Using Intervention Mapping to Develop and Facilitate Implementation of a Multifaceted Behavioural Intervention Targeting Physical Activity and Sedentary Behaviour in Stroke Survivors: Physical Activity Routines After Stroke (PARAS)

The benets of increased physical activity for stroke survivors include improved walking ability, balance and mood. However, less than 30% achieve recommended levels of physical activity, and high levels of sedentary behaviour are reported. We engaged stroke survivors, informal carers and healthcare professionals (HCPs) in a co-design process to develop an evidence-informed behavioural intervention targeting physical activity and sedentary behaviour for use by stroke rehabilitation teams. Mapping was used involved a systematic


Contributions To The Literature
There are recognised gaps in the literature and in practice on how to successfully support stroke survivors to engage in long-term physical activity and reduce sedentary behaviour. Current interventions are poorly described and target short-term improvements in physical activity without speci c links to activities of daily living.
Using intervention mapping, Physical Activity Routines After Stroke (PARAS) is an evidence-and theory informed co-designed intervention that addresses the limitations of existing interventions and incorporates training for healthcare professionals to facilitate delivery as part of routine care.

Background
Low levels of physical activity (1) and high levels of time spent sedentary (2) are common after stroke, irrespective of post-stroke disability. (3) Both are associated with an increased risk of cardiovascular disease and type 2 diabetes, (4) reduced life expectancy (5) and impact negatively upon mental health and well-being.
A wealth of research demonstrates the short-term bene ts of structured exercise interventions (commonly delivered in a group format) on cardiovascular risk factors (6) and function after stroke; (7) however, longer-term bene ts are under-researched. Furthermore, structured exercise interventions for stroke survivors often have little or no emphasis on everyday habitual levels of physical activity and sedentary behaviour. (7) Stroke survivors express preferences for undertaking previous or new activities that are meaningful, promote independence, and can be incorporated into their everyday lives, such as gardening (8) or walking to the shops. Consequently, structured exercise interventions with prescribed activities are not appropriate for a proportion of stroke survivors. Moreover, structured exercise interventions often do not focus on developing stroke survivors' self-management skills, which inhibits them from making sustained changes in behaviour beyond the intervention period. (8) The small number of randomised controlled trials (n=9) conducted that target long-term free-living physical activity after stroke show promise. (9) However, limitations in methodological quality and intervention design prevent any robust conclusions in this eld. (9) Speci cally, they lack adequate descriptions of intervention content (10) and delity assessment, (11) which restricts replicability and prevents successful implementation. There is also a dearth of interventions targeting reductions in sedentary behaviour alongside increasing physical activity of stroke survivors.
Furthermore, few interventions targeting physical activity and sedentary behaviour post-stroke have been developed with reference to theory, systematically developed or robustly evaluated. (9) The application of health behaviour change theory is critically important to gain a thorough understanding of the antecedents of the behaviours of interest, in order to develop targeted and effective interventions. (12)(13)(14) Intervention Mapping is a practical framework for systematic, evidence and theory-based planning for comparator group. (21) All of these interventions involved an element of supervised support that was tailored to individual needs. Both face-to-face and telephone contact were identi ed as promising modes of intervention delivery to engage stroke survivors in physical activity behaviour change. The number of contacts for promising interventions ranged from a single contact to 36 contacts, with the duration of interventions ranging from one single contact to twelve consecutive weeks. Nine promising behaviour change techniques were identi ed with reference to the BCT Taxonomy V1 (22) and considered for inclusion in our intervention: information about health consequences; information about social and environmental consequences; goal setting behaviour; problem-solving; action planning; feedback on behaviour; biofeedback; social support unspeci ed; and credible source.

