Abstract
Most long-term care systems (social care) for older people rely on some means testing, copayment system, private insurance, or other non-governmental funding to supplement state provision. In England, an estimated quarter of homecare delivery is funded privately. For many older people, the absence of state funding for their care is only part of the problem: they are also expected to search for care in a market characterised by complexity, plurality, and imperfect information. Surprisingly, there are few services available to support private funders to navigate the system. This paper examines willingness to pay for care navigation and seeks to classify heterogeneity of preferences for navigation support. A discrete choice experiment (DCE) survey was completed by 182 participants across England in 2020-21. The results of the random parameter logit model used to analyse preferences showed that people valued information about care options (quality, information, and finances), but they also wanted help to “think things through,” as processing information could be challenging. Generally, participants valued what the navigation service provided, more than how the services were organised and delivered. The study also used latent class analysis to identify four groups with similar preferences, with almost half of participants (48%) expressing high willingness to pay for a comprehensive navigation service. The other three classes represented those with preferences focused on a narrower set of attributes: fast access to information (20%), affordable help to “think things through” (18%), and information provided by their local council (14%). The study demonstrates the potential demand and likely take-up of navigation support if made available to people who pay privately for care. Future research needs to examine the barriers to market development for social care navigation services.
1. Introduction
The long-term care systems of most developed nations rely on private payments from individuals with care needs, operating alongside social insurance or state-financed provision [1, 2]. Depending on the system, this may include user charges (or copayments”), whereby older people receive services from the government but are liable for a supplemental fee, typically linked to financial means. However, for those not eligible even for part-funded care, services must be sourced privately. Very few countries have robust private insurance markets for care [3] due to a multiplicity of market failures and their general lack of appeal [2, 4]. Those unable to access state or social insurance-funded support are typically required to find and fund care themselves.
Older people who pay privately for their own support at home and/or in the community (“self-funders”) are neglected in both research and policy [5]. There are little available data to understand the size of this group. In the United States, industry data indicate that up to 70 per cent of revenue for nonhealth home support agencies comes from private payers, funded out-of-pocket, with another ten per cent arising from private care insurance [3]. In England, some estimates suggest that a quarter of home care hours are paid for privately [6], while the Health Survey for England data suggest that around 250,000 older people self-fund home care [5]. When including self-funding of equipment, adaptations, or support with low-level needs or preventive services, the number of self-funders is undoubtedly far higher.
An ageing population will only expand the number of self-funders further in coming decades, although the UK Government currently proposes a limit on lifetime care costs in England. Local authorities have a duty to assess applicants for social care against two separate eligibility criteria: whether care needs meet a required threshold (needs assessment) and whether the applicant has insufficient means to pay themselves (financial assessment). To receive state-funded support, both criteria must be satisfied, otherwise individuals will be required to pay independently for their care. Importantly, those failing to meet these criteria commonly also miss out on valuable support from local authorities to help them choose and arrange their care [7–9].
1.1. Self-Funders as Care Navigators
Accessing the right support at the right time to help manage a wide range of needs is essential for successful social care [10]. For state-funded service users, most receive some form of case management that will support the process by which care is sourced, commissioned, and delivered, in addition to assessments, risk planning, reviewing, and other allied functions. Yet, few such services exist for self-funders. The term social care navigation is used here to describe the process by which older self-funders make arrangements for social care, from information seeking and decision-making through to purchasing support.
Research on self-funder care navigation is sparse but tends to conclude that self-funders and their relatives are, generally, poorly supported and struggle to find and fund the care they need [5, 6, 11–13]. Older self-funders and their families are often left “feeling lost and confused” [14, p5] and have described entering an “unfamiliar world” [15, p9]. Finding care has been reported as stressful and upsetting, especially in a context of urgent care needs or where families were struggling to get the right advice at the same time as meeting the demands of work or their caring roles [14]. Research suggests that opportunities for information, advice, and guidance are missed because service users do not feel that local government would assist them in a meaningful way and that local authorities in practice have little incentive or capacity to engage self-funders and their families [15, 16]. Self-funders also appear unaware of the potential benefits of advice, such as financially planning for care [17]. Moreover, some self-funders face distinct disadvantages in navigating care systems relative to others. These can include weak social capital or community networks, and also geographical factors, such as the challenges of rural living [18].
