Planning and optimising a digital intervention to reduce older adults’ cognitive decline

Background By 2050 worldwide dementia prevalence is expected to triple, rising to 152 million. Affordable, scalable interventions are required to support protective behaviours such as physical activity, cognitive training and healthy eating. This paper outlines the development of ‘Active Brains’: a multi-domain digital behaviour change intervention to reduce cognitive decline amongst older adults, and key findings arising from this process. Methods A theory-, evidence- and person-based approach to intervention development was undertaken. Scoping reviews and behavioural analysis contributed to intervention planning. Optimisation involved qualitative interviews with 52 older adults with higher and lower cognitive performance scores. Data were analysed thematically and informed changes. Results The development process synthesised findings from planning and optimisation activities. Scoping reviews and qualitative interviews suggested that the same intervention content should be suitable for individuals with higher and lower cognitive performance. Qualitative findings revealed that maintaining independence and enjoyment motivated engagement in intervention-targeted behaviours, whereas managing ill health was a potential barrier. Social support for engaging in such activities could provide motivation, but was not desirable for all. These findings informed development of highly acceptable intervention content and functionality for target users. Conclusions A digitally-delivered intervention with minimal support appears acceptable and potentially engaging to older adults with higher and lower levels of cognitive performance. As well as informing our own intervention, the insights obtained through our development process may be useful for others working with, and developing interventions for, older adults and/or those with cognitive impairment. theory-, evidence- person-based approach to intervention development that numerous behaviour change contexts. systematic the current form, how it is expected to work. In doing so, transferable into the acceptability of a digital multi-domain intervention to reduce cognitive decline amongst older adults with a range of cognitive performance abilities.


Abstract
Background By 2050 worldwide dementia prevalence is expected to triple, rising to 152 million.
Affordable, scalable interventions are required to support protective behaviours such as physical activity, cognitive training and healthy eating. This paper outlines the development of 'Active Brains': a multi-domain digital behaviour change intervention to reduce cognitive decline amongst older adults, and key findings arising from this process.
Methods A theory-, evidence-and person-based approach to intervention development was undertaken. Scoping reviews and behavioural analysis contributed to intervention planning.
Optimisation involved qualitative interviews with 52 older adults with higher and lower cognitive performance scores. Data were analysed thematically and informed changes.
Results The development process synthesised findings from planning and optimisation activities.
Scoping reviews and qualitative interviews suggested that the same intervention content should be suitable for individuals with higher and lower cognitive performance. Qualitative findings revealed that maintaining independence and enjoyment motivated engagement in intervention-targeted behaviours, whereas managing ill health was a potential barrier. Social support for engaging in such activities could provide motivation, but was not desirable for all. These findings informed development of highly acceptable intervention content and functionality for target users.
Conclusions A digitally-delivered intervention with minimal support appears acceptable and potentially engaging to older adults with higher and lower levels of cognitive performance. As well as informing our own intervention, the insights obtained through our development process may be useful for others working with, and developing interventions for, older adults and/or those with cognitive impairment.

