The Impact of Nursing Home Residents’ Characteristics on Ratings of Importance of Autonomy Preferences in Daily Care over Time

Abstract

Objective:

Nursing home (NH) residents’ preferences for everyday living are the foundation for delivering individualized care. Yet, work has not examined the impact of demographic and clinical characteristics of NH residents on the stability of their preferences over time.

Method:

This study examined the rate of change in reports of importance of 27 autonomy-related everyday preferences from the Preferences for Everyday Living Inventory over 3-months and the demographic and clinical characteristics associated with change for nursing home residents (N = 255). Descriptive frequencies and tests of mean difference were utilized to examine differences between individuals reporting change in importance over time compared to those that did not report change.

Results:

Autonomy preferences in daily care remained stable over 3-months for the majority of residents. For residents that did report change on autonomy preferences, no systematic associations of demographic or clinical characteristics were found to be associated with change. Rather, change was associated with differential characteristics based on the preference.

Conclusion:

This study indicates that knowing a person’s demographic or clinical characteristics in care will not uniformly inform a caretaker’s understanding of the individual’s reports of importance for autonomy related preferences over time. Future work should explore the role of care environment on change in preference ratings over time.

Introduction

Evolving conceptualizations of optimal care in nursing homes (NHs) emphasize a need to “know the person” to deliver high-quality, individualized, holistic care (Edvardsson, Varrailhon, & Edvardsson, 2014). The Centers for Medicare and Medicaid Services (CMS) Quality of Life guidelines in the United States further call for a shift in focus to person centered care (PCC; Bowman, 2009). Although a centralized definition of PCC is lacking (American Geriatrics Society Expert Panel on Person-Centered Care, 2016; Kogan, Wilber, & Mosqueda, 2016), quality of life models and recent evidence support the notion that individualized care is important to enhancing well-being and positive outcomes through care delivery for individuals with and without dementia (McCormack & McCance, 2006; Radwin, 1995; Radwin & Alster, 2002; Robinson, Callister, Berry, & Dearing, 2008).

A foundation of individualized care plans within PCC is the assessment and integration of a person’s preferences for everyday living (Van Haitsma et al., 2014; Van Haitsma et al., 2015). Such psychosocial preferences represent expressions of how one would like his or her needs met. Building care around a person’s preferences has been shown to improve decision making (Whitlatch, 2012), quality of care (Simmons & Schnelle, 2004; Thompson & Smith, 1998), satisfaction with care (Applebaum, Straker, & Geron, 2000), and quality of life (Cohen-Mansfield, Marx, Thein, & Dakheel-Ali, 2010; Gerdner & Schoenfelder, 2010; Lawton, et al., 1998; Van Haitsma et al., 2015). Integration of preferences related to autonomy are particularly key, carrying implications for health, functioning, and well-being (Kasser & Ryan, 1999).

Initial evidence further demonstrates that reports of values and preferences over short (1 week) time frames (Feinberg & Whitlatch, 2001; Van Haitsma et al., 2014; Whitlatch, Feinberg, & Tucke, 2005) and over a 3-month time frame (Abbott, Heid, Kleban, Rovine, & Van Haitsma, in press) are consistent and reliable. However, providers often struggle to implement individualized care plans based on resident preferences (Abbott, Heid, & Van Haitsma, 2016). One outstanding concern by care staff is that despite overall stability in reports of preferences a person’s attributes, current mood, or clinical state may cause preference importance ratings to fluctuate more (Heid et al., 2014). In fact, NH staff describe resident level characteristics (i.e., behaviors, persistence, interest, mood, personal health, mental health, personal resources) as barriers to the delivery of consistent person-centered care (Abbott et al., 2016). For example, if a person is feeling “down” he or she may report (or be perceived to report) a specific preference as less important or change reported levels of importance over time. Yet, there is a gap in the literature examining how or if individuals that change their reported preferences on everyday preferences differ from those that report the same levels of importance over time on demographic and/or clinical characteristics. As a result, this study examines the stability of importance ratings of autonomy preferences in care and the demographic and clinical attributes of residents associated with change when change is present.

