Abstract
Objectives:
Persons with progressive cognitive impairment (CI) increasingly rely on surrogate decision-makers for everyday activities. Yet, little is known about changes in everyday preferences over time or about concordance between persons with CI and their care partners regarding longitudinal changes.
Methods:
The sample included 48 dyads of persons with CI (Clinical Dementia Rating Scale score ≥ 0.5) and their care partners. The Preferences for Everyday Living Inventory was used to assess importance of preferences among persons with CI at baseline and follow-up (mean 486 days). Care partners separately completed concurrent proxy assessments. Mixed random and fixed effects longitudinal models were used to evaluate changes in ratings and concordance levels between persons with CI and care partners.
Results:
There were significant gender differences regarding importance ratings of “autonomous choice” and “social engagement” preferences over time: women with CI rated these preferences as more important across time as a whole. Higher levels of neuropsychiatric symptoms were associated with less importance of “social engagement” preferences across time as a whole for persons with CI and a more negative discrepancy between persons with CI and care partner proxy assessments as time went on.
Conclusion:
This study yields new insights into predictors of longitudinal change in everyday preferences among persons with CI and their care partners. Although preferences were largely stable over time, there is increasing support for the relationship between differences in “social engagement” preferences and neuropsychiatric symptoms, which may have implications for monitoring and/or treatment in the context of cognitive impairment.
Keywords: decision-making, ethics, longitudinal change
Objectives
In approaching surrogate decision-making for persons unable to make decisions on their own, different decision-making standards are available to proxies: a person’s known wishes if available through previous discussions or advance directives; a substituted judgment based on a person’s previously held values and preferences if explicit wishes are unknown; a best interests approach if previous wishes, preferences, and values are not sufficiently known for the decision at hand.1 Ethics and practices regarding formulation, prioritization and application of decision-making standards may vary by culture and country. In the United States, when established explicit wishes and preferences are unknown, the legally recommended hierarchy has historically favored a preference for a substituted judgment if possible, and then a best interests approach if a substituted judgment is not possible.2
Given the pervasive cognitive impairment inherent in major neurocognitive disorders, surrogate decision-making becomes increasingly important in all aspects of life including clinical,3 research,4 and lifestyle decisions.5 The application of a substituted judgment standard in surrogate decision-making for a person with cognitive impairment allows autonomy to be preserved and negotiated into future decision-making for instances where a person is deemed incapable of making a decision at hand.3 The fidelity of a substituted judgment relies then on an accurate proxy understanding of a person’s previously held values and preferences as well as stability of those values and preferences over time.
Discrepancy between ratings of persons with cognitive impairment (CI) and proxy assessments by surrogate decision-makers may be common in diverse settings, including care-related preferences,6 quality of life assessments,7 and psychosocial preferences.8 We previously reported significant discrepancy in ratings of importance of “social engagement” preferences (e.g., regular contact with family, meeting new people, volunteering) between persons with CI and proxy assessments by care partners.9 A crucial aspect of these studies is the assumption that persons with CI can reliably articulate their values and preferences, which has been validated in other work looking at values and preferences, even among those with more advanced dementia.10
With respect to stability of values and preferences over time for persons with CI, it appears that there is at least moderate stability over one year in the assessment of everyday psychosocial preferences, although higher levels of cognitive impairment were associated with somewhat less stability (e.g., intraclass correlation for persons with Clinical Dementia Rating (CDR) global score of 1 = 0.49 and for CDR global score of 0.5 = 0.62).8 Proxy assessments of preferences and values appear to have some variability over time, however. For instance, care partner proxy ratings of the importance of care-related values and preferences for persons with CI significantly decreased over time.11 In looking at proxy assessment of changes in psychological suffering and quality of life for a person with dementia across time, changes in mental and physical health of the caregiver were noted to predict bias in proxy assessments.7 There remains, however, a paucity of data about the concordance of proxy ratings of the everyday psychosocial preferences for persons with CI over time and the factors driving changes in discrepancy over time.
The aims of the current study are: 1) to measure everyday preferences of persons with CI over time, 2) to assess changes in level of discrepancy in importance ratings over time for persons with CI and proxy ratings of care partners and 3) to address cognitive impairment severity, demographic factors, and neuropsychiatric symptom burden as predictors of longitudinal changes of both importance ratings of preferences for persons with CI and level of discrepancy with proxy ratings of care partners. Our hypothesis is that higher levels of neuropsychiatric symptoms and more severe cognitive impairment would be drivers of more significant changes across time.
Methods
Study Sample.
