NOU 2012: 16

Cost-Benefit Analysis

To table of content

10 Valuation of life and health

10.1 Introduction

From the terms of reference of the Committee:

The cost-benefit analysis guide offers general recommendations as to how one may seek to quantify the value of accident-reducing measures. This may, for example, be of relevance to cost-benefit calculations concerning safety measures within the transportation sector. The estimated value of a statistical life lost is presented in such a context. The Directorate for Health and Social Affairs recommended, in a report from 2007, the broad use of such a concept in intersectoral health impact assessments. The Committee shall examine what weight intersectoral cost-benefit assessment standards should carry in the evaluation of the impact on life and health, including within the health sector, and any ethical issues that may be raised thereby.

The terms of reference refer to the recommendations in the Ministry of Finance cost-benefit analysis guide (Ministry of Finance, 2005) as to how life and health may be valued and incorporated into the cost-benefit analysis of accident-reducing measures. The Ministry of Finance guide from 2005 is based on the recommendations made in the NOU 1997: 27 Green Paper, but the guide also includes updated recommendations on the economic valuation of statistical lives and life years. The present Green Paper is, as noted in Chapter 2, a supplement to the NOU 1997: 27 and NOU 1998: 16 Green Papers. The present Chapter therefore briefly outlines the assessments and recommendations made in the NOU 1997: 27 Green Paper in this area.

Life and health impacts may be included in the cost-benefit analysis of both preventive measures and treatment measures. However, it should be noted, to begin with, that any cost-benefit analysis of treatment measures will form part of the basis for making general capacity decisions, and not decisions at the level of individual patients. General capacity assessments are concerned with, for example, how much overall capacity one should have for a specific type of cardiac surgery. Such general assessments are referred to as first-order decisions at group level by the Lønning 2 Committee (NOU 1997: 18 Green Paper). Assessments as to what treatment shall be administered to each individual patient, in view of the available capacity, is referred to as second-order decisions in the NOU 1997: 18 Green Paper. Available capacity at group level shall thus, ideally speaking, be determined through well-founded first-order decisions, which define the framework for making second-order decisions for individual patients at the clinical level.

In Chapter 10.2, we briefly explain some key concepts used in the recommendations in the NOU 1997: 27 Green Paper, and frequently used when discussing life and health in a cost-benefit analysis context, whilst conclusions and recommendations from the NOU 1997: 27 Green Paper are discussed in Chapter 10.3. In Chapter 10.4, we outline the Committee’s interpretation of the terms of reference. In Chapter 10.5, we embark on a more detailed discussion of the meaning and applicability of the terms discussed in Chapter 10.2, based on recent research. Chapter 10.6 discusses practices and applications for the valuation of life and health in various countries, including Norway. The assessments and recommendations of the Committee are set out in Chapter 10.7 and Chapter 10.8, respectively.

10.2 Some key concepts

Ex-ante versus ex-post events – relevance to identifiability and willingness to pay studies

Ex ante means, in the context used in the NOU 1997: 27 Green Paper, that the accident or disease has not occurred, and hence that affected individuals cannot be identified with any certainty. Correspondingly, ex post means that the event has occurred and that the affected individuals can be identified. The reason why the NOU 1997: 27 Green Paper also highlights “a large number of individuals” and “small risks” as prerequisites for using economic valuations of life and health in cost-benefit analysis (cf. Chapter 2) is that this is the setting frequently used in studies of the willingness to pay (cf. Chapter 2) for reduced risk. Whether a large number of individuals and small risk for each individual are also necessary criteria for the ex-ante transfer of valuations and uses related to cost-effectiveness analysis (cf. below) is discussed in Chapter 10.5.

Cost-effectiveness analysis in health economics

The term cost-effectiveness analysis is used in the NOU 1997: 27 Green Paper and Ministry of Finance (2005) to designate an analysis of various measures that all offer the same benefits. Hence, the ranking of the measures will depend exclusively on the costs of such measures (cf. Chapter 2). In health economics analysis, this type of comparison of measures is often termed cost-minimisation analysis, whilst the term cost-effectiveness analysis is used to designate the analysis of measures offering different benefits that are measured on a common, comparable scale; for example quality-adjusted life years (cf. e.g. the discussion of “cost-effectiveness analysis” in Drummond et al., 2005, and Boardman et al., 2011). If such measures can be scaled up or down, measures that offer different benefits to begin with can be made identical as far as the magnitude of the benefits is concerned, and be ranked on the basis of costs only. Such scaling may be difficult in practice, but it follows from the above that it would not make much of a difference in theory whether one assumes that cost-effectiveness analysis means that the measures being compared offer identical or non-identical benefits – as long as these benefits are measured on a common, comparable scale.

If quality-adjusted life years (“QALY”; see below for a definition) is used as the health indicator, cost-effectiveness analysis is often termed cost-per-QALY analysis (Norwegian Directorate of Health, 2011). Unlike cost-benefit analysis (as a specific method of analysis), cost-effectiveness analysis does not include the economic valuation of health indicators. However, it is fairly common for health economics analysis to subsequently compare the estimated cost-effectiveness of a measure, e.g. in terms of NOK per QALY gained, with an estimated monetary value per QALY. Such values may be based on different premises (cf. Chapter 10.5, Box 10.1).

The value of a statistical life

The value of a statistical life (“VSL”) is defined in the NOU 1997: 27 Green Paper as the value of a one-unit reduction in the expected number of fatalities over a given period. VSL is often estimated as the sum of money each individual in the population is willing to pay for a given reduction in the risk of (premature) death, e.g. due to accident or air pollution (OECD, 2012). An estimated VSL represents the overall willingness to pay of a given population (here Norway’s population) for a risk reduction of the exact magnitude that can be expected to save one life. Reference is made to the NOU 1997: 27 Green Paper for a more comprehensive theoretical explanation of VSL.

The value of a statistical life year

The value of a statistical life year (“VOLY”) is discussed in the NOU 1997: 27 Green Paper as an alternative to using VSL for cost-benefit analysis purposes. The argument in favour of using VOLY is that life years can be a more precise unit of measurement in cases where the measure under evaluation applies to young or old individuals, where a large or small number of life years can be gained. Ministry of Finance (2005) recommends using VOLY for sensitivity analysis when measures affect the old and infirm, and a higher VSL value when measures affect children.

Measurement of health-related quality of life - Quality-adjusted life years

Quality-adjusted life years (“QALY”) are discussed in the NOU 1997: 27 Green Paper as a possible decision-making criterion for prioritisation between different patient groups and cases.

Statistical lives and statistical life years are concepts that lend themselves to measurement by counting. It is not equally straightforward either to observe or to count the health-related quality of life. A number of disease-specific measurement instruments are in existence, which are predominantly used in clinical contexts, but can also be used for economic analysis. Quality-adjusted life years (”QALY”) has been developed as a generic (joint) measurement instrument that can be used to compare measures to counter different diseases or injuries. Quality-adjusted life years are represented by an indicator ranging from zero to one, which indicator is based on various dimensions of the health-related quality of life. The quality indicator is multiplied by the duration of the benefit to become quality-adjusted life years. A quality-adjusted life year equal to 1 reflects a “full score” on all dimensions included in the measurement instrument used.

Reference is made to the NOU 1997: 27 Green Paper for an overview of quality of life measurement methods and various indices/instruments for the use of QALY in analysis. It is not necessary to calculate any monetary value of QALY when using QALY as a benefit indicator in cost-effectiveness analysis.

10.3 Conclusions and recommendations from the NOU 1997: 27 Green Paper

Chapter 12 of the NOU 1997: 27 Green Paper provides a thorough discussion of dealing with health risk changes in a cost-benefit analysis context. We reiterate the conclusions and recommendations from the NOU 1997: 27 Green Paper in Chapter 10.3, as many of these are deemed to remain intact. The recommendations in the NOU 1997: 27 Green Paper are divided into recommendations on accident risk analysis and recommendations on health sector analysis. Said division is premised on differences in the identifiability of individuals, differences in health risk and health-related quality of life, as well as differences in expected lifespan. Presentation of recommendations from the NOU 1997: 27 Green Paper is supplemented by comments and pointers to where the topics are discussed in the present Green Paper. Reference is made to the NOU 1997: 27 Green Paper for a more thorough presentation, and to Institute of Transport Economics (“TØI”) (2010) for an updated overview of the theoretical and empirical basis for the valuation of reduced accident risk.

10.3.1 Accident risk – recommendations from the NOU 1997: 27 Green Paper

From the NOU 1997: 27 Green Paper, p. 111:

It is not unproblematic to make recommendations on how the valuation of changes to the risks associated with life and health should be incorporated into cost-benefit analysis. One reason for this is that the technical basis for doing so remains unsettled in many respects, especially as far as empirical applications are concerned. However, an even more important consideration is that decisions relating to life and health will in most cases involve ethical trade-offs that are not necessarily meaningfully illuminated by economic theory. Nevertheless, the Committee deems it appropriate to make some recommendations on the valuation of changes to the risks associated with life and health.

Comments:

Unresolved aspects of the technical basis and empirical applications identified in the NOU 1997: 27 Green Paper are widely dispersed valuations of statistical lives due to data availability, choice of method, different types of risk, as well as the problems individuals experience in making rational choices and understanding complex problems that involve low probabilities. Such insensitivity to the magnitude of the risk reduction (“scope bias”) and validity problems in hypothetical valuation studies (“hypothetical bias”) remains unresolved (Andersson and Treich, 2012, Hultkrantz and Svensson, 2008, 2012), and hence the uncertainty associated with the valuation estimates is considerable (Institute of Transport Economics (“TØI”), 2010).

The ethical trade-offs discussed in the NOU 1997: 27 Green Paper have to do with the observation that valuation of life and health can hardly be related to one specific life, and should preferably be made in a context where the willingness to pay concerns a reduced probability of death and health impairment, without knowing the identity of the afflicted persons. Hence, the requirement is that life and health be valued in situations where a large number of individuals are facing small risks.

In 1997, the Committee took the view that the risks associated with life and health can on certain conditions (cf. the clarification outlined by the Committee in the next paragraph) be subjected to an economic valuation and incorporated into cost-benefit analysis, despite such valuation involving difficult ethical issues.

From the NOU 1997: 27 Green Paper, p. 111:

The Committee is of the view that cost-effectiveness analyses can often be appropriate for the assessment of changes in accident risk. If one chooses, in addition thereto, to perform a complete cost-benefit analysis, as a specific method of analysis, the Committee believes that the willingness to pay for risk changes should be included on the benefit side of the analysis. The Committee will nevertheless recommend that such valuation be limited to situations in which a large number of individuals are facing small risks of an undesirable outcome, cf. the discussion in Chapter 12.1.
In those cases where separate willingness to pay analysis is not performed, the Committee will recommend starting out from the estimate of NOK 10 million (at 1991 prices) per statistical life, cf. the discussion in Chapter 12.3.2. However, the Committee will not recommend any separate supplements being added with regard to altruism or gross (alternatively net) production value. This is because the theoretical basis for such supplements seems unclear, cf. the discussion in Chapter 12.5.2, and the general perspective that the Committee will in cases involving such doubt tend to err on the side of caution in choosing estimates. This implies that the Committee recommends an amount that is about NOK 4 million lower than that recommended by Elvik (1993) as the value of avoiding one traffic fatality.