Difficulties in navigating care are a source of significant stress, and there are also grounds to believe that decision-making is adversely affected. There are manifold concerns including needs being left unmet; the purchase of suboptimal care, mismatched to the person’s needs; poor value for money; inaccessible community services; additional pressures on family carers; and widening inequalities [18–20]. Though dated, there is some evidence that self-funders enter residential care earlier than those receiving state-funded support [19].
1.2. Help with Care Navigation
Previous studies have suggested that self-funders want more support with care navigation and in a form that would most likely require significant investment. For example, one study suggested that self-funders wanted tailored information over generic formats and desired personalised help from named individuals rather than impersonal remote contact from a multitude of people [14]. Evidence also suggests that self-funders want information on a broad range of different care features, including the availability of services, quality, and reports on client satisfaction and on finances [14]. While information alone may be regarded as necessary, it is unlikely to be sufficient to fully meet the criteria for good decision-making. A review of the skills needed to find and plan social care implies an almost inevitability that this could not be done without support [21].
However, attempts to improve care navigation for self-funders have floundered. In England, care “brokerage” services were established to support the growth of personal budgets (a user-controlled budget financed by government for care needs), but these did not fundamentally change care market structures. Furthermore, the Care Act 2014 paved the way for councils to take responsibility for greater support for self-funders when seeking social care and for “shaping” a local care market that would cater to their needs. Yet, little has changed as reforms were first delayed and then put into abeyance [22] because of a substantial retrenchment of social care investment from central government relative to expanding care needs. What remains is a system of largely paper- or web-based information, typically restricted to lists of local providers.
Any improvement in navigation support is likely to require new investment, and, arguably, user charges for those seeking support. Yet, with self-funders facing (often unanticipated) care costs, this raises the question of whether, and to what extent, self-funders would be willing to pay. Evidence hints at significant variations in preferences for different forms of social care navigation support. Little is known about these variations or the factors that may drive or moderate such differences.
2. Aims and Objectives
This study aimed to examine the preferences of older self-funders and informal carers for support in navigating community social care. In doing so, it sought to identify what features of navigation support in social care are most valued and to explore heterogeneity in preferences.
3. Methods
3.1. Study Design
Discrete choice experiments (DCEs) are widely used in health economics and allied fields to examine preferences for different service design options [23]. A DCE works by presenting descriptions of two (or more) hypothetical services, each with different design features (or “attributes”), with participants then asked to select their preferred option. By altering the attributes within each hypothetical service, and seeing the impact on participants’ preferred option, a DCE reveals people’s strength of preferences for differing design features [24]. Furthermore, by introducing a cost for each hypothetical service, an estimate of “willingness to pay” for attributes can be estimated [25].
In the present study, the DCE introduced the exercise by asking participants to imagine that they had social care needs and that they had learned that they would have to find, choose, and pay for any social care services themselves. The instructions went on to say that a “care navigation” service could help them in finding and choosing their care, if they wished. The study defined social care navigation as being support to help you with your care options and was colloquially described in text as being “like how a travel agent might help you to find a holiday, by providing you with information and helping you to choose”.
3.2. Defining Attributes and Levels
A DCE requires clear definition of the attributes which form the design features of the hypothetical service under examination [26, 27]. The attributes for this DCE were selected through a two-stage process. First, a “longlist” of evidence-informed attributes was formed from secondary analysis of archived qualitative data transcripts from a prior study with self-funders and the findings of a recent scoping review [28]. Second, the “longlist” was reviewed and prioritised at a workshop of service users (older self-funders paying for their care at home) and informal carers (family members of the older person). In this workshop, attendees were asked to individually rank their top attributes and then in small groups they discussed the attributes and agreed on terminology. During this activity, any new attributes that the group felt were missing were discussed and added. Third, the final attribute list, Box 1, was debated and approved by the study advisory group, comprising a range of academic, practitioner, and service user representatives. Finally, these attributes and levels were piloted with eleven older people using a “think aloud” interview approach. In this, the person worked through all sections of the DCE survey, including the wording of the attributes and levels, and spoke aloud their thoughts. The researchers then identified any ambiguous terminology or survey design issues and made improvements. A key change was the researcher team decided to offer all participants’ telephone support to complete the DCE survey, if useful to them.
