Introduction

During the past century, the increase in life expectancy in the majority of countries has turned population aging into a worldwide phenomenon, and it occurs at an even more accelerated pace in developing countries like Brazil.¹ This demographic change will impact health services with an increased demand for treatment of chronic degenerative non-communicable diseases. Dementias are among the most common of such diseases, and are potentially onerous to health systems² and stand out for their prevalence of comorbidity, high risk for incapacitation, and increase in high-cost and long-term care.³

Despite advances in the understanding of the neurobiological processes involved in the pathological course of dementias, the complexity of these diseases makes it clear that there is still a lot to be learned in terms of risk group identification, early diagnosis, and better management of behavioral and psychological symptoms of dementia, which serve as a large cause of stress and suffering for the patients as well as their families and caretakers.

Attachment is one of the psychosocial factors associated with dementias, and few studies have been conducted on the topic by the scientific community. As a result, there are great possibilities for intervention. Bowlby defines attachment as the behavior of seeking and maintaining proximity to another individual, emphasizing its contribution to the formation and maintenance of human species social groupings.⁴ This essentially systemic understanding perceives the individual in continuous and reciprocal attachment behavior during life through more or less stable relationship patterns among individuals.⁵ These patterns are especially responsible for the feelings of support and safety, and therefore can play a crucial role in dealing with stressful situations caused by the physiological, functional and cognitive alterations characteristic of dementia.

There are two main theoretical lines in the study of adult patterns of attachment: the three-category model and the four-category model. The three-category model is based on the classic studies by Ainsworth on childhood patterns of attachment,⁶ dividing the pattern of attachment into three types: safe, avoiding, and anxious/ambivalent. Hazan & Shaver,⁷ who are among the main authors of the three-group system, translated the descriptions of the patterns identified by Ainsworth using terms appropriate to an adult context and found such patterns in adult populations.

Meanwhile, the four-category model is based on a two-dimensional model of adult attachment, where the patterns of attachment are divided into four types: safe, worried, fearful, and avoiding. This model, proposed by Bartholomew and collaborators,^8-10 is based on theoretical propositions by Bowlby, in which the relationship of children with their caretakers leads to the formation of internal representations, both of the self and of others, that later guide social relations. In addition to describing the patterns of attachment in a more detailed manner, the four-category model can offer more personalized analyses, allowing for the possibility of changes in the patterns exhibited by an individual depending on instances of social relation.

Despite theoretical progress, attachment in adults is a recent area of research, with the majority of studies having been conducted with young adults. However, studies with the elderly have shown interesting results that can contribute to a better understanding of the cognitive and behavioral symptoms related to degenerative brain diseases.

In a Dutch study, Miesen investigated the connection between attachment behaviors in relation to family members, parental fixation (understood as the belief that long-deceased parents are still alive), and level of cognitive impairment in patients with Alzheimer’s disease. The results showed that the stronger the cognitive impairment, the lower the attachment behavior in relation to current family members, and the higher the parental fixation manifested. Miesen theorized that the cognitive impairments associated with Alzheimer’s disease complicate interactions with the environment, turning it into a constant source of stress and leading the elder to seek the proximity of family members (attachment behavior) as a way of feeling safe. Disease progression and the incapacity to recognize family members make the individual turn to older objects of attachment (parents), with parental fixation then becoming the attachment behavior itself, according to Miesen.¹¹ In an attempt to recreate Miesen’s study, Browne & Shlosberg found an association between the pre-morbid pattern of attachment and the manifestation of attachment behaviors (for example, following caretakers or calling for them when they are out of view), with the participants classified with an avoiding pattern of attachment manifesting these behaviors more than those classified as safe.¹²

Another study related the pre-morbid pattern of attachment to behavioral and psychological symptoms of dementia in individuals with dementia.¹³ The authors observed that individuals with an avoiding attachment pattern presented more paranoid delusions, while those classified with an ambivalent pattern exhibited more anxiety and anguish. The insecure attachment type was associated with a higher level of overburdening their caretakers.

Recently, a therapeutic approach using dolls has been employed to manage the behavioral and psychological symptoms of dementia in elders, based on the principles of attachment theory. The technique has shown to be effective in relieving emotional and behavioral symptoms, offering an increased sense of well-being and better interactions between elders with dementia and their external environment. Despite its promising results, the technique may not be effective on all patients,¹⁴ requiring more studies on the individual characteristics of the patients subjected to this therapeutic treatment, among them the patterns of attachment.