The sample included 48 dyads comprising persons with CI (Clinical Dementia Rating (CDR) global score ≥ 0.5) and their care partners. Persons with CI were recruited from the Massachusetts Alzheimer’s Disease Research Center (MADRC) Clinical Core longitudinal cohort; the recruitment scheme was described previously.9 All cohort participants have care partners who provide collateral information as part of MADRC procedures; in the MADRC, there are no specific requirements for frequency of contact between the cohort participant and care partner. All cohort participants with CDR global score ≥ 0.5 (i.e., persons with either mild cognitive impairment (MCI) or physician-determined clinically meaningful cognitive impairment without meeting formal MCI criteria) who presented to the MADRC with a care partner over a 24-month period from 2017-2019 were invited to participate in baseline assessments; those who agreed to participate were enrolled. Compared to those who chose to enroll, those who declined tended to be younger (average years of age 72 vs. 78), have higher levels of cognitive impairment (average global CDR score 0.94 vs. 0.65), and have more education (average years 17.6 vs 16.1).
Follow-up assessments were obtained over a 17-month period from 2018-2020. Research visits were postponed from March 2020 due to the COVID-19 pandemic. As of that time, approximately 47% of eligible cohort participants completing baseline assessments had completed the follow-up assessments. There were 15 cohort participants due for follow-up visits that were not able to be completed due to the COVID-19 pandemic. Additional reasons for not completing repeat measures included care partner not present at the study visit, cancelled appointment, cognitive impairment too severe to complete study measures, not enough time to complete the measures, refusal to participate, and death of the cohort participant prior to the follow-up visit. It appeared that those cohort participants who were eligible for follow-up but did not complete follow-up measures had higher levels of cognitive impairment (average global CDR score 0.72 vs. 0.57) than those cohort participants who completed the follow-up measures. Cohort participants provided written informed consent, and the Partners HealthCare Institutional Review Board reviewed and approved this study.
Measures – Predictors/Covariates.
MADRC cohort participants and their care partners complete annual evaluations that follow the Alzheimer’s Disease Centers Uniform Data Set (UDS) protocol.12 Briefly, the evaluation features a standard battery for cohort participants including ascertainment of demographic information, the CDR,13 behavioral and symptom measures (e.g., Neuropsychiatric Inventory brief Questionnaire [NPI-Q]14 and 15-item Geriatric Depression Scale [GDS-15]15), cognitive assessments (e.g., in memory, attention, executive function, fluency, Montreal Cognitive Assessment [MoCA]16, etc.), as well as a medical history and neurological examination.
Measures – Dependent Variables/Preferences Assessment.
Persons with CI completed the Preferences for Everyday Living Inventory (PELI), a validated preferences assessment tool for older adults with current preferences rated on a 4-point Likert scale ranging from 1 (very important) to 4 (not important at all).17 Care partners also completed proxy PELI assessments from the perspective of the persons with CI (i.e., as if acting as a surrogate decision-maker). PELI data were collected longitudinally at baseline and follow-up visits, roughly in line with annual assessments. Our previous work used exploratory factor analysis to derive potential latent sub-scale “domains” embedded within the 55-item PELI, and identified four factors, which were labeled: “autonomous choice” (involving 14 items), “social engagement” (15 items), “personal growth” (14 items), and “keeping a routine” (6 items); six items did not load sufficiently on any of these fours factors and were not included in further analyses.9
To account for missing data and differences in the number of items across the domains, a mean domain score was calculated for each participant per domain; this score was calculated as the average of the non-missing importance ratings of only those PELI items contributing to that respective domain. Missing preference data was only 1.5% of the total items that could have been completed (137/9,408). For this analysis, our assumption is that the numeric scales across these four domains are conceptually equivalent, i.e., a given numeric score on one domain measures approximately the same strength of preference for that domain as the same numeric score for the other domains. We regard this assumption as reasonable given the identically worded and scaled Likert-style response options for all PELI items. The “preference discrepancy score,” indexing disagreement between persons with CI and their respective care partners for a given domain, was computed by subtracting the value for the care partner from the value for the corresponding person with CI.
Statistical Analyses.
To account for correlation between domain scores, a within-subjects analysis of variance (ANOVA) was used to test for significant differences in the mean domain scores among the four domains of the PELI, separately at baseline and at follow-up, for the persons with CI. Tukey-Kramer adjusted post hoc tests were used to assess the significance of differences between pairs of domains, following a significant omnibus main effect of the four category domain factors. To assess for discrepancy in preferences assessment between persons with CI and care partners, a paired t-test comparing preferences of persons with CI versus the corresponding proxy assessments by their respective care partners on a given domain was performed separately at baseline and at follow-up. Because of the scaling of the PELI, lower scores denote assignment of higher importance and higher scores denote assignment of lower importance.
Mixed random and fixed effects longitudinal analyses were run across time in the study for each of the four PELI domains, each in a separate analysis, employing a backward elimination algorithm (p < 0.05 cut off) on an initial pool of fixed predictors and variances/covariances of random terms. Analyses were run to assess predictors of change in preference scores for persons with CI, and separately to assess predictors of changes in discrepancy scores between persons with CI and proxy ratings of care partners. By convention, during backward elimination, non-significant terms were retained in the model if higher order terms subsuming them (in our case, interactions involving them) were still in the model. The time predictor was the linear component of days in the study from baseline to follow-up assessment.
