Cost-effectiveness analyses of genetic and genomic diagnostic technologiesKatherine Payne, Sean P. Gavan, Stuart J. Wright and Alexander J. Thompson

Manchester Centre for Health Economics, Division of Population Health, Health

Services Research & Primary Care, The University of Manchester, Manchester, M13

9PL, UK

Correspondence to: Katherine Payne (katherine.payne@manchester.ac.uk)

AbstractDevelopments in next-generation sequencing technologies have driven the clinical

application of diagnostic tests that interrogate the whole genome, which offer the

chance to diagnose rare inherited diseases or inform the targeting of therapies. New

genomic diagnostic tests will compete with traditional approaches to diagnosis,

including the genetic testing of single genes, and other clinical strategies for the

available finite healthcare budget. In this context, decision analytic model-based cost-

effectiveness analysis is a useful method to help evaluate the costs versus benefits of

introducing new healthcare interventions such as genomic diagnostic tests. This Review

presents key methodological, technical, practical and organizational challenges that

must be considered by decision-makers responsible for the allocation of healthcare

resources to provide robust and timely information about the relative cost-effectiveness

of the increasing numbers of emerging genomic tests.

Introduction Historically, genetic tests have focused on identifying the causative single genetic

variant that results in an observed phenotype (clinical expression). This form of testing

usually occurs in the context of specialist genetic center that provide a diagnosis, risk

estimation, counselling, surveillance and support to people with inherited, often rare,

genetic conditions1. Of note, no single model of genetic service delivery currently exists,

and there is extensive variation within and across countries2. Advances in next-

1

generation sequencing technologies have yielded genome-wide diagnostic tests

(hereafter called ‘genomic tests’), which are based on either multi-gene panels, whole-

exome sequencing or whole-genome sequencing. Informed by both public and privately

funded research, genomic tests are emerging in countries with advanced market

economies4, such as the United States of America5,6, Canada7,8, France9 and the United

Kingdom10, and some genomic tests are starting to replace existing diagnostic

strategies, including in some instances genetic tests, in clinical practice3 (Table 1)11.

Practicing clinical geneticists understand the impact of the diagnostic odyssey on

patients and their families. Published evidence also highlights the detrimental effect that

not having a clear diagnosis of a condition that causes distressing symptoms can

have12. Genomic tests may offer an increased chance of diagnosis by moving the gene-

based diagnostic component to the front of the diagnostic care pathway13. Moreover in

some instances genomic tests have the potential to inform treatment decisions.

However, the existence of fixed annual healthcare budgets alongside the continued

development and emergence of new interventions to prevent or treat health conditions

provides decision-makers charged with the allocation of healthcare resources with an

enduring challenge. Evidence of the relative cost-effectiveness, which explicitly

quantifies the additional resource use (costs) and consequences (benefits and harms)

of a new intervention compared with other potential uses of a healthcare budget, can

provide decision-makers with useful information to guide resource allocation decisions

(Figure 1). Methods of economic evaluation, such as cost-effectiveness analysis (CEA)

and cost-benefit analysis (CBA), are available to obtain this evidence. In some

countries, such as the Australia and UK, national decision-making bodies use evidence

of cost-effectiveness to inform clinical guidelines or reimbursement decisions. However,

in contrast to most healthcare interventions, the nature of the decision-maker charged

with allocating finite healthcare resources to genetic and genomic technologies, and

associated diagnostic tests, is not well defined. Such decision-makers operate within

different levels of existing healthcare systems, and their roles and responsibilities differ

within and across countries.

This Perspective reviews the role of cost-effectiveness analysis in informing the

introduction of genetic and genomic tests for the diagnosis of inherited rare diseases or

2

targeted therapies, with an emphasis on evaluative frameworks employed by national

decision-making bodies such as the National Institute for Health and Care Excellence

(NICE) under the remit of NHS England. We review the key elements to consider in the

cost-effectiveness analysis (CEA) of genetic and genomic tests and services, before

specifically describing the potential of decision analytic model-based CEA, as well as its

methodological, technical, practical and organizational challenges. The emerging

economic evidence base for genetic tests, but lack of associated evidence for genomic

tests, is also summarized.

The cost of sequencing The increase in the clinical application of genomic tests is generally credited to the

decline in the cost of sequencing, which was facilitated by the move from Sanger

sequencing to next-generation sequencing technologies. According to data from the

National Human Genome Research Institute (NHGRI), up until the end of 2007, the cost

of sequencing a whole genome was estimated to be around US$10million , which by

late 2015 had fallen below US$1,50014. Large public and private funding programmes

have the potential to drive down costs further by sequencing genomes at high volume

and generating substantial economies of scale15. Consequently, the sequencing cost of

whole-genome sequencing is on a trajectory to break the $1,000 barrier in the very near

future16. However, most cost estimates within the literature (including those produced by

the NHGRI for whole-genome sequencing) tend to focus largely on the procurement

and resources used to run platforms but do not factor in downstream analysis and

interpretation of data that would be needed for use in routine clinical practice17,18,19.

Hence, the currently available cost estimates of whole-genome sequencing are of

limited use to decision-makers (Figure 1), who require a broader understanding of the

costs of introducing genomic technologies into clinical practice.

Importantly, the ‘cost’ of a technology differs from the ‘price’ advertised to payers in a

commercial setting. Prices are likely to be marked up for profit or to cross-subsidise

other tests on offer. Websites such as Genohub [https://genohub.com/] provide a useful

source of prices from a global network of laboratories offering next-generation

sequencing-based tests, including whole-genome sequencing (price range $1,105–

3

$16,420; prices as of October 2017) whole-exome sequencing (price range $534–

$7,637) and gene panels (price range $400–$5,800). The cost for a complete

‘sequencing service’, includes DNA preparation and is dependent on the number of

samples required; sequencing coverage and depth; number of reads; platform used;

general laboratory costs, such as overheads, skill-mix (for example, salary costs for the

relevant grade of molecular scientist or bio-informatician); data analysis and

interpretation and capacity to deliver sufficient number of tests.

Budget impact of genomic testsThe ‘budget impact’ of introducing genomic tests into a healthcare system, in terms of

knowing the direct healthcare resources required to implement widespread testing and

then potentially changing treatment, is harder to quantify than the cost of performing

sequencing and, to date, no accurate published figures exist for the total budget impact

of introducing genetic or genomic tests into practice. Budget impact estimates are a

function of the expected current and future eligible population size, which is measured

using expected prevalence or incidence rates of the condition of interest, multiplied by

the acquisition cost (if tests are offered at cost) or price of the technology used to treat

and/or diagnose20. However, population size will change with a test that improves

diagnostic yield, and tests may reveal incidental findings that can stimulate additional

healthcare costs19. Moreover, it is important to distinguish between the genomic

technology used to perform the sequencing (such as a gene panel or whole-exome

sequencing), the suitability and use of the technology as a diagnostic (that is, its

capacity to provide an actionable test result) and the mechanism used to offer the

relevant patient population the diagnostic service (model of service delivery)21. Even at

a seemingly simplistic level, the actual cost of providing services to deliver genomic

tests is not well documented. For example, in England, the cost of providing genomic

medicine is embedded within the specialist commissioned services22. However, some

evidence is starting to emerge for some example tests. The budget impact of

introducing whole-exome sequencing to the front of the diagnostic pathway for patients

with a high probability of having a rare inherited disease who attend special genetic

4

services in the UK, was estimated to lead to an increase pathway cost of £939 per

patient (or 52.5% increase) when compared with the usual testing strategy23.

Even if the population size is known with certainty, a fundamental challenge for

estimating the budget impact of genetic and genomic tests is the lack of a transparent

national tariff in some countries, and the use of charges, such as a flat rate per gene,

rather than the cost of delivering testing in terms of use of healthcare resources, in

others24.

Opportunity cost and economic evaluationAllocating resources for one intervention for a specified population means that those

same resources can no longer be used elsewhere for the care of other (known or

unknown) patient populations. Economic evaluation provides an explicit framework to

estimate the ‘opportunity cost’ of introducing a new intervention by comparing relevant

alternative uses of the healthcare budget in terms of their costs and consequences25.

Existing evaluation frameworks for genetic tests for rare inherited disorders26,27 are not

consistent with methods used in the evaluation of the economic impact of other

healthcare interventions. For example, medicines such as erlotinib for people with

locally advanced or metastatic non-small-cell lung cancer confirmed to be epidermal

growth factor receptor tyrosine kinase (EGFR-TK) mutation-positive, or diagnostic tests

such as testing EGFR–TK mutations in the tumors of adults with untreated NSCLC, are

assessed within the technology appraisal programme and the diagnostic assessment

programme, respectively, as part of national decision-making frameworks employed by

the National Institute for Health and Care Excellence (NICE) (Figure 1). NICE advocate

consideration of the opportunity cost for the healthcare budget and the use of

appropriate methods of economic evaluation to quantify the impact of a new technology

on healthcare costs and population outcomes28,29. In this context, two key tenets must

be remembered when using economic evaluation. First, the results should inform

resource allocation decisions at the population level rather than individual clinician–

patient decisions. Second, the results should be used as a source of information

towards evidence-based practice and cannot be seen as a substitute for careful

5

interpretation and appraisal by regional and local decision-makers charged with making

resource allocation decisions within their own jurisdiction.