Background
Fifty-million people worldwide currently have dementia (1). Cognitive impairment is even more common; Mild Cognitive Impairment (MCI) and Age-Associated Cognitive Decline (AACD) are estimated to affect nearly 20% of adults aged 60 and over (2,3). Around 10% of MCI and AACD cases convert to dementia each year (4). The annual global cost of dementia is nearly US$1 trillion with dementia prevalence expected to rise to 152 million by 2050 (5). Dementia and cognitive decline place unsustainable demand on health and social care systems worldwide, and pose substantial threat to individuals' independence and quality of life (6). Prevention and management of dementia are public health priorities (7).
Interventions targeting a single behaviour in individuals with and without existing cognitive impairment show some positive effects on cognitive outcomes (12)(13)(14)(15)(16). However, findings are mixed and often inconclusive, prompting investigation of multi-domain intervention strategies (7). Multidomain interventions have also shown mixed results (17)(18)(19). Despite positive effects of a face-to-face (group and individual) delivered programme addressing diet, physical activity, cognitive training and managing vascular risk (17), such interventions tend to be resource-intensive, prompting calls for scalable, cost-effective approaches (7). Understanding which intervention components are useful and how to improve cost-effectiveness is a key challenge (Livingston et al., 2017).
Digital health-behaviour interventions have excellent potential to deliver content efficiently, effectively and accessibly at low cost (20). There is currently limited evidence about the potential of digitally-delivered multi-domain interventions to protect cognitive health. One pre-post design study showed promising effects of a digital intervention addressing behaviours including physical activity, diet, smoking, alcohol intake, and sleep, but did not measure cognitive outcomes (21). Furthermore, there is limited evidence about whether digital-delivery of interventions is feasible, engaging and acceptable in this context. Potential barriers to feasibility and acceptability relate to users' cognitive capacity and digital literacy. It is important to explore whether individuals with cognitive decline have different preferences and requirements for intervention functionality. Furthermore, whilst older adults' digital literacy is rapidly growing, there is still wide variation in ability and/or willingness to engage with digital health material (22,23). It is therefore important to explore whether digital content and functionality can be made accessible and engaging for these users, and how best to achieve this. This paper outlines the development of 'Active Brains': a low-cost, digitally-delivered, multi-domain intervention to reduce cognitive decline amongst older adults. Active Brains is for 60-85 year-olds with and without existing cognitive impairment, and aims to reduce cognitive decline by addressing physical activity, cognitive training and healthy eating behaviours. Clear reporting of the development of new interventions avoids 'research waste' and duplication of ineffective, unfeasible or unacceptable interventions (24). In documenting this development process we aim not only to provide a clear account of how the intervention was created, but also to provide evidence about whether a digital intervention is a feasible, engaging and acceptable means of delivery in this context. The paper is broadly split into 'planning' and 'optimisation' sections that run through the 'Research Design and Methods' and 'Results' sections, reflecting the two phases of development. The 'Planning' section presents the theory-, evidence-and person-based 'Guiding Principles' and logic model that underpin intervention content and functionality. The 'Optimisation' section presents our qualitative findings about older adults' perceptions of cognitive health and associated protective health behaviours, as well as their feedback on all aspects of intervention content throughout development. We explain how these findings allowed Active Brains to be shaped by target users' expectations and preferences, whilst highlighting transferable insights and methods that could be applied across numerous behaviour change contexts.

Methods
Structure of the development process Active Brains was developed according to a theory-, evidence-and person-based approach to intervention development (25)(26)(27). The knowledge generated from these approaches was triangulated to inform 'guiding principles' (27) and a logic model outlining the programme theory underpinning the intervention. The development process was implemented in two phases: planning and optimisation.
Although described separately, in practice these phases occurred as an iterative cycle (Fig. 1).
Development focused on the physical activity and cognitive training intervention elements as the team had previously developed a healthy eating module that could be adapted for use in this context (28). Additional Table 1 summarises how each element of the Active Brains development process addresses recommended actions for intervention development (24). Table 1 The Active Brains Guiding Principles Key findings from Literature Key Design Objective Intervention Feature(s) Older adults with cognitive impairment tend to experience difficulties in the domains of memory, language, thinking and judgement. Difficulties not so extensive that the individual requires assistance with activities of independent daily living (49,53) Older adults with cognitive impairments that may affect Internet use are still actively engaging with technology (54) Good evidence of effectiveness and/or acceptability of various features/characteristics of interventions: -Simple goal setting and action planning with clear explanation of benefits/ importance (55) -Reinforcement/encouragement for achievements (50) -Self-monitoring of physical activity behaviours, e.g. using a pedometer (47,55) -Social support in the form of activity suggestions to be done with others/ local group recommendations (47,50); social element of cognitive training may also be beneficial (51) -Promotion of autonomy (50) Strength and balance exercises can be built into daily routines and activities (56) Minimising cognitive load and dependence on technology • Clear and simple layout, language and navigation procedures • Support provided for cognitive self-regulation (e.g. planning, reminders, prompts for periodic short-term and long-term selfmonitoring) for comment and iteration.
The following sections outline the methods employed in: 1) planning the intervention's theoretical framework; and 2) the empirical qualitative work conducted to optimise Active Brains. Respective findings/outcomes from each of these processes are reported in the results section.

