the interleukin gene IL28B in patients
being treated for chronic hepatitis C (Bota et al., 2013). This is used in an
economic analysis in Chapter 6.
The astute reader will note that the examples provided lend them-
selves to the tools of economic analysis that will be introduced in the
next two chapters. Several of these examples are the subject of application
of economic analysis and appear later in the book.
FACING THE CURRENT CHALLENGES OF GENOMIC

MEDICINE
To maximize the contribution of genomic medicine in current medical
practice, there are certain actions that should be taken and obstacles to
be overcome to expedite the integration of pharmacogenomics into
modern medical practice and, hence, to maximally exploit the benefits
of this new discipline for society (Figure 2.1). In the following
Public Economic
health Genome evaluation
genomics informatics in genomic
medicine
Health care
Genomics
Pharmacogenomics professionals
research
genomics
education
Stakeholder
analysis General
Genethics public
in genomic
medicine genomics
awareness
Figure 2.1 Schematic depiction, as a jigsaw puzzle, of the various disciplines that
constitute the multidisciplinary field of genomic medicine. Note the central role of
genomics research, which is depicted as the central piece in the puzzle that needs
to be complemented by the other pieces to fulfill the promise of genomic medicine.
paragraphs, we outline the various challenges and bottlenecks that

should be overcome and steps that have to be undertaken for pharma-
cogenomics to find its place in future medicine.
EDUCATION OF HEALTH PROFESSIONALS

A prior requirement for the introduction of pharmacogenomics in some
medical specialties is to ensure high-level genomics education for health
care professionals. An important bottleneck is to fill the gap between the
pharmacogenomic testing per se and the interpretation and utilization of
its results in a clinical setting, also known as translation of genomic results
into patient care (Patrinos, 2010; Reydon et al., 2012). In this context,
several international organizations have called for the integration of
pharmacogenomics and personalized medicine education in core medical

curricula (Gurwitz et al., 2003). Several medical, pharmacy, and health
science curricula are gradually integrating such courses for their under-
graduate and postgraduate students (Pisanu et al., 2014). Moreover, the
urgent need to incorporate genetic and pharmacogenomics education
into medical and pharmacy schools has been motivated by the challenging
ethical implications of personalized medicine (Frueh and Gurwitz, 2004).
At the same time, researchers in the field of public health genomics
are gradually starting to assess the attitudes of health care professionals and
the general public regarding pharmacogenomics and genomic medicine,
because these opinions may be used to harmonize the development of
genomic medicine, molecular diagnostics, and related disciplines with
those of other countries, establish the legal framework to control genetic
testing services, safeguard the interests of the citizens, and shape the over-
all genomic medicine policy environment. There are several studies
indicating that patients and the general public expect to receive pharma-
cogenomic services from health care professionals who can confidently
explain the test and interpret its implications for prescriptions. However,
respondents of these surveys indicate that a gap still exists between
patients’ high expectations and health care professionals’ knowledge (Mai
et al., 2011, 2014), although both stakeholders hold a very positive view
of genomic medicine (Mitropoulou et al., 2015). Interestingly, a stake-
holders’ awareness survey that queried stakeholders (academia, health care
professionals, industry, government) involved in pharmacogenomics in
Japan evidenced the same pattern of expectations, although they have also
expressed similar concerns regarding issues such as poor genetic awareness
from the general public and the possibility of breaching privacy (Tamaoki
et al., 2007). Although the level of pharmacogenomic education among
health care professionals is still not optimal, proper interventions at the
educational level will definitely have a very positive effect on increasing
genetics education, which would consequently facilitate the incorporation
of genetics into patient care.
DEMONSTRATION OF CLINICAL UTILITY

A key to implementation of a new technology in practice is evidence that
the new technology makes a difference. Clinical utility means assessing
the balance of the benefits and risks of a given technology. In its narrow-
est sense, clinical utility refers to the ability of a screening or diagnostic
test to prevent or ameliorate adverse health outcomes such as mortality,

morbidity, or disability through the adoption of efficacious treatments on
the condition of test results. Simply, if the benefits outweigh the risks
based on robust evidence, a technology is said to have clinical utility. In
1997, the National Institutes of Health Department of Energy Task
Force on Genetic Testing stated, “Before a genetic test can be generally
accepted in clinical practice, data must be collected to demonstrate the
benefits and risks that accrue from both positive and negative results”
(Holtzman and Watson, 1997). Evidence of clinical utility, with a few
exceptions, has been lacking, as evidenced by the majority of EGAPP
working group recommendation statements concluding that there is
insufficient evidence to recommend for or against use of a given genetic
test for a certain defined purpose. Evidence of clinical utility is ultimately
translated into guidelines that can effectively alter clinical practice when
effectively implemented. The lack of guidelines in genomic medicine is a
key contributor to its general lack of integration into practice. The afore-
mentioned CPIC guidelines are an intentional effort to address this lack
of evidence in the area of pharmacogenomics.
HEALTH CARE COSTS

Generally speaking, the cost-effectiveness of a pharmacogenomic test is a
crucial factor for its adoption in a clinical setting. Thus, to demonstrate its
economic benefits, pharmacogenomic testing requires evidence not only of
its clinical utility but also of its cost-effectiveness (Deverka et al., 2010).
Regarding their impact on different medical specialties, there are not many
reports indicating that certain drug treatments related to psychiatry (Perlis
et al., 2009b), cancer (Carlson et al., 2009), and chronic inflammatory dis-
eases (Priest et al., 2006) are cost-effective. For example, Perlis et al.
(2009b) have demonstrated cost-effectiveness and benefits of a putative
pharmacogenomic test for antidepressant efficacy in the treatment of a
major depressive disorder. It should be noted, however, that for this as well
as for every pharmacogenomic test, an equally important factor to deter-
mine cost-effectiveness is the pharmacogenomic marker allele frequencies
that vary significantly among different populations. The most characteristic
examples of the application of economic evaluation in pharmacogenomics
and genomic medicine are described in detail in Chapter 6.
Different reimbursement policies of the various health care systems,
particularly in developing countries with large fiscal deficits are another
important parameter for health care costs related to genomic medicine

(Mette et al., 2012). In other words, payers, particularly public insurance
funds, can obviate the rapid dissemination of pharmacogenomics (Ginsburg
and Willard, 2009; Mitropoulou et al., 2015). Conversely, in the United
States, the fragmented nature of the delivery and reimbursement systems leads
to different groups making different decisions, even when faced with the
same evidence. Williams (2007) provides a comprehensive review of this
problem in pharmacogenomics. Overall, pharmacogenomic testing and geno-
mic medicine in general may represent a valuable resource for health care
decision-makers, leading to increased quality of clinical care and increased
economic benefits for the pharmaceutical industry and public health,
although pharmacogenomic tests will probably prove to be more cost-
effective than cost-saving, particularly for certain disease/treatment modalities
and populations (Deverka and McLeod, 2008; Flowers and Veenstra, 2004;
Mette et al., 2012).
INSURANCE AND PRIVACY ISSUES

Considering the fact that pharmacogenomics and genomic medicine
strongly rely on interindividual genomic variability, these disciplines have to
deal with a number of issues related to genetic discrimination/stigmatization,
privacy, and possible implications for access to life and health insurance. In
1998, Matloff et al. (2000) surveyed the members of the National Society of
Genetic Counselors (NSGC) Special Interest Group (SIG) regarding cancer
and demonstrated that more than half of the respondents would not bill their
insurance companies for genetic testing, largely because of fear of genetic
discrimination. Similarly, in 2007, Huizenga et al. (2010), compared the
results of a new survey provided to NSGC SIG using the data of Matloff
et al. (2000) and pointed out a notable change in perceptions and behavioral
intent among cancer genetics professionals (CGPs) over time, indicating
that fear of genetic discrimination has become significantly less common
since 1998.
Although the data presented concern the broad concept of genetic
testing, pharmacogenomics has to deal with the same issues of genetic
stigmatization/discrimination (Robertson, 2001; Issa, 2002), particu-
larly because the social consequences arising from new disease labels,
namely “good, poor, or nonresponder” to a given treatment modality,
would touch on very sensitive issues of interpersonal stigmatization or
identity. Moreover, pharmaceutical companies could voluntarily ignore,
for economic reasons, patients with rare or complex genetic conditions

or those who are not responding to any known treatment, leading to
consequent deprivation of effective treatments for certain subpopula-
tions (Rothstein and Epps, 2001).
Also, an equally important issue that may raise concern, particularly
in the whole-genome sequencing era, is the storage of genomic infor-
mation in databases with the inherent danger of losing confidentiality
or privacy, because an enormous quantity of genotypic, phenotypic,
and demographic data regarding individuals are inter-related (Vaszar
et al., 2003; Potamias et al., 2014). To this end, protection of privacy
and confidentiality has to be ensured, particularly in the whole-genome
sequencing era, because pharmacogenomic tests can carry several types
of secondary information that represent a risk of psychosocial harm
and adverse insurance and/or employment implications. Moreover,
particular subpopulation groups may face genetic discrimination when
trying to access health care or health insurance (Smart et al., 2004).
The Genetic Information Non-discrimination Act (GINA), the Health
Insurance Portability and Accountability Act Privacy Rule, and the
Genomics and Personalized Medicine Act (GPMA) are all legislative
attempts in the United States trying to address these questions,
although they often lack clarity in critical issues regarding translation of
human genetic variation from the bench to the bedside (Lee and
Mudaliar, 2009).
CONCLUSIONS AND FUTURE PERSPECTIVES

The dawn of the postgenomics era has further revealed the role of
pharmacogenomics and genomic medicine as two of the most fundamental
disciplines of modern clinical practice toward achieving the goal of personal-
ized medicine. As with every new trend, these new disciplines have found
their way faster into some medical specialties, such as cardiology or oncology,
in which they are already applied for selecting and/or dosing a specific medi-
cation, whereas in other specialties, such as psychiatry or infectious diseases,
they are still lagging behind. Even though personalized medicine is not yet a
widespread practice, ongoing international efforts confirm that this concept
is close to becoming a reality (Kampourakis et al., 2014).
The delineation of genomic variation with complex pathways pertaining
to drug efficacy and toxicity and the etiology of complex genetic traits and
clinical conditions constitute a daunting challenge, because these phenotypes
are, by their nature, more complex and heterogeneous when compared

with bimodal traits. Although this may sound discouraging, there are some
strategies that, if used correctly, may lead us to translate genomics research
findings into genomic medicine applications.
In the whole-genome sequencing era, DNA sequence analysis allows
the identification of unique or very rare novel single nucleotide poly-
morphisms (SNPs) either in genes encoding for drug-metabolizing
enzymes and transporters or in the whole genome with direct implica-
tions in drug response (Mizzi et al., 2014). Preliminary evidence suggests
that there are several thousand variants in each individual affecting drug
response, hundreds of which are novel and whose identification is of
utmost importance in determining one’s personalized pharmacogenomics
profile (Mizzi et al., 2014). Moreover, some analytical approaches, such as
case-only genome-wide interaction (COGWI) (Pierce and Ahsan, 2010),
can provide a straightforward method for detecting pharmacogenomic
interactions related to treatment response in complex diseases, which might
provide a framework of operational and decisional criteria for setting the
course leading to widespread use of pharmacogenomics in public health.
Furthermore, there are novel concepts that emerge in the field of
pharmacogenomics, such as some mitochondrial variants that could be
considered putative pharmacogenomic markers (Pacheu-Grau et al., 2010),
or some genomic variants that could be correlated with variable responses
to drug therapy in organ transplantation. In the latter case, tacrolimus
constitutes perhaps the best established example of the CYP3A5 gene effect
on drug disposition and dosage, with the CYP3A53 allele shown to be
associated with a lower dose of tacrolimus to achieve therapeutic blood
concentrations. There are also other drugs used in organ transplantation
whose effectiveness has been correlated with a number of genomic
variants, such as azathioprine with TPMT alleles, cyclosporine with
ABCB1 alleles, sirolimus (like tacrolimus) with CYP3A5 alleles, and
corticosteroids with ABCB1 and IL10 polymorphisms (reviewed by
Girnita et al., 2008).
Several international organizations, policy and advocacy groups, and
research consortia have been formed, such as the Personalized Medicine
Coalition (PMC; http://www.personalizedmedicinecoalition.org; Abrahams
et al., 2005) and the European Alliance for Personalized Medicine (EAPM;
http://www.eapm.eu), which are nonprofit policy and advocacy groups
formed of pharmaceutical biotechnology, diagnostic and information
technology companies, health care providers and payers, patient advocacy
groups, industry policy organizations, academic institutions, and government

agencies, aiming to facilitate the use of personalized medicine approaches, to
help in the training of the public; to provide opinion leadership; and to con-
vey information to the media, government officials, and health care leaders.
Other initiatives, such as the Genomic Medicine Alliance (http://www.
genomicmedicinealliance.org), a newly established international research con-
sortium (Cooper et al., 2014), aim specifically to bridge the gap between
developed and developing nations by ensuring technology and knowledge
transfer. Such efforts would be particularly useful for developing nations to
defray health care costs and improve quality of life by minimizing adverse
drug reactions (Mitropoulos et al., 2011).
From the time of Hippocrates, personalized medicine has been
considered a component of good medical practice. However, as with
every new trend, extra caution should be taken to avoid abuse of or
underestimating the potential of this new technology. It should be
remembered that genomic information is only one type of information
that may be relevant to a clinical situation and that it does not occupy a
privileged position in relation to other types of information. The impor-
tance of understanding the patient’s goals for treatment has not been
given as much weight as it deserves. This “patient-centered” approach is
being increasingly emphasized but it is not new. Sir William Osler,
another icon of medicine, recognized its importance in care when he
said, “Care more for the individual patient than for the special features of
the disease. . .Put yourself in his place. . ..” We therefore favor the defini-
tion of personalized medicine proposed by Pauker and Kassirer in 1987
that embraces the concept of patient centeredness and equality of infor-
mation: “Personalized medicine is the practice of clinical decision-making
such that the decisions made maximize the outcomes that the patient
most cares about and minimizes those that the patient fears the
most, on the basis of as much knowledge about the individual’s
state as is available.”
In conclusion, the adoption of pharmacogenomics and genomic
medicine for the prescription of personalized medical interventions
still has to face a number of challenges, but the increasing knowledge
of the molecular basis of inherited conditions and of drug efficacy and
toxicity, accompanied by the growing attention of the pharmaceutical
industry and national health care policymakers, will probably accelerate
the pace toward the achievement of personalized health care and
precision medicine.
REFERENCES
Abrahams, E., Ginsburg, G.S., Silver, M., 2005. The personalized medicine coalition:
goals and strategies. Am. J. Pharmacogenomics 5, 345 355.
Alda, M., 2013. Personalized psychiatry: many questions, fewer answers. J. Psychiatry
Neurosci. 38, 363 365.
Arranz, M.J., de Leon, J., 2007. Pharmacogenetics and pharmacogenomics of schizophrenia:
a review of last decade of research. Mol. Psychiatry 12, 707 747.
Berg, A.O., Armstrong, K., Botkin, J., Calonge, N., Haddow, J., Hayes, M., et al., 2009.
Recommendations from the EGAPP working group: can UGT1A1 genotyping
reduce morbidity and mortality in patients with metastatic colorectal cancer treated
with irinotecan? Genet. Med. 11 (1), 15 20.
Bodin, L., Verstuyft, C., Tregouet, D.A., Robert, A., Dubert, L., Funck-Brentano, C., et al.,
2005. Cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase (VKORC1)
genotypes as determinants of acenocoumarol sensitivity. Blood 106, 135 140.
Bota, S., Sporea, I., Şirli, R., Neghină, A.M., Popescu, A., Străin, M., 2013. Role of
interleukin-28B polymorphism as a predictor of sustained virological response in
patients with chronic hepatitis C treated with triple therapy: a systematic review and
meta-analysis. Clin. Drug Investig. 33 (5), 325 331.
Carlson, J.J., Garrison, L.P., Ramsey, S.D., Veenstra, D.L., 2009. The potential clinical and
economic outcomes of pharmacogenomic approaches to EGFR-tyrosine kinase
inhibitor therapy in non-small-cell lung cancer. Value Health 12, 20 27.
Chouchana, L., Narjoz, C., Beaune, P., Loriot, M.A., Roblin, X., 2012. Review article:
the benefits of pharmacogenetics for improving thiopurine therapy in inflammatory
bowel disease. Aliment. Pharmacol. Ther. 35 (1), 15 36.
Cooper, D.N., Brand, A., Dolzan, V., Fortina, P., Innocenti, F., Lee, M.T., et al., 2014.
Bridging genomics research between developed and developing countries: the geno-
mic medicine alliance. Pers Med. 11, 615 623.
Deverka, P.A., McLeod, H.L., 2008. Harnessing economic drivers for successful clinical
implementation of pharmacogenetic testing. Clin. Pharmacol. Ther. 84, 191 193.
Deverka, P.A., Vernon, J., McLeod, H.L., 2010. Economic opportunities and challenges
for pharmacogenomics. Annu. Rev. Pharmacol. Toxicol. 50, 423 437.
Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working
Group, 2007. Recommendations from the EGAPP working group: testing for
cytochrome P450 polymorphisms in adults with nonpsychotic depression treated with
selective serotonin reuptake inhibitors. Genet. Med. 9 (12), 819 825.
Flowers, C.R., Veenstra, D., 2004. The role of cost-effectiveness analysis in the era of
pharmacogenomics. Pharmacoeconomics 22, 481 493.
Frueh, F.W., Gurwitz, D., 2004. From pharmacogenetics to personalized medicine: a vital need
for educating health professionals and the community. Pharmacogenomics 5, 571 579.
Gage, B.F., Eby, C., Johnson, J.A., Deych, E., Rieder, M.J., Ridker, P.M., et al., 2008.
Use of pharmacogenetic and clinical factors to predict the therapeutic dose of warfa-
rin. Clin. Pharmacol. Ther. 84, 326 331.
Geisler, T., Schaeffele, E., Dippon, J., Winter, S., Buse, V., Bischofs, C., et al., 2008. CYP2C19
and nongenetic factors predict poor responsiveness to clopidogrel loading dose after
coronary stent implantation. Pharmacogenomics 9, 1251 1259.
Ginsburg, G.S., Willard, H.F., 2009. Genomic and personalized medicine: foundations
and applications. Transl. Res. 154, 277 287.
Girnita, D.M., Burckart, G., Zeevi, A., 2008. Effect of cytokine and pharmacogenomic
genetic polymorphisms in transplantation. Curr. Opin. Immunol. 20, 614 625.
Gladding, P., Webster, M., Ormiston, J., Olsen, S., White, H., 2008. Antiplatelet drug
nonresponsiveness. Am. Heart J. 155, 591 599.
Gurwitz, D., Weizman, A., Rehavi, M., 2003. Education: teaching pharmacogenomics to
prepare future physicians and researchers for personalized medicine. Trends Pharmacol.
Sci. 24, 122 125.
Holtzman, N.A., Watson, M.S. (Eds.), 1997. Promoting safe and effective genetic testing
in the United States. Final report of the task force on genetic testing. ,http://www.
genome.gov/10001733..
Huizenga, C.R., Lowstuter, K., Banks, K.C., Lagos, V.I., Vandergon, V.O., Weitzel, J.N.,
2010. Evolving perspectives on genetic discrimination in health insurance among
health care providers. Fam. Cancer 9 (2), 253 260.
Issa, A.M., 2002. Ethical perspectives on pharmacogenomic profiling in the drug develop-
ment process. Nat. Rev. Drug Discov. 1, 300 308.
Kampourakis, K., Vayena, E., Mitropoulou, C., Borg, J., van Schaik, R.H., Cooper, D.N.,
et al., 2014. Key challenges for next generation pharmacogenomics. EMBO Rep. 15
(5), 472 476.
Klein, T.E., Altman, R.B., Eriksson, N., Gage, B.F., Kimmel, S.E., Lee, M.T., et al.,
2009. Estimation of the warfarin dose with clinical and pharmacogenetic data. N.
Engl. J. Med. 360, 753 764.
Lee, S.S., Mudaliar, A., 2009. Medicine. Racing forward: the genomics and personalized
medicine act. Science 323, 342.
Mai, Y., Koromila, T., Sagia, A., Cooper, D.N., Vlachopoulos, G., Lagoumintzis, G., et al.,
2011. A critical view of the general public’s awareness and physicians’ opinion of the
trends and potential pitfalls of genetic testing in Greece. Pers. Med. 8, 551 561.
Mai, Y., Mitropoulou, C., Papadopoulou, X.E., Vozikis, A., Cooper, D.N., van Schaik,
R.H., et al., 2014. Critical appraisal of the views of healthcare professionals with
respect to pharmacogenomics and personalized medicine in Greece. Pers. Med. 11
(1), 15 26.
Mallal, S., et al., 2008. HLA-B 5701 screening for hypersensitivity to abacavir. N. Engl. J.
Med. 358, 568 579.
Matloff, E.T., Shappell, H., Brierley, K., Bernhardt, B.A., McKinnon, W., Peshkin, B.N.,
2000. What would you do? Specialists’ perspectives on cancer genetic testing, prophy-
lactic surgery, and insurance discrimination. J. Clin. Oncol. 18, 2484 2492.
Mega, J.L., Close, S.L., Wiviott, S.D., Shen, L., Hockett, R.D., Brandt, J.T., et al., 2009.
Cytochrome p-450 polymorphisms and response to clopidogrel. N. Engl. J. Med.
360, 354 362.
Mette, L., Mitropoulos, K., Vozikis, A., Patrinos, G.P., 2012. Pharmacogenomics and
public health: implementing populationalized medicine. Pharmacogenomics 13 (7),
803 813.
Mitropoulos, K., Johnson, L., Vozikis, A., Patrinos, G.P., 2011. Relevance of pharmacoge-
nomics for developing countries in Europe. Drug Metabol. Drug Interact. 26, 143 146.
Mitropoulou, C., Mai, Y., van Schaik, R.H., Vozikis, A., Patrinos, G.P., 2015.
Documentation and analysis of the policy environment and key stakeholders in phar-
macogenomics and genomic medicine in Greece. Public Health Genomics 17,
280 286.
Mizzi, C., Mitropoulou, C., Mitropoulos, K., Peters, B., Agarwal, M.R., van Schaik, R.H.,
et al., 2014. Personalized pharmacogenomics profiling using whole genome sequencing.
Pharmacogenomics 15 (9), 1223 1234.
Mok, T.S., Wu, Y.L., Thongprasert, S., Yang, C.H., Chu, D.T., Saijo, N., et al., 2009.
Gefitinib or carboplatin paclitaxel in pulmonary adenocarcinoma. N. Engl. J. Med.
361, 947 957.
Pacheu-Grau, D., Gomez-Duran, A., Lopez-Perez, M.J., Montoya, J., Ruiz-Pesini, E.,
2010. Mitochondrial pharmacogenomics: barcode for antibiotic therapy. Drug Discov.
Today 15, 33 39.
Patrinos, G.P., 2010. General considerations for integrating pharmacogenomics into main-
stream medical practice. Hum. Genomics 4, 371 374.
Pauker, S.G., Kassirer, J.P., 1987. Decision analysis. N. Engl. J. Med. 316 (5), 250 258.
Perlis, R.H., Smoller, J.W., Ferreira, M.A., McQuillin, A., Bass, N., Lawrence, J., et al.,
2009a. A genome-wide association study of response to lithium for prevention of
recurrence in bipolar disorder. Am. J. Psychiatry 166, 718 725.
Perlis, R.H., Patrick, A., Smoller, J.W., Wang, P.S., 2009b. When is pharmacogenetic
testing for antidepressant response ready for the clinic? A cost-effectiveness analysis
based on data from the STAR D study. Neuropsychopharmacology 34,
2227 2236.
Picard, F.J., Bergeron, M.G., 2002. Rapid molecular theranostics in infectious diseases.
Drug Discov. Today 7, 1092 1101.
Pierce, B.L., Ahsan, H., 2010. Case-only genome-wide interaction study of disease risk,
prognosis and treatment. Genet. Epidemiol. 34, 7 15.
Piquette-Miller, M., Grant, D.M., 2007. The art and science of personalized medicine.
Clin. Pharmacol. Ther. 81, 311 315.
Pirmohamed, M., 2006. Warfarin: almost 60 years old and still causing problems. Br. J.
Clin. Pharmacol. 62, 509 511.
Pisanu, C., Tsermpini, E.E., Mavroidi, E., Katsila, T., Patrinos, G.P., Squassina, A., 2014.
Assessment of the Pharmacogenomics Educational Environment in Southeast Europe.
Public Health Genomics 17, 272, 279.
Potamias, G., Lakiotaki, K., Katsila, T., Lee, M., Topouzis, S., Cooper, D.N., et al., 2014.
Deciphering next-generation pharmacogenomics: an information technology perspec-
tive. OPEN Biol. 4 (7), 140071.
Priest, V.L., Begg, E.J., Gardiner, S.J., Frampton, C.M., Gearry, R.B., Barclay, M.L.,
et al., 2006. Pharmacoeconomic analyses of azathioprine, methotrexate and prospec-
tive pharmacogenetic testing for the management of inflammatory bowel disease.
Pharmacoeconomics 24, 767 781.
Relling, M.V., Hancock, M.L., Rivera, G.K., Sandlund, J.T., Ribeiro, R.C., Krynetski,
E.Y., et al., 1999. Mercaptopurine therapy intolerance and heterozygosity at the thio-
purine S-methyltransferase gene locus. J. Natl. Cancer Inst. 91, 2001 2008.
Reydon, T.A., Kampourakis, K., Patrinos, G.P., 2012. Genetics, genomics and society: the
responsibilities of scientists for science communication and education. Pers. Med. 9,
633 643.
Robertson, J.A., 2001. Consent and privacy in pharmacogenetic testing. Nat. Genet. 28,
207 209.
Rothstein, M.A., Epps, P.G., 2001. Ethical and legal implications of pharmacogenomics.
Nat. Rev. Genet. 2, 228 231.
Schalekamp, T., Brassé, B.P., Roijers, J.F., van Meegen, E., van der Meer, F.J., van Wijk,
E.M., et al., 2007. VKORC1 and CYP2C9 genotypes and phenprocoumon anticoa-
gulation status: interaction between both genotypes affects dose requirement. Clin.
Pharmacol. Ther. 81, 185 193.
Smart, A., Martin, P., Parker, M., 2004. Tailored medicine: whom will it fit? The ethics
of patient and disease stratification. Bioethics 18, 322 342.
Spear, B.B., Heath-Chiozzi, M., Huff, J., 2001. Clinical application of pharmacogenetics.
Trends Mol. Med. 7, 201 204.
Squassina, A., Manchia, M., Manolopoulos, V.G., Artac, M., Lappa-Manakou, C.,
Karkabouna, S., et al., 2010. Realities and expectations of pharmacogenomics and per-
sonalized medicine: impact of translating genetic knowledge into clinical practice.
Pharmacogenomics 11, 1149 1167.
Squassina, A., Manchia, M., Borg, J., Congiu, D., Costa, M., Georgitsi, M., et al., 2011.
Evidence for association of an ACCN1 gene variant with response to lithium treat-
ment in Sardinian patients with bipolar disorder. Pharmacogenomics 12, 1559 1569.
Stanulla, M., Schaeffeler, E., Flohr, T., Cario, G., Schrauder, A., Zimmermann, M., et al.,
2005. Thiopurine methyltransferase (TPMT) genotype and early treatment response
to mercaptopurine in childhood acute lymphoblastic leukemia. JAMA 293 (12),
1485 1489.
Tamaoki, M., Gushima, H., Tsutani, K., 2007. Awareness survey of parties involved in
pharmacogenomics in Japan. Pharmacogenomics 8, 275 286.
Tsermpini, E.E., Assimakopoulos, K., Bartsakoulia, M., Economou, G., Papadima, E.,
Mitropoulos, K., et al., 2014. Individualizing clozapine and risperidone treatment for
schizophrenia patients. Pharmacogenomics 15 (1), 95 110.
Vaszar, L.T., Cho, M.K., Raffin, T.A., 2003. Privacy issues in personalized medicine.
Williams, M.S., 2007. Insurance coverage for pharmacogenomic testing in the United
States. Pers. Med. 4 (4), 479 487.
CHAPTER 3
Economic Evaluation and

Genomic Medicine: What Can
They Learn from Each Other?
INTRODUCTION
The field of health economics is more about understanding utility and
effectiveness of medical interventions rather than focusing strictly on costs.
In its ideal application, health economics helps inform the decision-making
process to improve the quality of health outcomes while optimizing health
care expenditures.
There are three key features in health economics analyses as currently
applied: (i) they are more focused on the benefits received by the health care
system and society as a whole rather than the individual/patient, namely
price, health state, improvements in quality of life, expansion of life expec-
tancy, and resources saved, to name a few; (ii) the recipient of the medical
intervention is, in the majority of cases, not the most informed medical
decision-maker and, as such, does not have a complete picture of the poten-
tial benefits and harms of a given medical intervention or decision; and
(iii) most of the time, the recipients of medical interventions do not directly
pay for these treatments; rather, payment is received from a third party that
the recipient supports through taxes, medical premiums, or a shared model
of employer/employee contributions. As such, health economics aims to
better understand the value and costs of a certain medical intervention
compared with another by taking into assumption all the factors that impact
on patients, health care providers, the health care system in general, and,
ultimately, society.
Health economics uses various economic evaluation tools to evaluate
medical interventions, including:
• Cost-effectiveness analysis (CEA): This sort of analysis aims to deter-
mine whether a medical intervention for disease diagnosis, prevention,
and/or treatment improves clinical outcomes enough to justify the

additional costs compared with alternative approaches. It is quantitative

and relies on all the factors involved in the evaluation of medical inter-
ventions and/or health care technologies. It should be noted that CEA
is not a method to show which medical intervention or health care
technology reduces cost (cost benefit analysis), but rather it is used to
inform which medical interventions and/or health care technologies
provide the greatest value for a given amount of health care expendi-
ture. Another feature of CEA is that it valuates clinical events out-
comes, such as cost per life-year gained, but does not allow for direct
comparisons. For example, one cannot easily compare whether it is
cost-effective to spend a certain amount of money to prevent a rare
genetic disorder or cancer; for this, one should take into consideration
other factors, such as quality of life, life expectancy, accompanying
costs, and others.
• Cost utility analysis (CUA): This approach attempts to address the
aforementioned CEA limitations by measuring outcomes through a
metric called a quality-adjusted life-year (QALY) that allows for com-
parisons across medical interventions. For example, if fewer QALYs
are produced when the same amount is spent to prevent a rare genetic
disorder compared with that spent to prevent cancer, this adjustment
allows a better-informed societal medical decision to be made to
invest in one intervention compared with the other.
• Cost-minimization analysis (CMA): This approach aims to reduce
costs while keeping the outcome of a medical intervention the same.
To achieve this, CMA assumes that the outcome of two different
medical interventions is the same (clinical equipoise), which is rarely
the case. As such, this approach cannot be commonly applied to eval-
uate clinical interventions.
• Cost benefit analysis: This analysis gives a monetary value to every
aspect of health care and medical intervention, which can be very chal-
lenging in health care because health care providers are often very
reluctant to place a monetary value on health; as such, this approach is
truly difficult to perform accurately.
• Cost-threshold analysis: This type of analysis operates in reverse in
that the analyst defines a threshold of cost-effectiveness (usually
derived from the calculated cost-effectiveness of a given intervention).
Once this is defined the analyst asks the question, what are the neces-
sary performance characteristics of a given test or intervention such
that the cost of the test/intervention meets or exceeds the threshold.
Economic Evaluation and Genomic Medicine: What Can They Learn from Each Other? 29
This analytic approach is being used more frequently by developers to

understand whether their test or intervention is ready for clinical use
or needs additional development (Rubenstein et al., 2005; Gordon
et al., 2014). In genomics this has been used to assess when a genomic
risk panel or pharmacogenomics test would have sufficient discrimina-
tory power to justify its use in clinical care (Bock et al., 2014).
All these approaches consider the cost of the medical intervention itself
together with the accompanying costs, but they differ in how they measure
the outcome or utility of an intervention. The two more commonly used
approaches in the field are CEA and, most importantly, CUA.
GENOMICS IN HEALTH ECONOMICS

As indicated, application of economic evaluation in genomic medicine
is not so different compared with other aspects of medicine. There are
still some important elements that should be taken into consideration
during the various stages of analysis. Such elements are mostly the type
of genomic data and the various ethical, legal, and social issues that
pertain to these data.
Pharmacogenomics is a core component of genomic medicine and, as
such, it is used as an example to highlight the application of economic
evaluation in genomic medicine. Pharmacogenomics is a new discipline
in health care that attempts to enrich our understanding of how medicines
work in each individual based on genomic contributions to a medicine’s
safety and efficacy (reviewed by Squassina et al., 2010). The latter can lead
to a more efficient and effective approach to drug discovery. Furthermore,
pharmacogenomics may lead to a more diversified and targeted portfolio of
diagnostics and therapies, which, when used together, would yield greater
health benefits to society.
The term “pharmacogenetics” was first used in 1959 by F. Vogel to
describe a scientific discipline that sought to understand how an individual’s
genetic profile may influence their drug responses. Pharmacogenetics is
referred to as the study of the effect of genomic variations on drug
response in terms of both drug metabolism (pharmacokinetics) and drug
action (pharmacodynamics). Additionally, genetic variants have been shown
to explain what had previously been considered to be idiosyncratic adverse
drug events (ADE). In other words, this discipline aims to identify the best
medicine for a specific disease when the disease occurs in a patient popula-
tion with a particular genotype. Considering the fact that genetic factors
account for 20 95% of the observed responses to drug therapies, one

could understand the impact of this new discipline in modern medicine.
It is important to note that other factors such as age, food intake,
drug drug interactions, and the simultaneous presence of other diseases
(comorbidity) influence an individual’s drug response independent of, in
conjunction with, or in addition to genetic factors. Pharmacogenomics is
a broader term aiming to systematically assess the way genomic pathways
affect disease susceptibility, pharmacological function, and drug disposi-
tion and response. It aims not only to identify genomic biomarkers for
disease classification, staging, and diagnosis within the context of drug
responses but also to optimize drug discovery with the goal of achieving a
more desirable pharmacological response. Currently, the term “pharma-
cogenomics” is broadly used to cover all of these, and it impacts both
new and existing medicines. It is expected to have a major impact on the
translation of early-stage projects into medical treatments.
As indicated in the previous chapter, cardiology is among the medical
specialties in which pharmacogenomics is used. Warfarin is the most
commonly used anticoagulant, aiming to prevent and treat blood clots.
Anticoagulation caused by warfarin is due to the inhibition of vitamin K
epoxide reductase, an enzyme that activates vitamin K to produce antic-
oagulation factors II, VII, IX, and X. Warfarin is metabolized by the
CYP2C9 enzyme. Cardiologists commonly prescribe it for patients with
a history of atrial fibrillation, deep vein thrombosis, recurrent stroke, or
pulmonary embolism, as well as in cases of heart valve replacements.
A major challenge in treating patients with warfarin is that the optimal dose
varies significantly from individual to individual. If the prescribed dose is
too high, patients are at increased risk for serious bleeding; however, if
the dose is too low, patients are at increased risk for a thrombotic event
that could result in a stroke or other vaso-occlusive event. The highest
risk for these complications lies within the first 30 60 days after the
beginning of warfarin treatment.
Individual characteristics and behavior, such as age, sex, and diet, are
some of the factors that account for the variation in warfarin dose across
individuals, although these factors account for, at most, 50% of the inter-
individual variability (Flockhart et al., 2008). Importantly, genomic var-
iants in the CYP2C9 gene create variant alleles that have been found to
reduce the activity of CYP2C9, thus decreasing warfarin’s clearance. The
variant CYP2C9 2 allele decreases warfarin clearance by approximately
30% and the CYP2C9 3 allele by approximately 80% when compared
with the wild-type CYP2C91 allele. As a result, patients with a