Fixed terms in the analysis of preference scores for persons with CI were: GDS-15 score, NPI-Q score, CDR global score, age, gender, years of education, marital status, all at baseline for the person with CI, and the interaction of each of these predictors with time-in-study. Given low counts for marital status categories besides married/domestic partner, marital status at baseline was collapsed to two categories: 1) “married/domestic partner” and 2) “widowed,” “divorced,” “separated,” or “never married.” Fixed terms in the analysis of discrepancy between persons with CI and proxy care partner ratings were: baseline GDS-15 score (person with CI), NPI-Q score (person with CI), CDR global score (person with CI), age and gender (for both person with CI and care partner), and relationship status between the person with CI and care partner. Given low counts for relationship categories besides “spouse,” relationship status at baseline was collapsed to two categories: 1) “spouse” and 2) “child,” “friend,” or “sibling.” All statistical analyses were performed using SAS Version 9.4 (SAS, Cary, NC, USA).
Results
Participant characteristics are shown in Tables 1a (persons with CI) and 1b (care partners). The sample of persons with CI included just over half men (54%) and was predominately white (92%), highly educated (mean 16 years of education) and married (83%). Among persons with CI, approximately 19% had dementia, 31% had MCI, and 50% fell into the ADRC program’s “cognitively impaired, not MCI” category (i.e., global CDR score = 0.5 but without evidence of objective deficits on neuropsychological testing);18 approximately 85% had a global CDR score of 0.5 at baseline; mean CDR-Sum of Boxes score at baseline was 2.3; mean MoCA at baseline was 23.8 points. The mean time between baseline and follow-up assessment was 486 days (standard deviation 107 days, range 318-814 days). Most care partners were women (71%), white (92%), highly educated (mean 16 years of education), spouses/partners (75%), living with the person with CI (75%) and tended to have had long relationships with the person with CI (average of 46 years of having known the participant). If not living with the person with CI, all care partners had contact at least 3x/month.
Table 1a.
Description of persons with cognitive impairment: demographic and clinical characteristics at baseline and follow-up. All scores based on n = 48, unless otherwise noted.
| Baseline | Follow-up | |
|---|---|---|
| Age (mean years ([Standard Deviation, S.D.]) | 78.4 (7.2) | -- |
| Gender (women) (n [%]) | 22 (45.8%) | -- |
| Education (mean years [S.D.]) | 16.0 (2.9) | -- |
| Race (n [%]) | ||
| White | 44 (91.7%) | -- |
| Black or African American | 2 (4.2%) | -- |
| Asian | 1 (2.1%) | -- |
| American Indian or Alaska Native | 1 (2.1%) | -- |
| Marital Status (n [%]) | ||
| Married/Domestic Partner | 40 (83.3%) | -- |
| Widowed | 5 (10.4%) | -- |
| Divorced | 2 (4.2%) | -- |
| Separated | 1 (2.1%) | -- |
| Dementia [n (%)] | 9 (18.8%) | -- |
| Amnestic (including Alzheimer’s disease) | 9 (100%) | -- |
| Mild Cognitive Impairment (MCI) [n (%)] | 15 (31.3%) | -- |
| Amnestic | 10 (66.7%) | -- |
| Non-amnestic | 5 (33.3%) | -- |
| Cognitively impaired, not MCI [n (%)] | 24 (50%) | -- |
| Global Clinical Dementia Rating (CDR) Score (n [%]) | (n = 48) | (n = 45) |
| 0 | -- | 2 (4.4%) |
| 0.5 | 41 (85.4%) | 34 (75.6%) |
| 1 | 7 (14.6%) | 8 (17.8%) |
| 2 | -- | 1 (2.2%) |
| CDR-Sum of Boxes Score (mean ([S.D.]) | 2.3 (1.6) | 2.5 (2.4)† |
| Neuropsychiatric Inventory brief Questionnaire (NPI-Q) Score (mean ([S.D.]) | 2.7 (3.1)† | 2.9 (3.5)‡ |
| Montreal Cognitive Assessment (MoCA) Score (mean ([S.D.]) | 23.8 (4.6) | 24.6 (4.3)§ |
| Geriatric Depression Scale, 15-item (GDS-15) Score (mean ([S.D.]) | 2.1 (2.8)¶ | 2.3 (2.2)¶ |
| Days to Follow-Up (mean [S.D.]) | -- | 485.6 (106.5) |
Note. Percentages might not add up to 100% due to rounding.
n = 45.
n = 41.
n = 42.
n = 43.
Table 1b.
