Developing and implementing guidelines on culturally adapting cognitive tests: A qualitative illustration with the Addenbrooke’s Cognitive Examination Version III (ACE-III)

Background: We aimed to devise a methodology on developing and implementing guidelines for culturally adapting any cognitive test. We demonstrated this through developing guidelines for culturally adapting the Addenbrooke’s Cognitive Examination Version III (ACE-III) and utilising them to develop an ACE-III Urdu for a British South Asian population. Methods: We had a several stage qualitative study. We collated information from our systematic review on the translation and cultural adaptation of the ACE-III and its predecessors, and feedback from previous adaptors of the ACE-III to identify steps for cultural adaptation. We formatted these into question-by-question guidelines. These guidelines were used with feedback from focus groups with cognitively healthy older British South Asians and a consensus meeting with relevant experts, to develop an ACE-III Urdu. Results: Our systematic review found 32 adaptations and we received feedback from seven adaptors to develop guidelines for culturally adapting the ACE-III. Through the use of these guidelines, two focus groups with 12 participants, and a consensus meeting of two psychiatrists with a South Asian background and familiarity with cognitive assessments and cultural adaptation, the ACE-III Urdu was developed. Conclusions: We developed a set of guidelines for culturally adapting the ACE-III that can be used by future adapters for their own language or cultural context. We demonstrated how guidelines on cultural adaptation can be developed for any cognitive test and how they can be used to culturally adapt the measure.

influencing the perception of test questions and individual's responses to these questions [4]. Due to a frequently large number of questions within cognitive tests relying on the respondent's' culture, seen most commonly when assessing domains such as language and memory, we see this tendency for susceptibility more prominently in this type of assessment [5].
However, cognitive tests, like all health measures, are not designed for the populations they are being implemented across, having been developed in English for European countries; they do not accommodate for non-English speakers or ethnic populations residing there [3]. This can account for a higher rate of false positive and false negative scores across cognitive tests within these groups as compared to their Caucasian or English speaking counterparts, cultural bias in the administration and responses of the cognitive tests, and the compromising of the generalisability of these results [4,6,7]. This creates a disparity in the cognitive tests available to us versus the diverse populations in need.
Attempts at identifying a robust solution have proposed designing new cognitive tests but due to such a process being time consuming and complex it is not considered the most feasible option [8]. A common suggestion has been to adjust cut-off scores for different minority ethnic and non-English speaking groups but this has been criticised for reducing sensitivity, specificity and likelihood ratios [9]. Therefore, culturally adapting an existing cognitive test has been regarded as a preferred alternative.
At present, there is no universally standard procedure for undertaking the cultural adaptation of any given cognitive test [10]. Current approaches undertake the translation of health measures into a target language but this does not account for the influence of culture on the perception of a health measure beyond the requirement of fluency in the target language. Within cognitive tests we see questions assessing for orientation biased towards the western calendar or memory tasks requiring familiarity with western names and western history [5]. When cultural adaptation is not undertaken such bias occurs along with a loss of conceptual equivalence, [11] which results in cognitive tests not culturally suitable for minority ethnic or non-English speaking populations. Therefore, there must be a global consensus on undertaking thorough cultural adaptation of any health measure, including cognitive tests, before administering it to a target population that differs from the population it was originally designed for [4]. We propose the development of guidelines on culturally adapting any cognitive test according to a robust methodology to address this consensus, incorporating a review of previous literature [11] with the feedback of those that have already adapted the cognitive test. These guidelines would provide step by step instructions on how to culturally adapt every item of a cognitive test in accordance with evidence to allow for the retention of conceptual equivalence [11].
To illustrate how such guidelines can be formed we demonstrated the development and implementation of guidelines for culturally adapting the Addenbrooke's Cognitive Examination Version III (ACE-III) [12]. It is a gold standard tool for the diagnostic accuracy of cognitive impairment [13], consisting of 19 items that assess the cognitive domains attention, memory, fluency, language and visuospatial abilities. The ACE-III and its predecessors, the ACE [14] and ACE-Revised (ACE-R) [15] have been translated into a range of languages and incorporated into use across the globe. English versions of the ACE-III have also been adapted for the UK and the US. However, the ACE-III was originally designed for English speakers native to Australia, with a reliance on knowledge of the cultural background [12] and although cultural adaptation has been undertaken by adaptors to produce suitable adaptations [16,17,18] there are no existing standardised guidelines for the cultural adaptation of this cognitive test.
We endeavoured to develop such guidelines and implement them for a non-English speaking minority ethnic population within the UK, demonstrating the development and implementation process of such guidelines for any cognitive test. As South Asians are the UKs largest minority ethnic group, at over 6.3% of the overall population, we selected them as our target population to culturally adapt for. We