Conceptual Framework

Throughout life, individuals express needs, such as belongingness or safety; when a need is not met, an individual’s ability to maximize functioning is diminished (i.e., physiological, safety, belongingness and love, esteem, and self-actualization needs; Maslow, 1943). From a psychological standpoint, Deci and Ryan (2000) proposed that all humans have three innate needs: autonomy, competence, and relatedness. When these needs are supported, an individual’s psychological well-being is promoted; when they are not supported, an individual’s well-being is compromised (Deci & Ryan, 2000; Patrick, Knee, Canevello, & Lonsbary, 2007; Ryan & Deci, 2000). Preferences, thereby, reflect how an individual would like to have his or her needs met. The current study examines one particular psychological need of interest, namely, autonomy. The honoring of autonomy has repeatedly been linked to positive health and well-being outcomes in older adulthood including but not limited to improved quality of life and decreased mortality (Kasser & Ryan, 1999). In the NH context preferences for autonomy, such as having control over the type of bath one takes or the time that one goes to bed at night, are direct expressions of autonomous living that are often threatened by institutional constraints and policies (Persson & Wasterfors, 2009).

Preferences over time and the Impact of Individual Characteristics on Preference Ratings

Initial empirical evidence has demonstrated that cognitively capable NH residents (Van Haitsma et al., 2014) and individuals with mild to moderate dementia (Feinberg & Whitlatch, 2001; Whitlatch, Feinberg, & Tucke, 2005) can consistently and reliably report on their values and preferences in everyday care over short periods of time (one week). NH residents have also been shown to report stable levels of importance of preferences for everyday living over a three-month time frame (Abbott et al., in press). Furthermore, research on end-of-life preferences similarly finds stability in reports of more than 70% of preferences, with stability greatest among inpatients and seriously ill older individuals (Auriemma et al., 2014). However, even within these findings that estimate high levels of stability of 70% and above, up to 30% of individuals do report change. In order to deliver person-centered care to all residents, it is meaningful to explore factors that may impact this importance rating change for these individuals. Specifically, from a clinical standpoint, questions remain about whether clinical attributes of an individual resident can impact the stability in reports of preference importance over time for the proportion of individuals reporting change.

Residents in a qualitative study have indicated that preference importance may depend on a series of personal characteristics, including functional, cognitive, or sensory ability, and/or mood (Heid et al., 2014). Work has also demonstrated that one’s current state of functioning (i.e., vision, language comprehension, and continence) can impact the delivery of preference congruent care over time (Heid, Van Haitsma, Kleban, Rovine, & Abbott, 2015). Further, theory purports that preferences may change as one’s vulnerability or closeness to death approaches (Winter & Parker, 2007). As a result, resident demographic characteristics (i.e., gender or age) or clinical functioning at a given point in time (i.e., Activities of Daily Living (ADL) functional status, depressed mood, affect, anxiety), may influence change in reported importance over time. On the other hand, clinical characteristics may not have systematic impacts on preference ratings, as it was found that cognitive ability did not affect everyday preference importance over 3-months (Carey, Heid, & Van Haitsma, 2017) and systematic changes in end-of-life preferences have not been found when considering changes in health status (Auriemma et al., 2014). Given the call for person-centered care and the need to surmount barriers in implementing individualized care plans, it is critical to address this gap in understanding regarding the impact of resident characteristics on assessments of preferences for everyday living to maximize positive outcomes associated with autonomy.

Current Study

The purpose of this study was to examine the rate of change in reports of importance of 27 autonomy-related everyday preferences from the Preferences for Everyday Living Inventory – Nursing Home assessment battery over 3-months and the demographic and clinical characteristics associated with change. We focused on the impact of factors routinely documented or addressed by clinical staff (i.e., gender, age, length of stay, pain, depression, anxiety, positive and negative affect, perceived health, functional status, and facility satisfaction). We hypothesized that NH residents reporting change in preferences would demonstrate greater clinical vulnerability (i.e., less functional ability, more anxiety, more depression) than those not reporting change in preference importance ratings over 3-months.