Description of care partners: demographic characteristics at baseline. All scores based on n = 48, unless otherwise noted.
| Baseline | |
|---|---|
| Age (mean ([Standard Deviation, S.D.]) | 71.4 (13.6)† |
| Gender (women) (n [%]) | 34 (70.8%) |
| Education (mean ([Standard Deviation, S.D.]) | 15.9 (2.3)‡ |
| Race (n [%])§ | |
| White | 35 (92.1%) |
| Black or African American | 2 (5.3%) |
| Asian | 1 (2.6%) |
| Relationship to participant (n [%]) | |
| Spouse/Partner/Companion | 36 (75%) |
| Child | 8 (16.7%) |
| Friend | 2 (4.2%) |
| Sibling | 2 (4.2%) |
| Years known participant (mean ([S.D.]) | 45.6 (17.4)¶ |
| Lives with participant (yes) (n [%]) | 36 (75%) |
| If no, frequency of visits (n [%])# | |
| At least 3x/week | 5 (41.7%) |
| Weekly | 4 (33.3%) |
| At least 3x/month | 2 (16.7%) |
| Monthly | 1 (8.3%) |
| If no, frequency of telephone calls (n [%])# | |
| Daily | 5 (41.7%) |
| At least 3x/week | 2 (16.7%) |
| Weekly | 3 (25.0%) |
| At least 3x/month | 2 (16.7%) |
Note. Percentages might not add up to 100% due to rounding.
n = 46.
n = 35.
n = 38.
n = 44.
n = 12.
At baseline and follow-up assessments, the preferences domain that was ranked as most important among persons with CI (i.e., lowest mean domain score), was “social engagement,” followed by “autonomous choice,” “personal growth,” and “keeping a routine,” respectively (Tables 2a and 2b). There were significant differences in mean domain scores of importance by within-subjects ANOVA at baseline and follow-up (for baseline, domain main effect, F(3,47) = 12.06, p ≤ 0.0001; for follow-up, F(3,47) = 8.55, p = 0.0001). Post hoc testing (Tukey-Kramer) revealed statistically significant (p < 0.01) differences between pairwise domain comparisons “personal growth”/”social engagement,” “autonomous choice”/”keeping a routine,” and “social engagement”/“keeping a routine,” at baseline and follow-up (Tables 2a and 2b).
Table 2a.
Results of within-subjects analysis of variance (ANOVA) comparing the mean domain scores for the four domains of the Preferences for Everyday Living Inventory (PELI) at baseline for persons with cognitive impairment. F(3,47) = 12.06, p ≤ 0.0001. N = 48.
| Domain | Mean (S.D.)† | ANOVA: Mean difference, t-value, Tukey-Kramer adjusted p-value | |||
|---|---|---|---|---|---|
| Autonomous Choice | Social Engagement | Personal Growth | Keeping a Routine | ||
| Autonomous Choice | 2.13 (0.53) | -- | 0.18, 2.10, 0.17 | −0.07, −0.90, 0.81 | −0.27, −4.68, 0.0001 |
| Social Engagement | 1.95 (0.39) | -- | −0.25, −3.42, 0.007 | −0.45, −4.65, 0.0001 | |
| Personal Growth | 2.20 (0.40) | -- | −0.20, −2.26, 0.12 | ||
| Keeping a Routine | 2.40 (0.65) | -- | |||
Mean domain score for each domain is calculated as the average of the importance ratings of PELI items contributing to that domain. S.D. = Standard Deviation.
Note. PELI items are ranked on a 4-point Likert scale with 1 = very important, 2 = somewhat important, 3 = not very important, and 4 = not at all important.
Table 2b.
Results of within-subjects analysis of variance (ANOVA) comparing the mean domain scores for the four domains of the Preferences for Everyday Living Inventory (PELI) at follow-up for persons with cognitive impairment. F(3,47) = 8.55, p = 0.0001. N = 48.
| Domain | Mean (S.D.)† | ANOVA: Mean difference, t-value, Tukey-Kramer adjusted p-value | |||
|---|---|---|---|---|---|
| Autonomous Choice | Social Engagement | Personal Growth | Keeping a Routine | ||
| Autonomous Choice | 2.04 (0.44) | -- | 0.13, 2.02, 0.19 | −0.20, −2.63, 0.05 | −0.25, −3.51, 0.005 |
| Social Engagement | 1.91 (0.39) | -- | −0.33, −4.44, 0.0003 | −0.38, −4.37, 0.0004 | |
| Personal Growth | 2.24 (0.42) | -- | −0.05, −0.73, 0.89 | ||
| Keeping a Routine | 2.29 (0.56) | -- | |||
Mean domain score for each domain is calculated as the average of the importance ratings of PELI items contributing to that domain. S.D. = Standard Deviation.
Note. PELI items are ranked on a 4-point Likert scale with 1 = very important, 2 = somewhat important, 3 = not very important, and 4 = not at all important.
