Journal of Gerontological Nursing

Feature Article 

Preference Consistency: Veteran and Non-Veteran Nursing Home Resident Self-Reported Preferences for Everyday Living

Kimberly Curyto, PhD; Leah M. Dockler, PsyD; Kimberly S. Van Haitsma, PhD

Abstract

Understanding patient preferences is core to person-centered care. The consistency of everyday preference reporting was assessed comparing responses of Veteran (VA) and non-VA nursing home (NH) residents on the Preferences for Everyday Living Inventory (PELI) at baseline and 5 to 7 days later. Non-VA NH residents demonstrated higher perfect agreement than VA residents (66% vs. 56%, respectively) and higher acceptable agreement (95% vs. 88%, respectively). Multiple regression analyses examined significant predictors of reliability using demographics, cognitive functional variables, and interviewer ratings. In the VA group, higher perfect agreement was associated with residents who were less likely to have hearing deficits, better cognition, and better interviewer ratings related to energy, attention, and comprehension. In the non-VA group, higher perfect agreement was associated with residents who were younger and more independent with walking. Overall, higher agreement was associated with being female, non-VA, and having better cognition. Implications for future research and clinical practice are highlighted. [Journal of Gerontological Nursing, 46(3), 15–25.]

Abstract

Understanding patient preferences is core to person-centered care. The consistency of everyday preference reporting was assessed comparing responses of Veteran (VA) and non-VA nursing home (NH) residents on the Preferences for Everyday Living Inventory (PELI) at baseline and 5 to 7 days later. Non-VA NH residents demonstrated higher perfect agreement than VA residents (66% vs. 56%, respectively) and higher acceptable agreement (95% vs. 88%, respectively). Multiple regression analyses examined significant predictors of reliability using demographics, cognitive functional variables, and interviewer ratings. In the VA group, higher perfect agreement was associated with residents who were less likely to have hearing deficits, better cognition, and better interviewer ratings related to energy, attention, and comprehension. In the non-VA group, higher perfect agreement was associated with residents who were younger and more independent with walking. Overall, higher agreement was associated with being female, non-VA, and having better cognition. Implications for future research and clinical practice are highlighted. [Journal of Gerontological Nursing, 46(3), 15–25.]

The older adult population in the United States is projected to grow from 15% in 2014 to 24% in 2060 (Colby & Ortman, 2015). Approximately one (13%) in eight persons age ≥85 and approximately one (5%) in 20 persons ages 65 to 84 reside in long-term care facilities because of functional and/or cognitive limitations that require assistance with a variety of tasks, from bathing to medication management (Congressional Budget Office, 2013). The most common diagnoses in these settings are Alzheimer's disease or other dementias (50.4%), depression (48.7%), and diabetes (32.4%) (Harris-Kojetin et al., 2016).

In the past 3 decades, the culture within nursing homes (NHs) across the United States has been shifting from the medical model toward a holistic, person-centered care (PCC) model as a framework for delivery of health care based on an individual's values and preferences (Fazio, Pace, Flinner, & Kallmyer, 2018). Using an integrated theoretical preference model (Van Haitsma et al., 2019, p. 2), preferences are defined as “an expression of the attractiveness of an option that serves to fulfill a person's needs, is determined based on one's values, and directs behaviors to achieve goals.” PCC is essential in moving away from the treatment-oriented and deficit-driven nursing institution toward a home environment where self-efficacy and choice are promoted.

Evidence suggests that implementation of PCC approaches improves functional outcomes, quality of life (QOL), and worker experience. For example, when PCC approaches are implemented in NH settings, early evidence suggests that agitation and use of antipsychotic medications are significantly reduced in persons with dementia (Fossey et al., 2014; Li & Porock, 2014). Previous studies demonstrated that PCC is related to residents' ability to complete activities of daily living (ADL) (Sjögren, Lindkvist, Sandman, Zingmark, & Edvardsson, 2013), decrease in job dissatisfaction and burnout among direct care workers (Barbosa, Sousa, Nolan, & Figueiredo, 2015), and decreased resident boredom and helplessness (Brownie & Nancarrow, 2013). Koren (2010) reported that PCC interventions can improve QOL among residents as well as working conditions for staff through simple interventions, such as warmer temperatures in bathing areas. In addition, Epp (2003) identified multiple studies that found PCC approaches improved QOL, sleep patterns, and maintenance of self-esteem.

PCC core values (i.e., person-hood, respect and dignity, autonomy, choice and independence, and privacy) (Center for Excellence in Assisted Living, 2010) call for the assessment of individual resident preferences to deliver preference congruent care. The Preferences for Everyday Living Inventory (PELI) is a measure designed to assess specific preferences and level of importance across five domains of everyday living. The PELI was created for functionally limited homecare older adults, involving a professional advisory board and concept mapping to ensure that the tool measures a comprehensive list of everyday preferences and has good face validity, construct, and discriminant validity (Van Haitsma et al., 2013). A study by Curyto, Van Haitsma, and Towsley (2016) used cognitive interviewing with NH Veteran (VA) and non-VA residents and expert panel input to examine content validity, revise content, and adapt the language of the PELI for the NH population.

A follow-up study (Van Haitsma et al., 2014) compared response consistency of university students and NH residents across a 5- to 7-day period. The study found that perfect consistency over the trial period was similar in both groups (approximately 66%); personal care preferences were stable, whereas leisure activity preferences were less stable. A related study sought to determine why NH residents appear to change their minds about preferences over time (Heid et al., 2016). Using cognitive interviews, it was determined that residents' preferences depended on person- and environmental-level characteristics, such as functional ability, weather, quality of social interactions, potential barriers, and perception of the ability to have preferences met. The authors concluded that assessment of preferences was most dependent on contextual factors and should be completed following change in these factors (Heid et al., 2016).

