Public views on principles for health care priority setting: Findings of a European cross-country study using Q methodology

Highlights

• Little is known about views in society on principles to guide priority setting.

• We used Q methodology with members of the public across ten countries.

• We found five shared views on principles regarded appropriate for rationing.

• No single equity principle can be used to underpin priority setting in health care.

• This plurality of views in society should be reflected in priority setting processes.

Abstract

Resources available to the health care sector are finite and typically insufficient to fulfil all the demands for health care in the population. Decisions must be made about which treatments to provide. Relatively little is known about the views of the general public regarding the principles that should guide such decisions.

We present the findings of a Q methodology study designed to elicit the shared views in the general public across ten countries regarding the appropriate principles for prioritising health care resources. In 2010, 294 respondents rank ordered a set of cards and the results of these were subject to by-person factor analysis to identify common patterns in sorting. Five distinct viewpoints were identified, (I) “Egalitarianism, entitlement and equality of access”; (II) “Severity and the magnitude of health gains”; (III) “Fair innings, young people and maximising health benefits”; (IV) “The intrinsic value of life and healthy living”; (V) “Quality of life is more important than simply staying alive”.

Given the plurality of views on the principles for health care priority setting, no single equity principle can be used to underpin health care priority setting. Hence, the process of decision making becomes more important, in which, arguably, these multiple perspectives in society should be somehow reflected.

Introduction

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. Although this truth may be inconvenient and, in some countries, may even be contested (e.g. Greiner and von der Schulenburg, 2010), choices in the allocation of health care resources are, in essence, unavoidable. This means that decisions have to be made about which treatments are provided (and which patients will be helped), and which treatments are not provided (and which patients will be denied help). The contexts and ways in which such decisions are made differ between health care systems and sectors within health care. For instance, at the macro- or health care system level, it may be necessary to decide which drugs and other technologies to reimburse. At the meso- or hospital level, it may be necessary to set up rules for whom to treat first when there are waiting lists for particular services. At the micro or patient group level, it may be necessary to set a specific threshold for risk levels for health problems beyond which to start treatment (e.g. in case of high cholesterol). All such (explicit and implicit) choices are of course related and have implications for the final allocation of resources in the health care sector (Klein, 1993, Litva et al., 2002). The obvious question then is, on what basis such choices should be made? This question can provoke heated debate in the context of explicit priority setting (or rationing) as it involves difficult trade-offs and intrinsically normative questions regarding the aim(s) and value(s) of the health care sector. Typically, the aim of the health care sector will involve notions of efficiency and equity (Dolan et al., 2005). Ideally, therefore, allocation decisions reflect these two, potentially conflicting, notions.

The tension between equity and efficiency in the health care sector is apparent in a range of routine decisions and practices. For instance, in the prioritisation of patients on a waiting list, the aim to maximise overall health benefits from treatment may be at conflict with that of obtaining an equitable distribution of health and health care. As an illustration, a proposal in the Netherlands to treat employees more quickly than non-employees by creating additional treatment capacity, thus reducing waiting times for all (though unequally so) and saving lost production due to absence, was rejected. The main reason was that it was considered to be inequitable to treat employees quicker than non-employees, while their medical need (or capacity to benefit) was not necessarily higher (Brouwer and Schut, 1999). A similar tension may be observed in the controversy around user fees in the health care sector. While such fees may help to raise cost-consciousness and reduce moral hazard, they may also result in socio-economic differences in access to health care (Donaldson and Gerard, 1989).