Comparing ratings from persons with CI and proxy ratings from care partners, there was significant discrepancy observed for preferences related to “social engagement” at baseline and follow-up (paired t-test, p < 0.05; mean difference −0.15 and −0.24 respectively) (Table 3). This negative effect estimate indicates statistically significant underestimation of the importance of “social engagement” preferences by proxy care partner ratings compared to those of persons with CI. There were no significant differences between ratings of persons with CI and care partner ratings for any of the other three domains at baseline or at follow-up (Table 3).
Table 3.
Mean differences (standard deviation, S.D.) in preference domain scores between persons with cognitive impairment and proxy ratings of care partners for each domain of the Preferences for Everyday Living Inventory (PELI) at baseline and follow-up (mean days to follow-up = 485.6).†
| Domain | Baseline | Follow-Up | ||||
|---|---|---|---|---|---|---|
| Mean Difference (S.D.) | t-statistic | p-value‡ | Mean Difference (S.D.) | t-statistic | p-value‡ | |
| Autonomous Choice | 0.05 (0.57) | 0.58 | 0.57 | −0.06 (0.45) | −0.87 | 0.39 |
| Social Engagement | −0.15 (0.46) | −2.20 | 0.03 | −0.24 (0.46) | −3.60 | 0.0008 |
| Personal Growth | −0.05 (0.45) | −0.82 | 0.41 | −0.04 (0.42) | −0.60 | 0.55 |
| Keeping a Routine | 0.03 (0.69) | 0.31 | 0.76 | 0.01 (0.50) | 0.14 | 0.89 |
Domain score for each domain is calculated as the average of the importance ratings of PELI items contributing to that domain.
Comparisons were made with the paired sample t-test. Degrees of freedom = 47.
In the longitudinal analyses of change in importance ratings by persons with CI over time, the final model for “autonomous choice” showed women with CI rated these preferences as more important across time as a whole (i.e., across baseline & follow-up as an average), compared to men with CI (Table 4). There was a similar pattern in the final model for “social engagement” as well. Additionally, with the final model for “social engagement” preferences, higher levels of depressive symptoms for the person with CI as measured by the GDS-15 at baseline were associated with less importance of “social engagement” preferences across time as a whole. The analogous analyses for the other two domains showed no significant results.
Table 4.
Results of longitudinal mixed effects model for importance scores by persons with cognitive impairment for “autonomous choice” and “social engagement” domains of the Preferences for Everyday Living Inventory (PELI), measured at baseline and follow-up (mean days to follow-up = 485.6), showing fixed effect predictors retained in the final model.
| Autonomous Choice | Social Engagement | |||||
|---|---|---|---|---|---|---|
| Predictor | Regression Coefficient† | Standard Error | p-value | Regression Coefficient† | Standard Error | p-value |
| GDS-15 Score (Points) | -- | -- | -- | 0.04 | 0.02 | 0.04 |
| Participant Gender | −0.26 ‡ | 0.12 | 0.04 | −0.25 ‡ | 0.10 | 0.02 |
The regression coefficient is the unstandardized partial regression coefficient.
The regression coefficient is equivalent to the difference in adjusted means between women minus men.
In the longitudinal analyses of change in discrepancy of preference ratings between persons with CI and proxy care partner ratings over time, the final model for “social engagement” showed several significant effects (Table 5). Higher levels of neuropsychiatric symptoms for the person with CI measured by the NPI-Q at baseline were associated with a more negative discrepancy across time as a whole (i.e., toward underestimation); the interaction of depressive symptoms (as measured by the GDS-15 for the person with CI) and time (days) was a significant predictor such that a higher baseline GDS-15 score for the person with CI was associated with a steeper negative discrepancy decline across time (i.e., more depressive symptoms for the person with CI at baseline was associated with increasing underestimation by the care partner as time went on). The interaction of care partner gender and time (days) was also a significant predictor, such that women care partners showed a steeper negative discrepancy (i.e., toward increasing underestimation) as time went on compared to men care partners. The analogous analyses for the other three domains showed no significant results. Residuals from all longitudinal models reasonably conformed to normality assumptions of the significance tests.
Table 5.
Results of longitudinal mixed effects model for difference in the importance scores for the “social engagement” domain of the Preferences for Everyday Living Inventory (PELI) between persons with cognitive impairment and proxy ratings of care partners, measured at baseline and follow-up (mean days to follow-up = 485.6), showing fixed effect predictors retained in the final model.
| Predictor | Regression Coefficient† | Standard Error | p-value |
|---|---|---|---|
| Days between assessments (Time) | 0.0006 | 0.0002 | 0.20 |
| GDS-15 Score (Points) | 0.03 | 0.03 | 0.19 |
| GDS-15 Score interacting with Time | −0.0001 | 0.0001 | 0.0027 |
| NPI-Q Score (Points) | −0.05 | 0.02 | 0.01 |
| Care Partner Gender | 0.20‡ | 0.15 | 0.19 |
| Care Partner Gender interacting with Time | −0.0007 | 0.0003 | 0.02 |
The regression coefficient is the unstandardized partial regression coefficient.