The current pilot study aims to add to the existing literature by examining the consistency of responses of VA and non-VA residents in NH settings across a 5- to 7-day period. The VA NH population differs from the non-VA NH population in that it is a limited segment of the U.S. population with required service connected conditions, such as posttraumatic stress disorder, schizophrenia, and amputation, which involve psychiatric- and mobility-related disabilities served by a distinct health care organization with different policy and financial structure; primarily male residents; and a larger proportion of residents age ≤65 (McCarthy, Blow, & Kales, 2004) with traditionally higher rates of cognitive impairment (Shay & Burris, 2008). It is likely that these populations differ in person- and environmental-level characteristics that may affect preference reporting. Therefore, consistency in preference reporting was examined separately. The current study further examined factors associated with the reliability of preference responses using demographic, cognitive, functional, and interviewer rating variables. Variables associated with reliability were of interest for the VA Community Living Center (CLC) and non-VA NH groups separately. However, the current authors were interested in variables associated with reliability of preference reporting in the NH sample as a whole, and specific variables of interest required the total sample to capture more variability, such as examining how gender relates to preference consistency.

Method

Participants and Setting

Participants were recruited from the VA Western New York Healthcare System CLC (n = 42) and a comparison sample was recruited from an eastern Pennsylvania community NH (n = 37). Recruitment and consent procedures received Institutional Review Board approval for all sites. Criteria for participant selection included: age ≥55; mild impairment to intact cognition as measured by brief cognitive tests; a resident of the facility for 1 week or more; and medically stable as determined by their medical provider.

Measures

Demographics. Participants and/or their responsible party reported age, gender, education, ethnicity, and marital status for descriptive purposes (Table 1). Length of stay information was gathered from the medical record.

Participant Demographics

Table 1:

Participant Demographics

Sensory and Functional Impairment. Information regarding sensory impairment (i.e., hearing, vision) and level of assistance needed with 10 ADLs, including bed mobility, transfers, walking in the room/corridor, walking on/off the unit, dressing, eating, toileting, hygiene, and bathing, graded on a 5-point rating scale, where 0 = independent and 4 = total dependence, was gathered from the Minimum Data Set 2.0 Resident Assessment Instrument (MDS; Morris et al., 1990).

Cognition. The Modified Mini-Mental State Exam (3MS; Teng & Chui, 1987) and the Mini-Mental State Exam (MMSE; Folstein, Folstein, & McHugh, 1975) are validated measures for estimating general cognitive functioning in older adults (Tombaugh, McDowell, Kristjansson, & Hubley, 1996). Higher scores indicate a more intact cognitive functioning. The MMSE has 30 available points and a cut-off score of 22, which was used to determine eligibility in the non-VA NH group. The 3MS has 100 available points and an age and education adjusted cut-off score of greater than the second percentile, which was used to determine eligibility in the VA group. Cut-off scores for both measures allowed for a range of mildly impaired to cognitively intact participants. 3MS attention items (e.g., mental reversal, counting backwards) were summed separately as a quick measure of attention.

Psychosocial Preferences. The PELI was administered to assess preferences for everyday living. Questions covered topics from recreation and dining to personal care preferences in five domains: (a) social contact; (b) leisure and diversionary activities; (c) growth activities; (d) self-dominion; and (e) enlisting others in care. The 85-item PELI was included in the analysis. Participants were asked to rate each item's importance (i.e., “How important is it to you to…”) on a 4-point Likert scale from 1 (very important) to 4 (not important at all).

Interviewer Ratings. Interviewers also rated residents on seven questions indicating level of effort, distraction, understanding, discomfort, energy, enjoyment, and expression using a 4-point Likert scale from 1 (not at all) to 4 (very much). The authors (K.C., K.V.H.) developed these items to capture impressions observed during interviews. During the training process, the research authors (K.C., K.V.H.) and research interviewers (L.M.D. and others) compared ratings following interviews until agreement was achieved. Interrater reliability was not formally assessed.

Procedure

Participants were recruited from two NHs in the suburbs of eastern Pennsylvania for the non-VA NH sample and from the VA Western New York Healthcare System CLC for the VA sample. Social workers from each NH identified residents who were capable of being interviewed, English-speaking, and had a length of stay of 1 week or more. The attending physician then determined if residents had the capacity to consent for themselves and were medically stable. In the non-VA NH, social workers approached residents to gain their assent to be contacted by the research team and informed residents' responsible parties about the study, but they did not approach residents who lacked the capacity to consent. In the VA CLC, the research assistant contacted residents and when applicable due to lack of capacity to consent, she contacted the residents' responsible party.

The research authors (K.C., K.V.H.) trained the research interviewers in obtaining informed consent, administering the measures, administering the PELI, and completing interviewer ratings based on direct observations. Training involved observing the current authors, practicing among research staff, and being observed through the process, with post-interview discussion involving clarification and revision until proficient and agreement on interview ratings was achieved. In all sites, informed consent was obtained using interactive questioning during the consent process. If the research assistant believed the resident was unable to give consent, the process was stopped and the resident was excluded from the study in the non-VA NH and the responsible party was approached in the VA CLC. After a resident consented, the research assistant administered the MMSE (Folstein et al., 1975) in the non-VA NH or the 3MS (Teng & Chui, 1987) in the VA CLC to determine if residents were cognitively capable (MMSE score ≥22 or 3MS >2nd percentile). Consenting residents who met study criteria completed a baseline interview (T1), which comprised the 85-item PELI. Five to 7 days after completing the baseline interview, residents were re-interviewed with the PELI (T2). The final sample comprised 42 VA CLC participants and 37 non-VA NH participants.

