Advertisement

What Characteristics of Nursing Homes Are Most Valued by Consumers? A Discrete Choice Experiment with Residents and Family Members

  • Rachel Milte
    Correspondence
    Address correspondence to: Rachel Milte, School of Commerce, University of South Australia, GPO Box 2471, Adelaide, South Australia 5001, Australia.
    Affiliations
    Department of Rehabilitation, Aged and Extended Care, Flinders University, Adelaide, South Australia, Australia

    Cognitive Decline Partnership Centre, University of Sydney, Sydney, New South Wales, Australia

    Institute for Choice, University of South Australia, Adelaide, South Australia, Australia
    Search for articles by this author
  • Julie Ratcliffe
    Affiliations
    Institute for Choice, University of South Australia, Adelaide, South Australia, Australia

    Flinders Health Economics Group, Flinders University, Adelaide, South Australia, Australia
    Search for articles by this author
  • Gang Chen
    Affiliations
    Flinders Health Economics Group, Flinders University, Adelaide, South Australia, Australia

    Centre for Health Economics, Monash University, Melbourne, Victoria, Australia
    Search for articles by this author
  • Maria Crotty
    Affiliations
    Department of Rehabilitation, Aged and Extended Care, Flinders University, Adelaide, South Australia, Australia

    Cognitive Decline Partnership Centre, University of Sydney, Sydney, New South Wales, Australia
    Search for articles by this author
Open AccessPublished:December 01, 2017DOI:https://doi.org/10.1016/j.jval.2017.11.004

      Abstract

      Objectives

      To generate a scoring algorithm weighted on the preferences of consumers for assessing the quality of care in nursing homes (i.e., aged care homes or institutions) in six key domains.

      Methods

      A discrete choice experiment was undertaken with residents of nursing homes (n = 126) or family member proxies (n = 416) in cases where severe cognitive impairment precluded resident participation. Analysis was undertaken using conditional and mixed logit regression models to determine preferences for potential attributes.

      Results

      The findings indicate that all six attributes investigated were statistically significant factors for participants. Feeling at home in the resident’s own room was the most important characteristic to both residents and family members. Care staff being able to spend enough time with residents, feeling at home in shared spaces, and staff being very flexible in care routines were also characteristics identified as important for both groups. The results of the Swait-Louviere test rejected the null hypothesis that the estimated parameters between residents and family members were the same, indicating that data from these two groups could not be pooled to generate a single weighted scoring algorithm for the Consumer Choice Index-Six Dimension instrument. Preferences were therefore encapsulated to generate scoring algorithms specific to residents and family members.

      Conclusions

      This study provides important insights into the characteristics of nursing home care that are most valued by consumers. The Consumer Choice Index-Six Dimension instrument may be usefully applied in the evaluation, planning, and design of future services.