CYP2C91/1 genotype require a daily mean maintenance warfarin
dose that is higher than what is required for CYP2C91/ 2 heterozygote
and CYP2C91/3 heterozygote patients, with the latter requiring the
lowest dose. In various population groups, the variant alleles present with
varying frequencies, with CYP2C9 2 and CYP2C9 3 being more com-
mon in European Americans and less common in Asians and African
Americans, respectively. In addition, genomic variants in the VKORC1
gene, which encodes the production of vitamin K epoxide reductase,
have also been shown to affect warfarin treatment. There are five
VKORC1 variant combinations (e.g., haplotypes) that are associated
with altered VKORC1 gene expression and, as such, with different
warfarin dose requirements. The allelic frequencies of these VKORC1
haplotypes also vary in different populations. The combination of the
CYP2C9 and VKORC1 genomic variations appears to account for
another 30 40% of inter individual dose variation (Flockhart et al.,
2008). Remaining unexplained variability could be due to other genetic
variants, uncharacterized variants in other genomic loci, as well as other
personal or environmental factors yet to be identified. Currently avail-
able warfarin dosing calculators (e.g., http://warfarindosing.org) use a
combination of clinical and genetic factors and have been demonstrated
to be superior in predicting the stable warfarin dose when compared
with clinical judgment alone.
On average, one-third of the population carries one or both of
the CYP2C9 and VKORC1 genomic biomarkers that are shown to be
associated with slower warfarin metabolism, which in turn increases the
likelihood of excess anticoagulation and the associated risk of serious
bleeding. It is important to note that individuals who are “wild-type”
require slightly higher warfarin doses than the recommended starting
dose (6 mg/day compared with 5 mg/day). Currently, the appropriate
dose is determined by regularly monitoring the anticoagulation levels
through blood tests and decreasing or increasing the warfarin dose if the
international normalized ratio (INR) is too high or too low, respectively.
As such, pharmacogenomic testing could potentially identify the patients
who are likely to present with slower warfarin metabolism, which could
influence both the dose of warfarin and the recommended timing of
INR studies. This may be a cost-effective way to reduce bleeding events
from warfarin, as demonstrated in an economic modeling analysis based
on the results of a small prospective study (Meckley et al., 2010).
According to previously published reports in 2004 and 2005, side

effects from just three drugs were responsible for one-third of all emer-
gency hospitalizations of seniors (age 65 years or older) in the United
States who experienced adverse reactions to these medications. Warfarin
was one of these drugs, accounting for 58,000 emergency hospitalizations
per year. Also, the Adverse Event Reporting System of the US Food and
Drug Administration (FDA) provides evidence that warfarin is among the
top 10 drugs with the greatest number of serious adverse drug reactions.
Literature reports of major bleeding frequencies for warfarin vary from as
low as 0% to as high as 16%. On the basis of these data, the FDA added
a new black-box warning to the warfarin label in 2006. Also, in August
2007, the US FDA updated the warfarin product label to add pharmaco-
genetic information. In January 2010, the agency added specific instruc-
tions regarding how to use genotype to predict individualized doses; the
new label provides a concise table of dosing recommendations, stratified by
genotype. However, to date, the FDA black-box warning does not require
pharmacogenetic testing to be performed prior to initiation of warfarin.
An evidence-based practice guideline for pharmacogenetically informed
warfarin dosing has been published by the Clinical Pharmacogenomics
Implementation Consortium (Caudle et al., 2014).
TRIANGULATING GENOMICS WITH HEALTH TECHNOLOGY

ASSESSMENT AND HEALTH ECONOMICS
In the previous paragraph, we have briefly discussed various approaches in
economic evaluation and a theoretical example in the field of pharmaco-
genomics, demonstrating the application of economic evaluation in geno-
mics. The previous approaches can be fruitfully combined with health
technology assessment and health economics, giving rise to comparative
effectiveness research, which is closely related to the way genomic informa-
tion is being used by health care professionals.
Comparative effectiveness research includes a broad range of stake-
holders and stakeholder-related components, such as:
• Study design
• Stakeholder analysis (citizens/general public/patients, health care pro-
fessionals, policymakers) and their prioritization
• Genomic test comparisons
• Analysis of personalized medical decision-making process. This latter
component is highly relevant in the whole-genome sequencing era,
considering the fact that results may differ among different patient
groups on the basis of genome characteristics, which can directly
impact risk stratification as well as medical interventions.
ECONOMIC EVALUATION IN THE POSTGENOMIC ERA

Whole-genome sequencing, accompanied with competent data inter-
pretation and genetic counseling, will revolutionize genomic medicine
(Gullapalli et al., 2012; Kilpinen and Barrett, 2013). As such, its cost-
effectiveness depends on the outcomes being measured, such as number
of base pairs sequenced per monetary unit, the number of clinically
meaningful genomic variants identified, actionable findings (both inten-
tionally sought and incident) with their attendant clinical interventions,
and (most importantly) patient outcomes. Based on these assumptions, we
could consider a model to assess potential benefits and threats of whole-
genome sequencing as far as its application in genomic medicine is
concerned, namely the prevalence of the variants found in a certain
population group, the cost of the whole-genome sequencing, the cover-
age and error rate, the number of incidental findings, the severity of
the disease(s), and the accompanying costs and expected outcomes of the
consequent medical interventions.
Overall, one could identify the following key challenges for economic
evaluation in genomic medicine. First, the gradual decrease of whole-
genome sequencing costs to less than $1,000 may discourage targeted
genomic test development. This may, at some point, be true for devel-
oped countries, but it also may be an urgent need for developing and
low-resourced countries.
Second, data interpretation and storage accompanied by the need for
expert genetic counseling may create a market niche and, as such, a value
proposition. As such, health care cost reimbursement entities, both public
and private, in various health care systems should be adapted to accom-
modate this need.
Another challenge would be the development of policies to suffi-
ciently capture personalized medical outcomes from whole-genome
sequencing. Although molecular diagnostics does not necessarily increase
lifespan, individuals would most likely highly appreciate this information
because they can adapt their lifestyle to improve their quality of life.
Because such economic evaluation approaches to capture the benefit of
reporting incidental findings to patients are currently lacking, new models
and tools have to be developed based on recently described analyses, such

as conjoint analysis and discrete choice experiments, that model the value
of knowledge/information to patients (Basu and Meltzer, 2007; Grosse
et al., 2008; Regier et al., 2009). This can be particularly helpful consider-
ing the fact that whole-genome sequencing can be performed once in a
lifetime (e.g., at birth) and sequencing results can be stored and ideally can
be readily available for health care professionals to prescribe the proper
medical interventions to patients. The latter can be cost-effective because
no extra costs have to be spent for an adult patient to obtain the results of
a pharmacogenomic test for individualizing his/her anticoagulation treat-
ment because such data would already be available for him/her since birth.
Finally, the various components of next-generation sequencing involv-
ing targeted, whole-exome, and/or whole-genome sequencing should be
modeled, particularly because it is not known how patients and reim-
bursement entities will respond to the data that will become available
from genome sequencing, especially if such data impact their daily
lifestyle. Such components would also include the need for further invest-
ment in genomics services, genetic counseling, data storage and retrieval,
services to be reimbursed or not, incidental findings, and, ultimately,
new concepts for medical treatment. The astute reader will be aware
that current economic modeling approaches are not readily applicable to
the multitude of interventions and outcomes that could be impacted by
the availability of genomic information across the lifespan (see Chapters 7
and 9). However, it may not be necessary to model all relevant scenarios
because certain common scenarios that would use genomic information
could justify the cost of the once-in-a-lifetime sequencing approach,
meaning that all the other genomic information could be considered to
be value added without incremental cost. This assumes that the informa-
tion would be used appropriately and would not result in either harm to
patients or waste of resources, which is hardly a robust assumption in the
current health care delivery environment. Nonetheless, this approach may
be feasible, but it has yet to be accomplished.
REFERENCES
Basu, A., Meltzer, D., 2007. Value of information on preference heterogeneity and indi-
vidualized care. Med. Decis. Making 27 (2), 112 127.
Bock, J.A., Fairley, K.J., Smith, R.E., Maeng, D.D., Pitcavage, J.M., Inverso, N.A., et al.,
2014. Cost-effectiveness of IL28 genotype-guided protease inhibitor triple therapy
versus standard of care treatment in patients with hepatitis C genotypes 2 or 3 infec-

tion. Public Health Genomics 17, 306 319.
Caudle, K.E., Klein, T.E., Hoffman, J.M., Muller, D.J., Whirl-Carrillo, M., Gong, L.,
et al., 2014. Incorporation of pharmacogenomics into routine clinical practice: the
clinical pharmacogenetics implementation consortium (CPIC) guideline development
process. Curr. Drug Metab. 15 (2), 209 217.
Flockhart, D.A., O’Kane, D., Williams, M.S., Watson, M.S., 2008. Pharmacogenetic test-
ing of CYP2C9 and VKORC1 alleles for warfarin. Genet. Med. 10, 139 150.
Gordon, L.G., Mayne, G.C., Hirst, N.G., Bright, T., Whiteman, D.C., 2014. Australian
cancer study clinical follow-up study, Watson DI. Cost-effectiveness of endoscopic
surveillance of non-dysplastic Barrett’s esophagus. Gastrointest. Endosc. 79 (2),
242 256.e6.
Grosse, S.D., Wordsworth, S., Payne, K., 2008. Economic methods for valuing the out-
comes of genetic testing: beyond cost-effectiveness analysis. Genet. Med. 10 (9),
648 654.
Gullapalli, R.R., Lyons-Weiler, M., Petrosko, P., Dhir, R., Becich, M.J., LaFramboise,
W.A., 2012. Clinical integration of next-generation sequencing technology. Clin.
Lab. Med. 32 (4), 585 599.
Kilpinen, H., Barrett, J.C., 2013. How next-generation sequencing is transforming com-
plex disease genetics. Trends. Genet. 29 (1), 23 30.
Meckley, L.M., Gudgeon, J.M., Anderson, J.L., Williams, M.S., Veenstra, D.L.A., 2010.
Policy model to evaluate the benefits, risks and costs of Warfarin pharmacogenomic
testing. Pharmacoeconomics 28 (1), 61 74.
Regier, D.A., Friedman, J.M., Makela, N., Ryan, M., Marra, C.A., 2009. Valuing the
benefit of diagnostic testing for genetic causes of idiopathic developmental disability:
willingness to pay from families of affected children. Clin. Genet. 75 (6), 514 521.
Rubenstein, J.H., Vakil, N., Inadomi, J.M., 2005. The cost-effectiveness of biomarkers for
predicting the development of oesophageal adenocarcinoma. Aliment. Pharmacol.
Ther. 22 (2), 135 146.
Squassina, A., Manchia, M., Manolopoulos, V.G., Artac, M., Lappa-Manakou, C.,
Karkabouna, S., et al., 2010. Realities and expectations of pharmacogenomics and
personalized medicine: impact of translating genetic knowledge into clinical practice.
Pharmacogenomics 11 (8), 1149 1167.
Vogel, F., 1959. Moderne probleme der humangenetik. Ergebn. Inn. Med. Kinderheilkd.
12, 52 125.
CHAPTER 4
Introduction to the Technical

Issues of Economic Evaluation
INTRODUCTION
Cost-effectiveness analysis (CEA) is defined as an analytical technique
intended for the systematic comparative evaluation of the overall cost and
benefit generated by alternative therapeutic interventions for the manage-
ment of a disease (WHO Guide to Cost-effectiveness Analysis, 2003). The
cost associated with the different treatments is measured in monetary units,
whereas the benefit from the interventions is estimated in similar physical
units that can be determined objectively by suitable measurements.
Therapeutic interventions usually diminish the impact of the disease, elimi-
nate its symptoms, improve quality of life, and also prolong survival when
this is possible. However, society’s health care needs increase over time; the
associated costs are too great and the available resources are too limited to
meet current needs.
The idea of comparative evaluation of alternative interventions is fun-
damental in economics and is related to the way in which economists
assess the value of the relevant options. Economic evaluation, a major
part of which is CEA, is a standard approach to the problem of evaluating
alternative interventions through their opportunity cost. This term is used
to describe the process of evaluating different health technologies by con-
sidering the opportunity cost, meaning the best possible alternative
options at the societal and the individual levels (Palmer and Raftery,
1999). This cost should not be confused with the explicit cost of acquir-
ing a technology, which is constant and determined based on historical
information; instead, it should be evaluated periodically because it is a
function of current practices and changes over time as new treatments
become widely available to the general population.
The application of CEA allows such a comparison to be made, and
the limited resources that the health systems can allocate can be used
effectively based on optimization behaviors grounded in the tenets of

economics (Canning, 2009). As already mentioned, it is still (at a theoret-

ical level) a basic tool for the rational distribution and a central point of
reference for health economics methodology. This type of analysis serves
three primary goals: determination of the price of a technology; defini-
tion of the level of insurance compensation; and drafting of guidelines to
be used as guides for health care professionals when prescribing.
UNIQUE ASPECTS AND DIFFERENTIATION OF CEA

FROM ALTERNATIVE APPROACHES
First, this method focuses on evaluating an intervention in physical units
but is more interested in determining life expectancy (Ramsey et al.,
2005) and/or quality of life (in which case it is called “costutility analy-
sis”; Torrance, 1986; O’Brien and Gafni, 1996); this means that the
method deals primarily with the patients’ ultimate health outcomes instead
of intermediate indicators. Such intermediate/clinical indicators are of
interest only to the clinical scientist when making a diagnosis, proposing
treatment, or deciding on future research, but are not particularly useful
for economic evaluation except in the few circumstances when an inter-
mediate indicator is associated with a health outcome through a robust
chain of evidence.
As an example, the benefit achieved by administering insulin to a
patient with diabetes could be defined in terms of reducing the levels of
glycosylated hemoglobin; other such indicators could be the reduction of
hypertension or low-density lipoprotein (LDL) cholesterol in cardiac
patients. All of these indicators are aimed at reducing a risk factor and,
thus, at leading to increased survival or better quality of life. Instead, eco-
nomic evaluation deals with clinical intermediate indicators only to the
extent necessary to convert them into ultimate health outcomes (usually
through statistical models).
Another example is the measurement of the progression-free
survival (PFS) in oncology. This is of particular clinical interest because
it defines the period over which the patient is free from disease pro-
gression, but it is not used indiscriminately in economic evaluation
because it is often unable to demonstrate the physical equivalent of
clinical observation, such as prolongation of life or improved life
expectancy (Sullivan et al., 2011).
In this sense, economic evaluation “ignores” a drug’s mode of action,
its route of administration, or its pharmacokinetic properties. Therefore,
Introduction to the Technical Issues of Economic Evaluation 39
provided a certain technology has been approved by the competent

authorities for use in the general population, it is assessed by the single
indicator that is of the most interest for insurance carriers and for indivi-
duals in charge of health care budgets: patient survival (and the relevant
quality of life). With this “flexible” (but objective) approach, it is possible
to calculate the comparative cost and benefit generated by completely dif-
ferent health technologies targeting the same disease, as long as the out-
comes are measured on the same scale (months or years of life) and the
cost is presented in a common unit of measurement (h or $).
At the same time, economic evaluation is distinct from biostatistical
analysis, which determines the (clinically oriented) difference in survival
between alternative interventions. In this case, the indicator used when
dealing with patient data is the median survival (Collett, 2003) for each
sample; the statistical significance of their difference is estimated, usually
by use of survival curves, whereas economic evaluation deals exclusively
with mean survival (National Institute for Health and Clinical Excellence,
2011). The use of median survival is justified in statistical analysis because
mean survival is often falsely elevated. This happens because some patients
will live for a long time after responding to a treatment, thus increasing
mean survival, whereas median survival is quite lower. In such cases of
right-skewed samples (Gray et al., 2011), it would be incorrect to use the
mean because this would be overestimated and, therefore, so would the
relevant intervention’s effectiveness (and, accordingly, its cost).
In economic analysis, the mean is used instead because it is the most
suitable measure for calculating the actual economic effects of the adop-
tion of a new technology by society (Barber and Thompson, 1998).
In this sense, the mean does not accurately reflect, from a clinical stand-
point, the differentiation between interventions, nor does it examine how
the life-years gained are “shared” among the patients receiving treatment
in a statistical sample or in the population. However, one of the basic sta-
tistical properties of the mean is that it equalizes the distances (differences)
between the highest and lowest values, thus permitting an approximation
of the true overall benefit or cost of an intervention, whereas the median
as a robust estimator is not affected by outliers as a measure of central
tendency. The problems of asymmetry and of inductive analysis with
inclusion of the mean are addressed with the use of recursive simulation
techniques (bootstrapping when involving patient data or Monte Carlo
for distributions; Efron and Tibshirani, 1993). Extensive discussions of
these methods are beyond the scope of this book.
AN APPROPRIATE MEASURE FOR CEA

In the technical application of the method, the appropriate measure is the
incremental benefit obtained by a therapeutic intervention (usually an
innovative treatment) and the associated cost in comparison with the
treatment that is considered to be the gold standard for that particular dis-
ease, and the ratio of their differences (Drummond et al., 2005). The
concept of the marginal quantity, which is used in classical economics to
perform analyses associated with alternative options, is not used in health
economics for practical reasons because health technologies do not pro-
duce results infinitely divisible at the physical level. The mathematical
concept of the derivative for infinitesimal changes, which is a feature of
marginal quantities, can be adequately replaced by incremental quantities
without loss of generality for method application.
The size of this difference ratio, mentioned previously, is the major
criterion for rejecting or accepting a new treatment. The difference in
average cost-effectiveness ratios (ACERs) for the alternative treatments
may appear to be a consistent indicator, but it does not describe the
true benefits of the new technology (Detsky and Naglie, 1990). In a
usual case, the mean cost of the cheaper and older technology is lower
compared with that of a new innovative treatment, but its administra-
tion is associated with reduced survival for the patient and, therefore,
lower societal utility. The patients, their families, and the organizations
managing public health look to technological progress and the attendant
life prolongation, and so they need to know the additional amount that
must be invested to adopt the new technology. This incremental cost in
comparison with the benefit from the new technology (which is the
appropriate measure) partly reflects the rate of return on the industry’s
R&D investment, the reward for innovation, and society’s preference
for the adoption of new technologies under financially viable terms.
Therefore, and despite the fact that the new technology is associated
with greater mean costs, as expected, it is also associated with greater
benefit as the society requires.
Another issue arises when the mean cost is used for the comparison of
mutually exclusive alternative interventions. In some cases, the option of
nonintervention (the “do-nothing” option) for the treatment of a disease
is associated with benefits at no cost. In such a case, the calculation of the
mean cost will arbitrarily alter the result and the calculated indicator will
be misleading, as shown in Figure 4.1.
Cost
C
A
Benefit
Figure 4.1 Mean versus incremental cost and the “do nothing” option.
According to the diagram, the mean cost associated with option B

is lower than option C (because they lie on lines with different slopes);
therefore, it should be “preferred” over the latter based on the simplistic
criterion of mean cost per unit of benefit. Nevertheless, a society that
wishes to improve the level of health for its citizens might choose point C,
which is associated with greater benefit (survival). Additionally, point A
has zero cost because nonintervention is associated with a small benefit
and, by this reasoning, it should be preferred over the other two. In reality,
society will choose between option A and option C or a linear combina-
tion of the two if the two technologies were theoretically fully divisible,
considering that the border of the technologies’ productive capacities is
defined by the line segment connecting these two alternatives. Option B
can never constitute optimizing behavior, as already shown (Cantor, 1994),
because it lies within the convex polygon defined by the origin, option A
and option C. In such a case, a true decision criterion would be the slope
of the line defined by these points A and C. If this slope (threshold), which
indicates the additional cost of an additional year of life, were attractive
based on society’s priorities and strategic choices, then the new technology
would be adopted; otherwise, we would be led to the option of noninter-
vention (point A).
CEA DECISION CRITERIA

Let us consider two different treatments, T (new treatment) and S (standard
treatment), each associated with a specific effectiveness (E) and cost (C)
for the management of a disease. ET, ES, CT, and CS correspond to the
mean effectiveness of the new treatment, the mean benefit of the standard
treatment, the mean cost of the new treatment, and the mean cost of the
standard treatment, whereas ΔE 5 ET ES and ΔC 5 CT CS are the
definitions of the differences in cost and effectiveness, respectively.
Comparative evaluation will give the following four possible scenarios
(Drummond et al., 2005):
A. ΔE 5 ET 2 ES . 0 and ΔC 5 CT 2 CS . 0
This is the most common case. Usually, an increase in mean sur-
vival with the innovative treatment and a corresponding increase in
overall cost associated with its administration are observed. In recent
years, new discoveries have been made continuously as new drugs,
new biological and gene therapies, new diagnostic methods, targeted
treatments, or new procedures have been developed. Technology, in
most cases, significantly increases the cost of services because of the
great developing, purchasing, and operating expenses; the price of
technology (which is a fraction of the overall cost of the intervention)
tends to incorporate these costly processes. In such cases, there should
be a criterion by which to assess whether the increased cost is justi-
fied by the additional effectiveness (Figure 4.2, Box A).
B. ΔE 5 ET ES . 0 and ΔC 5 CT CS , 0
In this scenario, the new treatment dominates because it provides
more effectiveness and is associated with lower cost than the standard
treatment. In this case, the previous technology must be completely
discarded and the new treatment must be fully adopted by the health
system (Figure 4.2, Box B).
Medical and/or Cost of care Cost of care Cost of care

service outcomes decreased unchanged increased
Improved B A
Unchanged
Worsened D C
Figure 4.2 The value framework, adjusted. Boxes AD depict various scenarios
(see text for details).
C. ΔE 5 ET ES , 0 and ΔC 5 CT CS . 0
This unpromising scenario for the new technology includes
increased cost but lower effectiveness from its use compared with the
standard treatment. According to the applicable terminology, the new
treatment is dominated by the older one and should be discarded
because it is not in the society’s best interest and is not a good health
investment (Figure 4.2, Box C).
D. ΔE 5 ET ES , 0 and ΔC 5 CT CS , 0
In this fourth scenario, the difference in survival is for the standard
treatment, even though the use of the new treatment is associated
with resource savings. In scenario D (as in scenario A), the ultimate
decision to adopt or reject the new technology will be based on
weighing the savings (which is the desirable outcome) and the
reduced effectiveness (which is a negative consequence) when com-
paring treatments. This criterion is essentially the same as in the first
scenario, but the level of the indicator may differ (Figure 4.2, Box D).
A special case of these scenarios is the comparison of costs only when
clinical practice has failed to show a difference in favor of one of the
compared technologies. In this case, the treatment associated with the
lowest overall cost will be selected. This type of analysis is called cost-
minimization analysis and is not considered a complete economic analysis
because it does not examine the relative benefit of the treatments.
Nevertheless, we should note that failure to demonstrate a statistically sig-
nificant difference between treatments in a study does not necessarily
mean that the interventions are equivalent. “Statistical significance” is a
widely used concept but is based on an arbitrary level of significance
determined from the literature. Economic evaluation, however, is inter-
ested in the mean estimate of quantities (i.e., the mean expected differ-
ence between interventions) (Claxton, 1999). Let us imagine a very
simplistic example. Assume that we live in a world without costs; in
which case, the only thing we would want to compare between two ther-
apeutic interventions would be their effectiveness measured in “years of
life gained.” As in standard statistical practice, if the confidence interval of
the difference in survival between interventions includes zero, then we
may say that there is no statistically significant difference between treat-
ments, at a possibility of 95%, for the population from which the samples
were taken. In practice, however, we are not interested in inductive con-
clusions; rather, we need to decide whether to administer one treatment
or the other, disjunctively (i.e., with no intermediate alternative). In such
a case, we would select the treatment with the greatest expected mean
value, regardless of the confidence interval. Therefore, it is possible to
perform CEA even without a statistically significant difference in survival
between interventions (Briggs and O’Brien, 2001).
If, however, the equivalence of the interventions is assured, then com-
paring costs will reveal the maximizing option and will lead to the adop-
tion of the least expensive technology. In the marginal case where neither
cost nor benefit differs, although the patients’ quality of life is—or may
be reasonably assumed to be—equal, the economic analysis is unable to
provide an answer and other clinical and societal criteria must be consid-
ered to reach a decision. The cases described here are presented graphi-
cally in Figure 4.3 (Black, 1990).
The axes show the difference between cost and benefit for interven-
tions T and S; the differences are indicated with the Greek delta (Δ).
Quadrant I represents the first scenario. The new technology increases
survival as well as cost compared with the standard treatment. If the addi-
tional cost is not forbidding, then the new technology will be adopted;
however, if the cost is considered excessive, then it will be rejected.
There is also a “gray zone” where the additional benefit is associated with
higher cost and the result will be uncertain until the expense considered
II + New I
Δ- Cost
technology
rejected
?
New technology New technology

rejected accepted
Δ- Effectiveness – Δ- Effectiveness +
New technology
rejected
New technology
? accepted
New
Δ- Cost –
technology
accepted
III IV
Figure 4.3 The cost-effectiveness diagram.

acceptable for an additional year of life is fully quantified. Scenario B is

represented in quadrant IV, where T is cheaper and better. No additional
criteria are needed here; the new treatment should be selected immedi-
ately and reimbursement for S should be discontinued. The same logic is
followed for scenario C, where T is more expensive and less beneficial;
therefore, T should be rejected. Quadrant III represents scenario D,
where the resources saved by treatment T are associated with reduced sur-
vival. In this case, too, we would need a fully defined quantitative crite-
rion to compare treatments. If there is no difference between treatment
benefits, then we move along the vertical cost axis and we compare only
the costs of the interventions. If T is more expensive, then we are in the
upper area of quadrant I and in quadrant II, in which case the treatment
is rejected; otherwise, we are located in the lower part of quadrant IV and
in quadrant III, in which case we accept the treatment based on the cost-
minimization analysis mentioned previously.
Note that according to the diagram, the angles of the slopes (thresholds)
for scenario A and scenario D are identical, indicating that the additional
investment we are willing to make to obtain an additional year of life is equal
to the savings we will demand for losing 1 year of life. However, as a society,
we are continuously less willing to lose 1 year of life compared with standard
treatment (Morrison, 1998), and usually the amount of savings required is
quite larger than the amount we would be willing to pay to obtain an addi-
tional year of life; it can be as high as twice the first in the case of health ser-
vices (O’Brien et al., 2002). This case is represented in Figure 4.4.
Figure 4.4 reveals that the area of rejection of the new technology is
now greater in quadrant III because the indifference angle is smaller than in
quadrant I. The mathematical formula that quantifies the ratio of differences
is called incremental cost-effectiveness ratio (ICER) and is described as follows:
ICER 5 ΔC=ΔE or ICER 5 ðCT 2 CS Þ=ðET 2 ES Þ
By rearranging this formula, we find that:
1
ICER 5 ΔC 5 NNT 3 ΔC
ΔE
The number needed to treat (NNT) symbol is known from epidemi-
ology as the number of patients who need to follow the new treatment
for an additional year of life to be obtained (http://www.cebm.net/index.
aspx?o51044). By this logic, ICER is the product of the NNT multiplied
by the individual incremental cost.
+
II I
Δ- Cost
New technology
rejected
New technology
rejected
New technology
accepted
New technology
rejected
New
technology New technology
rejected accepted
New
technology
Δ- Cost –
accepted
III IV
Figure 4.4 The curved-threshold cost-effectiveness diagram.
The ratio of average cost differences (ACER), which we mentioned pre-

viously, is calculated here, and it is obviously not equal to the ICER in the
general case, except for specific values that make them equal to each other.
ACER 5 ðCT =ET Þ ðCS =ES Þ 6¼ ðCT 2 CS Þ=ðET 2 ES Þ or ACER 6¼ ICER
The amount a society is willing to pay to obtain 1 year of life is called
willingness to pay (WTP) and is represented by the Greek letter lambda (λ)
(Pauly, 1995), whereas the willingness to accept (WTA) expresses the savings
a society demands for losing 1 year of life, as mentioned previously. The
λ is the slope of the line dividing the cost-effectiveness diagram in two. If
the ICER is more than λ, then the new technology will be rejected;
however, if it is smaller than the new technology, then it is a socially ben-
eficial option because it is cost-effective. If the WTP is greater than the
ICER, then the formula can be rearranged as follows:
ΔC
λ . ICERηλ .
ηλχΔE . ΔC or λχΔE 2 ΔC . 0
ΔE
The increase in benefit is greater than the increase in cost for points
within the first quadrant, given that λ is by definition a positive number,
whereas if ΔE , 0 and ΔC , 0, then the term λχΔE is a negative num-
ber and the term 2ΔC is a greater positive number; therefore, the cost
(savings) is greater than the benefit (obtained by S) and treatment T is
again a cost-effective option.
In reality, ΔE and ΔC are not simply two data points (values); they
are actually distributions with an estimated mean and a standard devia-
tion (just like the ICER). Consequently, in some cases it is convenient
to alter the ICER equation with a suitable linear transformation.
Because the ICER is an index (ratio) of two other statistical measures
(ΔC and ΔE), it does not follow any known statistical distribution;
instead, it presents certain assessment difficulties when the difference
between treatments is small. In this case, the limits become infinite,
which complicates the statistical manipulation of the results. If S has a
marginally greater benefit, then the limit of ICER tends to negative
infinity; however, if T has slightly greater survival, then the ICER tends
to positive infinity. Let us assume, for example, a value of
ΔC 5 h10,000 and of ΔE 5 0.001 or ΔE 5 20.001. At these values,
the ICER obviously tends to infinity. The ICER presents one additional
difficulty: it does not provide clear information about which of the
treatments being compared should be adopted if we do not already
know the quadrant we are in, such as if the signs of ΔE and ΔC are
unknown. In this case, problems in estimating 95% confidence intervals
would arise if we were to classify negative ICER ratios together with
positive ones.
To overcome such difficulties in difference ratios, the following trans-
formation has been proposed:
λχΔE 2 ΔC 5 NMB
where NMB is the net monetary benefit arising from the difference in benefit
and in cost for the treatments being compared and for a given value of λ.
A therapeutic option in this case is considered acceptable if (and only if)
NMB . 0, regardless of the sign of ΔE.
A diagrammatic description for NMB and λ compared with the
ICER is given in the cost-effectiveness diagram (Figure 4.5, where bλ is
the NMB).
The diagram shows that the slope of the WTP (i.e., the λ) is greater
than the slope of the ICER, which proves that option A is more
+ New technology
II I
Δ- Cost
rejected B
WTP
bλ
New technology
rejected
λ A
ICER = ΔC / ΔE
New technology
rejected
New technology
New technology accepted
accepted
Δ- Cost –
III IV
Figure 4.5 The NMB in the cost-effectiveness diagram.
advantageous for the society. The society would be willing to marginally

accept up to option B, which is associated with greater cost than A for
equal intervention effectiveness. The vertical difference between the two
points expresses the net incremental benefit from the new technology. As
a result, the most significant advantage of NMB is the fact that it is
expressed in linear form, it is drawn as a straight line on inductive analy-
sis, and its statistical manipulation is easier because of its linearity and the
fact that it can be performed using the analytical approach of classical sta-
tistical distributions to calculate 95% confidence intervals (Willan and
Lin, 2001).
The variance (V) of the NMB can be defined with the following
equation:
V ðNMBÞ 5 λ2 V ðΔEÞ 1 V ðΔCÞ 2 2λCOVðΔE; ΔCÞ
and the 95% confidence interval can be defined as:
qffiffiffiffiffiffiffiffiffiffiffiffi
ðNMBÞ 6 zða=2Þ σ2NMB
where z represents the normal distribution, σ2 represents the variance,

and σ represents the standard deviation. The same concept of linear trans-
formation can also be applied to the health benefit. In this case, all quan-
tities are converted to health care terms (terms of benefit, not cost as
before). This approach is equivalent to the one we presented, but the
quantity we estimate is called NHB (net health benefit). Generally speaking,
the NBM is superior to the ICER, with the only difficulty being that the
calculations of variance and confidence intervals require that the λ be
known beforehand.
THE CONCEPT OF EXTENDED DOMINANCE

Based on what we have seen so far, if a treatment is more effective and
less expensive compared with a competitive treatment, then it is said to
dominate. However, we should examine another type of dominance that
occurs when comparing three or more options.
Let us start at the beginning. Usually, innovative health technologies
with much greater costs are developed over time. This case is depicted in
Figure 4.6.
Treatment comparison
High cost
Drug E
Time 3: Drug C
Drug D
LC Expensive?
Time 2: Drug B
Time 1: Drug A
Time 0: Do nothing High effectiveness

Figure 4.6 The concept of extended dominance.
It should be noted that the new health technologies (drug A, drug

B, and drug C) are associated with much greater costs when they pro-
vide greater effectiveness. Note that the transition from drug B to drug
C, for example, which achieves little additional effectiveness, requires
the sacrifice of a large amount of resources (large additional cost); this is
not the case for drug A and drug B, and even less so for “do nothing”
and drug A. This situation is quite common today as new technologies
are developed, and this puts into question the future survival of health
systems. To achieve a small additional benefit, very large amounts of
resources need to be sacrificed. Technically, in this case the ICER
between option “do nothing” and drug A is smaller than the ICER
between drug A and drug B, which is smaller than the ICER between
drug B and drug C. The condition of extended dominance occurs
when an option is dominated not by an alternative option, as seen in a
previous chapter, but by a linear combination of two other options. For
example, drug D is dominated by the linear combination of “do noth-
ing” and drug C at linear combination and is therefore “extendedly
dominated” (the same thing is true for the linear combination of A and
C). Visually, it lies to the left of the line connecting options “do noth-
ing” and “drug C,” whereas its effectiveness is between those two
options. In such a case, we could achieve the same benefit as that
achieved with drug D, but by expending much fewer resources if we
were to use a combination of technologies “do nothing” and “drug C.”
Note that this does not occur for any other drug between A, B, and C.
Visually, when any new technology lies to the right of any linear com-
bination of the other possible options, there is no extended dominance.
If it lies to the left (and “between” them with regard to effectiveness),
then there is extended dominance. So a simple way to check for this sit-
uation when comparing options is to make a diagrammatic representa-
tion of our options. If this situation does exist, then all extendedly
dominated (D) or dominated (E) options should not be adopted by the
health system. E is dominated because it is the least effective and most
costly of all the options.
Another analytical method is to record all the available options (e.g.,
in an Excel spreadsheet) by order of increasing effectiveness and indicate
the associated cost for each one. If the transition to a more effective tech-
nology is associated with lower cost, then we have a case of dominance.
After excluding these options, the ICER for the remaining options is cal-
culated. If each ICER is greater than the next, then we have a case of
extended dominance and those options should also be disregarded.
An example of the calculation of dominance and extended dominance

over five available options is given. Assume the following options:
Step 1: Rank order of all options by increasing benefit.
Option Benefit (years) Cost (h)
A 12 12,500
B 13 13,000
C 11 23,000
D 10 24,000
E 8 8,000
Result:
E 8 8,000
D 10 24,000
C 11 23,000
A 12 12,500
B 13 13,000
Step 2: Remove all directly dominated options.

Result:
Option D is directly dominated by option C because it offers less
benefit at greater cost. Option C is also directly dominated by option
A for the same reason. Therefore:
E 8 8,000
A 12 12,500
B 13 13,000
Step 3: Calculate the ICER values.