Different types of economic evaluation exist (Table 2)30-39. However, in terms of

application in practice, broadly two types of economic evaluation are used to quantify

the costs and consequences of a new intervention: cost-effectiveness analysis (CEA;

sometimes interchanged with the term ‘cost-utility analysis’ (CUA)) and cost-benefit

analysis (CBA). If economic evaluation is used to make ‘ought’ statements about what

decision-makers should do, it must have some underlying value judgements or

normative (‘what ought to be’) principles40,41,42. The normative underpinnings of CEA and

CBA are extra-welfarism and welfarism, respectively43, 44. To date the practical

application of extra-welfarism, such as in the context of NICE appraisal programmes,

supports the premise that decision-makers want to maximize health, which is measured

using quality-adjusted life-years (QALYs). Therefore, globally, CEAs have become the

type of economic evaluation recommended most frequently in guidelines to inform

resource allocation and/or reimbursement decisions45.

Trial-based versus model-based CEATwo approaches to produce the data to conduct a CEA exist: trial-based, whereby data

for decision-making are typically and almost exclusively collected from within a single

randomized controlled trial (RCT), or model-based, whereby data are collected from

multiple diverse sources and synthesized within an economic model.

Trial-based CEA

The timeliness for generating evidence from trial-based studies that collect cost and

consequences data at the individual patient-level can limit their use for decision-making

in practice46. Trial-based studies are also resource-intensive, are relevant only to the

trial protocol and have challenges when generalizing to other jurisdictions or patient

populations47. Using a trial-based CEA poses a specific practical challenge in the

context of genomic diagnostic tests. A genomic test must be embedded within a service

delivery model, which is likely to preclude the practical implementation of a

randomization process necessary for an RCT comparing diagnostic tests using next-

6

generation sequencing technologies with current diagnostic strategies. This approach is

further complicated by the prohibitively large research budget needed to follow up

patient outcomes and resource use over a sufficient time horizon to produce relevant

evidence for decision-makers.

Model-based CEA

Using a model-based CEA allows all interventions and relevant comparators

(comparators are defined as ‘relevant’ by the stated decision problem), such as current

clinical practice, to be included in the analysis using methods such as, for example,

network meta-analysis47. Results can be extrapolated to capture the impact of an

intervention over the lifetime of a cohort of patients and also quantify the impact of

uncertainty48,49,50. Uncertainty in a model-based study can arise for a number of reasons:

a methodological uncertainty of how to value costs and benefits that will occur in the

future using the appropriate discount rate; a structural uncertainty about which

diagnostic and treatment pathways should be represented in the model structure; a

parameter uncertainty about the ‘true’ value of each input parameter used within the

model50. Sensitivity analysis is, therefore, a key component of any model-based

economic evaluation, and methods such as probabilistic sensitivity analysis to measure

the impact of parameter uncertainty are viewed as prerequisites to a robust analysis 51,52.

Combined, the ability to represent many comparators, include the impact of uncertainty

and extrapolate to a long time horizon, have all contributed to decision-analytic model

based CEA globally becoming the preferred approach to quantify the incremental costs

and consequences of a new intervention27,28,53,54.

A key challenge for model-based CEA for genetic and genomic tests, and more broadly

for any diagnostic test, is whether evidence exists that spans the three component parts

of the intervention simultaneously: (1) the technology; (2) the diagnostic component; (3)

the technology within the health service delivery framework21. NICE describe such

evidence as being ‘end-to-end’, that is, starting from performing the original test all the

way to patient outcomes and costs29. The only known example of a published end-to-

end study in the context of genetic testing used a pragmatic trial design 55 with integrated

CEA56 to assess the use of genotyping tests for single-gene mutations (versus no

7

genotyping) when treating patients with azathioprine, and demonstrated the problem of

simply assuming that genetic information will mean clinicians appropriately modify their

treatment decisions. Where such studies are not available, assumptions must be made

as to how the component parts of the diagnostic technology interact, such as how

different diagnosis patterns from the new technology lead to changes in treatment

choices (and subsequent benefits and costs) when the set of two studies are connected

within an economic model.

Model-based CEA of genomic tests The key steps in a model-based CEA of a genetic or genomic test broadly mirror those

for diagnostic tests (Figure 2)25, 57-64. To ensure the resulting model-based CEA is

relevant for decision-making, the fundamental step is to explicitly define the decision

problem, underpinned by appropriate normative principles. That is, the following aspects

must be defined: the study perspective, the ‘time horizon’, the relevant study population,

the intervention to be evaluated and its relevant comparators.

The study perspective should be informed by the end user of the results of the CEA. If

the decision-maker (Figure 1) is based within a hospital then the relevant perspective

will be defined by the hospital budget. By contrast, a national level decision-maker

would need a study perspective that included all healthcare resource use, consistent

with informing how to spend the national healthcare budget. The study perspective is

most often constrained to the healthcare system because CEA are generally used to

inform the cost-effective use of a national budget for healthcare28,29. However, guidelines

used by some jurisdictions, such as The Netherlands, suggest broadening the viewpoint

to use a societal perspective, which would then mean patient costs are included in

addition to healthcare costs. The decision to use the societal perspective in The

Netherlands is based on a value judgement that costs outside of the healthcare sector

ought to be considered45.

The time horizon for a study should be sufficient to allow all relevant costs and

consequences to be incorporated which, in general, requires the lifetime time horizon

for the defined patient cohort to be used23. For example, a lifetime horizon would

capture the lifelong costs and consequences of using a test to target the use of a

8

medicine to treat cancer. Importantly, defining the decision problem also involves having

a clear understanding of the relevant patient population that will accrue the benefits

(and harms) of the intervention. However, defining the relevant population can be

problematic for genomic tests. For example, when considering the added value of

whole-genome sequencing, the relevant patient population may extend beyond the

index patient first tested and include the benefits (and harms) to family members. In this

instance, the decision problem could be specified to take account of externalities that

capture the additional impact of genomic testing on the family (sometimes referred to as

‘spillover’ effects)65. Including spillover effects introduces another problem that

compounds how to conceptually define a ‘lifetime’ as it is no longer readily apparent

which lifetime for which patient/s is appropriate and the potential value of the

intervention may span generations13.

Defining the decision problem requires an explicit statement of the intervention and the

relevant comparators, which may include strategies beyond the test itself. For example,

companion diagnostics (used in ‘test-treat’ strategies) may be compared with other

medicines that are prescribed without a test. Different tests may also be considered as

relevant comparators. For example, a range of technologies can be used to assess

EGFR-TK mutations for patients with NSCLC who are treated with different targeted

medicines (such as the tyrosine kinase inhibitors gefitinib or erlotinib), depending on the

identified mutation66. Numerous diagnostic applications of genetic and genomic tests

also exist, as listed in the UK Genetic Testing Network database

[https://ukgtn.nhs.uk/find-a-test/], and are emerging for disorders for which there are

currently no available treatments, as is often the case with rare inherited conditions. In

these cases, and specifically for a genome-based diagnostic test, where incidental

findings are possible18, there is a need to be explicit when defining the intervention

under evaluation about how (and whether) such findings will be reported to the clinician

ordering the test and the affected patient. Furthermore, the approach used to interpret

incidental findings needs to be known and described. Multi-disciplinary committees

comprising groups of specialist geneticists, counsellors and molecular scientists are

becoming common practice as an approach to decide on the reported test result in

9

some, but not all, clinical settings where next-generation sequencing technologies have

been used to produce a diagnostic test18.

Challenges of CEAs for genomic testsThe challenges of conducting model-based CEA for genetic or genomic tests can be

grouped into (a) whether a known treatment is available, such as in the case of a

companion diagnostic (linked test-treat strategies) or where subsequent preventative

strategies may be offered (for example, in the case of inherited cancers), and (b) if no

treatment options are currently known, such as in the case of diagnostic tests for some

rare inherited conditions (for example, in the case of hereditary ataxias, a group of rare,

neurodegenerative diseases)12.

Current economic evaluations of companion diagnostic tests tend to focus on the value

of the treatment or preventative strategy rather than quantifying the value of the

diagnostic test per se67. Some evaluation programmes have been developed specifically

to assess the added value of diagnostic tests; however, these currently still focus on

quantifying the value of the suggested treatment or preventative strategy in the CEA 29.

Box 1 shows an example of when a model-based CEA was used to assess the added

value of different companion diagnostic tests that were being used in clinical practice.

The challenges associated with conducting CEA of genomic-targeted ‘companion’

diagnostic tests have been well described67. They include, for example, the need for

clarity about whether the companion diagnostic test is designed to detect somatic or

germline mutations, and the careful framing of the study question and relevant clinical

pathways. It is also important to have analytical strategies in place to deal with the lack

of the necessary key requirements for a robust economic evaluation such as data on:

the clinical effectiveness; clinical utility; impact on health status and resource use; and

uptake of the test in practice.

By contrast, model-based CEA has not yet become routine to inform the introduction of

genetic or genomic tests for disorders with no available treatments, such as rare

inherited disorders that affect fewer than 1 in 2,000 individuals. This is likely to be the

result of the many notable challenges associated with using model-based CEA in this

context12,18,68-71. Although economic evaluations of genetic and genomic tests for rare

10

diseases have been published (see Supplementary information S1), model-based CEA

has yet to be used to inform the actual introduction of these diagnostic tests into clinical

practice. A summary of the notable challenges, when conceptualizing, designing and

using the results of model-based CEA for genomic tests, specifically when no treatment

options are available, will now be described and categorized according to four types of

challenge: methodological, technical, practical and organizational (Table 3).

Methodological challenges

Relevant methods must be selected to address the defined decision problem while

remaining consistent with the agreed normative framework. Table 3 outlines some key

methodological challenges and three of these are now described. A review conducted in

2015 explored the impact of using CBA (underpinned by welfarism) or CEA

(underpinned by extra-welfarism) on technology adoption decisions72. This study found

that the two methods often led to different recommendations on which health care

technology to adopt. The authors concluded that health economists do not currently

provide the users of economic evaluations with sufficient guidance about which method

(CBA or CEA) should be preferred, which should be the focus of future work.

