Synopsis

Research summary

Sedentary behaviour (SB) is the focus of considerable, clinical, policy and research interest, as the evidence supporting its detrimental effects on health and well-being increases.1–3 It is reported that these effects are independent of the level of moderate to vigorous physical activity (MVPA) someone might undertake.4–6 SB is defined as any waking behaviour characterised by low energy expenditure ≤ 1.5 metabolic equivalent of tasks while in a sitting or reclining posture.7

There are consistent reports that stroke survivors are more sedentary than other population groups.8–10 Longitudinal studies;11 systematic reviews (26 studies, 983 participants,12 12 studies, 367 participants13); and observational studies9 undertaken by the applicants report that stroke survivors are considerably more sedentary (in hours per day) and also have longer unbroken bouts of sedentary time than age-matched controls. Stroke survivors may be sedentary for up to 80% of the waking day;14 which does not improve over the first year after stroke; 19.1 hours per 24-hour day at 6 months; 19.3 hours/day at 12 months.11

There are important guidelines around levels of MVPA for people after stroke, but the majority of post-stroke survivors find these levels difficult to achieve.15 In line with recommendations,10,16,17 a complementary (but different) approach is to focus on reducing levels of SB, in terms of total amounts of time and the number of extended periods of time spent sitting/lying. This may be more achievable than a structured exercise programme,18 particularly in stroke survivors afraid of another stroke or falling. There were no evidence-based interventions available and a Cochrane Review19 provided confirmation of this during the programme.

Through five interlinked workstreams (WSs) we sought to develop and evaluate strategies to reduce SB in people after stroke (see Figure 1).

FIGURE 1 Inter-relationship between the different workstreams of the RECREATE programme. CN, cardiovascular; RCT, randomised controlled trial. (Figure not reproduced here; see the original article.)

The first three WSs progressed as planned; however, due to coronavirus disease discovered in 2019 (COVID-19) there was some disruption of activity of WS4 (relating to patient follow-up). However, we were in a position to begin WS5 as planned on 1 April 2020. Unfortunately, this could not take place; strenuous efforts were made to restart recruiting our n = 34 sites to undertake our definitive trial, but unfortunately due to the prioritisation of COVID-19 studies and disruption to service provision, this proved too challenging. We were supported throughout by our Programme Steering Committee (PSC) who recommended a review by National Institute for Health and Care Research (NIHR). At this review in May 2022, following presentation of alternative scenarios, it was agreed that the work would progress as an external pilot trial. Our revised aim was to recruit 300–350 participants across 12–15 sites and to continue with detailed process and economic evaluations, retaining the original timelines for the programme. As a consequence of the latter, a reduced number of participants were followed up for the previously planned 24 months. Despite careful planning, there was considerable delay in receiving data requested from NHS England (approximately 1 year for partial data release). This led to us requesting a 6-month unfunded extension to the work and some revision to the work we could undertake on this data set.

Workstream 1: update of the quantitative and qualitative evidence relating to sedentary behaviour

Some text in this section has been reproduced with permission from Mackie et al,20 Rawlings et al.21 and Johansson et al.22 These are Open Access articles distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

The aim of this WS was to ensure that the intervention was informed by up-to-date review of the quantitative and qualitative evidence relating to SB.

Objectives included:

• Review 1: to systematically review randomised controlled trials (RCTs) evaluation of interventions aimed at reducing SB in adults (all conditions including stroke), including:
  • assessing the effectiveness of interventions to inform development of the intervention(s) using meta-analysis
  • informing our knowledge of outcome measures used in evaluating SB interventions
  • utilising Michie’s Behaviour Change Taxonomy23 to identify specific behaviour change techniques (BCTs) contributing to effectiveness.24
• Review 2: to update the review by Benatti et al.,25 evaluating the effects of breaking up prolonged sitting time.
• Review 3: to systematically review qualitative explorations of SB, using a thematic synthesis approach focusing on peoples’ perceptions and experiences of SB and of participation in non-workplace interventions designed to reduce SB (in adult stroke and non-stroke populations).26
• Review 4: on advice from our Programme Management Group (PMG), we undertook an additional review to explore experiences of participating in interventions and the barriers and facilitators to delivery and participation in RCTs addressing the reduction of SB after stroke. For the published protocol, see https://doi.org/10.1136/bmjopen-2019-031291.

Review 1: a systematic review and meta-analyses of interventions to reduce time spent sedentary or break up prolonged sedentary events in adults

Methods

Ten electronic databases, relevant websites and previous reviews which focus on SB in adults were systematically searched (October 2017). No language restrictions were applied.

Papers were included if they reported RCTs including cluster RCTs (cRCTs), and randomised crossover trials in the adult population (clinical and non-clinical). Any study including a measure of SB (time spent sedentary or number of breaks in SB) was included even if reducing SB was not the primary outcome. Papers were excluded if they reported: the effectiveness of interventions delivered in schools, colleges, or workplaces; or studies which aim to investigate the acute (immediate) effects of breaking up sitting time as part of a supervised (usually laboratory-based) intervention. Two review authors conducted independent title/abstract screening and assessment of retrieved full texts. Risk of bias and quality of evidence were assessed by independent reviewers using the Cochrane Collaboration’s tool and Grading of Recommendations Assessment, Development and Evaluation (GRADE) system, respectively. Data from included studies including intervention BCTs were then extracted by two independent reviewers.

The review was prospectively registered on the Prospero database (CRD42018083751).