The final attribute list is presented in Table 1. Attributes 1–4 describe types of support the navigation service could provide, while attributes 5–7 describe organisational features. The final attribute defines the one-off cost of the service.
The choice questions for the DCE survey were formed using Ngene, a specialist piece of software for designing choice experiments. Two blocks of questions were created with respondents randomly assigned to receive one block of eight questions. An “opt-out,” whereby participants would receive no navigation service, was included in each choice set alongside two hypothetical navigation services, enabling an analysis of predicted take-up. The survey was available for either online or postal completion. Support was offered via telephone for both options. Figure 1 shows an example of a choice question from the DCE survey.
3.3. Measures
The survey also collected sociodemographic data, information on respondents’ experiences of social care, and health information including presence of long-term conditions. Two additional scales measured consumer variables around confidence in making purchases and financial wellbeing. The first used a five-item “information confidence” variable, constructed using a subscale of the Consumer Self-Confidence Scale [29] and rated on a five-item Likert scale from strongly agree to strongly disagree. Second, the self-reported perception of the extent to which someone’s financial situation provides them with security and freedom of choice was measured using the Consumer Financial Protection Bureau (CFPB) financial wellbeing scale [30].
3.4. Participants
The study took place across England and recruited two key groups of participants. First, older people (aged 65+) and families currently self-funding social care were sourced through home care providers, which, in turn, were identified through an advert circulated by the National Homecare Association, and through provider forums known to the research team. Interested providers were asked to pass on information and obtain consent from older people who self-funded their care or their family members if they were primarily responsible for payments and care arrangements. People were excluded if they lacked the capacity to consent or were not fluent in English.
Second, a sample of older people not currently using social care services, but with an emerging frailty or long-term condition, was sought. This was intended to examine the early formation of preferences amongst those likely to require navigation support in the future. Recruitment methods relied on established cohorts of research participants including the CARE75+ database [31] with identified frailty (excluding those currently receiving care), the DETERMIND cohort of people with dementia [32], and a panel of citizens from a local council who declared they were older people and had long-term conditions. During the study, the research team noted the lack of ethnic diversity in the sample characteristics and employed an outreach approach to identify additional participants. Researchers widely advertised the study, asking that older people specifically from ethnic minority groups who pay for care at home to contact the team.
Data collection began in August 2020 and completed in July 2021 and coincided with the COVID-19 pandemic. Ethical approval was granted by the NHS Health Research Authority Research Ethics Committee (20/SC/0213) in May 2020. This application included COVID-19 pandemic contingency plans to reduce risks to participants and researchers.
3.5. Data Analyses
Regression procedures suited to choose data were undertaken by the study health economist. An uncorrelated random parameter logit model was used to analyse the full data. Marginal willingness-to-pay (WTP) values were calculated to explore what respondents would be willing to pay for a discrete change in a level of a particular attribute. WTP values were calculated by dividing the coefficient for each attribute by the inverse of the coefficient for the cost attribute. The WTP values represent how much more or less a navigation service could cost before any improved or reduced uptake for the service resulting from an attribute was cancelled out by the increased or decreased cost. To examine preference variations, conditional logistic regression models were estimated for participants with high (vs low) scores on information confidence and financial wellbeing. Scatterplots of the willingness to pay values calculated from these models were used to illustrate differences in cost sensitivity. Further analyses were undertaken on differences in willingness-to-pay values for those currently using social care (vs. those with long-term conditions but not yet using social care) and older people completing the DCE (vs. their family members). Due to the small sample sizes in some of the subgroups, tests of poolability were not conducted. Finally, a latent class analysis was conducted on the whole sample in Latent Gold.
Further technical details of the steps taken in designing and analysing all the data are provided in the supplementary appendix.
4. Results
One hundred and eighty-two respondents completed the survey. Table 2 presents participant characteristics. Two-thirds were female, and almost all were white, despite efforts to be as inclusive as possible. The average age was 68 years old although there was a large spread in the sample, from 30 to 91 years old. All regions of England were represented, but the largest source of recruitment was from Yorkshire and Humber. Forty per cent were family members of people receiving support at home, and a smaller number (11%) were older people in receipt of care themselves. For both the family and older people, 80 per cent entirely paid for care themselves, with others jointly funding care with the council/other sources. Eighty-four per cent of respondents completed the survey online. As there was no evidence of scale heterogeneity relating to different survey delivery types (online or postal), data from each were combined in the central analysis.