Some commentators have suggested extending economic evaluations beyond health

status to include ‘personal utility’ to value the benefits of genetic or genomic tests.

Personal utility has been used to refer to the potential value resulting from the genomic

information derived from a test in its own right (positive or negative) regardless of

whether treatment decisions and the health of the population are affected by the test

result73-76. To date, however, no measure exists for personal utility to be used in an

economic evaluation. One published study took an empirical approach to understand

whether health status alone was sufficient to capture the consequences of genetic tests

and services77. The study suggested that the consequences could be captured using a

concept related to the capability to make an informed decision (in the study referred to

as ‘empowerment’). This conclusion should also be applicable to the evaluation of

genomic tests. However, the study did not produce an instrument that could measure

‘capability to make an informed decision’. The authors suggest, appropriately, that

further empirical research is required to understand the impact on opportunity cost if

11

non-health-related consequences are to be included as an additional factor to be

considered in CEA. Instruments have been developed that allow a broader

interpretation of extra-welfarism, such as the suite of measures of capability called

ICECAP (ICEpop CAPability measure)78,79, which capture the impact on well-being

(interpreted as ‘the ability to function’). NICE guidance currently recommends the

ICECAP measures as an option in the economic evaluation of social care

interventions78. However, using such measures of capability in a CEA does not currently

yield meaningful ways of presenting the results to decision-makers that are equivalent

to the presentation of QALY-based analyses, although work in this area is ongoing1,82.

Another methodological challenge being addressed is the potential need to move away

from estimating average costs and benefits for a population, and take account of the

relative cost-effectiveness at sub-group or the individual patient level when calculating

the relative cost-effectiveness of genomic-based interventions. Analytical approaches

calculating the value of heterogeneity (in, for example, patient or disease

characteristics) or value of preference heterogeneity use new theoretical frameworks

developed to assess the potential value of identifying cost-effective treatments for

defined sub-groups of patients, or individual patients given their preferences,

respectively. These approaches can in theory, therefore, compare the value of

‘individualized’ treatment decisions with the value of treatment decisions based on

traditional population-level cost-effectiveness analysis but, to date, there are no

published examples in the area of genomic tests.83,84

Technical challenges

Technical challenges refer to the need to account for the complexity of the decision

problem in the model-based CEA in terms of the characteristics of the relevant study

population, the number of relevant comparators and the potential diagnostic and, if

appropriate, subsequent treatment pathways. Extensive model conceptualization

techniques may be valuable to ensure that the structure of a de novo decision analytic

model is relevant to the prevailing decision problem (Figure 2)57. The most commonly

used models in published CEAs are state transition Markov models, which use a cohort-

based approach, sometimes combined with decision trees85,86. In theory, these types of

12

models can cope with any number of relevant comparators but, in practice, they are

generally limited to comparing two or three interventions. One published example, a

CEA of stratified breast screening programmes87, used optimization methods to inform

how to reduce the number of comparators by selecting only the optimal programmes for

inclusion in the analysis. An optimal programme was defined by solving the knapsack

problem88. The knapsack problem defines the decision problem in terms of how to ‘pack’

a set of items (individual screens in the breast screening programme) of defined value

(total number of QALYs) into a knapsack (the screening programme) that can hold a

fixed total amount (total screens in a lifetime) in such a way that the value of items

‘packed’ is maximized. Using optimization methods has advantages over current

methods that consider a large number of scenario analyses within a CEA to represent

the many potential alternative screening programmes89,90. The inherent characteristics of

cohort state transition Markov models mean they may not be sufficient to capture the

impact of comparing multiple interventions together with the extensive individual patient

variation in terms of underlying genotype and presenting phenotype58. For example,

when comparing a new genomic test, such as a multiple (105) gene panel test for the

diagnosis of inherent retinal dystrophies, with the many potential existing diagnostic

strategies that use Sanger sequencing of individual genes and existing non-genetic

diagnostic strategies requires a model to include many potential comparisons and

account for individual patient variation91.

More relevant but potentially complex models, such as discrete event simulation (DES),

can be used to reflect the multitude of care pathways, heterogeneity in patient-level

characteristics and time-related (temporal) health events that occur in the progression of

disease or the diagnosis and treatment pathway 59,60,92. Such models allow the health

experiences of individual patients within the model to depend on their characteristics

(such as age or gender) or other relevant individual-level factors (such as genotype or

phenotype). Using DES also facilitates the impact of a key stated advantage of genomic

tests (offering patients the opportunity of an earlier diagnosis compared with existing

strategies) to be estimated and assess whether the expected positive impact is

realized22 and personal communication Niall Davison, The University of Manchester).

However, the use of DES in this context remains unpublished93.

13

The majority of current applications of model-based CEA assume that all processes of

providing care and associated items of resource use are available with no limits,

regardless of the volume of the intervention required58. Effectively, there are no

‘constraints’ on the capacity of the system to provide healthcare services. The

assumption of infinite capacity may not be the case for a variety of reasons; there are

many potential types of constraints that are internal to a healthcare-system, such as

lack of equipment, data storage or healthcare workforce. Such ‘capacity constraints’ will

affect the ability of a healthcare system to reach its stated goal, such as the provision of

healthcare to realize patient benefits. In reality, all systems, including healthcare

systems, resemble networks of chains, in which each event is dependent on the

preceding event94. The discipline of mathematical optimization offers programming

techniques to select the best combination of inputs from a set of available alternatives,

to maximize output (for example, health gain) while taking into account capacity

constraints95,96. There is emerging interest in the combined use of decision analytic

modelling with mathematical optimization techniques to consider whether, and how,

model-based CEA of healthcare interventions can consider the impact of capacity

constraints when introducing a genetic or genomic diagnostic test into practice.

Practical challenges

Even if the methodological and technical challenges have been addressed, practical

challenges regarding research design may arise. A practical challenge associated with

the use of model-based CEA is the lack of the required robust data to populate the

model input parameters for outcomes, probabilities and resource use, in general, and

end-to-end evidence, specifically. New genomic sequencing technologies do not have a

defined development and regulatory pipeline that subsequently drives the available

evidence base for use in model-based economic evaluations. Key parameters, such as

the prevalence of the mutation and disease and the predictive value of the genomic test

are often not informed by a robust evidence base, meaning there is a need to rely on

information gained from further down the hierarchy of evidence97 to populate the model-

based CEA.

14

A further practical challenge that compounds the lack of accurate resource use data is

the absence of a transparent national pricing tariff for the genomic test 24. One

systematic review indicated the impact of this challenge when it aimed to identify

economic evaluations of genomic sequencing technologies but found only five studies

purportedly conducting cost analyses of sequencing technologies with insufficiently

reported detail on the actual methods used to identify resource use and attach unit

costs98. There are emerging examples of using micro-costing methods to understand

the cost of tests, such as the approach taken to identify the cost of an anti-drug antibody

and drug level test to target the use of adalimumab for people with rheumatoid

arthritis99. The lack of a nationally agreed unit cost effectively means that an analyst is

trying to perform a CEA of a technology for which the acquisition cost is not known. In

practice, the analysis can still be done using an estimated unit cost for the test, but the

results of a CEA may then not be useful to decision-makers, who will not know whether

the results apply in their setting.

Organizational challenges

Although mentioned last, organizational challenges present the most substantial hurdle

for decision-makers who want to use the results of model-based CEA of genomic tests

to inform treatment strategies or for diagnosing rare inherited diseases. Organizational

challenges refer to the apparent disconnect between the organization of healthcare

systems and funding streams. These differences mean that a decision-maker must

understand whether and how the results of a model based CEA are relevant to their

own setting based on the scope of the analysis as defined by the decision problem.

Both within and across countries, healthcare systems have different service provision

and staffing models for different sectors (community, general practice, hospital,

specialist) and a means of allocating funding to these different sectors. The existence of

‘silo’ budgets, inextricably linked to the silo approach to the management of disease,

creates another specific challenge when using the results of CEAs. Methods of

economic evaluation assume that money can be freely moved around a healthcare

system so that patient benefits can be maximized by funding the most efficient service

as identified by the results of the analysis. In practice, this is not the case, and this

15

organizational issue is particularly relevant for specialized services such as genomic

medicine for people with complex diseases, who require access to many types of

services across the healthcare system. The cost of a genomic-targeted diagnostic test

may, for example, fall to the clinician ordering the test from cardiology, as problems with

the patient’s heart were the presenting symptom, even if it emerges that the patient has

a syndromic condition. The results of a model-based CEA may show that achieving a

diagnosis can realize benefits across disparate and unconnected areas of the

healthcare service. However, a decision-maker may be in charge of one distinct budget,

such as for cardiology, and may not appreciate other benefits that are accrued.

On a global scale, genomics and associated Health Technology Assessment (HTA) is

the dominion of the prosperous and, perhaps unsurprisingly, focused within high-income

countries with relatively large healthcare budgets and formalized models of service

delivery. Some approaches have attempted to overcome global disparities in HTA. For

example, NICE International provides practical support and advice to any governments

and funding agencies about the process and methods of HTA in healthcare policy

decisions. The premise was that most countries could not afford to set up an equivalent

system to NICE owing to a lack of funds and expertise. In 2013, the Bill and Melinda

Gates Foundation funded a collaboration led by NICE International to develop the Gates

Reference Case, which outlined eleven principles to ensure economic evaluations

provide robust evidence for global health care decision-making100,101. NICE International

has now been replaced by the Global Health and Development Group

[https://www.imperial.ac.uk/global-health-innovation/ ] to continue to provide advice to

contribute to the effective, equitable use of resources and better global health.

