Preferences of Cognitively Impaired Patients and Patients Living with Dementia: A Systematic Review of Quantitative Patient Preference Studies

Abstract

Background:

Treatment decisions based on guidelines rather than patients’ preferences determine adherence to and compliance with treatment, which, in turn, could improve health-related outcomes.

Objectives:

To summarize the stated treatment and care preferences of people with dementia (PwD).

Methods:

A systematic review was conducted to assess the stated preferences of PwD. The inclusion criterion was the use of quantitative methods to elicit stated preferences, enabling a ranking of preferences.

Results:

Eleven studies revealed preferences for diagnostics, treatment decisions, patient-related outcomes, care services, end-of-life care, leisure activities, and digital life story work. PwDs prefer accurate, pain-free, and comfortable diagnostic procedures without radioactive markers as well as being accompanied by a caregiver. PwD’s quality of life (QoL), self-efficacy, and depression were equally most important for PwD and caregivers. However, PwD memory was only important for caregivers but not for PwD, and caregiver QoL was moderately important for PwD but least important for caregivers. Additionally, comfort and family involvement were most important for patients’ end-of-life care, whereas caregivers most preferred good communication and pain management. Also, preferences depend on the living situation: Patients living not alone prefer a regular care provider most, whereas those living alone only want to live nearby the caregiver. Preferences for leisure activities did not differ between past and present ratings, indicating that PwD prefer activities that have always been carried out.

Conclusion:

Only a few studies have applied quantitative methods to elicit the preferences of PwD. More research is needed to capture the stated preferences for the treatment, care, and support of PwD to improve health-related outcomes and the allocation of healthcare resources.

Keywords

Dementia decision making patient outcome assessment patient preference

INTRODUCTION

All over the world, life expectancy is increasing, leading to an aging society and an increase in age-associated illnesses, such as dementia [1]. Currently, there are approximately 46.8 million people with dementia (PwD) worldwide, and most of them suffer from Alzheimer’s disease [2]. Half of the affected people are older than 80 years old, and two-thirds are women [2]. Without a cure, it is estimated that there will be 131.5 million PwD in 2050 worldwide [3].

PwD requires a timely diagnosis and evidence-based patient-centered treatment to delay the progression of the disease. Attenuating functional and cognitive decline could improve or at least maintain patients’ health-related quality of life (HRQoL) [4 –6]. However, treatment guidelines are based on clinical evidence from different studies that aimed to improve, in most cases, clinical outcomes. Such outcomes are often not strictly associated with an improvement in health-related quality of life that could be noticed subjectively by the patients. In addition, other patient-related treatment characteristics, such as the mode of administration of drugs, are rarely considered in clinical studies. Therefore, the perspective of the affected patients and the patient preferences are rarely taken into account in both evidence-based guidelines and treatment and care decisions.

During the progression of dementia, PwD become increasingly cognitively and functionally impaired, resulting in an increasing need for care and further support. In the beginning stages of the diseases, approximately 75% of PwD are living at home, and the majority are supported by their close relatives [7]. Living next to their relatives at their own home is one major preference of many PwD [8]. In these times, PwD’s relatives provide informal care [9]. In most cases, the disease progresses to such a point that the informal caregivers cannot manage the intensity or the community-dwelling living situation, so institutionalization becomes inevitable [10]. Thus, it can be assumed that the priorities and preferences of PwD and their informal caregivers change over the course of the dementia [11]. Therefore, after receiving the diagnosis, PwD have to make many decisions, e.g., about their current and future living situation, therapy, level of care and support needed, and they need to define cornerstones for their patient decree for the time when they become unable to make decisions due to increasing cognitive impairment [12, 13]. All these fundamental decisions should be made in good consideration and with adequate support [14].

For optimum management of the dementia disease and to initiate individual, tailored, and patient-centered treatment, care, and support, a patient’s preferences for all affected domains and all stages of the disease must be known. However, there are considerable methodological challenges in validly eliciting the preferences of PwD. It is unclear how to best define preferences. Qualitative preference studies, such as interviews, are the most common method used to detect patient priorities [15]. In these studies, the quality of care, contact with relatives, person-centered caregiving, emphasizing dignity and respect were indicated as important for PwD [8]. However, decisions between alternatives that have different attributes or characteristics have not to be made, missing the opportunity to generate a numeric ranking of preferences, including a description of quantitative differences between various preferences and patient groups that were based on standardized methods. Within a complex healthcare system with scarce resources, it is important to know which aspects (attributes) of a treatment and which outcomes are most and least important from a patient’s perspective. Addressing the most preferred attributes could increase adherence to treatments and therapies, improve and individualize interventions, and finally improve patient-related outcomes [16]. Avoiding the least preferred attributes could save already scarce healthcare resources and preserve patients from low-value interventions, which, in turn, could improve patient-related outcomes. Those methods are grounded in microeconomy and pertain to the stated preference techniques, such as discrete-choice experiments, best-worst scaling, analytic hierarchy process and other types of survey questions using ranking and rating techniques [17, 18].

Therefore, the objective of this systematic review was to summarize the stated preferences of cognitively impaired people and PwD, considering all domains that could be affected by cognitive decline, such as the clinical status, symptoms, functioning, living situation, medical treatment, care, support, and patient-related outcomes.

MATERIALS AND METHODS

To identify and demonstrate patients’ stated preferences in cognitive impairment or dementia, a systematic review of quantitative studies was conducted.

Search strategy

The literature search was performed in PubMed, Science Direct, and Cochrane Library using a non-MeSH (medical subject headings) and MeSH search. The non-MeSH search used the key phrases “dementia” or “Alzheimer” or “Alzheimer’s disease” in combination with “preference*” or “priorit*” or “rank*”. MeSH terms were additionally used to find all studies that were linked to the terms in the PubMed database “mild cognitive impairment” or “dementia” in combination with the MeSH heading “patient preference”. Both search terms (non-MeSH and MesH) were combined using the OR operator. The systematic literature search included studies published between 1985 and April 12, 2019. The search strategy is presented in the Supplementary Material.