Key findings

Following removal of duplicates, searches identified 39,223 records, of which 85 studies (83 papers) met the inclusion criteria and were included in the narrative synthesis; 55 studies were included in the meta-analyses.

The majority of the results in meta-analyses were not statistically significant in favour of the interventions and the GRADE quality assessment showed that most evidence was not of a high quality. Only one intervention category using objective outcome measures showed effectiveness for reducing SB. This category comprised the provision of information, education, or support [advice/recommendations in written or verbal form (in one-to-one or group setting)] and counselling [mean difference −52.24 minutes/day; 95% confidence interval (CI) −85.37 to −19.10]. Only one intervention category using self-report outcome measures showed effectiveness for reducing SB. This category comprised the provision of information, education or support in conjunction with a form of structured/prescribed physical activity (standardised mean difference 0.15; 95% CI −0.23 to −0.07). However, this positive effect was not maintained at follow-up. No effect was found for interventions to break up long periods of SB. The most frequently occurring BCTs within effective interventions included action planning, goal-setting, social support and problem solving.

The findings from the systematic review and meta-analysis provided guidance for our work. Relating to our target population (stroke survivors) only one small community-based feasibility study (n = 33) was included, undertaken by our co-applicant English, the findings of which informed grant development. From the wider literature, there were indications that interventions might be effective in reducing time spent sedentary immediately post intervention and that intervention strategies to encourage sustainability of effects were needed. Behaviour change approaches including action planning/goal setting and structured/prescribed physical activity were highlighted as potentially effective intervention components. There was a dearth of studies focused specifically on clinical populations and older adults.

This review is published online: www.medrxiv.org/content/10.1101/2021.05.27.21256673v1

Review 2: evaluating the effects of breaking up prolonged sitting time

Methods

Five databases [MEDical Literature Analysis and Retrieval System (MEDLINE), Excerpta Medica dataBASE, Allied and Complementary Medicine Database (AMED), Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Cochrane library] were comprehensively searched from their inception until February 2018. Experimental trials which interrupted sitting time with frequent bouts of physical activity or standing in adults (≥ 18 years) were included. Comparison to a bout of prolonged sitting and a measure of at least one first or recurrent stroke risk factor was required to be included.

Key findings

Overall, 30 trials (35 articles) meeting the inclusion criteria were identified. Fifteen trials were completed with participants at an increased risk of having a first stroke and one trial with participants at risk of a recurrent stroke. Outcomes of hypertension and dysglycaemia were found to be more favourable following predominately light- to moderate-intensity bouts of physical activity or standing compared to sitting in the majority of trials with participants at risk of having a first stroke. In the one trial of stroke survivors, only outcomes relating to hypertension were significantly improved.

These findings are of importance taking into consideration hypertension is the leading risk factor for first and recurrent stroke.

This review has been published: https://doi.org/10.1371/journal.pone.0217981

Review 3: a systematic review of qualitative explorations of sedentary behaviour, using a thematic synthesis approach focusing on peoples’ views of sedentariness

Methods

A systematic search of seven databases [MEDLINE, AMED, Cochrane, PsycInfo^® (American Psychological Association, Washington, DC, USA), SPORTDiscus, CINAHL and Web of Science] was conducted in September 2017. Studies were included if they described primary research, with a focus on SB, using qualitative data collection methods, for example, data generated through interviews, focus groups, qualitative observational studies. Studies were excluded if: the focus of the intervention was on physical activity but did not explore SB; participants were aged < 18 years; or if only quantitative methods were used. Study selection followed independent screening of titles and abstracts, then full texts.

Data extraction was undertaken by two independent reviewers; methodological quality was also assessed independently using the National Institute for Health and Care Excellence qualitative appraisal guidance.27 A thematic synthesis was conducted.26

This review was prospectively registered with PROSPERO (CRD42017083436).

Key findings

Thirty individual studies capturing the experiences of 918 individuals were included. Eleven studies examined experiences and/or perceptions of SB in older adults (typically ≥ 60 years); 10 studies focused on SB in people experiencing a clinical condition (including 2 in stroke survivors n = 13 each); 4 explored influences on SB in adults living in socioeconomically disadvantaged communities; 2 examined university students’ experiences of SB; 2 on those of working-age adults; and 1 focused on cultural influences on SB. Three analytical themes were identified: (1) the impact of different life stages on SB; (2) lifestyle factors influencing SB and (3) barriers and facilitators to changing SB.

Sedentary behaviour is multifaceted and influenced by a complex interaction between individual, environmental and sociocultural factors. Knowledge of SB and the associated health consequences appears limited in adult populations, including survivors of stroke. There is a need for provision of accessible information about ways in which SB reduction can be meaningfully integrated in people’s daily lives. It was therefore important to consider the above factors and the multiple influences on sedentariness when designing and implementing our intervention targeting a reduction in SB.

This review has been published: https://doi.org/10.1186/s12889-019-7365-1

Review 4: a systematic review of process evaluations of interventions designed to reduce sedentary behaviour in adults

Methods

Ten electronic databases and reference lists from previous similar reviews were searched in March 2019, updated in May 2020 and October 2021. Studies were eligible if they were process evaluations of RCTs that measured SB as a primary or secondary outcome in adults (> 16; clinical and non-clinical populations). Studies of interventions delivered in schools, colleges, universities or workplaces were excluded, as were laboratory studies. Two reviewers independently screened titles and abstracts, and assessed full texts, with disagreements resolved by a third reviewer. Data were extracted by two independent reviewers, who also independently performed quality assessment using the Mixed Methods Appraisal Tool. A narrative approach to synthesis of qualitative and quantitative data was employed.