Realizing the benefits of genomic tests?There is substantial but disparate evidence to support the economic impact of genetic

tests and precision medicine more broadly, but limited examples exist in the area of

genomics. Precision medicine includes approaches that use genetics but also other

influential factors, such as the environment and patient characteristics, to identify the

most suitable treatment or preventative strategy for an individual102. Some 45 systematic

reviews have assimilated economic evaluations in the general area of precision

16

medicine and related areas, covering ‘test-treat’ strategies (of which, some used non-

genomic strategies) and genomic tests used for diagnosis, generally for single gene

disorders (see Supplementary information S1). Until 2016, there were no published

economic evaluations of new sequencing technologies, although examples of CEA,

some using diagnostic yield as the measure of effectiveness and some using QALYs,

are now emerging that evaluate the use of next-generation sequencing in the context of

gene panels, for exome sequencing and three hypothetical applications

(cardiomyopathy, colorectal cancer, healthy individuals) of returning incidental

findings23,103-108. This collective evidence base shows that the greatest focus is directed

towards evaluations in developed countries and that it is difficult to reach a collective

definitive view on whether healthcare systems can realize the potential benefits from

genomic tests. Collating this literature illustrates a substantial challenge for decision-

analysts and decision-makers who want to have an overall view of the supporting

evidence base on the economics of genomic tests: inconsistency in terminology. There

is no agreed or standardized use of terms within published papers, or electronic

databases, relevant to genomic-targeted diagnostic tests such as: precision medicine;

stratified medicine; personalized medicine; individualized medicine; targeted medicine;

pharmacogenetics; pharmacogenomics109. The early stages of any health technology

assessment or the conceptualization and development (Figure 2) of a decision analytic

model is to collate all available economic evidence, in the form of published economic

evaluations, to understand the current knowledge base on a topic and to identify

whether any existing decision analytic models can be used as the basis for a proposed

economic evaluation28,28. Examining the disparate nature of the results from 45

systematic reviews (see Supplementary information S1) reveals that it is virtually

impossible to identify all relevant evidence from one review given the inconsistency in

nomenclature. If proponents of the use of genomic diagnostic tests want to realize their

benefits in practice, an important first step will be to agree a set of consistent terms and

definitions. However, reaching agreement about the appropriate terms and definitions

across the broad community of stakeholders, which include clinicians, patients,

industrial partners who manufacture medicines and tests, and national assessment

17

agencies, will require a concerted effort by internationally recognized experts using

appropriate consensus methods.

ConclusionsDriven by the development of new sequencing technologies and national research

agendas, there will be an increase in the number of available genomic tests in the

developed world. Evaluation frameworks for genetic tests exist but it is reasonable to

suggest that these may not be sufficient given the potential new demand for genomics

within existing models of service delivery and beyond into mainstream medicine. Given

the finite nature of healthcare budgets, decision-makers need to understand the

opportunity cost of introducing new genomic diagnostic tests. Model-based CEA are a

useful tool embedded within health technology assessment programmes to provide

explicit information on the incremental costs and consequences of new interventions. To

inform decisions about efficient allocation of healthcare resources, model-based CEAs

must be supported and populated by robust studies, which ideally span the three

component parts of a genomic test: (1) the technology; (2) the diagnostic component;

(3) the model of service delivery to provide the diagnostic test and, if appropriate,

subsequent treatment options. The area of genetics has an emerging economic

evidence base. National research agencies have appropriately started funding the

development of new technologies to inform new genomic tests that may replace existing

genetic tests. It is now time to direct funding to support the empirical research needed to

develop the use of decision-analytic model based CEAs of genomic tests, while being

cognizant of the known methodological, technical, practical and organizational

challenges, to maximize the potential benefits to patient populations.

Box 1 | Example of using model-based CEA to inform clinical practice. This Box

shows an illustrative example when model-based cost effectiveness analysis was used

to inform the use of a test in practice.

Context National Institute for Health and Care Excellence Diagnostics Guidance 9 [https://www.nice.org.uk/guidance/dg9] published August 2013

18

Decision problem: What is the relative effectiveness and cost-effectiveness of epidermal growth factor receptor tyrosine kinase (EGFR–TK) mutation testing in patients with locally advanced or metastatic non-small cell lung cancer?

Technology: Ten ways of testing for EGFR mutations

Method of analysis: Model-based cost effectiveness analysis using a decision tree combined with a state transition Markov model

Costs: 2,012 (£)

Outcomes: Quality-adjusted life years

Decision rule: The NICE threshold of £20,000 to £30,000 per QALY gained

Recommendation: Five tests were recommended as options for detecting EGFR-TK mutations, when used in accredited laboratories participating in an external quality assurance scheme:

therascreen EGFR RGQ PCR Kit (CE-marked, Qiagen)

cobas EGFR Mutation Test (CE-marked, Roche Molecular Systems)

Sanger sequencing of samples with more than 30% tumour cells and therascreen EGFR RGQ PCR Kit for samples with lower tumour cell contents

Sanger sequencing of samples with more than 30% tumour cells and cobas EGFR Mutation Test for samples with lower tumour cell contents

Sanger sequencing followed by fragment length analysis and polymerase chain reaction (PCR) of negative samples.

There was insufficient evidence for the Committee to make recommendations on the following five methods:

high-resolution melt analysis

pyrosequencing combined with fragment length analysis

single-strand conformation polymorphism analysis

19

next-generation sequencing

therascreen EGFR Pyro Kit (CE-marked, Qiagen)

Figure 1 | Examples of decision-makers. This figures shows some illustrative

examples of the types of decision-makers who work within different levels of the system

for providing healthcare. This figure only shows some examples and is not a

comprehensive description of all the decision-makers within the healthcare system for

the UK.

Figure 2 | Key elements in the design and conduct of decision analytic model-based CEA. This flowchart should be read from the top to the bottom and each box

explains the relevant steps in the conduct of a model-based CEA with references

illustrating the key considerations for each step.

20

Table 1 | Example types of genetic and genomic tests Example test Type of

test11

Technology

Relevant population

Example source

Tests to inform treatment or for diagnosing disorders with known treatments

HLA-B 5701

genotype,

abacavir

hypersensitivity

, blood

genotyping

test for

known

disease or

treatment-

associated

variant

Qualitative

allele-

specific

Real-time

PCR

Individuals

with an

increased risk

of

hypersensitivit

y reactions to

the

antiretroviral

agent

abacavir,

based on the

presence of

the HLA-

B*57:01 allele

Mayo Clinic

[https://www.mayomedicallaboratories.com/

test-catalog/Overview/89346]

EGFR mutation

testing in

NSCLC

genotyping

test for

known

disease or

treatment-

associated

variant

Sanger

sequencing

Adult patients

with previously

untreated,

locally

advanced or

metastatic

NSCLC

Manchester Centre for Genomic Medicine

[http://www.mangen.co.uk/lab-services/view-

details.php?tag=55]

Testing to

identify at least

one G551D

mutation in the

CFTR gene to

indicate

treatment with

ivacaftor in

patients with

cystic fibrosis

genotyping

test for

known

disease or

treatment-

associated

variant

PCR-based

multiplex

assay

Test used to in

people with

CF to identify

those with a

specific

genetic

mutation

CFTR Elucigene Eu2v1

(http://www.elucigene.com/)

ivacaftor for the management of cystic fibrosis.

Formulary Submission Dossier.

Cambridge, MA:

Vertex Pharmaceuticals Inc. (2012)

Diagnostic test for disorders with no known treatment

Targeted

mutation

analysis for

Huntington

genotyping

test for

known

disease or

Sanger

sequencing

Patient and

family

members with

rare inherited

UK Genetic Testing Network

[https://ukgtn.nhs.uk/find-a-test/search-by-

disorder-gene/huntington-disease-203/]

21

disease treatment-

associated

variant

disorders

Retinal

degeneration

disorders 105

gene panel

disease-

targeted

panel tests

for genes

known to be

associated

with the

disorder or

treatment

Next-

generation

sequencing

gene panel

Patient and

family

members with

rare inherited

disorders

UK Genetic Testing Network

[https://ukgtn.nhs.uk/find-a-test/search-by-

disorder-gene/retinal-degeneration-disorders-

105-gene-panel-568/]

Epilepsy

disorders 110

gene exome

panel

exome

sequencing

test, which

interrogates

nearly all

protein-

coding

regions

Whole-

exome

sequencing

Patient and

family

members with

rare inherited

disorders

UK Genetic Testing Network

[https://ukgtn.nhs.uk/find-a-test/search-by-

disorder-gene/epilepsy-disorders-110-gene-

exome-panel-879/]

Whole genome

sequencing for

the diagnosis

of rare

inherited

disorders

(research

context only)

genome

sequencing

test, which

interrogates

nearly the

entire genetic

code

including

non-coding

regions

Whole-

genome

sequencing

Patient and

selected family

members with

rare inherited

disorders

(research

context only)

100,000 Genomes project

[https://www.genomicsengland.co.uk/the-

100000-genomes-project/]

There is no accepted definition of how a genetic test differs to a genomic test. For the purpose of this

paper we use the following definitions: genetic tests aim to identify germline or somatic mutations (or

pathogenic variants) that underlie high-risk, single-gene disorders, whereas genomic tests search for

germline or somatic mutations (or pathogenic or modifying variants) in coding and non-coding DNA

across the whole genome. Genomic tests are often applied to heterogeneous conditions, polygenic

conditions or conditions with unknown genetic causes, including conditions where epigenetic factors may

play a role in disease risk. CFTR, cystic fibrosis transmembrane conductance regulator; EGFR, epidermal

growth factor receptor; HLA, human leukocyte antigen; NSCLC, non-small-cell lung cancer.