Data extraction

Studies were included if 1) the articles were written in English or German language, 2) the patients had cognitive impairment or dementia, Alzheimer’s disease, or cognitively healthy participants from the general population who were asked hypothetically to imagine having dementia; and 3) the preferences had been elicited based on stated quantitative preference methods, such as discrete choice experiment, analytic hierarchy process, best-worst scaling, or other quantitative rating methods, enabling a ranking of stated preferences of PwD to elicit the most and least important preferences. Studies using a qualitative approach, such as interviews or open question surveys, were excluded. Additionally, we excluded systematic reviews and study protocols.

Overall, the search strategy yielded 578 articles. After removing 34 duplicates, 544 studies were retrieved for title and abstract screening. Two independent reviewers screened the titles and abstracts (SL, HW). In cases where the reviewers could not make a decision, a third reviewer was involved (AR, BM). Overall, 440 studies (see Flowchart) were excluded after title and abstract screening, and 93 were excluded after full text screening, resulting in the inclusion of eleven studies. The flow chart of the screening procedure is demonstrated in Fig. 1.

Fig. 1

Study Flow Chart (CONSORT).

Data description

We extracted basic information describing the studies, including year of publication, author, country, aim of the study, and the sample size. We furthermore extracted the quantitative method used to elicit stated preferences. The coefficients of demonstrated preferences were used to derive the relative importance of each attribute, generating the ranking of stated preferences, which was the main result of this review. In some cases, the ranking of attributes was demonstrated for different subgroups, such as patients, caregivers and dyads. In these cases, each ranking of each subgroup was elicited and presented.

Quality assessment of included studies

To assess the methodological quality of the included studies, the PREFS checklist (Purpose, Explanation, Findings, Significance) was used [19]. This checklist addresses the preference (Purpose), the possible selection bias (Respondents), the description of the used method (Explanation), the possible reporting bias (Findings), and the use of significance testing (Significance). We excluded the Respondents assessment because preference studies do not calculate how much respondents stopped or did not answer. According to this checklist, studies were ranked between zero, indicating a weak quality, and four, indicating the highest methodological quality (for details see Table 1) [19]. Two reviewers checked and judged the quality of the studies separately (SL, AR). If there was no consensus, a third reviewer became involved (BM).

Table 1

Valuation of the PREFS checklist for the selected studies

Study	Purpose: Is the purpose of the study related to preferences clearly stated?	Explanation: Are the methods of assessing preferences clearly explained?	Findings: Were all respondents included in the reported findings and analysis of preference results?	Significance: Were significance tests used to analyze the preference results?	Score/Grade
Bamford et al. [20]	x	x		x	3/4
Mühlbacher et al. [21]	x	x	x	x	4/4
Janus et al. [22]	x	x		x	3/4
Barrios et al. [25]	x	x		x	3/4
Smith et al. [26]	x	x		x	4/4
Groenewoud et al. [28]	x	x	x	x	4/4
Nieboer. et al. [27]	x	x	x	x	4/4
Cohen-Mansfield et al. [23]	x	x		x	3/4
O’Philbin et al. [24]	x	x		x	3/4
Mulqueen et al. [29]	x	x	x	x	4/4
Dening et al. [8]	x	x	x	x	4/4

Scale: 0 = poor, 1 = fair, 2 = good, 3 = very good, 4 = excellent.

RESULTS

Characteristics of the included studies

All eleven studies were published between 2010 and 2018. These studies considered a wide range of different attributes regarding the diagnostic procedure [20, 21], treatment preferences [22], leisure activities [23, 24], patient-reported outcomes [25, 26], care services [27, 28], and end-of-life care [8, 29]. One study was conducted in Canada, five were conducted in the United Kingdom (UK), one was conducted in Germany, three were conducted in the Netherlands, and two were conducted in the United States of America. The sample sizes of the included patients ranged between eight and 1,301. A description of the quality grading and characteristics of the studies is shown in Table 1 and Table 2, respectively.