      Keywords

      Introduction

      Long-term care costs remain a significant source of public expenditure, varying from 0.2% to 3% of the gross domestic product in member countries of the Organisation for Economic Co-operation and Development. Despite growth in home care services in most countries in the last decade, institutionally based care (such as nursing homes) accounts for the greatest proportion of aged care sector costs, typically representing 60% to 80% of total aged care expenditures [
      Organisation for Economic Co-operation and Development
      Long-Term Care for Older People.
      ,
      • Lueke S.
      • Hoffmann W.
      • Flebetaa S.
      Transitions between care settings in dementia: Are they relevant in economic terms?.
      ]. Personal contributions through “out-of-pocket” expenses are a significant contribution to total care costs, accounting for more than 30% of total spending in many countries [
      Organisation for Economic Co-operation and Development
      Long-Term Care for Older People.
      ,
      • Mulvey J.
      • Li A.
      Long-term care financing: options for the future.
      ]. Long-term care refers to care undertaken with the aim of maintaining well-being and independence of people living with functional and cognitive impairments and can encompass care undertaken in a person’s own home, in a group living setting, or in institutions [
      Organisation for Economic Co-operation and Development
      Long-Term Care for Older People.
      ]. Terms for facilities of this nature, however, differ across countries (e.g., residential aged care facility, skilled nursing facility, nursing home, and aged care home). Nevertheless, the term “nursing home” appears to be the most consistently used term across countries to refer to this type of care [
      • Howe A.L.
      • Jones A.E.
      • Tilse C.
      What’s in a name? Similarities and differences in international terms and meanings for older peoples’ housing with services.
      ]. With the aging of the population in Australia and internationally, there is increasing demand for accommodation and care services. Rising consumer expectations coupled with changes to the financing and structure of the sector in many countries have created an urgent need to develop a systematic and transparent mechanism for evaluating the effectiveness in meeting expected outcomes from the consumer perspective in nursing homes.
      One potential powerful mechanism for assessing the effectiveness of nursing home services is to measure and value the quality of care provided from the perspective of the consumer (residents and family members) [
      • Castle N.G.
      • Ferguson J.C.
      What is nursing home quality and how is it measured?.
      ]. Donabedian [
      • Donabedian A.
      Twenty years of research on the quality of medical care: 1964–1984.
      ] proposed a theoretical framework for indicators of quality of care including structures (i.e., organizational characteristics associated with provision of care), processes (i.e., tasks undertaken with or for the resident), and outcomes (i.e., the desired states the care is aiming to achieve). Although many definitions of the quality of care exist, traditionally in this context they have usually incorporated quality indicators of medical/clinical care, levels of psychosocial support, and fulfillment of the resident’s basic rights including dignity, autonomy, and privacy [
      • Castle N.G.
      • Ferguson J.C.
      What is nursing home quality and how is it measured?.
      ]. To date, the predominant concepts of quality of care in this context have been based on assessments provided by health professionals and/or aged care staff and have not strongly incorporated the views and preferences of consumers [
      • Castle N.G.
      • Ferguson J.C.
      What is nursing home quality and how is it measured?.
      ,
      • Nakrem S.
      • Vinsnes A.G.
      • Harkless G.E.
      • et al.
      Nursing sensitive quality indicators for nursing home care: international review of literature, policy and practice.
      ]. Concerns have been raised that such indicators produce a focus on paper compliance rather than promoting care processes and activities that enhance the resident’s well-being and quality of life (QOL) [
      • Rahman A.N.
      • Applebaum R.A.
      The nursing home minimum data set assessment instrument: manifest functions and unintended consequences—past, present, and future.
      ]. As such, it has been found that structural and clinical care–focused measures are not generally well correlated with improvement in QOL for residents [
      • Rahman A.N.
      • Applebaum R.A.
      The nursing home minimum data set assessment instrument: manifest functions and unintended consequences—past, present, and future.
      ], further compounding the negative effect of institutional structures on residents. This study uses a different approach to the measurement of quality of care in nursing homes through a focus on measuring performance against criteria that have been identified a priori as important to consumers. Standardized methods exist for incorporating changes in QOL into evaluations of the clinical and economic impact of services, through the use of the quality-adjusted life-year, which adjusts life-years gained by a measurement of the quality of those years [
      • Drummond M.F.
      • Sculpher M.J.
      • Torrance G.W.
      • et al.
      ]. This is usually through the use of generic preference-based health-related QOL instruments, which combine measurement of the health status of the individual with an “off-the-shelf” weighted scoring algorithm that indicates the desirability of that particular health state to members of the general population [
      • Makai P.
      • Brouwer W.B.
      • Koopmanschap M.A.
      • et al.
      Quality of life instruments for economic evaluations in health and social care for older people: a systematic review.
      ]. This has been considered appropriate for evaluating the effectiveness of health care interventions in countries where there is significant subsidy and funding of health care by governments ultimately using tax revenue from citizens [
      • Brazier J.
      • Ratcliffe J.
      The measurement and valuation of health for economic evaluation.