Comments:

The estimate of NOK 10 million at 1991 prices per statistical life was updated to NOK 15 million at 2005 prices in Ministry of Finance (2005) and updated to NOK 17.4 million at 2010 prices in Norwegian Government Agency for Financial Management (“DFØ”) (2010).

Altruism is here taken to mean a willingness to pay for a risk reduction that increases the welfare of others. The NOU 1997: 27 Green Paper assumes that production value as the result of changes in the production capacity of individuals is included in the measured willingness to pay for a risk reduction.

From the NOU 1997: 27 Green Paper, p. 112:

It may in some cases be argued that a statistical life should be accorded a value different from NOK 10 million, although no specific willingness to pay analysis has been carried out. The background to this is often an assessment as to the number of remaining life years or whether those affected seek the risk of their own volition, and not necessarily that the actual willingness to pay is higher than in the average case. One example may be traffic safety measures targeting children going to or from school. There may be reason to believe that many people will, with varying degrees of explicitness, wish to attribute a value in excess of NOK 10 million to each statistical life in such cases, because, inter alia, each life saved represents many remaining life years. It is likely, on the other hand, that the average willingness to pay may be fairly low for accident reduction relating to involvement in hobby activities like e.g. skydiving. The Committee will nevertheless recommend that the valuation of NOK 10 million per statistical life be applied in all cost-benefit analysis in such cases. However, cost-benefit analysis of changes to accident risk should include a description outlining what group of individuals would be encompassed by the measure. Such a description enables the decision maker to choose, if desirable, a different ranking of measures than would be implied by the value of a statistical life, when taken in isolation. This corresponds to the approach recommended by the Committee with regard to the presentation of distribution policy issues in Chapter 4.

Comments:

The question of whether one should use statistical life years in analysis instead of, or in addition to, statistical lives is discussed in Chapters 10.5 and 10.7.

Despite the assumption of a relatively low willingness to pay in cases where individuals seek a risk of their own volition, the NOU 1997: 27 Green Paper recommends the same valuation as for other types of risk. It is worth noting, in this context, that the Lønning 2 Committee (NOU 1997: 18 Green Paper) identified a lifestyle that impairs the effect of a measure as one of several “criteria that may be included when making discretionary prioritisations” in the health sector.

10.3.2 Cost-benefit analysis in the health sector – recommendations from the NOU 1997: 27 Green Paper

Willingness to pay studies are commonly used in the valuation of changes in risks associated with life and health caused by measures in other social sectors than the health sector. In the context of preventive measures, e.g. relating to accidents, one may ask individuals about their willingness to pay for reduced accident risk before the health impairment has materialised. However, the measures under assessment in the health sector are often related to diagnosis, treatment and rehabilitation. At those stages it will often be the case that the disease has already been contracted or the injury has already been sustained, or that groups with high risk exposure (for example associated with the inheritability of rare diseases) may be known, and hence those individuals who will benefit from the measure will be more readily identifiable.

From the NOU 1997: 27 Green Paper, p. 113:

The Committee is of the view that using willingness to pay within the health sector is in most cases considerably more problematic than with regard to accident risk. The most important reason for this is than one will rarely be in a position to fully base decisions on an ex ante perspective where a large number of individuals are facing small risks. It will typically be the case that some individuals have already contracted a disease, and using willingness to pay may seem unreasonable in such an ex-post situation. One example may be determining cardiac surgery capacity, which will affect both individuals who are completely healthy at the moment, individuals with a high probability of contracting cardiovascular disease and individuals who are already ill. It is difficult to see how willingness to pay studies can do much to solve the prioritisation problem in such a situation. In addition, it can be difficult to obtain useful findings from a willingness to pay survey because most people will have little experience with valuing health goods in monetary terms.

Comments:

Whether it is theoretically justifiable and ethically appropriate to use valuation of life and health in the analysis of treatments when such valuation is not based on studies of willingness to pay in an ex-post situation, but is instead based on the transfer of valuations estimated in an ex-ante situation, is discussed in Chapters 10.5 and 10.7.

From the NOU 1997: 27 Green Paper, p. 113:

The problems associated with using willingness to pay make it appropriate to use cost-effectiveness analysis within the health sector. However, it can often be difficult, as discussed in Chapter 12.6.1, to find cases where the benefits from various measures are the same, thus enabling the use of cost-effectiveness analyses. This means that quality-adjusted life years should be considered as a potential decision-making criterion.
In principle, the use of quality-adjusted life years offers a systematic method for prioritisation between different patient groups and cases. The method implies that everyone with the same disease is treated the same irrespective of income and wealth. However, this does not mean that the method solves the difficult distribution problems that will be encountered whenever scarce resources are to be allocated within the public health service. As far as the Committee is concerned, it seems both unrealistic and undesirable for various groups to be subjected to strict prioritisation on the basis of one single indicator like quality-adjusted life years. This is in line with views expressed by the Lønning Committee (NOU 1997:18), which notes that quality of life is a very difficult phenomenon to measure, and that quality-adjusted life years therefore cannot be the sole criterion when assessing the desirability of a health investment. Reporting of quality-adjusted life years may nevertheless provide information of interest to decision making, e.g. in the context of the evaluation of alternative medicine types. In addition, the use of quality-adjusted life years means that the effects of different treatment methods are described in a systematic and precise manner. The Committee therefore recommends a continued effort to make more systematic use of quality-adjusted life years or other disaggregated health indicators when performing health economics assessments.

Comments:

The recommendation in the NOU 1997: 27 Green Paper for a continued effort to make more systematic use of quality-adjusted life years (see above for an explanation of the concept) has largely been adhered to (cf. Chapter 10.6).

The NOU 1997: 27 Green Paper notes that the use of cost-effectiveness analysis can contribute to efficient resource use within a fixed budget limit, with no need for explicit economic valuation of life and health. This applies irrespective of whether one uses statistical lives, statistical life years or quality-adjusted life years as health indicator. Various aspects of the economic valuation of quality-adjusted life years are discussed in Chapter 10.6.

10.4 The Committee’s interpretation of the terms of reference

The duties stipulated in the terms of reference are to “examine what weight intersectoral cost-benefit assessment standards should carry in the evaluation of the impact on life and health, including within the health sector, and any ethical issues that may be raised thereby”. The Committee has deemed it necessary to clarify its interpretation of its duties as thus stipulated in the terms of reference, before presenting its assessments and recommendations.

10.4.1 Interpretation and delimitation of the terms of reference

Different meanings may be attributed to some of the concepts mentioned in the terms of reference. The Committee has therefore specified its interpretation of the concepts “intersectoral standards”, “cost-benefit assessment”, “life and health” and “what weight”, and the consequences thereof for the delimitation of the terms of reference.

10.4.1.1 Intersectoral standards

Intersectoral standards are here interpreted as uniform practices (standards) for using cost-benefit analysis in various sectors where life and health feature as major or minor effects. These may amount to recommendations on more homogeneous method use, more homogeneous health indicators and/or more uniform economic valuation of health indicators. Since the terms of reference mention the value of a statistical life, the terms of reference are interpreted as requesting an examination of the standardisation of the economic value used for analysis purposes.

10.4.1.2 Cost-benefit assessments

The Committee has applied a broad interpretation of the term “cost-benefit assessments”. This means that such term is used in the broader sense of economic analysis (i.e. not only as a specific method of analysis, cf. Chapter 2). Said interpretation is supported by the fact that little use is made of cost-benefit analysis, as a specific method, in the health sector, together with the fact that the health sector is expressly mentioned in the terms of reference.

Valuation of the benefit side is not necessary for cost-effectiveness analysis purposes, and hence standardisation of valuation estimates is, generally speaking, not required for such purposes. Nevertheless, the issue arises because it is, as mentioned above, not uncommon to combine cost-effectiveness analysis where the benefit side is measured in quality-adjusted life years, or in other standardised, comparable units, with threshold values intended to define acceptable limits for cost per benefit unit (e.g. per quality-adjusted life year). It should be noted that such practice makes cost-effectiveness analysis very similar to cost-benefit analysis (as a specific method of analysis).

10.4.1.3 Life and health

Life and health may be included in cost-benefit analysis in the form of various units of measurement. The value of a statistical life is expressly mentioned in the terms of reference. Statistical lives are the most commonly used economic valuation units in studies of the willingness to pay for reduced accident risk. In the health sector, the degree to which measures affect expected lifespans or not varies, and their objectives are often concerned both with extending lifespans and improving health, whilst some measures only influence health, and not lifespans. Statistical lives or life years are not suited for capturing the benefits of improved health. Since quality-adjusted life years are commonly used units of measurement in economic analysis within the health sector, and the health sector is expressly mentioned in the terms of reference, the Committee assumes that life and health in this context includes the units of measurement statistical lives, statistical life years and quality-adjusted life years. The Committee focuses on these aggregate units of measurements because intersectoral standards may be of relevance in relation thereto, but there is nothing to prevent more specific health indicators from being used for cost-benefit analysis purposes.

10.4.1.4 Weight in evaluations

The Committee has been requested to examine “what weight” intersectoral cost-benefit assessment standards should carry “in the evaluation of the impact” on life and health.

One possible interpretation of this is that one shall examine whether economic valuations could and should be used in the prioritisation of resources for measures with an impact on life and health. This includes whether economic theory justifies a recommendation to the effect that the authorities should establish threshold values for the maximum acceptable cost per QALY (or other health indicators) and, if applicable, what should be the level of these threshold values.

In Chapter 3, the Committee has specified that it considers the outcome of economic profitability assessments to be part of the basis for making decisions, and not a decision-making rule. In Chapter 3, the Committee has also emphasised that the economic profitability of a measure (as calculated through cost-benefit analysis) cannot be interpreted normatively as a matter of course, but should primarily be considered a descriptive calculation of the net willingness of the population to pay. The extent to which actual decisions should and shall be based on the net willingness of the population to pay is a matter that falls outside the scope of cost-benefit analysis. The question of what weight should be attributed to life and health when the authorities are to make actual decisions must be considered a political and ethical question that falls outside the scope of the terms of reference of the Committee. On the other hand, the Committee takes the view that it falls within the scope of the terms of reference to comment on the theoretical and empirical basis for calculating, introducing and using, if applicable, threshold values as mentioned above.

The NOU 1997: 27 Green Paper recommended a value of a statistical life for use in the analysis of measures that reduce accident risk. These are primarily measures that result in small changes in risk for a large number of persons. In the health sector, there is often a need for evaluating measures that will result in a major change in risk for a fairly small number of persons, and where the persons concerned may also to some extent be identified. One possible interpretation of the terms of reference is therefore that these call for an examination of the extent to which economic valuations can be used in the analysis of health sector measures like diagnosis, treatment, rehabilitation, alleviation, etc., which involve major changes in risks associated with life and health, where the target group in many cases can be identified.

10.4.1.5 The Committee’s delimitation of the terms of reference

The terms of reference are interpreted to mean that an examination of the following issues is requested: 1) To what extent could and should economic valuations be used as a basis for decision making in the analysis of measures that may entail major changes to risks associated with life and health, where it may, moreover, be known which individuals are affected by such changes in risk? 2) Should the economic values used in such analysis, if any, be the same for all sectors?