Qualitative focus group discussions
To complement and add context to the ndings of our systematic review, a series of interactive focus group discussions were conducted with stroke survivors and healthcare professionals involved with their care. The overall aim of the focus groups was to identify determinants of behavioural change for both stroke survivors and healthcare professionals and to explore the barriers and enablers to engagement in long-term physical activity and reduction in sedentary behaviour post-stroke. Focus group discussions were conducted in person, audio-recorded and transcribed verbatim. Transcripts were read, re-read and analysed following the conduct of each focus group discussion and any unsubstantiated issues or points were further explored during subsequent group discussions (i.e. topic guides were revised accordingly).
Data were analysed using the Theoretical Domains Framework (TDF) (23), to facilitate an exploration of behavioural determinants likely to predict and impact upon behaviour and behaviour change. Three researchers independently read, re-read and analysed transcripts of stroke survivor and healthcare professional focus groups. The rst researcher has a background in stroke physiotherapy and rehabilitation research (SAM), the second is a chartered health psychologist with expertise in health behaviour change and qualitative research methods (LA), and the third was a master's degree student.
The skill mix of the researchers ensured appropriate questions were asked and responses were further probed to generate a comprehensive understanding of what was required from an intervention. Analyses of the data involved assigning text segments to one or more domains of the TDF and generating themes within each domain. Given the explicit nature of the TDF, all focus group transcripts with healthcare professionals and stroke survivors were coded and analysed by hand and no qualitative software was required. Common themes across the stroke survivors and healthcare professionals were subsequently established. Regular meetings were held with all three researchers to discuss their independent coding and analyses and to discuss any discrepancies until a consensus on the nal domains and domainspeci c themes was reached.

Findings of focus groups discussions
Eighteen stroke survivors and twenty-four healthcare professionals (HCPs) (physiotherapists n=14, technical instructors n=8, physiotherapy assistants n=2) participated across seven focus groups. All 14 of the theoretical domains of the TDF were identi ed from the data generated from stroke survivor focus groups. The TDF domains and themes are presented in Table 1. The most commonly populated domains were 'environmental context and resources', 'beliefs about consequences' and 'beliefs about capabilities' (Stroke survivor TDF domains themes and related change objectives are presented in Table 2, Step 2).
Data generated from HCP focus groups populated seven theoretical domains: 'knowledge', 'skills', 'social/professional role and identity', 'belief about consequences', 'beliefs about capabilities', 'reinforcement' and 'environmental context and resources'. The most populated domains were 'environmental context and resources' and 'skills' (HCP TDF domains themes and related change objectives are presented in Table 3, Step 2).
Although the aim of the focus group discussions with stroke survivors was to explore the determinants of behavioural change in relation to long-term physical activity and reduction in sedentary behaviour poststroke, participants focused their discussions on physical activity. This highlighted a potential lack of understanding and awareness about sedentary behaviour and the importance of this in the context of stroke and stroke rehabilitation.
Lack of sustainable physical activity options were identi ed as a key barrier by stroke survivors to engagement in long-term physical activity post stroke. A lack of timely information and long-term support was also reported. Enablers to increasing physical activity and reducing sedentary behaviour were identifying meaningful, accessible, sustainable activities with social support and developing skills for self-monitoring physical activity and well-being. HCPs also identi ed environmental context and access to resources as barriers to promoting physical activity and reducing sedentary behaviour of stroke survivors, as well as a lack of skills to effectively support behaviour change.

