Abstract

Aiming to translate, culturally adapt, and psychometrically evaluate the Person-centred Practice Inventory - Staff (PCPI-S) for Portuguese healthcare professionals, this methodological study was conducted sequentially in two phases. Phase I followed the 10-steps recommendations from the ISPOR taskforce for translation and cultural adaptation of patient reported outcome measures. Phase II comprised a quantitative cross-sectional virtual survey of the translated PCPI-S with healthcare professionals, who were reached through snowball sampling from both primary and specialized care settings. The psychometric properties of the PCPI-S were determined by assessing reliability and construct validity. A sample of 304 healthcare professionals participated in Phase II. Ceiling effects were found. The overall internal consistency was excellent (> 0.9). The confirmatory factor analysis showed a good model fit after minor modifications, revealing construct validity, and supporting the theoretical framework. In conclusion, the three-factorial model of PCPI-S adjusted to the studied sample is a valid and reliable instrument to assess the perceptions of healthcare professionals on person-centred practice in various Portuguese clinical contexts. Considering the ceiling effects, the effect of social desirability should be explored.

Key words:
Patient-centered care; Psychometrics; Health personnel; Portugal

Resumo

Com o objetivo de traduzir, adaptar culturalmente e avaliar psicometricamente o Inventário para a Prática Centrada na Pessoa para profissionais de saúde (PCPI-S) em diversos contextos de prestação de cuidados, este estudo metodológico realizou-se em duas fases sequenciais. A Fase I seguiu as recomendações de dez etapas da taskforce da ISPOR para tradução e adaptação cultural de medidas de resultados auto reportados. A Fase II incluiu um estudo cross-sectional do PCPI-S traduzido com profissionais de saúde, que foram alcançados por meio de amostragem snowball em contextos de cuidados primários e diferenciados. A psicometria do PCPI-S foi analisada pela avaliação da confiabilidade e validade de construto. Uma amostra de 304 profissionais de saúde participou da Fase II. Efeitos de teto foram encontrados. A consistência interna geral foi excelente (> 0,9). A análise fatorial confirmatória mostrou um bom ajuste do modelo e validade de construto, refletindo o referencial teórico. Concluindo, o modelo tri-fatorial do PCPI-S ajustado à amostra estudada é um instrumento válido e fiável para avaliar as percepções dos profissionais de saúde sobre a prática centrada na pessoa em vários contextos de cuidados portugueses. Considerando os efeitos teto, a desejabilidade social deve ser explorada.

Palavras-chave:
Assistência centrada no paciente; Psicometria; Profissionais da saúde; Portugal

Introduction

Healthcare services have been transitioning from biomedical models of care delivery to a person-centered practice, with evidence of gains to both the healthcare service users and professionals, as well as to the organisation¹1 Gyllensten H, Björkman I, Jakobsson Ung E, Ekman I, Jakobsson S. A national research centre for the evaluation and implementation of person-centred care: content from the first interventional studies. Health Expect 2020; 23(5):1362-1375.. The importance of person-centred practice (PCP) to the sustainability of healthcare systems is recognized worldwide by politicians and healthcare managers and is becoming the gold standard worldwide towards ensuring quality of care in a costs-constrained society²2 World Health Organization. WHO global strategy on people-centred and integrated health services: interim report. Geneva: WHO; 2015.. Having ascertained PCP effectiveness, several research efforts are being driven towards prompting its implementation into mainstream healthcare practice. Yet, some challenges remain that hinder the successful implementation, long-term follow-up, and sustainability.

The absence of standardised measurement instruments that enable the assessment of person-centred processes and outcomes is a bottleneck with negative impact to its implementation³3 McCormack LA, Treiman K, Rupert D, Williams-Piehota P, Nadler E, Arora NK, Lawrence W, Street RL Jr. Measuring patient-centered communication in cancer care: a literature review and the development of a systematic approach. Soc Sci Med 2011; 72(7):1085-1095.. The assessment is further complicated by the need to account for the perspectives of all those involved in the therapeutical encounter, i.e., service users and healthcare professionals, as well as the specificities of the multi-level contexts where that encounter occurs, i.e. micro-context (care setting), meso-context (healthcare organizations) and macro-context (healthcare strategy & policy makers)⁴4 Krawczyk M, Sawatzky R, Schick-Makaroff K, Stajduhar K, Öhlen J, Reimer-Kirkham S, Mercedes Laforest E, Cohen R. Micro-meso-macro practice tensions in using patient-reported outcome and experience measures in hospital palliative care. Qual Health Res 2019; 29(4):510-521..

The evidence produced in the last decade on the theoretical underpinning of PCP and agreed definitions of PCP, allowed overcoming some of the difficulties in developing approaches to measurement and evaluation⁵5 Phelan A, McCormack B, Dewing J, Brown D, Cardiff S, Cook NF, Dickson CAW, Kmetec S, Lorber M, Magowan R, McCance T, Skovdahl K, Štiglic G, van Lieshout F. Review of developments in person-centred healthcare. Int Pract Dev J 2020; 10:3.. The Person-centred Practice Framework (PcPF) is one such theoretical structure to PCP. Anchored in a person-centeredness philosophy, the PcPF is a middle-range theory that acknowledges several domains towards PCP⁶6 McCormack B, McCance T, Bulley C, Brown D, McMillan A, Martin S. Fundamentals of person-centred healthcare practice. New Jersey: John Wiley & Sons; 2021.. Starting at the macro-level with policies and leadership strategies, it further depicts the prerequisites of the healthcare professional as a person, the care environment, including its culture, and the care processes. Within each of these domains, subsequent elements are identified towards accomplishing the ultimate outcome of a healthful culture, i.e., good care experiences, care involvement and well-being. A healthful culture therefore is one in which each person flourishes, i.e., recognise the value of the other and continuously develop and evolve in relationship despite their role in the organisation (e.g. manager, nurse, physician, patient)⁶6 McCormack B, McCance T, Bulley C, Brown D, McMillan A, Martin S. Fundamentals of person-centred healthcare practice. New Jersey: John Wiley & Sons; 2021..