This review was prospectively registered (PROSPERO ID: CRD42018087403).

Key findings

Seventeen process evaluation studies (50 reports) were included. Five interventions focused on reducing SB or sitting time, 12 aimed to increase physical activity or promote healthier lifestyles. Process evaluations indicated changes in SB outcomes were shaped by numerous factors including barriers (e.g. staffing difficulties and scheduling problems) and facilitators (e.g. allowing for flexibility) to intervention delivery; contextual factors (e.g. usual lifestyle and religious events) and individual factors (e.g. pain, tiredness, illness, age and individual preferences).

This review has been published: https://doi.org/10.1136/bmjopen-2021-053945

In addition, Nicola Cornwall completed her PhD thesis: exploring SB in stroke survivors with severe mobility disability living at home (available at: http://etheses.whiterose.ac.uk/id/eprint/25340).

This extensive review work confirmed that there was an evidence gap relating to reducing SB for people after stroke (content, dose and delivery mechanism). Some indications of useful components emerged from the wider literature: behaviour change approaches linked to structured physical activity. However, it was also highlighted that there seems to be limited understanding of the concept of SB and educational components will also be required. The process evaluation of evaluation studies indicated some challenges and the need for flexibility in intervention content and awareness of context factors. Infographics were created to help disseminate findings to a lay audience.

Workstream 2: qualitative exploration of sedentary behaviours in stroke survivors

Some text in this section has been reproduced with permission from Hall et al.28 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

The aim of this WS was to explore the real-world experiences and views of stroke survivors and their carers regarding SB; and the views of health professionals, using observations and qualitative investigation, to inform intervention development.

The behaviour change wheel (BCW); capability, opportunity, motivation model of behaviour; and use of BCTs underpinned the study and provided the theoretical basis for intervention development.29

Overview of methods

1. Non-participant observations in stroke units with a linked community stroke service in Yorkshire (Site A) and Scotland (Site B) over the equivalent of a 2-week period in each service.
2. Semistructured interviews with staff from each service.
3. Semistructured interviews with stroke survivors and their informal carers at either 6 or 9 months post stroke (community-based).

Analysis

Observations: Fieldnotes capturing treatment or interventions provided by staff and activities completed by or with stroke survivors were recorded contemporaneously. One researcher from each site completed thematic analysis of all observational data for that site. This was then reviewed by another senior researcher, and summary memos were developed.

Interviews: Data were analysed using Framework Analysis.30,31 Initially, six researchers independently open-coded six transcripts to generate a thematic framework; all transcripts were subsequently coded into this framework, developing the framework iteratively throughout this process. When coding into the thematic framework, six transcripts were double-coded and four of these were triple-coded. The final version of the framework included nine themes. Framework matrices were then generated, and the six researchers worked in pairs to produce summary memos for each theme.

The summary memos from the interview and observational data analysis were compared and reviewed by all researchers to agree the analytical themes.

Key findings

Non-participant observations

Observations were conducted in inpatient and community stroke service settings, totalling 69 hours (Site A) and 63.5 hours (Site B). Twenty-five stroke survivors, 5 carers and 78 staff members were recruited to this component of the study. Observation participants (stroke survivors) varied in mobility status with Barthel scores ranging from 7 to 20. Key findings included:

• In hospital, staff indirectly encourage stroke survivors to be less sedentary during their routine daily activities with patients, including therapy sessions, washing and dressing.
• The hospital environment generally does not encourage a reduction in SB, for example, ‘free’ space where standing and moving could be facilitated is often cluttered with equipment.
• Community care and therapy input (e.g. being seen by staff after discharge from hospital) varies significantly between different services and geographical locations – some stroke survivors would benefit from more staff time to support them to reduce SB once they have returned to the community.
• Community staff currently discuss lifestyle behaviours with stroke survivors – it may be possible to integrate discussions about SB into these conversations.

Interviews

Patient and carer interviews

Interviews focused on motivation, capability and opportunity to reduce and/or break up SB (stroke survivors), and motivation, capability and opportunity to encourage stroke survivors to reduce and/or break up SB (carers of stroke survivors). Sixteen stroke survivors (10 at 6 months post stroke and 6 at 9 months post stroke) and 9 carers of the stroke survivors from Site A, and 15 stroke survivors (10 at 6 months post stroke and 5 at 9 months post stroke) and 3 carers from Site B were sampled. All stroke survivors were ambulant and had the ability to provide informed consent. Six themes were constructed. Key findings are summarised below:

• The consequences of having a stroke, for example, weakness in the legs, fatigue and reduced confidence, can make it more difficult to stand and move after a stroke.
• Stroke survivors valued returning to activities that they enjoyed before their stroke and reducing SB was seen as a ‘steppingstone’ towards this.
• Existing health concerns, pre-stroke routines and awareness of the potential health consequences also influenced stroke survivors’ attitudes towards reducing SB.
• Carers recognised that stepping back from a ‘caring’ role (e.g. wanting to do things for the stroke survivor) over time may provide more opportunity for stroke survivors to increase their independence and reduce SB, for example, making their own meals and drinks. However, some felt reluctant to step back.
• Supporting the stroke survivor to reduce SB can be challenging for carers, depending on their circumstances including their own health and work status, and whether they live with the stroke survivor or have other caring responsibilities.