Table 2

Preference studies of people with cognitive impairment and dementia

Author, Year	Population, country	Aim of the study	Method	N	Results: Rankings
Diagnosis
Bamford et al., 2016 [20]	Patients over 60 years with mild/moderate dementia (MMSE > 12); UK	Assess preferences for diagnostic processes (PET and SPECT), comparison to controls	Card sorting exercise to importance (ranking)	68 patients, 59 caregivers, 30 controls (age-matched, no signs or symptoms of dementia)	patients*1. accuracy2. helpful staff3. comfort4. companion5. radiation dose6. space in the scanner7. duration8. distance to the next scanner9. fasting10. noise	caregivers, for patients1. accuracy2. helpful staff3. comfort4. companion5. radiation dose6. duration7. space in the scanner8. distance9. fasting/Noise	controls1. accuracy2. helpful staff3. comfort4. radiation dose5. space in the scanner6. fasting/noise7. duration8. distance/ companion
Mühlbacher et al., 2016 [21]	Above 60 years old general Population in UK and Germany	Preferences of older adults without memory problem for AD test technologies	DCE	1301	patients1. test imprecision 5% 2. test without radioactive marker3. test imprecision 15% 4. test cost5. test with radioactive marker6. no test7. spinal tab8. test imprecision 30%
Treatment
Janus et al., 2017 [22]	Healthcare professionals of 23 nursing homes, proxies of PwD living in the same nursing homes; the Netherlands	Treatment preferences of healthcare professionals and proxies of PwD	Best-worst-scaling (standard scores)	81 caregivers, 59 proxies	Pharmacological caregivers1. fastest results2. recommended by physicians and pharmacists3. most effective4. most appropriate for these complaints5. minimally burdensome for the patient6. patient’s preference7. most experience with this regimen8. recommended in guideline9. patient’s family asks for this treatment10. care organization has resources to offer this treatment11. caregivers request this treatment12. requires least amount of monitoring13. most feasible for caregiver	proxies of PwD1. most effective2. fastest result3. least negative effect4. physician’s preference5. minimally burdensome for the patient6. caregivers recommend this treatment7. patient’s preference8. most experience with this regimen9. requires the least amount of effort from me10. lowest cost
					14. least negative effect15. lowest cost16. requires the least amount of effort from staff
					nonpharmacological caregivers1. minimally burdensome for the patient2. most appropriate for these complaints3. patient’s preference4. most effective5. recommended by physicians and pharmacists6. care organization has resources to offer this treatment7. recommended in guideline8. patient’s family asks for this treatment9. requires the least amount of effort10. fastest results11. least negative effect12. caregivers request this treatment13. requires the least amount of effort from the staff14. lowest cost15. most feasible for caregiver16. most experience with this regimen	proxies of PwD1. least negative effect2. most effective3. minimally burdensome for the patient4. patient’s preference5. fastest result6. caregivers recommend this treatment7. physician’s preference8. most experience with this regimen9. lowest cost10. requires the least amount of effort from me
Patient-reported Outcomes
Barrios et al., 2016 [25]	Couples of summer 2014 HABIT session at MAYO clinic Rochester (MMSE≥27, USA	Ranking of behavioral outcomes of Patients with MCI and their caregivers, pre- and post-ranking after participation of the program	Ranking task	16 couples	Pre-intervention ranking patients:1. QoL of patient2. self-efficacy of patient3. QoL of caregiver4. depression of caregiver5. caregiver burden6. daily functioning7. anxiety of caregiver8. memory based-ADLs9. self-efficacy of caregiver10. anxiety of patient	Pre-intervention ranking caregivers:1. QoL of patient2. depression of patient3. self-efficacy of patient4. depression of caregiver5. QoL of caregiver6. self-efficacy of caregiver
					11. depression of patient12. distressing behaviorPost-intervention ranking patients:1. caregiver burden2. QoL of caregiver3. self-efficacy of patient4. QoL of patient5. anxiety of patient/depression of caregiver6. memory based-ADLs7. self-efficacy of caregiver8. daily functioning9. depression of patient10. anxiety of caregiver11. distressing behavior	7. memory based-ADLs8. daily functioning9. anxiety of caregiver10. anxiety of patient11. caregiver burden12. distressing behaviorPost-intervention ranking caregivers:1. QoL of patient2. QoL of caregiver3. self-efficacy of patient4. depression of caregiver5. daily functioning/self-efficacy caregiver6. memory based-ADLs7. depression of patient8. anxiety of partner9. caregiver burden10. distressing behavior11. anxiety of patient
Smith et al., 2018 [26]	HABIT program completers (2008-2014), USA	Outcome and treatment preferences of patients with MCI and their caregivers	Ranking task	88, only 29 Patient direct, 12 Patient and partner	patients1. patient QoL2. patient self-efficacy3. patient depression4. patient basic ADL5. patient memory ADL/caregiver QoL6. patient anxiety/caregiver burden7. caregiver self-efficacy8. caregiver anxiety/patient distressing behavior	caregivers for patients1. patient QoL2. patient self-efficacy/patients’ depression3. patient anxiety/patient memory ADL/patient basic ADL4. patient actual memory5. patient distressing behavior
					9. caregiver mood 10. patient actual memory	6. caregiver burden7. caregiver self-efficacy/caregiver anxiety8. caregiver QoL/caregiver mood
					patients and caregivers:1. patient QoL2. patient actual memory3. patient self-efficacy4. patient memory ADL/patient basic ADL5. partner QoL6. patient depression/partner self-efficacy7. patient anxiety/partner burden8. partner anxiety9. partner mood/patient distressing behavior
Activities
Cohen-Mansfield et al., 2017 [23]	PwD from seven home units and a senior day center; Toronto, Canada	Past and present preferences of PwD for leisure activities; influence of demographic and functional abilities	Ranking task reported by family and staff	77	family rating past (proxy):1. listening to music2. reading3. reminiscing4. participating in groups5. exercising6. viewing and/or producing art, painting, etc.7. playing physical games, such as ball or bowling8. cooking/baking9. playing trivia/word games	staff rating present (proxy):1. listening to music2. reminiscing3. viewing and/or producing art, painting, etc.4. reading5. participating in groups6. playing physical games such as ball or bowling7. cooking/baking8. exercising9. playing trivia/word games
					family rating present (proxy):1. listening to music2. reminiscing3. reading4. participating in groups5. exercising
					6. viewing and/or producing art, painting, etc.7. playing physical games, such as ball or bowling8. playing trivia/word games9. cooking/baking
O’Philbin et al., 2018 [24]	PwD living in the UK	Choices concerning digital life story work interventions and services	Online survey for patients (DCE character); online DCE for caregiver, 5 attributes	17 patients, 67 caregivers	patients1. setting: private one to one 71% 2. focus on learning how to use the app 65% 3. mainly use to show family and friends 65% 4. paying for the service 53% 5. no paying for the service 47% 6. being taught how to use it, be helped to create 41% 7. service of getting build a book 35% 8. setting: community group 29% 9. work individually at home 24% 10. enjoy memories/show to new people/having the service build the book for them 18% 11. written Instruction guide 12%	caregivers1. intervention setting2. elementary app usability3. follow-up assistance4. session focus5. price of intervention
Care service
Nieboer et al., 2010 [27]	50–65 years old general Population; long-term care scenarios, the Netherlands	Hypothetical preferences in the general elderly population for long-term care services, living situation (in comparison with living independently at home)	DCE, 10 attributes	1082	patients living alone:1. apartment in the proximity of care 2. regular care providers/coordinated care service delivery 3. 4 months shorter on waiting list 4. transportation service 5. +0.5 day extra social activity 6. elderly/nursing home 7. sheltered accommodation 8. out-of-pocket spendings 9. better punctuality 10. +4h extra care per week 11. individual preferences	patients living with partner:1. regular care provider 2. transportation service 3. out-of-pocket spendings 4. coordinated care services delivery 5. 4 months shorter on waiting list 6. better punctuality 7. +4h extra care per week / 0.5 day extra social activity 8. apartment in the proximity of care 9. individual preferences 10. sheltered accommodation 11. elderly/nursing home
Groenewoud et al., 2015 [28]	Dutch patient sample	Influences in patients’ health care decisions	DCE, 11 attributes	421 patients with AD	patients, outcome-orientated1. travel distance 5 km2. feel safe and comfortable 90% 3. 16 h of care per week4. good interpersonal treatment 75% 5. institution specialized in AD6. travel distance 20 km7. 3 staff members per 15 residents8. feel safe and comfortable 70% 9. deliver care as agreed: always10. institution with a special ward11. risk of pressure ulcers 2% 12. 10 h of personal care/week13. good interpersonal treatment 50% 14. 2 staff members per 15 residents15. waiting time 4 months16. communication staff 90% satisfied17. recommended by specialist18. risk of pressure ulcers 10% 19. waiting time 8 months20. deliver care as agreed: sometimes21. communication staff 70% satisfied22. recommended by family and friends	patients, trust-orientated1. travel distance 5 km2. travel distance 20 km3. 3 staff members per 15 residents4. waiting time 4 months5. feel safe and comfortable 90% 6. risk of pressure ulcers 2% 7. institution with a special ward8. institution specialized in AD9. 10 hours of personal care/week10. good interpersonal treatment 75% 11. 2 staff members per 15 residents12. 16 h of care per week13. risk of pressure ulcers 10% 14. communication staff 90% satisfied
						15. feel safe and comfortable 70% 16. deliver care as agreed: always17. good interpersonal treatment 50% 18. recommended by family and friends19. recommended by specialist20. communication staff 70% satisfied21. waiting time 8 months22. deliver care as agreed: sometimes
End-of-life care
Dening et al., 2012 [8]	PwD, caregivers and dyads attending memory assessment service, UK	Preferences for end-of-life care of PwD, family caregivers and influence of the caregiver on preferences of PwD	Nominal group technique	6 patients, 5 caregiver, 6 dyads	patients1. family contact2. independence3. safety4. dignity5. not being a burden6. choice of place of care7. pleasurable activities8. person-centered care9. in touch with the world10. comfortable environmentpatients and caregiver1. choice of place of care2. not being a burden3. respect4. no unnecessary prolonging life5. activities6. family contact7. legal preparation	preferences of caregivers1. to be in control2. quality of life3. quality of care4. comfortable death5. respect and dignity6. caregiver support
Mulqueen et al., 2017 [29]	PwD living in a residential care facility, Ireland	Preferences of PwD for end-of-life care, perceptions of their caregivers	Nominal group technique	6 patients, 6 caregivers	patients1. comfort2. family involvement3. familiar surrounding4. communication5. familiar staff	caregivers1. communication2. pain management3. advance care planning
						4. family involvement5. staff knowledge

