SP5A-CT-2007-044172

European Value of a Quality Adjusted Life Year

Instrument: Specific Targeted Research Project

Final Publishable Report

Period covered: 1 March 2007 to 31 August 2010 Project co-ordinator: Professor Cam Donaldson Project co-ordinator organisation: Newcastle University Project website: http://research.ncl.ac.uk/eurovaq/

1

Contents Page List of authors 5 Project summary 7

Chapter 1 Introduction 1.1 Policy relevance 9 1.2 State of the art 11

Chapter 2 Estimating a WTP-based monetary value of a QALY from existing contingent valuation studies of the value of prevented fatalities and serious injuries

17

2.1 Introduction 17 2.2 Data Sources 18 2.2.1 Types of data required 18 2.2.1.1 Estimating the Value of a Prevented Fatality 18 2.2.2 Adjusting the VPF 18 2.2.2.1 Estimating remaining life expectancy 19

2.2.2.2 Estimating a value per quality adjusted life year 19 2.2.2.3 Discounting 19

2.2.2.4 Value of a Serious Injury 20 2.3 Possible approaches to converting the value of a

statistical life year into a monetary value of a QALY 20

2.3.1 Approach 1 20 2.3.2 Approach 2 22 2.3.2.1 Approach 2a 25

2.3.2.2 Approach 2b 27 2.4 Results 29 2.5 Discussion 31 2.5.1 Summary of results 31 2.5.2 Missing data 31 2.5.3 Caveats 32 2.5.4 Comparisons with current „threshold‟ values 32 2.6 Conclusion 33

Chapter 3 Estimating a WTP-based value of a QALY through survey research: methods and preliminary results

35

3.1 Introduction 35 3.2 Conceptual overview 35 3.2.1 Chained 35 3.2.2 Direct 36 3.2.3 Complementarities and overlaps across both surveys 37

2

3.3 Questionnaire development 39 3.4 Sampling 39 3.5 Chained approach 44

3.5.1 The utility assessment component 44 3.5.1.1 The standard gamble procedure 44

3.5.1.2 The time trade-off procedure 47 3.5.2 The WTP component 50

3.5.2.1 Risk variant WTP questions 50 3.5.2.2 Time variant WTP questions 52

3.5.2.3 Additional WTP questions 53 3.5.2.4 The card sort procedure 54

3.5.2.5 Unwilling to pay anything at all 55 3.5.3 The chained questionnaire 56

3.5.3.1 The eight versions 56 3.6 Direct approach 57

3.6.1 Study design for direct questionnaire 57 3.6.2 Direct questionnaire 58

3.6.2.1 WTP questions 60 3.6.2.2 The four versions and grey block 60

3.6.2.2.1 Version 1 60 3.6.2.2.2 Version 2 60

3.6.2.2.3 Version 3 61 3.6.2.2.4 Version 4 61

3.6.2.2.5 Grey block 61 3.6.2.3 Exclusion criteria 61

3.6.2.4 Respondents unwilling to pay 62 3.6.2.5 The card sort procedure 63 3.7 Analytical strategy 64 3.7.1 The „preliminary case‟ 64

3.7.2 Sensitivity analyses 65 3.7.3 Consistency tests 65 3.8 Statistical and regression analyses 66 3.8.1 Chained approach 66

3.8.2 Direct approach 68 3.9 Results 70 3.9.1 Chained approach 70 3.9.2 Direct approach 72 3.10 Regression analysis 74 3.10.1 Chained approach 74 3.10.2 Direct approach 75 3.11 Discussion and conclusions 76

Chapter 4 Exploring attitudes to health care resource allocation amongst professionals and the public: results from Q methodology and Q block surveys

96

4.1 Introduction 96 4.2 Views on health care priority setting amongst the public

across 10 European countries 97

3

4.2.1 Introduction 97 4.2.2 Methods 99 4.2.3 Results 101 4.3 Views on health care priority setting amongst decision

makers across 10 European countries 108

4.3.1 Introduction 108 4.3.2 Methods 108 4.3.3 Results 108 4.4 Comparing public and decision makers’ views 112 4.5 Discussion 113 References 119

Figures

Figure 3.1 EuroVaQ: conceptual overview for monetary valuation of a QALY Figure 3.2 Example SG question Figure 3.3 SG iteration Figure 3.4 „Extreme‟ SG responses Figure 3.5 Example TTO question Figure 3.6 TTO iteration Figure 3.7 „Extreme‟ TTO responses Figure 3.8 Example risk variant WTP question Figure 3.9 Example time variant WTP question Figure 3.10 Summary of the WTP questions in chained questionnaire Figure 3.11 Example card sort procedure (shows risk variant WTP question) Figure 3.12 Overview of questionnaire versions 1-8 Figure 3.13 Design for chained approach, includes 3.13a, 3.13b and 3.13c Figure 3.14 Summary of WTP questions – direct questionnaire Figure 3.15 Design of direct approach Figure 3.16 Age, health and life expectancy Figure 3.17 Example of Question A Figure 3.18 Example of Question G Figure 4.1 Structure of the chapter and connection with appendices Tables

Table 2.1 Calculation of x and y Table 2.2 Value per Life Year (All values in Euros) Table 2.3 Value per Discounted Life Year (All values in Euros) Table 2.4 Value per QALY (All values in Euros) Table 2.5 Value per discounted QALY (All values in Euros) Table 3.1 Response rates to SSI panel invitations Table 3.2 Total number of starters and completion rates Table 3.3 Achieved sample – Chained questionnaire Table 3.4 Achieved sample – Direct questionnaire Table 3.5 Coefficients from the Heckman regression models for each question:

Chained – green health state Table 3.6 Coefficients from the Heckman regression models for each question:

Chained – yellow health state Table 3.7b Value of a QALY: Preliminary case – chained (Yellow health state)

4

Table 3.8a Value of a QALY: Chained trimmed – Green health state Table 3.8b Value of a QALY: Chained trimmed – Yellow health state Table 3.9 Value of a QALY: Preliminary case - direct Table 3.10 Value of a QALY: Direct trimmed Table 3.11 Coefficients from the Tobit regression models for each question: Direct Table 4.1 Response general public across ten countries Table 4.2 Statement rank scores general public Table 4.3 Response decision makers across ten countries Table 4.4 Statement rank scores decision makers

Appendices Appendix 2.1 Country-specific data and sources Appendix 2.2 Life expectancy gains version of Approach 1 Appendix 2.3 VPF estimates in home currency Appendix 3.1 Introduction to direct questionnaire Appendix 3.2 Example of Question A Appendix 3.3 Example of Question B Appendix 3.4 Example of Question D Appendix 3.5 Example of Question E Appendix 3.6 Example of Question F Appendix 3.7 Example of Question G Appendix 3.8 Example of Question I Appendix 3.9 Example of Question J Appendix 3.10 Example of Question L Appendix 3.11 Example of Question M Appendix 3.12 Example of Question N Appendix 3.13 Example of Question O Appendix 3.14 Example of Question P Appendix 4.I Technical appendices to the results of the Q study Appendix 4.II General public views in 10 countries Appendix 4.III Comparing results Appendix 4.IV Distribution of European views in the EuroVaQ sample

5

European Value of a Quality Adjusted Life Year

List of authors Cameron Donaldson1,2 Rachel Baker1,2 Helen Mason1 Mark Pennington1,17 Sue Bell1 Emily Lancsar1,3 Mike Jones-Lee1,3 John Wildman1,3 Angela Robinson4 Philomena Bacon4 Jan Abel Olsen5 Dorte Gyrd-Hansen6 Trine Kjaer6 Mickael Bech6 Jytte Seested Nielsen6 Ulf Persson7 Annika Bergman7 Christel Protière8 Jean Paul Moatti8 Stéphane Luchini9,8 Jose Luis Pinto Prades10 Awad Mataria11 Rana A Khatiba11 Yara Jarallah 11 Job van Exel12 Werner Brouwer12 Roman Topór-Madry13 Adam Kozierkiewicz13 Darek Poznanski13 Ewa Kocot13 László Gulácsi14 Márta Péntek14 Andrea Manca15 Samer Kharroubi15 Phil Shackley16

6

1Institute of Health and Society, Newcastle University, United Kingdom 2Yunus Centre for Social Business and Health, Glasgow Caledonian University, United Kingdom 3Ecomomics, Business School, Newcastle University, United Kindom

4School of Medicine Health Policy & Practice, University of East Anglia, United Kingdom 5Institute of Community Medicine, University of Tromso, Norway 6Institute of Public Health, University of Southern Denmark, Denmark 7Swedish Institute of Health Economics, Lund, Sweden 8INSERM, U-912 “Economic & Social Sciences, Health Systems & Societies”

(SE4S), Université Aix-Marseille, France 9GREQAM, Centre National de la Recherche Scientifique, France 10Department of Economy, Quantitative Methods and Economy History, Pablo de Olavide University, Seville, Spain 11Institute of Community and Public Health (ICPH), Birzeit University, Palestine 12Department of Health Policy and Management, Institute for Medical Technology Assessment, Erasmus University Rotterdam, The Netherlands 13Institute of Public Health, Jagiellonian University Medical College, Krakow, Poland 14Centre for Health Economics and Health Technology Assessment, Corvinus University Budapest 15Centre for Health Economics, University of York, United Kingdom 16School of Health and Related Research, University of Sheffield, United Kingdom 17School of Hygiene and Tropical Medicine, London, United Kingdom

7

Project Summary Proposal full title: European Value of a Quality Adjusted Life Year Proposal acronym: EuroVaQ Strategic objectives addressed: Different Member States will have different levels of affordability with respect to how much they can allocate to health care provision and, thus, to contribute to health production. Country-specific values of health, therefore, would lead to improved decision making and efficiency. This can be addressed by surveying members of the public about what factors should count when making such allocation decisions and also about their willingness to pay (WTP) for health gains. It is crucial that WTP-based values generated have been rigorously tested across cultures using a consistent methodological approach. To derive factors and values, a European-level research initiative was required in the interests of subsidiarity and coherence. Thus, this project addressed Task 1 – Developing methodologies to address cost-effectiveness in healthcare systems – within Research Priority 2.1 on Health determinants and the provision of high quality and sustainable health care services and pension systems (in particular in the context of ageing and demographic change) which is part of the more general Priority 2 on Providing health, security and opportunity to the people of Europe within the Policy-Oriented Research Work Programme of the Commission. Abstract: A major issue in economic evaluation of health care is that of the value to place on a quality adjusted life year (QALY), commonly used as a measure of health care effectiveness across Europe. This has come to the fore in several European countries, resulting from the creation of national health technology and pharmaceutical assessment agencies. Such agencies were established to make recommendations on technology adoption, addressing issues of affordability and sustainability of publicly-funded health care systems. Recommendations are most often made on the basis of QALYs produced relative to costs incurred. Methods of estimating cost per QALY, based on rigorous decision analytic models, are now very sophisticated. However, „threshold‟ values adopted (such as £20-40,000 per QALY above or below which a new therapy will be rejected or recommended for adoption in England) are essentially arbitrary, with little or no economic foundation. This critical policy issue is reflected in the growing interest across Europe in development of more sound methods to elicit such a value. The main aim of this research was, therefore, to develop robust methods to determine the monetary value of a QALY across a number of European Member States. This was addressed in two ways: through „modelling‟ such a value based on values of statistical lives currently used (or implicit values from adoption decisions in various fields) across Member States; and through survey research to test two methods of deriving a societal willingness-to-pay (WTP) based monetary value of a QALY. A further initiative, using Q methodology, involved surveying members of the

8

public and health care professionals about wider characteristics of health care beneficiaries that might count when making health care resource allocation decisions. A European-level research initiative was required in the interests of subsidiarity and coherence. Different Member States may have different levels of affordability of QALY production and different attitudes to what might determine who gets priority for health care resources. Likewise, there may be similar patterns in affordability and attitudes across countries. Country-specific and European-wide values would lead to improved decision making, and thus greater equity and efficiency in health care. But, it is crucial that WTP-based values generated and attitudes elicited have been rigorously tested across cultures using a consistent methodological approach.

9

Chapter 1 Introduction

1.1 Policy relevance To assess value for money in the health care systems of European Community Member States, robust and innovative methods are required to estimate the value placed on health gains produced by such systems. Thinking about such gains in terms of quality adjusted life years (or QALYs) - commonly used as a measure of health care effectiveness across Europe (Tarn et al., 2004) - the main aim of the project is to develop methods of estimating a monetary value of a QALY (Richardson and Smith, 2004; Smith and Richardson, 2005). This aim is supplemented by an exploration, using a particular methodology, of additional factors (particularly types of health gain and characteristics of beneficiaries) that members of the public and decision-makers feel should count in health care resource allocation. These research efforts add value, contributing substantially to the context and research objectives of Priority 2 on Providing health, security and opportunity to the people of Europe, by working towards a coherent approach to valuation of health gains which can cope with between-country differences but also be applied across countries as health policy becomes more integrated. By helping to ensure value for money, development of monetary values of a QALY ensure the most effective responses across countries to demands imposed by ageing populations and technological developments, thus contributing to the long-term objectives of accessibility, quality and sustainability. In order to ensure the sustainability of their health care systems, Member States have engaged in two main types of policy activity over the past 20-30 years. The first is health care reform; attempting to maintain coverage rates whilst changing the incentives within the system, such incentives being aimed at consumers, professionals and institutions (like hospitals). Experiences with such reforms are well-documented (European Observatory on Health Care Systems, 2002; Donaldson and Gerard, 2005). Second, and more recent, has been the development in many countries of national health technology and pharmaceutical assessment agencies, such as the Swedish Pharmaceutical Benefits Board, England and Wales‟ National Institute for Health and Clinical Excellence (NICE), the Norwegian Medicines Control Authority, the German

Institut fur Qualitat und Wirtschaftlichkeit im Gesundheitswesen (IQWiG) and the Dutch Health Care Insurance Board (CVZ). (Where such agencies have not been established, technology assessment still takes place through Ministries of Health, e.g. in Spain and Hungary.) This is also reflected in the creation of the European Network for Health Technology Assessment (EUnetHTA), which coordinates the Health Technology Assessment (HTA) efforts of 27 European countries of which 24 are Member States of the European Union. These institutions have been established in attempts to achieve some form of uniformity within systems which inevitably suffer from asymmetries of information and resulting inefficiencies, and to provide an efficient mechanism for evaluating new interventions arising from rapid technological development. Thus, more systematic approaches to HTA are being developed, and these require robust valuation of the potential benefits from interventions.

10

An outstanding issue in the development of methods of economic evaluation for use in national HTA is that of what value to place on a QALY. This arises because, in many such evaluations, the benefits of therapies are valued in terms of QALYs. When making decisions based on evaluations of single therapies, this poses empirical challenges for HTA agencies. As such agencies have been established to deal with the issue of affordability and sustainability of publicly-funded health care systems and, thus, make recommendations as to whether or not such systems should adopt the therapies evaluated, inevitably the appraisal process will involve judgements about whether the QALYs gained are worthwhile. The methods of economic evaluation based on rigorous decision analytic models are now very sophisticated (Drummond et al., 2005). However, the „threshold‟ values adopted (such as £20,000-40,000 per QALY above or below which a new therapy will be rejected or recommended for adoption in England and Wales and the commonly-mentioned Eur18000 in the Netherlands) are more or less “arbitrary” (National Institute for Health and Clinical Excellence, 2004; Rawlins and Culyer, 2004). Hence the growing interest in such agencies across Europe in development of more sound methods to elicit such a value. Furthermore, as other government departments/agencies (e.g. those dealing with transport safety) in many countries assign monetary values to the prevention of injury, illness and premature death, it would be in the interests of coherence and consistency in public policy more generally if comparable preference-based monetary values for QALYs could also be obtained from the population, meeting the Commission‟s goal of a more coherent policy vision with a clear evidence base. The main aim of this research was, therefore, to develop new and extend existing methods to determine how much it is reasonable to spend to achieve these gains in length and quality of life, that is, what monetary value should be placed upon a QALY. This has been addressed in two ways: first, and as reported in Chapter 2 below, through devising methods of „modelling‟ such a value based on those for statistical lives (or implicit from adoption decisions in various fields) currently used in public policy evaluation across member states; and, second, through development and testing two methods of deriving a WTP-based value of a QALY through surveys of members of the general public, as reported in Chapter 3. A valid measure of the value of a QALY would significantly enhance the abilities of HTA agencies to recommend policies to their Member States which allow them to cope with the challenges of an ageing population, technological innovation and major health burdens (such as mental disorders), in the most efficient manner, which encompasses the three goals of quality, accessibility and financial sustainability. This complements and takes into account actions launched under the Public Health Programme (non-R&D programme administered by the Directorate General for Health and Consumer Protection) by, again, helping to ensure that the most efficient public health policies are implemented, and also complements Priority 1 (Life sciences, genomics and biotechnology for health) by aiding the delivery of responsible choices reflecting societal values and ensuring that health care systems combat major diseases in a way that reflects greatest value for money (in terms of health gain to the population). Likewise, the products of this research will embrace and enhance the Commission‟s Key Actions of Ageing by, once again, ensuring value for money for health care systems from diagnostic, treatment and preventive

11

activities devised from these scientific endeavours. Indeed, whether a monetary value of a QALY should even vary according to some of the characteristics (such as age) listed above has been investigated in EuroVaQ through the application of Q methodology, a qualitative method new to health economics (Baker et al., 2006), as reported in Chapter 4.

1.2 State of the art

The concept of WTP has existed for a long time (Dupuit, 1844; Davis, 1963). However, not until the 1980s did Government Transport Departments worldwide consider using the method to value lives saved from safety projects, rather than the gross output („productivity‟) approach used previously (Jones-Lee, 1989). Arguably, the most natural measure of the extent of a person‟s preference for anything is the maximum amount that s/he would be willing to pay for it. Under what has naturally come to be known as the „willingness-to-pay‟ (WTP) approach to valuation of safety, one seeks to establish the maximum amounts that those affected would individually be willing to pay for (typically small) improvements in their own and others‟ safety. These amounts are then aggregated across individuals to arrive at an overall value for the safety improvement concerned, thus reflecting society‟s overall resource constraint. Estimating a WTP-based monetary value of a QALY can also be viewed as a group-aggregate WTP for marginal gains in health, at least in the case of a randomly-selected sample of the public. Indeed, this argument has been used in promoting an insurance-based approach to valuing publicly-provided health care; whereby respondents are informed of the probabilities of needing care as well as it being successful before providing a valuation (Gafni, 1991; O‟Brien and Gafni, 1996). The WTP method was first applied in health to value heart attack prevention (Acton, 1973). Subsequently, there were few studies in health, probably resulting from the view that such monetary valuation was unethical. In addition, the use of WTP to inform decisions about allocation of health care, which is supposed to be on the basis of (some notion of) need, may look problematic because WTP is obviously associated with ability pay. However, it has been shown that this need not impede the use of WTP in health economic evaluation (Donaldson, 1999) and that QALYs suffer from the same challenge of ability to trade in the valuation metric (whether risk or time) being associated with socio-economic grouping (Donaldson et al., 2002). Furthermore, using thresholds in decision-making implicitly reveals monetary values for QALY gains, but little work has been conducted to justify or challenge these. Since the early 1990s, the feasibility of using WTP in health economics has again been recognised (Gafni, 1991; Johannesson et al., 1991), and more studies undertaken (Olsen and Smith, 2001; Smith, 2003). Thus, in health, WTP methods historically addressed decision making dilemmas at two main levels: assessing relative utility of treatments for a given group of patients (involving elicitation of values from samples of such patients); and across disparate programmes funded by geographically-defined health organisations (involving elicitation from the community of WTP values for each programme at stake). Methods have been developed which work well in terms of WTP values reflecting patient preferences (Donaldson, 2002). In the latter area, methods have been more

12

problematic, but are improving (Olsen and Donaldson, 1998; Olsen et al., 2005). As in other public sector areas, results have been mixed on how sensitive WTP responses are to the size of the good (i.e. the health change/numbers treated) on offer to respondents (Olsen et al., 2004; Yeung et al., 2003; Smith, 2005) and to other aspects related to “framing” and programme information presented to respondents (Protière et al., 2004; van Exel et al., 2006). However, innovations in valuing safety improvements have shown promise in overcoming some of these issues (Carthy et al., 1999). Methods based on these developments, to be used in the proposed research, are described below. Through the 1990s, development of national-level technology assessment agencies (such as in the Health Care Insurance Board in the Netherlands, the Norwegian Medicines Control Authority and in the Hungarian Ministry of Health), led to calls for monetary values of a QALY to aid decision making at this third (i.e. national) level (Johannesson, 1995; Garber and Phelps, 1997). Implicit values of life, based on previous decisions, tend to vary greatly (Viscusi and Aldy, 2003). Some estimates have been made of the value of a QALY based either on modelling approaches or survey research. Modelling has essentially involved (1) taking an existing value of a prevented fatality (VPF), (2) assuming an age at which the fatality would have taken place (typically around 40 years), using actuarial data to project future survival prospects if the fatality were prevented, adjusting such future life years gained by age-related quality of life scores (where 0=death and 1=full health), to arrive at QALYs, and discounting these QALYs, and (3) dividing the VPF by the QALYs gained. This approach has been used in the US (Hirth et al., 2000) and is limited in that WTP for a reduction in risk of premature death is assumed to be the same as the WTP for the preservation of a given number of equally valued future QALYs. More sophisticated models for deriving a value of a QALY, previously applied to UK data only and devised by members of the project team (Mason et al., 2009), were applied to data from all participating countries, as reported in Chapter 2. The aim was to permit improved and European-relevant estimation of WTP-based values of a QALY in the short-term, whilst survey work was being designed and administered. However, it will be seen that the variation in values arising from this work simply reinforced the importance of the survey work to follow. Although survey work on the value of a QALY had begun to appear in the literature by the time of project commencement, this work had been limited. Typically, individuals had been asked about their WTP for health gains for which quality adjustment factors have been gained from another sample, without fully adjusting for uncertainty (i.e. by presenting scenarios involving certain gains in quality of life) and, in some cases, eliciting values from patients and not from members of the general public (Gyrd-Hansen, 2003; Byrne et al., 2005; King et al., 2005). Only one such estimate existed for a European country (Gyrd-Hansen, 2003). Given the limitations of the studies just referred to and the well-known difficulties that respondents have in providing sufficiently sensitive answers to questions which involve direct trade-offs between wealth and small reductions in risk, members of the research team had been working on the development of less-direct „chained‟ approaches which break down the valuation process into a series of more manageable steps (Pinto Prades, 2009; Baker et al., 2010). These were based on

13

work from the valuation of safety literature by Carthy et al. (1999), involving elicitation of a WTP value for a simpler and more imaginable „slight injury‟ scenario and then using „standard gamble‟ questions to link this scenario to others entailing a range of more serious consequences, up to and including premature death, so permitting calculation of a VPF. The Carthy et al. approach, which became known in EuroVaQ as the „chained‟ approach, involved a respondent being asked to value avoidance of a small, but certain, health detriment, and then provide a health state utility for the scenario using, for example, a standard gamble procedure (Pinto Prades et al., 2009; Baker et al. 2010). These early attempts to apply a chained approach to the value of a QALY have unearthed issues relating to methods of aggregation and measures of central tendency. For example, if a respondent says s/he is only willing to take a 1 in 100,000 risk (or less) of death to avoid being permanently in a minor health state, his/her WTP to avoid 12 months in that state would then be multiplied by a hundred thousand, potentially giving an astronomical figure for the value of a QALY1. In the UK Social Value of a QALY (SVQ) Project over 25 per cent of respondents gave responses to a 12-month scenario which, in combination, imply values of a QALY of more than £1m (Baker et al., 2010). Some respondents generated values of thousands of millions of pounds. It has been widely accepted that when such values are being used to guide public policy, it is the mean figure which should be used as the best indicator of social welfare. But clearly, a mean value for a QALY of £5 x 108, or even a figure one-thousandth as big as that (i.e. £500,000), would be totally anomalous in a world where the VPF is £1.4m (representing the prevention of a death which, on average, entails the loss of about 40 years of life expectancy) and the NICE cost per QALY threshold is £20,000-30,000. Other ways of managing the data were therefore devised. Rather than computing a ratio of WTP/QALY loss for each individual and then taking a mean, these alternative approaches involved taking a measure of central tendency (either mean or median) for WTP and the corresponding measure of central tendency for the QALY loss and then computing the ratio, rather more like a point estimate. Pinto Prades et al. (2009), having experienced the same problem, took the same approach. Essentially, these approaches, by diminishing, or even eliminating, impact of extreme responses on the valuations, lead to more conservative results. For example, combining a mean health state utility value of 0.1 with a mean WTP value for avoiding the same stomach condition for a period of one year of £1,870 led to a value of a QALY of £17,980 in SVQ, whilst in the Pinto Prades et al. study in Spain, although reversal health improvements were valued, a typical value of €27,192 was obtained from combining a mean health state utility gain of 0.05 with a mean WTP of €1,348 for that particular gain for 12 months. The more „terrestrial‟ values of a QALY obtained via these methods of aggregation led to the idea that if we could tailor QALY gains given to respondents to 0.1, then

1 Someone who says they are only willing to take a 1 in 100,000 risk of death to avoid the chronic

illness state is taken to be indexing that health state at 0.99999: i.e. a year spent in that state is taken to amount to the loss of 0.00001 of a QALY. So if 100,000 such people were each willing to pay, say, £300 to avoid the 12-month illness, they would between them be paying £30 million and their combined benefit would add up to just one QALY.

14

perhaps the resulting values of a QALY, when calculated via means of ratios obtained from individual respondents would be similarly conservative. This would mean developing the earlier approaches of Pinto Prades et al. and Baker et al. in two related ways. The first was to reverse the questions so that respondents answered health state utility questions first, followed by the WTP questions. Second, by doing this, the QALY gain presented to respondents in the WTP question could be tailored to a pre-determined size based on responses to the health state utility question. For example, if presented with a health state utility question resulting in an index (between 0 and 1) of 0.8, the loss from being in that state would simply be 0.2 (i.e. 1-0.8). To then present that person with a WTP question whereby they would experience a gain of 0.1 of a QALY, the respondent could then be asked (in a „time-variant‟ form of the question) about either their WTP to avoid being in that state for six months (i.e. 0.2 of a gain in health state utility over half a year, amounting to 0.1 of a QALY) or (in a „risk-variant‟ version) about their WTP to avoid a 50% chance of experiencing that loss for a year (i.e. 50% of a 0.2 gain in health state utility for a year amounting again to 0.1 of a QALY). As will be seen in Chapter 3, both time-variant and risk-variant versions were asked of separate groups of respondents. Given that some people may have still given very small health state utility values in the first set of questions, some of the lengths of time spent in such a state could be necessarily long (in order for the gain to amount to 0.1 of a QALY), and logically capped by the remaining life expectancy of the respondent. Likewise, some risk-variant questions were capped at 100%. Each of these cases would mean that some respondents would be valuing a QALY gain of less than 0.1. Other modifications were twofold. A gain of 0.1 of a QALY might still be seen as large, and thus subject to the problem of budget constraints; whereby a respondent simply states the maximum (a kind of threshold) they feel have available, a value which may have been the same as for a smaller QALY gain. This would lead to smaller values of a QALY based on the 0.1-QALY-gain scenario, but only because a given budget constraint, having been reached, would now be divided by a larger, as opposed to smaller, QALY gain. Therefore, some respondents had questions tailored to a 0.05 QALY gain. Furthermore, because challenges with health state utility values may have arisen from use of the standard gamble by Pinto Prades et al. and Baker et al., time trade-off questions as well as standard gambles were used. To provide a comparison to the WTP-based values of a QALY derived using the chained procedures outlined above, we also developed a more direct approach to eliciting values of QALY. The rationale for this method was simply to test whether we can ask the question „what is the monetary value of a QALY?‟ as directly as possible given the problems that we might face with more-conventional risk-based approaches and the fact that issues of dealing with small risks are still embedded within the chained approaches. Furthermore, the intention was that this „direct‟ approach would involve neither eliciting the utility value of health states nor trying to combine two different valuations. Obviously, it is difficult to simply ask people „what is the monetary value to you of a QALY?‟ However, various scenarios can be drawn diagrammatically, each of which represents a one-QALY gain. For example, one could start a questionnaire by asking

15

about gains of one QALY which are derived from gains in health-related quality of life only, before moving on to questions involving extension of life (by one QALY) and then others which involve more immediate life-saving scenarios. The further question addressed by doing this is whether different QALY-types have different value. In addition, we can also test whether the same gain is valued differently at different points in time and also build in more traditional questions involving uncertainty. Despite the logic of trying to ask such questions as directly as possible, this poses some challenges, however:

first, the QALY is never really explained to people. Scenarios can be presented that researchers know amount to a one-QALY gain and then people asked about their WTP to avoid them.

there is an issue of the degree to which this method can have embedded within it questions which are directly comparable to the chained method. With the chained method relying on presentation of relatively small QALY gains, it will be seen that conventional risk-based questions were employed (e.g. asking about eliminating a five percent chance of a one-QALY loss in order to result in a 0.05 QALY gain). This also has the advantage of us having such conventional questions used in the study as a basis for comparison against both direct and chained methods. To test for sensitivity to scale and to further match the chained approach, 10 percent as well as five percent chances of avoiding a one-QALY loss were presented to respondents.

the questions, by their nature, tend to be based on certain events. Again, this is to try to avoid the problems of understanding of risk and probabilities. As explained in the previous bullet, however, some examples of uncertain questions are included, partly as a check on how different magnitudes of QALY values might be interpreted.

the intention was to leave the notion of what it means to be in a health state valued at, say, 0.75 to the respondent to articulate for him/herself (although it will be seen that a little guidance was given in the introduction to the questionnaire).

due to the size of the QALY gains with which respondents are faced, it could be argued that they are more likely to face budget constraints, and, thus, hit a ceiling in their WTP responses which would be reflected in lower resulting values of QALY.

Finally, little is known across Europe about the views of senior decision makers and members of the public as to the relative importance of either different types of health gain (e.g. those arising from quality of life gains only or from extending or saving life) or the characteristics of beneficiaries from health gains (e.g. by age group). Through the work reported in Chapter 4, new data are produced on this by application of a method which is very new to health economics (i.e. Q methodology). Q methodology has previously been used as part of the SVQ project (Donaldson et al., 2010) to identify and describe the opinions that exist around health care priority setting. This study focused on the characteristics of the types of people who would receive care. The study found three main viewpoints about health care priority setting: a broadly egalitarian account; an efficiency/ health maximising view that strongly rejects prioritisation on the basis of income or contribution; and a third point of view distinguished by a concern for children together with a belief that decisions of

16

distributive justice should be made by experts. In a study which examined the types of health gain which arise from health care interventions, the results indicate that members of the public do distinguish between interventions that could be classified as life saving and interventions which produce gains in quality of life or gains in life expectancy which arise in the longer term (Mason, 2007). The study reported in Chapter 4 combines elements of both of these previous studies to examine respondents‟ views on how the QALY gain arises and the characteristics of the people who receive the QALY gain.

17

Chapter 2: Estimating a WTP-based monetary value of a QALY from existing contingent valuation studies of the value of prevented fatalities and serious injuries

2.1 Introduction

One possible way to elicit a monetary value of a QALY has been to „model‟ a WTP-based value of a QALY from the existing value of preventing a “statistical fatality” (VPF) currently used in public sector safety policy. Although such values are derived in a different context to that of health, there are similarities due to the values being based on surveys in which respondents are asked hypothetical questions about trade-offs between wealth and reduced risks of death. The appropriate way in which to derive an estimate of the money value of a QALY from a pre-determined VPF does, however, raise some fairly fundamental questions. Thus, for example, early (conventional) attempts to estimate the value of a QALY in this way simply involved dividing the VPF by the population mean remaining life expectancy or QALYs (Hirth et al., 2000). By contrast, more recent work has focused on the way in which the VPF varies with the age of those who will enjoy a reduction in the risk of premature death and, in particular, bases the estimate of the value of a QALY on the rate at which the VPF increases with increasing life expectancy (Baker et al., 2008).

The key difference between these two approaches is that the first yields aggregate WTP across a large group of people for small reductions in each individual‟s current-period risk of death which, when summed over the affected group, gives a total gain in life expectancy of one year. By contrast, the second approach produces aggregate WTP for a total gain in life expectancy of one year, where this gain can most naturally be thought of as resulting from reductions in the risk of death that are spread over the remainder of each affected individual‟s lifetime, but with the magnitude if the risk reduction also being an increasing function of the size of the individuals‟ hazard rate at each age (i.e. the probability of death at that age conditional on having survived to that point). Clearly, therefore, the question of which approach is the more relevant depends essentially on the manner in which a gain in life expectancy is generated – i.e. whether the gain is principally the result of reductions in the risk of death in early or in later years of adulthood. The modelling approaches derived are used to estimate country-specific WTP-based values of a QALY on the basis of existing values of statistical lives. We attempted to do this for the ten European countries participating in the EuroVaQ project. In the first section of the report the data requirements from each country are presented along with a table incorporating all of the data used in the calculations. The two different approaches which have been developed to convert the VPF into a monetary value of a QALY are then outlined. In this report the approaches are explained using UK data. Following this the results from each of the approaches are presented, with values for each country. The final section discusses the results and raises some caveats surrounding this work.

18

2.2 Data Sources

2.2.1 Types of data required All of the approaches described in Section 2.3 require information on the VPF, life expectancy for males and females, quality of life of the population and the discount rate for each country. Appendix 2.1 summarises the country specific values which were used and the sources from which these data were collected. 2.2.1.1 Estimating the Value of a Prevented Fatality There are two main approaches to deriving a VPF: a WTP based approach or a human capital approach. The WTP approach is favoured by economists as, under standard welfare economics, maximum WTP represents the theoretically correct measure of “strength of preference” or value for a commodity (Mishan, 1971; Pauly, 1995). The main way in which the WTP-based VPF is elicited is through the contingent valuation method, which is where an individual is asked to state a value for a good in question rather than reveal it as a result of real market based choices. Under the human capital approach the primary cost of a premature death is the lost future output which will no longer be produced as a result of a premature death. In some countries a small amount was then added to reflect the pain, grief and suffering of the victim and their family, this is usually described as the “welfare loss”. The VPF would then be expressed as the costs avoided (Mason et al., 2009). At present time only the UK and Sweden are estimates of the VPF based on directly commissioned contingent valuation studies. To illustrate how these studies derive the VPF assume that a group of 100,000 people benefit from a safety improvement that reduces the probability of premature death during the coming period by, on average, 1 in 100,000 for each member of the group. The expected number of fatalities within the group would thus be reduced by one. If the average WTP across the group for the safety improvement is £W the aggregate WTP is given by Wx£100,000. This figure is then referred to as the WTP-based VPF (Mason et al., 2009). The values for France and the Netherlands are also WTP-based but are derived from a combination of pre-existing contingent valuation studies rather than a specifically designed study. The VPF from Denmark is elicited using the human capital approach and incorporates a small adjustment for what is called “welfare loss”. The values from Hungary and Spain are based on the hedonic pricing method, using information on wage–job risk trade-offs. 2.2.2 Adjusting the VPF To be able to use the quality of life and life expectancy data in the approaches described in Section 2.3 it is necessary to perform some additional calculations which are presented below.

19

2.2.2.1 Estimating remaining life expectancy For each of the approaches an estimate of remaining life expectancy is required. This is used to reflect the life years lost from a fatality and facilitate the conversion of the VPF into a value per life year. Approach 1 compares the VPF with a representative life expectancy for the sample population used to elicit the VPF. Details of the ages and gender of the sample members are not available. Therefore, it was assumed that the sample was randomly drawn from the adult population. The mix of the sample population is assumed to reflect the adult population for each country as given by official population statistics. Approach 2 assesses the value per life year in relation to three specific ages - 18, 40 and 80. The life expectancy at 18 is the average of the male and female life expectancies at 18 weighted by the gender shares of the population at 18. The values at age 40 and 80 are similarly calculated. Life expectancy by age and gender is calculated using life tables. These tables give, by gender and age, the mortality rate between age x and age x+1 (the probability that a person aged x will die before reaching age x+1). Using the life tables to calculate remaining life expectancy is preferred to using life expectancy at birth information which is available from most countries as it allows for the introduction of quality of life data and discounting. 2.2.2.2 Estimating a value per quality adjusted life year The method for calculating expected QALYs is very similar to the life expectancy calculations explained above. The only modification to the methodology is to apply a quality of life weight to the probability of each potential year of life. The sum of the quality adjusted probabilities of living through each potential year of life gives the expected QALYs. The quality of life weights, along a scale from zero for death to one for full health, are taken from EQ-5D population norms which are specific to each country (see Appendix 2.1 for references for these values). The EQ-5D tariff does allow for states worse than death which would receive a value below zero. However, the population norms were estimated from a representative sample who considered themselves to be generally healthy and, as such, negative values are not relevant here. 2.2.2.3 Discounting Both the value per life year approaches and the value per QALY approaches have also been modified to take into account discounting. The discount rate used is specific to each country. In the UK and France the full discount rate is made up of two parts; a rate of pure time preference and a part which reflects diminishing marginal utility of consumption. However, theory shows that a WTP-based VPF can be expected to grow at much the same rate as the marginal utility of consumption declines and so is excluded from the calculations for reasons of double counting. However, the rate of pure time preference is included as it incorporates uncertainty about the future. For each of the other countries the discount rate is not separated

20

into two parts. Therefore, the discount rate is higher than those used for the UK and France which could lead to higher estimates of the monetary value of a QALY as a higher discount rate will increase the value. The relevant discount rate for each country is applied to the series of (quality adjusted) probabilities of living through each potential year of life. The discounted life expectancies and discounted QALY expectancies then replace life expectancies and QALY expectancies in the calculations. 2.2.2.4 Value of a Serious Injury Another way to calculate a monetary value of a QALY is to use the value of a serious injury (VSI) instead of the VPF. VSIs tend to be based on the valuation of different severities of serious injuries which range from injuries that will last only a few days, and require no hospital treatment, through to permanent paralysis and brain damage (Jones-Lee et al., 1995). It was anticipated before the project began that VSIs would be available from all participating countries. However, in attempting to source this information it was discovered that data only existed for half the countries. For the countries where a VSI did exist the way in which the values were calculated were very different making it more difficult to find a method by which the VSI could be converted into the value of a QALY. Hence, it was decided not to continue with this means of calculating a value of a QALY.

2.3 Possible approaches to converting the value of a statistical life year into a monetary value of a QALY

Two approaches to converting the VPF into a monetary value of a life year are described below. It is not possible to present the calculations using the data from each of the countries, thus, each of the approaches is explained using UK data. Of course, if country-specific calculations are required, the relevant data from Appendix 2.1 can be used to replace the UK values shown below. Each of the approaches below presents the calculation of a monetary value of a life year. To calculate the value of a QALY, the life expectancy data should be replaced with the quality adjusted life expectancy data. Similarly discounted values can be calculated by replacing the life expectancy data with the discounted life year or quality adjusted life year data.

2.3.1 Approach 1 The simplest way to think about this issue is to appreciate that the VPF is in fact an aggregate WTP across a large number of people for a risk reduction that can be expected to result in the prevention of one premature death during the coming year (Jones-Lee, 1989). In the UK the VPF is £1,427,340 (Department for Transport, 2007). On average, the avoidance of one premature death would result in the avoidance of the loss of 32 years of life expectancy. The WTP-based monetary value of each of these 32 years is, therefore, given by:

V1 =32

£1,427,340= £44,018 (1)

21

Under this approach the WTP based value of a life year would therefore be £44,018. One obvious criticism of the above approach is that it assumes that each future life year is of equal value, when it is more likely that, as far as any given individual is concerned, a year in the distant future is valued less highly than one in the near future. However, it is also possible to think of the resulting value as being based on an aggregation across a large group of people of their individual WTP for small risk reductions which will result in small gains in life expectancy that, when summed over the whole group, give aggregate WTP for a total gain in life expectancy of one year. For example, consider a large, representative group of n individuals each offered a

n

1 reduction in the risk of death during the coming year. To the extent that, taken

over the whole group, these risk reductions will prevent precisely one “statistical” fatality; the group‟s aggregate willingness to pay for the risk reductions will constitute

the WTP-based VPF. However, as shown in Appendix 2.2, given the reduction of n

1

in risk, each individual in the group will enjoy an increase in life expectancy that is

closely approximated by n

1 times his/her remaining life expectancy, Ei . With n being

large, these individual gains in life expectancy will be small (i.e. essentially marginal). However, summed over the whole affected group of n individuals, the total gain in life

expectancy will be equal to Enn

1E

n

1n ii

, that is, by the mean remaining life

expectancy for individuals in the group. It then follows immediately that WTP per year of aggregate gain in life expectancy is equal to the VPF divided by mean life expectancy. A more formal proof of this result is provided in Appendix 2.2. It is therefore clear that, based on this line of reasoning, exactly the same result (in this case, of £44,018 per life year gained) can be arrived at. This is an important point to recognise as although one can reasonably criticise earlier estimates of the value of a QALY, which have all been implicitly based on the “equally valued life years” assumption, the more sound reasoning underlying group aggregation, combined with the knowledge that the approaches lead to the same result, actually lends substantially greater credence to such earlier results. This having been said, it should be noted that there is a variant of the “group aggregation” approach which, although very closely related to the one that has just been developed, will nonetheless almost certainly produce a different result. Thus, suppose that rather than setting all of the individual risk reductions equal, these reductions are instead allowed to vary across individuals and are in fact set so that

individual gains in life expectancy are themselves all equal ton

1. In this case, as

shown in Appendix 2.2, the WTP-based value of an aggregate gain of one year of life

expectancy is given by the arithmetic mean of individual ratios i

i

E

M, where Mi is the

ith individual‟s marginal rate of substitution of wealth for risk, the mean of which is

(under the standard argument) the VPF.

22

Finally, it should be stressed that the gain in life expectancy that features in the “group aggregation” rationale that underpins both versions of Approach 1 is essentially the result of reductions in the current period risk of death. 2.3.2 Approach 2 Consider a large representative group of 40 year olds and a corresponding group of 50 year olds. Intuition, theory and empirical evidence all strongly suggest that – controlling for other factors such as income, health state, and so on – the WTP-based VPF for the 40 year olds will significantly exceed that for the 50 year old group quite simply because the 40 year olds have a larger remaining life expectancy than the older group and hence would experience a larger loss of expected lifetime utility as a result of premature death. To the extent that the value of a gain in life expectancy will depend directly on the resultant gain in expected lifetime utility, then the rate at which the VPF changes as we move from one age group to another should precisely parallel the rate at which people would be willing to trade off wealth against gains in life expectancy. Under Approach 2 the aim is therefore to base the value of a year‟s gain in life expectancy on the population mean of the annual rate at which the WTP-based VPF varies with life expectancy as one moves from one age group to another, controlling for other factors that are likely to influence the VPF. In seeking to establish the nature of the changes in survival probability that underpin the variations in life expectancy that are being valued under this approach it is important to appreciate that, ceteris paribus, the process of ageing by one year is tantamount to a unit rightward shift in the “starting point” (i.e. the origin) of an individual‟s “Gompertz Curve”, (or equivalently, a unit leftward shift in the curve itself), this curve being the relationship that expresses an individual‟s hazard rate as a function of his/her age where the hazard rate is the probability of death at a given age conditional on surviving to that age (Haybittle, 1998). On the assumption that the Gompertz curve takes an essentially exponential form then such a unit rightward shift in the origin will be tantamount to a proportionate increase in the hazard rate at each additional year of age for the individual concerned and the absolute impact will therefore increase as the hazard rate itself increases with age along the exponential Gompertz curve2. See Figure 2.1 for a diagrammatic presentation of this effect.

2 More specifically, with the Gompertz function taking the form eh(t) βt , where h(t) is the hazard

rate at time t and and β are positive constants, a unit rightward shift in the origin would yield a new

Gompertz function eth 1)β(t ˆ which is equivalent to multiplying all of the original hazard rates by

a factor eβ.

23

Figure 2.1 Gompertz curve showing the effect of increasing the starting age

by one year

Conversely, the effect of a one-year gain in life expectancy can be regarded as a proportionate reduction in each of the age-specific hazard rates, with the size of the reduction increasing in later years of life. In short, under Approach 2 one can think of the principal source of the gain in life expectancy as being the proportionate reduction in the larger, later-life hazard rates. This clearly contrasts markedly with Approach 1 under which, as already noted, the gain in life expectancy is exclusively the result of a reduction in the current period hazard rate. This then raises two further important points, namely:

To the extent that, ceteris paribus, total lifetime utility is an increasing but strictly concave function of length of life, then the utility gain from the current period hazard rate reduction that underpins Approach 1 will exceed the corresponding utility gain resulting mainly from the later-life hazard rate reductions that are implicit in Approach 2, given that both approaches involve a gain of one year in life expectancy as a result of the hazard rate reductions.

The question of which approach yields the more policy-relevant result clearly depends on the nature of the hazard rate reduction generated by the particular health-care or safety improvement programme under consideration. Thus, prima facie it would appear that if the value of a QALY was required as an input to the assessment of the desirability of say, a medical intervention whose effect would be more or less immediate, then Approach 1 would seem to be the appropriate source of such a value. By contrast, if the focus was on a reduction in air pollution or health effect (such as the beneficial impact of a protracted medical treatment), then to the extent that epidemiological evidence suggests that such effects will manifest themselves principally via proportionate changes in individual hazard rates (and will therefore have their

0 1 2

Hazard rate

Gompertz curve for

starting age x+1

Gompertz curve for

starting age x

Effect of

increasing

starting age

by 1 year

Years since starting age

24

main impact in later life-years), then Approach 2 would appear to be more relevant.

All of this having been said there is, however, one fairly serious complicating factor as far as Approach 2 is concerned and this relates to the way in which the VPF does in fact vary with age, and hence with remaining life expectancy. More specifically, growth of annual income over early years of adult life, together with market imperfections in the form of restrictions on borrowing against higher levels of future income might reasonably be expected to yield an inverted-U over the lifecycle for the typical individual‟s WTP for a current-period safety improvement. Nonetheless, it also seems reasonable to suppose that if life-cycle variations in income and wealth were controlled for, then this WTP for safety improvement would be an unambiguously increasing function of remaining life expectancy (and hence a decreasing function of age) and to a large extent these intuitions are confirmed by theory (Shepard and Zeckhauser, 1982). However, several empirical studies (Jones-Lee et al., 1985) have found that even when factors such as income are controlled for, WTP for safety improvement still follows an inverted–U life-cycle, so that at least over early years of adult life, individual valuation of safety is a strictly decreasing function of remaining life expectancy. To the extent that Approach 2 essentially treats the value of a life year as the rate at which the VPF increases with remaining life expectancy then it is clear that, applied to younger adult age groups, this approach will yield a negative value of a life year which is, to say the least, simply beyond the bounds of credibility. Is Approach 2 therefore fundamentally flawed? We think not. Rather, we would argue that the tendency for the valuation of safety to rise with age (and hence decline with remaining life expectancy) over early years of adult life, even when income effects are controlled for, is in fact a reflection of a fundamental change in attitude to risk and awareness of vulnerability to physical harm that would appear to be a common feature of the process of maturation for many people over the period from their late teens to their mid-twenties. It would therefore seem reasonable to argue that since the VPF-age relationship over early years of adulthood is largely a result of a fundamental change in preferences and attitudes rather than a change in an individual‟s future hazard rates then estimation of the value of a gain in life expectancy should be based only on the time interval over which the VPF is a decreasing function of age. In response to this argument it might, of course, be objected that in empirical studies such as that reported in Jones-Lee et al., 1985 the “inverted-U” WTP vs. age relationship does not peak before about the age of 40, which would appear to sit rather uncomfortably with the suggestion that for the typical individual the “recklessness of youth” will have passed by his or her mid-twenties. But in considering this issue it should be borne in mind that the point at which the inverted-U WTP vs. age relationship peaked in the Jones-Lee et al. study was to all intents and purposes driven by the specification of the regression relationship in which the non-linear effect of age (which was evident in the raw data) was captured by an “age minus mean age all squared” variable so that, setting aside the effect of the linear age variable (which was generally insignificant), it was inevitably the case that the “inverted-U” relationship reached a maximum at mean age. While it has to be conceded that this is something of a limitation of the estimated VPF-age relationships they are nonetheless arguably the best estimates currently available.

25

All things considered, therefore, it was felt appropriate to apply Approach 2 not only to the full available data set, but also to a subset of the data with those in the younger “problematic” age-group trimmed-out. More specifically, Approach 2a looks at the relationship between EandM ii from age

18 onwards, therefore incorporating the values for those aged between 18 and 40 as well as those in the older age group. By contrast, Approach 2b focuses exclusively on those aged over 40 on the basis that the lower values of life amongst the under 40s do indeed, reflect a problematic “recklessness of youth”. 2.3.2.1 Approach 2a

Since neither theory nor empirical evidence will allow us to specify the precise relationship between Mi vs. Ei we assume the following simple functional form: Mi = αEi

β (2) With the Mi-versus-age relationship estimated in three UK studies being essentially an inverted-U shape, peaking in middle age Jones-Lee et al., 1995; Jones-Lee et al., 1985; Carthy et al., 1999), then, based on these studies, and as a first approximation,

it would seem appropriate to take the average of M

M i - where M is the population

mean of Mi – as being about 0.45 for an 80 year old (for whom average remaining life expectancy is 8 years), about 1.36 for a 40 year old (with an average remaining life expectancy of 39 years) and about 0.7 for a 18 year old (with an average remaining

life expectancy of 60 years). Given this, and taking M = £1,427,340 along with a zero discount rate with no adjustment for declining quality of life in later years, would give, from equation (2): lnMi = lnα +βlnEi (3) Incorporating the data for a 18, 40 and 80 year it is necessary to calculate the β using

the formula

2x

xywhere ii EEx lnln and MiMy i lnln . Table 2.1 presents the

calculations used to estimate x and y for each of the three ages.

Table 2.1: Calculation of x and y

Age lnEi lnMi ii lnElnE lnMilnM i

18 4.109 13.814 0.836 -0.074

40 3.680 14.478 0.406 0.590

80 2.029 13.372 -1.243 -0.515

Mean 3.273 13.888

26

Using the values that are presented in Table 2.1, β can be calculated using equation (4):

β =

2x

xy (4)

β = 41.2

819.0

243.1406.0386.0

)515.0*243.1()590.0406.0()074.0*836.0(222

(5)

β = 0.339 (6) Substituting (6) into (4) lnMi = lnα +0.339lnEi (7) lnα = lnMi - 0.339lnEi (8) lnα =13.88 – (0.339*3.273) (9) α = 353,655 (10) Hence from equations (4), (6) and (10) Mi = 353,655Ei

0.339 (11) From which it follows that

i

i

dE

dM= 120,216Ei -0.66 (12)

Given that Ei varies across the UK population, there exists a density function giving the proportion of the population having each particular value of E i. The general form of this density function can be derived from statistical life tables. For the sake of simplicity it will be assumed that the density function for Ei takes the form:

(Ei) = 2

1

iE (13)

which entails that

60

8

2

1

iE dEi = 1 (14)

so that

60

8

i

2

3

xE3

2

= 1 (15)

Hence, 294.75 = 1 (16)

from which it follows that = 0.00339 (17)

27

From equations (12), (14) and (17) we therefore have

V2a = 60

8i

2

1

i

0.66-

i dE 0.00336E)x (120,216E (18)

= 60

8i

0.160-

i dE396E (19)

= 60

8

0.839

i472E (20)

= 23,199 (21) Hence, V2a = £23,199 Although across the whole age range the WTP per QALY value is positive, it is low because of the negative valuations which arise from those under the age of 40. 2.3.2.2 Approach 2b

As previously noted above, impacts of wealth and age such as possible risk seeking behaviour by younger individuals‟ results in the inverted U shaped relationship between Mi and Ei. If a policy maker were to wish to discount such aspects, then in trying to overcome them empirically, it is possible to focus only on those members of

society at or above middle age. As in Approach 2a, it is assumed that M

M i as being

about 0.45 for an 80 year old and about 1.36 for a 40 year old (for whom average remaining life expectancy is 39 years). Therefore, under Approach 2b the aim will be to arrive at an estimate of the mean of

i

i

dE

dMtaken over the subset of the population within this age group.

As in Approach 2a we take this relationship as having the following form: Mi = αEi

β (22) To solve equation (22) it is first necessary to take logs of both sides to make it into a linear equation. Thus equation (22) becomes: lnMi = lnα +βlnEi (23) When Ei = 8: lnα +βln8 = ln0.45x1,427,340 (24) When Ei = 39: lnα +βln39 = ln1.36 x1,427,340 (25)

28

Dividing (25) from (24) results in:

8

39lnβ

45.0

36.1ln

β = 64317.1

10599.1

β = 0.67309 (26) Substituting (26) into (24) to obtain α: lnα + (0.67309ln8) = ln0.45x1,427,340 lnα = 13.37 – 1.3595 α = 164,932 (27) Hence from equations (22), (26) and (27): Mi =164,932Ei

0.67309 (28)

From which it follows:

i

i

dE

dM(0.67309 x 164,932)Ei

-0.3269 (29)

= 111,013Ei-0.3269 (30) As in Approach 2a it is assumed the probability density function takes the form:

(Ei) = 2

1

iE (31)

which entails that

39

8

2

1

iE dEi = 1 (32)

so that,

39

8

i

2

3

xE3

2

= 1 (33)

Hence, 147.3 = 1 (34)

from which it follows that, = 0.00678 (35)

29

From equations (30), (32) and (35) we then have:

V2b= i

2/1

i

39

8

3269.0

i dEEx00678.0xEx013,111

(36)

= i

39

8

1731.0

i dEE86.747 (37)

= 39

8

17309.1

iE52.637 (38)

=40,029 (39) Hence, V2b = £40,029 2.4 Results

Table 2.2 presents the estimates of the value per life year from each country.

Table 2.2: Value per Life Year (All values in Euros)

Denmark France Hungary Netherlands Norway Spain Sweden UK

Approach 1 43,510 45,064 6,625/

37,235 64,991 56,454 76,723 56,537 63,371

Approach 2a 20,518 21,500 3,136/

17,577 32,176 27,659 40,492 26,558 33,562

Approach 2b 38,909 41,109 5,249/

29,274 59,016 51,729 62,465 58,206 56,791

Discounting was also incorporated using a relevant discount rate for each country (see Appendix 2.1 for a list of discount rates used). Table 2.3 shows the estimates of the discounted life year per country.

Table 2.3: Value per Discounted Life Year (All values in Euros)

Denmark France Hungary Netherlands Norway Spain Sweden UK

Approach 1 69,984 53,580 12,996/

72,849 97,533 104,934 128,871 90,075 81,154

Approach 2a 31,120 26,016 6,865/

38,483 48,345 45,143 70,517 40,219 42,953

Approach 2b 78,471 52,228 13,609/

72,895 104,294 318,983 131,838 143,888 82,091

30

Quality of life information was included by applying a quality of life weight to the life year data. This weight was based on EQ-5D population norm scores for each country (see Section 2.2.2.2). As also discussed in Section 2.2.2.2, there were no suitable quality of life scores for France or Norway, thus values of a QALY are not presented for these two countries. Table 2.4 presents values per QALY for each country.

Table 2.4: Value per QALY (All values in Euros)

Denmark France Hungary Netherlands Norway Spain Sweden UK

Approach 1 50,810 10,589/

59,356 76,640 107,764 70,323 80,591

Approach 2a 23,156 4,131/

23,153 36,937 51,154 31,785 39,473

Approach 2b 44,452 7,774/

43,631 69,399 84,310 70,119 68,359

Discounting was also introduced into the calculation of a value per QALY. Table 2.5 presents values per discounted QALYs for each country.

Table 2.5: Value per discounted QALY (All values in Euros)

Denmark France Hungary Netherlands Norway Spain Sweden UK

Approach 1 79,892 20,023/

112,234 180,295 178,527 110,961 102,373

Approach 2a 32,754 9,124/

51,145 55,274 92,488 50,712

50,524

Approach 2b 87,752 19,196/

104,290 122,598 171,476 168,152 77,884

These four tables present the estimates of the undiscounted and discounted value per life year and value per QALY for each country. As was noted in section 2.2.1 there was not sufficient data for Palestine or Poland to be able to estimate values. Thus, these tables present results for the remaining eight countries. For each country, values were elicited in the home currency (e.g. Norwegian Kroner for Norway) and then converted to Euros using exchange rates from the European Central Bank (European Central Bank, 2007). The results are presented here only in Euros for the purposes of comparing the countries. Country specific calculations in the home currency are presented in Appendix 2.3. Further discussion of the results is presented in the next section.