Primary Analyses

Consistency in PELI ratings across a 1-week period was assessed separately for the VA CLC and non-VA NH samples. Consistency was calculated three ways. First, Pearson correlations were obtained between the T1 and T2 preference importance ratings, then tested for significant differences between the two samples using Fisher's z test. Next, the percent “perfect agreement” between T1 and T2 was calculated for each sample. Perfect agreement meant that the respondent reported the exact same level of importance at each time period. Acceptable agreement between T1 and T2 was calculated when level of importance of ±1 deviation was reported across administrations. A z test of proportions was used to examine significant differences between groups. If significant differences were found, consistency of preference reporting would be reported separately for the VA CLC and non-VA NH samples.

In addition, several potential predictors of reliability were explored via multiple regression to understand what participant characteristics might be related to their measures of reliability. Several measures were selected to examine the hypotheses that cognitive impairment and participant effort or fatigue may contribute to unreliability in preference responses between T1 and T2. These measures included: 3MS percentile scores, 3MS attention item total, and MMSE total and adjusted scores. Interviewer ratings of participants' level of effort, distraction, understanding, discomfort, energy, enjoyment, and expression were also included in the analysis. Demographic variables were entered, such as age, gender, race, and marital status. In addition, hearing and vision impairment, length of stay in the NH, and MDS scores for need for assistance with ADLs were included as potential predictors of reliability.

The criteria of the percent of perfect and ±1 agreement on the PELI T1/T2 measure were each zero-order correlated with the above external predictor variables. Multivariate regressions identified significant zero-order predictors from an F-test and the adjusted R2 at the 0.05 level of significance. Multiple linear regressions were used to examine relationships among significant zero-order predictors, with each of the criterion measures. The purpose was to observe the proportion of predicted variance between the criteria and each significant predictor and estimate within the multiple regressions which of them contained unique or shared variances. t tests for uniqueness were run to indicate whether significant predictors contributed to unique variance at the 0.05 level of significance, or shared variance if the t test for uniqueness was not significant. Predictors of reliability were explored for the VA CLC and non-VA NH groups separately and assessed for the NH sample as a whole.

Results

Participant Differences

Groups from the VA CLC and non-VA NH had many similarities (Table 1). Most participants completed high school and were restricted in diversity as it relates to race and ethnicity, with no significant differences between groups. Most of the sample identified as Caucasian and a smaller proportion identified as African American. Average scores on the MMSE were similar in both groups, with greater variability in the VA group. The VA CLC group was more likely to be married or divorced/separated than the non-VA group, with most of the non-VA NH sample being widowed. The non-VA NH sample was significantly older (mean age = 81.6 years) than the VA CLC sample (mean age = 74.8 years). There were significantly more men in the VA CLC (97.6%) than the non-VA NH (32.4%) sample. In addition, average scores for length of stay were shorter for the VA group, whereas both groups demonstrated great variability in length of stay (Table 1).

Percentages of Agreement

Participants were administered the PELI at two time intervals 5 to 7 days apart. Measures of agreement were derived from these scales (i.e., 85 items) for each participant. Perfect agreement was defined as exact scores across the T1 and T2 PELI items. Acceptable agreement based on ±1 deviation across T1 and T2 for each participant was calculated. The average level of perfect agreement was 0.56 (SD = 0.11, range = 0.33 to 0.88), and the average percent acceptable agreement based on zero and ±1 deviation was 0.88 (SD = 0.06, range = 0.74 to 1.0) for the VA CLC group. The average level of perfect agreement was 0.66 (SD = 0.13, range = 0.38 to 0.97) and the average percent acceptable agreement based on zero and ±1 deviation was 0.95 (SD = 0.05, range = 0.76 to 1.0) for the non-VA NH group. When the VA CLC and non-VA NH groups were compared by average level of perfect agreement, there was a significant difference (t = 3.704, p < 0.001), where the VA CLC group had lower levels of perfect agreement between two trials. When groups were compared by average level of acceptable agreement, there was a significant difference (t = 5.385, p < 0.001), where the VA CLC group had lower levels of acceptable agreement between two trials.

Preferences that displayed the most stability (i.e., perfect agreement ≥70%) in the non-VA NH sample included: taking care of and locking personal belongings; bedroom and bed set up; choosing how and when to bathe; bedtime; mouth care; staff honoring privacy; staff showing respect and care; being involved and having family involved in care discussions; having regular contact with family and friends; group activities; using tobacco; using the computer; watching television and movies; keeping up with the news; shopping; playing, watching, or keeping up with sports; and going outdoors. Several of these preferences were also the most stable in the VA CLC sample, which included: choosing when to bathe; mouth care; staff honoring privacy; staff showing respect; being involved in care discussions; and using tobacco. Being around animals and exercise were also stable preference items in the VA CLC sample.

Preference items significantly less consistent in the VA sample in comparison with the non-VA NH sample (p < 0.05) included: bed set up; choosing caregivers' gender; using tobacco; regular contact with family and friends; group activities; shopping; and keeping up with the news. Although the preference for using tobacco was less consistent in the VA sample, it was rated significantly more important in the non-VA NH sample (p < 0.01). The VA sample also demonstrated higher importance ratings for consuming alcohol (p < 0.01) and outdoor activities (p < 0.01), and lower importance ratings for choosing bedtime (p < 0.01), spending one-on-one time with others (p < 0.01), volunteering (p < 0.05), and attending concerts and plays (p < 0.01) than the non-VA NH sample.