10.5 Interpretation and applicability of the terms VSL, VOLY and QALY

It is not necessarily unproblematic to use VSL to estimate the willingness to pay for lifesaving measures which involve non-marginal risk reductions, which pertain to groups with a different risk level than the general population to begin with and/or which can save lives that are identifiable in full or in part. Many measures in the health sector are characterised by exactly such circumstances. This will also have consequences for whether VOLY will be a good unit of measurement and for what use can be made of QALY. In the present Chapter we will examine the interpretation and applicability of VSL, VOLY and QALY in the cost-benefit analysis of life and health.

VSL

The value of a statistical life, VSL, is measured, as discussed above, by the willingness of individuals to pay for a small risk reduction. The VSL concept must be interpreted against that background. If a sufficiently large number of individuals experience a minor reduction in their risk of death (probability of premature death), these small risk reductions will in aggregate mean that society can expect fewer premature fatalities. However, the willingness of people to pay for small risk reductions will not necessarily be identical to their willingness to pay for lifesaving as such, and this imposes certain restrictions on the use of VSL to estimate the willingness of the population to pay for lifesaving measures in general.

A lifesaving measure may, for example, primarily relate to other people, whilst VSL is usually estimated on the basis of a marginal reduction in own risk. It may be known which persons’ lives the measure is expected to save, or who has a particularly high probability of being affected. The risk reduction for those involved may be non-marginal (for example from highly probable death to highly probable recovery). Their risk level to begin with may be very different from the risk level of a representative selection of the population (for example because the target group is seriously ill to begin with). All these factors may affect the willingness to pay, at times considerably so (see Hammitt and Treich, 2007, and Hammitt, 2012).

Performing a cost-benefit analysis of lifesaving measures by using VSL may nonetheless be of interest in such cases as well, provided that one evaluates alternative measures by adopting a fairly peculiar and hypothetical approach, i.e. what is often termed «behind the veil of ignorance» (Harsanyi, 1955, and Rawls, 1971). This involves envisaging that one is not yet aware of one’s own identity or how one will fare in life, and hence does not know whether one will become a cancer patient, have children, end up in a risk group for cardiovascular disease, etc. When considered from such a perspective, a new curative treatment for a specific group of cancer patients may, for example, be considered to involve a small risk reduction for each individual in the population, because nobody «yet knows» whether they will be afflicted themselves. This may conceivably be an interesting and relevant philosophical exercise for decision makers. Since neither the decision makers, nor the remainder of the population, are in actual fact behind any “veil of ignorance”, it must nevertheless be emphasised that the findings from such a cost-benefit analysis cannot be considered a measure of the actual net willingness of the population to pay in the situation in which the decision is in fact made. It may still provide an interesting indicator for decision makers charged with prioritising the use of society’s resources. It is under any circumstance important to note that VSL is a measure of the willingness to pay for a marginal risk reduction, and not for life and health as such.

A similar problem, although somewhat different in principle, arises when the change in risk applies to identified persons (Hammitt and Treich, 2007). A person who knows that her life will be saved by a measure will of course have an exceptionally high willingness to pay for such measure. Hence, a measure that is not economically profitable before the identities of those affected are known may nevertheless, as shown by Broome (1978), be profitable when measured after their identities have become known – because the change in risk has a very strong impact on their willingness to pay. Hammitt and Treich (2007) note that cost-effectiveness analyses can therefore, under certain assumptions, provide a more precise illustration of the welfare effects of such measures than can cost-benefit analysis. They also note that the findings from cost-benefit analysis for this type of measure can potentially be manipulated by tactical choice of the date of disclosing information in relation to the date of analysis.

VOLY

It is, as explained above, fairly straightforward to aggregate the willingness of individuals to pay for their own marginally reduced risk of death into a willingness to pay for a reduced number of expected premature fatalities in the population as a whole (“VSL”). It is, on the other hand, somewhat less clear how one might derive the value of a statistical life year (“VOLY”) from the value of a statistical life. In estimating VSL, people are usually asked about their willingness to pay for a reduced risk of death, but not about how many years they expect to have left to live to begin with, or how much relative weight they attach to each of these years (their pure rates of time preference). Consequently, it is not obvious how best to convert VSL into VOLY. It is possible, under highly simplifying assumptions, to arrive at the relationship between VSL and VOLY presented in Box 10.1. The formula in Box 10.1 is based on, inter alia, the assumption that individuals’ pure rate of time preference is constant over time, which assumption is questioned in recent literature (see for example Frederick, Loewnstein and O’Donoghue, 2002). Moreover, the formula makes implicit assumptions about the distribution of remaining life years and about time preferences amongst the respondents. The relationship between a marginally reduced risk of death and an expected life saved for society is more straightforward, and thus requires assumptions that are less strict, than the relationship between a marginally reduced risk of death and an expected life year saved for society.

VOLY can also be estimated directly via willingness to pay studies, cf. the discussion of applications in Norway in Chapter 10.6.1 and in the United Kingdom in Chapter 10.6.2.

Textbox 10.1 Assumptions underpinning conversion from VSL to VOLY and QALY

Make the following simplifying assumptions: 1) All individuals are identical, thus implying, inter alia, that the mean VSL in the population equals the VSL of each individual; 2) VOLY is constant over the lifespan of the individual; 3) Expected remaining lifespan T for each of the (identical) individuals is known; and 4) The discount rate applied by each individual (see Chapter 5 on discount rates), δ, is known and constant over time. We will then have the following relationship between VSL and VOLY (Boardman et al., 2011, and OECD, 2012):

Figure  

Assume, moreover, that it is possible to represent quality of life on a zero-to-one scale. Assume also that VSL can actually be interpreted as measuring the willingness of the population to pay for a life saved as such, and not only for marginal risk reductions (see Chapter 10.5). The willingness to pay for a QALY in a year t may then be calculated as wtVOLY (Boardman et al., 2011), where wt designates the quality of life in year t on the scale from zero to one. This implies that VOLY, which is an indicator that does not weigh life years by the health-related quality of life, is deemed to be equivalent to the value of a life year with perfect health (wt=1).

QALY

As discussed in Chapter 10.2, it is not necessary to calculate any monetary value for QALY in order to use it as a benefit indicator in a cost-effectiveness analysis. There are various methods for calculating QALY, which often attach weight to different factors and which are often based on different principles (cf. e.g. the NOU 1997: 27 Green Paper and Olsen, 2009). Different measurement methods may reflect what are, in principle, different ways of posing questions. QALY as a unit of measurement in relation to changes in life and health has been criticised from many quarters. Arnesen and Norheim (2003), Augestad et al. (2012) and Hendriksen (2012) may, for example, be mentioned in a Norwegian context. Nor is there a full consensus amongst researchers within the field as to which method is preferable. Drummond et al. (2009) discuss how QALY may be developed and improved as a health indicator. Norwegian Directorate of Health (2011) and Norwegian Medicines Agency (2012) examine which measurement instruments are most appropriate to use for QALY in a Norwegian cost-benefit analysis context. However, no decision has thus far been made as to the standardisation of methods and the weighting of the health-related quality of life in QALY. Calculation of QALY involves, inter alia, having to assess how important different types of afflictions and disabilities are in relation to each other, and such choices may of course be controversial. QALY should therefore be considered an indicator, rather than an exact measurement method for health benefits.

Although the use of QALY methodology has tended to be considered an approach different from the use of the willingness to pay for health goods in health economics analysis (cf. also the fact that cost-effectiveness analyses is more frequently used than cost-benefit analysis, as a specific method of analysis, in analysing health measures), one has also sought to estimate the value of a QALY through willingness to pay studies (cf. e.g. Gyrd-Hansen, 2003, Dolan et al., 2003, Dolan et al., 2008, and Bobinac et al., 2010). Hammitt (2002) provides a thorough examination of the theoretical differences between willingness to pay (“WTP”) and QALY. One may say, as a brief summary, that studies of the willingness to pay for a QALY suffer from the same methodological problems as studies of the willingness to pay for other non-market goods, and that issues relating to identifiability and non-marginal risk reductions are additional thereto. A person who knows that her life will be saved by a measure will obviously, according to Hammitt and Treich (2007), have an exceptionally high willingness to pay for such measure. This implies that the validity of such hypothetical valuations can often be questioned (cf. e.g. Smith and Richardson, 2005, Pinto-Prades et al., 2009, and Bobinac et al., 2012).

It is also established practice to convert VOLY to the value of a QALY (cf. e.g. Svensson and Hultkrantz, 2012, discussed in Chapter 10.6.2). Such practice is premised on a number of highly simplified assumptions; see Box 10.1. Firstly, it is assumed that VSL does in fact value life, and not risk reduction. Secondly, QALY is an indicator, and only measures aspects related to those dimensions that are included in the measurement instrument. When using the willingness to pay for statistical lives or life years, the valuation is not restricted to dimensions of any measurement instrument. Since the valuation of QALY is restricted by the dimensions of the measurement instrument, it is not possible to be entirely certain that 1 QALY corresponds to one year of “perfect” health, because it must be taken into consideration that the measurement instrument does not capture all aspects of health and life. Correspondingly, it is not possible to be certain that QALYt = wtVOLY cf. Box 10.1, because it is not possible to be certain that QALY and VOLY measure the same dimensions of life. In addition, it is not obviously the case that the willingness to pay for health changes can be assumed to be related to the willingness to pay for life in a straightforward manner.

QALY is often interpreted as expressing individual preferences. It may be appropriate to recall, in this context, that the aggregation of the preferences of various individuals is, generally speaking, theoretically problematic, cf. Chapter 3. Even if QALY had been sufficiently well measured to provide a precise description of the preferences of each individual, the questions of how best to aggregate QALY, and how to interpret such aggregated figure, would have remained.

Statistical lives, statistical life years and quality-adjusted life years are all aggregate health indicators. In principle, these can all be used for cost-benefit analysis purposes, irrespective of whether or not monetary values are attributed thereto. Quality-adjusted life years may be a more appropriate indicator for health gain than statistical life years when the quality of life during the life years gained is clearly inferior or varies considerably between alternative measures, in the same way that statistical life years may be a more precise indicator than statistical lives when the number of life years gained is small or varies considerably between measures. Although it may often be appropriate to use such aggregate health indicators, there is no reason to refrain from using more specific health indicators for cost-benefit analysis purposes (for example the expected change in the number of asthma cases amongst children).

Threshold values

A question that often arises when discussing health economics analysis is whether it is theoretically justifiable to recommend that the authorities establish threshold values for maximum acceptable costs per QALY or other health indicators. Such a threshold value would, if applied, be interpreted as a cap on how cost-intensive a measure can be per unit of health gain before its implementation is not recommended or permitted. Threshold values of this type may ensure that the same criterion is applied in different sectors and different contexts when determining which measures to implement. On the other hand, this also means a choice of decision-making criterion that is not necessarily in conformity with the views of the authorities.