Efficiency and equity are also central to debates regarding appropriate decision rules for reimbursement of new health technologies based on economic evaluations. A commonly applied decision rule is to assess whether the additional health benefits in terms of quality-adjusted life-years (QALYs) are ‘worth’ the additional costs. If the cost per QALY is below some (agreed upon) threshold, the treatment is taken to represent value for money and thus deemed eligible for funding. An important and much discussed question, however, is whether all QALYs should have equal value in these decisions (Bobinac et al., 2012, Brouwer et al., 2008, Dolan et al., 2005, Gerard and Mooney, 1993, Donaldson et al., 1988, Weinstein, 1988). Much empirical evidence suggests that people do not attach equal value (or weight) to different QALYs benefiting different groups of people. For instance, a QALY gained by a severely ill person may be valued (weighted) differently than a QALY gained in a person who is only mildly ill – the ‘severity of illness’ argument – and more so in a young person than in an old person – the ‘fair innings argument’ (Williams, 1997; Williams, 1988a; Nord, 2005; Donaldson et al., 2011; Donaldson et al., 1988). 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. In the UK, recently, some room has been created to accommodate these equity considerations in the decision making process, by allowing specific life prolonging interventions to be judged against a higher cost per QALY threshold (National Institute for Health and Clinical Excellence, 2009, Longson and Littlejohns, 2009, Scottish Medicines Consortium NS, 2010); it is however uncertain whether these measures reflect actual societal values (Linley and Hughes, 2013). In the Netherlands, as a general rule, higher thresholds are allowed for interventions aimed at increasingly severe illnesses (College voor Zorgverzekeringen, 2006, Van de Wetering et al., 2013).

If policy makers wish to reflect, or at least be responsive to, equity considerations or more general views on appropriate allocation of health care resources among the general public, more knowledge on such public preferences is necessary. However, robust research evidence reflecting the richness of the viewpoints among the public regarding the distribution of health and health care, is lacking (Buxton and Chambers, 2011, Donaldson et al., 2011). A number of studies have investigated public preferences, but they have typically focussed on specific treatments or patient groups, whilst a number of literature reviews have shown a wide variety of equity considerations and attitudes towards distribution of health care (Van de Wetering et al., 2013, Bobinac et al., 2012; 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 (Lancsar et al., 2011, Baker et al., 2010a).

The current study was part of the ‘European Value of a Quality adjusted life year’ (EuroVaQ) project (Donaldson et al., 2010), conducted in ten countries: Denmark, France, Hungary, Norway, Palestine, Poland, Spain, Sweden, the Netherlands and the UK. One of the principle aims of EuroVaQ was to develop and test robust methods to determine the monetary value of a QALY (Robinson et al., 2013, Pennington et al., 2013). The aim of study presented here was to describe the shared views in the general population across these ten countries regarding the prioritisation of health care.

Q methodology combines qualitative and quantitative methods and provides a scientific foundation for the systematic study of subjectivity (Stephenson, 1935, Brown, 1980, Watts and Stenner, 2012, McKeown and Thomas, 2013). Although the origins of Q methodology are in psychology, it is now relatively well established in health services research, with a rapidly growing number of published studies on a variety of health related topics: for instance, health behaviour and outcomes (Baker, 2006, Stenner et al., 2003), attitudes and beliefs (Vermaire et al., 2010, Van Exel et al., 2006, Eccleston et al., 1997), treatment adherence (Cramm et al., 2010, Tielen et al., 2008), coping and adaptation (Kraijo et al., 2012, Boot et al., 2009, Risdon et al., 2003), and professional views (Lobo et al., 2012, Wallenburg et al., 2010, Buljac et al., 2010).

In a Q methodology study respondents rank a set of opinion statements through a card sorting procedure known as a ‘Q sort’, and by doing so reveal their point of view toward the subject being studied. The rankings of the respondents are subject to correlation analysis, and the correlation between individual rankings is taken to indicate similarity between viewpoints. By-person factor analysis (Kline, 1994) is then used to identify significant clusters of correlations, which can be interpreted as distinguishable viewpoints. Q methodology is thus used to describe a ‘population of viewpoints’ and the correspondence and distinctions between them. By its nature – and in contrast to survey research – Q relies on relatively small purposive respondent samples (i.e. typically 25 to 40 respondents) conducting a large number of ‘tests’ (i.e. a full ranking of 30–50 statements). An important consequence is that, like qualitative findings, the results of a Q study may be generalised to the subject area from which the opinion statements were sampled, but not – as in survey research – to the population (Brown, 1980, Watts and Stenner, 2012). In other words, logical generalisations can be drawn about the nature of opinion and shared perspectives that exist on a given topic (here priority setting in health) but nothing can be said about how many people are associated with each viewpoint identified, or about their likely characteristics.

McKeown and Thomas, 2013, Watts and Stenner, 2012 and De Graaf and van Exel (2009), among others, provide comprehensive introductions to Q methodology. Brown et al. (2014) recently discussed some of the main critiques of Q methodology.