Methods

Participants and Procedures

Social work staff in 28 NHs in the suburbs of a major metropolitan East Coast region of the United States were provided with study information and asked to identify potentially eligible candidates based on specific criteria. Eligible residents were English speaking and had a length of stay of at least one week with an expected stay of at least 3 months. At each NH, the attending physician or director of nursing verified that referred, eligible residents were medically stable and had the cognitive capacity to consent for themselves or had a family member that could consent for them. A person was considered “not medically stable” if (s)he was considered at the end of life or had a medical condition that was acute or would keep him/her from physically being able to participate in an interview. A letter was provided to the medical team with this definition. Capacity to consent was determined both by the clinical impression of the person’s doctor or Director of Nursing and by using an iterative questioning format for consenting. After each subsection of the consent was reviewed, participants were prompted to repeat the main point to verify comprehension. Of the 581 residents identified by provider staff, 575 were deemed medically stable. Of these individuals, 255 provided consent, scored high enough on a cognitive screen, and participated in two waves of data collection (i.e., T1 and T2 which took place 3 months later).

Consent was obtained then a research assistant then administered the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975) as a cognitive screen. Residents were required to score at least a 13 on the MMSE to participate. The decision to use a cutoff of 13 on the MMSE was in part based on norms identified by Crum, Anthony, Bassett, and Folstein (1993) and on our team’s prior feasibility testing of the PELI-NH with individuals ranging from moderately cognitively impaired to those with no impairment (MMSE scores ranging from 13 to 30). At T1, participants completed the Preferences for Everyday Living Inventory-Nursing Home (PELI-NH) assessment and a battery of measures assessing pain, perceived health, and how they were feeling. At T2, 3 months later, the process was repeated; medical stability was re-confirmed, consent obtained, and MMSE administered prior to the interview. Comprehensive and quarterly Minimum Data Set (MDS) 3.0 data were collected for dates aligning with the T1 interview for each resident.

Our attrition rate from T1 to T2 was 25.4% which was due to death, transfer, change in cognitive ability, withdrawal, or change in medical stability over the 3 months. Those completing both T1 and T2 were not significantly different than those that only completed T1 on age, race, education, or length of stay. Participating residents had a mean age of 81 years (SD = 11.2) and were predominately widowed (44.3%) non-Hispanic Caucasian (76.9%) females (67.8%) with a high school education (48.2%). Average MMSE score at T1 was 24.6 (SD = 3.9) and average length of stay was 923.6 days (SD = 900.9). See Table 1 for additional sample demographics.

Table 1.

Sample Descriptive Statistics on Demographic and Clinical characteristics

Individual Characteristic	%(N) or M(SD)
Gender (female; %/n) *	68% (173)
Age (M/SD)	80.97(11.21)
Length of Nursing Home Stay	923.58(900.91) days
Pain	4.46(4.41)
Depressive symptoms	4.6(4.38)
Anxiety	11.72(11.43)
Positive affect	29.26(8.17)
Negative affect	18.08(7.18)
Perceived health	51.02(13.94)
Functional status	21.91(8.25)
Facility Satisfaction	7.63(3.28)

Note. N = 255

Measures

Pain.

Participants completed an 11-item adapted version of the Geriatric Pain Measure (GPM; Ferrell, Stein, & Beck, 2000), rating whether they had experienced symptoms of pain within the past two weeks as 1 (yes) or 0 (no). A total item score was computed (α = .88).

Depressive Symptoms.

Participants completed the Patient Health Questionnaire (PHQ-8; Kroenke, Strine, Spitzer, Williams, Berry, & Mokdad, 2009). Participants were asked if they were bothered over the past week by a list of eight symptoms, using a 4-point Likert scale of 0 (not at all) to 4 (nearly every day). A mean-item total score was computed (α = .73).

Anxiety.

A 25-item self-report measure was administered to participants to rate how often they had experienced anxiety-related symptoms within the past week from 0 (not at all) to 3 (nearly every day; Geriatric Anxiety Scale (GAS); Segal, June, Payne, Coolidge, & Yochim, 2010). A mean-item total score was computed (α = .90).

