Psychometric Evaluation of the Team Member Perspectives of Person-Centered Care (TM-PCC) Survey for Long-Term Care Homes

Person-centered care (PCC) is fundamental for providing high-quality care in long-term care homes. This study aimed to evaluate the psychometric properties of an 11-item Team Member Perspectives of Person-Centered Care (TM-PCC) survey, adapted from White and colleagues (2008). In a cross-sectional study, 461 staff from four long-term care homes in Ontario, Canada, completed the TM-PCC. Construct validity and internal consistency of the TM-PCC were examined with a principal component analysis and Cronbach’s alpha coefficient. Findings revealed a three-component structure with factor 1, Supporting Social Relationships; factor 2, Familiarity with Residents’ Preferences; and factor 3, Meaningful Resident–Staff Relationships. The TM-PCC, as compared to the original survey, presented with less components (i.e., did not address Resident Autonomy, Personhood, Comfort, Work with Residents, Personal Environment, and Management Structure), yet included one new component (Meaningful Resident–Staff Relationships). The TM-PCC has a similar internal consistency (Cronbach’s alpha coefficient 0.82 vs. White et al. 0.74–0.91). The TM-PCC can be used to assess PCC from the staff’s perspective in long-term care homes.


Introduction
Long-term care homes (also referred to as nursing homes) provide medical care and activities of daily living (ADL) support to over 1.4 million Americans and more than 224,000 Canadians [1,2]. Many of these seniors are frail and have complex medical conditions, and therefore depend on staff assistance [1,3]. Person-centered care (PCC) has been recognized as a fundamental element in providing high-quality care tailored to residents' needs [4,5]. PCC is generally defined as care which is responsive to the person's well-being, which includes meaningful social interactions, shared choice and decision-making, The second tool, the Person-Centered Care Assessment tool, was designed in Europe for completion by staff and includes 13 items on personalized care, and organizational and environmental support [28]. This tool has a high internal consistency reliability (Cronbach's α 0.83) and validity [28]. The original version (developed and tested in Australia) had a limited response rate of 21% in the first validation study and one of the subscales did not reach the Cronbach's alpha cut-off criterion of 0.7. This version was subsequently adjusted for use in Europe, and upon testing, had a high response rate (88%) and some evidence of reliability and validity [28].
In summary, while there are many PCC assessment tools available, few (<20) are validated and specific to long-term care homes. Because of the complex nature of long-term care in Canada, we deemed it necessary to develop a context-specific PCC tool. The Person-Directed Care Measure [27] has a high internal consistency and reliability, indicating redundancy amongst its items. The Person-Centered Care Assessment tool [28] is limited in its use because of under-representation in North American studies. A parsimonious, validated assessment tool specific to Canadian long-term care homes would aid in the feasible measurement and evaluation of PCC practices as perceived by staff.
We adapted the Person-Directed Care Measure [27] to develop the 'Team Member Perspectives of Person-Centered Care' (TM-PCC) survey, a brief and validated tool to assess and evaluate dimensions of PCC. The choice to base the TM-PCC on the Person-Directed Care Measure [27] was built on the fact that this tool is commonly used, yet needed to be shortened by removing redundant items [12]. There are several advantages to administering a short survey to long-term care home staff: increased likeliness of completion, respondents are more likely to be engaged throughout the survey, and less competition for staff's already limited availability. Additionally, the Person-Directed Care Measure [27] covers a wide range of PCC practices and has shown strong psychometric properties. Therefore, the primary objectives of this study were to adapt the Person-Directed Care Measure [27] to develop the TM-PCC survey and test the TM-PCC survey's psychometric properties with a group of long-term care home staff. Specifically, we aimed to evaluate internal consistency and construct validity. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Tri-Council Ethics Board at the agency (REB-118).

