Patient DOI 10.1007/s40271-014-0063-2

SYSTEMATIC REVIEW

A Systematic Review of Stated Preference Studies Reporting Public Preferences for Healthcare Priority Setting Jennifer A. Whitty • Emily Lancsar • Kylie Rixon • Xanthe Golenko • Julie Ratcliffe

Ó Springer International Publishing Switzerland 2014

Abstract Background There is current interest in incorporating weights based on public preferences for health and healthcare into priority-setting decisions. Objective The aim of this systematic review was to explore the extent to which public preferences and tradeoffs for priority-setting criteria have been quantified, and to describe the study contexts and preference elicitation methods employed. Methods A systematic review was performed in April 2013 to identify empirical studies eliciting the stated

Electronic supplementary material The online version of this article (doi:10.1007/s40271-014-0063-2) contains supplementary material, which is available to authorized users. J. A. Whitty (&) School of Pharmacy, Pharmacy Australia Centre of Excellence, The University of Queensland, 20 Cornwall Street, Woolloongabba, Brisbane, QLD 4102, Australia e-mail: [email protected] J. A. Whitty  K. Rixon  X. Golenko Centre for Applied Health Economics, School of Medicine and Population and Social Health Research Programme, Griffith Health Institute, Griffith University, Logan Campus, Meadowbrook, QLD, Australia E. Lancsar Centre for Health Economics, Faculty of Business and Economics, Monash University, Melbourne, VIC, Australia J. Ratcliffe Flinders Health Economics Group, School of Medicine, Faculty of Medicine, Nursing and Health Sciences, Flinders University, Adelaide, SA, Australia

preferences of the public for the provision of healthcare in a priority-setting context. Studies are described in terms of (i) the stated preference approaches used, (ii) the prioritysetting levels and contexts, and (iii) the criteria identified as important and their relative importance. Results Thirty-nine studies applying 40 elicitation methods reported in 41 papers met the inclusion criteria. The discrete choice experiment method was most commonly applied (n = 18, 45.0 %), but other approaches, including contingent valuation and the person trade-off, were also used. Studies prioritised health systems (n = 4, 10.2 %), policies/programmes/services/interventions (n = 16, 41.0 %), or patient groups (n = 19, 48.7 %). Studies generally confirmed the importance of a wide range of process, non-health and patient-related characteristics in priority setting in selected contexts, alongside health outcomes. However, inconsistencies were observed for the relative importance of some prioritisation criteria, suggesting context and/or elicitation approach matter. Conclusions Overall, findings suggest caution in directly incorporating public preferences as weights for priority setting unless the methods used to elicit the weights can be shown to be appropriate and robust in the priority-setting context. Abbreviations CUA Cost–utility analysis CV Contingent valuation DCE Discrete choice experiment MCDA Multicriteria decision analysis PTO Person trade-off QALY Quality adjusted life year QoL Quality of Life WTP Willingness to pay

J. A. Whitty et al.

Key Points for Decision Makers Choice-based stated preference methods have the potential to systematically elicit preference weights representing the values of society for consideration in priority-setting decisions. Research in this area is developing, but suggests some inconsistencies in the results regarding public views in the relative importance of prioritisation criteria. The priority-setting context and methodological approach to preference elicitation may both be important in explaining nuances and inconsistencies in priority-setting preferences.

1 Background Priority setting in healthcare is inevitable [1, 2]. Given the ageing population and concerns about the sustainability of healthcare funding, priority-setting decisions are likely to become more challenging in the future [3]. Where there is priority setting, there has to be some basis on which to set priorities. There has been substantial debate in healthcare over what this basis should be, and what criteria should be considered in priority-setting decisions (see, for example, [4–8]). Conventional cost–utility analysis (CUA) is considered the preferred approach to health economic evaluation by many healthcare decision makers when setting funding priorities [9]. CUA explicitly assumes overall societal health is the only desirable outcome from healthcare. However, it has long been recognised that it may be desirable to include a more holistic range of outcomes and benefits from healthcare in priority-setting decisions [10, 11]. The mechanism by which a broader range of benefits from healthcare could be considered in priority-setting decisions and systematically incorporated in a decisionmaking framework is receiving increasing attention [12]. Focus has fallen on two potential approaches. The first replaces CUA in health with cost–benefit analysis, employing a welfarist approach to measure the benefits of healthcare in monetary terms (e.g., via assessment of willingness to pay [WTP] for a health intervention or service) [13–17], thereby allowing the valuation of non-health outcomes to be considered. The second retains the extrawelfarist approach but weights the quality-adjusted lifeyear (QALY) within or alongside the framework of CUA, according to preferences for characteristics that might be

considered important in addition to the absolute health gain (such as who receives the health gain) [18–20]. This latter approach develops preference weights for criteria that are relevant to priority-setting decisions, and uses these weights alongside comparative costs and benefits in terms of health gain to weight the decision. The potential of weighting decision criteria has been recognised in health as one component of multicriteria decision analysis (MCDA) [21]. These approaches are still developing, and there is no clear indication as to which approach might be most appropriate. Regardless, these approaches to priority setting require preference weights for prioritisation criteria to be developed on a common scale for application in the priority-setting decision. The purpose of this study is two-fold. First, we explore the extent to which preferences of the public for priority setting, and the relative strength of importance of criteria that could be considered to weight health gain, have been identified in the peer-reviewed literature. Secondly, we seek to describe the contexts in which the studies have been undertaken and the approaches used to elicit preferences, with the intention of informing future research in this area. Our review has a broad focus on priority setting at any level, including setting priorities at the health system or healthcare services/programme level, as well as across different groups of people. It explores a wide range of potentially relevant criteria, both related to healthcare delivery (e.g., non-health benefits resulting from the process of delivery such as a reduction in waiting times) as well as characteristics of healthcare recipients. Thus, it extends previous reviews with a narrower focus on the relevance of personal characteristics [8, 22] or severity of illness [23], or the alignment between social values and assumptions inherent in the CUA framework [7, 24]. Further, our review only includes studies that have employed choice-based methods to quantify preferences on a common scale, distinguishing it from previous reviews [7, 8, 22–24]. Consequently, our review has the potential to inform the feasibility of weighting criteria for use in priority setting. 2 Methods 2.1 Systematic Review of Empirical Studies A systematic review of the published literature was performed to identify original empirical studies eliciting the stated preferences of members of the general public for the provision of healthcare in a priority-setting context. The study employed established methodological approaches for undertaking systematic reviews in healthcare [25, 26].

Public Preferences for Priority Setting

2.2 Identification of Studies All available dates in PubMed, the Cochrane Library, CINAHL, EconLit, Web of Knowledge, ProQuest and Informit databases were searched in April 2013. A complete list of databases and sources searched is provided in the electronic supplementary material. Additional material was identified in the reference lists of articles retrieved and by conducting citation searches on the articles that met the inclusion criteria. 2.3 Selection of Studies for Inclusion The intent was to identify studies that presented stated preference data based on a general public sample to hypothetical choice scenarios between two or more competing programmes, interventions, or patient populations, in a healthcare priority-setting context. The inclusion criteria are detailed in Table 1. To be included, studies needed to report original empirical data and be published in full (not abstract form) in the peer-reviewed literature. The study objectives were required to have an applied rather than methodological focus; for example, studies focusing

on framing issues or alternative approaches to data collection or analysis were excluded. Studies were required to have applied a choice-based stated preference approach and to derive an estimate of relative strength of preference. Titles and abstracts were screened by two authors (KR and JW) for eligibility against the inclusion criteria. Potentially eligible publications were then screened in full by two authors (XG, KR), with any disagreements adjudicated by a third author (JW). There is currently no single standardised method for assessing the quality of studies across the full range of methods that can be used to elicit stated preferences. To avoid imposing an untested framework of quality assessment on the literature identified for this review, a formal assessment of study quality was not undertaken. 2.4 Search Terms Search terms were developed using an iterative process. Searches of the PubMed and EBSCO databases were conducted using terms from the inclusion criteria (Table 1). The titles and key terms of the retrieved articles were then combined with the US National Library of

Table 1 Inclusion criteria To be included, papers had to meet all of the following criteria: 1. Publication Published in English in a peer-reviewed source. Theses and works that were not published in full (i.e., conference presentations, abstracts) were excluded 2. Empirical data with applied focus An original research paper presenting previously unpublished empirical data with an applied, rather than methodological, focus. This excluded reviews; papers focusing on instrument development, questionnaire design, or item construction; and papers that focused on methodological and technical issues of preference elicitation or data analysis rather than the healthcare topic being discussed 3. Stated preferences data Preferences were elicited in a hypothetical (stated) context, using a choice-based approach that required a trade-off to be made. An estimate of relative strength of preference was required. Examples include discrete choice experiment, allocation of points, pairwise choices, best– worst scaling, willingness to pay, willingness to assign, and person trade-off. Studies reporting percentage agreement with a statement/ alternative or non-choice-based approaches were excluded, including visual analogue scales, Likert-type scales, and ranking methods 4. Healthcare Preferences had to relate to a service where the primary purpose was diagnosis, treatment, and/or prevention of disease, illness, injury, or other physical or mental impairment. Studies relating predominantly to transport (including road safety) or the environment (including pollution or water quality) were excluded 5. Focus is priority setting for public resource allocation The preferences elicited had to relate to a priority-setting context; that is, the choice to provide one of two or more competing alternatives from public or social funds. The choice could relate to one programme, service, strategy, intervention, technology over another; or to treat/ benefit one population over another. Studies were also included if they identified preferences for one programme type, but delivered in two or more different ways 6. Preference population is general public The sample had to be drawn from a population that did not have expert knowledge or a specific vested interest in the topic. This excluded (1) patients or individuals who were being asked about the delivery of services for their own treatment; and (2) health professionals or administrators who were selected because of their knowledge/expertise in the area 7. Perspective of preferences was social and related to health as a public good The study must have had a focus on the public good. This excluded individual treatment preferences or personal purchasing decisions

J. A. Whitty et al.

Medicine’s MeSH Browser, and searches were re-run. This process continued until a set of appropriately sensitive search terms were identified. The search used Boolean AND commands to combine a range of terms representing each of the following four concepts: preferences, priority setting, stated preference methods, and healthcare. The search strategy used for the EBSCO databases (CINAHL and EconLit) is provided in the electronic supplementary material; other databases and sources sometimes required minor modifications. Filters or limits were used for language (English) and subjects (human). 2.5 Information Extraction and Synthesis Extraction of information and classification of the included references was conducted by two authors (XG and JW). Given the diversity of contexts and lack of comparability of criteria between studies, it was deemed inappropriate to synthesise preference data and a descriptive reporting approach was adopted. Studies were categorised and described in terms of (i) the method used to elicit preferences; (ii) the priority-setting level, that is, whether the focus was on choices about the whole health system, about healthcare delivery (such as health services, programmes, interventions or policies), or about patient groups; (iii) the contexts in which studies were undertaken; and (iv) the Fig. 1 Flowchart of search results

criteria identified as important to the public and their relative importance.