      ]. Nevertheless, several concerns exist regarding the application of such measures in evaluating social care interventions, such as nursing home care for older people [
      • Donaldson C.
      • Atkinson A.
      • Bond J.
      • et al.
      Should QALYs be programme-specific?.
      ]. First, generic QOL and health status measures (such as the EuroQol five-dimensional questionnaire and the six-dimensional health state short form), with their focus on mobility and function, are unlikely to be adequately sensitive to measure changes in people’s health states that can realistically occur with improvements to social care and are of value to the recipient [
      • Donaldson C.
      • Atkinson A.
      • Bond J.
      • et al.
      Should QALYs be programme-specific?.
      ]. In addition, there are questions of the appropriateness of using opinions of the general population as the basis of scoring the “value” or “benefit” of changes from a social care intervention, because many may not have interacted with these services or have direct experience of the types of limitations and functional problems that necessitate this care [
      • Coast J.
      Is economic evaluation in touch with society’s health values?.
      ]. The increasing trend for users to contribute directly to the cost of their care services, as governments struggle to balance the increasing demand for services with aging populations, calls into question the appropriateness of using general population judgments of the value of these services [
      Organisation for Economic Co-operation and Development
      Long-Term Care for Older People.
      ]. Therefore, there is a growing need to incorporate the preferences of people using nursing home services themselves into formal evaluations of service quality and effectiveness.
      There are few empirical studies of the preferences of older adults for nursing home services that can be used to generate an understanding of the value of different characteristics to the consumer [
      • Ryan M.
      • Netten A.
      • Skåtun D.
      • et al.
      Using discrete choice experiments to estimate a preference-based measure of outcome—an application to social care for older people.
      ]. The studies that exist have been predominantly focused on preferences for service inclusions in insurance schemes, or community-based versus nursing home services [
      • Ryan M.
      • Netten A.
      • Skåtun D.
      • et al.
      Using discrete choice experiments to estimate a preference-based measure of outcome—an application to social care for older people.
      ,
      • Brau R.
      • Lippi Bruni M.
      Eliciting the demand for long-term care coverage: a discrete choice modelling analysis.
      ,
      • Dixon S.
      • Nancarrow S.A.
      • Enderby P.M.
      • et al.
      Assessing patient preferences for the delivery of different community-based models of care using a discrete choice experiment.
      ,
      • Fernández-Carro C.
      Ageing at home, co-residence or institutionalisation? Preferred care and residential arrangements of older adults in Spain.
      ,
      • Kaambwa B.
      • Lancsar E.
      • McCaffrey N.
      • et al.
      Investigating consumers’ and informal carers’ views and preferences for consumer directed care: a discrete choice experiment.
      ], and have often been conducted with members of the general population, rather than with frail older adults who have direct experience of receiving these care services [
      • Brau R.
      • Lippi Bruni M.
      Eliciting the demand for long-term care coverage: a discrete choice modelling analysis.
      ,
      • Dixon S.
      • Nancarrow S.A.
      • Enderby P.M.
      • et al.
      Assessing patient preferences for the delivery of different community-based models of care using a discrete choice experiment.
      ,
      • Negrín M.A.
      • Pinilla J.
      • León C.J.
      Willingness to pay for alternative policies for patients with Alzheimer’s disease.
      ,
      • Nieboer A.P.
      • Koolman X.
      • Stolk E.A.
      Preferences for long-term care services: willingness to pay estimates derived from a discrete choice experiment.
      ]. Without a suitable instrument for empirically evaluating the effectiveness of innovations in nursing home care from a consumer perspective, quality improvement initiatives in this sector are missing an important component. Such an instrument will also facilitate decision making by providing a quantitative mechanism for maximizing the effectiveness and cost-effectiveness of innovations in nursing home care from the perspective of consumers. The Consumer Choice Index-Six Dimension (CCI-6D) instrument was designed to fill this gap in measuring and valuing the quality of nursing home care from the perspective of consumers.
      The CCI-6D comprises a descriptive system developed through a multistage process, including a comprehensive literature review, an in-depth qualitative study with people living with dementia and their family members (n = 41), and consultation with stakeholder groups, including a group of informal carers, clinicians, health service researchers, and representatives from aged care providers [
      • Milte R.
      • Shulver W.
      • Killington M.
      • et al.
      Quality in residential care from the perspective of people living with dementia: the importance of personhood.
      ]. This multistage process has been recommended as best practice for sourcing attributes for inclusion in stated preference studies [
      • Coast J.
      • Al-Janabi H.
      • Sutton E.J.
      • et al.
      Using qualitative methods for attribute development for discrete choice experiments: issues and recommendations.
      ]. The final attributes and levels included reflected the level of time care staff spent with residents, homeliness of shared spaces, homeliness of room setup, access to outside and gardens, frequency of meaningful activities, and flexibility with care routines (see Ref. [
      • Milte R.
      • Ratcliffe J.
      • Bradley C.
      • et al.
      Evaluating the quality of care received in long-term care facilities from a consumer perspective: development and construct validity of the Consumer Choice Index–Six Dimension instrument.
      ] for the instrument descriptive system). The aim of this study was to generate a weighted scoring algorithm for the CCI-6D for the measurement and valuation of the quality in nursing homes from the perspective of consumers.