If the threshold value is perceived as an estimate of the willingness of the population to pay, there is, in principle, little to distinguish such practice from attributing a monetary value to QALY and performing a cost-benefit analysis. If the threshold value is perceived as an actual cap on the willingness of the authorities to pay, the use of threshold values, or threshold value intervals, may turn such analysis into a decision-making tool, not only a component of a broader basis for making decisions. A practice in which a threshold value is linked to a form of automatic decision, with all measures calculated to have a cost-effectiveness in excess of the threshold value being implemented as a matter of course, has indeed met with some criticism. Such a practice may imply that less weight is accorded to other factors that the population and decision makers would like to emphasise, like e.g. the severity of a disease (NOU 1997:18 Green Paper and Weinstein, 2008). Grosse (2008) shows, in this context, that the willingness to pay for treatments that gain life years exceeds the willingness to pay for treatments for lesser afflictions. A number of authors also argue that the use of threshold values can give producers of pharmaceutical products and medical equipment an opportunity to adapt their prices and thus increase health sector costs (Drummond, 2003, and Grosse, 2008).

10.6 Practices and applications in various sectors and countries

Cost-benefit analysis of measures with an impact on life and health is carried out in many sectors of society. Such is the case both in Norway and in other countries (cf. Ministry of Finance, 2005, and OECD, 2012). However, the manner in which such analysis is performed varies, both in terms of the methods of analysis used and in terms of how the health effects are valued in monetary terms.

The health sector can, in the main, be said to use cost-effectiveness analysis1 with quality-adjusted life years as health indicator or other health indicators that capture changes in the health-related quality of life, cf. e.g. Norwegian Directorate of Health, 2011, and Norwegian Medicines Agency, 2012, whilst the transportation sector principally uses cost-benefit analysis, as a specific method of analysis, with statistical lives as health indicator (cf. e.g. Norwegian Public Roads Administration, 2006). Chapter 10.5 addresses problems and limitations associated with the interpretation and valuation of the terms VSL, VOLY and QALY. However, the discussion in the present Chapter shows that these terms are extensively used, and that such use is not always in line with the theoretical/methodological recommendations.

We will here examine recommended valuations in Norway and other countries, and we have structured the subchapters into prevention and treatment.

10.6.1 Valuations recommended and applied in Norway

10.6.1.1 Prevention of injury or disease

The 2005 edition of the Ministry of Finance cost-benefit analysis guide recommends using NOK 15 million at 2005 prices as the value of a statistical life. It also allows for higher values to be used if each life saved represents many remaining life years. The estimate is based on the European Commission’s VSL estimate of EUR 1.4 million at 2000 prices (approximately NOK 15 million at 2005 prices) for accidents (where the average age at death is about 40 years), and a somewhat lower value of EUR 1 million at 2000 prices (approximately NOK 11 million at 2005 prices) for a case of environmentally-related premature death (where the average age of fatalities is considerably higher), EC DG Environment, (2001). In the latter case, where the old and infirm die a few months, or possibly a few years, earlier than would otherwise have been the case, Ministry of Finance (2005) additionally recommends, as a sensitivity analysis, performing analysis using the value of a statistical life year (VOLY), which may be put at NOK 425,000 at 2005 prices. This figure is obtained from a study commissioned by the UK Department for Environment, Food and Rural Affairs (Chilton et al., 2004). Ministry of Finance (2005) also invites quality-adjustment of such value on the basis of QALY values for relevant diseases. The Norwegian Government Agency for Financial Management (“DFØ”) cost-benefit analysis handbook (2010) uses the Ministry of Finance estimate for VSL (converted into NOK 17.4 million at 2010 prices). In addition, it refers to the Norwegian Directorate of Health’s recommendation of NOK 500,000 at 2005 prices as the value of a statistical life year with perfect health (Norwegian Directorate of Health, 2007).

The transportation sector uses value estimates for risk reduction based on its own surveys. Norwegian Public Roads Administration Handbook 140 (Norwegian Public Roads Administration, 2006) stipulates, for example, a cost of NOK 26.5 million at 2005 prices for one traffic fatality. This cost estimate comprises both real economic costs (medical, equipment and administrative costs, as well as the loss of production imposed on society by accidents) and a welfare loss (the willingness to pay for reduced accident risk). The welfare loss represents 67 percent of the costs associated with fatalities. The loss of production is the value of the production lost as the result of fatalities and injuries, and is calculated on the basis of average labour income, specified by gender and age, less future consumption. The transportation bodies and Avinor have made, on the basis of a comprehensive Norwegian valuation study (Institute of Transport Economics (“TØI”), 2010), recommendations to the Ministry of Transport and Communications as to which unit prices should be used for cost-benefit analysis purposes in the transportation sector. These costs are approved by the Ministry of Transport and Communications. Institute of Transport Economics (“TØI”) (2010) estimates the value of statistical lives at NOK 22 million at 2009 prices, based on a stated choice study, and at NOK 39 million at 2009 prices, based on a contingent valuation study. Institute of Transport Economics (“TØI”) (2010) notes that this type of hypothetical valuation is subject to considerable uncertainty, and that the estimated VSL is approximately NOK 10 million at 2009 prices when the relative value of travel time and VSL are adjusted such as to make the value of travel time match the value of travel time from the value of time study conducted at the same time. Despite considerable variation in estimated VSL based on the various methods mentioned here, Institute of Transport Economics (“TØI”) (2010) concludes that «the current level of the value of statistical lives in the official valuations for the transportation sector, which is approximately NOK 26 million at 2009 prices, can continue to be applied». In order to deal with the uncertainty associated with estimated VSL, Institute of Transport Economics (“TØI”) (2010) indicates an uncertainty of (no less than) 20% in each direction. Approximately NOK 4 million at 2009 prices is added on top of the recommended VSL of approximately NOK 26 million at 2009 prices, to reflect the net loss of production, as well as medical, equipment and administrative costs, thus implying that the accident cost per traffic fatality is estimated at approximately NOK 30.2 million at 2009 prices by the Institute of Transport Economics (“TØI”) (2010).

There are two main reasons why the estimate in Norwegian Public Roads Administration (2006), which was adjusted upwards and recommended for continued application in Institute of Transport Economics (“TØI”) (2010), is about twice as high as the estimate recommended in the Ministry of Finance guide from 2005. These are summarised in Table 10.1. One reason is differences in what is included when estimating VSL. The Ministry of Finance assumes that the VSL from willingness to pay studies includes all benefits society derive from averting the loss of human lives. The Norwegian Public Roads Administration guide assumes that willingness to pay surveys provide data on the willingness of individuals to pay for intangible value and own consumption (referred to as welfare loss), and that tangible value like the net loss of production and medical, equipment and administrative costs (referred to as real economic costs) need to be added (Elvik, 1999). The Norwegian Public Roads Administration guide puts these costs at 33 percent of the cost of fatalities. The other reason is different estimates for the willingness to pay. The Norwegian Public Roads Administration estimates the willingness to pay at NOK 18.3 million at 2005 prices, whilst the Ministry of Finance estimate is NOK 15 million at 2005 prices.

Table 10.1 Elements of Norwegian estimates for, and applications of, the value of a statistical life (“VSL”) in the cost-benefit analysis of accident prevention measures

Elements of VSL

Ministry of Finance cost-benefit analysis guide (2005)

Norwegian Public Roads Administration Handbook 140 (2006)

Willingness to pay for averting a premature fatality

NOK 15 million at 2005 prices

NOK 18.3 million at 2005 prices

Net loss of production imposed on society in general by fatal accidents

Assumed to be included in the willingness to pay

Added on top

Medical costs not covered by individuals

Not discussed

Added on top

Equipment costs not covered by individuals

Not discussed

Added on top

Administrative costs

Not discussed

Added on top

In the report “Health Effects in Cost-Benefit Analysis” (Norwegian Directorate of Health, 2007), the Norwegian Directorate of Health recommends using NOK 500,000 at 2005 prices as the value of a statistical life year in perfect health for purposes of intersectoral cost-benefit analysis. The estimate is derived from the valuation of statistical lives lost in road traffic accidents and the assessment of various degrees of discounting (cf. Box 10.1). The estimate is rounded off to signal high uncertainty. The Norwegian Directorate of Health notes that the estimate does not include medical, equipment and administrative costs, the loss of production or the welfare loss of household members, unlike e.g. the method practised by the Norwegian Public Roads Administration. The Norwegian Directorate of Health estimate of NOK 500,000 at 2005 prices per QALY is used by the Norwegian Public Roads Administration to estimate positive health effects from increased physical activity associated with measures aimed at pedestrians and cyclists (Norwegian Public Roads Administration, 2011). The welfare loss from a minor injury in the transportation sector is estimated at NOK 467,000 at 2009 prices (Institute of Transport Economics (“TØI”), 2010), thus implying that this practice puts the said value at almost a par with the value of a quality-adjusted life year in the health sector.

10.6.1.2 Treatment and alleviation of injury or disease

The extent to which economic analysis is used as a basis for making decisions in the health sector in Norway is variable. To the extent that cost-benefit analysis is used, it is primarily in the form of cost-effectiveness analysis, and not in the form of cost-benefit analysis (as a specific method of analysis), and often with the benefit measured in terms of QALY.

Norway neither has any official QALY valuation, nor any so-called threshold value to determine what shall be classified as cost-efficient or not. However, in health economics analysis it is not uncommon for the analyst to define the cost-effectiveness of a project, measured in NOK per QALY gained, and comparing it to a QALY benchmark. Said benchmark then serves as a threshold defining how much a QALY can cost before the project is held not to be cost-efficient (cf. e.g. Norwegian Knowledge Centre for the Health Services, 2009). Such threshold values may be premised on various lines of reasoning, but can seem fairly random, and will at times lack both theoretical and empirical underpinnings (cf. e.g. Kristiansen, 2003, and Grosse, 2008).

The Norwegian Medicines Agency uses health economics analysis as a basis for deciding whether pharmaceutical products should qualify for pre-approved reimbursement under the government-funded prescription scheme. The pharmaceuticals industry seeks to include pharmaceutical products in the pre-approved reimbursement scheme, and a pharmacoeconomic analysis is required to be included in the applications. Such analyses shall be conducted pursuant to pharmacoeconomic analysis guidelines (Norwegian Medicines Agency, 2012). The guidelines stipulate no explicit economic valuation of life and health. The Norwegian Medicines Agency is authorised to grant pre-approved reimbursement status for pharmaceutical products that are considered cost effective and that fall below a defined expenditure limit. The expenditure limit, which is NOK 5 million in expected annual additional expenditure for the National Insurance System in the fifth reimbursement year, is a budgetary limit and is of no relevance to the cost-effectiveness assessment. Pharmaceutical products that are held to be cost effective and that involve expenditure in excess of the expenditure limit, are referred to the Ministry of Health and Care Services and included in the regular prioritisation decisions in the fiscal budget.

The Norwegian Directorate of Health has issued a consultative draft cost-benefit analysis guide for the health sector (Norwegian Directorate of Health, 2011). It addresses various methods of analysis and economic valuations of life and health for both preventive measures and treatments in the health sector. Norwegian Directorate of Health (2011) also illustrates how the cost-effectiveness of measures can be linked to the severity of the health condition. This was specified, in response to a commission from the National Council for Priority Setting in Health Care, in an evaluation of rare congenital diseases where many life years may be lost. The consultative draft suggested that both cost-benefit analysis, as a specific method of analysis, and cost-effectiveness analysis with QALY as health indicator can be used in the health sector. An economic valuation of NOK 500,000 per QALY was proposed for cost-benefit analysis, as a specific method of analysis, with NOK 100,000-1 million per QALY being proposed for sensitivity analysis purposes. The consultative comments submitted in respect of the guide are hardly indicative of a consensus, and a final version of the Norwegian Directorate of Health cost-benefit analysis guide has not yet been completed.