Methods
A several stage qualitative approach was undertaken to develop guidelines for translating and culturally adapting the ACE-III (See Fig. 1) and implement them to develop an ACE-III Urdu (See Fig.   2): Step 1: A systematic review Step 2: Receiving feedback from previous ACE-III adaptors Step 3: Collating the data to form guidelines Step 4: Implementing the guidelines with feedback from lay persons and experts' in the field Step 5: Developing the ACE-III Urdu Step 1: Systematic Review We conducted a systematic review of all existing primary publications of translations and cultural adaptations of the ACE-III as well as its predecessors, with the full methodology described elsewhere For each of the publications we extracted data on the version of the ACE that was culturally adapted, the language it was translated into, the country it was culturally adapted for and the section of the text that described its cultural adaptation process. These reported processes were broken down into individual steps and grouped by ACE-III question, allowing us to identify which questions were culturally dependent, how they had been culturally adapted, and what the rationale was behind the changes.
We also assessed the quality of the reported cultural adaptation of each publication utilising the Manchester Cultural Adaptation Reporting Questionnaire (MCAR) [23], which showed which publications had reported their cultural adaptation process in sufficient detail to be replicated by future adapters.
Step 2: Feedback from adaptors of the ACE-III We aimed to receive feedback from official adaptors of the ACE-III who had translated and culturally We distributed the questionnaires to the corresponding adaptors attached with a standardised email relaying the purpose of the questionnaires and a request for their completion. We also requested a check of our translation of their language version. After a two week period adaptors were sent a follow up email to act as a reminder. If adaptors did not initiate any form of contact after this no further contact was made.
Step 3: Data analysis and synthesis To develop guidelines for translating and culturally adapting the ACE-III we collated the information from our systematic review [23] and the feedback from adaptors of the ACE-III to identify sets of mutually exclusive steps that can potentially be undertaken to culturally adapt each question of the ACE-III.
From our systematic review we had broken down the cultural adaptation processes extracted from each publication according to ACE-III question. The adaptation processes for each question across publications were merged, with duplicates removed to identify the mutually exclusive steps that could be undertaken to adapt each question.
The questionnaires sent to adaptors were already organised by question and we followed the same methodology of merging the adaptors' feedback on the cultural adaptation process of each question.
We removed duplicating information such that for each question we had mutually exclusive steps that could be undertaken to adapt that question along with the adaptors' accompanying rationale.
The cultural adaptation steps for each ACE-III question identified from the systematic review and from the adaptors' feedback were merged. Duplicates were removed to identify overall mutually exclusive cultural adaptation steps for each question. Accompanying rationale was presented with these steps and the respective publications and adapted versions of the ACE-III were cited, resulting in a questionby-question set of guidelines.
Step 4: Implementation of the guidelines We conducted two focus groups within the British South Asian community of Greater Manchester. We aimed to recruit 12-14 laymen participants overall, fluent in speaking and writing Urdu, over the age of 60, able to give informed consent and who did not have a history of cognitive impairment.
Participants were voluntarily recruited via convenience sampling from the local Pakistani Community Day Centre and provided with an information sheet, available in English and Urdu. They were given 24 hours to decide if they wished to participate, after which they were contacted by a liaison at the Centre to confirm their participation and let them know the date and time of the focus group, which was also held at the centre. On the day of the focus group participants would be provided with consent forms and demographics sheets, available in English and Urdu.
Using the guidelines we produced several culturally adapted versions of all questions of the ACE-III, backed up by rationale, for the British Urdu speaking population from which the most suitable option would need to be selected. We presented these versions of the questions within our focus groups to receive their feedback on the questions' cultural appropriateness, which versions should be retained for a potential ACE-III Urdu and whether they proposed any changes or suggestions of their own. This feedback was audio recorded and transcribed.
We also conducted a consensus meeting with experts within the relevant fields, local to the Greater Manchester area. We aimed to recruit 2-4 experts on dementia, the cognitive testing process and the translation and cultural adaptation of these tests, who were familiar with the ACE-III, its rationale and how to administer it. These experts also had to be fluent in speaking and writing both English and Urdu and familiar with UK and South Asian cultures. The experts were recruited voluntarily via convenience sampling. The consensus meeting was held at the Centre for Primary Care and Health Services Research, at the University of Manchester, before which informed consent was obtained.
We presented the focus group feedback within the consensus meeting and experts determined which were the most appropriate and culturally suitable adaptations of each question of the ACE-III from the focus group suggestions. The consensus meeting was audio recorded, transcribed and the data was collated for each question of the ACE-III to determine how each would be culturally adapted.
Step 5: Developing the ACE-III Urdu To develop the ACE-III Urdu the template was acquired through NeuRA, allowing the ACE-III Urdu to retain the exact same format at the ACE-III. Urdu is read from right to left so the template was reversed horizontally such that questions were presented on the right side of the template and the scoring instructions on the left side. Standard information requested prior to the administration of the ACE-III and instructions for the implementation of the questions were translated into Urdu and typed out. Each ACE-III question was typed out and designed according to the suggestions confirmed within the consensus meeting.

