Tailoring STrAtegies for RelaTives for Black and South Asian dementia family carers in the United Kingdom: A mixed methods study

Abstract Objectives We culturally adapted STrAtegies for RelaTives (START), a clinically and cost‐effective intervention for dementia family carers, for Black and South Asian families. It had previously been delivered to family carers around the time of diagnosis, when most people with dementia had very mild, mild or moderate dementia. Methods We interviewed a maximum variation sample of family carers (phase one; n = 15 South Asian; n = 11 Black) about what aspect of START, required cultural adaptation, then analysed it thematically using the Cultural Treatment Adaptation Framework then adapted it in English and into Urdu. Facilitators then delivered START individually to carers (phase two; n = 13 South Asian; n = 8 Black). We assessed acceptability and feasibility through the number of sessions attended, score for fidelity to the intervention and interviewing family carers about their experiences. We used the Hospital Anxiety and Depression Scale. to examine whether immediate changes in family carers' mental health were in line with previous studies. Results In phase one we made adaptations to peripheral elements of START, clarifying language, increasing illustrative vignettes numbers, emphasising privacy and the facilitator's cultural competence and making images ethnically diverse. In phase two 21 family carers consented to receive the adapted intervention; 12 completed ≥5/8 sessions; four completed fewer sessions and five never started. Baseline HADS score (n = 21) was 14.4 (SD = 9.8) but for those who we were able to follow up was 12.3 (SD 8.1) and immediately post‐intervention was 11.3 (n = 10; SD = 6.1). Family carers were positive about the adapted START and continued to use elements after the intervention. Conclusions Culturally adapted START was acceptable and feasible in South Asian and Black UK‐based family carers and changes in mental health were in line with those in the original clinical trial. Our study shows that culturally inclusive START was also acceptable. Changes made in adaptations were relevant to all populations. We now use the adapted version for all family carers irrespective of ethnicity.


| INTRODUCTION
Family members providing care to a person with dementia are often psychologically distressed, with up to 40% showing clinical levels of depression and anxiety. 1,2 We developed the START (STrAtegies for RelaTives) intervention for family carers to prevent and manage these symptoms, and found that it is effective and cost-effective in reducing depression and anxiety in family carers of relatives with dementia in the short term and 6 years later. [3][4][5][6] It had previously been delivered to family carers around the time of diagnosis, when most people with dementia had very mild, mild or moderate dementia. The intervention is based on a manual and delivered one-to-one by supervised, trained facilitators, who are not clinically qualified. It works by addressing the carer's coping strategies and increasing emotion-focused coping or decreasing the use of dysfunctional coping over time. 7 The eight sessions comprise behavioural management, communication strategies, identifying and changing unhelpful thoughts, accessing emotional support, increasing pleasant events, relaxation, future planning, and developing a maintenance plan.
Black, South Asians and people from other minority ethnic groups often present to services at a later stage of dementia elated to culturally different understanding of dementia. 8 Reasons for this include feeling little can be done, previous negative experiences of services, stigma and shame, concerns about the breach of confidentiality, not regarding symptoms of dementia as illness, and a reluctance to reveal private information or share family care with people outside the family. [8][9][10][11][12][13] This may mean that people with dementia and their families miss out on support while being excluded from services that may feel unavailable or culturally irrelevant. 14 By the time people contact services, people living with dementia may have higher care needs and families may already be coping with behavioural changes Therefore, ethnicity and culture, as sources of shared norms and values, are important in the dementia care pathway and affect how people with dementia and their families accept or respond to health interventions and cultural adaptation may make them more acceptable Without an intervention being acceptable it cannot be effective but there is reason to think that if people will adhere to a dementia intervention with its core therapeutic elements still incorporated it will continue to be effective. 15 Although the START intervention and manual have been standardised and made available online with a training programme for trainers, 16 they have not been specifically adapted or tested specifically with carers of diverse ethnicities and cultures in the UK. In the original trial one-fifth of participants were from minority groups and they were noted to stop the intervention disproportionately, compared to other groups. 3 Therefore, we aimed to lay the foundations to widen access to START within the two largest minority ethnic groups of older people in the UK, Black and South Asian groups. In phase one we aimed to identify what aspects of the START intervention need to be culturally adapted for Black and South Asian dementia family carers. In phase two we aimed to test the culturally adapted START intervention for feasibility and acceptability, as well as to consider whether it was as effective as in the original START trial. 15

| METHODS
We conducted a mixed-methods study to tailor the START intervention for the UK's South Asian and Black communities' family carers of relatives with dementia and assess its implementation. The study received ethical approval from the London-West London & Gene Therapy Advisory Committee Research Ethics Committee (18/ LO/0369). All participants gave written informed consent.