Predictors Related to Perfect and Acceptable ±1 Agreements by Group

Several independent predictors emerged in the non-VA NH data for the perfect agreement and acceptable agreement criterion. A multiple regression model was conducted to predict non-VA NH perfect agreement from significant independent predictors, including age and ability to walk in corridors. The multiple regression model statistically signifi-cantly predicted perfect agreement (F [2, 17] = 6.80, p = 0.007, adj. R2 = 0.38) (Table 2), with a large size effect according to Cohen (1988). The independent predictors of resident age and walking in corridors both had significant t tests for uniqueness (p < 0.05). Non-VA NH residents who were older tended to produce lower perfect agreement scores, and residents with declined walking ability tended to produce lower perfect agreement in preference reporting across time points. The reverse was also true: those who were younger and demonstrated more independence with walking tended to produce higher perfect agreement scores.

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Non-Veteran Nursing Home Residents (n = 37)

Table 2:

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Non-Veteran Nursing Home Residents (n = 37)

Multiple regression was run to predict non-VA NH acceptable agreement scores from significant predictors of gender and level of expressing thoughts. The multiple regression model statistically significantly predicted acceptable agreement scores (F [2, 34]) = 9.83, p < 0.001, adj. R2 = 0.33) (Table 2), with a large effect size according to Cohen (1988). The t tests for uniqueness indicated that both predictors, gender and level of expression, were composed largely of two independent subsets of residents (p < 0.05). Non-VA NH residents who were men and rated lower on level of expression tended to produce lower ±1 acceptable agreement scores, and women with higher ratings on level of expression tended to produce higher acceptable agreement scores.

Several independent predictors emerged in the VA CLC data for the perfect agreement and acceptable agreement criteria. A multiple regression was run to predict VA CLC perfect agreement scores from significant independent predictors of hearing impairment, 3MS percentile score, and interviewer ratings for level of understanding, distraction, level of energy, and effort. The multiple regression model statistically predicted significantly perfect agreement scores for the group (F [6, 35] = 3.73, p = 0.006, adj. R2 = 0.29) (Table 3), with a large effect size according to Cohen (1988). Only the predictor level of understanding ratings had a significant t test for uniqueness (p < 0.05), which indicated the presence of residents who were unique. Residents shared their variance with the other predictors and were involved in all or several of the preditor relationships with the criterion. VA CLC participants who demonstrated lower rated level of understanding produced lower perfect agreement, and those with hearing impairment, lower cognitive functioning, lower ratings on level of effort and energy, and higher ratings on level of distraction tended to produce lower perfect agreement scores, and the opposite on these measures tended to produce higher perfect agreement scores in preference reporting across time points.

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Veteran Residents (n = 42)

Table 3:

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Veteran Residents (n = 42)

An additional multiple regression was run to predict VA CLC acceptable agreement from significant independent predictors of vision impairment, 3MS attention items, and MMSE total score. The multiple regression model statistically significantly predicted acceptable agreement (F [3, 38] = 5.62, p = 0.003, adj. R2 = 0.25) (Table 3), with a large effect size according to Cohen (1988). Only the predictor visual impairment made a unique contribution to the R2 based on t tests for uniqueness (p < 0.05). The other two predictors, 3MS attention and MMSE, shared many of the same residents in the multiple regression. VA CLC residents with visual impairment tended to produce lower ±1 acceptable agreement scores, but those with better scores on 3MS attention items and total scores for the MMSE tended to produce higher agreement measures. The reverse was also true.

Predictors Related to Perfect and Acceptable ±1 Agreements for All Residents

Several independent predictors emerged in the combined group data for non-VA NH and VA CLC residents for the acceptable agreement criterion. A multiple regression run to predict VA CLC and non-VA NH perfect agreement included two significant independent predictors of gender and VA status, and the multiple regression model statistically significantly predicted perfect agreement (F [2, 76] = 9.61, p < 0.001, adj. R2 = 0.18), with a moderate effect size. t tests for uniqueness found only unique significance for VA/non-VA status (p < 0.05), indicating lower perfect agreement scores for the VA group. The t test for uniqueness on gender was non-significant (t = 0.64, p = 0.525), which indicated a strong sharing of variance between gender and VA status and that VA status was entirely responsible for the adjusted R2 predictability.

A multiple regression was also run to predict VA CLC and non-VA NH acceptable agreement from significant independent predictors: gender, MMSE adjusted score, VA status, and interviewer ratings for level of distractibility and ability to express thoughts. The multiple regression model statistically significantly predicted acceptable agreement (F [5, 72] = 7.08, p < 0.001, adj. R2 = 0.28) (Table 4), with a large size effect according to Cohen (1988). Two of the predictors maintained unique significance, level of distractibility and MMSE adjusted score (p < 0.05). The ±1 acceptable percent of agreement on the PELI was associated with two independent groups of residents who tended to be rated less distracted while taking the test, with others who had higher MMSE adjusted scores, and with still other residents who had shared characteristics of being female, higher rated level of ability to express themselves well, and being non-VA.