Results
Step 1: Systematic Review The full results of the systematic review are described elsewhere [23].

Step 2: Feedback from adaptors of the ACE-III
Our search of the NeuRA website identified 17 fully adapted versions of the ACE-III for the languages Egyptian Arabic, Saudi Arabian Arabic, Chinese, Estonian, Hebrew, Hindi, Hungarian, Indian Kannada, Italian, Japanese, Marathi, Polish, Portuguese, Spanish, Tamil, Telugu and Indian Urdu. Three versions had been retained in English but culturally adapted for India, New Zealand and the US of America, resulting in a total of 20 ACE-III adaptations.
Of these adaptations the Estonian, Indian Kannada, Marathi, Japanese, Tamil and Telugu versions could not be translated into English due to a lack of resources in terms of translation applications and translators and were thus excluded from our analysis (30%). Questionnaires were developed for the remaining 14 adaptations (70%) and distributed to their respective adaptors, of which a total of seven questionnaires (35%) were returned to us fully completed.
The original Australian ACE-III was utilised by the Hindi, Hungarian and Spanish adaptors (15%) and the UK version of the ACE-III was utilised by the Egyptian Arabic, Hebrew and Welsh adaptors (15%) for their own adaptations. Polish adaptors utilised both (5%). Table 1 summarises which questions of the ACE-III were culturally adapted by which adaptors, thereby showing the frequency of reported cultural adaptation undertaken for each question. We can see that all adaptors culturally adapted questions 6, 7, 18 and 19 for memory and questions 10, 11 and 14 for language, and the majority had adapted question 2 for attention. In contrast, none of the adaptors had culturally adapted any of the questions assessing visuospatial abilities. This highlights which cognitive domains, and their respective questions, rely on culture and which would suffice with a simple translation into the target language.

Step 3: Data analysis and synthesis
For each question of the ACE-III, the individual cultural adaptation steps identified from our systematic review and from adaptors' feedback, along with rationale undertaken, were tabulated to form the guidelines (See example in Figure 3 and see Appendix A1). For each question the following was presented: How the question has been previously culturally adapted with the steps undertaken.
Examples compiled from publications and the questionnaires, citing the respective languages and adaptors of the ACE-III The rationale behind adapting the question and choosing the adapted replacement.
Step 4: Implementation of the guidelines Our focus groups had 12 voluntary participants, five female (41%) and seven male (59%), from ages 61-75 years (M=66.67, SD=6.44), from the Greater Manchester area. (See Table 2 for a breakdown of participant demographics). Participants came from varied socio-economic backgrounds.
Through our proposed suggestions for each question of the ACE-III Urdu, developed through the use of the guidelines, we determined items 5b, 8, 9, 15a, 15b, 15c and 16 would suffice with a direct translation whereas the remaining questions required further cultural adaptation that were deliberated over during these focus groups (See Appendix A2 for the proposed suggestions, developed using the guidelines).
Our consensus meeting was attended by two experienced old age psychiatrists who were both bilingual British Pakistanis who had lived in both the UK and Pakistan and were familiar with the cultures of both countries. They were also both involved in clinical and research work relevant to South Asian populations and were knowledgeable about cognitive assessments, the ACE-III and the translation and cultural adaptation of health questionnaires.

Step 5: Developing the ACE-III Urdu
The suggestions finalised within this consensus meeting and incorporated to form the ACE-III Urdu [19] can be seen in Table 3.

