Cross-cultural adaptation of the "Australian National University Alzheimer's Disease Risk Index" for the Brazilian population

Alzheimer's disease (AD) represents a major public health problem and it is therefore crucial that modifiable risk factors be known prior to onset of dementia in late-life. The "Australian National University - Alzheimer's Disease Risk Index" (ANU-ADRI) is one of the potential tools for primary prevention of the disease. Objective The aim of this study was to devise an adapted version of the ANU-ADRI for use in Brazil. Methods The instrument was translated from its original language of English into Portuguese and then back-translated into English by bilingual translators. It was subsequently reviewed and evaluated as to the degree of translation issues and equivalence. In this study, the ANU-ADRI was applied using individual (face-to-face) interviews in a public hospital, unlike the original version which is applied online by self-report. The final version (pretest) was evaluated in a sample of 10 participants with a mean age of 60 years (±11.46) and mean education of 11 years (±6.32). Results The intraclass correlation coefficient (ICC) (inter-rater) was 0.954 (P<0.001 for a confidence interval (CI) of 95%=[0.932; 0.969]). Cultural equivalence was performed without the need for a second instrument application step. Conclusion After cross-cultural adaptation, the language of the resultant questionnaire was deemed easily understandable by the Brazilian population.


INTRODUCTION
A lzheimer's disease (AD) is one of the most prominent public health issues. Around 8 million new cases of dementia are diagnosed each year, with AD being the most prevalent type. 1 A recent meta-analysis has shown that the global prevalence of this form of dementia has doubled every 20 years in a number of countries with aging populations. 2 The estimated global cost of the disease was US$ 604 billion in 2010, which is comparable to the economic burden of cancer and cardiovascular diseases. 3 Multiple factors are associated with the risk of developing AD. 4 The lifetime risk factors for AD may vary, with middle age representing a critical period for changes in some of these factors. 5 While a number of clinical or lifestyle-related factors such as a low educational level (less than 12 years), diabetes, hypertension in middle age, obesity in middle age, depression, dyslipidemia, and smoking are modifiable, biological or genetic factors such as age, gender, and apolipoprotein (APOE) ε4 genotypes are not. 6 The protective factors with stronger evidence are cognitive activities or reserve, physical activity, and engaging in other leisure activities that can stimulate social and cognitive aspects. 7 Known protective factors are: Mediterranean diet (high in Omega-3 fatty acids) and moderate alcohol consumption, both lifestyle-related. 8 Focusing on the risk of AD as opposed to diagnosis, the Australian National University -Alzheimer's Disease Risk Index (ANU-ADRI) could be used as one of several primary prevention strategies. 9 The original items of the ANU-ADRI were developed in English, precluding their use in other countries such as Brazil. Thus, the literature recommends the process of translation and cross-cultural adaptation using rigorous and widely used methods. 12 Finally, there are no papers published on cross-cultural adaptation of the ANU-ADRI for the Portuguese language. The instrument has not yet been used in specialized Neurology, Geriatrics or Psychiatry outpatient services or primary care (where patients are seen by general practitioners), and its applicability in Brazil has not yet been adequately assessed.Our objective in this study to devise an adapted version of the ANU-ADRI for use in Brazil.

METHODS
Ethical considerations. Permission to translate and adapt the instrument was granted by the researchers at the Centre for Research on Ageing, Health and Wellbeing (CRAHW) of the Australian National University (ANU) and by Dr. Kaarin J. Anstey (Email: kaarin.anstey@anu. edu.au). This research project was reviewed and approved by the institutional review boards of the "Universidade Federal de São Paulo" -UNIFESP and "Escola Paulista de Medicina" -EPM (applying institution). Subsequently, the "Centro de Educação em Saúde"(Health Education Center) Research Ethics Committee of the Health Department of Curitiba approved the project regarding the feasibility of access to the research venue. The approved research project is available on the "Plataforma Brasil" database (CAAE registry No. 38185614.5.1001.5505).
Design: cross-cultural adaptation study Instruments. A questionnaire was administered for sociodemographic and clinical data collection, after which the MMSE was applied for patient screening. The 84-item ANU-ADRI was the principal questionnaire. The Australian National University Alzheimer's Disease Risk Index (ANU-ADRI) is a self-report instrument to assess: 11 risk factors and 4 protective factors for Alzheimer's disease. These factors are shown in Figure 1.
The total score was obtained by summing the scores assigned by category to the risk or protective factors in Table 1. In this study, the ANU-ADRI was administered via individual face-to-face interviews as opposed to applying the original online self-report version.
Participants. A convenience sample comprised 10 patients selected from a public hospital in Curitiba, Brazil, between July and August 2015. All participants were aged >40 years with no evidence of severe or disabling disease at the outpatient service of "Hospital do Idoso Zilda Arns" (HIZA).
• TBI (No=0; Yes=4): self-report (history for TBI with loss of consciousness). The exclusion criteria were a history of any severe visual or hearing impairment or marked psychomotor disability (e.g. parkinsonism); a history of severe psychiatric or neurological disorders; severe clinical or psychiatric disorders; dementia and/or below expected score on the Mini-Mental State Examination (MMSE) adjusted for level of education (cut-off scores: 20 for illiterate persons; 25 for those with at least one year of schooling). 11 No patients were excluded from the study. All patients were instructed as to the study goals and provided written informed consent prior to instrument evaluation.
Procedures. The translation and cross-cultural adaptation were standardized. 12 The first step was a forward translation of the English version of the ANU-ADRI instrument into Brazilian Portuguese by two bilingual health professionals (DI and RNS), both of whom were Brazilian, independent, and aware of the main goal of the study. Translation focused on conceptual rather than linguistic equivalence of the items. The translations were compared by an expert review committee (two psychiatrists, MKB and VAC, and one geriatrician, AFJ) and a consensus version was formulated from these two translations.
In the next step, the consensus translation was back-translated into English by a third bilingual translator -a native English speaker with proficiency in the Portuguese language (ACD). Subsequently, this version was compared with the original English language instrument. The back-translation showed semantic equivalence and consistency in the translated items. The final step consisted of ensuring cultural equivalence and involved the same expert panel who participated in the semantic equivalence process. This step is justified by the fact that a word or statement used with a given intent in the original context may not convey the same meaning to the target population of the new version. The expert committee reviewed the questions with regard to content validity and prepared a summary version for pretesting in order to assess the acceptability and understanding of the instrument by the target population. The aim of this step was to identify the questions that were not understood or not answered by the target population. To that end, a "not applicable" answer choice was included alongside each question. The ANU-ADRI instrument was administered by five trained interviewers yielding data in the pretest phase.
Statistical analysis. When evaluating screening instruments, test indices may be inaccurate if samples are too small. The aim of testing reproducibility is to assess the random fluctuations noted for the same individual interviewed on multiple occasions. Each patient was assessed by a different interviewer in a random manner (inter-rater).
Intraclass correlation coefficient (ICC) was the specific method used to estimate inter-rater reliability, with values around 1 indicating good agreement between answers. 13 ICC is defined as "the proportion of variance of an observation due to between-subject variability in true scores" and in order to offset potential errors, a 95% confidence interval or standard error should be adopted. 14 The data were analyzed using the SPSS version 20.0 statistical package. Table 2 shows the sociodemographic characteristics. Mean pretest MMSE and ANU-ADRI scores of the sample are given in Table 3.