Planning Active Brains
The planning phase aimed to build the appropriate theory-, evidence-and person-based framework to underpin the Active Brains intervention. This involved: reviewing relevant literature, developing guiding principles, conducting a behavioural analysis and constructing a logic model.

Reviewing relevant literature
We conducted rapid scoping reviews of: 1) physical activity and/or sedentary behaviour interventions, and 2) cognitive training interventions, for older adults with and without cognitive impairment. We  Table 2). Additional literature was identified through reference-list searching and consultation within our team. Quantitative and qualitative papers were included. Initial searches returned over 9000 matches about cognitive training interventions. Therefore, we focused only on systematic reviews (n = 14). Data were extracted about research design, sample size and characteristics, and findings. Although each subsection contained a section addressing concerns, it was deemed important to bring shared concerns forward to the introductory material so people felt happy to proceed with the intervention content Some pages (and this applied throughout) perceived to be a bit cluttered/ busy with too much text which some found off-putting "I immediately look at this page and find it untidy and as a, not a struggle, but as a barrier there to reading it clearly and understanding it. I'm struggling to find what to click to go to next." (P0245) Focusing on identified problematic pages, we edited text to a minimum. Wherever possible text was bulletpointed and only key messages retained. If important to keep all text on a page, this was split over multiple pages where appropriate.
Getting Active Uncertainty about goal setting: some seemed unsure about exactly what they had set themselves goals to do even when goal setting process complete "… it's good to have goals, but I think the goals need to be specific. If you're asking people to achieve a goal that's very vague, I don't think they're enthusiastic and I think they give up and they probably give up the whole thing." (P0111) After revisiting the activity suggestions made in this section, it was considered that this uncertainty was likely to be arising from the fact that the activity suggestions and plans that people could select were a little too broadthese were amended to more specific options for people to choose from Many mistaking coloured, bolded text (to emphasise key messages in text) for hyperlinks and expected links to additional content "…anything else that's in blue, you think you can click on it"(P0102) We removed the colouring of these parts of text, but retained the bolding for emphasis. The minimisation of text to key messages only also addressed this issue.

Strength and Balance
Concern that information provided about the principles of how strength and balance exercises worked (including information about specific movements such as shifting weight to one side) was potentially risky if attempted by those who were less mobile.
"…if you have somebody with poor balance, it's just trying to ensure, how do you ensure that someone who shouldn't really be standing on one leg doesn't stand on one leg, despite what you've said about being safe." (P0106) The text in this section was reframed to describe the underlying principles of the exercises without reference to specific examples that may be dangerous if attempted by someone with poorer mobility/balance. Instead it now talks about how the suggested activities allow practice of movements to expand individuals 'comfort zone' in terms of movements they can make. Disagreement with advice that if users are unsure about whether suggested activities are suitable for them, then to consult with their GPusers not comfortable with the idea of taking up GP time with these types of queries.
"It's… I always find this information about checking with your doctor before you start interesting, because I very… well, I say that. I very seldom make plans to go and visit the doctor.
And I certainly wouldn't regarding this, I think." Revised to reassure users that the activities recommended were nothing outside of routine movements made in dayto-day life and that they were likely the best judge of whether they could safely/comfortably do these Also encouraged to of queries. regarding this, I think." (P0105) these Also encouraged to discuss with those who knew them well, and only seek advice from GP for serious concerns. The wording of some True/False quiz questions was considered confusing, e.g. one stated that the purpose of brain training games was to keep improving your score. If/when people answered 'True', they were surprised when their answer wasn't considered correct "Yeah, well that's, that automatically to me should be true but it's, you explaining it, but it's not clear. The point of brain training is for you to get better, any training is to get better, but it said it's false." (P0104) To ensure participants remained engaged with these quiz questions and did not take away the wrong message, wording/ feedback was revisited and amended where necessary. In this example, the feedback was clarified to state that whilst a good aim to try and improve scores, the important factor is continuing to practice these games whether or not your score improves Given different structure of the Brain Training section compared to other sections, a page preceding the Brain Training menu explained how to use menu page, but this created confusion "It's a bit confusing, this one. I don't know quite why. This bit I think might be more beneficial on the next page?" (J0112) This page was removed and the navigation around the Brain Training menu pages was revised to ensure that it was clear how users could access each element of the Brain Training module -most importantly the link to the Brain Training games.