      Methods

      Methodological Framework

      To generate a weighted scoring algorithm, two key methodological questions arise: 1) Whose values should be used? and 2) Which technique should be used to elicit these values? For the first question, potential sources of values include clinical and/or aged care staff involved in the care of residents, members of the general population, or people using the service themselves [
      • Rowen D.
      • Mulhern B.
      • Banerjee S.
      • et al.
      Estimating preference-based single index measures for dementia using DEMQOL and DEMQOL-Proxy.
      ]. For the reasons outlined previously, this study sought to incorporate consumers as the main source of values for the CCI-6D instrument, including residents and their family member carers. We have included family members as proxy participants when cognitive impairment precluded direct resident consent and participation. Family members often act as formal decision makers in cases where the decline in cognitive ability of the individual themselves necessitates support for decision making in health, care, and financial matters. Family members are often highly involved in choosing appropriate nursing homes and in supporting the ongoing care of residents [
      • Clarence-Smith B.
      Healthcare for people with dementia in care homes: family carer experiences.
      ,
      • Edwards H.
      • Courtney M.
      • Spencer L.
      Consumer expectations of residential aged care: reflections on the literature.
      ].
      The second key methodological question is which technique to use to elicit values for the CCI-6D instrument. Discrete choice experiments (DCEs) potentially have an advantage over other stated preference approaches for the elicitation of values in this context, including standard gamble and time trade-off, because they are framed in a less abstract way [
      • Brazier J.
      • Ratcliffe J.
      • Salomon J.A.
      • et al.
      ]. Participants are asked to make choices between alternative scenarios (in this case, reflecting characteristics of alternative nursing homes) and asked to indicate which scenario they would prefer. This type of choice situation is more reflective of how the selection of a nursing home is likely to be made in the real world [
      • Ryan M.
      • Gerard K.
      Using discrete choice experiments to value health care programmes: current practice and future research reflections.
      ,
      • Ryan M.
      • Scott D.
      • Reeves C.
      • et al.
      Eliciting public preferences for healthcare: a systematic review of techniques.
      ]. DCEs are particularly applicable to valuing characteristics of a social service [
      • Ratcliffe J.
      • Laver K.
      • Couzner L.
      • et al.
      Not just about costs: the role of health economics in facilitating decision making in aged care.
      ]. We therefore opted to use a DCE approach to measure quality in this context from the perspective of the consumer.

      Questionnaire Design

      The DCE to be used to generate preference weights was developed for completion by people living in nursing homes and their family members, assisted by a trained interviewer. The questionnaire comprised three main sections. Section A comprised a series of attitudinal statements relating to service provision and characteristics of a nursing home. Participants were asked to indicate how much they agreed or disagreed with each statement on a five-point Likert scale. Section B of the questionnaire contained the DCE, comprising a series of six questions involving a choice between two hypothetical nursing homes. The scenarios presented for consideration were based on the six salient attributes that form the basis of the CCI-6D instrument descriptive system.
      Three levels for each of the six attributes resulted in 729 possible scenarios (= 36) and a total of 265,356 possible pairwise choices ([729 × 728]/2). A D-efficient design with no prior parameter information (Dz-error, i.e., 0 priors assumed for all variables) was used to reduce the number of choice scenarios into a manageable number of 18 choice sets for presentation using the Ngene version 1.1.2 DCE design software package (ChoiceMetrics, Sydney, Australia) [
      ChoiceMetrics
      Ngene 1.1.2 User Manual and Reference Guide.
      ]. The resulting 18 scenarios were blocked into three versions of the DCE questionnaire each with six binary choice sets presented in each version. Participants were asked to indicate their preferred choice between a pair of hypothetical scenarios reflecting the characteristics of two alternative nursing homes in close proximity to each other within a geographical locality. Given that the main aim of the study was to determine preferred characteristics for nursing homes, a “forced choice” experiment was considered appropriate and no opt-out option was provided. Section C comprised a series of sociodemographic questions. The resident questionnaire is presented in the Appendix in Supplemental Materials (found at doi:10.1016/j.jval.2017.11.004).