There are some important tools for the study of attachment using psychological evaluation instruments and techniques. In Brazil, however, there is no validated instrument for the evaluation of attachment in elders, which hinders the advancement of studies in this area for this population.

Considering this context, the choice of cross-culturally adapting an instrument already used and recognized in a variety of foreign studies offers the possibility of reproducing studies and comparing results between population samples from different countries. In this sense, the Relationship Scales Questionnaire (RSQ), developed by Griffin & Bartholomew¹⁰ based on three previous scales of attachment, presents itself as an option to evaluate attachment.

The objective of this work was to present the process of translation and cross-cultural adaptation of the RSQ into Brazilian Portuguese and the results of its test-retest reliability obtained by applying a version of the RSQ in Brazilian Portuguese developed for interview application.

Methodology

The RSQ presents 30 declarative statements that express feelings and thoughts related to close relationships with people. Each of the statements must be answered according to a Likert scale ranging from 1 to 5, where 1 (“nothing like me” in the original) represents total disagreement with the statement and 5 represents total agreement (“totally like me”); answers 2 to 4 correspond to partial agreement (“somewhat like me”).

The translation of the RSQ was based on the principles laid out by the Task Force for Translation and Cultural Adaptation of the International Society for Pharmaeconomics and Outcomes Research (ISPOR).¹⁵ These principles are the result of a revision of 12 of the main guidelines for translation and cultural adaptation and propose a 10-step process, which served as guidelines for the entire translation process (Figure 1).

Figure 1 Flowchart of the translation stages of the Relationship Scales Questionnaire (RSQ) into Brazilian Portuguese based on the principles for translation and cross-cultural adaptation of the International Society for Pharmaeconomics and Outcomes Research (ISPOR).11 

The preparation phase began by contacting the authors of the original RSQ, who agreed to the development of the Brazilian version of the scale. Still during this stage, the professionals able to develop the necessary tasks for the upcoming stages were selected and consulted about their availability for the project.

In the second phase, two translations were made, conducted independently from one another by two researchers, both health professionals with experience in using psychological instruments, residents of Brazil, and fluent in English. The researchers also participated in the reconciliation phase, together with project coordinators, in order to develop the first version of the scale in Brazilian Portuguese, produced based on the two translations resulting from the prior phase.

In the fourth phase, the reconciled version of the RSQ in Brazilian Portuguese was handed to a professional translator, native of the United States, and a foreign collaborator, both fluent in Portuguese and English, who conducted two independent back-translations into English. These back-translations were compared to the original RSQ (phase 5). Adjustments were made based on the comparisons. The project’s coordinators judged that the stage of harmonizing the back-translations in different languages was not necessary since this was a process of translation into one specific language, and moved on to the seventh phase, in which the second version of the RSQ was applied to a sample of 20 healthy elderly individuals, contacted via the primary health network, only to verify possible problems with comprehension of the instrument during application.

After the satisfactory result of the pilot application, the coordinators submitted the applied version of the RSQ in Brazilian Portuguese for revision, applying appropriate orthographic corrections (phase 9), thus arriving at its final version, whose development process is here documented (phase 10).

To measure test-retest reliability, a sample of elders aged ≥60 years, enrolled in the Cerebral Aging Program (Programa de Envelhecimento Cerebral [PENCE]) of a Family Health Strategy (Estratégia de Saúde da Família [ESF]) facility in the city of Porto Alegre, state of Rio Grande do Sul, southern Brazil, was selected. The PENCE is longitudinal cohort study that aims to examine factors associated with mental health in a sample of elderly with disadvantaged socioeconomic status in Brazil.

Subjects participating in both the pilot application and the test-retest measurement were recruited by general practitioners in family health facilities from different neighborhoods in the city of Porto Alegre and by health care workers in the PENCE program, and were evaluated at Hospital São Lucas, Porto Alegre, Brazil. All subjects provided written consent to their participation in the study, answered a social-demographic questionnaire and were evaluated using Addenbrooke’s Cognitive Examination – Revised Version (ACE-R) and the Mini International Neuropsychiatric Interview (MINI) to exclude any possible cognitive impairments or psychiatric pathologies. The results of the ACE-R were evaluated according to years of education.¹⁶ Individuals were excluded from the sample if they presented: 1) cognitive impairment or dementia; 2) major psychiatric disorder or central nervous system disease; 3) alcohol or drug abuse or dependency; 4) serious physical illness that impaired their participation in the study. The demographic data of the studied population are presented in Table 1. The mean age was 71.7±6.6; participants were predominantly female and with low education level.