Exploration of intervention opportunities within existing pathways
To identify current stroke rehabilitation services and explore potential for delivering the intervention within existing pathways, a questionnaire was sent to local community stroke teams in the North East of England. Questions explored current and future sta ng, current service provision, practices around promoting and supporting physical activity and sedentary behaviour post-stroke and potential for participation in a future study. Community stroke services at seven NHS trusts in the North East of England were considered for inclusion in the exercise. Four of these trusts were already involved in another rehabilitation study led by the research team, therefore it was agreed that they would not be approached so not to over burden the teams.
The needs assessment highlighted that physical activity and sedentary behaviour are not adequately addressed post-stroke and HCPs do not feel equipped to target these behaviours effectively. The systematic review of randomised controlled trials targeting long-term physical activity and sedentary behaviour after stroke identi ed promising behaviour change techniques and intervention components to incorporate into a new intervention. The qualitative focus group discussions explored physical activity and sedentary behaviour of stroke survivors, and the promotion and long-term support of post-stroke physical activity behaviour by HCPs. Data were analysed using the TDF which enabled further identi cation of potential BCTs and selection of theory alongside the ndings of the systematic review. The needs assessment indicated that to increase physical activity and reduce sedentary behaviour after stroke, the intervention should be person-centred and adaptable to individual needs and preferences. A supported self-management approach was identi ed as a possible approach to target these requirements. Mapping of existing stroke rehabilitation pathways revealed there was potential to incorporate a physical activity and sedentary behaviour intervention and training for HCP into current practice.
Step 2: Identi cation of behavioural outcomes, and speci cation of performance and change objectives The needs assessment conducted in Step 1 identi ed the behaviours to be targeted by our intervention.
Target behaviours were identi ed as physical activity and sedentary behaviour of stroke survivors and HCPs knowledge about physical activity and the promotion of physical activity in the context of stroke, and skills to support behaviour change to optimise consultation behaviour. The two behavioural outcomes of the PARAS intervention and related performance objectives are reported in Table 1. Stroke survivor behavioural outcomes: To develop knowledge to raise awareness of the importance of physical activity in the context of stroke, and skills to increase and sustain activity levels and reduce sedentary behaviour in order to perform activities of daily living.

Performance objectives ·
Understands bene ts of physical activity and reducing sedentary behaviour after stroke · Requests support to increase physical activity and reduce sedentary behaviour at the most appropriate time · Selects and safely performs meaningful and sustainable physical activity and/or reduces sedentary behaviour · Identi es and utilises social support to maintain physical activity behaviour and reduce sedentary time · Applies behavioural goal setting, action planning, and coping planning to selected physical activities and/or reducing sedentary behaviour · Selects methods of self-monitoring physical activity and sedentary behaviour · Self-monitors physical activity and sedentary behaviour, behavioural goal attainment and associated con dence and well-being · Plans methods for maintaining physical activity or reducing sedentary behaviour HCP behavioural outcomes: To improve/increase knowledge about the bene ts of physical activity in the context of stroke and to develop skills to promote and sustain activity levels and reduce sedentary behaviour to enable stroke survivors to perform activities of daily living.
· Accepts supporting physical activity and reducing sedentary behaviour after stroke is bene cial for stroke rehabilitation and part of the HCP role · Supports stroke survivors to successfully engage in the PARAS intervention · Appropriately uses PARAS intervention resources to support stroke survivor engagement in the PARAS intervention · Appropriately uses behaviour change counselling techniques to support stroke survivor's identify reasons for physical activity behaviour change and maintenance Change objectives (i.e. aspects of behaviour individuals are required to learn, do or change) that need to be accomplished by stroke survivors and HCPs in order to achieve the behavioural outcomes and performance objectives were developed related to the TDF domains and sub-themes identi ed in Step 1. These are described in Tables 2 and 3.  To be able to access and effectively apply tools to support engagement in physical activity and reduction in sedentary behaviour e.g. self-monitoring tools To be able to access training on how to support physical activity and reduce sedentary behaviour post-stroke within restrictions of current job role Step 3: Selection of theory-based intervention content Selection of the theories/models to underpin the behaviour change intervention were informed by the ndings of steps 1 and 2.
Theoretical underpinning of stroke survivor component of intervention Two theories were selected to underpin the stroke survivor component of the multifaceted intervention, they were the Health Belief Model (24) and Self-Regulation Theory. (25) The Health Belief Model assumes an individual's belief in the personal threat of an illness together with their belief that the effectiveness of a health behaviour or action will determine whether they change their behaviour (or not).
Step 1 informed the selection of this model to target individual perceptions of stroke and stroke recurrence including the use and perceived bene ts and disadvantages of physical activity and inactivity. It was felt that this model would be appropriate particularly around challenging beliefs about the consequences of physical activity/inactivity and as such formulate reasons/intentions for engaging in physical activity.
Self-Regulation Theory assumes that behaviour is goal-directed or purposive. Findings from our systematic review and focus group discussions supported the need for speci c strategies to target volition as well as motivation in recognition that maintenance of physical activity for stroke survivors can be particularly challenging given the level of cognitive and physical effort required. Furthermore, inclusion of several speci c BCTs that target self-regulation e.g. goal setting-behaviour; problem-solving; action planning; feedback on behaviour were identi ed from the systematic review as promising.
Behaviour change techniques (BCTs) are the irreducible active ingredients of interventions targeting behaviour change, and are useful to inform, describe, deliver and evaluate behaviour change interventions. (26) TDF domains were identi ed from the data generated from stroke survivor focus group discussions and BCTs were selected with reference to those domains, supported by evidence from the systematic review (i.e. BCTs identi ed by the review as promising). When discrepancies occurred between the ndings of the qualitative study and the systematic review, members of the research team discussed these ndings and reached a consensus in terms of inclusion/exclusion of speci c BCTs. The outcome of the decision-making process is summarised in Table 4 which also describes how BCTs were operationalised to target increases in levels of physical activity and a reduction in sedentary behaviour.  The selection of BCTs incorporated in to the HCP component of the intervention was also informed by ndings from the qualitative focus group discussions conducted as part of step 1. The decision making process is summarised in Table 5 which also describes how BCTs were operationalised to support stroke survivors to engage in the PARAS intervention to engage in activities, improve their levels of physical activity and reduce sedentary behaviour. Repository of information providing details of local physical activity groups, support and resources Step 4: Development of the PARAS intervention Following the intervention mapping exercise in Step 3, a prototype intervention was developed and presented to stroke survivors and HCPs to further engage them in an iterative co-design process.