Positive and Negative Affect.

Participants self-reported on the 20-item Positive and Negative Affect Scale (PANAS) the extent to which they had felt a series of emotions in the past two weeks from 1 (very slightly or not at all) to 5 (extremely). Items fall on two subscales for which mean-item total scores were computed: positive affect (α = .83) and negative affect (α = .85; Watson, Clark, & Tellegen, 1988).

Perceived Health.

Five general self-rated health items were pulled from the SF-36 questionnaire for participants to rate their health in comparison to others they knew of the same age (Ware & Sherbourne, 1992). One item asked participants to rate their perceived general health on a scale of 1 (excellent) to 5 (poor), while the 4 other items asked about health compared to others on a scale of 1 (definitely true) to 5 (definitely false). Negative items were reverse coded and a mean item-total score was computed (α = .67).

Facility Satisfaction.

Four items asked participants to rate their overall satisfaction with their care provider on a 5-point Likert scale of 1 (yes, always) to 5 (no, never; Ohio Nursing Home Resident Satisfaction survey, 2012). A mean-item total score was calculated to determine the overall level of care satisfaction experienced by the resident (α = .79).

MDS 3.0 Chart Data.

Medicare and/or Medicaid certified NHs routinely provide comprehensive clinical assessments of residents’ functional capabilities (MDS 3.0; Saliba & Buchanan, 2009). Chart data from residents’ most recent comprehensive and quarterly assessments were extracted from medical records based on the date of T1 interview completion. Data included: demographic characteristics of age, gender, length of stay in days, and reports of ADLs as an indicator of functional ability. Ten-items assessed ADL capacity from 0 (independent) to 4 (total dependence, activity only occurred once or twice or did not occur), and a mean item-total score was computed (α = .90).

Dependent Variables.

Preference Importance Ratings.

Residents completed the Preferences for Everyday Living Inventory-Nursing Home version (PELI-NH; Van Haitsma et al., 2012; Curyto, Van Haitsma, & Towsley, 2016) at T1 and T2. The PELI-NH includes questions regarding a range of leisure pursuits and personal care preferences that fall into five conceptual domains: self-dominion/autonomy in care (e.g., taking care of personal belongings), enlisting others in care (e.g., be involved in discussions about your care), leisure and diversionary activities (e.g., watching TV, doing outdoor tasks), social contact (e.g., having regular contact with friends), and growth activities (e.g., keeping up with the news; Carpenter, Van Haitsma, Ruckdeschel, & Lawton, 2000). Residents rated the importance of their preferences from 1 (very important) to 4 (not important at all). Responses were reverse coded on all items to ease interpretation, with higher scores indicating greater importance (4 = very important) and lower scores indicating less importance (1 = not important at all). For the purposes of this focused study, we restricted analyses to the 27 preferences that assess preferences for autonomy in care as identified in Carpenter et al. (2000).

Analyses

Our goals were twofold: (1) to understand the rate of change in reports of importance of 27 autonomy-related everyday preferences from the PELI-NH over 3-months and (2) to examine whether individuals that change preference ratings differ on demographic and clinical characteristics. To address our research questions and hypotheses, in line with prior work (Van Haitsma et al., 2014; Abbott et al., in press), we used a clinically meaningful definition of change; change in importance was determined evident if a respondent reported a preference as either very or somewhat important at T1 in contrast to reporting a preference as not very or not at all important at T2, or vice versa. We created pattern variables of change for each of the 27 autonomy preferences on the PELI-NH to understand the proportion of the sample who reported different levels of importance at T1 and T2. Pattern variables were constructed with codes 1 (important) and 0 (not important), such that if the preference was important at T1 but unimportant at T2, the resultant variable was coded 10. If the preference was not important at T1 but important at T2, the variable was coded 01. If the person reported consistency of a preference being important it was coded 11 or if the person reported consistency of a preference being unimportant it was coded as 00. Frequencies were examined for these four groups to determine rate of change on a given preference by the sample from T1 to T2.