31

2.5 Discussion

2.5.1 Summary of results Two approaches have been developed to elicit the monetary value of a QALY based on pre-existing values of prevented fatalities. Approach 1 is based on making the assumption that each future life year is of equal value, while Approach 2 assumes that the relationship between the VPF and age is an inverted U shape, this is based on empirical evidence from the UK and Sweden. For each approach results were calculated to give estimates of the value of a life year, value of a discounted life year, value of a QALY and the discounted value of a QALY for each country. For all of these different estimates the values which arise from Approach 1 are higher than those from Approach 2. This is the case for each country. Discounting and quality of life adjustments lead to higher values than those which are based only on undiscounted life years. This is because the initial value is based on the ratio of the VPF to the number of life years. The introduction of quality adjustments or discounting results in the denominator falling (i.e. the number of discounted QALYs is less than the number of life years) and the ratio increases. Comparing the results from each country, Spain has the highest estimates of both the value per life year and value per QALY. The lowest values are from Hungary calculated using their lower VPF value. It may have been expected a priori that the resulting value per QALY would be higher for the Northern European Countries compared with those in the South and East. The results as shown in Tables 2.2-2.5 indicate that this is not necessarily the case. Examining the country specific data alongside the results it appears that there are two key variables which influence the results; these being the VPF and the discount rate. There is a wide range of VPFs which are used across the countries (see Appendix 2.1 for a full list of values). The highest VPF is for Spain and this corresponds with the highest estimates across all of the approaches and when discounting and quality of life are incorporated. The lowest VPF is from Hungary which corresponds with the lowest estimates of the undiscounted value per life year and QALY. However, Hungary has a high discount rate of 5.5% which pushed up the discounted estimates of the value of a life year and QALY. 2.5.2 Missing data There were a number of pieces of missing data and as a result of this it was not possible to calculate a full set of values for all ten countries. As was shown in Appendix 2.1 there was not sufficient data from Palestine or Poland to estimate any values. If data become available on the VPF in the future it would be possible to elicit the value per life year based on the approaches outlined in Section 2.3. An alternative would be to use data from a country which has a similar economic and population structure as a proxy in the calculations. The quality of life data used in this research is based on EQ-5D population norms which are specific to each country. EQ-5D data is used as it is in the form of utility values which can then be used to weight the life expectancy data. Population norm

32

values were not available for Norway and France so values per QALY and discounted QALY were not elicited for these countries. At present we are not aware of utility based quality of life data for these countries which could be substituted into the calculations. One possible way to overcome this would be to use population norms from another country which has similar population health characteristics. Alternatively, if SF-36 quality of life data existed for a country then this could potentially be converted into utility scores using the SF-6D. At present there is no official discount rate for Spain so a discount rate of 3.5% was assumed. This is the recommended discount rate from the European Commission as presented in the “Guidance on the Methodology for Carrying Out Cost-Benefit Analysis” (European Commission Directorate-General, 2006). 2.5.3 Caveats Underlying the conduct of this research is the implicit assumption that individuals‟ preferences for health gains from health care interventions are the same as their preferences for reductions in the risk of death through road safety improvements. Making this assumption allows for a direct conversion of the VPF into the value of a QALY. However, there is no empirical evidence available which has looked at whether individuals‟ preferences for road safety are the same as their preferences for health care. There are a number of different ways in which the VPF can be estimated; the two main ways being contingent valuation or human capital approaches. The preferred method by many economists is the contingent valuation method. This method elicits individuals‟ WTP for a reduction in the risk of death which, aggregated across a group, would results in the prevention of one premature death. However, the country specific VPFs which are used in this report are not all WTP based. Appendix 2.1 indicates which VPFs were elicited using the contingent valuation method. As the VPFs are calculated in different ways caution must be taken when interpreting the results and when comparing the results across the different countries. Approach 2 attempts to take into account that there is not a linear relationship between the VPF and age but instead that there is an „inverted U‟ shaped relationship which peaks around the age of 40. The two versions of Approach 2 use this relationship as the basis of the calculations of the value of a life year or QALY. Empirical studies from the UK and Sweden both support the existence of an „inverted U‟ relationship (Carthy et al., 1999; Persson et al., 2001). However, as no other country elicited a WTP-based VPF using a single study there is currently no data available on this from any of the other 8 countries. Therefore it was assumed in the calculation of Approach 2 for each country, that the distribution is the same as in the UK and also that the ratios of the VPF to age at age 18, 40 and 80 are the same as in the UK. 2.5.4 Comparisons with current „threshold‟ values As two different approaches have been developed to estimate the monetary value of a life year and QALY it is difficult to make a direct comparison between the results estimated in the study and the current threshold values used by each country. At

33

present only the UK, France, Sweden and the Netherlands have a threshold value which is in the public domain, although these are generally not explicitly acknowledged by the health care decision making body within each country. In the UK NICE does publicly state that their threshold is £25,000-30,000, above which a strong case must be made for an intervention with a higher cost per QALY to be accepted (National Institute for Health and Clinical Excellence, 2005). Taking the estimates of the value of a QALY (Appendix 2.3, Tables A3) the estimate calculated using Approach 2a is below the £25,000-30,000 but the values from Approach 1 and 2b are higher at £55,708 and £47,253 respectively. These values are somewhat higher than those currently used by NICE even allowing for inflation since the date of the original publication of the NICE recommendations. It is the case for each of the countries that approach 2a gives a lower estimate; this is an artefact of using the full age distribution with the assumption of an inverted U shaped relationship between age and the VPF. In the Netherlands a threshold value of €20,000 is often cited (Brouwer et al., 2008). The values estimated in this study from all three approaches are substantially higher There are a range of threshold values for Sweden with no explicit value. These values range from SKK400,000 up to SKK655,000 (Hjalte et al., 2005; Persson and Hjelmgren, 2003). The estimates calculated here are closer to the higher end of this range. However, this is not unexpected as the SKK655,000 threshold value is also based on converting the value of a statistical life in to the value of a QALY. There is also no official threshold for France, however, an unofficial value of €50,000 is used (Garber, 2000). This value is higher than the values per life year which have been calculated in this study using all three methods, although the values for Approach 1 and 2b are in the region of €40,000. 2.6 Conclusion

This report has set out two different approaches to estimating a monetary value of a QALY which can be used in policy making. Estimates of the value of a life year and value of a QALY have been elicited for 8 European countries using these approaches. The report has highlighted the types of issues which arise when attempting to estimate a monetary value of a QALY from secondary data sources. Fundamental to this research is the assumption that individuals‟ preferences for health gains from health care interventions are the same as their preferences for reductions in the risk of death through road safety improvements. However, at present there is no evidence to support this assumption. Although it was expected that the estimates from Approach 2a would be lower, the large variability within an approach was less expected. The variation in the discount rates and the VPF drive the estimates and highlight how sensitive this type of approach to estimating the value of a QALY is to the data that is entered into the model.

34

As a result of the problems highlighted here it is demonstrated that while this type of approach is useful in providing a first attempt at estimating a monetary value of a QALY, it shows the need for primary research to be conducted in this area.

35

Chapter 3: Estimating a WTP-based value of a QALY through survey research: methods and preliminary results

3.1 Introduction

As stated in the introduction, two main approaches to eliciting the value of a QALY - „chained‟ and „direct‟ - were tested. In this chapter, we start by providing a conceptual overview of the approaches. This then leads to more-specific descriptions of the designs for testing of the two approaches, each involving several sub-groups of our sample being randomly allocated different versions of the chained and direct questionnaires. This is followed by a guide through the questionnaires, issues of sampling and the analytic strategy, before offering results and discussion. 3.2 Conceptual overview

A conceptual overview of the surveys is displayed in Figure 3.1 on page 78 of this document. More details on the several different versions of the questionnaire are given in the following section. 3.2.1 Chained The chained approach set out to build upon previous research that had attempted to estimate WTP per QALY by breaking the exercise down into two distinct components. First, respondents would be asked to complete an utility assessment exercise in order that their utility value (between 0 and 1) for a given health state could be ascertained. Next they would be asked their WTP to avoid a given duration/risk of that health state. Combining the respondent‟s answers to both components then allows that respondent‟s WTP per QALY gained to be estimated (essentially by „multiplying up‟ their WTP for a known fraction of a QALY into one whole QALY). For example, if we know that a respondent is willing to pay £1000 to avoid one year with certainty in a health state with utility value of 0.95 (i.e. a loss of 0.05 or 1/20th of a QALY), we can estimate their WTP per QALY to be 20 * £1000 = £20,000 per QALY gained (assuming linearity). The basic principle behind the chained approach is that the „health losses‟ being considered in the WTP component of the exercise are not too large such that stated WTP values would likely be subject to „budget constraints‟. For example, most people would „value‟ the prevention of a certain catastrophic health event (such as going blind or losing the use of their limbs) to be greater than the sum they could possibly afford to pay; and of course, in reality, people are unaccustomed to being asked to pay to avoid the certainty of such large health losses. Rather, they are more accustomed to thinking about paying to insure themselves against the risk of some catastrophic event. Alternatively, people are more accustomed to considering the prospect of out-of-pocket payments to treat more minor ailments. Hence, the expected health losses people are used to considering in real purchasing decisions are relatively modest. Whilst keeping the QALY gains being valued „small‟ is a common feature of previous „chained‟ approaches to estimating WTP per QALY, the novel aspect of the research reported here is that we also strove to keep the QALY gains constant across

36

respondents. This will be explained in detail below. But, as outlined in section 1.2 above, we used two different types of WTP question: „risk variant‟ and „time variant‟ questions. Risk variant WTP questions ask respondents about their WTP to avoid some risk of a health state (that they had previously given a utility value for), with the risk allowed to vary across respondents in order to keep the QALY gain constant across respondents. So for example, for a respondent with a utility value for a given health state of 0.90, avoiding a 50% chance of that health state for one year would amount to an expected QALY gain of 0.05 QALYs (i.e. a 50% chance of avoiding the loss of 0.10 QALYs). In contrast, a respondent with a utility value of 0.8 for that same health state would only need to avoid a 25% chance of that health state for one year to gain 0.05 QALYs (i.e. a 25% chance of avoiding the loss of 0.20 QALYS). In this way „risk‟ varies in order to keep the QALY gain constant across respondents (in this case at 1/20th of a QALY). In time variant WTP questions, respondents are asked about their WTP to avoid some duration of a given health state (again that they have previously given a utility value for) and the duration varied in order to keep the QALY gain constant across respondents. For example, for the respondent with a utility value of the health state 0.90, avoiding 6 months in that state with certainty, would amount to a gain of 0.05 QALYs (0.10 * 6/12). The respondent with a utility value for the health state of 0.8 would need only avoid three months in that health state with certainty to gain 0.05 QALYs. As outlined in section 1.2 above, we elected here to „set‟ the QALY gains at 0.05 and 0.10. To the best of our knowledge no researchers have previously adopted such an „interactive‟ approach but, rather have used predetermined risks and/or durations in the WTP component of the chained approach. 3.2.2 Direct The direct questionnaire tested the notion of presenting health gains „directly‟ using a simple graphical and textual description. Most of the health gains presented were of one-QALY, avoiding the need to „multiply up‟ WTP values for smaller health gains to generate a value for one QALY. Hence all of the gains were for durations of at least one year and most involved no risk or probabilities. However, we sought to build in an „overlap‟ with the chained approach and included questions offering smaller health gains of similar magnitude (0.05 to 0.25 QALYs) to those in the chained questionnaire. Three of these questions presented the health gain using a risk format where respondents paid to avoid a fixed probability (5 or 10%) of a loss of one QALY. To keep the format simple we avoided the description of specific health states in the direct questionnaire. Health was described in terms of a simple „health thermometer‟ type scale from 0 (equivalent to dead) to 100 points (equivalent to full health). A detailed introduction explained the concept of the health thermometer to respondents and illustrated possible values with reference to specific health states described by the EQ5D instrument. The introduction also explained the graphical presentation of health profiles used in the questions. These simple graphs consisted of a y-axis representing health on the 0-100 point scale and an x-axis representing the lifetime of the respondent in years. The graphs for each question were customised to characteristics entered by the respondent at the start of the survey so that health gains could be presented in the context of the respondent‟s own age, reported health and life expectancy.

37

Health gains could be achieved via avoidance of losses in health status only or through gains in survival. In the case of the former, the health state utility losses presented to respondents were fixed at either 25 or 10 point losses for 4 or 10 years respectively. To overlap with the size of the health gains offered in the chained survey, avoidance of a 25 or 10 point loss over one year were also valued. Respondents were simply asked about their maximum WTP to avoid these states, each of which amounts to a one QALY gain, and these could be presented as though the one QALY loss would occur imminently or near the end of their life. For example, a QALY could be generated through receipt of an extra year of life in full health. However, logically this QALY can only be added on at the end of one‟s life (although in reality is more likely to be a result of an accumulation of several small reductions in risk to life over one‟s life span). Valuing such a gain is likely to be problematic, as, for many respondents, it might be seen as accruing so far into the future as to not be worth very much, even although it is a large gain in terms of physical magnitude when it arises. As a way around this, therefore, we attempted to bring forward the gain of one QALY by devising a „coma‟ scenario, whereby one could pay to avoid losing a QALY now. In order to avoid various emotional factors „contaminating‟ the valuations, respondents were told that they could „just pick up where they left off‟ in terms of relationships were they to have been in a coma. In another version the gain of one QALY was portrayed as arriving in a situation in which the respondent had to imagine s/he was in a state of terminal illness. In order to match the size of QALY gains asked about in the chained version and also to allow some conventional risk-based questions to be posed to respondents, three versions of the questionnaire asked about WTP to avoid either a 10 per cent or five per cent chance of being in a state that involved either a 25 (or 10) point loss in health for 4 (or 10) years. Finally, one WTP question involved payment over a 4-year period as opposed to full payment occurring now. 3.2.3 Complementarities and overlaps across both surveys Across both the chained and direct questionnaires, it can be seen that several sizes of QALY gain were valued. Most obviously, these range from QALY gains of 0.05 to 1. However, due to capping3, there are some gains valued which are less than 0.05. These gains are valued across a range of certain and uncertain scenarios, with some payments spread out. Most gains arise now and some at end of life and the gains also cover different QALY types – some enhancing quality of life only, others extending life and some involving threats to life itself. Nevertheless, the attempt to cover such a range of valuation scenarios poses methodological challenges. At one extreme, in the chained approach, because such small gains are being valued, the value-of-a-QALY estimates rely on assumptions of proportionality (e.g. that the value of one QALY is 20 times the value attached to 0.05 of a QALY). At the other extreme, most of the questions in the direct approach involve (what would generally be perceived as) large QALY gains. This is problematic with respect to the „budget constraint‟ issue whereby, beyond a threshold gain (which

3 Where respondents in the chained survey gave particularly high valuations for a health state it was not possible to offer them a risk of one year in that health state that would equate to 0.05 or 0.1 QALY loss. In these cases respondents were offered one year in that health state, for which the utility loss was 'capped' at less then the desired 0.05 or 0.1 QALY offered to other respondents.

38

could vary by individual) respondents will always state the same WTP because they reached their maximum at the threshold. In such situations the value of a QALY will be less the larger the QALY gain presented to respondents, as the same WTP is simply being divided across a greater number of QALYs. Other methodological challenges apply to very specific questions: first, if higher values result in the terminal illness question, this may be simply because people‟s marginal utility of income is substantially lower in that context compared with other scenarios, which might result in high values of a QALY; and, second, theory would predict that values arising from one-off payments now would be the same as those arising when payments are spread over four years. The design, therefore, through being so comprehensive, enabled a range of important issues to be addressed and values of a QALY to be elicited for several different scenarios. In broad terms, our hypotheses were:

H1: WTP for larger health gains would be greater than for smaller such gains; H2: One-QALY gains arising at the end of life would be valued less than gains of the same magnitudes arising more imminently; H3: A one-QALY gain arising from extension of life, as opposed to health improvement for existing years of life, would be valued higher. H4: WTP for a QALY would be higher under the chained as opposed to the direct approach. H5: WTP for terminal illness would give highest value of QALY due to effect on marginal utility of income

H1 is based on a standard test of scope. However, with non-linearities in the data, and following H4, resulting values of a QALY derived from larger health gains may indeed be less than those arising from presentation of smaller gains. H2 arises not only because of discounting but also because of the related notion that, although one extra year or more of survival represents a large gain, it is not valued highly by younger people when merely added to end of life. This applies even when the more imminent gains are presented in the form of health improvements for existing years of life and also when life extension is presented as arising closer to the present day. Nevertheless, all else equal, life-extending QALYs should be valued more highly than those arising only from health improvements to existing years (as in H3). H4 arises from experience with the earlier SVQ Project, in which values of QALY were large - although attempts are made within EuroVaQ to ensure that lower values are obtained. Furthermore, with smaller gains being valued in a risk-based format in the chained method, if respondents reach a WTP limit no matter the size of gain they are considering, the chained will lead to higher values of a QALY. Finally, H5 arises from the idea that, when placed in a situation of terminal illness, respondents will „blow the lot‟ on staying alive because they do not have to be concerned about their income beyond their demise. Of course, this could be countered by an argument that many would accept their fate (i.e. that they are going

39

to die soon anyway) and might rather ensure that such money is left to their loved ones or in some other form of legacy. 3.3 Questionnaire development Both questionnaires were piloted on UK respondents using a computer-based questionnaire similar in design to the final surveys. In addition the questionnaires were piloted on small groups in four other countries. The pilots took place with support and observation from researchers and extensive feedback from respondents. Both questionnaires started with a brief introduction to explain that the survey related to “valuing health” and offered assurances with regard to confidentiality of data etc. A set of demographic questions followed to ascertain:-

Age

Gender

Geographic area where respondent lived

Employment status4, nature and responsibility of employment, education and qualifications of head of household

Number of adults and children living in the household

Household and personal income Apart from the two entirely different valuation approaches used, one main difference in the questionnaires was that for the „chained‟ the respondent was always presumed to be in „good health‟, whilst for the „direct‟ the respondent‟s current health state was used to frame questions. Piloting highlighted the need to subdivide questions for both surveys across a number of different blocks to reduce respondent fatigue. 3.4 Sampling As is standard procedure with internet-based surveys, invitations were sent out via an internet survey provider, Survey Sampling International (SSI), to the 8 countries where the company held their own panels. Around 10% were accepted and the questionnaire started. The respondent would not have known the contents of the survey prior to accepting and „response rates‟ for each country were similar to those normally observed in such surveys – see Table 3.1 below.

4 Questions relating to „head of household‟ were not asked of all respondents; where the survey

company already held information on „head of household‟ for the respondent these questions were omitted.

40

Table 3.1: Response rates to SSI panel invitations5

Country

EuroVaQ questionnaire (SSI panel only) Average SSI response rate Invites Starters Response rate

UK 87312 4932 5.6% 6%

France 58386 7360 12.6% 10%

Spain 35850 2917 8.1% 6%

Denmark 31716 3169 10.0% 9%

Norway 29618 2288 7.7% 8%

Netherlands 61773 9544 15.5% 15%

Sweden 45752 5492 12.0% 8%

Poland 16977 3152 18.6% 16%

Hungary - -

Palestine - -

Respondents were also recruited using SSI's partner panels. Unfortunately data on the total numbers of invitations sent by SSI's partner panels is not available. After accepting an „invite‟, respondents were allocated to either the direct or chained questionnaire using a controlled randomisation process which ensured even numbers of participants completing each block of questions for the chained and direct surveys. Of those starting, a number were screened out where the quota for respondents from that demographic was completed and not all of the remaining respondents completed the questionnaire. In Table 3.2 below, column (b) includes the total number of respondents who started the survey (from both SSI's own and partner panels). The remaining columns show the numbers who completed and who dropped out prior to completing the survey (incompletes). As can be seen, around half the respondents starting a survey were screened out. The completion rates for the direct and chained surveys are shown in columns labelled „% completed‟.

5 The above data refers to responses at 10.02.2010. Further invites were sent out to boost responses

for some countries where responses were low in some demographic segments. We would expect the response rate to be lower for invites targeting population segments with low response rates after that date.

41

Table 3.2: Total number of starters and completion rates

The basic strategy was to recruit first from SSI panels and then move to SSI partner panels in order to fulfil any targets that were running behind. Increased incentives were also used for demographics that were particularly difficult to fill. The direct survey went live online on 23 November 2009 and the chained on 3 December 2009. With the exception of Palestine, recruitment to each closed on 28 February 2010. The survey in Palestine posed particular problems due to the translation. The translation was challenging, delaying the launch of the survey in Palestine. SSI‟s partner panel in Palestine were only able to obtain a small unrepresentative sample so further attempts were made to increase this, including SSI emailing respondents who were not members of any specific internet survey panel. Finally a programming error led, unfortunately, to respondents being directed to an English version of the chained survey rather than the Arabic translation. A partner specialising in phone recruitment was also engaged in Spain, but this failed to recruit required numbers and so this strategy was abandoned for other countries. The final sample was broadly representative but with significant under-representation of elderly females in most countries. In addition, in some countries the national distribution of socio-economic classes was not available and socio-economic class was substituted with household income. For those countries, it was evident that lower income groups were under represented. Table 3.3 and 3.4 below include a breakdown of the sample achieved per country.

42

Table 3.3 - Achieved sample – Chained questionnaire6

Netherlands UK France Spain Sweden Norway Denmark Poland Hungary

Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample

Male 18-25 6.4% 5.3% 6.4% 6.2% 6.4% 5.6% 6.4% 7.1% 6.4% 6.2% 6.4% 6.4% 6.4% 4.9% 8.8% 10.6% 8.0% 4.4%

Male 26-35 10.4% 8.4% 8.8% 8.3% 8.8% 9.1% 10.4% 10.3% 8.0% 7.2% 8.8% 10.8% 8.0% 6.8% 8.8% 9.4% 9.6% 8.9%

Male 36-45 10.4% 8.5% 9.6% 9.3% 9.6% 9.8% 9.6% 12.3% 8.8% 8.9% 9.6% 10.0% 10.4% 8.6% 8.8% 9.4% 7.2% 7.8%

Male 46-54 8.8% 8.2% 7.2% 7.6% 8.0% 7.4% 7.2% 8.5% 7.2% 6.8% 8.0% 8.4% 8.0% 8.0% 8.8% 9.0% 8.8% 8.0%

Male 55-64 6.4% 9.6% 6.4% 7.4% 5.6% 7.8% 6.4% 5.9% 8.0% 7.4% 8.0% 9.1% 7.2% 8.2% 5.6% 5.4% 6.4% 6.9%

Male 65+ 7.2% 7.3% 8.8% 8.4% 8.8% 6.7% 8.8% 5.0% 9.6% 7.7% 8.0% 8.8% 8.8% 11.6% 6.4% 2.3% 6.4% 1.9%

Female 18-25 6.4% 5.7% 6.4% 7.0% 6.4% 6.8% 5.6% 7.8% 6.4% 7.4% 6.4% 6.9% 5.6% 5.5% 8.8% 14.5% 7.2% 11.1%

Female 26-35 9.6% 8.8% 9.6% 9.0% 9.6% 12.4% 9.6% 14.4% 8.0% 8.5% 8.8% 10.0% 8.0% 8.1% 8.8% 12.5% 9.6% 15.8%

Female 36-45 9.6% 9.6% 9.6% 9.6% 9.6% 12.2% 9.6% 12.5% 8.8% 11.4% 9.6% 10.1% 10.4% 11.0% 8.8% 11.3% 7.2% 13.9%

Female 46-54 8.0% 10.0% 8.0% 9.0% 8.0% 10.0% 7.2% 8.8% 7.2% 9.4% 8.0% 9.5% 7.2% 8.5% 9.6% 9.1% 9.6% 12.0%

Female 55-64 6.4% 10.3% 7.2% 7.4% 6.4% 8.8% 7.2% 5.7% 8.8% 9.3% 7.2% 7.2% 7.2% 9.2% 6.4% 5.8% 8.0% 8.0%

Female 65+ 10.4% 8.3% 12.0% 10.9% 12.8% 3.4% 12.0% 1.7% 12.8% 9.9% 11.2% 2.7% 12.8% 9.7% 10.4% 0.5% 12.0% 1.4%

Region 1 18.4% 18.7% 16.0% 16.0% 12.8% 12.5% 17.6% 15.7% 18.4% 20.4% 36.8% 37.4% 30.4% 32.5% 12.8% 13.7% 28.8% 33.9%

Region 2 20.8% 21.3% 12.0% 11.1% 12.0% 14.1% 11.2% 11.8% 13.6% 12.3% 19.2% 18.3% 21.6% 22.0% 12.0% 11.7% 11.2% 12.2%

Region 3 20.8% 20.5% 16.0% 15.7% 7.2% 12.4% 16.0% 16.2% 17.6% 18.0% 20.0% 20.1% 8.8% 9.0% 12.0% 11.2% 9.6% 7.8%

Region 4 40.0% 39.6% 11.2% 10.5% 28.0% 13.1% 12.8% 12.7% 13.6% 20.3% 13.6% 13.8% 15.2% 14.8% 13.6% 15.0% 9.6% 9.7%

Region 5 13.6% 14.8% 16.8% 13.5% 14.4% 11.8% 20.8% 16.0% 10.4% 10.4% 24.0% 21.7% 11.2% 10.4% 12.0% 11.6%

Region 6 13.6% 14.7% 13.6% 16.2% 15.2% 17.3% 16.0% 12.8% 13.6% 15.2% 11.5%

Region 7 17.6% 17.3% 9.6% 18.3% 12.8% 14.5% 15.2% 14.0% 13.6% 13.2%

Region 8 10.4% 10.4%

S.G. A/Qt. 1 14.4% 13.8% 11.2% 13.5% 7.2% 14.3% 5.6% 12.4% 20.0% 10.5% 20.0% 12.9% 14.4% 20.6% 20.0% 10.7% 20.0% 8.6%

S.G. B/Qt. 2 12.8% 12.6% 9.6% 10.6% 12.8% 12.9% 5.6% 9.2% 20.0% 13.7% 20.0% 18.1% 14.4% 11.4% 20.0% 12.0% 20.0% 21.0%

S.G. C1/Qt. 3 24.0% 24.1% 13.6% 16.0% 16.8% 20.5% 9.6% 20.0% 20.0% 16.7% 20.0% 13.9% 33.6% 36.3% 20.0% 16.1% 20.0% 12.3%

S.G. C2/Qt. 4 25.6% 20.7% 19.2% 18.2% 28.0% 22.5% 15.2% 17.9% 20.0% 13.6% 20.0% 17.9% 16.0% 9.8% 20.0% 20.0% 20.0% 16.8%

S.G. D/Qt. 5 12.8% 12.5% 26.4% 23.0% 16.8% 14.8% 14.4% 12.6% 20.0% 33.4% 20.0% 24.8% 12.0% 12.8% 20.0% 27.8% 20.0% 28.0%

S.G. E/no inc. 10.4% 16.3% 20.0% 18.9% 18.4% 15.1% 49.6% 28.0% 12.0% 12.4% 9.6% 9.1% 13.4% 13.2%

6 Tables display the percentage of the entire country sample that should fall into each category to provide a nationally representative sample(‘Target’) and the actual proportion of the sample falling into that category(‘Sample’)

43

Table 3.4: Achieved sample - Direct questionnaire7

Netherlands UK France Spain Sweden Norway Denmark Poland Hungary Palestine

Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Target Sample Sample

Male 18-25 6.4% 4.5% 6.4% 5.8% 6.7% 5.6% 6.1% 7.1% 6.1% 5.1% 6.4% 7.2% 6.1% 4.4% 8.8% 9.5% 8.0% 6.7% 31.9%

Male 26-35 10.1% 7.1% 9.1% 7.8% 9.3% 8.9% 10.1% 11.4% 8.3% 6.7% 9.1% 10.7% 8.3% 6.4% 9.1% 11.5% 9.6% 13.5% 22.0%

Male 36-45 10.4% 7.3% 9.3% 9.6% 9.3% 8.6% 9.9% 12.0% 8.8% 8.4% 9.9% 9.7% 10.4% 8.3% 9.1% 12.5% 7.5% 8.0% 11.5%

Male 46-54 8.5% 8.2% 7.7% 8.9% 8.0% 7.6% 7.2% 7.9% 7.2% 6.8% 8.0% 8.3% 7.7% 9.3% 8.8% 9.2% 8.5% 11.1% 5.3%

Male 55-64 6.4% 14.2% 6.7% 7.9% 5.9% 9.7% 6.4% 7.2% 8.3% 9.1% 7.7% 8.0% 7.2% 8.4% 5.3% 7.4% 6.4% 7.1% 1.3%

Male 65+ 7.2% 8.1% 8.5% 7.7% 8.8% 6.4% 8.8% 6.2% 9.3% 8.4% 8.3% 9.0% 8.8% 12.8% 6.4% 1.6% 6.7% 4.1% 0.0%

Female 18-25 6.1% 4.3% 6.4% 6.4% 6.7% 5.9% 5.9% 7.8% 6.1% 5.5% 6.4% 6.9% 5.9% 4.7% 8.5% 13.1% 7.5% 7.2% 15.5%

Female 26-35 9.9% 8.3% 9.6% 9.3% 9.3% 9.4% 9.9% 13.6% 8.3% 8.2% 8.8% 9.1% 7.7% 8.5% 8.8% 11.4% 9.3% 14.1% 10.2%

Female 36-45 9.9% 9.3% 9.6% 9.7% 9.3% 11.6% 9.6% 10.7% 8.8% 10.5% 9.3% 9.9% 9.9% 11.7% 9.1% 10.2% 7.7% 8.6% 2.3%

Female 46-54 8.3% 9.4% 7.7% 7.9% 8.0% 11.5% 7.2% 8.5% 7.5% 10.3% 7.7% 8.4% 7.5% 7.9% 9.3% 8.2% 9.3% 9.5% 0.0%

Female 55-64 6.4% 11.8% 6.9% 8.1% 6.1% 10.2% 6.7% 6.3% 8.5% 10.5% 7.5% 8.1% 7.5% 8.1% 6.4% 5.3% 7.7% 7.3% 0.0%

Female 65+ 10.4% 7.6% 12.0% 10.8% 12.5% 4.7% 12.3% 1.3% 12.8% 10.4% 10.9% 4.7% 13.1% 9.6% 10.4% 0.1% 11.7% 2.8% 0.0%

Region 1 17.9% 20.4% 16.3% 16.6% 12.8% 11.8% 17.9% 14.8% 18.1% 18.4% 36.5% 36.8% 30.7% 29.4% 13.1% 15.1% 28.5% 34.1% 26.3%

Region 2 21.1% 20.0% 12.3% 12.6% 12.0% 12.7% 11.5% 12.3% 13.6% 13.4% 19.5% 18.1% 21.3% 22.0% 12.0% 11.3% 10.9% 10.6% 21.4%

Region 3 21.1% 22.2% 15.7% 15.3% 7.7% 8.5% 15.7% 15.6% 17.1% 16.3% 20.3% 19.2% 9.1% 10.8% 11.7% 11.5% 9.9% 7.3% 20.7%

Region 4 40.0% 37.4% 11.5% 10.2% 27.7% 20.0% 12.3% 13.1% 13.9% 18.5% 13.9% 14.7% 14.9% 15.6% 13.6% 14.4% 9.6% 9.0% 3.9%

Region 5 13.6% 14.0% 16.5% 15.6% 14.1% 11.1% 21.1% 17.0% 9.9% 11.3% 24.0% 22.2% 10.9% 9.2% 12.5% 13.0% 19.1%

Region 6 13.3% 13.0% 13.6% 15.7% 15.5% 17.7% 16.3% 16.3% 12.8% 13.7% 15.2% 11.9% 8.6%

Region 7 17.3% 18.3% 9.6% 15.6% 13.1% 15.5% 15.7% 13.6% 13.3% 14.1%

Region 8 10.1% 11.1%

S.G. A/Qt. 1 14.4% 12.2% 11.2% 13.3% 7.5% 10.8% 5.6% 16.0% 20.0% 10.7% 20.0% 14.7% 14.4% 18.0% 20.0% 11.1% 20.0% 10.2% 4.2%

S.G. B/Qt. 2 12.5% 10.8% 9.9% 10.9% 13.1% 13.5% 5.6% 9.4% 20.0% 15.9% 20.0% 20.2% 14.7% 11.0% 20.0% 10.9% 20.0% 22.2% 8.5%

S.G. C1/Qt. 3 24.0% 24.0% 13.9% 15.7% 16.8% 21.5% 10.1% 16.1% 20.0% 16.3% 20.0% 14.5% 33.3% 38.8% 20.0% 16.6% 20.0% 13.9% 12.2%

S.G. C2/Qt. 4 25.6% 19.1% 19.2% 19.0% 27.7% 27.6% 15.2% 18.0% 20.0% 15.2% 20.0% 17.8% 16.3% 10.6% 20.0% 18.7% 20.0% 15.9% 23.0%

S.G. D/Qt. 5 13.1% 14.2% 26.4% 22.4% 16.5% 11.9% 14.1% 11.6% 20.0% 32.7% 20.0% 22.5% 11.7% 12.0% 20.0% 28.0% 20.0% 26.3% 52.1%

S.G. E/no inc. 10.4% 19.7% 19.5% 18.7% 18.4% 14.8% 49.3% 29.0% 9.3% 10.3% 9.6% 9.5% 14.8% 11.4% 0.0%

7 Small differences in the target percentages across the two surveys are a product of integer rounding of the different sample sizes used for each survey.

44

Overall, we aimed to recruit a quota of 35,000 respondents. The aim was to recruit 2000 respondents per country for the chained approach and 1500 per country for the direct, and for these to be representative by age and gender, socio-economic status (or income where data on socio-economic breakdowns were unavailable) and region, although it became apparent during the tender process that this would not be possible for Palestine. The primary analysis reported here has been undertaken on the set of 'complete' answers – those respondents reaching the end of the survey. However we were also able to collect data from respondents who started the survey but did not complete it. The incomplete data contained only a small fraction of completed questions when compared to the complete data. This data was analysed to quantify to what extent respondents' demographic characteristics and their responses differed from those completing the survey. Differences in respondent characteristics between those completing the survey and those not completing it were small; respondents failing to complete the direct questionnaire had a mean age two years older than those completing it. The incomplete data was not analysed further. 3.5 Chained approach

3.5.1 The utility assessment component Every respondent completed a utility elicitation exercise on two EQ-5D health states, namely 211218 and 222229, referred to henceforth as the yellow and green states respectively. These health states were chosen as we wanted to include both mild and moderate health states and because yellow „dominates‟ the green (in that it is better on three dimensions and no worse on the other two). This provides us with a simple consistency test in that the utility value of the yellow state ought to be greater than that of green. We avoided using more severe health states as we wanted to avoid states that large numbers of respondents would consider to be worse than dead. Whilst all respondents valued both states, health states utilities were derived either by standard gamble (SG) or time trade off (TTO). Each is described in detail below. 3.5.1.1 The standard gamble procedure The SG is based on the „probability equivalence‟ method whereby respondents are faced with the certainty of an intermediate outcome and some probability (p) of a better outcome and some probability (1-p) of a worse outcome. The respondent‟s task is to set the probability „p‟ so that he is indifferent between the certain and uncertain outcomes. In health state utility assessment, the certain outcome is the „target‟ health state for sure (in this case yellow or green) and the better and worse outcomes normal health and death respectively. Assigning values of 1 and 0 to normal health and death respectively, and assuming the respondent is an Expected Utility (EU) maximiser, then it is easy to show that the value of „p‟ at the point of indifference is equal to the utility value of the target health state.

8 Health state 21121 represents some problems in the mobility and pain dimensions of HRQoL with no

problems in the self-care; usual activities and anxiety/depression dimension. 9 Health state 22222 represents some problems in all five dimensions.

45

Respondents were first taken through a „risk introduction‟ exercise designed to familiarise them with the format of the SG (and later risk variant WTP questions) and to try and consider what a risk of „X in 100‟ really meant. Respondents were introduced to the concept of „X in 100‟ by considering 100 „smiley‟ pink faces – of which „X‟ turned purple to denote suffering an illness. Respondents were first presented with the 100 smiley pink faces and asked to select one. The computer then selected one face at random and turned that purple to denote illness. Respondents were told that the chances of the face that they selected turning purple was „1 in 100‟. This exercise was repeated and the computer chose 20 faces at random to denote a risk of ‟20 in 100‟. The same „smiley face‟ format was then used for the SG questions themselves. Respondents were asked to consider two situations; with and without treatment. Without treatment they would be in the target health state for sure for the rest of their lives; whilst, with treatment, they faced some probability „p‟ of full health (EQ-5D state 11111) and some probability (1-p) of immediate death. In all cases, the starting values of „p‟ and „1-p‟ were 0.60 and 0.40 respectively. That is the „with treatment‟ option presented respondents with the prospect of a 60% chance of full health and a 40% chance of death (see Figure 3.2 below). The full health state (EQ-5D 11111) is denoted by the „pink health state‟.

Figure 3.2: Example SG question

Respondents were then asked to consider whether they would prefer not to have the treatment, prefer to have the treatment, or considered both options to be equally good. One of three things then happened depending on their response at this first iteration:

For those respondents preferring the „without treatment‟ option, the chances of full health with treatment increased to 85% (with the corresponding chances of death falling to 15%)

46

For those respondents preferring the „with treatment‟ option, the chances of full health with treatment decreased to 20% (with the corresponding chances of death rising to 80%).

For those respondents who thought both options were equally good, this was taken to signify indifference and they were considered to attach a utility value of 0.60 (i.e. „p‟) to the target health state.

This procedure continued until respondents were indifferent between the two options or they had reached the end of a devised algorithm with a given number of iterations, (see Figure 3.3 below).

Figure 3.3: SG iterations

When no point of indifference was reached before the end of the algorithm, the utility value was assumed to be the mid-point of the value of „p‟ where they had „switched‟ between preferring one option to preferring the other or vice versa. For example, if a respondent preferred the „without treatment‟ option when the chances of full health with treatment was 95% (and hence chances of death 5%), but preferred „with treatment‟ when the chances of full health with treatment were 99% (and hence chances of death 1%), that respondent was considered to have a utility value of 0.97 (i.e. [0.99+0.95]/2. In the majority of cases, respondents were faced with a maximum of four SG questions before their utility value could be estimated. The exception was where respondents continued to prefer the „without treatment‟ option even when the chances of full health with treatment were 99% (and hence chances of death 1%). These respondents were then asked a supplementary question to try and accurately estimate their true utility value of the target health state (see Figure 3.4 below).

47

Figure 3.4: „Extreme‟ SG responses

Again if respondents „switched‟ to preferring treatment, for example, between risks of death of 1 in 100 and 1 in 200, their utility value of that health state was considered to be 0.9925 (i.e. the midpoint of 0.99 and 0.995). Those respondents who did not accept a 1 in 10,000 risk of death were considered to have a utility value of „1‟ for that health state and were classified as „non–traders‟. We return to the issue of non-traders below. 3.5.1.2 The time trade-off procedure The TTO procedure is often used in health state utility assessment as an alternative to SG as it does not involve risk, and hence is often considered easier for respondents to complete. The TTO asks respondents to choose between a longer life in an impaired health state (the target state) or a shorter life in full health. The principle behind the TTO is one of „constant proportional trade off‟ i.e. respondents will always trade off the same proportion of available time in order to avoid the target state. For example, if respondents are faced with 10 years in the target state or X years in full health (where X<10) and are indifferent between the two situations, their utility value of the target health state is taken to be X/10. So, for example if the respondent is indifferent between spending 10 years in the target state and seven years in full health, their utility value for the target health state is considered to be 0.70. One difference between the TTO procedure used here and that commonly used in TTO exercises (see Dolan et al., 1996, for example) is that we set the duration of the TTO to correspond roughly to the respondent‟s likely life expectancy, rather than to some nominal number of years (such as 10 years used in the Dolan et al. study irrespective of the respondent‟s age). Respondents were first asked their age and a rough life expectancy was then estimated for them (assuming they would live to 80 years). So for example, respondents aged 52 were then told; „very roughly then we can say that people of your age are likely to live for another 30 years or so…‟ They were then asked to consider two different „lives‟; A and B. Life A was remaining life expectancy, Y (in this case 30 years), in the target health state and life B was X years (where X< Y) years in full health. In all cases, the starting value of „X‟ was set at 60% of Y, the respondent‟s remaining life expectancy (in this case X= 18 years). They were then

48

asked to consider whether they would prefer life A, prefer life B or they considered both lives to be equally preferable (see Figure 3.5 below).

Figure 3.5: Example TTO question

As in the SG, one of three things then happened depending on their response at this first iteration:

For those respondents preferring „life A‟, the time in full health in life B was increased to 85% of life expectancy (in this case 25 years and six months).

For those respondents preferring „life B‟, the time in full health in life B was decreased to 20% of life expectancy (in this case six years).

For those respondents who thought both lives A and B were equally preferable, the value of „X‟ was taken to signify their indifference value and they were considered to attach a utility value of 0.60 (in this case 18/30) to the target health state.

As in the SG, this procedure continued until respondents were indifferent between lives A and B or they had reached the end of a devised algorithm with a given number of iterations (see Figure 3.6 below).

49

Figure 3.6: TTO iterations

It is important to note that the TTO algorithm was designed to replicate exactly that used in the SG procedure (by simply swapping chances of full health in the SG with proportion of life expectancy in the TTO). So, for example, respondents reaching indifference at the first iteration in either the SG or TTO would be considered to have a utility value of 0.60 for that health state. Again when no point of indifference was reached before the end of the algorithm, the utility value was assumed to be the mid-point of the value where they had „switched‟ between preferring one life to the other or vice versa. As in the SG procedure, respondents were faced with a maximum of four TTO questions before their utility value could be estimated. The exception was where respondents continued to prefer „Life A‟ even when the years spent in full health in life B were 99% of life expectancy (in this case 29 years, 8 months and 12 days) They were then asked whether they would be willing to give up any time in order to avoid the target health state (see Figure 3.7 below).

Figure 3.7: „Extreme‟ TTO responses

Those respondents who indicated that they would be willing to give up some time were invited to write in the number of weeks they would be willing to give up. Those who said they would not be willing to give up any time at all were considered to have a utility value of „1‟ for the target health state and are classified below as „non traders‟.

50

3.5.2 The WTP component As above, the „chained‟ approach to estimating WTP per QALY breaks the exercise down into two distinct components. The „first‟ component (although see below for details of the question order used in the survey design) is the elicitation of utility values for the health state - described in section 3.5.1 above. The second is the elicitation of respondent‟s WTP to avoid some risk/duration of that health state. Respondents were first presented with an introductory screen as set out below (emphasis included as presented to respondents):

„THE NEXT QUESTIONS ARE A LITTLE DIFFERENT. Now we want you to think about what value you would place on avoiding the [target] health state. For each of these questions you will be asked if you would be willing to pay something to avoid being in the [target] health state. When

you are thinking about what it would be worth to you to avoid this, please try to forget about any loss of income that might happen as a result of being in the [target] health state - please suppose that your income is unaffected - and just focus on how that state would affect your quality of life. When you state your willingness to pay it is very important that you consider what you actually would be willing to pay to avoid being in the [target] health state. When answering the following questions, think about how big an impact the extra payment would have on your household‟s budget and keep in mind that the money you spend to avoid being in the [target] health state could be spent on other things‟.

The exact nature of the question that followed then depended on the type of WTP question involved; risk variant, time variant or „one year with certainty‟ - see below. Henceforth, all WTP questions that relate to the green health state (EQ-5D state 22222) are prefixed by „G‟ and those that relate to the yellow health state (EQ-5D state 21121) are prefixed by „Y.‟ 3.5.2.1 Risk variant WTP questions In the majority of risk variant WTP questions respondents were asked to think about the value they would put on avoiding some chance of suffering the target health state for one year. The „chances‟ were set according to a) the respondents‟ utility value derived in the utility assessment component and b) whether the QALY gain was „set‟ at 0.05 or 0.10 QALY. So, for example a respondent who attached a utility value of 0.60 to the yellow health state would be presented with a 13% chance of that health state in a 0.05 QALY gain question and 26% chance of that state in a 0.10 QALY gain question. As the QALY gain for one year with certainty is 0.4 QALYs (from 0.6 up to 1), multiplying by 13% and 26% yields 0.05 and 0.10 respectively (after rounding to whole numbers). An example risk variant WTP question is given in Figure 3.8 below.

51

Figure 3.8: Example risk variant WTP question

Respondents were first asked whether they would be willing to pay something for the treatment, even a small amount. Respondents who were willing to pay something were then presented with a „card sort‟ procedure (see 3.5.2.4 below). Respondents who were unwilling to pay anything at all were asked supplementary questions designed to elicit the reasoning behind such a response (see 3.5.2.5 below). As above, the majority of the risk variant questions asked the respondent about their WTP to avoid some chance of the target health state for one year. These questions are referred to below as Gj and Gd (relating to the green health state and 0.05 and 0.10 QALY gains respectively) and Yi and Yh (relating to the yellow health state and 0.05 and 0.10 QALY gains respectively). A comparison of responses to a) Gj and Gd and b) Yi and Yh then offer straightforward tests of sensitivity to scope. That is, all other things equal, we would expect the responses to the 0.10 QALY gain question to be roughly double those to the 0.05 QALY gain question and, hence, estimates of WTP per QALY to be roughly equivalent. If, on the other hand, respondents were, on average, willing to pay the same for the 0.05 and 0.10 QALY gains, the estimated WTP per QALY derived via the 0.05 QALY gain question would be roughly twice that estimated via the 0.10 QALY gain question. There was one other risk variant WTP question included that was a bit different to those detailed above. Question Gf related to the green health state and a 0.10 QALY

gain, but used a 10-year duration, rather than one year and adjusted the risk accordingly (i.e. essentially divided the risk presented in question Gd by 10). This question was included as we were keen to test the impact of presenting respondents with a longer duration of the health state. This was considered important as one of the advantages of using a risk-based chained approach is that it could potentially be used to value the prevention of chronic health states.

52

3.5.2.2 Time variant WTP questions The other main set of WTP questions were time variant WTP questions in which the respondent was asked to pay to avoid some duration of the certainty of a health state. In this case durations were set according to a) the respondents‟ utility value derived in the utility assessment component and b) whether the QALY gain was „set‟ at 0.05 or 0.10 QALY. So, for example a respondent who attached a utility value of 0.60 to the green health state would be presented with a 3-month duration of that health state in a 0.10 QALY gain question and 1.5-month duration (expressed as 1 month and 15 days) in a 0.05 QALY gain question. As the QALY gain with certainty for 12 months is 0.4 QALYs (from 0.6 up to 1), we are essentially dividing that duration by 4 and 8 to arrive at QALY gains of 0.10 and 0.05 respectively. An example time variant WTP question is given in Figure 3.9 below.

Figure 3.9: Example time variant WTP question

Respondents were again first asked whether they would be willing to pay something for the treatment, even a small amount. Again, respondents who were willing to pay something were then presented with a „card sort‟ procedure (see 3.5.2.4 below) and those who were unwilling to pay anything at all a supplementary question (see 3.5.2.5 below). The time variant WTP questions are referred to below as Gm, and Gn (relating to the green health state and 0.05 and 0.10 QALY gains respectively) and Yq and Yr (relating to the yellow health state and 0.05 and 0.10 QALY gains respectively). Similar to above, comparing responses to a) Gm and Gn and b) Yq and Yr offer straightforward tests of sensitivity to scope. Again, all other things equal, we would expect the responses to the 0.10 QALY gain question to be roughly double those to the 0.05 QALY gain question and, hence, estimates of WTP per QALY to be roughly equivalent. It should also be obvious that Gm and Gn are the time variant counterparts of the risk variant questions Gj and Gd. Likewise, Yq and Yr are the time variant counterparts

53

of the risk variant questions Yi and Yh. A comparison of responses to a) Gj and Gm, b) Gd and Gn, c) Yi and Yq, and d) Yh and Yr, then offer a test of sensitivity of WTP responses to the „framing‟ of the WTP question. All other things equal, we would expect the responses to risk variant and time variant to be roughly equivalent and, hence, estimates of WTP per QALY to be roughly equivalent. 3.5.2.3 Additional WTP questions In addition to the risk and time variant WTP questions described above, „one year with certainty‟ questions were included in the chained survey. These questions were included to allow more overlap between the chained and direct study designs. These questions are fundamentally different than those described above in that the QALY gain valued in the WTP component is allowed to vary across respondents (as individuals will have different utility values for the health state in question). They also differ in that the QALY gain being valued is potentially much greater than in the questions outlined above that „set‟ the gain at 0.05 and 0.10 QALYs. For example, the median valuation of the state 22222 (the green state) in the MVH study (Dolan et al., 1996) was approx 0.60, indicating that avoiding a year in that health state with certainty would amount to 0.40 QALYs, and of course many respondents will value the state much lower than that. The one year with certainty WTP questions are referred to below as Ga and Yb relating to the green and yellow health states respectively. As the green health state is strictly worse than the yellow state we would expect respondents to pay more to avoid a year in the green health state than a year in the yellow state. Finally, one additional question was asked referred to below as Ye. This question was similar to Yb in that it involved paying to avoid the yellow health state for one year with certainty. The difference here was that, rather than ask for a „one off‟ payment as in the remainder of the WTP questions, this question asked the respondent to state their WTP each year for a period of four years. This question was included as we were keen to assess respondents‟ sensitivity to the duration of the payment period. All other things equal, we would expect the total WTP over four years to be roughly equivalent to the one off payment and, hence, estimates of WTP per QALY to be roughly equivalent. Please refer to Figure 3.10 below for a summary of all WTP questions for the chained questionnaire.

Figure 3.10: Summary of the WTP questions in chained questionnaire

QALY value

WTP question Green health state

EQ5D 22222 Yellow health state

EQ5D 21121

0.05 100% time Gm Yq

0.05 1 year risk Gj Yh

0.10 100% time Gn Yr

0.10 1 year risk Gd Yi

0.10 10 year risk Gf

Varies One year with certainty Ga Yb

Varies One year with certainty –four annual payments

Ye

54

3.5.2.4 The card sort procedure For all WTP questions, those respondents who indicated that they were willing to pay at least something for the treatment were then taken through a „card sort‟ procedure. Respondents were presented with a series of 15 money amounts in random order and asked to decide whether they definitely would pay that amount, definitely wouldn‟t pay that amount or whether they were unsure whether or not that would pay that amount. For each amount in turn, respondents were invited to „click and drag‟ the card into one of three boxes; „definitely would pay‟, „definitely wouldn‟t‟ or „unsure‟. Once respondents had clicked and dragged all 15 amounts into the boxes the interactive programme then presented back to them their highest ‟definitely would pay‟ amount and lowest „definitely would not pay‟ amounts. They were then asked whether they would be willing to pay more than their highest „definitely would pay‟ amount and, if so, were invited to write in that amount (in the example given, somewhere between £7500 and £15,000) and that amount was taken to be the best estimate of their maximum WTP for the treatment. If they were unwilling to pay more than their highest „definitely would pay‟ card, then that amount was taken to be the best estimate of their maximum WTP for the treatment.

Figure 3.11: Example card sort procedure (shows risk variant WTP question)

The card sort procedure was programmed so that respondents were not allowed to be inconsistent in their responses. That is, if the respondents had placed a higher amount in the definitely would pay than they had in the definitely wouldn‟t pay, the programme flagged this up to respondents and invited them to rearrange the amounts in the boxes. Likewise, respondents were constrained to write in an amount that was between their highest definitely would pay amount and lowest definitely

55

would not pay. Again the programme alerted respondents who entered values outside this range to their „mistake‟ and prompted them for an amount in between the stated amounts. An important design feature of the card sort procedure used in the chained survey was that the set of cards presented to respondents aimed to keep the „range‟ constant in terms of implied WTP per QALY. That is the card sort range presented in the 0.10 QALY questions was exactly twice that presented in the 0.05 QALY questions. This was done in order to try and reduce any „framing‟ effects on WTP per QALY introduced solely as a result of the chosen card sort amounts. So, for example, a respondent placing the lowest four cards into the „definitely would pay‟ box would be indicating the same WTP per QALY irrespective of the magnitude of those QALY gains. Whilst it could be argued that this is „leading‟ respondents to give consistent WTP per QALY values, we would argue that it is giving the approach the „best shot‟ at uncovering robust WTP per QALY estimates. As previous research has shown the WTP per QALY estimates to vary with size of QALY gain on offer, we consider a „best shot‟ approach here is justifiable. In the case of the „one year with certainty‟ questions, where the magnitude of QALY gain varied across respondents, the range of card sort amounts presented varied accordingly. Thus, respondents in the UK version of the chained survey were presented with:

1. In 0.05 QALY gain questions, amounts ranging from £5 to £15,000 (i.e. from £100 to £300,000 per QALY);

2. In 0.10 QALY gain questions, amounts ranging from £10 to £30,000 (i.e. from £100 to 300,000 per QALY);

3. In the „one year with certainty‟ questions, amounts that varied with their utility value of the health state in order to keep range from £100 to £300,000 per QALY.

The card sort amounts were converted from UK currency to the prevailing local currency using Purchasing Power Parity conversion rates for 2008 and rounded to two significant figures. 3.5.2.5 Unwilling to pay anything at all

As above, whenever a respondent indicated that they would not be willing to pay anything, even a small amount, they were presented with a supplementary question to ascertain the reason behind such a response. Respondents were presented with a series of possible reasons, derived from a combination of previous experience and the comments made during piloting. These comprised:

It wouldn't be too bad/I could live with it.

I would get better anyway, so it is not worth paying for the treatment.

I do value the treatment, but I cannot afford to pay anything for it.

I do value the treatment, but do not want to pay because the government should provide health care.

The risk is low, I would take the chance10.

Other (please specify below).

10 This option was only included for risk questions.

56

Respondents were asked to tick all statements that applied to them. Any respondent ticking the „other‟ category was then invited to write in their reason in a free text format. These responses have been translated but have not yet been analysed. Four of the remaining five options could be taken to be „legitimate‟ reasons for being unwilling to pay anything at all. That is, the respondent is stating that they cannot afford to pay anything for the treatment or that they consider the benefits of the treatment to be trivial. The response „I do value the treatment, but do not want to pay because the government should provide health care‟ is rather different in that it could be interpreted as a „protest‟ response i.e. the respondent is using a zero bid in order to make a statement about how they believe health care ought to be provided. We return to this issue in section 3.6.2.4 below. 3.5.3 The chained questionnaire

Piloting of the chained questionnaires helped inform the final wording of the survey and the number of questions it was plausible to ask of one respondent. It was concluded that – along with completing utility assessment exercises on two health states (either by SG or TTO) - 5 was the maximum number of WTP questions a respondent could reasonably be expected to answer. To ensure adequate coverage of all questions above and the inclusion of certain within-respondent consistency tests (see below), we elected to develop eight different versions of the survey which were subsequently programmed and hosted on the internet. As none of the key results are presented by version below, familiarity with the versions is not essential to understanding the results. For completeness, however, we outline the eight versions below. 3.5.3.1 The eight versions To ensure that the largest range of scenarios was covered and the testing of the methodological issues outlined above, eight versions of the chained questionnaire were devised. A summary is given in Figure 3.12 below.

Figure 3.12: Overview of chained questionnaire versions 1-8

Figure 3.13a (p79) gives a reminder of the individual WTP questions: Figure 3.13b (p80) then shows the structure of versions 1-4 in detail whilst Figure 3.13c (p81) shows the structure of versions 5-8. The main difference is that the SG was used in the utility assessment component in versions 1-4 whilst the TTO was used in

57

versions 5-8. Versions 1, 2, 5 and 6 included risk variant WTP questions whilst versions 3, 4, 7 and 8 included time variant WTP questions. In this way, respondents answering both risk or time variant questions in the WTP component could have done either SG or TTO in the preceding utility assessment component. Respondents were initially randomised to one of the eight versions. Within each version they were further randomised to begin by answering questions relating to either the green or yellow health state. Those randomised to yellow first would complete an utility assessment exercise (either SG or TTO) on the yellow health state and then go on to answer the set of WTP questions relating to the yellow health state (the set depending on which of the eight versions the respondent was randomised to). They would then go on to complete an utility assessment exercise on the green health state before answering the set of WTP questions relating to the green health state (again the set depending on version). Those randomised to green first did the opposite. Further, the order in which the WTP questions were presented was randomised wherever possible in order to avoid introducing any systematic „ordering‟ bias. See Figures 3.13b and 3.13c for details. All versions finished with the respondent completing a self-reported EQ-5D. 3.6 Direct approach

3.6.1 Study design for direct questionnaire As can be seen in Figure 3.14 below, the direct approach involved trying to value a QALY via 13 different scenarios. Most of these scenarios present one-QALY gains for certain, for example, in scenario A respondents are asked to value avoiding a 0.25 QALY (25 point) loss for four years. More conventional risk-based questions were also asked, allowing us to value smaller QALY gains more in line with the chained approach.

Figure 3.14: Summary of WTP questions in Direct questionnaire

QALY gain Question

Health Loss /Gain Duration When Probability

1 A 25 point loss 4 years 1 year's time Certainty

1 B 25 point loss 4 years End of life Certainty

1 F 10 point loss 10 years 1 year's time Certainty

1 G 10 point loss 10 years End of life Certainty

1 I Months of life 12+ months End of life Certainty

1 J Months of life 12+ months 1 year's time Certainty

1 L Months of life 12+ months 1 year's time Certainty

1 P 25 point loss 4 years 1 year's time Certainty

0.25 D 25 point loss 1 years 1 year's time Certainty

0.1 E 10 point loss 1 years 1 year's time Certainty

0.1 M 25 point loss 4 years 1 year's time 10% risk

0.1 O 10 point loss 10 years 1 year's time 10% risk

0.05 N 25 point loss 4 years 1 year's time 5% risk

58

The 13 scenarios were distributed across four main versions of the questionnaire, labelled 1, 2, 3 and 4, or a fifth version referred to as the „grey block‟. Figure 3.15 (p82) illustrates the versions and the ordering of the questions, with pairs of questions (shaded and adjacent) being asked in random order. Risk based questions, which might have been more cognitively challenging were nearly always placed at the end of the version. In Version 2 the two risk questions were followed by the „terminally-ill‟ question. There was also a staged payment question (P) which was placed after other questions to limit confusion over the timing of payments. Randomisation across the four versions was controlled to ensure similar completion numbers for each version across each subgroup of the nationally representative sample for each country. 3.6.2 Direct questionnaire A comprehensive introduction (Appendix 3.1 attached) was presented to show respondents how they might rate their health on a visual analogue scale or „health thermometer‟, following which respondents were asked to enter the age to which they expected to live (life expectancy) and to indicate how healthy they thought they were (health state). The latter was done by the respondent moving the point on a „health thermometer‟, 100 representing best imaginable health for their age and zero representing death (see Figure 3.16 below). Respondents could not indicate a health state below zero and their current age was auto-filled from the demographics section.

Figure 3.16: Age, health and life expectancy

Each respondent was allocated to one of the four main versions and their current age, life expectancy and health state were used throughout the questionnaire to generate questions and accompanying graphs specific to them. Respondents indicating poor health or a low life expectancy (relative to current age) were excluded from questions which would have presented a scenario taking them to either a very low health level (below 10 points) or beyond their life expectancy. However, where multiple questions would have been excluded, the respondent was instead directed to the 'grey block' - this version contained health loss/gains of one-year duration only.

59

Figures 3.17 and 3.18 below illustrate how questions were presented to the respondent. In both examples, the respondent was aged 52 and had stated that s/he expected to live until 80 and had rated their health at 90 - this is illustrated by the pale blue shaded area of the graph. Figure 3.17 relates to Question A and the darker blue area of the graph indicates a 25 point drop in health for four years, starting in one year‟s time.

Figure 3.17: Question A

In the next example (Figure 3.18) the health loss posed by Question G is 10 points for 10 years, later in life. The respondent‟s current age, life expectancy and health state is the same, again illustrated by the pale blue shaded area, but this time the darker blue area of the graph indicates a 10 point drop in health for 10 years.

Figure 3.18: Example of graph in Question G

60

3.6.2.1 WTP questions

The basis for all 13 questions is set out below and an example of each has been included at the end of this document (see Appendices 3.2 – 3.14). The structure of each question was the same throughout:

1. A written scenario accompanied by a graph to help illustrate gain/loss/risk. 2. Respondents were asked to say if they would be willing to pay to avoid a

loss/risk or achieve a gain.

If they replied „Yes‟, respondents were routed to a „card sort‟ similar to that in the chained questionnaire to elicit their WTP.

If they replied „No‟, respondents were asked to say why by selecting one or more reasons from a list of set responses. A free text box was also provided if respondents wanted to add their own reason, and, for one question (L) which related to „postponement of death‟, respondents who agreed to pay for the health gain were also asked to say why they were prepared to pay such an amount.

3. Versions 1 and 2 also included an explanation of risk much the same as that described at 3.5.1.1 above for the chained questionnaire.

3.6.2.2 The four versions and grey block

3.6.2.2.1 Version 1

The first two questions both referred to a 25 point drop in health for four years. The drop in health in Question A would happen in 12 months time, whereas the drop in Question B would occur later in life. If a WTP was indicated for either Question A or B, the respondent was asked to say why they valued one „more‟, „less‟ or „equal to‟ the other.

Question O referred to a 10% risk of a 10 point drop in health for 10 years starting in 12 months time.

The scenario in Question P was the same as Question A, except that respondents were able to pay in four annual instalments.

3.6.2.2.2 Version 2

The first two questions both offered „additional life‟: Question I was straightforward and simply asked whether the respondent was willing to pay for X-months11 of time at the end of life, whereas Question J asked if the respondent was willing to pay to avoid being in a coma for X-months11

of time „now‟. If a WTP was indicated for either Question I or J, the respondent was asked to say why they valued one „more‟, „less‟ or „equal to‟ the other.

Two „risk‟ questions, M and N, then followed. Although both questions referred to a „risk‟ of a 25 point drop in health over four years, the risk in Question M was 10% whilst that in Question N was only 5%.

11

The duration offered to respondents was customized to their own VAS health rating so that the gain (at that health) amounted to one QALY. Hence those reporting a health of 50 points were offered 24 months (at health 50).

61

Question L asked respondents if, diagnosed with a terminal illness, they would be willing to pay to extend their life by X-months11.