Development of Guiding Principles
Guiding principles aim to maximise the acceptability of an intervention amongst target users and, therefore, to enhance engagement and effectiveness. Each guiding principle comprises: 1) a design objective outlining a user/context-specific behavioural need, issue or challenge; and 2) intervention features that address the design objective (27). To draft provisional guiding principles, we drew on our understanding of target users obtained from the scoping reviews and from our research team, including PPI members. These guiding principles were iteratively developed as new data emerged, e.g., from the behavioural analysis and qualitative interviews.
Developing Active Brains programme theory A programme theory explicitly describes how an intervention is expected to achieve its intended outcomes, and the anticipated mechanisms through which this occurs (29). The behavioural analysis and logic model contributed to this process.

Behavioural Analysis
The behavioural analysis aimed to identify behaviours to be targeted by Active Brains and their potential barriers and facilitators. We recorded relevant evidence from scoping reviews, team

Optimising Active Brains
The optimisation phase aimed to seek feedback on draft intervention material, and to explore the acceptability and feasibility of the digital content and functionality amongst older adults with higher and lower levels of cognitive performance. Due to the vast quantity of relevant literature available to inform initial content development, primary qualitative research was delayed in favour of planning and drafting initial content (34). This allowed us to seek feedback sooner, and meant we could still explore target users' experiences to inform changes required.

Think-aloud interviews
Forty-one adults (22 female, mean age = 70.5 years, range 61-80) were recruited from GP practices across the South of England and from Join Dementia Research (JDR; an online database for matching UK community-dwelling individuals to relevant studies) to take part in think-aloud interviews. We employed purposive sampling whereby we attempted to obtain maximum variation in terms of gender, age, education level, socio-economic status and cognitive performance score. Participants were excluded if they were already reasonably physically active (score > 30 on Godin Leisure Time Exercise Questionnaire (35)), had diagnosed dementia, a severe uncontrolled mental health condition, or terminal illness. As part of the screening process, participants completed a brief cognitive assessment (online Baddeley verbal reasoning task (36)), which determined whether they were identified as a participant with 'lower cognitive performance' or 'higher cognitive performance'. Lower cognitive performance was defined as a score falling more than one standard deviation below the 'normative score', determined by a database of scores from a large (n > 10,000) pre-existing cohort of older adults. Although this single test was not indicative of cognitive impairment, this categorization enabled us to sample views from those with higher (n = 20) and lower levels (n = 21) of cognitive performance.
Each participant took part in one think-aloud interview in which they worked through the prototype Active Brains intervention with an interviewer. The participant was encouraged to vocalise all immediate thoughts and feelings toward the content. This allowed insight into target users' immediate reactions to elements of the intervention. As it was unfortunately not possible for participants to access the brain training games, we provided screenshots to show the types of task this involved. These existing games have been extensively used in ongoing cohort studies, and study investigators report them being well liked and engaged with. Following the think-aloud interview, there were semi-structured interview questions about participants' general views of the intervention: what they liked/disliked, found helpful/difficult, would like to change etc. All interviews were audiorecorded and transcribed verbatim.
Data were analysed to understand user views on the intervention content and inform potential changes. We collated all positive and negative comments pertaining to specific intervention elements into a 'table of changes' (Additional Table 3). After discussing the frequency and significance of positive and negative comments, we coded the importance of possible changes by deciding whether any amendment was likely to enhance the persuasiveness, acceptability and likelihood of changing behaviour (37). For example, we considered: whether multiple people provided the same feedback; if the potential change aligned with our guiding principles and/or expert opinion; and whether theory and/or evidence suggested the change would make the desired behaviour more likely. We prioritised changes by their relevance to behaviour change or ability to prevent disengagement. If changes were low-priority, they were implemented only if relatively quick and easy. Interviews continued alongside this analysis to allow iterative modification of content prior to the next batch of interviews. Once it seemed that no further important changes were required, we considered that data saturation had been reached (37).