Tool Development
The original 64-item Person-Directed Care Measure [27] includes eight PCC dimensions. The measure was designed for staff to rate their own care practices. Each item was rated on a Likert-type scale as 1 = 'Never', 2 = 'Rarely', 3 = 'Sometimes', 4 = 'Almost all the time', 5 = 'All the time'.
An expert panel consisting of long-term care home staff, residents and families, administrators and researchers selected a subset of 11 items from the Person-Directed Care Measure [27]. The 11 items were selected based on the perceived importance of which the item specifically addressed PCC practices, validation of the item's importance in the literature, and on the individual item's psychometric performance. A key resource was the work of Bangerter and colleagues (2015), which surveyed long-term care home residents to identify their preferences with respect to care interactions and which emphasized the importance of care partners knowing the residents' preferences [29]. In addition, the expert panel and the literature review supported the decision to create three new items to capture the relationships between staff and residents; an important aspect of PCC not fully addressed in the Person-Directed Care Measure [27]. Specifically, White and colleagues (2008) addressed supportive relationships, which focused on friends and family, but did not include the relationships between residents and staff in their Person-Directed Care Measure [27]. Other tools have incorporated items on the staff-resident relationship [20,22,23], further strengthening the importance for inclusion. Table 2 provides an overview of supporting references for the items [21,[29][30][31][32][33][34][35].
The final TM-PCC survey included a total of 14 items exploring PCC practices from the perspective of the staff. Each item was rated similarly to the original survey on a Likert-type scale as 1 = 'Never', 2 = 'Rarely', 3 = 'Sometimes', 4 = 'Almost all the time', 5 = 'All the time'. Estimated completion time was five to ten minutes. The TM-PCC survey is designed to be completed by care team members, specifically registered nurses (RNs), licensed practical nurses (LPNs) and certified nursing assistants (CNAs). Although CNAs provide the majority of residents' care, all care members were included in the sample, as they collaborate as a team. RNs are typically involved when residents' care needs are not well-defined, when health conditions are not well-controlled or require frequent monitoring, and when outcomes are unpredictable [36]. LPNs support residents with well-defined care needs with little fluctuation in conditions and when there is a low risk of negative outcomes [36]. CNAs provide assistance with residents' activities of daily living and provide the most direct care time [14]. The readability of items was assessed using Flesch reading easy score, which is widely established as an accurate measure [37]. Readability was rated as 'standard' by the Flesch reading easy score suggesting that the content of the items was understandable at an 8th to 9th grade level [37].

Sampling Strategy
The staff sample was randomly selected from 4 long-term care homes in the province of Ontario, Canada (long-term care homes were assigned numbers and units were assigned letters). Eight of the fifteen surveyed units were dementia care units. Research assistants approached 750 care staff (RNs, LPNs and CNAs) to obtain written consent and invited them to conduct the TM-PCC survey. Data were collected from 1 July 2014 to 30 June 2015. As this was part of a larger study, more details on the study methods and sampling are described elsewhere [38].

Analytical Strategy
Only complete data were included in the analysis. Of the invited care staff (n = 750), 461 completed a TM-PCC (response rate 61.5%). Eighty-two (17.8%) surveys were not usable due to missing information. Of the remaining surveys, 379 (82.2%), descriptive statistics were calculated for respondents' demographics and for each item on the TM-PCC, as well as the subscales reflecting PCC dimensions. Principal component analysis was chosen a priori to evaluate construct validity because it is a variable reduction technique for large samples when variables are highly correlated; it generates component scores which are a linear combination of the observed variables [39]. Unlike factor analysis, it does not assume underlying latent constructs in the data [39]. Factors were extracted based on the Kaiser criterion (eigenvalues of factors >1), inspection of the scree plot, and percentage of variance explained (>10%) [40]. Items were assigned by component loadings of 0.40 or greater and deleted if they loaded on more than one component where the difference was <0.20 [40]. Results are presented with orthogonal rotation for interpretability and to preserve variable communalities [41]. Internal consistency was examined using a Cronbach's alpha coefficient. All data analysis was performed using SAS 9.4 software (SAS Institute, Cary, NC, USA).