Derived from the PcPF, the Person-Centred Practice Inventory (PCPI) was developed to capture the perspectives of person-centredness from staff as well as healthcare service users. Specifically focusing on the staff version (i.e., PCPI-S), the instrument has been developed with the main goal of assisting quality improvement of healthcare practices towards person-centredness. It is composed of 59 items that allow the self-assessment of the domains of pre-requisites, care environment and person-centred processes on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree).

The instrument is intended to be suitable for all healthcare professionals across various healthcare settings. It is a valid and reliable measure that reflects the PcPF theoretical structure, where all factor loadings were found to be statistically significant (P < 0.05) and ranging from 0.417 to 0.921⁷7 Slater P, McCance T, McCormack B. The development and testing of the Person-centred Practice Inventory-Staff (PCPI-S). Int J Qual Health Care 2017; 29(4):541-547.. It has been translated to many languages with evidence of psychometric robustness across culturally different contexts⁸8 Bing-Jonsson PC, Slater P, McCormack B, Fagerstrom L. Norwegian translation, cultural adaption and testing of the Person-centred Practice Inventory - Staff (PCPI-S). BMC Health Serv Res 2018; 18(1):555.

9 Weis ML, Wallner M, Köck-Hódi S, Hildebrandt C, McCormack B, Mayer H. German translation, cultural adaptation and testing of the Person-centred Practice Inventory-Staff (PCPI-S). Nursing Open 2020; 7(5):1400-1411.^-1010 Balqis-Ali NZ, Saw PS, Jailani AS, Fun WH, Mohd Saleh N, Tengku Bahanuddin TPZ, Sararaks S, Lee SWH. Cross-cultural adaptation and exploratory factor analysis of the Person-centred Practice Inventory-Staff (PCPI-S) questionnaire among Malaysian primary healthcare providers. BMC Health Serv Res 2021; 21(1):32..

The Individualized Care Scale (ICS)¹¹11 Suhonen R, Leino-Kilpi H, Välimäki M. Development and psychometric properties of the Individualized Care Scale. J Eval Clin Pract 2005; 11(1):7-20. has been translated and validated to the Portuguese healthcare context to assess the perceptions about the individuality of care from the perceptions of both nurses and patients. Although this instrument was found to be valid to the Portuguese population with good internal consistency¹²12 Amaral AFS, Suhonen R. Translation and validation of the individualized care scale. Int J Caring Sci 2014; 7(1):90., and is grounded on a theoretical model, it does not allow to capture the perspectives of other healthcare professionals rather than nurses and, has a particular focus on the hospital setting. The PCPI-S will complement this measurement instrument has it allows to gather the multi-professional perspective at the various levels of an institution, as well as, transversely across healthcare settings (i.e., primary and differentiated care), which is essential to improve person-centeredness of healthcare.

This study aimed to translate, culturally adapt, and psychometrically evaluate the Person-centered Practice Inventory - Staff (PCPI-S) for Portuguese healthcare professionals in various healthcare contexts.

Method

Study design

This methodological study followed a two-stage sequential research design entailing the translation and cultural adaptation of the PCPI-S from English to European Portuguese (i.e., phase I) and the subsequent psychometric evaluation with a cross-sectional quantitative approach (i.e., phase II). The study was approved by The Ethics Committee of The Health Sciences Research Unit: Nursing (Ref. Nr. 674/05-2020).

Phase I: Translation and cultural adaptation

The principles of good practice in translation and cultural adaptation approved by the International Society for Pharmacoeconomics and Outcome Research (ISPOR) guided the phase I processes¹³13 Wild D, Grove A, Martin M, Eremenco S, McElroy S, Verjee-Lorenz A, Erikson P; ISPOR Task Force for Translation and Cultural Adaptation. Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: report of the ISPOR task force for translation and cultural adaptation. Value Health 2005; 8(2):94-104.. The work was conducted according to ten recommended steps as described in detail in Chart 1.

The cognitive debriefing in step seven was conducted according to Willis¹⁴14 Willis GB. Cognitive interviewing revisited: a useful technique, in theory. Methods for testing and evaluating survey questionnaires. In: Presser S, Rothgeb JM, Couper MP, Lessler JT, Martin E, Martin J, Singer E. Methods for testing and evaluating survey questionnaires. New Jersey: John Wiley & Sons; 2004. p. 23-43.. Following a non-probabilistic convenience sample enhanced by a snowball approach, participants were identified and invited to participate in the study. Prospective participants received an email with the study’s written information and were asked to reply with a suitable date and time for the interview. They then received a link to the virtual meeting platform. Accordingly, at the beginning of each interview, the participant received verbal information about the study that explained the nature of their participation. They were informed they could leave at any time without justification and were granted anonymity as no personal identification was collected, and confidentiality of perspectives offered was assured. Each participant who agreed to participate in the individual cognitive interview was granted access to a virtual form containing the informed consent, socio-demographic variables, and the harmonized version of the PCPI-S. Having completed the written informed consent, the participant was instructed on the reflexive activity and the interviewer explained the process by referring to questions drawn from the nationwide census. The interview was initiated with an introductory question to gather the perspectives of each participant about person-centered practice. Thereafter each item from the PCPI-S was read aloud. The interviewer carried out retrospective probing, on specific wordings or when there was a pause while reading the item and asked for alternative formulations when the participant considered the item problematic. All sessions were audio-recorded, transcribed verbatim and analyzed for discrepancies and similarities of interpretation of concepts across professions and healthcare settings, as well as cultural inadequacies¹⁴14 Willis GB. Cognitive interviewing revisited: a useful technique, in theory. Methods for testing and evaluating survey questionnaires. In: Presser S, Rothgeb JM, Couper MP, Lessler JT, Martin E, Martin J, Singer E. Methods for testing and evaluating survey questionnaires. New Jersey: John Wiley & Sons; 2004. p. 23-43..