Result:
Option Benefit Cost
(years) (h)
E 8 8,000 1,000a
A 12 12,500 1,125
B 13 13,000 500
a
This ICER is calculated in comparison to the “do
nothing” option, the cost and benefit of which are set
to zero.
Step 4: Check if each ICER is smaller than the next; otherwise,

remove the extendedly dominant options.
Result:
Option A has an ICER greater than the next option; therefore, it
is dominated by a linear combination of E and B. For example,
(50% 3 E 1 50% 3 B) gives (8/2 1 13/2) QALYs (5 10.5 QALYs)
at a cost of h8,000 3 50% 1 h13,000 3 50% (5 h10,500), whereas
the “purchase” of 10.5 QALYs with option A requires a cost of
(10.5/12) 3 h12,500 5 h10,937 and is therefore more expensive, as
expected. For a more realistic version of the example, let us imagine
that we were to give treatment B to 50% of our patients and treatment
E to the remaining 50% of patients.
The available options would then be as follows:
E 8 8,000
B 13 13,000
In conclusion, the ultimate choice between options E and B

depends on the λ, that is, on whether society is willing to pay an
additional h5,000 for 5 life-years per patient. It should be noted, how-
ever, that although extended dominance is an attractive concept in
economics because it allows us to determine how to achieve the same
benefit with the lowest possible cost, it also has a few disadvantages.
For example, to achieve the same benefit with a linear combination of
technologies, we need to use a health technology with a demonstrably
higher benefit (B) for a certain percentage of patients and a less effec-
tive (but cheaper) technology (E) for another segment of patients.
This raises the ethical issue of equal patient treatment and does not
provide an accurate criterion for identifying the patients who are to
receive one or the other treatment. The patients themselves, who care
primarily for their own welfare and not at all for financial models, will
try in every case to be included in the patient segment that receives
treatment B. In addition, the owner of technology A, who will obtain
demonstrably higher benefit compared with option E, will not readily
resign themselves to being excluded from the market based on the
abstract concept of “linear option combinations” and will attempt by
any legal means to be included in the state reimbursement lists and
maintain their market share.
SELECTION OF ALTERNATIVE OPTIONS DURING ECONOMIC

EVALUATION
Selecting one of the alternative options under comparison is important
for applied research. Such a comparison may be strictly limited to include
only a new technology and the technology that was its main competitor
until then. A comparison may also be made against no treatment or
against the best supportive care. If justified by the type of study, then the
comparison may also include the least expensive treatment (e.g., a generic
drug) or the most effective treatment in terms of survival. It is obvious
that an alternative should be selected with consideration of the practices
followed in each individual country (Drummond et al., 2005).
VALUING THE QUALITY OF LIFECOSTUTILITY ANALYSIS

Costutility analysis uses various indices and tools to measure the quality
of the patient’s life to adjust the result according to patient quality of life.
A common measure of effectiveness is the “quality-adjusted life-year”
(QALY). Quality is often measured on a scale of 01, or of 0100,
where 0 is the “worst possible” and 100 is the “highest or best possible”
state of health. A “QALY” is a period of 1 year weighted by the quality
of life that the patient is experiencing when suffering from a disease or
when improving as a result of a treatment. For example, if a prostate can-
cer patient is found to have 75% quality of life, then 1 year of life with
prostate cancer is equivalent to 0.75 years of life with perfect health (0.75
QALYs). If the patient improves to 90% after treatment, then 1 year of
life after treatment is equivalent to 0.9 years of life in perfect health, and
the treatment benefit is 0.15 years of life.
Various methodological tools are used to value a patient’s health state
and quality of life. Some of these are specialized for specific diseases,
whereas others seek to evaluate a patient’s general state of health. Some
are based on simple indices and others are more comprehensive but also
more difficult to assess. The subjects in such studies are usually patients,
but they may also be health professionals, such as nurses or physicians, or
the general population. Quality assessment may be done directly or indi-
rectly through the use of certain characteristics of the treatment groups
and the creation of empirical utility functions by professional investiga-
tors. Examples of such efforts are the EuroQol EQ-5D (http://www.
euroqol.org/eq-5d-references/reference-search.html), the Health Utility
Index (HUI; http://fhs.mcmaster.ca/hug), the Quality of Welfare Scale

(QWB), and the SF-36 (http://www.sf-36.org/tools/sf36.shtml). Because
of the importance of the quality of life, and because this type of analysis
will (in theory) facilitate broad comparisons between different medical
interventions by reducing them all to a common measure of value (the
QALY), costutility analyses are becoming more and more common,
and many organizations such as the UK National Institute of Health and
Care Excellence (NICE) encourage their use. A simple example of an
ICER calculation in such analyses is given.
Survival Quality QALYs Cost (h)
(years) of life
Treatment A 5 50% 550% 5 2.5 10,000
Treatment B 4 60% 4 3 60% 5 2.4 11,000
ICER 5 ð11; 000 10; 000Þ=ð2:5 2:4Þ 5 h10; 000 per QALY
This example should make it clear that 1 QALY is equal to 1 calendar

year only for people in perfect health. Let us repeat that 1 QALY is “1
year of life for a person in perfect health” or, more accurately, “the equiv-
alent of 1 year of life in perfect health.”
We consider the methods used for quality assessment. The careful
reader will certainly note that such assessments are not free of disadvan-
tages and are heavily subjective. The latter should not surprise us because
economics, by its nature, is the study of the “production of useful com-
modities” (health services in our case) under specific conditions of “pref-
erence” by the “consumers” (patients).
The simplest method of utility assessment is the visual analog scale
(Figure 4.7). In this case, we ask the patients to evaluate their health on a
100-point scale like the one in this chapter. The advantage of this method
0 10 20 30 40 50 60 70 80 90 100
Worst imaginable Best imaginable

health state health state
Figure 4.7 The visual analog scale.
is its calculation simplicity because it allows even a relatively inexperi-

enced investigator to assess the patient’s quality of life. Its disadvantage is
the fact that it does not offer the subject the option to “trade” states of
health (no questions like: “Would you prefer state X or state Y?”) and
therefore tends to make the subject give more pessimistic answers.
The second type of analysis is the typical game (Figure 4.8). In this
case, we ask the patients to take part in a “game.” We ask whether they
would prefer their current state of health (option 1) or to participate in a
game in which they would have P chances of being completely cured and
(1 2 P) chances of dying (option 2). When we determine the possibility P,
which makes the patients indifferent to the two options (their current state
versus the game), then essentially the relevant quality of life has been
assessed. For example, if patients become indifferent to whether they con-
tinue living with their specific condition or they are given a 90% possibility
of becoming perfectly healthy and a 10% possibility of dying, then their
quality of life is determined to be 90%.
Another method of quality assessment is the time trade-off method.
In this case, we ask the patients how many years of their life they would
be willing to sacrifice to become completely healthy. For example, if they
are indifferent to whether they live with their condition for 10 years or
live without the condition for 6 years (i.e., they would be willing to sacri-
fice 4 years to regain their health), then their quality of life is estimated to
be 60%. This is schematically represented (Figure 4.9). Although this
approach is simple and intuitive, in certain cases (e.g., rheumatoid arthritis)
many patients may not be willing to trade their condition for years of life;
P=1 Current state

Treatment A of health
Patient
P
Perfect health
Treatment B
1–P
Death
Figure 4.8 The typical game.

Quality of life
Complete health
With the disease
a b Time
Quality of life = a/b
Figure 4.9 The time trade-off.
therefore, a different method of quality of life assessment may be necessary

with the use of a different instrument or another “anchor point” instead
of death.
Finally, another interesting method is quality assessment through a
questionnaire. We examine the generic EQ-5D questionnaire, but the
others follow the same approach. With the EQ-5D, we ask the subjects
questions involving five dimensions of their health (hence “5D”). These
are mobility, self-care, daily activities, pain, and anxiety. The questions are
answered using a simple three-level scale of “no problems,” “some pro-
blems,” and “severe problems.” The full questionnaire is available online
at www.euroqol.org.
It should be obvious to the reader that certain answers are meaningless
or require further investigation; such an example would be a patient who
is bedridden with intense pain but expresses feeling no anxiety. The scor-
ing system of the English version of the questionnaire is described. There
are no scores available for some countries and so the investigator can select
to use another country as a basis, either the United Kingdom or some
country that they believe (or have ascertained) has prevailing preferences
that are not very different from those of the country for which there are no
scores available. One can easily search the Internet for the scoring algo-
rithm of similar instruments (as well as of the EQ-5D) in Excel spreadsheet
format. Starting from perfect health (1), one begins to automatically “lose
points” for answering “2” to any question (plus the initial loss of 20.081;
for further details go to http://www.euroqol.org/about-eq-5d/valuation-
of-eq-5d/eq-5d-5l-value-sets.html). The number of additional points lost
is given next to each question. If the patient replies “3” (severe problems)
even to only one question, then that patient loses an additional 20.269
points. The final estimate will be the patient’s quality of life (which will be
a number between 0 and 1). We should note that because such analyses are
based on statistical methods, the relevant scores will incorporate the flaws
of these methods. For example, there may be cases with scores less than
zero, cases where the loss of welfare due to the disease is nonlinear, and
other cases. A detailed presentation of the advantages and disadvantages of
such instruments is beyond the scope of this work.
In conclusion, if one wants to estimate QALYs using a clinical ques-
tionnaire (e.g., PASI for psoriasis or DAS28 for rheumatoid arthritis),
then statistical methods should be used to convert clinical estimates to
QALYs based on the literature. To do this, one would enter the clinical
estimates into an equation that includes utility and therefore calculate the
QALYs (e.g., an equation like QALY 5a b 3 PASI, where a and b are
calculated for specific samples through statistical regression).
A CONCEPTUAL APPROACH TO QALY

We should examine the QALY a little more, because this concept permits,
at least at a theoretical level, extensive comparisons to be made between
interventions originating from very different fields of clinical research.
For convenience, we might imagine that the QALY has some of the same
properties as money. For example, money can be used as a means of valu-
ation, because every product or service is measured in monetary units. By
this rationale, we can compare commodities by value. The QALY has the
same function (in its most ambitious form). A therapeutic procedure in
cardiology, a surgical operation, the treatment of oncology patients, the
treatment of patients with rheumatoid arthritis, and others are actually
completely different activities—in fact, they are essentially separate minor
scientific fields. However, the theoretical concept of the QALY allows us
to compare them and check which activity is the most “rewarding” in
QALYs, that is, it offers the most QALYs at the lowest “price.” This idea
lays the ground for the concept of allocative efficiency. One basic ques-
tion in economics is, “where should the available resources be invested to
obtain the best possible result?” and its answer is the answer to the impor-
tant issue of allocative efficiency. A reasonable approach would be to
invest more resources into “productive activities” (conditions from which
many QALYs can be obtained) and less money into the others if we were
aiming at some kind of maximization of the total QALYs with no regard
to who will benefit from these. A different approach would be to think

less “socially” and to invest money into areas where fewer QALYs are
produced, because that is probably where the “difficult” conditions
belong (scourges, e.g., aggressive forms of cancer, renal failure patients,
etc.) and should therefore be given priority. Yet another approach would
be to somehow distribute the money “horizontally” to all conditions
(e.g., by disease prevalence), thus giving all patients a chance at a future
and at life. Another approach would be to invest our money in younger
patients regardless of condition.
It is not clear which approach would let us invest our money “effec-
tively according to our preferences,” because to define our “preferences”
we must consider a number of factors such as equality, maximization,
clinical practice, the degree of societal consensus on the value of life,
moral imperatives, and national culture. From a technical point of view,
we could say that society’s “welfare equation” must be fully known and
determined to make such a decision. We also need to answer questions
such as “for whom is this result better?”; “who will be the one to deter-
mine the result?”; and “who will be the one to implement it properly, as
part of a leadership with its own needs, possibly its own separate goals,
and perhaps a short and difficult political term?”
We avoid further details regarding this issue because they might be
tiresome to the reader, but we should point out that this particular step is
often overlooked. Frequently, matters of economics are treated with a
rough, “modern,” and summary mathematical approach whose robust
assumptions are expressed in vague scientific terms and equations. The
purpose of this comment is to introduce to the reader the notion that
economics is primarily a social science and that mathematics, although an
elegant discipline, cannot possibly consider all the social parameters of a
decision involving citizens’ health care.
In view of this and to define a typical analytical approach for the
QALY, nine different assumptions about it are listed:
1. A decision needs to be made regarding the allocation of resources
2. The results of the various interventions can be measured (health
condition or improvement, survival time or duration of disease,
etc.)
3. The available resources are limited and the adoption of each interven-
tion entails costs for the system
4. An important goal of the decision-makers is to maximize the popula-
tion’s health based on economic and other restrictions
5. Health is defined in accordance with the length and quality of life

(how long and how well one lives)
6. Preferences regarding the states of health can be measured
7. Each patient is neutral regarding risk (risk-neutral) and has a utility
additive over time. In practice, the first half of this assumption sti-
pulates that the utility offered to the patient by a potential uncertain
treatment that includes favorable and unfavorable outcomes (such as
“death” and “complete cure”) is equal to the utility provided for
the “average” of the intervention. In other words, patients should
be displeased by an unfavorable development to the same extent
that they are pleased by a favorable development, and those two
outcomes are balanced. This might not be the case because people
are strongly averse to negative developments, and it is rather
unlikely that the possibility of a favorable outcome will offset their
displeasure before they make their final decision. However, society
may be considered to be neutral with regard to risk, which is a rea-
sonable approach. For example, society is interested in statistical
measures of the success of a therapeutic intervention and does not
care about which individuals specifically will survive and for how
long (in contrast, of course, to the patients themselves who wish to
minimize any risks of an intervention with regard to their own
lives). The second part of the assumption (that of additivity) implies
that the patients will add the utility they receive each day, with no
satisfaction or dissatisfaction about their outlook for the rest of their
lives. For example, if one patient’s remaining life span is 1 week (at
the equivalent of “perfect” health) and another patient’s remaining
life span is 10 weeks (also at the equivalent of “perfect” health),
then the second patient’s utility will be 10-times that of the first
patient, according to the assumption. However, this might not be
the case because the second patient has a greater absolute survival
reservoir and can therefore make arrangements for several life activi-
ties and complete activity cycles. This means that this patient will
probably enjoy greater utility than 10-times that of the less fortunate
patient.
8. Preferences may be added between individuals. This is also an ambi-
tious assumption. According to this assumption, the welfare of a social
group can be expressed collectively by adding the individual levels of
welfare. We should keep in mind that often, due to pure irrationality,
the evolution of a society takes place through conflict between social
groups, between special-interest groups and the rest of the society, and
so on, so the unquestioning acceptance of additivity should be evalu-
ated according to the problem to be solved and with consideration to
the available data in each case.
9. A QALY is a QALY regardless of who gains or loses it. The last three
axioms are obviously simplistic and research is being done to rebut
some of their features so that they will more clearly reflect the real
world and the decision of the decision-makers. For example, an ambi-
tious approach would be to establish the “weight” of a QALY
obtained from cardiology, oncology, or other field and then additively
maximize some type of social welfare equation. In any case, such
issues are rather delicate and require broad participation of many soci-
etal partners to make decisions based on mathematical models.
In concluding this practical summary, there is another “narrower”
approach to the QALY in applied research that is more commonly used
but that is less interesting from a theoretical viewpoint. This second
approach involves the investigation of differences in the quality of life of
patients with comparative treatments for the same condition. This type of
approach is quite reasonable because many new innovative treatments are
not different from older ones with regard to survival but are easier to
administer and have fewer side effects. These are all important factors and
should be taken into account.
ASSESSING THE QUALITY OF AN ECONOMIC ANALYSIS

The methods described here are important to critically appraise as one
plans an economic analysis. Lack of attention to these important issues
can result in studies that give misleading or inaccurate results. As eco-
nomic analyses become increasingly common, concern has been raised
about the “. . .validity, methodological quality, and utility of health eco-
nomic analyses as well as the potential for bias and misuse” (Ofman et al.,
2003). This is a serious problem given that many of the decision-makers
who rely on these studies are not adequately equipped to critically evalu-
ate the studies, as evidenced by the results of a survey performed by the
European Network on Methodology and Application of Economic
Evaluation Techniques (Goetghebeur and Rindress, 1999). This has
resulted in the construction of a number of instruments designed to criti-
cally assess economic studies, although the utility of such tools is limited
by their reliance on subjective items and a lack of testing of the tools
Questions Points Yes No

1. Was the study objective presented in a clear, specific, and measurable manner? 7
2. Were the perspective of the analysis (societal, third-party payer, etc.) and reasons for its
selection stated? 4
3. Were variable estimates used in the analysis from the best available source (i.e., randomized
control trial—best; expert opinion—worst)? 8
4. If estimates came from a subgroup analysis, were the groups prespecified at the beginning
of the study? 1
5. Was uncertainty handled by (1) statistical analysis to address random events and (2) sensitivity
analysis to cover a range of assumptions? 9
6. Was incremental analysis performed between alternatives for resources and costs? 6
7. Was the methodology for data abstraction (including the value of health states and other
benefits) staled? 5
8. Did the analytic horizon allow time for all relevant and important outcomes? Were benefits
and costs that went beyond 1 year discounted (3%–5%) and justification given for the
discount rate? 7
9. Was the measurement of costs appropriate and the methodology for the estimation of
quantities and unit costs clearly described? 8
10. Were the primary outcome measure(s) for the economic evaluation clearly stated and did they 6
include the major short-term was justification given for the measures/scales used?
11. Were the health outcomes measures/scales valid and reliable? If previously tested valid and
reliable measures were not available, was justification given for the measures/scales used? 7
12. Were the economic model (including structure), study methods and analysis, and the
components of the numerator and denominator displayed in a clear, transparent manner? 8
13. Were the choice of economic model, main assumptions, and limitations of the study stated
and justified? 7
14. Did the author(s) explicitly discuss direction and magnitude of potential biases? 6
15. Were the conclusions/recommendations of the study justified and based on the study results? 8
16. Was there a statement disclosing the source of funding for the study? 3
TOTAL POINTS 100
Figure 4.10 The quality of health economic studies instrument.

Source: Reprinted from Chiou et al. (2003) with permission.
construct validity (Ofman et al., 2003). This led Chiou et al. (2003) to
develop and validate a Quality of Health Economic Studies (QHES)
instrument that the authors claim will simplify the assessment of quality of
health economic evaluations. The QHES uses 16 explicit elements that
are answerable using a yes/no approach combined with a weighted scor-
ing system with a total possible score of 100. A full evaluation of the
QHES instrument is beyond the scope of the chapter; however, a copy of
the tool is provided here (Figure 4.10).
CONCLUSIONS
The reader will note that the elements included in Figure 4.10 reflect the
content of this chapter very well. This tool not only is useful in the critical
appraisal of published economic studies but also can serve as a guide in the
design of economic studies. Careful attention to the methods outlined in
this chapter and the elements from the QHES tool will improve the valid-
ity and reduce the bias of economic analyses. This approach will yield
results that are more relevant to and trusted by the intended audience.
REFERENCES
Barber, J.A., Thompson, A., 1998. Analysis and interpretation of cost data in randomised
controlled trials: review of published studies. BMJ 317, 11951200.
Black, W.C., 1990. The CE plane: a graphic representation of cost-effectiveness. Med.
Decis. Making 10, 212214.
Briggs, A.H., O’Brien, B.J., 2001. The death of cost-minimization analysis? Health Econ.
10 (2), 179184.
Canning, D., 2009. Axiomatic foundations of cost effectiveness analysis. Working Paper Series.
PGDA Working Paper No.51 ,http://www.hsph.harvard.edu/pgda/working.htm..
Cantor, S., 1994. Cost-effectiveness analysis, extended dominance, and ethics. A quantita-
tive assessment. Med. Decis. Making 14, 259265.
Chiou, C.F., Hay, J.W., Wallace, J.F., Bloom, B.S., Neumann, P.J., Sullivan, S.D., et al.,
2003. Development and validation of a grading system for the quality of cost-
effectiveness studies. Med. Care 41 (1), 3244.
Claxton, K., 1999. The irrelevance of inference: a decision-making approach to the
stochastic evaluation of health care technologies. J. Health Econ. 18 (3), 341364.
Collett, D. (Ed.), 2003. Modelling Survival Data in Medical Research, second ed.
Chapman & Hall/CRC Press, Boca Raton, FL.
Detsky, A.S., Naglie, I.G., 1990. A clinician’s guide to cost-effectiveness analysis. Ann.
Intern. Med. 113 (2), 147154.
Drummond, M., Sculpher, M., Torrance, G., 2005. Methods for the Economic
Evaluation of Health Care Programmes. Oxford University Press, Oxford.
Efron, B., Tibshirani, R.J., 1993. An Introduction to the Bootstrap. Chapman & Hall/
CRC Press, Boca Raton, FL.
EuroQol Group Association, EuroQol EQ-5Dt. ,http://www.euroqol.org/fileadmin/
user_upload/Documenten/PDF/Products/Sample_UK__English__EQ-5D-3L_Paper_
Self_complete_v1.0__ID_23963_.pdf..
Goetghebeur, M.M., Rindress, D., 1999. Towards a European consensus on conducting
and reporting health economic evaluations—a report from the ISPOR inaugural
European conference. Value Health 2, 281287.
Gray, A., Clarke, P., Wolstenholme, J., 2011. Applied Methods of Cost-Effectiveness
Analysis in Healthcare. Oxford University Press, Oxford.
Morrison, G.C., 1998. Understanding the disparity between WTP and WTA:
endowment effect, substitutability, or imprecise preferences? Econ. Lett. 59,
189.
National Institute for Health and Clinical Excellence. Guide to the Methods of
Technology Appraisal, March 2011. ,http://www.nice.org.uk/media/B52/A7/
TAMethodsGuideUpdatedJune2008.pdf..
O’Brien, B.J., Gafni, A., 1996. When do the dollars make sense? Toward a conceptual
framework for contingent valuation studies in health care. Med. Decis. Making 16,
288299.
O’Brien, B.J., Gertsen, K., Willan, A.R., Faulkner, L.A., 2002. Is there a kink in consu-
mers’ threshold value for cost-effectiveness in health care? Health Econ. 11 (2),
175180.
Ofman, J.J., Sullivan, S.D., Neumann, P.J., Chiou, C.F., Henning, J.M., Wade, S.W.,
et al., 2003. Examining the value and quality of health economic analyses: implica-
tions of utilizing the QHES. J. Manag. Care Pharm. 9 (1), 5361.
Palmer, S., Raftery, J., 1999. Opportunity cost. BMJ 318 (7197), 15511552.
Pauly, M.V., 1995. Valuing Health Care Benefits in Money Terms. Cambridge University
Press, Cambridge.
Ramsey, S., Willke, R., Briggs, A., Brown, R., Buxton, M., Chawla, A., et al., 2005.
Good research practices for cost-effectiveness analysis alongside clinical trials: the
ISPOR RCT-CEA task force report. Value Health 8 (5), 521530.
Sullivan, R., Peppercorn, J., Sikora, K., et al., 2011. Delivering affordable cancer care in
high-income countries. Lancet Oncol. 12 (10), 933980.
Torrance, G.W., 1986. Measurement of health state utilities for economic appraisals:
a review. J. Health Econ. 5, 130.
WHO guide to cost-effectiveness analysis, 2003. ,www.who.int/choice/publications/
p_2003_generalised_cea.pdf..
Willan, A., Lin, D.Y., 2001. Incremental net benefit in randomized clinical trials. Stat.
Med. 20, 15631574.
CHAPTER 5
Advanced Methodological Aspects

in the Economic Evaluation
INTRODUCTION
In the previous chapter we have provided an overview of various technical
issues of economic evaluation. In this chapter, we describe some methodo-
logical issues involved in the economic evaluation of health services. The
handling of each of these factors may be extremely important because it can
lead to entirely different evaluation results for similar health technologies. The
decision-makers must be aware of the assumptions used in each economic
model and understand how the assumptions can impact the results of the
model. Only then will they be able to use the analysis results as a practical
tool for drafting and implementing political decisions.
MODEL TYPES
Generally speaking, there are two types of analysis used in economic evalua-
tion. The first uses patient data (raw data), when the relevant records are
available, to calculate all basic economic quantities. In this case, we use
techniques borrowed from statistics (which will be explained in the next
chapter). A typical case is economic evaluation performed in parallel with a
clinical trial. In the second type, we create models simulating the course of
the disease with data from the literature. In certain cases, we might use data
both from the literature and from the available clinical trial results. For
example, in certain cases of chronic diseases, we would be interested in the
economic evaluation performed in parallel with a clinical trial (short-term),
but we would also be interested in economic evaluation in the longer-term
for the same patients; this would necessitate combining data from clinical
studies with data from available databases (registries). We also need to deter-
mine how our results would be affected if we combined our clinical trial
data with cost data from different countries.

In this case, some kind of modeling is necessary. When speaking about

“modeling” or “models,” we refer to a visual representation (usually
presented graphically) that describes the course of a disease or a clinical
activity or its treatment, with all possible intermediate options. Models
usually need to be simple so that they can be understood, but they should
also be sufficiently complex so that they incorporate certain basic features
of reality. In practice, the creation of a model is a matter of experience
and skill, but it should also result naturally from the type of question we
wish to investigate and should be validated by the clinical scientists who
work with the analyst.
The most basic form of a model is the decision tree. Such models include
“decisions” (represented as squares), “transition possibilities” (represented as
circles), “conclusions” (represented as small triangles pointing to the left), and
“alternative options” (represented as branches). Such analyses are used if it is
not actually important “when” something happens, but we are more inter-
ested in the possible outcomes that could occur. This means that such models
do not include time as a distinct parameter. A feature of these models is that
options are mutually exclusive (e.g., “surgery” or “conservative treatment,”
etc.), and the sum of the alternative options should equal 100% (e.g., the
“surgery” option has a 5% probability of “death,” 25% probability of “compli-
cations,” and 70% probability of “perfect health;” 5% 1 25% 1 70% 5 100%).
The next type of model is the Markov model. This model consists of
specific “health states,” “transition possibilities” from one state to another,
and “cycles” such as the time scale in which patients are periodically
evaluated. Markovian models are used when we would like to determine
when an event takes place and usually have a long-term scope (e.g., in
cardiology studies). For example, a model with three health states, each of
which was more severe than the next (“patient [stable disease]”, “disease
progression,” “expired”) and a yearly cycle, would calculate how many
patients each year would fall into one of these three discrete states. It is
obvious that, as time goes by, fewer and fewer individuals would be in
the first state and more and more would be in the last state. In fact, in
some cases such models only “move” in one direction (e.g., a patient
with “disease progression” has 0% probability of returning to “patient”)
and are called “absorbing models.” In the typical models, the probability
of transition from one state to the next remains constant over time, which
is not the case in the clinical sciences; therefore, mathematical modifi-
cations have been applied that address variations in the probability of tran-
sition between cycles. This approach is reasonable because the passing of
Advanced Methodological Aspects in the Economic Evaluation 67
time (at least) generally increases the patient’s odds of moving more
rapidly to a less favorable state of health (e.g., death by old age), regardless
of whether there have been any other adverse developments in the parti-
cular disease being examined.
THE SIZE OF THE ICER OR THE λ

As discussed in detail in Chapter 4, the ICER is an index that calculates
the additional cost required per unit of benefit when comparing a new
treatment with the most effective alternative for therapeutic interventions
used for the same disease. It is intended primarily to provide information
during the decision-making process in the case of more expensive and
more effective treatments, which is the most common scenario.
Nevertheless, the ICER calculation by itself does not allow conclusions to
be drawn about the cost-effectiveness of the various options. As mentioned
previously, such conclusions require a quantitative criterion (measured in
h/year or $/year), below which an option is considered effective and above
which the option is rejected. According to the neoclassical theory of the
economics of welfare, a threshold may be defined for ICER to be used as a
criterion in deciding which interventions maximize the effectiveness of any
health investment at a given time (Weinstein and Zeckhauser, 1973). This
type of neoclassical approach can be generalized and used not only for the
comparison of interventions for the same disease but also for the compari-
son of therapeutic interventions aimed at different diseases. Here, the
various health interventions are tabulated in a league table based on the
mean or median ICER, and then those with the lowest ICER are financed
first, then those with higher ICER, and so on, until the entire budget is
expended. For example, in a review of the literature involving 242 studies
(Greenberg et al., 2010), median ICER values for five different cancer
types were determined (Table 5.1).
Table 5.1 Ranking table for various types of cancer

Cancer type Median ICER (2008 US$)
Colorectal cancer 22,000
Breast cancer 27,000
Lung cancer 32,000
Prostate cancer 34,500
Hematological malignancies 48,000
According to this model, all the patients with the lowest ICER (colorectal
cancer patients) would be financed first, then breast cancer patients, and so
on. In such a case, certain expensive treatments would be left without
reimbursement because they would be significantly more expensive than
others and therefore uneconomical for the society, and resources would be
insufficient to cover all needs.
A full utilization of the model would require the existence and awareness
of the society’s welfare equation, a desire by all citizens to maximize this
equation, and the establishment of an independent authority that would
implement the citizens’ wishes through this fully mathematicized equation.
It is, however, intuitively obvious that all of these represent a theoretical
construct that does not match the reality or the needs of health care
decision-makers, nor does it account for political and special interest
influences. Furthermore, it is difficult (and possibly pointless) to compare dif-
ferent forms of cancer, let alone completely different disease types that require
different patient treatments, that are associated with different ICER values,
and also that have different severities, as mentioned in the previous chapter.
Furthermore, the development of the model assumes that the society
is using a fixed budget, that a fixed efficiency of scale in health services,
which is not affecting the ICER, may be assumed when a program is
scaled up or down, that full information is disclosed for all available treat-
ments, and that the individual programs are completely independent.
In many cases, the assumption of a fixed budget may not be true
because governments may invest more resources into an intervention
program that can appreciably improve the level of health for the popula-
tion. If the budget is indeed fixed, then the acceptable ICER will change
over time as new treatments are developed, which will alter the relative
position of the respective diseases in the league table. According to a
different version of the model, the ICER could remain fixed provided
the budget is variable. In that case, the introduction of new treatments
would have no effect on the ICER because the budget would be adjusted
accordingly by the equivalent percentage of the additional burden caused
by the introduction of the new interventions.
Furthermore, the assumption of fixed economies of scale may be stip-
ulated only if the fixed cost of a program is zero or negligible, so that any
changes in its implementation scale will lead to similar health results. The
assumption of full disclosure is a simplification because no health system,
insurance carrier, or private enterprise can ever be fully informed. In
almost every case, it is not certain what benefits will arise and what will
be the full economic consequences of the application of a program until

those are verified by real data.
Nevertheless, this type of analysis allows economical evaluation of the
interventions, which would prevent unnecessary investments in health care
if a medical intervention was not cost-effective. Another advantage of this
method is the fact that it offers the necessary information to make deci-
sions at a political level and to reveal any trade-off in cost and effectiveness,
as well as the economic burden to society for each individual option.
Therefore, this analysis can be a useful information tool that can be com-
bined with other information for the purposes of decision-making but not
as a sole comprehensive decision-making criterion independent from soci-
ety. From a limited and intentionally “naı̈ve” economic viewpoint, accord-
ing to the invariable practice of the exact science of economics (Friedman,
1953), the purpose of these analyses is to reveal relationships that actually
exist (in this case, the added cost of the new intervention necessary to
obtain an additional year of life), just like “physical” laws are determined
in the paradigm of the exact sciences, and not to deal with ethical, moral,
or humanistic issues that are not part of the usual professional activities of a
health economist; instead, they involve choices expressed collectively by
the society through its institutional representatives or through other infor-
mal means of social demands in modern democratic states.
Therefore, an alternative approach is recommended. According to this
approach, we establish a society’s WTP for 1 additional year of life. If the
acceptable WTP limit is predetermined, then all interventions with an
ICER lower than the limit would be reimbursed and all the others would
be considered ineffective. This model obviously raises ethical issues about
the way a society treats its patients because the diseases that are being most
exhaustively studied by science at any given time are usually associated with
more expensive treatments, innovative medicines, and higher ICERs. At
the same time, the arguments against the previous model also hold for this
model. In practice, it is difficult (if not impossible) to find clear examples of
optimizing behavior, because such a solution is inconsistent with Arrow’s
theorem for a social choice problem (Arrow, 1963). Generally speaking,
there cannot be a social utility function that mathematically describes a
maximization behavior for the entire population of a modern democracy,
where individualism and moral deviations or deficiencies predominate. In
many cases, such decisions regarding the distribution of resources for health
care are contingent on each country’s individual political and historical
background and are not determined by economic models.
The estimation of this indicator remains a subject of extensive debate,

and even large organizations such as the UK National Institute for Health
and Care Excellence (NICE) have yet to announce a clear decision on its
“correct” size (Towse, 2009). In various other countries, however, WTP
values have been proposed for the “purchase” of 1 year of life to provide a
transparent criterion for this difficult undertaking. According to the World
Health Organization, the desired value for the indicator is approximately
three-times the average per capita income of the country (Eichler et al.,
2004). For the United Kingdom, a value between d40,000 and d60,000 is
the maximum accepted value in most cases. A value between $50,000 and
$100,000 is considered cost-effective, a value less than h20,000 is consid-
ered particularly attractive, and values more than h100,000 are considered
particularly costly and are rejected (Devlin and Parkin, 2004).
Figure 5.1 presents the possibility of rejection of a treatment by ICER
size based on a logarithmic regression model from the UK data. A value of
h40,000h60,000 has a 50% chance of rejection by NICE. Of course, this
indicator can be adjusted for special groups such as children or patients
with rare diseases, because the social nature of health care and the social
imperative for population-wide coverage regardless of economic status or
public expenditures will prevail in such cases. A more recent study has
determined the value of λ for various countries, as seen in Figure 5.2.
1
Probability of rejection
Model 1
Model 2
0.5 Model 3
Model 4
0
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 90,000 100,000
Cost-effectiveness ratios
Source: Devlin and Parkin. Health economics 2004; 13: 437–452
Figure 5.1 Probability of acceptance depending on the ICER.
How much does a year of life cost (€)?