Stroke survivor co-design workshops
We conducted three co-design workshops with stroke survivors (n=21). The aim of these workshops was to elicit views on intervention content, form and mode of delivery of the intervention. Prototypes of the intervention tools (workbook, physical activity diary, information on apps accessible on mobile phone) and a range of pedometers were circulated during workshops to facilitate discussion and generate feedback. During the rst two workshops (n=13 stroke survivors) a feedback form was used to collate opinions/information (Appendix B). To support the involvement of stroke survivors with aphasia (impairment of language), the third workshop was delivered with the North East Aphasia Research User Group (https://www.neta.org.uk/) (n=8 stroke survivors). This workshop was delivered in an aphasia friendly format (i.e. using strategies to enable understanding of language) and verbal rather than written feedback was collated. To enable stroke survivors with aphasia and cognitive di culties to take part in future testing of the intervention, aphasia friendly consent forms and information sheets (Appendix C and D) were developed speci cally for the study and taken to the aphasia research group for feedback prior to use. All three workshops were audio-recorded and transcribed verbatim to capture feedback generated and facilitate the intervention development processes.

Key ndings of stroke survivor co-design workshops
A detailed overview of workshop ndings with stroke survivors is provided in Appendix E. In summary, participants reported a preference for the intervention to be supported by HCPs and delivered either at home or in a community outpatient setting. A preference was reported for at least two sessions, with the rst session delivered face-to-face and subsequent sessions delivered either face-to-face or by telephone.
The majority (>75%) of stroke survivors either strongly agreed or agreed that the content of the intervention workbook was well organised, easy to follow and appropriate. The physical activity diaries were reported to be well designed and considered easy to use by most participants. However, mixed views were received on whether other stroke survivors would use the diaries and the commercially available apps discussed, although it was agreed that this required further testing in a feasibility study. Eight commercially available pedometers that have been used successfully in other physical activity studies (29)(30)(31) were presented to stroke survivors during the workshop. The CSX 301S 3D simple pedometer was considered the most appropriate from those shown and was the only pedometer to be voted by all participants as easy to use and something they would be likely to use.