Phase II: Psychometric evaluation

Quantitative cross-sectional research was carried out to generate a dataset for psychometric validation of the Portuguese version of the PCPI-S. The sampling process followed a convenience and snowball strategy to reach as many participants as possible and maximize variation with respect to the healthcare setting, professional category, and length of professional experience. A virtual version of the translated PCPI-S was disseminated through the Portuguese professional organizations of nursing, physiotherapy, occupational therapy, as well as professional groups of psychologists and physicians. Additionally, the members of the research team disseminated the study invitation through their professional network and personal social networks. The research team sent out reminders every two-weeks during the three-month data-collection period. Healthcare professionals who were actively delivering care in a Portuguese healthcare setting were invited to answer the virtual survey, which was composed of three sections: i) information about the study and request for informed consent, ii) socio-demographic and professional variables, and iii) the translated and culturally adapted PCPI-S. Anonymity and confidentiality were assured at this stage since no personal or Internet-related elements were collected from respondents that could act as identifiers. Through the written information, participants were informed about the research project, the meaning of their participation, their right to withdraw from the study at any time without any repercussions, and their right to access the study findings by contacting the principal investigator. Respondents made their informed consent explicit by answering the first section, which was set as mandatory before proceeding to answer the remaining sections. Data collection occurred from January to March 2022, during which time a 10:1 ratio of respondent to item was achieved¹⁵15 Nunnally JC. An overview of psychological measurement. In: Wolman BB. Clinical Diagnosis of Mental Disorders. Boston: Springer; 1978. p. 97-146.. To maximize this ratio and in line with other PCPI-S validation procedures (e.g., Bing-Jonsson et al.^8, Weis et al.⁹9 Weis ML, Wallner M, Köck-Hódi S, Hildebrandt C, McCormack B, Mayer H. German translation, cultural adaptation and testing of the Person-centred Practice Inventory-Staff (PCPI-S). Nursing Open 2020; 7(5):1400-1411.), analyses were made independently for the three constructs (prerequisites, care environment, person-centered processes).

Data analysis

The software package SPSS statistics version 24.0 and SPSS AMOS version 21.0 were used to conduct data analysis. Descriptive statistics of respondents’ characteristics and items were determined, including the items’ floor and ceiling effects. Missing data analysis for the 59 items was performed. Reliability analysis included the internal consistency, using Cronbach’s alpha for both total scale and individual constructs and domains. Values between 0.70 and 0.79 were considered evidence of acceptable internal reliability, between 0.80 and 0.89 indicated good internal reliability, and values of 0.90 or higher showed excellent internal reliability¹⁶16 Evers A, Hagemeister C, Høstmælingen A, Lindley P, Muñiz J, Sjöberg A. EFPA review model for the description and evaluation of psychological and educational tests. Test review form and notes for reviewers Version 42. 2013.. Construct validity was determined by (i) item-construct and item-total correlations, and (ii) exploratory (EFA) and confirmatory factor analysis (CFA). Correlations were judged as inadequate if < 0.20, sufficient if ranging from 0.20-0.34, good if ranging from 0.35-0.49 and excellent if ≥ 0.50¹⁶16 Evers A, Hagemeister C, Høstmælingen A, Lindley P, Muñiz J, Sjöberg A. EFPA review model for the description and evaluation of psychological and educational tests. Test review form and notes for reviewers Version 42. 2013..

Kolmogorov-Smirnov and Shapiro-Wilk tests of normality were calculated to determine data distribution characteristics. The Kaiser-Meyer-Olkin (KMO) measure of sample adequacy of 0.929 (≥ 0.6) and a significant Bartlett’s test of sphericity (χ²2 World Health Organization. WHO global strategy on people-centred and integrated health services: interim report. Geneva: WHO; 2015. (1711) = 12 163.432, ρ < 0.001), indicated adequacy of data reduction procedures. EFA using the extraction method of principal component analysis with orthogonal varimax rotation was used. The number of factors was identified by the Kaiser criterion of Eigenvalue ≥ 1 and Scree Plot¹⁷17 Marôco J. Análise estatística com o SPSS Statistics. Lisboa: ReportNumber; 2018..