Longitudinal qualitative feasibility study
This second element of the qualitative work commenced once the majority of prioritised modifications to the prototype had been made. Eligibility, sampling and recruitment procedures were the same as in the think-aloud interviews. Eighteen older adults (12 female, mean age = 69.1 years, range 62-76) took part, seven of whom had participated in the think-aloud interviews. The eighteen participants were classified evenly across the lower and higher cognitive performance groups (n = 9 in each group).
Participants were invited to use Active Brains for three weeks (timings were 'sped-up' to allow access to all sections) and were given a diary to keep notes about their experiences. Participants took part in one semi-structured interview each during this time. The interview asked participants about their experiences of engaging with the intervention and any relevant activities they tried. They were prompted to discuss certain features or elements that they particularly liked and/or found helpful or disliked and/or found difficult. Towards the end, there were questions about participants' perceptions and understandings of cognitive health, and their views on social support for engaging in new activities. All interviews were audio-recorded and transcribed verbatim.
All data were tabulated and analysed as described in the think-aloud study. In addition, inductive thematic analysis (38) was conducted on the data from the second part of the interview examining perceptions and understandings of cognitive health and social support.

Planning Active Brains
Reviewing relevant literature Given the wealth of existing reviews on the topics of interest, the findings were not formally synthesised for writeup. However, several findings pertinent to our research questions are summarised in Table 1, which also illustrates how they informed intervention guiding principles. There was no substantial evidence that the intervention's physical activity recommendations should differ for older adults with MCI/AACD compared to a general older-adult population. The cognitive training intervention evidence suggested training multiple cognitive-domains to be the optimum choice for both cognitively-healthy older adults and those with cognitive impairment (e.g. 39,40). Regarding physical activity interventions, those with and without cognitive impairment shared similar attitudes towards physical activity, and recognised similar barriers (e.g. remembering, social isolation), facilitators (e.g. accessibility of activity options, simple activities) and preferred activities (e.g. walking) (41,42). There was only a small amount of evidence about intervention features that may be acceptable and engaging for both groups. Acceptable intervention features amongst those with cognitive impairment often overlapped with those frequently used in interventions for older adults in general (e.g. planning features; 43).
Otherwise, there was little evidence about whether engagement with intervention features was likely to differ between groups, so we aimed to explore this within our primary qualitative work.

Development of Guiding Principles
The finalised Active Brains guiding principles (Table 1)  The full behavioural analysis is presented in Additional Table 4. Active Brains targeted nine behaviours: initial engagement with the online intervention; increasing physical activity; reducing sedentary behaviour; uptake of strength and balance activities; uptake of brain training; healthy changes to eating behaviours; reviewing behaviours and revising goals; integration of recommended activities into daily routines, and; maintaining engagement with the online intervention. These behaviours were further broken down into 19 sub-behaviours required to enact each behaviour. Mapping these behaviours, their determinants, and intervention features onto the BCW and TDF illustrates that Active Brains employs 36 BCTs to deliver seven intervention functions (modelling, education, persuasion, training, enablement, environmental restructuring, incentivisation) to target thirteen behavioural domains (intentions, optimism, emotion, knowledge, skills, beliefs about consequences, beliefs about capabilities, goals, social influences, environmental context and resources, reinforcement, memory, attention and decision processes and behavioural regulation).