Table I Threshold incremental cost-effectiveness ratios
in selected countries
Country Threshold value Threshold value
in local currency in Euro
Australia AUS$42,000–$76,000 € 24,700–€ 44,700
per QALY per life year
Canada CAN$20,000–$100,000 € 12,700–€ 63,300
per QALY per QALY
England/Wales £20,000–£30,000 € 22,700–€ 34,100
per QALY per QALY
The Netherlands €20,000–€80,000 € 20,000–€ 80,000
per QALY per QALY
New Zealand NZ$ 3000–15,000 € 1400–€ 7200
per QALY per QALY
United States US$ 50,000 per QALY € 34,400
per QALY
Notes: Local threshold values were converted into Euro using market
exchange rates.
Abbreviation: QALY, quality-adjusted life year.
Figure 5.2 The amount of λ in various countries.
It should be noted, however, that the determination of λ has direct

effects on health system budgets and therefore should not be done inde-
pendently of each economy’s available funds.
With regard to λ, we should also keep in mind the following:
• Shadow prices: The λ is supposed to reflect the shadow price or the
opportunity cost for health care resources used for the purchase of health
programs. Nevertheless, from a wider economic analysis perspective
(e.g., the social perspective), it is difficult to calculate the budget
precisely, which consequently prevents the exact determination of the λ.
• Budget size: In most cases, we usually simplify by assuming that the
size of the budget does not affect the marginal cost of the resources
spent for health care. This indicates that the advantages lost in one
area by withdrawing resources compared with another area where
these resources are invested are the same for each euro invested regard-
less of sector or level of consumption.
• ICER: Because the λ represents the shadow price of the budget (or the
opportunity cost), it should be equal to the ICER for the last health
care program selected for funding. However, all health care programs
are associated with uncertainty; therefore, a deterministic value for λ
can never be defined, only a distribution of values. In fact, considering
that new programs are constantly introduced for evaluation and reim-
bursement, even the distribution of the λ will constantly change.
• The λ as a decision rule: The use of the λ as a decision rule for the reim-
bursement of health care programs entails two basic assumptions: constant
returns on scale and full divisibility. Constant returns mean that the λ is
unrelated to the size of the implemented program. In its simplified form,
full divisibility means that we can purchase any health care program on a
scale as small as needed; this is a simplified form and full-divisibility for-
mulation. Unfortunately, the conditions described here for the correct
calculation of the λ are not yet met by any health system in the world.
ESTIMATES OF EFFECTIVENESS
For economic evaluation, one needs to accurately and reliably estimate
the effectiveness of each median intervention. In many cases one can refer
to various sources to estimate the effectiveness of a treatment, and each of
these sources may have different advantages and disadvantages. The rank-
ing of such sources reflects, in part, the reliability of the data collected
and, therefore, the quality of the analysis results.
The best known source is the clinical trial, which belongs to a wider class
of studies called “controlled experiments.” Clinical trials are scientific experi-
ments involving people who suffer from a disease that assess the difference in
response between a new treatment and an alternative, for example, different
therapy regimens are selected and compared with each other (http://www.
who.int). Clinical trials are conducted in four phases with each one assessing
different aspects of the technology undergoing investigation, such as safety,
maximum tolerated dose of a drug, effectiveness, and comparison of the drug
with other alternative treatments in a large population sample. They are
considered the most reliable source of data and are currently an obligatory
crucible through which each treatment must pass to be used in the general
population. They are characterized by high internal reliability thanks to the
very strict patient inclusion criteria and the standardized documentation of
the results, they may apply double-blind randomization, and they have high
validity thanks to support by modern statistical techniques that minimize
sampling errors and bias. Their disadvantage is their low external reliability
because general population patients have lower compliance to treatment, may
suffer from other concomitant diseases, and may present different charac-
teristics than the study reference population.
The second source of data is meta-analysis. Meta-analysis is a statistical
technique aimed at summarizing results obtained from clinical trials
(Petiti, 1994). It does not constitute original research but draws its infor-
mation from the clinical trials included in the analysis. The most impor-
tant advantage of meta-analysis is that it can draw conclusions at greater
statistical power than the individual studies, because it includes greater

numbers of patients and can reveal statistically significant relationships. Its
disadvantages are that data from studies with different designs must be
treated similarly despite differences in study design, and that it incorpo-
rates the errors of the individual clinical studies. Also, it does not address
the shadow reference problem, and many meta-analyses do not actually
fulfill the criteria for a high-quality analysis and degenerate into a simple,
misguided statistical technique. Specific methodologies have been
designed to minimize these types of problems, but they will not be
described here because they are not within the scope of this chapter.
Databases (also known as data warehouses) are repositories of data and
records accumulated in the daily operations of large organizations.
Depending on the organization, they may include data encompassing much
of the care provided to patients or insured persons. They are characterized
by high external validity because they include large numbers of subjects,
but with the disadvantage that they do not provide detailed information
regarding clinical effectiveness and may include inaccurate records. A signif-
icant limitation of these organizational databases is the type of data cap-
tured. Much of the detailed patient information is captured in text blurbs
that require manual review or sophisticated natural language processing that
is subject to error and bias. Even the structured data (such as diagnosis and
medication codes) are generated for reasons more aligned with reimburse-
ment than accurately reflecting the patient’s disease process. Coding systems
also have systematic deficiencies that impact the ability to extract needed
data. This is particularly true in the case of rare disease. Finally, different
organizations may use different structured coding systems that have imper-
fect matching that interferes with data aggregation (e.g., Canada and most
of the rest of the developing world uses ICD-10 but the United States is
still using ICD-9). In countries without a national health service, such as
the United States, the availability of insurance or claims data is held apart
from clinical data and regulations inhibit the ability to aggregate these data,
leading to data voids in epidemiologic studies (Prokosch and Ganslandt,
2009; Choi et al., 2013; Meineke et al., 2014).
Medical records are records kept by treating physicians for therapeu-
tic or scientific purposes, and they may be electronic or physical. In
countries with sophisticated electronic systems, patient records from
different centers may be linked or, more commonly, kept separately at
each institution. Their greatest advantage is that they record actual
patient data instead of assumptions regarding the administration of drugs
or treatments. Their disadvantage is that, in most cases, they do not

include detailed information, and the information they do contain may
not be in the same format for all patients. In medicine, drugs may be
used “off label” (Casali, 2007), a fact that complicates the efforts to
investigate the effectiveness of individual therapeutic regimens. Such
use may be justified when there is evidence that off-label administration
may be effective, even if this has not been proven in long-term, expan-
sive clinical trials. In all of these cases, the use of medical records
reduces the degree of uncertainty and provides reliable information.
Clinical databases (also called disease-specific databases or registries)
focus on patients with specific conditions and are characterized by large
follow-up periods, but they have low external validity and one-
dimensional information.
Finally, expert panels (such as the Delphi method) are a qualitative
method for determining effectiveness based on the opinions of expert
physicians from each field (Cialkowska et al., 2008). Their greatest advan-
tage is that they are very useful when data are insufficient or not available,
or when data collection is too expensive. This is important because clini-
cians must make decisions regarding treatment of their patients even
when evidence is absent or insufficient. However, they are characterized
by high bias and it is difficult to decide who is an “expert” in any given
case. The professional guideline development process in the United States
relies heavily on the panel model. A 2011 publication from the Institute
of Medicine of the US National Academies entitled “Clinical Practice
Guidelines We Can Trust” (Greenfield and Steinberg, 2011) outlines a
series of recommendations to ensure that practice guidelines are as reliable
as possible. It should be noted that the value of information related to
assessing effectiveness has a small life span and it is necessary to evaluate
treatments regularly, because the emergence of new information can
change current medical assessments of regimens and treatments.
In any case, the new tendency when drafting economic studies is to
synthesize all available data into a unit model; in that case, we include
clinical study data, literature, and databases to reach specific economic
evaluation results.
STUDY PERSPECTIVE AND COST DETERMINATION

The concept of the perspective refers to the institution for which any
economic consequences from the alternative therapeutic interventions are
valued (Muennig, 2008). The study perspective is a key factor when deter-
mining the cost categories that will ultimately be involved in the analysis.
The approach selected will not always be the same; it will differ
depending on the purpose of the analysis. For example, if the issue
involves distributing resources between various sectors of the economy
(such as education, health, defense), then the analysis to be used will be
the cost-benefit analysis to determine all the consequences of the rele-
vant options. If, instead, the analysis focuses on distributing resources
between different sectors of health care (prevention, treatment, etc.) or
between different interventions for the treatment of a specific condi-
tion, then the consequences to be measured will be more limited. The
selection of analysis method is also affected by the person or institution
performing it (patient, hospital, insurance carrier, etc.) and by the avail-
ability of relevant data. In practice, economic evaluation is used exten-
sively for “narrow-type” analyses (e.g., alternative interventions within
a disease), and less so for interventions involving different sectors of the
economy. In this case, economic theory assumptions do not fully reflect
the individuals’ selections and economic evaluation is commonly used
as a tool and not as a complete system of values with which to make
the relevant decisions.
For example, if the study concerns an insurance carrier, then the anal-
ysis is performed from the perspective of the insurance fund and supplier
charges are examined; however, if the study focuses on all possible conse-
quences, then the perspective is social (Table 5.2). The social perspective
Table 5.2 Analysis perspective and cost determination

Fields of analysis Determination of economic consequences
perspectives
Society All types of medical and nonmedical costs

Loss of production and productivity. Invisible
(hidden) costs
Social security or private Charges related to reimbursement of suppliers and
insurance users
Health care service Various types of variable costs affecting health care
providers expenditures
Patients All types of costs, focusing on own payments and
loss of productivity
Employer/Company Charges related to insurance costs and loss of
productivity
includes all possible consequences with no regard for who pays for them.
If the analysis concerns an employer, then it would include the charges
for insurance costs and loss of employee productivity. In this case, costs
such as transfer to the hospital, feeding costs, and others, which are
covered by the employee, will not be included in the analysis, whereas
costs related to loss of productivity or temporary personnel would be
included. Currently, the social perspective is considered to be the most
appropriate for economic analyses because it includes all of the relevant
economic flows within a system. However, this should not be interpreted
to mean that the tools of economic analysis are not useful for decision-
making from other stakeholder perspectives.
When examining the inclusion of costs, we use a process through
which the resources expended for the production of medical interventions,
the human-hours consumed for providing health care, and the loss to the
productive process due to the patient’s inability to work and contribute to
the national product are calculated.
Specifically, the concept of cost includes the following (Phillips, 2005):
• Direct costs: The actual cost consumed for the intervention
• Direct health care costs: The cost caused by health care suppliers (the
total expenditures for monitoring, treatment, diagnostic tests, medi-
cation, etc., which result from the treatment)
• Direct nonhealth care costs: Expenditures arising for the patient as a
result of the disease as well as the treatment-seeking process (home
help costs, travel expenses, special diet expenses, etc.)
• Indirect cost: Financial losses that are a result of the disease and do not
include the costs for providing treatment
• Indirect cost essentially refers to the loss of productivity because of
the disease, either because of work absenteeism or because of reduced
productivity (presenteeism)
• It reflects the value of the goods that the patient could have pro-
duced if he/she had not become sick
• It usually includes lost productivity, free time, time expended by
relatives providing assistance, and other time lost
• Finally, this type of cost includes lost productivity due to premature
death
• Invisible (intangible) cost: A term that describes difficult-to-measure
consequences of the disease and its treatment
• It is due to the pain, discomfort, reduced quality of life, or other
social or moral consequences of the disease (or its treatment)
It should be noted that the concept of intervention “cost” in economic

evaluation refers to the “total resources” expended to treat the disease and
is not limited to the cost of a specific technology (e.g., the price of the
drug) being valued against an alternative. This cost may vary significantly
depending on the institution’s perspective, and it can increase considerably
as we move to broader analyses of the consequences of the disease.
Treatments in oncology and cardiology follow a specific pattern of adminis-
tration, are given in regimens together with many other drugs, and are
associated with toxicity and side effects with various probabilities of occur-
ring and very high management costs. In this case, a simple comparison of
the price of two drugs is usually misleading because it does not take into
account the effect their administration has on the overall burden to the
system through utilization of all the relevant resources, such as hospitaliza-
tion days and medication given to treat toxicities. A theoretical framework
for cost calculation is given in Table 5.3.
Cost measurement in economic evaluation is a process associated with
disputes concerning which cost variables should be included and which
should not. Any variability in the cost categories included will result in
incomparability between different economic evaluation studies. Discussions
about costs have led to the development of various guidelines aimed at
achieving consistency.
THE DISCOUNT RATE

Discounting is a quantitative technique for adapting the value of future flows
to current values. It may appear to be a simple process, but it is actually a
major issue in cost-effectiveness analysis, with significant consequences on
the valuation of alternative health interventions (Katz and Welch, 1993;
Hillman and Kim, 1995; Smith and Gravelle, 2001; Cohen, 2003).
Currently, no consensus exists among researchers regarding the appropriate
discount rate. As a rule, the selections made by economic institutions at dif-
ferent time points with regard to investing or consuming resources for health
care or other sectors are very important, both at the individual and at the
national level. Since its beginnings, economics has dealt with this important
issue, and a simple and understandable model was developed to study it
(Shane et al., 2002). We will not go into an elaborate presentation of the axi-
omatic foundation of such a model, but we should mention that, thanks to
its simplicity, it was quickly developed for use in comparing economic
choices made at different time points. In brief, the model could reduce all
Table 5.3 Estimation of the cost of a disease

Direct Direct non- Indirect costs Invisible costs
health care costs health care costs
Therapeutic Patient travel to Loss or reduction Compromised
interventions and from of income due quality of life
hospital, clinic to temporary,
or doctor’s partial or
practice complete
disability
Laboratory Family Loss of income Social and
examinations members’ and productivity professional
travel and stay for family problems
expenses members in
order to care for
the patient at
home
Medical Home help and Loss or reduction Adverse
commodities nursing care in employee psychosocial
productivity and circumstances
employer and
societal
production
Use of diagnostic Own payments Disruption of
technologies by the patient family
cohesion
Time spent by
medical staff and
other health
care
professionals
Stay and servicing
psychological variables of a technical analysis to a simple coefficient without

trying to explain human nature (Shane et al., 2002).
So, using a technical approach, we conclude that flow discounting is
justified for two main reasons: price inflation and because economic insti-
tutions maintain definite time preferences with regard to holding funds
and, consequently, to the cost of investment programs.
Inflation is the rate of price changes over time. Let P be the general
price level and let T be a unit of time (usually one calendar year); infla-
tion I would then be defined as:
dP=P
I5
T
Prices will change every year; therefore, any health expenditures made
today will have greater value compared with future expenditures. From a
different viewpoint, one could claim that as the national product grows
over time, a fixed flow of expenditure will account for a smaller percent-
age of the total income and therefore reflect a smaller burden, which
should be calculated by the discount method. In this case, the discount
rate reflects the rate of economic growth.
Furthermore, even in a world without inflation, a person would rather
have an amount of money today than the same amount at any future
time. In other words, they would rather enjoy benefits today than in the
future. This approach does not reflect any specific economic theories, but
rather it reveals the effect of psychological and social factors outside
conventional limits of rational thought in individuals or patients. A person
will consider future cash flows uncertain; therefore, short-term cash flows
are more significant than long-term flows.
To define in practice the preferred discount rate, one possible
approach would take into account the interest rate of a financial product
free from market risk, such as long-term Treasury bonds, based on the
rates of individual time preferences. A second alternative would be to
determine the funds’ shadow price, which takes into account the social
time preference rate. With this analysis, the social time preference interest
rate is determined by the total loss of current private consumption in
relation to the long-term flow of benefit that will be obtained by under-
taking a public program, specifically the health care intervention program
under study. The two methods are associated with advantages and disad-
vantages. A full theoretical presentation of the respective arguments is
beyond the scope of this chapter.
Similar arguments are presented for benefit discounting, which has
also caused disputes among economists (Drummond and Torrance, 2005).
Those opposed to benefit discounting argue that it is difficult to conceive
of a “long-term investment in health” or an exchange of cash flows by
individuals or society. They also maintain that flow discount implies that
less weight is given to future health care benefits, which means that the
current generation is given priority over future generations. The final
argument is that people have different understanding and approaches to
the monetary cost consumed at different time periods and to the health
benefits obtained by health care intervention programs.
On the opposing side, the supporters of the discount technique
claim that, after an intervention, the patient will “consume” health care
benefits over the long-term; therefore, a discount of future flows appears
to be a reasonable approach. Furthermore, nondiscount leads to logical

inconsistencies because a health program that provides a small and insig-
nificant perpetual benefit would be considered preferable (if there was
no discount), regardless of the level of the initial investment.
Additionally, and within reasonable margins, it may be assumed that the
people’s conduct reveals behaviors that “trade” health over time, such as
the avoidance of unhealthy habits (smoking, alcohol consumption, etc.),
which deprive the person from a certain utility but transfer an amount
of health to the future.
In the relevant literature, the balance leans in favor of discounting and
all leading organizations that issue guidelines on economic evaluation use
some discount rate for the benefits obtained, along with the cost. The
rate is not the same for all countries, thus reflecting a different degree of
aversion to future risk or uncertainty. The larger this rate is for costs and
benefits, the greater weight is given to short-term consequences, whereas
a low or zero rate expresses neutrality toward future risk. In many cases,
using a high rate can be dangerous for the public health because many
prevention programs have a long-term scope. A high rate might lead to
such a program being classified as “ineffective,” even if it can protect
future generations from a dangerous disease.
To circumvent all problems associated with the rate, a safe solution
would be to list the results with no discount and then perform various
sensitivity analyses using an interest rate ranging from 0% to 10% for cost
and benefits. In practice, adopting such an interest rate is not a purely
economic matter, but rather an evaluative judgment made by the society
through its institutional representatives. A comparative table of discount
rates for various countries within and outside Europe is given in
Table 5.4. In most cases, the rate is 35% and the sensitivity analysis is
between 0% and 10% (International Society for Pharmacoeconomic and
Industrial Research; http://www.ispor.com).
Regarding the technical aspect, the present discounted value (PDV)
for an investment made at a specific time n is calculated by the following
formula:
C
PDV 5
ð11rÞn
where C is the cost, r is the discount rate, and n is the year.

Table 5.4 Comparison of discount rates for cost and benefits in various countries
Cost Benefit
Baltic countries 5% 5%
Belgium 3% 1.5%
Canada 5%; SA: 03% 5%; SA: 03%
Finland 3% and 0% 3% and 0%
France 0%, 3% and 5% 0%, 3% and 5%
Germany 5%, SA (0%, 3%, 10%) 5%, SA (0%, 3%, 10%)
Hungary 5%; SA: 36% 5%; SA: 06%
Ireland YES YES
Italy 3%; SA: 08% 3%; SA: 08%
New Zealand 3.5%, SA: (0%, 5%, 10%) 3.5%, SA: (0%, 5%, 10%)
Norway 2.5%, SA: 08% 2.55%, SA: 08%
Poland 0% and 5% 0% and 5%
Portugal 5% 5%, 0%
Scotland 6%; SA: 0%—full 1.5%; SA: 0%—full
treasury discount rate Treasury discount rate
Spain 6% 6%
Sweden 3%; SA: 05% 3%; SA: 05%
The Netherlands 4% 1.5%
United Kingdom 3.5%; SA: 06% 3.5%; SA: 06%
United States Expert recommendation Expert recommendation
Switzerland 2.5%, 5%, 10% 2.5%, 5%, 10%
China Yes and no, with Yes and no, with
justification justification
Austria 5%, SA: 310% 5%, SA: 310%
Brazil 5%, SA: 010% 5%, SA: 010%
Cuba Yes, as recommended by Yes, with justification
state experts
Slovakia 7% 7%
Thailand 3% 3%
Mexico 5%, SA: 37% 5%, SA: 37%
SA, sensitivity analysis.
Source: Data adapted from http://www.ispor.com, International Society for Pharmacoeconomics and
Outcome Research (ISPOR, 2010).
If the cost of the intervention is spread equally over n years, then the
formula is generalized as follows:
C C C C
PDV 5 1 1 2 1 3 1?1
ð11rÞ ð11rÞ ð11rÞ ð11rÞn
or

1 1 1 1
PDV 5 C 1 1 2 1 3 1?1
ð11rÞ ð11rÞ ð11rÞ ð11rÞn
or

1 2 ð11rÞ2n
PDV 5 C
r
An example of the technique’s application for a 10-year investment at
a rate of 5%, beginning at year 1, is given in Table 5.5.
In this example, an amount of h10,000 invested annually over 10 years is
equivalent to h7,722 in present values. The result can be directly calculated
using the formula as follows:

1 2 ð11rÞ2n 1 2 ð110:05Þ210
PDV 5 C 5 h1;000 5 h7;722
r 0:05
For a technical manipulation of inflation and time preferences for a
health program, we can either appreciate all future flows using a single,
projected percentage of the inflation (provided it is the same for all
elements of the cost), or we can appreciate none of the flows and use a
lower single interest rate that will express time preferences only. However,
if the elements composing the cost have different rates of price change,
then either we appreciate them at the appropriate inflation rate and then
discount them at a generic rate (inflation 1 time preference) or we do not
Table 5.5 Calculation of the PDV for a 10-year health investment projecta
Year Interest rate Cost (h) Discounted cost (h)
1 0.952 1,000 952
2 0.907 1,000 907
3 0.864 1,000 864
4 0.823 1,000 823
5 0.784 1,000 784
6 0.746 1,000 746
7 0.711 1,000 711
8 0.677 1,000 677
9 0.645 1,000 645
10 0.614 1,000 614
Total 7.722/10 years 5 0.722 10,000 7,722
a
(r 5 5%).
appreciate and we instead use a weighted total discount rate that expresses
only time preference. In practice, for reasons of simplicity, prices are taken
as fixed (without inflation) and only one rate, expressing time preference
between alternative treatments, is used. In many cases, this assumption is
methodologically consistent because many interventions have brief time
scales and their price changes are negligible. Alternatively, we might say
that the discount rate reflects the combined effect of inflation, growth of
the economy, and social time preference.
SENSITIVITY ANALYSIS
The term “sensitivity” essentially refers to the way in which our results
change when we change our model’s assumptions. If sensitivity is high, then
the results vary greatly when we change certain assumptions; these assump-
tions must be very robustly established for our model to have any validity.
To put it concisely and intuitively, we could say that economic evalua-
tion is beset by many kinds of uncertainties. First, there is uncertainty about
the structure of the model we have created (structural uncertainty). For
example, is our approach to the issue the correct one? Does our analysis
reflect reality? Is our model of clinical practice incorrect? To answer such
questions, we need to critically evaluate our model and perform various
sensitivity analyses, changing its structure to judge its value.
Another source of uncertainty is heterogeneity between the various
subgroups to which the model refers (variability due to heterogeneity). If
we want to compare the cost and effectiveness of a new cardiology drug
with the standard treatment for the management of acute myocardial
infarction, then perhaps we should perform separate analyses for men and
women, young and elderly patients, patients with previous infarctions, and
so on, because these groups will give different results. Because heterogene-
ity is important for economic models, it should be included in economic
evaluation. Such subanalyses (men vs. women, etc.) are sensitivity analyses
for heterogeneity.
Uncertainty also exists between patients. We cannot be sure how long
a patient will actually live after a procedure, and even “similar” patients
do not survive for the same length of time, nor does their care cost the
same. This type of uncertainty is called first-order uncertainty. This would
theoretically decrease if data from studies with large samples or valid
meta-analyses were available. Realistically, there will always be some first-
order uncertainty because nature is inherently stochastic.
The uncertainty that is associated with the exact value of a statistical

parameter (and that is estimated by the standard deviation) is called second-
order uncertainty. This type of uncertainty is of significant interest.
For example, if we estimate that the cost of the treatment of a
chemotherapy patient is h10,000 6 2,000 (mean 6 standard deviation),
then we have entered second-order uncertainty.
Sensitivity analysis in this case is a technique that estimates the effect
that different values of an independent variable have on the end results
(Jain et al., 2011). Sensitivity analysis is very important when examining
the robustness and validity of our conclusions based on the significance
of the initial parameters (Meltzer, 2001; Yoder, 2008). It needs to be
performed mostly for three purposes:
• First, when the evaluator wants to determine the range of values in
which the proposition of the economic model is valid
• Second, to increase the model’s reliability when the input data are
elastic (e.g., when estimates are used)
• Third, to make the model evaluation convenient to the end user.
The sensitivity analysis methodology consists of three steps. First, the
uncertainty parameters are determined. Second, the range of variation is
determined. Third, the results are calculated based on the most likely
prediction as well as the “direction” of the results. This means that we exam-
ine whether one of the interventions is superior to the other according to a
certain statistical probability—usually the 95% or 99% significance level.
The most common forms of sensitivity analysis are:
• Single sensitivity analysis: Single analysis explores ICER variations
when a single variable of the model—a different one each time—is
altered. The variation values are usually within the variation range of
the confidence interval, or alternatively they can include all the values
found in the literature.
• Multiple sensitivity analysis: A multiple analysis is performed to assess
simultaneous changes in two or more variables, such as effectiveness
and cost. Similarly, the variation values are obtained from the confi-
dence intervals or determined from the literature.
• Probabilistic sensitivity analysis: Probabilistic sensitivity analysis (PSA) deals
with the significant problem of statistical estimation of quantities, as in the
example of the chemotherapy patient we mentioned previously, and
should always be included in any reliable economic analysis. For example,
when the variables examined are strongly correlated, are uncertain, or fol-
low distributions, then single sensitivity analysis is not appropriate. The
same is true for data originating from different sources.
Based on the reasoning used so far, the ICER was calculated in a

deterministic way—with only one point estimate—with no uncer-
tainty when drawing conclusions. In practice, the ICER has a proba-
bilistic nature because the types of costs and the benefit from each
intervention follow theoretical or empirical distributions (Briggs and
Fenn, 1998). Furthermore, the introduction of a new intervention
into the health system entails assuming certain risks, in which case
this type of analysis is indicated for handling the uncertainty associated
with those options. PSA takes into account the mean value, the stan-
dard deviation, and the distribution of each variable, creating thou-
sands of results under computer simulation by selecting random
cases based on the defined assumptions. It then summarizes and dis-
plays the results obtained, despite its limitations, through the use of
the acceptability curve (O’Hagan et al., 2000; Fenwick et al., 2004;
Barton et al., 2008).
We present a conceptual example of the necessity and understanding
of the method. We assume that the reader is familiar with the concept of
distribution. In this simple example, we assume that the cost consists of
only one factor (e.g., only the drug). If the cost comprised many factors,
then we would need to generalize the method for each one separately
and obtain the total sum of the individual distributions.
Let us assume that we are comparing two second-line treatments for
lung cancer. Assume that the survival for these two patient groups (and
the respective costs) is estimated as follows:
ET 5 10 months
ES 5 9 months
CT 5 h10,000
CS 5 h9,000
Simple calculations give the ICER based on the mean value, as shown
in Figure 5.3.
According to this, the ICER is h1,000 per month or h12,000 per
year. If society’s willingness to pay (λ) is even slightly higher, such as,
h12,001 or h12,002, then the new treatment is considered 100% cost-
effective, as can be seen from the diagram (Figure 5.4). The opposite is
true if the λ is marginally less than h12,000: there is a 100% chance that
the treatment is not cost-effective. It is obvious that such an approach is
quite “strict” and does not leave any room for error.
In practice, we will not restrict ourselves to an estimate of the ICER,
but we will perform many experiments to calculate more than one ICER
based on the distributions available for our variables. We should note that
ET = 10 months ES = 9 months
CT = €10,000 CS = €9,000
ICER = (€10,000–9,000) / (10–9) = €1,000 per month

Figure 5.3 An example of ICER calculation.
Point representation of the ICER

100% acceptance of the treatment as cost-effective
λ or wtp
+
ΔC
€
(€1000, 1 month)
ys
– +
ΔE
–
Figure 5.4 Point representation of the ICER.
the selection of the distribution is important, because its characteristics

will affect the realization of specific values on cost and effectiveness. Two
experiments (out of the thousands we would actually run in the case of a
real analysis) are presented in Figure 5.5.
Experiment 1
ET =11 months ES = 8 months
CT = €12,000 CS = €6,000
ICER = (€12,000–6,000) / (11–8) = €2000 per month
Experiment 2
ET = 14 months ES = 10 months
CT = €14,000 CS = €10,000
ICER = (€14,000–10,000) / (14–10) = €1000 per month

Figure 5.5 A probabilistic approach to the ICER.
Correlated Not correlated

The second variable The second variable
interval is limited interval is not limited at all
Figure 5.6 The concept of correlation in probabilistic analysis.
Note that in the two experiments we ran, the ΔC, the ΔE, and the
ICER are different each time, which means that we have moved on from
deterministic results to stochastic analysis (i.e., analysis that includes uncer-
tainty). We should, however, note that in this simple example we did not
take into account any correlation in the data, but we arbitrarily assumed that
there is no correlation. For that reason, we allow variables to move indepen-
dently along their entire distribution. In reality, correlation often appears in
the data; for example, if the cost is high, then survival may also be high.
This concept of correlation is represented schematically in Figure 5.6.
Regardless of correlation (positive or negative), by this process we
create many ICERs (as many as 5,000 or more) and plot them in a diagram
as shown in Figure 5.7, where each point represents one experiment
and, therefore, one calculation of the ICER (the diagram is based on hypo-
thetical results).
This diagram prompts the following question: How can I handle all of
this information and represent it concisely? The last step of this analysis is
to assume various values of λ within a reasonable range and use these to
find the percentage of points that are cost-effective. In this way, instead of
relying on an unrealistic approach in which the intervention would be
considered 100% cost-effective at exactly h1 above the ICER we calcu-
lated and not cost-effective at h1 below it, again with a probability of
100%, the analysis will now take a probabilistic nature. Let A, B, and C
symbolize various values of λ, which we assume (in applied analysis the
Stochastic representation of the ICER
Incremental cost
800
600 €
400
ys
200
–200
–400
–0.40 –0.20 0 0.20 0.40 0.60 0.80 1.00 1.20
Incremental effect
Figure 5.7 Dot plot of ICER representation.
Calculation of the acceptance curve
Incremental cost Tenecteplase - Alteplase

B
A
C
800
600 €
400
ys
200
–200
–400
–0.40 –0.20 0 0.20 0.40 0.60 0.80 1.00 1.20
Incremental effect
Figure 5.8 Derivation of the acceptability curve.
range is usually h0 to h50,000 or h100,000). The diagram (Figure 5.8)

answers the following questions: If the λ is equal to A, then how many
points are cost-effective out the total (percentage)? If the λ equals B, then
how many points are cost-effective?
Cost-effectiveness acceptability curve

1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 10,000 20,000 30,000 40,000

λ or wtp
Figure 5.9 Representation of the acceptability curve.
The curve summarizing this information is called the cost-effectiveness

acceptability curve (CEAC) and is the output of the probabilistic approach.
The scale on the vertical axis is 0100%, and the horizontal axis represents
various values of λ. Based on this curve, we are able to inform budget
managers regarding the percentage of probabilistic analysis experiments in
which the new treatment is cost-effective compared with the standard as
soon as they inform us of their own λ. A hypothetic example of such a
curve is shown in Figure 5.9.
It should be noted that plotting a CEAC does not always provide infor-
mation regarding which treatment is the optimal option and cost-effective
compared with an alternative. Under certain conditions (e.g., when com-
paring more than two options, when the cost and benefit of the treatments
follow a specific correlation pattern, when the cost is highly asymmetrical,
etc.), a percentage higher than 50% in the CEAC diagram might not indi-
cate a cost-effective option. The curve showing the probability that the
optimal option (the one with the greatest NMB) is cost-effective for a specific λ
is called cost-effectiveness acceptability frontier, and it is what we are most
interested in. In this case, the correct way to find the optimal/cost-effective
option is to calculate the mean NMB for various values of λ (“find the
option with the highest mean NMB for various λ based on the PSA and
then find the probability that this is cost-effective”). In this case, there will
be areas of λ where one treatment is considered cost-effective and other
areas where another treatment is considered cost-effective. Usually, how-
ever, when comparing two options in practical research, the CEAC is the
correct basis for judging different options.
• Distribution selection in PSA: The success of PSA essentially relies on

the correct selection of distributions for each case. Considering that
there are standard computer programs with integrated distributions,
the selection of a distribution by the investigators may be somewhat
arbitrary in some cases. However, if a correct approach is used, consis-
tent with the statistical properties of the data distribution, then the
number of possible selections is significantly limited. For example, if
we wish to perform a PSA on a probability that is bounded by 0 on
the left and by 1 on the right, then we would normally select a
suitable statistical distribution like the beta (Claxton et al., 2005). If
this probability is derived from the survival analysis coefficients on a
log-hazard scale, then selection of the multivariable normal distribu-
tion is preferable (Claxton et al., 2005). For right-skewed costs, the
gamma or log-scale distributions are commonly used. In cases of data
correlation (e.g., differences in cost and benefit), a normal bivariate
analysis is usually followed, whereas recent data correlation is also
patterned through copula forms (Daggy et al., 2011). We should note
that the development of a probabilistic model is indeed a challenge for
economists; nevertheless, such an approach is considered more reliable
and less arbitrary than simple sensitivity analysis.
THE VALUE OF INFORMATION AND BAYESIAN ANALYSIS

According to the modern approach to economic evaluation, an analysis
would ideally use as much of the available information as possible when
constructing a model. In addition to study data (e.g., from a new clinical
trial), such an approach will also incorporate data from other sources, includ-
ing the subjective opinion of experts in a scientific field. According to the
Bayesian analysis approach, the study we will conduct is simply “one more
step on the road to knowledge.” This suggests a synthesis of pre-existing
knowledge with new knowledge obtained through our own particular study
to obtain new, “all-inclusive” knowledge. Such an analysis would first deter-
mine our prior perceptions (in the form of “prior” statistical distributions)
and combine them with the new data available to obtain the final results: our
end perceptions (through posterior distributions). In the literature and in
applied research, Bayesian analysis is a new field of study, but the acceptabil-
ity curve is interpreted more naturally in such a field of analysis.
Let us assume that, ideally, we have created an economic model
comparing the cost and effectiveness of two options using some kind of
What should we do with the 40% of patients which are expected to

receive a noncost-effective treatment (for λ=12,000)?
1.0
0.9
Probability of being cost-effective
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 10,000 20,000 30,000 40,000

Value of threshold ratio (€)
Figure 5.10 The EVPI concept.
Bayesian analysis that incorporates existing societal information with regard

to the two therapeutic interventions. Let us also assume that we have per-
formed a PSA and we know that society is willing to purchase an additional
year of life at h12,000 (λ 5 12,000). As Figure 5.10 shows, the probability
that the new option is cost-effective compared with the standard treatment
is 60% for λ 5 h12,000. Although the new treatment should be adopted,
we expect that 40% of patients will receive a noncost-effective treatment.
In this case, we might be willing—as a society—to allocate some resources
for scientific research to convert the 60% estimate to a more “certain” esti-
mate (i.e., closer to 100% or 0%). If the result of further research, based on
new data, pushes this percentage closer to 100% (e.g., up to 90%
[for λ 5 12,000]), then the percentage of patients expected to receive a
noncost-effective treatment would be reduced to 10%. If, instead, this
probability tended to 0% (e.g., 5%), then the standard treatment should be
adopted with probability 95% and the percentage of patients expected to
receive a noncost-effective treatment would be reduced to 5%. This means
that, in those areas where the probability is close to 50% for a specific λ, it
is particularly worthwhile to expend resources to reduce this uncertainty.
The value of the resources we are willing to expend to obtain the
additional information can be determined objectively and calculated using
an analysis called expected value of perfect information (EVPI).
We should not confuse our willingness to reduce this uncertainty with

the λ. The λ does not add anything to scientific knowledge. It simply
indicates how much we are willing to give to “purchase” 1 year of life,
and it is taken as fixed here.
At a technical level, what we do is “run” a PSA, convert the difference
in benefit between treatments into h based on the λ, and then compare the
benefit and cost from each experiment of the 1,000 or 5,000 experiments
we usually run in a Monte Carlo simulation (see also previous chapter). If
the benefit is negative, then we estimate the total negative benefit and divide
it by the number of experiments. One example of an EVPI calculation is
given in Table 5.6. Let us say that we are comparing treatments TH1 and
TH2 and we know that λ is h10,000/QALY. The PSA included only 20
experiments to simplify the calculations, but the analysis would be the same
even if we did 1,000 repetitions. So, assume we ran the 20 repetitions and
we obtained the first four columns of the table. The next step is to determine
the ΔE (the difference in effectiveness) for each experiment and multiply it
by 10,000. Next, we subtract the ΔC (difference in cost) from the ΔE for
each experiment and obtain the net monetary benefit (NMB). Note that,
out of the 20 experiments, 7 (35%) show that option B, and not option A, is
the cost-effective option. We know this now that we have full knowledge of
the experiments. How much money would we be willing to pay to avoid
this “failure” in 35%? If we add all those who had negative benefit and divide
by 20, then we calculate the value of full information. In this example, this
value is h639 per patient. By calculating the disease prevalence, the time
needed to develop the technology, the discount rates, and other factors, we
can determine the total resources we would be willing to invest in research.
We should note that, although the concept of the EVPI is extremely
interesting, we unfortunately have to go one step further. Based on the
analysis, we found that we would invest up to h639 per patient, BUT the
analysis does not say which parameters we should measure to reduce their
uncertainty. In this case, we should determine which variables we should
give priority to (e.g., we should conduct research to reduce the uncer-
tainty with regard to days of hospitalization for the condition, survival,
rate of side effects, etc.). Such an analysis is called partial EVPI and is now
the leading edge in modern economic evaluation. Mathematic models
used in this area are usually complex, and attempts are being made to
develop algorithms that will provide reliable results in a reasonable time
frame (such an analysis, in its full form, may need several days, even with
a fast computer).
Table 5.6 EVPI calculation
Cost (h) Cost (h) Effect Effect INB
TH1 TH2 TH1 TH12 (for λ 5 10,000/ In this particular experiment, which
year) treatment should I choose based on
the perfect knowledge I have now?
Experiment 1 31,898 20,145 4.19 3.1 2802 TH2