AD, Alzheimer’s disease; ADL, Activities of Daily Living; DCE, discrete choice experiment; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; PET, positron emission tomography; PwD, people with dementia; QoL, quality of life; SPECT, single photon emission computed tomography

Diagnostic procedure

Based on a card sorting exercise, Bamford et al. [20] revealed preferences for diagnosing dementia using positron emission tomography (PET) or single photon emission computed tomography (SPECT) in the UK in 2016. Patients over 60 years (n = 68) with mild or moderate dementia (Mini-Mental State Examination > 12) ranked accuracy (rank 1), helpful staff (rank 2), and comfort during scans as most important (rank 3). The accuracy of the diagnostic procedure was perceived as more important than the circumstances of the scan, such as radiation dose (rank 5), space in the scanner (rank 6), duration of the scanning procedure (rank 7), or noise (rank 10). Most elicited attributes were similarly preferred between patients, caregivers, and healthy controls. However, compared to healthy controls, the presence of a companion, such as a caregiver, was very important for PwD and caregivers but not for healthy controls (rank 4 versus rank 8).

Mühlbacher et al. [21] elicited similar preferences for the general population older than 60 image having dementia in the UK and Germany (n = 1,301). Based on a discrete choice experiment (DCE), the authors investigated preferences for dementia test technologies. The preferences did not differ significantly between the two countries. The respondents reported that a test imprecision of 5% (Coeff: 10.0) and a test without a radioactive marker (Coeff: 9.0946) were most important, followed by test imprecisions of 15% (Coeff: 6.0402) and the cost for the test (Coeff: 5.3259). The absence of the test (Coeff: 1.9769), a spinal tap (Coeff: 1.7451), and a test imprecision of 30% (Coeff: 0.0) were least important.

Treatment

Janus et al. [22] elicited preferences for pharmacological or nonpharmacological treatments for caregivers (n = 81) and proxies (n = 59) of PwD using best-worst scaling techniques. Considering pharmacological treatment, caregivers reported that the fastest results (standard score: 0.36), recommendations by physicians and pharmacists (standard score: 0.33) and the most effective treatment (standard score: 0.29) were most important, but treatments with the lowest costs (standard score: –0.43) and the least amount of effort for themselves (standard score: –0.51) were less important. For the proxies, the most effective treatment (standard score: 0.58), the fastest result (standard score: 0.53), and the least negative effect (standard score: 0.36) are most important, whereas the treatment with the lowest cost (standard score: –0.75) was the least important. Concerning the preferences for nonpharmacological treatments, caregivers reported that minimally burdensome treatments for the patient (standard score: 0.35), the most appropriate treatment for the complaints (standard score: 0.27) and the patient’s preferences (standard score: 0.26) were most important, but the most feasible treatment (standard score: –0.36) and the most widely used treatment (standard score: –0.48) were least important. For the nonpharmacological treatment, the proxies of PwD reported that the least negative effect (standard score: 0.67) and the most effective treatment (standard score: 0.45) were most important, whereas the lowest cost (standard score: –0.57) and the treatment that required the least amount of effort (standard score: –0.67) were least important.

Patient's and caregiver’s outcomes

Barrios et al. [25] investigated the preferences for behavioral outcomes of seven patients with mild cognitive impairment (MCI) and their caregivers with a ranking task in the United States in 2016. Participants were recruited from a memory training program that includes cognitive rehabilitation and cognitive stimulation training and were asked for their preferences before and after the training program [30]. Before starting the training program, PwD ranked their own quality of life (QoL) (rank 1) and self-efficiency (rank 2) as most important, followed by caregiver QoL (rank 3) and depression of the caregiver (rank 4). Whereas PwD and caregivers equally preferred patients’ QoL and self-efficacy as most important, all other preferences significantly differed between both groups. For caregivers, patients’ depression was the second most important, but this factor was rather unimportant for patients themselves (rank 11). Furthermore, patients rated caregivers’ QoL and anxiety as more important and the self-efficacy and depression of the caregiver as less important compared to the rating the caregivers made by themselves. Notably, the rankings slightly changed over time: caregivers’ outcomes were more important for PwD after the intervention than they were before. Patients ranked caregiver burden as most important after the intervention, whereas this attribute was ranked as the 5th most important before receiving the intervention. Furthermore, after the intervention, caregivers rated their own QoL as the second most important preference, but it was only moderately important before the intervention (rank 5).