Table 3 presents the central findings from the DCE analysis. A first notable observation is that the specific navigation services provided (attributes 1–4) were markedly more important to participants than the organisational features of the service (attributes 5–7). The most highly valued attribute was for information about the quality of social care services (a mean willingness to pay of £285), closely followed by “help to think things through” (£212). Both information about the availability of services and financial information were valued at just below £200.
Amongst service characteristics, the most important feature was that the service was provided by a single individual which the participant gets to know (rather than multiple people). There was a small willingness-to-pay value (£69) for the service to be available immediately, rather than having to wait a few weeks. Participants did not strongly value differences in who delivered the service, with a marginal preference for charitable organisation compared to councils or private-sector provision.
Analysis also estimated the likely acceptance of different service designs costing different amounts. An “optimal” service, with all attributes set at the most highly valued option (shown in Table 3), would be accepted by 91 per cent of participants if available without charge. The predicted acceptance rate declines with the cost charged to 84 per cent, 69 per cent, and 57 per cent for optimally designed service costing £100, £300, and £600, respectively.
4.1. Variation in Preferences: The Role of Information Confidence and Financial Wellbeing
Willingness-to-pay estimates for each attribute were calculated separately for respondents with high and low values on the “information confidence” and “financial wellbeing” scales. The “high” and “low” categorisation was achieved simply by dichotomising above/below the median value. Figures 2 and 3 present an “at-a-glance” visualisation of the variation in preferences in the form of scatterplots. Each point on the scatterplot represents the estimated willingness to pay for an attribute with each of the levels available (see Table 1). Perfectly equal preferences between participants with high/low values would be indicated by each point lying on a straight line at a 45o angle through the origin. Deviations from this would suggest that the two compared groups have different preferences.
Figure 2 shows a scatterplot of willingness-to-pay estimates for each attribute for those with higher (y axis) and lower (x axis) information confidence scores. It reveals relatively little variation in preferences. Respondents who were highly confident with information had a slightly greater preference for information on quality. Those who were less confident with information had a greater preference for information on availability. On average, those with high information confidence were willing to spend just six per cent more than those with low information confidence on navigation support. Further details can be seen in the supplementary tables.
More variation was evident with respect to financial wellbeing (Figure 3). Most notably, those with greater financial wellbeing were willing to pay 50 per cent more overall than those with low financial wellbeing. Those with low financial wellbeing were more likely to value “help to think things through” and preferred the navigation support to be provided by the council rather than by private organisations. Further details can be found in the supplementary tables.
Additional exploratory analyses of the estimated willingness to pay were undertaken on potential subgroups. First, those responding as current self-funders were compared with those who had long-term conditions but were not yet using care. Second, those responding as older people using care were compared with family members of older people using care.
Respondents currently receiving care had willingness-to-pay values which were 19.7 per cent higher than those planning for care in the future due to a long-term health condition. The latter group was also more likely to value help with their finances than those currently receiving care and placed a higher “intrinsic value” on the navigation service. This meant they were more likely to value any help with care navigation, regardless of its specific design features.
There was no evidence of variation between older people and their family carers, indicating similar navigation preferences between them.
4.2. Variation in Preferences: Latent Classes
Diversity in preferences was further examined through latent class analysis (LCA). LCA examines whether the sample as a whole can be disaggregated into separate “classes,” within which each participant has similar preferences for navigation support. This allows a typology of preferences to be established. Four latent classes were identified. Table 4 presents willingness to pay for each attribute for each class. A more comprehensive output can be found in the supplementary information.
Class 1 represented nearly half (48 per cent) of the sample. Members of Class 1 had the highest willingness to pay for all attributes, compared with the other classes. They also had the highest “intrinsic valuation” for navigation support (represented by the “alternative specific constant” in Table 4), which, as noted above, indicates that this group tended to highly value navigation support, regardless of its specific design features. Even so, this group particularly prioritised help to “think things through,” suggesting that information alone was insufficient for them. This group also placed a relatively high premium on the service characteristics: for example, they had a strong preference for support from a service with charity status, for support to be provided by one individual they got to know, and for the service to be available without delay.