22

Table 2 | Types of economic evaluationEconomic evaluation (acronym)a

Normative underpinning

Consequences Presentation of results

Example decision rule

Application to inform decision-making

Generally recommended for evaluation of healthcare programmes?

Cost-

effectivenes

s analysis

(CEA)b

Extra-

welfarism

Clinical

effectiveness

eg.

Life years

gained (LYG)

calculated from

the proportion of

treatment

responders

measured using

a clinical

outcome

Calculate

incremental

costs and

incremental

benefits

and, if

appropriate

ICERc

ICER <

pre-

defined

threshold

eg.

£20,000

per LYG

Quantifies impact on

health using a

measure of eg. life-

years gained within

for example NICE

Technology

Appraisals or

Diagnostics

Assessment

programmes

[https://www.nice.org

.uk/]

Yes as data

on

effectiveness

often

available

from trials

Cost-utility

analysis

(CUA) b

Extra-

welfarism

Quality adjusted

life years

(QALYs)

calculated by

combining

remaining years

of life with a

quality

adjustment to

reflect health

status from

using a generic

measure (eg.

https://euroqol.o

rg/) and

published

preference

weights30

Incremental

costs and

incremental

benefits

and, if

appropriate

ICERc

ICER <

pre-

defined

threshold

eg.

£20,000

per QALY

Incremen

tal net

benefit

(INB)31

Quantifies the impact

on health using eg.

QALYs within for

example NICE

Technology

Appraisals or

Diagnostics

Assessment

programmes

Yes as can

be used to

compare

across

programmes

23

Cost-benefit

analysis

(CBA)

Welfarism Monetary

benefits

estimated from

a contingent

valuation

method eg.

Willingness to

pay survey32

Total costs

and

monetary

benefits for

each

intervention

Benefit

(£) / cost

(£) for

interventi

on A >

Benefit

(£) / cost

(£) for

interventi

on B

Benefit

(£) > cost

(£)

Quantifies impact of

non-healthcare

public programmes33

and suggested as an

option within NICE

public health

programme to allow

consequences other

than health status to

be measured

[https://www.nice.org

.uk/]

No because

of

methodologic

al and ethical

challenges34

Cost-

minimizatio

n analysis

(CMA)

None None as

benefits

assumed to be

equal

Total costs

for each

intervention

Cost for

interventi

on A <

cost for

interventi

on B

Pragmatic approach

advocated when

differences in benefit

between

technologies can be

assumed to be zero

or when the

differences in cost so

large that any

difference in benefits

unlikely to change

the decision

No because

of inherent

uncertainty

when

estimates of

benefits and

costs are

combined

comparing

costs alone

can produce

biased

results35,36

Cost-

consequenc

es analysis

(CCA)

None A suite of

consequences

from pre-

defined

outcomes that

can include

measures of:

clinical

effectiveness;

health status;

capability; non-

health or

process

measures (eg.

Total costs

for each

intervention

and total

consequenc

es (for each

included

outcome)

A suite of

ICERsc for

each

consequenc

e, if

None Pragmatic approach

that leaves decision-

makers to decide on

a case-by-case basis

how trade-offs

should be made

within NICE public

health or Medical

Technology

Evaluation

Programmes

[https://www.nice.org

.uk/]

No because

opportunity

cost not

consistently

considered

across

programme

s38,39

24

quality of a

genetic

counselling

episode)36

relevant,

can be

presented

a same general approach to quantify the costs of each relevant technology by identifying

the relevant types of resources. Relevance in this context is defined by the chosen

study perspective and time-horizon, and using unit costs for each resource to calculate

the total cost.b CEA is, in some situations, such as within the NICE programmes, used instead of the

perhaps more accurate term CUA, which is what NICE is actually advocating as they

suggest quality adjusted life years (QALYs). This paper adopts this approach.c Incremental Cost Effectiveness Ratio (ICER) = difference in costs (eg. Cost

intervention A – Cost intervention B) / difference in effectiveness (eg. Life-years gained

for intervention A – life-years gained for intervention B)

25

Table 3 | Typology of example challenges for CEA of genomic testsType of challenge Example of challenge Description of challenge

Methodological Selecting the

appropriate evaluative

framework

Is the standard extra-welfarist view and use of CEA appropriate or

should the distinct theoretical approach reflecting the welfarist view

and use of CBA be adopted to allow consequences other than

health gain, such as the value of diagnostic information from the

genomic-targeted diagnostic test, to be valued?

Relevant study

perspective

Is the standard recommendation to focus on the use of healthcare

services appropriate when the genomic-targeted diagnostic test

may provide information that affects the use of other services such

as education or employment?

Relevant time horizon Is a lifetime sufficient when the impact of a genomic-targeted

diagnostic test may extend to infinite time horizons that are not

limited by the lifespan of one individual?

Defining the relevant

study population

Is the standard definition of a patient (the person receiving the

technology) appropriate when there could be ‘spillover’ effects to

family members (currently alive or to be born) as a result of

information from a genomic-targeted diagnostic test?

Valuing consequences Is identifying and measuring the impact on health status alone

sufficient to capture the (good and bad) consequences of a

genomic-targeted diagnostic test?

Technical Variation in the

individual characteristics

of the relevant study

population

The use of cohort state transition Markov models, sometimes

combined with decision trees, cannot easily capture the impact of

individual patient variation within a population with different

genotypes and phenotypes

Number of diagnostic,

and if appropriate,

subsequent treatment

pathways

The use of cohort state transition Markov models, sometimes

combined with decision trees, cannot easily account for multiple

comparators often needed when evaluating a new genomic-targeted

diagnostic test

Capture impact of

reduced time to

diagnosis

The use of cohort state transition Markov models, sometimes

combined with decision trees, cannot account for the impact of

reduced time to achieve a diagnosis, which is often a proposed

benefit of a genomic-targeted diagnostic test

Capture impact of

capacity constraints

Decision analytic model-based CEA currently assume limitless

capacity within healthcare systems, which is often not a reasonable

assumption when introducing a genomic-targeted diagnostic test to

populations for whom a diagnosis was not previously available

Practical Availability of data There is often a lack of data available to populate decision analytic

model-based CEA

National tariff of test No national tariff for genomic-targeted tests exist

26

cost

Organizational Complex healthcare

systems

Decision analytic model-based CEA assume that money saved, and

benefits accrued, are transferable but this is often challenging in

complex healthcare systems that comprise: an overarching funding

mechanism (public, private, insurance); service and staffing model

for providing care for different sectors (community, general practice,

hospital, specialist); and a means of allocating funding to these

different sectors

Generalizability of

results

Decision analytic model-based CEA are only relevant to the defined

decision problem and decision-makers who want to use the results

must decide if the focus of the analysis is relevant to their own

jurisdiction

Expensive nature of

health technology

assessment

Decision analytic model-based CEA conducted within national

health technology assessment processes requires significant

funding and expertise that is not available to all, which may

contribute to the inequity in access to new genomic-targeted

diagnostic tests across the world

27

References *highlighted references

1. Donnai, D. & Elles, R. Integrated regional genetic services: current and future provision.

BMJ  322, 1048 (2001) 

2. Godard, B., Kääriäinen, H., Kristoffersson, U., Tranebjaerg, L., Coviello, D., Aymé, S.

Provision of genetic services in Europe: current practices and issues. European Journal of

Human Genetics 11 (Suppl 2), S13–S48 (2003)

3. Venter, J.C., et al. The Sequence of the Human Genome. Science 40 (291), 1304 (2001)

4. Pohlhaus, J.R. & Cook-Deegan, R.M. Genomics research: world survey of public

funding. BMC Genomics 10(9), 472 (2008)

5. NIH National Human Genome Research Institute Genome Sequencing Program (GSP).

https://www.genome.gov/10001691/nhgri-genome-sequencing-program-gsp/.

6. Genomics platform at Astrazeneca. https://www.astrazeneca.com/media-centre/press-

releases/2016/AstraZeneca-launches-integrated-genomics-approach-to-transform-drug-

discovery-and-development-22042016.html.

7. Large-Scale Applied Research Project Competition – Genomics and Personalized

Health (2012) https://www.genomecanada.ca/en/programs/large-scale-science/past-

competitions/large-scale-research-project-competitions/2012-large

8. Large-Scale Applied Research Project Competition – Genomics and Precision Health

(2017) https://www.genomecanada.ca/en/request-applications-2017-large-scale-applied-

research-project-competition-genomics-and-precision

9. Médecine France génomique 2025.

http://www.gouvernement.fr/sites/default/files/document/document/

2016/06/22.06.2016_remise_du_rapport_dyves_levy_-_france_medecine_genomique_2025.pdf

10. Genomics England. https://www.genomicsengland.co.uk/the-100000-genomes-project/.

11. Rehm, H.L. Evolving healthcare through personal genomics. Nature Reviews Genetics

18, 259-267 (2017)

12. Daker-White, G., Ealing, J., Greenfield, J., Kingston, H., Sanders, C., Payne, K. Trouble

with ataxia: A longitudinal qualitative study of the diagnosis and medical management of a

group of rare, progressive neurological conditions. Sage Open Medicine 28(1), 2050312113505560 (2013)

13. Spackman, E., Hinde, S., Bojke,L., Payne, K., Sculpher, M. Using cost-effectiveness

analysis to quantify the value of genomic-based diagnostic tests: recommendations for practice

and research. Genetic Testing & Molecular Biomarkers (2017)

28

14. NIH National Human Genome Research Institute – The cost of sequencing a human

genome. https://www.genome.gov/sequencingcosts/

15. Generation Genome Annual Report of the Chief Medical Office (2016).

https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/631043/

CMO_annual_report_generation_genome.pdf

16. Service, R.F. The race for the $1000 genome. Science 311(5767), 1544-1546 (2006)

17. Mardis, E.R. The $1,000 genome, the $100,000 analysis? Genome Med 2(11), 84

(2010)