Staff interviews

Interviews focused on staff capability, opportunity and motivation to encourage patients to reduce and/or break up SB. The sample included 16 staff from Site A, and 14 staff from Site B, from both the inpatient and community stroke service setting, from therapy, nursing and support roles.

Four themes were identified; key findings are summarised below:

• All staff recognised the importance of standing and moving. However, their awareness of the definition of ‘sedentary behaviour’ and the associated health consequences varied.
• Encouraging reduced SB was viewed as positive, since staff felt that doing so could help increase independence.
• Reported challenges to promoting reductions in SB included limited time/high volume of workload and concerns regarding increased falls risk.
• Staff indicated that additional training about the consequences of SB and how to reduce it would be required.

This work has been published: https://doi.org/10.1186/s12889-020-09113-6 (patients and carers) and https://doi.org/10.1080/09638288.2021.1955307 (staff).

These findings indicate that the physical and social environment, perceptions of stroke survivors’ physical and psychological capability to move, and routinised practices enacted by staff facilitate an expectation to be sedentary in the inpatient setting. This is often carried over when stroke survivors leave hospital. However, staff, stroke survivors and carers recognised the value of reducing SB after stroke and wanted to learn about safe and appropriate and methods for doing this. Opportunities exist to incorporate a reduction in SB within existing routine activities of stroke survivors, carers and staff.

Stroke survivor and staff personas were developed to aid knowledge mobilisation; these and other findings informed the coproduction study (WS3) to develop an intervention intended to reduce and break up SBs in stroke survivors.

The outputs of the first two WSs were considered by the PMG and in keeping with our original rationale agreed that intervention implementation should begin in the inpatient setting with clarification and reinforcement post discharge.

Workstream 3: development of an intervention to reduce sedentary behaviour through use of coproduction methods

The aim of this WS was:

• to establish and facilitate a series of interactive, face-to-face, coproduction workshops comprising key stakeholders (stroke survivors, carers, inpatient and community healthcare professionals, and public health practitioners)
• to coproduce a tailored intervention (to commence in the inpatient setting and continue post discharge when stroke survivors return to the community) designed to reduce and/or break up SB (sitting, lying, reclining) in individuals following a stroke.

Objectives:

• to collate, synthesise and disseminate the multiple perspectives about SB from WS1 and WS2 to the stakeholders
• develop a logic model specifying key elements of the proposed intervention such as mechanisms of action (including BCTs, approach to implementation, mediators and expected outcomes)
• develop an intervention and supporting materials, incorporating behaviour change strategies to reduce SB in stroke inpatient and community dwelling survivors.

Coproduction group members

The groups included nursing, therapy, activity and volunteer staff based in either the inpatient or community settings, and stroke survivors and their carers, including WS2 participants who had expressed interest in continued involvement and others identified by clinical staff in the relevant services. For stroke survivors, time since stroke varied from 4 to 15 months (an average of 10 months) at the point of the first workshop. Overall, 43 people participated in workshops including 17 staff, 14 stroke survivors, 6 carers and 6 researchers. This totalled 262 participant hours across all participants excluding researchers. At least three researchers were present at each workshop to facilitate small-group discussions and activities.

Summary of coproduction workshops

The intervention development process took place concurrently in Yorkshire and Scotland from October 2018 to February 2019.

Following the first workshop where the concept of SB and the intent of the coproduction process was outlined, the workshops broadly followed the three stages of the BCW:29

• Stage 1: Understand the behaviour (workshop 2)
• Stage 2: Identify intervention options (workshop 3)
• Stage 3: Identify intervention content (BCTs) and implementation options (workshops 4/5).

The focus, content and outputs of each workshop are outlined in Table 1.

While the workshops in Yorkshire and Scotland followed the same format/content, the meetings varied slightly based on local discussions. Two researchers attended workshops 2–5 at both sites to ensure consistency.

Each coproduction workshop included three sections:

1. Inform the group about the project. Key findings from WS1 (using infographics; see Appendix 1) and WS2 (using stroke survivor, carers and staff personas) were fed into the workshops to allow the intervention development to be informed by up-to-date relevant evidence. Feedback from and updates on developments following previous workshops were also provided.
2. Utilise knowledge and experience of group members. This involved conducting a series of activities using coproduction materials specifically designed for use during the meeting including engagement with personas, opportunity cards and solutions in practice activities. Materials were refined by the research team on an iterative basis, ahead of each workshop.
3. Evaluate the process using verbal summaries, feedback and questions.

Between workshops, the Acceptability, Practicability, Effectiveness, Affordability, Spill-over effects, and Equity criteria29 were applied to coproduced intervention components to develop a prototype intervention.

Participant feedback

At the end of each workshop, participants completed an evaluation form about the session. This allowed us to gather information on what participants thought about engaging with the process, but also to improve subsequent workshops. Both positive and constructive feedback, received from all participants at both sites, indicated that the experience was: ‘entertaining’, ‘stimulating’, ‘realistic’, ‘thought-provoking’, ‘enjoyable’, ‘exhausting’, ‘challenging’, ‘interesting’, ‘valuable’, ‘exciting’, ‘inspiring’, ‘structured’ and ‘difficult’.

Deliverables completed: The output of the coproduction study was an agreed, tailored intervention with delivery strategies: Get Set Go – Standing and moving more after stroke [see Appendix 2 for description of the intervention, including a provisional logic model (see Appendix 2, Figure 7) and Template for Intervention Description and Replication checklist32 (see Appendix 2, Table 11)].