      Participants

      Participants were recruited from 17 nursing homes across Australia, including both metropolitan and rurally located facilities. Recruitment occurred over a 13-month time period, between January 2015 and February 2016. The study was approved by the Flinders University Social and Behavioural Research Ethics Committee (project number 6706). Before completing the questionnaire, residents were administered the Psychogeriatric Assessment Scales-Cognitive Impairment Scale (PAS-Cog) by a trained research nurse [
      • Jorm A.F.
      • Mackinnon A.J.
      • Henderson A.S.
      • et al.
      The Psychogeriatric Assessment Scales: a multi-dimensional alternative to the categorical diagnoses of dementia and depression in the elderly.
      ]. The PAS-Cog is a standardized instrument that assesses memory and other cognitive functions, with excellent reliability and validity [
      • Jorm A.F.
      • Mackinnon A.J.
      • Henderson A.S.
      • et al.
      The Psychogeriatric Assessment Scales: a multi-dimensional alternative to the categorical diagnoses of dementia and depression in the elderly.
      ]. It is scored on a scale between 0 and 21, where a higher score indicates greater cognitive impairment. Residents with no to mild cognitive impairment (indicated by a PAS-Cog score between 0 and 9) were then asked to complete the questionnaire themselves because a previous work we had undertaken had supported the validity of undertaking DCE with this population [
      • Milte R.
      • Ratcliffe J.
      • Chen G.
      • et al.
      Cognitive overload? An exploration of the potential impact of cognitive functioning in discrete choice experiments with older people in health care.
      ]. When residents had a more severe level of cognitive impairment, a family member proxy was approached to complete the questionnaire on behalf of the resident. Eligibility criteria included that the residents had been living in a nursing home for at least 12 months, and that they were not currently receiving palliative care. For those residents who required a proxy to answer on their behalf, a suitable person needed to be available to act as a proxy, which was defined as a person who had a close relationship with the individual and who visited the person regularly and assisted with making decisions on their behalf—usually a spouse, sibling, or offspring of the individual. After informed consent, the participants took part in a face-to-face interview with trained data collectors. The interviews were generally undertaken in the resident’s room, a sitting room, or other private area. If a proxy participant was unable to attend a face-to-face interview because of remoteness of location, or other logistical issues, arrangements were made for them to participate in the study via postal survey supplemented by telephone interview.

      Data Analysis

      The data from the DCE were analyzed within a random utility theory framework. The utility function can be specified as follows:
      Uijt=xijtβi+εijt,


      where Uijt is the utility individual i derives from choosing alternative j in choice scenario t, x is a vector of observed attributes (i.e., the CCI-6D dimensions and corresponding levels), β is a vector of coefficients reflecting the desirability of the attributes, and eijt is an error term. Two econometric approaches were used to estimate this utility function, including the classical conditional logit model and a mixed logit model that could be used to capture potential preference heterogeneity [
      • McFadden D.
      • Train K.
      • Mixed M.N.L.
      models for discrete response.
      ]. In the mixed logit model, the desirability of attributes constitutes a vector of average preferences of the population for each attribute (β) and the individual’s specific preference components (η) (i.e., βi = β + ηi), whereas in the conditional logit model, only average preferences are estimated (i.e., βi = β). The estimated coefficients and their statistical significance (or otherwise) indicate the relative importance of the different attributes on individual preferences. A positive sign on a coefficient indicates that as the level of that attribute increases, so does the utility derived, and the converse applies for a negative sign on a coefficient. Using established methods, the estimated coefficients were then rescaled onto a 0 to 1 scale (where a score of 0 equates to the least preferred care home, and 1 the most preferred care home) to provide a scoring algorithm for the CCI-6D instrument [
      • Coast J.
      • Flynn T.N.
      • Natarajan L.
      • et al.
      Valuing the ICECAP capability index for older people.
      ].
      The Swait-Louviere test was applied to test whether the responses from residents and family members could be pooled together [
      • Swait J.
      • Louviere J.
      The role of the scale parameter in the estimation and comparison of multinomial logit models.
      ]. Conditional and mixed logit regression models were compared using the Bayesian information criterion (BIC), which is commonly used for model selection in random utility framework [
      • Hensher D.A.
      Accounting for scale heterogeneity within and between pooled data sources.
      ,
      • Lancsar E.
      • Louviere J.
      Conducting discrete choice experiments to inform healthcare decision making.
      ]. The extent to which participants exhibited dominant preferences was also investigated. A dominant preference pattern implies that the scenario with the best level of one particular attribute is always chosen, irrespective of the levels of the remaining attributes presented [
      • Lancsar E.
      • Louviere J.
      Deleting “irrational” responses from discrete choice experiments: A case of investigating or imposing preferences?.
      ].

      Results

      Descriptive Statistics

      A total of 1323 people living in 17 nursing homes in four Australian states were assessed for eligibility. Nine hundred and one (68%) resident and or family member proxies were identified as eligible to participate, and of these 545 (60%) consented. Three resident participants failed to complete the DCE (section B) and were therefore excluded from the data analysis. A total of 126 (23%) residents completed the DCE questionnaire themselves. The remaining 416 participants were proxies, usually a close family member who was asked to complete the questionnaire on behalf of the resident where the resident had a level of cognitive impairment that precluded their own participation. Characteristics of the study participants are presented in Table 1. Most of the study participants (75.6%) had been living in the care home for more than 2 years. The reported PAS-Cog score indicates the level of cognitive impairment among residents in the sample and includes both residents who self-participated and those for whom a proxy participated. Mean (SD) PAS-Cog score was 13 [
      • Rahman A.N.
      • Applebaum R.A.
      The nursing home minimum data set assessment instrument: manifest functions and unintended consequences—past, present, and future.
      ], indicating a moderate level of cognitive impairment for the study sample.
      Table 1Demographic characteristics of the sample
      Characteristicn (%)
      Results are presented as n (%) unless otherwise specified.
      Age (y), mean ± SD68 ± 14
      Sex
        Male162 (30)
        Female380 (70)
      Participant
        Family member416 (77)
        Resident126 (23)
      How long have you (or your family member) been in nursing home?
        ≤24 mo132 (24)
        >24 mo410 (76)
      Highest educational qualification attained
      Missing responses n = 2.
        No qualifications104 (19)
        Completed high school157 (29)
        Undergraduate degree or professional qualification173 (32)
        Postgraduate qualification74 (14)
        Others32 (6)
      Born in Australia
      Missing responses n = 2.
        Yes429 (79)
        No111 (20)
      PAS-Cog score of admitted resident
       0–4 (no cognitive impairment)95 (18)
       5–9 (mild cognitive impairment)100 (19)
       10–15 (moderate cognitive impairment)85 (16)
       16–21 (severe cognitive impairment)262 (48)
      How difficult did you find this questionnaire to complete?
      Missing responses n = 5.
        Very difficult40 (7)
        Moderately difficult136 (25)
        Slightly difficult170 (31)
        Not difficult191 (35)
      PAS-Cog, Psychogeriatric Assessment Scales-Cognitive Impairment Scale.
      low asterisk Results are presented as n (%) unless otherwise specified.
      Missing responses n = 2.
      Missing responses n = 5.