Class 2, representing 20 per cent of respondents, had the lowest “intrinsic valuation” of all classes. That is, members of this class would be least likely to pay money for navigation support, regardless of its design. Nevertheless, members of Class 2 placed a relatively high valuation on information about quality, but they did not value help to “think things through” or had particularly strong preferences over how the service was organised.
Respondents in Class 3, representing 18 per cent of the sample, had a positive intrinsic valuation, indicating demand for navigation support in general, but were unlikely to pay significant sums for specific design features. Members of this class would be prepared to pay a modest cost for help “to think things through.”
Respondents in Class 4 (representing just 14 per cent of respondents) had strong preferences for the information attributes, but in sharp contrast to other classes, they also had a strong preference that this came from the council, and an aversion to private organisations. In addition, in contrast to other groups, they would prefer to wait longer for support. This latter finding may indicate that they would not wish to rush any decisions.
5. Discussion
Consumerism has played a powerful role in shaping public services across past decades, with the centrality of quasimarkets dominating reforms in social care in the UK and internationally [33, 34]. In England, these pressures dovetailed with the independent living movement and a discourse of “choice and control” to form fertile ground for enthusiasts of rational actor economics [17]. In theory, self-funders of social care are the archetypal form of consumer, armed with the resources to make choices in the marketplace for support, and matching information to their preferences.
However, as outlined above, navigation in social care markets is not so simple. It is likely that poor navigation is related to poor social care outcomes. In healthcare, there is good evidence of this association, and international attention is now given to suitable definition and measurement of “navigation” in care contexts, with new studies underway [35]. This may give rise to new evidence-informed interventions to support decision-making and in implementing decisions. Yet, in social care, evidence is sadly lacking. The present study is the first to examine what self-funders may, hypothetically, be willing to pay for specialist support with navigation. In doing so, it revealed the strength in preferences for help and variations between different self-funder groups.
The study finds that information about quality was the most highly valued navigation attribute, and this is not unexpected given the informational deficiencies and fragmentation facing older people in need of care [5, 14]. The importance of information is supported by wider market research evidencing that older people tend to consult fewer sources of information when making consumer decisions [36]. Yet, this new study demonstrates that providing information alone is insufficient for most self-funders and that they typically value information alongside help to “think things through.” Other research has highlighted the frustration that self-funders face when professionals provide information but will not help in the process of making decisions [28]. Elsewhere, evidence from “cash-for-care” schemes that attempt to mirror the decisional autonomy of self-funding for state-supported clients (by devolving care budgets to the end user) has found that older people faced multiple anxieties about the responsibilities of organising their own support [37]. More generally, the need for decision-making assistance also chimes with consumer psychology studies that find that older people are more likely (than younger counterparts) to make decisions that are “just good enough,” rather than making the best possible choices for them [38]. Taken together, the research provides a powerful narrative that self-funders should receive more assistance in making decisions.
This study also finds important variations in self-funder preferences for navigation support. Those with poorer subjective financial wellbeing were significantly less likely to be willing to pay for a navigation service, despite arguably having a greater need for it. Interestingly, this group was more likely than those with higher financial wellbeing to prefer navigation support from their council, rather than from a private agency. Since it is the local council that administers financial eligibility criteria, it might have been anticipated that council advice may not be trusted, in which they have a financial motive (e.g., for self-funders to avoid depleting their resources, and so becoming eligible for state-funded care). Certainly, those obligated to self-fund can face significant financial anxiety intertwined with the process of purchasing their care [22, 28]. These findings, however, imply that local authorities are a trusted place for those with financial anxieties, rather than private-for-profit organisations.
The latent class analysis suggests that there are four groups of individuals who share similar preferences for care navigation. The first, and largest, class has a high valuation of navigation support and would be willing to pay a significant price for support. Although this is based on only hypothetical exercises (these are “stated” rather than “revealed” preferences), it nevertheless supplies strong evidence that there is a market for navigation services that may attract user charges to fund their work. In England, there are few services that offer a chargeable but comprehensive navigation package. Limited “brokerage” or “direct payment support” services exist but are mostly targeted at people receiving state-funded “cash-for-care” budgets. Unfortunately, there is limited knowledge of older people’s views of these options [39–41].