18. Phillips, K.A., Pletcher, M.J. & Ladabaum, U. Is the ``$1000 Genome'' really $1000?

Understanding the full benefits and costs of genomic sequencing. Technol Health Care 23(3), 373-9 (2015)

19. Weymann, D. et al. The cost and cost trajectory of whole‐genome analysis guiding

treatment of patients with advanced cancers. Mol Genet Genomic Med 5(3), 251–260 (2017)

20. Sullivan, S.D., et al. Budget Impact Analysis—Principles of Good Practice: Report of the

ISPOR 2012 Budget Impact Analysis Good Practice II Task Force. Value Health 17, :5–14

(2013)

21. *Payne, K., Eden. M., Davison, N., Bakker, E. Towards health technology assessment of

whole genome sequencing: challenges and solutions. Personalized Medicine 14(3), (2017) This

paper provides a detailed overview of how to focus the health technology assessment of whole

genome sequencing based diagnostics using the ‘PICO’ framework

22. NHS England – Spotlight on specialised services.

https://www.england.nhs.uk/wp-content/uploads/2017/09/spotlight-on-specialised-

services.pdf

23. Sagoo, G,. Norbury, G., Mohammed, S., Kroese, M. The budget impact and cost-

effectiveness of introducing whole-exome sequencing-based virtual gene panel tests into

routine clinical genetics. Cambridge: PHG Foundation (2017)

24. Payne, K. Fish and chips all round? Regulation of genetic-based technologies. Health

Economics 18(11), 1233-1236 (2009)

25. *Drummond, M.F.,, Sculpher, M.J., Torrance, G.W., O'Brien, B.J. & Stoddart, G.L.

Methods for the economic evaluation of health care programmes. Oxford University Press Inc:

New York (2005) A core textbook that explains methods of economic evaluation.

26. UK Genetic Testing Network Genetic test evaluation.

https://ukgtn.nhs.uk/resources/genetic-test-evaluation/

29

27. Centers for Disease Control and Prevention ACCE Model Process for Evaluating

Genetic Tests. https://www.cdc.gov/genomics/gtesting/acce/

28. National Institute for Health and Care Excellence. Guide to the methods of technology

appraisal (2013)29. National Institute for Health and Care Excellence. Diagnostic assessment programme

manual (2011)30. Dolan, P. et al. A social tariff for EuroQol: results from a UK general population survey.

University of York Discussion Paper 138, (1995)

31. Stinnett, A.A. & Mullahy, J. Net health benefits: a new framework for the analysis of

uncertainty in cost-effectiveness analysis. Medical Decision Making 18(2 Suppl), S68-80 (1998)

32. Smith, R. D. Construction of the contingent valuation market in health care: a critical

assessment. Health Econ 12, 609–628 (2003)

33. The Green Book: appraisal and evaluation in central government see

https://www.gov.uk/government/publications/the-green-book-appraisal-and-evaluation-in-

central-governent

34. Smith. R.D. & Sach, T.H. Contingent valuation: what needs to be done? Health

Economics Policy and Law 5, 91-111 (2010)

35. Briggs, A.H. & O'Brien, B.J. The death of cost-minimization analysis? Health Econ 10(2), 179-84 (2001)

36. Dakin, H & Wordsworth, S. Cost-minimisation analysis versus cost-effectiveness

analysis, revisited. Health Econ 22(1), 22-34 (2013)

37. Coast, J. Is economic evaluation in touch with society's health values? Br Med J 329, 1233-1236 (2004)

38. Brazier, J., McCabe, C., Edlin, R. Health economics and cost consequences analysis: a

step back in time. BMJ Rapid Response (2005)

39. Claxton, K.P., Sculpher, M.J., Ades, T. Cost consequences: implicit, opaque and anti

scientific. BMJ Rapid Response (2005)

40. Culyer, A.J. The normative economics of health care finance and provision. Oxford

Review of Economic Policy 5(1), 34-58 (1989)

41. Culyer, A.J., Evans, R.G. Mark Pauly on welfare economics: Normative rabbits from

positive hats. J Health Econ 15, 243–51 (1996)

42. Hurley, J.E. An Overview of the Normative Economics of the Health Sector in Handbook

of health economics 1(part A), 55-118 (2000)

30

43. Payne K & Thompson AJ. Economics of pharmacogenomics: rethinking beyond QALYs?

Pharmacogenomics 11(3), 187-95 (2013)

44. Brouwer, W.B., Culyer, A.J., van Exel, N.J., Rutten, F.F. Welfarism vs. extra-welfarism. J

Health Econ 27(2), 325-38 (2008)

45. International Society for Pharmacoeconomics and Outcome Research

(ISPOR). Pharmacoeconomic Guidelines around the World. 

http://www.ispor.org/PEguidelines/index.asp

46. *Sculpher, M.J., Claxton, K., Drummond, M., McCabe, C. Whither trial-based economic

evaluation for health care decision making? Health Econ 15(7), 677-87 (2006) This paper

provides a thorough explanation of the value of model-based economic evaluation

47. Dias, S., Sutton, A.J., Ades, A.E., Welton, N.J. Evidence synthesis for decision making

2: a generalized linear modeling framework for pairwise and network meta-analysis of

randomized controlled trials. Med Decision Making 33(5), 607-17 (2013)

48. *Briggs A, Sculpher M, Claxton K. Decision Modelling for Health Economic

Evaluation. Oxford: Oxford University Press; 2006 A core textbook that explains methods of

model-based economic evaluation.

49. Zechmeister-Koss, I., Schnell-Inderst, P.  Zauner, G. Appropriate Evidence Sources for

Populating Decision Analytic Models within Health Technology Assessment (HTA). A

Systematic Review of HTA Manuals and Health Economic Guidelines. Med Decision Making

34(3), 288–299 (2014)

50. Philips, Z., Bojke, L., Sculpher, M., Claxton, K., Golder, S. Good practice guidelines for

decision-analytic modelling in health technology assessment: A review and consolidation of

quality assessment. Pharmacoeconomics 24, 355-371 (2006)

51. Claxton, K. Exploring uncertainty in cost-effectiveness analysis. Pharmacoeconomics

26, 781-798 (2008)

52. Claxton, K., et al., Probabilistic sensitivity analysis for NICE technology assessment: not

an optional extra. Health Econ 14(4), 339-47 (2005)

53. Brennan, A. & Akehurst, R. Modelling in health economic evaluation. What is its place?

What is its value? Pharmacoeconomics 17(5), 445-59 (2000)

54. Buxton, M., et al. Modelling in economic evaluation: an unavoidable fact of life. Health

Econ 6, 217–27 (1997)

55. Newman, W., et al. A pragmatic randomised controlled trial of thiopurine

methyltransferase (TPMT) genotyping in the management of patients, prior to azathioprine

treatment: The TARGET study. Pharmacogenomics 12(6), 815-26 (2011)

31

56. Thompson, A.R., Newman, W.G., Elliott, R.A., Tricker, K., Payne, K. The cost-

effectiveness of a pharmacogenetic test: a trial-based evaluation of TPMT testing for

azathioprine. Value in Health 17, 22-33 (2014)

57. Roberts, M., et al. Conceptualizing a model: a report of the ISPOR-SMDM modeling

good research practices task force-2. Value Health 15, 804–811 (2012)

58. Brennan, A., Chick, S.E. & Davies, R. A taxonomy of model structures for economic

evaluation of health technologies. Health Economics. 15(12), 1295-310 (2006)

59. Siebert, U., et al. State-transition modeling: a report of the ISPOR-SMDM modeling good

research practices task force-3. Value Health 15, 812–820 (2012)

60. Karnon, J., et al. Modeling using discrete event simulation: a report of the ISPOR-SMDM

modeling good research practices task force-4. Value Health 15, 821–827 (2012)

61. Briggs, A.H., Claxton, K. & Sculpher, M.J. Decision Modelling for Health Economic

Evaluation. Oxford Handbooks in Health Economic Evaluation (2006)

62. Eddy, D.M., et al. Model transparency and validation: a report of the ISPOR-SMDM

modeling good research practices task force-7. Value Health 15, 843–850 (2012)

63. Briggs, A.H.,, et al. Model parameter estimation and uncertainty: a report of the ISPOR-

SMDM modelling good research practices task force-6. Value Health 15, 835–842 (2012)

64. Fenwick, E ., Claxton, K. & Sculpher. M. Representing uncertainty: the role of cost-

effectiveness acceptability curves. Health Economics 10(8), 779-87 (2001)

65. Basu, A. & Meltzer, D. Implications of spillover effects within the family for medical cost-

effectiveness analysis. J Health Econ 24(4), 751-73 (2005)

66. EGFR-TK mutation testing in adults with locally advanced or metastatic non-small-cell

lung cancer (NICE DG9). https://www.nice.org.uk/guidance/dg9

67. *Shabarrudin, F.H., Fleeman, N., Payne, K. Economic evaluations of personalized

medicine: existing challenges and current developments. Pharmacogenomics and Personalized

Medicine 8, 115–126 (2015) A good overview of the issues when performing economic

evaluations of pharmacogenetics and pharmacogenomics

68. *Annemans, L., Redekop, K., Payne, K. Current methodological issues in the economic

assessment of personalized medicine. Value Health 16(6 Suppl), S20-6 (2013) This paper

provides a detailed overview of the methodological challenges with model based economic

evaluations of personalized medicine

69. *Buchanan, J., Wordsworth, S., Schuh, A. Issues surrounding the health economic

evaluation of genomic technologies. Pharmacogenomics 14(15), 1833-47 (2013) This paper