Healthcare professional feedback
An online questionnaire was completed by four North East community stroke teams (n=11 HCPs) to elicit feedback on the prototype intervention. These teams had previously expressed an interest in reviewing the intervention and taking part in a future feasibility study. The Template for Intervention Description and Replication (TIDieR) checklist (10) was used to present the components of the prototype intervention to the stroke teams.
Findings: Feedback in relation to the intervention design and content was largely positive (Appendix F). Team 2, 3 and 4 strongly agreed or agreed with the suitability of the intervention, tools and mode of delivery. Team 2 were uncertain about whether they could deliver the intervention within their team because they reported discharging patients to other rehabilitation services (i.e. follow-up reviews might not be possible). Team 4 raised concerns about their con dence in being audio-recorded delivering the intervention for delity assessment and feedback in the context of a feasibility study.
Team 1 were less certain about the content of the intervention in terms of delivery expressing a need for training. There were also some issues highlighted by Team 1 in terms of whether their patients would be suitable for the intervention because it was felt that patients may have been discharged from their service if they were su ciently mobile to take part in the intervention. A further meeting was held with Team 1 to discuss concerns and provide more detail that could enable a more informed decision regarding potential participation in a feasibility study of the intervention (e.g. to further emphasise that the intervention was not aimed at 'high functioning' patients). Following this meeting, Team 1 agreed they could potentially deliver the intervention.
Step 5: Formulation of an implementation plan An important consideration during the development of the PARAS intervention was implementation of the intervention and that it targeted all three pillars of high quality care: patient experience; safety and effectiveness. (32) To increase the likelihood of implementation, the APEASE criterion: affordability; practicability; effectiveness and cost-effectiveness; acceptability; side effects/safety and equity (33) were considered and applied in the nal intervention design. The nal intervention components and APEASE criteria are provided in Table 6.
Step 6: Development of an evaluation plan In order to further develop and optimise the PARAS intervention within community stroke settings, the most appropriate next step was to undertake a feasibility study to inform the iterative development of the intervention (in accordance with the MRC Framework). A protocol was developed for a feasibility study  Intervention toolkit including: stroke survivor workbook; repository of local/national information on PA choices; selfmonitoring tools (activity diary, pedometer (3DFitBud-Counter-Walking-Pedometer, 3D active, UK) and instructions, app advice); laminated goal summary sheet and fridge magnet pen; laminated bene ts, outcomes and activities cards to aid discussion between stroke survivor and HCP and support people with speech and language problems Components provided to/used by healthcare professionals · Consent form and participant information sheet · HCP training brochure · Dictaphone Affordability: Portable document format (PDF) les of all the intervention tools were created, printed out and stored in a workbook le meaning extra patient speci c sheets could be added to individual's les (e.g. physical activity diary). This process allowed iterative changes to be made without large costs of reprinting manuals. Rather than creating a website with large costs linked to maintenance and development, it was decided to trial a paper-based version of the intervention initially which could be developed online at a later date.
The pedometer selected had a relatively low price point (£16.99) to enable increased sued with NHS settings.
Practicability: HCPs were provided with a PARAS kitbag holding all the intervention tools so they could deliver the intervention then and there rather than having to for example nd out information about available resources and get back to participants at a later date.
Acceptability: all components tested at co-design workshops and developed iteratively in response to feedback Equity: The stroke survivor intervention tools were designed to be inclusive so individuals with speech and language or cognitive di culties would not be excluded as is the case in the majority of stroke research studies. Acceptability, affordability and practicability: Supported self-management was identi ed as the most appropriate mode of delivery for the stroke survivor component following our needs assessment and co-design workshops. This type of intervention appears more sustainable than for example face-to-face structured group exercise which presents with a number of environmental and resource related barriers.
Acceptability and effectiveness: Although HCPs working in community stroke care will have some experience of goal setting etc. qualitative workshops identi ed there were training needs in this area and it was acceptable to target these needs.
Who: For each category of intervention provider (e.g. psychologist, nursing assistant), describe their expertise, background and any speci c training given.