Experiment 2 29,570 18,742 3.95 3.19 23,274 TH2
Experiment 3 30,582 20,028 4.04 2.86 1,294 TH1
Experiment 4 30,915 19,509 4.1 3.04 2793 TH2
Experiment 5 29,416 20,397 3.92 2.98 360 TH1
Experiment 6 28,669 18,784 4.11 2.96 1,602 TH1
Experiment 7 28,827 20,497 3.82 2.92 704 TH1
Experiment 8 28,634 20,157 4.07 3.02 1,994 TH1
Experiment 9 29,575 19,543 3.9 3.06 21,624 TH2
Experiment 10 29,709 20,756 3.92 3.15 21,262 TH2
Experiment 11 29,590 19,013 4.51 2.4 10,518 TH1
Experiment 12 29,410 19,621 4.27 2.49 8,091 TH1
Experiment 13 31,507 20,591 4.46 2.88 4,925 TH1
Experiment 14 28,548 19,690 4.09 2.69 5,106 TH1
Experiment 15 28,131 19,771 3.79 2.89 569 TH1
Experiment 16 30,443 21,095 3.94 3.2 21,968 TH2
Experiment 17 31,165 21,164 3.8 3.1 23,059 TH2
Experiment 18 30,380 20,126 4.24 2.66 5,510 TH1
Experiment 19 31,124 21,198 4.04 2.86 1,861 TH1
Experiment 20 28,985 20,151 3.73 2.74 1,035 TH1
Average 29,854 20,049 4.04 2.91
EVPI 5 h639
CONCLUSIONS
This chapter, like the previous one, has presented the theory and metho-
dology behind high-quality economic analysis in health care. In the sub-
sequent chapters, we illustrate the application of economic analysis to
genomic medicine interventions that are emerging into clinical practice,
using some key examples from the published literature. Furthermore, we
describe the special provisions required to accurately model the evaluation
of the analysis costs and resource use, and the analysis outcomes and effec-
tiveness in genomic medicine applications.
REFERENCES
Arrow, K.J., 1963. Social Choice and Individual Values, second ed. Wiley, New York, NY.
Barton, G.R., Briggs, A.H., Fenwick, E.A., 2008. Optimal cost-effectiveness decisions:
the role of the cost-effectiveness acceptability curve (CEAC), the cost-effectiveness
acceptability frontier (CEAF), and the expected value of perfection information
(EVPI). Value Health 11 (5), 886897.
Briggs, A.H., Fenn, P., 1998. Confidence intervals or surfaces? uncertainty on the cost-
effectiveness plane. Health Econ. 7, 723740.
Casali, P.G., 2007. The off-label use of drugs in oncology: a position paper by the
European Society for Medical Oncology (ESMO). Ann. Oncol. 18 (12), 19231925.
Choi, I.Y., Kim, T.M., Kim, M.S., Mun, S.K., Chung, Y.J., 2013. Perspectives on clinical
informatics: integrating large-scale clinical, genomic, and health information for clini-
cal care. Genomics Inform. 11 (4), 186190.
Cialkowska, M., Adamowski, T., Piotrowski, P., Kiejna, A., 2008. What is the Delphi
method? Strengths and shortcomings. Psychiatr. Pol. 42 (1), 515.
Claxton, K., Sculpher, M., McCabe, C., et al., 2005. Probabilistic sensitivity analysis for
NICE technology assessment: not an optional extra. Health Econ. 14 (4), 339347.
Cohen, B.J., 2003. Discounting in costutility analysis of healthcare interventions:
reassessing current practice. Pharmacoeconomics 21 (2), 7587.
Daggy, J.K., Thomas III, J., Craig, B.A., 2011. Modeling correlated healthcare costs.
Expert Rev. Pharmacoecon. Outcomes Res. 11 (1), 101111.
Devlin, N., Parkin, D., 2004. Does NICE have a cost-effectiveness threshold and what
other factors influence its decisions? a binary choice analysis. Health Econ. 13 (5),
437452.
Drummond, M.F., Torrance, G.W., 2005. Methods for the Economic Evaluation of
Health Care Programmes. Oxford University Press, Oxford.
Eichler, H.G., Kong, S.X., Gerth, W.C., Mavros, P., Jonsson, B., 2004. Use of cost-
effectiveness analysis in health-care resource allocation decision-making: how are cost-
effectiveness thresholds expected to emerge? Value Health 7 (5), 518528.
Fenwick, E., O’Brien, B.J., Briggs, A.H., 2004. Cost-effectiveness acceptability curves—
facts, fallacies and frequently asked questions. Health Econ. 13, 405415.
Friedman, M., 1953. Essays in Positive Economics. University of Chicago Press, Chicago, IL.
Greenberg, D., Earle, C., Fang, C.H., Eldar-Lissai, A., Neumann, P.J., 2010. When is
cancer care cost-effective? a systematic overview of cost-utility analyses in oncology.
J. Natl. Cancer Inst. 102 (2), 8288.
Greenfield, S., Steinberg, E., 2011. Clinical Practice Guidelines We Can Trust. Institute
of Medicine, National Academies Press, Washington, DC.
Hillman, A.L., Kim, M.S., 1995. Economic decision making in healthcare. A standard
approach to discounting health outcomes. Pharmacoeconomics 7 (3), 198205.
Jain, R., Grabner, M., Onukwugha, E., 2011. Sensitivity analysis in cost-effectiveness
studies: from guidelines to practice. Pharmacoeconomics 29 (4), 297314.
Katz, D.A., Welch, H.G., 1993. Discounting in cost-effectiveness analysis of healthcare
programmes. Pharmacoeconomics 3 (4), 276285.
Meineke, F.A., Stäubert, S., Löbe, M., Winter, A.A., 2014. Comprehensive clinical
research database based on CDISC ODM and i2b2. Stud. Health Technol. Inform.
205, 11151119.
Meltzer, D., 2001. Addressing uncertainty in medical cost-effectiveness analysis implications
of expected utility maximization for methods to perform sensitivity analysis and the use
of cost-effectiveness analysis to set priorities for medical research. J. Health Econ. 20 (1),
109129.
Muennig, P., 2008. Cost Effectiveness Analysis in Health Care, second ed. Jossey-Bass,
San Francisco, CA.
O’Hagan, A., Stevens, J.W., Montmartin, J., 2000. Inference for the cost-effectiveness
acceptability curve and cost-effectiveness ratio. Pharmacoeconomics 17 (4), 339349.
Petiti, D., 1994. Meta-Analysis, Desicion Analysis, and Cost-Effectiveness Analysis.
Oxford University Press, New York, NY.
Phillips, C.G., 2005. Health Economics. An introduction for Health Professionals.
Blackwell Publishing Ltd, Massachusetts.
Prokosch, H.U., Ganslandt, T., 2009. Perspectives for medical informatics. Reusing the
electronic medical record for clinical research. Methods Inf. Med. 48 (1), 3844.
Shane, F., Loewenstein, G., O’Donoghue, T., 2002. Time discounting and time preference:
a critical review. J. Econ. Lit. 40 (2), 351401.
Smith, D.H., Gravelle, H., 2001. The practice of discounting in economic evaluations of
healthcare interventions. Int. J. Technol. Assess. Health Care 17 (2), 236243.
Towse, A., 2009. Should NICE’s threshold range for cost per QALY be raised? Yes. BMJ
338 (b181).
Weinstein, M., Zeckhauser, R., 1973. Critical ratios and efficient allocation. J. Public Econ.
2, 147157.
Yoder, J.L., 2008. The importance of cost inputs and sensitivity analyses in a cost-
effectiveness analysis. Clin. Drug Investig. 28 (7), 461462.
CHAPTER 6
Economic Evaluation in the

Genomic Era: Some Examples
from the Field
INTRODUCTION
The preceding chapters have explored the theory and methods behind
high-quality economic analysis in health care. Although some examples
from genomic medicine have been used to illustrate theoretical aspects of
the methods, this chapter uses examples from the published literature to
examine the application of economic analysis to genomic medicine inter-
ventions that are emerging into clinical practice. As has been noted previ-
ously in the book, these are early days for the field, so the number of
published studies is still relatively small. Nonetheless, the studies that have
been published represent a diverse approach to the application of eco-
nomic analysis to genomic medicine instances that serve the purposes of
this chapter well.
This chapter is organized to illustrate some of the opportunities and
challenges introduced in the prior chapters. The examples emphasize
examination of the critical aspects of the analysis that can impact the
validity of the analysis. The chapter finishes with an exploration of the
opportunity economic analysis presents in emerging countries looking to
implement genomic medicine. The reader should bear in mind that this
chapter is not an exhaustive review of the few existing studies of eco-
nomic evaluation in genomic medicine, but rather a collection of the
most representative studies to demonstrate the importance of economic
evaluation in genomic-guided therapies.
USING PHARMACOGENOMICS TO PREVENT

ADVERSE EVENTS
Adverse drug events (ADE) are a major contributor to morbidity, mortal-
ity, and costs of care (Classen et al., 1997, 2010). Considerable effort has

been expended to identify preventable causes of ADE and to create sys-

tems that reduce risks due to poor reliability of delivery systems (such as
computerized order entry). It has long been known that patients respond
differently to medications as a result of environmental and individual fac-
tors, including genomic variation.
One of the most well-known examples involves the drug abacavir.
Abacavir is a synthetic carbocyclic nucleoside analog with inhibitory
activity against human immunodeficiency virus (HIV-1). In combination
with other antiretroviral agents, it is indicated for the treatment of HIV-1
infection. Serious and sometimes fatal hypersensitivity reactions have
been associated with abacavir. Studies of patients who experienced an
abacavir-associated ADE identified an association between the ADE and a
specific genetic variant in the HLA complex, HLA-B 57:01. Patients
who carry the HLA-B 57:01 allele are at high risk for experiencing a
hypersensitivity reaction to abacavir (Hughes et al., 2008; Mallal et al.,
2008; Saag et al., 2008). Approximately 0.5% of patients who are HLA-
B 57:01-negative will develop hypersensitivity, whereas more than 70%
who are HLA-B 57:01-positive will develop hypersensitivity. The FDA
issued an alert in July 2008 about this and information was added to the
box warning (FDA, 2014).
This is a very straightforward case for the application of economic
analysis. The first was performed in 2004 by Hughes et al. (2004). The
patient level data for abacavir ADE were obtained from a large HIV clinic
and the analysis included costs, such as cost of genetic testing, cost of
treating abacavir hypersensitivity, and the cost and selection of alternative
antiretroviral regimens. The investigators used a probabilistic decision ana-
lytic model that compared testing with no testing and tested the model
using Monte Carlo simulations. They concluded that based on the choice
of comparators, the testing strategy ranged from dominant (less expensive
and more beneficial compared with no testing) to an incremental
cost-effectiveness ratio (ICER) of h22,811. The study was limited by
very small numbers of events, resulting in a very wide range of odds ratios
(7.9 with 95% confidence intervals 1.5 41.4). The study was done from
the health system perspective because it did not include data to allow for
a societal perspective. Several subsequent analyses have been performed,
all of which have determined testing prior to the use of abacavir as being
cost-effective and potentially cost-saving under some assumptions.
A study by Schackman et al. (2008) used a simulated model of
HIV disease based on the prospective randomized evaluation of DNA
Economic Evaluation in the Genomic Era: Some Examples from the Field 99
screening in a clinical trial study. The study modeled three different

approaches: (i) triple therapy including abacavir; (ii) genetic testing prior
to triple therapy with tenofovir substituted for abacavir for patients who
carry the HLA B 57:01 allele; and (iii) triple therapy with tenofovir
substituted for abacavir for all patients. Abacavir and tenofovir were
assumed to have equal efficacy, and the cost of the tenofovir treatment
was $4 more than the abacavir treatment. Outcomes were quality-
adjusted life-years (QALY) and lifetime medical costs. The authors con-
cluded that the genetic testing strategy was preferred and resulted in a
cost-effectiveness ratio of $36,700/QALY compared with no testing
(the tenofovir strategy was found to increase cost with no improvement
in outcomes and thus was dominated). The authors stressed that the
model was robust if the assumption that abacavir and tenofovir had equiv-
alent efficacy and abacavir therapy were less expensive. This is a very
important conclusion because it demonstrates that the result of an analysis
is sensitive to changing conditions in the health care system and thus may
not remain “true” in the face of these changing conditions.
Kauf et al. (2010) studied this from the US health care system perspec-
tive in a very well-defined patient cohort. They used two approaches in
their evaluation: (i) a short-term model looking at the efficiency of
genetic testing compared with no testing prior to initiation of abacavir
and (ii) a lifetime discrete event simulator that compared testing prior to
abacavir use versus initiation with a tenofovir-containing regimen. The
authors carefully described the sources of the data needed for their analy-
ses and identified gaps in data when expert opinion was used. They used
both sensitivity and scenario analyses to assess the impact of parameter
uncertainty. For the short-term analysis, the outcome was cost per
patient. For the long-term analysis, outcomes were presented as QALY.
The results of the short-term analysis showed that while the cost per
patient increased slightly by $17 for the tested group, the model predicted
that testing would result in the avoidance of 537 ADE per 10,000
patients. In the lifetime model, the testing prior to abacavir use domi-
nated the tenofovir-containing regimen (based on improved effectiveness
and decreased cost) in the base case and in sensitivity analyses.
The latter two studies score well on the Quality of Health Economic
Studies (QHES) (Ofman et al., 2003) instrument. One can see the
challenge that the results could pose to a reader trying to answer the
question of whether genetic testing prior to abacavir is cost-effective. It
requires the reader to understand the scenarios used for the analysis and
to determine the relevance of the results to their specific situation. The

fact that all of the studies concluded that testing was cost-effective across a
variety of scenarios and assumptions provides some reassurance, although
in reality given the FDA recommendation, cost-effectiveness takes the
back seat to considerations of patient safety. However, the modeling pro-
vides some information about the relative efficiencies of various strategies
that could be useful to systems looking to implement the FDA recom-
mendations. This is explored later in the chapter.
BALANCING ADVERSE EVENTS AND EFFICACY

The preceding case was unusual in that the genomic marker was associ-
ated solely with a risk for an ADE. In contrast, most medications require
balancing the benefits against the potential risks associated with the ther-
apy. The nineteenth century English physician Peter Latham recognized
this balance with his pithy quote, “Poisons and medicine are oftentimes
the same substance given with different intents.” There are a number of
factors that can alter an individual’s risk/benefit ratio, including germline
genomic variants. In this section, we examine some examples of pharma-
coeconomic analyses that explore the economic implications of the
balance between efficacy and harm.
Thiopurine medications (including 6-mercatopurine [6-MP], 6-
thioguanine [6-TG], and azathioprine [AZA]) interfere with purine
metabolism and are used for a variety of indications, including acute lym-
phoblastic leukemia (ALL) and inflammatory and autoimmune diseases,
and as immunosuppressants in organ transplant recipients. These medica-
tions are metabolized by the enzyme thiopurine methyltranferase
(TPMT) encoded by the gene TPMT. Variants in the TPMT gene that
alter the composition of the protein can result in altered enzymatic activ-
ity (Relling et al., 2011, 2013). Individuals with reduced or absent
TPMT activity are at increased risk for clinically significant myelosuppres-
sion compared with those with normal enzyme activity (a harm). This
has led to guidelines for reducing the dose and enhancing monitoring for
signs of myelosuppression in individuals known to have reduced TPMT
activity or who have TPMT gene variants known to lead to reduced
enzymatic activity (Relling et al., 2011, 2013). Myelosuppression is gen-
erally considered an adverse event; however, in the case of ALL, it is the
goal of the treatment, in so far as myelotoxicity is needed, to eliminate
the malignant lymphoblasts and induce remission. The endpoint of the
induction therapy in ALL is elimination of at least 99.9% of blasts from

the blood and peripheral bone marrow, which necessitates a significant
degree of generalized myelosuppression. In this context, myelosuppression
can be looked at as both a beneficial and potentially harmful outcome.
Stanulla et al. (2005) examined an important outcome of ALL therapy,
minimal residual disease (MRD) in a population of pediatric patients trea-
ted with standard therapies that included 6-MP who had undergone
TPMT genotyping. MRD is the strongest predictor of relapse of ALL, so
it is a very important intermediate outcome of treatment. The genotype
information collected was not used to adjust the dose of 6-MP. Analysis
of the study data showed that patients with a genotype that predicted
reduced TPMT activity had an MRD rate of 9.1% compared with 22.8%
in the group predicted to have normal TPMT activity (a statistically
significant result). This study emphasized the importance of considering
outcomes that reflect both efficacy and harm. This caveat is illustrated in
the following examples.
One of the earliest published cost-effectiveness studies of TPMT geno-
typing prior to use of 6-MP compared with standard of care was performed
by the Institute for Prospective Technological Studies (van den Akker-van
Marle et al., 2006). The analysis assumed a genotyping cost of h150 and
yielded a mean cost per life-year gained of h2100, which increased to
h4300 when the usual 3% discount rate was applied. The authors concluded
that TPMT genotyping was cost-effective compared with standard care and
expected the cost-effectiveness to improve as the genotyping cost was
expected to decrease. They recommended that TPMT genotyping should
be seriously considered for implementation. Let us assume we are the
decision-maker who has received this information. What more do we want
to understand about the analysis before we act on the recommendation?
Most of the book to this point has focused on the importance of the
methods of the analysis, and a careful examination of the methods of this
chapter is instructive. One of the first statements in the methods section
of this chapter is, “. . .not much information on the parameters for the
TPMT model was specifically available for children with ALL.” This is a
problem given that the analysis is specifically about the use of TPMT gen-
otyping in children with ALL. The authors used data from other pharma-
coeconomic studies of thiopurines (many in adults with inflammatory
disease) and expert interviews. The authors explicitly stated the analysis
was performed from the societal perspective but only included direct costs
(so costs related to time lost from work, e.g., were not included).
Appropriate price indexing was used and a discount rate for future costs
was applied. The authors appropriately considered differences in allele fre-
quency based on ethnicity and included realistic ethnic distributions in
the model. The authors made clear that the only complication of treat-
ment considered was the most severe adverse event, myelosuppression.
The estimates of the frequency were derived primarily from adult studies
in which 6-MP was used for inflammatory bowel disease or other inflam-
matory conditions. Applying adult experience to pediatric patients is a
problem. In addition, the indication for treatment can make a huge dif-
ference in the balance of risk and benefit, as is illustrated in the following
section of this chapter. A key parameter for the model is the cost of an
episode of myelosuppression. The authors identified a number of sources
for costs associated with inpatient and outpatient treatments, developing
an average cost of the treatment from the perspective of three different
national health systems and then making a determination of the average
cost to be used for purposes of the model. The sources were cited and
the calculations used were provided. Another challenging parameter to
determine is how many of the adverse events can be attributed to the
presence of a TPMT variant. Although presence of decreased TPMT
activity is known to be associated with increased risk of myelosuppression,
the nature of the treatment means that even those with normal activity
are at risk for an adverse event. In fact, given that the majority of patients
have normal TPMT activity, the absolute number of adverse events in
this population is higher than in the population with decreased activity.
The authors concluded that, for the purposes of the model, they would
attribute 32% of all cases of myelosuppression to a TPMT variant. Stated
another way, for every ADE occurring in a patient with decreased
TPMT activity, two would occur in patients with normal activity. It is
important to state again that the estimates of ADE risk were obtained in
studies of adults for disorders other than ALL. Finally, the authors noted that
severe ADEs have the potential to result in loss of life. As could be
expected in a study that focuses on children, the prevention of a child-
hood death leads to a large number of life-years gained. This must be
tempered by the fact that ALL can also lead to premature death, so realis-
tically estimating the average life expectancy of all treated patients com-
pared with peers without ALL is very important. We return to this point
because it is critical in the balance of risk and efficacy. Given the uncer-
tainty in many of the model parameters, the authors used sensitivity
analyses to test the impact of the assumptions made.
In discussing the results, the authors note that TPMT genotyping

can reduce health care costs through the avoidance of myelosuppression
episodes compared with no genotyping. However, these costs saved,
“. . .are a minor consideration compared with the potential for saving
lives.” Reinforcing this point, the authors noted that a series of univariate
sensitivity analyses for the model’s parameters had a modest effect on the
cost-effectiveness estimates, with the notable exception of mortality pre-
vented. Changing this parameter 10-fold (from prevention of death in 1
in 1,000 treated patients to prevention of death in 1 in 10,000 treated
patients) changed the estimates of cost-effectiveness from h2100 per
life-year gained to h47,600 per life-year gained. The authors also per-
formed a multivariate sensitivity analysis that gave the boundaries of cost-
effectiveness ratios (per life-year gained without applying discounting)
between a cost-saving strategy and h53,000. The authors note that even
the extreme cases are at or only modestly above the cost-effectiveness
threshold usually used (US$50,000 per life-year gained).
This is a well-done high-quality economic analysis, so the conclusion
that TPMT genotyping should be routinely implemented is reasonable.
However, a key assumption is made that is not critically considered in the
analysis: the impact of a potential reduction in efficacy. The authors
assumed that the alterations in management would reduce ADEs with no
impact on the treatment response for the primary disease, ALL. As noted
previously (Stanulla, 2005), this assumption may not be correct. If we
assume that the rate of MRD is approximately twice that for patients
with a wild-type TPMT genotype compared with a variant genotype,
then a much larger number of patients would experience relapse based on
their MRD, thus incurring increased medical costs as a result of more
intensive therapies, including bone marrow transplantation. More impor-
tantly, this could result in more children who succumb to their underly-
ing disease. The authors did not provide the ALL mortality rate used for
the model because it was assumed to be the same in both groups. If we
use a long-term event-free survival rate of 90% for children approximately
8 years of age (as in the study), that would mean that out of a cohort of
100,000 treated ALL patients we would expect 10,000 to die from their
disease (for the purposes of this exercise, we assume that none of these
deaths are due to ADE). Using the study estimate of prevention of 1 death
due to an ADE in 1,000 treated patients, we expect that the genotyped
cohort would experience 9,900 deaths (10,000 deaths—100 deaths pre-
vented by genotyping). However, if we assume that efficacy of treatment
is decreased by 10%, then the mortality rate would increase from 10% to
11%. This means that the number of deaths in the genotyped group
would now be 10,900 (11,000 deaths—100 deaths prevented by genotyp-
ing), 900 more deaths than in the nongenotyped group. The impact on
efficacy would have to be reduced to a mere 1% difference for the num-
ber of deaths to be equal in the two groups, a number that seems unreal-
istically low given the potential impact on MRD. Clearly, the truth is
“still out there;” however, this should illustrate the importance of consid-
ering both efficacy and harm for the purposes of economic analyses.
THE DISEASE MAKES A DIFFERENCE

Staying with the example of thiopurines and TPMT, this section examines
the impact that a different treatment indication has on the economic anal-
ysis. Thiopurines are a mainstay of treatment for inflammatory diseases,
with AZA being the preferred member of the class. We look at an analysis
performed in the context of Crohn’s disease (Dubinsky et al., 2005) and
compare that with the ALL example.
This study examines both efficacy and prevention of adverse events. In
contrast to ALL, myelosuppression is not an expected outcome of treat-
ment but it represents a pure adverse event with no associated efficacy.
However, if the focus is only on prevention of the adverse event, then
conservative dosing could decrease treatment efficacy. Therefore, there
still exists a balance between efficacy and adverse events, but the tolerance
for myelosuppression in inflammatory disease is lower than in ALL, shift-
ing the balance toward prevention of the adverse event. TPMT genotyp-
ing is used to identify individuals who have an increased susceptibility to
myelosuppression, whereas monitoring of white blood cell counts and
disease activity scales were used to assess the efficacy of treatment. Three
different approaches were compared with standard of care: (i) TPMT gen-
otyping alone; (ii) intensive monitoring; and (iii) TPMT genotyping with
intensive monitoring. Values for the decision model were obtained from
standard resources (for costs), a literature review, and expert opinion
when data were not available. The perspective was that of a third-party
payer. One- and two-way sensitivity analyses were used to test the model.
All three of the alternative treatments provided higher efficacy at a lower
cost than the standard of care.
The TPMT genotyping-only arm provided the lowest cost of the
three alternative treatments, a result that persisted in the sensitivity analyses.
The model did not identify a significant difference related to adverse events,
which the authors attribute to a low event rate. Interestingly, the efficacy
was similar for all three of the alternative therapies, which was unexpected
given that the monitoring and TMPT genotyping plus monitoring arms
allowed response-based dosing of AZA, which would be expected to result
in higher efficacy. It was not clear from the article if the assumptions for
efficacy were based on published data (ideally from prospective clinical
trials) or relied on expert opinion. These two results raise some questions
about the assumptions used in the model, an issue that the authors
acknowledge as a weakness given the reliance on expert opinion as opposed
to clinical trial data. The lack of robust clinical trial evidence represents an
ongoing challenge for economic analysis, as evidenced by the fact that mul-
tiple analyses over nearly 15 years still have not yielded a definitive answer
regarding the cost-effectiveness of TPMT genotyping prior to the use of
thiopurines for any indication.
PERSPECTIVES ON PERSPECTIVE
Economic analyses must specify from what perspective they are being
performed. The perspective used is important if the result is to be of use
to the intended stakeholder. To illustrate this, we use the example of
tumor-based screening for Lynch syndrome.
Colorectal cancer (CRC) is one of the most common cancers world-
wide. Approximately 2 5% of CRC is due to germline mutations in one
of several DNA mismatch repair genes, a condition called Lynch syn-
drome. Individuals who carry such a mutation have a lifetime risk of devel-
oping CRC that approaches 80%, and they are at higher risk for several
other cancers. Identification of a mutation carrier allows initiation of ear-
lier and more frequent colonoscopy and consideration of prophylactic sur-
gery, which has been shown to reduce the risk of developing certain types
of cancer associated with Lynch syndrome. Identification of an index case
allows identification of at-risk family members who have not yet devel-
oped cancer. The most effective way to identify Lynch syndrome in
a patient presenting with CRC is to screen the tumor for changes
that are indicative of a mutation (e.g., absence of staining for one of
the mismatch repair proteins or DNA microsatellite instability). Tumor-
based screening was recommended based on a systematic evidence review
and synthesis by the Evaluation of Genomic Applications in Practice and
Prevention Working Group (EGAPP, 2009). Soon thereafter, economic
analyses (Mvundura et al., 2010) were performed that demonstrated that

universal tumor-based screening for Lynch syndrome was cost-effective
from a societal perspective if at-risk family members were identified and
screened.
Based on this evidence, the chapter author’s institution elected to
implement tumor-based screening. However, questions arose about the
most cost-efficient way to screen—a question that the published analyses
did not address. We constructed decision analytic models and used a com-
bination of budget impact and cost-consequences (or cost-comparison)
analyses performed from the perspective of our local health care delivery
system (Gudgeon et al., 2011). The primary outcomes of these analytic
models included the absolute and relative effectiveness and costs of the
different screening protocols by determining intermediate endpoints such
as total costs of testing, expected number of cases detected, and cost per
case detected. Secondary outcomes, to be defined by stakeholders as
we proceeded through decision-making were expected to include various
budget impact metrics.
Test performance values were taken primarily from the published liter-
ature, although additional data from the referral laboratory (cost and test
dependence), our institution’s electronic data warehouse (local patient
characteristics), and gray data made available by a large research project
were used as needed. To estimate specific outcomes pertinent to local
decision-makers, we began the analyses open to all testing protocols we
believed to be reasonable. Given the rapidly changing nature of the test-
ing protocol, the team monitored new developments and, when appro-
priate, added new testing protocols to the model. This proved to be a
critical decision because the most efficient protocol incorporated an
emerging test. This emphasizes the importance of not “locking down”
models in a field that changes as rapidly as genomics.
The screening approach with the maximum sensitivity is to sequence
the four mismatch repairs in all patients with CRC who have a positive
screening test of the tumor. Although this will detect the maximum
number of patients with Lynch syndrome, the costs are prohibitive
(poor efficiency). The preferred screening approach had only a slight
decrease in sensitivity compared with sequencing, but the cost per case
detected was dramatically different (sequencing, US$13,355; immunohis-
tochemistry with reflex testing, US$10,412).
Local stakeholders were interested in additional metrics to help
them determine the expected impact on clinical resources and budget,
including the number of CRC patients treated annually within our deliv-
ery system and the number for which we expect to have tumor tissue
available for screening. Our models allowed us to estimate the percent of
total test costs that would fall under bundled hospital payments (approxi-
mately 75%) and percent of screened patients who would be eligible for
sequencing, thus requiring consent and genetic counseling (5 6%). After
implementation of screening, a question arose as to whether screening
should be restricted to patients aged 50 years or younger, given that
Lynch syndrome usually results in tumors appearing at a younger age.
Models were used to assess the sensitivity of screening and cost per case
detected using a variety of age cutoffs. Applying cutoffs reduced the cost
of the screening program with a concomitant decrease in sensitivity.
Presentation of the data allowed leadership to consider both costs and
clinical impact, and it was ultimately decided not to impose an age cutoff
for screening (Gudgeon et al., 2013). Finally, the health care system col-
lects data from the screening of all of the key parameters of the model,
allowing testing of the performance of the screening program, allowing
for continuous improvement.
DEFINING PERFORMANCE CHARACTERISTICS

New technologies are frequently introduced into practice with limited
evidence regarding utility. This not only leads to questions about whether
the technology is ready to be used but also renders cost-effectiveness anal-
ysis heavily reliant on assumptions. One way to address this issue is to cre-
ate a model that tries to answer the question, what is the magnitude of
improvement of the new technology compared with current care needed
to meet a defined threshold of cost-effectiveness? We recently performed
a threshold analysis in the context of chronic hepatitis C.
Chronic hepatitis C is a major public health issue. Failure to achieve
sustained viral response (SVR) with treatment can lead to end-stage liver
disease, necessitating expensive interventions such as liver transplantation.
Viral genotype is a strong predictor of treatment response, with hepatitis
C virus (HCV) genotype 1 showing relative resistance to treatment. It has
also been shown that variants in the patient’s IL28B gene can also predict
treatment response, with certain variants leading to a less favorable
response. Treatment in HCV genotype 1 has been dramatically improved
through the addition of protease inhibitors to standard of care, with
the rate of SVR in treatment-naı̈ve patients increasing from 40 50% to
70 80%. Multiple cost-effectiveness (CE) studies have been completed

for the new regimens. A systematic review found 11 CE studies of triple
therapy in treatment-naı̈ve and pretreated patients with genotype 1 overall
and using IL28B guidance for better treatment response, with all study
results (most from the payer/health system perspective) supporting its
cost-effectiveness (San Miguel et al., 2014).
Response rate in HCV genotypes 2 and 3 are much higher than for
genotype 1, with SVR being achieved in approximately 80% of patients
treated with standard therapy. As with HCV genotype 1, the IL28B status
of the patient can affect treatment response, although to a lesser degree.
Despite the high response rate, there is still a substantial minority of
patients who do not achieve SVR and are at risk for development of
end-stage liver disease. This represents a potential opportunity for the use
of protease inhibitors, but can the cost be justified? Could the IL28B
status be used to identify patients more likely to experience failure of
standard therapy? This was recently modeled (Bock et al., 2014).
This analysis used cohort simulations on decision tree models to esti-
mate SVR rates, treatment costs, and outcomes for multiple treatment
arms that were incorporated into a Markov model to predict triple
therapy treatment costs, short-term and long-term health care costs,
quality-adjusted life-expectancies, cost-effectiveness thresholds, and the
relative percent changes in observed SVR rates needed to reach the
cost-effectiveness thresholds. The results specify ranges of improvement
in SVR rates needed for the IL28B-guided triple therapy to be cost-
effective for each IL28B variant compared with the SOC therapy for
HCV genotype 2 and genotype 3 patients. Because the estimated ranges
of improvement in SVR rates for each IL28B variant are relatively small
in comparison with current response rates with standard therapy, they
support the study’s conclusion that IL28B-guided triple therapy for HCV
genotype 2 and genotype 3 patients is likely to be cost-effective when
long-term outcomes are considered.
COST-EFFECTIVENESS ANALYSIS IN GENOMIC MEDICINE

AND THE DEVELOPING WORLD
Although cost-effectiveness studies have mostly been done from the
perspective of developed countries, an argument can be made that the
analyses could be much more important to developing countries where
the consequence of expending scarce resources on technology that is not

cost-effective has a much higher opportunity cost (Mitropoulos et al.,
2011). Unfortunately, there is also a shortage of health economists and
modelers in these countries.
In a recent economic evaluation study involving elderly atrial fibrilla-
tion patients using warfarin treatment, it was shown that 97% of elderly
Croatian patients with atrial fibrillation belonging to the
pharmacogenomics-guided group did not have any major complications,
compared with 89% in the control group, and, most importantly, the
ICER of the pharmacogenomics-guided versus the control groups was cal-
culated to be just h31,225/QALY (Mitropoulou et al., 2015). These data
suggest that, contrary to other developed countries, pharmacogenomics-
guided warfarin treatment represents a cost-effective therapy option for the
management of elderly patients with atrial fibrillation in Croatia, which
may very well be the case for the same and other anticoagulation treatment
modalities in neighboring countries.
Apart from the differences in current drug prices and resource utiliza-
tion in different countries, another important parameter to determine the
cost-effectiveness of a certain medical intervention in different health care
systems is the variable frequencies of the pharmacogenomic biomarkers.
As such, one should bear in mind that a pharmacogenomics-guided
medical intervention that is not cost-effective in a certain country may be
cost-effective in another country, even if no significant cost differences
exist between these two countries because of the higher frequency of a
pharmacogenomic biomarker in the general population.
For example, the CYP2D6 7 allele frequency is significantly higher in
the Maltese population compared with the Caucasian average, whereas the
same is true for the CYP2C9 3 allele frequency in the Serbian population
(unpublished data). This suggests that there may be significant implications
in the cost-effectiveness of risperidone and warfarin treatments, respec-
tively, in these two countries. Another example is screening for the HLA-
B 15:02 allele, which increases the risk of developing Stevens-Johnson
syndrome (SJS) and toxic epidermal necrolysis (TEN) when treated with
the antiepileptic drugs (AEDs) carbamazepine (CBZ) and phenytoin.
A recent systematic review and meta-analysis including 16 studies found
considerable variation among different racial/ethnic populations in the
relationship between HLA-B 15:02 and CBZ-induced SJS and TEN, as
illustrated by a summary odds ratio of 79.84 (95% CI, 28.45 224.06),
with the following strong relationships warranting screening for