TM-PCC Survey Results
A total of 750 staff were invited to participate in the surveys. Several participants (289, 38.5%) did not consent, leaving a total of 461 survey responses: 33 (7.1%) RNs, 79 (17.1%) LPNs and 349 (75.7%) CNAs. The majority of participants were employed on a full-time basis (83.1%). The ratio of full-time and part-time did not appear to vary by long-term care home. With respect to years of participant experience working in long-term care homes, the number of staff with over ten years of experience ranged from 5.6% to 32.4% across long-term care homes. A significant proportion of participants had no more than three years of experience in the current long-term care home (26.0-53.9%). Generally, most staff were female (89.8-94.7%). Most staff appeared to be between 25 and 45 years of age (35.5-84.0%).
Areas of PCC which all staff rated highest included item 4 (i.e., "I quickly help __ of my residents to the toilet when they request or need help"), with a score of 4.2-4.4 out of five across four long-term care homes (Table 3). Item 12 (i.e., "I look after the same residents from day to day") was rated 4.1-4.3 out of five. Item 15 (i.e., "I am able to build fulfilling relationships with residents") was rated 4.0-4.3 out of five. Lastly, item 16 (i.e., "I can learn from residents and their family members and incorporate this caring into my daily routine") was rated 4.0-4.2 out of five. Item 8 rated the lowest (i.e., "I help __ of my residents stay connected to previous associations") with a rate of 2.2-2.8 out of five. All TM-PCC survey scores are described in Table 3.

Component Structure
In the first principal component analysis, three items (item 5: "I know when __ of my residents need to use the toilet, even if they cannot speak", item 6: "I can calm __ of my residents if they feel agitated or upset", and item 11: "I spend time talking or just being with __ of my residents") each loaded on two components. The difference in the loading was <0.2, suggesting that these three items could be potentially removed. Although these items are conceptually important, when removing these items the eigenvalues and inspection of the scree plot suggested a three-component structure for the TM-PCC survey. The component pattern matrix is displayed in Table 4. The cumulative variance explained by the three components was 61.5%.  There were three items which loaded on the third component. This component contributed 11.1% to the total explained variance and the eigenvalue was 1.2. Component loadings of item 9: 'I look after the same residents from day to day', item 10: 'I am able to build fulfilling relationships with residents', and item 11: 'I can learn from residents and their family members and incorporate this caring into my daily routine', ranged from 0.63 to 0.84. As a result, this component was named Meaningful Resident-Staff Relationships.

Internal Consistency Reliability
Communality estimates for each item ranged from 0.41 to 0.77. Cronbach's alpha coefficient for all 11 items was 0.82. Internal consistency for the three components was generally high and ranged from 0.62 to 0.83. Average scores in each construct were high and ranged from 3.56 to 4.17 out of five. The descriptive statistics and Cronbach's alpha coefficient for each of the components are described in Table 5.