      Attitudinal Questions

      The responses of the participants to the attitudinal questions are presented in Table 2. A large proportion of residents indicated that they either “strongly agreed” or “agreed” to statements “C: I am (my family member is) able to make my room here ‘my (their) own’” (98%); “D: It is important to me that I have (my family member has) access to therapists to provide physical exercise and keep me (them) walking” (94%); and “F: It is important to me that I can have access to specialist services from this facility” (95%). For other statements participant responses were more divided, for example, in relation to whether it would be better if aged care homes provided care for only people with dementia or people without dementia separately, or whether they would be willing to pay an additional bond upon entry into a nursing home facility for dementia-specific care.
      Table 2Responses to attitudinal questions
      QuestionResponse, n (%)
      Strongly agreeAgreeNeither agree nor disagreeDisagreeStrongly disagree
      A: I receive enough information from care staff regarding my care and health232 (43)220 (41)30 (6)46 (9)14 (3)
      B: I receive enough information from doctors regarding my care and health (n = 541)152 (28)177 (33)70 (13)99 (18)43 (8)
      C: I am able to make my room here “my own” (i.e., bring in own furniture, pictures, etc.)323 (60)205 (38)7 (1)5 (1)2 (0.4)
      D: It is important to me that I have access to therapists to provide physical exercise and keep me walking325 (60)183 (34)21 (4)9 (2)4 (1)
      E: I would want to be able to walk by myself, even if there was a risk I could fall and injure myself (n = 540)110 (20)165 (30)52 (10)119 (22)94 (17)
      F: It is important to me that I can have access to specialist services from this facility (e.g., dental, speech pathology, and geriatrician)330 (61)186 (34)15 (3)7 (1)2 (0.4)
      G: Changes to my medication or health care should be explained to me (n = 541)346 (64)172 (32)17 (3)4 (1)2 (0.4)
      H: It is more important that care home staff have a caring attitude than a high level of training (n = 541)207 (38)185 (34)92 (17)45 (8)12 (2)
      I: It would be better if aged care homes provided care only for people with dementia or cognitive impairment, rather than for people with and without cognitive impairment together (n = 539)91 (17)64 (12)62 (11)202 (37)120 (22)
      J: It is important to me that I am able to access morning or afternoon tea for myself or my family whenever I want (n = 541)178 (33)233 (43)78 (14)39 (7)13 (2)
      K: I would be willing to pay an additional $100,000 bond upon entry into nursing home care to receive dementia-specific care (n = 539)50 (9)83 (15)102 (19)147 (27)155 (29)

      DCE Estimates

      The number of participants exhibiting a dominant response pattern for each attribute was relatively low (<10%) for most of the attributes (for more details, see Appendix Table 1 in Supplemental Materials found at doi:10.1016/j.jval.2017.11.004). Among the six attributes presented, the amount of time care staff were able to spend with residents had the highest number of dominant responses.
      The Swait-Louviere pooling test rejected the null hypothesis that the estimated parameters between residents and family members were the same while allowing scale factors to vary between the two sources of data [
      • McFadden D.
      • Train K.
      • Mixed M.N.L.
      models for discrete response.
      ]. The BIC values further suggested that the conditional logit estimates were preferable to the mixed logit estimates for both resident and family member samples. As such, residents and family members were analyzed separately and only the preferred conditional logit estimates are reported here. The mixed logit estimates are presented in Appendix Table 2 in Supplemental Materials found at doi:10.1016/j.jval.2017.11.004.
      The conditional logit estimates for both residents and family members and their respective rescaled scoring algorithms are presented in Table 3. As can be seen, except for one level (“sometimes”) in each of the five attributes (“care staff time,” “feeling at home in shared spares,” “feeling at home in own room,” “access to outside and gardens,” and “meaningful activities”), all other attributes and levels were statistically significant, indicating that the attributes are all important in determining preferences for nursing homes. For residents, the higher levels of each attribute were associated with higher coefficients, indicating greater positive preferences for that attribute. For responses from family members, all attributes and levels were statistically significant, except the “sometimes” level for attributes “feeling at home in shared spares” and “feeling at home in own room” and the “whenever they want” level for the “access to outside and gardens” attribute. Family members also exhibited inconsistent preferences for one attribute (“access to outside and gardens”) with the coefficient for “access whenever” greater than the coefficient for “access sometimes” for residents, whereas the reverse is true for the family member proxy participants. The coefficients were rescaled on a 0 to 1 scale to provide the preference-weighted scoring algorithm for the CCI-6D on the basis of the 1) resident and 2) family member responses, which are also presented in Table 3.
      Table 3Conditional logit estimates and rescaled coefficients on residents and family members
      LevelConditional logit estimatesRescaled coefficient on 0–1 scale
      ResidentFamilyResidentFamily
      CoefficientSECoefficientSE
      How much time are care staff able to spend with my family member?
       Rarely−0.3620.071
      P < 0.01.
      −0.8190.047
      P < 0.01.
      0.002−0.081
       Sometimes0.0880.0700.1710.043
      P < 0.01.
      0.1060.121
       Always0.2740.072
      P < 0.01.
      0.6480.051
      P < 0.01.
      0.1490.219
      Does your family member feel “at home” in the shared spaces in this place?
       Rarely−0.2620.075
      P < 0.01.
      −0.3510.044
      P < 0.01.
      0.0250.015
       Sometimes−0.0020.077−0.0400.0440.0850.078
       Always0.2640.073
      P < 0.01.
      0.3910.044
      P < 0.01.
      0.1460.166
      Is your own room here set up to make you feel “at home”?
       Rarely−0.5800.079
      P < 0.01.
      −0.6330.044
      P < 0.01.
      −0.048−0.043
       Sometimes−0.0360.0670.0100.0360.0770.089
       Always0.6160.091
      P < 0.01.
      0.6230.051
      P < 0.01.
      0.2280.214
      Is there access to outside and gardens in this aged care home?
       Cannot−0.5570.076
      P < 0.01.
      −0.2620.048
      P < 0.01.
      −0.0430.033
       Sometimes0.1220.0750.2250.047
      P < 0.01.
      0.1140.132
       Whenever they want0.4350.077
      P < 0.01.
      0.0370.0490.1860.094
      How often does the aged care home offer my family member things to do that make them feel valued?
       Rarely−0.2220.081
      P < 0.01.
      −0.2440.045
      P < 0.01.
      0.0340.037
       Sometimes0.0740.0820.0970.049
      P < 0.05.
      0.1030.106
       Often0.1480.080
      P < 0.1.
      0.1470.041
      P < 0.01.
      0.1200.116
      How flexible are staff with the care routines?
       Not much−0.2470.078
      P < 0.01.
      −0.2360.040
      P < 0.01.
      0.0290.038
       A little−0.1270.075
      P < 0.1.
      −0.0880.037
      P < 0.05.
      0.0560.069
       Very0.3740.080
      P < 0.01.
      0.3240.043
      P < 0.01.
      0.1720.153
      Log likelihood−393.768−1266.676
      BIC874.7882635.359
      Observations14384918
      Note. All attributes were effects-coded.
      BIC, Bayesian information criterion; SE, standard error.
      low asterisk P < 0.01.
      P < 0.05.
      P < 0.1.