Confirmatory factor analysis (CFA) was conducted for each PCPI-S domain corresponding to the PcPF domains of prerequisites, care environment and person-centered processes. Skewness (> 2) and kurtosis (> 7) were not relevant issues across items (cf. Marôco¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010.). Accordingly, the maximum likelihood estimation method recommended for categorical, non-normally distributed data was used¹⁹19 Hu Lt, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equation Model Multidiscip J 1999; 6(1):1-55.. Outliers were determined through the Mahalanobis distance for each data case. The items’ factor loadings were examined and the threshold for acceptability defined upon the sample size achieved¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010.. Correlated items with a modification index (MI) greater than 4 were reviewed to potential modification, giving priority to MIs greater than 11¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010.. Modifications to the original structure of each domain were defined in congruence with the criteria used in previous validation studies of the PCPI-S into Norwegian⁸8 Bing-Jonsson PC, Slater P, McCormack B, Fagerstrom L. Norwegian translation, cultural adaption and testing of the Person-centred Practice Inventory - Staff (PCPI-S). BMC Health Serv Res 2018; 18(1):555. and German⁹9 Weis ML, Wallner M, Köck-Hódi S, Hildebrandt C, McCormack B, Mayer H. German translation, cultural adaptation and testing of the Person-centred Practice Inventory-Staff (PCPI-S). Nursing Open 2020; 7(5):1400-1411. in the following order 1) correlated errors across items within the same factor, 2) correlated errors across items across factors, and 3) correlated factor loadings of items to factors. Modifications were only allowed if congruent with the theoretical framework.

The instrument model was refined continuously and iteratively until it was considered to have acceptable fit. Following the recommendations from the literature, model fit was attained if at least one criterion from each component revealed acceptable fit. The goodness-of-fit model was assessed through several fit indices from each category of three model fits (absolute, incremental, and parsimonious fit). In this study, the fitness indices used were Root Mean Square Error of approximation (RMSEA; acceptable fit ≤ 0.06), root mean square residual (SRMR; acceptable fit < 0.08), Tucker-Lewis Index (TLI; acceptable fit ≥ 0.9), Comparative Fit Index (CFI; acceptable fit ≥ 0.9), chi-square (χ²2 World Health Organization. WHO global strategy on people-centred and integrated health services: interim report. Geneva: WHO; 2015.)/degrees of freedom (df) (acceptable fit ≤ 5).^{18, 19}

Results

Translation and cultural adaptation

The work on developing the explanations of the instrument’s concepts, and the subsequent reconciliation of the two forward translated versions allowed the identification of language differences related to the culture and to the healthcare system. Attaining the definition of the construct of ‘Appropriate skill mix’, the Portuguese healthcare system does not comprise staff working as non-registered nurses and the multidisciplinary designation of ‘staff’ was found to be ambiguous. Non-registered nurses in the Portuguese healthcare system were then considered to be culturally equivalent to the operational assistants at the hospital wards and no such role was identified in relation to primary care. Concerning the designation of ‘staff’, it was culturally equivalent to the multidisciplinary team, who would present different members and structures depending on the context of care delivery.

These two steps of the translation and cultural adaptation of the instrument also revealed to be challenging with respect to ensuring the focus on a language of person-centeredness congruent with the PcPF and relevant across healthcare settings (e.g., physical environment: healthcare environment vs. clinical environment). The person-centered processes construct of ‘being sympathetically present’ was found to be particularly difficult to adapt culturally, while keeping its alignment with the philosophy of person-centeredness. Specifically, the literal translation of ‘sympathetic presence’ was not considered to bear the original meaning of the concept (i.e., an engagement that recognizes the uniqueness and value of the individual, by appropriately responding to cues that maximize coping resources through the recognition of important agendas in their life). Empathetic presence would be a more commonly used expression in Portuguese healthcare discourse and is understood by both healthcare professionals and service users. Yet, it is not endorsed by the original theoretical framework. The culturally adapted back-translation was therefore developed to ‘solidary presence’.

Cognitive debriefing

Fifteen healthcare professionals participated in the individual cognitive debriefing with each virtual interview having one hour duration on average. Each participant reviewed 20 items on average following the instrument structure. The sample was mostly composed of women (n=12, 80%) and included the participation of physicians, nurses, physiotherapists, and psychologists. Participants schooling ranged from 16 to 24 years with a PhD being the highest academic degree. Ten participants (67%) worked at a specialized healthcare practice context and five (33%) from the primary healthcare context. Their professional experience ranged between 11 and 24 years.

Cognitive debriefing revealed understandability and suitability of the items. Yet, some issues that can compromise cultural equivalence were identified and grouped in relation to i) lack of inclusiveness in relation to healthcare professionals’ discourse, practice, or healthcare context; ii) ambiguity of interpretation, iii) social desirability and iv) readability.

The interpretation of the items’ meaning amongst the different professional categories was found to be similar, although different terms were used to portray the same meaning. Lack of inclusiveness in relation to the healthcare professionals’ discourse, practice or healthcare context was identified for the words ‘care’ and ‘team’, and in specific items. Chart 2 depicts the preferred wording for specific designations in relation to the healthcare professionals and the items where it was found to be problematic.

Across healthcare professionals the designation ‘care’ was identified as being closer to the nursing profession, which was even recognized by nurses as a potential misfit with other professions discourses. Psychologists identified ‘intervention’ as preferrable to ‘care’. Likewise, physiotherapists would be more familiar with ‘rehabilitation plan’ while physicians at the specialized settings identified more easily with ‘diagnose and treatment’. Importantly, primary care physicians were more comfortable with using the original wording compared with specialist physicians, without recommending amendment of the item.