The Active Brains intervention logic model
A summary version of the Active Brains logic model is shown in Fig. 2. Additional Fig. 1 shows the full version with intervention processes mapped on to BCW, TDF and BCTs. The Active Brains digital intervention comprises three online modules that become available sequentially: 'Active Lives' (physical activity) is available immediately; 'Brain Training' (cognitive training) is available after 4 weeks; and 'Eat for Health' (healthy eating) is available after 8 weeks. 'Active Lives' is further divided into three sub-modules: 'Getting Active', 'Strength and Balance' and 'Breaks from Sitting' with recommendations about which to start with tailored to users' baseline activity and capability. Within each module, users can access: information addressing common concerns, instruction about recommended activities, goal setting and review for chosen activities, and tailored motivational feedback on progress. Reminder emails are sent to motivate users to continue with their activities and to encourage them to revisit online content. Additional support from a central facilitator (for one arm of Active Brains trial) comprises up to three 10-minute phone calls at two-week intervals, plus additional email support if required. This can be used to discuss behavioural changes participants are attempting, and to support them with use of the online intervention content. The facilitator employs the CARE (Congratulate, Ask, Reassure, Encourage) approach to provide support in a broadly standardised format (44). After seven months, the Active Brains 'booster section' allows users access to additional resources for embedding recommended activities into daily life. It also introduces the brain training 'boosters' to maintain the benefits of the initial intensive training period.

Optimising Active Brains
The findings of the qualitative work are described below. These fed back into ongoing iteration of the guiding principles, and behavioural analysis and also informed required intervention changes. Table 2 summarises feedback and the resulting changes implemented. Feedback on the Active Brains prototype was encouraging. Users were particularly positive about what they considered to be more novel activities including strength and balance training, and brain training games. Less positive feedback included sections where users found navigation confusing, a lack of specificity surrounding physical activity goals, and a desire to address health-related concerns earlier. We analysed feedback for differences between those with lower and higher cognitive performance scores to determine whether different intervention features or characteristics may be more engaging or desirable for those with lower cognitive performance. There was no evidence of any substantive differences.

Think-aloud interviews
Longitudinal qualitative feasibility study Part one: Table of changes analysis Collating feedback into a second table of changes confirmed that the amendments based on the initial thinkaloud interviews were well received, with the original issues no longer being raised. In general, there were a smaller number of negative comments about the intervention content, but a few remaining points were identified and addressed (Table 3). There were no substantial differences in the views expressed by individuals with higher and lower levels of cognitive performance. Table 3 Summary of feedback and amendments resulting from longitudinal interviews Summary of issue identified Example Change implemented Some voiced opinion that they felt physical activity content wasn't relevant to them as they perceived themselves to already be physically active (despite not meeting exclusion criteria for existing high levels of physical activity) "For me, it was the actual activities, the actual physical bits, weren't terribly challenging." (P0146) Additional messages added in to introductory and early physical activity content to emphasise that even those who are already active can use content to help them increase activity, and to stress importance of continuing with/ increasing activities they already do and enjoy. A few suggestions that it would be useful to have more explicit suggestions about ways to stay motivated with making behavioural changes -particularly about how to use social support to do so "The one thing I think you could do a bit on is finding the incentive to do all these things, so that we've got to do them to keep going, but you tend to put them off because you're doing other things at the time." (P0265) Extra pages added to give examples of motivational strategies, including ways to involve others (e.g. weekly step-count competition with friends/family) and activities to boost motivation (e.g. 'Reasons to be Active' card). Some perceived healthy eating content to be largely in line with what they were already doing. Furthermore, expert advice from within the team recommended placing greater emphasis on the 'foods for brain health' as novel/interesting element.
"I read it through with interest and I thought: Oh, well, I do that; I eat that. I agree with all of that and that's what I do; but there was nothing in there that I felt that I didn't already do." (P0250) Restructure of the healthy eating content so that the initial information and goal setting centres around specific foods beneficial for cognitive health, with additional more general healthy eating advice presented after this.
Part two: inductive thematic analysis The inductive thematic analysis generated three overarching themes, comprising several subthemes. These were: 1)'knowledge and understanding of brain health', including subthemes 'the meaning of brain health', 'perceived availability of information about brain health' and 'knowledge of determinants of brain health'; 2) 'motivators and barriers', including the subthemes 'motivations for achieving/maintaining good brain health', 'motivators for engaging in helpful behaviours', and 'barriers to engaging in helpful behaviours'; and finally 3) 'the role of social support' including subthemes 'desirability of social support' and 'motivational mechanisms of social support'. Each theme is briefly summarised with illustrative quotes from the data. These findings helped to refine the intervention guiding principles and behavioural analysis.
Knowledge and understanding of brain health This theme suggests that, for older adults, 'good brain health' is largely about maintaining independence and remaining able to do the activities one wishes to do. More than half of participants also discussed retention of specific cognitive skills such as good memory and decision-making.
"If you've got good brain health, then you can carry on with your daily life: cooking, managing your finances, managing your social life -you know, day-to-day things, really." (P0229, female, 68, higher cognitive performance) A large proportion of individuals felt that, whilst information about cognitive health and how to protect it is available, it often requires one to actively look for it. Many also mentioned the availability of information about body health, but not necessarily about brain health.
"So you do need to know about it. But you have to make the effort to either read a newspaper or look at the news, or get your brain active yourself." (P0265, female, 69, lower cognitive performance) Despite this, nearly three-quarters of participants named typically promoted strategies for maintaining cognitive health, such as brain training activities and puzzles. Half of participants also acknowledged the role of healthrelated behaviours, such as physical activity, in maintaining cognitive health.