3 Results 3.1 Description of Included Studies The database searches identified 1,172 citations (Fig. 1). A further 156 records were identified for consideration through other sources, including citation searches on the included papers, a review of the reference lists of included papers, and previous reviews identified in the searches. Following a review of titles and abstracts, 905 were excluded as irrelevant (682) or duplicates (223). A further 382 were excluded following a review of the full text against the inclusion criteria. Thus, a total of 41 papers across 39 studies met the inclusion criteria and were included in the review, including two papers resulting from a single German study [27, 28] and two further papers reporting the social value of a QALY in the UK [18, 19]). Details of the papers included are provided in Table 2. The included papers were published between 1993 and 2012. The number of publications in this field has increased over time (see Fig. 2). Most of the results were published after 2000 (n = 30, 77 %), and the number of publications

Year

2006

2010

2010

2002

2002

2005

2011, 2012

Study

Akkazieva et al. [29]

Baker et al. [18] (NB: same study as Lancsar, et al. [19]

Bosworth et al. [30]

Bryan et al. [44]

Corso et al. [50]

Costa-Font and Rovira [58]

Diederich et al. [27, 28]

Germany

Spain

USA

UK

USA

UK

Hungary

Country

DCE

CV; willingness to assign

Spain sample (n = 66), small deliberative groups of the general public that had previously participated in focus groups Random sample of German population over 18 years (n = 2,031)

CV; WTP

DCE

DCE

PTO with QALY grid

DCE

Method

1,104 English-speaking US residents aged over 18 years

Random sample, n = 909

General adult public of Hertfordshire, UK

2 National surveys: public prevention survey, public treatment survey (1,500 each)

Sample of adults living in England (n = 587), designed to be nationally representative

86 patients from a rheumatology outpatient centre at the Flo´r Ferenc County Hospital, Budapest, Hungary

Population

2011: age as a criterion for priority setting, and if preference depends on the interviewee’s own age, sex, socioeconomic status, or health status. 2012: acceptance of possible prioritization criteria (medical, socioeconomic and lifestyle situation of the patient); use of DCEs for representing the ‘voice of the patient’

Preferences for competing healthcare programmes

Resource allocation for prevention versus treatment

Assumptions of QALY model (is societal value proportional to size of benefit)

Assess preferences for publicly funded health policies designed to prevent or treat major health threats

Estimate relative QALY weights

Healthcare system reforms in Hungary

Study focus

People

Programme

Programme

People

Policy

People

Health system

Level

Table 2 Details of included studies (n = 39 studies, employing 40 separate preference elicitation tasks, reported in 41 papers)

Generic patient groups

Competing programmes within Catalan health service

Prevention vs treatment

Generic patient groups

Prevention vs treatment

Generic patient groups

Reform

Context

Medical aspects, lifestyle and socioeconomic status: severity of disease (health status; QoL); age; family status; occupational status; lifestyle

Ten competing programmes providing health and nonhealth (equity and process) benefits; public funding/taxes

Prevention or treatment; outof-pocket cost

Prevention or treatment: risk source; illness addressed; age-group affected; number affected; health gain (deaths averted), timing of health gain; out-of-pocket cost Benefit: number of people, chance of success, survival, QoL after treatment

Age and severity of illness: age at onset (health gain expressed as percentage health improvement but fixed, not traded)

Efficiency; market vs controlled; access to additional services; physician choice; patient empowerment; evidence base

Prioritisation criteria explored

Public Preferences for Priority Setting

Year

2005

2012

2011

2009

2004

2002

2000

Study

Dolan et al. [24]

Edlin et al. [49]

Eisenberg et al. [59]

Green and Gerard [31]

Gyrd-Hansen [32]

Gyrd-Hansen and Slothuus [33]

Jan et al. [34]

Table 2 continued

Australia

Denmark

Denmark

UK

USA

UK

UK

Country

Representative sample of South Australians (n = 231)

Random sample of noninstitutionalised Danes (n = 1,895)

3,201 individuals randomly selected across all regions of Denmark

Generally representative sample of the population in the Southampton City area, UK (n = 259)

C.S. Mott Children’s Hospital National Poll on Children’s Health January 2009. Representative sample of US adults 18 years and over (n = 2,132)

582 respondents across 17 areas of England and Wales

600 members of the UK general population

Population

DCE

DCE

DCE

DCE

PTO

Pairwise choices and Thurstone scores

Trade-off

Method

South Australian community preferences about aspects of their public hospital services

Quality improvement

Preferences of the general public over scenarios describing healthcare interventions against generic social value judgments Valuations regarding: (1) preferences for equality in final health status/vs maximisation of overall health in the population, (2) preferences for treating all patients vs some patients

Age preferences related to health programmes affecting survival as well as programmes affecting nonfatal illnesses

Responsibility versus public preferences for reducing inequalities

Public preferences over maximising health and reducing inequalities in health

Study focus

Service

Health system

People

People

People

People

People

Level

Hospital services

Quality improvement

Generic patient groups

Generic patient groups

Generic programme

Generic patient groups

Generic patient groups

Context

Travel time to hospital; public transport/parking facilities; Medicare levy; waiting time for elective surgery; waiting times in casualty; rate of complications

Quality attributes: availability of new treatments; screening programmes; choice of hospital; subsidised treatment in private hospitals; focus on prevention; extra tax payment; maximum out-ofpocket payment

Severity; health gain (type not defined, specified as small, moderate, large); value for money; other treatment available Severity of illness (health states) before and after treatment—infers trade between equity and health gain (QoL)

Age; number benefiting

Health gain (QALY); responsibility (of individual or error caused by health system); inequality in lifetime health (fair innings argument)

Life-time health prospect (life expectancy and QoL, QALY), equality of health

Prioritisation criteria explored

J. A. Whitty et al.

Year

2011

2012

1997

2010

2011

2010

1993

Study

Lancsar et al. [19] (NB: same study as Baker et al. [18])

Lim et al. [35]

Lindholm et al. [51]

Louviere and Flynn [48]

Mentzakis et al. [36]

Nieboer et al. [37]

Nord [60]

Table 2 continued

Norway

The Netherlands

Canada

Australia

Sweden

South Korea

UK

Country

Survey 2: n = 25 convenience sample employees of National Institute of Public Health in Oslo (Survey 1 does not meet selection criteria for review)

General population subsample aged 50–65 years) drawn from the Dutch Survey Sampling International panel (n = 1,082)

213 respondents based on a convenience sample of university students

204 respondents, Australian population

104 randomly selected 50-year-old public employees who had been invited to participate in a health screening programme

South Korea, Internet panel, n = 800

Sample of adults living in England (n = 587), designed to be nationally representative

Population

PTO

DCE

DCE

BWS

CV; WTP

DCE

DCE

Method

Significance of the resulting health state for prioritising between different people

Long-term care services for varying types of patients

Preferences for funding drugs used to treat orphan diseases

Public perceptions and preferences for healthcare reform in Australia

WTP for a community-based intervention against cardiovascular disease (including screening) run for 10 years in Sweden

Preference for health-carer resource allocation in South Korea

Calculating distributional weights for QALYs using the Hicksian compensating variation approach to welfare measurement

Study focus

People

Service

Intervention

Health systems

Programme

People

People

Level

Generic patient groups

Long-term care

Orphan drugs

Reform

Cardiovascular disease prevention

Generic patient groups

Generic patient groups

Context

Health gain (QoL, expressed as health state after treatment); number treated

Participants provided choices for four hypothetical patient groups: hours care per week; organised social activities; transportation service; living situation; care provider; individual preferences; coordinated care service delivery; punctuality; waiting list; co-payment per week

Orphan drugs: frequency of disease (rare or common); cost of treating a patient; budget impact; severity of disease before treatment; health gain (life expectancy)

15 healthcare reform principles

Severity of disease (QoL without treatment, life-years remaining without treatment); health gain (life expectancy, QoL, also modelled as QALY); patient’s household income Effectiveness (% reduction in the average cholesterol level in the population); annual tax

Age and severity of illness: age at onset; age at death if untreated; QoL if untreated; health gain (life expectancy, QoL modelled as QALY)

Prioritisation criteria explored

Public Preferences for Priority Setting

Year

1996

2012

1998

2012

2004

2009

2000

2011

Study

Nord et al. [61]

Norman et al. [38]

Olsen and Donaldson [52]

Oremus et al. [53]

Protie`re et al. [54]

Quintal [62]

Ratcliffe [47]

Richardson et al. [63]

Table 2 continued

Australia

UK

Portugal

France

Canada

Norway

Australia

Australia

Country

Broadly representative of Australian population, higher proportion with tertiary education (n = 484)

Convenience sample of academic and non-academic employees of a British University (n = 303)

Two Portuguese municipalities representative of local population (n = 70)

Representative sample of the general population in the south-eastern region of France (n = 303)

Representative random sample of Canadians (n = 500)