The word ‘team’ was probed to elicit the constituting members being considered when answering the item. It became evident that it was interpreted differently across healthcare professionals’ and within the same professional group depending on the practice setting. This was evident regardless of the context being primary or specialized care. Healthcare professionals compared experiences of working at different practice settings, mentioning that at one of the settings they would interpret the ‘team’ as being the nursing team, whereas in the other they would consider the multidisciplinary team. Primary care physicians considered the multidisciplinary team across items. Nurses and physicians at the specialized healthcare settings had varying interpretation across items. This variability was particularly identified in the care environment domain. For example, item 19 ‘I recognize when there is a lack of knowledge and skills in the team and its impact on the provision of care’, ‘team’ referred to each healthcare professional group, i.e., nursing team, physicians. While item 26 ‘I work in a team that values my contribution to person-centered care’ and 27 ‘I work in a team that encourages everyone to contribute to person-centered care’, ‘team’ referred to the multidisciplinary team. Psychologists usually considered the other psychologists as being the team members, whereas physiotherapists considered the multidisciplinary team.

Item 10 was judged to be inadequate by psychologists and the other healthcare professionals reflected on how allocating extra time to a specific encounter would compromise person-centeredness in encounters that followed. To fit the practice of professionals across categories while stressing person-centeredness, the item was consensually adapted to ‘I strive to ensure that the time I spend with the person is of quality’ (the original version was ‘I go out of my way to spend time with the people receiving care’).

Item 25 ‘My opinion is sought in clinical decision-making forums (e.g., ward rounds, case conferences, discharge planning)’ lacked inclusiveness in relation to primary healthcare practice, while being associated with routines of inpatient care contexts. Amendments were consensually adapted to ‘My opinion is requested in multidisciplinary therapeutic decision-making meetings (e.g., clinical service meeting, clinical case conferences, discharge planning).

Item 49 ‘I work with the person to set health goals for their future’ was sensitive for healthcare professionals working in oncological care settings. The respondents stressed the focus on future health objectives in comparison with the person’s future. This item was therefore consensually harmonized to ‘I work with the person to set future health goals’.

Ambiguity of interpretation across healthcare professionals’ categories was found in relation to the word ‘others’ and ‘person’ or ‘people’. On items 4, 5 and 16, ‘others’ was usually reflected upon to either relate to healthcare colleagues or to patients and their families. Respondents suggested changing to the singular form to stress the focus on respect and consideration in relationships regardless of the persons interacting in the relation. On item 50 ‘I enable people receiving care to seek information about their care from other healthcare professionals’, the word ‘others’ was continuously questioned and reflected to be other healthcare professionals. Accordingly, the item was rephrased to ‘I sensitize the person to obtain information about their care from other health professionals’, to reduce ambiguity of interpretation.

On items 9, 10, 36, 41, 46, 47, 50, 53, 54, and 55, ‘people’ was replaced by ‘person’ to enhance interpretation as service user. On item 8, ‘I strive to deliver high quality care to people’, ‘people’ was considered redundant and decreased readability, which led to the deletion of the word from the item. The positioning of item 29 ‘The contribution of colleagues is recognized and valued’ in the sequence of the previous items, was identified as influencing its interpretation towards reducing the ambiguity in relation to the person recognizing or valuing the contribution.

Social desirability was identified as potentially leading to positively scored items. While testing the items, respondents reflected recurrently on how they aimed for the ideal described in the item, although they were not consistently able to achieve it in their daily practice. Thus, they would answer ‘neither agree, nor disagree’ or ‘agree’ instead of ‘strongly agree’. To give an example, when reflecting on the item 4 ‘I make sure I hear and acknowledge the other’s perspectives’, Respondent 13 (R13) reflected as follows:

Interviewer: Ok, and what would you select as an answer to this item?

Concerning the items readability, specific items were considered complicated. Namely item 28 ‘My colleagues positively role model the development of effective relationships’, which was consensually amended to ‘My colleagues set a good example of effective peer relationships’. Similarly, item 41 ‘I am recognized for the contribution that I make to people having a good experience of care’ was adapted to ‘I am recognized for providing a positive care experience to the person’. Item 56 ‘I ensure my full attention is focused on the person when I am with them’ was also considered to be confusing, with participants suggesting its rephrasing to ‘I make sure I fully focus my attention on the person I’m with’. On item 19 ‘I recognize when there is a lack of knowledge and skills in the team and its impact on care delivery’, three of the respondents considered that the item was addressing two different elements.

Person-centered care as phrased on items 26, 27, and 37 was systematically read in the Portuguese plural form of ‘care’. When asked to reflect on this, respondents identified the plural form to be part of their natural discourse as healthcare professionals, whereas the singular form was interpreted to be mostly in line with characteristic of being careful. Person-centered care was therefore adapted to the plural form of care.

Psychometric evaluation

A total of 308 healthcare professionals completed the inventory, of which four cases were considered dropouts as none of the items were answered. The sample (n = 304) was 86% composed of women (n = 261) and 14% (42) men. Participants were mostly nurses (45%) followed by occupational therapists, (26%), physicians (15%), physiotherapists (9%), and psychologists (2%). Healthcare professionals had an average of 17.3 schooling years (range 4-37) and 14.3 years of professional experience (range 0-46).

Missing data analysis revealed a negligible percentage of 0.314 for all items and missing at random as per ρvalues ≥ 0.01 but < 0.05 (Little’s MCAR test: χ²2 World Health Organization. WHO global strategy on people-centred and integrated health services: interim report. Geneva: WHO; 2015. = 2275, df = 2131, ρ = 0.02). Multiple imputation was therefore used for the identified missing values, which were included in the analysis.

Item mean scores ranged from 2.65 (item 40) to 4.66 (item 8) and were mostly positively scored (Table 1). Ceiling effects were identified for most of the items with exception of 10 that are identified in Table 1, with most of these belonging to the domain of the care environment.

