Cultural Adaptation and Validation of the Portuguese Version of the CANHELP Lite Bereavement Questionnaire

Background: Satisfaction with care is an important outcome measure in end-of-life care. Validated instruments are necessary to evaluate and disseminate interventions that improve satisfaction with care at the end of life, contributing to improving the quality of care offered at the end of life to the Portuguese population. The purpose of this study was to perform a cross-cultural adaptation and psychometric analysis of the Portuguese version of the CANHELP Lite Bereavement Questionnaire. Methods: Methodological research with an analytical approach that includes translation, semantic, and cultural adaptation. Results: The Portuguese version comprised 24 items. A panel of experts and bereaved family members found it acceptable and that it had face and content validity. A total of 269 caregivers across several care settings in the northern region of Portugal were recruited for further testing. The internal consistency analysis of the adapted instrument resulted in a global alpha value of 0.950. The correlation between the adapted CANHELP questionnaire and a global rating of satisfaction was of 0.886 (p < 0.001). Conclusions: The instrument has good psychometric properties. It was reliable and valid in assessing caregivers’ satisfaction with end-of-life care and can be used in both clinical and research settings.


Introduction
The quality of care provided to individuals at the end of life, and to their family members, has become an important health and social issue across the world [1]. Quality end-of-life care is considered a right for all citizens and the responsibility of every government [2]. Thus, improving 1.
Linguistic translation of the instrument into European Portuguese by three translators, native in Portugal and fluent in American English. Two translators were unaware of the concept under study, but the third was an expert in the palliative care field.

2.
Synthesis and review of the first translation.

3.
Back-translation of the reviewed version by two other translators, natives of the USA and fluent in Portuguese. They did the semantic analysis and had an agreement result of 100.0%.

4.
Review of all translations by a panel of experts (one expert in research and methodological studies, two experts in palliative care, and one expert in bereavement, in addition to all translators and back-translators). The original developers of the questionnaire were in close contact with the panel of experts during this part of the process. 5.
Assessment of face and content validity through a pre-test on a sample of 20 bereaved family members, which aimed to verify the understanding of the items, evaluate the ease of using the response set, and estimate the response time. The instrument was easily understood, and the bereaved family members answered when requested. The average answer time was 10 min for the entire questionnaire. After the face validity of the instrument was analyzed, its psychometric properties were examined. The literature proposes the inclusion of 5-10 participants per item for the sample [33]. The inclusion of 10 participants per item was estimated and the final sample was composed of 269 bereaved family members.

Participants, Setting and Procedure
The participants were bereaved family members knowledgeable about the healthcare provided during the last three months of life to an adult decedent (aged 18 years and older). To identify potential participants, the death certificates of all adults (aged 18 years and older) that were available in the national electronic death certificate registration were analyzed. Records of decedents where the cause of death codes were associated with external or accidental causes, suicides, pregnancy complications, or medical and surgical complications, as well as death certificates that had an uncertain cause of death were excluded.
In Portugal, death certificates do not have information regarding family members and thus, it was necessary to cross information and to search for a phone number on the health system national registration using the health national registration number provided on the death certificate. A first phone call was made to inform the participants about the study's aims and the voluntary nature of their participation. They were also given guarantees of data confidentiality and anonymity. Informed consent was obtained from each bereaved family member. After that, a phone interview was scheduled at a date of the bereaved family member's preference. Subsequently, the instrument was applied through a phone interview conducted by trained interviewers. All interviews were conducted three to 12 months after death. As the period of bereavement is recognized as a sensitive time, this interval was decided to respect the family bereavement period but also to guarantee that the memory of the experience was recent enough to be considered reliable.

Data Analysis
Descriptive analysis was performed through means and standard deviation, or numbers and percentages of the demographic and clinical variables. The Kolmogorov-Smirnov test was performed to check the distribution of the variables enrolled. A bivariate correlation by Pearson was used.
The assumptions used in the original version were followed and the internal consistency of the scale was obtained using Cronbach's alpha. Cronbach's alpha if item deleted and corrected item-total correlation were also obtained. According to the literature, alpha values under 0.50 are unacceptable, from 0.50 to 0.60 are questionable, from 0.60 to 0.70 are acceptable, from 0.70 to 0.80 are good, from 0.80 to 0.90 are very good, and over 0.90 are considered excellent [34,35].
Criterion-related validity was determined through the concurrent use of a single-item, global rating of caregivers' general satisfaction with the care received in the last three months of life (scored between 1 and 5, meaning that 1 was the worst level of satisfaction possible and 5 the best level of satisfaction possible). This option is justified by the fact that there isn't a validated tool in Portugal to assess bereaved family members' satisfaction with the care received in the last months of life regardless of the existing pathology or setting of care. This was performed using the Pearson correlation coefficient.
Convergent-discriminant validity was performed using an exploratory factorial analysis considering the principal components method with orthogonal rotation (varimax method with Kaiser normalization).
It was not possible to perform test-retest reliability, as the purpose of the instrument was to assess the bereaved families' satisfaction with the care received in the last three months of life. As this might be considered a sensitive matter, we believe that respecting human dignity should have priority over the interests of the research.
The data were analyzed using SPSS, version 23, for Windows. A level of significance of 0.05 was assumed.