The interview version of the RSQ in Brazilian Portuguese was applied in person to a sample of 43 elders. The retest was applied via telephone 12 to 18 months after the first application and was responded by 29 elders.

Results

The sentences in the Likert scale were translated, respectively, as “nada a ver comigo,” “tudo a ver comigo” and “mais ou menos a ver comigo.” In relation to the statements, the translation process did not require many adaptations (Table 2). Only 6 of the 30 items required the replacement of words or expressions to maintain the conceptual and semantic equivalence of the instrument. The expression “romantic partner,” present in items 11, 21 and 29, was translated as “companheiro(a),” a word whose use to denote a love partner, regardless of the formalization status of the relationship, is already established in Brazilian Portuguese. Meanwhile, the expression “others are reluctant” (item 25), whose literal translation is of difficult comprehension to the majority of the Brazilian population, was translated in the first version as “os outros tem resistência.” However, after the back-translation it was observed that this first construction drifted from the sentence’s original meaning, and therefore we opted to use the translation “os outros evitam” in the instrument’s final version.

Adaptations were also needed for the verbal expression “to merge,” present in items 4 and 18, since its literal translation is rarely used in reference to interpersonal relations in the Brazilian context. Therefore, in the first version the expression used was “entregar completamente,” in item 4, which proved not to satisfy the equivalence criteria, being then replaced in the final version with “unir.” Meanwhile, in question 18, the translation considered most appropriate to the sentence’s context was “relacionar profundamente.”

No comprehension difficulties were observed in relation to the declarative statements during the pilot application of the instrument. However, self-application proved to be a less than adequate format for this specific demographic group, considering the reading difficulties presented by elders with lower levels of education and those with visual impairments, both common within this age group. Therefore, the decision was made to develop another version to be applied as a structured interview.

The structured interview version was developed, starting with instructions for the interviewer on how to fill out the answers. The interviewer is oriented to read each one of the items and ask the interviewee to answer each one with “yes,” “no,” or “more or less.” If the answer is “more or less,” the interviewer should mark 3 on the Likert scale. If the answer is “yes,” the interviewer should ask the interviewee to further choose between “a lot” or “totally”, and then mark 4 or 5, respectively, according to the answer given. If the answer is “no,” the interviewer must ask the interviewee to choose between “a little” or “nothing,” and mark 2 or 1, respectively.

The statements did not suffer any significant alterations in relation to the translation, the only change being to grammatical agreement, from first person singular in the original and self-applied versions to third person singular in the interview version, and each statement beginning with “O(A) Sr(a). é uma pessoa que...” (“You are a person that...”). The expectation is that this new version minimizes the difficulties encountered during the pilot application and proves to be more adequate for use in the elderly Brazilian population.

Considering the possibility of conducting studies with elders who already present some type of cognitive impairment that hinders their self-assessment, we suggest that the instrument should be applied to a third party related to the patient as well. To this purpose, a separate version of the RSQ, specifically designed for informants, was developed based on the self-applied version.

Just as in the interview version, only grammatical agreement was modified in order to refer to a third person, with each statement beginning with “O(A) Sr(a). [NOME] é uma pessoa que...” (“Mr. or Ms. [NAME] is a person that...”), in which the interviewer must say the name of the person about whom they wish to gather information. The instructions for the interviewer remain the same as in the interview version. In both versions, the interview and the informant versions, the statements should be read exactly as they are written in order to minimize any biases caused by differences between written and spoken languages.

In order to assess test-retest reliability, statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS) version 21.0, adopting a significance level of 5% (p<0.05). Quantitative variables were described as mean and standard deviation, with the scores being calculated using the Griffin & Bartholomew model for types of attachment¹⁰ and compared using the paired sample t-test (Table 3).

Although the measure for secure attachment showed differences in the answers between the first and second applications, the other measures of attachment did not present differences, suggesting reliability of the final version of the instrument. We highlight some limitations of this study, such as the small sample size and the long interval between the applications. Another aspect that should be emphasized in studies on levels of attachment in elderly individuals is that, although there is a tendency for styles of attachment to remain stable throughout one’s life, the occurrence of significant life events and environmental changes may lead to modifications.

Conclusion

The Brazilian adaptation of the RSQ is the first step in the development of a standardized attachment evaluation instrument for individuals over 60 years of age. We highlight that the RSQ is designed as a research instrument and is not intended for clinical or diagnostic purposes. All questionnaires are available online as supplementary material. Thus, competent professionals can now use this tool in Brazilian Portuguese for the purposes of scientific research.