Multiple Regression Between Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors for All Residents (N = 79)

Table 4:

Multiple Regression Between Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors for All Residents (N = 79)

In summary, VA CLC and non-VA NH residents who are Veterans tended to produce lower perfect agreement, and those with better scores on the MMSE and lower ratings of distraction tended to produce higher acceptable agreement measures.

Discussion

The PELI was designed to assess resident preferences for everyday living and supports the core values of PCC, which stresses the importance of choice. The current article reports the consistency of responses on the PELI over a 1-week period among VA CLC and non-VA NH residents and which factors are associated with the reliability of preference responses. Findings are consistent with previous work demonstrating that consistency of preference ratings over a 1-week period when using exact agreement measures is lower than desired (approximately 60%). Exact agreement in a sample of college students compared with community NH residents suggested that stability of reported preferences is similar in younger and older populations living in group residences (Van Haitsma et al., 2014). Furthermore, the increase in consistency in reported preferences when allowing a variance is consistent with previous reports on using the PELI with NH residents and college students (Van Haitsma et al., 2014), increasing consistency to 86% to 87% when allowing for preferences to be reported as either important or not important. Findings are also in line with the work of Cohen-Mansfield and Jensen (2007a), who assessed the reliability of community-dwelling older adults' reports of preferences for self-care over a 2-week period. Exact agreement for item content was 73%, and acceptable ±1 agreement was 93%. Carpenter, Kissel, and Lee (2007) demonstrated similar reliability of care preferences using the PELI among community-dwelling older adults with and without dementia over a 1-week period (intraclass correlation coefficient = 0.62 to 0.73).

Understanding the consistency of assessed preferences is central when delivering preference-based care. The change in PELI responses over 1 week may reflect that preferences fluctuate. Alternatively, lower agreement scores may reflect measurement error and the need for improved measurement. Cognitive interviewing techniques in previous work revised the PELI to include wording understood by NH residents to decrease this potential misunderstanding due to unfamiliar language as a source of measurement error. However, the format/method of interview or interviewer may have created differences across time points. Measurement errors might be due to (a) errors in recording or coding of responses, (b) uncontrolled aspects of the interview (e.g., fatigue, distraction), or (c) influence of other variables (e.g., illness, use of as-needed pain medication that day). In addition, participant responses on PELI items are significantly skewed and have a severe restriction in range, which could contribute to instability. For example, the preference item for use of tobacco was more highly rated and less stable in the VA CLC sample than in the non-VA NH sample.

It is important to understand factors associated with the reliability of reported preferences. In the current study, residents who were more cognitively impaired were less consistent, similar to previous findings that community-dwelling older adults with mild dementia were less stable than individuals without dementia (Carpenter et al., 2007). In addition, interviewer ratings of higher distractibility and lower levels of expression were related to inconsistencies in preference reporting in both groups. The VA CLC sample's reliability of reported preferences was also related to interviewer ratings of level of energy, effort, and level of understanding. It is not clear why these ratings were significantly related to VA CLC residents' and not non-VA NH residents' preference ratings. This outcome may be related to measurement error due to different interviewers rating each group and/or use of ratings without established reliability and validity.

The level of consistency was significantly different among VA CLC residents and non-VA NH residents despite equivalent scores on cognitive tests. Results indicate that regardless of setting, male residents and Veterans are less consistent in reporting preferences over time. For the VA CLC sample, vision impairment was found to be a unique contributor of unreliability for acceptable agreement and hearing impairment was a contributor of unreliability when using perfect agreement. Although hearing impairment could have impacted residents' ability to hear and understand the interviewer questions, it is not clear why vision impairment would contribute to instability in preference reporting. Low vision could affect residents' ability to engage in activities that they may value and their perception of the ability for preference fulfillment, already noted to affect ratings of preference importance. It is also possible that hearing impairment had the same effect on change in reported preferences, affecting perceived opportunity for fulfillment. Previous work (Heid et al., 2016; Heid, Van Haitsma, Kleban, Rovine, & Abbott, 2017) has noted that NH residents indicate the stability of preferences depends on change in health status and function in addition to actual and perceived opportunity for preference fulfillment and true change in level of interest. For the non-VA NH sample, being older and demonstrating more impairment in walking was related to inconsistent preference reporting. Specific to non-VA NH residents, those older adults with lower ability to walk independently were associated with less consistent preference reporting. Results for contributors of reliability in the non-VA NH sample are consistent with previous work in other NH samples that demonstrated that reliability of preference item content was found to vary with age and instrumental ADL status (Cohen-Mansfield & Jensen, 2007b). Further explanation for these differences warrants exploration.

Limitations

Although efforts were made to include residents with diversity, the pilot sample was limited and future work is needed with NH residents demonstrating greater diversity in characteristics, such as age, race, ethnicity, gender, and physical and cognitive functioning. The current study used small convenience samples in the VA CLC and non-VA NH settings, limiting the power needed to detect additional contributors to reliability, especially characteristics specific to each sample such as gender and VA status. The small sample also limits the ability to generalize results to the broader populations of VA CLC and non-VA NH residents. Although these two samples were similar based on a brief cognitive test, they were recruited using slightly different procedures, and the VA CLC sample included residents who required consent from a legally authorized representative due to lack of ability to consent. Although this need for representative consent accounted for <5% of the VA pilot sample, it could limit the ability to make comparisons between groups. Important factors and characteristics that may be related to the reliability of preference reporting were absent, including differences in religion and other characteristics that would affect values and goals conceptualized to affect preferences (Van Haitsma et al., 2019); medical and psychiatric diagnoses; medications used that could impact attention and communication; and standardized, comprehensive measures of neurocognitive functioning, among other potential factors.