Second, the change groups (10 and 01) were combined and the stable groups were combined (00 and 11). Where sufficient numbers of the sample reported a change in importance (i.e., ≥ 20% of the sample) groups were compared with chi-square and t-tests to determine if groups differed on gender, age, length of stay, pain, depression, anxiety, positive and negative affect, perceived health, functional status, and facility satisfaction. To account for the number of comparisons made with these data, an omnibus p value adjustment is advised. All results reaching significance of p < .05 are presented, but we heed caution in interpreting results not reaching significance of p < .001. See Table 1 for overall sample descriptives.

Results

First examining the rate of change, we see high levels of stability in preference ratings (as reported elsewhere; Abbott et al., in press; see Table 2). More specifically, here it is evident that on 19 out of 27 preferences for autonomy, 80% or more of the sample reported no change in preference ratings from T1 to T2, indicating that the distinction between important versus not important for these preferences remained consistent across three months. The majority of this consistency is in ratings of autonomy preferences as important and remaining important over time. On 8 items, however, despite high overall stability within the sample, we do see greater rates of change, with between 20–32% of the sample changing from important to not important or vice versa. These preferences include: choose what name you would like me to use when I greet you, follow a routine when you wake up in the morning, choose what time of day to bathe, take a nap when you wish, adjust the lighting in your room, choose when to eat, choose where to eat, and order take-out food.

Table 2.

Patterns of Change in Importance by Autonomy PELI-NH Item

Autonomy PELI-NH Item: “How important is it to you to…”	11 No Change: Important	10 Change: Important to Not Important	01 Change: Not Important to Important	00 No Change: Not Important	% change
01B. Choose what name you would like me to use when I greet you?a	57% (144)	11% (29)	16% (40)	16% (41)	27% (69)
02. Choose when to get up in the morning?	73% (185)	7% (18)	12% (31)	8% (20)	19% (49)
03. Follow a routine when you wake up in the morning?	70% (179)	12% (30)	11% (28)	7% (17)	23% (58)
04. Choose how often to bathe?	87% (221)	4% (11)	6% (15)	2% (6)	10% (26)
05. Choose what time of day to bathe?	69% (176)	10% (25)	11% (28)	10% (25)	21% (53)
06. Choose between a tub bath, shower, bed bath, or sponge bath?	80% (203)	9% (22)	8% (19)	4% (10)	16% (41)
07. Choose what clothes to wear?	73% (186)	8% (19)	9% (24)	10% (26)	17% (43)
08. Choose how to care for your mouth?	87% (226)	3% (8)	6% (14)	2% (5)	8% (22)
09. Choose how often to care for your nails?	73% (18)	6% (16)	11% (29)	9% (24)	18% (45)
10. Choose how to care for your hair?	84% (215)	6% (14)	7% (17)	3% (7)	12% (31)
11. Take a nap when you wish?	52% (133)	11% (27)	18% (45)	19% (49)	28% (72)
12. Set up your room the way you want?	75% (191)	7% (18)	12% (31)	6% (15)	19% (45)
13. Take care of your personal belongings or things?	93% (236)	3% (8)	2% (6)	2% (5)	5% (14)
14. Keep your room at a certain temperature?	86% (220)	6% (16)	5% (12)	2% (6)	11% (28)
15. Adjust the lighting in your room?	76% (104)	11% (27)	10% (25)	4% (9)	20% (52)
16. Choose your own bedtime?	79% (201)	7% (17)	10% (26)	4% (11)	17% (43)
17. Follow a routine when you go to bed?	69% (177)	9% (24)	10% (25)	11% (29)	19% (49)
18B. Set up your bed for comfort?	83% (212)	6% (16)	8% (20)	2% (4)	14% (36)
21B. That your daily caregiver knows your needs when going to the bathroom?	79% (202)	7% (18)	6% (15)	7% (17)	13% (33)
27B. Do what helps you feel better when you are upset?	85% (216)	5% (13)	5% (13)	1% (3)	10% (26)
32. Have privacy?	79% (201)	6% (15)	9% (23)	6% (15)	15% (38)
33. Lock things up to keep them safe?	71% (181)	8% (20)	7% (18)	14% (35)	15% (38)
34. Be involved in choosing your roommate?b	67% (117)	8% (14)	10% (18)	14% (25)	13% (32)
35. Choose what to eat?	81% (207)	9% (23)	6% (15)	4% (9)	15% (38)
36. Choose when to eat?	52% (132)	15% (39)	17% (43)	16% (40)	32% (82)
37. Choose where to eat?	57% (146)	13% (32)	14% (36)	15% (39)	27% (68)
40. Order take-out food?	34% (87)	11% (28)	10% (26)	44% (112)	21% (54)