The Norwegian Knowledge Centre for the Health Services performs health economics analyses as a basis for decision making for the Ministry of Health and Care Services, the health trusts, the Norwegian Directorate of Health, the Norwegian Medicines Agency and the National Council for Priority Setting in Health Care. These analyses often use QALY as a health indicator, but other health indicators are also used if QALY is not available. When drawing conclusions, the Norwegian Knowledge Centre for the Health Services will often compare the estimated cost per QALY to a threshold value of NOK 500,000 per QALY, without this being related to the seriousness of the condition. A measure is often referred to as «cost-efficient» if its estimated cost-effectiveness is below the threshold value. An example is the analysis of the inclusion of vaccination against rotavirus (which causes gastroenteritis) in the Norwegian child immunisation programme (Norwegian Knowledge Centre for the Health Services, 2009). Said analysis concluded that such vaccination is cost-efficient if one includes the loss of production caused by parents being absent from work when their children get ill, and: “The conclusion with regard to the cost-effectiveness of introducing the vaccination is based on a threshold value for the willingness to pay per quality-adjusted life year gained of NOK 500 000.” In addition to illustrating how threshold values are used for health economics analysis purposes, the analysis shows that whether and, if applicable, how loss of production is included in an analysis can have a major impact on findings.

The National Council for Priority Setting in Health Care performs an advisory function on issues relating to quality and prioritisation in health and care services. The Council has advised on a large number of cases, and health economics analysis often forms part of the basis for making decisions. Its advice on the introduction of new health measures is premised on three prioritisation criteria: seriousness, benefits and cost-effectiveness (cf. NOU 1997: 18 Green Paper). In 2012, the Council has, for example, considered advising against the introduction of vaccination against rotavirus infection in the national child immunisation programme on the grounds that it is not a particularly serious disease for children. This was despite the vaccination having been found to be cost-efficient when including the loss of production (cf. the above discussion of threshold value).

10.6.1.3 Distributional considerations and other non-efficiency-related priorities

Cost-benefit analysis only forms part of the basis for making decisions about the prioritisation of resources. It is therefore important to note that the findings from a cost-benefit analysis are not tantamount to a decision/prioritisation. Weight will also be attached to other considerations than cost-effectiveness/economic profitability when making actual decisions.

The Lønning committees (NOU 1987: 23 and NOU 1997: 18 Green Papers) discuss prioritisation in the health sector. The NOU 1997: 18 Green Paper (the Lønning 2 Committee) concludes that three main priorities should apply in the health sector: 1) the severity of the condition, 2) the benefits from the measure, and 3) the cost-effectiveness of the measure. The principles outlined by the Lønning committees have been codified in the health legislation. The principles also mean that the Storting has concluded that cost-effectiveness is not the sole relevant prioritisation criterion within the health sector (cf. overview of the basic principles underpinning prioritisation, Norwegian Directorate of Health, 2012). An assessment exclusively focused on cost-effectiveness and linked to so-called cost-effectiveness “threshold values” through a form of “automatic decision-making”, which thus fails to consider cost-effectiveness alongside the two other criteria, would mean that some relevant priorities are left out.

The weight attached to the severity of health implications is reflected in, for example, objectives like the zero vision in the road sector (”a vision for a transportation system that does not cause any loss of life or permanent injury”), in recommended valuations when a large versus a small number of life years are lost (Ministry of Finance, 2005), and in how seriousness is emphasised in the evaluation of resource use devoted to health measures (National Council for Priority Setting in Health Care).

10.6.2 Valuations recommended and applied in other countries

10.6.2.1 Prevention of injury or disease

A project funded by the European Commission, Directorate-General for Transport and Energy, reviews the use of cost-benefit analysis, as a specific method of analysis, in the evaluation of traffic safety measures (SafetyNet, 2009). The report compares the valuations of a statistical life in various countries (cf. Chart 10.1). Norway’s value of a statistical life is obtained from the Norwegian Public Roads Administration’s Handbook 140 (Norwegian Public Roads Administration, 2006).

Figure 10.1 Official estimates for the value of a statistical life in the traffic sector in selected countries. Unit: Euros at 2002 prices (Purchasing Power Parity-adjusted).

Figure 10.1 Official estimates for the value of a statistical life in the traffic sector in selected countries. Unit: Euros at 2002 prices (Purchasing Power Parity-adjusted).

Source SafetyNet (2009).

All countries that make fairly extensive use of cost-benefit analysis are included in the comparison. The Norwegian Public Roads Administration’s estimate for the value of a statistical life is the second highest in the world, according to this comparison.

The OECD has recently issued a meta analysis on the valuation of the risk of death in the environmental, health and transportation sectors (OECD, 2012). The report is based on data gathered and analysed over a four-year period. The database is, according to OECD (2012), the world’s largest database of studies showing VSL based on stated preferences with regard to a small reduction in the risk of death. Based on the meta analysis, a VSL range of USD 1.8 million to 5.4 million at 2005 prices is recommended for an adult, with a point estimate of USD 3.6 million at 2005 prices (OECD, 2012), as far as the EU 27 member states are concerned. For transfers between countries, it is recommended that VSL be adjusted on the basis of differences in PPP-adjusted GDP, and that an income elasticity of the willingness to pay for VSL of 0.8 be used. Such a transfer to Norwegian conditions results in a range of NOK 20-60 million at 2012 prices, with a point estimate of approximately NOK 40 million at 2012 prices. For reasons of equality, OECD (2012) recommends no income adjustment for uses within a single country.

Willingness to pay for the life of a child is discussed in OECD (2004) in the context of the risk of environmentally-related death. Children do not have the cognitive or financial capacity to state such a value, and if one uses the willingness of parents to pay it is necessary to examine whether altruistic values are included in a consistent manner. OECD (2012) does not recommend general adjustment of VSL on the basis of age. OECD (2012) nonetheless recommends, based on existing empirical studies from Europe and the United States (US EPA, 2003, and OECD, 2010), using a VSL that is 1.5 – 2 times higher for children than for adults.

OECD (2012) provides a survey on how VSL is used in different countries and sectors. In the United States, for example, government bodies responsible for environment, transportation, food and medicines all make use of a different VSL (Hammitt and Robinson, 2011). The transportation sector in the United Kingdom makes use of a VSL corresponding to that used in the transportation sector in Norway (UK Department for Transport, 2009). Both countries add the loss of production, as well as medical and equipment costs (cf. 10.6.1.1).

According to OECD (2012), the focus in most countries that issue official recommendations on the valuation of life and health is on VSL, and not on VOLY. If VOLY is used in cost-benefit analysis, the findings from such analysis may differ from those obtained if using VSL, depending on, inter alia, the average age of the persons who benefit from the risk reduction; see also the discussion in Chapter 10.5 and Box 10.1. In other words, an analysis using VSL and an analysis using VOLY may differ in their findings, even if the number of lives saved is the same and VSL and VOLY have both been correctly measured. The countries using VOLY primarily derive such value from VSL, which involves, inter alia, making assumptions with regard to subjective discount rates, average individuals, etc. (cf. Box 10.1). Only a small number of studies estimate VOLY directly by using contingent valuation or choice experiments. EU countries predominantly use VOLY for sensitivity analysis. OECD (2012) identifies the United Kingdom as an example of a country using VOLY in its main analysis (i.e. not only in sensitivity analysis) of air pollution measures. Moreover, the United Kingdom uses a lower VOLY if the state of health of the target group is highly precarious and the lifespan lost is only a few months. Hence, the use of VOLY in the United Kingdom has a clear parallel in the use of QALY in the analysis of treatments in the health sector, where weight is also attached to the health-related quality of life of the target group. Svensson and Hultkrantz (2012) estimate an economic value of SEK 1.2 million per QALY (range SEK 750,000 – 3.2 million) based on Swedish VSL estimation studies.

10.6.2.2 Treatment and alleviation of injury or disease

Internationally, it is common practice to use cost-effectiveness analysis with quality-adjusted life years (QALY) as a health indicator to conduct economic analysis of health sector measures. International practice has, however, to some extent been more explicit than Norwegian practice when it comes to adopting threshold values defining what constitutes a cost-effective measure. NICE (National Institute for Health and Clinical Excellence) in England does, for example, use a range of GBP 20,000-30,000 per QALY when making its recommendations to the National Health Service in England (NICE, 2010). The NICE Citizens Council has concluded that treatments costing in excess of the upper limit of GBP 30,000 per QALY can be recommended in cases where the health condition is very severe and the treatment is of a lifesaving nature (NICE Citizens Council, 2010). The NICE lower limit is not interpreted in a corresponding manner. Nord (2009) notes that practice in countries like France and Germany is less uniformly focused on cost-effectiveness than the NICE model in England (Pouvourville, 2013, and Caro and Nord et al., 2010).

The World Health Organisation (WHO, 2012) proposes GDP per capita as a basis for grading the cost-effectiveness of health measures. Measures with a cost per QALY gained of less than GDP per capita are thus classified as “very cost effective”. Measures with a cost per QALY gained of between one and three times GDP per capita are categorised as “cost effective”, whilst measures with a cost per QALY gained of more than three times GDP per capita are categorised by the WHO (2012) as “not cost effective”.

Both Grosse (2008) and Weinstein (2008) provide overviews and discuss how USD 50,000 per QALY has become established as a threshold value referred to in health economics analysis contexts in the United States. Both conclude that the provenance of the said figure is unclear. Grosse notes that the amount may have its origin in the actual cost of dialysis and a notion that such measure ought to be offered to Medicare patients with terminal renal failure. Grosse also proposes updating the interval from USD 20,000 – 100,000 per QALY at 1982 prices (Kaplan and Bush, 1982) to USD 40,000 – 100,000 per QALY at 2008 prices. Despite the existence of such threshold value estimates, the use of threshold values denominated in USD per QALY, to determine which health measures are cost effective is outlawed in the United States (US Public Law, 2010).

10.7 The assessments of the Committee

Above, the Committee has interpreted the terms of reference as follows: 1) To what extent could and should economic valuations be used as a basis for decision making in the analysis of measures that may entail major changes to risks associated with life and health, where it may, moreover, be known which individuals are affected by such change in risk? 2) Should the economic values used in such analysis, if any, be the same for all sectors? We will here discuss each of these two issues in turn.

10.7.1 Economic valuation as a basis for decision making

In Chapter 3, the Committee has emphasised that it considers the outcome of economic profitability assessments to form part of the basis for making decisions, and not as an outright decision-making rule. The Committee has also specified in Chapter 3 that the economic profitability of a measure (as calculated in a cost-benefit analysis) cannot necessarily be interpreted normatively, and must primarily be considered a descriptive estimate of the net willingness to pay of the population. The extent to which actual decisions should and shall be based on the net willingness to pay of the population is, on the other hand, a question that falls outside the scope of cost-benefit analysis. It is important to keep these underlying premises in mind when considering the discussion below.