Representative population living in Troms county (n = 150): 75 in Tromse, 25 in a smaller city and 50 in the rural area

Broadly representative of the Australian population (n = 552)

Australians living in Melbourne and four country towns (n = 176)

Population

PTO

Discrete choice approach: allocation of livers

PTO

CV; WTP

CV; WTP

CV; WTP

DCE

PTO

Method

Relationship between social value, individual assessment of health improvement and the severity of illness

Allocation of donor liver grafts for transplantation

Inequality in the distribution of health gains across regions and to test the assumption of variation in preferences across regions

Assess the general public’s maximum (M) WTP for an increase in annual personal income taxes to fund unrestricted access to AD medications Impact of information on nonhealth attributes on WTP for multiple healthcare programmes

WTP in increased taxation for three different healthcare programmes: a helicopter ambulance service, more heart operations and more hip replacements

Population preferences regarding the allocation of health gain between hypothetical groups of potential patients; derives equity weights

Significance of age and duration of effect in social evaluation of healthcare

Study focus

People

Generic programmes

Liver transplantation

Generic patient groups

People

People

Competing programmes

AD medication

Competing programmes

Generic programmes

Generic projects

Context

Programme

Intervention

Programme

People

People

Level

Health gain (QoL); number treated

Length of waiting time; health gain (life expectancy); age; personal responsibility; primary or re-transplant

Four efficacy scenarios: treating cognitive decline vs modifying disease progression; chance of adverse effects; additional tax Expansion of three new health programmes: heart operations, breast cancer treatment, helicopter ambulance service; household contribution to social security sickness fund/ donation Number avoiding disease; geographical equality of distribution

Health programmes providing health gains as life extension and/or QoL vs QoL only; health gain (QALY); additional tax

Gender; smoking status; income; healthy lifestyle; carer status; current total life expectancy; health gain (life expectancy)

Age or health gain (life expectancy); number benefiting

Prioritisation criteria explored

J. A. Whitty et al.

Germany

Australia

1996

2006

2003

2010

2012

2007

Ryynanen et al. [40]

Schwappach and Strasmann [45]

Schwappach [46]

Scuffham et al. [41]

Singh et al. [64]

Tang et al. [55]

Taiwan

UK

Germany

Finland

The Netherlands

2009

Ringburg et al. [39]

Country

Year

Study

Table 2 continued

Random sample (n = 1,817)

Representative sample of UK general population (n = 1,030)

Two convenience samples recruited from the UK and Australia (n = 50 each)

Convenience sample of undergraduate students (n = 154)

T1: n = 843; T2: n = 716; internet panel not representative sample

Four groups of subjects (postgraduate students n = 8, randomly selected from telephone directory n = 47, medical and nursing school students n = 104 and postgraduate nursing students n = 36)

Convenience sample (n = 136)

Population

CV; WTP

PTO

DCE

Discrete choice approach: allocation of fixed budget

Discrete choice approach: allocation of points

DCE (though not named as such by authors)

DCE

Method

WTP for a drug abuse treatment programme by the general public in Taiwan

Whether the public value safety-related healthcare improvements more highly than the same improvements in contexts where the healthcare system is not responsible

Elicit preference weights surrounding health-system attributes

Preferences for resource allocation

Preferences for resource allocation

Attitudes towards prioritisation in medicine

WTP for lives saved by Helicopter Emergency Medical Services (HEMS)

Study focus

Programme

Intervention/ service

Health systems

People

People

People

Service

Level

Drug-abuse treatment

Competing services

Reform

Generic patient groups

Generic programmes

Generic patients

Helicopter emergency

Context

Number drug users treated; youths vs all; tax/insurance contribution or voluntary donation

Causation (prevention of harms caused by a range of contexts categorised as related to healthcare, individuals themselves, or nature); number treated (health gain fixed)

Health system goals: life expectancy; infant mortality rate; patient choice; direct access to specialist; patient information; waiting time for surgery; missing out on treatment due to cost; additional tax contribution

Health-related lifestyle; socioeconomic status; age; health gain (life expectancy after treatment; QoL after treatment); receipt of extensive medical care in past

Age; initial and post-treatment QoL; health gain (life expectancy); prevalence; treatment costs

Prognosis; age; severity of disease; responsibility; socioeconomic status; cost of treatment

Number of additional lives saved; number of noise disturbances during daytime; number of noise disturbances during night time

Prioritisation criteria explored

Public Preferences for Priority Setting

2012

1996

2011

2000

Watson et al. [42]

Weaver et al. [56]

Whitty et al. [43]

Zarkin et al. [57]

USA

Australia

Central African Republic

Scotland (UK)

Country

Pilot study in two communities (n = 393): the Triad area in North Carolina (n = 197) and Brooklyn, New York (n = 196)

Public respondents (n = 161 non-representative) and 11 decision makers (current or past PBAC and ESC members)

National survey in Central African Republic representative sample (n = 1,263)

Random sample of residents of Dumfries and Galloway (n = 68)

Population

CV; WTP

DCE

CV; WTP

DCE

Method

WTP for drug-abuse treatment

Evaluating the consistency of public and decision-maker preferences for the public subsidy of pharmaceuticals

WTP for child survival

Priority-setting exercise in a Scottish healthcare organization: National Health Service (NHS) Dumfries and Galloway

Study focus

Programme

People

Service

Service

Level

Drug-abuse treatment

Generic pharmaceuticals

Competing quality improvements

Competing services

Context

Number drug users treated; women vs all; voluntary contribution

Severity of illness with current treatment; chance of success; health gain (life expectancy, QoL); cost to Government

Seven quality improvements: Facility maintenance; supervision of personnel; drugs to treat diarrheal disease; drugs to treat acute respiratory infection; drugs to treat malaria; drugs to treat intestinal parasites; drugs to treat sexually transmitted diseases; user fee

Location of care; public involvement in decision making; use of technology; service availability; patient involvement in own care; management of care; evidence of clinical effectiveness; health gain (type not specified, described as size of gain to number); risk avoidance; priority area

Prioritisation criteria explored

AD Alzheimer’s disease, BWS best–worst scaling, CV contingent valuation, DCE discrete choice experiment, ESC Economics Sub-Committee, PBAC Pharmaceutical Benefit Advisory Committee, PTO person trade-off, QALY quality-adjusted life-year, QoL quality of life, WTP willingness to pay

Year

Study

Table 2 continued

J. A. Whitty et al.

Public Preferences for Priority Setting Fig. 2 Number of publications by year (n = 41)

has jumped from two or three per year between 1993 and 2007, to eight in 2012. The majority of studies originated in Europe (including the UK) (59.0 %) and Australia (17.9 %), with some from the USA (10.3 %), and a smaller proportion from Canada (5.1 %), Asia (5.1 %) and Africa (2.6 %) (Table 3). 3.2 Methods Used to Elicit Preferences Studies employed eight different methods to elicit preferences under three broad approaches (Table 4). More than half the studies (n = 23, 59.0 %) used a discrete choice approach. The Discrete Choice Experiment (DCE) was the most commonly applied single method, employed in 18 (46.2 %) studies [19, 27–44]. The earliest identified application of a DCE approach was in 1996 [40]; although, the authors did not label it as such. Three studies employed an approach in which participants were asked to allocate a fixed number of points or transplants, or a fixed budget between competing alternatives [45–47]. In addition, one study applied a best–worst scaling variant of a choice Table 3 Number of studies by region, split by B2000, [2000 (n = 39) Region (n = 39)

B2000

[2000

Total (% of studies)

Africa

1

0

1 (2.6 %)

Asia

0

2

2 (5.1 %)

Australia Canada

2 0

5 2

7 (17.9 %) 2 (5.1 %)

Europe (other than UK)

4

11

15 (38.5 %)

UK

1

7

8 (20.5 %)

USA

1

3

4 (10.3 %)

Total

9

30

39 (100 %)

experiment [48], and one study used pairwise choices with an analysis based on Thurstone scores [49]. Contingent valuation (CV) approaches were applied in nearly one quarter (9, 23.1 %) of studies [50–58], including one study which elicited willingness to assign from public funds [58]. Other trade-off techniques were used by seven studies (17.9 %) employing a person trade-off (PTO) method [18, 59–64] and one study evaluating a trade-off between overall lifetime health prospects and equality of distribution of health [65]. The increase in studies published since 2000 was mostly accounted for by an increased application of DCE methods. 3.3 Priority-Setting Levels, Contexts and Criteria In order to determine the breadth and depth of the preferences being investigated, the studies under review were assessed on three domains: level, context, and criteria. Level refers to the unit of analysis or aggregation, that is, whether the focus was on the level of (i) the whole health system; (ii) the point of healthcare delivery, such as health services, programmes, interventions, or policies; or (iii) patient groups. Context refers to the topic of the research and the priority-setting choice that was to be made. The criteria were the variables or attributes on which participants based their choices between different options. Table 5 summarises the level of priority setting and the contexts in which the relevant studies were conducted. Four (10.2 %) studies related to prioritising between goals or characteristics at the level of the health system; 16 (41.0 %) studies related to prioritising between the provision of different programmes, services or interventions; and 18 (48.7 %) studies prioritised the provision of

J. A. Whitty et al. Table 4 Preference elicitation methods employed, split by B2000, [2000 Method (n = 39 studies) Discrete choice approaches

Contingent valuation Trade-off approach

Variant

B2000

[2000

Totala (% of studies)

Discrete choice experiment

2

16

18 (46.2 %)

Allocation of points

1

2

3 (7.7 %)

Best–worst scaling

0

1

1 (2.6 %)

Pairwise choice

0

1

1 (2.6 %)

Willingness to pay

3

5

8 (20.5 %)

Willingness to assign

0

1

1 (2.6 %)

Person trade-off

2

5

7 (17.9 %)

Other trade-off approach

0

1

1 (2.6 %)

