Validity and Reliability of the Portuguese-Performance Assessment of Self-Care Skills (P-Pass) for Individuals with Disabilities

Background: Due to the increasing need to measure health outcomes and to ensure a more independent life, it is essential to validate instruments capable of assessing the functionality of activities of daily living. The Performance Assessment of Self-Care Skills (PASS), based on three occupational performance constructs (independence, safety and adequacy), allows health professionals to delineate their interventions more effectively according to the needs of each patient. The purpose of this study was to create a Portuguese version of the PASS (P-PASS) and to measure the performance of people with disabilities. Methods. Linguistic validation was performed through the translation/back-translation procedure. Next, the scale to be completed by health professionals was submitted to a clinical review by a panel of experts. Subsequently, it was applied to 98 people with physical and/or cognitive disabilities. The data collected were then analyzed and the results evaluated in accordance with the three PASS constructs. The results were also analyzed in terms of age and gender.

number of older people in Portugal, from 708,569 to 2,010,064. This growth is expected to continue at least until 2040 [2]. In 2017, we were the country with the fourth highest percentage of people aged 60 years or older (27.9%) and, in 2050, we are expected to occupy the third position, with 41.7% of older people. Furthermore, elderly dependence is increasing and, currently, Portugal is among those countries that have the lowest expectation of a healthy life at age 65, especially women [3]. Therefore, efforts to improve the quality of life during old age should be one of the main challenges of today's society.
Several authors point to the existence of a signi cant association between the level of disability in activities of daily living (ADL) and the increased probability of having depression, as well as a reduction in the quality of life [4]. In addition, basic (BADL) and instrumental ADL (IADL) enable people to participate in their immediate community. In older people, the greater the disability in ADL the greater may also be the decline in cognitive abilities [5]. In addition, many older people experience disability in IADL as their age increases [6]. Therefore, early identi cation of the main de ciencies in the independence of individuals in their daily living activities enables more appropriate interventions to prevent disability [7]. Consequently, it is crucial to increase knowledge regarding the assessment of BADL and IADL functionality, and in the use of assessment tools, thus enabling the improvement of the health care provided.
Multiple measurement instruments are available for evaluating BADL and IADL. Some instruments have already been validated for the Portuguese population, for example, the Barthel Index [8] and the Lawton Index [9]. However, there are few validated performance instruments appropriate for the older Portuguese population. One potential instrument, the Performance Assessment of Self-Care Skills (PASS) [10,11], is a performance-based, criterion-referenced observational tool designed to measure occupational performance of daily life tasks. As an observer-rater instrument, its original version consists of 26 core tasks, categorized in four domains: functional mobility (5 tasks), basic activities of daily living (3 tasks), and instrumental activities of daily living with a physical emphasis (4 tasks) and with a cognitive emphasis (14 tasks). For each task, PASS also rates three occupational performance constructs: independence, safety, and adequacy. The PASS is designed to evaluate patients' abilities to live independently in their community. It is particularly suited for cultural adaptation because its manual provides a template for developing test items that are relevant for a country's population. The purpose of this study was to validate the PASS for the Portuguese population (P-PASS) by adapting and developing test items relevant to the Portuguese population and establishing the validity (content, construct) and reliability (internal consistency, interrater) of the new instrument.

Methods
Validity and reliability studies were conducted in Portugal in 2019. The studies involved two phases: (i) translation and cultural validation of the P-PASS, including content validity; and (ii) reliability of the P-PASS, followed by construct validity.
Phase I: Translation and cultural validation of P-PASS