Provider of stroke survivor component
A healthcare professional (HCP) who is a credible source (e.g. well informed on stroke rehabilitation) and plays a key role in the stroke survivors community rehabilitation e.g. physiotherapist, occupational therapist, nurse.

Provider of HCP component
Health psychologist with experience in delivering behaviour change interventions in long-term conditions, research physiotherapist with 20 years clinical experience and 10 years research experience in developing and delivery physical activity and rehabilitation stroke interventions Affordability, practicability and acceptability: PARAS focus groups identi ed delivery of the stroke survivor component should be by a healthcare professional with experience working in stroke. Using healthcare professionals embedded within community stroke teams meant these individuals already had specialist core stroke skills meaning training was not required in this area alongside training in PARAS delivery. As the intervention was designed to be delivered within usual care this meant there were not additional salary costs. Initially consideration was made to include technical instructors and rehab assistants however on discussion with these individuals it was felt they would prefer to support the delivery rather than lead on the delivery and that they were not happy to be audio-recorded as part of the delity assessment.
Acceptability and effectiveness: As the providers of the HCP training had developed the intervention and were experienced in this eld from both a therapy and a psychology perspective they were thought to be the most credible source to deliver the training.
Practicability: At this feasibility stage it was decided that two members of the research team would deliver the HCP training face-to-face. This allowed the research team to highlight any iterative changes required to the training programme before scaling.
How: Describe the modes of delivery (e.g. face-to-face or by some other mechanism, such as internet or telephone) of the intervention and whether it was provided individually or in a group

Stroke survivor component
First session face-to-face, follow-up sessions either face-to-face or remotely by phone dependent on patient choice.

HCP training component
Face-to-face for initial training, then email and phone contact to provide feedback Acceptability: The modes of delivery were assessed as acceptable from our needs assessment, codesign workshops and questionnaires.
Affordability and practicality: Our qualitative work indicted that this mode of delivery of the stroke survivor intervention was practical. As the community HCP involved in delivering the stroke survivor component were already working with the stroke survivors participating in the study it was practical for them to initially see the participants face-to-face. To lower travel costs the option of providing the review sessions by phone was provided Side effects/safety: Our needs assessment and codesign workshops provided evidence that the intervention training and delivery methods would be safe with minimal side effects. As the stroke survivors were already being seen by a community stroke team with specialist skills it was felt that this team would be able to effectively identify any risks associated with taking part in the intervention and potential changes in physical/sedentary behaviour.
Effectiveness: It was hypothesised that all the components of the stroke survivor intervention could be delivered effectively within the two or more sessions.
It was also hypothesised that all elements of the stroke survivor intervention delivery could be taught effectively within three hours with email and phone contact for support during delivery Affordability: The supported self-management approach for the stroke survivor component provided a more affordable but at the same time potentially effective method of delivery than for example a face-to-face exercise intervention. Affordability: To enable effectiveness our needs assessment identi ed that a person centred individual tailored approach was required for the stroke survivors. This approach is potentially more expensive than a group based approach, however The HCP training was tailored according to personal needs during the face-to-face training, email and telephone support. All participating HCPs received feedback on delivery that was tailored to their individual learning needs.
the increased potential for effectiveness should outweigh these costs.
Equity: the stroke survivor component was designed to allow a person-centred tailored approach that would not exclude any stroke survivor who has the potential to move more or sit less.  Acceptability: The main issue brought up by the HCPs during the devlopment phase was the need to audio-record the intervention delivery. When it was discussed that this approach was to detrmine whether our training programme was appropaite the HCPs stated they thought this was acceptable. Whether this was actually the case will be further tested in a feasbility study.
training on delivery of programme to HCPs Treatment delity strategies for monitoring and improving receipt of programme Assess participants understanding of programme, use of cognitive skills and ability to perform behavioural skills through completion of workbook and analysis of audio-recorded sessions Treatment delity strategies for monitoring and improving enactment of programme skills Review workbook completion and achievement of goals.