Ethical Procedures
First, a research protocol was designed. It was then submitted, analyzed, and approved by the Portuguese Data Protection Authority (number 9607/2016), ethics committee (number P361/09-2016), and the assistant secretary of state and health. A collaboration protocol was also signed between the research team and the Directorate General of Health.

Study Population
All 269 participants were bereaved Portuguese family members from the north region recruited through the identification provided in the health system national registration after analysis of the death certificates. The majority of the participants were female (78.4%) and married (59.5%), and were most often daughters of the deceased (47.6%). The average age was 58.2 years (sd = 13.3 years; Healthcare 2020, 8, 27 5 of 11 minimum = 18 years and maximum = 91 years). The interview was conducted, on average, 295.93 days after the death of the relative (sd = 76.6 days; minimum = 93 days and maximum = 364 days).

Reliability
Considering the 24 items of the scale, the analysis of internal consistency reliability revealed an alpha of 0.950. The global alpha when removing items were also analyzed ( Table 1). Although Cronbach's alpha would increase if items 14, 16, and 23 were deleted, it was decided to retain them, as the alpha obtained is already considered excellent. The English and Portuguese item description of the CANHELP Lite Bereavement Questionnaire is available in Table 2.

Construct Validity
During the translation and cultural adaptation phase, it was observed that the scale was answered quickly and that there were no suggestions for alterations in items or terms considered inadequate or unnecessary, which shows a good understanding and acceptance of the items by the participants.
Concurrent validity was determined using the Pearson correlation coefficient, and a value of 0.886 (p < 0.001) was obtained.
Regarding convergent-discriminant validity, an exploratory factorial analysis considering the principal components method with orthogonal rotation (varimax method with Kaiser normalization) was performed. The initial exploratory factor analysis of our study proposed four factors (Table 3). Table 3. Exploratory factorial analysis considering the principal components method with orthogonal rotation (varimax method with Kaiser normalization). The CANHELP Lite Bereavement Questionnaire proposes the evaluation of five quality subscales, namely relationship with the doctors, characteristics of doctors and nurses, illness management, communication and decision making, and your involvement. Despite these considerations, the instrument remains unidimensional as the first factor explains 53.061% of the total variance of the items and the second factor explains 6.545%, indicating unidimensional adequacy (Figure 1).

Item
The CANHELP Lite Bereavement Questionnaire proposes the evaluation of five quality subscales, namely relationship with the doctors, characteristics of doctors and nurses, illness management, communication and decision making, and your involvement. Despite these considerations, the instrument remains unidimensional as the first factor explains 53.061% of the total variance of the items and the second factor explains 6.545%, indicating unidimensional adequacy ( Figure. 1).

Discussion
The current study provides evidence for the reliability and validity of the Portuguese version of the CANHELP Lite Bereavement Questionnaire. To the best of our knowledge, this study is the first cross-cultural adaptation and psychometric analysis of this instrument in Portugal. A rigorous methodology for the translation, cross-cultural adaptation, and validation processes following the recommended guidelines has been used in this study.
The cross-cultural adaptation and psychometric testing of this instrument were achieved satisfactorily following the recommendations of the international guidelines [36,37]. Regarding this, the Portuguese version of the CANHELP Lite Bereavement Questionnaire showed satisfactory psychometric properties. The face and content validity of the instrument was acceptable, which indicates that all items are relevant for the construct measurement. The internal consistency analysis was satisfactory (Cronbach's alpha = 0.950), suggesting excellent homogeneity of the items. These values are in line with those reported in the original studies [29,30].
In terms of concurrent criterion-related validity, there was adequate correlation with the single-item global rating of caregiver satisfaction with the care received in the last three months of life, which might indicate that the Portuguese version of the CANHELP Lite Bereavement Questionnaire is adequate to measure satisfaction with the care received in the last months of life.
An after-death assessment of the caregivers' satisfaction with the care received in the last months of life would allow us to identify, evaluate, and disseminate interventions that improve satisfaction with care at the end of life, which would also, indirectly, improve the quality of care at the end of life. For that, it is necessary to have valid and reliable instruments. Thus, we believe that the translation and validation of the Portuguese version of the CANHELP Lite Bereavement Questionnaire will contribute to improving care at the end of life in the Portuguese population.
This study followed a rigorous methodology and was developed across care settings, which allow us to be confident regarding the generalizability of the results produced here. Nevertheless, there are some limitations. Effects related to memory might be considered a limitation of this study, as it is a retrospective evaluation by caregivers. In this context, the evaluation of satisfaction with care could be affected by the amount of time between the death and the evaluation. Proxy reports after end-of-life care can be obtained at varying time intervals post-death. Regarding this, it was decided to conduct the interview three to 12 months after the patient's death. This is a conventional time frame for evaluating this construct, as similar periods of time were used in other studies [38][39][40]. However, we recognize that other intervals may provide more accurate data. Future research should explore the optimal timing for the collection of data from bereaved caregivers. Also, the test-retest reliability was not performed, which is the reason why the instruments' sensitivity to change was not evaluated.

Conclusions
The Portuguese version of the CANHELP Lite Bereavement Questionnaire is a valid and reliable instrument for the assessment of satisfaction with end-of-life care and can be used in both clinical and research settings. This is the first instrument available in Portugal to assess bereaved family members' satisfaction with end-of-life care and it could be very important as it might be a valuable contribution to improve the Portuguese health system.