Discussion
Given the importance of PCC practices to enhance care quality for residents in long-term care homes, few validated tools are developed specifically to measure PCC in these settings. The need to develop a brief, valid and reliable tool to measure a wide range of PCC practices led to the development of the TM-PCC survey. The development was based on theoretical and evidence-based items, founded in the Person-Directed Care Measure [27], as well as a literature review, and decided by an expert panel to be specific to PCC in long-term care homes. The goal of this study was to develop and validate a tool to measure PCC practices to be able to assess implementation and evaluation of PCC interventions. TM-PCC survey data were collected from 461 staff in four long-term care homes to test the survey for its psychometric properties.
In terms of respondents, there was high variability across units and homes in terms of years of experience. Other staff characteristics such as age, gender and shift type did not appear to vary much across units or homes, with a predominantly female staff employed on full-time shifts. A lack of research on the variability of long-term care home staff prevents larger comparisons. However, these demographics findings reflect what is generally reported in practice and suggest that the results are generalizable to other long-term care homes.
The analysis provided strong evidence to support the internal consistency and construct validity of the TM-PCC survey. A principal component analysis revealed a three-factor structure, with the cumulative variance explaining 61.5%. The internal consistency was high with a Cronbach's alpha of 0.82. These findings compare favorably with the Person-Directed Care Measure [27], which demonstrated a Cronbach's alpha coefficient of 0.74-0.91 and variance explained of 61%.
For the TM-PCC, the first component, Supporting Social Relationships presented with high loading from four items related to helping residents stay connected to friends and family. It explained 36.5% of the total variance. Residents value meaningful relationships and are dissatisfied when there are no such opportunities [42]. Similar to the TM-PCC survey, Sidani and colleagues (2014) included social relationships as part of a larger holistic care construct. Additionally, social environment and individual aspects of the person were determined as important principles central to PCC [5]. Maintaining social relationships is a recognized aspect of care models, building on relational and humanistic theories [42,43].
The second TM-PCC survey component, Familiarity with Residents' Preferences presented with a high loading on four items related to residents' preferred habits, food and music. This component explained 13.9% of the total variance. Knowing and accommodating residents' preferences is central to the idea of PCC [5,13]. The third component, Meaningful Resident-Staff Relationships, had high loadings from three items related to consistent care assignment, learning from residents and building relationships with residents. It contributed 11.1% to the total explained variance. In a systematic review by Kogan and colleagues [5], facilitating enriched relationships was an important value within the definition of PCC. Emotional caring and mutual sharing of personal information is central to residents feeling satisfied with their care and well-being [42]. De Brouwer and colleagues (2017) included constructs on relationships in their scale. The nature of the relationship between staff and residents is a determinant of care delivery: understanding the resident as a person engages staff to discover residents' preferences and provide responsive care [42]. Developing meaningful relationships depends on staff having sufficient time to engage with residents, as well as requires consistent care assignments, relating to the construct on organizational support [42].
A few limitations need to be noted for this study. Although the TM-PCC survey did not include items related to resident autonomy, shared decision-making, or resident engagement in activities, it did emphasize elements of PCC such as meaningful social interactions and recognizing residents' preferences [4,5]. Another limitation of the study relates to the absence of test-retest validity testing. When compared with other tools designed by researchers, the TM-PCC survey performs moderately well. The internal consistency and percent variance explained were high when compared to other tools [15,16,19,21,22,26]. Additionally, the TM-PCC employed Likert-type scales which are simple to construct and easy to complete for staff; however, this style of questioning can be subject to central tendency bias (i.e., where participants avoid extreme response categories) which could make it difficult to monitor improvement in scores over time. As well, the choice to perform a principal component analysis allowed for reducing the number of observed variables into a smaller number that accounted for most of the variance in the dataset and made no assumptions concerning an underlying causal structure. However, this strategy required a large sample size (i.e., ideally five times the number of survey items) and dropped some items, which is why we included multiple items for each construct we hoped to measure [39].
Additionally, there was a significant number of staff who chose not to participate in the study which could have biased results. Of the participants who did not consent, ten staff indicated some demographic information; however, this is not a large enough sample to explain why some staff chose not to consent. Because completing research surveys requires time away from residents, it is unclear if staff who chose not to consent were less motivated to provide person-centered care than staff who completed surveys. This further emphasizes the need for surveys to be concise in order to maximize completion rates and ease the burden on care staff. As well, the TM-PCC evaluated staff perspectives only. Future tools could expand upon resident and family perspectives.
We also found that staff scores were generally high, suggesting a potential social desirability bias. One could emphasize to staff that the survey's purpose is to identify areas where one can improve rather than judging an individual's performance. This likely had minimal effects on the validity of the tool since decreased variability in scores would decrease the risk of a type 1 error. As the survey was only administered to four long-term care homes in one Canadian province, results may vary among countries, long-term care home models, more or less experienced staff, different staffing ratios, or variations in PCC operationalization. In this study, we did not focus on non-nursing staff responses (i.e., other staff on the team such as kinesiologists, housekeeping, physicians, etc.). Finally, perceptions from the participants about the acceptability of completing the measure were not sought. However, this is valuable information and will be looked at in further research for the future utility of TM-PCC.
In summary, the TM-PCC survey is a brief, validated tool to assess PCC practices in long-term care homes from a staff's perspective. The survey includes similar constructs as other tools and has a high internal consistency and construct validity. The three components in the adapted TM-PCC survey mirror key principles of PCC that were described in a literature review completed by Kogan and colleagues (2015).
It can be used in long-term care homes interested in monitoring or implementing PCC interventions. Using validated tools such as the TM-PCC survey promotes accurate representations of the PCC delivered in the care settings as well as staff's perceptions towards PCC. The results of this study can help identify staff, teams and leadership in long-term care homes to identify practices where teams are confident in PCC, as well as those where supports may be needed.

Conclusions
The TM-PCC survey was considered a psychometrically valid tool for use in long-term care homes. The use of valid, standardized tools allows for the comparison of care between different care settings as well as the comparison of changes in care practices within the same setting.