9

31

Total methods

40a

a The 39 studies employed a total of 40 preference elicitation tasks (with the Social Value of a QALY study employing both a discrete choice experiment and person trade-off task in the same sample[18, 19]). Hence, the number of methods exceeds 100 % of the total number of studies

Table 5 The priority-setting levels and contexts The priority-setting level (n = 39 studies) Health systems

Healthcare policy/programme/service/intervention

Patient groups

Context

B2000

[2000

Total

Reform

0

3

3

Quality improvement

0

1

1

Sub total

0

4

4 (10.2 %)

Alzheimer’s disease drugs

0

1

1

Cardiovascular disease prevention

1

0

1

Competing programmes

1

2

3

Competing quality improvements

1

0

1

Competing services

0

2

2

Drug-abuse treatment

1

1

2

Helicopter emergency

0

1

1

Hospital services Long-term care

1 1

0 0

1 1

Orphan drugs

0

1

1

Prevention vs treatment

0

2

2

Sub total

5

11

16 (41.0 %)

Generic patients

2

10

12

Generic pharmaceuticals

0

1

1

Generic programmes/projects

0

5

5

Liver transplantation

1

0

1

Sub total

4

14

19 (48.7 %)

9 (23.1 %)

30 (76.9 %)

39 (100 %)

Total

healthcare between different patient groups. The following sections discuss the contexts and the criteria for which preferences have been considered within each of the three levels. 3.4 Health Systems Four studies exploring trade-offs between the preferred goals and characteristics for delivery of a health system were identified [29, 33, 41, 48]. As shown in Table 5, the

context for these studies was either health reform or quality improvement. Akkazieva et al. [29] reported preferences in Hungary for a health system that was not cost constrained, was based on solidarity, and empowered patients. They also found that participants would accept not having a choice of physician in order to avoid co-payments. Scuffham et al. [41] found a range of characteristics representing the level of health, equity, responsiveness and healthcare financing affected preferences in Australia and the UK. In another

Public Preferences for Priority Setting

Australian study, Louviere and Flynn [48] used best–worst scaling (Case 1) to identify preferences for 15 healthreform criteria, finding quality and safety to be the most highly valued, and a culture of reflective improvement and innovation to be the least valued. However, they also identified a relatively poor self-reported level of participant understanding around some of the principles explored, with between 47 and 85 % of respondents indicating they understood each criteria. In a larger DCE study, GyrdHansen and Slothuus [33] reported that Danish citizens were willing to trade between a range of quality attributes used to described the health system, and were willing to pay for health-system improvements; however, the amount they were willing to pay was sensitive to the payment vehicle (i.e., additional income tax or maximum user charges). Three of the studies that addressed health-system priority-setting preferences used convenience samples, suggesting the findings of these studies cannot be generalised [29, 41, 48]. Nevertheless, overall, the four studies indicate that the relevant criteria to explore need to be carefully aligned with the nuances of the country and culture in which a study is undertaken. They also highlight the importance of careful explanation of health-system concepts to participants and that framing might (or is perhaps likely to) impact on preferences in this context. 3.5 Healthcare Delivery Sixteen studies were identified that quantified preferences for policies, programmes, services or interventions. Studies explored the importance of both health and non-health benefits associated with healthcare. 3.5.1 Health Benefits Several studies have explored the relative value of different types of health benefit; namely, prevention, treatment, disease modification, symptom control and safety. Preferences for prevention versus treatment were explored in two studies, with conflicting findings. Bosworth et al. [30] reported the marginal utility associated with avoiding deaths to be about twice as high for prevention than treatment policies. Conversely, Corso et al. [50] reported a significantly greater WTP out of pocket for a treatment programme (US$665) than for a prevention programme (US$223), even though they were described to avert the same number of deaths. Preferences for prevention versus treatment might be dependent on the condition being treated. Bosworth et al. [30] found participants were more likely to support cancer prevention policies and less likely to support cancer treatment policies than equivalent policies targeting other conditions.

In a further CV study exploring preferences for the public provision of medications targeting Alzheimers’ disease, Oremus et al. [53] found the Canadian public is willing to pay increased taxes to fund unrestricted access to medications, with WTP varying by medication specification. WTP was higher for medications that modify disease as opposed to treat symptoms and for those that are associated with less adverse effects. Whilst reduced adverse effects or safety is important, priorities may vary according to the cause of harm. Singh et al. [64] used the PTO method to assess the relative preference of the UK population between interventions that prevented harms caused by a range of contexts categorised as related to healthcare, individuals themselves, or occurring by nature. They found preferences differed substantially according to differences in context, rather than according to the overarching harm category. Participants valued interventions to prevent hospital-acquired infections most highly, but valued avoiding injury due to drug errors similarly to avoiding disease from naturally occurring genetic disorders. Interventions to prevent injury to staff, related to lifestyle, and designed to avoid sports injuries were less valued. The authors concluded that preferences depend subtly on context and as a consequence suggest the use of results from public preference surveys to directly inform policy is premature. More recently, Mentzakis et al. [36] undertook a pilot DCE in a convenience sample to explore preferences for prioritising public funding for orphan drugs. Orphan drugs were described as those treating a rare condition (\5,000 cases annually in a population of 10 million). Their results suggested severity and treatment effectiveness were of similar importance in the context of both rare and common diseases, with preferences weights for these attributes being larger than for the other criteria. However, preferences for the treatment of rare and common diseases were found to be similar; respondents were not willing for the government to spend more for drugs or to pay more per life year gained for a rare disease. Several studies have explored preferences for targeting a specific disease or condition. In an early CV study exploring WTP through taxation for three different programmes, Olsen and Donaldson found no statistically significant difference in the publics’ WTP for a helicopter ambulance (median 200 Norwegian kroner [NOK] per annum) or heart operation (median 200NOK) programme [52]. However, participants were willing to pay less (150NOK) for a hip-replacement programme, despite the hip programme giving the better QALY gain to society. The mean WTP ranged from 0.2 to 6.7 NOK/QALY gained depending on the underlying assumptions (e.g., discounting rate) and the programme in question. In another national survey using CV conducted in the Central African

J. A. Whitty et al.

Republic and published in 1996, Weaver et al. [56] found the median WTP for seven quality improvements differed across the targeted illness, ranging from US$7.98 for drugs to treat malaria to US$16.61 for drugs to treat diarrheal diseases. Taken together, these two studies suggest preferences are affected by the disease or circumstance being targeted. Finally, two studies prioritising between healthcare programmes identified WTP for drug-abuse treatment programmes delivered for different patient groups. Whilst both studies found the public were willing to pay social health insurance premiums or voluntary contributions for a drug-abuse treatment programme [55, 57], one study in Taiwan identified a preference to target programmes to youths [55]. The second study in the US found equal WTP for a programme targeting women as for all drug users [57]. Neither study found a value associated with a programme that would target a larger number (5,000 over 1,000 drug users in Taiwan, 500 over 100 drug users in the US). This is against conventional expectations, but in line with known methodological limitations of the CV method [66]. Overall, the majority of studies investigating the relative strength of preferences for priority setting between healthcare programmes, services or interventions support the notion that the size of health benefit alone is not all that is important to the public in priority setting, and reinforce the importance of context. The type of health benefit (e.g., prevention, harm reduction, etc.), the condition being targeted and the population likely to benefit may all impact preferences. Whilst the studies providing this evidence are of mixed quality in terms of the sample from whom preferences were elicited, most studies employed large potentially representative samples [30, 50, 53, 55, 56, 64], albeit derived from a single country. 3.5.2 Non-Health Benefits A range of studies exploring preferences around the provision of health services or programmes found that nonhealth benefits are valued alongside the health benefits provided; although, often to a lesser extent. For example, in a study assessing ‘willingness to assign’ from public funding, Costa-Font and Rovira [58] found that although programmes promoting health received the higher valuation, programmes promoting equity and process-related benefits, such as reduced treatment waiting times, were also valued highly. Protie`re et al. [54] found healthcare programmes were more valued by participants in a CV study when they were provided with additional information on the process of treatment and the quality of care; suggesting that people are willing to pay an additional social

insurance contribution for these characteristics in a public programme in addition to the health benefits achieved. In the context of choice between the provision of competing public hospital services, Jan et al. found an improvement in complication rates increased the probability of choosing a hospital; waiting times for casualty, waiting times for elective surgery, and, anomalously, parking and transport facilities reduced the probability a hospital service was chosen. Travel time and the required increase in the Medicare (tax) levy to fund the service did not statistically significantly affect choice of hospital [34]. Watson et al. [42] explored the potential of a DCE, described using ten priority-setting attributes, to develop weights to prioritise between healthcare services. Attributes related to the process of care, the involvement of the public and patients in decision making and care, the evidence base, whether the service represented a risk reduction, health gain, and the level of the priority area addressed. All attributes except risk avoidance were significant, with a large health gain to many (type of gain not defined), care provided in teams, use of the latest cuttingedge technology, and 24-hour service availability found to be the most important characteristics of the service. Addressing local priorities was valued more highly than addressing national priorities. Nieboer et al. [37] explored the preferences of a general population subsample aged 50–65 years in the Netherlands around criteria of importance in the delivery of long-term care services. Participants responded to a DCE related to four hypothetical patient scenarios varying by nine process characteristics and an out-of-pocket co-payment for the service user. The authors found that having the same person delivering care and transportation services was associated with greatest value; punctuality and room for individual preferences was associated with a low value. The authors conclude their results suggest long-term care services represent different value for different types of patients and that the value of a service depends upon the social context. In a DCE study identifying preferences and WTP for a helicopter emergency service, Ringburg et al. [39] found a marginal WTP of €3.43 per month per household for one additional life saved in a month. They also identified disutility associated with noise disturbance, with the disutility associated with noise disturbance at night being eight times as great as that for noise disturbance during the day. However, the weight assigned to avoiding noise disturbance was small relative to saving a life. Whilst these studies highlight the importance of nonhealth benefits in the public’s preferences for priority setting, with the exception of Nieboer et al. [37] and Protie`re et al. [54], these studies used small and/or convenient