| Phase one-Changes to the manual
Family carers were recruited from the London Dementia Network, NHS Memory Services in London which were part of mental health trusts who were identified in trusts and gave consent to be contacted by researchers. We visited Bradford based community organisations to recruit carers. We conducted one-to-one qualitative interviews with South Asian (n = 15) and Black (n = 11) carers. One Patient and Public Involvement (PPI; TB) representative reviewed the topic guides, provided feedback on interview analysis, and made suggestions regarding changes to the manual, such as highlighting that dementia is a disease of the brain rather than something "mental" that is "all in your head." We incorporated changes from the interviews in START (carer and therapist version) words and illustrations using the knowledge generated.
We translated START materials in writing and audio-recording into Urdu (a common South Asian language in the UK) to enable access to those without literacy or those who speak Urdu, not English. A translation agency initially translated a session, but we found that the translation was not appropriately worded, and when asked to retranslate this did not improve. Instead an Urdu-speaking psychology graduate (SJ) translated it, and NM supervised and edited this. We refined the manuals iteratively through SJ delivering the sessions to NM, which were then audio-recorded in Urdu. There was no one word for Dementia in Urdu, so we spelt "Dementia" phonetically.

| Phase two-Intervention delivery and evaluation
We trained three facilitators to deliver START, aiming to deliver to 8-10 South Asian and 8-10 Black carers in English, and three South Asian carers in Urdu. We recruited carers from three NHS trusts memory services in London throughout 2019 who were identified to participate in services after they presented and agreed to be contacted for research. They were then contacted by the research team and gave informed consent to the study. We purposively sampled people with different demographic characteristics, relationship to their relatives with dementia (e.g., spouse or child), and living situations. We collected demographic information from all participants (age, sex, ethnicity, relationship to the person living with dementia, level of education, work and whether they lived with the person with dementia).
The clinically supervised facilitators were two women and one man, all had a psychology degree, one was bilingual and could deliver the intervention in Urdu. We interviewed facilitators after they had stopped delivering START sessions.
We evaluated acceptability through adherence and participants' views of the intervention We defined adherence as attending ≥5 sessions. 3 We collected views through qualitative interviews using a topic guide (see appendix) after the final START session and 1 year late, also asking if they continued to use the intervention and if so what components. We audio-recorded and transcribed interviews verbatim and continued until interviews yielded no new substantive information (theoretical sufficiency).
We audio-recorded a randomly chosen session. Two independent researchers rated the facilitator's delivery fidelity using a sessionspecific checklist and gave an overall fidelity score for each session ranging from one to five, with five being high. 3 We measured the carers' mood at baseline (January-December

| Analysis
In phase one, two researchers independently double-coded all phase one qualitative interviews, using thematic analysis. We analysed the interviews about changes using the Cultural Treatment Adaptation Framework (CTAF; Figure 1) to guide the thematic analysis of phase one interviews 21 covering: � the core elements of the intervention (e.g., coping strategies, oneto-one delivery, relaxation exercises), � the intervention's engagement, that is, the ability of the intervention to reach participants and involve them successfully in the intervention, � and the delivery of the intervention, that is, making core components understandable and acceptable.
In phase two, to consider the feasibility and acceptability of the adapted intervention, we described the number of sessions completed and the fidelity. We described the demographics of the carers in the intervention study, the HADS and HSQ mental health score their means, and the differences in mean scores, as well as the number of cases of both anxiety and depression on the HADS, at baseline, immediately post-intervention, and 1-year post-intervention. We

| Demographics
In phase one, we interviewed 26 dementia family carers; 15 from South Asian backgrounds and 11 Black carers (see Table 1 for detailed demographics). We recruited people from London or Bradford, of either sex, from a range of countries, with a range of first languages, relationships with the person for whom they cared for and living situations, education levels and employment status. We interviewed all but two South Asian carers in English. We interviewed one carer in Urdu and one in Sylheti with an interpreter. Some carers had received START and were familiar with it. We explained the structure to others and asked them to look at one or two sessions in depth.
We did not make any changes to the core components of the intervention but to peripheral elements that render core components more acceptable or understandable, such as materials and semantics,  Table A1 with detailed summary of themes and illustrated them with quotes).