RESULTS
Cross-cultural equivalence was achieved with no need for a second step in the instrument evaluation process. The intra-class correlation coefficient (inter-rater) of the ANU-ADRI was 0.954 (p<0.001, 95% CI=[0.932; 0.969]).
Most of the individuals interviewed (90%) needed the interviewer's assistance to read the items concerning "Physical Activity" and "Depression" domains, yet had no difficulty choosing the answers. None of the 84 questions were deemed "not applicable" by the respondents.
The mean time for administering the ANU-ADRI was 25 (±5) minutes. The final adapted Portuguese version can be found as (Supplementary material).

DISCUSSION
The cross-cultural adaptation process is a complex task, as it entails following a series of steps until functional equivalence is achieved. The most adequate adapted version for the older Brazilian population underwent minor changes in the items concerning the Physical Activity domain based on the International Physical Activity Questionnaire (IPAQ) scale adapted for older Brazilian individuals 15 and validated for older women 16 and men. 17 Table 2. Sociodemographic characteristics of the sample.

Sociodemographic characteristics Categories (%) Participants (n=10) Mean (SD) Range (years)
Age 60 (11.46) (41-77) Education 11 (6.32) (0-18) Cultural equivalence presupposes a literal match between the original and adapted versions, and should address the impact a term would have in the cultural setting of the target population sample. The expressions used in item 50 ("shoveling snow") and item 60 ("doubles tennis") of the source questionnaire were excluded following cultural adaptation as they would not be as meaningful in the adapted version for the Brazilian population. To allow the use of IPAQ-based questions for older individuals, examples were included of activities that are common for persons in this age group.
It is important to know the particularities of the sample considering the fact that respondents' gender, age group, and level of education could have influenced the performance of the instrument. Older adults with a low educational level had more difficulty understanding the items related to the "Depression" domain based on the CES-D scale validated in Brazil 18 compared to adults with a higher level of education.
A randomized, controlled study including a sample of middle-aged adults (mean age=55 years) with a high educational level (mean=18 years of education) revealed a lower mean ANU-ADRI score (-1.38) than that found in our sample (6.3). 19 The author of the cited study noted that the ANU-ADRI should be tested in "target samples with lower levels of education and higher ANU-ADRI risk scores".
In the present study, the mean administration time was longer (25±5 minutes) than that of the original ANU-ADRI, reported to be around 15±5 minutes. The short version of the ANU-ADRI can be considered an alternative whenever the original version is deemed too lengthy or "not applicable" due to limitations in admin-istration time. 20 The assessment of older people via questionnaires (either self-administered or in the form of interviews) is a particularly difficult task due to the inaccuracy of the information given and susceptibility to recording or recall bias. In the present study, the ANU-ADRI was administered using individual interviews as opposed to the self-report approach of the original online version. The original version, the short form and the Portuguese version are available for printout from the internet. 21 In conclusion, after the cross-cultural adaptation of the ANU-ADRI, the wording of the instrument was found to be easily understandable by the Brazilian population.
Furthermore, the final adapted version can facilitate international cooperation projects that employ this instrument.
In the near future, the authors of the present study intend to publish the validity and test-retest reliability of the ANU-ADRI with a larger sample (n=100) of participants from the same institution in Brazil.

Author contribution. Marcus Kiiti Borges and Vanessa de
Albuquerque Citero participated in the study design, obtained the data and wrote the manuscript. Alessandro Ferrari Jacinto revised the versions of the questionnaire and the paper. Borges