A final methodological advance is also noted with this research. Almost all usage of DCE methods in the health and care landscape has focused on support that is either state-funded or else supported by social or personal insurance systems. In either case, the notion of “willingness to pay” is abstract, since there is no fee imposed on individual users at the point of use. For example, when examining preferences for GP surgeries in Britain, DCE analysts have resorted to proxy measures of willingness to pay, such as using waiting time, the idea of out-of-pocket expenses, or travel times. By contrast, in the current example, the notion of a cost constraint is directly relevant to the preferences being tested, since by definition, self-funders would most likely be expected to meet costs themselves for any additional navigation support. The validity and interpretability of the results are thus enhanced.
There are several important limitations to note. The DCE method is a rigorous approach to understanding the strength of preferences for service attributes but is only a hypothetical exercise. It is not clear whether the results here would reflect real-life choices. In healthcare, there is credible evidence that real-life choices show reasonable association with those predicted by DCEs [42]. However, the research team can only speculate as to whether such results hold for social care. A further limitation is a modest sample size, which although powerful for main willingness-to-pay estimates may nevertheless be suboptimal for comparison of different respondent groups. Relatedly, the COVID-19 pandemic greatly impacted the data collection phase of this study. The research team was limited in avenues for recruitment. They could not access older people currently receiving care by attending in-person groups, and it is notable the sample included few people from Black or Asian communities. The research cannot, therefore, make confident assertions about whether preferences may differ for these groups.
Recommendations for future research would examine heterogeneity in self-funder decision-making and navigation styles in more detail. At the time of writing, a qualitative follow-up study of these participants is examining pathways and decision-making styles. Other research could helpfully move from research on experiences and preferences to actual design and testing of support systems to improve navigation. A review of decision-making aids [21] has found a clear gap in research to support social care navigation, despite a plethora of such systems available in healthcare decisions.
6. Conclusion
This study explored the preferences of older self-funders and their families when choosing and arranging care. Significantly, these findings show that the most highly valued service attribute was for information about the quality of social care services, closely followed by “help to think things through” when making decisions. Earlier research has shown that local authorities provide care management support to nonself-funders which can lead to potential benefits in choice and personalised care. In the future care reforms, English self-funders are likely to have a closer relationship with the local council, as new care caps bring new roles and functions. As part of these reforms, care navigation support to self-funders may prove an important part of planning and commissioning care.
Data Availability
The data for this study are held at the University of York. As per the data management plan submitted to NHS HRA ethics, the data will not be shared openly. Specifically, data will not be available to anyone other than the immediate research team, since it is project-specific information.
Ethical Approval
Ethical approval was granted by the NHS Health Research Authority Research Ethics Committee (20/SC/0213) in May 2020. As this study ran during the COVID-19 pandemic, we included contingency plans in our successful application. We avoided all face-to-face contact, therefore reducing the risks to participants and researchers.
Disclosure
The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
We would like to acknowledge and thank the research participants who kindly gave up time to share their experiences and thoughts with us. People with lived experience (of finding and choosing care for an older person) were involved throughout the study, from designing the survey attributes to making sense of the findings. We acknowledge that some early findings from the DCE were presented as part of an oral presentation, in line with our funder requirements for impact. A big thank you to all our public involvement experts by experience for their useful insights and contributions to our study. This study received funding from the National Institute of Health and Care Research School for Social Care Research (C969/CM/UYYB-P140). This document presents independent research funded by the National Institute for Health and Care Research (NIHR) School for Social Care Research.
Supplementary Materials
Supplementary Table 1: technical appendix contains further technical details of the steps taken in designing and analysing all the data. It includes text and two small tables which show the statistical models used. Supplementary Table 2: information confidence (conditional logistic regression model) shows both the statistical modelling data and attributes of a care navigation service for people who have higher or lower confidence scores when receiving information. Supplementary Table 3: information confidence and willingness to pay show both the willingness-to-pay values and attributes of a care navigation service for people who have higher or lower confidence scores when receiving information. Supplementary Table 4: financial wellbeing (conditional logistic regression model) shows both the statistical modelling data and attributes of a care navigation service for people who have higher or lower financial wellbeing scores. Supplementary Table 5: financial wellbeing and willingness-to-pay values £ show both the willingness-to-pay values and attributes of a care navigation service for people who have higher or lower financial wellbeing scores. Supplementary Table 6: willingness to pay (confidence intervals £) for four latent classes (full data) shows a comprehensive breakdown of willingness-to-pay values for each of the four latent classes. (Supplementary Materials)