      Discussion

      This is the first study of the preferences of people living in nursing homes and their family members for characteristics of quality of care. Although the results of this study provide insight into the value consumers attribute to different characteristics of care, the main aim of this study was to generate preference weights for a weighted scoring algorithm to measure quality of nursing home care, the CCI-6D. We found that all the items of the CCI-6D were important determinants of preferences for nursing homes. This supports the findings of our previous qualitative work with people living with dementia and their family members in a larger population and with an empirical focus [
      • Milte R.
      • Shulver W.
      • Killington M.
      • et al.
      Quality in residential care from the perspective of people living with dementia: the importance of personhood.
      ]. It indicates that these characteristics are relevant to consumers of these services, and supports their inclusion in an instrument to evaluate improvements and innovations in nursing home care.
      Previous studies have indicated the preferences of the general population and community-dwelling older people to avoid institutionalization and remain in their own home for as long as possible [
      • Dixon S.
      • Nancarrow S.A.
      • Enderby P.M.
      • et al.
      Assessing patient preferences for the delivery of different community-based models of care using a discrete choice experiment.
      ,
      • Fernández-Carro C.
      Ageing at home, co-residence or institutionalisation? Preferred care and residential arrangements of older adults in Spain.
      ]. Nevertheless, there is evidence that institutionally based care becomes increasingly acceptable as a care option when people consider increasing frailty, cognitive impairment, and palliative care needs, because of perceived improved access to skilled care and strategies for symptom relief, and reduced burden on family members [
      • Nieboer A.P.
      • Koolman X.
      • Stolk E.A.
      Preferences for long-term care services: willingness to pay estimates derived from a discrete choice experiment.
      ,
      • Gott M.
      • Seymour J.
      • Bellamy G.
      • et al.
      Older people’s views about home as a place of care at the end of life.
      ,
      • Tucker S.
      • Brand C.
      • Sutcliffe C.
      • et al.
      What makes institutional long-term care the most appropriate setting for people with dementia? Exploring the influence of client characteristics, decision-maker attributes, and country in 8 European nations.
      ,
      • Chau P.H.
      • Kwok T.
      • Woo J.
      • et al.
      Disagreement in preference for residential care between family caregivers and elders is greater among cognitively impaired elders group than cognitively intact elders group.
      ,
      • Wolff J.L.
      • Kasper J.D.
      • Shore A.D.
      Long-term care preferences among older adults: A moving target?.
      ]. These studies have, however, focused on evaluation of preferences for long-term care strategies more generally rather than nursing home care specifically.
      The Swait-Louviere test rejected the null hypothesis of equal parameters between resident and family member respondents, and therefore the preferences of these groups were presented separately. In addition, the conditional logit estimates presented as the BIC values indicate these to be preferable to the mixed logit estimates in both groups. Although preferences were not the same, several synergies were evident for resident and proxy family member participants. The preferences for both groups followed a logical progression, with improved levels for each attribute associated with higher coefficients, indicating greater positive preferences. The only exception to this general finding was in relation to the attribute “access to outside and gardens,” for which family member proxies indicated that a moderate level of freedom in access was preferred. In contrast, a high level of access to outside and gardens was one of the attributes most valued by residents themselves. The reasons for this discrepancy are not completely clear but may be explained by a concern among family members about safety and risk injury when residents are given unfettered access to outside space. Nursing homes may therefore need to use innovative strategies to balance access to outside and gardens against concerns that family members may have about exposure to risk, to a level that is acceptable to all.
      We developed preference-weighted scoring algorithms for the CCI-6D instrument, which can be used to measure the extent to which a nursing home provides quality of care in key aspects of daily life within nursing homes from the perspective of residents and their family members. Although both scoring algorithms provide interesting perspectives, we recommend that the resident-specific scoring algorithm should be used as the preferred algorithm for the CCI-6D instrument on the basis that residents have live experience of the service under consideration and may therefore be considered the “primary” service users. In addition, the mixed logit estimates (see Appendix Table 2 in Supplemental Materials) indicated that preferences among the resident sample were generally homogeneous, indicating similarity in preferences among residents. For the family member respondents, however, the statistical significance of the SDs indicated greater heterogeneity in the estimated preferences (i.e., evidence of greater differences in stated preferences between members of the sample). Resident preferences also exhibited greater consistency across all attributes (as compared with the proxy respondents who showed inconsistency in preferences for the “access to outside and gardens” attribute).
      Although the resident-specific scoring algorithm is preferred for the aforementioned reasons, we encourage potential users to select the algorithm that they consider best meets the aims of the strategy or intervention they propose to evaluate. For example, in some situations the perspective of family members of residents with more severe cognitive impairment may be considered as particularly important, and thus the algorithm based on the family member preferences may be used.
      There are some limitations to this study that should be acknowledged. We chose to adopt a consumer-driven approach for the design of the CCI-6D instrument descriptive system, and thus content of the instrument was derived from items identified to be of importance to people with dementia and their family members. Nevertheless, the concepts of importance they identified were somewhat difficult to operationalize because of their qualitative and psychosocial nature. Thus, the items of the descriptive system and their levels (e.g., “rarely” or “sometimes”) could be considered open to interpretation of the individual. The CCI-6D instrument was, however, found to discriminate between types of care environments and items of QOL in residents in our validation study [
      • Milte R.
      • Ratcliffe J.
      • Bradley C.
      • et al.
      Evaluating the quality of care received in long-term care facilities from a consumer perspective: development and construct validity of the Consumer Choice Index–Six Dimension instrument.
      ]. In addition, it should be noted that DCE studies can be subject to bias; notable for this study is the concept of “status quo bias,” where participants choose a scenario that is familiar to their current situation, rather than one they truly prefer, which has been identified as occurring in a number of previous preference studies [
      • Salkeld G.
      • Ryan M.
      • Short L.
      The veil of experience: Do consumers prefer what they know best?.
      ]. If this were in play in this study, the true preferences for the attributes in the CCI-6D instrument may still be hidden. Nevertheless, attempts were made to reduce this bias through the study design by using the interviewer administration, where participants were coached on the hypothetical nature of the task and to choose what they preferred, and in case any of the participants referred to choosing what they received currently after coaching, the interview was ceased. Further investigation of the presence of status quo bias when undertaking DCE in this sample, as well as other useful insights regarding their decision-making process during DCE, could be elicited using qualitative methods such as a “think-aloud study” [
      • van Someren M.
      • Barnard Y.
      • Sandberg J.
      ]. In addition, the present study was conducted in an Australian setting and thus preferences may be similar or different across cultures and different countries—given the lack of information available on preferences of consumers for long-term care, we consider this an important space for future research.