| Materials and semantics
Throughout the manuals, we simplified language to increase clarity and allow for lower level of English literacy. We modified graphics to be ethnically diverse, and names in vignette to be ethnically neutral.
We referred to carers as "family member caring for a relative with dementia" as the term "carer" implied for some people being paid to care rather than a relative. When discussing communication strategies, we changed the word "assertive" to "clear."

| Cultural examples and themes
We added a sentence in the beginning to emphasise cultural competence: "I may come from a different background to you, and we will have different experiences, but I have experience of working with people from a range of backgrounds." We added testimonials around the stressful nature of caring, as several interviewees said that many people in the South Asian community did not acknowledge this. We modified a vignette to illustrate feelings of guilt about "not being a good daughter" associated with wanting a break from caring.

| Confidentiality and wider family
People from both the Black and South Asian community had concerns about confidentiality as often the dementia diagnosis is not known outside the immediate family, so we added a prompt to highlight that everything said is confidential, as in all health settings.
Because more than one family member was likely to be caring, we added a prompt at each session's end: "What from this session would you want to share with other family members involved in caring for your relative?" We said that while the same person had to attend every session, some sessions could be attended by other relatives; and provided or directed carers to resources from the voluntary sector to be shared with family members.

| Dementia and services
We added more in the section "overview of memory loss" highlighting dementia as a "physical" illness and not "just a mental health issue," considering the stigma attached to mental health issues. We added a prompt for facilitators when discussing future care options such as admission to a care home, to highlight that "No one plans to have a member of their family in long-term care" but this may sometimes be safest and around short-term respite options "as little as an afternoon."

| Demographics
Thirty-six carers were referred, 21/30 (70%) of those who were eligible and contactable consented -13 were South Asian and 8 Black (see Table 2 for carer characteristics, Figure 2 for flow through the study and one session (South Asian). They stopped because two could not fit sessions around childcare, one wished to attend a group and one was unwell.
We initially delivered all sessions face-to-face. The last three participants had sessions seven and eight delivered via telephone while using the START manual for reference, due to the COVID-19 lockdown in March 2020. We audio-recorded 12 participants' sessions and their mean fidelity score was 4.4/5 (SD 0.5); four participants were not audio-recorded as they did not complete the selected sessions.

| Follow up of carers who received the adapted START intervention
Immediately post-intervention, 10/16 carers were followed up with quantitative measures and a qualitative interview. The carers who had not begun START declined follow-up (4) and the fifth carer was too unwell for an interview. The others were ill (n = 1), declined Twelve carers were not followed up: five who never started the intervention did not want follow-up, three others declined, three did not respond and one carer was too ill.

| Quantitative outcomes
The baseline HADS score amongst all participants (n = 21) was 14.4 (SD = 9.8; range 0-33), and for those who started the intervention   What went well when delivering START All three described enjoying delivering the intervention, and that they felt it helped them build confidence in delivering interventions and therefore benefitted their career.
Going into a stranger's home, was a bit scary at first. One facilitator discussed remote delivery at the start of the COVID-19 pandemic and did not think it was detrimental as they had met participants face-to-face previously but it did bring new challenges.
And one of them… used to come to the office, but obviously he couldn't. They can't do the sessions in front of the person they care for, so they went and sat in the car, but the exercise was a stretching exercise which just doesn't work.

Delivery of the intervention
The ethnicity of the facilitator and their cultural competence mattered to some participants as they felt that some challenges were culturally driven and could be better understood by the facilitator who shared that knowledge. Other carers felt that talking to someone outside their cultural group was easier, or the facilitator's ethnicity did not make a difference.
I think just understanding the culture helped and also, she was so friendly… she made it really easy to talk to her… she got it.