Discussion
The combination of the systematic review and adaptors' feedback provided us with detailed information on the cultural adaptation of the ACE-III, which was utilised to develop question-byquestion guidelines. We compiled extensive adaptation processes to develop the guidelines, which present culturally adapted versions of questions of the ACE-III. We could potentially have incorporated more cultures and languages through translators but due to limited resources we were unable to produce questionnaires for 6 of the existing adaptations and we had a poor return rate of questionnaires (35%), reducing the amount of additional information we could have received and compiled into the guidelines.
Therefore we must acknowledge that the usefulness of these guidelines, and any guidelines for any health measure developed via these methods, is dependent on how many language and cultural versions of that health measure have already been developed and how accessible these versions are.
They are also limited by how many current adaptors of these versions can be requested to and are willing to provide rich data on the rationale behind culturally adapting questions that is often not conveyed through publications alone.
However, the methods for forming these guidelines are designed in such a way to allow for necessary updates as more information on cultural adaptation is acquired.
In addition, we were still able to account for 22 international languages and cultural contexts within our guidelines. Through the frequency of cultural adaptation across questions, evidenced in our systematic review [23] and from the feedback of the adaptors, we were also able to determine which questions would most likely require cultural adaptation and which could suffice with a simple translation and the guidelines highlight this. With the accompanying rationale these guidelines would allow future adaptors to conduct their own cultural adaptation of the ACE-III and we have demonstrated this through our cultural adaptation of the ACE-III for a British Urdu speaking population.
While utilising the guidelines to develop potential questions for an ACE-III Urdu [19] we acknowledged British South Asians' preference for using English words that are spelt with Urdu letters, such as, 'county', 'bell' and 'ball', as opposed to translating words into Urdu. This is attributed to the mixing of English and Urdu that occurs within British Urdu speaking communities. We also noted the influence of the structure in which sentences are presented in Urdu, and proposed the rephrasing of questions to avoid confusion. This can be seen with the elaboration of 'What is the season?' to 'Which of the four seasons is it?' due to the Urdu word for weather and season being the same.
We presented these suggestions within our focus groups with lay persons, allowing us to gather feedback from men and women who represented a vast array of educational backgrounds within the British Urdu speaking community. Throughout the discussion participants were able to follow the rationale provided by the guidelines when proposing suggestions for the ACE-III Urdu and there was a notable insistence on cultural adaptation for questions assessing memory and language, with little focus on fluency and visuospatial abilities.
Following this we conducted a consensus meeting with experts to review suggestions for the ACE-III Urdu proposed during the focus groups. There was a general consensus with the suggestions proposed, barring a few items. Within 'Question 2: Attention' experts preferred 'ball' be translated directly into Urdu and in the case of 'Question 7: Memory -Retrograde' experts decided to retain these questions as they were. The proposed suggestions were deemed too easy and the original questions were at a specific level of difficulty that was required to measure retrograde memory.

Conclusions
The guidelines are the first of their kind, and we have provided an in depth account of the novel approach we undertook to develop them. This was not restricted to published literature but incorporated the first hand experiences of cultural adaptation by existing adaptors of the ACE-III, which accounted for adaptations that may not have had corresponding publications. Furthermore, instead of adhering to general guidelines on cultural adaptation, developing guidelines designed for the cultural adaptation of a specific cognitive test allows for familiarity with the chosen assessment, in this case the ACE-III.
We have also demonstrated the usability of these guidelines by implementing them to devise an ACE-III Urdu [19] that is the first version of this language that can be implemented in the cultural context of the UK. The next step was to assess its suitability within the target populations in terms of cultural appropriateness and understanding, resulting in its cultural validation, which has been described with the full methodology, results and finalised ACE-III Urdu elsewhere [19]. Following this, further efforts should and are being taken to determine its performance in the detection of dementia within older Urdu speakers in the UK through a psychometric validation.
The implications of this methodology can be taken forward to develop guidelines in the same manner for other existing health measures, not limited to cognitive tests, enhancing the current standard of cross cultural research.

Declarations
Ethics Approval and Consent to Participate: Ethical approval was waived for this research by the University of Manchester Research Ethics Committee (UREC). This was because the portion of the research requiring participation of members of the public was an act of public involvement for research on developing an Urdu version of the ACE-III that was used in a cultural validation study [19].
However, informed consent was still obtained from all participants through a signed consent form.

Consent for Publication: Consent for publication was obtained verbally as the standard consent
forms provided by UREC at the time of this research did not have a statement specifically regarding publication. It was also simpler to explain the publication process, the anonymising of their published data and obtaining consent for it through speaking in Urdu. However, we did obtain consent to use their quotes anonymously through a signed consent form.