The regression coefficient is equivalent to the difference in adjusted means between women minus men at baseline when days = zero.
Conclusions
This study expands on characterization of everyday experiences of older adults with CI and presents novel findings regarding longitudinal changes in preferences assessments and discrepancy with proxy ratings of care partners. These data suggest that for persons with CI, everyday preferences are largely stable over time, although gender and depressive symptoms of the person with CI did appear to correlate significantly with importance ratings for some of the domains over time as a whole. Further, there appeared to be relative consistency in the level of/lack of discrepancy in assessments between persons with CI and care partners as proxies, with neuropsychiatric symptoms of the person with CI significantly associated with discrepancy scores across time as a whole and depressive symptoms of the person with CI and care partner gender significantly associated with differential changes in level of discrepancy as time went on for “social engagement” preferences.
There has been previous work exploring changes in everyday preferences over various time points, in various settings, using various analyses and various versions of the PELI. For example in the nursing home environment, one study found an association of higher consistency in importance ratings over a one-week period with women and higher cognitive scores;19 another study found at least 70% stability in the majority of preference ratings over a three-month period.20 Further, there appeared to be similar consistency in PELI importance ratings over a one-week period between younger university students and older nursing home residents,21 although a different study found PELI importance ratings to be significantly less consistent over a one-year period for people with a CDR global score of 1 relative to those with a CDR global score of 0.5.8
Our previous work identified “social engagement” preferences to be rated as the most important for a sample of community-dwelling older adults with CI, with women tending to have higher importance ratings of “social engagement” preferences, while depressive symptoms were significantly associated with lower importance.22 The results of the current study build on this foundation by highlighting that not only are depressive symptoms and gender associated with baseline importance ratings of “social engagement” but these factors are also associated with differences in importance ratings for “social engagement” across time as a whole: women with CI tended to have higher importance ratings for “social engagement” over time, while persons with CI and a higher level of depressive symptoms at baseline had lower ratings for “social engagement” over time. Our study did not find a significant association between level of cognitive impairment and change in importance ratings over time; however, there may be insufficient variability in our sample to detect such differences (e.g., 85% of persons with CI had a CDR global score of 0.5 at baseline) as well as non-enrollment of/attrition for those with higher levels of CI.
Regarding discrepancy between persons with CI and proxy ratings over time, there appears to be limited investigation thus far; for example, one study showed proxy ratings by caregivers of perceived importance for care-related values significantly decreased over time and another study showed changes in mental and physical health of caregivers predicted bias in proxy assessments of changes in psychological suffering and quality of life for persons with dementia across time.7,11 Our previous work showed significant discrepancy between persons with CI and proxy ratings by care partners in the domain of “social engagement” with depressive symptoms of the person with CI and care partner gender significantly associated with levels of discrepancy.9 The results of the current study expand on these findings by showing an ongoing discrepancy in “social engagement” preferences over time. Further, it appears that care partner gender as well as depressive symptoms for the person with CI are associated with not only baseline discrepancy in “social engagement” preferences but also differential changes in the level of that discrepancy over time. Care partner characteristics such as levels of distress and depression were not included in the current study.
It appears that higher level of depressive symptoms for the person with CI at baseline are related to less perceived importance of “social engagement” preferences over time. A cautious interpretation of these findings is that depressive symptoms may lead to less emphasis over time on social engagement-related activities, possibly because of the anhedonic and amotivational characteristics of depression or lower reward derived from social activity in depression. What is not clear from these data, however, is whether a decreased importance of “social engagement” ratings does in fact correlate with an overall decreased social engagement per se. For instance, previous work using the PELI in the nursing home environment found varying levels of congruence between importance ratings for preferences related to recreation and leisure activities and actual participation in those activities.23
Changes over time in individual core human values appear to be mediated by both age and gender differences.24 For instance, older women were found to have more variability in ranking of values related to “conservation” (e.g., preserving traditional practices) over a 3-year period compared with older men.24 One thought with respect to the results of the current study is that differences over time in ratings of importance of “social engagement” and “autonomous choice” preferences by persons with CI may reflect an inherent difference between men and women in how core values evolve over time. Further, this dynamic relationship between gender, values, and time may extend to the association elucidated in the current study between the interaction of care partner gender and time with discrepancy in assessment of “social engagement” preferences over time.