Smith et al. [26] analyzed the preferences of people with MCI (n = 29) within the same memory training program of Barrios et al. but based on a different and larger sample using a ranking task. In this analysis, patient preferences significantly differ from the analysis of Barrios et al. [25], which could be due to bias and a given uncertainty in eliciting preferences using small sample sizes. However, patients’ QoL (rank 1), self-efficacy (rank 2). and depression (rank 3) were most important for people with MCI (n = 29). Whereas patients’ QoL and self-efficacy still represented the most important stated preference, PwD’s depression was much higher ranked (patients: rank 3) compared to the findings of Barrios et al. [25] (patients pre-intervention: rank 11). Caregiver outcomes were less important for people with MCI. The QoL of the caregiver was ranked as moderately important for the patients (rank 5) but least important for the caregivers (rank 8). Furthermore, there were significant differences concerning the actual memory capacity between the ranking solicited from the PwD and the ranking by the caregivers. Memory was more important for the couples (dyads, patients and their caregivers) and caregivers (rank 2 and rank 3) than it was for PwD themselves (rank 10). Another difference was observed in the preferences concerning depression. Both PwD and caregivers ranked this outcome as more important in the absence of their caregiver or PwD, respectively. When couples were interviewed, both jointly ranked these attributes lower. In contrast to Barrios et al. [25], Smith et al. [26] added the attribute “actual memory performance” to the decision model. However, this additional attribute appears to be not responsible for the different results of similar studies (Barrios and Smith) because differences were observed mainly in the ranking for caregivers’ QoL and mood.

Activities

The study of Cohen-Mansfield et al. [23] examined proxy-reported preferences concerning the leisure activities of n = 77 PwD given by staff and family members. They ranked the activities separately for the past and present preference of the PwD. Listening to music (rank 1) and reminiscing (rank 2) were ranked as the highest preference for the PwD rated by family members and staff. Participating in group activities (rank 4 for family members and rank 5 for staff) and exercising (rank 5 for family members and rank 8 for staff) were moderately attractive, and playing trivia/word games and cooking/baking were least preferred for PwD (rank 9). However, females ranked cooking/baking much higher than males, revealing gender-specific differences in such preferences. The ranking of preferred activities in the past does not differ from the present rating (correlation 0.97). Differences in the preference for exercising were observed between family (5th position) and staff (8th position).

The study of O’Philbin [24] analyzed the preferences for digital life story work (LSW) service interventions of PwD and their caregivers in the UK. They were asked to be encouraged to perform digital LSW with the result of their own book. The PwD n = 17 completed an online survey with a DCE character, whereas the caregiver n = 67 joined a DCE to elicit their preferences. PwD in the UK most preferred a private setting (rank 1) and a session that focused on learning how to use the given app (rank 2). Presentation to friends and family seems to be the most common use of a digital LSW (rank 3); 53% of the respondents reported that they would pay for the service (rank 4), whereas 47% would not pay for the service (rank 5). Enjoying memories with the book, showing it to new people and having a service to get a book built (rank 10) were less preferred. The least preferred attribute was an instruction guide for the work (rank 11). In addition, for caregivers, the private setting is the most important attribute (Coeff: –0.802, willingness to pay (WTP): 41.45£), followed by a usability of the app (elementary Coeff: 0.180 WTP: –9.32£, advanced Coeff: –0.180) and follow-up assistance (Coeff: –0.150, WTP: 7.75£). A session focus (Coeff: –0.032) and a price for the intervention (Coeff: –0.019) were less preferred. In this study, the caregivers responded not to the patients but to themselves. The preferences of both groups did not differ significantly. PwD and caregivers paid special attention to ensure that the intervention takes place in a one-to-one setting.

Care services

Nieboer et al. [27] identified the preferences for long-term care services in a general Dutch elderly population (n = 1,082) wherein patients imagined having dementia using a DCE. Elicited preferences differed between patients living with a partner and those living alone. The most preferred attribute of patients living with a partner was a regular care provider (Coeff: –0.493, WTP: 88€), followed by transportation service (Coeff = –0.306, WTP: 55€), out-of-pocket spending for the care (Coeff: –0.280), coordinated care service (Coeff = –0.220, WTP: 39€), and short time on waiting list for all services (Coeff: –0.212, WTP: 38€). For patients without a partner, living near the caregiver was the most important attribute (Coeff: 0.498, WTP: 177€), followed by having a regular care provider (Coeff = –0.434, WTP: 154€), time on a waiting list (Coeff = –0.294, WTP: 154€), and transportation services (Coeff = –0.247, WTP: 88€). For people living alone, all services were found to be relatively more important than for those living not alone.

The attributes regular care provider and coordinated care service were ranked equally between both groups. Living in proximity of the caregiver was most important for patients living alone but was not important for patients with a partner. Furthermore, social activities were more important for those living alone (Coeff: 0.227 versus 0.144). Timeliness of the care provider (Coeff: –0.122), 4 hours of additional care per week (Coeff: 0.117) and individual preferences in care (Coeff: 0.068) were ranked as less important for the general elderly in the Netherlands.

Groenewoud et al. investigated preferences in patients’ health care decisions by using a DCE for a Dutch sample of n = 421 PwD in 2015. The representatives were divided into two subgroups: outcome-orientated and trust-orientated responders. The results of the groups differed only for a few preferences. However, a low travel distance (up to 5 km) to the health care provider was most preferred for both groups (Coeff: 1.035 outcome-orientated and 1.516 trust-orientated). For the outcome-oriented respondents, safety (Coeff: 0.923), 16 h extra care per week (Coeff: 0.779) and interprofessional treatment (Coeff: 0.709) were most preferred. Less preferred attributes for this subgroup were “Delivered care as agreed: sometimes” (Coeff: 0.131), “satisfied communication with staff 70% ” (Coeff: 0.084) and recommendations for a health care provider from family and friends (Coeff: 0.025). The trust-oriented subgroup most preferred a “travel distance to 20 km” (Coeff: 0.889) and “3 staff members per 15 residents” (Coeff: 0.769). In addition, the attribute “Waiting time 4 months” (Coeff: 0.699) was important. The least preferred attributes were “satisfied communication with staff 70% ” (Coeff: 0.250), waiting time of up to 8 months (Coeff: 0.201) and “delivered care as agreed: sometimes” (Coeff: 0.130).