This work has been published: https://doi.org/10.1186/s40814-020-00667-1

Workstream 4: feasibility study

The aim of this WS was to determine the feasibility of conducting a large-scale cRCT: refining implementation strategies for the intervention developed in WS3 and exploring and clarifying trial procedures. Our mixed-methods study in three services addressed the following objectives:

1. Explore the feasibility of delivering each intervention component: clarifying methods of delivery; identifying contextual influences on implementation; examining the fidelity of intervention delivery and identifying local individual/organisational barriers and facilitators to intervention delivery.
2. Explore the core resource use associated with delivering the intervention.
3. Assess acceptability of the intervention to patients, their family/friends and/or carers and staff.
4. Clarify and refine the proposed logic model for the intervention.
5. Clarify and test trial procedures by assessing:
   1. patient eligibility criteria, recruitment processes and data collection
   2. measures of fidelity with the intervention and their acceptability
   3. the appropriateness of outcomes measures (ease of use, level of missing data, outcome returns)
   4. assess use of patient completed Client Service Receipt Inventory (CSRI) to capture data for health economic analysis in the definitive trial.
6. Determining whether intervention, implementation methods and outcome assessments need revision before the definitive trial.

Methods

Study design

A prospective, single-arm, multicentre feasibility trial, alongside a qualitative study, to test data collection methods and evaluate feasibility and refine the intervention.

Setting

We sought to recruit stroke services, inpatient (not hyperacute) with linked early supported discharge (ESD)/community-based services.

Three stroke services (two acute Trusts with their linked community services in Yorkshire, and one hospital Trust in Scotland) participated in this feasibility work. To replicate the proposals for a future cRCT, all stroke survivors in these services had potential to be exposed to the intervention as it was implemented across the stroke service [stroke unit and community service(s)].

Implementation groups were created by the clinical staff in inpatient and community settings to operationalise intervention implementation.

Intervention

Inpatient and community staff teams were trained in intervention implementation by researchers face to face. The intervention comprised four main components:

• Education aimed at staff, stroke survivors and carers.
• Staff guidance (assessment tools) and materials (posters, prompt magnets) to support stroke survivors integrating standing and moving more into daily routines.
• A guide (with monitoring forms) to support stroke survivors to stand and move more after stroke, including information for their family, friends and informal carers.
• Environmental adaptation.

A key component of the intervention was therapy assessment of ability to stand by way of a ‘traffic light’ system. Red: requiring physical assistance from two or more people to stand and move; amber: requiring physical assistance or supervision from one person when standing or moving; green: able to stand or move independently, may need assistance on uneven surfaces or stairs. A prescription for standing and moving was provided to stroke survivors assessed as amber or green. Stroke survivors would be re-assessed regularly and their progress and prescription amended as required with detailed information and monitoring forms in the accompanying guide.

Prototype forms for capturing the intervention data (delivery and adherence) were also developed and utilised.

Participant recruitment

Trial, observations and semistructured interviews with stroke survivors

For outcome measurement assessment, we proposed to recruit up to 30 stroke survivors and carers across 3 services. Purposive sampling of stroke survivor interviews considered stroke severity and living arrangements. Stroke survivors were excluded if they were in receipt of end-of-life care, or discharged to an address outside participating stroke service remit, and were eligible if they met all of the following:

1. aged 16+ at time of stroke
2. diagnosis of new stroke and receiving rehabilitation in a participating stroke service
3. able to stand independently or with the help of one person [functional ambulation category (FAC)33 3–6]
4. have been/are being discharged home, not a care home
5. able and willing to provide written informed consent or consultee declaration (England) or welfare attorney, welfare guardian or nearest relative consent (Scotland). (Informed consent required for interviews.)

Carers were eligible if they were:

1. aged 16+
2. a family member/close friend of a participating stroke survivor
3. able and willing to provide written informed consent.

Semistructured interviews with staff: inpatient and community clinical staff were purposively selected based on their level of involvement with the intervention, profession and grade.

Quantitative methods

Outcome measures were selected in discussion with our patient and public involvement (PPI) group to ensure selection of outcomes of relevance to those affected by stroke and of importance in the main trial (e.g. to inform health economics). Self-reported outcome measures were collected using questionnaires at baseline in the stroke service and by post at 4 and 6 months post registration to assess potential usefulness. These included the following questionnaires:

Stroke survivors:

• Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS)34
• Nottingham Extended Activities of Daily Living (NEADL)35
• Fatigue Assessment Scale (FAS)36
• World Health Organization Disability Assessment Schedule (WHODAS 2.0 36-item)37–41
• Measure of Older Adults’ Sedentary Time (MOST)42
• A purposely developed mechanisms of change (MOC) assessment

To inform health economics:

• EuroQol-5 Dimensions, five-level version (EQ-5D-5L) and visual analogue scale (VAS)43
• The CSRI44,45 to capture resource use

Carers:

• The Caregiver Burden Scale (CBS)46
• Carer MOC assessment

Staff:

• MOC assessment

To assess feasibility of use, participants were asked to wear an activity monitor (activPAL47), attached to the thigh, and complete a sleep diary every morning, for 8 days at baseline (from recruitment), and again at 4 and 6 months post registration.