The Cronbach’s alpha for the whole inventory revealed excellent internal consistency (α = 0.96). Only the domain of Prerequisites had a good internal consistency (α = 0.89). Here the construct of ‘professionally competent’ had a value lower than 0.7 (α = 0.56). Particularly on this construct, the inter-item correlations were only sufficient ranging from 0.28 to 0.33¹⁶16 Evers A, Hagemeister C, Høstmælingen A, Lindley P, Muñiz J, Sjöberg A. EFPA review model for the description and evaluation of psychological and educational tests. Test review form and notes for reviewers Version 42. 2013.. In the Care Environment domain (α = 0.93), the constructs of ‘skill mix’ (α = 0.65) and ‘potential for innovation and risk taking’ (α = 0.62) also scored below 0.7 on the Cronbach’s alpha, revealing acceptable internal consistency.

The item-scale correlations were generally good ranging from 0.40 to 0.85, with the exception for the item 19 ‘I recognize when there is a lack of knowledge and skills in the team and its impact on care delivery’, with an inadequate correlation value (0.20).

EFA with principal component analysis and orthogonal varimax rotation revealed 10 components in line with Eigenvalue ≥ 1. Altogether, these factors accounted for an explained variance of 65,21%, which was considered adequate.

To keep in line with the CFA adequate ratio of 10:1 of respondent to item, the factor analysis was conducted according to the original domains (i.e., Prerequisites 17:1, Care Environment 12:1, Care Processes 19:1)¹⁵15 Nunnally JC. An overview of psychological measurement. In: Wolman BB. Clinical Diagnosis of Mental Disorders. Boston: Springer; 1978. p. 97-146.. Given the sample size greater than 250 participants, factor loadings of 0.35 were considered acceptable¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010..

Following the confirmatory factor analysis, modifications were performed to the domains of Prerequisites and Care environment according to the first criterion of correlated errors across items within the same construct for items with correlated errors with MI greater than 11¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010.. Specifically, for the domain of Prerequisites, the measurement errors of items 4 and 5 within the construct Developed Interpersonal Skills were correlated (MI = 16.259). Within the construct Being Committed to the Job, the measurement error of items 9 and 10 (MI = 18.635) were correlated, as well as the measurement errors of items 9 and 11 (MI = 17.246), items 11 and 12 (MI = 5.925), and items 10 and 12 (MI = 14.402). Within the construct Clarity of Beliefs and Values, the measurement errors of items 16 and 18 (MI = 6.259), as well as the measurement errors of items 17 and 18 (MI = 7.459). Concerning the domain of Care Environment, the measurement errors of items 31 and 32 within the construct Power Sharing were correlated (MI = 20.161), as well as the measurement errors of items 34 and 35 (MI = 45.985) within the construct Potential for Innovation and Risk Taking, and the measurement errors of items 39 and 40 (MI = 12.719) within the construct Supportive Organizational Systems. The fit statistics for each domain of the adjusted PCPI-S indicate good model fit and are displayed in Table 2.

Overall, factor loadings for all items on each construct were considered acceptable ranging from .42 on Item 34 to .87 on Item 59 (Table 3). Factor loadings between constructs were all acceptable for the domains of Prerequisites and The Person-Centered Care Processes. Yet in the Care Environment domain, some factor loadings were below 0.4, specifically between Skill Mix and Effective Staff Relationship (.33), Skill-Mix and Potential for Innovation and Risk Taking (.32), Skill-Mix and Supportive Organizational Systems (.31), and Effective Staff Relationship and Physical Environment (.30).

The composite reliability for all constructs across domains improved from the original to the adjusted model. The values ranging from 0.66 to 0.90 in the Prerequisites, 0.66 to 0.91 in the Care Environment domain, and 0. 84 to 0.95 in the Person-Centered Processes, indicate improved internal consistency of the adjusted model across the three domains.

The average variance extracted (AVE) indicated acceptable convergent validity (i.e., above 0.5, according to Marôco¹⁸18 Marôco J. Análise de equações estruturais: fundamentos teóricos, software & aplicações: Lisboa: ReportNumber; 2010.) for all constructs across domains, with the exception of the construct Professionally Competent (AVE = 0.40) in the domain of the Prerequisites, and the construct Potential for Innovation and Risk Taking (AVE = 0.43) in the domain of the Care Environment.

Finally, discriminant validity was found for most of the constructs within the Prerequisites and the Care Environment. The exceptions were in The Prerequisites, between Professionally Competent and Developed Interpersonal Skills, and between Professionally Competent and Commitment to the Job. Similarly, in the Care Environment, discriminant validity was not found between the constructs Power Sharing and Potential for Innovation and Risk Taking, Power Sharing and Supportive Organizational Systems, and Potential for Innovation and Risk Taking and Supportive Organizational Systems. Lastly, concerning the domain Person-Centered Care Processes, discriminant validity was only found between the constructs Working with Patients’ beliefs and Values and Providing Holistic care, and Shared Decision Making and Providing Holistic Care.

Discussion

The systematic and rigorous translation and cultural adaptation of an instrument is a crucial prerequisite for its validity and reliability¹¹11 Suhonen R, Leino-Kilpi H, Välimäki M. Development and psychometric properties of the Individualized Care Scale. J Eval Clin Pract 2005; 11(1):7-20.. Importantly, language is one of the culture cornerstones and together with the person’s values, education and in a specific context, it will shape the meaning of the words and the interpretation of the phenomena being assessed²⁰20 Acquadro C, Patrick DL, Eremenco S, Martin ML, Kulis D, Correia H, Conway K; International Society for Quality of Life Research (ISOQOL) Translation and Cultural Adaptation Special Interest Group (TCA-SIG). Emerging good practices for translatability assessment (ta) of patient-reported outcome (PRO) measures. J Patient Rep Outcomes 2017; 2(1):8..