Availability of data and materials: Not Applicable
Competing Interests: The authors declare that they have no competing interests Funding: There were no sponsorship or funding arrangements relating to this research.

Author Contributions:
The research question was formulated and the study designed by WW and NM. The study was carried out by NM, with assistance from AM and MP, supervised by WW. The qualitative data analysis was conducted by NM, with assistance from MWW, supervised by WW. All authors contributed to the writing of the article. Egyptian Arabic   ii.
'Hospital' and 'county' will be spelt using Urdu letters. iii.
Only accept dates in the English calendar.
Participants and experts agreed with the proposed rationale from the guidelines. ii.
'Ball' is directly translated into Urdu Participants and experts agreed with the proposed rationale from the guidelines for i and ii. For iii, participants said 'bell' and 'ball' sound far that 'ball will be directly translated as the Urdu syllable.

3: Attention
Use the word 'minus', spelt using Urdu letters and the Urdu translation for 'take away'.
Participants and experts agreed with the proposed rationale from the guidelines. 4: Memory Replace the letter 'P' with the Urdu letter ‫چ‬ (chay).
Participants agreed with both proposed suggestio from the guidelines. Between the letters ‫چ‬ (chay) and ‫گ‬ (gaaf), the experts after debate due to its unique sound as the the similar sounding letter ‫ک‬ (kaaf). 6: Memory i.
The first name Haroon is used. The last name Butt is used. ii.
The original ACE-III address will be retained and spelt using Urdu letters.
Participants agreed with all the proposed suggesti from the guidelines. ii.
The fourth question will be replaced with 'Name of the princess who died in a car crash in the 1990s'.
Participants agreed with the proposed suggestio from the guidelines. The second question was retained according to t was said in the focus groups. Experts ruled that B should be aware of the first female Prime Mini prominence. She would have been Prime Minister a elderly would have initially immigrated to the UK her. The fourth question was replaced with a new Despite following the rationale of the guidelines 'N and 'Name of the city where (a 'Wonder of the Wor too easy. It was also agreed that these repla conceptual equivalence of the question. 'Name of the princess who died in a car crash in th relating to a well know historical death relevant to t 10: Language i. ii.
Book is retained. iii.
Kangaroo is replaced with a goat. iv.
Penguin is replaced with a peacock. v.
Anchor is replaced with scissors. vi.
Camel is retained.
Participants agreed with the proposed suggestio from the guidelines.
For iii, experts decided that though sheep are mor Urdu speakers would be familiar and able to recogn For iv, experts selected a peacock as British Urd familiar with it as opposed to a parrot. For v, exper are a common household object.
For vii, experts selected a dohl due to familiarity wi vii.
Harp is replaced with a dohl. viii.
Rhino is replaced with a bear. ix.
Barrel is replaced with a suitcase. x.
Crown is replaced with a cap. xi.
Crocodile is replaced with a tortoise. xii.
Accordion is replaced with a trumpet.
For viii, experts proposed the new suggestion of a b is a better known wild animal but still unique in the could be confused with other big cats such as a t relative to the cultural context of the UK.
For ix, experts selected a suitcase as it is a form purpose.
For x, experts proposed the new suggestion of a form of headwear.
For xi, experts proposed the new suggestion of a to known wild animal that would be better recognised elderly.
For xii, a trumpet was selected out of the proposed from the focus groups as it was considered the mos recognisable by British Urdu speaking elderly.

13: Language
The following questions were asked regarding the images: 'Which one is related to the head', 'Which one is found in the desert', 'Which one has a shell on it' and 'Which one is related to travel'.
All questions were developed by the authors NM an images that were finalised, following the guideline Participants and experts agreed with the proposed 14: Language The words used in the Indian Urdu ACE-III were retained.
Participants and experts agreed with the proposed rationale from the guidelines 17: Visuospatial Abilities The letters ‫م،ی،ا‬ ، ‫و‬ were selected.
Participants and experts agreed with the proposed rationale from the guidelines 18: Memory Refer to Question 6: Memory.
Refer to Question 6: Memory. 19: Memory Refer to Question 6: Memory. The names Jamal Butt and Haroon Khan replaced the original names for recognition.
Refer to Question 6: Memory. Participants agreed with the proposed suggestio from the guidelines. Overview of the process of utilising the guidelines to culturally adapt the ACE-III for British South Asians