32

provides a detailed overview of the methodological challenges with model based economic

evaluations of genomics

70. *Rogowski, W., et al. Concepts of 'personalization' in personalized medicine:

implications for economic evaluation. Pharmacoeconomics 33(1), 49-59 (2015) This paper

provides a detailed overview of the different methodological challenges with economic

evaluations of personalized medicine that depend on whether personalization is defined by

preferences or physiological characteristics

71. Phillips, K.A., Sakowski J.A., Liang, S., Ponce, N.A. Economic Perspectives on

Personalized Health Care and Prevention. Forum for Health Economics & Policy 6(2), (2013)

72. Buchanan, J., & Wordsworth, S. Welfarism versus extra-welfarism: can the choice of

economic evaluation approach impact on the adoption decisions recommended by economic

evaluation studies? Pharmacoeconomics 33, 571 - 579 (2015)

73. ACMG Board of Directors. Clinical utility of genetic and genomic services: a position

statement of the American College of Medical Genetics and Genomics. Genetics in Medicine

17(6), 505-507 (2005)

74. Grosse, S.D., Khoury, M.J. What is the clinical utility of genetic testing? Genetics in

Medicine 8(7), 448-450 (2006)

75. Foster, M.W., Mulvihill, J.J., Sharp, R.R. Evaluating the utility of personal genomic

information. Genetics in Medicine 11(8), 570-574 (2009)

76. Mills, R.A., Haga, S.B., Ginsburg, G.S. Genetic testing: clinical and personal utility.

Virtual Mentor 14(8), 604-609 (2012)

77. Payne, K., McAllister, M., Davies, L.M. Valuing the economic benefits of complex

interventions: when maximising health status is not sufficient. Health Economics 22(3), 258-71

(2013)

78. Coast J, Smith R, Lorgelly P. 2008a. Welfarism, extra-welfarism and capability: the

spread of ideas in health economics. Social Science and Medicine; 67: 1190-1198

79. ICECAP capability measures.

http://www.birmingham.ac.uk/research/activity/mds/projects/HaPS/HE/ICECAP/index.aspx

80. National Institute for Health and Care Excellence. The social care guidance manual

(2016)

81. Mitchell, P.M., Venkatapuram, S., Richardson, J., Iezzi, A., Coast, J. Are Quality-

Adjusted Life Years a Good Proxy Measure of Individual Capabilities? PharmacoEconomics

35(6), 637-46 (2017)

33

82. Mitchell, P.M., Roberts, T.E., Barton, P.M., Coast, J. Assessing sufficient capability: A

new approach to economic evaluation. Social Science & Medicine 139, 71-9 (2015)

83. Basu, A., Meltzer, D,. Value of Information on Preference Heterogeneity and

Individualized Care. Med Decis Making 27; 112-117 (2007)

84. Espinoza, M.A., Manca, A., Claxton, K., Sculpher, M.J. The value of heterogeneity for

cost-effectiveness subgroup analysis: conceptual framework and application. Med Decis Making

34(8), 951-64 (2014)

85. Sonnenberg, F.A., Beck, J.R. Markov models in medical decision making: a practical

guide. Med Decis Making 13(4), 322-38 (1993)

86. Briggs, A. & Sculpher, M. An introduction to Markov modelling for economic evaluation.

Pharmacoeconomics 13(4), 397-409 (1998)

87. Donten, A., Gray E., Payne. Using mathematical optimisation in model-based cost-

effectiveness analyses: a case study of a stratified breast screening programme. Health

Economics Study Group (2016)

88. Cormen, T.H., Leiserson, C.E., Rivest, R.L., Stein, C. Introduction to Algorithms.

Cambridge, Massachusetts: The MIT Press 2009

89. Vilaprinyo, E., et al., Cost-Effectiveness and Harm-Benefit Analyses of Risk-Based

Screening Strategies for Breast Cancer. PLoS One 9,: e86858 (2014)

90. Gray E, Donten A, Karssemeijer N, van Gils C, Astley S, Evans GE, Payne K. Evaluation

of a National Stratified Breast Screening Programme in the United Kingdom: A model-based

cost-effectiveness analysis. Value in Health dx.doi.org/10.1016/j.jval.2017.04.012 (2017)

91. Retinal degeneration 105 gene panel. https://ukgtn.nhs.uk/find-a-test/search-by-

disorder-gene/retinal-degeneration-105-gene-panel-568/

92. Caro, J.J, Möller, J., Karnon, J., Stahl, J., Ishak, J. Discrete Event Simulation for Health

Technology Assessment. Chapman & Hall (2015)

93. Salleh, S., Thokala, P., Brennan, A., Hughes, R., Dixon, S,. Discrete event simulation-

based resource modelling in health technology assessment. Pharmacoeconomics 35, 989-1006

(2017)

94. Goldratt, E.M. Essays on the Theory of Constraints. Great Barrington, MA: North River

Press (1998)

95. Williams, H.P. Model Building in Mathematical Programming. New York, NY: John Wiley

and Sons (1999)

96. Hillier, F.S. & Lieberman, G.J. Introduction to Operations Research. New York, NY:

McGraw-Hill (2010)

34

97. Atkins, D., et al., Systems for grading the quality of evidence and the strength of

recommendations I: Critical appraisal of existing approaches The GRADE Working Group. BMC

Health Serv Res 4, 38 (2004)

98. *Frank, M., Prenzler, A., Eils, R., Graf von der Schulenburg, J.-M., Lander, E., Linton, L.,

Ness, B. Van. Genome sequencing: a systematic review of health economic evidence. Health

Economics Review 3(1), 29 (2013) Provides examples of the challenges associated with

understanding the costs of genomic sequencing

99. Jani M, Gavan S, Chinoy H, Dixon WG, Harrison B, Moran A, Barton A, Payne K. A

microcosting study of immunogenicity and TNFi drug level tests for therapeutic drug monitoring

in clinical practice. Rheumatology 55(12), 2131–2137 (2016)

100. Bill and Melinda Gates Foundation Methods for Economic Evaluation Project (MEEP).

Final Report and Appendices. London: NICE (2014)

101. Bill and Melinda Gates Foundation Methods for Economic Evaluation Project (MEEP).

The Gates Reference Case: What it is, why it’s important, and how to use it. London: NICE

(2014)

102. Genetics Home Reference: What is precision medicine?

https://ghr.nlm.nih.gov/primer/precisionmedicine/definition

103. Health Technology Assessment of next-generation sequencing.

http://repository.ubn.ru.nl/handle/2066/173274

104. Schofield, D., et al. Cost-effectiveness of massively parallel sequencing for diagnosis of

paediatric muscle diseases. Genomic Medicine 2(1), 4 (2017)

105. Stark, Z., et al. Prospective comparison of the cost-effectiveness of clinical whole-

exome sequencing with that of usual care overwhelmingly supports early use and

reimbursement. Genet Med 19(8), 867-874 (2017)

106. Buchanan, J., et al. Using Genomic Information to guide ibrutinib treatment decisions in

Chronic Lymphocytic Leukaemia: A Cost-Effectiveness Analysis. Pharmacoeconomics 35 (8),  845–858 (2017)

107. Doble, B., Thomas, J., Thomas, D., Fellowes, A, Fox S, Lorgelly, P. Cost-effectiveness

of precision medicine in the fourth-line treatment of metastatic lung adenocarcinoma: An early

decision analytic model of multiplex targeted sequencing. Lung Cancer 107, 22-35 (2017)

108. Bennette,C.S., et al. The cost-effectiveness of returning incidental findings from next-

generation genomic sequencing. Genetics in Medicine 17(7), 587-595 (2015)

109. *Pearson, E.R. Personalized medicine in diabetes: the role of 'omics' and biomarkers.

Diabet Med 33(6), 712-7 (2016) This paper shows trends in the use of different terms in the

35

context of personalized or precision medicine

36

AcknowledgementsWe would like to acknowledge the help of Hunter Moore with the systematic review of

published economic evaluations of precision medicine in the Supplementary Appendix.

We would also like to acknowledge the expert advice on the applications of genomic

tests from Graeme Black, George Burghel, Georgina Hall and Bill Newman.

K.P. has a research programme supported by grants awarded from: National Institute

for Health Research; Medical Research Council (MRC); Engineering and Physical

Sciences Research Council (EPSRC); Lupus UK; The Swedish Foundation for

Humanities and Social Sciences. S. W. is funded by a Wellcome Trust PhD

Studentship. A. J. T. is funded by a grant awarded to The University of Manchester for

the Manchester Molecular Pathology and Innovation Centre (MMPathIC) by the Medical

Research Council (MRC) and Engineering and Physical Sciences Research Council

(EPSRC). S. P. G. is supported by two grants awarded to The University of Manchester

for MASTERPLANS, funded by the Medical Research Council (MRC) [grant number

MR/M01665X/1] Stratified Medicine Programme and by Lupus UK. The views

expressed in this article are those of the authors and not the funding bodies.

37

Further informationDuke Center for Applied Genomics and Precision Medicine

https://precisionmedicine.duke.edu

EuroQoL https://euroqol.org/

Genetics Home Reference https://ghr.nlm.nih.gov/primer/testing/costresults

Manchester Centre for Genomic Medicine http://www.mangen.co.uk/lab-

services/MolecularPathology.php

Genohub. https://genohub.com/

The Mayo Clinic http://www.mayomedicallaboratories.com

NHS England specialised services

https://www.england.nhs.uk/commissioning/spec-services/

NIH National Human Genome Research Institute

https://www.genome.gov/sequencingcosts/

National Institute for Health and Care Excellence https://www.nice.org.uk/

UK Genetic Testing Network – find a test https://ukgtn.nhs.uk/find-a-test/

38

GLOSSARY

Budget impact:  The total budget impact of a new technology in terms of the cost

falling on the budget-holder. Budget impact should be calculated by considering: the

perspective of the specific health care decision maker; the size and characteristics of

the population; the current and new treatment mix; the effectiveness and safety of the

new and current treatments; the resource use and costs for the treatments and

symptoms as would apply to the population of interest.