Motivators and barriers
Two different types of motivation were identified within participants' accounts. The first were motivations to maintain good brain health in order to avoid cognitive decline and its anticipated negative consequences, such as loss of independence, poor quality of life, and interference with relationships. This was often accompanied by accounts of friends or family with dementia and their strong wish to avoid this.
"It's a tremendous thing, for me anyway, because I've seen other people go through it. I don't want to, […] It is frustrating for other people as well as for yourself. I think it's important not just for you, but it's also important for the rest of the family, and to be able to pass the memories on as well." (P0225, female, 65, lower cognitive performance) The second type of motivation related to factors that encouraged individuals to engage in behaviours important for maintaining cognitive health. The overwhelming sentiment was that enjoyment is the main motivator. Even when individuals acknowledged that behaviours were beneficial for brain health, this seemed an 'added bonus' rather than the primary motivator.
"…yeah, you know, I do a lot of things like maths games. And crosswords and stuff like that every day, so I don't know if that actually helps but I just find them interesting." (P0138, male, 70, higher cognitive performance) Barriers to engaging in activities to support cognitive health were not discussed extensively, but the most common difficulty mentioned was managing other health conditions. "I'm quite hampered with physical activity because I've got arthritis and am registered disabled so, to be honest, physical activity is so difficult for me. That's where these exercises come in, really and it's mostly what I can do." (P0261, male, 62, higher cognitive performance) The role of social support Participants who discussed involving others in healthy lifestyle activities mentioned several mechanisms through which this provided motivation for beginning and maintaining activities. This included creation of action plans with others, being accountable to others and sharing encouragement and new ideas.
"I think, if you're going swimming or something once a week, it's nice if someone says, 'Are you ready to go?' 'Shall we go today?' rather than you think: Oh, do I really want to go today? If there's two of you or three of you wanting to go, you encourage each other." (P0229, female, 68, higher cognitive performance) However, it was widely acknowledged that individuals' preferences and circumstances determine whether involvement of others is possible, or even desirable. More than half of participants expressed that they would be happy (or sometimes prefer) to do such activities alone. "I'm quite happy with my own company. I mean, I enjoy doing things with other people, and I go to yoga and I get on with everybody there, and I've got quite a few friends that go, but I would go whether they went or not." (P0129, female, 67, lower cognitive performance).