three racial/ethnic subgroups: (i) Han-Chinese, 115.32 (18.17 732.13);
(ii) Thai, 54.43 (16.28 181.96); and (iii) Malaysian, 221.00
(3.85 12,694.65). Among individuals of white or Japanese race/ethnicity,
no patients with SJS or TEN were carriers of the HLA-B 15:02 allele
(Tangamornsuksan et al., 2013). This variability can impact estimates of
cost-effectiveness even in a single country. An example is a cost-
effectiveness analysis of three ethnic groups in Singapore with different
allele frequencies of HLA-B 15:02. Genotyping for HLA-B 15:02 and
providing alternate AEDs to those who test positive were found to be
cost-effective for Singaporean Chinese and Malays, but not for
Singaporean Indians, based on ICERs of $37,030/QALY for Chinese
patients, $7,930/QALY for Malays, and $136,630/QALY for Indians
(Dong et al., 2012).
This suggests that pharmacogenomic economic evaluation studies
must be replicated in every country to inform policymakers prior to the
implementation of a pharmacogenomic-guided medical intervention to
evaluate its cost-effectiveness based on characteristics specific to each
country (Snyder et al., 2014). This is a daunting expectation, given the
rapidly increasing number of pharmacogenomics guidelines of developed
country regulatory agencies such as the US Food and Drug
Administration and multinational organizations such as the Clinical
Pharmacogenetics International Consortium, and given that many devel-
oping countries do not have the necessary resources and expertise to per-
form the analyses. This problem could be minimized by the
construction of generic economic models that would allow input of cer-
tain key variables such as allele frequency, test, and treatment costs that
vary by country. This model could be used by less experienced indivi-
duals who have access to the country-specific variables to generate a first
approximation of cost-effectiveness, allowing prioritization between dif-
ferent emerging tests. The decrease of genotyping costs for the once-in-
a-lifetime determination of an individual’s personalized pharmacoge-
nomics profile using next-generation sequencing technologies including
whole-genome sequencing (Mizzi et al., 2014) would gradually result in
a cost-effective pharmacogenomics-guided treatment for drugs bearing
pharmacogenomic testing recommendations on their labels, because the
single test could provide information that would inform prescribing
choices over the lifetime of the patient.
CONCLUSIONS
Genomic cost-effectiveness analysis has the potential to inform assess-
ments about the value of current and emerging technologies and priori-
tize value-based decisions about adoption and investment. Efforts to close
evidence gaps can strategically target areas of greatest need and potential
health and cost impacts. Economic evaluations from the perspective of
the relevant stakeholder can provide information and guidance for
decision-making and policy-making. Results from these evaluations can
include estimated ranges and threshold levels for key outcome variables to
achieve desirable real-world results. Economic analyses should adhere to
quality standards such as the QHES tool. Most importantly, given the rap-
idly changing nature of the evidence in genomics, economic models must
be developed that are flexible and adaptable. Ultimately, high-quality and
robust models must be developed that can be utilized by experienced sta-
keholders without high-level training in economics to encourage routine
use of these models to assist in decision-making.
REFERENCES
2014. Cost-effectiveness of IL28B genotype-guided protease inhibitor triple therapy
versus standard of care treatment in patients with hepatitis C genotypes 2 or 3 infection.
Public Health Genomics 17, 306 319.
Classen, D.C., Pestotnik, S.L., Evans, R.S., Lloyd, J.F., Burke, J.P., 1997. Adverse drug events
in hospitalized patients. Excess length of stay, extra costs, and attributable mortality.
JAMA 22 (277), 301 306.
Classen, D.C., Jaser, L., Budnitz, D.S., 2010. Adverse drug events among hospitalized
medicare patients: epidemiology and national estimates from a new approach to
surveillance. Joint Comm. J. Qual. Patient Saf. 36 (1), 12 21.
Dong, D., Sung, C., Finkelstein, E.A., 2012. Cost-effectiveness of HLA-B 1502 genotyp-
ing in adult patients with newly diagnosed epilepsy in Singapore. Neurology 79,
1259 1267.
Dubinsky, M.C., Reyes, E., Ofman, J., Chiou, C.F., Wade, S., Sandborn, W.J., 2005. A
cost-effectiveness analysis of alternative disease management strategies in patients with
Crohn’s disease treated with azathioprine or 6-mercaptopurine. Am. J. Gastroenterol.
100, 2239 2247.
Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working
Group, 2009. Recommendations from the EGAPP working group: genetic testing
strategies in newly diagnosed individuals with colorectal cancer aimed at reducing
morbidity and mortality from lynch syndrome in relatives. Genet. Med. 11, 35 41.
FDA Information for Healthcare Professionals: Abacavir (marketed as Ziagen)
and Abacavir-Containing Medications. ,http://www.fda.gov/Drugs/DrugSafety/
PostmarketDrugSafetyInformationforPatientsandProviders/ucm123927.htm . (accessed
31.07.14).
Gudgeon, J.M., Williams, J.L., Burt, R.W., Samowitz, W.S., Snow, G.L., Williams, M.S.,
2011. Lynch syndrome screening implementation: business analysis by a healthcare
system. Am. J. Manag. Care 17, e288 e300.
Gudgeon, J.M., Belnap, T.W., Williams, J.L., Williams, M.S., 2013. Impact of age cutoffs
on a Lynch syndrome screening program. J. Oncol. Pract. 9, 175 179.
Hughes, C.A., Foisy, M.M., Dewhurst, N., Higgins, N., Robinson, L., Kelly, D.V., et al.,
2008. Abacavir hypersensitivity reaction: an update. Ann. Pharmacother. 42, 387 396.
Hughes, D.A., Vilar, F.J., Ward, C.C., Alfirevic, A., Park, B.K., Pirmohamed, M., 2004.
Cost-effectiveness analysis of HLA B 5701 genotyping in preventing abacavir hyper-
sensitivity. Pharmacogenetics 14, 335 342.
Kauf, T.L., Farkouh, R.A., Earnshaw, S.R., Watson, M.E., Maroudas, P., Chambers, M.
G., 2010. Economic efficiency of genetic screening to inform the use of abacavir sul-
fate in the treatment of HIV. Pharmacoeconomics 28, 1025 1039.
Mallal, S., Phillips, E., Carosi, G., Molina, J.M., Workman, C., Tomazic, J., et al., 2008.
PREDICT-1 study team. HLA-B 5701 screening for hypersensitivity to abacavir. N.
Engl. J. Med. 358, 568 579.
Mitropoulos, K., Johnson, L., Vozikis, A., Patrinos, G.P., 2011. Relevance of pharmacoge-
nomics for developing countries in Europe. Drug Metabol. Drug Int. 26, 143 146.
Mitropoulou, C., Fragoulakis, V., Bozina, N., Vozikis, A., Supe, S., Bozina, T, et al.,
Economic evaluation for pharmacogenomic-guided warfarin treatment for elderly
croatian patients with atrial fibrillation. In Press 2015.
Mvundura, M., Grosse, S.D., Hampel, H., Palomaki, G.E., 2010. The costeffectiveness of
genetic testing strategies for lynch syndrome among newly diagnosed patients with
colorectal cancer. Genet. Med. 12, 93 104.
Ofman, J.J., Sullivan, S.D., Neumann, P.J., Chiou, C.F., Henning, J.M., Wade, S.W.,
et al., 2003. Examining the value and quality of health economic analyses: implica-
tions of utilizing the QHES. J. Manag. Care Pharm. 9, 53 61.
Relling, M.V., Gardner, E.E., Sandborn, W.J., Schmiegelow, K., Pui, C.H., Yee, S.W.,
et al., 2011. Clinical Pharmacogenetics Implementation Consortium. Clinical phar-
macogenetics implementation consortium guidelines for thiopurine methyltransferase
genotype and thiopurine dosing. Clin. Pharmacol. Ther. 89, 387 391.
Relling, M.V., Gardner, E.E., Sandborn, W.J., Schmiegelow, K., Pui, C.H., Yee, S.W.,
et al., 2013. Clinical Pharmacogenetics Implementation Consortium guidelines for
thiopurine methyltransferase genotype and thiopurine dosing. Clin. Pharmacol. Ther.
Advance online publication 20 February 2013. http://dx.doi.org/10.1038/clpt.2013.4.
Saag, M., Balu, R., Phillips, E., Brachman, P., Martorell, C., Burman, W., et al., 2008.
Study of hypersensitivity to abacavir and pharmacogenetic evaluation study team.
High sensitivity of human leukocyte antigen-B 5701 as a marker for immunologically
confirmed abacavir hypersensitivity in white and black patients. Clin. Infect. Dis. 46,
1111 1118.
San Miguel, R., Gimeno-Ballester, V., Mar, J., 2014. Cost-effectiveness of protease inhibi-
tor based regimens for chronic hepatitis C: a systematic review of published literature.
Expert Rev. Pharmacoecon Outcomes Res. 14, 387 402.
Schackman, B.R., Scott, C.A., Walensky, R.P., Losina, E., Freedberg, K.A., Sax, P.E.,
2008. The cost-effectiveness of HLA-B 5701 genetic screening to guide initial antire-
troviral therapy for HIV. AIDS 22, 2025 2033.
Snyder, S.R., Mitropoulou, C., Patrinos, G.P., Williams, M.S., 2014. Economic evaluation of
pharmacogenomics: a value-based approach to pragmatic decision-making in the face of
complexity. Public Health Genomics 17, 256 264.
Stanulla, M.(1), Schaeffeler, E., Flohr, T., Cario, G., Schrauder, A., Zimmermann, M.,
et al., 2005. Thiopurine methyltransferase (TPMT) genotype and early treatment
response to mercaptopurine in childhood acute lymphoblastic leukemia. JAMA 293,
1485 1489.
Tangamornsuksan, W., Chaiyakunapruk, N., Somkrua, R., Lohitnavy, M., Tassaneeyakul, W.,
2013. Relationship between the HLA-B 1502 allele and carbamazepine-induced Stevens-
Johnson syndrome and toxic epidermal necrolysis: a systematic review and meta-analysis.
JAMA Dermatol. 149, 1025 1032.
van den Akker-van Marle, M.E., Gurwitz, D., Detmar, S.B., Enzing, C.M., Hopkins, M.M.,
Gutierrez de Mesa, E., et al., 2006. Cost-effectiveness of pharmacogenomics in clinical
practice: a case study of thiopurine methyltransferase genotyping in acute lymphoblastic
leukemia in Europe. Pharmacogenomics 7, 783 792.
CHAPTER 7
Special Requirements
for Economic Evaluation
and Health Technology
Assessment in Genomic Medicine
INTRODUCTION
From the previous chapter, it is obvious that genomic interventions could
enable improved patient stratification and tailor-made treatment interven-
tions. To date, however, most genetic tests focus on a gene-by-gene
approach, rarely detecting multiple genomic variants at high resolution
and accuracy. Technological breakthroughs have led to the development
and early adoption of genomic tests with the potential to meet these crite-
ria, namely whole-exome and whole-genome sequencing. The latter
technologies can simultaneously detect at a genome-wide scale all of an
individual’s genomic variants and, as such, have huge potential in genomic
medicine (Mardis, 2008; Mizzi et al., 2014). Hybrid technologies such as
next-generation sequencing panels sequence multiple genes known to be
associated with a given condition, such as hereditary cancer (Laduca et al.,
2014) or intellectual disability (Flore and Milunsky, 2012). These have the
advantages of improved sequencing quality and depth compared with
whole-exome and genome sequencing while minimizing issues related to
incidental findings (ACMG IF report). As an example for pharmacoge-
nomic testing, PGRNSeq (designed by the Pharmacogenomics Research
Network) is a next-generation sequencing platform that assesses sequence
variation in 84 proposed pharmacogenes (Gordon et al., 2012).
To date, only a small fraction of drugs have pharmacogenomic infor-
mation on their labels and are approved by the US Food and Drug
Administration (http://www.fda.gov) and the European Medicines
Agency (http://www.ema.europa.eu). Also, there is a variable rate of
adoption of genomics in various countries worldwide, and this rate is

lower in developing countries with limited resources. Such discrepancies

arise, in part, due to the lack of evidence for clinical effectiveness of
genetic tests. Recently, the use of whole-genome sequencing brought up
an additional concern regarding how incidental findings should be com-
municated to the patient. On top of this, there is a very limited pace of
translating genomic research findings into genomic medicine applications,
which almost always lack the relevant economic evaluation evidence
(Snyder et al., 2014).
Economic evaluation in genomic medicine has some specific require-
ments by nature. As such, certain approaches used in standard health eco-
nomics, such as CEA or quality-adjusted life-years (QALYs), may not
adequately capture the outcomes in genomic medicine and pharmacoge-
nomics, whereas other elements related to the economic evaluation
approach, such as stakeholder attitudes, societal perspective, and heteroge-
neity of cost data in different countries, are integral to the evaluation but
complicate the economic evaluation.
The preceding paragraph argues that existing economic evaluation
methods may not be adequate for the evaluation and technology assess-
ment of genomic medicine interventions. In particular, it should be clari-
fied which economic evaluation approach should be used for a particular
intervention and/or technology to be adopted into routine clinical
practice; this is a task that becomes more complicated as more
advanced—and reciprocally more expensive—genomic tests, such as
array-based genetic screening or next-generation sequencing methods,
emerge. To this end, the battery of tools for health economists will be
further enriched for them to provide their expertise to translate genomics
data into the clinic (Van Rooij et al., 2012).
In brief, there are four main items that should be taken into consider-
ation for economic evaluation in genomic medicine and health technol-
ogy assessment, namely analysis costs, analysis method, outcome of
analysis, and analysis effectiveness. These items are described in detail in
the following paragraphs.
ANALYSIS COSTS
The costs of each medical intervention constitute the cornerstone of all
economic evaluations. In contrast to cost data used in classic economic
evaluation, in genomic medicine there is often uncertainty regarding
which costs should be collected, the time that they should be collected,
Special Requirements for Economic Evaluation in Genomic Medicine 117
and how costs vary between different laboratories and health care systems.
There are a variety of factors that lead to this problem. For example, in
the United States, reimbursement for genetic tests has been based on a
“stack” of procedure-based codes (e.g., DNA extraction and purification,
amplification, sequencing, etc.). Different laboratories could use different
procedures to generate results, meaning that the same genetic test can
have markedly different costs and reimbursement due to the specific pro-
cedure that was used. Another issue is that a germline genetic test result
does not change over the course of a person’s lifetime and thus could be
used multiple times for care decisions. An example of this is the pharma-
cogene CYP2D6, which is estimated to be involved in the oxidative
metabolism of 25% of the drugs used in patient care (Samer et al., 2013).
How would one perform a cost-effectiveness analysis for CYP2D6 if the
horizon is the patient’s life span?
In genomic medicine, there are direct and indirect costs that should
be taken into consideration for economically evaluating a genomics-
related intervention (Table 7.1). This implies that health economists
involved in such economic evaluation should ensure that all relevant
costs are included in their analysis. Such costs would include patient
recruitment (Giacomini et al., 2003; Rogowski et al., 2010), sample
Table 7.1 Main costs items involved in economic evaluation in Genomic Medicine
Direct costs
• Patient recruitment, sample collection (blood, saliva, buccal swabs, etc.)
• Nucleic acid isolation
• Genetic testing, including amplification and purification
• Nature of genomic variant tested (germline vs. somatic)
• Data analysis, including data storage and analysis using informatics solutions
and genomic databases
• Frequency of data analysis (on the basis of novel genomics research findings)
• Accreditation, Quality Control of genetic testing platforms and assays
• Genetic counseling and results communication
• Post-testing actions (including treatment options, follow-up testing,
monitoring, etc.)
• Training of personnel, health care professionals
• Infrastructure acquisition and maintenance
Indirect costs
• Patient’s productivity loss
• Family costs (travel, accommodation, productivity loss)
collection (blood, saliva, buccal swabs, etc.) and delivery, genomic testing
(Giacomini et al., 2003), genetic data analysis, which includes the appro-
priate genome informatics tools and databases (Feero et al., 2013), report-
ing of genetic test results to the interested parties, namely patients and
their physicians, and genetic counseling (Veenstra et al., 2000; Payne,
2007; Faulkner et al., 2012; Singer and Watkins, 2012). Those costs
should be also accompanied by the costs accrued by the medical interven-
tions dictated by the genetic test results, such as costs of treatment
(from which the costs from the adverse drug reactions and follow-up tests
avoided due to the genetic testing should be deducted). The latter costs
can be significant. For example, a targeted resequencing or a microarray-
based approach, both of which are truly expensive tests on a stand-alone
basis, to identify the genetic basis of a rare mental disorder could be cost-
effective if the costs for follow-up testing can be avoided (Wordsworth
et al., 2007). As noted in Chapter 6, in the case of chronic hepatitis C,
even a small increment of improvement due to a genomic intervention
can be cost-effective if the consequences of treatment failure (end-stage
liver disease hepatocellular carcinoma) are very costly (Bock et al., 2014).
Another important cost item involves drug response and/or disease pro-
gression monitoring. In the case of warfarin treatment in which monitor-
ing, for example, CYP2C9 genotyping, is relatively cheap (Veenstra et al.,
2000), genomic interventions to individualize treatment may not be cost-
effective. At this point, it must be noted that this might not be the case in
developing countries, where genotyping and follow-up treatment costs
may vary (Mitropoulou et al., 2015). As such, genomic tests that are more
likely to be cost-effective are those for conditions for which monitoring is
expensive and cumbersome (Veenstra et al., 2000) or when adverse events
and treatment failures that could be prevented by the use of genomic test-
ing are very expensive (Kauf et al., 2010; Bock et al., 2014). Unfortunately,
there are not yet established guidelines and reimbursement rates for
genomic-based interventions and testing because the genomic technologies
either are too new for reimbursement rates to have been established, such
as whole-genome sequencing, or are integrated as part of a cost item for an
entire treatment modality, in which case costs may vary considerably
depending on the genetic testing laboratory cost policy (Payne, 2009; Van
Rooij et al., 2012; Singer and Watkins, 2012; Deverka et al., 2012;
Malhotra et al., 2012). The latter differences may be the result of the use of
either laboratory-developed assays versus commercially available and
quality-certified genotyping kits (De Leon, 2009; Djalalov et al., 2011) or
the variable prices among different countries (van den Akker-van Marle
et al., 2006). In the latter case, it is noteworthy that genome-guided warfa-
rin treatment may be cost-effective in developing countries (Mitropoulos
et al., 2011) that do not have dedicated anticoagulation monitoring clinics
but not cost-effective in developed countries (Mitropoulou et al., 2015),
which makes it difficult if not impossible to generalize economic evaluation
results between different health care systems.
The nature of the genomic variant being tested is also an important
parameter that should be taken into consideration in economic evaluation
studies pertaining to genomic medicine. Genomic variants may be either
germline, in which case genetic testing should only be performed once
during a patient’s lifetime (De Leon, 2009), or acquired/somatic, which in
some cases requires recurrent genetic analyses, like in the case of chronic
lymphocytic leukemia (Knight et al., 2012). In addition, for new technolo-
gies such as next-generation sequencing, data analysis with its attendant
costs may need to be repeated more than once because genomic data
interpretation is frequently updated with new genomics research results,
allowing for novel genotype phenotype correlations on data re-evaluation
(Gurwitz et al., 2009; Rogowski et al., 2010).
Apart from the direct costs indicated previously, there are also indirect
costs such as the productivity costs (e.g., time lost from work) or the costs
for a patient to seek the proper treatment, which reflects the time that will
be lost from work. To return to the warfarin cost-effectiveness example,
although the cost-effectiveness may be marginal, as noted by Meckley et al.
(2010), from the patient perspective having two to three fewer INR
measurements represents a significant reduction in life disruption that, if
quantified, could impact the cost-effectiveness conclusion. Another
example would be the use of a gene expression panel to stratify risk of
recurrence in endocrine receptor positive, node-negative breast cancer
patients to determine which patients are less likely to benefit from adjunct
chemotherapy based on a low recurrence risk (Vataire et al., 2012; Carlson
and Roth, 2013). These patients could choose to forego chemotherapy
with its attendant morbidity with a modest impact on cancer-related
outcomes, supporting this as a cost-effective (or possibly even cost-saving)
intervention (Rouzier et al., 2013).
Finally, education and training costs are additional cost items in genomic
medicine economic evaluation studies (Reydon et al., 2012; Kampourakis
et al., 2014), although these costs may be difficult to calculate and/or model
accurately because of the complexity of genetic tests performed in a single
instrument (e.g., real-time PCR or sequencer) or the economy of scale

(multiple instruments working in parallel) that would contribute in reducing
the overall analysis costs.
ANALYSIS METHOD
According to the few available economic evaluation studies in genomic
medicine, it seems that no golden rule is currently available, contrary to
the majority of economic evaluation studies of classical (nongenomic)
interventions that take into consideration the direct impact of a medical
intervention on the health care system and public health (Mette et al.,
2012). The information provided by genomic analysis, particularly whole-
genome sequencing performed very early in life, can have long-term
implications that are not taken into consideration with classical economic
evaluation studies. Consider identifying an α-synuclein gene variant
leading to Parkinson disease in young asymptomatic patients. In this case,
classical studies fail to estimate the long-term cumulative costs and effects
for such a patient (Rogowski et al., 2007, 2009; Grosse et al., 2010).
Also, the timing of an economic evaluation study may be an equally
important parameter because the genetic testing costs are rapidly decreasing,
whereas their specificity and accuracy are steadily increasing (Conti et al.,
2010; Goddard et al., 2012; Lin et al., 2012). This fact may lead to patient
subgroup stratification with direct impact on individualized treatments, or
the incorporation in the clinical practice of microarray-based genetic screen-
ing tests (e.g., AmpliChip [Roche Diagnostics], DMET1 [Affymetrix],
etc.) that were previously used for research purposes only.
Cost-effectiveness of a specific genomic technology and/or a genome-
based intervention can only be assessed in the context of a specific clinical
application involving a certain patient subpopulation. As such, when
performing economic evaluation for such a technology or intervention, it
is of utmost importance to carefully identify and select the comparators.
For example, one should assess both genetic and nongenetic testing in
addition to genetic in conjunction with nongenetic testing (Sanderson
et al., 2005; Hall et al., 2012); in the latter case, combining a cheap
conventional screening approach with an expensive genetic test may
prove to be cost-effective, like in the case of combining immunohis-
tochemistry with DNA sequencing, respectively, for the identification
of patients with Lynch syndrome among newly diagnosed patients
with colorectal cancer (Mvundura et al., 2010). In the case of warfarin,
the impact of nongenetic patient factors has at least as great an impact on

warfarin dose as does the pharmacogenomics variants and, as such, they
are included in warfarin dosing algorithms such as those found on warfar-
indosing.org. A study of warfarin dosing based on genomic factors alone
compared with clinical factors alone would not be deemed safe, which is
why all warfarin pharmacogenomics studies have compared a conven-
tional dosing algorithm with a conventional plus genomic algorithm.
It is important to carefully evaluate the appropriate perspective from
which the economic analysis is performed, because the results will vary
depending on the stakeholder perspective that is chosen. Although most
cost-effectiveness studies are done from the societal perspective, the utility
of the results for a given stakeholder perspective may be limited based on
the health care environment of the stakeholder.
Finally, economic evaluation studies should be restricted to genomic
medicine interventions that are approved by regulatory bodies because
the strength of scientific evidence about the genomic biomarker associ-
ated with the resulting clinical outcome(s) may directly impact the cost-
effectiveness of a therapeutic intervention (e.g., the cost-effectiveness of
Oncotype DX varies depending on whether it is being applied in breast
or colon cancer; Alberts et al., 2013; Holt et al., 2013).
OUTCOME OF ANALYSIS
There are several methodological issues in economic evaluation studies in
genomic medicine that arise when measuring the outcomes of a genomic
intervention, such as the type of outcome measure to be used in the anal-
ysis, personal utility, and the importance of individual outcomes. The last
of these is particularly relevant to genomic medicine because, although
the probability of an individual outcome is considered in economic analy-
ses, the predictive nature of genomic information provides more detailed
information specific to individual patients. This has led some to conclude
that new types of studies are needed to truly capture the impact of
genomic medicine (Hu, 2012; Garraway, 2013). If true, then this will
inevitably impact economic analyses as well.
In economic evaluation studies for genomic medicine interventions,
preference-based outcome measures (e.g., the EuroQol five-dimensional
[EQ-5D] questionnaire) should be used because they can be more easily
and uniformly compared among different population subgroups by col-
lecting data across a broad range of health-related quality-of-life domains.
However, QALYs measured using these measures do not reflect all possi-
ble health states applicable for genomic interventions (e.g., an asymptom-
atic patient with a predisposing genomic variant).
Genomic information, especially in the case of incidental findings, may
offer significant benefits and, at the same time, harms for a person. Some
genomic information lacks clinical utility (that is, there are no medical
actions that can be taken that can prevent or treat the condition identified
by the genomic test). This type of information can affect a person’s well-
being and decision-making (Foster et al., 2009; Roth et al., 2011; Garau
et al., 2013), a concept referred to as personal utility. There are factors that
have a positive impact on personal utility, such as prognostic or diagnostic
information that explains a certain genotype phenotype correlation, ends
a diagnostic odyssey, or rationalizes a therapeutic intervention. Also, geno-
mic information can be reassuring and can relieve anxiety (Asch et al.,
1996; Caughey, 2005; Grosse et al., 2010), allowing patients and their rela-
tives to better plan their lives and make lifestyle decisions, such as in the
case of APOE E4 allele testing as a predisposing factor in Alzheimer’s dis-
ease (Grosse et al., 2009, 2010; Payne et al., 2012). At the same time, there
are factors that have a negative impact on personal utility. Anxiety may be
increased if a genetic test result indicates that a patient is a nonresponder
or is at increased risk for development of adverse reactions to a certain
medical treatment, when no treatment alternatives exist (Conti et al.,
2010; Khoury et al., 2011), or when mutations of unknown significance
or other incidental findings are false-positives results (Jarrett and Mugford,
2006; Foster et al., 2009; Payne et al., 2012; Patrinos et al., 2013). In other
words, if a patient is genomically diagnosed with a predisposition for
development of Alzheimer’s disease prior to the development of symp-
toms, this does not provide them with any clinical utility as traditionally
defined because there are no effective treatments to prevent development
of the disease. However, it considerably impacts their personal utility,
sometimes in a positive way through adoption of a healthier lifestyle or
purchase of long-term care insurance, but it can also lead to anxiety,
despair, and pursuit of ineffective therapies that incur cost but provide no
benefit (Roberts et al., 2011). Such result may also worsen family dynam-
ics, lead to fear of discrimination and stigmatization (Grosse et al., 2008),
and impact reproductive decisions (Jarrett and Mugford, 2006; Foster
et al., 2009; Rogowski et al., 2010). False-negative genetic test results can
also lead to false reassurance and encourage unhealthy behaviors, such as
smoking and drinking (Grosse et al., 2008).
It is obvious that although personal utility is an important measure in

this context, it is very difficult to incorporate into economic evaluations
because we cannot accurately measure personal utility. Returning to the
Alzheimer’s disease example, some early work utilizing willingness to pay
for genetic testing is being used to explore the valuation of personal
utility (Kopits et al., 2011). Discussions among experts in genomics and
economics are needed to develop new approaches that could potentially
overcome these issues (Buchanan et al., 2013).
ANALYSIS EFFECTIVENESS
Economic evaluations also incorporate some measure of the effectiveness
of the intervention being evaluated, such as the unpredictable behavior of
patients and their physician. The complexity of genomic effectiveness
data makes the task of measuring the effectiveness of the analysis a
challenging one.
It is well-understood that the cost-effectiveness of a certain genomic
intervention is dependent on how patients and their physicians use these
interventions and how they respond to the genomic test results (e.g.,
whether patients change their lifestyle or alter their treatment modalities).
As such, information regarding behavioral responses should be included
into economic calculations to evaluate genomic technology and a geno-
mic medicine intervention (Asch et al., 1996; Caughey, 2005; Rogowski
et al., 2007; 2010; Conti et al., 2010). However, very little information
regarding how to model behavioral attitudes is available (Conti et al.,
2010). One cannot be sure whether all patients will want to have their
genome, in part or as a whole, tested, whether they will they adjust their
behavior to comply with the advice of a genetic counselor (Rogowski,
2009; Deverka et al., 2012), or whether physicians will take information
from genomic tests into account when making clinical decisions (Conti
et al., 2010). In some case, physicians confess that their genetics literacy is
very limited to make informed decisions based on genetic test results
(Mai et al., 2011, 2014). Some investigators such as those referenced in
the Oncotype DX and Alzheimer’s disease examples are using study
designs that capture patient and physician choices in response to genomic
information, thus generating data that can be used in subsequent eco-
nomic analyses. These analyses will have much more relevance to real-
world decision-making compared with studies that rely on untested
assumptions about patient and physician behaviors.
The level of evidence linking complex phenotypes and genomic data

to measurable health outcomes is generally very limited (Van Rooij et al.,
2012; Guzauskas et al., 2013), which makes it a challenging task to incor-
porate the information provided by genomic technologies into economic
evaluations in genomic medicine. For example, although there are more
than 150 CYP2D6 pharmacogene variants that can be used as pharmaco-
genomic biomarkers to more than 200 of the most commonly prescribed
drugs in oncology, psychiatry, and other fields, it is mostly unclear which
of these biomarkers are most clinically relevant (Van den Akker-van
Marle et al., 2006; EGAPP, 2007). Also, the current performance of
CYP2D6 genotyping assays is highly variable, leading to uncertainty
about the analytic validity of the testing (Fang et al., 2014). Other issues
that may arise include the uncertainty of how to incorporate diagnostic
characteristics into economic analyses, such as the analytic performance
of commercially available versus laboratory genetic diagnostic tests, which
may vary considerably (Garau et al., 2013) and for which there are limited
published data.
CONCLUSIONS
From the previous paragraphs, it is obvious that economic evaluation in
genomic medicine and health technology assessment has significant differ-
ences when compared with classical health economics and poses a num-
ber of challenges for health economists. To date, there are some solutions
to these challenges that have already been proposed and could be incor-
porated in future economic evaluations in genomic medicine. The goal
will be to introduce complex genomic effectiveness data into economic
analyses as soon as next-generation sequencing is incorporated into clini-
cal practice to facilitate informed decisions about implementation,
resource allocation, and investment.
Also, it is of paramount importance to incorporate personal utility of
genomic tests into the economic evaluation studies. In the United States
and in other countries where the genetic tests are reimbursed, a flat cost-
based scheme is used, leaving very little flexibility to consider personal
utility of a genetic test. However, this may be more relevant for regula-
tory bodies rather than health care providers, whose aim is to improve
public health and, as such, personal utility may not be so relevant (Grosse
et al., 2008). Early efforts to value personal utility should be increased to
develop effective solutions to this challenge.
Ideally, economic evaluation should be readily available and applicable

to convince policy-makers to engage in nationwide research projects to
use genomic approaches to relieve populations from major health bur-
dens, such as β-thalassemia in countries in the Mediterranean belt and
type II diabetes in Middle Eastern countries and China. Such genomic
research findings would improve the quality of life of citizens and at the
same time potentially reduce the overall health care expenditure. This
way, economic evaluation would enable researchers to compare the
expected benefits and cost-savings from implementing the results from
this genomics research with the expected costs of undertaking this
research, and it can highlight those cases for which new scientific evi-
dence would be particularly valued by decision-makers. The pace of
genomic technology innovation, especially in the dawn of the next-
generation sequencing era, suggests that investing in expensive projects in
the short-term may yield important results that would improve the quality
of life and result in cost-savings in the longer-term. Also, such studies
may strengthen the evidence of the effectiveness of specific genomic
interventions by collecting additional data. Such studies may involve
whole-genome sequencing analysis of large population groups from cer-
tain countries, such as the 100K genome project in the United Kingdom,
in Saudi Arabia, and elsewhere.
Finally, new methods and models may be required to resolve some of
the challenges outlined in the previous paragraphs and to improve the
quality of economic evaluations, which should be the focus of future
work in this field. Resolving these methodological issues will enable
decision-makers and policymakers to make better informed adoption
decisions, which may help accelerate the translation of pharmacoge-
nomics and genomics research into genomic medical practice.
REFERENCES
Alberts, S.R., Yu, T., Behrens, R.J., et al., 2013. Real-world comparative eco-
nomics of a 12-gene assay for prognosis in stage II colon cancer. J. Clin. Oncol.
31 (4), 391.
Asch, D.A., Hershey, J.C., Pauly, M.V., Patton, J.P., Jedrziewski, M.K., Mennuti, M.T.,
1996. Genetic screening for reproductive planning: methodological and conceptual
issues in policy analysis. Am. J. Public Health 86 (5), 684 690.
2014. Cost-effectiveness of IL28 genotype-guided protease inhibitor triple therapy ver-
sus standard of care treatment in patients with hepatitis C genotypes 2 or 3 infection.
Public Health Genomics 17, 306 319.
Buchanan, J., Wordsworth, S., Schuh, A., 2013. Issues surrounding the health economic
evaluation of genomic technologies. Pharmacogenomics 14 (15), 1833 1847.
Carlson, J.J., Roth, J.A., 2013. The impact of the Oncotype Dx breast cancer assay in
clinical practice: a systematic review and meta-analysis. Breast Cancer Res. Treat. 141
(1), 13 22.
Caughey, A.B., 2005. Cost-effectiveness analysis of prenatal diagnosis: methodological
issues and concerns. Gynecol. Obstet. Invest. 60 (1), 11 18.
Conti, R., Veenstra, D.L., Armstrong, K., Lesko, L.J., Grosse, S.D., 2010. Personalized
medicine and genomics: challenges and opportunities in assessing effectiveness, cost-
effectiveness, and future research priorities. Med. Decis. Making 30 (3), 328 340.
De Leon, J., 2009. Pharmacogenomics: the promise of personalized medicine for CNS
disorders. Neuropsychopharmacology 34 (1), 159 172.
Deverka, P.A., Schully, S.D., Ishibe, N., Carlson, J.J., Freedman, A., Goddard, K.A., et al.,
2012. Stakeholder assessment of the evidence for cancer genomic tests: insights from
three case studies. Genet. Med. 14 (7), 656 662.
Djalalov, S., Musa, Z., Mendelson, M., Siminovitch, K., Hoch, J., 2011. A review of eco-
nomic evaluations of genetic testing services and interventions (2004 2009). Genet.
Med. 13 (2), 89 94.
Fang, H., Liu, X., Ramı́rez, J., Choudhury, N., Kubo, M., Im, H.K., et al., 2014.
Establishment of CYP2D6 reference samples by multiple validated genotyping plat-
forms. Pharmacogenomics J. 14, 564 572.
Faulkner, E., Annemans, L., Garrison, L., Helfand, M., Holtorf, A.P., Hornberger, J.,
et al., 2012. Challenges in the development and reimbursement of personalized
medicine—payer and manufacturer perspectives and implications for health economics
and outcomes research: a report of the ISPOR personalized medicine special interest
group. Value Health 15 (8), 1162 1171.
Feero, W., Wicklund, C., Veenstra, D.L., 2013. The economics of genomic medicine:
insights from the iom roundtable on translating genomic-based research for health.
JAMA 309 (12), 1235 1236.
Flore, L.A., Milunsky, J.M., 2012. Updates in the genetic evaluation of the child with global
developmental delay or intellectual disability. Semin. Pediatr. Neurol. 19 (4), 173 180.
Foster, M.W., Mulvihill, J.J., Sharp, R.R., 2009. Evaluating the utility of personal geno-
mic information. Genet. Med. 11 (8), 570 574.
Garau, M., Towse, A., Garrison, L., Housman, L., Ossa, D., 2013. Can and should value
based pricing be applied to molecular diagnostics? Pers. Med. 10 (1), 61 72.
Garraway, L.A., 2013. Genomics-driven oncology: framework for an emerging paradigm.
J. Clin. Oncol. 31 (15), 1806 1814.
Giacomini, M., Miller, F., O’Brien, B.J., 2003. Economic considerations for health insur-
ance coverage of emerging genetic tests. Community Genet. 6 (2), 61 73.
Goddard, K.A., Knaus, W.A., Whitlock, E., Lyman, G.H., Feigelson, H.S., Schully, S.D.,
et al., 2012. Building the evidence base for decision making in cancer genomic medi-
cine using comparative effectiveness research. Genet. Med. 14 (7), 633 642.
Gordon, A.S., Smith, J.D., Xiang, Q., Metzker, ML., Gibbs, RA., Mardis, ER., et al.,
PGRNseq: a new sequencing-based platform for high-throughput pharmacoge-
nomic implementation and discovery. Abstract presented at the 62nd Annual
Meeting of the American Society of Human Genetics: November 6 8, 2012;
San Francisco, CA. Program #244. ,http://www.ashg.org/2012meeting/abstracts/
fulltext/f120122669.htm..
Grosse, S.D., Kalman, L., Khoury, M.J., 2010. Evaluation of the validity and utility of
genetic testing for rare diseases. Adv. Exp. Med. Biol. 686, 115 131.
Grosse, S.D., McBride, C.M., Evans, J.P., Khoury, M.J., 2009. Personal utility and geno-
mic information: look before you leap. Genet. Med. 11 (8), 575 576.
Grosse, S.D., Rogowski, W.H., Ross, L.F., Cornel, M.C., Dondorp, W.J., Khoury, M.J.,
2010. Population screening for genetic disorders in the 21st century: evidence, eco-
nomics, and ethics. Public Health Genomics 13 (2), 106 115.
Grosse, S.D., Wordsworth, S., Payne, K., 2008. Economic methods for valuing the outcomes
of genetic testing: beyond cost-effectiveness analysis. Genet. Med. 10 (9), 648 654.
Gurwitz, D., Zika, E., Hopkins, M.M., Gaisser, S., Ibarreta, D., 2009. Pharmacogenetics
in Europe: barriers and opportunities. Public Health Genomics 12 (3), 134 141.
Guzauskas, G.F., Garrison, L.P., Stock, J., Au, S., Doyle, D.L., Veenstra, D.L., 2013.
Stakeholder perspectives on decision-analytic modeling frameworks to assess genetic
services policy. Genet. Med. 15 (1), 84 87.
Hall, P.S., McCabe, C., Stein, R.C., Cameron, D., 2012. Economic evaluation of geno-
mic test-directed chemotherapy for early-stage lymph node-positive breast cancer.
J. Natl. Cancer Inst. 104 (1), 56 66.
Holt, S., Bertelli, G., Humphreys, I., Valentine, W., Durrani, S., Pudney, D., et al., 2013.
A decision impact, decision conflict and economic assessment of routine Oncotype
DX testing of 146 women with node-negative or pNImi, ER-positive breast cancer
in the UK. Br. J. Cancer 108 (11), 2250 2258.
Hu, V.W., 2012. Subphenotype-dependent disease markers for diagnosis and personalized
treatment of autism spectrum disorders. Dis. Markers. 33 (5), 277 288.
Jarrett, J., Mugford, M., 2006. Genetic health technology and economic evaluation: a
critical review. Appl. Health Econ. Health Policy. 5 (1), 27 35.
Kampourakis, K., Vayena, E., Mitropoulou, C., Borg, J., van Schaik, R.H., Cooper, D.N.,
et al., 2014. Key challenges for next generation pharmacogenomics. EMBO Rep. 15
(5), 472 476.
Kauf, T.L., Farkouh, R.A., Earnshaw, S.R., Watson, M.E., Maroudas, P., Chambers, M.G.,
2010. Economic efficiency of genetic screening to inform the use of abacavir sulfate in
the treatment of HIV. Pharmacoeconomics 28 (11), 1025 1039.
Khoury, M.J., Gwinn, M., Dotson, W.D., Bowen, M.S., 2011. Is there a need for
PGxceptionalism? Genet. Med. 13 (10), 866 867.
Knight, S.J.L., Yau, C., Clifford, R., Timbs, A.T., Sadighi Akha, E., Dréau, H.M., et al.,
2012. Quantification of subclonal distributions of recurrent genomic aberrations in
paired pre-treatment and relapse samples from patients with B-cell chronic lympho-
cytic leukemia. Leukemia 26 (7), 1564 1575.
Kopits, I.M., Chen, C., Roberts, J.S., Uhlmann, W., Green, R.C., 2011. Willingness to
pay for genetic testing for Alzheimer’s disease: a measure of personal utility. Genet.
Test. Mol. Biomarkers 15 (12), 871 875.
Laduca, H., Stuenkel, A.J., Dolinsky, J.S., Keiles, S., Tandy, S., Pesaran, T., et al., 2014.
Utilization of multigene panels in hereditary cancer predisposition testing: analysis of
more than 2,000 patients. Genet Med. 16, 830 837.
Lin, X., Tang, W., Ahmad, S., et al., 2012. Applications of targeted gene capture and
next-generation sequencing technologies in studies of human deafness and other
genetic disabilities. Hear. Res. 288 (1 2), 67 76.
Mai, Y., Koromila, T., Sagia, A., Cooper, D.N., Vlachopoulos, G., Lagoumintzis, G., et al.,
2011. A critical view of the general public’s awareness and physicians’ opinion of the
trends and potential pitfalls of genetic testing in Greece. Pers. Med. 8 (5), 551 561.
Mai, Y., Mitropoulou, C., Papadopoulou, X.E., Vozikis, A., Cooper, D.N., van Schaik, R.H.,
et al., 2014. Critical appraisal of the views of healthcare professionals with respect to phar-
macogenomics and personalized medicine in Greece. Pers. Med. 11 (1), 15 26.
Malhotra, A.K., Zhang, J.P., Lencz, T., 2012. Pharmacogenetics in psychiatry: translating
research into clinical practice. Mol. Psychiatr 17 (8), 760 769.
Mardis, E.R., 2008. The impact of next-generation sequencing technology on genetics.
Trends Genet. 24 (3), 133 141.
Meckley, L.M., Gudgeon, J.M., Anderson, J.L., Williams, M.S., Veenstra, D.L.A., 2010.
Policy model to evaluate the benefits, risks and costs of warfarin pharmacogenomic
testing. Pharmacoeconomics 28 (1), 61 74.
Mette, L., Mitropoulos, K., Vozikis, A., Patrinos, G.P., 2012. Pharmacogenomics and
public health: implementing populationalized medicine. Pharmacogenomics 13 (7),
803 813.
Mitropoulos, K., Johnson, L., Vozikis, A., Patrinos, G.P., 2011. Relevance of pharmaco-
genomics for developing countries in Europe. Drug Metabol. Drug Interact. 26 (4),
143 146.
Mitropoulou, C., Fragoulakis, V., Bozina, N., Vozikis, A., Supe, S., Bozina, T., et al.,
2015. Economic evaluation for pharmacogenomic-guided warfarin treatment for
elderly croatian patients with atrial fibrillation. Pharmacogenomics. (in press).
Mvundura, M., Grosse, S.D., Hampel, H., Palomaki, G.E., 2010. The cost-effectiveness
of genetic testing strategies for lynch syndrome among newly diagnosed patients with
colorectal cancer. Genet. Med. 12 (2), 93 104.
Patrinos, G.P., Baker, D.J., Al-Mulla, F., Vasiliou, V., Cooper, D.N., 2013. Genetic tests
obtainable through pharmacies: the good, the bad and the ugly. Hum. Genomics 7
(1), 17.
Payne, K., 2007. Towards an economic evidence base for pharmacogenetics: consideration
of outcomes is key. Pharmacogenomics 9 (1), 1 4.
Payne, K., 2009. Fish and chips all round? Regulation of DNA-based genetic diagnostics.
Health Econ. 18 (11), 1233 1236.
Payne, K., McAllister, M., Davies, L.M., 2012. Valuing the economic benefits of complex
interventions: when maximising health is not sufficient. Health Econ. 22 (3),
258 271.
Recommendations from the EGAPP Working Group: testing for cytochrome P450,
2007. Polymorphisms in adults with nonpsychotic depression treated with selective
serotonin reuptake inhibitors. Evaluation of genomic applications in practice and
prevention (EGAPP) working group. Genet Med. 9 (12), 819 825.
Reydon, T.A., Kampourakis, K., Patrinos, G.P., 2012. Genetics, genomics and society:
the responsibilities of scientists for science communication and education. Pers. Med.
9 (6), 633 643.
Roberts, J.S., Christensen, K.D., Green, R.C., 2011. Using Alzheimer’s disease as a model
for genetic risk disclosure: implications for personal genomics. Clin. Genet. 80 (5),
407 414.
Rogowski, W.H., 2007. Current impact of gene technology on healthcare. A map of
economic assessments. Health Policy 80 (2), 340 357.
Rogowski, W.H., 2009. The cost-effectiveness of screening for hereditary hemochroma-
tosis in Germany: a remodeling study. Med. Decis. Making 29 (2), 224 238.
Rogowski, W.H., Grosse, S.D., John, J., Kääriäinen, H., Kent, A., Kristofferson, U., et al.,
2010. Points to consider in assessing and appraising predictive genetic tests. J.
Community Genet. 1 (4), 185 194.
Rogowski, W.H., Grosse, S.D., Khoury, M.J., 2009. Challenges of translating genetic tests
into clinical and public health practice. Nat. Rev. Genet. 10 (7), 489 495.
Roth, J.A., Garrison Jr., L.P., Burke, W., Ramsey, S.D., Carlson, R., Veenstra, D.L., 2011.
Stakeholder perspectives on a risk-benefit framework for genetic testing. Public
Health Genomics 14 (2), 59 67.
Rouzier, R., Pronzato, P., Chéreau, E., Carlson, J., Hunt, B., Valentine, W.J., 2013.
Multigene assays and molecular markers in breast cancer: systematic review of health
economic analyses. Breast Cancer Res. Treat. 139 (3), 621 637.
Samer, C.F., Lorenzini, K.I., Rollason, V., Daali, Y., Desmeules, J.A., 2013. Applications
of CYP450 testing in the clinical setting. Mol. Diagn. Ther. 17 (3), 165 184.
Sanderson, S., Zimmern, R., Kroese, M., Higgins, J., Patch, C., Emery, J., 2005. How
can the evaluation of genetic tests be enhanced? Lessons learned from the ACCE
framework and evaluating genetic tests in the United Kingdom. Genet. Med. 7 (7),
495 500.
Singer, D.R.J., Watkins, J., 2012. Using companion and coupled diagnostics within strat-
egy to personalize targeted medicines. Pers. Med. 9 (7), 751 761.
Snyder, S., Mitropoulou, C., Patrinos, G.P., Williams, M.S., 2014. Economic evaluation
of pharmacogenomics: a value-based approach to pragmatic decision-making in the
face of complexity. Public Health Genomics 17, 256 264.
van den Akker-van Marle, M.E., Gurwitz, D., Detmar, S.B., Enzing, C.M., Hopkins, M.M.,
Gutierrez de Mesa, E., et al., 2006. Cost-effectiveness of pharmacogenomics in clinical
practice: a case study of thiopurine methyltransferase genotyping in acute lymphoblastic
leukemia in Europe. Pharmacogenomics 7 (5), 783 792.
Van Rooij, T., Wilson, D.M., Marsh, S., 2012. Personalized medicine policy challenges:
measuring clinical utility at point of care. Expert Rev. Pharmacoecon. Outcomes
Res. 12 (3), 289 295.
Vataire, A.L., Laas, E., Aballea, S., Gligorov, J., Rouzier, R., Chereau, E., 2012. Cost-
effectiveness of a chemotherapy predictive test. Bull. Cancer 99 (10), 907 914.
Veenstra, D.L., Higashi, M.K., Phillips, K.A., 2000. Assessing the cost-effectiveness of
pharmacogenomics. AAPS PharmSci. 2 (3), E29.
Wordsworth, S., Buchanan, J., Regan, R., Davison, V., Smith, K., Dyer, S., et al., 2007.
Diagnosing idiopathic learning disability: a cost-effectiveness analysis of microarray
technology in the national health service of the United Kingdom. Genomic Med. 1
(1 2), 35 45.
CHAPTER 8
A New Methodological Approach