Companion diagnostic (also sometimes referred to as the test in a ‘test-treat’

strategy): A diagnostic test (typically an in-vitro diagnostic) co-developed alongside a

pharmaceutical and stated explicitly within the product label as it is essential for the safe

and effective use of the corresponding medicine.

Complementary diagnostic (also sometimes referred to as the test in a ‘test-treat’

strategy): A diagnostic test (typically an in-vitro diagnostic) developed to guide the use

of a pharmaceutical. The test may be mentioned in the medicinal product label but no

specific test is named

Conceptual model: An abstract representation of a phenomenon of interest, often

illustrated diagrammatically, to assist in determining the final structure of a de novo

decision analytic model in a model-based economic evaluation.

Cost-benefit analysis (CBA): A type of economic evaluation that compares the

relative costs and consequences of different courses of action in which the

consequences are measured using an approach that captures the impact in terms of

money, such as willingness-to-pay.

Cost-consequences analysis (CCA): A type of economic evaluation that compares

the relative costs and consequences of different courses of action in which the

consequences are measured using different outcomes and presented separately.

39

Cost-effectiveness analysis (CEA): A type of economic evaluation that compares the

relative costs and consequences of different courses of action in which the

consequences are measured using outcomes that capture the impact on clinical

effectiveness.

Cost-effectiveness threshold: The additional cost that must be imposed on the budget

for health care to displace one QALY elsewhere within the health care system.

Interventions which increase cost but with an incremental cost per QALY below the

threshold are typically viewed as being cost-effective. NICE uses a threshold range

thought to be between £20,000 and £30,000 per QALY gained.

Cost-minimization analysis (CMA): A type of economic evaluation that compares the

only the relative costs of different courses of action as it is assumed that the

consequences are equal.

Cost-utility analysis (CUA): A type of economic evaluation that compares the

relative costs and consequences of different courses of action in which the

consequences are measured using quality adjusted life years (QALYs). Often used

synonymously with cost-effectiveness analysis.

Decision analytic model: A series of mathematical relationships that represent the

progression of a patient’s diagnosis or disease and the impact of a health technology on

diagnosis and/or disease progression. The output of a decision analytic model can be

expressed in terms of the expected outcomes of interest for each alternative comparator

strategy.

Decision problem: An explicit statement of the resource allocation decision under

consideration.

40

Decision tree: A decision analytic modelling technique that simulates a cohort of

patients following a pre-defined pathway with associated probabilities, costs and

outcomes. Decision trees do not typically incorporate a time-component.

Discrete event simulation: A decision analytic modelling technique that simulates the

histories of individual patients over time, characterized by the specific events that they

may experience.

Economic evaluation: Defined by Drummond et al (2015) as “…the comparative

analysis of alternative courses of action in terms of both their costs and consequences ”.

An alternative typically refers to a health technology and the consequence is typically

expressed in terms of health benefits.

End-to-end evidence: Evidence that captures all elements of a test-and-treatment

strategy, by following a patient (i) from their observed test result, (ii) to a specific

treatment decision, and then to their final (health and resource) outcomes

Externalities: Occur when the effect of a production or individual consumption of goods

and services imposes costs or benefits on others. Externalities can be positive (benefit)

or negative (harm) and are sometimes called ‘spill over’ effects.

Extra-welfarism: A set of normative principles which guides the conduct, design and

interpretation of an economic evaluation. Extra-welfarism is typically taken to underpin

the use of cost-effectiveness analysis as the method of economic evaluation and the

Quality-Adjusted Life-Year as the metric of benefit/outcome.

Health technology: A technology that is used to create an intervention used in the

delivery of health care. For example, a medicine, diagnostic test, or medical device.

Incremental cost: The difference in cost between two alternative interventions.

41

Incremental cost-effectiveness ratio: The ratio of incremental costs to incremental

health benefits.

Incremental QALY: The difference in QALYs between two alternative interventions.

Incremental net benefit: incremental net benefit can be measured in monetary units

(incremental net monetary benefit) or units of health gain (incremental net health

benefit). If measured in monetary units (eg, dollars or euros), the incremental net benefit

(INB) is the monetary difference between expected net benefit of the new intervention

and the expected net benefit of the relevant comparator.

Normative: what ‘ought to be’; a statement underpinned by value judgments, rather

than a positive statement of ‘what is’.

National Institute for Health and Care Excellence (NICE): The decision-making

authority responsible for making recommendations regarding the allocation of

population health care resources in England.

Opportunity cost: The benefit forgone from the next-best use of a specific resource.

The opportunity cost of resource allocation decisions for health care can be expressed

in the health benefits forgone.

Perspective: The scope of the costs that should be included in an economic evaluation.

The perspective is typically defined by the budget constraint of the decision-maker.

Examples include a health care system perspective and a societal perspective. Also

helps to determine the relevant outcome chosen for analysis.

Probabilistic sensitivity analysis: A form of sensitivity analysis used where

uncertainty is propagated through the: (i) characterization of input parameters as

probability distributions and (ii) sampling of values for parameters using Monte Carlo

simulation.

42

Quality-adjusted life year (QALY): A generic outcome measure of health benefit,

calculated by multiplying each year of life by a weight that represents its health-related

quality of life. Weights are calculated according to the reference points of one (full

health) and zero (death); states worse than death are possible.

Reference case: A pre-specified preferred criteria for conducting an economic

evaluation. A reference case is typically an expression of a decision-maker’s value

judgement.

State transition Markov models: A type of decision analytic model that conceptualizes

a problem by defining relevant health states through which a cohort of patients transition

over time.

Time horizon: The scope of the costs and consequences that should be included in an

economic evaluation, from the present until a defined point in the future. The time

horizon for a study should be sufficient to allow all relevant costs and consequences to

be incorporated which, in general, requires a ‘lifetime time’ horizon to be used. The

lifetime is taken to be that of the last dying patient within the analysis cohort.

Uncertainty: Understanding the impact of uncertainty is a key component of any

economic evaluation. There are different types of uncertainty such as parameter,

methodological and decision. The impact of parameter and methodological uncertainty

can be captured using sensitivity analysis, such as probabilistic sensitivity analysis or

scenario analysis, respectively. Decision uncertainty is the probability that an incorrect

decision is made, in the context of resource allocation decisions for health care.

Welfarism: A set of normative principles which guides the conduct, design and

interpretation of an economic evaluation. Welfarism places individual utilities at the

heart of the evaluative space, is typically taken to be consistent with the use of cost-

benefit analysis and the use of ‘willingness to pay’ as the metric of benefit/outcome.

43

Author biographies

Katherine Payne holds a personal Chair in health economics at The University of

Manchester and leads a research group (the Manchester Centre for Health Economics

Precision Medicine Economics Team: MCHE-PMET) that applies and develops

methods of economic evaluation and valuation to understand the economic impact of

precision medicine and genomic technologies.

https://www.research.manchester.ac.uk/portal/Katherine.Payne.html

Sean P. Gavan is a research fellow in health economics at the Manchester Centre for

Health Economics, The University of Manchester. Sean has an interest in the economic

evaluation of precision medicine and the generation of evidence to support the uptake

of diagnostic health technologies within healthcare systems.

https://www.research.manchester.ac.uk/portal/sean.gavan.html

Stuart J. Wright is a Wellcome Trust funded PhD student at the Manchester Centre for

Health Economics, The University of Manchester. Stuart is interested in evaluating the

economic evidence for the introduction of precision medicine into clinical practice with a

particular focus on the impact of capacity constraints within healthcare systems.

https://www.research.manchester.ac.uk/portal/stuart.wright-2.html

Alexander J. Thompson is a research fellow in health economics and at the

Manchester Centre for Health Economics, The University of Manchester. Alex is

interested in developing value-frameworks and evaluating the economic evidence for

the introduction of precision medicine into clinical practice.

https://www.research.manchester.ac.uk/portal/alexander.thompson.html

44

Key points Informed by public and privately funded programmes, advances in next-

generation sequencing technologies have yielded genomic-targeted diagnostic tests,

based on either multi-gene panels, whole-exome sequencing or whole-genome

sequencing, are starting to replace some genetic tests in clinical practice.

Cost-effectiveness analysis of new genomic technologies, that explicitly

quantifies the costs and consequences of the new technology compared with other

potential uses of a healthcare budget, can provide decision-makers with information to

guide resource allocation decisions.

This review first summarizes what is known about, and describes the challenges

when estimating the magnitude of, the budget impact of genetic and genomic tests and

services. Existing literature has focussed on the downward-trend in the cost of

sequencing. We highlight the importance of estimating the opportunity cost (by

considering both the cost and consequences) of introducing new genomic tests.

The review then describes the potential role of decision analytic model-based

cost-effectiveness analysis, when compared with trial-based cost-effectiveness

analysis, and then outlines the key steps to consider in the evaluation of genetic and

genomic tests.

To inform decisions about the efficient allocation of healthcare resources, model-

based CEAs must be supported and populated by robust studies, which ideally span the

three component parts of a genomic test: the technology; the diagnostic component; the

model of service delivery to provide the diagnostic test and, if appropriate, subsequent

treatment options.

An emerging economic evidence base for genetic tests has been identified. It is

now time to direct funding to support the empirical research needed to develop the use

of decision-analytic model based CEAs of genomic tests, while being cognizant of the

known methodological, technical, practical and organizational challenges, to maximize

the potential benefits to patient populations.

45