In the current study, the translation and cultural adaptation processes of the PCPI-S from English to European Portuguese language was a demanding process to retain the conceptual meanings of the original instrument in relation to the language, healthcare professions’ categories and contexts. The challenges found were similar to the ones identified in the Norwegian validation study⁸8 Bing-Jonsson PC, Slater P, McCormack B, Fagerstrom L. Norwegian translation, cultural adaption and testing of the Person-centred Practice Inventory - Staff (PCPI-S). BMC Health Serv Res 2018; 18(1):555. and more recently in the Malaysian cross-cultural adaptation¹⁰10 Balqis-Ali NZ, Saw PS, Jailani AS, Fun WH, Mohd Saleh N, Tengku Bahanuddin TPZ, Sararaks S, Lee SWH. Cross-cultural adaptation and exploratory factor analysis of the Person-centred Practice Inventory-Staff (PCPI-S) questionnaire among Malaysian primary healthcare providers. BMC Health Serv Res 2021; 21(1):32., both aiming at retaining the interdisciplinarity of the instrument. In the Portuguese study, the interdisciplinarity and the context of practice (i.e., primary care vs. specialized care) added further complexity in reaching a common discourse.

The iterative harmonization processes between the forward- and back-translated versions and the other existing versions, as well as the cognitive debriefing, were essential procedures in ensuring an inclusive discourse that was accepted and recognized across healthcare professionals without compromising the conceptual equivalence.

Another aspect that benefited the current study in terms of reaching a conceptual structure aligned with the original, was the parallel discussion of the items with the Swedish team, who were undertaking the same procedure for translating and culturally adapting the PCPI-S to Swedish. This continuous process kept alive the issue of ensuring comparability across cultures, while ensuring each country’s cultural and linguistic accuracy²⁰20 Acquadro C, Patrick DL, Eremenco S, Martin ML, Kulis D, Correia H, Conway K; International Society for Quality of Life Research (ISOQOL) Translation and Cultural Adaptation Special Interest Group (TCA-SIG). Emerging good practices for translatability assessment (ta) of patient-reported outcome (PRO) measures. J Patient Rep Outcomes 2017; 2(1):8.^,2121 Beaton DE, Bombardier C, Guillemin F, Ferraz MB. Guidelines for the process of cross-cultural adaptation of self-report measures. Spine 2000; 25(24):3186-3191..

Stressing the ISPOR recommendations, retaining the conceptual meaning supplants the literal translation¹³13 Wild D, Grove A, Martin M, Eremenco S, McElroy S, Verjee-Lorenz A, Erikson P; ISPOR Task Force for Translation and Cultural Adaptation. Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: report of the ISPOR task force for translation and cultural adaptation. Value Health 2005; 8(2):94-104.. This aspect is of particular importance in relation to the challenge of reducing terms’ ambiguity ²⁰20 Acquadro C, Patrick DL, Eremenco S, Martin ML, Kulis D, Correia H, Conway K; International Society for Quality of Life Research (ISOQOL) Translation and Cultural Adaptation Special Interest Group (TCA-SIG). Emerging good practices for translatability assessment (ta) of patient-reported outcome (PRO) measures. J Patient Rep Outcomes 2017; 2(1):8.. Altogether, the cognitive debriefing of the varied healthcare professionals revealed equivalent interpretations of item meaning, despite the use of less inclusive terms in relation to the healthcare professionals background and practice setting. Even though many terms were changed to a meaningfully rich translation (e.g., ‘care’ to ‘practice’), their interpretation was kept in line with the conceptual understanding of the items.

Yet, specific terms in specific items, might represent different things for different respondents, depending on their healthcare context. One such term is ‘team’. Should this aspect be important to the study, a clarification of the term might be needed in the introduction of the data collection instrument or examples can be given for each specific item¹⁴14 Willis GB. Cognitive interviewing revisited: a useful technique, in theory. Methods for testing and evaluating survey questionnaires. In: Presser S, Rothgeb JM, Couper MP, Lessler JT, Martin E, Martin J, Singer E. Methods for testing and evaluating survey questionnaires. New Jersey: John Wiley & Sons; 2004. p. 23-43.. Although different interpretations arose concerning the team elements, they did not compromise the conceptual meaning of ‘team’, which consistently referred to the other elements, with whom a healthcare professional connects in order to achieve holistic person-centered care⁶6 McCormack B, McCance T, Bulley C, Brown D, McMillan A, Martin S. Fundamentals of person-centred healthcare practice. New Jersey: John Wiley & Sons; 2021..

The other challenge identified in the cognitive debriefing related to answering the item according to the ideal of care or their competence to perform, rather than on the actual practice or caring procedures. This challenge was corroborated by the psychometric evaluation, where ceiling effects were present in forty-nine items across domains.