Page 4/21
To create the P-PASS, approval for cultural adaptation was obtained from the PASS authors. The original PASS version encompasses 26 items in four domains: functional mobility (FM), basic activities of daily living (BADL), cognitive instrumental activities of daily living (CIADL) and physical instrumental activities of daily living (PIADL) (see Table 1).
Content validity of the PASS was based on four questionnaires and is reported elsewhere [12,13,14,15]. Test-retest and inter-rater reliability for the PASS were acceptable. (For more information on the PASS see Chisholm et al. [16,17]). For the purposes of task observation and scoring, each of the 26 tasks is broken down into subtasks. Each subtask indicates the action that is to be carried out by patients and the quality/adequacy of the outcome to be achieved by the action. Three distinct occupational performance constructs are measured by the PASS: (i) independence (type/amount of assistance needed to initiate, execute, and complete a task); (ii) safety (risks to people or environment); and (iii) adequacy (performance skill/quality or acceptability of action/product). A table was developed to help the evaluator to assign a score from 0 to 3 for each construct (see Table 2). The summary task independence score is obtained by averaging the independence scores for the corresponding subtasks. The summary scores for the safety and adequacy constructs each re ect the total task.
As a rst step in cultural adaptation and validation of the P-PASS [18,19], a translation from American English to European Portuguese was independently conducted by two Portuguese translators, who translated the 57 page protocol, including test conditions, task instructions, task/subtask pro le, and scoring grid for each of the 26 tasks. A third translator translated the test manual. Differences between the two Portuguese translations were resolved through consensus to yield the P-PASS Version 1. Next, an English language native translator performed the back translation, which was compared to the original PASS to verify semantic equivalence. Subsequently, for content validity, a panel of health care professionals analyzed the relevance of the 26 tasks, including the way in which they were operationalized, for the adult Portuguese population. To test the content validity, the P-PASS was applied to a group of lay people consisting of old or physically/cognitively disabled adults.

Part 2: Reliability and construct validity study
To test the reliability of the Portuguese version, we addressed internal consistency and interrater reliability between two professionals assessing the same participants. Internal consistency was assessed with Cronbach's alpha coe cient, based on the original structure provided by the authors. This coe cient should have scores between 0.7 and 0.9 [20]. The agreement between professionals reached beyond chance was tested using Cohen's kappa [21]. Following Landis and Koch, negative values indicate no agreement, values from 0 to 0.20 slight agreement, from 0.21 to 0.40 fair agreement, from 0.41 to 0.60 moderate agreement, from 0.61 to 0.80 substantial agreement, and from 0.81 to 1 an almost perfect agreement [22].
Construct validity was assessed by performing known-group analyses and by factor analysis (structural validity). Known group analysis consisted of age and gender to ascertain whether they could be considered as two determining variables of independence, safety and adequacy in the performance of different tasks. A sample was constituted of participants of both genders, different ages and with various clinical conditions. The inclusion criteria were adults (18-60) and older adults (> 61) with some type of physical and/or cognitive disability. The exclusion criteria were people who were not able to express themselves clearly and coherently orally or in writing.
For descriptive statistics, means and standard deviations were calculated for each of the tasks for independence, safety, and adequacy. It should be noted that this study was research oriented, (i.e., its purpose was to establish the validity and the reliability of the P-PASS), so all tasks were evaluated for all patients. In a clinic setting, however, we would only select the relevant tasks for a particular patient for clinical measurement and follow-up. In this study, even if participants did not regularly perform a P-PASS task for any reason, we still assessed whether they could or could not perform that task. This is why, in the current study, all tasks were assessed for all participants.
For age, ANOVA was performed for three age groups (<60, 61-79, ≥80). If a signi cant difference occurs, a post-hoc Scheffé's method was used. For gender, the Student's t-test for independent samples was used.
To test PASS structure, we decided to perform a con rmatory factor analysis, a multivariate statistical procedure used to test how well a set of measured variables represent the prede ned structure. We used different t statistics, including the model Chi-square (c2), the Tucker Lewis Index (TLI), the Comparative Fit Index (CFI), and the Root Mean Square Error of Approximation (RMSEA). The acceptable cut-offs to a good t are, in terms of p-values, respectively, >0.05 (c 2 ), ≥0.95 (TLI), ≥0.90 (CFI), and <0.08 (RMSEA) [24]. In the case of failure of this procedure, to further test construct validity we used principal component analysis with varimax rotation aiming at having the easiest interpretation with eigenvalues close to 1.0, after con rmation with Kaiser-Meyer-Olkin Measure (KMO) of Sampling Adequacy and Bartlett's test of sphericity [23,24]. KMO, indicating the proportion of data variance explained by the underlying factors, should be as close to 1 as possible, and signi cant levels <0.05 show that the factor analysis may be useful. Data were processed using SPSS and AMOS version 25.
The study was approved by the Ethics Committee of the Faculty of Medicine of the University of Coimbra and all participants gave informed consent after the study objectives were clari ed.