Note. PELI-NH = Preferences for Everyday Living Inventory-Nursing Home.

^aQuestion numbers are not consecutive as they refer to the item number on the PELI-NH survey

^bOnly 174 residents responded to this question as it did not apply for the remainder of the residents surveyed.

Second when we examine group differences for individuals that report change versus those that do not on these 8 preferences, we do not find systematic associations with demographic or clinical characteristics of the resident as hypothesized. However, we do find distinct differences by preference (see Table 3). For the preference of choosing what name one would like to have used when greeted, we see that those reporting change in the importance of this preference over time are significantly older (m = 84.65) than those that do not report change in this preference over time (m = 79.70). For the preference of following a routine when one wakes up in the morning those that report change report significantly less positive affect (m = 27.03) than those that report no change in importance in this preference over three months (m = 29.95). For the preference of choosing what time of day to bathe individuals that report change are significantly more likely to be male (43%, n = 23), older (m = 83.58), and report greater depressive symptoms (m = 5.72) than those that do not report change in importance of this preference over time (29%, n = 48 males; m = 80.26 years old; m = 4.40 depressive symptoms). And, for the preference of being able to order take out food, individuals that report change in importance ratings report significantly lower perceived health scores (m = 47.69) than those that do not report change in importance of this preference over time (m = 51.91).

Table 3.

Group differences for Residents that Change Importance Ratings Over 3 Months Compared to Residents who do not Change Importance Ratings Over 3 Months

Autonomy PELI-NH Item: “How important is it to you to…”	Individual characteristic [X2(df)/t(df)]
	Gender	Age	Length of Stay	Pain	Depressive Symptoms	Anxiety	Positive Affect	Negative Affect	Perceived Health	Functional Ability	Facility Satisfaction
01B. Choose what name you would like me to use when I greet you?a	0.82(1)	−3.20(252)**	0.70(246)	−1.51(248)	−1.30(249)	0.75(247)	1.01(252)	0.51(252)	0.62(252)	−0.20(248)	0.29(252)
03. Follow a routine when you wake up in the morning?	0.23(1)	0.16(252)	0.63(246)	−1.17(248)	−0.43(249)	−0.23(247)	2.42(252)*	−0.31(252)	−0.43(252)	0.29(248)	−1.55(252)
05. Choose what time of day to bathe?	4.08(1)*	−2.24(102.85)*	0.04(246)	−0.50(248)	−1.96(249)*	0.84(247)	0.45(252)	−0.44(252)	0.50(252)	−1.22(248)	−1.28(252)
11. Take a nap when you wish?	0.08(1)	−0.57(252)	0.25(246)	1.14(248)	0.55(249)	0.71(247)	0.27(252)	0.01(252)	−1.06(252)	0.32(248)	0.75(252)
15. Adjust the lighting in your room?	0.82(1)	0.63(253)	−0.21(247)	0.94(249)	0.19(250)	1.88(248)+	1.19(253)	1.06(253)	0.42(253)	−1.59(249)	0.27(253)
36. Choose when to eat?	0.99(1)	−0.37(252)	−0.60(246)	1.37(248)	0.75(249)	0.42(247)	1.71(252)+	0.38(252)	−0.30(252)	−1.59(248)	0.61(252)
37. Choose where to eat?	0.01(1)	0.04(251)	0.93(245)	0.54(247)	0.36(248)	0.228(246)	1.80(251)+	−0.26(251)	0.50(251)	−1.89(247)+	−0.81(251)
40. Order take-out food?	0.24(1)	−0.90(251)	0.05(245)	−1.38(247)	−0.58(248)	0.16(246)	−0.43(251)	0.16(251)	1.98(251)*	0.55(247)	0.87(251)

⁺p < .10

^*p < .05

^**p < .01

^***p < .001

Note. PELI-NH = Preferences for Everyday Living Inventory-Nursing Home.