In the Malaysian study, attempting to handle the identified ceiling effect, authors modified the rating of the Likert options from the original opinion-interpreted rating (i.e., agree vs. disagree) to a focus on the frequency (i.e., never, rarely, sometimes, often, very often)¹⁰10 Balqis-Ali NZ, Saw PS, Jailani AS, Fun WH, Mohd Saleh N, Tengku Bahanuddin TPZ, Sararaks S, Lee SWH. Cross-cultural adaptation and exploratory factor analysis of the Person-centred Practice Inventory-Staff (PCPI-S) questionnaire among Malaysian primary healthcare providers. BMC Health Serv Res 2021; 21(1):32.. Another strategy to enhance coherence of response to the rating system has been the inclusion of a timeframe to the instrument (e.g., in the last month)¹⁴14 Willis GB. Cognitive interviewing revisited: a useful technique, in theory. Methods for testing and evaluating survey questionnaires. In: Presser S, Rothgeb JM, Couper MP, Lessler JT, Martin E, Martin J, Singer E. Methods for testing and evaluating survey questionnaires. New Jersey: John Wiley & Sons; 2004. p. 23-43.. Social desirability is a common phenomenon among healthcare professionals known to influence their responses about their practice²²22 Althubaiti A. Information bias in health research: definition, pitfalls, and adjustment methods. J Multidiscip Healthc 2016; 9:211-217.. Future studies might also consider exploring the extent to which the items are sensitive to this type of response bias towards tendentially positive scores.

The results from the psychometric evaluation replicated the theoretical structure of the instrument. However, the undertaken approach to model adjustment deviated slightly from other psychometric evaluations, where inter-construct correlations were established (e.g., Weis et al⁹9 Weis ML, Wallner M, Köck-Hódi S, Hildebrandt C, McCormack B, Mayer H. German translation, cultural adaptation and testing of the Person-centred Practice Inventory-Staff (PCPI-S). Nursing Open 2020; 7(5):1400-1411.). Such a procedure had the potential of improving the lower factor loadings registered between some constructs in the Care Environment construct, where modification indices were the greatest. Even though these relationships might be off interest to understand the construct validity of the instrument, the undertaken conservative and theoretical adjustment within the isolated constructs and confined to correlated errors was enough to achieve acceptable fit statistics.

The CFA results further revealed convergent validity in relation to the original model, as indicated by the acceptable AVE and CR scores. However, without breaching the overall construct validity, some constructs had low discriminant validity, that can be understood from the theoretical point of view. Particularly concerning the construct person-centered processes, the specificities being captured by the construct ‘Working with Patients’ beliefs and Values’ might not be sufficiently distinct from those captured by the construct ‘Providing Holistic Care’. A similar phenomenon might be occurring in relation to the constructs ‘Shared Decision Making’ and ‘Providing Holistic Care’. Following the instrument’s originally intended purpose, i.e., a tool for practice improvement towards person-centeredness, the lack of discriminant validity between these constructs might not be problematic as they theoretically overlap. Although rephrasing the items to enhance the concept they attempt to capture, might improve the understanding of their specificities. However, if the ambition is to develop an instrument for outcome measurement, this overlapping feature and lack of discriminant validity might be explored with the potential to reduce the overall number of items. The length of the inventory was frequently negatively judged during the cognitive debriefing and might also be prompting further response bias towards positively scored items.

To the generalization of the study results a few aspects should be considered. The lack of a priori robust sample size determination might be considered a limitation in the current study^{23, 24}. The lack of agreement concerning the methods for sample size estimation led the research team to closely follow other other validation procedures of the same instrument into various languages. Moreover, the results on the KMO and the Bartlett test of sphericity are indicative of its adequacy to proceed with confirmatory factor analysis.

Another point worth of consideration when interpreting the measurement structure of the instrument is related to the lack of analyses that account for the interaction between the three constructs. Despite the evidence reinforcing the effectiveness of the PCPI-S and its alignment with the theoretical grounds on the PcPF²⁵25 McCance T, McCormack B, Slater P, McConnell D. Examining the theoretical relationship between constructs in the person-Centred practice framework: a structural equation model. Int J Environ Res Public Health 2021; 18(24):13138., data aggregation from future studies should pursue this goal for the Portuguese population.

Altogether, the translated and validated version of PCPI-S to the Portuguese healthcare context adds to the other versions of the instrument to fill the gap concerning the need for theoretically grounded instruments that allow assessing the multiple perspectives of healthcare professionals across settings²⁶26 Edvardsson D, Sandman PO, Rasmussen B. Swedish language person-centred climate questionnaire-patient version: construction and psychometric evaluation. J Adv Nurs 2008; 63(3):302-309..

Conclusion

In conclusion, the three-factorial model of PCPI-S adjusted to the studied sample is a valid and reliable instrument to assess the perceptions of healthcare professionals on person-centered practice in various Portuguese healthcare contexts, which reflects the theoretical structure of the original instrument. The selection of this instrument for research purposes should take into consideration its strength as a theory-driven assessment tool originally intended for quality improvement purposes. While anchored in the consolidated PcPF, the instrument further allows the capturing of perspectives of different healthcare professionals across healthcare contexts. Additionally, the complementary use of other resources derived from the PcPF will allow for the triangulation of results with the perspectives of healthcare service users.

Further refinements to the items phrasing and rating system, might enhance construct validity, necessarily demanding additional psychometric evaluation. Considering the ceiling effects, the effect of social desirability on the responses to the inventory should be explored along with future psychometric evaluations.

Acknowledgements

Bárbara Oliveira, assistant professor, University of Coimbra, Faculty of Medicine (UC, FMUC). Jeanette Melin, researcher at RISE Metrology, Gothenburg, Sweden. Carolina Lunde, senior lecturer, Department of Psychology, University of Gothenburg, Sweden.

References

Publication Dates

History