End-of-life care

Mulqueen et al. [29] examined PwDs’ preferences for end-of life care. The respondents were n = 6 patients living in residential care facilities and their n = 6 caregivers in Ireland in 2017, and they used the nominal group technique to elicit preferences. There were considerable differences between patients’ and caregivers’ preferences. Comfortable environment (quiet and no worries) (rank 1) and family involvement (rank 2) were most important for the patients, followed by family surrounding (rank 3), communication (rank 4), and familiar staff (rank 5). However, communication (rank 1) and pain management (rank 2) were most important for caregivers. Advance care planning (rank 3), family involvement (rank 4), and knowledge of care staff (rank 5) played a minor role for the caregivers.

Dening et al. [8] used the nominal group technique to analyze the preferences of PwD and their caregivers and the influence of the caregiver on the preferences of the PwD in 2012. The responding dyads attended a memory assessment service in the UK. Patient’s ranked family contact (rank 1), independence (rank 2), and safety (rank 3) as most important but comfortable environment as least important (rank 10). The preferences of the caregiver were elicited for themselves and depended on the experience the caregivers had made during the care process. Therefore, control (rank 1) and QoL (rank 2) were most important for the caregivers.

For patients and caregivers together, the place of care (rank 1) and not being a burden (rank 2) were ranked as most important, whereas legal preparations were unimportant for the dyads (rank 7).

DISCUSSION

An increasing number of patient preference studies have been published over the last two decades [31]. However, the number of published preference studies examining patients with dementia is still particularly rare. The studies analyzed herein covered a few domains of the entire treatment, care and support pathway as well as patient-related outcomes in dementia and differed mostly with regard to the methods used. In most cases, preferences were elicited using qualitative methods, such as interviews or focus group discussion, which precludes the calculation of coefficients or a numeric rank for attributes. Only a few studies used ranking methods. Some studies were based on hypothetical scenarios asking respondents to imagine having dementia and state the preferences that they thought they would have under this condition [21, 27]. Thus, only a few studies used stated preference methods, such as DCE, analytical hierarchy process or best-worst scaling, in patients actually suffering from dementia.

Elicited preferences regarding the diagnostic procedure revealed that accuracy, explanation by staff, comfort during the procedures, and being accompanied by a caregiver were most important for both PwD and caregivers. For patients alone, the accuracy is more important than the costs.

The most important patient-related outcomes were the QoL and self-efficiency of the PwD followed by the QoL and depression of the caregivers. However, there were significant differences in patient-reported outcomes between PwD and caregivers. Whereas depression was rather unimportant for PwD themselves, PwD rated caregiver’s QoL and anxiety as more important, and the self-efficacy and depression of the caregiver was less important compared to the rating the caregivers made themselves. Furthermore, preferences differ significantly between different studies, highlighting considerable uncertainty. This difference is highlighted by the studies of Barrios and Smith, reporting different rankings of attributes, although the list of attributes and methods were similar. Smith et al. elicited preferences of n = 88 PwD compared to n = 8 in Barrios et al. Therefore, the results of Smith may have a higher relevance and generalizability. However, both studies demonstrated that for PwD, their own QoL was most important. Although QoL is the most important and preferred outcome of PwD, less than 10% of intervention studies in dementia used this outcome as an indicator of a favorable intervention [26]. Most of the studies focused on clinical outcomes, such as cognition, mortality, morbidity or psychiatric behavior, and functional impairment. Hence, QoL should be considered in future studies. Some aspects that appear to be relevant for QoL are the preferred leisure activities. Participants mostly prefer activities that they practiced in their past lifetime [23]. Thus, caregivers report preferred activities independently of the presence of dementia. Furthermore, dementia-specific activities to improve PwD memory are least important. Perhaps PwD could feel exposed or ashamed while practicing those activities and therefore tend to avoid such activities. This possibility is highlighted by the study of O’Philibin [24]. Digital LSW is preferred in a one-to-one setting and would be used to show family and friends. In most cases, LSW could be used as a therapeutic aspect [32] but derived from their preferences; for PwD, it seems to be more active.

It could be argued that PwD living with a partner differ in their preferences compared with those living alone. The ranking of patients without a partner differed in cases of transportation service or in the living situation. PwD without a partner living close to the caregiver was ranked highest, verifying that PwD wanted to live as long as possible in their own homes [33]. However, there are also similarities between the groups, for example, the preference for receiving care by the same professional caregiver or nurse. There is some evidence that such approaches are safe, beneficial for improving patient and caregiver outcomes and cost-effective [34 –36]. Additionally, a short time on a waiting list for a care institution is of mediocre importance for both PwD living alone and PwD living with a partner, underlining that patients and caregivers are sensitized to the fact that serious health events could tremendously change the living and care situation of PwD.

Some preferences deal with the financial situation of PwD. A negative WTP indicates that respondents living with their partner prefer to spend money for extra home care rather than spending money for institutionalization. For PwD, distance to the health care provider, safety, and staffing were more important than the quality of care. This outcome seems to indicate that PwD prefer very close and intensive support from care professionals. Preferences were similar for end-of-life care. Interpersonal relations such as family contact or involvement as well as comfort and independence were most important. Patients and caregivers mutually influence each other if preferences are jointly elicited. Place of care, i.e., at home was most important, but the recommendation of family plays a minor part. Furthermore, preferences differ across countries. Respondents from Ireland most preferred a comfortable environment in end-of-life care, whereas this factor was least preferred in the UK. The studies of Dening and Mulqueen differ in some aspects, which could be due to the different sample sizes used in these studies. Nevertheless, family contact or involvement was important for respondents in both studies.