3.6.2.2.3 Version 3

The first two questions both referred to a 10 point drop in health for a period of 10 years. The drop in health in Question F would happen in 12 months time, whereas the drop in Question G would happen later in life. Again, if a WTP was indicated for either Question F or G, the respondent was asked to say why they valued one „more‟, „less‟ or „equal to‟ the other.

The next two questions both related to a one-year drop in health starting in 12 months time. In Question D this was a 25 point drop in health, whereas in Question E this was a 10 point drop.

3.6.2.2.4 Version 4

Questions A and F both referred to a drop in health in 12 months time: „A‟ was a 25 point drop in health for four years and „F‟ a 10 point drop for 10 years.

Question I (additional time at end of life) and Question J (coma question) offered the respondent X-months11 of time and again, if a WTP was indicated for either question, respondents would be asked to say why they valued one „more‟, „less‟ or „equal to‟ the other.

The „terminal illness‟ question (L) followed.

3.6.2.2.5 Grey block

As previously mentioned, those respondents who indicated poor health or a low life expectancy (relative to their current age) would have been directed to the grey block (see criteria at 3.6.2.3 below). The grey block consisted of 2 sets of 2 questions. The first 2 questions related to a one-year drop in health starting in 12 months time. One related to a 25 point drop in health (Question D) and the other a 10 point drop (Question E). The second 2 questions offered X-amount11 of „additional life‟ either in 12 months time (Question J) and at the end of life (Question I).

3.6.2.3 Exclusion criteria

To ensure questions were relevant to individual respondents, exclusion criteria applied throughout the questionnaire. Respondents would be excluded from all versions and redirected to the demographics12 if their health state was less than 20 points. If their life expectancy was less than 2 years, respondents would be redirected to the demographics/attitudinal questions12 and, if their life expectancy was less than six years, they would be redirected to the grey block. Exclusion criteria were also applicable to each version of the questionnaire:-

Version 1 o Excluded from Question O if life expectancy less than 12 years.

12

The intention was to divert these respondents to the end of the survey to collect demographic data. Unfortunately, a problem with the online survey led to these respondents being excluded.

62

o Excluded from Question B if life expectancy less than nine years. o Redirected respondent to grey block if current health was less than 35

points.

Version 2 o Excluded from Questions M and N if current health was less than 35

points.

Version 3 o Excluded from Question G if life expectancy less than 15 years. o Excluded from Question F of life expectancy less than 12 years. o Excluded from Question D if health state less than 35 points. o Redirected respondent to grey block if health state less than 35 points

and life expectancy less than 12 years.

Version 4 o Excluded from Question F if life expectancy less than 12 years. o Excluded from Question A if health state less than 35 points.

Grey block o Excluded from Question D if health state less than 35 points. o Excluded from Question J if life expectancy less than 100/health +1

years.

The intention was to ensure that all health gains were complete at least one year before the respondent expected to die and that no health losses reduced a respondent‟s health to below 10 points. In addition we did not offer respondents the same health gain now and at the end of life if the absolute difference between the start of the two health gains was three years or less for that respondent. The final exclusion for the grey block excludes Question J unless the coma duration ends at least a year before the respondent expects to die. 3.6.2.4 Respondents unwilling to pay

If a respondent indicated that they were unwilling to pay, for any question, they were asked to say why by way of a drop down response screen. The basic options presented were similar to those offered in the chained questionnaire, which included:

It wouldn't be too bad/I could live with it.

I would get better anyway, so it is not worth paying for the treatment.

I do value the treatment, but I cannot afford to pay anything for it.

I do value the treatment, but do not want to pay because the government should provide health care.

The risk is low, I would take the chance13.

Other (please specify below). For some questions in the direct alternative options were presented, some of which were taken from comments made during the pilot exercise. By way of example, those presented for Question G, which related to a 10 point drop in health for 10 years, later in life, are shown below:

It wouldn't be too bad/I could live with it.

13

This option was only included for risk questions.

63

I would get better anyway, so it is not worth paying for the treatment.

I may not live until that age, so it is not worth paying for treatment now.

I may be in poor health at that age, so it is not worth paying for treatment.

I do value the treatment, but I cannot afford to pay anything for it.

I do value the treatment, but do not want to pay because the government should provide health care.

Other (please specify). The response, “I do value the treatment, but do not want to pay because the government should provide health care” was included for all questions. As with the chained questionnaire, this was the main filter to identify „protestors‟. The questions themselves are to a certain extent hypothetical and „protestors‟ are seen as those respondents who are unwilling to „play the game‟. The treatment of „protestors‟ in the data analysis is mentioned in section 3.7.1 below. Further analysis of the responses for those electing not to pay; the reasons respondents gave for their response to Question L; and the responses to questions comparing equivalent health gains now and in the future are not presented in this report but will be the subject of future publications. 3.6.2.5 The card sort procedure Respondents agreeing to pay for a health gain were asked to undertake a 'card sort' exercise. This was similar to the one used in the chained survey but with some subtle differences. For the direct survey the potential variation in the size of the health gains was not as large as in the chained survey14. Hence it was decided to use a single set of 19 cards for all questions. The amounts offered to respondents varied from £10 to £300,000. Again, the card sort values were converted from UK currency to the prevailing local currency using Purchasing Power Parity conversion rates for 2008. The resulting values were rounded to two significant figures. Further discretionary rounding was undertaken for values close to round figures. Respondents were asked to sort the 19 cards into three trays according to whether they would pay that amount, would not pay that amount, or were unsure. A summary was then provided stating the maximum value the respondent would pay and the minimum value they would not pay. The respondent was then requested to state the maximum amount that they would be prepared to pay. Values outside the range bracketed by the minimum and maximum values in the summary returned an error comment requesting that the respondent either re-entered a value or re-arranged the cards.

14

The questions offering a year in the Green or Yellow health state in the chained survey represented health gains from 0 to 1 QALY depending on the respondent's valuation of that health state.

64

3.7 Analytical strategy

3.7.1 The „preliminary case‟ Given the propensities for respondents to protest to WTP-type questions, to provide responses which in the strictest economic terms might be defined as inconsistent (see below) and for implausible values to be stated, we defined a preliminary case for valuing a QALY. For all versions of both chained and direct questionnaires, the preliminary case comprised responses which:

were complete (in the sense that respondents managed to complete the whole questionnaire and did not abandon it part way through);

exclude protestors (who we defined as those respondents who stated that they were not willing to pay and gave „government should pay‟ exclusively as the reason why in the subsequent question);

are untrimmed (thus initially leaving in some respondents who gave implausibly high WTP values); and

include inconsistent responses. As outlined in describing chained and direct approaches above, respondents who were unwilling to pay anything at all for the treatment were asked to give their reasoning and one response category was; „I do value the treatment, but do not want to pay because the government should provide health care‟. Whilst this is clearly a subjective decision, we elected here to use the „government should provide‟ statement to distinguish between „true‟ and „protest‟ zeros. As we considered that a „protest‟ zero was not a true zero WTP response, we excluded „protestors‟ from the analysis. Respondents could, however, tick multiple statements. This raised the issue of whether a respondent who ticks „government should provide‟ and at least one other „legitimate‟ reason ought to be classified as a protestor or not - again this is subjective. We elected to classify a „protestor‟ in our „preliminary case‟ for analyses as a respondent who ticked „government should pay‟ only and to explore alternative assumptions in our sensitivity analyses. Specifically for the chained approach, a further exclusion is of non-traders when calculations of the value of a QALY are made at the level of the individual before calculating a mean (the „means of ratios‟ approach). As detailed in section 3.5 above, respondents who accepted no risk of death in the SG or traded off no time at all in the TTO were classified as „non traders‟. According to the principles underlying the SG and TTO approaches, respondents are then considered to have attached a utility value of „1‟ to that health state. Whilst we could have elected not to ask such respondents the subsequent WTP questions (as they had indicated that there was no utility loss associated with that state), we decided to elicit WTP values from all respondents including „non traders‟. The problem this poses is that, if „non-traders‟ go on to pay a positive amount to avoid that health state, they effectively have an infinite WTP per QALY as they are paying to avoid something that involves zero utility loss.

65

This is obviously problematic for the analyses, so the aggregate results presented below exclude „non-traders‟ in the utility assessment component. The main justification for the preliminary case is to take all responses at face value and also because not all respondents faced consistency tests. In the latter case dropping respondents who failed a test that only they happened to face may in itself appear inconsistent. 3.7.2 Sensitivity analyses Obviously, some aspects of the above preliminary case are controversial, although there are no „right‟ answers as to what to do about inclusions and exclusions. With respect to issues of completeness, no additional calculations were made to test the sensitivity of results to different rules. However, for other aspects, it was possible to do this. Most of these sensitivity tests are not reported here, but will form the subject matter of subsequent academic papers. To illustrate the impact of such analyses, the results of one sensitivity test is reported here. It will be seen that some respondents gave implausibly high WTP values. To assess the impact of this, and to minimise exclusions, the top one per cent of WTP values in each country was excluded. Other main sensitivity tests to be reported in the future are twofold:

The above filtering of zero responses represents the standard way of dealing with such values. They tend to be labelled as true zeros. However, it might be the case that all zeros are protests but that not all respondents admit this. Likewise it could be that all are in fact true, or at least should be interpreted as such in that people who say „government should pay‟ simply do not wish to contribute. Thus, in addition to excluding protestors in the way defined above, two further sensitivity tests were carried out, one which includes all zeros and one which excludes all zeros.

For non-traders in the chained approach, it is not possible to conduct a sensitivity test - as their inclusion renders WTP-for-a-QALY estimates infinite, and, therefore, implausible when ratios are calculated at the individual level. However, an alternative is to leave in such non-traders but take the mean of the health state utility assessments and the mean of the WTP value for the given scenario and then compute a ratio of these values (a „ratio of means‟ method of aggregation).

3.7.3 Consistency tests The tendency within contingent valuation studies is to err on the side of „generosity‟ to respondents and, thus, take their responses at face value. This was the approach with the preliminary case. However, unlike many other studies, some unambiguously better goods were offered to respondents in some scenarios when compared with others. Here, we merely point out those instances, but, again, they will be reported in future academic papers. In the chained approach, it is clear that avoiding the green health state (EQ-5D 22222) is unambiguously better than avoiding the yellow state (EQ-5D 21121).

66

Therefore, in the chained approach, we can check first of all whether, on average, the health state utility scores reflect this, expecting that, on average the green state is valued lower than the yellow because it is more serious. We can then examine these same data at the level of the individual, identifying the numbers of respondents who we define as strictly inconsistent, having given a higher health state utility score for the green state when compared with yellow. Then, there are various checks for consistency with respect to the WTP stage of the chained approach which will be undertaken. These are as follows:

the WTP to avoid the certainty of the green state for one year should be greater than WTP to avoid the certainty of the yellow state for one year;

in cases where respondents are valuing both a 0.05 and 0.1 QALY gain, WTP for the former would be expected to be less than for the latter;

in cases where respondents are faced with a health state for certain for one year, as well as some risk of that same health state, they should attach a higher value to the former than the latter.

In the direct approach only three such tests were conducted, comparing scenarios N (a 25 point loss for four years with a five per cent risk of occurrence) and M (a 25 point loss for four years with a 10 per cent risk of occurrence) in Version 2; E (a 10 point loss for one year for certain) and D (a 25 point loss for one year for certain) in Version 3 and comparing A (a 25 point loss for four years for certain with payment required now) and P (a 25 point loss for four years for certain with payment each year over four years) in Version 1. WTP for M should be greater than for N, WTP for D greater than E and WTP for A should be greater than for P. 3.8 Statistical and regression analyses

3.8.1 Chained approach As outlined earlier in this chapter, the basic principle underlying the chained approach was to estimate WTP per QALY by breaking the exercise down into two distinct components; an utility assessment component and a WTP component. So, if we know that a respondent is willing to pay £1000 to avoid one year with certainty in a health state with utility value of 0.95, we can estimate their WTP per QALY to be £20,000 per QALY gained (again assuming linearity). Whilst at first glance it may appear that we need simply multiply the respondent‟s WTP responses to the 0.05 and 0.10 QALY questions by 20 and 10 respectively in order to estimate their WTP per QALY, the analysis is somewhat more complicated than that for the reason set out below. Whilst we set out to keep QALY gains constant across respondents (the exceptions to this were the „one year with certainty‟ questions), it was not always possible to keep the QALY loss constant across respondents due to „capping‟ when respondents gave very high utility values for the health state. For example, for a respondent with a utility value of the yellow health state of 0.999, there is no risk of that health state (not greater than 100%) for one year they could be faced with that would amount to 0.05 (or 0.10) QALYs. Rather, the risk would be capped at 100% with an expected QALY gain of 0.001. Similarly, in a time variant WTP question, a respondent with a

67

utility value of 0.999 would need to avoid spending 100 years in that state with certainty in order to gain 0.10 QALYs, which is clearly implausible. Risk variant WTP questions were then capped at 100% and time variant questions at life expectancy and hence, it was not always possible to present all respondents with the same QALY gain. The implication of this is that the WTP responses to 0.05 and 0.10 QALY gain questions cannot simply be multiplied up by 20 and 10 respectively. And, of course, the QALY gains varied in the „one year with certainty‟ questions in any case. Thus, the QALY gain that respondents had actually valued had to be calculated in each case and this, combined with their WTP response, allowed an individual WTP per QALY to be estimated for each respondent. The QALY gain respondents had valued was calculated from their utility value of that health state and the risk and/or duration they had been presented with in the WTP component (which the programme always recorded). Their maximum willingness to pay was taken to be the highest amount they had placed in their „definitely would pay‟ pile or some higher amount they had written into the open text box (see 3.5.2.4 above). Means and medians of these individual WTP per QALY estimates are given in the results section below. Due to the two-stage nature of the chained procedure, the Heckman (1976) selection model is used to estimate the WTP per QALY for each of the 12 questions in the chained survey. The model consists of two parts a regression equation and a selection equation. The Heckman model assumes that there exists an underlying regression equation:

But the dependent variable is not always observed, instead the dependent variable for observation j is observed only if the selection equation is true:

Where

In this case the dependent variable is the log of WTP per QALY (retaining the zero

values by generating the log from ln(1+ raw WTP)) in PPP USD; in the regression equation is a vector of explanatory variables (including country dummies for the

summary models); the dependent variable in the selection equation is „gtrader‟ or „ytrader‟ a binary variable to denote trading; is the vector of explanatory variables

in the selection equation. The regression equation is estimated using OLS incorporating the mills ratio from the selection equation, which is estimated using probit. As indicated, this method was chosen because of the special characteristics of the survey questions. Respondents are taken through two steps; first, deciding whether they were willing to trade any risk of death in the SG question or any time in the TTO to avoid a health state; and, secondly, specifying the amount they are willing to pay to avoid some risk/duration of that health state. Tables 3.5 (pp83-85) and 3.6 (pp86-

68

88) each present the results for six questions, relating to the green and yellow health states respectively. We present summary regressions for all countries combined for all questions. As far as possible the explanatory variables have been kept the same for all Heckman estimates, the exceptions to this are due to specification error i.e. where

in which case standard OLS is more appropriate – see Heckman (1979). This applies to questions Gj, Yh and Yi in the summary estimates this is because there are no non-traders to select from. The income quintiles for the summary estimates are derived from global household income converted to PPP USD. Therefore any interpretation can be made independently from the country dummies. The missing income dummy has been included to avoid the loss of observations due to lack of income data; it represents the residual of respondents and is not part of the quintile categorisation. The base case for income is the top income quintile and as for all the base case categories, should be read as „zero‟. The age group 34-44 years is the base case for age groups; secondary education (pre degree) is the base case for education; and better than „green/yellow‟ the base case for current health state. The current health state variables are obtained from the EQ5D Tariff scoring system introduced by EuroQol Group (Dolan et al., 1996) derived in this survey from the five variables: EQ5D1 to EQ5D5. Gender is 1 for male; OECDCOEFFICIENT is a household composition composition variable based on the modified OECD coefficient (Hagenaars et al., 1994) for household composition (values above 10 were constrained to value of 10). The base case for the country dummies is the UK (Palestine has been excluded from these models). There are two intercepts to these estimates labelled „Constant 1‟ and „Constant 2‟. Constant 1 belongs with the selection step of the model.

3.8.2 Direct approach The analysis of the direct questionnaire was relatively straightforward in that each respondent was offered the same size health gain in each question. Where it was not credible to offer the respondent the health gain (for instance where the respondent's stated remaining life expectancy was less than the duration of the health gain) respondents were not offered the question. Responses to each question were aggregated by country and the mean and median calculated. Responses were also converted to international dollars at Purchasing Power Parity rates to allow aggregation across all 10 countries prior to calculating the mean and median. In the preliminary analysis all respondents opting not to pay for the health gain were assumed to have a zero valuation with one exception. Respondents selecting the option 'I do value the treatment, but do not want to pay because the government should provide health care' and only this option, where treated as protesters and a missing value was assumed for their response. Consequently the preliminary data included a number of zero valuations. It also included a few very high responses which would exert a strong influence on the mean. In order to reduce the sensitivity of the mean to very large values we elected to 'trim' the data by removing the highest 1% of responses. Trimming was undertaken by question and country – hence all responses to question A from a particular country

69

were ranked and the top 1% changed to missing values. Trimming was undertaken on each of the subsequent sensitivity analyses. The assumptions made in interpreting the responses of those electing not to pay for a health gain was tested in two sensitivity analyses. In the first analysis all respondents electing not to pay were assumed to value the health gain at zero. In the second analysis all respondents electing not to pay were assumed to be protestors and assigned missing values in the analysis. A final sensitivity analysis examined the effect of including only those respondents who reported higher willingness to pay for larger health gains. Three of the four versions contained pairs of questions in which one health gain was strictly larger than another. These were questions D and E in Version 3; questions M and N in the Version 2; and questions A and P in Version 1 (respondents were requested for a payment per year over four years for the heath gain in P hence we would expect a lower value than the one-off payment for the same health gain in A). The analysis examined the mean and median values for the subset of respondents from the trimmed dataset who reported a larger value for the strictly superior health gain. Respondents completing Version 4, where no test was available, were excluded. Basic regression analysis was undertaken to examine the impact of respondent characteristics on the mean WTP. We had collected a number of socio-demographic data on respondents including occupation and marital status. We restricted inclusion of socio-demographic variables in the regression models to those for which there were strong a priori reasons to include. The variables included were age; gender; household composition as captured by the OECD coefficient; household income, respondents‟ education level and the respondents‟ reported health. Age was entered using six age band categories. Three questions offered health gains at the end of respondents‟ lives and for these questions the number of years elapsing before the health gain commenced were` entered for each respondent. Income was entered using five quintiles plus a category for those electing not to declare their income. Some respondents elected not to declare their income but did select an income band. Annual incomes were imputed for these respondents based on mid band figures or median values from the reported income data for bands open ended at the top or bottom. Incomes for all respondents were converted to international dollars at purchasing power parity rates and the entire 10 country sample was split into quintiles. Respondents were assigned to three education levels: secondary, middle and tertiary according to whether their education level indicated that they had completed a university degree (tertiary) or some non-compulsory education (middle). The regression model chosen had to be flexible enough to deal with the large number of zero valuations. Options included a negative binomial regression, a tobit regression or a two part model in which the first part predicted the likelihood of observing a non zero WTP response. The tobit regression treats zero responses as censored negative values and has been frequently used in the past to model WTP data. The tobit was chosen for it‟s simplicity and common application. Prior to modelling the data was „trimmed‟ – responses for each question were ranked by country, and the top 1% of responses in each country were changed to missing.

70

Tobit regression was undertaken on the entire 10 country sample for each question separately. A dummy variable was entered for each country allowing the mean WTP for each question to vary by country. This was effectively a fixed effects model. The regressions quantified the impact of age, gender, education level, household size and income on WTP for health gains. 3.9 Results: Value of a QALY estimates

3.9.1 Chained approach The number of respondents agreeing to pay for the health gain offered varied from 67% to 79% across the 12 questions. In general fewer respondents were prepared to pay to avoid a period in the yellow health compared to the green health state but the lowest percentage agreeing to pay was observed for question Gf where respondents were offered a small risk of 10 years in the green health state. A consistently small fraction of those who elected not to pay indicated that the government should pay for health care as their only reason and were treated as protestors in the preliminary case. From the chained approach, WTP-based values of a QALY, using the preliminary case (i.e. complete and protestors, untrimmed and including inconsistent responses), are displayed, by country and for all nine countries collectively, in Table 3.7 (pp89-90). It can be seen that for some questions and in some countries WTP per QALY estimates are very high; as high as US $4m for certain WTP questions relating to the green health state when aggregated across countries with one estimate of US $20m in the Dutch sample. Such high values occur less frequently in those WTP questions relating to the yellow health state. Table 3.8 (pp91-92) shows these same results, but with WTP trimmed. Now, it can be seen that the results are more „well-behaved‟, the highest aggregated value for WTP questions relating to the green and yellow health states being US $34,000 and US $77,000 respectively. Looking at the results in more detail, we turn first to the risk variant questions. Recall that in the risk variant questions Gj, Gd, Yh and Yi respondents were asked to think about the value they would put on avoiding some chance of suffering the target health state for one year. The „chances‟ were set according to a) the respondents‟ utility value derived in the utility assessment component and b) whether the QALY gain was „set‟ at 0.05 or 0.10 QALY. Questions Gj and Gd relate to the green health state and 0.05 and 0.10 QALY gains respectively whilst Yh and Yi relate to the yellow health state and 0.05 and 0.10 QALY gains respectively. In this way, a comparison of responses to a) Gj and Gd and b) Yh and Yi, offered straightforward sensitivity to scope tests. Had respondents been, on average, willing to pay twice as much for the QALY gain that was twice as big, then the estimated WTP per QALY would be roughly equivalent. If, on the other hand, respondents were, on average, willing to pay the same for the 0.05 and 0.10 QALY gains, the estimated WTP per QALY derived via the 0.05 QALY gain questions (Gj and Yh) would be roughly twice that estimated via their 0.10 QALY counterparts (Gd and Yi). The last column of Table 3.8 shows that the ratio of mean (medians) for WTP per QALY derived via Gj and Gd is 1.14:1 (1.10:1), indicating that, whilst the general direction of the aggregate results indicates insensitivity to scope, some degree of sensitivity exists (as neither ratio is close to 2). The corresponding ratio of means (medians) of WTP per QALY

71

derived via Yh and Yi at 1.38:1 (1.59:1) is, however, closer to two, indicating there is less sensitivity to the magnitude of the QALY gain in those WTP questions involving the yellow health state. Recall that the other main set of WTP questions were time variant WTP questions in which the respondent was asked to pay to avoid some duration of the certainty of a health state. In this case durations were set according to a) the respondents‟ utility value derived in the utility assessment component and b) whether the QALY gain was „set‟ at 0.05 or 0.10 QALY. Questions Gm and Gn relate to the green health state and 0.05 and 0.10 QALY gains whilst Yq and Yr relate to the yellow health state and 0.05 and 0.10 QALY gains respectively. Again, had respondents been, on average, willing to pay the same for the 0.05 and 0.10 QALY gains, the estimated WTP per QALY derived via the 0,05 QALY gain questions (Gm and Yq) would be roughly twice that estimated via their 0.10 QALY counterparts (Gn and Yr). The last column of Table 3.8 shows that the ratio of mean (medians) for WTP per QALY derived via Gm and Gn is 1.50:1 (1.27:1). The ratio of means (medians) of WTP per QALY derived via Yq and Yr is 1.62 (1.25). So again, some degree of sensitivity has been shown in the aggregate results. Focusing on means, for both health states the ratios are closer to 2 than in their risk variant counterparts, indicating that there was less sensitivity to the magnitude of the QALY gain when duration was varied than when risk was varied. Perhaps the most striking manifestation of insensitivity to scope, however, can be seen by comparing the results of Ga and Yb i.e. the one year with certainty questions with those above. It was highlighted previously that the QALY gains involved in avoiding the green health state in particular for one year were potentially much larger than those involved in the 0.05 and 0.10 QALY questions. Insensitivity to scope would then predict that the estimated WTP per QALY derived via Ga and Yb would be considerably lower than in those questions described above. The last column of Table 3.8 shows that is indeed the case. The overall median WTP per QALY estimate of $3019 derived via Ga is considerably lower than those derived via Gj, Gd, Gm and Gn of $8211, $7473, $6382 and $5015 respectively. Likewise, the overall median WTP per QALY estimate of $2959 derived via Yb is considerably lower than those derived via Yh, Yi, Yq and Yr of $7295, $4584, $4687 and $3723 respectively. Hence, whilst the comparisons of 0.05 and 0.10 QALY questions seemed to show at least a degree of sensitivity to the magnitude of the QALY gains, the results of the one year with certainty questions show that asking about QALY gains that are potentially much larger results in estimates of WTP per QALY that are considerably lower. This raises the possibility that respondents may have been budget-constrained when answering the one year with certainty questions. As outlined in section 3.5.2.2. above, a comparison of responses to a) Gj and Gm, b) Gd and Gn, c) Yh and Yq, and d) Yi and Yr, offer a test of sensitivity of WTP responses to the „framing‟ of the WTP question i.e. whether risk variant or time variant questions were asked. If setting the QALY gains using either risk or time did not affect responses, we would expect the responses to risk variant and time variant to be roughly equivalent and, hence, estimates of WTP per QALY to be roughly equivalent. The last column of Table 3.8 shows that the median WTP per QALY estimated via the risk variant questions Gj, Gd, Yh and Yi of $8211, $7473, $7295, and $4584 respectively are in each case higher than their time variant counterparts

72

Gm, Gn, Yq and Yr of $6,382, $5015, $4687 and $3723 respectively. Thus, it would appear as if estimated WTP per QALY is not independent of the framing used in the WTP question in that, all else equal, using a risk variant format appears to yield higher estimates than a time variant format. Turning now to the remaining WTP questions, recall that question Gf related to the green health state and a 0.10 QALY gain, but used a 10 year duration, rather than the one year duration used in Gd and adjusted the risk accordingly (i.e. essentially divided the risk presented in question Gd by 10 whilst multiplying the duration by 10). A comparison of WTP per QALY estimates derived via Gf and Gd show the means to be somewhat higher in the former, but medians are lower. Finally, question Ye was similar to Yb in that it involved paying to avoid the yellow

health state for one year with certainty. The difference here was that, rather than ask for a „one off‟ payment as in the remainder of the WTP questions, this question asked the respondent to state their WTP each year for a period of four years. All other things equal, we would expect the total WTP over four years to be roughly equivalent to the one off payment and, hence, estimates of WTP per QALY to be roughly equivalent. A comparison of WTP per QALY estimated via Ye and Yb, however, shows the former to be considerably greater than the latter with means (medians) of $77,323 ($7807) compared with $28,878 ($2959). This finding could either be interpreted as a lack of sensitivity to the payment period or as evidence that budget constraints were indeed biting in the „one off‟ payment scenario which were slackened by the ability to pay over four years. In summary, leaving aside the one year with certainty questions (where the QALY gains arguably went against the general principle of what we set out to do here - namely keep the magnitude of the QALY gains small) and returning to Table 3.8 the overall range of mean WTP per QALY is $18,247 (via Yr) to $77,323 (via Ye). The overall range of median WTP per QALY is $3723 (via Yr) to $8211 (via Gj). Many readers will consider the median WTP per QALY estimates to be conservative and they are certainly lower than commonly used „thresholds‟. For example, in the UK, NICE use a threshold WTP per QALY of around £30,000 per QALY gained-a threshold considerably higher than the median values derived here.

3.9.2 Direct approach The number of respondents agreeing to pay for the health gain offered varied from 53% to 79% across the 13 questions. Lower proportions were observed for the questions offering health gains at the end of life, with the lowest proportion for question I. Respondents electing not to pay and indicating only that they thought the government should pay constituted ca. 6-9% of respondents across the 13 questions. The remaining respondents indicated at least one reason that would suggest they valued the health gain at zero. From the direct approach, WTP-based values of a QALY, again using the preliminary case (i.e. complete and protestors, untrimmed and including inconsistent responses), are displayed, by country and for all 10 countries collectively, in Table 3.9 (p93). It can be seen that the questions offering smaller health gains generated larger mean and median WTP values for one QALY - mean WTP per QALY was higher for the

73

question offering a 0.25 QALY gain (Question D) than for any offering one QALY, and values were higher still for questions offering a gain of 0.1 QALY (Questions E, M and O). The risk-based question offering an expected gain of 0.05 QALYs (Question N) generated the highest mean values per QALY. The difference was less marked in the medians but questions offering a gain of one QALY all generated lower median values per QALY than any question offering a smaller fraction of a QALY. Respondents valued the „10 point 10 year‟ health gains (Questions F and G) similarly to the „25 point four year‟ health gain occurring now (Question A). The „25 point four year‟ health gain at the end of life (Question B) generated an unusually large mean WTP. This was the result of one Spanish respondent who indicated a maximum WTP of 80,000,000 euros. Extensions of life were valued more highly than improvements in quality of life. Mean responses for increases in quality of life appeared to be insensitive to the timing of the gain – mean WTP for Question G was similar to that for Question F. This was not the case for median values. Across countries median values for increases in quality of life now were roughly double the medians for improvements in quality of life at the end of life. Both means and medians were sensitive to timing for increases in longevity. Two questions offered scenarios in which respondents gained an increase in longevity in the near future – the coma and the terminal illness questions (Questions J and L respectively). Both generated higher mean and median WTP values than the increase of one QALY at the end of life. It is possible that respondents gave lower values per QALY when offered larger health gains due to a budget constraint. We explored this possibility by offering a Question P in which the health gain from Question A was duplicated but the respondent was asked to pay in four yearly instalments. Not surprisingly Question P generated larger mean and median values per QALY than Question A. However, it was possible that some respondents misread this question and believed that they were being asked to pay in one lump sum. The impact of this would be to raise the mean and median WTP per QALY for Question P. It was evident that the means were highly influenced by a few very high responses and this may have been responsible for the insensitivity of the means to the timing or scale of the health gain. Scrutiny of the median responses shows more sensitivity to the size of the health gain and to the timing of the health gain. Median responses for the health gains at the end of life were all noticeably smaller than for the corresponding health gain now. Lower medians for health gains at the end of life would be generated by lower positive responses; they would also arise if a larger number of people elected not to pay for the health gain and were ascribed a value of zero. Table 3.10 (p94), shows these same results, but with WTP trimmed. The impact of trimming the top 1% of the data is to reduce the overall means by around one third to one half. In general the impact of trimming is greater for questions offering smaller health gains. The impact on Question B is much larger due to the removal of a single extremely large value. The mean for Question B is now in line with the means for the other questions offering improvements in quality of life, with lower means for improvements later in life. Again, means for extension of life equivalent to one QALY are higher than means for an improvement in quality of life equivalent to one QALY.

74

The highest trimmed mean occurs for the smallest health gain, with trimming reducing the overall mean across all 10 countries from $176,000 to $82,000. The impact of trimming on median values is minimal. In summary, the raw mean values are heavily influenced by a few respondents giving very high values. Trimming generally has a larger impact on the questions offering smaller health gains, perhaps because erroneous responses are scaled up to generate values per QALY for these questions. After trimming, means for the questions offering one QALY range from $4854 to $20,719 across all 10 countries. Smaller health gains generated higher means per QALY ranging from $27,977 for question D offering a 25 point health gain for one year to $82,347 for question N offering a 5% chance of a health gain of 25 points for four years. The results support the hypotheses we laid out on page 32 of the report although for some the evidence is weak. We do observe larger mean responses for larger health gains but any sensitivity to scale is weak with many respondents failing to distinguish the scale of the health gain offered or expressing higher willingness to pay for smaller health gains (H1). Similarly a small fall in the trimmed mean is observed where the same health gain is offered later in life (H2). Whilst the medians show more sensitivity to the timing and size of the health gain this is due primarily from greater numbers of respondents electing not to pay for smaller health gains or gains at the end of life. Increases in longevity do appear to be valued greater than increases in quality of life (H3) with the increases in longevity offered now (either through avoiding a coma or delaying death) providing the highest means and medians. The chained approach has generated higher mean and median values than the direct approach as predicted (H4). However, where questions in the direct survey offered a smaller health gain similar in size to those in the chained (questions E, M, N and O) the mean and median responses were similar in magnitude. The question offering respondents a postponement from early death generated the highest mean and median values for questions offering a one QALY, again as expected (H5). Economic theory would predict that respondents would be prepared to sacrifice a significant proportion of their wealth to delay death, given that the alternative would leave few opportunities to spend the money. Nevertheless, the median values for this question were relatively modest, with the median amongst the 71% electing to pay being $5319.

3.10 Regression analysis

3.10.1 Chained approach In terms of selection estimates in each case the coefficient for the constant is positive, indicating that there is a tendency to „trade‟ in the SG or TTO in the absence of any of the other explanatory variables in the model. The sign associated with each coefficient belonging to the explanatory variables indicates the effect on the probability of trading. For example a positive coefficient for „Tertiary‟ education signifies that trading increases with education, a negative coefficient for „Hungary‟ suggests that the probability of trading is reduced if the respondent lives in Hungary. A typical respondent will occupy only a single category within each of the groups. For example looking at the summary results, for a male respondent answering

75

question Ga with the following characteristics - aged 25-34 years; income of US $24,366 - $33,701; from the Netherlands; having secondary education; full health; OECDCOEFFICIENT = 1; answering a standard gamble question - their probability of trading is given by:

If the same respondent lives in Hungary the probability of trading falls to:

Where is the cumulative distribution function (CDF). This probability estimate then

becomes an explanatory variable in the regression equation. From the second stage regression estimates, as we would expect, WTP rises with respondents‟ income with those respondents with missing income tending to be willing to pay the least. WTP tends to rise with age but there is a non-linear progression, the exception to this being question Ga. On the whole, all the coefficients for the country dummies for all questions have the same sign, indicating general consistency across questions; the exception being France where two questions have a negative sign (although these are not however statistically significant). For gender (males) the sign for the coefficient is negative in all but two cases, again these are not significant, indicating that males are willing to pay less than females. The coefficient for OECDCOEFFICIENT varies in both sign and magnitude but is rarely significant, in general WTP rises as household size falls. For the three education dummies there is strong evidence of increasing WTP with education. The same is true for the three health dummies, increasing WTP with (good) health. Lastly the magnitude of the coefficient for constant 2 indicates that the intercept for WTP lies around US $500. The model specification results appear on the third sheet of each Table 3.5 (p85) and 3.6 (p88): N represents the number of observations used in the model with the censored observations representing respondents who answered the question but did not trade The parameter estimate chi2(df) is the Wald test. This tests the hypothesis that all the coefficients in the model except the constant are zero. This is rejected in every case. The likelihood-ratio test is given by chi2_c, tests the joint hypothesis that the Heckman selection model is computationally better than the alternative i.e. independent probit and separate OLS, thereby justifying the use of the Heckman selection model. Whilst some of these are not significant, others are strongly significant indicating that the choice of explanatory variables should be more carefully chosen, non-selection by model of the explanatory variables accounts for this. The log likelihood is the goodness of fit. Due to space constraints the correlation matrices have not been presented.

3.10.2 Direct approach The results of the regression analysis for each question using a Tobit model are presented in Table 3.11 (p95). For each question the base case was a Swedish female aged 36-45 with a household income in the middle quintile of the distribution across all 10 countries and with a middle education level. The variable 'years prior'

76

captures the number of years before the health gain commenced for the three questions offering health gains towards the end of life. Respondent's health is entered on a 0 to 100 scale as the health (VAS) variable. A similar pattern emerges across the 13 questions in terms of the impact of respondent characteristics on their willingness to pay. We see the expected gradient in education across all thirteen questions with respondents with a tertiary education giving higher values after adjusting for income. Similarly respondents whose household income places them in a higher quintile generally give higher values although this trend was weak when comparing the lowest and second lowest quintiles. There is a noticeable jump in the coefficient for respondents in the highest quintile for increases in longevity now (the coma question J and the terminal illness question L), which may be indicative of the impact of lowered budget constraints in this group. A strong gradient across age is evident with older respondents prepared to pay considerably more for health gains. However, this trend is reversed for questions J and L. Older respondents give lower means than younger respondents to avoid a coma or postpone a terminal illness but they give a higher mean to increase their longevity at the end of life. The latter may reflect the fact that gains at the end of life are experienced sooner for older respondents. For the three questions offering health gains at the end of life there was no strong relationship between the time elapsing before the start of the health gains and the willingness to pay. This variable may have been strongly correlated with age. Overall it appears that older respondents reverse the overall trend and place higher values on gains in quality of life than gains in longevity. In general men pay more than women but this trend is reversed for questions J and L which may reflect considerations of family and carer commitments on the part of female respondents when valuing gains in longevity now. In general household size produced the expected trend with respondents from larger households giving lower means. Respondents rating their own health higher also gave higher values for health gains, and this effect was particularly noticeable for questions offering increases in longevity now. This is surprising given that respondents reporting a higher health were offered a longevity gain of smaller duration to keep the overall gain constant at one QALY. There were three exceptions to the relationship between reported health and willingness to pay for health gains and these occurred for the questions offering certain health gain of 10 points in size. This gain may have appeared relatively smaller to respondents with high health scores. There were noticeable differences in mean WTP values across the 10 countries. After conversion to international dollars at purchasing power parity rates and adjusting for household income, respondents in the UK, France and the Netherlands consistently gave lower mean WTP values. The Danes generally gave the highest values closely followed by the Spanish.

3.11 Discussion and conclusions The work presented here has made significant advances on earlier studies which have used survey methods to estimate WTP-based values of QALY. In particular the methodological advances on the chained approach have allowed us to present results on what might be argued to be the more-theoretically-relevant form of means

77

and medians of individuals‟ ratios. The means presented will be seen by policy makers as in the vicinity of thresholds currently in use. Medians would be perceived as low. However, given that the original conception of the chained approach involved a willingness-to-accept compensation stage to be combined with WTP, the mean of which would then be combined with a health state utility assessment, it could be argued that an extra 30/50% could be added to the chained values. The direct approach results in lower values of a QALY, but provides interesting insights into values attached to different QALY types. In particular, it raises questions as to whether a premium should be attached to QALYs gained in terminal illness relative to other health gain scenarios.

78

Figure 3.1: EuroVaQ: conceptual overview for monetary valuation of a QALY

79

Figure 3.13a: Design for chained approach to monetary valuation of a QALY

QALY value WTP question Green health state

EQ5D 22222

Yellow health state

EQ5D 21121

0.05 100% time Gm Yq

0.05 1 year risk Gj Yh

0.10 100% time Gn Yr

0.10 1 year risk Gd Yi

0.10 10 year risk Gf

Varies One year with certainty Ga Yb

Varies One year with certainty –four annual payments Ye

RESPONDENT ALLOCATED TO 1 OF 8 VERSIONS – Figure 3.13b and 3.13c below

80

Figure 3.13b Design for chained approach to monetary valuation of a QALY RESPONDENT RANDOMLY ALLOCATED TO START WITH EITHER GREEN OR YELLOW STATE

GREEN HEALTH STATE – EQ5D 22222 YELLOW HEALTH STATE - EQ5D 21121

STANDARD

GAMBLE

ALWAYS FIRST

THEN 2 OR 3 QUESTIONS IN RANDOM ORDER

ALWAYS FIRST

THEN 2 OR 3 QUESTIONS IN RANDOM ORDER

Version 1

SG 22222 WTP RISK VARIANT SG 21121 ONE YEAR

CERTAINTY WTP RISK VARIANT

Standard

Gamble

exercise

1 year risk

QALY=0.05

1 year risk

QALY=0.10

Standard

Gamble

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.05

1 year risk

QALY=0.10

Gj Gd Yb Yh Yi

Version 2

SG 22222 ONE YEAR

CERTAINTY WTP RISK VARIANT SG 21121

ONE YEAR

CERTAINTY WTP RISK VARIANT

Standard

Gamble

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.10

10 year ⅟10th risk

QALY=0.10

Standard

Gamble

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.10

Ga Gd Gf Yb Yi

Version 3

SG 22222 ONE YEAR

CERTAINTY WTP TIME VARIANT SG 21121 WTP TIME VARIANT

Standard

Gamble

exercise

1 year 100%

QALY varies

100% time

QALY=0.05

100% time

QALY=0.10

Standard

Gamble

exercise

100% time

QALY=0.05

100% time

QALY=0.10

Ga Gm Gn Yq Yr

ALWAYS FIRST

2 QUESTIONS RANDOM ORDER

ALWAYS LAST

Version 4

SG 22222 WTP TIME VARIANT SG 21121 WTP TIME VARIANT ONE YEAR

CERTAINTY

Standard

Gamble

exercise

100% time

QALY=0.05

100% time

QALY=0.10

Standard

Gamble

exercise

100% time

QALY=0.05

100% time

QALY=0.10

1 year 100%

4 yearly pays

QALY varies

Gm Gn Yq Yr Ye

81

Figure 3.13c: Design for chained approach to monetary valuation of a QALY

RESPONDENT RANDOMLY ALLOCATED TO START WITH EITHER GREEN OR YELLOW STATE

GREEN HEALTH STATE - EQ5D 22222 YELLOW HEALTH STATE- EQ5D 21121

TIME TRADE

OFF

ALWAYS FIRST

2 OR 3 QUESTIONS RANDOM ORDER

ALWAYS FIRST

2 OR 3 QUESTIONS RANDOM ORDER

Version 5

TTO 22222 WTP RISK VARIANT TTO 21121 ONE YEAR

CERTAINTY WTP RISK VARIANT

Time

trade-off

exercise

1 year risk

QALY=0.05

1 year risk

QALY=0.10

Time

trade-off

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.05

1 year risk

QALY=0.10

Gj Gd Yb Yh Yi

Version 6

TTO 22222 ONE YEAR

CERTAINTY WTP RISK VARIANT TTO 21121

ONE YEAR

CERTAINTY WTP RISK VARIANT

Time

trade-off

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.10

10 year ⅟10 risk

QALY=0.10

Time

trade-off

exercise

1 year 100%

QALY varies

1 year risk

QALY=0.10

Ga Gd Gf Yb Yi

Version 7

TTO 22222 ONE YEAR

CERTAINTY WTP TIME VARIANT TTO 21121 WTP TIME VARIANT

Time

trade-off

exercise

1 year 100%

QALY varies

100% time

QALY=0.05

100% time

QALY=0.10

Time

trade-off

exercise

100% time

QALY=0.05

100% time

QALY=0.10

Ga Gm Gn Yq Yr

ALWAYS FIRST

2 QUESTIONS RANDOM ORDER

ALWAYS LAST

Version 8

TTO 22222 WTP TIME VARIANT TTO 21121 WTP TIME VARIANT ONE YEAR

CERTAINTY

Time

trade-off

exercise

100% time

QALY=0.05

100% time

QALY=0.10

Time

trade-off

exercise

100% time

QALY=0.05

100% time

QALY=0.10

1 year 100%

4 yearly pays

QALY varies

Gm Gn Yq Yr Ye

82

Figure 3.15: Design for direct approach to monetary valuation of a QALY

83

Table 3.5: Coefficients from the Heckman regression models for each question: Chained – green health state Explanatory Log Ga Per QALY Log Gj Per QALY Log Gd Per QALY Log Gm Per QALY Log Gn Per QALY Log Gf Per QALY

Variables Coef.(Se) Coef.(Se)¥ Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se)

Missing Income -0.8259***(0.1333) -1.4033***(0.2254) -1.0717***(0.1509) -1.0777***(0.1560) -1.1264***(0.1356) -1.2669***(0.2404)

<USD 17,063 Q1 -1.4828***(0.1451) -1.6030***(0.2329) -1.4620***(0.1578) -1.4386***(0.1632) -1.4889***(0.1425) -1.6042***(0.2590)

<USD 24,365 Q2 -1.0570***(0.1351) -1.1642***(0.2220) -0.9101***(0.1495) -1.0189***(0.1536) -1.0895***(0.1346) -1.1867***(0.2399)

<USD 33,701 Q3 -0.6933***(0.1287) -1.1510***(0.2148) -0.7409***(0.1436) -0.6494***(0.1478) -0.7173***(0.1293) -0.9682***(0.2294)

<USD 48,032 Q4 -0.4534***(0.1250) -0.5030*(0.2074) -0.3908**(0.1391) -0.4099**(0.1434) -0.3143*(0.1250) -0.6406**(0.2242)

>USD 48,032 (base case) <25 yrs -0.0364(0.1327) -0.4782*(0.2225) -0.5293***(0.1496) -0.3088*(0.1500) 0.1024(0.1295) -0.7422**(0.2432)

25-34 yrs -0.3079**(0.1149) -0.3727*(0.1893) -0.3500**(0.1272) -0.4082**(0.1291) -0.1269(0.1126) -0.5148*(0.2067)

35-44 yrs (base case) 45-54 yrs -0.1558(0.1139) 0.2455(0.1893) 0.0355(0.1272) 0.065(0.1311) 0.1327(0.1142) 0.2915(0.2044)

55-64 yrs -0.131(0.1219) 0.9376***(0.2051) 0.3924**(0.1366) 0.3005*(0.1379) 0.2755*(0.1208) 0.7317***(0.2179)

65 yrs + 0.2347(0.1323) 1.4151***(0.2195) 0.6743***(0.1469) 0.6289***(0.1523) 0.7161***(0.1329) 0.8481***(0.2370)

Netherlands -0.1325(0.1569) 0.2395(0.2562) 0.0281(0.1742) 0.1192(0.1766) 0.1688(0.1549) 0.2317(0.2858)

UK (base case) France -0.2414(0.1558) 0.3239(0.2497) 0.1893(0.1706) 0.4092*(0.1762) 0.0588(0.1544) 0.4727(0.2788)

Spain 0.9241***(0.1558) 1.4543***(0.2586) 1.1872***(0.1733) 1.5418***(0.1779) 1.2123***(0.1552) 1.5785***(0.2805)

Sweden 0.6354***(0.1543) 1.2678***(0.2572) 0.8860***(0.1727) 0.7844***(0.1769) 0.6106***(0.1547) 1.0128***(0.2778)

Norway 0.8641***(0.1658) 1.0106***(0.2820) 0.9254***(0.1891) 0.7845***(0.1935) 0.7447***(0.1688) 1.5105***(0.2990)

Denmark 0.5574***(0.1536) 1.2463***(0.2602) 0.8971***(0.1759) 0.5127**(0.1766) 0.6041***(0.1542) 1.0577***(0.2792)

Poland 0.5420**(0.1738) 2.1644***(0.2647) 1.5304***(0.1845) 1.3979***(0.1892) 1.2345***(0.1652) 1.4402***(0.3137)

Hungary 0.7260***(0.1749) 1.9861***(0.2673) 1.2261***(0.1853) 1.6718***(0.1926) 1.4216***(0.1686) 1.2242***(0.3104)

Gender -0.045(0.0742) -0.2441*(0.1224) -0.0665(0.0825) 0.0195(0.0849) -0.1084(0.0733) -0.3576**(0.1347)

OECDCOEFFICIENT -0.1056(0.0795) -0.0856(0.1297) -0.0687(0.0867) -0.1268(0.0898) -0.0759(0.0775) -0.0841(0.1417)

Middle -0.4873***(0.1162) -0.3983*(0.1901) -0.5016***(0.1279) -0.3543**(0.1305) -0.3909***(0.1148) -0.4896*(0.2067)

Secondary (base case) Tertiary 0.3923***(0.0816) 0.2565(0.1342) 0.3353***(0.0903) 0.2219*(0.0926) 0.3676***(0.0808) 0.2191(0.1468)

Full health 0.0753(0.0773) 0.0752(0.1264) 0.0133(0.0855) 0.1431(0.0882) 0.1188(0.0765) 0.0528(0.1384)

Better than Green (base case) Worse than Green -0.4978***(0.1446) -0.7829***(0.2377) -0.3454*(0.1580) -0.6613***(0.1638) -0.7186***(0.1431) -0.0516(0.2513)

Constant 2 8.2054***(0.2341) 7.5742***(0.3793) 8.0917***(0.2568) 7.5620***(0.2678) 7.5908***(0.2320) 7.1893***(0.4291)

84

Table 3.5 continued Explanatory Log Ga Per QALY Log Gj Per QALY Log Gd Per QALY Log Gm Per QALY Log Gn Per QALY Log Gf Per QALY

Variables Coef.(Se) Coef.(Se)¥ Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se)

Green Trader Missing Income -0.0117(0.0649) _ 0.1191(0.0792) -0.3054***(0.0753) -0.3611***(0.0950) -0.3730***(0.1132)

<USD 17,063 Q1 0.1133(0.0739) _ 0.1351*(0.0663) -0.1533(0.0846) -0.1323(0.1079) -0.2017(0.1274)

<USD 24,365 Q2 0.0969(0.0704) _ 0.1365*(0.0676) -0.0111(0.0816) -0.0187(0.1021) -0.1042(0.1197)

<USD 33,701 Q3 0.0319(0.0663) _ 0.1098(0.0844) -0.0207(0.0783) -0.0654(0.0973) -0.1029(0.1167)

<USD 48,032 Q4 0.0436(0.0658) _ 0.0631(0.0590) 0.0838(0.0778) 0.1047(0.0985) -0.0118(0.1174)

>USD 48,032 (base case) <25 yrs 0.2546***(0.0721) _ 0.2709***(0.0704) 0.3378***(0.0855) 0.4130***(0.1077) 0.3856**(0.1250)

25-34 yrs 0.2543***(0.0594) _ 0.1938***(0.0557) 0.1416*(0.0676) 0.1966*(0.0839) 0.2784**(0.1008)

35-44 yrs (base case) 45-54 yrs 0.0798(0.0544) _ 0.022(0.0658) -0.0646(0.0645) -0.0075(0.0805) 0.0932(0.0932)

55-64 yrs 0.0792(0.0596) _ -0.0495(0.0686) -0.0272(0.0690) -0.0107(0.0849) 0.0445(0.0992)

65 yrs + 0.0429(0.0646) _ -0.0702(0.0607) 0.0472(0.0789) 0.1514(0.1012) 0.105(0.1109)

Netherlands 0.1099(0.0767) _ 0.1744*(0.0786) 0.0979(0.0919) 0.199(0.1188) -0.0064(0.1386)

UK (base case) France 0.0217(0.0750) _ 0.1992(0.1561) -0.0925(0.0883) 0.047(0.1153) -0.0963(0.1348)

Spain 0.0382(0.0743) _ 0.0253(0.1182) -0.1492(0.0864) -0.1273(0.1091) -0.0619(0.1351)

Sweden 0.071(0.0773) _ 0.0112(0.1165) 0.0982(0.0937) 0.0566(0.1145) -0.0899(0.1343)

Norway 0.0993(0.0828) _ -0.0818(0.1278) -0.0969(0.0971) -0.1685(0.1201) -0.0375(0.1482)

Denmark 0.1124(0.0782) _ -0.1121(0.0675) 0.1364(0.0942) 0.0147(0.1143) 0.1462(0.1445)

Poland 0.1422(0.0875) _ -0.1462*(0.0708) 0.1006(0.1006) 0.0268(0.1256) -0.1475(0.1501)

Hungary -0.0108(0.0840) _ -0.0699(0.1005) 0.11(0.0998) 0.1165(0.1244) -0.0376(0.1485)

Gender -0.0452(0.0376) _ 0.0163(0.0330) -0.1711***(0.0433) -0.1130*(0.0544) -0.1850**(0.0637)

OECDCOEFFICIENT 0.0665(0.0399) _ -0.0987***(0.0287) 0.0011(0.0474) 0.023(0.0598) 0.1174(0.0712)

Middle 0.0224(0.0550) _ 0.1094(0.0611) -0.0362(0.0645) 0.0029(0.0808) -0.0622(0.0943)

Secondary (base case) Tertiary -0.0148(0.0420) _ -0.0603(0.0355) 0.1007*(0.0488) 0.2380***(0.0619) 0.083(0.0722)

Full health -0.0366(0.0392) _ -0.0002(0.0375) -0.1318**(0.0461) -0.0216(0.0579) -0.0441(0.0669)

Better than Green (base case) Worse than Green 0.0979(0.0716) _ 0.0974(0.0585) -0.0787(0.0832) -0.0923(0.1015) -0.0352(0.1175)

Std gamble question 0.2025***(0.0353) _ 0.0198(0.0565) 0.4095***(0.0444) 5.9085(3190.79) 0.5582***(0.0663)

Constant 1 1.0294***(0.1179) _ 1.4394***(0.1031) 1.5500***(0.1400) 1.3570***(0.1728) 1.2665***(0.2069)

85

Table 3.5 continued Log Ga Per QALY Log Gj Per QALY Log Gd Per QALY Log Gm Per QALY Log Gn Per QALY Log Gf Per QALY

Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se)

N 9051 4392 8931 9718 9674 4426

N_censored 526 _ 285 543 379 266

chi2(df) 457.3275(24) (24) 437.0561(24) 373.1442(24) 482.9439(24) 221.32(24)

Log likelihood -23472.1481 -12220.557 -24111.4226 -27550.5026 -25994.5514 -12784.44

chi2_c 523.2404*** _ 767.8279*** 0.044 2.0574 0.3336

athrho -1.9153***(0.0645) _ -3.4580**(1.0921) 0.023(0.1002) 0.0976(0.0613) 0.0924(0.1288)

lnsigma 1.2279***(0.0081) _ 1.3256***(0.0084) 1.3665***(0.0074) 1.2384***(0.0074) 1.4283***(0.0112)

rho -0.9575 _ -0.998 0.023 0.0973 0.0922

sigma 3.414 _ 3.7643 3.9215 3.4501 4.1717

lambda -3.269 _ -3.7568 0.0901 0.3357 0.3845

Heckman Selection Model: Regression dependent variable in log of PPP USD; selection dependent variable: Whether the respondent traded. Income quintiles global in PPP USD

*Denotes significance p-value < 0.05 ; ** p-value <0.01; *** p-value <0.001: at 5% level ¥ Denotes OLS regression

86

Table 3.6: Coefficients from the Heckman regression models for each question: Chained – yellow health state Explanatory Log Yb Per QALY Log Yh Per QALY Log Yi Per QALY Log Yq Per QALY Log Yr Per QALY Log Ye Per QALY

Variables Coef.(Se) Coef.(Se)¥ Coef.(Se)¥ Coef.(Se) Coef.(Se) Coef.(Se)

Missing Income -0.8508***(0.1307) -0.6974**(0.2554) -0.6743***(0.1723) -1.2157***(0.1575) -1.3677***(0.1387) -1.5555***(0.1934)

<USD 17,063 Q1 -1.4950***(0.1377) -1.1731***(0.2593) -0.9995***(0.1796) -1.8425***(0.1660) -1.8057***(0.1461) -1.7911***(0.2022)

<USD 24,365 Q2 -0.8415***(0.1307) -1.0582***(0.2481) -0.6229***(0.1716) -1.2466***(0.1558) -1.1713***(0.1379) -1.0381***(0.1931)

<USD 33,701 Q3 -0.6078***(0.1244) -0.4162(0.2413) -0.2219(0.1654) -0.8889***(0.1505) -0.7801***(0.1326) -0.5218**(0.1853)

<USD 48,032 Q4 -0.4210***(0.1210) -0.1127(0.2361) -0.0515(0.1623) -0.5533***(0.1455) -0.5006***(0.1284) -0.5449**(0.1800)

>USD 48,032 (base case) <25 yrs 0.1582(0.1302) -0.2112(0.2469) -0.4417**(0.1700) -0.2319(0.1517) 0.0503(0.1327) 0.0318(0.1881)

25-34 yrs -0.1177(0.1105) -0.1844(0.2126) -0.3402*(0.1464) -0.2098(0.1310) -0.1815(0.1154) -0.2344(0.1604)

35-44 yrs (base case) 45-54 yrs 0.0113(0.1108) 0.5655**(0.2132) 0.2363(0.1471) 0.1349(0.1335) 0.1117(0.1180) 0.0592(0.1672)

55-64 yrs 0.2236(0.1192) 1.2271***(0.2293) 0.6607***(0.1574) 0.3569*(0.1407) 0.2368(0.1240) 0.015(0.1739)

65 yrs + 0.5170***(0.1281) 1.6541***(0.2440) 1.2714***(0.1703) 0.8966***(0.1541) 0.7782***(0.1357) 0.4099*(0.1930)

Netherlands 0.193(0.1530) 0.2235(0.2852) 0.3446(0.2006) 0.1383(0.1783) -0.1361(0.1581) 0.2363(0.2249)

UK (base case) France 0.1725(0.1511) 0.7391**(0.2807) 0.6058**(0.1974) 0.2543(0.1799) -0.1946(0.1588) 0.1622(0.2239)

Spain 1.1151***(0.1512) 1.5995***(0.2869) 1.4802***(0.1995) 1.5385***(0.1811) 1.1234***(0.1591) 1.5186***(0.2274)

Sweden 0.8906***(0.1512) 1.2311***(0.2878) 1.0053***(0.2006) 0.7737***(0.1789) 0.4507**(0.1580) 0.8050***(0.2241)

Norway 1.0409***(0.1634) 1.4873***(0.3222) 1.3167***(0.2229) 0.7546***(0.1961) 0.6047***(0.1728) 0.7916**(0.2476)

Denmark 0.8575***(0.1531) 1.4998***(0.2930) 1.1794***(0.2012) 0.6148***(0.1786) 0.4330**(0.1568) 0.9149***(0.2219)

Poland 1.0953***(0.1617) 2.6073***(0.2915) 1.9908***(0.2070) 1.7391***(0.1909) 1.1506***(0.1686) 1.2532***(0.2286)

Hungary 1.1847***(0.1615) 2.0255***(0.2939) 1.3980***(0.2125) 1.7009***(0.1942) 1.2013***(0.1717) 1.6457***(0.2340)

Gender -0.0225(0.0722) -0.3907**(0.1369) -0.2928**(0.0945) 0.0253(0.0856) -0.1909*(0.0753) -0.0861(0.1049)

OECDCOEFFICIENT -0.144(0.0761) -0.0136(0.1432) 0.0257(0.1007) -0.2615**(0.0883) -0.2659***(0.0790) -0.2307*(0.1065)

Middle -0.4338***(0.1115) -0.4890*(0.2080) -0.4612**(0.1442) -0.2563(0.1325) -0.3774**(0.1173) -0.1931(0.1631)

Secondary (base case) Tertiary 0.5143***(0.0790) 0.4061**(0.1501) 0.2545*(0.1039) 0.2632**(0.0939) 0.2970***(0.0829) 0.5013***(0.1166)

Full health -0.0072(0.0866) 0.0235(0.1662) 0.0546(0.1150) 0.0762(0.1032) 0.0408(0.0913) 0.0532(0.1273)

Better than Yellow (base case) Worse than Yellow -0.3593***(0.0978) -0.3914*(0.1861) -0.2988*(0.1295) -0.1603(0.1158) -0.181(0.1022) -0.2403(0.1418)

Constant 2 7.3035***(0.2332) 6.2763***(0.4343) 6.5029***(0.3043) 7.6493***(0.2765) 8.1346***(0.2422) 8.5908***(0.3373)

87

Table 3.6 continued

Explanatory Log Yb Per QALY Log Yh Per QALY Log Yi Per QALY Log Yq Per QALY Log Yr Per QALY Log Ye Per QALY

Variables Coef.(Se) Coef.(Se)¥ Coef.(Se)¥ Coef.(Se) Coef.(Se) Coef.(Se)

Yellow Trader Missing Income -0.1386*(0.0678) _ _ -0.2360***(0.0670) -0.3478***(0.0864) -0.1777(0.0932)