Discussion
This paper presents a theory-, evidence-and person-based approach to intervention development that could be applied across numerous behaviour change contexts. We have provided a systematic account of how and why the intervention took its current form, and how it is expected to work. In doing so, we have provided valuable transferable insights into the acceptability of a digital multi-domain intervention to reduce cognitive decline amongst older adults with a range of cognitive performance abilities.
This study has provided preliminary empirical evidence that a digital multi-domain behaviour change intervention appears acceptable and engaging amongst UK community-dwelling older adults. Our findings reinforce and extend existing literature. Importantly, there were no substantive differences between those with higher and lower cognitive performance scores regarding their preferences or requirements for the Active Brains intervention, or their ability and willingness to engage with the digital intervention. This finding reinforced our judgement (informed by the initial literature review) that the Active Brains digital intervention was accessible and engaging for people with lower cognitive performance scores and did not need to be tailored according to cognitive status. This also aligns with previous findings (39,40), suggesting that the same activity recommendations might be suitable for those with and without cognitive impairment, given that they appear to share similar motivations, barriers, and attitudes. Our findings extend limited evidence about preferred intervention features (43), by demonstrating that those suitable for a general older adult population appear engaging and acceptable for those with lower cognitive performance scores too.
Similarly, our findings about older adults' perceptions of cognitive health generally align with the existing literature: maintaining independence (45) and enjoyment of activities (46) seem key motivators amongst this group. We found that an awareness of, and desire to avoid, the consequences of dementia also seemed to motivate cognition-protective behaviours, whilst other health conditions arose as a possible barrier. Additionally, our findings extend understandings about the value of social support. Whilst they concur that, for many, social support is an important motivator of behaviours such as physical activity (47), they reveal that for many it is not considered necessary or desirable. This has important implications for offering social support within interventions, i.e. it should be available but not a compulsory element. Furthermore, as social support didn't appear to be a primary motivation for engagement with the intervention, this lends additional support to the potential feasibility and acceptability of a digital-delivery format. Overall, our findings largely concur with the literature that informed our preliminary intervention guiding principles and behavioural analysis and so largely confirmed the priorities for intervention functions and features. The qualitative data also provide valuable, detailed feedback that has informed the optimisation of Active Brains to maximise the likelihood of intervention engagement and effectiveness.
This paper addresses numerous calls to more clearly articulate the intervention development process, and the resulting intervention's expected mechanisms of action (33). It is important to test these mechanisms to provide new evidence about behavioural determinants and the most effective intervention functions to target them (33), particularly in multi-domain interventions. This information enables advancement of theoretical understandings of behaviour change in diverse contexts (48). The extensive qualitative research is a further strength of this rigorous intervention development process. The qualitative interviews provided in-depth understanding of target users' preferences and life-context to maximise acceptability. These participants represent a wide-ranging community-based group, purposively sampled from JDR as well as primary care. As participants' understandings of cognitive health were sought during the period that they had access to Active Brains, their views could have been influenced by their experience of the intervention. However, these questions explored perceptions of cognitive health in a broad sense rather than how they related to specific intervention content. Indeed, participants' responses more often involved accounts of their own beliefs and experiences than reference to Active Brains.
With the development of Active Brains now complete, the next step is a feasibility trial (n = 360, in progress) to test the intervention and trial procedures. A fully powered trial (n = 20,000) then aims to determine whether the Active Brains intervention can successfully reduce the incidence of dementia amongst older adults over a 5-year period. The systematic development process provides scope for later process evaluation to test the mechanisms through which the intervention is expected to work.

Conclusion
This study has begun to address the need for a rigorously-developed, low-cost, multi-domain behaviour change intervention for maintaining older adults' cognitive health. It presents the theory-, evidence-and person-based framework that arose from the planning of the intervention, as well as the primary qualitative evidence that helped to optimise acceptability of intervention content and functionality. As well as facilitating optimisation of intervention content, the qualitative data contribute a greater understanding of older adults' perceptions of brain health, and the barriers and facilitators to engaging in preventative behaviours. In doing so, this study has provided evidence that a digitally-delivered intervention with minimal support appears acceptable and potentially engaging to older adults with higher and lower levels of cognitive performance. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Further examples and extracts from data/analysis are available in additional files.

Competing interests
The authors declare that they have no competing interests.

Additional Files
Additional