What was helpful or not (from the core components of the intervention)
Overall, the carers interviewed discussed how the START intervention had made them feel less alone in their caring role and had had a positive impact on them and the person they were caring for 1 year later. Carers also felt that the intervention came at the right for them.
I found out I wasn't alone, and I learnt these tech-

Delivery of the intervention
At the 1-year post-intervention, carers described that they preferred the one-to-one delivery of the intervention at the time they received it, but a year on they may have preferred a group setting, due to the progression of their relative's dementia and also because they felt ready to share their experiences.

What has been gained or changed
Some carers said they shared components of the manuals with other relatives or friends who were caring for people living with dementia.
One carer also discussed how they now had a social worker and a paid carer helping to look after their relative as a direct result of the intervention.
My wife has used some of the stuff as well. We've gone through it, and my daughters as well.
[  15 We had envisaged that we would produce different tailored manuals for each of the groups, however after the changes suggested we did not feel this to be necessary. We found that the changes suggested, such as making names culturally neutral and pictures ethnically diverse, emphasising confidentiality, that caring was hard, and that facilitators were culturally competent, could benefit people from different backgrounds.
We are now using this culturally adapted version of START for everyone. This is because as the interviewees did not suggest changing the core components, which are those we regard as therapeutic, we do not expect these changes to impact the effectiveness of the intervention although it might improve the acceptability in all groups. 15 Fidelity to the intervention was high. Of those who began sessions, 75% were adherent suggesting the intervention was acceptable. This is higher adherence than in the original START trial, where 61% of minority ethnic carers attended ≥5 sessions. 3  The themes in the study did not appear to be unique to minority ethnic carers but some were more common in these groups. They wanted facilitators with cultural competence, the intervention delivered with consideration of their other caring and religious duties and had concerns about privacy and confidentiality. Familism was more common, with carers for example, more frequently than our experience has been in the majority white population prioritising their duty to look after family members, who are children, for their parent in school holidays, over an intervention to help them cope.
Again, this occurs in all groups. The ethnicity of the facilitator was important to some carers, but some people wanted carers of the same and some of different ethnicity. We have not asked this question of white UK carers.

| Strengths and limitations
We were able to interview people with a range of demographic characteristics and reach theoretical sufficiency although there will WEBSTER ET AL.
-11 of 13 always be some views that are not captured. Our qualitative analysis used a culturally appropriate framework and two independent raters for reliability. We were, however, only able to follow up with 10 people recruited after the intervention, and 1 year later this dropped to nine. As seven out of 10 of those who had case level depression completed the intervention we know that it was acceptable both for some people who were depressed and some who were not. This was partly related to the beginning of COVID-19 when people stopped services, were unused to communicating online, including filling in questionnaires, or the person with dementia and carers were ill. We do not know if those not seen were systematically different.
For the qualitative interviews in both phases, the researchers who collected and analysed the data were from a variety of cultural backgrounds, including sometimes similar backgrounds to the participants who were interviewed. They considered the effect their ethnicity and gender had and as laid out some participants were explicit that they preferred someone of the same characteristics as themselves and others liked them to be different. Overall the researchers were t a mix of insider and outsider researchers, and there are pros and cons to both positions, 23 including in terms of minimising bias.
We used the HADS and HSQ to measure both anxiety and depression and health status, and both of these questionnaire have been used in studies with dementia carers previously. 24,25 We used the English versions of both measures which may be a limitation, and though each, particularly the HADS, has been translated and validated in other languages and used within different cultures, the evidence on their cross-cultural validity is scarce. 26 We had no control group in this trial, unlike the original RCT where the control group's mental health deteriorated, thus accentuating the between-group difference. Nevertheless, it is appropriate to use an acceptability and feasibility trial rather than a full controlled trial is not necessary to test an intervention that has not modified its therapeutic components. 15

| CONCLUSIONS
The culturally adapted START intervention for family carers from Black and South Asian British backgrounds has the potential to help underserved UK populations. It was feasible and acceptable, with similar effectiveness to the original trial. Clinicians should consider the calendar for religious holidays, childcare, and other caring responsibilities when they offer START. There are many differences within as well as between ethnicities and the START changes address issues relevant to the majority population, so the changes benefit all ethnicities. This is now the available online version irrespective of ethnicity.