^aQuestion numbers are not consecutive as they refer to the item number on the PELI-NH survey

Post-hoc descriptive examination of the data reveals that for each of these preferences where we see significant differences between the change group and the non-change group on individual characteristics, we find that equal proportions of individuals are increasing in importance ratings over time as decreasing in importance (see Table 2). Therefore, instability in responses is due to a shift in importance in either direction.

Discussion

Recent efforts to deliver person-centered care to older adults in NHs call for the assessment of preferences to guide individualized care planning. However, the implementation of systematic preference assessments has been limited in practice. One barrier to implementation is a lack of knowledge about how individual characteristics of residents may impact their ratings of preferences over time (Abbott et al., 2016). Given the critical impact of supporting autonomy-related preferences to provide PCC, this study examined the stability of preference importance ratings related to autonomy over time and how those that report change in importance levels differ on key demographic and clinical characteristics. Results reveal high levels of stability of importance ratings of autonomy preferences assessed. For only eight preferences we find that ≥ 20% of the sample reported change over time. Where a portion of the sample does report change, we do not find systematic differences between the group that changed their importance ratings and the group that did not change importance ratings on individual level characteristics, rather differences are preference specific. These finding carry implications for research and practice.

An initial finding is that the importance level on the majority of autonomy-related everyday preferences for NH residents did not change over time. Only for eight out of 27 (30%) of autonomy preferences assessed did more than 20% of the sample report change over time: choose what name you would like me to use when I greet you, follow a routine when you wake up in the morning, choose what time of day to bathe, take a nap when you wish, adjust the lighting in your room, choose when to eat, choose where to eat, and order take-out food. The apparent stability parallels earlier work showing that the majority of everyday preference importance ratings remain consistent/stable over time (Abbott et al, in press; Van Haitsma et al., 2014), work that fails to find differences in importance stability by resident cognitive status (Carey et al., 2017), and even work on end-of-life care preferences (Auriemma et al., 2014). Where we do see change it may be that these preferences are more unstable within certain care circumstances (i.e., the way food delivery happens may impact one’s reported importance for choosing when to eat, where to eat, or ordering take-out food) or for certain types of individuals (i.e., older individuals).

To this latter point, when we examine group differences for individuals that report change versus those that do not on preferences showing greater change at the sample level, we do not find systematic associations with demographic or clinical characteristics of the resident, but we do find distinct differences by preference. Individuals that changed their importance for choosing what name to be greeted by and choosing what time of day to bathe were older than those showing no change. Age may impact the importance of these preferences over time because one’s viewpoint as they grow older may change—valuing autonomy and respect for these preferences differentially (either increasing or decreasing importance ratings). Being older may result in a change in importance as a reaction to clarifying their social interaction conventions. Those reporting change in importance of following a routine when one wakes up in the morning report less positive affect than those reporting no change. This finding may indicate that how a person is feeling emotionally impacts how (s)he wants to be treated in the morning. Depressive symptoms similarly are greater for those that change their preference about choosing what time of day to bathe. The apathy associated with a more depressed, less positive state may result in change in the importance ratings of these two autonomy-related preferences (i.e., when one wakes up, time of day to bathe). Additionally, those that report change on the preference importance of choosing what time of day to bathe are more likely to be male, which may indicate a differential meaning associated with autonomy with bathing for men compared to women over time. And finally, for those that report change in the importance of ordering take-out food, perceived health scores are lower, which may mean a changing importance of this preference as one’s perceived health declines. In each of these instances, knowing something about the individual’s characteristics can inform the clinical team of the potential likelihood of an individual changing importance over time. It may mean that the importance of these preferences and associated specific ways such preferences are met (i.e., exact time one wants to be bathed; see the Full PELI at preferencebasedliving.com) need to be reassessed more frequently for individuals exhibiting the identified characteristics. Or in all cases where more change is evident (i.e., 8 preferences showing greater than 20% of the sample changing), it may mean that further work needs to consider how the care environment influences one’s expression of importance. Groups did not differ on length of stay in the facility or reported satisfaction with the facility, but other factors of the environment may have an effect. Change in importance may represent a change in perceived need for that preference being met (Heid et al., 2014). For some individuals, if care is not responsive to needs, one may increase reported importance as a way of advocating for one’s autonomy to be met, while responsive care may invoke an acceptance of care and less perceived need to express importance. A changing care environment would then result in a changing importance rating.