Higher neuropsychiatric symptom burden for the person with CI as measured by the NPI-Q score showed a trend toward greater underestimation of importance of “social engagement” preferences by care partners over time as a whole. A preliminary explanation of these findings may include factors for both persons with CI and care partners. For instance, it may be that greater neuropsychiatric symptom intensity impacts ratings of “social engagement” preferences over time for persons with CI, which is then not appreciated by care partners in their proxy ratings. Further, it may be that greater neuropsychiatric symptom intensity for persons with CI impacts care partners’ ability to accurately predict ratings. For example, previous work in surrogate decision-making has suggested that environments that tax the cognitive resources of a proxy create more bias in surrogate assessments.25 It is not clear, however, whether a baseline level of discrepancy is associated with changes in neuropsychiatric symptoms over time, which may present a clearer target for therapeutic intervention (e.g., efforts to resolve baseline discrepancy may lessen burden of depressive symptoms over time).
Strengths of this study included comprehensive assessment of everyday preferences in a well-characterized sample, longitudinal analyses, and use of gold-standard cognitive and neurobehavioral assessments. Study limitations included a relatively small sample size, limited racial diversity, and high levels of education compared to older adults nationally. Overall, this work adds to the growing body of support for the relationship between ratings of the perceived importance of “social engagement” preferences with neuropsychiatric symptom burden and demographic factors, namely the gender of persons with CI and that of their care partners. These results also suggest that addressing “social engagement” may have implications for the monitoring and/or treatment of neuropsychiatric symptoms, particularly, depression in the context of cognitive impairment.
Key Points:
1) The primary question addressed by this study is how do everyday preferences change over time for persons with cognitive impairment and how does concordance between persons with cognitive impairment and their care partners change over time in assessment of everyday preferences?
2) In this longitudinal study, there were significant gender differences regarding importance ratings of “autonomous choice” and “social engagement” preferences over time. Higher levels of neuropsychiatric symptoms were associated with less importance of “social engagement” preferences across time as a whole for persons with cognitive impairment and a more negative discrepancy between persons with cognitive impairment and care partner proxy assessments as time went on.
3) This study yields new insights into predictors of longitudinal change in everyday preferences among persons with cognitive impairment and their care partners and lends increasing support to the relationship between differences in “social engagement” preferences and neuropsychiatric symptoms, which may have implications for monitoring and/or treatment in the context of cognitive impairment.
Acknowledgements
Grateful acknowledgement is made of the contributions of Emily Merrill in data collection and processing as well as the MADRC participants and study staff. This study was supported by the Dupont Warren Fellowship, Harvard Medical School, Department of Psychiatry (JMW); Livingston Award, Harvard Medical School, Department of Psychiatry (JMW); Alzheimer’s Association Clinician Scientist Fellowship (JMW); National Institutes of Health Loan Repayment Award, L30 AG060475 (JMW); grant P30 AG062421 from the National Institutes of Health (BTH). This work was previously presented at the Alzheimer’s Association International Conference 2020 (Virtual Event).
This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health Award UL 1TR002541) and financial contributions from Harvard University and its affiliated academic healthcare centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health.
Footnotes
No Disclosures to Report.
References
- 1.Berger JT, DeRenzo EG, Schwartz J. Surrogate decision making: reconciling ethical theory and clinical practice. Ann Intern Med. 2008;149(1):48–53. http://www.ncbi.nlm.nih.gov/pubmed/18591637. [DOI] [PubMed] [Google Scholar]
- 2.Saks ER, Dunn LB, Wimer J, Gonzales M, Kim S. Proxy consent to research: the legal landscape. Yale J Health Policy Law Ethics. 2008;8(1):37–92. http://www.ncbi.nlm.nih.gov/pubmed/18564493. [PubMed] [Google Scholar]
- 3.Wilkins JM. Narrative Interest Standard: A Novel Approach to Surrogate Decision-Making for People With Dementia. Gerontologist. 2018;58(6):1016–1020. doi: 10.1093/geront/gnx107 [DOI] [PubMed] [Google Scholar]
- 4.Wilkins JM, Forester BP. Informed consent, therapeutic misconception, and clinical trials for Alzheimer’s disease. Int J Geriatr Psychiatry. 2020;35(5):430–435. doi: 10.1002/gps.5262 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wilkins JM. More Than Capacity: Alternatives for Sexual Decision Making for Individuals With Dementia. Gerontologist. 2015;55(5):716–723. doi: 10.1093/geront/gnv098 [DOI] [PubMed] [Google Scholar]