If analysis is to serve as a decision-making tool in the sense that public sector measures are to be prioritised directly on the basis of their estimated economic profitability, it will be of essential importance to value as many effects as possible, also those for which economics does not appear to provide much valuation guidance – because the absence of an explicit valuation of an impact will be tantamount to applying a value of zero. However, the situation is different when analysis is only intended to provide decision makers with the best possible basis for making decisions. The primary function of analysis will then be, as stated in the NOU 1997: 27 Green Paper (p. 6), “toclarify andelucidate the consequences of alternative measures”. The issue of how far one should go in attributing value to effects must therefore be considered in view of the extent to which economic valuation contributes to enhanced understanding of the consequences of the measure on the part of decision makers.

If one refrains from valuing life and health impacts in monetary terms, cost-benefit analysis can be carried out in the form of cost-effectiveness analysis or cost-effect analysis (see Chapter 2). If all consequences are valued and one conducts a full cost-benefit analysis, as a specific method of analysis, the findings will provide an indication as to the net willingness of the population to pay for the various measures, although this cannot readily be interpreted as a normative specification of the relevant measure’s contribution to society’s welfare.

A reasonable implication of these considerations is that monetary valuation of life and health impacts, for purposes of including these in cost-benefit analysis, as a specific method of analysis, should be limited to cases where theoretical and empirical evidence justifies an assumption to the effect that the chosen values reflect, directly or indirectly, the marginal willingness of the population to pay for the relevant impacts. Other valuation principles than aggregate individual willingness to pay, for example government-stipulated cost limits, will measure something else, and hence not be compatible with the interpretation of cost-benefit analysis, as a specific method of analysis, adopted in the present Report.

We will therefore, in the discussion below, attach weight to whether we find theoretical and empirical support for estimating the willingness of the population to pay for life and health impacts, also in the health sector. For those impacts where the Committee is of the view that such theoretical and empirical support does not exist or is, in our assessment, too weak, there will be a strong case for using cost-effectiveness analysis or cost-effect analysis, inasmuch as it will be very difficult to interpret the findings from a cost-benefit analysis (as a specific method of analysis).

If economic valuation is theoretically and empirically justified, it nevertheless remains to consider whether such valuation will provide decision makers with a better understanding of the impacts of alternative measures than can be achieved through forms of analysis that do not involve such valuation (cf. the fact that it is not necessary to value impacts in order to perform cost-benefit analysis, in the more general sense of the term used in the present Report). Whether economic valuation and cost-benefit analysis, as a specific method of analysis, offer a better basis for making decisions than does cost-effectiveness analysis, or cost-effect analysis, will need to be examined in each individual case.

VSL

The value of a statistical life, VSL, is measured, as noted above, by aggregation of the willingness of many individuals to pay for a small risk reduction. It is therefore (cf. Chapter 10.5) not necessarily unproblematic to use VSL to estimate the willingness to pay for lifesaving measures that involve non-marginal risk reductions. This is of particular concern for groups with a different risk level than the general population to begin with, and/or where the lives saved are identifiable, in full or in part. Many measures within the health sector will be characterised by exactly such circumstances.

If a sufficiently large number of individuals experience a minor reduction in their risk of death (probability of premature death), these small risk reductions will in aggregate mean that society can expect fewer premature fatalities. However, the willingness of people to pay for small risk reductions will not necessarily be in conformity with their willingness to pay for lifesaving as such (i.e. saving the lives of identified individuals), and this imposes certain restrictions on the use of VSL in the cost-benefit analysis of lifesaving measures in general.

A lifesaving measure may, for example, primarily relate to other people, whilst VSL is usually estimated on the basis of a marginal reduction in own risk. It may be known which persons’ lives the measure is expected to be able to save, or who has a particularly high probability of being affected. The risk reduction for those involved may be non-marginal (for example from highly probable death to highly probable recovery). Their risk level to begin with may be very different from the risk level of a representative selection of the population. If one seeks to estimate the willingness to pay for reduced risk in groups that are already afflicted, or groups that have a higher risk than the “average individual” in the overall population, all these factors may affect their willingness to pay, at times considerably so. Using VSL to carry out cost-benefit analysis, as a specific method of analysis, of lifesaving measures may nevertheless be of interest in such cases as well. One may, in order to determine the willingness to pay to be used in such cost-benefit analysis, assume that alternative measures are examined by adopting a fairly peculiar and hypothetical approach, i.e. what is often termed «behind the veil of ignorance» (cf. Chapter 10.5.). This involves envisaging that one is not yet aware of one’s own identity or how one will fare in life, and hence does not know whether one will become a cancer patient, have children, end up in a risk group for cardiovascular disease, etc. Since neither the decision makers, nor the remainder of the population, are in actual fact behind any “veil of ignorance”, it must nevertheless be emphasised that the findings from such a willingness to pay assessment cannot be considered a measure of the actual net willingness of the affected population to pay in the situation in which the decision is in fact made.

VOLY

It is, as explained above, fairly straightforward to aggregate the willingness of individuals to pay for their own marginally reduced risk of death into a willingness to pay for a reduced number of expected premature fatalities in the population as a whole (“VSL”). It is, on the other hand, somewhat less clear how one might derive the value of a statistical life year (“VOLY”) from the value of a statistical life. In estimating VSL, people are usually asked about their willingness to pay for a reduced risk of death, but not about how much relative weight they attach to each of the life years gained (their pure rates of time preference). Consequently, it is not obvious how best to convert VSL into VOLY.

The formula in Box 10.1 is based on a number of strict assumptions. Consequently, it is not obvious how best to convert VSL into VOLY. However, there is, generally speaking, no reason to expect the values of remaining life years for different age brackets to exhibit the structure indicated by the conversion from VSL to VOLY in Box 10.1. Consequently, there is no good theoretical or empirical justification for deriving VOLY estimates from a VSL estimated over all age brackets. The relationship between a marginally reduced risk of death and an expected life saved for society is more straightforward, and thus requires assumptions that are less strict, than the relationship between a marginally reduced risk of death and an expected life year saved for society.

VOLY estimates must instead be based on surveys of the willingness to pay for reduced risk within the same age bracket. However, general methodological problems associated with willingness to pay studies suggest that the validity of such valuations is open to doubt (cf. Chapter 10.5), and one may experience different valuations of identical health effects and identical valuations of different health effects if findings from such valuation studies are used for analysis purposes (cf. Chapter 10.6.1).

Identifying a VOLY estimate that may serve as an intersectoral standard for all age brackets is a task that involves as yet unresolved ethical, theoretical and empirical problems.

QALY

As discussed in Chapter 10.2, it is not necessary to calculate any monetary value for QALY in order to use it as a benefit indicator in a cost-effectiveness analysis. There are various methods for calculating QALY, which often attach weight to different factors and which are often based on different principles (cf. e.g. the NOU 1997: 27 Green Paper and Olsen, 2009). Different measurement methods may reflect what are, in principle, different ways of posing questions. QALY as a unit of measurement in relation to changes in life and health has been criticised from many quarters. Nor is there a full consensus amongst researchers within the field as to which method is preferable. Calculation of QALY involves, inter alia, having to assess how important different types of afflictions and disabilities are in relation to each other, and such choices may of course be controversial. QALY should therefore be considered an indicator, rather than an exact measurement method for health benefits.

In principle, it is possible to determine the willingness to pay for a reduction in the risk of a specific health impairment for a specific number of years (although it has turned out to be challenging in practice, and it also remains subject to the assumptions of small risks and large populations, cf. Chapter 10.5). If the willingness to pay for various states of health for a specific period of time is determined, it is also possible, in principle, to normalise such willingness by reference to a given state of health (for example the best one, with all states of health being represented by an index between zero and one). However, there is, generally speaking, no reason to expect the willingness to pay for a risk-reducing measure to have a linear form, thus enabling it to be inferred from a time-independent index and the duration of the state of health, as (implicitly) assumed in the economic valuation of quality-adjusted life years in Box 10.1.

The value of a QALY has been sought estimated through willingness to pay studies, but it may generally be noted that the validity of willingness to pay studies has been questioned and that the estimates are deemed to be highly uncertain (cf. Chapter 10.5). It is also established practice to convert VSL, typically via VOLY, into the value of a QALY. Such practice is premised on a number of highly simplified assumptions; see Box 10.1. Firstly, it is assumed that VSL does in fact value life, and not risk reduction. In addition, it is not obviously the case that the willingness to pay for health changes can be assumed to be related to the willingness to pay for life in a straightforward manner. Consequently, the conversion from VSL to valuation of QALY would appear to add further uncertainty.

Estimating the value of a QALY, either based on estimates from willingness to pay studies or converted from VSL estimates, is associated with both theoretical and empirical weaknesses. Consequently, there is no theoretical and empirical basis for stipulating any recommended economic value of a QALY.

10.7.2 Should the economic values used, if any, be the same for all sectors?

Cost-benefit analysis, as a specific method of analysis, is intended to provide a measure of the net willingness to pay. In order for this interpretation to be as correct and comparable as possible across analyses from different sectors, identical consequences should be evaluated identically in different sectors. The key consideration is for the findings to be understandable and fairly straightforward to interpret for decision makers, thus enabling the analysis to make a real contribution to improved decisions.

Methodological problems discussed above imply that theoretical and empirical evidence does not provide much support for establishing recommended VOLY and QALY figures. Attaching an economic value to a statistical life appears less challenging, from a theoretical perspective, than attaching an economic value to statistical life years or quality-adjusted life years. In addition, the availability of studies valuing statistical lives is better, as reflected in the existence of meta analyses (cf. e.g. OECD, 2012). Incidentally, OECD (2012) recommends carrying out national willingness to pay studies, rather than transferring values from other countries.

When it is unclear whether economic values incorporated in a cost-benefit analysis actually reflect the willingness of the population to pay, the findings from such analysis are difficult to interpret, and thus it also becomes unclear how said analysis can contribute to improved decisions. However, the Committee observes that valuations of life and health, in the form of VSL, VOLY and valuations of QALY, are in practice included in cost-benefit analysis in different ways in different sectors. This is, generally speaking, unfortunate.

If one wishes to make use of economic values for the willingness to pay for statistical life years and QALY, despite the difficulties of estimating these, both in practice and in principle, and wants such values to be reasonably consistent with the VSL estimate one uses, it is likely, in practice, that one will engage in some sort of calibration, with VOLY and the value of a QALY being derived on the basis of VSL (cf. Box 10.1), despite such calibration resting on a number of critical assumptions.

The Committee finds it theoretically and empirically justifiable to recommend an intersectoral VSL estimate (see below for further details). The Committee does not, on the other hand, find it theoretically and empirically justifiable to recommend an intersectoral standard estimate for VOLY or the value of QALY. This is partly because we find that the theoretical basis for such a valuation is weak, and partly because the empirical basis for estimating the willingness to pay appears to be considerably weaker for VOLY and QALY than for VSL.