Discussion
Low levels of physical activity and high levels of sedentary behaviour are common following stroke (35) and are associated with cardiovascular health, mental health and quality of life. (36) An intervention development process, informed by the MRC guidelines for the development and evaluation of complex interventions, (19) using Intervention Mapping as a framework (20) was undertaken to address this problem as part of the routine care pathway (37). An initial needs assessment identi ed a lack of effective theory-and-evidence informed interventions targeting long-term free-living physical and sedentary behaviour in stroke survivors. (9) Interventions that showed promise were limited by inadequate study design and a lack of comprehensive description to facilitate replicability. Furthermore, a lack of delity assessment limits understanding of the components of interventions delivered that are associated with positive outcomes.
We conducted a series of qualitative focus groups to identify determinants of behavioural change, and to explore the barriers and enablers to engagement in long-term physical activity and reduction in sedentary behaviour post-stroke. This enabled identi cation of behavioural domains and associated sub-themes within those domains that could be targets for a new intervention. The needs assessment highlighted the need for a timely, sustainable, person-centred intervention to support physical activity and sedentary behaviour after stroke. This led to the decision to use supported self-management and the subsequent mapping of the components of the intervention during co-design workshops to allow iterative development of the nal intervention. Consideration of the APEASE criteria was undertaken to facilitate the development of an implementation plan. A feasibility study protocol was developed to evaluate the intervention, and this study is currently underway.
Historically, structured exercise has been the most common mode of targeting low levels of physical activity after stroke. (7) Although structured exercise can lead to short-term changes in function, how this mode of delivery impacts on long-term health and well-being has not been established. Perhaps more importantly, our qualitative research mirrored previous ndings indicating that many barriers exist to this approach in terms of implementation e.g. resources, training, access, costs making it unsustainable for many stroke survivors. (13) Furthermore, structured exercise does not account for individual physical activity needs and preferences. Our qualitative work indicated that stroke survivors wish to partake in activities that provide meaning to their lives and allow them to recapture activities they engaged in prior to experiencing a stroke. This may be through structured exercise, but more commonly reported was engagement in day-to-day activities such as washing the car, shopping or playing with grandchildren. It was therefore important that the intervention was person centred rather than 'one size ts all' as with previous structured exercise research trials.
Developing an intervention that targets and addresses the needs of the end user is paramount and was the reason for early engagement with stroke survivors, informal carers and HCPs. Co-design and coproduction is vital to ensure the voices of patients and healthcare professionals are heard and valued.
There is now an expectation within self-management in stroke and at a governmental level that personheld experience is incorporated into healthcare intervention design. (38) The early use of co-design can potentially enable future implementation, with those taking part in the process becoming champions for the intervention. (39) Engaging with patients with aphasia during co-design is complex and as a result is often not undertaken. Aphasia is a common communication problem affecting approximately one third of stroke survivors. (40) In previous self-management interventions in stroke, up to 46% of studies have excluded individuals with aphasia limiting extrapolation of ndings to large numbers of stroke survivors. (41) One of the strengths of our intervention development process was early engagement with a group of stroke survivors with aphasia, and their views were incorporated in to the intervention content and design. This ensured that the intervention developed was suitable for the large proportion of stroke survivors with speech and language problems.
Continuous engagement with stroke survivors, carers and healthcare professionals during the developmental process was undertaken to increase the likelihood of developing an intervention that could be successfully implemented into practice. Alongside user views, we also applied the APEASE criteria to consider the social context of intervention delivery to further facilitate implementation. (33) Our systematic review highlighted that the majority of RCTs and pilot studies in this eld have been led by research teams, not clinicians, and attempts have not been made to embed testing within existing clinical pathways and settings. (9,37,(42)(43)(44)(45)(46)(47) The PARAS intervention was developed with implementation into the clinical care pathway in mind, with minimal adaptation as implementation of research