<USD 17,063 Q1 -0.0699(0.0725) _ _ -0.1684*(0.0730) -0.1883*(0.0953) -0.194(0.0994)

<USD 24,365 Q2 -0.0821(0.0681) _ _ -0.0173(0.0704) -0.0275(0.0914) -0.0856(0.0950)

<USD 33,701 Q3 0.0426(0.0672) _ _ -0.0094(0.0681) -0.0123(0.0886) 0.0323(0.0939)

<USD 48,032 Q4 -0.0655(0.0636) _ _ -0.0136(0.0657) -0.0866(0.0846) 0.0688(0.0919)

>USD 48,032 (base case) <25 yrs 0.3603***(0.0758) _ _ 0.2852***(0.0720) 0.3290***(0.0929) 0.4050***(0.1021)

25-34 yrs 0.1717**(0.0587) _ _ 0.1095(0.0580) 0.1399(0.0739) 0.1275(0.0786)

35-44 yrs (base case) 45-54 yrs -0.0315(0.0550) _ _ -0.0781(0.0563) -0.0669(0.0725) -0.1265(0.0763)

55-64 yrs -0.0434(0.0591) _ _ -0.0406(0.0599) -0.0361(0.0766) -0.008(0.0821)

65 yrs + 0.0817(0.0663) _ _ 0.0726(0.0692) 0.0536(0.0884) 0.1435(0.0987)

Netherlands 0.2485**(0.0813) _ _ 0.0313(0.0803) 0.2019(0.1049) -0.2399*(0.1151)

UK (base case) France -0.1071(0.0734) _ _ -0.2384**(0.0762) -0.1522(0.0979) -0.4136***(0.1108)

Spain 0.0475(0.0765) _ _ -0.2044**(0.0769) -0.1739(0.0975) -0.4114***(0.1110)

Sweden -0.0093(0.0753) _ _ 0.0392(0.0813) 0.1786(0.1037) -0.1619(0.1189)

Norway 0.0579(0.0835) _ _ -0.1354(0.0852) -0.1878(0.1058) -0.3226**(0.1246)

Denmark 0.2309**(0.0816) _ _ 0.1767*(0.0849) 0.1828(0.1054) -0.0461(0.1216)

Poland 0.1006(0.0849) _ _ 0.0502(0.0873) -0.0664(0.1086) -0.05(0.1240)

Hungary 0.1885*(0.0844) _ _ 0.0651(0.0873) 0.0779(0.1094) -0.0319(0.1241)

Gender -0.1037**(0.0374) _ _ -0.0863*(0.0375) -0.063(0.0483) -0.0553(0.0515)

OECDCOEFFICIENT 0.0427(0.0404) _ _ 0.0348(0.0405) 0.0728(0.0536) 0.0172(0.0548)

Middle -0.0569(0.0560)

-0.0109(0.0571) 0.0452(0.0733) 0.0213(0.0768)

Secondary (base case) Tertiary 0.0840*(0.0418) _ _ 0.0522(0.0417) 0.1852***(0.0537) 0.1309*(0.0575)

Full health 0.0588(0.0457) _ _ -0.0205(0.0452) -0.0202(0.0589) -0.0667(0.0634)

Better than Yellow (base case) Worse than Yellow -0.0626(0.0494) _ _ 0.0648(0.0516) -0.0052(0.0659) -0.0106(0.0704)

Std gamble question 0.4213***(0.0373) _ _ 0.3862***(0.0378) 6.0092(2280.04) 0.4035***(0.0521)

Constant 1 1.0093***(0.1202) _ _ 1.2308***(0.1232) 1.1059***(0.1570) 1.4045***(0.1716)

88

Table 3.6 continued Log Yb Per QALY Log Yh Per QALY Log Yi Per QALY Log Yq Per QALY Log Yr Per QALY Log Ye Per QALY

Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se) Coef.(Se)

N 9771 3781 6544 9898 9602 5450

N_censored 884 _ _ 846 571 450

chi2(df) 505.1623(24) (24) (24) 416.7345(24) 479.2143(24) 276.856(24)

Log likelihood -26009.7755 -10648.8382 -17844.0819 -28074.6089 -25826.6214 -14997.5606

chi2_c 2.5808 _ _ 0.2671 3.1564 0.0006

athrho 0.1039(0.0560) _ _ 0.0486(0.0820) 0.109(0.0558) -0.003(0.1218)

lnsigma 1.1894***(0.0077) _ _ 1.3757***(0.0075) 1.2499***(0.0075) 1.2849***(0.0100)

rho 0.1036 _ _ 0.0486 0.1085 -0.003

sigma 3.285 _ _ 3.9579 3.4898 3.6142

lambda 0.3402 _ _ 0.1923 0.3788 -0.0107

Heckman Selection Model: Regression dependent variable in log of PPP USD; selection dependent variable: Whether the respondent traded. Income quintiles global in PPP USD

*Denotes significance p-value < 0.05 ; ** p-value <0.01; *** p-value <0.001: at 5% level ¥ Denotes OLS regression

89

Table 3.7a Value of a QALY: Preliminary case – chained (green health state)15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Hungary Total

Ga N 1,047 902 1,076 1,078 1,115 854 1,143 758 790 8,763

Mean 28,187 76,883 63,846 116,347 190,166 128,529 72,999 34,430 61,958 88,242

Median 2,845 2,553 1,537 3,336 3,062 5,724 4,600 2,852 1,791 3,035

Gj N 508 463 574 501 505 369 477 515 521 4,433

Mean 32,549 30,170 35,760 126,969 792,671 98,506 115,013 49,813 32,964 146,397

Median 7,967 7,092 4,950 12,829 8,004 15,472 15,564 10,748 7,671 8,803

Gd N 1,025 889 1,139 1,059 1,066 767 987 907 957 8,796

Mean 47,132 45,245 137,175 145,004 297,032 105,006 51,895 59,831 39,349 106,714

Median 7,398 7,365 4,843 7,412 7,362 10,637 11,499 7,738 4,342 7,551

Gm N 1,128 960 1,173 1,140 1,117 814 1,140 968 954 9,394

Mean 20,200,000 59,035 55,686 385,330 10,200,000 79,554 75,140 1,118,126 29,598 3,837,764

Median 5,640 3,064 3,301 8,005 4,314 7,659 7,589 5,681 7,594 6,588

Gn N 1,120 962 1,156 1,150 1,132 826 1,159 982 961 9,448

Mean 10,200,000 28,590 24,501 168,106 33,825 38,765 136,527 572,108 17,038 1,320,432

Median 5,122 3,066 3,293 7,338 3,235 7,659 7,704 5,159 4,299 5,159

Gf N 479 424 542 530 547 417 543 361 416 4,259

Mean 2,640,908 33,796 83,252 115,041 58,424 86,409 109,787 14,300,000 28,582 1,574,197

Median 4,553 3,064 3,293 7,338 5,392 10,637 11,499 5,159 3,081 5,647

15 Responses were converted to international dollars at Purchasing Power Parity rates to allow aggregation across all ten countries prior to calculating the

mean and median

90

Table 3.7b Value of a QALY: Preliminary case – chained (yellow health state)15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Hungary Total

Yb N 1,067 921 1,132 1,119 1,109 839 1,051 930 1,014 9,182

Mean 1,122,777 82,991 356,524 267,874 123,494 139,657 215,099 1,125,724 47,177 386,923

Median 2,845 2,089 1,537 3,002 3,019 6,116 3,450 2,579 1,576 2,978

Yh N 442 397 478 445 432 291 397 465 468 3,815

Mean 38,217 27,435 25,600,000 35,711 522,457 38,527 50,954 38,918 29,434 3,292,094

Median 4,378 3,064 3,344 8,005 5,392 10,745 11,615 9,921 4,478 7,365

Yi N 765 651 831 806 774 528 769 757 721 6,602

Mean 1,507,397 87,871 43,809 53,310 162,594 25,949 36,889 710,515 15,890 303,994

Median 4,553 3,064 3,277 6,671 3,235 7,621 7,704 7,441 3,081 5,122

Yq N 1,132 967 1,152 1,112 1,109 808 1,136 964 946 9,326

Mean 20,200,000 30,766 42,096 126,709 224,592 73,995 1,082,313 1,110,216 25,091 2,754,621

Median 4,553 3,064 3,293 8,005 4,313 7,659 7,589 5,159 5,731 5,159

Yr N 1,098 950 1,100 1,103 1,105 801 1,152 958 941 9,208

Mean 10,400,000 15,617 18,929 72,811 27,123 32,488 30,270 566,359 16,564 1,322,896

Median 3,415 3,830 2,745 6,671 3,235 7,659 5,751 4,127 3,081 3,984

Ye N 580 528 648 617 575 422 616 604 592 5,182

Mean 8,002,221 191,029 349,322 237,047 373,508 286,562 298,261 958,300 280,835 1,231,043

Median 7,588 7,251 6,064 12,809 8,626 15,092 11,499 7,738 8,513 7,925

91

Table 3.8a: Value of a QALY: Chained trimmed – green health state15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Hungary Total

Ga N 1,038 894 1,067 1,068 1,104 846 1,132 751 783 8,682

Mean 12,862 12,421 11,770 31,616 23,261 45,518 20,906 22,132 14,506 21,399

Median 2,845 2,363 1,537 3,202 3,055 5,401 4,312 2,579 1,791 3,019

Gj N 504 459 569 496 500 365 473 510 516 4,392

Mean 24,892 23,267 26,890 38,162 35,200 37,427 57,389 40,023 26,132 34,097

Median 7,904 6,317 4,574 12,669 7,842 15,472 15,409 10,748 7,671 8,211

Gd N 1,017 882 1,129 1,050 1,056 760 978 898 948 8,717

Mean 22,211 21,182 24,852 50,083 34,824 33,685 34,063 35,773 19,617 30,581

Median 7,398 6,775 4,391 7,347 6,862 10,637 11,499 7,738 4,299 7,473

Gm N 1,119 953 1,163 1,130 1,107 807 1,131 959 946 9,314

Mean 27,418 29,308 25,965 52,876 28,805 41,298 42,118 29,188 21,791 33,236

Median 4,554 3,064 3,294 8,005 4,313 7,659 6,899 5,171 7,176 6,382

Gn N 1,111 954 1,146 1,140 1,122 818 1,149 973 952 9,364

Mean 18,623 15,897 16,613 33,789 19,287 26,399 31,456 22,434 13,222 22,057

Median 4,556 3,065 3,293 7,338 3,235 7,659 7,704 5,159 4,285 5,015

Gf N 475 421 538 525 542 413 538 358 412 4,222

Mean 22,911 25,245 28,884 49,342 27,471 39,136 47,191 39,076 19,519 33,498

Median 3,988 3,064 3,293 7,338 5,392 10,637 11,499 5,159 3,081 5,394

92

Table 3.8b: Value of a QALY: Chained trimmed – yellow health state15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Hungary Total

Yb N 1,058 914 1,122 1,109 1,099 831 1,041 921 1,005 9,099

Mean 22,951 22,762 18,666 33,665 33,555 43,438 43,584 24,425 20,060 28,878

Median 2,845 2,042 1,537 3,002 2,986 5,910 3,285 2,579 1,549 2,959

Yh N 439 394 474 441 428 288 393 460 464 3,781

Mean 23,133 23,285 20,115 25,629 27,696 29,474 37,309 32,104 21,181 26,386

Median 3,557 3,064 3,327 8,005 5,392 10,745 11,499 9,921 4,478 7,295

Yi N 759 646 824 799 767 523 762 750 715 6,544

Mean 16,819 14,848 13,927 28,049 16,908 21,602 25,328 23,104 11,565 19,129

Median 4,378 3,064 3,245 6,671 3,235 7,621 7,704 6,798 3,081 4,584

Yq N 1,124 960 1,143 1,102 1,100 801 1,127 955 938 9,250

Mean 27,345 20,525 19,952 39,356 33,142 41,003 41,316 25,716 19,117 29,726

Median 4,553 3,064 3,293 8,005 4,313 7,659 7,589 5,159 5,731 4,687

Yr N 1,089 942 1,091 1,093 1,095 794 1,142 949 932 9,127

Mean 15,738 13,228 11,317 26,299 18,292 24,757 24,796 18,601 10,938 18,247

Median 3,415 3,256 2,745 6,671 3,235 7,659 5,749 3,611 3,081 3,723

Ye N 575 523 642 611 569 418 610 598 586 5,132

Mean 68,643 60,342 102,109 70,841 63,128 113,730 68,996 46,688 108,345 77,323

Median 7,588 6,127 5,489 11,860 8,626 14,974 11,499 7,738 7,881 7,807

93

Table 3.9: Value of a QALY: Preliminary case - direct15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Palestine Hungary Total

Question A N 954 808 932 958 854 728 996 816 153 892 8,091

Mean 9,697 6,154 4,838 19,283 11,533 16,491 17,549 6,078 8,714 4,242 10,705

Median 1,138 1,531 768 1,334 1,617 2,659 2,875 1,084 251 860 1,433

Question B N 470 400 438 452 439 353 494 420 63 421 3,950

Mean 7,126 2,997 2,387 198,042 6,843 28,799 18,750 7,135 2,911 2,457 30,824

Median 80 8 110 667 216 1,064 575 671 301 329 334

Question D N 516 378 478 495 442 387 535 467 71 451 4,220

Mean 67,239 28,646 22,800 65,013 39,085 49,456 87,635 33,129 33,489 13,352 46,392

Median 3,415 3,064 1,866 5,337 4,745 6,382 9,199 4,128 402 2,866 4,255

Question E N 527 408 494 502 462 382 538 485 77 466 4,341

Mean 169,048 43,899 44,902 121,557 89,184 80,823 168,948 181,908 134,816 24,095 106,659

Median 2,846 720 1,647 5,670 4,044 8,510 11,499 5,160 1,005 3,081 3,582

Question F N 925 762 939 963 814 751 968 850 156 833 7,961

Mean 11,270 6,881 6,023 17,666 15,296 10,991 17,678 10,023 8,175 5,399 11,361

Median 1,138 766 549 1,334 1,078 2,127 2,875 1,032 126 716 1,098

Question G N 461 344 450 471 400 360 477 450 65 387 3,865

Mean 17,068 22,926 6,533 12,115 9,912 10,051 16,979 7,638 8,071 3,358 11,732

Median 285 230 165 734 539 1,064 1,150 516 75 308 516

Question I N 1,025 833 1,026 991 862 723 1,027 830 152 951 8,420

Mean 10,106 9,718 4,226 23,964 15,864 10,427 12,209 6,563 15,522 4,419 10,962

Median 0 31 55 334 302 532 149 310 75 179 155

Question J N 1,004 827 999 1,001 850 722 1,007 810 147 933 8,300

Mean 13,160 19,705 7,616 36,350 21,485 29,892 25,557 9,440 16,297 7,002 18,754

Median 1,138 1,532 768 3,335 2,264 5,319 4,600 1,548 201 1,075 1,725

Question L N 996 812 1,001 973 845 715 994 822 144 909 8,211

Mean 26,201 42,179 13,638 37,598 34,794 39,449 42,737 17,910 14,252 9,958 28,802

Median 1,138 2,298 549 2,668 2,696 5,319 4,600 2,580 126 1,075 2,149

Question M N 506 388 509 493 429 340 511 419 58 448 4,101

Mean 78,735 57,272 32,985 316,169 213,767 113,943 146,941 58,237 10,249 45,534 118,423

Median 1,707 1,532 2,196 8,005 5,392 10,637 8,624 5,160 653 3,582 3,582

Question N N 505 390 511 491 434 337 510 417 56 443 4,094

Mean 431,825 50,850 54,612 356,105 116,769 157,056 184,209 106,539 21,587 81,520 175,857

Median 0 0 2,196 8,005 4,313 10,637 4,600 6,192 1,608 3,582 3,235

Question O N 447 370 418 448 428 337 481 421 64 400 3,814

Mean 65,167 29,960 26,929 132,427 61,290 104,811 200,949 66,955 40,354 23,322 81,045

Median 2,846 1,532 3,293 7,338 5,392 10,637 11,499 7,740 2,010 3,582 5,319

Question P N 467 391 434 451 451 340 485 419 65 419 3,922

Mean 21,811 9,993 8,759 50,207 28,365 101,411 63,478 26,416 9,716 7,921 34,068

Median 2,276 1,532 2,196 3,202 3,062 5,106 4,600 3,096 603 1,433 3,062

94

Table 3.10: Value of a QALY: Direct trimmed15

Country Netherlands UK France Spain Sweden Norway Denmark Poland Palestine Hungary Total

Question A N 946 801 924 950 846 721 987 808 151 884 8,018

Mean 5,735 4,638 3,169 9,657 6,837 11,664 13,631 4,140 2,442 3,138 6,907

Median 1,138 1,379 768 1,334 1,617 2,659 2,875 1,032 201 824 1,334

Question B N 467 398 435 449 436 350 490 416 62 418 3,921

Mean 3,828 2,704 1,811 8,005 4,392 10,811 7,169 3,875 1,727 2,046 4,854

Median 74 0 110 667 216 1,064 575 568 226 315 329

Question D N 512 375 474 491 438 383 530 463 70 447 4,183

Mean 23,090 18,255 15,164 47,717 22,702 38,085 55,484 18,527 5,259 10,875 27,977

Median 3,415 3,064 1,756 5,337 4,313 6,382 9,199 4,128 402 2,293 4,128

Question E N 523 405 490 498 458 379 534 481 76 462 4,306

Mean 41,939 25,692 24,209 81,812 41,698 63,500 93,650 45,149 70,485 18,907 49,681

Median 2,846 306 1,647 4,336 3,235 8,510 11,499 5,160 1,005 3,081 3,450

Question F N 918 756 931 955 807 744 959 842 154 825 7,891

Mean 6,824 5,416 4,023 12,030 6,103 8,864 12,148 4,469 1,757 2,855 6,989

Median 1,138 766 549 1,334 1,078 2,127 2,875 1,032 100 716 1,078

Question G N 458 342 447 467 397 357 473 446 64 384 3,835

Mean 5,375 4,249 3,998 7,876 5,017 7,603 8,857 3,842 975 2,321 5,461

Median 285 207 165 734 539 957 1,150 516 75 308 516

Question I N 1,020 828 1,020 985 856 718 1,021 824 151 944 8,367

Mean 4,755 5,198 2,526 13,924 7,623 7,552 7,888 3,424 10,567 2,289 6,218

Median 0 23 47 334 302 532 149 310 75 179 153

Question J N 997 821 992 992 843 715 999 803 145 925 8,232

Mean 9,268 9,494 4,881 30,063 16,150 19,533 19,949 6,386 9,592 4,498 13,349

Median 1,138 1,532 604 3,335 2,157 5,319 4,025 1,548 201 1,075 1,659

Question L N 989 806 994 966 838 709 986 815 143 902 8,148

Mean 14,210 24,439 8,941 31,931 23,761 33,032 36,086 11,302 9,011 6,565 20,719

Median 1,138 2,298 549 2,668 2,588 5,319 4,600 2,580 100 1,075 2,064

Question M N 503 385 505 489 425 337 507 415 57 444 4,067

Mean 38,474 29,072 26,160 138,684 39,952 73,924 75,524 30,824 6,903 23,128 52,916

Median 1,707 1,532 2,196 7,338 5,392 10,637 8,049 5,160 301 3,582 3,450

Question N N 502 388 508 487 431 334 506 414 55 440 4,065

Mean 51,225 44,006 42,833 243,968 58,538 84,579 112,945 56,462 14,672 37,656 82,347

Median 0 0 2,196 8,005 4,313 10,637 4,600 6,192 1,608 3,582 3,152

Question O N 444 367 415 444 424 334 477 417 63 396 3,781

Mean 29,794 25,196 20,908 66,444 36,945 62,439 78,484 38,273 28,873 19,578 42,354

Median 2,561 1,532 3,293 7,004 5,392 10,637 11,499 7,740 2,010 3,582 5,160

Question P N 463 388 430 447 447 337 480 415 64 415 3,886

Mean 8,823 8,412 7,738 26,540 11,443 27,444 18,814 12,248 7,983 6,665 13,972

Median 2,276 1,532 1,756 2,935 3,062 5,106 4,600 3,096 603 1,433 2,959

95

Table 3.11: Coefficients from the Tobit regression models for each question - Direct15

96

Chapter 4: Views on health care priority setting amongst the public and decision makers across 10 European countries

4.1 Introduction This chapter represents EuroVaQ deliverable D23 and reports on the work conducted in Workpackage 7 (WP7): the “Q methodology study”, which was a co-operation between all project partners under coordination of Erasmus University Rotterdam. The ultimate objectives of WP7 were to elicit and describe, across the 10 countries participating in the EuroVaQ study:

1. the views on health care priority setting amongst the public 2. the views on health care priority setting amongst decision makers.

The structure of this chapter is as follows. Section 4.2 addresses objective 1 and presents the views on health care priority setting amongst the public across the 10 countries participating in the EuroVaQ project. Section 4.3 addresses objective 2 and reports the views amongst decision makers. Section 4.4 discusses the consistency and discrepancy between the two sets of results and section 4.5 concludes. This chapter refers to four appendices (see Figure 4.1):

1. Appendix 4.I (“Technical appendix to the main results of the Q study”) provides a more detailed account of the development and pilot testing of the study materials and the analysis techniques used in Sections 4.2 and 4.3.

2. Appendix 4.II (“General public views in 10 countries”) presents the results of

separate, national-level factor analyses of the data collected in each of the participating countries, which are also used in section 4.2.

3. Appendix 4.III (“Comparing results”) reports methods and detailed statistical

results for section 4.4, which compares the views on health care priority setting amongst the public across 10 countries (section 4.2) with the results from individual countries (appendix 4.II) and decision makers (section 4.3).

4. Appendix 4.IV (“Distribution of European views in the EuroVaQ sample”)

shows how the views amongst the public (section 4.2) were distributed across the 10 countries participating in the EuroVaQ project.

97

Figure 4.1 Structure of the chapter and connections with appendices

4.2 Views on health care priority setting amongst the public across 10

European countries

4.2.1 Introduction

In any country, the resources available to the health care sector are ultimately finite and typically insufficient to fulfil all the demands and needs for health care in the population. This means that decisions have to be made about which treatments to provide (and to whom), and consequently, which not (and to whom not). Deciding on how „best‟ to allocate health care resources, one has to answer intrinsically normative questions regarding the aim(s) of the health care sector. In the EuroVaQ project (European Value of a Quality adjusted life year website, 2010) a main goal was to obtain (the methodology to assess) monetary valuations of quality adjusted life years (QALYs). The need for such a value is apparent when one considers the conventional theory behind economic evaluations, which normally entail some way of ensuring that the benefits related to some intervention outweigh the associated costs. Therefore, many countries have adopted some form of economic evaluation in the decision making process on funding health care interventions. These evaluations assist in determining the relative value for money of treatments and usually take the form of a cost-utility analysis (CUA), in which the benefits of treatments are expressed in terms of quality adjusted life years (QALYs). An important reason for choosing a single, generic outcome measure in CUAs is that it facilitates the comparison of the relative value for money of treatments for different diseases and patient groups using different technologies. The efficiency of different

Views on health care priority

setting amongst the public

across ten European

countries (n=294)

Views on health care priority

setting amongst the public

across ten European

countries (n=294)

Views on health care priority

setting amongst decision

makers across ten European

countries (n=110)

Views on health care priority

setting amongst decision

makers across ten European

countries (n=110)

Section 4.2Section 4.2

Section 4.3Section 4.3

Technical appendix to the

main results of the Q study

Technical appendix to the

main results of the Q study

Appendix 4.IAppendix 4.I

General public views

in 10 countries

General public views

in 10 countries

Appendix 4.IIAppendix 4.II

Comparing resultsComparing results

Appendix 4.IIIAppendix 4.III

Distribution of European

views in the EuroVaQ sample

Distribution of European

views in the EuroVaQ sample

Appendix 4.IVAppendix 4.IV

DiscussionDiscussion

Section 4.5Section 4.5

Comparing public and

decision makers‟ views

Comparing public and

decision makers‟ views

Section 4.4Section 4.4

98

interventions is then expressed as an incremental cost per QALY ratio (the incremental cost effectiveness ratio or ICER). Assessing whether the benefits outweigh the costs then involves assessing whether additional QALY benefits are „worth‟ the additional costs or the „monetary value of a QALY‟. If the costs per QALY are below the value per QALY the treatment represents value for money. An important question, however, is whether all QALYs should have equal value in these decisions. This question is addressed in part by certain of the empirical estimations of WTP per QALY within the EuroVaQ project (specifically the „Direct‟ survey), but will receive in-depth attention as the focus of WP7. There is growing evidence that people do not attach equal value (or weight) to different QALYs and have preferences over the equitable distribution of health and health care. For instance, a QALY gained in a severely ill person may be valued (weighted) differently than a QALY gained in a person who is only mildly ill. Such considerations may have to do with equity preferences in distribution of health and health care and it is clear that they are influential within the health care sector and may even constitute the basis for social health care systems coming into being. This casts doubts on the validity of current decision rules, which typically attach (more or less) one and the same value to gained QALYs. Differential weighting of QALYs has been subject of debate for some time (Donaldson et al., 1988). It is therefore important to consider to what extent the reasoning underlying current decision rules reflect the preferences of members of the societies they serve. After all, when choices between treatments are made on behalf of society it is important that these choices align with public preferences for the distribution of health benefits. The relevance of equity considerations is also evident in policy debates regarding the valuation of benefits of end-of-life care and the treatment of „rare‟ diseases. Following new guidance for appraisal of end-of-life technologies, committees of the National Institute for Health and Clinical Excellence (NICE) in the UK are now permitted to recommend therapies for provision by the National Health Service (NHS) which have cost per QALY ratios above the standard cost-effectiveness threshold (National Institute of Health and Clinical Excellence, 2009. Longson and Littlejohns (2009) have calculated implicit QALY weights based on the decisions that have invoked this supplementary guidance and show that the committees have accepted an implicit QALY weight of 1.7. In other words, health gains from end-of-life care were considered to be of greater social value than other types of health gain. The Scottish Medicines Consortium (SMC) has taken similar steps by including a number of „decision modifiers‟ in their guidance, which may be called into play when the cost per QALY of a treatment is relatively high (Scottish Medicines Consortium NS, 2010). In the Netherlands, higher thresholds are allowed for interventions aimed at increasingly severe illnesses (Zorgverzekeringen, 2006). Although such amendments to the common decision rule of applying the same value to all types of QALYs have been established by policy makers, current guidelines are not grounded in robust research evidence reflecting the richness of the points of view among the public regarding the distribution of health and health care. Transparent and accountable committees and Citizens Councils provide societal legitimacy to

99

resource allocation decisions but such legitimacy can only be achieved if societal views on the distribution of health and health care are known and represented. Whilst a number of studies have investigated public preferences, they have typically focussed on specific treatments or patient groups, and a number of literature reviews have shown a wide variety of equity considerations and attitudes towards distribution of health care (Schwappach, 2002; Dolan et al., 2005; Tsuchiya and Dolan, 2005; Smith and Richardson, 2005). In-depth studies of public opinions regarding the relative value of health gains and incorporating the full range of relevant issues are rare; the recent social value of a QALY (SVQ) project in the UK is a notable example (Baker et al., 2010). Given the importance of issues of distributional equity and social value judgements in health care, but also to place its own findings into context, the EuroVaQ project investigated the points of view among the public and decision makers on the relative value of health gains in 10 countries across Europe. The results of this study are reported here.

4.2.2 Methods

The views of the public and decision makers about the prioritisation of health and health care were explored using Q methodology, a method which combines elements of qualitative and quantitative methods and provides a scientific foundation for the systematic study of subjectivity (Stephenson, 1935 and 1953; Brown, 1980; Smith, 2001; Watts and Stenner, 2005; Cross, 2005). Q methodology is well established in health research, with a rapidly growing number of published studies using the method during the past decade [e.g. Stainton, 1991; Stenner et al., 2000; Stenner et al., 2003; Baker, 2006; Eccleston et al.,1997; van Exel, 2006; Risdon et al., 2003), and more recently was introduced in health economics (Baker et al., 2006). In a Q methodological study participants rank a set of opinion statements and by doing so reveal their point of view toward the subject being studied (Smith, 2001; Cross, 2005). This ranking is achieved through a card sorting procedure known as a „Q sort‟. The rankings of the participants are subject to correlation analysis, under the assumption that correlation between individual rankings indicates similarity in viewpoint. By-person factor analysis is used to identify significant clusters of correlations, which are interpreted as viewpoints. Each point of view is then described using the weighted average ranking of the statements (as defined by the participants who are statistically significantly correlated with the factor, weighted by their correlation coefficients with the factor) (Stephenson, 1935; Smith, 2001). Q methodology is thus used to describe a population of viewpoints and not as is in conventional by-item factor analysis a population of people (Risdon, 2003). The analysis in fact concerns an inversion of conventional factor analysis, in the sense that Q methodology correlates people instead of items (i.e., the participants are treated as the variables). Consequently, the results may be generalised to the subject area from which the opinion statements were sampled, not to the population. Our study was conducted in three steps described below. These are described in detail in Appendix 4.1 and so are only outlined here.

100

Step 1: Collection and selection of opinion statements A set of statements for participants to rank order in their Q sorts was selected in several iterative stages. A literature search was conducted to identify papers relevant to the topic of study yielding 53 relevant papers. A long-list of issues pertinent to priority setting in health was compiled, categorised and discussed by the EuroVaQ team who considered comprehension, omission, and interpretation in different cultural or socio-economic contexts. Following pilot work and continuing discussion among the research team an initial set of 37 statements was reduced to a final set of 34 (see Table 4.2, pp115-116). Step 2: Administration of Q sorts Q sorts are individuals‟ rank ordering of the full set of statements, which are placed onto a sorting grid (see Figure 4.I.1 in Appendix 4.I). A relatively small respondent sample is necessary for Q analysis and in this study respondents were identified and invited to take part by EuroVaQ project partners in the 10 participating countries. The sampling frame was purposive and aimed to identify respondents who are likely to hold different views about priority setting in health. Partners were asked to recruit approximately equal numbers of respondents in six groups stratified by gender and age (younger than 25 and no children; 25 to 65; 65 and older). In addition, and where possible, the sample should also comprise of respondents with different socio-economic and health characteristics. Materials were designed such that Q sorts could be administered online or in person and most partners chose the online facility with the exception of Palestine and The Netherlands. In the UK a mixed approach was used. Whether in person or online, respondents were given similar guidance on the completion of the card sorting exercise, beginning by sorting the cards into one of three piles: agree, disagree and neutral. Next they were directed to place the cards from the three piles onto the sorting grid from most agree to most disagree, in doing so drawing finer distinctions between them. Step 3: Analysis of Q sorts In Q methodology by-person factor analysis is used to identify „shared viewpoints‟ amongst participants. Each individual‟s Q sort is correlated with every other individuals Q sort to bring together groups of Q sorts which are similar. A dedicated software package, PQMethod v2.11 (software and a manual can be downloaded from http://www.lrz.de/~schmolck/qmethod/), was used to conduct the analysis. In the first stage of the analysis a correlation matrix of all of the Q sorts is computed. This correlation matrix is then subject to factor analysis in order to identify groups of respondents with mutually high correlation coefficients, using common factor analysis techniques (in this study centroid factor analysis followed by varimax rotation). The final stage of the analysis is to identify the most appropriate factor solution for interpretation. To aid interpretation, each factor can be represented by an idealised Q sort, which represents the way in which a person with a correlation coefficient of 1 with that factor would have ranked the 34 statements. In the interpretation of the factor solution each factor is laid out and the characterising and distinguishing statements for the factor are examined. A statement is considered to be characterising for a factor if it was positioned in the outer two columns of the idealised Q-sort (i.e., if it had a rank score of -4, -3, +3 or +4). A statement is considered to be distinguishing for a factor if its score in the idealised Q-sort of that

101

factor was statistically significantly different from its score in all other factors. Attention is also paid to consensus statements, that is, those statements for which the score was not statistically significantly different between any pair of factors. As outlined above, data were collected for all 10 countries. These data could then be analysed in two main ways: pool the data and produce a European level analysis or analyse the data at the individual country level (so producing 10 analyses). In this chapter we present the results of the pooled European level analysis for the data from the public and decision makers. The country specific data from the public was also analysed and the results are presented in Appendix 4.II. Appendix 4.I provides a more detailed account of the research method used and the development and pilot testing of the study materials.

4.2.3 Results

A total of 294 members of the general public across the 10 countries of study provided useful responses for our analysis. A number of online respondents (33; 10.0%) were excluded because they did not give the exercise sufficient time to give meaningful responses and two in-person respondents (0.6%) were excluded because they revealed a lack of comprehension of the exercise. Table 4.1 presents the breakdown of respondents across the countries.

Table 4.1 Response general public across 10 countries

Country Interview mode Number Final sample

% of total

Denmark web 27 25 8.5%

France web 33 28 9.5%

Hungary web 58 47 16.0%

Norway web 25 23 7.8%

Palestine in person 20 20 6.8%

Poland web 32 29 9.9%

Spain web 30 29 9.9%

Sweden web 31 23 7.8%

The Netherlands in person 30 30 10.2%

UK in person / web 30 / 10 40 13.6%

Total 329 294

By-person factor analysis showed that the data supported a maximum structure of five factors. The statistical features of each factor solution were examined in detail. The 5-factor solution was selected because it had a clearly interpretable account for each factor and was supported by the written/ verbal comments respondents provided while completing the Q sort. The five factors together explain 52% of the total variance in the Q sorts. The salient positive and negative statements are listed in Table 4.2 on pp115-116.

102

General public point of view 1 (GP1) “Egalitarianism, entitlement and equality of access” The first of the five factors is an account which is relatively familiar from the health economics and other literatures and could be described as an egalitarian view of health care priority setting. Health care is seen as a basic right and the focus of the first factor is entitlement and access to health services. The two most important statements (#4, #29 with score +4; Table 4.2 on p115-116) emphasise that personal characteristics are not at issue. Since everyone is equally entitled to care, age, gender, income, social and economic status should be of no relevance to the distribution of health care resources. This is substantiated by the comments of respondents:

“I do not think that persons should be prioritised because of age, gender or income. This would give vulnerable groups bad odds, and would reduce equity across age and income. This would be a rejection of the principle of solidarity in society.” [Denmark_id22] “Access to health care should be equal. It would be unjust if privileged people, who already have a higher quality of life, would receive better treatment.” [France_id22] “It is unjust to discriminate against people in this way, in matters of health and life, one‟s income should not count.” [Hungary_id16] “A person‟s health status should be the most important criterion, because this is what treatment is about.” [Poland_id25] “Everybody has the same value and deserves equal treatment.” [Sweden_id10]

The egalitarian position is reinforced by opposition to the purchase of priority treatment by those with sufficient resources (#24), even if doing so would not affect others.

“Just simply to prevent a health care system at two tiers.” [France_id33]

In all other factors this statement is accepted, or else there is ambivalence towards the issue. The placing of statement #24 is, therefore, significant in the interpretation of this factor and strengthens the account described. It is unsurprising, and consistent with the rest of the narrative, that statements #3 and #5 are the most rejected statements in factor 1. This rejection of priority based on income, perhaps more unexpected, also goes for people from lower income groups (#16).

“The most important principle in a democratic health system is the solidarity principle.” [Denmark_id18] “In my opinion the health care system should be a solidarity system. Those who are in good health help those who are not, whether they have contributed or not.” [France_id11] “If people have contributed more, that is because they were able to.” [France_id18] “Ability to work or social status should not decide who gets the best help in a democratic society” [Norway_id17] “We need to meet everyone‟s basic health needs, irrespective of income or wealth.” [Spain_id24] “Treatment should not be related to income. Perhaps people from lower income groups have more health problems. Peoples‟ health problems should be attended to, but not because of their income position.” [The Netherlands_id29] “I don‟t think it matters two hoots about how much money you earn, you all live in the country and contribute in some way, and that doesn‟t mean taxes or like in other ways as well. You should all benefit from the system.” [UK_id21]

Having dependents (#7, #31), personal responsibility for health and lifestyle (#21, #25) and past use of health services (#34) are not considered relevant for priority setting either in this account.

103

The only exception to “equal access for all” is in the fact that some attention is given to medical need in setting priorities (#29). Doctors should be the ones to judge this (#12), but overall there is support that saving life (#8, #17) and prevention (#27) should matter. It is noteworthy that in this egalitarian account, relative to the other accounts, least weight is put on effectiveness of care (#15, #19) and that scarcity or opportunity cost is „sidestepped‟ somewhat (#6).

“Additional resources should be provided in order to avoid such considerations.” [Denmark_id22] “Every person has the right to life, regardless of the nature of their circumstances.” [Palestine_id12] “If there is a possibility of survival, even if it is 1% or less, one should hold on to it as its better than death.” [Palestine_id18]

Factor 1 is defined by participants from all 10 countries and 66 respondents have significant, pure factor loadings, with 2 respondents from Denmark, 16 from France, 6 from Hungary, 1 from Norway, 8 from Palestine, 4 from Poland, 9 from Spain, 3 from Sweden, 10 from the Netherlands and 7 from the UK. General public point of view 2 (GP2)

“Efficiency, severity and the magnitude of health gains” The second of the five factors is an account which focuses on prioritising health care towards patients in greatest need and treatments that will generate the most health benefits. The emphasis is hence on health-related features of patients and not on their personal, social or economic characteristics (#29, #4).

“Disease does not differentiate between income or sex and thus should priority to treatment not be affected by such characteristics.” [Palestine_id15] “Health care should be neutral to other things than need for care.” [The Netherlands_id18]

„Need‟ in this factor is related to severity of illness and encompasses concerns about worsening health (#18) and illnesses that are life threatening (#8), as well as health before (#1) and after treatment (#22).

“The first and most important thing is to save a person‟s life. All priority questions come after this.” [Hungary_id21] “Severity should be the main argument.” [Hungary_id56] “One should look at the degree of seriousness, even if not acute. Other health services must wait in acute situations where life can be saved.” [Norway_id7] “Everybody deserves care, but if someone is dying this person should be helped first and others should wait.” [The Netherlands_id18]

Efficiency (maximising benefit relative to cost) is important in this account and so priority is given to those treatments that generate the greatest health gain (#15, #19).

“As the health care budget is limited, the cheaper treatment with the same effectiveness should be financed.” [Hungary_id27] “The amount of money to spend on health care is limited, so it should be managed as efficiently as possible.” [Hungary_id50]

Because health care priorities should be based on medical need and the size of health gains achieved through treatment, past use of health services (#34) or culpability (#21, #25) have no role to play in this account.

“One should not focus on the causes of a particular disease, but on the effects of treating it.” [Poland_id8] “Everyone has a right to get treatment regardless of drinking, smoking and leading stressful lifestyle or staying at home and listening to Radio Maria.” [Poland_id14]

104

It is clear that, for this viewpoint, personal, social and economic characteristics are not relevant to health care priorities. There is particularly strong opposition to prioritising health care on the basis of age (#23, #14). The argument that health care for younger people should be prioritised because they would enjoy health benefits over a longer period of time is rejected (#26).

“Human life does not decrease in value over time.” [Hungary_id34] “You cannot discriminate access to health care by age. What is next? By race?” [Spain_id17]

No account should be taken of personal contribution to the health service (#3, #5), like in the other factors, neither should a special case be made for those in low income groups (#16).

“I do not see why people with more money should be given priority treatment.” [France_id7] “In case of equally serious illnesses, I find it unfair to prioritise patients by income.” [Hungary_id27] “Everybody should have the same health care, no matter what. The value of a human being is always equally great and has nothing to do with money. These are two totally different things” [Sweden_id23] “There are differences between people in willingness and ability to contribute, but in the end everyone should be treated equally.” [The Netherlands_id12]

Despite this, there is no opposition here to privately funded health care and people who are able to pay for their treatment should be allowed to do so (#24).

“I think buying health care should be allowed, it increases competition and decreases queues.” [Sweden_id22]

Factor 2 is also defined by participants from all 10 countries and 42 respondents have significant, pure factor loadings, with 3 respondents from Denmark, 2 from France, 11 from Hungary, 1 from Norway, 3 from Palestine, 7 from Poland, 5 from Spain, 3 from Sweden, 4 from the Netherlands and 3 from the UK. General public point of view 3 (GP3) “Fair innings, young people and maximising health benefits” The most prominent feature of factor 3 is the relevance of patients‟ age in priority setting. This preference for treating the young clearly distinguishes this account from the previous two factors. Three statements in the top nine ranked statements are significantly distinguishing statements prioritising younger over older patients (#14, #23, #26). In support of this, statement #4, generally accepted in other factors, is rejected in this factor since priorities can be set on the basis on the basis of age.

“Because the 80 year old has attained a long life, while the 30 year old has attained a lot less. Prioritising the 80 year old would be like giving money to those who are well off instead of those who don‟t have enough.” [Norway_id13] “A person who is 30 years old is still young and might have a critical position at work, perhaps a spouse and children, whereas an 80 year old person does not have such responsibilities (perhaps retired, children grown up and spouse dead).” [Palestine_id2] “Age is actually only a factor, it is reasonable that a younger person receives more resources than a 95 year old. However, the person‟s sex and income should not be a factor, the severity of disease and the patient‟s suffering should be the only factors.” [Sweden_id5]

There are two reasons for privileging the young, both of which are legitimate in the account represented by this factor. Statements #14 and #23 represent the „fair innings‟ argument. Statement #26 relates priorities for young people to „prospective health‟, a health maximising argument which emerges more generally in this factor (#15, #19, #32).

105

“Improving health is the goal of the health service. Good use of resources will maximise health.” [Norway_id10] “Everyone should have as many years as possible in as good health as possible.” [Norway_id11] “This would lead to the best possible equality between generations.” [Norway_id25]

The preference for treating younger people and maximising health gains aligns with the higher importance attached to prevention (#27) relative to all other accounts. And because investment should be directed towards those interventions that generate the most benefit, treatments should not be provided on a first come first served basis (#28) or, as in many of the other accounts, depend on patients‟ economic circumstances (#16, #5), financial contributions to the health service (#3, #5) or ability to purchase private health care by those with sufficient income (#24).

“If this must be fair, then one must also pay a bigger fine, if you drive too fast and if you earn more.” [Denmark_id10] “Give according to ability, receive according to need.” [Norway_id11] “It is of course a matter of fairness that all people are equally valued.” [Sweden_id5] “Everybody is worthy of treatment, irrespective of their background or income.” [Sweden_id5]

Factor 3 is defined by participants from six countries and 16 respondents have significant, pure factor loadings, with 1 respondent from Denmark, 1 from Hungary, 5 from Norway, 3 from Palestine, 4 from Sweden and 2 from the UK. General public point of view 4 (GP4)

“The intrinsic value of life and healthy living” The value of life and the personal responsibility for taking care of one‟s own health are prominent features of the ideals depicted by this account. It permits priority setting based on a number of personal characteristics, which is a considerable departure from the account in factor 1 which stipulated that everyone is just as worthy of treatment as everyone else. An important distinction made here is that priority setting on the basis of lifestyle is supported, apparent from the agreement with statement #25 and the strong rejection of statement #21.

“If one has explicitly chosen to live unhealthily, then it is ones own problem.” [Denmark_id26] “People must take responsibility for their actions and it is also true about health related things.” [Hungary_id22] “Everyone is responsible for their health and their burden on the system. Responsibly thinking people don‟t harm themselves and the society purposely.” [Hungary_id43] “Unhealthy lifestyle, in so far as dependent on patients will, should not burden overall costs of health care.” [Poland_id21] “Every person is responsible for his life. There is sufficient proof that smoking cigarettes is associated with the risk of lung cancer. If, despite that fact, somebody smokes, than this person is largely responsible for the disease and should suffer the consequences of his actions.” [Poland_id23]

In this account, the importance of personal responsibility for health appears to be linked to a belief that life is intrinsically valuable. The size of the health gain from treatment is important here as it is in all other accounts (#15, #19). Prevention is also ranked highly (#27), both in relation to individual lifestyle and population health. Life should be preserved, even when quality of life is poor (#17), and rescuing people from certain death should take priority over all other kinds of health care (#8).

“For me, the respect of life is above all.” [Hungary_id33] “Available resources are limited, their use should bring as much benefit as possible.” [Hungary_id52]

106

“At a time when people‟s health in the west is generally getting poorer and poorer, preventing poor health should be one of the things with the highest priority!” [Norway_id3] “In our civilization human life is believed to be of highest value.” [Poland_id21] “Prevention is cheaper than treatment. Healthy life style should be promoted.” [Poland_id23]

The high value attached to preserving life also shows from reactions to other statements: “12 people hold their lives rather than 1, even if it‟s only one month longer.” [Hungary_id24, #20] The position of statement #11 was somewhat puzzling in this context, but participants appear to have framed it in relation to personal responsibility and health gain:

“Why should a person with a bad quality of life be positively discriminated, if it was this person‟s own fault or responsibility that he / she got in this situation.” [Hungary_id22] “It‟s inefficient spending of health care budget. It‟s like pouring money into a leaky bucket.” [Hungary_id43]

This belief in life having inherent value is not connected to a strong priority for younger people, since statements #14, #23 and #26 achieve only moderate agreement. The preferences for prioritising parents with dependent children over similar people without dependents (#31), and the fact that it is the only factor not strongly rejecting priority for people with a partner over those who are single (#7), refers to a „family effect‟ of health.

“People who are suffering from an illness and also have care of children should be prioritised, since the illness will also greatly affect the children and their quality of life, development and feelings of security.” [Norway_id3]

Finally, it is notable the statements #3, #5 and #24 do not meet with the same strength of opposition in this account and all three statements are distinguishing in this respect for factor 4.

“If there is opportunity and need, why should I not get better care if I pay more and the others are not harmed?” [Hungary_id6] “If one does not harm others, one is in his right.” [Hungary_id31]

Factor 4 is defined by participants from four countries and 20 respondents have significant, pure factor loadings, with 1 respondent from Denmark, 14 from Hungary, 2 from Norway and 3 from Poland. General public point of view 5 (GP5)

“Quality of life above all else” Quality of life is the key issue in this account. Statement #17 is the most important positive statement in this factor and distinguishes it from all other factors.

“It makes no sense to spend a lot of resources on a patient whose quality of life thereafter will be closer to „surviving‟ than to „living‟. And not just for the money.” [France_id23] “Such people would only generate further costs and their life, also for them, may turn out to be merely suffering.” [Poland_id11] “I think quality of life is much more important than length of life. If quality of life is really bad, everyone around the patient suffers, while the patient may not even be aware he is still alive.” [The Netherlands_id16]

Statements #8 and #33, both of which suggest that life-saving treatments are the most important, are firmly rejected, supporting the notion that quality and not simply extension of life is the most important issue. The placing of these statements also significantly distinguishes this factor from all others.

“It is better to live less long but in better health.” [France_id5; The Netherlands_id17]

107

“We should avoid engaging in relentless treatment.” [France_id13] “Personally I see no benefit in prolonging life if it has no quality.” [Spain_id10] “You must consider more than survival only.” [Sweden_id27] “Quality of life is subjective. Some people don‟t want to live longer. The important thing is what the person involved wants, what he thinks of his quality of life, and what he decides on that basis.” [The Netherlands_id4] “Rescuing people who may then continue to live in a vegetative state puts a heavy burden on their family, their carers and the health care budget.” [The Netherlands_id4]

In conjunction, these three statements (#8, #17, #33) reveal a belief that quality and not length of life should be the aim of health services. The emphasis on quality of life in this account relates specifically to prognosis as a result of treatment and not simply the prioritising of patients who are severely ill and where little (or less) can be done to improve quality of life (#11, #18). Rather, the focus should be on treatments that restore health to an acceptable level (#22).

“Saving a life that is not worth living doesn‟t make sense.” [Denmark_id7] “If you can improve quality of life only by a small amount, what is the point of giving these patients priority over others? Moreover, I see little difference between a low and a slightly higher but still low quality of life.” [France_id23] “It‟s about improving your quality of life, in fact if you have a poor quality of life I would say that you shouldn‟t get the treatment possibly.” [UK_id7] “Because quality of life is more important than simply staying alive.” [UK_id39]

There is some support for the role of doctors in priority setting (#12) in this context

“Competent doctors can judge the relevance of treatment. It is no use putting lots of effort in treating a person who is in a bad health state and suffering, if there is no perspective for returning to a normal life.” [France_id17]

Prevention of ill health is important in this account (#27), and is often seen, by respondents, in terms of personal responsibility (#21, #25).

“This would be cheaper and in a lot of areas would be better instead of waiting until the „damage‟ is done. If we look at obesity, it is harder to lose weight when you have become fat than it is to prevent an obesity problem.” [Denmark_id15] “It is much smarter to prevent than trying to cure. It will save burden on the patient and his family, and money as well.” [France_id5; Spain_id11] “More economical and less suffering for society and most fair.” [Sweden_id6] “Prevention saves a lot of suffering, improves quality of life and gives people the opportunity to progress.” [The Netherlands_id3] “Everyone should live their life in a way that promotes health and minimises the need for care.” [The Netherlands_id4]

In line with other factors, personal financial contribution and economic productivity should be irrelevant to health care decision making (#3, #5).

“Everybody should have equal right to public benefits. Just because you don‟t make a lot of money should not mean that you are less entitled to treatment.” [Denmark_id15] “Classifying peoples‟ lives according to their economic capacity terrifies me.” [France_id17] “How much people earn and can therefore pay in tax differs. A person who has spent a life as a cleaner has no less right to health services than a person who had better educational opportunities and became a lawyer.” [Norway_id12] “Health is affected by knowledge, finances and resources. Even though some people, by chance and for various reasons, have better access to these factors than others, they have no right to priority over those with limited access. Do we want a society with room for „soft‟ values like compassion, fellowship and care or instead “hard” values like „survival of the fittest‟?” [Norway_id18]

Factor 5 is defined by participants from all 10 countries and 39 respondents have significant, pure factor loadings, with nine respondents from Denmark, 6 from France,

108

1 from Hungary, 2 from Norway, 2 from Palestine, 3 from Poland, 2 from Spain, 2 from Sweden, 6 from the Netherlands and 6 from the UK. Appendix 4.I provides a more detailed account of the data collection and European-level data analysis of the data from the public and decision makers and Appendix 4.II presents the results of separate country-level analyses of the general public data. In Appendix 4.III the five views presented in this section are compared with those found in the ten individual countries and amongst decision makers (see section 4.3). Finally, Appendix 4.IV shows how the five views were distributed across the 10 countries participating in the EuroVaQ project. 4.3 Views on health care priority setting amongst decision makers across 10

European countries

4.3.1 Introduction

In addition to eliciting the views of members of the public in each country, Q sorts were also completed with an additional sample of “decision makers”. The definition of decision makers was not rigid given the range of health care systems across each of the partner countries but was inclusive of people who are involved in preparing and/or making decisions about health budget allocations and exclusive of politicians, opinion leaders from the media and members of stakeholder groups.

4.3.2 Methods

Partners were asked to recruit up to 10 decision makers in their country. The same study materials were used for the decision makers as for the general public. In Palestine and the Netherlands the Q sorts were completed in face to face interviews, in the other countries decision makers completed the online survey. Otherwise the methods are identical to those used in the general public Q study reported above.

4.3.3 Results

Across the 10 countries, 110 decision makers completed the Q sort. Table 4.3 presents the breakdown of respondents across the countries.

Table 4.3 Response decision makers across 10 countries

Country Interview mode Number % of total

Denmark web 5 4.5%

France web 6 5.5%

Hungary web 12 10.9%

Norway web 23 20.9%

Palestine in person 3 2.7%

Poland web 10 9.1%

Spain web 13 11.8%

Sweden web 10 9.1%

The Netherlands in person 13 11.8%

UK web 15 13.6%

Total 110

109

For the decision makers a five factor solution proved the most appropriate and amenable to interpretation consistent with respondents‟ comments. The factor arrays for all five factors are presented in Table 4.4 on pp117-118. There was a good deal of overlap between the views expressed by decision makers with those described above from the general public. A few areas of distinction are drawn out in the discussion. Decision makers‟ point of view 1 (DM1) For the first of the five factors health-related features of treatments and patients are important. High priority is given to those treatments that generate the biggest health gain (#19) including when deciding between two different types of treatment (#15).

“Assuming these are for the same condition, the treatment giving the highest health gain would be better, provided this does not reduce the number of people with the condition having access to the treatment” (Uk_id01dm)

In this account access to health care should be based on health needs and not on the personal or social characteristics of the patient such as their location or economic situation (#29) or the amount of health care they have previously received (#34). Delivery of health services should also be based on health needs and not based on a simple waiting list or “first come first served” approach (#28). Despite this emphasis on health need decision makers associated with this factor do not believe that doctors should be the ones to judge which patients get priority based on their medical expertise (#12). There is rejection of the idea that a person‟s financial status should influence their access to treatment (#29; #3; #5) this includes positive discrimination towards those in lower socioeconomic groups (#16). This factor is also against individuals paying for treatment even if it does not affect others negatively as can be seen by the high placing of statement #24 which distinguishes this factor and is rejected by most others. Prevention is an important feature of this first factor with an emphasis on the need to prevent ill health rather than just attempt to cure people once illness occurs (#27).

“Preventing the disease from happening is the first most important from of prevention and saves many economic, social and national efforts” (Palestine_id05dm)

Factor 1 is defined by 11 defining Q sorts: 2 from France, 1 from Palestine, 7 from Spain and 1 from the UK. Decision makers‟ point of view 2 (DM2) A key belief associated with Factor 2 is that life saving is important and should take priority over other forms of health care (#8) even if the quality of saved lives would be bad (#17). This resonates in respondent‟s open ended comments:

“Life-saving operation is the primary goal of health care. Namely if it is not done, all other treatments lose sense.” (Hungary_id01dm) “First, we have to save life, we can only think about life quality, treatment and ranking afterwards.” (Hungary_id03dm)

This emphasis on saving lives distinguishes this factor from the other four (#8) and comments suggest and underlying rationale that people have to be alive in order to benefit in any way from treatment. Like factor 1, there is an emphasis on health outcomes as the main focus of the health care system. Maximising health gain is

110

therefore important in this account and the combination of statements #19, #15 and #32 suggest that this could be linked to the effectiveness of the treatment, and seeking to provide treatments which are most effective. The size of the expected health gain is less important if the outcome of the treatment is not certain with decision makers preferring to give priority to treatments with a certain outcome (#2). The placing of statement #21 (relating to patient‟s lifestyle) in the left hand tail of the grid may be related to perceptions about the effectiveness of treatments. If an individual has an unhealthy lifestyle the benefits that they receive from treatment may not be as great as those who have a healthier lifestyle. Factor 2 is defined by 10 Q sorts: 1 from Denmark, 4 from Hungary, 1 from Norway, 1 from Palestine, 1 from Poland, 1 from Sweden and 1 from Spain. Decision makers‟ point of view 3 (DM3) The account described by factor 3 is concerned with equal access to health care for all. This could be described as an egalitarian view of health care and is relatively familiar within the health care prioritisation literature and is a standard of many health care systems across Europe. The statements in the tails of the distribution illustrate that access to health care is a basic right based solely on health need and not on the personal or social characteristics of patients. This includes prioritisation based on age (#23), gender or socioeconomic factors (#29; #4), dependents (#7) or the amount of health care a person has received in the past (#34).

“Society must distribute health service according to equality and need from the common purse, even though some contribute more than others” (Norway_id01dm) “I believe this is a core ethical principal of health care provision. That it should be based on need not geographical or other circumstances. This is not to say everyone gets the same but rather the level of provision matches the level of need within an area or population group. Too often access to care is inversely proportionate to need.” (UK_id10dm) “We are all equal and the whole society should be influenced by fairness and equality. No one should be denied or receive less health care because of age, sex or income.” (Sweden_id08dm)

The rejection of prioritisation based on an individual‟s contribution through taxes or premiums (#3, #5) is a strong feature of this factor, the comments from respondents highlight an understanding that those who might be in most need of health care being those who can contribute least financially;

“It‟s not fair, if kids, elderly and people on social security should have inferior health treatment, because they contribute less than the working group. Additionally it would hit the lowest income groups the hardest, who are struggling with the biggest health problems.” (Denmark_id01dm) “Everyone has the right to health care and the least contributors may be the most in need of health care” (Palestine_id01dm)

In keeping with the ideal of equal access to health care this factor is the only one to agree that a person‟s lifestyle, and whether this may have contributed to an illness, should not be considered relevant to priority setting (#25; #21). Factor 3 has 42 defining Q sorts and is the only factor to be represented by decision makers across all countries, with 1 from Denmark, 3 from France, 1 from Hungary, 8 from the Netherlands, 2 from Norway, 7 from Palestine, 3 from Poland, 6 from Sweden, 1 from Spain and 8 from the UK.

111

Decision makers‟ point of view 4 (DM4) The most prominent feature of factor 4, setting it apart from the other viewpoints, is the relevance of a patient‟s age in the provision of health care. As can be seen in Table DM F4 3 statements relating to age are in the right hand „most agree‟ tail of the grid, all of which are distinguishing for this factor (#14; #23; #26). These statements present two main explanations on why priority should be given to younger people. The first seen in statements #23 and #14 (and the quotation below) is a „fair innings‟ argument.

“Everyone should have a chance to live up to mean age of the population” (Poland_id12dm)

The second is a health maximisation argument as can be seen in statement #26 and relates to a proposition that younger people have the potential to benefit from health improvements for longer. This health maximising argument is also consistent with the view (both here and in other factors) that treatments which produce the biggest health gain should be given priority (#15) and should be directed at those who would benefit the most (#32). Although age should be taken into account, other personal characteristics such as having a partner (#7) or a rare disease (#30) should not (perhaps because there is no overall health benefit to be gained by prioritising these groups). Positive discrimination towards those on lower incomes is also rejected (#16). In contrast to factor 2, those who define factor 4 believe that quality of life is important and treatment should not only be about saving lives or increasing life expectancy. The placing of statement #8, which is distinguishing for this factor, highlights the view that treatment should be about more than rescuing people from certain death and this is corroborated by a disagreement with statement #33 and the positive placing of statement #17. There are eight defining Q sorts for Factor 4: 1 from Spain, 1 from Poland, 6 from the Netherlands. Decision makers‟ point of view 5 (DM 5) As with most of the other factors, those who define factor 5 believe that in general everyone should have equal access to health care. Patient characteristics including, age, gender, disease type, or income should not be taken into account when deciding on who should receive treatment. This is demonstrated by the high levels of agreement given to statements #4 and #29 and the disagreement of statements #23 and #30. Despite this rejection of prioritisation based on personal characteristics, decision makers associated with this factor do believe that lifestyle is important with lower priority given to those who may be in some way responsible for their illness. This can be seen by the strong agreement with statement #25 which is distinguishing for this factor and is supported by the disagreement with statement #21.