We take the opportunity to reiterate, in this context, that there is no theoretical justification for interpreting VSL as a willingness to pay for lives as such. The latter implies that if such an interpretation is nevertheless applied, the informational value of the valuation will be limited, and thus generate findings that are difficult to interpret for decision makers. This is an argument in favour of attaching more weight to cost-effectiveness analyses and/or cost-effect analyses than to cost-benefit analysis, as a specific method of analysis, for projects where lives saved are an important consequence and the prerequisites for using VSL (small risk reductions and unidentified individuals) are not in place.

VSL

Institute of Transport Economics (“TØI”) (2010) estimates the value of statistical lives in relation to reduced risk of road transport accidents in Norway under various hypothetical valuation methods. A stated choice study estimates VSL at NOK 22 million at 2009 prices and a contingent valuation study estimated VSL at NOK 39 million at 2009 prices. Institute of Transport Economics (“TØI”) (2010) notes that there is considerable uncertainty associated with this type of hypothetical valuation, and observes that the estimated VSL is approximately NOK 10 million at 2009 prices when the relative value of travel time and VSL are adjusted to bring the value of travel time into conformity with the value of travel time from the value of time study carried out at the same time. Such calibration can be considered an attempt at estimating a VSL with a more «correct» relative value, and thus possibly less influenced by hypothetical bias (cf. e.g. Sælensminde, 2003). The VSL estimate from the stated choice study, at NOK 22 million at 2009 prices, which values reduced risk simultaneously with reduced travel time, does in itself represent a method that attempts to estimate relative valuations. This is in contrast to the VSL estimate from contingent valuation, at NOK 39 million at 2009 prices, which values reduced risk in isolation, without relating it to travel time. Institute of Transport Economics (“TØI”) (2010) concludes, despite considerable variation in estimated VSL based on the three different methods mentioned here, that «the current level of the value of statistical lives in the official valuations for the transportation sector, which is approximately NOK 26 million at 2009 prices, can continue to be applied». In order to deal with the uncertainty associated with estimated VSL, Institute of Transport Economics (“TØI”) (2010) indicates an uncertainty of (no less than) 20% in each direction. Approximately NOK 4 million at 2009 prices is added on top of the recommended VSL of approximately NOK 26 million at 2009 prices, to reflect the net loss of production, as well as medical, equipment and administrative costs, thus implying that the accident cost per traffic fatality is estimated at approximately NOK 30.2 million at 2009 prices.

OECD (2012) recommends, based on a meta analysis, a VSL range of USD 1.8 million to 5.4 million at 2005 prices for an adult, with a point estimate of USD 3.6 million at 2005 prices (OECD, 2012), as far as the EU 27 member states are concerned. For transfers between countries it is recommended that VSL be adjusted on the basis of differences in PPP-adjusted GDP, and that an income elasticity of demand of 0.8 be used. Such a transfer to Norway results in a range of NOK 20-60 million at 2012 prices, with a point estimate of approximately NOK 40 million at 2012 prices. OECD (2010) recommends using a VSL for children that is 1.5 – 2 times higher than the average VSL.

The NOU 1997: 27 Green Paper examined whether production gains should be added on top of the estimated willingness to pay for VSL (cf. the overview in Chapter 10.3). The NOU 1997: 27 Green Paper concluded that the theoretical basis for such a supplement was unclear. The Committee relies on the conclusions from the NOU 1997: 27 Green Paper in this respect, and assumes that any net production gains can be held to be included in the measured willingness to pay. This interpretation means that the Institute of Transport Economics (“TØI”) (2010) willingness to pay estimate from a stated choice study, in the amount of NOK 22 million at 2009 prices, should be considered as an estimate of the willingness to pay inclusive of production gains. Adjusted for GDP per capita developments (cf. Chapter 4 on income adjustment of the willingness to pay over time) and measured at 2012 prices, such estimate becomes somewhat higher. Besides, the Institute of Transport Economics (“TØI”) (2010) estimate is at the lower end of the VSL range estimated for Norway on the basis of the values from OECD (2012). An upwards adjustment of the estimate to NOK 30 million at 2012 prices will to some extent compensate for part of the uncertainty associated with the estimates.

OECD (2012) recommends using a higher value of statistical lives in the analysis of measures targeting children than in the analysis of measures targeting adults (cf. Chapter 10.6.2). Ministry of Finance (2005) also allows for this. This is based on the premise that the willingness to pay for risk-reducing measures targeting children exceeds the willingness to pay for such measures targeting adults.

VOLY

The discussion in the present Chapter shows that conversion of VSL into VOLY is based on unreasonably strict assumptions. In addition, the estimates of the willingness to pay for VSL are subject to considerable uncertainty. Moreover, the empirical foundation for basing value estimates directly on surveys of the willingness to pay for VOLY is weak. This means, all in all, that the Committee does not find theoretical and empirical support for recommending a standard VOLY estimate.

A weak theoretical and empirical basis for estimating the willingness to pay for statistical life years, implies that a monetary value that is nonetheless attributed to statistical life years will be of limited informational value and be difficult to interpret for decision makers. This is an argument in favour of attaching more weight to cost-effectiveness analyses and/or cost-effect analyses than to cost-benefit analysis, as a specific method of analysis, whenever the specific nature of the measure indicates that statistical life years is an appropriate unit of measurement.

QALY

The discussion in the present Chapter shows, as was the case with VOLY, that conversion of VSL into an economic value of a QALY is based on unreasonably strict assumptions. Furthermore, the estimates of the willingness to pay for VSL are subject to considerable uncertainty and the empirical foundation for basing estimates for the value of a QALY directly on surveys of the willingness to pay is weak.

Empirical evidence suggests that the willingness to pay depends on seriousness (cf. Chapter 10.6). If one arrives, despite a weak theoretical and empirical basis, at an economic value of a QALY (cf. Box 10.1) that may seem consistent with VSL in terms of magnitude, one cannot necessarily assume that such an imputed value of a QALY is applicable to a measure irrespective of factors like, for example, the number of life years gained. Moreover, one needs to be aware, as a general observation, that effect measurements like QALY are variables that only provide an indication as to the magnitude of the impacts, and not precise measurement instruments. This means, all in all, that the Committee does not find theoretical and empirical support for recommending a standard value for a QALY. As with VOLY, this is an argument in favour of attaching more weight to cost-effectiveness analyses and/or cost-effect analyses than to cost-benefit analysis, as a specific method of analysis, whenever the specific nature of the measure indicates that quality-adjusted life years are an appropriate unit of measurement.

Threshold values

The values discussed above are based on the willingness of the population to pay. Such values would have been a good basis for establishing threshold values if the authorities wanted to prioritise resource use based on precisely what the population is willing to pay for. However, this principle is unlikely to be in conformity with the principles laid down in the report of the Lønning 2 Committee. Besides, willingness to pay is not necessarily, as noted in Chapter 3, aligned with welfare, and it can be argued that willingness to pay attaches relatively minor weight to benefits accruing to low-income groups. Consequently, it is unlikely to be theoretically justifiable to recommend that the Norwegian authorities should apply the values discussed here as caps on the cost per unit of health gain.

It is likely that a VSL estimate can nonetheless serve as an interesting point of reference for the analysis of measures within the health sector and other sectors. If, for example, one is able to identify measures with a cost per expected life saved that is considerably less than what the population is assumed to be willing to pay, this will indicate that one may be able to reap major gains at a low cost. What information decision makers should attach weight to when making their final decisions must, however, be deemed to fall outside the scope of the work of this Committee.

10.8 Summary recommendations

  • Health indicators will have to be chosen on the basis of the specific character of the public measures in question. It will, for example, be more appropriate to use statistical life years than statistical lives when expected remaining life years deviate sharply between alternative measures. Correspondingly, it will be more appropriate to use indicators for quality-adjusted life years when improved health-related quality of life is an important consequence. It may also be relevant to use specific health indicators.

  • It is not necessary to attribute an economic value to the health indicators statistical lives, statistical life years or quality-adjusted life years in order to include these in cost-effectiveness analysis or cost-effect analysis.

  • It is proposed an economic value of a statistical life (VSL) at NOK 30 million at 2012 prices. It is recommended that this be applied to all sectors (cf. intersectoral standard in the terms of reference).

  • In the analyses of measures specifically targeting the safety of children one may apply, by way of supplementary analysis, a higher value of a statistical life than for the general population. An appropriate level is twice the VSL of the general population.

  • In principle, the value of equivalent consequences should be the same irrespective of sector, also for other health-related benefit indicators, like value of a statistical life year (VOLY) and quality-adjusted life years (QALY). However, the Committee is of the view that the technical basis for estimating the willingness to pay for these is currently not sufficiently established to merit the recommendation of intersectoral standard values for VOLY and QALY.

  • It is proposed that the economic value of VSL be adjusted in line with the growth in GDP per capita (cf. Chapter 4 on real price adjustments).

  • For measures where effects on life and health represent a main consequence, especially where the measures imply significant risk changes for individuals and/or where the identity of those especially affected is known, it will often be more appropriate to use cost-effectiveness analysis or cost-effect analysis than cost-benefit analysis (as a specific method of analysis).

10.9 Bibliography

Andersson, H. & Treich, N. (2011). The value of a statistical life. In De Palma et al. (ed.), A Handbook of Transport Economics. Edward Elgar Publishing.

Arnesen, T. M. & Norheim, O. F. (2003). Quantifying quality of life for economic analysis: time out for time trade off. Med Humanities, 29, pp. 81-86.

Augestad, L. A. et al. (2012). Learning Effects in Time Trade-Off Based Valuation of EQ-5D Health States. Value in Health, 15 (2), pp. 340-345.

Bobinac, A. et al. (2010). Willingness to Pay for a Quality-Adjusted Life-Year: The Individual Perspective. Value in Health, 13 (8), pp. 1046-1055.

Bobinac, A. et al. (2012). Get more, pay more? An elaborate test of construct validity of willingness to pay per QALY estimates obtained through contingent valuation. Journal of Health Economics, 31 (1), pp. 158-168.

Boardman, A. E., Greenberg, D. H., Vining, A. R. & Weimer, D. L. (2011). Cost-benefit analysis: Concepts and practice. Fourth edition. Prentice Hall.

Broome, J. (1978). Trying to value a life. Journal of Public Economics, 9, pp. 91-100.

Caro, J. & Nord, E. et al. (2010). The efficiency frontier approach to economic evaluation of health-care interventions. Health Economics, 19 (10), pp. 1117-1127.

Chilton, S., Covey, J., Jones-Lee, M., Loomes, G. & Metcalf, H. (2004). Valuation of Health Benefits Associated with Reductions in Air Pollution. London: DEFRA.

Dolan, P., Edlin, R., Tsuchiya, A., Armitage, C., Hukin, A., Brazier. J., Ibbotson, R., Bryan, S., Eiser, D. & Olsen, J. A. (2008). The relative societal value of health gains to different beneficiaries. Report for the NIHR Methodology Programme, RM03/JH11, www.hta.ac.uk/nihrmethodology/outlines/1577.pdf

Dolan, P., Tsuchiya, A., Armitage, C., Brazier, J., Bryan, S., Eiser, D., Olsen, J. A., & Smith, M. (2003). What is the Value to Society of a QALY? A protocol to determine the relative value of a QALY according to various health and non-health characteristics. www.publichealth.bham.ac.uk/nccrm/publications.htm.