In addition to the methodological limitations of the quantification of patients’ preferences, the dependence of the sample size and the setting where preferences were elicited, this review revealed that preferences were not stable over time, e.g., before and after a disease-specific intervention or adverse life event [37]. Yu et al. highlighted the tremendous heterogeneity in the revealed preferences across different studies, demonstrating that preferences depend on the eliciting methods used [38]. Therefore, further research is needed to assess patient preferences in dementia diseases capturing several domains of diagnostics, treatment, care, and support as well as patient- and caregiver-related outcomes. In addition, it is of vital importance to assess the association between sociodemographic and clinical factors and different patterns of preferences. Knowing these preferences would allow us to more easily individualize treatment, care, and support in routine care to increase adherence to such interventions by a simultaneous improvement of PwD quality of life. These factors should be evaluated in future studies using designs with a high methodological quality and an adequate sample size. In addition, further research is needed to evaluate whether innovative and simpler quantitative measures, such as analytical hierarchy process, are more suitable for eliciting stated preferences in cognitively impaired patients or PwD.

Implications

The results of this systematic review do not exactly portray clinical practice, treatment, and care decisions. However, they should be considered for the development of clinical guidelines to optimize the treatment and care of PwD. Treatment and care decisions should correspond to the demonstrated PwDs’ and their caregivers’ preferences, such as minimal burdensome treatments for the patients, proximity to the caregiver or exactness in diagnostic tests, to improve the outcomes of both groups. Measuring individual preferences should be implemented in all phases of the disease. Considering these preferences will help to integrate PwD in decision-making processes concerning their therapy and care at all stages of the dementia disease. Participation from patients and their caregivers in the decision-making process would improve the success of treatments and other interventions [39]. Thus, preferences reported by patients with MCI and PwD can inform and guide the development of new medicines and therapies. As in most other diseases, in MCI and dementia, the patient is the best expert regarding his or her own health and wellbeing and should be involved as an active partner in all decisions regarding treatment, care and living situation.

Limitations

We were not able to perform a formal meta-analysis of patient preferences because the literature search yielded only a few studies with heterogeneous objectives, measures, and populations, causing an inconsistent reporting of patient preferences. Recall bias of PwD is a further limitation. It is unclear how long PwDs are able to make informed treatment decisions due to their declining cognitive abilities [40]. Consequently, it is important to explore the preferences of PwD who are still capable of judging options and express treatment and care preferences to their physicians and caregivers.

The sample sizes of the studies retrieved were small, varying from eight to 1,301. These small sample sizes limit the generalizability of the reported results. For statistically robust conclusions, the sample size of preference studies should be much higher. For example, for a patient preference study based on a discrete choice experiment, a sample size of at least n = 300 is recommended [41, 42]. In addition, an exclusion criterion for our review was the use of languages other than German or English. Hence, our strategy may have excluded preference studies in other languages.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/19-1299r2).

Footnotes

The supplementary material is available in the electronic version of this article: .

References

Prince

, Comas-Herrera

, Knapp

, Guerchet

, Karagiannidou

(2016) World Alzheimer Report 2016. Improving healthcare for people living with dementia: Coverage, quality and costs now and in the future. Alzheimer’s Disease International, London.

Bickel

Die Häufigkeit von Demenzerkrankungen. Deutsche Alzheimer Gesellschaft e.V. Selbsthilfe Demenz, https://www.deutsche-alzheimer.de/fileadmin/alz/pdf/factsheets/infoblatt1_haeufigkeit_demenzerkrankungen dalzg.pdf, Accessed April 9, 2019.

Hibbeler

(2015) Demenz – weltweit fast zehn Millionen Neuerkrankungen. Deutsches Ärzteblatt, Volume: Jg. 112, Section: Heft 37.

Prince

, Bryce

, Ferri

(2011) World Alzheimer Report 2011. The benefits of early diagnosis and intervention. Alzheimer’s Disease International, London.

Black

, Rabins

(2010) Quality of life in dementia: conceptual and practical issues. In Dementia, Burns A, Ames, O’Brien J, eds. Edward Arnold Publishers, London, pp. 293-304.

Robinson

, Tang

, Taylor

(2015) Dementia: timely diagnosis and early intervention. BMJ 350, h3029.

(2018) Pflegestatistik Pflege im Rahmen der Pflegeversicherung Deutschlandergebnisse 2017. Statistisches Bundesamt.

Dening

, Jones

, Sampson

(2013) Preferences for end-of-life care: a nominal group study of people with dementia and their family carers. Palliat Med 27, 409–417.

Michalowsky

, Thyrian

, Eichler

, Hertel

, Wucherer

, Flessa

, Hoffmann

(2016) Economic analysis of formal care, informal care, and productivity losses in primary care patients who screened positive for dementia in Germany. J Alzheimers Dis 50, 47–59.

10.

McLennon

, Habermann

, Davis

(2010) Deciding to institutionalize: why do family members cease caregiving at home? J Neurosci Nurs 42, 95–103.

11.

Wolters

, Tinga

, Dhana

, Koudstaal

, Hofman

, Bos

, Franco

, Ikram

(2019) Life expectancy with and without dementia: a population-based study of dementia burden and preventive potential. Am J Epidemiol 188, 372–381.

12.

Krajcovicova

, Marecek

, Mikl

, Rektorova

(2014) Disruption of resting functional connectivity in Alzheimer’s patients and at-risk subjects. Curr Neurol Neurosci Rep 14, 491.

13.

Bharath

, Joshi

, John

, Balachandar

, Sadanand

, Saini

, Kumar

, Varghese

(2017) A multimodal structural and functional neuroimaging study of amnestic mild cognitive impairment. Am J Geriatr Psychiatry 25, 158–169.

14.

Burggren

, Brown

(2014) Imaging markers of structural and functional brain changes that precede cognitive symptoms in risk for Alzheimer’s disease. Brain Imaging Behav 8, 251–261.

15.

Institute for Quality and Efficiency in Health Care (IQWiG), Allgemeine Methoden Version 5.0, Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen, https://www.iqwig.de/download/Allgemeine-Methoden_Version-5-0.pdf, Accessed June 27, 2019.

16.

Hummel

, Volz

, van Manen

, Danner

, Dintsios

, Ijzerman

, Gerber

(2012) Using the analytic hierarchy process to elicit patient preferences: prioritizing multiple outcome measures of antidepressant drug treatment. Patient 5, 225–237.