for Cost-Effectiveness Analysis in
Genomic Medicine
INTRODUCTION
As mentioned in Chapters 4 and 5, economic evaluation compares the
costs and health effects of an intervention aiming to maximize health
benefits from available resources (Cartwright, 1999). So far, for innovative
interventions, the most commonly used formal investigation is implemen-
ted via the determination of incremental cost-effectiveness ratio (ICER;
O’Brien and Briggs, 2002; see Chapter 4). ICER is defined as the ratio
of incremental cost (ΔC ) of an intervention divided by the incremental
effectiveness (ΔE ), measured either as additional life-years (LY) or addi-
tional quality-adjusted life-years (QALYs). Based on the classical decision-
making approach of the cost-effectiveness analysis, ICER is compared
with a fixed amount that a policymaker is willing to pay (λ) for an addi-
tional LY or QALY. In this methodology context, if ICER is below λ,
the new intervention meets the basic criterion for reimbursement by the
payers (McCabe et al., 2008). In practice, this methodology assumes that
the policymakers are willing to pay this amount, regardless of the amount
of QALYs “purchased,” and in that way are ignoring the budget con-
straint. However, certain limitations must be considered in this approach
because: (i) there is no clear linkage between λ and the budget affordabil-
ity of health services (Towse, 2009) and (ii) it assumes, silently, an instant
adjustment of budget to absorb the cost of innovation (Gafni and Birch,
2006). Thus, the adoption of the existing process has led to the adoption
of costly decisions (Sendi et al., 2002; Gyrd-Hansen, 2005; Birch and
Gafni, 2006) and consequently increased expenditures for health care
systems (Gafni and Birch, 2006). Even if the budget does not face hard
constraints, it is reasonable to assume that there is an upper budget bound,
especially today in the presence of a global economic crisis. In addition,

λ has been traditionally determined one-dimensionally by existing theory,

without any consideration of the size of actual ΔE. Hence, λ has been
considered insensitive to the actual ΔE as a criterion for reimbursement.
In other words, in this methodology context, technology with a big
influence on human life for a single person (namely with high ΔE in
terms of QALYs compared with the standard therapy) would be reim-
bursed proportionally equally (e.g., up to h50,000 per QALY gained)
with those technologies that would provide only a modest incremental
health benefit (small ΔE ). A similar concept applies to those technologies
that are incrementally less effective but also less expensive compared with
the standard ones (negative ΔE ). The theory assumes that on the condi-
tions of constrained resources and a fixed willingness to accept (wta),
cost-saving interventions could be reimbursed by permitting more
efficient resource reallocation across health care sectors (O’Brien et al.,
2002; Nelson et al., 2009). This argument definitely imposes the need for
accounting the cost of new interventions by the policymakers. However,
there might be a lowest accepted bound difference between the standard
and a new intervention because there are moral issues that prevent the
adoption of a new, although far less effective, technology.
Several concerns have been raised regarding the methodological limita-
tion of decision-making in the cost-effectiveness analysis, and there might
be a need for an alternative approach (Gafni, 1998; Donaldson et al.,
2002a,b; Barton et al., 2008; Eckermann et al., 2008; Whitehead and Ali,
2010). A comprehensive decision-making process should not be insensitive
to the real impact on patients or to the payer’s budget. Previous attempts to
improve the decision-making process focused mainly on the problem of
maximization of QALYs across health care sectors given a total health bud-
get and using either linear or integer programming, depending on the
model assumptions (perfect divisibility, deterministic or stochastic approach
for outcomes, etc.; Weinstein and Zeckhauser, 1973; Stinnett and Paltiel,
1996). In such a case, some limitations must be addressed (Sendi et al.,
2002; Birch and Gafni, 2006). For instance, the actual determination of λ
in this kind of model requires the explicit knowledge of societal utility
function, which has to be maximized, and the explicit knowledge of prac-
tical importance for QALYs, which come from different therapeutic areas
(Wailoo et al., 2009). Of course in theory a “QALY is a QALY,” but in
practical applications it is somewhat difficult to compare an intervention in
cardiology with one in oncology or rheumatology or other field to extract
a common measure for comparisons among them. However, only if these
A New Methodological Approach for Cost-Effectiveness Analysis in Genomic Medicine 133
prerequisites were fully satisfied would the true value of λ or the desired
mix of budget across several disease-specific areas have been estimated.
Mathematical programming can incorporate some restrictions referred to
as ethical or technical issues (minimum budget spent on a specific disease,
divisibility, etc.), but an amount of exogenous information for the pattern
of budget allocation must also be provided as an input to solve the maximi-
zation problem. These are some crucial problems that face practical
research today. In this chapter, we propose a new simple way of thinking
and an alternative methodological approach for decision-making that takes
into account the budget constraint and relaxes the constant return on scale
assumption of λ.
NEW METHODOLOGICAL FRAMEWORK OF THE DECISION-

MAKING PROCESS IN GENOMIC MEDICINE
Let us introduce or repeat some simple notations:
S 5 standard/conventional intervention; T 5 new intervention;
E 5 effectiveness; C 5 cost
Bt 5 the total available budget the state is willing to pay for a patient’s
lifespan (t) (measured in a monetary unit such as h or $)
Ct 5 the total cost per patient of the new intervention
Cs 5 the total cost per patient of the standard/conventional
intervention
C(T )py 5 the average cost per patient per year of the new
intervention
C(S )py 5 the average cost per patient per year of the standard/conven-
tional intervention
Et 5 the effectiveness of the new intervention (measured in QALYs)
Es 5 the effectiveness of the standard/conventional intervention (mea-
sured in QALYs)
Cs 5 the total cost per patient of the standard/conventional
intervention
ΔE 5 the difference in the effectiveness between the standard/con-
ventional and the new intervention
ΔEmax 5 the maximum expected difference in effectiveness between
the standard/conventional and the new intervention that could be
additionally reimbursed given the budget
ΔC 5 the difference in the total cost between the standard/conven-
tional and the new intervention per patient
Given that we are referring to the “standard/conventional” interven-

tion, which has already been adopted from the health care system, it must
be obvious that Eq. (8.1) must be holding true:
Bt 5 CðSÞpy Es ð8:1Þ
For instance, if the state is willing to pay h10,000 for a patient during
the remaining lifetime (which is expected to be 2 years), the mean cost
per year for this patient that could be covered is limited to h5,000 per
year. Figure 8.1 depicts the relation between the average cost per patient
per year and the effectiveness of the standard intervention via the rational
function Eq. (8.1).
An increased effectiveness in a given budget Bt from the substitution
of standard intervention with a new intervention with equal total cost
must be followed by a decrease in the average cost per patient per year
and vice versa. The area under the curve is equal for each point across
the curve and expresses the total cost per patient. When a new and more
Cpy
C(S)py Budget constraint
C(T)py
Es Et E
Figure 8.1 Effectiveness and average cost per year for a given budget constraint.
Cpy: average cost per patient per year; C(S)py: average cost per patient per year for the
standard intervention; C(T)py: average cost per patient per year for the new intervention;
budget constraint refers to the available budget per patient for his/her life span; Es: effec-
tiveness of the standard intervention; Et: effectiveness of the new intervention.
costly intervention is introduced in the market, it must satisfy the follow-

ing equation:
CðT Þpy Et $ CðSÞpy Es ð8:2Þ
The reader must keep in mind that Eq. (8.2) is satisfied even if C(T)py
is equal, lower, or higher than C(S)py.
For instance, let us assume that the total cost of a new intervention has
been determined at 20,000 against 10,000 for the standard one. So, we have:
C(T)pyEt 5 20,000 and C(S)pyEs 5 10,000. If Et 5 2 LYs and Es 5 1 LY,
then C(S)py 5 C(T)py; if Et 5 3 and Es 5 1 LY, then C(S)py . C(T)py,
and so on.
This is depicted in Figures 8.2 and 8.3. Figure 8.2 depicts the case
where the most expensive and cost-infeasible interventions due to a specific
budget have equal, lower, or higher average costs per patient per year.
Figure 8.3 depicts the budget constraint expansion required for affording
new—more costly—interventions. Every single curve that is more distant
from the axes reveals an increased expenditure or a higher budget.
Cpy
ΔCpy>0
ΔCpy=0
C(S)py Budget constraint
ΔCpy<0
Es E
Figure 8.2 Cost-infeasible interventions with lower, equal, and higher average cost
per year. Cpy: average cost per patient per year; C(S)py: average cost per patient per
year for the standard intervention; ΔCpy: the difference in the average cost per year
per patient between the standard and new intervention; budget constraint refers to
the available budget per patient for his/her life span; Es: effectiveness of the standard
intervention.
In a general form, the mean cost per patient per year for the new
intervention could be expressed as:
Ct
CðT Þpy 5 or
Et
Ct
CðT Þpy 5 or
Es 1 ΔE
ð8:3Þ
Cs 1 ΔC
CðT Þpy 5 or
Es 1 ΔE
CðSÞpy Es 1 ΔC
CðT Þpy 5
Es 1 ΔE
Cpy
Budget constraint
map
C(T2)py
C(S)py
C(T1)py
Es Et1 Et2
Figure 8.3 New interventions and budget constraint expansion path. C(S)py: average
cost per patient per year for the standard intervention; C(T1)py: average cost per
patient per year for the new T1 intervention; C(T2)py: average cost per patient per
year for the new T2 intervention; budget constraint refers to the available budget
per patient for his/her life span; Es: effectiveness of the standard intervention;
Et1: effectiveness of the new T1 intervention; Et2: effectiveness of the new
T2 intervention.
Let us assume that the threshold for a QALY is unknown but equal to
λ. In addition to that, we know that ΔC 5 λ ΔE. Thus, we have:
CðSÞpy Es 1 λΔE
CðTÞpy 5 ð8:4Þ
Es 1 ΔE
From Eq. (8.4) it must be obvious that we have connected an incre-
mental quantity (λ) with an average one (C(T )py).
Thus, we can determine the expansion path of the average cost per
patient per year when a new, more costly intervention is introduced as a
function of ΔE, taking into account the (fixed) incremental threshold of λ.
If C(S)py 5 λ, then the graphical representation of C(T )py and Et is a
straight line. If C(S)py . λ (C(S)py , λ), then the graphical representation
is a downward (upward) curve with an asymptotic plateau of λ. Figure 8.4
depicts, for illustrative purposes, the case where C(S)py , λ. Figure 8.4
depicts that the threshold approach leads to decisions that result in
increased expenditures and thus raises concerns about the funding sustain-
ability. Figure 8.5 shows the opposite case. Given a fixed budget constraint
that has been exhausted, λ has a decreasing trend, is not constant across
Cpy
λ
C(T)py
Equilibrium given the
budget constraint
Budget constraint
map
E
Figure 8.4 Average cost and expanded budget constraint given a fixed λ. Cpy: aver-
age cost per patient per year; λ: willingness to pay; C(T)py: average cost per patient
per year for the new intervention; budget constraint refers to the available budget
per patient for his/her life span; E: effectiveness.
Decreasing λ in a
given budget constraint
Es Et E
Figure 8.5 Willingness to pay as a function of a fixed budget constraint and effec-
tiveness. λ: willingness to pay; E: effectiveness; Es: effectiveness of the standard/con-
ventional intervention; Et: effectiveness of the new intervention.
ΔE, and becomes asymptotically zero. Thus, λ tends to be zero in the case
of—potential—new interventions with high ΔE, because the society is
unable to afford the incremental cost of innovation. If the ΔE tends to be
small, then λ could be high; nonetheless, society probably is unwilling to
pay a high incremental cost for a small ΔE.
RELAXATION OF FIXED λ
In this paragraph, we go a step further concerning λ. The presented
approach is sensitive to the actual incremental effectiveness of a new
intervention. Let us assume that the budget B2t is fixed. The model
assumes there is a maximum expected incremental effectiveness (ΔEmax)
that could be additionally reimbursed. Hence, we could introduce a non-
linear S-shape function that determines λ as a function of actual ΔE,
where ΔE $ 0. As shown in Figure 8.6, the willingness to pay threshold
belongs to the S-shape family, ensuring that λ is different across ΔE in
every single point of the cost-effectiveness plane. For a very small ΔE
(i.e., a “me too” drug), and without any substantial differentiation point
from existing interventions, λ is expected to be low. If ΔE is fair or small,
then a higher λ with an increased marginal utility for the additional
QALYs can be expected. This is in accordance with economic theory
ΔC
Very innovative drugs
almost zero marginal utility
of QALYs, high λ
Innovative drugs
decreased marginal utility
=4
5, b
of QALYs, high λ h
ac
a= pro
ap
ne ant λ
4
=
l i
b
t t
igh cons
2,
tra
a=
S
=4
Fair drugs
,b
1.3
increased marginal utility
of QALYs, small λ
a=
“Me too” drugs
small λ
ΔEmax ΔE
Figure 8.6 A nonlinear S-shape function that determines the willingness to pay as a
function of actual ΔE. ΔE: difference in the effectiveness between the standard/con-
ventional and the new intervention; ΔC: difference in the total cost between the
standard/conventional and new intervention per patient; ΔCmax: available budget
(per patient) is willing to afford the budget holder to capture the maximum
expected effectiveness (ΔEmax); a and b are parameters of the equation:
b
1 2 eð2αΔE Þ
λ5 ΔCmax
ΔE
and practical evidence because a small increase in life expectancy has an

increased marginal utility (Kvamme et al., 2010) and the excess budget
has a decreased marginal utility (Gafni and Birch, 2006). If a new inter-
vention is very innovative in terms of ΔE, then λ must be diminishing
because of diminishing marginal utility of QALYs and increased marginal
utility of limited budget (Gafni and Birch, 2006).
In mathematical terms, relaxing the assumption of constant λ and
expressing it as a function of ΔE, λ 5 f(ΔE ), Eq. (8.4) becomes:
CðSÞpy Es 1 λðΔEÞ
CðT Þpy 5 ð8:5Þ
Es 1 ΔE
Thus, the path of average cost, the budget constraint map, and the
path of λ changes depend on the form of function of the willingness to
pay threshold. Let us assume that λ has a flexible form derived from the
Weibull family (Yin et al., 2003):
1 2 eð2αΔE Þ
b
λ5 ΔCmax ð8:6Þ
ΔE
where:
ΔCmax 5 The maximum available budget (per patient) the budget
holder is willing to afford to capture the maximum expected differ-
ence in effectiveness (ΔEmax).
ΔEmax has been assumed and a and b are constants. In such a case, the
straight line converts to a line segment (bounded by 0 to the left and
ΔEmax to the right). The new model partly incorporates the classical
model because we could find appropriate values for the parameters a and
b, which produce an approximately straight line segment or, to postulate
it, a line in the cost-effectiveness plane. It is paramount to consider that
in this model the budget upper limit is predetermined, and Figure 8.6
reveals the path of willingness to pay as we get closer to the budget
constraint.
A SIMPLE INDICATIVE EXAMPLE

Let us assume that the survival (measured in QALYs) of standard inter-
vention for a specific disease is estimated at 1 year (Es 5 1) and the total
cost until death is h10,000 (Es 5 h10,000). To keep it simple, let us
assume that the maximum amount that a budget holder is willing to pay
is h50,000 per patient (ΔCmax 5 h50,000 per patient) if, and only if, 1
more year is captured for each patient (ΔEmax 5 1). The function that
describes the willingness to pay path for a 5 5 and b 5 4 is:
1 2 eð25ΔE Þ
4
λ5 50;000
ΔE
assuming that a and b are parameters that have been estimated from real
data. If a new intervention has a ΔE that is higher than expected (i.e.,
ΔE 5 1.5), then λ is estimated as λ 5 ΔCmax/ΔE or λ 5 50,000/1.5 5
h33,000 per LY. In that case, the manufacturer must accept this λ due to
budget constraints of the health care system and the inability of budget
holders to absorb the cost of such an innovative (in term of ΔE) inter-
vention. In the extreme scenario where the actual ΔE was even higher, λ
tends to be zero, as depicted in Figure 8.5, but the benefit of new
intervention disperses in the society on behalf of the patients. In the usual

case where ΔE is less than 1 year, λ is estimated by Eq. (8.6).
For instance, let us assume an innovative technology with Et 5 1.8 (or
ΔE 5 0.8) and Ct 5 43,000. Thus, Eq. (8.6) estimates λ 5 54,438. ICER
is then estimated as ICER 5 ΔC/ΔE or ICER 5 33,000/0.8 5 41,250
per QALY. In this case, society is able to afford the new intervention and
is willing to reimburse it because the ICER is lower than society’s will-
ingness to pay for this survival benefit of ΔE 5 0.8. It is noteworthy that
λ was higher than ΔCmax 5 50,000 in Et 5 1.8 and the budget criteria
were met. Thus, the model incorporates the budget criteria, the prefer-
ences, and the cost-effectiveness criteria at the same time.
AN APPLICATION OF THE NEW MODEL IN

GENOMIC MEDICINE
To test the aforementioned concept using a genomic medicine scenario,
we consider an economic evaluation analysis comparing pharmacogenomic
(PGx)-guided treatment with conventional nonpharmacogenomic
(N-PGx)-guided treatment. For the purpose of our study, we performed
an economic evaluation of a pharmacogenomic-guided warfarin treatment
in elderly patients with atrial fibrillation, considering that both PGX and
N-PGx patients groups were homogeneous and stratified according to sex
and age and then randomly assigned to the two different treatment arms.
In the PGx patient group, genotyping analysis included
CYP2C9 2, 3, and VKORC1-1173C . T pharmacogenomic biomarkers
(see Chapter 3) prior to the prescription of warfarin, and the prescribed
warfarin dose was calculated based on the genotype and the algorithm
published in the nonprofit website http://www.WarfarinDosing.org.
Subsequently, the doses were adjusted depending on the measured INR
values. In patients with the “wild-type” (normal) CYP2C9 1/ 1 and
VKORC1-1173 CC genotypes, we applied a double dose of estimated
doses during the first 2 days of treatment because it is known that these
alleles are the slowest to achieve the target INR. After that, the doses
were adjusted depending on the INR measurement. In the N-PGx con-
trol group consisting of the same number of patients with the same indi-
cation for anticoagulation and with the same criteria for entry in the
study, warfarin was introduced by a fixed dose according to the standard
criteria without pharmacogenomic analysis and doses were then adjusted
depending on the INR values.
Based on the aforementioned available data and international litera-

ture, we constructed a pharmacoeconomic model to compare, from an
economic point of view, PGx-guided and N-PGx warfarin therapy in the
treatment of elderly patients with atrial fibrillation in a 1-year time
period. Our pharmacoeconomic model was a decision tree populated
with cost data, in line with current treatment guidelines on patient man-
agement, outcomes, and economic consequences. The model structure
was identical for each of the two assessed strategies.
In short, results showed that the total cost per patient was estimated to
be more than twofold higher for the PGx group (h540) compared with
the N-PGx control group (h230). This cost difference was due to the
genotyping costs. In terms of QALYs gained, the true difference in
QALYs was estimated at 0.01 in favor of the PGx group. The ICER of
the PGx versus the N-PGx control groups was estimated at (ΔC/
ΔE 5 h310/0.01) h31,000/QALY. Also, the results from the probabilistic
sensitivity analysis indicate that the treatment for the PGx group was
more expensive but also more effective than that for the N-PGx control
group.
Nonetheless, the cost-effectiveness of PGx is a subjective assessment
and depends on the willingness to pay per QALY gained with it. A con-
venient way of illustrating the results is the cost-effectiveness acceptability
curve, which shows the chances that a treatment is cost-effective relative
to another for different levels of willingness to pay. The cost-effectiveness
acceptability curve shows the probability (on the y-axis) that PGx may be
cost-effective compared with N-PGx for a range (on the x-axis) of maxi-
mum monetary values that a decision-maker might be willing to pay per
QALY. According to Figure 8.7A for the classical model (assuming linear shar-
ing between additional cost and effectiveness units without upper limit across ΔE),
the probability of PGx being cost-effective at h60,000 per QALY, is more
than 80%.
In the case of the new model, assuming ΔEmax 5 0.027 or 0.01 and
1 2 eð25ΔE Þ
3
λ5 310
ΔE
where ΔCmax 5 h310.0, different acceptability curves are determined.
The results of the probabilistic analysis are shown in Figure 8.7B. It must
be noted that the scenarios with ΔEmax 5 0.027 are too “ambitious” and
“demanding” because they require that PGx must have a difference of
0.027 QALYs against N-PGx to invest all the additional h310, and thus
(A)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
5. 0
9. 0
13 0
17 0
21 00
25 00
29 00
33 00
37 00
41 0
45 00
49 00
53 00
57 00
61 00
65 00
69 0
73 00
77 00
81 00
85 00
89 00
93 00
97 0
00
00
00
00
0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
1.
Willingness to pay for a QALY (€)
(B)
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
5. 0
9. 0
13 0
17 00
21 00
25 00
29 00
33 00
37 00
41 00
45 00
49 00
53 00
57 00
61 00
65 00
69 00
73 00
77 00
81 00
85 00
89 00
93 00
97 00
00
00
00
00
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
1.
Willingness to pay for a QALY (€)

Classical model ΔEmax= 0.027 QALYs ΔEmax = 0.01 QALYs
Figure 8.7 Cost-effectiveness acceptability curve for PGx group versus the N-PGx
control group using the classical model (A) and the new model (B). ΔEmax: maximum
expected difference in effectiveness between the N-PGx and PGx groups, which
could be additionally reimbursed given that the budget is 0.01 and 0.027 QALYs in
the two scenarios, respectively. λ: ðð1 2 eð25ΔE Þ Þ=ΔEÞ310 for the new models.
3
the probability of PGx being cost-effective is lower across willingness to

pay for a QALY against the classical model that is more conservative. On
the contrary, the scenario with ΔEmax 5 0.01 represents a society with
low expectations against the innovation, because the amount of h310 will
be invested even if the new technology (PGx) has only an incremental

difference of 0.01 QALYs against N-PGx (closer to classical model).
From this example, it must be clear that the probability of a new technol-
ogy being cost-effective is a function of the shape of λ across willingness
to pay (namely the parameters a and b) and the maximum amount a bud-
get holder is willing to pay to capture the maximum expected additional
effectiveness (namely is a function of ΔEmax and ΔCmax).
IMPLICATIONS
Today, the cost of innovation in health care is expected to be covered by
the third-party payers and usually leads to uncontrolled growth in health
care expenditure. In view of the scarcity of resources, economic evalua-
tion provides a criterion for the final decision concerning the adoption of
certain new interventions but has certain inconsistencies and drawbacks.
In this analysis, a new methodological approach was proposed taking into
account the budget constraint, the effectiveness of a new intervention,
and the willingness to pay in a flexible way.
In this new cost-effectiveness analysis model, it was assumed that the
budget is exogenous and has been set by the budget holders. It must be
noted that the health care budgets frequently depend on historical, politi-
cal, and social criteria without taking into account the economic models
and their assumptions (Schwappach, 2002). Furthermore, uncertainty in
the determination of costs and outcomes (Ramsey et al., 2005), lack of
data or knowledge for preferences (Weyler and Gandjour, 2011), lack of
training of policymakers (Veney et al., 1997), established status quo, and
other policy relevant issues set restrictions in a rational decision-making
process for the health maximization problem. Thus, the quantitative
determination of constraints and the application of prominent instruments
such as mathematical programming (Flessa, 2000) or other methodologi-
cal approaches (Sendi and Briggs, 2001) seem difficult to be implemented
practically for the optimum allocation of the health care budgets.
In this model, a far less ambiguous approach was adopted to estimate
the equilibrium that is driven by the de facto budget availability. The
model allows the determination of λ (via the budget constraint and the
S-curve for λ) to vary across ΔE. This argument of varying determina-
tion of λ is in accordance with related literature (Bridges et al., 2010;
Zhao et al., 2011). There are some notable differences between the pro-
posed new and classical models. In the proposed new model, the
maximum incremental effectiveness that could be additionally reimbursed

by payers is predetermined, taking into account the budget constraints.
The model presented here describes, in a dynamic manner, the link
between the willingness to pay threshold toward the budget limit, which
can be particularly helpful in developing and resource-limited countries,
where there are often significant budget limitations. In addition, the equi-
librium between λ, ΔE, and Bt is in accordance with a variable return on
scale assumption of λ, taking into consideration the size of ΔE and,
partly, the degree of innovation for a new intervention. Moreover, the
classical approach (fixed λ) has been incorporated as a particular case,
given that we could find appropriate coefficients in the Weibull equation
that produce almost a straight line segment in the cost-effectiveness plane.
The methodology, proposed in this chapter, shares the spirit of meth-
ods used in microeconomic theory. The model flexibly demonstrates how
willingness to pay changes as the budget remains fixed and allows ΔE to
vary, how the budget must be expanded to afford a new technology given
a specific willingness to pay path, fixed or not, or a combination of these.
This approach has some degree of generalization because it could possibly
be considered as a “standard intervention” in the bundle of interventions
that have already launched in the market and have been reimbursed by
the payers. Also, it should be noted that this model does not take into
account differences in the incidence of pharmacogenomic biomarkers,
which may vary significantly among different populations and ethnic
groups (Georgitsi et al., 2011).
The analysis focuses mainly on the presentation of results deterministi-
cally and gives less attention to estimating the equilibrium in a stochastic
manner. In such a case, the solution of the model could be estimated
with some degree of uncertainty depending on the assumption made for
the parameter of interest or the data availability. Inevitably, the implemen-
tation of the solution for practical purposes becomes much more complex
when taking into account the uncertainty of all variables.
CONCLUSIONS
The model presented in this chapter provides a new conceptual
methodology for the betterment of our understanding in decision-
making process, as well as a practical tool for educational purposes. The
simplification of the model’s assumptions and the calibration of the
mathematical equations to reflect more accurately the real-life setting

should very well be the scope of future research.
Overall, this new model would be particularly useful to perform eco-
nomic evaluation studies of novel genomic medicine interventions in dif-
ferent countries and health care systems, taking into account health care
budget differences, so that resource allocation in these countries could be
optimized.
REFERENCES
Barton, G.R., Briggs, A.H., Fenwick, E.A., 2008. Optimal cost-effectiveness decisions:
the role of the cost-effectiveness acceptability curve (CEAC), the cost-effectiveness
acceptability frontier (CEAF), and the expected value of perfection information
(EVPI). Value Health 11 (5), 886897.
Birch, S., Gafni, A., 2006. The biggest bang for the buck or bigger bucks for the bang:
the fallacy of the cost-effectiveness threshold. J. Health Serv. Res. Policy 11 (1),
4651.
Bridges, J.F., Onukwugha, E., Mullins, C.D., 2010. Healthcare rationing by proxy: cost-
effectiveness analysis and the misuse of the $50,000 threshold in the US.
Pharmacoeconomics 28 (3), 175184.
Cartwright, W.S., 1999. Methods for the economic evaluation of health care programmes,
second edition. By Michael F. Drummond, Bernie O’Brien, Greg L. Stoddart,
George W. Torrance. Oxford: Oxford University Press, 1997. J. Ment. Health Policy
Econ. 2 (1), 43.
Donaldson, C., Birch, S., Gafni, A., 2002b. The distribution problem in economic evalu-
ation: income and the valuation of costs and consequences of health care programmes.
Health Econ. 11 (1), 5570.
Donaldson, C., Currie, G., Mitton, C., 2002a. Cost effectiveness analysis in health care:
contraindications. BMJ 325 (7369), 891894.
Eckermann, S., Briggs, A., Willan, A.R., 2008. Health technology assessment in the cost-
disutility plane. Med. Decis. Making 28 (2), 172181.
Flessa, S., 2000. Where efficiency saves lives: a linear programme for the optimal alloca-
tion of health care resources in developing countries. Health Care Manage. Sci. 3 (3),
249267.
Gafni, A., 1998. Willingness to pay. What’s in a name? Pharmacoeconomics 14 (5),
465470.
Gafni, A., Birch, S., 2006. Incremental cost-effectiveness ratios (ICERs): the silence of the
lambda. Soc. Sci. Med. 62 (9), 20912100.
Georgitsi, M., Viennas, E., Gkantouna, V., Christodoulopoulou, E., Zagoriti, Z., Tafrali,
C., et al., 2011. Population-specific documentation of pharmacogenomic markers and
their allelic frequencies in FINDbase. Pharmacogenomics 12 (1), 4958.
Gyrd-Hansen, D., 2005. Willingness to pay for a QALY: theoretical and methodological
issues. Pharmacoeconomics 23 (5), 423432.
Kvamme, M.K., Gyrd-Hansen, D., Olsen, J.A., Kristiansen, I.S., 2010. Increasing mar-
ginal utility of small increases in life-expectancy? Results from a population survey. J.
Health Econ. 29 (4), 541548.
McCabe, C., Claxton, K., Culyer, A.J., 2008. The NICE cost-effectiveness threshold:
what it is and what that means. Pharmacoeconomics 26 (9), 733744.
Nelson, A.L., Cohen, J.T., Greenberg, D., Kent, D.M., 2009. Much cheaper, almost as
good: decrementally cost-effective medical innovation. Ann. Intern. Med. 151 (9),
662667.
O’Brien, B.J., Briggs, A.H., 2002. Analysis of uncertainty in health care cost-effectiveness
studies: an introduction to statistical issues and methods. Stat. Methods Med. Res. 11
(6), 455468.
O’Brien, B.J., Gertsen, K., Willan, A.R., Faulkner, L.A., 2002. Is there a kink in consu-
mers’ threshold value for cost-effectiveness in health care? Health Econ. 11 (2),
175180.
Ramsey, S., Willke, R., Briggs, A., et al., 2005. Good research practices for cost-
effectiveness analysis alongside clinical trials: the ISPOR RCT-CEA Task Force
report. Value Health 8 (5), 521533.
Schwappach, D.L., 2002. Resource allocation, social values and the QALY: a review of
the debate and empirical evidence. Health Expect. 5 (3), 210222.
Sendi, P., Gafni, A., Birch, S., 2002. Opportunity costs and uncertainty in the economic
evaluation of health care interventions. Health Econ. 11 (1), 2331.
Sendi, P.P., Briggs, A.H., 2001. Affordability and cost-effectiveness: decision-making on
the cost-effectiveness plane. Health Econ. 10 (7), 675680.
Stinnett, A.A., Paltiel, A.D., 1996. Mathematical programming for the efficient allocation
of health care resources. J. Health Econ. 15 (5), 641653.
Towse, A., 2009. Should NICE’s threshold range for cost per QALY be raised? Yes. BMJ
338, b181.
Veney, J.E., Williams, P., Hatzell, T., 1997. In search of failure: guidelines for ministries of
health. J. Health Popul. Dev. Ctries. 1 (1), 115.
Wailoo, A., Tsuchiya, A., McCabe, C., 2009. Weighting must wait: incorporating equity
concerns into cost-effectiveness analysis may take longer than expected.
Pharmacoeconomics 27 (12), 983989.
Weinstein, M., Zeckhauser, R., 1973. Critical ratios and efficient allocation. J. Public
Econ. 2, 21472157.
Weyler, E.J., Gandjour, A., 2011. Empirical validation of patient versus population prefer-
ences in calculating QALYs. Health Serv. Res.
Whitehead, S.J., Ali, S., 2010. Health outcomes in economic evaluation: the QALY and
utilities. Br. Med. Bull. 96, 521.
Yin, X., Goudriaan, J., Lantinga, E.A., Vos, J., Spiertz, H.J., 2003. A flexible sigmoid
function of determinate growth. Ann. Bot. 91 (3), 361371.
Zhao, F.L., Yue, M., Yang, H., Wang, T., Wu, J.H., Li, S.C., 2011. Willingness to pay per
quality-adjusted life year: is one threshold enough for decision-making? results from a
study in patients with chronic prostatitis. Med. Care 49 (3), 267272.
CHAPTER 9
Conclusions and Future