Analysis

Analysis focused on descriptive statistics and CI estimation rather than formal hypothesis testing of outcomes. Analysis was conducted on the full sample of participants regardless of protocol compliance or study withdrawal. The mean and variability of self-reported outcome measures were used to inform sample size estimation for the main trial. Analysis also considered recruitment and retention, and feasibility and acceptability of follow-up methods.

Qualitative methods

Observations

Observations of intervention delivery were conducted at each site to assess fidelity.

Implementation groups

Meetings functioned to refine implementation strategies and fidelity assessment, including: intervention overview and components; training and implementation plans; local context; problem solving. Meetings were conducted every few weeks over 6 months. Each lasted up to 1 hour, was led by a researcher and audio-recorded.

Semistructured interviews

A sample of stroke survivors (and carers) were interviewed to obtain feedback on outcome assessment and intervention acceptability. Interviews took place in stroke survivors’ homes, 2–4 months after intervention commenced. A sample of staff who had delivered the intervention were interviewed to obtain feedback on its acceptability, including barriers and facilitators to intervention delivery. Interview topic guides were informed by similar studies in SB (e.g. Ezeugwu and Manns 201848), observations, previous WSs and (for staff) Normalisation Process Theory (NPT).49

Analysis

Staff interview data were subject to formal qualitative analysis, all transcripts were entered into NVivo (QSR International, Warrington, UK) and the five key steps of thematic analysis outlined by Richie et al. were followed.30 This involved six researchers in total, led by two, who created a thematic framework and summary memos, leading to final themes and findings.

Tables of Change (TOC)50 facilitated the intervention refinement process and documented changes (incorporating all collected data). For each site, tables for implementation and content were populated with interview, observation and implementation group data into prescribed headings (intervention component; comment source; negatives; positives; possible change; reason for change; decision). This was then combined into one table (intervention component; main issues; potential solutions) and discussed as a team to determine how the intervention should be refined.

Results

To note: The 4- and 6-month data collection period was affected by the emerging COVID-19 pandemic; this impacted quantitative data collection, particularly that related to the activPAL (concern over infection risk).

Participant recruitment

Thirty-nine stroke survivors were recruited across three sites between July 2019 and January 2020, 14 of whom had a participating carer (36%). Stroke survivor demographics were similar between those screened and registered (by age, gender, ethnicity) (see Appendix 3, Table 12). There were some differences in demographics between sites, for example, in cognitive impairment levels and comorbidities. There was one eligibility violation (1/39, 2.6%) due to a stroke survivor’s FAC score not being 3–6, discovered on the day of registration. Participant flow is summarised in a Consolidated Standards of Reporting Trials (CONSORT) diagram (see Appendix 3, Figure 10). Baseline data are summarised in Appendix 3, Tables 13–15.

Numbers analysed

Follow-up was completed at 4 and 6 months in English sites, and at 4 months in the Scottish site. Of the 39 stroke survivors, 22 (56%) returned their 4-month follow-up questionnaire. Of the 25 stroke survivors in English sites, 10 (40%) returned their 6-month follow-up questionnaire. Follow-up rates are summarised in Appendix 3, Tables 16 and 17. One stroke survivor died during the study (3%). Six stroke survivors withdrew (15%), from two sites, 65–183 days after registration. At 4 months, 7 of 14 carers returned the questionnaire (50%). At 6 months, 3 of 11 (27%) carers at English sites returned the questionnaire. No carers died; one withdrew (7%). The primary outcome (NEADL35) was analysed for all stroke survivors who returned their questionnaire.

Intervention data

The prototype forms for capturing the intervention delivery and adherence were poorly completed with data available from only 6 of 39 patients (see Appendix 3, Table 18).

Outcomes

Of all stroke survivors, help with questionnaires was needed for 82% (32/39) at baseline, 20% (8/39) at 4 months, 4% (1/39) at 6 months.

At baseline, 69% (27/39) of stroke survivors found it easy to complete questionnaires, although 49% (19/39) thought there were too many questions. Carers found it easy to complete questionnaires and thought the number of questions was acceptable.

For participant-reported outcome measures, the WEMWBS,34 NEADL35 and FAS36 were fully completed for almost all stroke survivors who returned their questionnaires (see Appendix 3, Table 19). The WHODAS 2.037–41 was partially completed for 28% (11/39) at baseline and 18% of stroke survivors at follow-up (7/39). The MOST42 was partially completed for 10% at baseline (4/39), 26% (10/39) at 4 months and 16% (4/39) of stroke survivors at 6 months. Where questionnaires were partially completed, pro-rating was rarely possible, except for the WHODAS 2.037–41 questionnaire.

The MOC questionnaire completion was poor. This was fully completed by 54% at baseline (21/39), 44% at 4 months (17/39) and 28% stroke survivors at 6 months (7/25).

The CBS46 was completed by 10 carers at baseline (71%, 10/14), 2 at 4 months (14%, 2/14) and 2 at 6 months in England (18%, 2/11). Carer MOC was fully completed by 64% carers at baseline (9/14), 43% at 4 months (6/14) and 9% at 6 months (1/11).

Staff MOC was fully completed by 87% staff at baseline (153/176), 75% immediately after training (132/176) and 13% at 4 months after training (23/176).

Due to the nature of data collection (by post) and the onset of the COVID-19 pandemic, we were unable to explore why questionnaire completion was limited. The MOC questionnaire was long and complex which may have contributed to the lower-than-anticipated completion rates.