Results
Translation and cultural validation of P-PASS; Content validity During the translation and cultural validation of P-PASS, the expert panel decided that one CIADL task (bingo) should be removed because it was not relevant for independent living of the older adult Portuguese population. 'Bill paying by check', 'checkbook balancing', 'mailing bills', 'obtaining critical information from auditory media', 'obtaining critical information from visual media', ' ashlight repair', and 'oven use' were combined or adapted to yield 'bill paying by ATM', 'obtaining critical information from the TV', 'changing TV command batteries', and 'microwave use'. Thus, the P-PASS consists of 22 task items: ve FM, three BADL, ten CIADL, and four PIADL. Table 3 shows these 22 tasks included in the P-PASS.
The independence, safety, and adequacy scoring system from the original pass was retained. When administered to a group of older adults and disabled adults no major concerns were raised about the content of the P-PASS. Testing construct validity, in relation to age ( When analyzing the impact of gender (Table 6) on the performance of the 22 tasks for independence, safety and adequacy, in general, no signi cant differences were found between men and women. There were, however, two exceptions for independence, with men scoring higher than women: 'bill paying by ATM' (p=0.035) and 'changing TV command batteries' (p=0.052).

Reliability
To test the structural validity for each dimension, we performed a con rmatory factor analysis attempting to t the results from the Portuguese PASS data in what concerns the items that should belong to each domain of activities. The diagrams and main statistical results are in a supplementary le 1. As evidence from these results, for each dimension and using the c 2 test, we always found a p-value less than 0.05, which means a poor t model. This could be more or less acceptable, as the sample size may be considered not large enough.
However, using TLI and CFI indicators, we also have an evidence of not t as p-values were, respectively, 0.801 and 0.825 (independence), 0.866 and 0.890 (safety), and 0.787 and 0.813 (adequacy). In addition, using the RMSEA measure, the associated p-values were 0.151 (independence), 0.133 (safety) and 0.142 (adequacy), all higher than 0.08.
Therefore, in order to deeper understand the behavior of the data from the Portuguese PASS, applying principal components factor analysis, we obtained very good scores for KMO (independence: 0.944; safety: 0.924; adequacy: 0.928) and for signi cance associated to Bartlett's test of sphericity (independence: <0.001; safety: <0.001; adequacy: <0.001). In general, all factors followed the criteria for a good factor structure. However, the BADL task 'trimming toenails', independent of the number of factors selected, always appeared alone in a sole factor. Therefore, we decided to drop it from this analysis.
We also decided that a model with three factors was appropriate for all three constructs, obtaining cumulative percent of variances, respectively, equal to 80.8% (independence), 77.1% (safety) and 77.1% (adequacy). This was the preferred number of factors to overcome the trade-off between interpretation, cumulative percent of variance and eigenvalues close to 1.0. A supplementary le 2 presents, for each construct, the task loadings associated to each factor, as well as the corresponding eigenvalues, percent of variance and cumulative percent of variance.
Looking at the independence construct, we found primarily FM and BADL tasks together in Factor 1 and CIADL and PIADL tasks in Factors 2 and 3, respectively. However, contrary to the original structure, 'stovetop use' belonged to the PIADL factor and the tasks 'use of sharp utensils', 'microwave use' and 'taking out garbage' belonged to the rst factor together with FM and BADL tasks.
Concerning safety, an almost similar situation to independence was found. That is, Factor 1 was formed by FM and BADL tasks, Factor 2 by mostly PIADL tasks and Factor 3 by CIADL tasks. However, 'medication use' was considered as belonging to the rst factor, the tasks 'stair use' and 'use of sharp utensils' as belonging to the PIADL factor, and, appropriately, the tasks 'stovetop use' and 'microwave use' as belonging to CIADL factor.
Finally, considering the adequacy construct, Factor 1 included CIADL tasks, Factor 2 encompassed FM, PIADL and BADL tasks, and Factor 3, PIADL, CIADL, and a BADL task. Except for 'dressing' the Factor 2 tasks involved moving in the environment. For Factor 3, the tasks focused on meal preparation, except for 'changing bed linens' and 'oral hygiene.'