Drummond, M. (2003). The societal value of a QALY – Where science and decision-making find it difficult to meet. In Bringedal, Iversen & Kristiansen (eds.), The value of life and health. How much should society be willing to pay for health improvements? (English language article in Norwegian language publication. Norwegian title of publication: Verdien av liv og helse. Hvor mye bør samfunnet være villig til å betale for helseforbedringer?) Occasional Papers 2003: 6, HERO, University of Oslo.

Drummond, M. et al. (2005). Methods for the economic evaluation of health care programmes. Third Edition. Oxford University Press.

Drummond, M. et al. (2009). Toward a consensus on the QALY. Value in Health, 12, Supplement 1, pp. 31-35.

EC DG Environment (2001). Recommended interim values for the value of preventing a fatality in DG environment cost benefit analysis. Report from EU DG Environment.

Elvik, R. (1993). Economic valuation of welfare loss caused by traffic accidents (In Norwegian only. Norwegian title: Økonomisk verdsetting av velferdstap ved trafikkulykker). Institute of Transport Economics (“TØI”) Report 203/93. Institute of Transport Economics, Oslo.

Elvik, R. (1999). Is the Cost Calculation Committee on the wrong track? (In Norwegian only. Norwegian title: Er kostnadsberegningsutvalget på villspor?) Samferdsel, No. 7, 1999.

Frederik, S., Loewenstein, T. & O'Donoghue, J. (2002). Time Discounting and Time Preference: A Critical Review. Journal of Economic Literature, 40 (2), pp. 351-401.

Grosse, S. D. (2008). Assessing cost-effectiveness in healthcare: History of the $50,000 per QALY threshold. Expert Rev. Pharmacoeconomics Outcomes Res., 8 (2), pp. 165-178.

Gyrd-Hansen, D. (2003). Willingness to pay for a QALY. Health Economics, 12, pp. 1049–1060.

Hammitt, J. K. (2002). QALYs Versus WTP. Risk Analysis, 22 (5), pp. 985-1001.

Hammitt, J. K. & Robinson, L. A. (2011). The income elasticity of the value per statistical life: Transferring estimates between high and low income populations. Journal of Benefit-Cost Analysis, 2 (1), pp. 1-27.

Hammitt, J. K. & Treich, N. (2007). Statistical vs. identified lives in benefit-cost analysis. Journal of Risk and Uncertainty, 35, pp. 45-66.

Harsanyi, J. C. (1955). Cardinal welfare, individualistic ethics, and interpersonal comparison of utility. Journal of Political Economy, 63, pp. 309-321.

Hendriksen, K. (2012). EQ5D – not as good as it is made out to be? – and what might we do about it? (In Norwegian only. Norwegian title: EQ5D - dårligere enn sitt rykte? - og hva kan vi eventuelt gjøre med det?) Presentation at the Health Economics Conference 2012, Sundvolden.

Hultkrantz, L. & Svensson, M. (2008). The Value of Lives (In Swedish only. Swedish title: Värdet av liv). Ekonomisk Debatt, 36 (2), pp. 5-16.

Hultkrantz, L. & Svensson, M. (2012). The value of a statistical life in Sweden. A review of the empirical literature. Forthcoming in Health Policy.

Institute of Transport Economics (2010). The Norwegian Valuation Study – Summary Report (In Norwegian only. Norwegian title: Den norske verdsettingsstudien – sammendragsrapport). Institute of Transport Economics (“TØI”) Report 1053/2010, Institute of Transport Economics, Oslo.

Johnson, F. R. (2012). Why not real economics? Pharmacoeconomics, 30 (2), pp. 127-131.

Kaplan, R. M. & Bush, J. W. (1982). Health-related quality of life measurement for evaluation research and policy analysis. Health Psych., 1, pp. 61-80.

Kristiansen, I. S. (2003). How much should society be willing to pay for health improvements? In Bringedal, Iversen & Kristiansen (eds.), The value of life and health. How much should society be willing to pay for health improvements? (Norwegian language article in Norwegian language publication. Norwegian title of article: Hvor mye bør samfunnet være villige til å betale for helseforbedringer? Norwegian title of publication: Verdien av liv og helse. Hvor mye bør samfunnet være villig til å betale for helseforbedringer?) Occasional Paper 2003: 6, HERO, University of Oslo.

Ministry of Finance (2005). Cost-Benefit Analysis Guide (In Norwegian only. Norwegian title: Veileder i samfunnsøkonomiske analyser).

NICE (2010). Measuring effectiveness and cost effectiveness: the QALY. National Institute for Health and Clinical Excellence, 20 April 2010. http://www.nice.org.uk/newsroom/features/measuringeffectivenessandcosteffectivenesstheqaly.jsp.

NICE Citizens Council (2010). Citizens Council report on departing from the threshold published. National Institute for Health and Clinical Excellence, 12 June 2010. http://www.nice.org.uk/newsroom/features/CitizensCouncilReport.jsp

Nord, E. (2009). Economic evaluation of health measures; Theory, practice and value issues. In Haug, Kaarbøe & Olsen (eds.). A health service without borders (Norwegian language article in Norwegian language publication. Norwegian title of article: Økonomisk evaluering av helsetiltak; Teori, praksis og spørsmål om verdier. Norwegian title of publication: Et helsevesen uten grenser). Cappelen Academic Publishers, 2009.

Norwegian Directorate of Health (2007). Health Effects in Cost-Benefit Analysis (In Norwegian only. Norwegian title: Helseeffekter i samfunnsøkonomiske analyser). Report IS-1435, Norwegian Directorate of Health.

Norwegian Directorate of Health (2011). Cost-Benefit Analysis in the Health Sector – A Guide (In Norwegian only. Norwegian title: Samfunnsøkonomiske analyser i helsesektoren – en veileder). Consultative draft. Report IS-1915, Norwegian Directorate of Health.

Norwegian Directorate of Health (2012). Health Sector Prioritisation – Basic Principles, Status and Challenges (In Norwegian only. Norwegian title: Prioriteringer i helsesektoren – verdigrunnlag, status og utfordringer). Report IS-1967, Norwegian Directorate of Health.

Norwegian Government Agency for Financial Management (“DFØ”) (2010). Cost-Benefit Analysis Handbook (In Norwegian only. Norwegian title: Håndbok for samfunnsøkonomiske analyser). Norwegian Government Agency for Financial Management, Oslo.

Norwegian Knowledge Centre for the Health Services (2009). Cost-effectiveness of including vaccination against rotavirus in the Norwegian child immunisation programme (In Norwegian only. Norwegian title: Kostnadseffektivitet av å inkludere vaksinasjon mot rotavirus i det norske barnevaksinasjonsprogrammet). Norwegian Knowledge Centre for the Health Services Report No. 31–2009.

Norwegian Medicines Agency (2012). Pharmacoeconomic Analysis Guidelines (In Norwegian only. Norwegian title: Retningslinjer for legemiddeløkonomiske analyser). Valid from 1 March 2012. Norwegian Medicines Agency, Oslo.

Norwegian Public Roads Administration (2006). Handbook 140 Impact Assessments (In Norwegian only. Norwegian title: Håndbok 140 Konsekvensanalyser). Norwegian Public Roads Administration, Oslo.

Norwegian Public Roads Administration (2011). Forwarding letter – New unit prices for cost-benefit analysis (In Norwegian only. Norwegian title: Oversendelsesbrev – Nye enhetspriser for samfunnsøkonomiske analyser), letter to the Ministry of Transport and Communications, 24 January 2011.

NOU 1987: 23 Green Paper; Guidelines for Prioritisations within the Norwegian Health Service (In Norwegian only. Norwegian title: Retningslinjer for prioriteringer innen norsk helsetjeneste). Oslo: Government Administrative Services, 1987.

NOU 1997: 18 Green Paper; Prioritisations revisited. Review of Guidelines for Prioritisations within the Norwegian Health Service (In Norwegian only. Norwegian title: Prioritering på ny. Gjennomgang av retningslinjer for prioriteringer innen norsk helsetjeneste). Oslo: Government Administrative Services, 1997.

NOU 1997: 27 Green Paper; Cost-Benefit Analysis. Principles for Profitability Assessments in the Public Sector (In Norwegian only. Norwegian title: Nytte-kostnadsanalyser. Prinsipper for lønnsomhetsvurderinger i offentlig sektor). Oslo: Government Administrative Services, 1997.

NOU 1998: 16 Green Paper; Guidance on Using Profitability Assessments in the Public Sector (In Norwegian only. Norwegian title: Veiledning i bruk av lønnsomhetsvurderinger i offentlig sektor). Oslo: Government Administrative Services, 1998.

OECD (2004). The Valuation of Environmental Health Risk to Children: Synthesis Report. OECD, Paris.

OECD (2010). Valuing environmental-related health impacts – with special emphasis on children (VERHI-Children), Final report, OECD, Paris.

OECD (2012). Mortality risk valuation in environment, health and transport policies. OECD, Paris.

Olsen, J. A. (2009). Principles in health economics and policy. Oxford University Press, 2009.

Pinto-Prades, Loomes, G. & Brey, R. (2009). Trying to estimate a monetary value for the QALY. Journal of Health Economics, 28 (3), pp. 553-562.

Pouvourville,G. (2013). Pricing and reimbursement for drugs in France: What role for cost-effectiveness analysis? Forthcoming in Schlander, M. (ed.) Economic evaluation of health care programs. Springer Verlag.

Rawls, J. (1971). A Theory of Justice. Cambridge, Mass. Harvard University Press

SafetyNet (2009). Cost-benefit analysis. http://ec.europa.eu/transport/road_safety/specialist/knowledge/pdf/cost_benefit_analysis.pdf. Downloaded on 7 August 2012.

Sculpher, M. & Claxton, K. (2012). Real economics needs to reflect real decisions. A response to Johnson. Pharmacoeconomics, 30 (2), pp. 133-136.

Smith, R. D. & Richardson, J. (2005). Can we estimate the social value of a QALY: Four core issues to resolve. Health Policy, 74 (1), pp. 77-84.

Svensson, M. & Hultkrantz, L. (2012). The Willingness to Pay for a QALY. Results from value of life estimates in Sweden. Örebro University Business School, Dept. of Economics, Working Paper.

Sælensminde, K. (2003). Embedding effects in valuation of non-market goods. Transport Policy, 19 (1), pp. 59-72.

UK Department for Transport (2009). Guidance Document – Expert, TAG unit 3.4.1: The Accidents Sub-Objective. UK Department for Transport (DfT), London, July 2009.

US EPA (2003). Children’s health valuation handbook. US, EPA, Washington, D.C.

US Public Law (2010). The Patient Protection and Affordable Care Act. Public Law 111-148. 3-23-2010, United States.

Weinstein, M. C. (2008). How much are Americans willing to pay for a quality-adjusted life year? Medical Care, 46 (4), pp. 343-345.

WHO (2012). Cost-effectiveness threshold. http://www.who.int/choice/costs/CER_thresholds/en/index.html. Downloaded on 9 August 2012.

Footnotes

1.

Using cost-benefit analysis, as a specific method of analysis, instead of cost-effectiveness analysis has also been discussed in the health sector (cf. e.g. Norwegian Directorate of Health, 2011, Johnson, 2012, and Sculpher and Claxton, 2012).

Go to front page