“I will give biased example – the smoker poisons himself and in my opinion it is not obvious that he would deserve the treatment in the same degree.” (Poland_id01dm) “Some components of the life style should be taken into account in the treatment process, in particular such as cigarette smoking or alcohol drinking.” (Poland_id06dm)

Individuals should be allowed to pay for treatment if it does not affect others negatively (#24) and in keeping with the other factors, treatments which produce the

112

biggest health gain should be given priority (#19; #15) and preventing ill health is as important as providing a cure for illness (#27). There is some support from this factor that treatment should not just be about extending life (#33) but there is no strong view on prioritising treatments which improve quality of life (17). There are five defining sorts for factor 5: 1 from the Netherlands, 3 from Poland, 1 from Spain. 4.4 Comparing public and decision makers’ views

In addition to considering the general public (GP) and decision makers‟ (DM) views independently, there is also the opportunity to compare their views. The results of the two analyses were compared using a narrative approach alongside statistical analyses. Details of these methods are presented in Appendix 4.III. For reasons of brevity this section will only highlight the key similarities and differences between the solutions. In both the GP and DM solutions there is an „egalitarian‟ factor. From the factor descriptions it can be seen that the overall accounts given by GP1 (see Table 4.2, pp115-116 ) and DM3 (see Table 4.4, pp117-118) focus on equal access to health care for all and the irrelevance of personal characteristics such as age, income and socio economic status in determining who receives priority for health care. This descriptive comparison of the 2 factors is supported by a correlation coefficient of 0.83 (see Appendix 4.III) indicating a high level of similarity in the placing of the statements. DM3 also correlates highly with GP2 (0.89). There are clear points of contact between GP2 and the accounts described by GP1 and DM3: all three reject the prioritisation of health care based on patients‟ characteristics, however for GP2, the focus is prioritising health care based on patient need and maximising the health benefit. Life saving is important for both decision makers (DM2) and the public (GP4) and there was a high correlation between these two factors (0.81). The combination of the high rank score achieved by statement #8 along with a low score of statement #17 is evident in both factors indicating a support for giving priority to treatments which rescue people from certain death even if the quality of life after treatment is poor. Living a healthy lifestyle was also important with both factors placing statement #21 at the bottom of the grid. The key differences between the DM and the GP solutions are in the way in which whole accounts are constructed and, although the same issues are important to both DM and GP, it is the combination with other statements which makes the two factor solutions somewhat different. Consider the issues: quality of life, age and personal responsibility. Quality of life is important in both the decision makers and the general public factor solutions, encompassing part of the accounts given in DM4 and GP5. For the general public quality of life was the main emphasis of the factor while for the decision makers this was part of an account focusing on age, prioritising health care for young people, and quality of life after treatment. For the general public age was the focus of GP3. The account given by DM5 relates access to health care to need

113

rather than the personal characteristics of patients but holds that people should also be responsible for their own health. There is also a preference for certain over uncertain health gains of the same expected magnitude (statement #2). These issues are not seen in combination in the general public accounts. 4.5 Discussion The EuroVaQ Q study (Workpackage 7) represents a significant piece of work eliciting societal views regarding the appropriate principles for prioritising health care resources. Such research is important in order to develop health policies that reflect the views of the public. By examining the views held by both members of the public and health care decision makers, areas of consensus and divergence can be identified. Applying techniques of Q methodology 294 members of the public and 110 decision makers completed card sorts in 10 countries. Analysed separately these two datasets led to five public accounts and five decision makers accounts about how health and health care should be prioritised, which were then compared using narrative as well as statistical methods. There was a good deal of overlap between the accounts generated by the general public respondents and those derived from the decision makers. There was no evidence of a point of view existing only in one group and not in the other, although some individual statements (e.g. #2) were regarded differently. Nonetheless, the construction of each „whole account‟ was different between the factor solutions and the combination of beliefs comprising a factor was distinct in each dataset. Our finding that there is a plurality of societal views on a topic such as this is unsurprising. Statements in the Q set include issues of life and death and touch on fundamental political and religious beliefs. Hence the egalitarian account presented in GP1 and DM3 is an account which consistent with a somewhat left-wing view of the welfare state, concerning equality and solidarity. The inherent value of life described by GP4 may, for some people, be connected with religious views, and the focus on quality of life in terms of „a life worth living‟ in GP5 and DM4 with liberal views. Such views, whilst they may be at odds, should all be represented in decision making processes. The results presented in Appendix 4.I (e.g. Tables 4.I.7 and 4.I.8) and Appendix 4.II show that the same plurality of views among the public exists at the national level, but also that the support for the five views discussed in section 4.2 seems to differ across the participating countries. EU1, EU2 and EU5 appear to be prominent in all countries, while support for EU3 and EU4 may be limited in some. The analysis of the distribution of the five viewpoints across the 10 countries presented in Appendix 4.IV, and in particular Figure 4.IV.6, however shows that all the European level viewpoints have a significant role in the public debate on prioritisation of health care at the national level. The similarities we have identified between the views of decision makers and members of the public offers some reassurance as regards policy making, and the distribution of the views among the public provides some guidance to how these views could be weighted in this process. However, the nature of views about specific

114

decisions, especially in times of austerity, remains an open question. Whether or not decision makers and their publics would find as much common ground around actual investment and disinvestment decisions, as there would appear to be around the appropriate principles of policy, is an important matter for future research.

115

Table 4.2: Statement rank scores general public

Statement Point of view

GP1 GP2 GP3 GP4 GP5

1 If two groups of patients can benefit from a treatment equally and group A‟s health is fairly good and group B‟s health is poor, group B deserves priority.

+1* +1 +2 -1 -1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 0 0 0 -1

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -4 -4 -2* -4

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4* +4* -2* +1* +2*

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -4 -3 -3 0* -3

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

-2* 0 +1 0* +1

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-3 -2 -2 0* -2

8 Rescuing people from a certain death should take priority over all other kinds of health care. +2* +3* +1* +4* -3*

9 Treatment of illnesses that put the highest burden on patients‟ families should receive higher priority. +1 0 -1 +1 0

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

0 +1 -1 0 0

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 0 -1 -3* -2*

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +3 +1 0 +2 +2

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0 +1 0 -1 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

0 -1* +3* 0 +1*

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2* +3 +4 +3 +3

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-3 -3 -2 -3 -2

116

Statement Point of view

GP1 GP2 GP3 GP4 GP5

17 There is no sense in saving lives if the quality of those lives will be really bad. -3 -1 -1 -2 +4*

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 +2* +1 +2 0

19 Priority should be given to those treatments that generate the most health. +1* +2 +2 +3 +3

20 It is more important to extend one person‟s life by one year than to extend 12 people‟s lives by one month. -2 -1 +1* -1 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+2 +2 +2 -4* -2*

22 Priority should be given to treatments that restore health to an acceptable level, there‟s no use in improving health when the final result is still a very poor state of health.

+1 +1 -1 +1 +2*

23 Younger people should be given priority over older people, because they haven‟t had their fair share of health yet. -2 -3* +3* -1 0*

24 People should not be allowed to buy themselves priority treatment, even if it doesn‟t affect others negatively. +3* -2 -2 -4* -1*

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-2* -2 -3 +3* +1*

26 Priority should be given to younger people, because they may benefit from treatment for longer. 0 -4* +2* -1 0*

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +3 +2* 0* +4 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

0 0 -4* -2* -1*

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4 +4 +4 +2 +4

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-1* 0* -3 -3 -3

31 Parents with dependent children should be given priority over similar people without dependents. -1* -2* 0 +2* +1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

-1 -1 +3* +1 +1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -1 -1 0* -2 -4*

34 The amount of health care people have had in the past should not influence access to treatments in the future. +2 +3* +1 +1 +2

Note: * indicates statement is distinguishing (p<.05).

117

Table 4.4: Statement rank scores decision makers

Statement Point of view

DM1 DM2 DM3 DM4 DM5

1 If two groups of patients can benefit from a treatment equally and group A‟s health is fairly good and group B‟s health is poor, group B deserves priority.

+1 -1 +1 0 +2

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 +3* +1 +1 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -1 -4 -2 -2

4 Patient characteristics like age, gender or income should play no role in prioritising between people. -1 -1 +3* 0* +4*

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -2* -1 -4* -1 0

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

+2 +2 0 +1 +1

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-2 -2 -3 -3 -2

8 Rescuing people from a certain death should take priority over all other kinds of health care. +1 +4* +1 -4* -1*

9 Treatment of illnesses that put the highest burden on patients‟ families should receive higher priority. +1 0 -1 -1 +1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

-2 -1 0 +1 +1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

0 -2 0 -1 0

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. -3* 0 +2* 0 0

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. -1 -2 0 -1 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

+1 +1 -1* +4* +1

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2 +4 +3 +4 +2

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-4 -4 -2* -3 -1*

17 There is no sense in saving lives if the quality of those lives will be really bad. +2 -3* -1 +2 -1

118

Statement Point of view

DM1 DM2 DM3 DM4 DM5

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

0 +1 +2 +2 +2

19 Priority should be given to those treatments that generate the most health. +4 +3 +2* +1* +3

20 It is more important to extend one person‟s life by one year than to extend 12 people‟s lives by one month. -1 0 0 0 +1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

0* -3 +3* -2 -3

22 Priority should be given to treatments that restore health to an acceptable level, there‟s no use in improving health when the final result is still a very poor state of health.

+2 +1 +1 0 -2*

23 Younger people should be given priority over older people, because they haven‟t had their fair share of health yet. -1* 0* -3* +3* -4*

24 People should not be allowed to buy themselves priority treatment, even if it doesn‟t affect others negatively. +3* -3 0* -2 -3

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

0* +1 -3* +2 +4*

26 Priority should be given to younger people, because they may benefit from treatment for longer. -1 0 -2 +3* -2

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +4 +2 +2 +2 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-3* -4* -1 +1* 0

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +3 +2 +4 0 +3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-2 -2 -2 -3 -4

31 Parents with dependent children should be given priority over similar people without dependents. 0* +2* -2 -2 -1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

+1 +3 +1* +3* -1*

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -3 0* -1* -4 -3

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3 +1* +4* -1* +2

Note: * indicates statement is distinguishing (p<.05).

119

References

Acton J (1973) Evaluating public programs to save lives: the case of heart attacks.

RAND Corporation (R95ORC), Santa Monica Baker R (2006) Economic rationality and health and lifestyle choices for people with

diabetes. Social Science and Medicine 63: 2341-53. Baker R, Bateman I, Donaldson C, Jones-Lee M, Lancsar E, Loomes G et al. (2010)

Weighting and valuing quality adjusted life years using stated preference methods: preliminary results from the Social Value of a QALY (SVQ) Project. Health Technology Assessment, forthcoming.

Baker R, Thompson C and Mannion R (2006) Q methodology in health economics. Journal of Health Services and Policy Research 11: 38-45.

Baker R et al (2008) Valuing lives equally: Defensible premise or unwarranted compromise? Journal of Risk and Uncertainty 36: 125-138.

Brouwer WBF et al (2008) The New Myth. The Social Value of the QALY. Pharmacoeconomics 26(1): 1-4.

Brown S. (1980) Political Subjectivity. Applications of Q Methodology in Political Science. New Haven: Yale University Press

Byrne MM, O‟Malley K and Suarez-Almazor (2005) Willingness to pay per quality adjusted life year in a study of knee osteoarthritis. Medical Decision Making 25: 655-666.

Carthy T, Chilton S, Covey J, Hopkins, L, Jones-Lee M, Loomes G, Pidgeon N and Spencer A (1999) On the contingent valuation of safety and the safety of contingent valuation: part 2 – the CV/SG „chained‟ approach. Journal of Risk and Uncertainty 17: 187-213.

College voor Zorgverzekeringen (2006) Pakketbeheer in de praktijk. Diemen Cross RM (2005) Exploring attitudes: the case for Q methodology Health Education

Research 20: 206–213. Davis RK (1963). Recreation planning as an economic problem. Natural Resources

Journal 3: 239-249. Department for Transport (2007), Highways Economics Note No.1:2005. Department

for Transport: London. Dolan P, Gudex C Kind P and Williams A (1996) The time trade-off method: results

from a general population study, Health Economics 5, 141-154. Dolan P, Shaw R, Tsuchiya A, Williams A (2005) Qaly maximisation and people‟s

preferences: a methodological review of the literature. Health Economics 14: 197-208.

Donaldson C, Atkinson A, Bond J, Wright K. (1988) Should QALYs be programme specific. Journal of Health Economics 7: 239-257.

Donaldson C (1999) Valuing the benefits of publicly-provided health care: does „ability to pay‟ preclude the use of „willingness to pay‟? Social Science and Medicine 49: 551-563.

Donaldson C (2001) Eliciting patients‟ values by use of „willingness to pay‟: Letting the theory drive the method. Health Expectations 4: 180-188.

Donaldson C, Birch S and Gafni A (2002) The pervasiveness of the „distribution problem‟ in economic evaluation in health care. Health Economics 11: 55-70.

Donaldson C and Gerard K (with Jan S, Mitton C, Wiseman V) (2005) Economics of Health Care Financing: the Visible Hand. Palgrave Macmillan, London.

120

Drummond MF, Stoddart GL, O‟Brien BJ, Sculpher M and Torrance GW (2005) Methods for the Economic Evaluation of Health Care. Oxford University Press, Oxford.

Dupuit J (1844) On the measurement of utility of public works. Translated by RH Barback in International Economic Papers 2 (1952): 83-110 from “De la Mesure de l‟Utilite des Travaux Publics” Annales des Ponts et Chausees, 2nd Series Vol 8 (Another English version in Arrow JJ and Scitovsky T, eds, Readings in Welfare Economics, George Allen and Unwin, London, 1969).

Eccleston C, Willams ACD, Stainton Rogers W. (1997) Patients' and professionals' understandings of the causes of chronic pain: blame, responsibility and identity protection. Social Science and Medicine 45(5):699-709.

European Central Bank (2007) Euro Foreign Exchange Reference Rates. [cited 8/10/2007: www.ecb.int/stats/eurofxref/.]

European Commission Directorate-General Regional Policy (2006) Guidance on the Methodology for Carrying out Cost-Benefit Analysis

European Observatory on Health Care Systems (2002) Health Care Systems in Eight Countries: Trends and Challenges London School of Economics and Political Science.

European Value of a Quality adjusted life year (EuroVaQ): project website. 2010; Available from: http://research.ncl.ac.uk/eurovaq/.

EuroQoL website: http://www.euroqol.org/home.html Gafni A (1991) Willingness to pay as a measure of benefits: relevant questions in the

context of public decision making about health care programmes. Medical Care 29: 1246-52.

Garber AM and Phelps CE (1997) Economic foundations of cost-effectiveness analysis. Journal of Health Economics 16: 1-31.

Garber AM. Advances in cost-effectiveness analysis of health interventions. In: Culyer AJ, Newhouse JP, editors. Handbook of Health Economics: Elsevier; 2000

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

Hagenaars AJM, De Vos K, and Zaidi A. (1994). Poverty statistics in the late 1980s: Research based on micro-data Luxembourg: Office for official publications of the European Union.

Hjalte K, Hjelmgren J, Johansson F, Person U. Betalningsviljan för ett kvalitetsjusterat levnadsår- en pilotstudie. IHE e- rapport 2005:1

Haybittle J. (1998) The use of the Gompertz function to relate changes in life expectancy to the standardized mortality ratio. International. Journal of Epidemiology 27(5): 885-889.

Hirth RA, Chernew ME, Miller E, Fendrick MF and Weissert WG (2000) Willingness to Pay for a Quality-Adjusted Life Year: In Search of a Standard, Medical Decision Making 20: 332-342.

Heckman JJ (1976): "The Common Structure of Statistical Models of Truncation, Sample Selection, and Limited Dependent Variables and a Simple Estimator for Such Models," Annals of Economic and Social Measurement, 5, 475-492.

Heckman JJ (1979): "Sample Selection Bias as a Specification Error," Econometrica 47, 153-161.

Johannesson M (1995) The relationship between cost effectiveness analysis and cost benefit analysis. Social Science and Medicine 41: 483-489.

121

Johannesson M, Jonsson B and Borgquist L (1991) Willingness to pay for anti-hypertensive therapy – results of a Swedish pilot study. Journal of Health Economics 10: 461-474.

Jones-Lee M (1976) The value of life: an economic analysis. London and Chicago: Martin Robertson and University of Chicago Press.

Jones-Lee M (1989) The economics of safety and physical risk. Oxford: Blackwell. Jones-Lee M, Hammerton M and Philips PR (1985) The value of safety: Results of a

national sample survey. Economic Journal, 95: 49-72. Jones-Lee MW (1989) The Value of Safety and Physical Risk, Blackwell, Oxford. Jones-Lee M, Loomes G and Philips PR (1995) Valuing the prevention of non-fatal

road injuries: contingent valuation vs. standard gamble. Oxford Economic Papers 47: 676-695.

Kind P, Hardman G and Macran S (1999) UK population norms for EQ-5D. Centre for Health Economics, University of York: York.

King JT, Tsevat J, Lave JR and Roberts M (2005) Willingness to pay for a quality adjusted life year: implications for societal health care resource allocation. Medical Decision Making 25: 667-677.

Longson C and Littlejohns P (2009) Update Report On The Application Of The „End-Of-Life‟ Supplementary Advice In Health Technology Appraisals: National Institute for Health and Clinical Excellence

Mason, H (2007) Monetary valuation of health outcomes for use in national policy formulation. PhD Thesis; Newcastle University

Mason H, Jones-Lee MW and Donaldson C (2009) Modelling the monetary value of a QALY: a new approach based on UK data. Health Economics 18: 933-950.

Mishan E (1971) Evaluation of life and limb: a theoretical approach. Journal of Political Economy 79: 687-705.

National Institute for Health and Clinical Excellence (2004) Guide to the Methods of Technology Appraisal. National Institute for Clinical Excellence, London.

National Institute for Health and Clinical Excellence (2005) Social value judgements. Principle for the development of NICE guidance.

National Institute for Health and Clinical Excellence (2009) Appraising life-extending, end of life treatments. London.

O‟Brien B and Gafni A (1996) When do the “dollars” make sense? Toward a conceptual framework for contingent valuation studies in health care, Medical Decision Making 16: 288-299.

Olsen JA, Donaldson C (1998) Helicopters, hearts and hips: using willingness to pay to set priorities for public sector health care programmes. Social Science and Medicine 46:1-12.

Olsen JA, Donaldson C, Shackley P and the EuroWill Group (2005) Implicit vs explicit ranking: on inferring ordinal preferences for health care programmes based on differences in willingness-to-pay. Journal of Health Economics 24: 990-996.

Olsen JA, Donaldson C and Periera J (2004a) The insensitivity of „willingness to pay‟ to the size of the good: new evidence for health care. Journal of Economic Psychology 25: 445-460.

Olsen JA, Smith RD (2001) Theory versus practice: A review of „willingness-to-pay‟ in health and health care. Health Economics 10(1): 39-52.

Pauly M (1995) Valuing health care benefits in money terms, in Valuing health care: costs, benefits and effectiveness of pharmaceuticals and other medical technologies, F. Sloan, Editor. Cambridge University Press: Cambridge.

122

Persson U et al (2001) The value of a statistical life in transport: Findings from a new contingent valuation study in Sweden. Journal of Risk and Uncertainty 23(2): 121-134.

Persson U, Hjelmgren J. Hälso- och sjukvården behöver kunskap om hur befolkningen värderar hälsan. Läkartidningen, 2003: (43) 3436-7

Pinto Prades JL, Loomes G and Brey R (2009) Trying to estimate a monetary value of a QALY. Journal of Health Economics 28: 553-562

Protière C, Donaldson C, Luchini S, Moatti JP and Shackley P (2004) The impact of information on non-health attributes on willingness to pay for multiple health care programmes, Social Science & Medicine 58: 1257-1269.

Rawlins MD, Culyer AJ (2004). National Institute for Clinical Excellence and its value judgments. British Medical Journal 329: 224-227.

Richardson J, Smith RD (2004) Calculating society‟s willingness-to-pay for a QALY: key questions for discussion. Applied Health Economics and Health Policy 3 (3): 125-126.

Risdon A, Eccleston G, McCracken L. (2003) How can we learn to live with pain? A Q-methodological analysis of the diverse understandings of acceptance of chronic pain. Social Science & Medicine 56: 375-386.

Schwappach DLB. (2002) Resource allocation, social values and the QALY: a review of the debate and the empirical evidence. Health Expectations 5: 210-222.

Scottish Medicines Consortium NS. (2010) Guidance to the Manufacturers for Completion of New Product Assessment Form (NPAF)

Shepard DS and Zeckhauser RJ (1982) Life-cycle consumption and willingness to pay for increased survival, in The Value of Life and Safety, M. Jones-Lee, Editor. North Holland: Amsterdam.

Smith NW. (2001) Current systems in psychology: history, theory, research, and applications. Wadsworth

Smith RD (2003) Construction of the contingent valuation market in health care: a critical assessment. Health Economics 12(8): 609-628.

Smith RD (2005) Sensitivity to scale in contingent valuation: the importance of the budget constraint. Journal of Health Economics 2005; 24: 515-529.

Smith RD, Richardson J (2005) Can we estimate the „social‟ value of a QALY? Four core issues to resolve. Health Policy 74: 77-84.

Stainton Rogers W. (1991) Explaining health and illness: an exploration of diversity. London: Harvester Wheatsheaf

Stenner PHD, Dancey CP, Watts S. (2000) The understanding of their illness amongst people with irritable bowel syndrome: A Q-methodological study. Social Science & Medicine 51: 439–452.

Stenner PHD, Cooper D, Skevington SM. (2003) Putting the Q-into quality of life; The identification of subjective constructions of health-related quality of life using Q-methodology. Social Science & Medicine 57: 2161–2172.

Stephenson W (1935) Correlating persons instead of tests. Character Pers. 4: 17-24.22.

Stephenson W (1953) The study of behaviour: Q technique and its methodology. Chicago: University of Chicago Press

Szende A, Williams A (2004) eds. Measuring Self-Reported Population Health: An international perspective based on EQ-5D. SpringerMed Publishing Ltd. Hungary.

Tarn TYH and Smith MD (2004) Pharmacoeconomic guidelines around the world. ISPOR Connections 10(4): 5.

123

Tsuchiya A, Dolan P. (2005) The QALY model and individual preferences for health states and health profiles over time: a systematic review of the literature. Medical Decision Making 25: 460-467.

van Exel NJA, Brouwer WBF, van den Berg B, Koopmanschap MA (2006) With a little help from an anchor: Evidence of starting point bias in contingent valuation of informal caregiver time inputs. Journal of Socio-Economics 35(5): 836-853

van Exel J, de Graaf G, Brouwer WBF. (2006) "Everyone dies, so you might as well have fun!" Attitudes of Dutch youths about their health lifestyle. Social Science and Medicine 63: 2628-2639.

Viscusi WK and Aldy JE (2003) The Value of a Statistical Life: a Critical Review of Market Estimates Throughout the World. Working Paper 9487, National Bureau of Economic Research, Cambridge Mass.

Watts S, Stenner P. (2005) Doing Q methodology: theory, method and interpretation. Qualitive Research in Psychology 2: 67-91.

Yeung RYT, Smith RD, McGhee SM (2003) Willingness to pay and size of health benefit: an integrated model to test for „sensitivity to scale‟ Health Economics 12(9): 791-796.

Cou

ntry

-spe

cific

dat

a a

nd s

ourc

es

Cou

ntry

V

PF (v

alue

in

Eur

os)

Val

ue e

stim

ated

thro

ugh

a sp

ecifi

cally

des

igne

d st

ated

pr

efer

ence

surv

ey (c

ontin

gent

va

luat

ion

met

hod)

?

Qua

lity

of li

fe (w

ith r

efer

ence

) D

isco

unt

Rat

e Po

pula

tion

and

Life

Tab

les (

with

ref

eren

ce)

Den

mar

k 1,

396,

460

No

Val

ue e

stim

ated

usi

ng H

uman

Cap

ital

App

roac

h

Unp

ublis

hed

mat

eria

l pro

vide

d by

Jan

Søre

nsen

, CA

ST, U

nive

rsity

of S

outh

ern

Den

mar

k

Bas

ed o

n da

ta c

olle

cted

200

0-20

01.

3%

http

://w

ww

.stat

bank

.dk/

stat

bank

5a/d

efau

lt.as

p?w

=102

4

Fran

ce

1,50

0,00

0

No

Bas

ed o

n re

sults

of s

ever

al c

ontin

gent

va

luat

ion

stud

ies

No

data

ava

ilabl

e 1%

IN

ED (I

nstit

ut n

atio

nal d

'étud

es d

émog

raph

ique

s)

ww

w.in

ed.fr

/en/

Hun

gary

178,

560

1,00

1,40

2

No

Val

ues e

stim

ated

usi

ng h

edon

ic p

ricin

g

(com

pens

atin

g w

age

diff

eren

ce)

Szen

de A

and

Will

iam

s A. (

eds)

(20

04)

5.5%

w

ww

.ksh

.hu

Net

herla

nds

2,20

0,00

0

Oth

er st

ated

pre

fere

nce

met

hod

Pe

rson

al c

omm

unic

atio

n fr

om E

lly S

tolk

2.

5%

WW

W.C

BS.

NL

Nor

way

1,

955,

671

U

nkno

wn

No

data

ava

ilabl

e 4%

w

ww

.ssb.

no

Pale

stin

e N

o da

ta

avai

labl

e

No

data

ava

ilabl

e N

o da

ta

avai

labl

e N

o da

ta a

vaila

ble

Pola

nd

No

data

av

aila

ble

N

o da

ta a

vaila

ble

No

data

av

aila

ble

ww

w.st

at.g

ov.p

l/gus

/inde

x_EN

G_H

TML.

htm

Spai

n 2,

500,

000

No

Val

ue e

stim

ated

usi

ng H

edon

ic w

age

pric

ing

Szen

de A

and

Will

iam

s A. (

eds)

(20

04)

No

data

av

aila

ble

ww

w.in

e.es

Swed

en

1,86

3,19

8

Yes

Sz

ende

A a

nd W

illia

ms A

. (ed

s) (

2004

) 3%

Po

pula

tion

data

: d

ate

crea

ted

13/0

4/20

07

http

://w

ww

.scb.

se/te

mpl

ates

/tabl

eOrC

hart_

___1

5928

5.as

p

UK

2,

064,

868

Y

es (D

epar

tmen

t of T

rans

port,

200

7)

Kin

d et

al (

1999

) 1.

5%

Popu

latio

n da

ta:

ww

w.st

atis

tics.g

ov.u

k/st

atba

se/s

sdat

aset

.asp

?vln

k=74

91&

Mor

e=Y

. Li

fe ta

bles

: ww

w.g

ad.g

ov.u

k

Appendix 2.1

References for data in Appendix 2.1 VPF Denmark

Trafikministeriets (2004), Nøgletalskatalog - til brug for samfundsøkonomiske analyser på transportområdet.

France COMMISSARIAT GENERAL DU PLAN BOITEUX (M), BAUMSTARK (L) CGPC.- Rapport relatif aux suites à donner aux préconisations du groupe de travail du Commissariat général du plan présidé par Marcel Boiteux sur le choix des investissements et le coût des nuisances dans le domaine des transports. Transports : choix des investissements et coût des nuisances. Rapport dit '' Rapport Boiteux 2''. Paris, Documentation française (La), 2001.- 325 p., tabl.

Hungary Adorjan, R: Value of human life. Investigation of an unconventional economic issue in Hungary. Budapest University of Economics, 2004

Kaderjak, P, et al: A csökkenő halálozási és baleseti kockázat közgazdasági értéke Magyarországon (Economic value of decreasing risk of mortality and injuries in Hungary) Közgazdasági Szemle (Economic Review), vol. LII., March 2005 (p. 231-248.)

Netherlands

Institute for road safety research http://www.swov.nl/rapport/R-2005-04.pdf Norway

Personal communication from Jan Abel Olsen Spain

Antoni Riera Font, Aina M. Ripoll Penalva y Joseps Mateu Sbert Estimación del valor estadístico de la vida en España: una aplicación del modelo de salarios hedónicos, Hacienda Pública Española, 181, 1/2007, pp. 29-48.

Sweden

Vägverket: Vägverkets samhällsekonomiska kalkylvärden. Publication 2006:127, Borlänge

UK

Department for Transport. Highways Economics Note 1: 2005. 2007. London.

Appendix 2.1

Discount Rate Denmark Data from the following websites accessed 13/04/2007: http://www.fm.dk/1024/visPublikationesForside.asp?artikelID=2628&mode=hele http://www.sns.dk/Nationalparker/Organisation/FG_20041007/Vejledning%20_okonomiske.pdf http://glwww.mst.dk/default.asp?Sub=http://glwww.mst.dk/udgiv/publikationer/2006/87-7052-141-7/html/kap02.htm

France Commissariat General au Plan http://www.plan.gouv.fr/intranet/upload/actualite/Rapport%20Lebegue%20Taux%20actualisation%2024-01-05.pdf

Hungary

Personal Communication with expert from Hungarian Central Bank Netherlands – Data from the following website accessed 13/04/2007

http://www.minfin.nl/binaries/minfin/assets/pdf/actueel/kamerstukken/2007/03/irf07-90.pdf

Norway

Personal communication from Jan Abel Olsen

Spain Guidance on the methodology for carrying out cost benefit analysis. European Commission. Directorate General Regional Policy. Working Document No. 4. 08/2006.

Sweden

Läkemedelsförmånsnämnden (LFN), The Swedish Pharmaceutical Benefit Board Guidelines for economic evaluation Ref: www.lfn.se

UK

Great Britain HM Treasury (2004) The Green Book. London, TSO.

Appendix 2.1

Life expectancy gains version of Approach 1

Suppose that a large representative group of individuals is offered small individual reductions, pi, i

= 1,2……….n, in the risk of death during the coming year. Denoting the ith individuals marginal rate

of substitution of wealth for the risk of death during the coming year by mi it follows that aggregate

willingness to pay for the risk reduction will be given by:

V = n

ii pm (A1)

Furthermore, the risk reductions will necessarily result in gains in life expectancy for each of the n

individual’s. More specifically, as shown in Jones-Lee(1976) [20] denoting the ith individuals

remaining life expectancy by Ei and probability of surviving the coming year by pi then:

i

i

i

i

P

E

P

E 1

(A2)

However, for the typical individual pi will be very close to 1 and Ei will exceed 30, so that to a

reasonable approximation we can write:

iEP

E

i

i

(A3)

It therefore follows that the individual gains in life expectancy, Ei , resulting from the individual

risk reductions, pi, will be given by:

Ei = Ei pi, i=1,2,….n. (A4)

Now suppose that the individual risk reductions are all set equal to

niE

1. It then follows

immediately from equation (A4) that taken across the whole group of n individuals, the aggregate

gain in life expectancy will be given by:

n

ni

in

i EEδE

1 (A5)

= 1 (A6)

That is, the aggregate gain in life expectancy will be exactly one year.

But from equation (A1), with pi = iE

1

EM

i

iV

(A7)

that is,

V = E

M (A8)

Appendix 2.2

Where M and E denote, respectively, the arithmetic means of Mi and Ei so that if the group of n

individuals is indeed representative then the WTP based value of an aggregate gain of one year of

life expectancy is given by the population mean of Mi (which is under the standard WTP definition,

the VPF) divided by the population mean of Ei.

The key assumption underpinning the above argument is that all of the affected individuals enjoy the

same risk reduction. Suppose that instead the individual risk reduction are allowed differ and are in

fact set so that all n individuals enjoy the same gain in life expectancy. More specifically suppose

that the risk reductions are set so that:

n........21,i,nE

1δpi

i

(A9)

It then follows from (A4) and (A8) that:

n.........2,1,i,n

1δEi (A10)

Taken over the whole group of n individuals, the aggregate gain in life expectancy would therefore

be exactly one year. In turn, from (A1) and (A9), aggregate willingness to pay would be given by:

nE

M

nV

i

i (A11)

that is

E

Mn

1V

i

i

n (A12)

In this case it is therefore clear that the VOLY will be given by the arithmetic mean of i

i

E

M , as

opposed to the arithmetic of Mi divided by the arithmetic mean of Ei as in the previous case.

Notice also that the mean of the ratio i

i

E

M will be equal to the ration of the means of Mi and Ei

only in exceptional circumstances.

Appendix 2.2

VPF estimates in home currency

Table A1. Value per Life Year

Denmark

(DKK)

France

(Euro)

Hungary

(HUF)

Netherlands

(Euro)

Norway

(NOK)

Spain

(Euro)

Sweden

(SKK)

UK

(GBP)

Approach 1 324,225 45,064 1,653,889 /

9,270,676 64,991 433,004 76,723 518,249 43,805

Approach 2a 152,897 21,500 782,855 /

4,388,198 32,176 212,147 40,492 243,448 23,199

Approach 2b 289,942 41,109 1,310,455 /

7,308,307 59,016 396,763 62,465 533,545 39,257

Table A2. Discounted Value per Life Year

Denmark

(DKK)

France

(Euro)

Hungary

(HUF)

Netherlands

(Euro)

Norway

(NOK)

Spain

(Euro)

Sweden

(SKK)

UK

(GBP)

Approach 1 514,051 53,580 3,244,511 /

18,186,719 97,533 804,847 128,871 825,672 56,098

Approach 2a 231,898 26,016 1,713,955 /

9,607,373 48,345 346,246 70,517 368,666 29,691

Approach 2b

584,745

52,228 3,397,496 /

18,198,253 104,294 2,446,598 131,838 1,318,953 56,746

Appendix 2.3

Table A3. Value per QALY

Denmark

(DKK)

France

(Euro)

Hungary

(HUF)

Netherlands

(Euro) Norway

Spain

(Euro)

Sweden

(SKK)

UK

(GBP)

Approach 1

378,624 2,643,576 /

14,818,254 76,640 107,764 644,614 55,708

Approach 2a 172,554 1,031,196 /

5,780,245 36,937 51,154 291,353 27,286

Approach 2b 331,252 1,940,844 /

10,892,412 69,399 84,310 642,746 47,253

Table A4. Discounted Value per QALY

Denmark

(DKK)

France

(Euro)

Hungary

(HUF)

Netherlands

(Euro) Norway

Spain

(Euro)

Sweden

(SKK)

UK

(GBP)

Approach 1

595,760 4,998,632 /

28,019,239 180,295 178,527 1,017,122 70,765

Approach 2a 244,071 2,277,858 /

12,768,265 55,274 92,488 464,855 34,925

Approach 2b 653,898 4,792,212

26,036,006 122,598 171,476 1,541,369 53,838

Appendix 2.3

Introduction to Direct questionnaire 1

2

3

4

Appendix 3.1

5

6

7

Appendix 3.1

8

9

10

 

 

Appendix 3.1

Direct questionnaire – Question A – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.2

Direct questionnaire – Question B – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.3

Direct questionnaire – Question D – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.4

Direct questionnaire – Question E – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.5

Direct questionnaire – Question F – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.6

Direct questionnaire – Question G – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.7

Direct questionnaire – Question I – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.8

Direct questionnaire – Question J – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.9

Direct questionnaire – Question L – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.1

0

Direct questionnaire – Question M – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.1

1

Direct questionnaire – Question N – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.1

2

Direct questionnaire – Question O – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.1

3

Direct questionnaire – Question P – Respondent: Aged 52, Life expectancy 80, Health state 90

If ‘Yes’

If ‘no’

Continue to card sort

Ap

pen

dix

3.1

4

1

Technical appendices to the results of the Q study

Contents

4.I Technical appendices to the results of the Q study ......................................... 3

4.I.1 Development of the study materials ............................................................. 3

4.I.1.1 Exploration of the topic area.................................................................. 3

4.I.1.2 Development of the statement set ......................................................... 9

4.I.1.3 Development of the interview materials .............................................. 11

4.I.2 Data collection ............................................................................................ 20

4.I.3 Analysis ...................................................................................................... 21

4.I.4 Results ....................................................................................................... 24

4.I.4.1 Study sample ...................................................................................... 24

4.I.4.2 Points of view about the prioritisation of health care ........................... 25

Index of tables

Table 4. I.1 Result from the literature review 5 Table 4. I.2 Theoretical structure for the development of the statement set 10 Table 4. I.3 Statement sample (Q set) 12 Table 4. I.4 Instructions to participants 15 Table 4. I.5 Correlations between factors in consecutive factor solutions (n=294) 27 Table 4. I.7 Factor loadings general public sample (n=294) 29 Table 4. I.8 Explained variance, and defining and associated Q sorts by factor, per country 37 Table 4. I.9 Z- and Rank-scores of statements 38

Index of figures

Figure 4. I.1 Score sheet for ranking statements 17 Figure 4. I.2 Correlations between consecutive factor solutions 26

Appendix 4.I

2

Figure 4. I.3 Idealised Q sort for GP1: “Egalitarianism, entitlement and equality of access” 41 Figure 4. I.4 Idealised Q sort for GP2: “Efficiency, severity and the magnitude of health gains” 42 Figure 4. I.5 Idealised Q sort for GP3: “Fair innings, young people and maximising health benefits” 43 Figure 4. I.6 Idealised Q sort for GP4: “The intrinsic value of life and healthy living” 44 Figure 4. I.7 Idealised Q sort for GP5: “Quality of life above all else” 45

Appendix 4.I

3

4.I Technical appendices to the results of the Q study

4.I.1 Development of the study materials

4.I.1.1 Exploration of the topic area

The first step in conducting a Q methodology study is developing the Q set. This set

of opinion statements forms the actual research instrument and therefore it is

important that it is representative of the subject area of study. In order to produce a

comprehensive overview of characteristics that the general public and decision

makers may possibly take into account in health care priority setting, from which to

select the statements for the Q set, a review of the literature on the social value of a

QALY was conducted by two researchers from Erasmus University Rotterdam in July

and August 2007. Four fairly recent reviews were available at the time and thus used

as a starting point:

Schwappach D. Resource allocation, social values and the QALY: a review of the

debate and the empirical evidence. Health Expectations 2002;5: 210-222

Dolan P, Shaw R, Tsuchiya A, Williams A. QALY maximisation and people’s

preferences: a methodological review of the literature. Health Economics

2005;14: 197-208

Tsuchiya A, Dolan P. The QALY model and individual preferences for health

states and health profiles over time: a systematic review of the literature. Medical

Decision Making 2005;25: 460-467

Smith RD, Richardson J. Can we estimate the ‘social’ value of a QALY?: Four

core issues to resolve. Health Policy 2005;74(1): 77-84

Appendix 4.I

4

In addition, three common databases -EconLit, PubMed and Google Scholar- were

searched for further references. Two researchers retrieved references to articles

published between the years 2001 (continuing where the first review mentioned

above stopped) and 2007 (the time of this review), using the following specific

(combinations of) search terms:

EconLit: age/health/QALY, efficiency/health/QALY, equity/health/QALY,

inequality/health/QALY, priority setting/health, public preferences/QALY/health,

public preferences/QALY, resource allocation/health/QALY, social

value/health/QALY, social values/health, value of life/health

Pubmed: QALY, equity priority setting, inequality QALY, health-care rationing

QALY, methodology QALY, priority setting QALY, public preference QALY,

social value QALY, QALY value, resource allocation QALY, resource allocation

rationing, resource allocation priority setting, resource allocation quality, social

values resource allocation, social values rationing, willingness to pay QALY,

QALY weights, resource allocation rationing survey, resource allocation

rationing statistics, resource allocation rationing numerical data.

Google Scholar: WTP QALY, social value QALY

Together, these sources resulted in a list of 201 references to potentially useful

articles. After selection of unique references to full-length accessible articles in

English language and inspection of the abstracts to confirm relevance, 49 references

were retained and retrieved for further analysis, in addition to the four review articles

mentioned above (see Table 4. I.1).

Appendix 4.I

5

Table 4. I.1 Result from the literature review

1. Abellan-Perpinan, J.M., Pinto-Prades, J.L. Health state after treatment: a reason for discrimination? Health Economics. 1999;8(8): 701-707

2. Baker, R. and Robinson, A. Responses to standard gambles: are preferences ‘well constructed’? Health Economics. 2004;13: 37-48

3. Bleichrodt, H., Doctor, J. and Stolk, E. A nonparametric elicitation of the equity-efficiency trade-off in cost-utility analysis. Journal of Health Economics. 2005;24: 655-678

4. Bowling A. Health care rationing: the public's debate. BMJ. 1996;312(7032): 670-674

5. Brouwer, W.B.F., van Exel, N.J.A. and Stolk, E.A. Acceptability of less than perfect health states. Social Science & Medicine. 2005;60: 237-246

6. Browning, C.J. and Thomas, S.A. Community values and preferences in transplantation organ allocation decisions. Social Science & Medicine. 2001;52: 853-861

7. Bryan, S., Roberts, T. Heginbotham, C. and McCallum, A. QALY-maximisation and public preferences: results from a general population survey. Health Economics. 2002;11: 679-693

8. Choudhry, N., Slaughter, P., Sykora, K., Naylor, CD. Distributional dilemmas in health policy: large benefits for a few or smaller benefits for many? Journal of Health Services Research & Policy. 1997;2(4): 212-216

9. Coast, J., Donovan, J., Litva, A., Eyles, J., Morgan, K., Shepherd, M., Tacchi, J. “If there were a war tomorrow, we’d find the money”: contrasting perspectives on the rationing of health care. Social Science & Medicine. 2002;54: 1839-1851

10. Cookson, R and Dolan, P. Public views on health care rationing: a group discussion study. Health Policy. 1999;49: 63-74

11. Corso, P.S., Hammitt, J.K., Graham, J.D., Dicker, R.C., Goldie, S.J. Assessing preferences for prevention versus treatment using willingness to pay. Medical Decision Making. 2002;22(5): S92-101

12. Costa-Font, J. and Rovira, J. Eliciting preferences for collectively financed health programmes: the ‘willingness to assign’ approach. Applied Economics. 2005;37: 1571-1583

13. Devlin, N. and Parkin, D. Does NICE have a cost-effectiveness threshold and what other factors influence its decisions? A binary choice analysis. Health Economics. 2004;13: 437-452

14. Devlin, N., Appleby, J. and Parkin, D. Patients’ views of explicit rationing: what are the implications for health service decision-making? Journal of Health Services Research & Policy. 2003;8(3): 183-186

15. Dolan, P., Cookson, R., Ferguson, B. Effect of discussion and deliberation on the public’s views of priority setting in health care: focus group study. BMJ. 1999;318: 916-919

16. Dolan, P., Shaw, R. A note on a discussion group study of public preferences regarding priorities in the allocation of donor kidneys. Health Policy. 2004;68: 31-36

Appendix 4.I

6

17. Dolan, P., Tsuchiya, A. Health priorities and public preferences: the relative importance of past health experience and future health prospects. Journal of Health Economics. 2005;24: 703-714

18. Donaldson, C., Birch, S. and Gafni, A. The distribution problem in economic evaluation: income and the valuation of costs and consequences of health care programmes. Health Economics. 2002;11: 55-70

19. Edwards, R.T., Boland, A., Wilkinson, C., Cohen, D., Williams, J. Clinical and lay preferences for the explicit prioritisation of elective waiting lists: survey evidence from Wales. Health Policy. 2003;63: 229-237

20. Gyrd-Hansen, D. Willingness to pay for a QALY. Health Economics. 2003;12: 1049-1060

21. Gyrd-Hansen, D. Willingness to pay for a QALY: theoretical and methodological issues. Pharmacoeconomics. 2005;23(5): 423-432

22. Johri M, Damschroder LJ, Zikmund-Fisher BJ, Ubel PA. The importance of age in allocating health care resources: does intervention-type matter? Health Economics. 2005;14(7): 669-678

23. King, D and Maynard, A. Public opinion and rationing in the UK. Health Policy. 1999;50: 39-53

24. King, J.T., Tsevat, J., Lave, J.R. and Roberts, M.S. Willingness to pay for a Quality-adjusted life year: implications for societal health care resource allocation. Medical Decision Making. 2005;25: 667-677

25. Kuiper, R., Fontani, P. Big life & death survey: medical ethics in nine themes [Grote leven & dood enquête: medische ethiek in negen thema’s; in Dutch]. Quest. 2007;4: 20-28

26. Matschinger, H. and Angermeyer, M.C. The public’s preferences concerning the allocation of financial resources to health care: results from a representative population survey in Germany. European Psychiatry. 2004;19: 478-482

27. Mortimer, D. On the relevance of personal characteristics in setting health care priorities: a comment on Olsen, Richardson, Dolan and Menzel (2003). Social Science & Medicine. 2005;60: 1661-1664

28. Mortimer, D. The value of thinly spread QALYs. Pharmacoeconomics. 2006;24(9): 845-853

29. Mossialos E, King D. Citizens and rationing: analysis of a European survey. Health Policy. 1999;49(1-2): 75-135

30. Nord, E. Severity of illness and priority setting: worrisome lack of discussion of surprising finding. Journal of Health Economics. 2006;25(1): 170-172

31. Nord, E. The trade-off between severity of illness and treatment effect in cost-value analysis of health care. Health Policy. 1993;24(3): 227-38

32. Olsen, J.A. et al. The moral relevance of personal chracteristics in setting health care priorities. Social Science & Medicine. 2003;57: 1163-1172

33. Oudhoff, J.P., Timmermans, D.R., Knol, D.L., Bijnen, A.B., Van der Wal, G. Prioritising patients on surgical waiting lists: a conjoint analysis study on the priority judgements of patients, surgeons, occupational physicians, and general practitioners. Social Science & Medicine. 2007;64(9): 1863-1875

34. Pinto-Prades, J.L., Abellan-Perpinan, J.M. Measuring the health of populations: the veil of ignorance approach. Health Economics. 2005;14(1): 69-82

35. Protiere, C., Donaldson, C., Luchini, S., Moatti, J.P. and Shackley, P. The impact of information on non-health attributes on willingness to pay for multiple health care programmes. Social Science & Medicine. 2004;58: 1257-1269

Appendix 4.I

7

36. Reed Johnson, F. and Backhouse, M. Eliciting stated preferences for health-technology adoption criteria using paired comparisons and recommendation judgments. Value in Health. 2006;9(5): 303-311

37. Robinson, R. Limits to rationality: economics, economists and priority setting. Health Policy. 1999;49: 13-26

38. Rodriguez-Miguez, E. and Pinto-Prades, J-L. Measuring the social importance of concentration or dispersion of individual health benefits. Health Economics. 2002;11: 43-53

39. Schomerus, G., Matschinger, H. and Angermeyer, M.C. Preferences of the public regarding cutbacks in expenditure for patient care. Social Psychiatry and Psychiatric Epidemiology. 2006;41: 369-377

40. Schwappach, D.L.B. Does it matter who you are or what you gain? An experimental study of preferences for resource allocation. Health Economics. 2003;12: 255-267

41. Stolk, E.A., Pickee, S.J., Ament, A.H.J.A. and Busschbach, J.J.V. Equity in health care prioritisation: an empirical inquiry into social value. Health Policy. 2005;74: 343-355

42. Stolk, E.A., van Donselaar, G., Brouwer, W.B.F. and Busschbach, J.J.V. Reconciliation of economic concerns and health policy. Pharmacoeconomics. 2004;22 (17): 1097-1107

43. Tsuchiya, A. QALYS and ageism: philosophical theories and age weighting. Health Economics. 2000;9: 57-68

44. Tsuchiya, A. The Value of health at different ages. Discussion paper 184. York, Centre for health economics, University of York. 2001

45. Tsuchiya, A., Dolan, P. and Shaw, R. Measuring people’s preferences regarding ageism in health: some methodological issues and some fresh evidence. Social Science & Medicine. 2003;57: 687-696

46. Ubel, P.A. How stable are people's preferences for giving priority to severely ill patients? Social Science & Medicine. 1999;49(7): 895-903

47. Ubel, P.A., Spranca, M.D., Dekay, M.L., Hershey, J.C., Asch, D.A. Public preferences for prevention versus cure: What if an ounce of prevention is worth only an ounce of cure? Medical Decision Making. 1998;18: 141-148

48. Werntoft, E., Hallberg, I.R., Edberg, A.K. Older people's reasoning about age-related prioritization in health care. Nursing Ethics. 2007;14(3): 399-412

49. Wilmot, S., Ratcliffe, J. Principles of distributive justice used by members of the general public in the allocation of donor liver grafts for transplantation: a qualitative study. Liver Transplantation. 2003;9(8): 878-880

Appendix 4.I

8

All selected articles were inspected for characteristics that the general public and

decision makers may take into account in health care priority setting and a rough

long-list of such characteristics was compiled. A team of five researchers from

Newcastle University and Erasmus University Rotterdam met in September and

October 2007 to discuss this long-list of characteristics with the aim to structure and

condense it. After an interactive process, consensus was reached on a categorised

list of 23 potentially relevant characteristics for prioritisation of health care divided.1

In order to make sure that this categorized list of statements included all issues of

relevance in each participating country, it was sent to all partners in the EuroVaQ

project. Partners, all health economists, were asked to comment on the theoretical

structure that emerged from the literature review and to assess whether the 23

characteristics reflected the issues entering current debates in their country about

priority setting in health care among the public, decision makers, stakeholder

organisations and in the media. Although no additional characteristics were

proposed, comments were made concerning the level of detail in their formulation,

e.g. ‘severity’ versus ‘severity at baseline’ and ‘severity in the expectation’, and

different possible interpretations in the socio-economic or cultural context of the

countries involved in the study. For instance, accessibility of treatment may relate to

financial barriers at the individual or national level as well as to geographical/physical

barriers (e.g. urban vs. rural areas, or checkpoints and closed borders as in the case

of Palestine). A small number of minor changes were made to the formulation and

wording of statements following partners’ comments. However, since Q methodology

1 Gender and race were excluded ex-ante because legislation in most countries will not allow priority

setting using these characteristics.

Appendix 4.I

9

specifically allows respondents to interpret statements themselves, many of these

more specific issues, if considered relevant by respondents, would be revealed and

addressed during their Q sorts and post-sort comments. The final theoretical

structure for the development of the Q set is shown in Table 4. I.2.

4.I.1.2 Development of the statement set

Based on the comprehensive overview shown in Table 4. I.2 and a Q set from related

work conducted by Newcastle University in the context of the social value of a QALY

(SVQ) project2

, two researchers from Erasmus University Rotterdam drafted a list of

opinion statements with at least two statements reflecting each characteristic, which

resulted in a long-list of 80 statements.

In a number of rounds during November 2007, five researchers from Newcastle

University and Erasmus University Rotterdam worked and commented on the long-

list and condensed it to a comprehensive draft Q set of 37 statements. These

statements were then discussed with a small panel at Erasmus University Rotterdam,

consisting of two health economists not involved with the project and two non-

academics. The panel provided comments on the understandability of the statements

and the clarity of language for a lay sample of respondents. The statement related to

proportional shortfall (‘If people stand to lose a large proportion of their future health

they should be prioritised over those who stand to lose just a small proportion’) was

considered to be too complex as part of a set of 37 statements, and remarks were

2 Baker R, Bateman I, Donaldson C, Jones-Lee M, Lancsar E, Loomes G, et al. Weighting and valuing

quality-adjusted life-years using stated preference methods: preliminary results from the Social Value

of a QALY Project: Health Technology Assessment 2010:14 vol 27.

Appendix 4.I

10

made about the number of statements related to age. In addition, comments about

language led to minor changes in wording of a few statements.

Table 4. I.2 Theoretical structure for the development of the statement set

Category Characteristic

A. Patient 1. Age(ism) / fair innings

2. Socio-economic status

3. Prior health consumption / previous health profile

4. Payment / contribution

5. Having dependent adults or children

6. Equality / all patients equal (no prioritization on patient characteristics)

B. Illness 7. Severity: life threatening vs. mild / stable vs. progressive / chronic

8. Rule of rescue / pain / relief intervention

9. Probable cause / culpability: genetic / congenital, bad luck (could have happened to anyone), avoidable, lifestyle / self-inflicted, medical negligence / fault of others

10. Rarity

C. Treatment 11. Efficiency

12. Waiting lists / waiting time

D. Health effects of treatment

13. Size of the effect

14. Length vs. quality of life

15. Certainty of effect occurring

16. Distribution of fixed health gains / threshold effect

17. Start-point before / end-point after treatment

18. Direction of the effect: health gain / loss avoidance (prevention)

19. Proportional shortfall / prospective health / prognostic difference

20. Capacity to benefit: functioning and capabilities regained after treatment

E. Non-health effects of treatment

21. Social support / family effect / care giving effect

22. Productivity (work)

23. Health effects should be leading

This draft set of 37 statements was presented to the EuroVaQ project team at their

meeting in Barcelona, early December 2007. All partners attending the meeting

(n=26) ranked the statements using the draft interview materials (see below) and

Appendix 4.I

11

commented on the statements using a dedicated response sheet, and all issues

raised as well as those from the panel in Rotterdam were discussed in a dedicated

session. Following the meeting, three statements were removed from the draft set.

There was agreement with the Rotterdam panel that the proportional shortfall

statement would not be effective and general consensus that statements referring to

ageism and maximizing benefits were over-sampled. Furthermore, a number of

statements were reworded for reasons of balance.

The final set of 34 statements, presented in Table 4. I.3, was printed on cards and

randomly assigned a number for purpose of identification.

4.I.1.3 Development of the interview materials

The interview materials were compiled on the basis of available examples from

previous studies by the research team. These materials included instruction for

participants, a score sheet and a post-sort questionnaire.

The instructions to participants describe the purpose of the ranking exercise to

respondents and provide step by step guidance through the ranking exercise, so that

the exercise can be performed individually. Table 4. I.4 shows the instructions as

they were handed out to respondents on a paper sheet during in-person interviews.

The wording was altered slightly when the interview was conducted online.

Appendix 4.I

12

Table 4. I.3 Statement sample (Q set)

Characteristic Statement Number

A. Patient

1. Age(ism) / fair innings Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

14

Younger people should be given priority over older people, because they haven’t had their fair share of health yet.

23

Priority should be given to younger people, because they may benefit from treatment for longer.

26

2. Socio-economic status Access to health care should be based on need, not on geographical, social or economic circumstances.

29

In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

16

3. Prior health consumption / previous health profile

The amount of health care people have had in the past should not influence access to treatments in the future.

34

4. Payment / contribution People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

3

People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively.

24

5. Having dependent adults or children Parents with dependent children should be given priority over similar people without dependents.

31

If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

7

6. Equality / all patients equal (no prioritization on patient characteristics)

Patient characteristics like age, gender or income should play no role in prioritizing between people.

4

B. Illness

7. Severity: life threatening vs. mild / stable vs. progressive / chronic

If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

18

Appendix 4.I

13

Characteristic Statement Number

8. Rule of rescue / pain / relief intervention Rescuing people from a certain death should take priority over all other kinds of health care.

8

9. Probable cause / culpability: genetic / congenital, bad luck (could have happened to anyone), avoidable, lifestyle /self-inflicted, medical negligence/ fault of others

People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

25

Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

21

10. Rarity Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

30

C. Treatment

11. Efficiency When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

15

If a treatment adds one month to the life of a patient and costs 7,500€3 6 , one should consider whether the money could have been better spent on other health care.

12. Waiting lists / waiting time For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

28

D. Health effects of treatment

13. Size of the effect Priority should be given to those treatments that generate the most health. 19

14. Length vs. quality of life It is more important to provide treatments that prolong life than treatments that improve quality of life.

33

There is no sense in saving lives if the quality of those lives will be really bad. 17

15. Certainty of effect occurring If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

2

16. Distribution of fixed health gains / threshold effect

It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month.

20

3 In non-Euro countries, equivalent, rounded values in local currency (d.d. December 2007) were used.

Appendix 4.I

14

Characteristic Statement Number

A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

10

17. Start-point before / end-point after treatment Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

22

Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

11

If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

1

18. Direction of the effect: health gain / loss avoidance (prevention)

It is more important to prevent ill health than it is to cure ill health once it occurs. 27

19. Proportional shortfall / prospective health / prognostic difference

Not included

20. Capacity to benefit: functioning and capabilities regained after treatment

People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

32

E. Non-health effects of treatment

21. Social support / family effect / care giving effect Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

9

People who depend heavily on members of their family or neighbours for care should be treated with priority.

13

22. Productivity (work) People who are in paid work and so contribute financially to society should be prioritized over people who do not work.

5

23. Health effects should be leading Doctors should be the ones to judge which patients get priority on the basis of their medical expertise.

12

Appendix 4.I

15

Table 4. I.4 Instructions to participants

INSTRUCTIONS

These instructions will guide you through the study step by step. Please read through each step

completely before you start carrying it out, and please finish each step completely before proceeding

to the next one.

This study is about decision making in health care. Even though spending on health varies between

countries, all countries are faced with the same problem. That is, the health care budget is never

sufficient to do everything that could possibly be done. Because of this, choices must be made about

which health services and treatments to provide and, therefore, which not to provide. There is a lot of

debate about how decisions like these should be made. We are interested in your views about how

health care decisions should be made in your country. The statements on the 34 numbered cards are

things people have said about how health care decisions should be made. Later on we will ask you to

what extent you agree with these statements.

1. Place the large score sheet in front of you on a table. The 34 numbered cards contain

statements about how health care decisions should be made in your country. This study is

about people’s individual opinions; there are no right or wrong answers. The numbers on the

cards (from 1 to 34) are to help you to complete the response sheet and apart from that do not

have any meaning.

2. Read through the 34 statements carefully and at the same time split them up into three piles: a

pile for statements with which you agree (and place them to your right), a pile for statements

with which you disagree (and place them to your left), and a pile for statements with which you

neither agree nor disagree, do not consider relevant or are unclear to you (and place them in the

middle).

3. Take the pile containing the statements you agree with (to your right) and read them through

once again. Select the two statements which you AGREE WITH MOST and place them in the

extreme right column of the large score sheet, below the “9”. It does not matter which of them

you place at the top or at the bottom. Next, from the remaining pile select the three statements

which you now AGREE WITH MOST and place them in the three spaces below the “8”.

Proceed until all statements you agree with have been placed on the score sheet.

4. Take the pile containing the statements you disagree with (to your left) and read them through

once again. Select the two statements which you DISAGREE WITH MOST and place them in

Appendix 4.I

16

the extreme left column of the score sheet, below the “1”. It does not matter which of them you

place at the top or at the bottom. Proceed until all statements you disagree with have been

placed on the score sheet.

5. Finally, take the remaining pile and read through these statements once again too. Place the

cards in the remaining spaces on the score sheet, just like you feel it should be done.

6. When you have finished placing the cards on the score sheet, read them all through as a final

check and change positions if you feel like it.

7. When you are completely ready, please copy the numbers on the cards onto the response

sheet, exactly like they are on the large score sheet.

8. Please complete the remaining questions on the response sheet.

You have finished the study.

Thank you very much for participating!

The score sheet presents respondents with a suggested distribution for the Q sorting

task. As Figure 4. I.1 shows, it provides a continuum for ranking the statements,

typically a sentiment like importance, agreement or preference, and usually takes the

form of a quasi-normal distribution, with between 7 and 11 columns, which is flatter

when respondents are expected to have strong, well articulated opinions on the topic

at issue. In the present study, respondents were asked to rank the statements

according to agreement (see also Table 4. I.4) on a 9-column score sheet ranging

from ‘disagree most’ to ‘agree most’ (Figure 4. I.1).

Appendix 4.I

17

Figure 4. I.1 Score sheet for ranking statements

DISAGREE

MOSTAGREE

MOST

1 765432 98

Appendix 4.I

18

The post-sort questionnaire asked respondents about some basic socio-demographic

characteristics, primarily for the purpose of monitoring the sampling strategy (see

below).

The draft interview materials were pilot tested in two rounds. The EuroVaQ project

team, largely academic health economists, tested the study materials (i.e. statement

set and interview materials) by performing the Q sorting task during their meeting in

Barcelona early December 2007. Following this meeting, additional pilot tests were

organised with academic health economists at Erasmus University Rotterdam and

convenience samples of non-academics in the Netherlands, the UK and members of

the international Q methodology association (ISSSS; www.qmethod.org).