Discussion
In general, the purpose of our study was achieved: to validate the PASS for the Portuguese population (P-PASS) by adapting and developing test items relevant to the Portuguese population and establishing the validity (content, construct) and reliability (internal consistency, interrater) of the new instrument. Validity and reliability scores were very good and consistent with the original PASS as well as other translations/adaptations.
During the development of P-PASS, and following the advice of the expert panel, we decided to remove one task (bingo) from the original PASS because it did not have current relevance to independent living with this particular Portuguese population. Seven other tasks were combined or changed to re ect current practices in the Portuguese population, but the content remained the same. For example, the tasks 'obtaining critical information from the media (audio)' and 'obtaining critical information from the media (visual)' were combined into 'obtaining critical information from TV', which encompasses the visual and hearing components. The task 'oven use', while signi cant, was considered di cult to evaluate, as it requires very speci c and di cult-to-obtain testing conditions, especially when the measurement is not made in the residence of the patients, and was changed to 'microwave use.' Likewise, 'bill paying by check,' 'checkbook balancing,' and 'mailing bills' (money management) were changed to 'bill paying by ATM.' Lastly, the home maintenance task of ' ashlight repair' was changed to 'changing TV command batteries.' The operationalization of the aforementioned tasks into the Portuguese version was carried out through the process for constructing new items available in the PASS user manual and the PASS scoring system was retained. This type of procedure was similar to that used by other authors during the validation of the PASS scale [25].
The P-PASS results for the study participants indicated that the outcome measures as shown in Table 4 re ect the disability status of the participants. Scores of 3 on the P-PASS are consistent with the skills needed for living in the community. However, 11 of the mean independence scores are below 2, indicating that the evaluators needed to provide the participants verbal encouragement, verbal cues, verbal instructions, gestures, a change of the context or task or demonstrations to initiate, continue or complete the tasks (see Table 2). This is consistent with the settings from which 82 of the 98 participants were recruited -nursing or care homes.
As for the results obtained for all three measurement constructs (independence, safety and adequacy), the tasks 'indoor walking', 'oral hygiene', 'taking out garbage', 'telephone use', 'obtaining critical information from TV' and 'use of sharp utensils' required less assistance for independence, safety, and or adequacy from the evaluator compared to other tasks. In the literature, some results agree with ours for these tasks [26][27][28]. Regarding the task 'use of sharp utensils', it was found that our sample had a superior performance to that described in the literature, whereas the task 'stovetop use' is known to be severely negatively affected, especially in the elderly [26][27][28]. When only independence was considered, the tasks that needed the least amount of assistance were: 'indoor walking', 'bed mobility', 'bathtub and shower mobility', 'dressing', 'taking out the garbage', 'telephone use', 'microwave use', and 'home safety'. In the task 'bathtub and shower mobility', the results are in accordance with the original studies [17].
However, in the current study 'medication management' was one of the lowest scores for independence in the CIADL group, which is not in agreement with these studies [17]. However, the studies of Njegovan et al. [28] indicate that in older people with cognitive decline, this task emerges as the one that has its functionality most affected. Clinically, some of the items requiring less assistance could be chosen to begin an assessment if these are tasks the patient needs to perform, wants to perform, or is expected to perform in the community. More di cult tasks (those with lower scores) could then be added until the patient is unable to initiate, maintain, or complete a task.
Tasks with lower scores, re ecting the need for greater assistance from the evaluator were the FM task 'stair use', the BADL task 'trimming toenails', the PIADL task 'changing bed linens', the CIADL 'bill paying by ATM', and 'stovetop use'. Similar results are con rmed in the literature [17]. The authors Jefferson et al. [26] and Millán-Calenti et al. [27] mention that household management activities yield lower performance scores because they are the rst ones where limitations arise in the event of incapacity. In the task 'cleanup after meal preparation', a lower safety score than that indicated by the PASS authors was obtained by our sample [17]. The task that, in our study, presented the lowest score in the construct of safety in the CIADL group was 'stovetop use', which is in accordance with the literature [17]. Likewise, Jefferson et al. [26] report that meal preparation activities are the rst in which people with disabilities experience a loss of functionality. This is also supported by Njegovan et al. [28], who state that elderly people with greater cognitive impairment experience a rapid loss of independence in these types of tasks. Clinically, if the assessment is being conducted to determine if the patient can remain in the community or needs institutional care, the assessment could begin with some of these tasks and continue until the patient can initiate, maintain, or complete relevant tasks for the discharge environment.
Looking at gender, in the tasks 'bill paying by ATM' and 'changing TV command batteries', male participants presented higher scores than female participants for independence. These ndings are in line with those reported in the literature, where women with disabilities due to chronic disease have a higher incidence of incapacity in IADL [27,29,30]. These authors argue that as women live longer, they experience a greater number of chronic, but not fatal, diseases, which result in limitations of functionality [27,11]. It is unclear why women did not do better on the 'microwave use' item, however, given that 84% of the participants lived in care homes, perhaps the women were unfamiliar with microwaves.
Regarding the age variable, overall, younger participants performed better than their older counterparts. This conclusion was also reported in the studies of Millán-Calenti et al. [27,31], in which the presence of cognitive incapacity or dementia affected functionality in the tasks. Concomitantly, Maciel and Guerra [32] argue that the higher the age, the worse a person's functionality.
Finally, the use of exploratory factor analysis may be controversial as, in theory, it is not the best strategy when the structure is previously known. The con rmatory factor analysis is much more suitable. However, the authors of the original PASS version did not search by factors under each construct; they only tested the unidimentionality of the three constructs [33]. In addition, taking into account our sample size, exploratory factor analysis is considered "adequate" by the COSMIIN checklist [34]. In contrast, cultural adaptions led us to change some tasks to be assessed by the PASS in Portugal. Our exploratory factor analysis had only, as purpose, to better understand the results.
With our data, unidimensionality may not be absolutely discarded as, with a 1-factor solution, we already obtained 64.7%, 65.1% and 57.5% for, respectively, the constructs 'independence', 'safety' and 'adequacy', as presented in the supplementary le 2 attached to this paper. However, a 3-factor solution naturally showed us better percent of variance and a better way to interpret the data.