Public Preferences for Priority Setting

samples that were not representative of the general population [39, 42, 58]. 3.6 Patient Groups Eighteen studies prioritised the treatment of groups of individuals, where the groups were described by a range of criteria or characteristics, and the criteria were traded against each other in the priority-setting decision. This included several studies (e.g., [43]) whose priority question was between programmes or interventions; but, because these were labelled generically and described only in terms of the people who would be treated, they are included in this section rather than in the previous section on programmes. Studies prioritising patient groups were often allocating generic or unspecified interventions or healthcare between groups, with the exception of one study which allocated liver transplants [47]. A range of prioritisation criteria was explored, but criteria were described differently in different studies. Further, although some researchers recognised the multi-attribute potential of the DCE method [38], perhaps understandably no study explored all the criteria of potential interest simultaneously. Generally, it is therefore not possible to assess trade-offs between all criteria and their relative weight in priority setting in a common context and population. The criteria explored in identified preference studies and evidence supporting or refuting their use in prioritisation are summarised below. 3.6.1 Evidence to Prioritise by Improvement in Health or Health Gain Many papers have focussed on the question of what criteria should be considered in priority setting alongside health outcome, and all papers that have explicitly considered health gain as a criteria in the trade-off have clearly indicated that an improvement in health or health gain is consistently highly valued by the public in priority setting [19, 30–32, 35, 36, 38, 42, 43, 45–47, 49]. However, definitions of health gain differ across studies with some studies focusing only on life expectancy (e.g., Norman et al. [38]), others focusing on quality of life (QoL) (e.g., Gyrd-Hansen [32]), and others defining gain in terms of both life expectancy and QoL in combination as QALYs (e.g., Lancsar et al. [19], Baker et al. [18]). For example, Green and Gerard [31] found a large rather than very small health gain (but did not define the type of health gain) to be the most important attribute for the UK public when prioritising patients for resource allocation, Norman et al. [38] found a preference to prioritise according to health gain, described in terms of life expectancy, Schwappach [46] found health state after treatment to be the most important

priority-setting criterion, and Lancsar et al. [19] found health gain described in terms of QALYs to be the most important criterion for the UK public. Most studies have also found a range of additional criteria to be important for prioritising healthcare or health gains between individuals, either alongside or independent of the overall level of health gain [38, 43]. In large, potentially representative samples (and one smaller sample [43]) originating from European, Asian and Australian studies, investigators have variously found a relatively high social value for health gain if it were accrued to nonsmokers, carers, individuals with a low income, younger individuals, individuals with a lower life expectancy, individuals for whom there was no alternative treatment available, or those with more severe illness [18, 31, 32, 35, 38]. Norman et al. [38] reported a wide distribution of equity weights for health gain for different patient groups described by some of these criteria in an Australian study, highlighting the substantial relevance of criteria outside of health gain in their study. Further, in one part of the UK Social Value of a QALY study, a QALY grid approach based on PTO questions was used to derive the relative value of age and severity of illness as priority-setting criteria compared with health gain (described as a QALY gain). This study found the same health gain given to subgroups of different age and illness severity was valued between 2.75 and 4.0 times as highly for the most as compared with least priority groups [18]. However, the literature is not entirely consistent in finding criteria other than health gain to be important considerations for priority setting. In an exception, Lancsar et al. [19] used a novel DCE analytic approach to derive distributional weights for the QALY in the UK. This study was undertaken as part of the UK Social Value of a QALY study, in the same sample as the Baker et al. [18] PTO study. The results of the Lancsar et al. DCE study do not generally support weighting QALYs for criteria other than health gain (specifically severity of illness or age) explored in the study, except in a small number of cases, where a relatively small weight might be appropriate. Reasons why the findings of Lancsar et al. [38] may diverge from the finding of other studies might include the very varied criteria and their definitions used across studies. For example, Norman et al. [38] did not consider severity of illness or age, Green and Gerard [31, 32] and Gyrd-Hansen [31, 32] considered severity of illness but not age; and Lancsar et al. [19] did not consider some criteria that were found to be relevant in the other studies. There were also differences in the way health gain was defined (in terms of QALYs [19], life expectancy [38], QoL [32] or undefined [31]). Studies defining health gain in terms of only QoL or life expectancy are unlikely to reflect actual healthcare prioritisation considerations which often aim to both improve QoL and

J. A. Whitty et al.

extend life. Regardless of differences in methodology between studies, Norman et al. [38] found their equity weights were sensitive to the value of the gain in life expectancy; as the gain increases, equity weights tend towards the conventional QALY-type model where it does not matter to whom the gain accrues, consistent with the Lancsar et al. [19] findings. Thus, the results of both the Lancsar et al. [19] and Norman et al. [38] studies imply that the size of a health gain on offer affects the derivation of preferences around patient characteristics and equity weights, with the consideration of criteria alongside health gain being more important when a health gain is relatively small. This relates to a more general point that in the extreme where health gain is held constant (e.g., in PTO studies, as was the case with the Baker et al. [18] study) this can artificially increase the importance of the attributes that are allowed to vary. Bryan et al. [44] used discrete choice methods to investigate whether the marginal value of improvements in health gain is proportional to the level of improvement, described by improvements in chance of treatment success, patient QoL after treatment, and number of patients treated. This is a key assumption underlying the QALY as a measure of health outcome and therefore CUA. The authors report their findings broadly support this assumption, although preferences were not perfectly proportional by health gain. However, these indicators of health gain were not traded against other criteria. Nord et al. found that the duration of benefit and number benefiting matters, with participants in a PTO preferring to give a shorter health gain to a larger number of people [61]. Participants valued extending the life of ten patients by 10 years equally to extending the life of 5.5 patients by 20 years. This is consistent with a time preference and the application of conventional discounting in economic evaluation. 3.6.2 Severity of Illness—Before and After Treatment Most studies that have explored the impact of severity of illness before treatment on preferences for prioritisation of patients for healthcare have reported it to be an important criterion [18, 27, 32, 35, 40, 43]. These studies were of varying quality, with a mixture of large, representative [27, 32, 35] and small or convenience samples [40, 43]. This finding is consistent with that of Shah [23], who reviewed the literature on severity of illness and found the majority of studies suggest that on average people are willing to sacrifice some health gain to prioritise the severely ill. However, as highlighted by Shah, the studies exploring severity of illness are somewhat inconsistent in their definition of severity and methodological approach. One study in the current review presented an exception to the overall

finding in favour of the relevance of severity of illness. Lancsar et al. [19] found severity of illness was relatively less important than health gain in most circumstances tested, and found, where severity was important, that less weight was given to the most rather than less severe. Diederich et al. [27] found severity of illness (expressed as health status) to be a more important criterion for prioritisation than age, socioeconomic status or lifestyle. However, they did not assess the relative importance of criteria, including severity of illness, against health gain. Baker et al. [18] reported a preference to prioritise health improvements for those with a moderate severity of illness (described as 20–60 % QoL) over those with a very poor or very good level of health. However, in their PTO, they did not directly trade severity of illness against health gain. Other studies have explicitly traded severity of illness and some measure of health gain. Green and Gerard [31] found treating patients who were severely affected to be less than half as important for the UK public as providing a large rather than very small health improvement. In a Korean sample, Lim et al. [35] reported a willingness to forego 0.39 QALYs to prioritise patients with a shorter lifespan, and a lesser 0.07 QALYs to prioritise patients with a lower QoL (precise extent of detriments in lifespan and QoL were unspecified). Exploring the impact of initial and final health states on preferences for priority setting, GyrdHansen [32] found respondents were inclined to give priority to those in a more severe health condition; this was particularly the case when the severity was described by selected health dimensions (e.g., pain/discomfort). Thus, the type of health dimension affected might affect social preferences for the treatment of others differently to individual preferences for the treatment of oneself [32]. Some authors have also explored the importance of severity of illness after treatment for preferences. When exploring the impact of severity of a health state after treatment on preferences for saving a life, Richardson et al. [63] found a preference not to leave people in a severe health state, after adjusting for the level of health improvement. 3.6.3 Age Recipient age has been widely explored as a potential prioritisation criterion, with mixed findings. Some studies have not supported age as a relevant or important prioritisation criterion when considered relative to other potential prioritisation criteria. For example, Diederich et al. [27, 28] found the importance weight for age as a criterion to be small on a normalised scale relative to severity of illness (expressed as health status) or existing restrictions in QoL, whilst Lancsar et al. [19] did not generally find support for weighting QALY gain by age (age of onset or age of death).