The paper and web materials were then pilot tested in study samples in the UK and

the Netherlands. In the UK, 30 people attended one of five focus group meetings

organized by Newcastle University. During these meetings, participants performed

the Q sort individually, followed by group discussion about the topic, individuals’

explanations for their ranking of statements and the method of study. The focus

group interviews proved successful, no further changes to the materials were

necessary and these data are retained in the main analysis. In the Netherlands, 30

individual interviews were held by Erasmus University Rotterdam. Participants

performed the Q sort individually and explained their ranking of the statements. Next,

participants were asked to phrase in their own words three to five statements

selected by the interviewer, mostly statements that were placed in the centre of the

score sheet (columns 3 to 7). The purpose of this was to get an idea of how the

statements were interpreted by respondents, and whether this generally coincided

Appendix 4.I

19

with what was intended from the theoretical structure. The results of both parts of the

interview confirmed that the statements and the interview materials were

understandable and clear, and these data are also included.

Finally, the 30 people who participated in the pilot test in the Netherlands were

invited to conduct the same Q sort online4 as well, two weeks later, in order to get

feedback on the web application used5

and the comparability of materials collected

through in person and web interviews. Twenty people responded to the invitation and

results were compared between the in-person and online Q sorts (see appendix 4.III

for an explanation of methods of comparison). First, correlations were computed

between the pair of Q sorts performed by each respondent. Setting apart two

respondents whose Q sorts, for whatever reason, appeared unrelated (correlation

<0.40), the mean correlation between the in-person and online Q sorts was 0.74

(range 0.44 to 0.93), which can be classified as fair to good. Analysis of the pooled

data showed that the Q sorts of 16 of the 18 respondents loaded on the same factor.

Finally, comparison of two-factor solutions for each data-set (n=18) showed that the

correlations between the corresponding factors were 0.86 and 0.84, which indicates

that, despite some individual variation, the results from the in-person and online Q

sorts were very similar. All in all, the outcome of the comparisons was considered

satisfactory and respondents gave a few suggestions to improve the lay-out and

clarity of the web interview, which were incorporated.

4 See: www.yourviewonhealth.com/healthdecisions/index.html

5 FlashQ (www.hackert.biz/flashq/home/).

Appendix 4.I

20

The results of the various pilot tests were presented and discussed with the partners

in the EuroVaQ project during their meeting early May 2008 in Krakow.

The final step in the preparation of the study materials was the translation of the

statement set, the instructions and the post-sort questionnaire into the languages of

all the participating countries. Dedicated forms were developed containing all the text

to be translated. Partners from all participating countries were asked to translate the

materials into their language, have it back translated into English, and to finalise the

translation by comparing the original and the back-translated versions.

4.I.2 Data collection

The partners in the EuroVaQ project were asked to conduct interviews with purposive

samples of members of the general public and decision makers in the health care

sector and were offered the possibility to conduct the interviews in-person or online.

Partners conducting in person interviews were asked to recruit 20 respondents, those

conducting web interviews 25 useable responses (see below).

The aim of the sampling strategy was to recruit respondents who were likely to hold

different points of view to one another regarding health care priority setting. Based on

the literature review discussed in the previous section of this appendix, it was

expected that age, gender, socio-economic status and health status would be the

most relevant sampling characteristics. Because of the limited size of the sample

required, partners were asked to recruit approximately equal numbers of respondents

in six groups stratified by two of those characteristics, i.e. gender (female; male) and

Appendix 4.I

21

age (younger than 25 and no children; 25 to 65; 65 and older). Furthermore, in as far

as possible, the sample should represent respondents with different health and

socio-economic status although this information is less readily accessible, ex ante.

Age, gender, health and socio-economic status (i.e. level of education; self-assessed

wealth) were recorded during the post-sort interview in order to monitor the sampling

strategy.

Most interviews were conducted between September and November 2008.

4.I.3 Analysis

The data collected in 10 countries in samples of the general public and decision

makers was analyzed at country and aggregate European level6

. The purpose of

each analysis was to identify the views presented by groups of respondents who

ranked the 34 opinion statements in a similar way, under the assumption that this

indicates similarity in point of view about prioritisation of health care. The results of

general public and decision makers were compared at the European level (see

appendix 4.III).

The data were recorded to represent rank scores between -4 for statements placed

in column1 of the score sheet () and +4 for statements placed in column 9 and

6 In line with the rest of the report the aggregated findings are, for reasons of brevity, referred to as

“European” despite the inclusion of Palestine and the exclusion of many European countries.

Appendix 4.I

22

analyzed using PQMethod 2.11, a dedicated software package for Q factor analysis7

.

In each sample, first the correlation matrix of Q sorts is computed. In other words,

contrary to most conventional clustering methods, in Q analysis persons are

correlated instead of tests and the correlation matrix therefore shows the similarity in

the ranking of statements between participants. Next, the correlation matrix was

subject to factor analysis in order to identify groups of respondents with mutually high

correlation coefficients, using common factor analysis techniques (i.e. centroid factor

analysis followed by varimax rotation). Then, the maximum factor structure supported

by the data was identified and the statistical features of all supported factor solutions

were inspected.

For the purposes of factor interpretation, an idealised ranking of the statements was

computed for each factor. This is the normalised, weighted average ranking of the

statements in that factor, based on the Q sorts defining the factor with their factor

loading (correlation coefficient between factor and Q sort) as the weight. As is

common, a Q sort was considered to be a defining variable for a factor if the Q sort

correlated statistically significantly (p<.05) with that factor8 and, in order to disregard

confounded Q sorts, the factor explained more than half of the common variance9

7 Software and a manual can be downloaded from

.

Weighted rank scores were computed for all statements and each factor by

averaging the product of rank score and correlation coefficient over defining

http://www.lrz.de/~schmolck/qmethod/.

8 I.e., the correlation coefficient of the Q sort with the factor should exceed 0.336, i.e. the ratio of the

multiplier for the 5%-level statistical significance divided by the square root of the number of

statements (= 341.96 ).

9 I.e., the square of the correlation coefficient between Q sort and factor should exceed the sum of the

squares of correlation coefficients of the Q sort with all other factors.

Appendix 4.I

23

variables. The idealised ranking of statements for a factor is determined by ordering

the weighted rank scores of the statements from high to low. This idealised Q-sort

(see Table 4. I.8 and Figure 4. I.3 for examples) represents the way in which a

person with a correlation coefficient of 1 with that factor would have ranked the

statements. Finally, the weighted rank scores of statements for each factor were

normalized, with a mean of 0 and a standard deviation of 1 (i.e. Z scores), in order to

allow for comparison of scores between factors.

The idealised Q sort of each factor from each factor solution was then interpreted,

focussing on the characterising and distinguishing statements. A statement was

considered to be characterising for a factor if it was positioned in the outer two

columns of the idealised Q-sort (i.e., if it had a rank score of -4, -3, +3 or +4; see

Table 4. I.8 and Figure 4. I.3 for examples). A statement was considered to be

distinguishing for a factor if its score in the idealised Q-sort of that factor was

statistically significantly different from its score in all other factors. Attention was also

paid to consensus statements, that is, those statements for which the score was not

statistically significantly different between any pair of factors.

In each of the 12 analyses (i.e., ten general public country-level, one general public

European level and one decision makers European level) the factor solution that was

considered to present the most clearly interpretable account for each factor was

selected and described, using both the statistical results and the explanations of the

respondents defining a factor in order to confirm and clarify the interpretation of the

statistical results. In chapter 4, where the results of the European level analyses are

Appendix 4.I

24

presented, some explanations are cited in the results section to illustrate and support

the description of that particular viewpoint.

4.I.4 Results

4.I.4.1 Study sample

A total of 329 members of the general public were interviewed, resulting in 294

(89.4%) responses considered useful for analysis. A number of online respondents

(33; 10.0%) were excluded because their completion time was very quick, indicating

that they could not have given the exercise sufficient time to give meaningful

responses10

and two in-person respondents (0.6%) were excluded because they

revealed a lack of comprehension of the exercise. The majority of participants

completed the interview online (82%). Participants were fairly well distributed over the

ten countries, varying from 20 (7%) in Palestine to 47 (16%) in Hungary (see Table

4.1 in Chapter 4).

All categories in the sampling matrix were covered satisfactorily. Participants were

40 years of age on average, ranging from 17 to 86 years, 59% were female and 52%

had children. The level of education was fairly high (i.e. 10% low; 48% middle; 42%

high) and 62% was employed (i.e. 32% public / 30% private sector; 3% unemployed;

10 Defined as taking less than 10 minutes to read through and rank-order the 34 statements. This time

limit was determined by asking four staff members of partner institutes to conduct the exercise

conscientiously, but as quickly as possible.

Appendix 4.I

25

4% housewife; 12% retired; 19% student). Participants rated their health as good11

and their wealth as similar to the average person in their country12

.

4.I.4.2 Points of view about the prioritisation of health care

By-person factor analysis showed that the 294 Q sorts supported a maximum

structure of five factors. The statistical features of each factor solution were examined

in detail. Table 4. I.5 and Figure 4. I.2 show the correlations between consecutive

solutions, from a one factor solution to the maximum supported structure of five

factors.

The correlations show that two factors remain robust. Factor 1 in the two factor

solution (2/1) has very high correlations with the first factor in the three, four and five

factor solutions (0.99, 0.98 and 0.95 respectively; see Table 4. I.5), which indicates

that the content of this factor –shown as 2/1, 3/1, 4/1 & 5/1 in Figure 4. I.1- is virtually

the same in each factor solution. Likewise, factor 2 in the two factor solution (2/2)

has very high correlations with the last factor in the three, four and five factor

solutions (0.96, 0.94 and 0.93 respectively; 2/2, 3/3, 4/4 and 5/5 in Figure 4.I.2).

11 The question to respondents was: “Please give your health a grade. Place an ‘X’ on the line below,

on the spot that best reflects how healthy you feel in general. The ‘0’ represents the worst health state

you can imagine. The ‘10’ represents the best health state you can imagine.” Respondents rated their

health on a visual analogue scale ranging from 0 to 10, with endpoint labels ‘worst conceivable health’

and ‘best conceivable health’. The average score was 8.1 (range 0-10), 9% scored 5 or lower.

12 The question to respondents was: “How wealthy would you say you are as compared to the average

person in your country?” The answer categories were ‘much less wealthy’; ‘somewhat less wealthy’;

‘about as wealthy’; ‘somewhat wealthier’ and ‘much wealthier’, scored as 1 to 5, respectively. The

average score was 3.2 (range 1 to 5) and the extreme categories were both selected by 4%.

Appendix 4.I

26

Figure 4. I.2 Correlations between consecutive factor solutions 13

13 A factor diagram is a simplified visual representation of the hierarchical factor structure in a data set

(Goldberg, 2006). Each row of the factor diagram presents a consecutive factor structure for the data

(from separate analyses of the data), here from the one factor structure at the top to the five factor

structure at the bottom. The boxes in each row represent the factors, the width of the boxes their

percentage explained variance. The arrows between boxes indicate the most important correlations

(>.70) between factors in consecutive factor structures (i.e., between-structures comparison), and the

numbers next to the arrows the corresponding correlation coefficient.

Appendix 4.I

27

Table 4. I.5 Correlations between factors in consecutive factor solutions (n=294)

Factors 1/1 2/1 2/2 3/1 3/2 3/3 4/1 4/2 4/3 4/4 5/1 5/2 5/3 5/4 5/5

1/1 0.94 0.73 0.91 0.79 0.67 0.88 0.83 0.66 0.64 0.88 0.86 0.56 0.62 0.66

2/1 0.51 0.99 0.64 0.46 0.98 0.73 0.51 0.42 0.95 0.91 0.43 0.45 0.45

2/2 0.46 0.81 0.96 0.42 0.67 0.75 0.94 0.48 0.45 0.61 0.74 0.93

3/1 0.57 0.42 0.99 0.68 0.45 0.39 0.95 0.90 0.38 0.40 0.42

3/2 0.67 0.56 0.79 0.93 0.61 0.56 0.63 0.64 0.89 0.61

3/3 0.38 0.56 0.63 1.00 0.46 0.37 0.50 0.62 0.97

4/1 0.61 0.47 0.35 0.96 0.84 0.33 0.41 0.37

4/2 0.55 0.51 0.61 0.76 0.83 0.50 0.52

4/3 0.59 0.47 0.50 0.38 0.99 0.60

4/4 0.43 0.34 0.44 0.59 0.98

5/1 0.76 0.39 0.41 0.42

5/2 0.37 0.46 0.41

5/3 0.32 0.40

5/4 0.59

% explained variance 32 24 16 20 11 13 19 9 8 12 14 12 7 8 11

Cumulative % explained variance 32 40 44 48 52

Note: Correlations shown in Figure 4. I.2 (>0.70) in bold. Factors 5/1, 5/2, 5/3 5/4 and 5/5 correspond with ponts of view GP1, GP2, GP3, GP4 and GP5 respectively in Chapter 4.

Appendix 4.I

28

Next, a judgement was made that the accounts represented by the factors shown as

5/2, 5/3 and 5/4 in Figure 4. I.2 were sufficiently coherent, theoretically interesting,

and consistent with the comments respondents provided while completing their Q

sort. Therefore, five factors were retained as the final solution for the Q sorts

obtained from the general public across the ten participating countries. The five

factors together explained 52% of the total variance in the Q sorts, ranging from 7 to

14% (see Table 4. I.5). Full descriptions of the factors are presented in chapter 4 (as

public points of view GP1, GP2, GP3, GP4 and GP5 respectively) and are not

repeated here, but further statistical information is provided regarding the factor

solution that was selected.

Table 4. I.6 contains the full set of factor loadings, for each of 294 respondents

across each of the five factors, with defining sorts (explained in section 4.I.3)

highlighted by asterisk. The first column (labelled Q sort) lists respondents’ unique

identifiers, the first two letters of which represent a country code.

In Table 4. I.7 frequencies of defining and associated Q sorts per country are shown.

Associated Q sorts are those with statistically significant (p<.05) factor loadings.

Summarising the information in this table, the frequencies show that, for some

factors, there are no defining Q sorts for certain countries. Factor 4, for example is a

strong account in Hungary and has a number of significant, but not many defining, Q

sorts in other countries. Figure 4. I.3 to Figure 4. I.7 present the idealised Q sorts for

each of the five factors and Table 4. I.8 contains the Z- and rank scores on which

they are based.

Appendix 4.I

29

Table 4. I.6 Factor loadings general public sample (n=294)

Q sort Factor

5/1 5/2 5/3 5/4 5/5

1 dk_id1 0.37 0.20 0.11 0.09 0.73 *

2 dk_id2 0.14 0.50 0.10 0.05 0.52

3 dk_id3 0.24 0.57 * 0.26 0.24 0.38

4 dk_id4 0.16 0.50 0.29 0.32 0.27

5 dk_id5 -0.10 -0.21 0.26 0.39 0.57 *

6 dk_id6 -0.10 0.05 0.23 0.51 0.52

7 dk_id7 -0.26 0.13 0.36 0.18 0.62 *

8 dk_id8 0.12 -0.13 0.27 0.35 0.24

9 dk_id9 0.44 0.43 0.31 0.08 0.52

10 dk_id10 0.18 0.09 0.59 * 0.15 0.28

11 dk_id11 0.06 -0.12 0.14 0.42 0.61 *

12 dk_id12 0.11 0.27 -0.07 0.12 0.48 *

13 dk_id13 0.25 0.50 0.10 0.26 0.45

14 dk_id14 0.03 0.16 0.31 0.04 0.64 *

15 dk_id15 0.26 -0.16 0.35 0.14 0.57 *

16 dk_id17 0.18 -0.04 0.24 0.12 0.56 *

17 dk_id18 0.64 * 0.25 0.36 0.25 0.27

18 dk_id19 0.19 0.51 * -0.02 -0.06 0.40

19 dk_id20 0.05 -0.28 0.12 0.33 0.65 *

20 dk_id22 0.75 * 0.43 0.14 -0.09 0.02

21 dk_id23 0.46 0.48 0.04 0.15 0.46

22 dk_id24 0.21 0.45 * 0.02 -0.10 0.27

23 dk_id25 -0.18 -0.18 0.09 0.31 0.07

24 dk_id26 -0.14 0.08 0.17 0.55 * 0.35

25 dk_id27 0.48 0.50 0.40 0.14 0.03

26 fr_id2 0.58 * 0.38 -0.05 0.05 0.17

27 fr_id3 0.58 * 0.14 0.20 0.42 0.03

28 fr_id4 0.61 * 0.56 0.07 0.20 0.02

29 fr_id5 0.16 0.02 -0.21 0.26 0.65 *

30 fr_id6 0.54 * -0.02 -0.03 0.03 0.13

31 fr_id7 0.49 0.51 * 0.12 0.04 0.06

32 fr_id9 0.26 0.51 * 0.20 0.02 0.18

33 fr_id10 0.57 0.38 0.30 0.22 0.32

34 fr_id11 0.69 * 0.36 0.22 0.00 0.27

35 fr_id12 0.29 0.29 0.28 0.50 0.34

36 fr_id13 0.19 0.33 0.10 0.21 0.65 *

Appendix 4.I

30

Q sort Factor

5/1 5/2 5/3 5/4 5/5

37 fr_id14 0.66 * 0.27 0.27 -0.10 0.15

38 fr_id16 0.53 * -0.10 0.06 0.18 0.08

39 fr_id17 0.50 0.23 0.20 -0.05 0.62 *

40 fr_id18 0.62 * 0.34 0.19 -0.08 0.26

41 fr_id19 0.44 0.43 0.42 0.25 0.07

42 fr_id20 0.64 * 0.48 0.30 0.20 0.00

43 fr_id21 0.02 0.14 -0.13 -0.11 0.70 *

44 fr_id22 0.73 * 0.17 0.41 -0.07 0.29

45 fr_id23 0.07 -0.21 0.25 0.25 0.65 *

46 fr_id24 0.45 * 0.10 0.21 0.16 0.25

47 fr_id26 0.46 * 0.42 -0.07 0.08 0.12

48 fr_id28 0.03 0.11 -0.01 0.10 0.49 *

49 fr_id29 0.56 * 0.23 0.32 0.18 0.15

50 fr_id30 0.65 * 0.39 0.48 0.09 -0.03

51 fr_id31 0.29 0.08 0.47 0.11 0.39

52 fr_id32 0.70 * 0.42 0.08 -0.06 0.01

53 fr_id33 0.50 * 0.21 0.24 0.17 0.30

54 hu_id1 0.52 0.10 0.26 0.60 * 0.07

55 hu_id2 0.36 0.65 * 0.23 0.07 -0.10

56 hu_id3 0.31 0.46 0.06 0.22 0.36

57 hu_id4 0.45 0.40 -0.19 0.30 0.40

58 hu_id5 0.42 0.27 0.08 -0.03 0.52 *

59 hu_id6 0.14 0.11 -0.11 0.64 * 0.43

60 hu_id7 0.17 0.24 0.33 0.03 0.17

61 hu_id9 0.24 0.12 -0.24 0.47 0.54

62 hu_id11 0.45 0.69 * 0.04 0.11 0.35

63 hu_id13 0.30 0.30 0.27 0.32 0.03

64 hu_id15 0.47 0.19 0.04 0.59 * 0.27

65 hu_id16 0.37 * 0.13 -0.02 0.22 0.17

66 hu_id18 0.09 0.13 -0.47 * 0.42 0.15

67 hu_id20 0.54 * 0.19 0.36 0.33 0.11

68 hu_id21 0.07 0.67 * 0.05 0.13 -0.01

69 hu_id22 0.15 -0.01 0.06 0.74 * 0.25

70 hu_id24 0.12 0.03 -0.33 0.46 * -0.08

71 hu_id25 0.22 0.31 0.05 0.37 0.03

72 hu_id27 0.06 0.48 * 0.29 0.20 0.29

73 hu_id28 0.47 * 0.20 0.17 0.36 -0.07

Appendix 4.I

31

Q sort Factor

5/1 5/2 5/3 5/4 5/5

74 hu_id29 0.64 * 0.54 0.12 -0.02 -0.04

75 hu_id30 0.13 0.35 * 0.15 -0.03 0.07

76 hu_id31 0.17 0.33 -0.04 0.59 * -0.08

77 hu_id33 0.36 0.17 0.09 0.78 * 0.07

78 hu_id34 0.30 0.50 * 0.23 0.10 0.09

79 hu_id35 0.44 0.33 0.33 0.41 0.32

80 hu_id36 0.56 0.44 0.36 0.16 0.30

81 hu_id37 0.21 0.29 -0.23 0.51 0.30

82 hu_id38 0.57 * 0.48 -0.13 0.21 -0.10

83 hu_id39 0.10 0.51 * 0.07 0.28 0.29

84 hu_id40 0.47 0.62 0.26 0.21 0.36

85 hu_id41 -0.18 -0.29 0.00 0.63 * 0.08

86 hu_id42 0.32 0.51 * 0.00 0.32 0.09

87 hu_id43 0.03 -0.08 0.29 0.59 * 0.15

88 hu_id45 0.20 0.43 -0.03 -0.29 -0.30

89 hu_id46 -0.17 0.17 0.08 0.68 * 0.02

90 hu_id47 0.53 * 0.41 -0.01 0.07 -0.02

91 hu_id48 0.13 0.04 0.14 0.74 * 0.22

92 hu_id49 -0.12 0.28 0.22 0.55 * 0.08

93 hu_id50 0.31 0.66 * 0.03 0.41 -0.04

94 hu_id51 0.50 0.58 * 0.23 0.08 0.02

95 hu_id52 -0.08 0.09 0.06 0.66 * 0.05

96 hu_id53 0.47 0.50 0.04 0.35 -0.01

97 hu_id55 0.36 0.26 -0.05 0.57 * -0.01

98 hu_id56 -0.09 0.56 * 0.26 0.33 0.19

99 hu_id57 0.09 0.14 0.43 0.53 0.32

100 hu_id58 0.45 0.49 0.37 0.26 0.16

101 nl_id1 0.41 * 0.22 0.16 -0.17 0.14

102 nl_id2 0.40 0.36 0.01 0.31 0.14

103 nl_id3 0.23 0.23 -0.02 0.05 0.61 *

104 nl_id4 0.25 0.12 -0.06 0.36 0.55 *

105 nl_id5 0.39 -0.05 0.28 0.12 0.45

106 nl_id6 0.10 0.29 0.25 0.18 0.47 *

107 nl_id7 0.55 * 0.14 0.17 -0.07 0.33

108 nl_id8 0.36 0.34 0.18 -0.20 0.37

109 nl_id9 0.40 * 0.13 0.01 -0.25 0.06

110 nl_id10 0.54 0.09 0.11 0.29 0.54

Appendix 4.I

32

Q sort Factor

5/1 5/2 5/3 5/4 5/5

111 nl_id11 0.35 0.29 0.02 0.22 0.40

112 nl_id12 0.34 0.56 * 0.05 -0.06 0.08

113 nl_id13 0.31 0.33 0.28 0.03 0.44

114 nl_id14 0.56 * 0.18 0.02 0.08 0.34

115 nl_id15 0.55 0.34 0.28 0.40 0.27

116 nl_id16 -0.23 -0.22 -0.10 0.30 0.75 *

117 nl_id17 0.08 0.19 -0.15 0.24 0.68 *

118 nl_id18 0.20 0.56 * -0.18 0.31 0.17

119 nl_id19 0.72 * 0.38 0.12 0.07 0.18

120 nl_id20 0.51 0.12 0.33 -0.26 0.46

121 nl_id21 0.30 0.16 0.05 -0.23 0.05

122 nl_id22 0.65 * 0.05 -0.07 0.09 0.32

123 nl_id23 0.24 0.01 0.13 0.20 0.70 *

124 nl_id24 0.45 0.52 0.02 0.09 0.62

125 nl_id25 0.64 * 0.14 0.18 0.10 0.40

126 nl_id26 -0.04 0.39 * -0.14 -0.03 0.09

127 nl_id27 0.34 0.58 * 0.31 0.17 0.21

128 nl_id28 0.51 * 0.26 0.28 -0.03 0.30

129 nl_id29 0.67 * 0.01 0.23 -0.17 0.32

130 nl_id30 0.71 * 0.04 0.18 0.31 0.05

131 no_id1 0.34 0.23 -0.06 0.46 0.42

132 no_id2 0.23 0.52 0.08 0.40 0.35

133 no_id3 0.12 -0.03 0.27 0.59 * 0.33

134 no_id4 -0.19 -0.04 0.26 0.31 0.39

135 no_id5 0.00 0.19 0.16 0.49 * 0.33

136 no_id6 -0.02 0.20 0.60 * 0.30 0.14

137 no_id7 0.17 0.49 * 0.23 0.13 -0.01

138 no_id8 -0.08 0.46 0.40 0.47 0.14

139 no_id9 0.47 0.52 0.11 0.21 0.30

140 no_id10 0.24 0.05 0.72 * -0.02 0.28

141 no_id11 0.24 0.00 0.58 * 0.33 -0.03

142 no_id12 0.20 -0.09 0.49 -0.02 0.68 *

143 no_id13 -0.08 0.13 0.66 * 0.00 0.21

144 no_id14 -0.25 -0.15 0.48 0.47 0.43

145 no_id15 0.49 -0.13 0.40 0.30 0.18

146 no_id16 0.22 0.32 0.54 -0.10 0.50

147 no_id17 0.66 * 0.47 0.17 -0.03 0.29

Appendix 4.I

33

Q sort Factor

5/1 5/2 5/3 5/4 5/5

148 no_id18 0.19 0.30 0.06 -0.07 0.61 *

149 no_id19 0.31 -0.03 0.28 0.13 0.02

150 no_id20 0.13 0.23 0.53 0.50 0.17

151 no_id23 0.31 0.47 0.42 0.03 0.15

152 no_id24 0.24 0.17 0.53 0.13 0.43

153 no_id25 0.10 0.05 0.67 * -0.07 0.04

154 pa_id1 0.30 0.07 0.11 0.13 0.46 *

155 pa_id2 0.14 0.09 0.38 * 0.23 0.00

156 pa_id3 0.57 * 0.44 -0.08 0.01 -0.03

157 pa_id4 -0.04 -0.23 0.39 * 0.10 0.28

158 pa_id5 0.42 0.37 0.01 0.13 -0.17

159 pa_id6 0.50 * 0.15 0.35 0.15 0.14

160 pa_id7 0.48 * 0.18 0.24 0.26 -0.11

161 pa_id8 0.04 0.38 * 0.24 0.05 -0.04

162 pa_id9 0.44 * -0.02 0.02 0.16 0.04

163 pa_id10 -0.19 0.29 0.53 * 0.14 0.01

164 pa_id11 0.49 * 0.23 0.21 0.07 0.28

165 pa_id12 0.58 * 0.42 0.15 0.30 0.06

166 pa_id13 0.55 * 0.17 -0.13 0.14 0.13

167 pa_id14 0.26 0.05 0.25 0.30 -0.07

168 pa_id15 0.40 0.51 * 0.24 0.06 0.17

169 pa_id16 0.05 0.07 0.23 -0.02 0.44 *

170 pa_id17 0.21 0.13 0.13 0.24 0.13

171 pa_id18 0.57 * 0.16 -0.20 0.06 0.02

172 pa_id19 0.52 0.61 * -0.05 0.00 -0.22

173 pa_id20 0.34 0.31 0.03 -0.19 0.16

174 po_id2 0.40 -0.27 0.20 0.23 0.41

175 po_id3 0.23 0.57 * 0.14 -0.25 0.15

176 po_id4 0.55 0.48 0.39 0.21 0.31

177 po_id5 0.35 0.40 0.31 0.11 0.24

178 po_id6 0.41 0.44 -0.02 0.19 0.29

179 po_id7 0.19 0.29 -0.09 -0.10 0.51 *

180 po_id8 0.16 0.48 * 0.02 0.09 0.03

181 po_id9 0.46 0.61 * -0.05 0.03 -0.12

182 po_id10 0.10 0.44 * 0.31 0.28 -0.02

183 po_id11 -0.18 0.13 0.14 0.36 0.59 *

184 po_id12 0.33 0.38 0.11 0.54 0.24

Appendix 4.I

34

Q sort Factor

5/1 5/2 5/3 5/4 5/5

185 po_id13 0.43 0.36 0.09 0.33 0.17

186 po_id14 -0.06 0.34 * 0.09 0.08 -0.03

187 po_id15 -0.06 0.26 0.14 0.30 0.04

188 po_id16 0.26 0.61 * 0.03 0.20 -0.17

189 po_id17 0.48 0.43 -0.08 0.41 0.28

190 po_id18 0.33 0.46 0.02 0.46 0.35

191 po_id19 0.18 0.39 0.19 0.40 0.01

192 po_id21 0.11 0.44 0.13 0.61 * 0.00

193 po_id22 0.28 0.64 * 0.32 0.00 0.20

194 po_id23 0.18 0.16 0.11 0.58 * 0.27

195 po_id24 0.52 * -0.02 0.44 0.16 0.05

196 po_id25 0.37 * 0.00 -0.01 -0.18 0.17

197 po_id26 0.40 * 0.27 -0.02 -0.13 -0.02

198 po_id27 0.50 0.03 0.32 0.50 0.05

199 po_id28 0.21 -0.02 -0.11 0.11 0.30

200 po_id29 0.57 * 0.01 0.00 0.05 0.05

201 po_id31 -0.14 0.30 0.20 0.12 0.59 *

202 po_id32 0.15 0.27 0.05 0.48 * 0.29

203 se_id1 0.59 0.55 0.17 -0.05 0.17

204 se_id2 0.23 0.43 0.56 * 0.11 0.22

205 se_id4 0.15 0.49 0.38 0.20 0.42

206 se_id5 0.04 0.17 0.56 * 0.42 0.18

207 se_id6 0.27 -0.08 0.35 0.07 0.62 *

208 se_id8 0.38 0.53 0.49 0.18 0.16

209 se_id10 0.53 * 0.32 0.06 0.00 0.15

210 se_id11 0.27 0.66 * 0.20 0.17 0.09

211 se_id12 0.34 0.09 0.43 0.33 0.51

212 se_id13 0.46 * 0.17 0.15 0.24 -0.07

213 se_id14 0.40 0.12 0.40 0.32 -0.05

214 se_id16 0.18 0.18 0.48 * 0.13 0.01

215 se_id18 0.01 0.28 0.76 * 0.07 0.28

216 se_id19 0.02 -0.13 0.15 0.30 0.26

217 se_id20 0.01 0.01 0.28 0.40 0.35

218 se_id21 0.51 0.47 0.49 -0.03 0.19

219 se_id22 0.05 0.49 * 0.31 0.17 0.21

220 se_id23 0.25 0.42 * 0.08 0.04 0.28

221 se_id24 0.20 0.40 -0.12 0.29 0.50

Appendix 4.I

35

Q sort Factor

5/1 5/2 5/3 5/4 5/5

222 se_id26 0.40 0.19 0.34 0.16 0.55

223 se_id27 0.13 0.28 0.45 0.07 0.60 *

224 se_id30 0.59 * 0.13 0.22 0.11 0.07

225 se_id31 0.25 0.44 0.33 0.45 0.22

226 sp_id1 0.50 * 0.23 0.04 0.14 0.38

227 sp_id2 0.29 0.44 0.33 0.06 0.59

228 sp_id3 0.54 * 0.34 0.06 0.03 -0.03

229 sp_id4 0.57 * 0.35 -0.11 -0.10 0.02

230 sp_id5 0.20 -0.31 0.16 0.34 0.50

231 sp_id7 0.64 * 0.44 0.22 -0.15 0.17

232 sp_id8 0.54 0.42 -0.09 -0.16 0.40

233 sp_id9 0.41 0.42 0.17 0.38 0.24

234 sp_id10 0.15 -0.10 0.30 -0.09 0.63 *

235 sp_id11 0.32 0.22 0.00 0.09 0.43 *

236 sp_id12 0.36 * 0.16 0.20 0.13 0.01

237 sp_id13 0.46 -0.11 0.29 0.05 0.44

238 sp_id14 0.41 0.54 * 0.14 0.08 0.28

239 sp_id15 0.22 0.43 0.28 -0.22 0.43

240 sp_id16 0.74 * 0.47 -0.09 0.06 0.02

241 sp_id17 0.23 0.70 * -0.07 0.31 0.22

242 sp_id18 0.45 0.32 0.01 0.32 0.02

243 sp_id19 0.38 0.52 * 0.16 0.13 0.00

244 sp_id20 0.42 0.69 * 0.09 -0.03 -0.05

245 sp_id21 0.23 0.16 0.32 0.22 -0.04

246 sp_id22 0.62 * 0.31 0.06 0.02 0.23

247 sp_id23 0.59 0.59 -0.11 0.15 0.25

248 sp_id24 0.48 * 0.33 0.04 0.09 0.28

249 sp_id25 0.63 * 0.30 0.28 -0.10 0.10

250 sp_id26 0.42 0.61 * 0.22 0.31 0.21

251 sp_id27 0.47 0.46 -0.07 0.16 0.41

252 sp_id28 0.56 0.43 0.33 -0.09 0.17

253 sp_id29 0.28 0.22 0.00 0.18 -0.18

254 sp_id30 0.60 0.62 -0.10 0.13 -0.07

255 uk_id1 0.28 0.20 0.29 0.20 0.49

256 uk_id2 0.54 * 0.15 -0.23 -0.04 0.42

257 uk_id3 0.07 0.42 0.11 0.25 0.65 *

258 uk_id4 0.32 0.33 -0.01 -0.03 0.53 *

Appendix 4.I

36

Q sort Factor

5/1 5/2 5/3 5/4 5/5

259 uk_id5 0.57 * 0.32 -0.10 0.06 0.30

260 uk_id6 0.52 * 0.41 -0.09 0.17 -0.05

261 uk_id7 -0.05 -0.01 0.05 0.29 0.70 *

262 uk_id8 0.17 -0.26 0.27 0.29 0.31

263 uk_id9 0.57 * 0.36 -0.06 0.05 0.41

264 uk_id10 0.45 0.32 0.22 0.43 0.35

265 uk_id11 0.28 -0.07 0.41 0.31 0.55

266 uk_id13 0.26 0.46 0.12 -0.10 0.49

267 uk_id14 0.33 0.60 0.47 0.02 0.25

268 uk_id15 0.55 * 0.02 0.10 0.25 0.26

269 uk_id16 0.35 0.43 * -0.04 0.10 0.00

270 uk_id17 0.21 0.34 0.16 -0.17 0.65 *

271 uk_id18 -0.02 0.15 0.19 0.30 0.35

272 uk_id19 0.23 0.26 0.19 0.07 0.27

273 uk_id20 0.34 0.29 0.62 * 0.18 0.24

274 uk_id21 0.60 * 0.18 0.20 0.22 0.39

275 uk_id22 0.22 0.26 0.44 0.43 0.27

276 uk_id23 0.49 * -0.05 0.04 -0.04 0.37

277 uk_id24 0.14 0.25 0.48 0.31 0.54

278 uk_id26 0.31 0.30 -0.31 0.14 0.30

279 uk_id27 0.04 0.37 0.49 0.34 0.24

280 uk_id28 0.29 0.37 0.22 0.33 0.51

281 uk_id29 0.30 0.47 0.46 0.15 0.34

282 uk_id30 0.11 0.54 * 0.08 0.06 0.46

283 uk_id31 -0.11 0.15 0.49 * 0.23 0.37

284 uk_id32 0.43 0.52 0.26 -0.04 0.36

285 uk_id33 0.39 0.43 -0.28 0.06 0.21

286 uk_id34 0.18 0.56 0.18 -0.14 0.49

287 uk_id35 -0.10 -0.17 0.04 0.33 0.60 *

288 uk_id36 0.09 0.33 0.02 0.39 0.35

289 uk_id37 0.33 0.61 0.42 0.07 0.35

290 uk_id38 0.04 0.03 0.19 0.48 0.49

291 uk_id39 0.02 0.44 0.25 -0.06 0.68 *

292 uk_id40 0.28 0.64 * 0.16 0.03 0.52

293 uk_id41 0.30 0.41 0.24 0.25 0.45

294 uk_id42 0.14 0.48 0.02 0.43 0.33

Note: * Q-sort is defining variable for factor.

Appendix 4.I

37

Table 4. I.7 Explained variance, and defining and associated Q sorts by factor, per country

Q sort Factor Total

1 2 3 4 5

Explained variance (%) 14 12 7 8 11 52

Defining / associated Q sorts* 66 / 135 42 / 126 16 / 53 20 / 59 39 / 104 183 / 477

▪ Denmark 2 / 6 3 / 10 1 / 5 1 / 5 9 / 17 16 / 43

▪ France 16 / 20 2 / 12 - / 4 - / 2 6 / 8 24 / 46

▪ Hungary 6 / 20 11 / 21 1 / 4 14 / 23 1 / 7 33 / 75

▪ Norway 1 / 4 1 / 6 5 / 13 2 / 7 2 / 8 11 / 38

▪ Palestine 8 / 12 3 / 6 3 / 4 - / - 2 / 2 16 / 24

▪ Poland 4 / 12 7 / 16 - / 2 3 / 9 3 / 5 17 / 44

▪ Spain 9 / 21 5 / 17 - / - - / 2 2 / 9 16 / 49

▪ Sweden 3 / 8 3 / 10 4 / 12 - / 3 2 / 7 12 / 40

▪ The Netherlands 10 / 20 4 / 9 - / - - / 2 6 / 15 20 / 46

▪ UK 7 / 12 3 / 19 2 / 9 - / 6 6 / 26 18 / 72

Note: * Number of Q sorts statistically significantly (p<.05) associated with the factors exceeds the number of participants due to confounded variables.

Appendix 4.I

38

Table 4. I.8 Z- and Rank-scores of statements

Statement Factor

1 2 3 4 5

Z Rank Z Rank Z Rank Z Rank Z Rank

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

0.26 +1 0.70 +1 0.94 +2 -0.27 -1 -0.50 -1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

-0.20 0 0.14 0 -0.12 0 -0.12 0 -0.04 -1

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-1.78 -4 -1.40 -4 -1.77 -4 -0.57 -2 -1.75 -4

4 Patient characteristics like age, gender or income should play no role in prioritizing between people.

2.03 +4 1.74 +4 -0.73 -2 0.21 +1 0.59 +2

5 People who are in paid work and so contribute financially to society should be prioritized over people who do not work.

-1.77 -4 -1.36 -3 -1.54 -3 -0.20 0 -1.49 -3

6 If a treatment adds one month to the life of a patient and costs €7,500, one should consider whether the money could have been better spent on other health care.

-0.98 -2 0.20 0 0.37 +1 -0.17 0 0.46 +1

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-1.11 -3 -1.13 -2 -1.24 -2 -0.11 0 -1.11 -2

8 Rescuing people from a certain death should take priority over all other kinds of health care.

0.90 +2 1.46 +3 0.21 +1 1.98 +4 -1.63 -3

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

0.11 +1 -0.27 0 -0.14 -1 0.36 +1 0.21 0

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

-0.12 0 0.29 +1 -0.17 -1 0.11 0 0.04 0

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-0.32 -1 -0.15 0 -0.25 -1 -1.48 -3 -0.71 -2

Appendix 4.I

39

Statement Factor

1 2 3 4 5

Z Rank Z Rank Z Rank Z Rank Z Rank

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise.

1.14 +3 0.24 +1 -0.04 0 0.94 +2 1.04 +2

13 People who depend heavily on members of their family or neighbours for care should be treated with priority.

-0.16 0 0.22 +1 -0.03 0 -0.25 -1 -0.03 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

-0.09 0 -0.62 -1 1.46 +3 0.03 0 0.53 +1

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

0.95 +2 1.37 +3 1.54 +4 1.62 +3 1.40 +3

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-1.14 -3 -1.29 -3 -1.19 -2 -1.34 -3 -1.08 -2

17 There is no sense in saving lives if the quality of those lives will be really bad.

-1.40 -3 -0.28 -1 -0.50 -1 -1.19 -2 1.81 +4

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

0.63 +1 1.02 +2 0.51 +1 0.63 +2 0.15 0

19 Priority should be given to those treatments that generate the most health.

0.77 +1 1.09 +2 1.15 +2 1.29 +3 1.05 +3

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month.

-0.75 -2 -0.39 -1 0.33 +1 -0.43 -1 -0.26 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

0.99 +2 1.12 +2 1.18 +2 -1.75 -4 -0.71 -2

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

-0.08 +1 0.30 +1 -0.15 -1 0.36 +1 0.94 +2

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet.

-0.65 -2 -1.39 -3 1.33 +3 -0.32 -1 0.08 0

Appendix 4.I

40

Statement Factor

1 2 3 4 5

Z Rank Z Rank Z Rank Z Rank Z Rank

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively.

1.61 +3 -1.05 -2 -0.88 -2 -1.67 -4 -0.27 -1

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-0.67 -2 -1.26 -2 -1.47 -3 1.23 +3 0.55 +1

26 Priority should be given to younger people, because they may benefit from treatment for longer.

-0.27 0 -1.42 -4 0.95 +2 -0.29 -1 0.07 0

27 It is more important to prevent ill health than it is to cure ill health once it occurs.

1.32 +3 0.72 +2 -0.13 0 1.71 +4 1.45 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-0.13 0 -0.12 0 -1.74 -4 -1.24 -2 -0.64 -1

29 Access to health care should be based on need, not on geographical, social or economic circumstances.

1.93 +4 2.00 +4 1.46 +4 1.08 +2 1.65 +4

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-0.58 -1 -0.06 0 -1.46 -3 -1.29 -3 -1.25 -3

31 Parents with dependent children should be given priority over similar people without dependents.

-0.58 -1 -0.87 -2 0.17 0 1.14 +2 0.25 +1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

-0.44 -1 -0.34 -1 1.18 +3 0.28 +1 0.22 +1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life.

-0.52 -1 -0.57 -1 -0.02 0 -0.76 -2 -1.98 -4

34 The amount of health care people have had in the past should not influence access to treatments in the future.

1.08 +2 1.35 +3 0.78 +1 0.50 +1 0.96 +2

Appendix 4.I

41

Figure 4. I.3 Idealised Q sort for GP1: “Egalitarianism, entitlement and equality of access”

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritized over people who do not work.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Priority should be given to those treatments that

generate the most health.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Patient characteristics like age, gender or income should play no role in prioritizing between

people.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Parents with dependent children should be given

priority over similar people without dependents.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Appendix 4.I

42

Figure 4. I.4 Idealised Q sort for GP2: “Efficiency, severity and the magnitude of health gains”

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Parents with dependent children should be given

priority over similar people without dependents.

There is no sense in saving lives if the quality of

those lives will be really bad.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who are in paid work and so contribute

financially to society should be prioritized over people who do not work.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to those treatments that

generate the most health.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritizing between

people.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Appendix 4.I

43

Figure 4. I.5 Idealised Q sort for GP3: “Fair innings, young people and maximising health benefits”

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Patient characteristics like age, gender or income should play no role in prioritizing between

people.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Parents with dependent children should be given

priority over similar people without dependents.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Priority should be given to those treatments that

generate the most health.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who are in paid work and so contribute

financially to society should be prioritized over people who do not work.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

There is no sense in saving lives if the quality of

those lives will be really bad.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Rescuing people from a certain death should take priority over all other kinds

of health care.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Appendix 4.I

44

Figure 4. I.6 Idealised Q sort for GP4: “The intrinsic value of life and healthy living”

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Parents with dependent children should be given

priority over similar people without dependents.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Priority should be given to those treatments that

generate the most health.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to younger people, because

they may benefit from treatment for longer.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Patient characteristics like age, gender or income should play no role in prioritizing between

people.

People who are in paid work and so contribute

financially to society should be prioritized over people who do not work.

Appendix 4.I

45

Figure 4. I.7 Idealised Q sort for GP5: “Quality of life above all else”

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People who are in paid work and so contribute

financially to society should be prioritized over people who do not work.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Rescuing people from a certain death should take priority over all other kinds

of health care.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to those treatments that

generate the most health.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Parents with dependent children should be given

priority over similar people without dependents.

Patient characteristics like age, gender or income should play no role in prioritizing between

people.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Appendix 4.I

1

General public views in 10 countries

Contents

4.II General public views in 10 countries................................................................ 3

4.II.1 Denmark ................................................................................................... 4

4.II.2 France .................................................................................................... 13

4.II.3 Hungary .................................................................................................. 20

4.II.4 Norway ................................................................................................... 29

4.II.5 Palestine ................................................................................................. 38

4.II.6 Poland .................................................................................................... 45

4.II.7 Spain ...................................................................................................... 54

4.II.8 Sweden ................................................................................................... 65

4.II.9 The Netherlands ..................................................................................... 74

4.II.10 The UK ................................................................................................... 81

Index of tables

Table 4.II.1 Rank scores of the 34 statements for the three factors in Denmark .............................. 8

Table 4.II.2 Rank scores of the 34 statements for the two factors in France ..................................16

Table 4.II.3 Rank scores of the 34 statements for the three factors in Hungary .............................24

Table 4.II.4 Rank scores of the 34 statements for the three factors in Norway ...............................33

Table 4. II.5 Rank scores of the 34 statements for the two factors in Palestine ...............................41

Table 4. II.6 Rank scores of the 34 statements for the three factors in Poland ................................49

Table 4. II.7 Rank scores of the 34 statements for the four factors in Spain ....................................59

Table 4. II.8 Rank scores of the 34 statements for the three factors in Sweden ..............................69

Table 4. II.9 Rank scores of the 34 statements for the two factors in the Netherlands ....................77

Table 4. II.10 Rank scores of the 34 statements for the four factors in the UK ..................................85

Index of figures

Figure 4.II.1 Correlations between consecutive factor solutions for Danish data ............................... 4

Figure 4.II.2 Idealised Q sort for point of view 1 in Denmark ............................................................10

Figure 4.II.3 Idealised Q sort for point of view 2 in Denmark ............................................................11

Figure 4.II.4 Idealised Q sort for point of view 3 in Denmark ............................................................12

Appendix 4.II

2

Figure 4.II.5 Correlations between consecutive factor solutions for French data .............................13

Figure 4.II.6 Idealised Q sort for point of view 1 in France ...............................................................18

Figure 4.II.7 Idealised Q sort for point of view 2 in France ...............................................................19

Figure 4.II.8 Correlations between consecutive factor solutions for Hungarian data .......................20

Figure 4.II.9 Idealised Q sort for point of view 1 in Hungary .............................................................26

Figure 4.II.10 Idealised Q sort for point of view 2 in Hungary .............................................................27

Figure 4.II.11 Idealised Q sort for point of view 3 in Hungary .............................................................28

Figure 4.II.12 Correlations between consecutive factor solutions for Norwegian data .......................29

Figure 4.II.13 Idealised Q sort for point of view 1 in Norway ..............................................................35

Figure 4.II.14 Idealised Q sort for point of view 2 in Norway ..............................................................36

Figure 4.II.15 Idealised Q sort for point of view 3 in Norway ..............................................................37

Figure 4.II.16 Correlations between consecutive factor solutions for Palestinian data ......................38

Figure 4. II.17 Idealised Q sort for point of view 1 in Palestine ............................................................43

Figure 4. II.18 Idealised Q sort for point of view 2 in Palestine ............................................................44

Figure 4. II.19 Correlations between consecutive factor solutions for Polish data ..............................45

Figure 4. II.20 Idealised Q sort for point of view 1 in Poland ...............................................................51

Figure 4. II.21 Idealised Q sort for point of view 2 in Poland ...............................................................52

Figure 4. II.22 Idealised Q sort for point of view 3 in Poland ...............................................................53

Figure 4. II.23 Correlations between consecutive factor solutions for Spanish data ...........................54

Figure 4. II.24 Idealised Q sort for point of view 1 in Spain .................................................................61

Figure 4. II.25 Idealised Q sort for point of view 2 in Spain .................................................................62

Figure 4. II.26 Idealised Q sort for point of view 3 in Spain .................................................................63

Figure 4. II.27 Idealised Q sort for point of view 4 in Spain .................................................................64

Figure 4. II.28 Correlations between consecutive factor solutions for Swedish data ..........................65

Figure 4. II.29 Idealised Q sort for point of view 1 in Sweden .............................................................71

Figure 4. II.30 Idealised Q sort for point of view 2 in Sweden .............................................................72

Figure 4. II.31 Idealised Q sort for point of view 3 in Sweden .............................................................73

Figure 4. II.32 Correlations between consecutive factor solutions for Dutch data ...............................74

Figure 4. II.33 Idealised Q sort for point of view 1 in the Netherlands .................................................79

Figure 4. II.34 Idealised Q sort for point of view 2 in the Netherlands .................................................80

Figure 4. II.35 Correlations between consecutive factor solutions for the UK data .............................81

Figure 4. II.36 Idealised Q sort for point of view 1 in the UK ...............................................................87

Figure 4. II.37 Idealised Q sort for point of view 2 in the UK ...............................................................88

Figure 4. II.38 Idealised Q sort for point of view 3 in the UK ...............................................................89

Figure 4. II.39 Idealised Q sort for point of view 4 in the UK ...............................................................90

Appendix 4.II

3

4.II General public views in 10 countries

This Appendix presents the results of the separate analyses of the general public

data collected in each of the 10 participating countries.

For each country, in alphabetical order, the following information is presented:

The number of participants, the factor structures supported by their ranking of the

34 statements and the chosen factor solution.

The factor diagram1

A description of the factors, based on the factor arrays

.

2 and the distinguishing

statements for each factor3

A table with the statement rank scores for each factor.

.

Figures displaying the idealised Q sort for each factor in the score sheet.

1 The factor diagram is a simplified visual representation of the hierarchical factor structure in the data. Each row presents a consecutive factor structure for the data (from separate analyses of the data), from the one-factor structure at the top to the maximum supported structure by the data at the bottom. The boxes in each row represent individual factors, and their width the percentage explained variance. The arrows between boxes highlight the most important correlations between factors in consecutive factor structures (i.e., between-structures comparison), the numbers the corresponding correlation coefficient. 2 This is the idealised Q sort for each factor, i.e. the weighted average ranking of the statements computed on the basis of the individual rankings of the statements of Q sorters loading significantly on that factor with their factor loadings as weight. The statement scores, ranging from -4 (‘disagree most’) to +4 (‘agree most’), can be used to arrange the statements on the score sheet, as used by the respondents. 3 The distinguishing statements are those that are ranked statistically significantly differently from all other factors in the solution. These statements highlight what is really different about this factor as compared to the other factors.

Appendix 4.II

4

Denmark

25 participants from Denmark completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4.II.1). The statistical features of each factor

solution were examined. The three-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4.II.1 presents the factor

scores of the statements, Figure 4.II.2 to Figure 4.II.4 the idealised Q sorts for each

point of view.

Figure 4.II.1 Correlations between consecutive factor solutions for Danish data

Appendix 4.II

5

Before describing each factor in more detail and focussing on the distinguishing

features of each account, there are a number of issues about which the three Danish

factors are all in agreement. That it is more important to prevent ill health that to cure

it (#27) and that treatment should go to those people who benefit most from it (#32)

are both ranked highly in all three factors. There is also consensus that contribution

to the system, income, having a partner or rarity of the disease should not matter for

priority (#3; #7; #16 and #30 have factor scores of 3 or lower across all factors). In

what follows, and with the consensus issues in mind, each factor is described and

attention is paid to the matters that separate each of the accounts from the other

three.

The account presented in factor 1 is that prioritisation of care should be about

generating the most health. This is evident from the placing of statements #19 and

#15, but also from the importance attached to prevention (#27). Participants

explained that prevention promotes health, and therefore should have priority.

Similarly, treatments should go to the patients who benefit most from it (#32).

Factor 1 - “Health care should be about health”

While generating health is very important, the people who make up factor 1 express

a clear preference for improving quality of life over prolonging length of life (#33).

This is supported by the placement of statements #17 and #22, indicating that quality

of life after treatment should also be taken into account.

Characteristics or contributions of patients should not be of importance. Access to

health care should be based on need (#29) and not depend on past health

Appendix 4.II

6

consumption (#34), personal and socio-demographic characteristics (#4; #29), or

having dependents (7, 31). Peoples’ socio economic status (#16) or financial

contribution should also not be taken into account (#3; #5), but buying priority

treatment is not completely rejected (#24).

Factor 2 is a similar to factor 1 in the importance attached to health, but puts more

focus on promoting quality of life, which is clearly found to be more important than

longevity (#17; #33; #1). Another distinguishing aspect in this factor, related to the

importance given to quality of life, is that culpability is considered relevant. When

people have not taken personal responsibility for their own health, they should not

burden others when they become ill and thus receive lower priority in the provision of

care (#21, #25).

Factor 2 - “Promote quality of life”

Doctors -rather than politicians- should be given a strong role in prioritising health

care (#12), because they supposedly are best able to judge whether treatment is

meaningful from the perspective of the patient and efficient from the perspective of

the health care budget. This latter notion of efficiency is also put forward in the

placement of statements #15 and #6.

In factor 1 there is least support for equality in access, which can be seen from

distinguishing low scores for statements #4 and #29, although like in the other factors

priority on the basis of income or financial contribution to the system is rejected (#3;

#5). There is a distinct preference for treating the young (#23; #14), largely from a

‘fair innings’ motivation, and people with dependent children (#31). Although no clear

Appendix 4.II

7

explanation was given for the latter, in the context of this factor it possibly relates to

the preference for young people -which young parents presumably are.

According to the people making up factor 3, access to care should be based on need

(#29) and doctors should be the ones to judge which patients are most entitled (#12).

Saving lives should receive priority over anything else (#8) and prolonging life is

distinctly more important than improving quality of life when compared to other

factors (#33; #17; #22) –overall there seems to be a slight preference for treating the

elderly (#14; #26).

Factor 3 “Access based on need”

In line with the accounts in the other factors, prioritisation on the basis of personal

characteristics (#4; #29) or financial status (#3; #5) is rejected. Socio-economic

status (#16) and claims on health services in the past (#34) should also not be

considered.

In contrast to the other factors, buying priority treatment (#24) is disapproved, as is

prioritisation on the basis of lifestyle and personal responsibility (#21; #25) –which

distinguishes this factor from the previous factor in particular.

Appendix 4.II

8

Table 4.II.1 Rank scores of the 34 statements for the three factors in Denmark

Statement 1 2 3

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1 -1* 0

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-3 -3 -4

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4 -1** +4

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -4 -2** -3

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

0 +2** -1

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-3 -3 -3

8 Rescuing people from a certain death should take priority over all other kinds of health care. 0 0 +3**

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. 0 0 0

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

+1 0 0

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1* -2 -2

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +2** +4 +4

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. -1 -1 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

0* +1** -1*

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+3 +3 +1

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-4 -2 -4

17 There is no sense in saving lives if the quality of those lives will be really bad. +2 +4 -2**

Appendix 4.II

9

Statement 1 2 3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 0 +2

19 Priority should be given to those treatments that generate the most health. +4** +1 +1

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -2* -1* +1*

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+1* -4** +2*

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

+1 +1 -1**

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -1 +2** -1

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. -2* -1* +1**

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-1** +3** -3**

26 Priority should be given to younger people, because they may benefit from treatment for longer. -1* +1* -1*

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +2 +3 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

0** -2 -2

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +3 +1* +3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-2 -3 -2

31 Parents with dependent children should be given priority over similar people without dependents. -2** +2** 0**

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

+2 +2 +1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -3* -4* +2**

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3 0** +2

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

10

Figure 4.II.2 Idealised Q sort for point of view 1 in Denmark

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

There is no sense in saving lives if the quality of

those lives will be really bad.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Priority should be given to those treatments that

generate the most health.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Parents with dependent children should be given

priority over similar people without dependents.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Appendix 4.II

11

Figure 4.II.3 Idealised Q sort for point of view 2 in Denmark

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Patient characteristics like age, gender or income should play no role in prioritising between

people.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Priority should be given to those treatments that

generate the most health.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

There is no sense in saving lives if the quality of

those lives will be really bad.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Parents with dependent children should be given

priority over similar people without dependents.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Appendix 4.II

12

Figure 4.II.4 Idealised Q sort for point of view 3 in Denmark

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

There is no sense in saving lives if the quality of

those lives will be really bad.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Parents with dependent children should be given

priority over similar people without dependents.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to those treatments that

generate the most health.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to younger people, because

they may benefit from treatment for longer.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Appendix 4.II

13

France

28 participants from France completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4.II.5). The statistical features of each factor

solution were examined. The two-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4.II.2 presents the factor

scores of the statements, Figure 4.II.6 and Figure 4.II.7 the idealised Q sorts for each

point of view.

Figure 4.II.5 Correlations between consecutive factor solutions for French data

Appendix 4.II

14

The account presented in factor 1 is that provision of health care should be based on

health need and not on the personal characteristics of the patients. This includes the

age, gender or geographical location (#29; #4) of a person as well as any previous

use of the health care system (#34). As indicated by statements #21 and #25 a

person’s lifestyle and whether this was a contributor to an illness should not give

them lower priority for treatment.

Factor 1 – “equal access to health care”

Similarly, a person’s financial contribution should not be taken into account as is

shown by the rejection of statements #3 and #5 and people should not be able to buy

themselves priority treatment (#24) even if this does not affect the care of others.

For factor 1 there is some preference for providing treatments which save lives (#8)

even if quality of life following treatment will be poor (#1). The placing of statement

#33 at the lower end of the distribution suggests that this is view is focused on saving

lives rather than prolonging life as the participants on factor 1 disagree that it is more

important to provide treatments which prolong life than those which improve quality of

life. Possibly related to this support for treatments which save lives, is the

disagreement with statement #6 that the opportunity cost of providing a treatment

which may only add one month to the life of a patient should be considered.

For the respondents who make up factor 2, quality of life is important. The provision

of treatment should not just be about saving lives (#28) if the quality of life a patient

will have following treatment would be poor (#17). This view is further supported by

Factor 2 – “quality of life”

Appendix 4.II

15

the placing of statement #33 in the ‘most disagree’ position with treatments that

improve quality of life being just as important as those which prolong life. However

the positioning of statements #11 and #22 suggest that while quality of life is

important priority should not be given to treatments that would only restore people to

a poor level of health.

Similar to factor 1 there is the belief that access to health care should not be based

on patient characteristics such as age, gender or socioeconomic status (#4; #29) but

within factor 2 there is not such a strong objection to those who have made a

financial contribution to society receiving lower priority (#3; #5). For non emergency

treatments, patients should be treated according to health need and not on a first

come first served basis (#2; #18). The decision on who receives priority for treatment

should be made by doctors based on their medical expertise (#12).

For this factor prevention is important as can be seen by the placing of statement #27

in the ‘most agree’ position.

Appendix 4.II

16

Table 4.II.2 Rank scores of the 34 statements for the two factors in France

Statement 1 2

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1** -1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -3

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4* +3

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -4** -2

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

-3** 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-2 -2

8 Rescuing people from a certain death should take priority over all other kinds of health care. +2** -3

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. +1** +2

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future. 0 -1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1** -3

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +2** +4

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. +1 +1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

-1** +1

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2 +3

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-2 -2

17 There is no sense in saving lives if the quality of those lives will be really bad. -3** +3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former +2* +1

Appendix 4.II

17

Statement 1 2

should be treated with priority.