      Conclusions

      This study has provided important insights into the characteristics of nursing homes that are most valued by consumers. Although the preferences of residents and proxy family members were not the same, several synergies in preferences were evident. The CCI-6D instrument may be usefully applied for assessing the quality of nursing home care from the consumer perspective. The instrument may also be incorporated into an economic evaluation framework to inform the planning and design of future aged care services.
      Source of financial support: This study was supported by funding provided by the National Health and Medical Research Council Partnership Centre on Dealing with Cognitive and Related Functional Decline in Older People (grant no. GNT9100000). The contents of the published materials are solely the responsibility of the administering institution, Flinders University, and the individual authors identified, and do not reflect the views of the National Health and Medical Research Council or any other funding bodies or the funding partners. G. Chen was supported by a grant funded by the financial support of Cancer Council SA’s Beat Cancer Project on behalf of its donors and the state government of South Australia through the Department of Health together with the support of the Flinders Medical Centre Foundation, its donors, and partners.

      References

        • Organisation for Economic Co-operation and Development
        Long-Term Care for Older People.
        OECD Publishing, Paris2005
        • Lueke S.
        • Hoffmann W.
        • Flebetaa S.
        Transitions between care settings in dementia: Are they relevant in economic terms?.
        Value Health. 2014; 17: 679-685
        • Mulvey J.
        • Li A.
        Long-term care financing: options for the future.
        Benefits Q. 2002; 18: 7-14
        • Howe A.L.
        • Jones A.E.
        • Tilse C.
        What’s in a name? Similarities and differences in international terms and meanings for older peoples’ housing with services.
        Ageing Soc. 2012; 33: 547-578
        • Castle N.G.
        • Ferguson J.C.
        What is nursing home quality and how is it measured?.
        Gerontologist. 2010; 50: 426-442
        • Donabedian A.
        Twenty years of research on the quality of medical care: 1964–1984.
        Eval Health Prof. 1985; 8: 243-265
        • Nakrem S.
        • Vinsnes A.G.
        • Harkless G.E.
        • et al.
        Nursing sensitive quality indicators for nursing home care: international review of literature, policy and practice.
        Int J Nurs Stud. 2009; 46: 848-857
        • Rahman A.N.
        • Applebaum R.A.
        The nursing home minimum data set assessment instrument: manifest functions and unintended consequences—past, present, and future.
        Gerontologist. 2009; 49: 727-735
        • Drummond M.F.
        • Sculpher M.J.
        • Torrance G.W.
        • et al.
        Methods for the Economic Evaluation of Health Care Programmes. 3rd ed. Oxford University Press, Oxford, UK2005
        • Makai P.
        • Brouwer W.B.
        • Koopmanschap M.A.
        • et al.
        Quality of life instruments for economic evaluations in health and social care for older people: a systematic review.
        Soc Sci Med. 2014; 102: 83-93
        • Brazier J.
        • Ratcliffe J.
        The measurement and valuation of health for economic evaluation.
        in: Heggenhougen K. International Encyclopedia of Public Health. Academic Press, Oxford, UK2008
        • Donaldson C.
        • Atkinson A.
        • Bond J.
        • et al.
        Should QALYs be programme-specific?.
        J Health Econ. 1988; 7: 239-257
        • Coast J.
        Is economic evaluation in touch with society’s health values?.
        BMJ. 2004; 329: 1233-1236
        • Ryan M.
        • Netten A.
        • Skåtun D.
        • et al.
        Using discrete choice experiments to estimate a preference-based measure of outcome—an application to social care for older people.
        J Health Econ. 2006; 25: 927-944
        • Brau R.
        • Lippi Bruni M.
        Eliciting the demand for long-term care coverage: a discrete choice modelling analysis.
        Health Econ. 2008; 17: 411-433
        • Dixon S.
        • Nancarrow S.A.
        • Enderby P.M.
        • et al.
        Assessing patient preferences for the delivery of different community-based models of care using a discrete choice experiment.
        Health Expect. 2015; 18: 1204-1214
        • Fernández-Carro C.
        Ageing at home, co-residence or institutionalisation? Preferred care and residential arrangements of older adults in Spain.
        Ageing Soc. 2016; 36: 586-612
        • Kaambwa B.
        • Lancsar E.
        • McCaffrey N.
        • et al.
        Investigating consumers’ and informal carers’ views and preferences for consumer directed care: a discrete choice experiment.
        Soc Sci Med. 2015; 140: 81-94
        • Negrín M.A.
        • Pinilla J.
        • León C.J.
        Willingness to pay for alternative policies for patients with Alzheimer’s disease.
        Health Econ Policy Law. 2008; 3: 257-275
        • Nieboer A.P.
        • Koolman X.
        • Stolk E.A.
        Preferences for long-term care services: willingness to pay estimates derived from a discrete choice experiment.
        Soc Sci Med. 2010; 70: 1317-1325
        • Milte R.
        • Shulver W.
        • Killington M.
        • et al.
        Quality in residential care from the perspective of people living with dementia: the importance of personhood.
        Arch Gerontol Geriatr. 2016; 63: 9-17
        • Coast J.
        • Al-Janabi H.
        • Sutton E.J.
        • et al.
        Using qualitative methods for attribute development for discrete choice experiments: issues and recommendations.
        Health Econ. 2012; 21: 730-741
        • Milte R.
        • Ratcliffe J.
        • Bradley C.
        • et al.
        Evaluating the quality of care received in long-term care facilities from a consumer perspective: development and construct validity of the Consumer Choice Index–Six Dimension instrument.
        Ageing Soc. 2017; (http://dx.doi.org/10.1017/S0144686X17000861): 1-23
        • Rowen D.
        • Mulhern B.
        • Banerjee S.
        • et al.
        Estimating preference-based single index measures for dementia using DEMQOL and DEMQOL-Proxy.
        Value Health. 2012; 15: 346-356
        • Clarence-Smith B.
        Healthcare for people with dementia in care homes: family carer experiences.
        J Dement Care. 2009; 17: 36-38
        • Edwards H.
        • Courtney M.
        • Spencer L.
        Consumer expectations of residential aged care: reflections on the literature.
        Int J Nurs Pract. 2003; 9: 70-77
        • Brazier J.
        • Ratcliffe J.
        • Salomon J.A.
        • et al.
        Measuring and Valuing Health Benefits for Economic Evaluation. 2nd ed. Oxford University Press, Oxford, UK2017
        • Ryan M.
        • Gerard K.
        Using discrete choice experiments to value health care programmes: current practice and future research reflections.
        Appl Health Econ Health Policy. 2003; 2: 55-64
        • Ryan M.
        • Scott D.
        • Reeves C.
        • et al.
        Eliciting public preferences for healthcare: a systematic review of techniques.
        Health Technol Assess. 2001; 5: 1-186
        • Ratcliffe J.
        • Laver K.
        • Couzner L.
        • et al.
        Not just about costs: the role of health economics in facilitating decision making in aged care.
        Age Ageing. 2010; 39: 426-429
        • ChoiceMetrics
        Ngene 1.1.2 User Manual and Reference Guide.
        ChoiceMetrics, Sydney, Australia2014
        • Jorm A.F.
        • Mackinnon A.J.
        • Henderson A.S.
        • et al.
        The Psychogeriatric Assessment Scales: a multi-dimensional alternative to the categorical diagnoses of dementia and depression in the elderly.
        Psychol Med. 1995; 25: 447-460
        • Milte R.
        • Ratcliffe J.
        • Chen G.
        • et al.
        Cognitive overload? An exploration of the potential impact of cognitive functioning in discrete choice experiments with older people in health care.
        Value Health. 2014; 17: 655-659
        • McFadden D.
        • Train K.
        • Mixed M.N.L.
        models for discrete response.
        J Appl Econom. 2000; 5: 447-470
        • Coast J.
        • Flynn T.N.
        • Natarajan L.
        • et al.
        Valuing the ICECAP capability index for older people.
        Soc Sci Med. 2008; 67: 874-882
        • Swait J.
        • Louviere J.
        The role of the scale parameter in the estimation and comparison of multinomial logit models.
        J Mark Res. 1993; 30: 305-314
        • Hensher D.A.
        Accounting for scale heterogeneity within and between pooled data sources.
        Transp Res Part A Policy Pract. 2012; 46: 480-486
        • Lancsar E.
        • Louviere J.
        Conducting discrete choice experiments to inform healthcare decision making.
        Pharmacoeconomics. 2008; 26: 661-677
        • Lancsar E.
        • Louviere J.
        Deleting “irrational” responses from discrete choice experiments: A case of investigating or imposing preferences?.
        Health Econ. 2006; 15: 797-811
        • Gott M.
        • Seymour J.
        • Bellamy G.
        • et al.
        Older people’s views about home as a place of care at the end of life.
        Palliat Med. 2004; 18: 460-467
        • Tucker S.
        • Brand C.
        • Sutcliffe C.
        • et al.
        What makes institutional long-term care the most appropriate setting for people with dementia? Exploring the influence of client characteristics, decision-maker attributes, and country in 8 European nations.
        J Am Med Dir Assoc. 2016; 17 (e9–15): 465
        • Chau P.H.
        • Kwok T.
        • Woo J.
        • et al.
        Disagreement in preference for residential care between family caregivers and elders is greater among cognitively impaired elders group than cognitively intact elders group.
        Int J Geriatr Psychiatry. 2010; 25: 46-54
        • Wolff J.L.
        • Kasper J.D.
        • Shore A.D.
        Long-term care preferences among older adults: A moving target?.
        J Aging Soc Policy. 2008; 20: 182-200
        • Salkeld G.
        • Ryan M.
        • Short L.
        The veil of experience: Do consumers prefer what they know best?.
        Health Econ. 2000; 9: 267-270
        • van Someren M.
        • Barnard Y.
        • Sandberg J.
        The Think Aloud Method: A Practical Guide to Modelling Cognitive Processes. Academic Press, London1994