Public Preferences for Priority Setting

Others have found age to be an important prioritisation criterion. Dolan and Tsuchiya [65], when using a trade-off approach, found the value of an additional QALY decreases as the age of the recipient increases from 50 to 70 years (relative weights 7.11 for a 50-year-old and 1.00 for a 70-year-old). In a large, representative US sample, Eisenberg et al. [59] found most respondents favoured programmes benefiting 100 children aged 10 years over as many as 1,000 adults aged 60 years, inferring a relative valuation of 10:1 for treating 10-year-olds versus 60-yearolds; however, they did not estimate an average trade-off (representing a preference weight) across their sample. Nord et al. [61] found the participants in a PTO prioritised younger patients, for both life-saving and health-improving interventions. For example, extending the lives of four 20-year-olds was considered to be equivalent to extending the lives of ten 60-year-olds. Baker et al. [18] reported a preference to prioritise adults aged 20–60 years over either older adults (60–80 years) or younger individuals (aged 0–20 years). However, none of these studies [18, 59, 61, 65] explicitly allowed age to be traded against the size of the health gain. Schwappach [46] also found a large effect for age, with results indicating ‘‘that [between the ages of 20 and 60 years] for every year persons in one group are older than in the other, [gain in] life expectancy produced by treatment has to be nearly one additional year as compared to the younger group to compensate’’. Ratcliffe [47] found participants were willing to give 1.49 fewer liver grafts to a group of patients in need of a transplant rather than a competing group, for each 1-year increase in the age difference between the two groups. In the context of prioritisation of a range of policies aimed at prevention or treatment, Bosworth et al. [30] reported participants were more likely to support prevention or treatment policies that target children than adults, and Ryynanen et al. [40] found preferences prioritising children and giving a negative priority to the elderly. However, some of these studies only included a very restricted range of other criteria in the trade-off (e.g., Ryynanen et al. [40]) and some others did not include interaction terms between age and health gain (e.g., Schwappach and Strasmann [45] and Schwappach [46]), which might be anticipated to be relevant. When weights were derived for different age categories, older adults receive a lower priority; however, the relative ranking of children and younger adults is inconsistent across studies. Diederich et al. [27, 28], although finding age relatively unimportant compared with other criteria, did find differences across age groups with a preference rank order of 43 years [ 25 years [ 68 years [ 87 years. Schwappach and Strasmann [45] ranked children [ employable age [ teenager [ seniors. Nord et al. identified monotonic preferences with a rank order 10 years

[ 20 years [ 60 years [ 80 years [61]. Schwappach [46] and Ratcliffe [47] assumed linear monotonic preferences for age, and Eisenberg et al. [59] and Dolan and Tsuchiya [65] each only tested preferences for two ages. 3.6.4 Socioeconomic Status Diederich et al. [27] found little support from the German public for prioritising based on family or occupational status, as compared with criteria related to medical need (severity of illness or level of QoL restriction). Conversely, Ryynanen et al. [40] and Schwappach [46] found a preference to prioritise patients of lower socioeconomic status in Finland and Germany, respectively. Lim et al. [35] found the public in South Korea were willing to prioritise those with a lower household income, forgoing 0.83 QALYs to treat those with a one quartile lower income. Further, Norman et al. [38] found a willingness to prioritise life-year gains to those with a lower income. 3.6.5 Carer Status Norman et al. [38] found a direct preference to prioritise life gain to carers. Other studies have not directly explored the implication of carer status for preferences; however, it is possible that carer status is indirectly inferred by participants and correlated with preferences for other characteristics; for example, it might in part explain preferences observed in favour of treating adults of child-bearing age. 3.6.6 Lifestyle/Responsibility Several studies have explored the impact of lifestyle and/or responsibility for illness, which are two distinct but closely related concepts, on preferences for prioritising patients. Overall, studies are fairly consistent in suggesting these concepts may be relevant considerations in priority setting in some contexts, but some studies suggest them to be relatively minor considerations as compared with some other prioritisation criteria. Diederich et al. [27] found no support for prioritising based on a healthy versus an unhealthy lifestyle, as compared with criteria related to medical need (severity of illness or level of QoL restriction). Conversely, Norman et al. [38] found a relatively high social value to be associated with providing additional life years to non-smokers and to a much lesser extent those who lived a generally healthy lifestyle (in terms of diet and exercise) as compared with their counterparts. In a small German convenience sample, Schwappach [46] also reported a relatively high preference to treat those who have a healthy lifestyle (defined as not smoking, healthy eating, moderate alcohol intake, and avoidance of high-risk activities; but not

J. A. Whitty et al.

defined in terms of whether the disease was self-induced), equating to approximately the same value as an additional 10 years’ life expectancy. Edlin et al. [49] explored the impact of responsibility, in terms of both whether an individual could be considered ‘blameworthy’ and whether a health-system error was responsible for illness, on preferences for priority setting. They concluded that public preferences supported responsibility to be a significant but not pivotal consideration in priority setting. Ryynanen et al. [40] reported a preference to negatively prioritise those with selfacquired disease, but no impact of an illness acquired by negligent behaviour on prioritisation preferences (the difference between these two constructs was not clearly defined by the authors). In the specific context of allocation of liver transplants, Ratcliffe [47] observed a strong preference to prioritise naturally occurring rather than alcoholic liver disease (equating to willingness to forego 9.5 years in survival in the competing group to do so). A preference to prioritise individuals with illness acquired through health-system error, and to assign a lower priority to those with lifestyle-related disease is consistent with the findings of Singh et al. [64], who reported similar findings in the context of prioritising programmes to prevent harm. 3.6.7 Prior Medical Care In a small German convenience sample, Schwappach [46] found a relatively strong preference to treat those who had not received a high level of previous life-saving, complex or costly medical intervention, equating to approximately the same value as an additional 13 years’ life expectancy. This is converse to the findings of Ratcliffe [47], who reported a significant preference to prioritise liver transplants to those requiring a re-transplantation rather than first transplantation, trading nearly two additional life years in order to do so. 3.6.8 Availability of Effective Alternatives Green and Gerard [31] found a significant but relatively small preference to prioritise patients for whom there was no effective alternative treatment available. None of the other identified studies explored the impact of this criterion on prioritisation preferences of the public. 3.6.9 Cost or Cost Effectiveness of Treatment Few studies have explored the impact of the cost of the treatment on prioritisation decisions (as distinct from a participant’s WTP for provision of a healthcare programme). However, Ryynanen et al. [40] found that how

‘expensive’ a treatment was did not impact on preferences for priority setting. Schwappach and Strasmann [45] found respondents preferred to allocate ‘priority points’ when the treatment programme had below-average costs to patients. Conversely, when the cost is borne by the public funder, Green and Gerard [31] found achieving very good rather than very poor value for money to be almost equally valued as providing a large rather than very small health improvement. 3.6.10 Disease Prevalence Schwappach and Strasmann [45] found respondents preferred to allocate ‘priority points’ to patients when the programme targeted common (15,000–20,000 patients per year) rather than rare (1,000 patients per year) diseases. This is consistent with the findings of Mentzakis et al. [36] that the public do not want to prioritise orphan drugs used to treat rare as opposed to common diseases. 3.6.11 Equality Based on a small convenience sample in Norway, Nord [60] reported participants in a PTO emphasised equality in the value of life and entitlement to treatment rather than the level of health that could be achieved after treatment. Quintal [62] reported findings from a PTO in which most (70–80 %) participants were willing to trade health (in terms of the number of people avoiding disease) in exchange for an improvement in the geographical equality of the distribution on the gain, with most people willing to forego 10–30 % of total health gain to retain geographical equality. 3.6.12 Waiting Times In the context of allocating liver grafts, Ratcliffe [47] reported a preference to allocate to those who had waited longest, with participants willing to trade approximately 3 months’ survival to treat people who had been waiting for an additional month.

4 Discussion The findings from this review suggest a developing literature on the preferences and relative trade-offs the public are willing to make in healthcare priority setting. The number of studies in this area has been relatively steady until the last 5 years, since which time there appears to have been an exponential growth in reporting. This growth suggests an increasing interest by researchers and/or policy makers in considering trade-offs and developing

Public Preferences for Priority Setting

prioritisation weights from the public’s perspective. The findings of this study indicate that the contexts explored are diverse and confirm the importance of a wide range of process and non-health-related characteristics in priority setting, alongside health outcomes. A substantial body of the literature relates to prioritisation between hypothetical patient groups, and to some extent this literature is linked to underlying distributive assumptions (such as utilitarianism versus egalitarianism) and evaluation frameworks (such as cost–utility analysis employing the QALY as an outcome measure) in healthcare. Although results of selected studies support a generally utilitarian (Lancsar et al. [19]) or egalitarian (Nord [60]) viewpoint, the vast majority suggest public preferences lie on the continuum between these extremes. Overwhelmingly, studies that included health gain (defined in terms of life expectancy, QoL, or both life and quality extensions via QALYs) as a criterion supported the importance of health gain relative to other criteria for prioritising healthcare. Severity of illness without treatment was also often, though not exclusively, considered to be relatively important. However, inconsistency in the implications of the data between studies was often observed. It seems possible that when criteria related to recipients, such as age, socioeconomic status and lifestyle are evaluated using preference methods that force a trade-off relative to (multiple) other prioritisation criteria, they may become relatively less important. This is speculative, but perhaps highlights the benefits of using a multi-criteria, choicebased preference elicitation method to evaluate preferences for prioritisation. This review has provided an overview of the methodological approaches to preference elicitation specifically in a priority-setting context. We find a growth in the application of the DCE method, which is aligned with the popularity of the method in healthcare more broadly [67, 68]. Our focus solely on studies that present a choice, resulting in the prioritisation criteria being evaluated in the context of a trade-off, is an important aspect of our review. We identified studies that used a WTP approach in which characteristics relevant to priority setting were traded against a payment vehicle, and which could potentially be applied in a cost–benefit evaluation framework to inform priorities. We also identified studies that assessed relative trade-offs between different non-monetary prioritisation criteria on a common scale, some of which assessed trade-offs for health gain against other prioritisation criteria. The findings of this review suggest some substantial inconsistencies relating to some criteria explored (such as recipient age and socioeconomic status). Overall, it would seem that environment, context, sample and the other criteria that are included in the choice are likely to impact the relative trade-offs revealed in the studies. Other differences across

studies likely to impact implied trade-offs are the different methods used and quality of the studies undertaken. The review findings suggest any attempt to develop prioritisation weights for application in priority setting would need to be country and context specific. Further, such attempts need to undertake preliminary research to identify the best approach to elicit preferences in the required context, and give careful consideration to including all relevant criteria for assessing trade-offs between different alternatives. Given current interest in the possible application of MCDA or QALY weights in priority setting, an exploration of the potential to use the preferences derived from these studies in decision making would be of interest. Few identified studies translated their preference data into weights for potential use in decision making. There were three key exceptions. In two studies, distributional or equity weights were estimated [19, 38]. Findings from the two studies were inconsistent, in that Lancsar et al. [19] reported small weights and Norman et al. [38] reported relatively large weights; however, key differences in the studies discussed above make the finding that they produced different results unsurprising. In the third study, Watson et al. [42] attempted to use their preference weights to rank competing services in a priority-setting exercise in Scotland. However, others found preferences to differ by subtle context and caution against using preferences in priority setting until a greater understanding of their nuances can be garnered [64]. It is possible that we may have missed relevant published studies that could expand on these findings; although, our comprehensive and systematic search approach was designed to limit the potential for omission of relevant work. A key challenge in eliciting preferences for priority setting is the question of whose preferences should be considered. The ideal preference population depends on the decision context. However, when decisions may affect more than one patient or population subgroup, it is generally considered preferable to use preferences from the general public without self-interest, ideally elicited under a ‘veil of ignorance’ [69], to inform priority setting [70, 71]. Therefore, this review was limited to studies in the general public. Ideally, samples would be large and selected so as to be representative of the wider population. This is important, as several studies included in this review observed preference heterogeneity around the criteria that might be important to consider in priority setting [38]; therefore, having a sample that accurately reflects the population of interest on characteristics that might impact preferences is important. Further, Norman et al. [38] found respondents tended to favour treating individuals who were similar to themselves. However, many of the studies in this review used a convenience sample (e.g., [29, 36, 39–41, 46, 47, 60]), and even those studies aiming for a representative