19 Priority should be given to those treatments that generate the most health. +1 +1

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -1 0

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else. +3** -1

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

0** +2

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. 0 0

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. +3** 0

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-3** 0

26 Priority should be given to younger people, because they may benefit from treatment for longer. -1 -1

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +1** +4

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-2** -4

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4** +2

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-1 -2

31 Parents with dependent children should be given priority over similar people without dependents. 0** +1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

0 -1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -2** -4

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3* +2

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

18

Figure 4.II.6 Idealised Q sort for point of view 1 in France

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

There is no sense in saving lives if the quality of

those lives will be really bad.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to those treatments that

generate the most health.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Parents with dependent children should be given

priority over similar people without dependents.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Appendix 4.II

19

Figure 4.II.7 Idealised Q sort for point of view 2 in France

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to those treatments that

generate the most health.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

There is no sense in saving lives if the quality of

those lives will be really bad.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to younger people, because

they may benefit from treatment for longer.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Appendix 4.II

20

Hungary

47 participants from Hungary completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4.II.8). The statistical features of each factor

solution were examined. The three-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4.II.3 presents the factor

scores of the statements, Figure 4.II.9 to Figure 4.II.11 the idealised Q sorts for each

point of view.

Figure 4.II.8 Correlations between consecutive factor solutions for Hungarian data

Appendix 4.II

21

Factor 1 presents a predominantly egalitarian viewpoint. The emphasis is on

providing access to health care which is based on health need rather than any

personal characteristics of the patient, such as age, gender or whether they have

dependents (#29; #4; #7; #31; #23; #26). It also includes the amount of health care

people have had in the past (#34). In particular a person’s financial contribution

should not be taken into account as is shown by the rejection of statements #3 and

#5 including positive discrimination to those in lower income groups (#16).

Factor 1

Health need is more important in the decision on who receives treatment. Patients

should be treated on a according to need rather than on a first come first served

basis on a waiting list (#28) with treatments which would rescue people from certain

death given priority over other health care treatment (#8). The size of the health gain

should also be taken into account with funding going to treatments that give the

biggest health gain (#15; #19).

For factor 2 the focus is on the size of the health gain that can be received from

treatment. The placing of statements #15 and #19 at the ‘most agree’ end of the

distribution highlights that the health outcome itself should be taken into

consideration when allocating health care resources. Preventing ill health in the first

place is also ranked highly (#27).

Factor 2

In contrast to the previous factor, those participants who load onto factor 2 believe

that some personal characteristics could be taken into account when deciding on

Appendix 4.II

22

who should receive treatment. In particular, a person’s lifestyle. If a person is seen

as being in some way responsible for their illness this should be taken into account

with lower priority for people who have led an unhealthy life. This can be seen in the

placing of statement #25 as something with which they most agree and supported by

statement #21 with which they disagree most. In addition to lifestyle, respondents

who load on factor 2 gave some support to parents with dependent children being

given priority over those without dependent children (#31).

A distinguishing feature of this factor is the more neutral attitude to whether a

person’s previous financial contribution is important when prioritising treatments (#3;

#5). However, there should be no positive discrimination towards those in lower

income groups (#16). There is agreement that people can pay for treatment (#24).

For factor 2 there appears to be a preference for providing treatments which save

lives as is indicated by the position of statement #8 but that quality of life following

treatment is also important. Statement #11, which is distinguishing for factor 2,

highlights the idea that there is some acceptable minimum quality of life and priority

should not be given to those whose quality of life is very low if it can only be

improved by a small amount. This is supported by the placing of statements #17 and

#22 that quality of life should be considered alongside the decision to rescue

someone from a certain death.

To some extent the account provided by factor 3 is that provision of health care

should be based on health need and not patient characteristics (#4; #29). In

Factor 3

Appendix 4.II

23

particular, a patient’s age (#26; #14) or whether they have a partner (#7) should not

be used to prioritise between people. The amount of health care a person has

previously received should not influence their future access to treatment (#34). This

relates back to health need being important in how health care resources are used.

If a patient continues to need treatment, then this should be available to them.

Decisions on who receives priority for treatment should be made by doctors based on

their medical expertise (#12). Providing treatments which rescue people from certain

death should also be given priority (#8).

A distinguishing feature of this factor is the role of an individual’s financial

contribution in receiving priority for treatment. There is some support from this factor

that people who work and pay taxes or make some financial contribution to society

should receive priority over people who do not (#5; #3). It is this contribution to

society which appears to be important here as those who load onto this factor also

agreed with statement #24 that people should not be allowed to buy themselves

priority treatment.

Appendix 4.II

24

Table 4.II.3 Rank scores of the 34 statements for the three factors in Hungary

Statement 1 2 3

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1** -2 0

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4** -1** +1

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4 0** +3

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -4** -1** +2**

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

+1 +1 -3**

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-3 0** -4

8 Rescuing people from a certain death should take priority over all other kinds of health care. +3 +4 +4

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. 0 +1 +1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

+1 +1 -1**

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 -3** -2

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +2 +2 +4**

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0 -1 -1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

-1* 0** -3*

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+3 +3 +2**

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-3 -3 -2

17 There is no sense in saving lives if the quality of those lives will be really bad. -1 -2 -2

Appendix 4.II

25

Statement 1 2 3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 +1 +1

19 Priority should be given to those treatments that generate the most health. +2 +3* +2

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. 0 -1 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+1* -4** +1*

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

+2 +2 0**

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2 -1* -2

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. 0** -4** +1

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-2** +3** 0**

26 Priority should be given to younger people, because they may benefit from treatment for longer. -2** 0** -4**

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +2 +4** +2

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-3 -2 -3

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4** +2 +3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-1 -3** -1

31 Parents with dependent children should be given priority over similar people without dependents. -2** +2** 0**

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

0 +1** -1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -1 -2 0**

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3 0** +3

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

26

Figure 4.II.9 Idealised Q sort for point of view 1 in Hungary

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Parents with dependent children should be given

priority over similar people without dependents.

There is no sense in saving lives if the quality of

those lives will be really bad.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Priority should be given to those treatments that

generate the most health.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Appendix 4.II

27

Figure 4.II.10 Idealised Q sort for point of view 2 in Hungary

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Rescuing people from a certain death should take priority over all other kinds

of health care.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Priority should be given to those treatments that

generate the most health.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Parents with dependent children should be given

priority over similar people without dependents.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Appendix 4.II

28

Figure 4.II.11 Idealised Q sort for point of view 3 in Hungary

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Priority should be given to younger people, because

they may benefit from treatment for longer.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

There is no sense in saving lives if the quality of

those lives will be really bad.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Parents with dependent children should be given

priority over similar people without dependents.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to those treatments that

generate the most health.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Appendix 4.II

29

Norway

23 participants from Norway completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4.II.12). The statistical features of each factor

solution were examined. The three-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4.II.4 presents the factor

scores of the statements, Figure 4.II.13 to Figure 4.II.15 the idealised Q sorts for

each point of view.

Figure 4.II.12 Correlations between consecutive factor solutions for Norwegian data

Appendix 4.II

30

Before describing each factor in more detail and focussing on the distinguishing

features of each account, there are a number of issues about which the three

Norwegian factors are all in agreement. Whether or not people work and so

contribute financially to society is not considered relevant to health care priority

setting (see statement #5). The positive positioning of statements #15 and #19 in all

three factors indicates consensus around prioritising treatments that generate the

biggest health benefits. And there is general objection to both the prioritisation of

treatments for rare conditions (#30) and the notion that prolongation of life is more

important than improving quality of life through health care (#33). In what follows,

and with these consensus issues in mind, each factor is described and attention is

paid to the matters that distinguish each of the accounts from the other two.

The key feature that makes the account described by factor 1 distinct is a belief that

the young should be prioritised over the old (#23 ; #14 and #26), not only because

their capacity to benefit may be greater with a longer lifespan but also based on a

‘fair innings’ argument. These statements are not supported in the accounts

presented by factors 2 and 3. Health gain is important in prioritising health care in

this as in other factors (#15; #19; #1; #32). There is support for prioritising some

groups of people over other groups of people, and the notion that patient

characteristics should play no part (#4) has little resonance. It is consistent,

therefore, that the operation of waiting lists is rejected (#28). Whilst prioritisation is

acceptable with regard to some characteristics, others should not be taken into

account and financial contribution through taxes or premiums should not enter into

priority setting in this account (#3) nor should productivity (#5). Whether or not

Factor 1 - “The young”

Appendix 4.II

31

patients can be held in some way responsible for their own illnesses, for example

due to their lifestyle, should not result in lower priority (#25).

Factor 2 presents a strong account that has two central features, firstly that everyone

is equal when it comes to health care and secondly that resources should be directed

to the treatments and services that deliver the greatest health benefits. This

message can be seen in the four most important statements for this factor (see

Factor 2 - “Health as a human right: equal access and maximum benefits”

Table

4.II.4). Treatment is important whatever the age of the patients and regardless of

their economic situation or financial contribution to the health service. This is the

only one of the three factors that does not regard an additional year of life to

someone who will otherwise die at 30 with more importance than for someone who

would die at 80 (#14) and rejects all statements about prioritising the young (#14;

#23; #26) (although, interestingly, parents with dependent children are considered a

priority (#31) as are people who depend on their family (#13).

Factor 3 shares some views with 1 and 2 – in a concern for the just allocation of

resources (#29) and maximising health gain. In this account, though, effectiveness

would appear to be more important than equality, revealed by the importance placed

on #32 which is a significant, distinguishing statement with much less resonance in

the other factors. The significant emphasis on prevention (#27) might also stem from

a concern with effectiveness and the size of the health gain over time. The

importance placed on parents with dependent children distinguishes this account

(#31). It fits, then, that waiting lists are not regarded as a good way of setting

Factor 3 - “Prevention, health gain and family”

Appendix 4.II

32

priorities, which should take into account other issues. Private payment for treatment

is not ruled out in factor 3. Statement #24**is important and is rejected by this

account alone, implying that people should be allowed to buy treatment if it doesn’t

affect others.

Appendix 4.II

33

Table 4.II.4 Rank scores of the 34 statements for the three factors in Norway

Statement 1 2 3

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+2** -1 0

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 +1 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -4 -2*

4 Patient characteristics like age, gender or income should play no role in prioritising between people. 0** +4** -2**

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -3 -3 -2

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

+2 0 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-4* -3 -1

8 Rescuing people from a certain death should take priority over all other kinds of health care. -2** +3 +2

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. 0 0 0

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

-2 0* -1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 -1 -1

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. 0* +1* -2**

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0* +2 +1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

+3 -1** +2

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+3 +3 +3

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-1** -3 -3

17 There is no sense in saving lives if the quality of those lives will be really bad. -1 -1 -1

Appendix 4.II

34

Statement 1 2 3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 +1 +2

19 Priority should be given to those treatments that generate the most health. +2 +3 +2

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -1 -1 +1*

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+1 +2 -1*

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

-1* 0 +1

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. +4** -4** 0**

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. +1 +1 -4**

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-3 -2 0**

26 Priority should be given to younger people, because they may benefit from treatment for longer. +3** -2** +1**

27 It is more important to prevent ill health than it is to cure ill health once it occurs. 0** +2* +4*

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-3 0** -4

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4 +4 +3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-2 -2 -3

31 Parents with dependent children should be given priority over similar people without dependents. +1* +2* +3**

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

+2* 0** +4*

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -2 -2 -3

34 The amount of health care people have had in the past should not influence access to treatments in the future. +1 +1 +1

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

35

Figure 4.II.13 Idealised Q sort for point of view 1 in Norway

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Priority should be given to those treatments that

generate the most health.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Parents with dependent children should be given

priority over similar people without dependents.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Appendix 4.II

36

Figure 4.II.14 Idealised Q sort for point of view 2 in Norway

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to those treatments that

generate the most health.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Rescuing people from a certain death should take priority over all other kinds

of health care.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Appendix 4.II

37

Figure 4.II.15 Idealised Q sort for point of view 3 in Norway

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Priority should be given to those treatments that

generate the most health.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Rescuing people from a certain death should take priority over all other kinds

of health care.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

There is no sense in saving lives if the quality of

those lives will be really bad.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Parents with dependent children should be given

priority over similar people without dependents.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Appendix 4.II

38

Palestine

20 participants from Palestine completed the paper Q sort. Their data supported a

maximum of five factors (see Figure 4.II.16). The statistical features of each factor

solution were examined. The two-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4. II.5 presents the factor

scores of the statements, Figure 4. II.17 to Figure 4. II.18 the idealised Q sorts for

each point of view.

Figure 4.II.16 Correlations between consecutive factor solutions for Palestinian data

Appendix 4.II

39

Before describing each factor in more detail and focussing on the distinguishing

features of each account, there are a number of issues about which the two

Palestinian factors are in agreement. That doctors should decide on priorities (#12),

that priority should be given to treatments that generate the most health (#19), that

lifestyle should not be relevant (#21) and that it is more important to prevent ill health

that to cure it (#27) are all ranked highly in both factors. There is also consensus that

contribution to the system should not matter for priority (#3; #5) and both factors

strongly reject the statement that there is no sense saving lives if the quality after

treatment is really bad (#17). In what follows, and with the consensus issues in mind,

each factor is described and attention is paid to the matters that separate each of the

accounts from the other three.

The account presented in Factor 1 is about equal access to care, based on need and

irrespective of personal characteristics and financial contribution. This is evident from

the high ranking of statements #4 and #29 and the low ranking of #3 and #5. Lifestyle

should also not be relevant to priority (#21; #25) and taking account of age is also

rejected (#14; #23).

Factor 1 - “Egalitarian perspective”

Health care should be concerned with saving lives (#8), even if the quality of these

lives after treatment will be really bad (#17). In line with this, statement #20 is

rejected. Life is considered as being precious and each life as equally valuable. From

the participants explanations religion appears to play an important role in this.

There is a preference for treating illnesses that cause substantial burden on the

families of patients (#9), which participants connect to the financial/economic position

of the family and should be considered in conjunction with agreement to statement

Appendix 4.II

40

#16. This can also be seen from the low placing of statements #7 and #31, which

clearly indicate that patients with a partner (#7) or dependent children (#31) should

not receive priority. Participants motivated this from the egalitarian perspective.

This factor is focussed on health gain from treatment. Priority should be given to

those treatments that generate the most health (#15; #19) and to the patients that

benefit most from treatment (#32), which is interpreted both in terms of severity of

illness and capabilities regained after treatment.

Factor 2 - “Priority to those who benefit most”

The people who make up this account show a preference for treating young people

(#14; #23). The explanation given by participants is that they expect that younger

patients will have family and professional responsibilities and 80 year olds most likely

not, and therefore the young patients and their environs will benefit most from

treatment. The distinguishing higher scores for statements #7 (having a partner) and

#31 (having dependent children) fit in this same reasoning. The low score for

statement #5 (which refers to having paid work) is notable in this context, but the

interpretation given was that priority based on patient’s ability to pay (see also #3;

#24) would exclude vulnerable families from health services. Therefore, doctors

should decide who benefits most from treatment (#12), not the individual based on

his financial situation.

The low scores of statements #4 and #29 clearly distinguishes factor 2 from the

egalitarian account in factor 1. The low placement of statements #28 and #11 also

supports the preference for prioritising certain groups, based on severity of illness

and benefit from treatment.

Appendix 4.II

41

Table 4. II.5 Rank scores of the 34 statements for the two factors in Palestine

Statement 1 2

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1 +1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

-1* 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-3 -3

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4** 0

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -2 -4

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

0 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-2** +1

8 Rescuing people from a certain death should take priority over all other kinds of health care. +3** -1

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. +2** -1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future. 0 0

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1** -3

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +2 +2

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0 +1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

-3** +2

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2** +4

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

+1** -2

17 There is no sense in saving lives if the quality of those lives will be really bad. -4* -2

Appendix 4.II

42

Statement 1 2

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

0** +2

19 Priority should be given to those treatments that generate the most health. +2* +3

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -4** -2

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else. +3 +3

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

0 0

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2** +1

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. +1* -1

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-1 -1

26 Priority should be given to younger people, because they may benefit from treatment for longer. -1 0

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +4 +4

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-2** -4

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +3** -3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

+1** -2

31 Parents with dependent children should be given priority over similar people without dependents. -3** +1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

0** +3

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -1 -1

34 The amount of health care people have had in the past should not influence access to treatments in the future. +1** +2

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

43

Figure 4. II.17 Idealised Q sort for point of view 1 in Palestine

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

There is no sense in saving lives if the quality of

those lives will be really bad.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Parents with dependent children should be given

priority over similar people without dependents.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to younger people, because

they may benefit from treatment for longer.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Priority should be given to those treatments that

generate the most health.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Appendix 4.II

44

Figure 4. II.18 Idealised Q sort for point of view 2 in Palestine

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Priority should be given to those treatments that

generate the most health.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

There is no sense in saving lives if the quality of

those lives will be really bad.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Appendix 4.II

45

Poland

29 participants from Poland completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4. II.19). The statistical features of each factor

solution were examined. The three-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4. II.6 presents the factor

scores of the statements, Figure 4. II.20 to Figure 4. II.22 the idealised Q sorts for

each point of view.

Figure 4. II.19 Correlations between consecutive factor solutions for Polish data

Appendix 4.II

46

For those respondents who load onto factor 1 there is a general rejection of the use

of a patient’s personal characteristics to make decisions on who should be prioritised

for treatment. This can be seen by the high level of agreement which was given to

statements #4 and #29. Opposition to prioritisation according to age is evident from

the rejection of statements #26 and #23. One personal characteristic which does

seem to be supported by this factor is that parents with dependent children should be

given some priority as seen by statement 31 which is distinguishing for this factor. In

addition statements #7 and #13 which also relate to having dependents are towards

the middle of the factor and are distinguishing.

Factor 1

A person’s financial status or their contribution through taxes should not be taken into

account (#3; #5) but they can pay for treatment (#24). There is some indication that

the size of the health gain is important in this factor. When choosing between two

treatments priority should be given to the one which generates the biggest health

gain (#15). This view is somewhat contradicted by the rejection of statements #19

and #32.

Quality of life is also important; treatments should not just be about saving lives if the

quality of life will be poor (#17).

Factor 2 is concerned with access to health care being based on health need rather

than patient characteristics as can be seen with the placing of statement #29 in the

‘most agree’ position. Beyond the agreement with the general principle that access

Factor 2

Appendix 4.II

47

should be based on need, more specific statements relating to financial contribution

(#5; #3), patient age (#26), type of disease (#30), whether they have dependents

(#31; #7; #9) or whether their lifestyle contributed to the illness (#25; #21) were all

rejected.

Medical need is more important in the decision on how to prioritise health care

resources. Priority should be given to treatments that generate the most health (#19;

#15). The severity of a person’s condition should also be taken into account and

priority given to those whose condition is worsening over those in a more stable

condition (#18; #1). In keeping with this emphasis on medical need, a distinguishing

statement for this factor is that doctors should be the ones to make the decision on

who receives treatment based on their medical expertise (#12). Prevention is also

important in this factor (#27).

As with both of the previous factors, there is a general rejection of making decisions

on who should receive priority for treatment based on the personal characteristics of

the patients (#29; #34; #4; #23). However, in the account provided by factor 3 the

lifestyle of a person can be taken into account when making decisions on who should

receive treatment. Individuals should take personal responsibility for their health and

if illness is a result of an unhealthy lifestyle then this should be taken into account.

This can be seen in statements #25 and #21 both of which are distinguishing for this

factor.

Factor 3

Appendix 4.II

48

As can be seen from the least important statements for this factor (see Table 4. II.6),

an individual’s financial contribution should not be taken into the decision on whether

to provide treatment (#5; #3). No distinction should be made between rich and poor

people and positive discrimination should not occur in the provision of treatment

(#16). This is complementary to the statements respondents on factor 3 agreed with

in that treatment should be based on need and not any personal characteristics.

However, the respondents who load onto this factor did agree that people should be

allowed to pay for treatment as long as it does not affect others (#24).

The health gain which arises from treatment is important in this factor with resources

being directed at treatments which result in the largest health gain (#25) and towards

those who would receive the greatest benefit from treatment (#32). The quality of life

people have following treatment is important and saving lives is not the only

consideration if quality of life after will be bad (#33). This is supported by the positive

positioning of statement #32 which states that people who benefit more from

treatment should be given priority. Whilst there is support for providing treatments

which result in an acceptable level of health, it does not lead to the rejection of

statements #8 and #17 and the provision of treatments which save lives.

Appendix 4.II

49

Table 4. II.6 Rank scores of the 34 statements for the three factors in Poland

Statement 1 2 3

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1 +2 +1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

+1 0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-3 -3 -3

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4* +1* +3*

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -3 -4 -4

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

+1 -2* 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

0** -2 -1

8 Rescuing people from a certain death should take priority over all other kinds of health care. +2 +2 0*

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. -1 -2 -1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

+2 +1 +1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 0* -2

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +1 +3** +1

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0* -1 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

0* -1* +2**

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+4 +2** +4

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-2 -1 -2

17 There is no sense in saving lives if the quality of those lives will be really bad. -4** +1** -1**

Appendix 4.II

50

Statement 1 2 3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+3 +2 +1

19 Priority should be given to those treatments that generate the most health. -2** +3* +1*

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -1 -1 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+2* +4* -3**

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

-1* 0* +2*

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2 -1* -3

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. -4* 0** -2*

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

0** -3** +2**

26 Priority should be given to younger people, because they may benefit from treatment for longer. -3 -2 0**

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +1* +3 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

0 0 -1

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +3 +4 +4

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-1 -3 -2

31 Parents with dependent children should be given priority over similar people without dependents. +2** -4** 0**

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

-2** +1 +2

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. 0 0 -4**

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3 +1** +3

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

51

Figure 4. II.20 Idealised Q sort for point of view 1 in Poland

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Parents with dependent children should be given

priority over similar people without dependents.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to those treatments that

generate the most health.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Appendix 4.II

52

Figure 4. II.21 Idealised Q sort for point of view 2 in Poland

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Parents with dependent children should be given

priority over similar people without dependents.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to those treatments that

generate the most health.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

There is no sense in saving lives if the quality of

those lives will be really bad.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Appendix 4.II

53

Figure 4. II.22 Idealised Q sort for point of view 3 in Poland

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

There is no sense in saving lives if the quality of

those lives will be really bad.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to those treatments that

generate the most health.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Priority should be given to younger people, because

they may benefit from treatment for longer.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Parents with dependent children should be given

priority over similar people without dependents.

Appendix 4.II

54

Spain

29 participants from Spain completed the online Q sort. Their data supported a

maximum of four factors (see Figure 4. II.23). The statistical features of each factor

solution were examined. The four-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4. II.7 presents the factor

scores of the statements, Figure 4. II.24 to Figure 4. II.27 the idealised Q sorts for

each point of view.

Figure 4. II.23 Correlations between consecutive factor solutions for Spanish data

Appendix 4.II

55

Before describing each factor in more detail and focussing on the distinguishing

features of each account, there are a number of issues about which the four Spanish

factors are all in agreement. That it is more important to prevent ill health that to cure

it (#27) and that access should be based on need and not other characteristics (#29)

are both ranked highly in all four factors. There is also consensus that consideration

should be given to maximising health gain (statements #15 and #19 having factor

scores of 7 or higher across all factors). In what follows, and with the consensus

issues in mind, each factor is described and attention is paid to the matters that

separate each of the accounts from the other three.

The two statements of most importance to factor 1, (in the positive tail of the factor

array) are also consensus statements (#27 and #29). The positioning of statement

#29 when coupled with #4 (significantly distinguishing this factor at p<0.01) indicates

the rejection of any non-health characteristics of beneficiaries in health care

prioritisation which should instead focus on health and need. Consistent with this

overarching view, factor 1 particularly opposes prioritisation according to age and two

different arguments for giving priority to the young over the old are firmly rejected (a

health gain maximisation argument in statement #26 and a ‘fair innings’ argument in

#23). The health-related characteristics of patients, such as severity, prognosis or

quality of life is regarded as relevant, as shown by agreement with statement #1 and

by the positioning of statements #8 and #33. Interestingly, on first examination, the

placing of these latter two statements could be regarded as counter-intuitive. One

statement (#8) presents a ‘rule of rescue’ type of argument and prioritises life-saving

treatments over all other kinds of health gain, but disagreement with #33 indicates

Factor 1 - “Prevention, health-related need and anti-ageism”

Appendix 4.II

56

that prolonging life is not the only consideration and quality of life is also important.

However, taken together this factor presents a discourse which is focussed on health

and quality of life from a health perspective - other distributional issues do not enter

into it (although the proposition that doctors should be the ones to dictate priorities

(#12) has little resonance here so it is not focussed on medicine, but health more

broadly).

As in other factors, prevention and access according to need are important to factor

2. In contrast with the views about life-saving treatments in F1, here there is a strong

belief that lives should only be saved, or prolonged, if quality of life is good (#17,8)

(#33). Statement #33 meets with the most disagreement and distinguishes F2.

Although quality of life is important, priority should not be given to patients with very

low quality of life if it cannot be improved by very much (#11) or if saving lives results

in low quality (#17). The very low factor scores attached to statements #5, #3 and

#16 reveal strong views about the irrelevance of income and financial contribution to

health care, or to society more broadly, in the prioritisation of health care between

different groups of beneficiaries.

Factor 2 - “Quality of life, the young, and the irrelevance of income and contribution”

Examination of the distinguishing statements for factor 2 reveals an additional feature

of this account – it is the only account in which people with dependent children (#31)

are given priority. This does not appear to be about the burden of care, since #9 and

#13 have little resonance, but rather would appear to be preference for directing

resources towards children and young people (see the positive factor scores for

statements #14 and 26).

Appendix 4.II

57

The key, outstanding feature of factor 3 is the rejection of statements #13, #31 and

#9 in columns 1 and 2. These are all significant, distinguishing statements for factor

3 and relate to issues of dependence, carers and families. According to this account,

such things should not be considered in health care priority setting. Statement 7,

which is a consensus statement, also relates to social factors/ family and meets with

strong disagreement. In the positive columns 8 and 9 are statements that are

positive in other factors too, although there is a notable emphasis on life-saving in #8

and #14. Like factors 2 and 4, there is objection to the notion of people buying

private health care in this account. Unlike the other factors, there is some positive

support for statement #12 (placed in column 7) which argues that doctors should

judge which patients get priority. It would appear that this account regards health

care priority setting as distinct and separate from issues of social and family

circumstance and the burden of care.

Factor 3 - “The burden on carers and families is not relevant to priorities”

Statement 24 is the most important to this factor and, together with #5, #4 and #3

indicates that contribution, either through insurance, taxation or private payments

should not be a determinant of health care priorities. Whilst the magnitude of the

health benefits is important, priorities should be set around need and severity of

disease. Young people should not be given priority (#23) (#4) (#26*) according to

this account and neither should those who are socio economically deprived (#16). In

fact, for factor 4, non-emergency care should not be prioritised, but should operate a

waiting list system (#28). This message of equality of access is extended to those

Factor 4 - “health care priorities should not favour the rich, the poor, or the young and

should be directed towards greatest need in terms of severity”

Appendix 4.II

58

who may be in some way responsible for their own illness and the idea that such

patients should receive lower priority than others is rejected. This liberal view

distinguishes factor 4 and is not important for any of the other three factors (#25).

Appendix 4.II

59

Table 4. II.7 Rank scores of the 34 statements for the four factors in Spain

Statement 1 2 3 4

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+3 0 +1 0

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 -1 0 -1

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-2 -3 -2 -3

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +3** -1** +1 +2

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -3* -4 -2* -4

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

-2* 0 0 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-3 -2 -3 -2

8 Rescuing people from a certain death should take priority over all other kinds of health care. +3* -1* +4* +1*

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. +1 +1 -3** +1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

0 +1 -2 -1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

+1* -3* 0* +3**

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. 0 +1 +2* 0

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. +2 -1 -4** +1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

0 +4 +3 0

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2 +2 +2 +2

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-2 -3 -1* -3

17 There is no sense in saving lives if the quality of those lives will be really bad. -2 +3** -1 -2

Appendix 4.II

60

Statement 1 2 3 4

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 0 +1 +1

19 Priority should be given to those treatments that generate the most health. +2 +2 +2 +3

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -1 0 0 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+1 0 +2 +1

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

-1 +1 +1 0

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -4 0 0 -4

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. -1** +3 +3 +4

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-1 -1 -1 -3**

26 Priority should be given to younger people, because they may benefit from treatment for longer. -3 +2** -4 -2*

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +4 +3 +4 +3

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

+1 -2 -1 +2

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4 +4 +3 +4

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

0 -2* 0 -1

31 Parents with dependent children should be given priority over similar people without dependents. 0 +2** -3* -2

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

-1 -2 -1 -1

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -4 -4* -2 0**

34 The amount of health care people have had in the past should not influence access to treatments in the future. +2 +1 +1 +2

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

61

Figure 4. II.24 Idealised Q sort for point of view 1 in Spain

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to younger people, because

they may benefit from treatment for longer.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

The amount of health care people have had in the

past should not influence access to treatments in the

future.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Priority should be given to those treatments that

generate the most health.

Rescuing people from a certain death should take priority over all other kinds

of health care.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Parents with dependent children should be given

priority over similar people without dependents.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Appendix 4.II

62

Figure 4. II.25 Idealised Q sort for point of view 2 in Spain

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Patient characteristics like age, gender or income should play no role in prioritising between

people.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Priority should be given to those treatments that

generate the most health.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Parents with dependent children should be given

priority over similar people without dependents.

There is no sense in saving lives if the quality of

those lives will be really bad.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Appendix 4.II

63

Figure 4. II.26 Idealised Q sort for point of view 3 in Spain

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Priority should be given to younger people, because

they may benefit from treatment for longer.

Parents with dependent children should be given

priority over similar people without dependents.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to those treatments that

generate the most health.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Appendix 4.II

64

Figure 4. II.27 Idealised Q sort for point of view 4 in Spain

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

There is no sense in saving lives if the quality of

those lives will be really bad.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Rescuing people from a certain death should take priority over all other kinds

of health care.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Parents with dependent children should be given

priority over similar people without dependents.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Priority should be given to those treatments that

generate the most health.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Appendix 4.II

65

Sweden

23 participants from Sweden completed the online Q sort. Their data supported a

maximum of five factors (see Figure 4. II.28). The statistical features of each factor

solution were examined. The three-factor solution was selected because it had a

clearly interpretable account for each factor and was supported by the comments

respondents provided while completing the Q sort. Table 4. II.8 presents the factor

scores of the statements, Figure 4. II.29 to Figure 4. II.31 the idealised Q sorts for

each point of view.

Figure 4. II.28 Correlations between consecutive factor solutions for Swedish data

Appendix 4.II

66

Before describing each factor in more detail and focussing on the distinguishing

features of each account, there are a number of issues about which the three

Swedish factors are all in agreement. For all factors there is agreement that access

to health care should be based on need not geographical, social or economic

circumstances (#29). Some specific personal characteristics which all factors agree

on is that people with partners should not receive priority for treatment (#7) and that

an individual’s financial contribution should not be taken into account (#3 and #5). In

what follows, and with the consensus issues in mind, each factor is described and

attention is paid to the matters that separate each of the accounts from the other two.

The characteristics of Factor 1 are “need and equality”. Respondents associated with

factor 1 believe that a person in need of health care should be prioritised and also

that everyone should have equal right to access health care resources. This can be

seen in the placing of statements #29, #34, #4 and #21 in the ‘most agree’ position

within the factor array. Complementary to this is the disagreement that an

individual’s financial contribution should allow them to receive priority for treatment

(#3; #5).

Factor 1 – “Need and equality”

The distinguishing statements of this factor: #21 “Whether an illness is the result of

an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment

as everyone else” and statement # 25 “People who are in some way responsible for

their own illness should receive lower priority than people who have the same illness

simply due to chance” highlight the importance of equality and that everybody are

equally entitled to health care.

Appendix 4.II

67

For factor 2 health gain and maximising the size of the health effect is important.

Statement 15, which is the top ranked statement for this factor, highlights the view of

those who load onto this factor that funding should be given to treatments that

generate the biggest health gain and should also be targeted at those who will gain

the most from treatment (#32).

Factor 2 - “Health gain and fairness”

Although the placing of statement 29 indicates that personal characteristics should

not be used to prioritise treatment between people there appears to be some

exceptions made for younger people. Statement #4 is distinguishing for factor 2,

being placed in the ‘most disagree’ end of the factor array. Combined with statement

#14 which is placed in the 8 position, suggests younger people should be given

priority. However given that statements #23 and #26 are in the middle of the

distribution this preference may only be when a certain type of health gain is received

rather than a general preference for prioritising younger people.

Factor 3 is an account that is concerned with treatment effectiveness and prevention.

Priority should be given to treatments that generate the most health (#19) is a

distinguishing statement for this factor, as is statement #27 which focuses on

prevention. In combination these two statements may indicate that the respondents

who load onto this factor are considering cost effectiveness for the health care

system.

Factor 3 - “Prevention and effectiveness”

Appendix 4.II

68

As seen in all factors the statements which form the lower section of the factor array

relate to the rejection of an individual’s personal circumstance being used to prioritise

between people.

Appendix 4.II

69

Table 4. II.8 Rank scores of the 34 statements for the three factors in Sweden

Statement 1 2 3

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+2 +2 0*

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

+1 0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -4 -4

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +3 -3** +3

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -3 -3 -4**

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

+1 0 0

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-3 -4 -3

8 Rescuing people from a certain death should take priority over all other kinds of health care. +1 +2 +1

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. -1 +1** -2

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

+1 -1 +1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 0** -3

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +2 0* +3

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. 0* +1** -2*

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

0** +3 +2

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+3 +4 +1**

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-2 -2 -3

17 There is no sense in saving lives if the quality of those lives will be really bad. -1 -1 +1*

Appendix 4.II

70

Statement 1 2 3

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+2 +2 0**

19 Priority should be given to those treatments that generate the most health. +1 +2 +4**

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. 0 0 -1

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+3** +1** -1**

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

0 0 -1*

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2** +1* +2*

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. -2 -2 -2

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-4** -2 -1

26 Priority should be given to younger people, because they may benefit from treatment for longer. 0** +1* +2*

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +2** -1** +4**

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

0 -2** 0

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4 +4 +3

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-2 -3 -2

31 Parents with dependent children should be given priority over similar people without dependents. -1 -1 0

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

-1** +3** +1**

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -3 -1 -1

34 The amount of health care people have had in the past should not influence access to treatments in the future. +4* +3 +2

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

71

Figure 4. II.29 Idealised Q sort for point of view 1 in Sweden

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Priority should be given to those treatments that

generate the most health.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Parents with dependent children should be given

priority over similar people without dependents.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Appendix 4.II

72

Figure 4. II.30 Idealised Q sort for point of view 2 in Sweden

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

There is no sense in saving lives if the quality of

those lives will be really bad.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Priority should be given to those treatments that

generate the most health.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Appendix 4.II

73

Figure 4. II.31 Idealised Q sort for point of view 3 in Sweden

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

Priority should be given to those treatments that

generate the most health.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

Parents with dependent children should be given

priority over similar people without dependents.

There is no sense in saving lives if the quality of

those lives will be really bad.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Appendix 4.II

74

The Netherlands

30 participants from the Netherlands completed the paper Q sort. Their data

supported a maximum of four factors (see Figure 4. II.32). The statistical features of

each factor solution were examined. The two-factor solution was selected because it

had a clearly interpretable account for each factor and was supported by the

comments respondents provided while completing the Q sort. Table 4. II.9 presents

the factor scores of the statements, Figure 4. II.33 and Figure 4. II.34 the idealised Q

sorts for each point of view.

Figure 4. II.32 Correlations between consecutive factor solutions for Dutch data

Appendix 4.II

75

There are a number of issues about which the two Dutch factors are in agreement.

That the amount of health care people have had in the past should not matter (#34),

that doctors should decide on priorities (#12) and that more efficient treatments

should be preferred (#15) are ranked highly in both factors. There is also consensus

that having a partner (#7), being from a lower socio-economic group (#16) or

suffering from a rare disease (#30) should not matter for priority. In what follows, and

with the consensus issues in mind, each factor is described and attention is paid to

the matters that separate each of the accounts from the other three.

The account presented by factor 1 is an egalitarian perspective. Access to healthcare

should be based on need, not on any personal or socio-demographic characteristics

of patients (#4; #29). Having a partner (#7) or dependent children (#31) and possible

responsibility for illness (#21; #25) are rejected as reason for prioritisation. Patients’

history of health services use (#34) should also not be taken into account, because

“people should not be punished for bad luck”. Doctors should be the ones to decide

about priorities (#12), as they have the proper training and access to patient level

information, otherwise patients should be treated on a first come first served basis

(#28) because that is most fair. More than in the other factor, saving lives (#8; #17)

and looking after those worst off (#1) are considered to be an important responsibility

of health services.

Factor 1 – “The egalitarian perspective”

The strong rejection of statements #3 and #5, indicating that a patient’s financial

contribution to the system should not matter for priorities, aligns with this egalitarian

perspective. Like with illness, it is usually not peoples’ choice to be unemployed or

Appendix 4.II

76

financially worse off. People should also not be allowed to buy themselves priority

treatment (#24). Equal access to public services like health care and education is

considered a basic principle of a democratic society.

The people who make up factor 2 find health and quality of life important. Prevention

is ranked highest (#27), simply because it impedes or postpones illness and the

associated loss of health and quality of life. Prevention is also seen as an important

personal responsibility and people who do not look well after themselves should

receive lower priority and perhaps wait a bit longer for treatment than those who are

not to blame for their disease (#21; #25). In any case, priority should go to treatments

that generate most health (#19) and bring the biggest bang for the buck (#15).

Factor 2 – “Health care is about health and quality of life”

Quality of life is also important in this account. Treatments that improve quality of life

are preferred over those that perhaps only prolong life (#33). Saving lives is not seen

as the ultimate goal of health care (#8), but quality of life after treatment should also

be considered (#17; #22); as participants explained, when quality of life is really low,

the patient may be suffering severely or not be conscious of being alive, while the

family experiences a heavy burden.

Another distinction with the account presented by the previous factor is less

opposition to people buying priority treatment (#24) and priority for people in paid

work (#5). On the other hand, priority treatment to people from lower socio-economic

groups is strongly rejected (#16). Treatment on a first come first served basis is not

encouraged (#28).

Appendix 4.II

77

Table 4. II.9 Rank scores of the 34 statements for the two factors in the Netherlands

Statement 1 2

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1** -1

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4** -2

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +3** +1

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -3** -2

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

-1 -1

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-4* -3

8 Rescuing people from a certain death should take priority over all other kinds of health care. +1** -3

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. -3** 0

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future. 0 +1

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-2 -1

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +3 +3

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. -1* +1

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

+1 +1

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2 +2

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-3 -4

17 There is no sense in saving lives if the quality of those lives will be really bad. -1** +2

Appendix 4.II

78

Statement 1 2

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+1 0

19 Priority should be given to those treatments that generate the most health. +1** +4

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. -1 0

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else. +2** -3

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

0** +2

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2 -1

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. +2** -1

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

0** +3

26 Priority should be given to younger people, because they may benefit from treatment for longer. 0 +1

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +2** +4

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

+3** -2

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4** +2

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-2 -2

31 Parents with dependent children should be given priority over similar people without dependents. -2** 0

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

0 0

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. -1** -4

34 The amount of health care people have had in the past should not influence access to treatments in the future. +4 +3

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

79

Figure 4. II.33 Idealised Q sort for point of view 1 in the Netherlands

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

There is no sense in saving lives if the quality of

those lives will be really bad.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Parents with dependent children should be given

priority over similar people without dependents.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Priority should be given to those treatments that

generate the most health.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Appendix 4.II

80

Figure 4. II.34 Idealised Q sort for point of view 2 in the Netherlands

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Priority should be given to those treatments that

generate the most health.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Priority should be given to younger people, because

they may benefit from treatment for longer.

There is no sense in saving lives if the quality of

those lives will be really bad.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Parents with dependent children should be given

priority over similar people without dependents.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Appendix 4.II

81

The UK

32 participants from the UK completed the paper Q sort and 10 the online Q sort,

making a total of 42. Their data supported a maximum of six factors (see Figure 4.

II.35). The statistical features of each factor solution were examined. The four-factor

solution was selected because it had a clearly interpretable account for each factor

and was supported by the comments respondents provided while completing the Q

sort. Table 4. II.10 presents the factor scores of the statements, Figure 4. II.36 to

Figure 4. II.39 the idealised Q sorts for each point of view.

Figure 4. II.35 Correlations between consecutive factor solutions for the UK data

Appendix 4.II

82

The most important statements in factor 1 relate to an egalitarian position of equality

of access and entitlement to health care. Distribution of health care should be driven

by medical need alone (#29; #12; #18 and #28).

Factor 1

The emphasis placed on equality and priority on the basis of need is reinforced by

several of the statements rejected in this account. Disagreement with #28

emphasises that patients should be prioritised on the basis of severity of illness and

the support of doctors’ judgements (#12) implies that need is medically defined and

so consistent with priority for the seriously ill. Rarity of disease is not regarded as an

appropriate criterion for priority setting and this view is shared across the other

factors. Similarly, contribution to the health services (via taxes) is not considered

relevant when distributing health resources. Turning to the statements which

distinguish this factor, statements #21 and #25 are worthy of comment. These relate

to lifestyle and the responsibility of patients for their illness. Their placing in factor 1

reveal an objection to including issues attribution of the cause of illness in health care

priorities.

The account represented by factor 2 is quite different to factor 1. Although the top

ranked statements #12 and #29 are supported in other factors, here there is a focus

on efficiency and maximising benefits that distinguishes this factor from all of the

other accounts. Statement 15 is the clearest demonstration of this view but the

placing of other statements support its interpretation. Statement #26 is supported

only in this factor (significantly distinguishing at p<.01) and contributes to the

Factor 2

Appendix 4.II

83

message about Taken together with qualitative data, the prioritisation of prevention

(#27) is also linked to efficiency in terms of cost savings

Whilst there is consensus that contribution to the national purse is no grounds for

subsequent prioritisation of health care resources, in this account people should be

allowed to purchase health care as long as doing so does not affect others negatively

(#24) – a distinguishing view which sets this factor apart. There is support for young

people and parents with dependents children in this factor too (#26; #31). This would

appear to be strongly linked to the maximisation of benefits (i.e. younger people have

a longer life expectancy and so overall may benefit more from treatment) since

similar statements prioritising young people for other reasons are not salient in this

account (#23 ; #14).

Lastly there is concern about quality of life in factor 2 revealed by the distinctive

placing of #22 and #33.

Factor 3 is a rights based account. Health care access should not be related to

personal characteristics (#29; #4). There is particular emphasis on income (#3; #5;

#16) and age (#23; #26) and statements prioritising on these characteristics are

rejected most strongly. The ‘anti-ageism’, anti-discrimination perspective

distinguishes this factor. Neither arguments of fair innings (#23) and efficiency (#26)

are sufficient to warrant special treatment for the young over the old.

Factor 3

Appendix 4.II

84

People who can afford it should not be permitted to buy health care even though the

argument is qualified in the statement by the fact that others will not be adversely

affected (#24) – which would indicate that this is an account founded on beliefs and

principles.

Factor 4 shares some points of view with other factors. Statement #29 achieves

agreement across all four factors and there is consensus around the rejection of #3

and #30. Statement #34 is generally supported although not statistically significant.

Turning to the distinguishing statements, there are two which appear to contradict

each other (#17; #8). Agreement with statement #17 would imply that quality of life is

important and that preservation of life is not sacrosanct if quality of life is poor.

However, statement #8 (which finds agreement only in factor 4 and is significantly

distinguishing p<.01) states that rescuing people from a certain death should take

priority over all other kinds of health care. This embodies the notion of the ‘rule of

rescue’ and the distinction between the two statements is the mention in #17 of

quality of life being poor and some kind of minimum acceptable threshold below

which life is not worth living.

Factor 4

Factor 4 is the only factor that does not agree with statement #12 – that doctors are

the ones to judge – and that doesn’t support the adage that prevention is better than

cure (#27).

Appendix 4.II

85

Table 4. II.10 Rank scores of the 34 statements for the four factors in the UK

Statement 1 2 3 4

1 If two groups of patients can benefit from a treatment equally and group A’s health is fairly good and group B’s health is poor, group B deserves priority.

+1 0 +1 +2

2 If one treatment results in one life year gained for certain and another in a 50% chance of gaining two life years, priority should be given to the first type of treatment.

+1 -1 0 0

3 People who have contributed more (e.g. through premiums or taxes) to the health care system should be treated with priority over people who have contributed less.

-4 -4 -4 -4

4 Patient characteristics like age, gender or income should play no role in prioritising between people. +4 -3 +4 -2

5 People who are in paid work and so contribute financially to society should be prioritised over people who do not work. -4 -1 -4 -1

6 If a treatment adds one month to the life of a patient and costs 7.500 Euros, one should consider whether the money could have been better spent on other health care.

0 0 0 +1*

7 If two patients are waiting for a transplant organ, one with partner and the other single but otherwise identical, the first organ to become available should go the patient with partner.

-2 -3 -1** -3

8 Rescuing people from a certain death should take priority over all other kinds of health care. -1 -1 -1 +2**

9 Treatment of illnesses that put the highest burden on patients’ families should receive higher priority. 0 +1 +1 +1

10 A treatment which benefits patients in the short-term should have priority over a treatment with similar benefits for patients in the future.

-1 -2 -2 -3

11 Priority should be given to people whose quality of life is low over those whose quality of life is moderate, even if treatment can only improve their quality of life by a small amount.

-1 -1 0 0

12 Doctors should be the ones to judge which patients get priority on the basis of their medical expertise. +3 +4 +2* -2**

13 People who depend heavily on members of their family or neighbours for care should be treated with priority. +1 0 -1 0

14 Adding one year to the end of life for someone who will otherwise die at age 30 is more important than adding one year to the life of someone who otherwise would die at age 80.

-1 +1 -2 +3

15 When having to choose between two treatments that both cost the same, funding should be given to the treatment that results in the biggest health gain.

+2 +3 +1 +4

16 In general, if people from different income groups are suffering from the same condition, people from low income groups should be given priority.

-2 -2 -3 -1*

17 There is no sense in saving lives if the quality of those lives will be really bad. 0 0 0 +3**

Appendix 4.II

86

Statement 1 2 3 4

18 If two people have the same current condition but the health of one of the two is worsening while that of the other is stable, the former should be treated with priority.

+3 0* +2 +2

19 Priority should be given to those treatments that generate the most health. +2 +2 +1 +2

20 It is more important to extend one person’s life by one year than to extend 12 people’s lives by one month. 0 -2 -1 0

21 Whether an illness is the result of an unhealthy lifestyle should not be relevant, everyone is just as worthy of treatment as everyone else.

+2** -1 +1** -2

22 Priority should be given to treatments that restore health to an acceptable level, there’s no use in improving health when the final result is still a very poor state of health.

+1* +2* 0* -3*

23 Younger people should be given priority over older people, because they haven’t had their fair share of health yet. -2* +1 -3** +1

24 People should not be allowed to buy themselves priority treatment, even if it doesn’t affect others negatively. -2 -3* +3** -1

25 People who are in some way responsible for their own illness should receive lower priority than people who have the same illness simply due to chance.

-3** +1 +2 +1

26 Priority should be given to younger people, because they may benefit from treatment for longer. -1 +3** -3* -1

27 It is more important to prevent ill health than it is to cure ill health once it occurs. +2** +3 +3 -1**

28 For non-emergency treatments where there are waiting lists, patients in need of care should be treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the illness).

-3** +1 +2 0**

29 Access to health care should be based on need, not on geographical, social or economic circumstances. +4 +4 +4 +4

30 Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than more common ones.

-3 -4 -2 -4

31 Parents with dependent children should be given priority over similar people without dependents. 0 +2* 0 +1

32 People who benefit more from a treatment, because it is more effective for them, should receive priority over people who benefit less from this treatment.

+1 0 -1 0

33 It is more important to provide treatments that prolong life than treatments that improve quality of life. 0* -2 -2 -2

34 The amount of health care people have had in the past should not influence access to treatments in the future. +3 +2 +3 +3

Note: statement scores range from -4 (‘disagree most’) to +4 (‘agree most’). * distinguishing statement (p<.05). ** distinguishing statement (p<.01).

Appendix 4.II

87

Figure 4. II.36 Idealised Q sort for point of view 1 in the UK

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

There is no sense in saving lives if the quality of

those lives will be really bad.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to those treatments that

generate the most health.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to younger people, because

they may benefit from treatment for longer.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Rescuing people from a certain death should take priority over all other kinds

of health care.

Parents with dependent children should be given

priority over similar people without dependents.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

Appendix 4.II

88

Figure 4. II.37 Idealised Q sort for point of view 2 in the UK

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

Priority should be given to those treatments that

generate the most health.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Rescuing people from a certain death should take priority over all other kinds

of health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Priority should be given to younger people, because

they may benefit from treatment for longer.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Parents with dependent children should be given

priority over similar people without dependents.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

There is no sense in saving lives if the quality of

those lives will be really bad.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

Appendix 4.II

89

Figure 4. II.38 Idealised Q sort for point of view 3 in the UK

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to younger people, because

they may benefit from treatment for longer.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to those treatments that

generate the most health.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Rescuing people from a certain death should take priority over all other kinds

of health care.

There is no sense in saving lives if the quality of

those lives will be really bad.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

Appendix 4.II

90

Figure 4. II.39 Idealised Q sort for point of view 4 in the UK

DISAGREE MOST

AGREE MOST

-4 -3 -2 -1 0 +1 +2 +3 +4

Priority should be given to people with rare diseases, even when these diseases do not necessarily cause more health damage than

more common ones.

Priority should be given to treatments that restore health to an acceptable level, there’s no use in

improving health when the final result is still a very

poor state of health.

Patient characteristics like age, gender or income should play no role in prioritising between

people.

It is more important to prevent ill health than it is to cure ill health once it

occurs.

If one treatment results in one life year gained for certain and another in a

50% chance of gaining two life years, priority should

be given to the first type of treatment.

Treatment of illnesses that put the highest burden on patients’ families should receive higher priority.

If two people have the same current condition but

the health of one of the two is worsening while that of the other is stable, the former should be treated

with priority.

The amount of health care people have had in the

past should not influence access to treatments in the

future.

Access to health care should be based on need, not on geographical, social

or economic circumstances.

People who have contributed more (e.g. through premiums or

taxes) to the health care system should be treated with priority over people

who have contributed less.

A treatment which benefits patients in the short-term

should have priority over a treatment with similar

benefits for patients in the future.

Doctors should be the ones to judge which

patients get priority on the basis of their medical

expertise.

Priority should be given to younger people, because

they may benefit from treatment for longer.

People who benefit more from a treatment, because

it is more effective for them, should receive

priority over people who benefit less from this

treatment.

Parents with dependent children should be given

priority over similar people without dependents.

Priority should be given to those treatments that

generate the most health.

There is no sense in saving lives if the quality of

those lives will be really bad.

When having to choose between two treatments that both cost the same,

funding should be given to the treatment that results in the biggest health gain.

If two patients are waiting for a transplant organ, one with partner and the other

single but otherwise identical, the first organ to become available should

go the patient with partner.

Whether an illness is the result of an unhealthy lifestyle should not be

relevant, everyone is just as worthy of treatment as

everyone else.

People should not be allowed to buy themselves priority treatment, even if it

doesn’t affect others negatively.

People who depend heavily on members of

their family or neighbours for care should be treated

with priority.

If a treatment adds one month to the life of a

patient and costs 7,500€, one should consider

whether the money could have been better spent on

other health care.

If two groups of patients can benefit from a

treatment equally and group A’s health is fairly

good and group B’s health is poor, group B deserves

priority.

Adding one year to the end of life for someone who will otherwise die at age 30 is

more important than adding one year to the life of someone who otherwise

would die at age 80.

It is more important to provide treatments that

prolong life than treatments that improve

quality of life.

In general, if people from different income groups are suffering from the

same condition, people from low income groups should be given priority.

It is more important to extend one person’s life by one year than to extend 12

people’s lives by one month.

Younger people should be given priority over older people, because they

haven’t had their fair share of health yet.

Rescuing people from a certain death should take priority over all other kinds

of health care.

People who are in paid work and so contribute

financially to society should be prioritised over people who do not work.

Priority should be given to people whose quality of life

is low over those whose quality of life is moderate, even if treatment can only improve their quality of life

by a small amount.

People who are in some way responsible for their

own illness should receive lower priority than people who have the same illness

simply due to chance.

For non-emergency treatments where there are

waiting lists, patients in need of care should be

treated on a first come first served basis and not be prioritised in other ways (e.g. the severity of the

illness).

Appendix 4.II

1

Comparing results

Contents

4.III Comparing results ............................................................................................ 2

4.III.1 Methods of comparison ............................................................................ 2

4.III.2 Comparison of national with European general public views .................... 4

4.III.3 Comparison of decision maker with European general public views ...... 17

Index of tables

Table 4. III.1 Rank scores of statements for European and national points of view .......................... 6

Table 4. III.2 Correlations between factor arrays ............................................................................... 7

Table 4. III.3 Correlations between the five European points of view and the 294 Q sorts ............... 8

Table 4. III.4 Example of an analysis of pooled factor arrays .......................................................... 16

Table 4. III.5 Statement rank scores for general public and decision makers points of view ........... 18

Table 4. III.6 Correlations between general public and decision makers factor arrays .................... 19

Table 4. III.7 Correlations of the general public views with the 110 decision makers Q sorts ......... 19

Table 4. III.8 Example of an analysis of general public and decision makers pooled factor arrays . 22

Index of figures

No figures in this appendix.

Appendix 4.III

2

4.III Comparing results

The objective of this appendix is to investigate how the different results obtained in

the Q methodology study relate to each other. The five European points of view

about the prioritisation of health care presented in section 4.2 of this report will be

compared with the points of view generated in the ten national-level analyses

presented in Appendix 4.II (see 4.III.2) and the five decision makers’ points of view

presented in section 4.3 of this report (see 4.III.3). But first, possible methods for

comparison of results from multiple Q studies will be discussed.

4.III.1 Methods of comparison

The comparisons that will be made here concern the results from multiple studies

using the same interview materials in the sense that different groups of participants

(i.e. general public and decision makers samples from Denmark, France, Hungary,

Norway, Palestine, Poland, Spain, Sweden, the Netherlands & the UK) conducted

the exact same Q sort. All participants ranked the same set of opinion statements

according to the same condition of instruction. There are four methods for comparing

this type of results1

1 Based on entries on Q listserv (

using the correlation matrix, factor loadings and factor arrays

from the underlying Q studies as principal sources of information, these are: visual

inspection, correlation analysis, second-order analysis and pooled samples analysis.

http://listserv.kent.edu/archives/q-method.html) and the work of

Robyn et al. (R. Robyn (Ed.) The Changing Face of European Identity. London: Routledge. 2005). We

also wish to acknowledge Noori Akhtar-Danesh and Peter Schmolck for their suggestions.

Appendix 4.III

3

Visual inspection

A quick and straightforward way to compare the points of view from different samples

is comparing the results from the individual studies visually. Bearing in mind that the

number of factors may differ between studies, it is possible to inspect the factor

arrays. If the ranking of statements is similar between two factors from different

studies, these factors may be similar. If, in addition, the distinguishing statements are

similar between studies, the factor solution as a whole may be similar. Using visual

inspection alone will result in a narrative comparison between the factor solutions

based on various data sources.

Correlation analysis

Another quick way to compare results between studies is to compute correlation

coefficients between factor arrays (i.e. the Z- or Rank-scores of statements per

factor). Correlation between factor arrays will show whether factors are similar in

content; if Q-sort patterns are similar between factors, their correlation will be positive

(and statistically significant).

Second-order analysis

A third option for comparing the results between Q studies is to insert the factor

arrays from one study as additional Q sorts in the data file of a second study or to

join together the factor arrays from multiple studies and to re-analyse this new data

set. In the first case, sometimes also called spiking, the correlation matrix shows the

statistical correspondence between the factors from the first study and the Q sorts

from the second study. The factor arrays of the second-order factors show the

Appendix 4.III

4

similarities and distinctions in point of view between both studies. For instance,

factors may emerge that are associated with one or more factors from the first study

and Q sorts from the second study, showing comparable points of view. It is also

possible that factors emerge that are associated with either a factor from the first

study or only Q sorts from the second study, indicating that there was a distinct point

of view in the corresponding study group. When factor arrays are combined, the

correlation matrix will show the statistical correspondence between the factors from

the underlying studies and the factor arrays of the second-order factors provide a

more in-depth view on the similarities and distinctions between the points of view

from the underlying studies.

Pooled samples analysis

A final option for comparing results between Q studies is to join together the Q sorts

from multiple studies and to analyse this pooled data set. The factor arrays resulting

from this analysis can be interpreted on their own and be compared to those from the

underlying studies (e.g. using visual inspection, correlation analysis or spiking).

4.III.2 Comparison of national with European general public views

Table 1 presents the factor arrays of the European and the national points of view

about the prioritisation of health care. As a reminder, the five European points of view

were:

GP1: “Egalitarianism, entitlement and equality of access”

GP2: “Efficiency, severity and the magnitude of health gains”

Appendix 4.III

5

GP3: “Fair innings, young people and maximising health benefits”

GP4: “The intrinsic value of life and healthy living”

GP5: “Quality of life above all else”.

The overview in Table 4.III.1, for instance, shows the very positive sentiment in

almost all points of view toward statement 29 (‘access to health care should be

based on need, not on geographical, social or economic circumstances’) and the

disagreement between viewpoints on statements such as 17 (‘There is no sense in

saving lives if the quality of those lives will be really bad’) and 23 (‘Younger people

should be given priority over older people, because they haven’t had their fair share

of health yet’).

Because visual inspection of so many factor arrays may be somewhat inconvenient,

Table 4. III.2 shows the correlations between all these points of view. The

correlations between the 29 national points of view show a range between 0.01 and

0.84 (average of 0.50). This indicates that there is considerable overlap between

points of view between countries, but also that in certain countries views exist that

are not present in some other countries. The correlations of the five European with

the 29 national points of view show that the European factors represent the national

ones very well. The highest correlation of national with European factors,

representing the best match between a national and one of the European points of

view, ranges between a moderate 0.60 and an excellent 0.97 (with 4 of the 29

national viewpoints with correlations in the range 0.60-0.70, 11 in 0.70-0.80, also 11

in 0.80-0.90 and 3 >0.90).