ActivPAL data

Of those recruited (n = 39), one participant declined to wear an activPAL accelerometer. At baseline, an activPAL and sleep diary were administered to 97% of participants (n = 38). Of the administered activPALs, 90% (n = 34) were returned, and data from 71% (n = 27) passed the initial quality check. The majority of files which did not pass the initial quality check (five of seven) were from one site.

During the 4- and 6-month data collection period, all research activity was paused due to the COVID-19 pandemic and infection risk from wearing reused activPALs, thus impacting further meaningful data analysis.

Safety

One participant was admitted to hospital due to a fall (1/39, 3%), on two separate occasions (length of stay= 12 days for visit 1 and unknown for visit 2). There were no related unexpected serious adverse events (RUSAEs) reported during the study.

Health economic outcomes

At baseline, there were 38 and 37 out of 39 participants in total with complete EQ-5D-5L and VAS43 data, respectively. Overall, the response rates for each item of the EQ-5D-5L,43 in the cohort of participants that received and returned the questionnaire, were acceptable and exceeded 90%. There was an observed change in EQ-5D-5L and VAS43 values over the three time points, indicating the potential for the instrument to detect changes in health status over time in this participant population.

At baseline, all 39 participants returned the CSRI44,45 questionnaire. Twenty-two out of 39 and 10 out of 25 participants (English sites only at 6 months) returned the CSRI44,45 questionnaire at 4 and 6 months, respectively.

There was a high degree of missing data for some specific resource use questions. For example, for the ‘emergency department attendance’ question, the percentage of missing responses was 41% and, of those who responded, 5% reported use of the services but number of contacts were completely missing. The same pattern of missingness was observed for ‘outpatient hospital appointments’ and ‘hospital treatment as a day patient’ where participants reported ‘Yes’ for using a service but did not state the number of contacts, and also for level of unpaid help from family and friends.

Qualitative findings

Data from implementation groups (25 hours) highlighted that:

• inpatient and community staff had different training needs
• releasing nursing staff proved difficult
• timing of training required careful planning and could run sequentially.

The approach to intervention refinement (informed by qualitative data collection and the TOC approach) is summarised in Appendix 3, Figure 11. A table was populated across the sites outlining the changes that were required to both the content and the implementation of intervention.

Interviews and observations

Further findings from interviews and observations are summarised here and detailed in Appendix 3. Data from 28 staff interviews resulted in three overarching themes: (1) practising and sustaining knowledge; (2) staff roles and teamwork; and (3) fitting an intervention into routine practice. Patients (n = 12) were generally positive about the notion of standing and moving more after their stroke but described specific fears and barriers. Observations (99 hours) identified important barriers (e.g. limited staff availability; competing priorities; staff doing things for patients; lack of space) and facilitators for standing and moving (e.g. staff enthusiasm; daily routines providing opportunity for movement; good understanding of patient ability; good communication between inpatient and community). Observations also showed intervention components being used in practice including how sites approached this differently and varied in the time taken.

Limitations

Interpretation of the follow-up data was limited due to impact of the COVID-19 restrictions, with high dropouts at the 4- and 6-month time points.

Implications for the main trial (workstream 5)

Based on findings from the feasibility study, considerable changes were made prior to the main trial (WS5).

Intervention refinements

Although staff and patients were positive about the purpose of Get Set Go (GSG), the feasibility of intervention delivery across sites was variable. Refinements in both intervention content and implementation were made following review of the TOC. These included:

• Intervention components: scaled back.
• Training: increased focused on clarifying staff roles, and emphasis on importance of communication across staff teams.
• Patient guides: wording simplified, and accessibility increased for people with aphasia, informed by feedback from our stroke survivor programme co-applicant. The developed aphasia-friendly guide was preferred by all patients and that version was taken forward and used in the following WS.
• Intervention data forms: simplified and produced in booklets to avoid staff having to manage many paper materials.

A referral form was also added to enhance communication between inpatient and community teams. Refinements to the intervention were reflected in the logic model.

Trial processes

Although not the prime focus, recruitment to the feasibility study was slower and more challenging than anticipated. Amendments to the screening process were undertaken to simplify it and reduce completion time for sites. Consent forms and information sheets for stroke survivors with language/cognitive impairments were further refined to improve accessibility for this patient group.

To improve follow-up rates, we planned additional resources, including priming messages before questionnaires were sent and reminders after questionnaires were sent by Short Message Service or e-mail. The questionnaire pack was also shortened to improve completion and return rates. Following discussions with our patient group, the PSC, and PMG, we decided to move from the more general WHODAS 2.0 to an assessment of extended activities of daily living via the NEADL for the primary outcome, as the NEADL would more appropriately capture the effects of the intervention which is focused on participants sitting less and moving more. In moving more, participants are encouraged to move more within the context of their daily lives and therefore this is reflected in the NEADL such as making drinks and doing housework. MOC questionnaires were removed and the 36-item WHODAS 2.0 questionnaire was replaced with the brief WHODAS 2.037,38 version. To encourage the return of the activPAL device, the wording of the activPAL information sheets was amended.

Workstream 5: external pilot cluster randomised controlled trial evaluating Get Set Go with embedded process and economic evaluations

Previous WSs progressed according to planned timelines; unfortunately, the COVID-19 pandemic began just as we were ready to start WS5. Our work schedule was disrupted by the pandemic. Following discussions with our PSC and the NIHR, the trial team amended the design to a large external pilot trial that aimed to recruit 300–400 participants across 15 stroke services. In order to keep to original timelines, a reduced number of participants were followed up for the originally planned 24 months. This trial was therefore not powered to detect a difference on the primary outcome.