Future Direction

Future work is needed to examine the reliability of the PELI and associated contributors using a larger, more diverse sample. Using a larger, more diverse sample would allow for the power needed to examine which preferences are stable and which are dynamic and truly change over time, and also to further understand contributors of stability in specific groups and for specific preferences.

Previous work has demonstrated similar consistency for >3 months in the non-VA NH population (Abbott, Heid, Kleban, Rovine, & Van Haitsma, 2018). It will be important to examine stability and reliability of the PELI over a longer period, including greater diversity in the non-VA and VA NH population, to guide recommendations on the frequency of preference assessment. Understanding contributors of stability and change in preferences will help determine which and when preferences should be reassessed and for which subgroups. For example, both groups reported stable preferences for staff showing respect, privacy, and being involved in care discussions; these preferences will be important to address in any NH setting and may only need to be assessed annually or at significant change. This finding is consistent with previous work that demonstrated preferences in the domain of enlisting others in care to be the most consistent (Abbott et al., 2018). In contrast, some preference items were significantly less consistent in the VA CLC sample compared to the non-VA NH sample, including preferences for choosing caregiver gender, bed set up, using tobacco, regular contact with family and friends, group activities, shopping, and keeping up with the news. It is recommended that when assessing VA preferences, these items be reassessed more frequently, especially after a significant change in sensory, affective, cognitive, or physical functioning. It was notable that there were certain preferences rated more important by the VA population in the current pilot sample. It will also be important to learn more about which preferences are more or less important, as well as consistent, in different groups, which will help focus preference assessment and engagement in preference congruent activities and care.

Qualitative work is needed to further understand reasons for the differences in the stability of preference reporting based on gender and VA status and help better understand the relationship between stability of preference reporting and sensory impairments and cognitive and functional deficits. To further understand impact of distractibility, expression, level of energy, and level of effort, it will be important to examine the potential impact of common physical and mental health factors and diagnoses, such as depression, dementia, and diabetes, on stability of preference reporting. Conditions such as depression can impact changes in function, perceived opportunity for preference fulfillment, and level of interest—all factors found to be related to changes in preference reporting (Heid et al., 2016).

Conclusion

The current results indicate the importance of understanding the true stability of preferences for everyday living, such as which preferences are stable, which are more dynamic, and which truly change and under what conditions. The current study also highlights the benefit of understanding contributors to reliability and in which situations preference reassessment will be required. Preliminary results suggest that it will be important to reassess preferences after a significant change in health status, cognitive or physical level of functioning, or when vision or hearing impairment develops. In addition to cognitive, sensory, and functional deficits contributing to change in preference reporting, it is likely that these limitations impact resident perception of the ability to have important preferences fulfilled and thus their ratings of preference importance.

Resident adjustment of expectations of the care environment when it no longer matches their current abilities/disabilities is adaptive. However, it will be important to adjust the care environment to the level of functioning of the resident to meet important preferences and enhance health care professionals' ability to deliver PCC. Providing person-centered preference congruent care may have the impact of increasing stability of preference importance ratings. Improved measurement of NH resident preferences and their fulfillment may serve as a quality improvement indicator and support intervention studies around implementation of preference congruent care, integrating preferences in care planning and the impact on important clinical outcomes.