Appendix 4.III

6

Table 4. III.1 Rank scores of statements for European and national points of view

Nr European level Denmark France Hungary Norway Palestine Poland Spain Sweden NL UK

GP 1

GP 2

GP 3

GP 4

GP 5

1 2 3 1 2 1 2 3 1 2 3 1 2 1 2 3 1 2 3 4 1 2 3 1 2 1 2 3 4

1 +1 +1 +2 -1 -1 +1 -1 0 +1 -1 +1 -2 0 +2 -1 0 +1 +1 +1 +2 +1 +3 0 +1 0 +2 +2 0 +1 -1 +1 0 +1 +2

2 0 0 0 0 -1 0 0 0 0 0 0 0 0 0 +1 0 -1 0 +1 0 0 0 -1 0 -1 +1 0 0 0 0 +1 -1 0 0

3 -4 -4 -4 -2 -4 -3 -3 -4 -4 -3 -4 -1 +1 -4 -4 -2 -3 -3 -3 -3 -3 -2 -3 -2 -3 -4 -4 -4 -4 -2 -4 -4 -4 -4

4 +4 +4 -2 +1 +2 +4 -1 +4 +4 +3 +4 0 +3 0 +4 -2 +4 0 +4 +1 +3 +3 -1 +1 +2 +3 -3 +3 +3 +1 +4 -3 +4 -2

5 -4 -3 -3 0 -3 -4 -2 -3 -4 -2 -4 -1 +2 -3 -3 -2 -2 -4 -3 -4 -4 -3 -4 -2 -4 -3 -3 -4 -3 -2 -4 -1 -4 -1

6 -2 0 +1 0 +1 0 +2 -1 -3 0 +1 +1 -3 +2 0 0 0 0 +1 -2 0 -2 0 0 0 +1 0 0 -1 -1 0 0 0 +1

7 -3 -2 -2 0 -2 -3 -3 -3 -2 -2 -3 0 -4 -4 -3 -1 -2 +1 0 -2 -1 -3 -2 -3 -2 -3 -4 -3 -4 -3 -2 -3 -1 -3

8 +2 +3 +1 +4 -3 0 0 +3 +2 -3 +3 +4 +4 -2 +3 +2 +3 -1 +2 +2 0 +3 -1 +4 +1 +1 +2 +1 +1 -3 -1 -1 -1 +2

9 +1 0 -1 +1 0 0 0 0 +1 +2 0 +1 +1 0 0 0 +2 -1 -1 -2 -1 +1 +1 -3 +1 -1 +1 -2 -3 0 0 +1 +1 +1

10 0 +1 -1 0 0 +1 0 0 0 -1 +1 +1 -1 -2 0 -1 0 0 +2 +1 +1 0 +1 -2 -1 +1 -1 +1 0 +1 -1 -2 -2 -3

11 -1 0 -1 -3 -2 -1 -2 -2 -1 -3 -1 -3 -2 -1 -1 -1 -1 -3 -1 0 -2 +1 -3 0 +3 -1 0 -3 -2 -1 -1 -1 0 0

12 +3 +1 0 +2 +2 +2 +4 +4 +2 +4 +2 +2 +4 0 +1 -2 +2 +2 +1 +3 +1 0 +1 +2 0 +2 0 +3 +3 +3 +3 +4 +2 -2

13 0 +1 0 -1 0 -1 -1 0 +1 +1 0 -1 -1 0 +2 +1 0 +1 0 -1 0 +2 -1 -4 +1 0 +1 -2 -1 +1 +1 0 -1 0

14 0 -1 +3 0 +1 0 +1 -1 -1 +1 -1 0 -3 +3 -1 +2 -3 +2 0 -1 +2 0 +4 +3 0 0 +3 +2 +1 +1 -1 +1 -2 +3

15 +2 +3 +4 +3 +3 +3 +3 +1 +2 +3 +3 +3 +2 +3 +3 +3 +2 +4 +4 +2 +4 +2 +2 +2 +2 +3 +4 +1 +2 +2 +2 +3 +1 +4

16 -3 -3 -2 -3 -2 -4 -2 -4 -2 -2 -3 -3 -2 -1 -3 -3 +1 -2 -2 -1 -2 -2 -3 -1 -3 -2 -2 -3 -3 -4 -2 -2 -3 -1

17 -3 -1 -1 -2 +4 +2 +4 -2 -3 +3 -1 -2 -2 -1 -1 -1 -4 -2 -4 +1 -1 -2 +3 -1 -2 -1 -1 +1 -1 +2 0 0 0 +3

18 +1 +2 +1 +2 0 +1 0 +2 +2 +1 +1 +1 +1 +1 +1 +2 0 +2 +3 +2 +1 +1 0 +1 +1 +2 +2 0 +1 0 +3 0 +2 +2

19 +1 +2 +2 +3 +3 +4 +1 +1 +1 +1 +2 +3 +2 +2 +3 +2 +2 +3 -2 +3 +1 +2 +2 +2 +3 +1 +2 +4 +1 +4 +2 +2 +1 +2

20 -2 -1 +1 -1 -1 -2 -1 +1 -1 0 0 -1 -1 -1 -1 +1 -4 -2 -1 -1 -1 -1 0 0 -1 0 0 -1 -1 0 0 -2 -1 0

21 +2 +2 +2 -4 -2 +1 -4 +2 +3 -1 +1 -4 +1 +1 +2 -1 +3 +3 +2 +4 -3 +1 0 +2 +1 +3 +1 -1 +2 -3 +2 -1 +1 -2

22 +1 +1 -1 +1 +2 +1 +1 -1 0 +2 +2 +2 0 -1 0 +1 0 0 -1 0 +2 -1 +1 +1 0 0 0 -1 0 +2 +1 +2 0 -3

23 -2 -3 +3 -1 0 -1 +2 -1 0 0 -2 -1 -2 +4 -4 0 -2 +1 -2 -1 -3 -4 0 0 -4 -2 +1 +2 -2 -1 -2 +1 -3 +1

24 +3 -2 -2 -4 -1 -2 -1 +1 +3 0 0 -4 +1 +1 +1 -4 +1 -1 -4 0 -2 -1 +3 +3 +4 -2 -2 -2 +2 -1 -2 -3 +3 -1

25 -2 -2 -3 +3 +1 -1 +3 -3 -3 0 -2 +3 0 -3 -2 0 -1 -1 0 -3 +2 -1 -1 -1 -3 -4 -2 -1 0 +3 -3 +1 +2 +1

26 0 -4 +2 -1 0 -1 +1 -1 -1 -1 -2 0 -4 +3 -2 +1 -1 0 -3 -2 0 -3 +2 -4 -2 0 +1 +2 0 +1 -1 +3 -3 -1

27 +3 +2 0 +4 +3 +2 +3 +3 +1 +4 +2 +4 +2 0 +2 +4 +4 +4 +1 +3 +3 +4 +3 +4 +3 +2 -1 +4 +2 +4 +2 +3 +3 -1

28 0 0 -4 -2 -1 0 -2 -2 -2 -4 -3 -2 -3 -3 0 -4 -2 -4 0 0 -1 +1 -2 -1 +2 0 -2 0 +3 -2 -3 +1 +2 0

29 +4 +4 +4 +2 +4 +3 +1 +3 +4 +2 +4 +2 +3 +4 +4 +3 +3 -3 +3 +4 +4 +4 +4 +3 +4 +4 +4 +3 +4 +2 +4 +4 +4 +4

30 -1 0 -3 -3 -3 -2 -3 -2 -1 -2 -1 -3 -1 -2 -2 -3 +1 -2 -1 -3 -2 0 -2 0 -1 -2 -3 -2 -2 -2 -3 -4 -2 -4

31 -1 -2 0 +2 +1 -2 +2 0 0 +1 -2 +2 0 +1 +2 +3 -3 +1 +2 -4 0 0 +2 -3 -2 -1 -1 0 -2 0 0 +2 0 +1

32 -1 -1 +3 +1 +1 +2 +2 +1 0 -1 0 +1 -1 +2 0 +4 0 +3 -2 +1 +2 -1 -2 -1 -1 -1 +3 +1 0 0 +1 0 -1 0

33 -1 -1 0 -2 -4 -3 -4 +2 -2 -4 -1 -2 0 -2 -2 -3 -1 -1 0 0 -4 -4 -4 -2 0 -3 -1 -1 -1 -4 0 -2 -2 -2

34 +2 +3 +1 +1 +2 +3 0 +2 +3 +2 +3 0 +3 +1 +1 +1 +1 +2 +3 +1 +3 +2 +1 +1 +2 +4 +3 +2 +4 +3 +3 +2 +3 +3

Appendix 4.III

7

Table 4. III.2 Correlations between factor arrays

DK1 DK2 DK3 FR1 FR2 HU1 HU2 HU3 NO1 NO2 NO3 PA1 PA2 PO1 PO2 PO3 SP1 SP2 SP3 SP4 SE1 SE2 SE3 NL1 NL2 UK1 UK2 UK3 UK4

GP1 0.70 0.21 0.84 0.91 0.52 0.84 0.31 0.61 0.46 0.82 0.26 0.78 0.42 0.57 0.72 0.60 0.75 0.54 0.64 0.80 0.77 0.47 0.61 0.84 0.43 0.73 0.42 0.76 0.22

GP2 0.79 0.14 0.74 0.75 0.45 0.91 0.37 0.60 0.26 0.84 0.34 0.71 0.36 0.76 0.75 0.63 0.84 0.31 0.59 0.74 0.84 0.49 0.52 0.71 0.39 0.78 0.29 0.69 0.34

GP3 0.49 0.39 0.51 0.48 0.34 0.52 0.31 0.11 0.86 0.39 0.71 0.22 0.60 0.33 0.52 0.43 0.26 0.51 0.37 0.26 0.57 0.89 0.59 0.39 0.29 0.59 0.54 0.12 0.61

GP4 0.51 0.62 0.44 0.29 0.51 0.50 0.97 0.51 0.17 0.49 0.71 0.34 0.44 0.49 0.25 0.71 0.44 0.34 0.30 0.17 0.36 0.37 0.56 0.32 0.60 0.43 0.58 0.36 0.41

GP5 0.78 0.81 0.38 0.39 0.86 0.57 0.54 0.21 0.55 0.54 0.55 0.25 0.44 0.27 0.43 0.78 0.40 0.76 0.32 0.35 0.55 0.46 0.73 0.54 0.86 0.64 0.72 0.59 0.53

DK1 0.64 0.66 0.82 0.46 0.44 0.50 0.73 0.47 0.54 0.52 0.50 0.75 0.76 0.66 0.56 0.52 0.58 0.80 0.58 0.79 0.75 0.70 0.79 0.53 0.68 0.44

DK2 0.12 0.71 0.33 0.66 0.14 0.46 0.28 0.56 0.5 0.40 0.90 0.19 0.61 0.14 0.63 0.24 0.10 0.26 0.40 0.63 0.30 0.75 0.33 0.71 0.29 0.51

DK3 0.82 0.49 0.83 0.36 0.66 0.43 0.77 0.36 0.64 0.49 0.59 0.72 0.47 0.58 0.39 0.56 0.62 0.73 0.49 0.66 0.73 0.35 0.81 0.34 0.61 0.22

FR1 0.81 0.20 0.61 0.53 0.79 0.31 0.73 0.47 0.58 0.69 0.51 0.70 0.47 0.56 0.69 0.76 0.53 0.52 0.71 0.32 0.76 0.30 0.68 0.27

FR2 0.59 0.47 0.40 0.47 0.52 0.46 0.34 0.52 0.31 0.36 0.66 0.40 0.72 0.34 0.29 0.54 0.34 0.59 0.44 0.81 0.68 0.58 0.56 0.36

HU1 0.46 0.82 0.44 0.73 0.46 0.65 0.75 0.70 0.72 0.49 0.65 0.69 0.84 0.58 0.64 0.73 0.51 0.84 0.37 0.67 0.34

HU2 0.16 0.44 0.71 0.29 0.41 0.41 0.20 0.68 0.36 0.37 0.24 0.12 0.30 0.34 0.54 0.25 0.60 0.36 0.59 0.26 0.32

HU3 0.60 0.59 0.18 0.63 0.18 0.43 0.48 0.32 0.56 0.14 0.53 0.41 0.45 0.22 0.30 0.49 0.31 0.50 0.21 0.48 0.15

NO1 0.27 0.52 0.21 0.39 0.43 0.25 0.63 0.35 0.31 0.55 0.77 0.60 0.43 0.37 0.55 0.57 0.26 0.59

NO2 0.65 0.40 0.64 0.64 0.62 0.80 0.49 0.51 0.76 0.77 0.47 0.56 0.71 0.45 0.76 0.39 0.72 0.38

NO3 0.16 0.58 0.34 0.28 0.65 0.41 0.49 0.24 0.18 0.43 0.69 0.51 0.22 0.53 0.49 0.59 0.17 0.55

PA1 0.49 0.62 0.43 0.67 0.19 0.54 0.61 0.60 0.28 0.39 0.54 0.21 0.57 0.21 0.56 0.30

PA2 0.39 0.45 0.49 0.28 0.42 0.31 0.22 0.47 0.49 0.54 0.35 0.43 0.57 0.40 0.26 0.20

PO1 0.62 0.16 0.34 0.40 0.72 0.36 0.39 0.54 0.19 0.63 0.25 0.52 0.27

PO2 0.61 0.39 0.68 0.65 0.76 0.55 0.61 0.74 0.32 0.74 0.36 0.57 0.30

PO3 0.64 0.57 0.41 0.44 0.66 0.52 0.67 0.64 0.75 0.65 0.58 0.61 0.46

SP1 0.74 0.44 0.44 0.64 0.42 0.61 0.32 0.68 0.38

SP2 0.51 0.49 0.64 0.51 0.67 0.46 0.60 0.46 0.49

SP3 0.55 0.39 0.50 0.66 0.27 0.43 0.23 0.52 0.32

SP4 0.62 0.41 0.37 0.68 0.31 0.58 0.26 0.74 0.29

SE1 0.79 0.44 0.84 0.48 0.62 0.42

SE2 0.49 0.37 0.59 0.62 0.25 0.74

SE3 0.71 0.66 0.64 0.63 0.47 0.43

NL1 0.66 0.49 0.76 0.37

NL2 0.52 0.68 0.51 0.37

Appendix 4.III

8

Table 4. III.3 shows the results of a second-order analysis of the original data-set of

294 Q sorts to which the factor arrays of the five European points of view were added

(i.e. spiking). The highest correlation of a Q sort with any one of the five European

points of view varies between 0.25 and 0.87 (average 0.62). For 65% of the Q sorts

the correlation is higher than 0.60 on at least one factor and for 7% it is lower than

0.50 on all factors, indicating that the large majority of Q sorts is well represented.

Table 4. III.3 Correlations between the five European points of view and the 294 Q sorts

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

1 dk_id1 0.57 0.56 0.32 0.39 0.81

2 dk_id2 0.44 0.62 0.32 0.34 0.58

3 dk_id3 0.55 0.72 0.46 0.48 0.61

4 dk_id4 0.47 0.58 0.52 0.46 0.48

5 dk_id5 0.60 0.10 0.31 0.51 0.68

6 dk_id6 0.13 0.21 0.43 0.59 0.57

7 dk_id7 0.60 0.15 0.54 0.29 0.66

8 dk_id8 0.24 0.15 0.27 0.38 0.31

9 dk_id9 0.72 0.72 0.48 0.40 0.72

10 dk_id10 0.38 0.29 0.65 0.32 0.39

11 dk_id11 0.24 0.19 0.28 0.52 0.72

12 dk_id12 0.31 0.47 0.80 0.29 0.52

13 dk_id13 0.55 0.71 0.33 0.47 0.61

14 dk_id14 0.31 0.33 0.40 0.25 0.71

15 dk_id15 0.34 0.14 0.45 0.37 0.64

16 dk_id17 0.32 0.22 0.32 0.22 0.65

17 dk_id18 0.81 0.64 0.50 0.51 0.56

18 dk_id19 0.43 0.55 0.15 0.18 0.46

19 dk_id20 0.17 -0.10 0.28 0.49 0.65

20 dk_id22 0.82 0.68 0.31 0.14 0.21

21 dk_id23 0.71 0.74 0.32 0.43 0.62

22 dk_id24 0.36 0.48 0.15 0.70 0.38

23 dk_id25 -0.17 -0.14 0.14 0.28 0.60

24 dk_id26 0.90 0.21 0.34 0.71 0.47

25 dk_id27 0.72 0.70 0.50 0.37 0.31

26 fr_id2 0.71 0.61 0.18 0.24 0.32

27 fr_id3 0.63 0.45 0.32 0.49 0.28

28 fr_id4 0.77 0.78 0.28 0.41 0.28

Appendix 4.III

9

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

29 fr_id5 0.30 0.24 0.30 0.49 0.68

30 fr_id6 0.47 0.25 0.90 0.30 0.17

31 fr_id7 0.67 0.70 0.30 0.22 0.26

32 fr_id9 0.51 0.60 0.36 0.22 0.28

33 fr_id10 0.80 0.73 0.47 0.52 0.54

34 fr_id11 0.82 0.63 0.42 0.25 0.46

35 fr_id12 0.49 0.52 0.52 0.69 0.61

36 fr_id13 0.49 0.52 0.31 0.51 0.74

37 fr_id14 0.70 0.52 0.42 0.11 0.32

38 fr_id16 0.46 0.26 0.12 0.26 0.15

39 fr_id17 0.69 0.55 0.39 0.27 0.71

40 fr_id18 0.74 0.59 0.36 0.10 0.38

41 fr_id19 0.64 0.64 0.52 0.42 0.34

42 fr_id20 0.81 0.71 0.47 0.39 0.29

43 fr_id21 0.16 0.22 0.12 0.8 0.60

44 fr_id22 0.84 0.56 0.52 0.17 0.47

45 fr_id23 0.21 0.70 0.33 0.39 0.69

46 fr_id24 0.54 0.39 0.34 0.30 0.41

47 fr_id26 0.59 0.56 0.15 0.24 0.24

48 fr_id28 0.22 0.21 0.12 0.21 0.49

49 fr_id29 0.67 0.54 0.44 0.30 0.32

50 fr_id30 0.77 0.64 0.59 0.24 0.22

51 fr_id31 0.46 0.35 0.57 0.19 0.52

52 fr_id32 0.80 0.64 0.21 0.15 0.20

53 fr_id33 0.69 0.50 0.39 0.39 0.47

54 hu_id1 0.60 0.44 0.39 0.64 0.38

55 hu_id2 0.61 0.75 0.36 0.21 0.12

56 hu_id3 0.59 0.66 0.27 0.43 0.55

57 hu_id4 0.60 0.64 0.16 0.54 0.52

58 hu_id5 0.59 0.49 0.30 0.24 0.60

59 hu_id6 0.32 0.37 0.9 0.81 0.59

60 hu_id7 0.38 0.39 0.35 0.17 0.32

61 hu_id9 0.39 0.41 0.60 0.59 0.62

62 hu_id11 0.74 0.86 0.34 0.39 0.54

63 hu_id13 0.44 0.51 0.31 0.42 0.30

64 hu_id15 0.60 0.55 0.27 0.77 0.54

65 hu_id16 0.46 0.29 0.90 0.32 0.28

66 hu_id18 0.10 0.19 -0.22 0.38 0.19

67 hu_id20 0.62 0.51 0.47 0.50 0.37

68 hu_id21 0.33 0.64 0.14 0.24 0.20

69 hu_id22 0.26 0.22 0.28 0.81 0.45

70 hu_id24 0.70 0.12 -0.24 0.41 0.40

Appendix 4.III

10

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

71 hu_id25 0.36 0.40 0.25 0.39 0.22

72 hu_id27 0.37 0.51 0.51 0.41 0.38

73 hu_id28 0.51 0.47 0.25 0.40 0.14

74 hu_id29 0.79 0.74 0.29 0.20 0.17

75 hu_id30 0.28 0.35 0.21 0.11 0.20

76 hu_id31 0.31 0.45 0.10 0.64 0.17

77 hu_id33 0.47 0.43 0.36 0.85 0.37

78 hu_id34 0.54 0.59 0.33 0.31 0.29

79 hu_id35 0.68 0.59 0.56 0.59 0.59

80 hu_id36 0.80 0.76 0.54 0.42 0.53

81 hu_id37 0.41 0.46 0.00 0.59 0.44

82 hu_id38 0.61 0.62 0.20 0.36 0.16

83 hu_id39 0.39 0.63 0.28 0.44 0.49

84 hu_id40 0.79 0.86 0.49 0.51 0.61

85 hu_id41 -0.19 -0.17 0.10 0.57 0.13

86 hu_id42 0.54 0.64 0.21 0.49 0.26

87 hu_id43 0.14 0.12 0.37 0.64 0.38

88 hu_id45 0.16 0.29 0.0 -0.25 -0.23

89 hu_id46 0.5 0.22 0.29 0.60 0.18

90 hu_id47 0.61 0.52 0.11 0.22 0.15

91 hu_id48 0.29 0.32 0.29 0.82 0.47

92 hu_id49 0.12 0.31 0.43 0.60 0.22

93 hu_id50 0.57 0.75 0.24 0.56 0.27

94 hu_id51 0.71 0.78 0.38 0.22 0.26

95 hu_id52 0.6 0.20 0.11 0.71 0.22

96 hu_id53 0.63 0.73 0.19 0.51 0.23

97 hu_id55 0.43 0.47 0.13 0.68 0.22

98 hu_id56 0.29 0.54 0.47 0.46 0.31

99 hu_id57 0.37 0.38 0.57 0.61 0.56

100 hu_id58 0.69 0.72 0.57 0.46 0.43

101 nl_id1 0.51 0.43 0.19 0.2 0.29

102 nl_id2 0.57 0.54 0.19 0.41 0.32

103 nl_id3 0.40 0.43 0.18 0.34 0.62

104 nl_id4 0.41 0.41 0.16 0.60 0.66

105 nl_id5 0.49 0.31 0.36 0.31 0.56

106 nl_id6 0.32 0.41 0.40 0.43 0.61

107 nl_id7 0.58 0.41 0.30 0.14 0.41

108 nl_id8 0.50 0.46 0.28 0.40 0.43

109 nl_id9 0.41 0.29 0.5 -0.10 0.70

110 nl_id10 0.66 0.49 0.34 0.56 0.67

111 nl_id11 0.54 0.55 0.27 0.38 0.46

112 nl_id12 0.49 0.62 0.22 0.13 0.20

Appendix 4.III

11

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

113 nl_id13 0.54 0.50 0.50 0.25 0.54

114 nl_id14 0.64 0.47 0.16 0.28 0.48

115 nl_id15 0.77 0.72 0.47 0.59 0.53

116 nl_id16 -0.40 -0.90 0.60 0.37 0.63

117 nl_id17 0.28 0.36 0.12 0.41 0.68

118 nl_id18 0.47 0.69 0.6 0.39 0.29

119 nl_id19 0.87 0.71 0.30 0.30 0.36

120 nl_id20 0.64 0.40 0.44 0.20 0.44

121 nl_id21 0.36 0.24 0.14 -0.70 0.10

122 nl_id22 0.66 0.42 0.13 0.16 0.40

123 nl_id23 0.44 0.32 0.24 0.41 0.80

124 nl_id24 0.74 0.77 0.37 0.39 0.71

125 nl_id25 0.71 0.54 0.36 0.33 0.55

126 nl_id26 0.90 0.34 -0.30 0.11 0.15

127 nl_id27 0.63 0.77 0.45 0.38 0.46

128 nl_id28 0.63 0.55 0.39 0.25 0.47

129 nl_id29 0.68 0.39 0.24 0.6 0.40

130 nl_id30 0.71 0.44 0.32 0.35 0.28

131 no_id1 0.50 0.53 0.17 0.57 0.59

132 no_id2 0.56 0.68 0.36 0.62 0.58

133 no_id3 0.31 0.24 0.43 0.60 0.43

134 no_id4 -0.10 0.0 0.44 0.37 0.41

135 no_id5 0.25 0.34 0.36 0.63 0.51

136 no_id6 0.24 0.32 0.68 0.37 0.33

137 no_id7 0.39 0.49 0.38 0.19 0.16

138 no_id8 0.25 0.48 0.55 0.52 0.37

139 no_id9 0.74 0.77 0.36 0.44 0.47

140 no_id10 0.41 0.24 0.78 0.90 0.41

141 no_id11 0.31 0.18 0.70 0.31 0.20

142 no_id12 0.39 0.17 0.61 0.18 0.71

143 no_id13 0.18 0.22 0.73 0.50 0.31

144 no_id14 -0.10 -0.30 0.61 0.47 0.50

145 no_id15 0.47 0.19 0.51 0.39 0.39

146 no_id16 0.49 0.45 0.69 0.16 0.63

147 no_id17 0.86 0.77 0.34 0.28 0.49

148 no_id18 0.39 0.42 0.27 0.25 0.68

149 no_id19 0.32 0.12 0.31 0.16 0.16

150 no_id20 0.36 0.39 0.68 0.63 0.41

151 no_id23 0.59 0.59 0.54 0.28 0.37

152 no_id24 0.49 0.37 0.73 0.31 0.56

153 no_id25 0.22 0.13 0.64 -0.10 0.17

154 pa_id1 0.42 0.32 0.30 0.34 0.47

Appendix 4.III

12

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

155 pa_id2 0.26 0.22 0.53 0.29 0.21

156 pa_id3 0.61 0.60 0.13 0.13 0.12

157 pa_id4 0.40 -0.90 0.34 0.11 0.32

158 pa_id5 0.45 0.44 0.17 0.19 -0.10

159 pa_id6 0.60 0.41 0.38 0.32 0.30

160 pa_id7 0.54 0.44 0.29 0.37 0.17

161 pa_id8 0.21 0.41 0.26 0.16 0.12

162 pa_id9 0.44 0.22 0.11 0.21 0.11

163 pa_id10 0.40 0.24 0.57 0.26 0.14

164 pa_id11 0.62 0.50 0.32 0.27 0.44

165 pa_id12 0.73 0.71 0.33 0.47 0.33

166 pa_id13 0.56 0.42 -0.40 0.32 0.24

167 pa_id14 0.32 0.26 0.30 0.29 0.70

168 pa_id15 0.62 0.66 0.37 0.34 0.41

169 pa_id16 0.21 0.19 0.28 0.17 0.49

170 pa_id17 0.28 0.34 0.21 0.34 0.22

171 pa_id18 0.53 0.43 -0.14 0.17 0.12

172 pa_id19 0.60 0.65 0.10 0.11 -0.10

173 pa_id20 0.43 0.37 0.90 -0.20 0.19

174 po_id2 0.44 0.12 0.28 0.37 0.47

175 po_id3 0.45 0.60 0.21 -0.5 0.22

176 po_id4 0.82 0.78 0.59 0.50 0.56

177 po_id5 0.57 0.59 0.52 0.30 0.39

178 po_id6 0.62 0.65 0.17 0.34 0.40

179 po_id7 0.35 0.44 0.12 0.19 0.51

180 po_id8 0.37 0.50 0.14 0.19 0.12

181 po_id9 0.59 0.69 0.15 0.17 0.70

182 po_id10 0.33 0.48 0.39 0.35 0.19

183 po_id11 0.11 0.24 0.35 0.50 0.67

184 po_id12 0.54 0.59 0.41 0.71 0.49

185 po_id13 0.56 0.57 0.32 0.47 0.36

186 po_id14 0.12 0.31 0.12 0.70 0.6

187 po_id15 0.12 0.31 0.22 0.37 0.19

188 po_id16 0.44 0.71 0.18 0.28 0.2

189 po_id17 0.65 0.68 0.21 0.59 0.47

190 po_id18 0.59 0.69 0.34 0.63 0.54

191 po_id19 0.36 0.47 0.31 0.41 0.25

192 po_id21 0.34 0.52 0.38 0.68 0.31

193 po_id22 0.57 0.72 0.44 0.26 0.41

194 po_id23 0.36 0.40 0.31 0.62 0.46

195 po_id24 0.56 0.35 0.44 0.28 0.25

196 po_id25 0.28 0.16 0.10 -0.10 0.24

Appendix 4.III

13

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

197 po_id26 0.40 0.39 0.11 -0.10 0.50

198 po_id27 0.58 0.41 0.46 0.57 0.29

199 po_id28 0.24 0.12 0.40 0.16 0.25

200 po_id29 0.52 0.31 0.60 0.15 0.10

201 po_id31 0.14 0.29 0.42 0.29 0.65

202 po_id32 0.34 0.43 0.36 0.61 0.45

203 se_id1 0.77 0.72 0.34 0.19 0.35

204 se_id2 0.54 0.56 0.74 0.21 0.36

205 se_id4 0.50 0.59 0.56 0.41 0.61

206 se_id5 0.32 0.36 0.73 0.47 0.36

207 se_id6 0.40 0.26 0.38 0.34 0.74

208 se_id8 0.65 0.71 0.62 0.39 0.46

209 se_id10 0.59 0.47 0.18 0.20 0.26

210 se_id11 0.59 0.77 0.38 0.32 0.31

211 se_id12 0.53 0.42 0.56 0.53 0.74

212 se_id13 0.47 0.41 0.29 0.32 0.11

213 se_id14 0.46 0.39 0.46 0.44 0.21

214 se_id16 0.29 0.36 0.48 0.24 0.24

215 se_id18 0.33 0.39 0.87 0.21 0.44

216 se_id19 0.11 0.30 0.22 0.32 0.32

217 se_id20 0.22 0.21 0.44 0.45 0.46

218 se_id21 0.73 0.69 0.62 0.21 0.37

219 se_id22 0.34 0.56 0.37 0.36 0.40

220 se_id23 0.47 0.60 0.25 0.28 0.41

221 se_id24 0.43 0.52 0.14 0.50 0.62

222 se_id26 0.61 0.52 0.51 0.39 0.76

223 se_id27 0.42 0.44 0.64 0.32 0.68

224 se_id30 0.59 0.45 0.27 0.24 0.29

225 se_id31 0.50 0.64 0.51 0.64 0.54

226 sp_id1 0.66 0.55 0.24 0.29 0.47

227 sp_id2 0.65 0.66 0.56 0.35 0.68

228 sp_id3 0.58 0.55 0.19 0.90 0.12

229 sp_id4 0.59 0.53 0.4 0.60 0.14

230 sp_id5 0.25 -0.20 0.33 0.39 0.48

231 sp_id7 0.78 0.66 0.38 0.13 0.29

232 sp_id8 0.69 0.66 0.18 0.90 0.40

233 sp_id9 0.66 0.70 0.34 0.54 0.48

234 sp_id10 0.30 0.12 0.37 0.14 0.67

235 sp_id11 0.45 0.43 0.24 0.27 0.49

236 sp_id12 0.49 0.37 0.23 0.20 0.16

237 sp_id13 0.55 0.25 0.35 0.22 0.51

238 sp_id14 0.65 0.73 0.34 0.36 0.44

Appendix 4.III

14

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

239 sp_id15 0.48 0.54 0.31 0.9 0.48

240 sp_id16 0.82 0.74 0.14 0.28 0.21

241 sp_id17 0.53 0.81 0.24 0.45 0.38

242 sp_id18 0.54 0.52 0.20 0.44 0.24

243 sp_id19 0.54 0.62 0.29 0.24 0.16

244 sp_id20 0.58 0.74 0.27 0.13 0.14

245 sp_id21 0.28 0.27 0.39 0.32 0.13

246 sp_id22 0.66 0.55 0.24 0.22 0.37

247 sp_id23 0.77 0.82 0.17 0.32 0.39

248 sp_id24 0.63 0.59 0.26 0.26 0.37

249 sp_id25 0.74 0.54 0.38 0.11 0.22

250 sp_id26 0.71 0.86 0.39 0.54 0.45

251 sp_id27 0.69 0.73 0.25 0.35 0.51

252 sp_id28 0.73 0.65 0.49 0.19 0.35

253 sp_id29 0.29 0.36 0.10 0.19 -0.30

254 sp_id30 0.70 0.74 0.11 0.28 0.12

255 uk_id1 0.51 0.47 0.48 0.41 0.62

256 uk_id2 0.54 0.39 0.3 0.18 0.46

257 uk_id3 0.44 0.57 0.39 0.44 0.71

258 uk_id4 0.48 0.55 0.20 0.21 0.58

259 uk_id5 0.66 0.57 0.14 0.28 0.42

260 uk_id6 0.61 0.61 0.30 0.26 0.14

261 uk_id7 0.14 0.17 0.26 0.50 0.72

262 uk_id8 0.24 0.40 0.30 0.32 0.36

263 uk_id9 0.69 0.61 0.16 0.31 0.52

264 uk_id10 0.69 0.64 0.42 0.62 0.59

265 uk_id11 0.49 0.31 0.53 0.48 0.65

266 uk_id13 0.57 0.57 0.27 0.18 0.60

267 uk_id14 0.64 0.74 0.60 0.31 0.47

268 uk_id15 0.54 0.34 0.29 0.37 0.41

269 uk_id16 0.47 0.52 0.12 0.21 0.17

270 uk_id17 0.46 0.46 0.37 0.14 0.68

271 uk_id18 0.17 0.26 0.43 0.32 0.39

272 uk_id19 0.43 0.44 0.28 0.24 0.35

273 uk_id20 0.60 0.49 0.73 0.39 0.46

274 uk_id21 0.76 0.55 0.43 0.44 0.53

275 uk_id22 0.47 0.47 0.63 0.52 0.48

276 uk_id23 0.50 0.24 0.14 0.21 0.41

277 uk_id24 0.46 0.46 0.71 0.54 0.68

278 uk_id26 0.41 0.46 -0.20 0.24 0.32

279 uk_id27 0.37 0.47 0.67 0.43 0.39

280 uk_id28 0.56 0.60 0.46 0.61 0.73

Appendix 4.III

15

Q sort Point of view

GP1 GP2 GP3 GP4 GP5

281 uk_id29 0.63 0.67 0.61 0.36 0.56

282 uk_id30 0.46 0.65 0.31 0.31 0.54

283 uk_id31 0.15 0.19 0.68 0.32 0.46

284 uk_id32 0.69 0.67 0.47 0.22 0.52

285 uk_id33 0.51 0.55 -0.30 0.19 0.27

286 uk_id34 0.47 0.59 0.37 0.13 0.56

287 uk_id35 0.60 0.30 0.17 0.36 0.61

288 uk_id36 0.26 0.47 0.33 0.46 0.51

289 uk_id37 0.65 0.76 0.64 0.32 0.57

290 uk_id38 0.23 0.28 0.33 0.57 0.59

291 uk_id39 0.36 0.52 0.43 0.23 0.76

292 uk_id40 0.62 0.81 0.37 0.32 0.67

293 uk_id41 0.56 0.61 0.49 0.44 0.61

294 uk_id42 0.47 0.61 0.24 0.62 0.54

Table 4. III.4 shows the results of an analysis of pooled factor arrays, which gives an

impression of how the five European and 29 national points of view relate. As a

means of illustration a three-factor solution was imposed. Although these meta

clusters have not been interpreted, from their correlations across the five European

points of view it can be inferred that cluster A most probably represents concerns for

equality of access and the magnitude of health gains, cluster B fair innings and

maximising benefits, and cluster C a high valuation of health and quality of life. The

correlations with the national views illustrate how these clusters are represented

across the national samples in each country.

Appendix 4.III

16

Table 4. III.4 Example of an analysis of pooled factor arrays

Point of view Meta clusters

A B C

GP1: “Egalitarianism, entitlement and equality of access” 0.88 * 0.28 0.15

GP2: “Efficiency, severity and the magnitude of health gains” 0.91 * 0.06 0.26

GP3: “Fair innings, young people and maximising health benefits” 0.21 0.86 * 0.19

GP4: “The intrinsic value of life and healthy living” 0.28 -0.02 0.82 *

GP5: “Quality of life above all else” 0.27 0.38 0.74 *

DK1 0.64 * 0.35 0.49

DK2 -0.05 0.37 0.81 *

DK3 0.79 * 0.31 0.17

FR1 0.83 * 0.33 0.09

FR2 0.35 0.27 0.64 *

HU1 0.84 * 0.24 0.34

HU2 0.19 -0.02 0.83 *

HU3 0.66 * -0.09 0.21

NO1 0.19 0.86 * 0.18

NO2 0.81 * 0.17 0.33

NO3 0.10 0.47 0.66 *

PA1 0.79 * 0.04 0.09

PA2 0.29 0.39 0.39

PO1 0.63 * 0.07 0.24

PO2 0.75 * 0.40 0.05

PO3 0.47 0.19 0.73 *

SP1 0.79 * 0.05 0.29

SP2 0.24 0.54 * 0.48

SP3 0.61 * 0.27 0.13

SP4 0.79 * 0.17 0.02

SE1 0.77 * 0.39 0.25

SE2 0.32 0.74 * 0.26

SE3 0.43 0.50 0.49

NL1 0.75 * 0.37 0.18

NL2 0.24 0.23 0.76 *

UK1 0.71 * 0.35 0.37

UK2 0.14 0.51 0.63 *

UK3 0.74 * 0.10 0.27

UK4 0.15 0.47 * 0.42

Note: * p<.01

Appendix 4.III

17

4.III.3 Comparison of decision maker with European general public views

Table 4. III.5 presents the factor arrays of the European general public and decision

makers’ points of view about the prioritisation of health care. Visual inspection of the

statement rank scores shows, for instance, that the difference between GP1 and

DM3 is limited (i.e. the difference in score of 26 of the 34 statements is zero or one).

Furthermore, it is interesting to see that the rankings of the statements referring to

the role of doctors in priority setting (#12) and lifestyle / culpability (#21; #25) seem to

be quite different between decision makers and the general public. There also seems

to be less resistance among decision makers to take account of productivity in

prioritisation; statement 29 (‘Access to health care should be based on need, not on

geographical, social or economic circumstances’) is scored lower and statement 5

(‘People who are in paid work and so contribute financially to society should be

prioritised over people who do not work’) higher.

While interesting, it remains difficult to draw a good overall comparison between

these ten points of view based on visual inspection alone. Table 4. III.6 shows the

correlations between the factor arrays of the five general public and five decision

makers points of view, with correlation coefficients higher than 0.80 between DM2

and GP4 and DM3 and GP1 as well as GP2. Factor DM1 is moderately correlated

with all general public factors, DM4 with GP5 and DM5 with GP2, GP4 and GP5. This

suggests there is some difference in point of view between decision makers and

general public.

Appendix 4.III

18

Table 4. III.5 Statement rank scores for general public and decision makers points of view

Statement General Public Decision Makers

GP1 GP2 GP3 GP4 GP5 DM1 DM2 DM3 DM4 DM5

1 +1 +1 +2 -1 -1 +1 -1 +1 0 +2

2 0 0 0 0 -1 0 +3 +1 +1 0

3 -4 -4 -4 -2 -4 -4 -1 -4 -2 -2

4 +4 +4 -2 +1 +2 -1 -1 +3 0 +4

5 -4 -3 -3 0 -3 -2 -1 -4 -1 0

6 -2 0 +1 0 +1 +2 +2 0 +1 +1

7 -3 -2 -2 0 -2 -2 -2 -3 -3 -2

8 +2 +3 +1 +4 -3 +1 +4 +1 -4 -1

9 +1 0 -1 +1 0 +1 0 -1 -1 +1

10 0 +1 -1 0 0 -2 -1 0 +1 +1

11 -1 0 -1 -3 -2 0 -2 0 -1 0

12 +3 +1 0 +2 +2 -3 0 +2 0 0

13 0 +1 0 -1 0 -1 -2 0 -1 0

14 0 -1 +3 0 +1 +1 +1 -1 +4 +1

15 +2 +3 +4 +3 +3 +2 +4 +3 +4 +2

16 -3 -3 -2 -3 -2 -4 -4 -2 -3 -1

17 -3 -1 -1 -2 +4 +2 -3 -1 +2 -1

18 +1 +2 +1 +2 0 0 +1 +2 +2 +2

19 +1 +2 +2 +3 +3 +4 +3 +2 +1 +3

20 -2 -1 +1 -1 -1 -1 0 0 0 +1

21 +2 +2 +2 -4 -2 0 -3 +3 -2 -3

22 +1 +1 -1 +1 +2 +2 +1 +1 0 -2

23 -2 -3 +3 -1 0 -1 0 -3 +3 -4

24 +3 -2 -2 -4 -1 +3 -3 0 -2 -3

25 -2 -2 -3 +3 +1 0 +1 -3 +2 +4

26 0 -4 +2 -1 0 -1 0 -2 +3 -2

27 +3 +2 0 +4 +3 +4 +2 +2 +2 +3

28 0 0 -4 -2 -1 -3 -4 -1 +1 0

29 +4 +4 +4 +2 +4 +3 +2 +4 0 +3

30 -1 0 -3 -3 -3 -2 -2 -2 -3 -4

31 -1 -2 0 +2 +1 0 +2 -2 -2 -1

32 -1 -1 +3 +1 +1 +1 +3 +1 +3 -1

33 -1 -1 0 -2 -4 -3 0 -1 -4 -3

34 +2 +3 +1 +1 +2 +3 +1 +4 -1 +2

Appendix 4.III

19

Table 4. III.6 Correlations between general public and decision makers factor arrays

General public Decision makers

DM1 DM2 DM3 DM4 DM5

GP1: “Egalitarianism, entitlement and equality of access” 0.50 0.28 0.83 0.09 0.37

GP2: “Efficiency, severity and the magnitude of health gains” 0.49 0.35 0.89 0.04 0.54

GP3: “Fair innings, young people and maximising health benefits” 0.51 0.58 0.54 0.46 0.16

GP4: “The intrinsic value of life and healthy living” 0.43 0.81 0.35 0.32 0.63

GP5: “Quality of life above all else” 0.66 0.39 0.54 0.63 0.57

Table 4. III.7 shows the correlations between the five views among the general public

about the prioritisation of health care and the 110 decision makers. Apart from three

decision makers (from three countries) with correlation coefficients close to zero with

all factors, 61% correlates moderately to good (>0.60) with at least one of the five

factors, and 7% low (<0.40) with all factors.

Table 4. III.7 Correlations of the general public views with the 110 decision makers Q sorts

Decision makers General public

GP1 GP2 GP3 GP4 GP5

1 dk_dm1 0.73 0.86 0.39 0.25 0.37

2 dk_dm2 0.71 0.82 0.43 0.47 0.52

3 dk_dm3 0.49 0.58 0.41 0.06 0.43

4 dk_dm4 0.09 0.10 0.06 0.11 0.19

5 dk_dm5 0.29 0.34 0.07 -0.29 0.25

6 fr_dm1 0.53 0.46 0.44 0.24 0.49

7 fr_dm2 0.39 0.44 0.40 0.26 0.29

8 fr_dm3 0.80 0.65 0.47 0.31 0.34

9 fr_dm4 0.53 0.46 0.61 0.36 0.48

10 fr_dm5 0.58 0.61 0.31 0.07 0.35

11 fr_dm6 0.60 0.45 0.68 0.21 0.57

12 hu_dm1 0.31 0.44 0.54 0.71 0.32

13 hu_dm2 0.21 0.34 0.00 0.58 0.56

14 hu_dm3 0.17 0.47 0.47 0.62 0.31

15 hu_dm4 0.52 0.51 0.69 0.41 0.37

16 hu_dm5 0.32 0.50 0.47 0.42 0.29

17 hu_dm6 0.61 0.61 0.51 0.61 0.64

Appendix 4.III

20

Decision makers General public

GP1 GP2 GP3 GP4 GP5

18 hu_dm7 -0.24 -0.11 -0.22 -0.08 -0.27

19 hu_dm8 0.31 0.54 0.49 0.62 0.54

20 hu_dm9 0.13 0.18 0.24 0.63 0.18

21 hu_dm10 0.07 0.17 0.32 0.51 0.42

22 hu_dm11 0.50 0.36 0.75 0.52 0.38

23 hu_dm12 0.17 0.29 0.35 0.74 0.52

24 nl_dm1 0.37 0.39 0.13 0.31 0.53

25 nl_dm2 0.50 0.47 0.16 0.25 0.52

26 nl_dm3 0.55 0.78 0.41 0.25 0.51

27 nl_dm4 0.43 0.48 0.00 0.27 0.49

28 nl_dm5 0.68 0.39 0.24 0.06 0.40

29 nl_dm6 0.36 0.41 0.17 0.43 0.69

30 nl_dm7 0.37 0.33 0.54 0.22 0.29

31 nl_dm8 -0.17 -0.21 0.22 0.14 0.34

32 nl_dm9 0.47 0.55 0.30 0.53 0.82

33 nl_dm10 -0.05 0.05 0.30 0.40 0.60

34 nl_dm11 0.16 -0.04 0.47 0.32 0.56

35 nl_dm12 0.25 0.34 0.08 0.17 0.24

36 nl_dm13 0.71 0.78 0.47 0.36 0.67

37 nl_dm14 0.71 0.68 0.26 0.04 0.31

38 nl_dm15 0.49 0.57 0.57 0.00 0.32

39 nl_dm16 0.06 -0.05 -0.03 0.00 0.11

40 nl_dm17 0.35 0.34 0.51 0.43 0.56

41 nl_dm18 0.11 0.11 0.62 0.39 0.68

42 nl_dm19 0.61 0.68 0.33 0.31 0.54

43 nl_dm20 0.44 0.47 0.52 0.19 0.65

44 nl_dm21 0.57 0.58 0.39 0.32 0.46

45 nl_dm22 -0.12 -0.19 0.21 0.13 0.34

46 nl_dm23 0.63 0.62 0.24 -0.02 0.29

47 no_dm1 0.67 0.67 0.43 0.21 0.21

48 no_dm2 0.51 0.57 0.71 0.70 0.53

49 no_dm3 0.51 0.55 0.55 0.22 0.36

50 pa_dm1 0.65 0.69 0.26 0.14 0.48

51 pa_dm2 0.63 0.69 0.39 0.31 0.51

52 pa_dm3 0.70 0.65 0.29 0.33 0.29

53 pa_dm4 0.59 0.81 0.34 0.29 0.28

54 pa_dm5 0.28 0.37 0.30 0.34 0.43

55 pa_dm6 0.70 0.63 0.37 0.40 0.52

56 pa_dm7 0.31 0.14 0.41 0.03 0.48

57 pa_dm8 0.50 0.60 0.30 0.42 0.16

58 pa_dm9 0.66 0.62 0.53 0.46 0.46

59 pa_dm10 0.24 0.24 0.43 0.58 0.16

60 po_dm1 0.21 0.29 0.25 0.56 0.42

Appendix 4.III

21

Decision makers General public

GP1 GP2 GP3 GP4 GP5

61 po_dm2 0.52 0.55 0.34 0.32 0.38

62 po_dm3 -0.09 -0.06 0.16 0.54 0.03

63 po_dm4 0.69 0.59 0.34 0.41 0.36

64 po_dm5 0.46 0.50 0.17 0.63 0.54

65 po_dm6 0.36 0.33 0.17 0.76 0.46

66 po_dm7 0.56 0.69 0.27 0.61 0.47

67 po_dm8 0.55 0.61 0.34 0.56 0.44

68 po_dm9 0.44 0.63 0.28 0.41 0.26

69 po_dm10 0.68 0.75 0.31 0.34 0.46

70 po_dm11 0.16 0.33 0.19 0.23 0.17

71 po_dm12 0.34 0.27 0.49 0.48 0.71

72 po_dm13 0.44 0.49 0.47 0.70 0.63

73 se_dm1 0.42 0.46 0.65 0.41 0.68

74 se_dm2 -0.11 0.04 0.34 0.26 0.13

75 se_dm3 0.43 0.45 0.71 0.37 0.48

76 se_dm4 0.47 0.64 0.54 0.41 0.54

77 se_dm5 0.46 0.55 0.66 0.34 0.50

78 se_dm6 0.47 0.59 0.73 0.32 0.43

79 se_dm7 0.57 0.59 0.47 0.18 0.39

80 se_dm8 0.72 0.73 0.51 0.24 0.34

81 se_dm9 0.33 0.44 0.56 0.37 0.25

82 se_dm10 0.60 0.64 0.64 0.51 0.57

83 sp_dm1 0.17 0.10 0.12 0.38 0.07

84 sp_dm2 0.26 0.32 0.32 0.22 0.50

85 sp_dm3 0.51 0.44 0.43 0.37 0.54

86 sp_dm4 0.27 0.18 0.34 0.24 0.50

87 sp_dm5 0.56 0.49 0.50 0.39 0.70

88 sp_dm6 0.38 0.31 0.54 0.28 0.71

89 sp_dm7 0.11 0.42 -0.03 0.15 0.43

90 sp_dm8 0.13 0.16 0.46 0.37 0.48

91 sp_dm9 0.11 0.22 0.55 0.55 0.51

92 sp_dm10 0.48 0.51 0.43 0.46 0.66

93 sp_dm11 0.60 0.79 0.35 0.28 0.38

94 sp_dm12 0.41 0.40 0.47 0.28 0.60

95 sp_dm13 0.21 0.17 0.51 0.28 0.63

96 uk_dm1 0.58 0.57 0.52 0.42 0.59

97 uk_dm2 0.51 0.62 0.41 0.12 0.65

98 uk_dm3 0.66 0.69 0.41 0.46 0.71

99 uk_dm4 0.41 0.40 0.47 0.56 0.70

100 uk_dm5 0.61 0.69 0.46 0.37 0.64

101 uk_dm6 0.62 0.74 0.40 0.27 0.45

102 uk_dm7 0.62 0.61 0.51 0.44 0.41

103 uk_dm8 0.56 0.71 0.21 0.37 0.49

Appendix 4.III

22

Decision makers General public

GP1 GP2 GP3 GP4 GP5

104 uk_dm9 0.67 0.69 0.48 0.65 0.61

105 uk_dm10 0.64 0.70 0.48 0.24 0.54

106 uk_dm11 0.62 0.55 0.56 0.43 0.49

107 uk_dm12 0.40 0.31 0.34 0.45 0.71

108 uk_dm13 0.59 0.48 0.35 0.14 0.52

109 uk_dm14 0.41 0.60 0.34 0.34 0.67

110 uk_dm15 0.31 0.44 0.57 0.47 0.44

The analysis of the pooled factor arrays of general public and decision makers

presented in Table 4. III.8 confirms that DM3 aligns with GP1 and GP2, as also

evident from the correlations between factor arrays (see Table 4. III.6), but it also

clearly shows that DM4 -cluster G- is a distinct point of view (with distinguishing

statements #8 (+4) and #17 (-2)).

Table 4. III.8 Example of an analysis of general public and decision makers pooled factor

arrays

Point of view Meta clusters

D E F G

GP1: “Egalitarianism, entitlement and equality of access” 0.80 * 0.22 0.15 0.05

GP2: “Efficiency, severity and the magnitude of health gains” 0.85 * 0.15 0.32 -0.02

GP3: “Fair innings, young people and maximising health benefits” 0.32 0.48 * 0.16 0.26

GP4: “The intrinsic value of life and healthy living” 0.20 0.11 0.92 * 0.11

GP5: “Quality of life above all else” 0.31 0.57 0.33 0.40

DM1 0.38 0.52 0.29 0.25

DM2 0.15 026 0.64 * 0.33

DM3 0.90 * 0.31 0.16 0.11

DM4 -0.02 0.27 0.19 0.84 *

DM5 0.34 0.28 0.51 * 0.07

Appendix 4.III

Distribution of European views in the EuroVaQ sample

Contents

4.IV Distribution of European views in the EuroVaQ sample ................................... 2

4.IV.1 Method of determining factor membership ............................................... 2

4.IV.2 Data .......................................................................................................... 6

4.IV.3 Statistical results ....................................................................................... 6

List of tables

Table 4. IV.1 Abbreviated descriptions of the five general public views ............................................. 3

Table 4. IV.2 Results of the assignment procedure (n=32,861) ......................................................... 9

List of figures

Figure 4. IV.1 Sample question for self-categorisation to abbreviated factor descriptions ............. 5

Figure 4. IV.2 Sample question for untying highest scores on more than one factor ...................... 5

Figure 4. IV.3 Highest score given to one of the five factor descriptions ........................................ 8

Figure 4. IV.4 Same score given to all factor descriptions .............................................................. 8

Figure 4. IV.5 Factor membership in the survey sample ................................................................. 9

Figure 4. IV.6 Distribution of the five points of view across the ten countries ............................... 10

Appendix 4.IV

4.IV Distribution of European views in the EuroVaQ sample

The objective of this appendix is to explore how the five views among the public

about the prioritisation of health care from section 4.2 are distributed across the ten

countries participating in the EuroVaQ study. In the next section we first discuss how

factor membership can be determined, followed by the data that were used in 4.IV.2

and the results in 4.IV.3.

4.IV.1 Method of determining factor membership

Different methods have been proposed in order to use the results of a Q study to

design a survey instrument to determine the ‘factor membership’ or factor association

of a large sample of respondents.1

In this study the approach used was self-

categorisation to abbreviated factor descriptions. Participants in the survey were

presented with summary descriptions of the Q factors and asked to indicate the

degree to which each description is similar to their own point of view on prioritisation

of health care.

In an iterative process, five researchers from Newcastle University and Erasmus

University Rotterdam summarised the factor descriptions from section 4.2, focussing

on the most characterising and distinguishing statements for each factor. Because

the questions were part of a large online survey, the length of each summary was

1 Baker R, J van Exel, H Mason, M Stricklin. Connecting Q & surveys: a test of three methods to

explore factor membership in a large sample. Operant Subjectivity, forthcoming.

Appendix 4.IV

limited to 75-100 words in order to limit the burden for participants. Table 4. IV.1 lists

the final versions of the abbreviated descriptions as used in the survey.

Table 4. IV.1 Abbreviated descriptions of the five general public views

EU PoV1: “Egalitarianism, entitlement and equality of access”

I think access to health care is a basic right for all citizens. It should not be related to individual,

social, family circumstances or lifestyle, it is more important that everyone in need of care is equally

entitled to treatment. All life has the same value regardless of quality of life before and after

treatment, past use of health services, financial contributions to the health services in the past, or

ability to pay for private treatment.

EU PoV2: “Efficiency, severity and the magnitude of health gains”

I think everyone is equally worthy of treatment, but it is important to consider need and how much

patients will benefit. People with worsening health or who would otherwise die should take priority

over others. However, the characteristics of patients, such as their age, gender or income should not

be used to prioritise between them. There’s nothing wrong with people paying for treatment in the

private sector as long as it doesn’t affect the treatment of others.

EU PoV3: “Fair innings, young people and maximising health benefits”

I think that the health budget should be spent on treatments that generate the most health. Younger

people should be given priority because they haven’t had their fair share of life yet and they are

likely to benefit more from treatment than older people. Access to health care should not be based

on where people live, their lifestyle, their financial contribution to the health services in the past, or

their ability to pay for private treatment.

EU PoV4: “The intrinsic value of life and healthy living”

I think life itself is valuable. Health services should be about saving lives and preventing illness. It is

also individuals’ own responsibility to lead a healthy lifestyle. This would create a more healthy

society and prevent costs to the health care system. Parents with dependent children should be

given priority. People should be allowed to pay privately for treatments as long as it doesn’t affect

the treatment of others.

EU PoV5: “Quality of life above all else”

I think that health care should be about the quality of life of patients. There’s no use prolonging lives

if treatment won’t restore quality of life to an acceptable level. In general people should have equal

access to treatment but priority should be given to prevention and treatments that generate the most

health. Individual characteristics of patients, such as their age, gender or income and financial

contributions to the health care system in the past should not be used to prioritise between them.

Appendix 4.IV

The question was introduced to respondents as follows:

This study is about decision making in health care. No matter how large, the

health care budget will never be enough to do everything that could possibly be

done. Because of this, choices must be made about which health services and

treatments to provide and, therefore, which not to provide. Different choices will

mean that some patients get treatments (because they are provided) and other

patients will not (because treatments are not provided). A survey of members of

the public showed five points of view about how to make these choices in health

care. We are interested in how much you agree with each of these points of view.

After that, they received an instruction:

You will now be shown short descriptions of each of these five points of view.

Please indicate your agreement with them by ticking the box below the

description.

The five abbreviated descriptions of the Q factors were presented to respondents

individually and consecutively, in random order. Together with each description

respondents were given a seven point Likert scale and asked to indicate to indicate

their agreement with each factor summary on a scale labelled ‘very unlike my point of

view’ to ‘very much like my point of view’. shows a sample question from the online

survey.

Appendix 4.IV

Figure 4. IV.1 Sample question for self-categorisation to abbreviated factor descriptions

For determining factor membership it is important to have a unique highest score on

one of the five factor descriptions. Therefore a follow-up question was presented to

respondents with a tie on their highest score, as shown in Figure 4. IV.2 (in this case

two factors).

Figure 4. IV.2 Sample question for untying highest scores on more than one factor

Appendix 4.IV

4.IV.2 Data

Data were available from the main EuroVaQ survey administered online in ten

European countries, i.e. Denmark, France, Hungary, Norway, Palestine, Poland,

Spain, Sweden, the Netherlands, and the UK. The sample consisted of 38,074

respondents, representative for each country in terms of age, gender, level of

education and socio-economic status. More details about the questionnaire and the

data collection can be found in Chapter 3.

4.IV.3 Statistical results

Inspection of the response patters showed that 2,512 respondents (6.6%) gave all

factor descriptions a score lower than five. In other words, these respondents rated

all factors less than ‘a little like my point of view’ (see Figure 4. IV.3). This was an

encouraging result, as it suggests that the large majority of respondents agree with at

least one of the factors, indicating that they see their point of view on prioritisation of

health care represented in the results of the primary Q study. Respondents who

scored all descriptions less than five were dropped from further analysis because it

was considered inappropriate to match respondents to a point of view that they rated

as less than ‘a little like my point of view’.

A further 4,265 respondents (11.2%) gave the exact same score to all factor

descriptions, i.e. did not differentiate between the points of view. These respondents

would have been asked to choose one factor summary as most like their point of

Appendix 4.IV

view, as shown in Figure 4. IV.2, but their initial scores across all five summaries

were tied. It is possible (although perhaps unlikely) that such respondents have

precisely the same level of agreement with all factor summaries, but it is also

plausible that some respondents ‘clicked through’ the survey quickly and without

sufficient consideration. Because it is not possible to distinguish respondents giving

genuine answers from those unwilling to participate in the exercise, these

respondents were also excluded from the analysis.

In all, 5,213 respondents (13.7%) were dropped because they gave all factors a

score lower than five and/or the same score to all factor summaries, leaving a sample

of 32,861 for analysis.

Further inspection showed that 13,958 respondents (42.5%) were ‘pure types’ (i.e.

gave their highest score to a single factor), while 18,903 respondents (57.5%) tied on

their highest score. The latter respondents were assigned to one of the five factors

using their response to the follow-up question (see Figure 4. IV.2). This procedure

resulted in the distribution over points of view shown in Table 4.IV.2 and Figure

4.IV.5.

Appendix 4.IV

Figure 4. IV.3 Highest score given to one of the five factor descriptions

Figure 4. IV.4 Same score given to all factor descriptions

Appendix 4.IV

Table 4. IV.2 Results of the assignment procedure (n=32,861)

Pure types Matched to points of view

GP1 GP2 GP3 GP4 GP5 Total

GP1 6,924 6,924 21.1%

GP2 1,998 1,998 6.1%

GP3 738 738 2.2% 42.5%

GP4 2,039 2,039 6.2%

GP5 2,259 2,259 6.9%

Not a ‘pure type’ 7,902 3,637 717 3,245 3,402 18,903 57.5%

Total 14,826 5,635 1,455 5,284 5,661 32,861

45.1% 17.1% 4.4% 16.1% 17.2% 100.0%

Figure 4. IV.5 Factor membership in the survey sample

Figure 4. IV.6 shows how the five views among the public about the prioritisation of

health care are distributed across the ten countries participating in EuroVaQ.

Appendix 4.IV

Figure 4. IV.6 Distribution of the five points of view across the ten countries

17,5%

22,5%

17,8%14,4%

18,0%

6,7%

18,6%

13,3%

26,1%

17,3%

15,4%

11,1% 20,4%

16,7%

16,7%

19,8%

15,6%

15,4%

11,5%

20,0%

4,4%4,9%

4,4%

5,6%

10,8%

4,3%

5,3%

3,6%

2,7% 4,5%

14,2% 12,3%

20,7%

16,4%

23,6%

19,8%

17,3%

14,8%

21,5%16,4%

48,5% 49,1%

36,7%

46,9%

30,8%

49,4%

43,2%

53,0%

38,3%41,9%

0%

25%

50%

75%

100%

Den

mar

k

Franc

e

Hun

gary

Nor

way

Pales

tine

Polan

d

Spa

in

Swed

en

The N

ethe

rland

s

The U

K

GP1: “Egalitarianism, entitlement and equality of access”

GP2: “Efficiency, severity and the magnitude of health gains”

GP3: “Fair innings, young people and maximising health benefits”

GP4: “The intrinsic value of life and healthy living”

GP5: “Quality of life above all else”

Appendix 4.IV