J. A. Whitty et al.

sample often found their sample did not reflect the population of interest across all characteristics (e.g., [38, 43]), limiting the generalisability of their findings. This illustrates a key potential challenge to implementing preference weights in decision making through either equity weights or MCDA. If we accept that valuation of non-health outcomes, process and distributional characteristics in decision making is desirable and the population whose preferences are of interest can be agreed on (both these points are contentious), deriving preferences simultaneously across all criteria of interest and from a sufficiently representative sample is challenging. Overall, the findings suggest caution in directly incorporating public preferences as weights for priority setting unless the methods used to elicit the weights can be shown to be appropriate and robust in the priority setting context. Further research is required to elaborate on the studies here and more fully understand the methodological as well as healthcare context and participant-specific characteristics that impact preferences in priority setting. We were unable to formally assess the quality of included studies and this is a limitation of the review. The development of quality criteria for all stated preferences studies, regardless of methodological approach, would be valuable in assisting future reviews of this nature. Whilst several checklists are now available to guide an assessment of the quality of DCE studies [68, 72, 73], this is not the case for all methods included in this review. We therefore relied on peer review publication as a proxy for quality. This is not ideal, and future research could aim to extend the existing DCE checklists to develop quality criteria that could be applied across all stated preference studies. The need to develop such criteria is further highlighted by this review, as it is apparent that to enable a greater understanding and ability to compare and interpret the findings of preference studies in a meaningful way, increased consistency in the methodological approach used in preference elicitation studies is required. It appears necessary to make the context clear to participants; generic studies without a healthcare context may not be generalisable to specific priority-setting contexts. The preference elicitation methods used, the definition of criteria to be valued, and the framing of questions to derive preferences need to be consistent, if study findings are to be compared. Consistency would allow us to build a body of knowledge that with time and similar findings could provide a sufficient level of evidence on the public’s priority setting preferences to lend strong support to their consideration in decision making. Importantly, the literature suggests that the omission of criteria of relevance might explain some of the inconsistent findings to date. As a consequence, methods that allow for the simultaneous consideration of multiple criteria alongside health gain may be preferable to derive weights (e.g., the

DCE), as compared with those that allow only a small number of trade-offs to be explored (e.g., the PTO or CV). Further, given the inconsistency in findings around some criteria, we would suggest that mixed methods studies that attempt to validate the implications of preference studies would be beneficial. Whilst an external gold standard indicating the public’s priority-setting preferences is not available for validation, studies could, for example, present the implications of their findings to participants and investigate the stability of their preferences in response to this feedback. In conclusion, many criteria have been explored across a diverse range of priority setting contexts and found to be important considerations for the public. However, few studies have translated their preference data into weights for potential use in decision making. Further, the prioritysetting context and methodological approach to preference elicitation may both be important in explaining nuances and inconsistencies in priority-setting preferences. Therefore, caution is required in directly incorporating public preferences as weights for priority setting unless the methods used to elicit the weights can be shown to be appropriate and robust in the priority-setting context. Acknowledgments Funding source

None declared.

None declared.

Conflict of interest

None declared.

Author contributions This review was undertaken in response to an invitation from The Patient. JW, EL, KR and JR developed the review scope and methods. KR, XG and JW selected studies for inclusion, extracted information and classified the studies. JW synthesised the review findings and drafted the manuscript. KR and XG assisted with manuscript drafting. EL and JR assisted with interpretation of the findings from the literature synthesis. All authors critically reviewed the manuscript for intellectual content and approved the final version. JW is overall guarantor for the paper.

References 1. Ham C. Priority setting in health care: learning from international experience. Health Policy. 1997;42(1):49–66. doi:10.1016/ S0168-8510(97)00054-7. 2. Sabik LM, Lie RK. Priority setting in health care: lessons from the experiences of eight countries. Int J Equity Health. 2008;7:4. doi:10.1186/1475-9276-7-4. 3. Australian Government Productivity Commission. Impacts of advances in medical technology in Australia. Melbourne: Australian Government Productivity Commission; 2005. 4. American College of Physicians. How can our nation conserve and distribute health care resources effectively and efficiently? Philadelphia American College of Physicians; 2011. 5. Coast J, Smith R, Lorgelly P. Should the capability approach be applied in health economics? Editorial. Health Econ. 2008;17(6):667–70. doi:10.1002/hec.1359.

Public Preferences for Priority Setting 6. Littlejohns P, Weale A, Chalkidou K, Faden R, Teerawattananon Y. Social values and health policy: a new international research programme. J Health Org Manage. 2012;26(3):285–92. 7. Schwappach DL. Resource allocation, social values and the QALY: a review of the debate and empirical evidence. Health Expect. 2002;5(3):210–22. 8. Olsen JA, Richardson J, Dolan P, Menzel P. The moral relevance of personal characteristics in setting health care priorities. Soc Sci Med. 2003;57(7):1163–72. 9. Stafinski T, Menon D, Philippon DJ, McCabe C. Health technology funding decision-making processes around the world: the same, yet different. Pharmacoeconomics. 2011;29(6):475–95. doi:10.2165/11586420-000000000-00000. 10. Mooney G, Jan S, Wiseman V. Examining preferences for allocating health care gains. Health Care Anal. 1995;3(3):261–5. 11. Ryan M, Scott DA, Reeves C, Bate A, van Teijlingen ER, Russell EM, et al. Eliciting public preferences for healthcare: a systematic review of techniques. Health Technol Assess. 2001;5(5):1–186. 12. Brazier J. Measuring and valuing health benefits for economic evaluation. Oxford: Oxford University Press; 2007. 13. Diener A, O’Brien B, Gafni A. Health care contingent valuation studies: a review and classification of the literature. Health Econ. 1998;7(4):313–26. doi:10.1002/(SICI)1099-1050(199806)7: 4\313:AID-HEC350[3.0.CO;2-B. 14. Donaldson C, Birch S, Gafni A. The distribution problem in economic evaluation: income and the valuation of costs and consequences of health care programmes. Health Econ. 2002;11(1):55–70. 15. Olsen JA, Smith RD. Theory versus practice: a review of ‘willingness-to-pay’ in health and health care. Health Econ. 2001;10(1):39–52. 16. Birch S, Donaldson C. Valuing the benefits and costs of health care programmes: where’s the ‘extra’ in extra-welfarism? Soc Sci Med. 2003;56(5):1121–33. 17. McIntosh E, Donaldson C, Ryan M. Recent advances in the methods of cost–benefit analysis in healthcare. Matching the art to the science. Pharmacoeconomics. 1999;15(4):357–67. 18. Baker R, Bateman I, Donaldson C, Jones-Lee M, Lancsar E, Loomes G, et al. Weighting and valuing quality-adjusted lifeyears using stated preference methods: preliminary results from the Social Value of a QALY Project. Health Technol Assess. 2010;14(27):1–162. doi:10.3310/hta14270. 19. Lancsar E, Wildman J, Donaldson C, Ryan M, Baker R. Deriving distributional weights for QALYs through discrete choice experiments. J Health Econ. 2011;30(2):466–78. doi:10.1016/j. jhealeco.2011.01.003. 20. Cookson R, Drummond M, Weatherly H. Explicit incorporation of equity considerations into economic evaluation of public health interventions. Health Econ Policy Law. 2009;4(2):231–45. doi:10.1017/S1744133109004903. 21. Devlin N, Sussex J. Incorporating multiple criteria in HTA: methods and processes. London: Office of Health Economics; 2011. 22. Stafinski T, Menon D, Marshall D, Caulfield T. Societal values in the allocation of healthcare resources: is it all about the health gain? Patient. 2011;4(4):207–25. doi:10.2165/11588880000000000-00000. 23. Shah KK. Severity of illness and priority setting in healthcare: a review of the literature. Health Policy. 2009;93(2–3):77–84. doi:10.1016/j.healthpol.2009.08.005. 24. Dolan P, Shaw R, Tsuchiya A, Williams A. QALY maximisation and people’s preferences: a methodological review of the literature. Health Econ. 2005;14(2):197–208. doi:10.1002/hec.924. 25. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;338:b2535. doi:10.1136/bmj.b2535.