References

  • Abbott, K. M., Heid, A. R., Kleban, M., Rovine, M. J. & Van Haitsma, K. (2018). The change in nursing home residents' preferences over time. Journal of the American Medical Directors Association, 19(12), 1092–1098 doi:10.1016/j.jamda.2018.08.004 [CrossRef] PMID:30287264
  • Barbosa, A., Sousa, L., Nolan, M. & Figueiredo, D. (2015). Effects of person-centered care approaches to dementia care on staff: A systematic review. American Journal of Alzheimer's Disease and Other Dementias, 30(8), 713–722 doi:10.1177/1533317513520213 [CrossRef] PMID:24449039
  • Brownie, S. & Nancarrow, S. (2013). Effects of person-centered care on residents and staff in aged-care facilities: A systematic review. Clinical Interventions in Aging, 8, 1–10 doi:10.2147/CIA.S38589 [CrossRef] PMID:23319855
  • Carpenter, B. D., Kissel, E. C. & Lee, M. M. (2007). Preferences and life evaluations of older adults with and without dementia: Reliability, stability, and proxy knowledge. Psychology and Aging, 22(3), 650–655 doi:10.1037/0882-7974.22.3.650 [CrossRef] PMID:17874963
  • Center for Excellence in Assisted Living. (2010). Person-centered care in assisted living: An informational guide. Retrieved from https://www.theceal.org/component/k2/item/644
  • Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
  • Cohen-Mansfield, J. & Jensen, B. (2007a). Dressing and grooming: Preferences of community-dwelling older adults. Journal of Gerontological Nursing, 33(2), 31–39 doi:10.3928/00989134-20070201-07 [CrossRef] PMID:17310661
  • Cohen-Mansfield, J. & Jensen, B. (2007b). Self-maintenance habits and preferences in elderly (SHAPE): Reliability of reports of self-care preferences in older persons. Aging Clinical and Experimental Research, 19(1), 61–68 doi:10.1007/BF03325212 [CrossRef] PMID:17332723
  • Colby, S. & Ortman, J. (2015). Projections of the size and composition of the U.S. population: 2014 to 2060. Retrieved from https://www.census.gov/content/dam/Census/library/publications/2015/demo/p25-1143.pdf
  • Congressional Budget Office. (2013). Rising demand for long-term services and supports for elderly people. Retrieved from https://www.cbo.gov/sites/default/files/113th-congress-2013-2014/reports/44363-ltc.pdf
  • Curyto, K., Van Haitsma, K. & Towsley, G. (2016). Cognitive interviewing: Revising the preferences for Everyday Living Inventory for use in the nursing home. Research in Gerontological Nursing, 9(1), 24–34 doi:10.3928/19404921-20150522-04 [CrossRef]
  • Epp, T. D. (2003).Person-centered dementia care: A vision to be refined. Retrieved from http://www.livingdementia.com/downloads/docs/person-centered_dementia_care.pdf
  • Fazio, S., Pace, D., Flinner, J. & Kallmyer, B. (2018). The fundamentals of person-centered care for individuals with dementia. The Gerontologist, 58(Suppl. 1), S10–S19 doi:10.1093/geront/gnx122 [CrossRef] PMID:29361064
  • Folstein, M. F., Folstein, S. E. & McHugh, P. R. (1975). “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198 doi:10.1016/0022-3956(75)90026-6 [CrossRef] PMID:1202204
  • Fossey, J., Masson, S., Stafford, J., Lawrence, V., Corbett, A. & Ballard, C. (2014). The disconnect between evidence and practice: A systematic review of person-centred interventions and training manuals for care home staff working with people with dementia. International Journal of Geriatric Psychiatry, 29(8), 797–807 doi:10.1002/gps.4072 [CrossRef] PMID:24535885
  • Harris-Kojetin, L., Sengupta, M., Park-Lee, E., Valverde, R., Caffrey, C., Rome, V. & Lendon, J. (2016). Long-term care providers and services users in the United States: Data from the National Study of Long-Term Care Providers, 2013–2014. Vital Health Statistics, Series 3, Analytical and Epidemiological Studies, 38, 1–105 PMID:27023287
  • Heid, A. R., Eshraghi, K., Duntzee, C. I., Abbott, K., Curyto, K. & Van Haitsma, K. (2016). “It depends”: Reasons why nursing home residents change their minds about care preferences. The Gerontologist, 56(2), 243–255 doi:10.1093/geront/gnu040 [CrossRef] PMID:24814829
  • Heid, A. R., Van Haitsma, K., Kleban, M., Rovine, M. J. & Abbott, K. M. (2017). Examining clinical predictors of change in recreational preference congruence among nursing home residents over time. Journal of Applied Gerontology, 36(11), 1351–1369 doi:10.1177/0733464815617288 [CrossRef] PMID:26620057
  • Koren, M. J. (2010). Person-centered care for nursing home residents: The culture-change movement. Health Affairs (Project Hope), 29(2), 312–317 doi:10.1377/hlthaff.2009.0966 [CrossRef] PMID:20056692
  • Li, J. & Porock, D. (2014). Resident outcomes of person-centered care in long-term care: A narrative review of interventional research. International Journal of Nursing Studies, 51(10), 1395–1415 doi:10.1016/j.ijnurstu.2014.04.003 [CrossRef] PMID:24815772
  • McCarthy, J. F., Blow, F. C. & Kales, H. C. (2004). Disruptive behaviors in Veterans Affairs nursing home residents: How different are residents with serious mental illness?Journal of the American Geriatrics Society, 52(12), 2031–2038 doi:10.1111/j.1532-5415.2004.52559.x [CrossRef] PMID:15571538
  • Morris, J. N., Hawes, C., Fries, B. E., Phillips, C. D., Mor, V., Katz, S. & Friedlob, A. S. (1990). Designing the national resident assessment instrument for nursing homes. The Gerontologist, 30(3), 293–307 doi:10.1093/geront/30.3.293 [CrossRef] PMID:2354790
  • Shay, K. & Burris, J. F. (2008). Setting the stage for a new strategic plan for geriatrics and extended care in the Veterans Health Administration: Summary of the 2008 VA state of the art conference, “The changing faces of geriatrics and extended care: Meeting the needs of veterans in the next decade”. Journal of the American Geriatrics Society, 56(12), 2330–2339 doi:10.1111/j.1532-5415.2008.02079.x [CrossRef] PMID:19093933
  • Sjögren, K., Lindkvist, M., Sandman, P. O., Zingmark, K. & Edvardsson, D. (2013). Person-centredness and its association with resident well-being in dementia care units. Journal of Advanced Nursing, 69(10), 2196–2205 doi:10.1111/jan.12085 [CrossRef] PMID:23331281
  • Teng, E. L. & Chui, H. C. (1987). The modified mini-mental state (3MS) examination. The Journal of Clinical Psychiatry, 48(8), 314–318 PMID:3611032
  • Tombaugh, T. N., McDowell, I., Kristjansson, B. & Hubley, A. M. (1996). Mini-mental state examination (MMSE) and the modified MMSE (3MS): A psychometric comparison and normative data. Psychological Assessment, 8(1), 48–59 doi:10.1037/1040-3590.8.1.48 [CrossRef]
  • Van Haitsma, K., Abbott, K. M., Arbogast, A., Bangerter, L. R., Heid, A. R., Behrens, L. L. & Madrigal, C. (2019). A preference-based model of care: An integrative theoretical model of the role of preferences in person-centered care. The Gerontologist. Advance online publication. doi:10.1093/geront/gnz075 [CrossRef] PMID:31152589
  • Van Haitsma, K., Abbott, K. M., Heid, A. R., Carpenter, B., Curyto, K., Kleban, M. & Spector, A. (2014). The consistency of self-reported preferences for everyday living: Implications for person-centered care delivery. Journal of Gerontological Nursing, 40(10), 34–46 doi:10.3928/00989134-20140820-01 [CrossRef] PMID:25199153
  • Van Haitsma, K., Curyto, K., Spector, A., Towsley, G., Kleban, M., Carpenter, B. & Koren, M. J. (2013). The Preferences for Everyday Living Inventory: Scale development and description of psychosocial preferences responses in community-dwelling elders. The Gerontologist, 53(4), 582–595 doi:10.1093/geront/gns102 [CrossRef] PMID:22936532