Perspectives
INTRODUCTION
The previous chapters outlined the various technical aspects and selected
applications of economic evaluation in genomic medicine. If applied
correctly, Cost-effectiveness analysis (CEA) and Cost utility analysis
(CUA) are promising technical analyses, but they are also characterized by
certain drawbacks. Some of these can be resolved if handled appropri-
ately; others involve political issues and conflicts of interest characteristic
of the health care sector in general. There are also challenges associated
with the nature of the subject itself.
Lack of Education
In many cases, health care decision-makers cannot understand the meth-
odological aspects of economic evaluation if they are not familiar with
them, which may lead to a lack of trust in the results. For example, they
are unable to understand basic concepts, such as the quality-adjusted life-
year (QALY), or the fact that a cost-effective treatment does not neces-
sarily mean savings in resources and that it could be just the “socially
acceptable option”—much less the concept of the “acceptability curve.”
They may also conclude that the results are not relevant because the anal-
yses were not done from the perspective of “their system.”
This problem is soluble in that policymakers can familiarize themselves
with the principles of economic analysis and can utilize experts conver-
sant with this scientific field and “translate” research findings into a lan-
guage that can be understood by the general public.
This Field Is Highly Technical but Not Very “Objective”

In recent years, CEA has become a highly technical field with many links
to mathematics and statistics, which means that nonexperts find it difficult
to understand, evaluate, or contribute to it. This problem is further

aggravated by the fact that conventional medical curricula do not include

such subjects, and doctors first come in contact with this field only after
completing their formal studies. Obviously, in this case valid doubts are
raised regarding the reliability of the analyses because those responsible for
applying the method’s conclusions are unable to validate them. Such an
overly technical analysis also has an additional problem: it allows the
health economists to modify a statistical approach and to be led to a
completely different result depending on the assumptions implemented
for their model. Lack of evidence for interventions, which is a particular
challenge in genomic medicine, increases the reliance on assumptions
that, if incorrect, can lead to misleading results. Application of sensitivity
analysis can only partially account for this problem.
There Is No Standard Methodology for Study Performance/

Presentation
Some have also claimed that economic evaluation suffers from a lack of
standardized methodology in performing the published studies. Even
among the pioneers of the field, one finds examples with differing results
depending on the assumptions made, a fact indicating that it is difficult to
compare economic studies, particularly if the publication does not ade-
quately present the methods and assumptions to allow independent
analysis. The standard literature exacerbates this problem with limitations
on word count, tables, and figures, although the ability to publish supple-
mental materials online can ameliorate this issue. Clinical effectiveness is
connected to physical reality and can be perceived easily, whereas the
standardized quantification of abstract economic quantities (e.g., loss of
productivity due to death) is much more difficult. We should mention
here that criticism of this issue has been productive and efforts are being
made by the scientific community to standardize the methodology. In
addition, many reputable journals specify important conditions with
regard to paper presentation to ensure that they are clear, ethical, repro-
ducible, and comprehensible. If someone were to follow the evolution of
the subject over the past 10 15 years, they would conclude that simplistic
and occasionally misleading analyses have been replaced by more papers
that are sophisticated, more definite, and objective, especially in the
leading journals.
Of course, one should remember that the nature of the subject itself
is complex and, in theory, attempts to answer a much more ambitious
question than a clinical study because it combines cost and effectiveness
Conclusions and Future Perspectives 151
data; furthermore, it attempts to standardize, in a unit model, conclusions

of uncertainty involving the entire society.
There Is Bias Because of Funding

This is a potential concern because companies funding such studies are for-
profit enterprises and seek to depict their products as useful for their “clien-
tele” (the medical community or the insurance funds). The same argument
could also hold for clinical or other trials and is the subject of increased scru-
tiny. In particular, creation of clinical guidelines by professional societies has
been noted to be dependent on nonobjective methods that are subject to
bias. The recent publication from the Institute of Medicine (Clinical
Guidelines We Can Trust; Graham et al., 2011) outlines an approach to
reduce the risk of bias in the guidelines. Similar recommendations could be
created for economic analysis to address these concerns (Berger and Olson,
2013). Journals have increased the requirements for disclosure of potential
conflict of interest and funding sources to enhance transparency and reduce
the publication of biased analyses. Also, as the subject “matures,” the compa-
nies themselves understand that an impartial analysis is more beneficial in the
long-term for their products’ image than any other approach. In fact, some
companies are using economic analyses to set prices based on cost-
effectiveness thresholds. Of course, the authors’ commitment to the study
sponsors is also a limitation, because they are asked to sign confidentiality
agreements and waive the right to publish their findings without authoriza-
tion. This can result in unfavorable results not being submitted for publica-
tion, although journals have also been shown to be less likely to publish
studies with negative results (Knight, 2003).
It Is Based on Multiple Assumptions

In many cases, CEA is “accused” of being dependent on too many assump-
tions for its models; therefore, its results cannot be used in real life by clini-
cal scientists or policymakers. Economic evaluation often collects data
from many sources to reach a conclusion. The unsuspecting reader might
believe that the results of the analysis are a “universal truth” even when the
model is beset with uncertainties. For example, in many cases we are inter-
ested not only in comparing the effectiveness of two interventions in a
clinical study but also in the long-term estimation of the incremental cost-
effectiveness ratio (ICER) for the same patients. In such cases, it is neces-
sary to combine data from the clinical study with observational or registry
data, in which case the degree of uncertainty increases. One solution to

this problem is to perform a CEA based on the clinical trial patient data
and then perform a supplemental sensitivity analysis to calculate the long-
term results. Another good and reliable solution is to explicitly declare the
assumptions of statistical models and the way that these affect the results. It
is always important to describe what programs were used to perform the
analysis, such as Excel or other programs.
It Uses a Different Perspective than the One Commonly Used

by Clinical Scientists or Policymakers
The estimation of the cost-effectiveness ratio is a primary goal of eco-
nomic evaluation; however, what politicians care about in most cases is
“reduced expenditures,” not “increased expenditure based on social crite-
ria.” The politician’s primary goal is not to exceed the available budget,
whereas the clinical scientist cares about prescribing the treatment that
will maximize the patient’s outcomes, including survival. The health
economist stands between the two, trying to mediate the two viewpoints.
Reproduction of Authority by the Medical Community

CEA is an “innocent” technical field of analysis, but it can actually and
indirectly set some limits on a doctor’s “authority” over patients and soci-
ety; therefore, its widespread application in practice has met with resis-
tance. For the first time, we stand at the brink of an important
development: a scientist outside the medical community (an economist)
“criticizes,” through their results, the practice of clinical scientists without
having the clinical background to correctly understand the consequences
of the theories they propose. In the United States, this was seen in the
enabling legislation that created and funded the Patient-Centered
Outcomes Research Institute (PCORI), which specifically prohibited
funding for cost-effectiveness studies. Although this was a political deci-
sion, it reflected an underlying concern that cost-effectiveness would
interfere with the practice of medicine at the discretion of physicians and
would interfere with the development and implementation of new drugs
and technologies. We believe that productive use of economic evaluation
is achieved when it is performed and evaluated with the help of scientists
from many different specialties, so that the study results have greater
validity and greater probability of being accepted by a large number of
health professionals.
It Is not “Ethical”
Economic evaluation seeks to maximize social welfare through statistical
approaches, whereas clinical scientists seek to maximize the welfare of
their patient (personalized medicine). In this sense, the clinical scientist’s
approach can never fully coincide with the economist’s because it has
been said that it is “unethical” to prescribe a treatment that we know not
to be the most effective one based on pharmacoeconomic criteria. To this
end, there are several ethical issues that are complex and therefore beyond
the scope of this book.
It Increases Costs
It has also been said that economic evaluation is mostly used by compa-
nies to convince insurance funds to reimburse their expensive treatments
without specifically indicating which competing treatment should be
rejected from the reimbursement system to meet the relevant budget
goals. Studies proposing immediate discontinuation of reimbursement of
specific treatments that are not cost-effective are rare or nonexistent. The
analysis usually proceeds in only one direction: the introduction of even
more treatments. This puts tremendous pressure on insurance funds,
which collapse under the burden of the economic obligations created by
the expectations of patient, providers, and policymakers.
Despite the negative statements regarding the disadvantages of eco-
nomic evaluation, it should be clear from this book that there is potential
for its utilization in the current political climate. Preparing an economic
evaluation is now a prerequisite for a medication to be reimbursed in
most European countries, which is a significant sign of the acceptance of
this approach. In several countries, the first steps have already been taken
and we are quickly proceeding in the direction of full implementation.
This development creates a basis that will help all shareholders in the
health care sector take into account the cost of the treatments they offer,
along with their benefit.
We believe that for new treatments it is very important to prepare
economic evaluations, both to estimate the cost-effectiveness ratio and to
determine their burden on the health system. This is not as important for
already existing health technologies, because these are established in clini-
cal practice and any studies performed are not expected to lead to signifi-
cant changes in practice.
For a new treatment, things are quite different. In cases of “mar-

ginal” innovation from a new technology, economic evaluation could be
used as a means of pressure on the industry to reduce prices; however,
in the case of actually innovative products, it would be a means to
reward the innovators. An important step would be the founding of an
academic, autonomous agency that would evaluate health technologies
and would provide its conclusions to the political authorities. In this
case, a careful reading of economic evaluations could potentially form a
natural nucleus from which to begin negotiations between state and
industry.
CONCLUSIONS
Economic evaluations are of utmost importance for genomic medicine to
demonstrate not only the cost-effectiveness of a certain intervention that
directly impacts on the national health care expenditure but also, most
importantly, its potential to improve the quality of life of the patients.
This is an emerging and highly promising field (Snyder et al., 2014) con-
sidering the very few economic evaluation studies in pharmacogenomics
and genomic medicine.
Furthermore, there is an urgent need at the moment to perform eco-
nomic evaluation studies in various developing countries to demonstrate
the cost-effectiveness of the individualization of certain therapeutic inter-
ventions that may not necessarily be cost-effective in developed countries.
For example, individualization of warfarin treatment has been shown to be
cost-effective for elderly Croatian patients (Mitropoulou et al., 2015), sug-
gesting that, contrary to other developed countries, pharmacogenomics-
guided warfarin treatment not only represents a cost-effective therapy
option for the management of elderly patients with atrial fibrillation in
Croatia but also may be the case for the same and other anticoagulation
treatment modalities in neighboring countries.
This fact dictates such economic evaluation studies to be replicated in
every health care system. To this end, a generic economic evaluation
model would be extremely useful to assess cost-effectiveness of individual-
ized treatment modalities. Such a tool could be used to perform a cost-
effectiveness threshold analysis across a variety of input conditions, such as
drugs, costs, and biomarkers, and would enable entities with no specific
expertise in economic analysis but with access to the necessary parameters
and cost items (see also Chapter 7) to reach a rapid decision regarding
whether a certain genome-based therapeutic intervention is cost-effective

in various settings/countries.
Considering the large economic evaluation evidence gaps in genomic
medicine, there is an urgent need to adequately address them to measure
the value of current and emerging genomic technologies, which would
allow appropriate value-based decisions about adoption and investment in
genomic medicine interventions. In essence, realizing the huge potential
of genomic medicine to benefit society, it may be of equal importance to
engage not only in innovative scientific discovery in genomics research
alone but also, most importantly, in scientific research to produce break-
throughs in the field of economic evaluation of genomic medicine, dem-
onstrating its value to those who will support and fund its integration
into mainstream medical practice.
REFERENCES
Berger, A.C., Olson, S., 2013. The Economics of Genomic Medicine: Workshop
Summary. Institute of Medicine of the National Academies, National Academies
Press, Washington, DC.
Graham, R., Mancher, M., Miller-Wolman, D., Greenfield, S., Steinberg, E., 2011.
Clinical Practice Guidelines We Can Trust. Institute of Medicine of the National
Academies, National Academies Press, Washington, DC.
Knight, J., 2003. Negative results: null and void. Nature 422, 554 555.
Mitropoulou, C., Fragoulakis, V., Bozina, N., Vozikis, A., Supe, S., Bozina, T., et al.,
2015. Economic evaluation for pharmacogenomic-guided warfarin treatment for
elderly Croatian patients with atrial fibrillation. Pharmacogenomics.
Snyder, S.R., Mitropoulou, C., Patrinos, G.P., Williams, M.S., 2014. Economic evalua-
tion of pharmacogenomics: a value-based approach to pragmatic decision-making in
the face of complexity. Public Health Genomics 17, 256 264.
INDEX
Note: Page numbers followed by “f ” and “t” refer to figures and tables, respectively.
A expected value of perfect information

Abacavir, 14, 98 100 (EVPI) concept, 91 93, 92f, 94t
Acenocoumarol, 12 13 mathematic models, 93
ACERs. See Average cost-effectiveness net monetary benefit (NMB), 93
ratios (ACERs) Adverse drug events (ADE), 29 30,
Acute lymphoblastic leukemia (ALL), 12, 100 104
100 104 balancing, 100 104
Acute myocardial infarction, 83 prevention, 97 100
ADE. See Adverse drug events (ADE) AEDs. See Antiepileptic drugs (AEDs)
Advanced methodological aspects, in Agency for Healthcare Research and
economic evaluation Quality (AHRQ), 5
discount rate, 77 83, 81t AHRQ. See Agency for Healthcare
estimates of effectiveness, 72 74 Research and Quality (AHRQ)
ICER/λ, size of, 67 72 ALL. See Acute lymphoblastic leukemia
budget size, 71 (ALL)
health system budgets, 71 Alzheimer’s disease, 122
probability of acceptance and, 70f cost effectiveness of, 123
shadow prices, 71 Analysis costs, 116 120
types of cancer, 67, 67t Analysis effectiveness, 123 124
methodological issues in, 65 Analysis method, 120 121
model types, 65 67 Antiepileptic drugs (AEDs), 109 110
absorbing models, 66 67 Antiretroviral drug therapies, 14
Markov model, 66 67 APOE E4 allele test, 122
mathematical modifications, 66 67 Atrial fibrillation, 109
sensitivity analysis, 83 91 warfarin therapy for, 141 142
for heterogeneity, 83 Average cost-effectiveness ratios (ACERs),
ICER calculation, 85 87, 86f, 87f, 40, 46
89f Azathioprine (AZA), 12, 100 101,
multiple analysis, 84 104 105
probabilistic sensitivity analysis (PSA),
84 B
single analysis, 84 Bayesian analysis, 91 94
uncertainty, 83 84 Biostatistical analysis, 39
study perspective and cost
determination, 74 77, 75t C
cost of disease, estimation of, 78t Cancer genetics professionals (CGPs), 18
direct costs, 76 Carbamazepine (CBZ), 109 110
health care sectors, 75 Case-only genome-wide interaction
indirect cost, 76 (COGWI), 20
invisible (intangible) cost, 76 CEA. See Cost-effectiveness analysis (CEA)
value of information and Bayesian CGPs. See Cancer genetics professionals
analysis, 91 94 (CGPs)
157
158 Index
Chronic hepatitis C, 14, 107 108 in genomic medicine, 108 110

Chronic inflammatory diseases, 17 cost-infeasible interventions, 135, 135f
Circumcision, 9 10 decision-making process, new
Clinical effectiveness, 150 methodological framework of,
Clinical Pharmacogenetics Implementation 133 138
Consortium, 10 11 implications, 144 145
Clinical trials, 72 incremental cost-effectiveness ratio
Clinical utility, demonstration of, 16 17 (ICER), 131 132, 141 142
Clopidogrel, 12 13 marginal utility, 138 139
CMA. See Cost-minimization analysis mathematical programming,
(CMA) 132 133, 144
Colorectal cancer (CRC), 105 107 new interventions and budget
Commodities constraint expansion path, 136f
combinations of, 2 3 new model, application of, 141 144
consumption of, 1 2 pharmacogenomic biomarkers, 145
Comparative effectiveness research, 32 33 relaxation of fixed λ, 138 140
Controlled experiments, 72 simple indicative example, 140 141
Cost determination, 74 77, 75t willingness to pay (λ), 138f, 139f
cost-benefit analysis, 75 highly technical, 149 150
direct costs, 76 marginal quantity, 40
health care costs, 76 reproduction of authority by medical
nonhealth care costs, 76 community, 152
health care sectors, 75 unique aspects and differentiation of,
indirect cost, 76 from alternative approaches, 38 39
invisible (intangible) cost, 76 vs. biostatistical analysis, 39
Cost measurement, in economic Cost-effectiveness plane, 138 139,
evaluation, 77 144 145
Cost of care, 5f Cost-effectiveness ratio, 152 153
Cost utility analysis, 28, 38, 53 57 Cost-infeasible interventions, 135, 135f
EuroQol EQ-5D, 53 54 Cost-minimization analysis (CMA), 28,
Health Utility Index, 53 54 43 44
quality-adjusted life-year (QALY), Cost-savings, 125
53 54 Cost-threshold analysis, 28 29
questionnaire, 56 Coumarinic oral anticoagulants (COAs),
time trade-off method, 55 56, 56f 12 13
typical game, 55, 55f CRC. See Colorectal cancer (CRC)
visual analog scale, 54 55, 54f CYP2C9, 13 14, 30 31
Cost-benefit analysis, 28 CYP2C9 gene, 13, 30 31, 118 119
Cost-effectiveness, 17, 90 91, 98 101, CYP2C9 2 allele, 30 31
103, 105, 107 108, 111, 119, 123 CYP2C9 3 allele, 30 31, 109 110
Cost-effectiveness analysis (CEA), 27 28, CYP2D6, 116 117, 124
37, 44f, 154 CYP2D6 7 allele, 109 110
appropriate measure for, 40 41 Cytochrome P450 isoenzymes, 13 14
assumptions-based, 151 152
decision criteria, 41 49 D
definition, 37 Databases, 73 74
and funding, 151 Decision tree, 66
Index 159
Decision-making, 69, 132 133 Evidence-based health economics (EBHE),

ICER, 67 1, 4 6
methodological framework of, 133 138 Evidence-based medicine (EBM), 4 6
Definition of economic evaluation, 3 4 Expected value of perfect information
Developing countries (EVPI) concept, 91 93, 92f, 94t
cost-effectiveness analysis, 108 109 Extended average lifespan, 7
health care costs in, 17 18 Extended dominance, concept of, 49 52,
Diagnostic characteristics, incorporation of, 49f
124
Direct costs, 76, 117t F
health care costs, 76 FDA. See Food and Drug Administration
nonhealth care costs, 76 Financial costs, 7
Discount rate, 77 83, 81t Food and Drug Administration (FDA),
Drafting of economic evaluations, 6 8 10 11, 32
Drug response and/or disease progression Funding, 151
monitoring, 118 119 Future perspectives, 149
E G
Economic evaluation of health services, Genetic discrimination/stigmatization,
scope of, 1 4 18 19
Economic stakeholders, 2 3, 5 Genetic Information Non-discrimination
Education and training costs, 119 120 Act (GINA), 19
Education, lack of, 149 Genetic testing, 11, 13 14, 16, 18 19
Effectiveness, estimates of, 72 74 insurance and privacy issues, 18 19
clinical databases, 74 National Institutes of
clinical trials, 72 Health Department of Energy
controlled experiments, 72 Task Force on, 16 17
databases, 73 Genomic era, economic evaluation in, 97
expert panels, 74 balancing adverse events and efficacy,
medical records, 73 74 100 104
meta-analysis, 72 73 cost-effectiveness analysis in genomic
Epidermal growth factor receptor (EGFR), medicine and developing world,
11 108 110
EQ-5D questionnaire, 56 diseases, 104 105
Ethics in genomics, 153 performance characteristics, defining,
European Alliance for Personalized 107 108
Medicine (EAPM), 20 21 perspectives on perspective, 105 107
European Medicines Agency (EMA), pharmacogenomics and adverse events
10 11, 115 116 prevention, 97 100
European Network on Methodology and Genomic medicine, 9
Application of Economic central dogma of, 9
Evaluation Techniques, 60 61 clinical utility, demonstration of, 16 17
EuroQol EQ-5D, 53 54 current challenges of, 14 15
Evaluation, defined, 3 economic evaluation in, 15f
Evaluation of Genomic Applications in education of health professionals, 15 16
Practice and Prevention (EGAPP), future perspectives, 19 21
11, 13 14, 16 17 health care costs, 17 18
160 Index
Genomic medicine (Continued) probability of acceptance, 70f

insurance and privacy issues, 18 19 shadow prices, 71
pharmacogenomics and, 10 types of cancer, 67, 67t
applications, 10 14 IL28B, 107 108
cardiology examples, 12 13 Incremental cost-effectiveness ratio
infectious disease examples, 14 (ICER), 45 47, 50, 98, 109 110,
oncology examples, 11 12 131 132
psychiatric examples, 13 14 Incremental effectiveness, 131 133,
stakeholder analysis in, 15f 138 141
Genomics and Personalized Medicine Act Indirect cost, 76, 117t
(GPMA), 19 Inflammatory bowel disease (IBD), 12
Genomics awareness, 15f, 16 Inflation, 78 79, 82 83
INR. See International normalized ratio
H (INR)
HCV genotypes 2 and 3, 108 Insurance and privacy issues, 18 19
Health care costs, 17 18 Insurance funds, 153
Health economics, 27 Intangible cost, 76
cost-benefit analysis, 28 International normalized ratio (INR), 31,
cost-effectiveness analysis (CEA), 27 28 141
cost-minimization analysis (CMA), 28 Invisible cost, 76
cost-threshold analysis, 28 29 Irinotecan, 11
cost-utility analysis (CUA), 28
economic evaluation tools used in, 27 29 L
genomics in, 29 32 Lynch syndrome, 105 107, 120 121
key features in, 27
triangulating genomics with health M
technology assessment and, 32 33 Marginal quantity, concept of, 40
Health economists, 9 Mean measurement, 39
Health Insurance Portability and Medical errors, 7
Accountability Act Privacy Rule, Medical records, 73 74
19 Meta-analysis, 72 73
Health professionals, education of, 15 16 Microarray-based genetic screening tests,
Health Resources and Services 120
Administration (HRSA), 5 Microeconomic theory, 145
Health Utility Index, 53 54 Minimal residual disease (MRD),
Hepatitis C virus (HCV) genotype 1, 100 101, 103 104
107 108 Model types, in economic evaluation,
HLA-B 15:02 allele, 109 110 65 67
Human immunodeficiency virus (HIV-1), absorbing models, 66 67
98 99 chronic diseases, 65
Hybrid technologies, 115 decision tree, 66
Markov model, 66 67
I mathematical modifications, 66 67
ICER/λ size, 67 72 Modern unhealthy lifestyle, 7
budget size, 71 Monte Carlo simulations, 98
decision-making process, 67 MRD. See Minimal residual disease (MRD)
health system budgets, 71 Myelosuppression, 100 104
Index 161
N infectious disease examples, 14

National Institute for Health and Care oncology examples, 11 12
Excellence (NICE), 70 psychiatric examples, 13 14
National Society of Genetic Counselors challenges for, 14 15
(NSGC), 18 definition, 10
Net health benefit (NHB), 48 49 genomic medicine and, 10
Net monetary benefit (NMB), 47 49, 48f, short-term analysis, 99
93 Pharmacokinetics, 29 30
NMB. See Net monetary benefit (NMB) Phenprocoumon, 12 13
Nonpharmacogenomic (N-PGx)-guided Phenytoin, 109 110
treatment, 141 142 Policymakers, 152
Number needed to treat (NNT) symbol, Population allele frequency, 109 110
45 Postgenomic era in economic evaluation,
33 34
O Preference-based outcome measures,
Oncotype DX, cost effectiveness of, 123 121 122
One- and two-way sensitivity analyses, 104 Present discounted value (PDV), 80 82, 82t
Outcome of analysis, 121 123 Principle of convexity, 2
Psoriasis, 12
P Public’s expectations, 7
Patient Centered Outcomes Research
Institute (PCORI), 5, 152 Q
Personal utility, 122 123 QALY. See Quality-adjusted life-years
Personalized medicine, 9 10, 13 14, 19 21 Quality of Health Economic Studies
Personalized Medicine Coalition (PMC), (QHES), 60 61, 61f, 99 100
20 21 Quality of Welfare Scale (QWB), 53 54
PGRNSeq, 115 Quality-adjusted life-year (QALY), 28,
Pharmacodynamics, 29 30 53 54, 98 99, 109, 131 133,
Pharmacoeconomic analyses, 100 102 140 141
Pharmacogenetics, 29 32 allocative efficiency, 57 58
Pharmacogenomic testing, 10 11, 17 18 analytical approach, 58 60
acenocoumarol for, 12 13 conceptual approach, 57 60
azathioprine for, 12
clopidogrel for, 12 13 R
cost-effectiveness of, 17 Recursive simulation techniques, 39
coumarinic oral anticoagulants (COAs) Research projects, engaging, 125
for, 12 13
irinotecan for, 11 S
phenprocoumon for, 12 13 Selective serotonin reuptake inhibitor
trastuzumab for, 11 (SSRI), 13 14
warfarin for, 12 13, 30 Sensitivity analysis, 83 91
Pharmacogenomics (PGx)-guided acceptability curve
treatment, 141 142 derivation of, 89f
Pharmacogenomics, 15f, 29 30, 32 representation of, 90f
adverse events prevention, 97 100 common forms of, 84 85
applications of, 10 14 cost-effectiveness acceptability curve
cardiology examples, 12 13 (CEAC), 90
162 Index
Sensitivity analysis (Continued) typical game, 55, 55f

ICER in, 85 87, 86f, 87f, 89f visual analog scale, 54 55, 54f
multiple, 84 cost-effectiveness analysis (CEA), 37, 44f
probabilistic, 84, 88f appropriate measure for, 40 41
single, 84 average cost-effectiveness ratios
stochastic analysis, 88 (ACERs), 40, 46
uncertainty, 83 84 decision criteria, 41 49
Serious bleeding, 30 32 examples, 38
6-mercatopurine (6-MP), 100 102 incremental cost-effectiveness ratio
6-thioguanine (6-TG), 12, 100 101 (ICER), 45 47
Social welfare, maximizing, 153 mean effectiveness, 41 42
Special Interest Group (SIG), 18 mean measurement, 39
Special requirements for economic net health benefit (NHB), 48 49
evaluation and health technology net monetary benefit (NMB), 47 49,
assessment, 115 48f
analysis costs, 116 120 new vs. standard treatment, 41 42,
costs items involved in economic 44 45
evaluation, 117t number needed of treat (NNT), 45
drug response/disease progression physical units, intervention of, 38
monitor, 118 119 primary goals, 37 38
genome-guided warfarin treatment, process of, 37
118 119 therapeutic intervention, 40
productivity costs, 119 unique aspects and differentiation of,
analysis effectiveness, 123 124 from alternative approaches, 38 39
analysis method, 120 121 value framework, 42f
comparators, identification/selection vs. biostatistical analysis, 39
of, 120 121 willingness to accept (WTA), 46 47
warfarin dosing algorithms, 120 121 willingness to pay (WTP), 46 47
outcome of analysis, 121 123 economic analysis, quality assessment of,
Stakeholder, 32 33, 121 60 61
awareness, 16 extended dominance, 49 52, 49f
Stevens-Johnson syndrome (SJS), 109 110 available options, 51 52
Stochastic analysis, 88 linear option combinations, 52
Study performance/presentation, 150 151 results, 52
Sustained viral response (SVR), 107 108 QALY, conceptual approach to, 57 60
SVR. See Sustained viral response (SVR) allocative efficiency, 57 58
analytical approach, 58 60
T preferences factors, 58
Technical issues, of economic evaluation, 37 Tenofovir, 98 99
alternative options, selection of, 53 Thiopurine, 12, 100 102, 104 105
cost utility analysis, 53 57 Thiopurine methyltranferase (TPMT), 12,
EuroQol EQ-5D, 53 54 100 101
Health Utility Index, 53 54 genotyping, 101 105
quality-adjusted life-year (QALY), Toxic epidermal necrolysis (TEN), 109 110
53 54 Trastuzumab, 11
questionnaire, 56 Triple therapy, 98 99, 107 108
time trade-off method, 55 56, 56f Tumor-based screening, 105 106
Index 163
U W
UGT1A1 28, 11 Warfarin, 12 13, 30 32
Unhealthy habits, avoidance of, 79 80 for atrial fibrillation, 141 142
Unilateral consumption, 2 concept of, 1 6
United States Preventive Services Task dosing algorithms, 120 121
Force (USPSTF), 5 genome-guided warfarin treatment,
118 119
V Whole-exome sequencing, 115
Value framework, 5f, 42f Whole-genome sequencing, 19 20,
Value of information and Bayesian analysis, 32 34, 115 116, 118 119, 125
91 94 Willingness to accept (WTA), 46 47,
Value-based health care, 4 131 132
V-erb-b2 erythroblastic leukemia viral Willingness to pay (WTP), 46 47, 138f, 139f
oncogene homolog 2 nonlinear S-shape function, 139f
(HER2/ERBB2), 11 relaxation of fixed λ, 138 140
Vitamin K epoxide reductase (VKORC1), WTA. See Willingness to accept (WTA)
13, 30 31 WTP. See Willingness to pay (WTP)

Economic Evaluation in Genomic Medicine

Uploaded by

Copyright:

Available Formats

Economic Evaluation in Genomic Medicine

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

Economic Evaluation in Genomic Medicine

Uploaded by

Copyright:

Available Formats

What is the subject of the textbook?

What is the subject of the textbook?

What are some of the main concepts discussed in the textbook?

What are some of the main concepts discussed in the textbook?

Economic Evaluation in

AMSTERDAM • BOSTON • HEIDELBERG • LONDON

For information on all Academic Press publications

Typeset by MPS Limited, Chennai, India

Publisher: Christine Minihane

The subject of this textbook is the emerging scientific field of applied

Economic Evaluation in Health

THE SCOPE OF ECONOMIC EVALUATION OF

Economic Evaluation in Genomic Medicine © 2015 Elsevier Inc.

the behavior of all these actors should be determined by certain simple

of commodities associated with different costs, and that there is a specific

economic evaluations, we seek a complete, thorough, and accurate compari-

EVIDENCE-BASED MEDICINE AND EVIDENCE-BASED

Medical and/or Cost of care Cost of care Cost of care

Figure 1.1 The value framework. Value is represented as a relationship between

opposite to those initially expected. For example, the adoption of practices

THE NECESSITY OF DRAFTING ECONOMIC EVALUATIONS

• The resolution to provide full health care coverage to the population

Genomic Medicine Today:

Economic Evaluation in Genomic Medicine © 2015 Elsevier Inc.

(http://download.yutorah.org/2013/1053/793670.pdf). At present, there is a

PHARMACOGENOMICS AND GENOMIC MEDICINE

(http://www.pharmgkb.org/page/cpic). Currently, pharmacogenomics

Azathioprine, 6-mercaptopurine, and 6-thioguanine are drugs that are

half a century, coumarinic oral anticoagulants (COAs), namely warfa-

implicated CYP2D6 genomic variants as a predictor of risperidone-induced

Infectious Disease Examples

FACING THE CURRENT CHALLENGES OF GENOMIC

paragraphs, we outline the various challenges and bottlenecks that

EDUCATION OF HEALTH PROFESSIONALS

pharmacogenomics and personalized medicine education in core medical

DEMONSTRATION OF CLINICAL UTILITY

test to prevent or ameliorate adverse health outcomes such as mortality,

HEALTH CARE COSTS

important parameter for health care costs related to genomic medicine

INSURANCE AND PRIVACY ISSUES

for economic reasons, patients with rare or complex genetic conditions

CONCLUSIONS AND FUTURE PERSPECTIVES

are, by their nature, more complex and heterogeneous when compared

groups, industry policy organizations, academic institutions, and government

Economic Evaluation and

Economic Evaluation in Genomic Medicine © 2015 Elsevier Inc.

additional costs compared with alternative approaches. It is quantitative

This analytic approach is being used more frequently by developers to

GENOMICS IN HEALTH ECONOMICS

account for 20 95% of the observed responses to drug therapies, one

with the wild-type CYP2C91 allele. As a result, patients with a

According to previously published reports in 2004 and 2005, side

TRIANGULATING GENOMICS WITH HEALTH TECHNOLOGY

ECONOMIC EVALUATION IN THE POSTGENOMIC ERA

and tools have to be developed based on recently described analyses, such

versus standard of care treatment in patients with hepatitis C genotypes 2 or 3 infec-

Introduction to the Technical

Economic Evaluation in Genomic Medicine © 2015 Elsevier Inc.

economics (Canning, 2009). As already mentioned, it is still (at a theoret-

UNIQUE ASPECTS AND DIFFERENTIATION OF CEA