26. Higgins J, Green S. Cochrane handbook for systematic reviews of interventions, Version 5.1.0 [updated March 2011]. The Cochrane Collaboration; 2011. 27. Diederich A, Swait J, Wirsik N. Citizen participation in patient prioritization policy decisions: an empirical and experimental study on patients’ characteristics. PLoS ONE. 2012;7(5):e36824. doi:10.1371/journal.pone.0036824. 28. Diederich A, Winkelhage J, Wirsik N. Age as a criterion for setting priorities in health care? A survey of the German public view. PLoS ONE. 2011;6(8):e23930. doi:10.1371/journal.pone. 0023930. 29. Akkazieva B, Gulacsi L, Brandtmuller A, Pentek M, Bridges JFP. Patients’ preferences for healthcare system reforms in Hungary: a conjoint analysis. Appl Health Econ Health Policy. 2006;5(3):189–98. 30. Bosworth R, Cameron TA, DeShazo JR. Is an ounce of prevention worth a pound of cure? Comparing demand for public prevention and treatment policies. Med Decis Making. 2010;30(4):E40–56. doi:10.1177/0272989x10371681. 31. Green C, Gerard K. Exploring the social value of health-care interventions: a stated preference discrete choice experiment. Health Econ. 2009;18(8):951–76. 32. Gyrd-Hansen D. Investigating the social value of health changes. J Health Econ. 2004;23(6):1101–16. doi:10.1016/j.jhealeco.2004. 02.002. 33. Gyrd-Hansen D, Slothuus U. The citizen’s preferences for financing public health care: a Danish survey. Int J Health Care Finance Econ. 2002;2(1):25–36. 34. Jan S, Mooney G, Ryan M, Bruggemann K, Alexander K. The use of conjoint analysis to elicit community preferences in public health research: a case study of hospital services in South Australia. Aust N Z J Public Health. 2000;24(1):64–70. 35. Lim MK, Bae EY, Choi SE, Lee EK, Lee TJ. Eliciting public preference for health-care resource allocation in South Korea. Value Health. 2012;15(1 Suppl):S91–4. doi:10.1016/j.jval.2011. 11.014. 36. Mentzakis E, Stefanowska P, Hurley J. A discrete choice experiment investigating preferences for funding drugs used to treat orphan diseases: an exploratory study. Health Econ Policy Law. 2011;6(3):405–33. 37. Nieboer AP, Koolman X, Stolk EA. Preferences for long-term care services: willingness to pay estimates derived from a discrete choice experiment. Soc Sci Med. 2010;70(9):1317–25. doi:10. 1016/j.socscimed.2009.12.027. 38. Norman R, Hall J, Street D, Viney R. Efficiency and equity: a stated preference approach. Health Econ. 2012;. doi:10.1002/hec. 2827. 39. Ringburg AN, Buljac M, Stolk EA, van Lieshout EMM, van Beeck EF, Patka P, et al. Willingness to pay for lives saved by helicopter emergency medical services. Prehosp Emerg Care. 2009;13(1):37–43. doi:10.1080/10903120802472004. 40. Ryynanen OP, Myllykangas M, Vaskilampi T, Takala J. Random paired scenarios—a method for investigating attitudes to prioritisation in medicine. J Med Ethics. 1996;22(4):238–42. doi:10. 1136/Jme.22.4.238. 41. Scuffham PA, Whitty JA, Taylor M, Saxby RC. Health system choice: a pilot discrete-choice experiment eliciting the preferences of British and Australian citizens. Appl Health Econ Health Policy. 2010;8(2):89–97. 42. Watson V, Carnon A, Ryan M, Cox D. Involving the public in priority setting: a case study using discrete choice experiments. J Public Health (Oxf). 2012;34(2):253–60. doi:10.1093/pubmed/ fdr102. 43. Whitty JA, Scuffham PA, Rundle-Thiele SR. Public and decision maker stated preferences for pharmaceutical subsidy decisions. Appl Health Econ Health Policy. 2011;9(2):73–9. doi:10.2165/ 11537150-000000000-00000.

J. A. Whitty et al. 44. Bryan S, Roberts T, Heginbotham C, McCallum A. QALYmaximisation and public preferences: results from a general population survey. Health Econ. 2002;11(8):679–93. 45. Schwappach DL, Strasmann TJ. ‘‘Quick and dirty numbers’’? The reliability of a stated-preference technique for the measurement of preferences for resource allocation. J Health Econ. 2006;25(3):432–48. doi:10.1016/j.jhealeco.2005.08.002. 46. Schwappach DLB. Does it matter who you are or what you gain? An experimental study of preferences for resource allocation. Health Econ. 2003;12(4):255–67. 47. Ratcliffe J. Public preferences for the allocation of donor liver grafts for transplantation. Health Econ. 2000;9(2):137–48. doi:10. 1002/(sici)1099-1050(200003)9:2\137:aid-hec489[3.3.co;2-t. 48. Louviere JJ, Flynn TN. Using best–worst scaling choice experiments to measure public perceptions and preferences for healthcare reform in Australia. Patient. 2010;3(4):275–83. doi:10.2165/ 11539660-000000000-00000. 49. Edlin R, Tsuchiya A, Dolan P. Public preferences for responsibility versus public preferences for reducing inequalities. Health Econ. 2012;21(12):1416–26. doi:10.1002/Hec.1799. 50. Corso PS, Hammitt JK, Graham JD, Dicker RC, Goldie SJ. Assessing preferences for prevention versus treatment using willingness to pay. Med Decis Making. 2002;22(5):S92–101. doi:10.1177/027298902237713. 51. Lindholm LA, Rosen ME, Stenbeck ME. Determinants of willingness to pay taxes for a community-based prevention programme. Scand J Soc Med. 1997;25(2):126–35. 52. Olsen JA, Donaldson C. Helicopters, hearts and hips: using willingness to pay to set priorities for public sector health care programmes. Soc Sci Med. 1998;46(1):1–12. 53. Oremus M, Tarride JE, Raina P, Thabane L, Foster G, Goldsmith CH, et al. The general public’s willingness to pay for tax increases to support unrestricted access to an Alzheimer’s disease medication. Pharmacoeconomics. 2012;30(11):1085–95. doi:10. 2165/11594180-000000000-00000. 54. Protie`re C, Donaldson C, Luchini S, Moatti JP, Shackley P. The impact of information on non-health attributes on willingness to pay for multiple health care programmes. Soc Sci Med. 2004;58(7):1257–69. 55. Tang CH, Liu JT, Chang CW, Chang WY. Willingness to pay for drug abuse treatment: results from a contingent valuation study in Taiwan. Health Policy. 2007;82(2):251–62. doi:10.1016/j. healthpol.2006.09.007. 56. Weaver M, Ndamobissi R, Kornfield R, Blewane C, Sathe A, Chapko M, et al. Willingness to pay for child survival: results of a National Survey in Central African Republic. Soc Sci Med. 1996;43(6):985–98. 57. Zarkin GA, Cates SC, Bala MV. Estimating the willingness to pay for drug abuse treatment: a pilot study. J Subst Abuse Treat. 2000;18(2):149–59. 58. Costa-Font J, Rovira J. Eliciting preferences for collectively financed health programmes: the ‘willingness to assign’ approach. Appl Econ. 2005;37(14):1571–83. doi:10.1080/ 00036840500181695. 59. Eisenberg D, Freed GL, Davis MM, Singer D, Prosser LA. Valuing Health at different ages: evidence from a nationally

60.

61.

62.

63.

64.

65.

66.

67.

68.

69. 70.

71.

72.

73.

representative survey in the US. Appl Health Econ Health Policy. 2011;9(3):149–56. doi:10.2165/11587340-000000000-00000. Nord E. The relevance of health state after treatment in prioritizing between different patients. J Med Ethics. 1993;19(1):37–42. doi:10.1136/Jme.19.1.37. Nord E, Street A, Richardson J, Kuhse H, Singer P. The significance of age and duration of effect in social evaluation of health care. Health Care Anal. 1996;4(2):103–11. doi:10.1007/ Bf02251210. Quintal C. Aversion to geographic inequality and geographic variation in preferences in the context of healthcare. Appl Health Econ Health Policy. 2009;7(2):121–36. doi:10.2165/10899530000000000-00000. Richardson JR, McKie J, Peacock SJ, Iezzi A. Severity as an independent determinant of the social value of a health service. Eur J Health Econ. 2011;12(2):163–74. doi:10.1007/s10198-0100249-z. Singh J, Lord J, Longworth L, Orr S, McGarry T, Sheldon R, et al. Does responsibility affect the public’s valuation of health care interventions? A relative valuation approach to health care safety. Value Health. 2012;15(5):690–8. doi:10.1016/j.jval.2012. 02.005. Dolan P, Tsuchiya A. It is the lifetime that matters: public preferences over maximising health and reducing inequalities in health. J Med Ethics. 2012;38(9):571–3. doi:10.1136/medethics2011-100228. Whitty JA. Insensitivity to scope in contingent valuation studies: new direction for an old problem. App Health Econ Health Policy. 2012;10(6):361–3. doi:10.2165/11635250-00000000000000. de Bekker-Grob EW, Ryan M, Gerard K. Discrete choice experiments in health economics: a review of the literature. Health Econ. 2012;21(2):145–72. doi:10.1002/hec.1697. Lancsar E, Louviere J. Conducting discrete choice experiments to inform healthcare decision making: a user’s guide. Pharmacoeconomics. 2008;26(8):661–77. Rawls J. A theory of justice. Rev ed. Cambridge: Belknap Press of Harvard University Press; 1999. Scuffham PA, Whitty JA, Mitchell A, Viney R. The use of QALY weights for QALY calculations: a review of industry submissions requesting listing on the Australian Pharmaceutical Benefits Scheme 2002–4. Pharmacoeconomics. 2008;26(4):297–310. Robinson A, Parkin D. Recognising diversity in public preferences: the use of preference sub-groups in cost-effectiveness analysis. A response to Sculpher and Gafni. Health Econ. 2002;11(7):649–51. Bridges JF, Hauber AB, Marshall D, Lloyd A, Prosser LA, Regier DA, et al. Conjoint analysis applications in health—a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force. Value Health. 2011;14(4):403–13. doi:10. 1016/j.jval.2010.11.013. Johnson FR, Lancsar E, Marshall D, Kilambi V, Muhlbacher A, Regier DA, et al. Constructing experimental designs for discretechoice experiments: report of the ISPOR Conjoint Analysis Experimental Design Good Research PracticesTask Force. Value Health. 2013;16:3–13.

A systematic review of stated preference studies reporting public preferences for healthcare priority setting.

There is current interest in incorporating weights based on public preferences for health and healthcare into priority-setting decisions...
848KB Sizes 2 Downloads 3 Views