Participant Demographics

VariableVA CLC Residents (n = 42)Non-VA NH Residents (n = 37)χ2t test
Age (years) (mean, SD, range)74.79 (11.1) (55 to 94)81.62 (11.83) (55 to 101)2.646*
Gender (male) (n, %)41 (97.6)12 (32.4)37.857***
Education (% high school) (n, %)33 (78.6)33 (89.1)1.613
Ethnicity (n, %)
  Not Hispanic or Latino100 (42)100 (37)
Race (n, %)
  Caucasian39 (92.9)36 (97.3)0.806
  African American3 (7.1)1 (2.7)
Marital status (n, %)
  Married11 (26.2)**2 (5.4)**16.233**
  Divorced/separated9 (21.4)4 (10.8)
  Widowed11 (26.2)**26 (70.3)**
  Never married11 (26.2)5 (13.5)
Length of stay (days) (mean, SD, range)393.2 (853.5) (5 to 3,663)809.5 (820) (15 to 3,115)2.203*
Veteran status (n, %)100 (42)3 (8.1)67.754***
MMSE total score (0 to 30) (mean, SD, range)7 (3.2) (19 to 30)26.3 (1.9) (22 to 30)1.163
MMSE adjusted score for age and education (mean, SD, range)26.8 (3.1) (19 to 31)26.8 (2.2) (21 to 31)<0.001***
Perfect agreement scores (%) (mean, SD, range)56.15 (0.11) (32 to 76)65.75 (0.13) (43 to 94)3.704***
Acceptable ±1 agreement scores (%) (mean, SD, range)88.5 (0.06) (70 to 99)95.1 (0.05) (74 to 100)5.385***

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Non-Veteran Nursing Home Residents (n = 37)

Significant Individual Regressions Between Percent Agreement and PredictorsMultiple R
F (p)Adj. R2Adj. R2F (d, f)p Value
Perfect agreement predictor
  Age7.03 (0.012)0.140.386.80 (2, 17)0.007
  Walking in corridors5.28 (0.034)0.18
Acceptable agreement predictor
  Gender6.34 (0.017)0.130.339.83 (2, 34)<0.001
  Level of expressing thoughts12.54 (0.001)0.24

Multiple Regression Between Perfect and Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors in Veteran Residents (n = 42)

Significant Individual Regressions Between Percent Agreement and PredictorsMultiple R
F (p)Adj. R2Adj. R2F (d, f)p Value
Perfect agreement predictor
  Hearing impairment5.47 (0.024)0.100.293.73 (6, 35)0.006
  3MS percentile score4.97 (0.032)0.09
  Level of understanding4.50(0.040)0.08
  Level of distraction4.23 (0.046)0.07
  Level of energy6.78 (0.013)0.12
  Level of effort4.90 (0.033)0.09
Acceptable agreement predictor
  Vision impairment5.68 (0.022)0.100.255.62 (3, 38)0.003
  3MS attention items8.40 (0.006)0.15
  MMSE total score5.92 (0.020)0.11

Multiple Regression Between Acceptable ±1 Percentage Agreement With Baseline and Repeated PELI Ratings and Potential Predictors for All Residents (N = 79)

PredictorSignificant Individual Regressions Between Perfect Percent Agreement and PredictorsMultiple R
F (p)Adj. R2Adj. R2F (d, f)p Value
Gender16.29 (<0.001)0.160.287.08 (5, 72)<0.001
Level of expressing thoughts4.22 (0.043)0.04
Level of distraction10.12 (0.004)0.09
MMSE adjusted score6.38 (0.014)0.07
Veteran status15.39 (<0.001)0.16
Authors

Dr. Curyto is CLC Psychologist, Research Scientist, Veterans Administration, Western New York Healthcare Service, Batavia, New York; Dr. Dockler is Postdoctoral Geropsychology Fellow, Veterans Administration, Pittsburgh Healthcare System, Pittsburgh, Pennsylvania; and Dr. Van Haitsma is Associate Professor, Director, Program for Person Centered Living Systems of Care, College of Nursing, Pennsylvania State University, University Park, Pennsylvania.

The authors have disclosed potential conflicts of interest. The content and views presented are those of the current authors and do not represent the views of the Department of Veterans Affairs or the United States Government and do not necessarily represent the official views of the National Institute of Nursing Research or the National Institutes of Health. The current project is based on work supported in part by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development and the National Institute of Nursing Research (R21NR011334).

The authors thank Leslie McCulloch, Karen Eshragi, Christina Duntzee, and Borah Kim for their efforts at data collection, cleaning, analysis, and reporting.

Address correspondence to Kimberly Curyto, PhD, CLC Psychologist, Research Scientist, Veterans Administration, Western New York Healthcare Service, 222 Richmond Avenue, Batavia, NY 14020; e-mail: kimberly.curyto@va.gov.

Received: June 12, 2019
Accepted: October 25, 2019

10.3928/00989134-20200129-06

Sign up to receive

Journal E-contents