- 6.Reamy AM, Kim K, Zarit SH, Whitlatch CJ. Understanding discrepancy in perceptions of values: Individuals with mild to moderate dementia and their family caregivers. Gerontologist. 2011;51(4):473–483. doi: 10.1093/geront/gnr010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Schulz R, Cook TB, Beach SR, et al. Magnitude and causes of bias among family caregivers rating Alzheimer disease patients. Am J Geriatr Psychiatry. 2013;21(1):14–25. doi: 10.1016/j.jagp.2012.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Carpenter BD, Kissel EC, Lee MM. Preferences and life evaluations of older adults with and without dementia: reliability, stability, and proxy knowledge. Psychol Aging. 2007;22(3):650–655. doi: 10.1037/0882-7974.22.3.650 [DOI] [PubMed] [Google Scholar]
- 9.Wilkins JM, Locascio JJ, Gunther JM, et al. Differences in Assessment of Everyday Preferences Between People With Cognitive Impairment and Their Care Partners: The Role of Neuropsychiatric Symptoms. Am J Geriatr psychiatry. 2020;28(10):1070–1078. doi: 10.1016/j.jagp.2020.01.189 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Miller LM, Whitlatch CJ, Lyons KS. Shared decision-making in dementia: A review of patient and family carer involvement. Dementia. 2016;15(5):1141–1157. doi: 10.1177/1471301214555542 [DOI] [PubMed] [Google Scholar]
- 11.Reamy AM, Kim K, Zarit SH, Whitlatch CJ. Values and preferences of individuals with dementia: Perceptions of family caregivers over time. Gerontologist. 2013;53(2):293–302. doi: 10.1093/geront/gns078 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Weintraub S, Salmon D, Mercaldo N, et al. The Alzheimer’s Disease Centers’ Uniform Data Set (UDS): The neuropsychologic test battery. Alzheimer Dis Assoc Disord. 2009;23(2):91–101. doi: 10.1097/WAD.0b013e318191c7dd [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 1993;43(11):2412–2414. [DOI] [PubMed] [Google Scholar]
- 14.Kaufer DI, Cummings JL, Ketchel P, et al. Validation of the NPI-Q, a brief clinical form of the Neuropsychiatric Inventory. J Neuropsychiatry Clin Neurosci. 2000;12(2):233–239. doi: 10.1176/jnp.12.2.233 [DOI] [PubMed] [Google Scholar]
- 15.Yesavage JA, Sheikh JI. GERIATRIC DEPRESSION SCALE (GDS): Recent Evidence and Development of a Shorter Version. Clin Gerontol. 1986;5(1-2):165–173. doi: 10.1300/J018v05n01_09 [DOI] [Google Scholar]
- 16.Nasreddine ZS, Phillips NA, Bédirian V, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53(4):695–699. doi: 10.1111/j.1532-5415.2005.53221.x [DOI] [PubMed] [Google Scholar]
- 17.Van Haitsma K, Curyto K, Spector A, et al. The preferences for everyday living inventory: scale development and description of psychosocial preferences responses in community-dwelling elders. Gerontologist. 2013;53(4):582–595. doi: 10.1093/geront/gns102 [DOI] [PubMed] [Google Scholar]
- 18.Besser L, Kukull W, Knopman DS, et al. Version 3 of the National Alzheimer’s Coordinating Center’s Uniform Data Set. Alzheimer Dis Assoc Disord. 2018;32(4):351–358. doi: 10.1097/WAD.0000000000000279 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Curyto K, Dockler LM, Van Haitsma KS. Preference Consistency: Veteran and Non-Veteran Nursing Home Resident Self-Reported Preferences for Everyday Living. J Gerontol Nurs. 2020;46(3):15–25. doi: 10.3928/00989134-20200129-06 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Abbott KM, Heid AR, Kleban M, Rovine MJ, Van Haitsma K. The Change in Nursing Home Residents’ Preferences Over Time. J Am Med Dir Assoc. 2018;19(12):1092–1098. doi: 10.1016/j.jamda.2018.08.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Van Haitsma K, Abbott KM, Heid AR, et al. The consistency of self-reported preferences for everyday living: implications for person-centered care delivery. J Gerontol Nurs. 2014;40(10):34–46. doi: 10.3928/00989134-20140820-01 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wilkins JM, Locascio JJ, Gunther JM, et al. Predictors of the Importance of Everyday Preferences for Older Adults with Cognitive Impairment. Int psychogeriatrics. 2021;1–8. doi: 10.1017/S1041610220003956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Heid AR, Van Haitsma K, Kleban M, Rovine MJ, Abbott KM. Examining Clinical Predictors of Change in Recreational Preference Congruence Among Nursing Home Residents Over Time. J Appl Gerontol. 2017;36(11):1351–1369. doi: 10.1177/0733464815617288 [DOI] [PubMed] [Google Scholar]
- 24.Milfont TL, Milojev P, Sibley CG. Values Stability and Change in Adulthood: A 3-Year Longitudinal Study of Rank-Order Stability and Mean-Level Differences. Pers Soc Psychol Bull. 2016;42(5):572–588. doi: 10.1177/0146167216639245 [DOI] [PubMed] [Google Scholar]
- 25.Epley N, Keysar B, Van Boven L, Gilovich T. Perspective taking as egocentric anchoring and adjustment. J Pers Soc Psychol. 2004;87(3):327–339. doi: 10.1037/0022-3514.87.3.327 [DOI] [PubMed] [Google Scholar]