ndings into rehabilitation settings has been previously shown to be slow, with evidence often not in uencing practice. (48) Our needs assessment highlighted that the intervention should be multi-faceted, targeting both the behaviour of the stroke survivors and the behaviour of the healthcare professionals promoting physical activity and providing support. This is a novel approach in this eld, where most of the research has focused exclusively on stroke survivors. Our qualitative work indicated that the stroke survivors have a preference to be supported by a healthcare professional, therefore it was important to consider behavioural change counselling strategies for use by healthcare professionals to enable this support. It could be argued that healthcare professionals already have skills to support these long-term behaviour changes given the importance of lifestyle in the context of stroke, however observational data on habitual physical activity and sedentary behaviour post-stroke (1) indicates the contrary and the ndings of our qualitative study highlighted the need for healthcare professional training in this area. Previous research further suggests that perceptions on physical activity post-stroke vary between stroke survivors, informal carers and healthcare professionals, (8) therefore training on how to deliver person-centred support to enable meaningful engagement in physical activity, which is more likely to result in long-term change, may be required.
The TDF was selected as a framework for analysis of our qualitative data to identify behavioural determinant of behavioural change, and selection of behaviour change techniques in order to facilitate intervention mapping. The main advantages of the TDF are that it provides a robust theoretical basis for implementation studies and can be used to identify barriers and enablers of behaviour and aid behaviour change intervention development. (49) The TDF has been used extensively to understand behaviour in clinical populations including stroke. (50) Occasionally issues are identi ed at the boundaries between the domains of the TDF but having three independent researchers reviewing the data facilitated consensus on occasions where data fell into more than one domains.
It is anticipated that the application of complex intervention development processes will increase the likelihood of future effectiveness and implementation of the intervention in healthcare settings. However, several limitations associated with our developmental process should be acknowledged. Stroke survivors that took part in the initial qualitative focus group discussions were required to travel, meaning only those who had access to transport or were mobile could attend. In addition, invitations to participants in these groups were advertised mainly at local stroke groups or patient carer panels which may have limited representation of a general stroke population. Although we advertised for informal carers to attend the focus groups, only three took part and information contributed was minimal and did not enable formal analyses. Therefore, this limited our understanding from a carer perspective. These limitations are being addressed, and as such the feasibility study will explore intervention views from a broader representation of stroke survivors. The participants will be recruited from community stroke services and will not need to travel to take part in the research (it will be delivered in the home).
The healthcare professionals recruited may not have been representative, which limits generalisability.
Purposive sampling was not undertaken and community stroke teams self-selected to attend focus groups. The majority were physiotherapists and assistants, rather than from the broad range of disciplines who may also have been suitable to deliver the intervention e.g. nurses, occupational therapists, speech and language therapists, exercise of referral/ tness instructors. Whether other healthcare professionals of different disciplines would be willing to deliver the intervention will be explored during the feasibility study.

Conclusions
Effectively targeting complex behaviours such as physical activity and sedentary behaviour post-stroke requires systematic and iterative development of evidence and theory informed interventions. Alongside effectiveness, the likelihood of adoption, implementation and sustainability of an intervention should also be considered during intervention development. Here we have presented the development of an intervention targeting long-term habitual physical activity and sedentary behaviour post-stroke. Throughout the developmental process there was active engagement of stroke survivors, their carers and healthcare professionals to increase likelihood of the acceptability and effectiveness of the intervention and long-term implementation. The PARAS intervention is currently being testing in three North East community stroke services and the results of this feasibility study will further inform the development of the mode, form and content. Following the MRC guidelines for the development and evaluation of complex interventions, the most promising intervention will then be further evaluated assessing e cacy and cost-effectiveness and process.