Practically, the overall stability of preference importance ratings for autonomy in care suggests that organizational efforts to invest staff time to assess resident preferences for autonomous decision-making in care is meaningful; reassessment would not necessarily need to occur each quarter (3 months). As found prior with the full battery of the PELI-NH items (Abbott et al., in press), for the vast majority of autonomy-related preferences care plans formed initially will remain reflective of authentic resident wishes for care between quarterly care planning sessions. Yet, future research should test how long of an interval can be assumed for stability in care; scholars could potentially utilize MDS 3.0 Section F data to examine annual reports and/or reports after a change in condition. Findings reported here also uniquely demonstrate that no systematic associations of demographic or clinical characteristics were found to be associated with reports of changing importance of autonomy preferences. Yet, for select identified preferences showing more change, knowing something about a person’s individual characteristics may inform practitioners of how a resident will rate the importance of autonomy preferences over time.

This study is strengthened by its use of data from a unique clinical NH population. However, it is not without limitation. This sample included Caucasian and African-American participants, yet it was not representative of other racial or ethnic groups and it excluded individuals who were not medically stable. Results may differ for different groups of individuals and/or for those who are managing a terminal illness. Second, this study is limited by its duration of follow-up; although assessed at two points in time, 3 months apart, the vulnerability of a NH population makes long-term follow-up difficult. Future work could consider stability of autonomy preferences in other long-term care settings such as assisted living facilities or adult day settings to see if similar patterns of findings are evident. Third, this study only explored the impact of a select set of individual characteristics on preference importance ratings for preferences for autonomy in care and only used a 4-item battery for assessment of self-perceived health; future work should explore additional clinical attributes of residents and a more comprehensive assessment of health to determine if similar patterns exist for other types of preferences. Fourth, this work also did not explore the impact of change in clinical state over time on the change in preference ratings over time; such an investigation may further our understanding of the impact of clinical attributes on preference ratings. Fifth, we were limited in our ability due to small cell sizes to differentially examine individuals that change upwards in importance ratings versus those that report decreases in importance, but we see almost equal proportions of individuals shifting upward and downward in importance for preferences where change is reported; future work with larger samples should explore the potential unique associations of direction of preference importance change with individual attributes. Sixth, we collected data from multiple providers; future work should explore if there are provider-level variables that dictate differences in importance ratings (i.e., culture of care for empowerment of residents). Finally, the PELI-NH assesses the importance of individuals’ preferences for quality improvement purposes. Parallel work using the Full PELI (see preferencebasedliving.com) or another similar instrument where specifics of a person’s preference (i.e., the exact time of day one would like to be bathed) rather than just importance should be completed to see if the way a person would like to have preferences met changes over time if (s)he possesses specific demographic or clinical characteristics.

Conclusion

Overall, this study furthers our understanding of the assessment of autonomy-related preferences in care for NH residents. While prior work indicates that preferences can change in response to individual characteristics (Heid et al., 2014), we find overall stability across 3-months on the majority of everyday preferences related to autonomy and that individual characteristics do not systematically impact the report of importance of autonomy preferences by residents over time when change is present, rather differences are preference specific. Knowing something about an individual demographically and clinically may help inform understanding of stability of importance for preferences that do show greater levels of fluctuation over time. This work directly addresses a key cited barrier in delivering preference-based care—staff perceptions that residents often change their minds about preferences in care (Abbott et al., 2016)—and ultimately advances our understanding of the delivery of person-centered care in NHs.