Conclusions
The P-PASS is a culturally and linguistically valid instrument, which can be applied to the Portuguese population. The P-PASS can be seen as a reference criterion for those who are living independently in the community. Community living individuals should be able to perform all 22 tasks independently, safely and adequately. Having such a tool allows for a more detailed measure of health outcomes, making it possible to administer more rapid and effective interventions for the elderly or disabled. Likewise, given that the assessment is based on observation of everyday task performance, it can be easily understood by the various health professionals that interact with the patient. Furthermore, this scale enables the development of scienti c studies into the impact of various pathologies on the independence, safety, and adequacy of task performance, enabling more person-centered interventions.

Availability of data and materials
The data that support the ndings of this study are available from the corresponding author upon reasonable request Authors' contributions PLF and MD contributed to the study concept and participated in its design and coordination. ALS performed study procedures and collection of data and was responsible for drafting the manuscript; PLF was responsible for the analysis of data; PLF and MD were responsible for editing of the manuscript; MBH and JCR revised the manuscript. All authors read and approved the nal manuscript.

Ethics approval and consent to participate
The Ethics Committee of the Faculty of Medicine of the University of Coimbra approved the study and all participants gave their previous informed consent to participate in the study and for publication. There was, under no circumstances, any interference with the health professional decision regarding the bestsuited medical approach to each patient.  Pagar contas com multibanco money management: bill paying by ATM 11 Utilizar o telefone/telemóvel telephone use 12 Gerir a medicação medication management 14 Obter informação importante da televisão obtaining critical information from TV 15 Mudar as pilhas de um comando de televisão home maintenance: changing TV command batteries 18 Ter segurança em casa home safety 19 Utilizar o micro-ondas meal preparation: microwave use 20

Tables
Utilizar o fogão meal preparation: stovetop use 21 Utilizar utensílios afiados meal preparation: use of sharp utensils