17.

Torrance

, Furlong

, Feeny

, Boyle

(1995) Multi-attribute preference functions. Health Utilities Index. Pharmacoeconomics 7, 503–520.

18.

Ryan

, Farrar

(2000) Using conjoint analysis to elicit preferences for health care. BMJ 320, 1530–1533.

19.

Joy

, Little

, Maruthur

, Purnell

, Bridges

(2013) Patient preferences for the treatment of type 2 diabetes: a scoping review. Pharmacoeconomics 31, 877–892.

20.

Bamford

, Olsen

, Davison

, Barnett

, Lloyd

, Williams

, Firbank

, Mason

, Donaldson

, O’Brien

(2016) Is there a preference for PET or SPECT brain imaging in diagnosing dementia? The views of people with dementia, carers, and healthy controls. Int Psychogeriatr 28, 123–131.

21.

Mühlbacher

, Johnson

, Yang

, Happich

, Belger

(2016) Do you want to hear the bad news? The value of diagnostic tests for Alzheimer’s disease. Value Health 19, 66–74.

22.

Janus

, van Manen

, van Til

, Zuidema

, IJzerman

(2017) Pharmacological and non-pharmacological treatment preferences of healthcare professionals and proxies for challenging behaviors in patients with dementia. Int Psychogeriatr 29, 1377–1389.

23.

Cohen-Mansfield

, Gavendo

, Blackburn

(2019) Activity preferences of persons with dementia: an examination of reports by formal and informal caregivers. Dementia (London) 18, 2036–2048.

24.

O’Philbin

, Woods

, Holmes

(2018) People with dementia and caregiver preferences for digital life story work service interventions. A discrete choice experiment and digital survey. Aging Ment Health 24, 353–361.

25.

Barrios

, Gonzalez

, Hanna

, Lunde

, Fields

, Locke

, Smith

(2016) Priority of treatment outcomes for caregivers and patients with mild cognitive impairment: preliminary analyses. Neurol Ther 5, 183–192.

26.

Smith

, Chandler

, Fields

, Aakre

, Locke

DEC

(2018) A survey of patient and partner outcome and treatment preferences in mild cognitive impairment. J Alzheimers Dis 63, 1459–1468.

27.

Nieboer

, Koolman

, Stolk

(2010) Preferences for long-term care services: willingness to pay estimates derived from a discrete choice experiment. Soc Sci Med 70, 1317–1325.

28.

Groenewoud

, Van Exel

, Bobinac

, Berg

, Huijsman

, Stolk

(2015) What influences patients’ decisions when choosing a health care provider? Measuring preferences of patients with knee arthrosis, chronic depression, or Alzheimer’s disease, using discrete choice experiments. Health Serv Res 50, 1941–1972.

29.

Mulqueen

, Coffey

(2017) Preferences of residents with dementia for end of life care. Nurs Older People 29, 26–30.

30.

Hort

, O’Brien

, Gainotti

, Pirttila

, Popescu

, Rektorova

, Sorbi

, Scheltens

(2010) EFNS guidelines for the diagnosis and management of Alzheimer’s disease. Eur J Neurol 17, 1236–1248.

31.

Johnson

, Lancsar

, Marshall

, Kilambi

, Mühlbacher

, Regier

, Bresnahan

, Kanninen

, Bridges

(2013) Constructing experimental designs for discrete-choice experiments: report of the ISPOR conjoint analysis experimental design good research practices task force. Value Health 16, 3–13.

32.

McKeown

, Clarke

, Repper

(2006) Life story work in health and social care: systematic literature review. J Adv Nurs 55, 237–247.

33.

e.V. DAG, Projekt Allein lebende Demenzkranke – Schulung in der Kommune., https://www.deutsche-alzheimer.de/ueber-uns/unsere-aufgaben-und-aktivitaeten/abgeschlossene-projekte/allein-lebende-demenzkranke.html#c3756, Accessed July 23, 2019.

34.

Reilly

, Miranda-Castillo

, Malouf

, Hoe

, Toot

, Challis

, Orrell

(2015) Case management approaches to home support for people with dementia. Cochrane Database Syst Rev 1, CD008345.

35.

Thyrian

, Hertel

, Wucherer

, Eichler

, Michalowsky

, Dreier-Wolfgramm

, Zwingmann

, Kilimann

, Teipel

, Hoffmann

(2017) Effectiveness and safety of dementia care management in primary care: a randomized clinical trial. JAMA Psychiatry 74, 996–1004.

36.

Michalowsky

, Xie

, Eichler

, Hertel

, Kaczynski

, Kilimann

, Teipel

, Wucherer

, Zwingmann

, Thyrian

, Hoffmann

(2019) Cost-effectiveness of a collaborative dementia care management-Results of a cluster-randomized controlled trial. Alzheimers Dement 15, 1296–1308.

37.

Straton

, Wang

, Meoni

, Ford

, Klag

, Casarett

, Gallo

(2004) Physical functioning, depression, and preferences for treatment at the end of life: the Johns Hopkins Precursors Study. J Am Geriatr Soc 52, 577–582.

38.

, Enkh-Amgalan

, Zorigt

(2017) Methods to perform systematic reviews of patient preferences: a literature survey. BMC Med Res Methodol 17, 166.

39.

Gorenoi

, Schönermark

, Hagen

(2007) Maßnahmen zur Verbesserung der Compliance bzw. Adherence in der Arzneimitteltherapie mit Hinblick auf Therapieerfolg [Measures to improve compliance and adherence in drug therapy with regard to therapy success]. Schriftenreihe Health Technology Assessment (HTA) in der Bundesrepublik Deutschland,

40.

Trachsel

, Mitchell

, Biller-Andorno

Nikola

(2013) Decision-making incapacity at the end of life: conceptual and ethical challenges. Bioethica Forum 6, 26–30.

41.

Orme

(2006) Getting started with conjoint analysis: strategies for product design and pricing research. Research Publishers, LLC.

42.

Bridges

, Hauber

, Marshall

, Lloyd

, Prosser

, Regier

, Johnson

, Mauskopf

(2011) Conjoint analysis applications in health—a checklist: a report of the ISPOR good research practices for conjoint analysis task force. Value Health 14, 403–413.