External pilot cluster randomised controlled trial

Primary objective

The primary objective was to undertake a preliminary exploration of whether the developed intervention is likely to improve the ability to complete extended activities of daily living in the first year after stroke.

Secondary objectives

A key secondary objective was to explore whether the intervention reduces SB. Other secondary objectives were:

1. to explore core resource use and costs associated with delivering the intervention, including impacts on wider care
2. to explore whether the intervention reduces cardiovascular risk factors
3. to explore whether the intervention reduces disability
4. to explore whether the intervention improves health and well-being outcomes, such as health-related quality of life and mental well-being
5. to explore whether the intervention reduces carer strain
6. to explore and understand how the intervention is experienced and understood by recipients and providers
7. to explore potential moderators and mediators of the intervention effect
8. to explore the intervention implementation (e.g. intervention adherence, compliance, the process, benefits and challenges).

Trial methods

The trial protocol is available at: http://doi.org/10.1136/bmjopen-2023-074607.

Summary of trial design

The study is a pilot cRCT in stroke survivors evaluating the GSG intervention plus usual care (UC) compared to UC across 15 sites, including embedded process and economic evaluations.

National Health Service stroke services in the UK (clusters) were randomised on a 1 : 1 basis to deliver the GSG intervention alongside UC or UC alone.

Stroke survivors completed outcome measures (questionnaires and/or assessments, including wearing an activPAL monitor47) at 6 and 12 months following registration (see Appendix 4 for activPAL protocol considerations). A subset of participants also completed outcome measures at 24 months post registration.

Randomisation and blinding

A computer-generated minimisation programme incorporating a random element was used to randomise clusters to either the GSG plus UC arm or the UC arm. Randomisation was completed on a 1 : 1 basis and stratified by:

• Sentinel Stroke National Audit Programme (SSNAP)51 grade: A or B/C/D/E, based on the latest available audit data at the time of randomisation
• stroke unit composition: single or combination of trusts, and
• stroke unit size (acute): ≤ 24 and > 24 beds, based on national acute audit data (type 2 and type 3 beds combined).

Recruitment, setting and participants

The trial was conducted in UK NHS stroke services. A stroke service was defined as an acute and/or rehabilitation stroke unit with a linked community service over a defined geographical area.

Stroke survivors were eligible for recruitment if they:

• were aged > 16 years at time of stroke
• had a clinical diagnosis of new or recurrent ischaemic or haemorrhagic (excluding subarachnoid haemorrhage) stroke
• required manual contact of no more than one person to stand to prevent falling (manual contact consists of continuous or intermittent light touch to assist balance or co-ordination, i.e. not to support body weight)
• planned to live in the community post discharge, and
• provided informed consent/consultee declaration.

Stroke survivors were excluded if they were:

• receiving palliative care, or
• due to be discharged outside the defined geographical area of the associated community service(s) participating in the trial.

We also aimed to recruit a carer (a relative/friend regularly engaging with a stroke survivor participant) of each recruited stroke survivor. To ensure study populations were representative of the clinical population, accessible versions of the information sheets and consent forms were developed during the feasibility work with guidance from stroke survivors.

Intervention

Get Set Go is a whole-service intervention, commencing in the inpatient stroke unit setting and continuing post discharge for at least 12 weeks, with components including education (on SB and environmental adaptation), guidance and materials for staff to support stroke survivors integrating standing and moving into daily routines, and a guide with monitoring forms for stroke survivors and their carers (see Appendix 2). Implementation was led by site champions.

Control services continued with UC as determined by local policy and practices.

Statistical analysis methods

No interim analyses were conducted, and all our exploratory analyses were planned prior to the final analysis. Analyses were conducted on the intention-to-treat (ITT) population, with participants included in the analysis according to allocation regardless of non-compliance with the intervention. All participants were included in the analysis, except three participants registered in error (see below). Participant-reported questionnaires were analysed in line with scoring manuals including guidance on handling missing items. Where no direction was given, scores were prorated if ≥ 50% of the items were completed. Where multiple responses to a question were recorded, the ‘worst’ response was used in the analysis. We used descriptive statistics and 95% CI estimations to evaluate the outcomes.

Given the small number of clusters, for the primary and secondary patient-reported questionnaire and activity monitor endpoints, we used a two-stage cluster-level analysis, adjusting for the stratification factors, following guidance of Hayes and Moulton.52 As a sensitivity analysis, for the primary end point, mixed-effects regression models with repeated measures at 6 and 12 months were completed including all participants with data available for at least one follow-up time point.

Analyses were adjusted for cluster-level randomisation stratification factors, baseline outcome score, type of stroke (cerebral infarction or primary intracerebral haemorrhage) and time since stroke. We undertook appropriate regression diagnostics to check the validity of the statistical modelling.

Complete case analysis was used. Multiple imputation was not deemed appropriate due to the high proportion of missing outcomes.

Other outcomes including carer outcomes, self-reported falls and blood pressure were summarised descriptively by arm using mean, standard deviation (SD) and 95% CI for continuous outcomes and number and proportion for categorical outcomes.

A small study within a trial (SWAT) was also undertaken.

Key results

Study summary

Figures 2 and 3 present the overall trial CONSORT flow diagram depicting the flow of clusters and participants through the study from screening to analysis.

FIGURE 2 Flow diagram of study participants and clusters from cluster randomisation to baseline. (Figure not reproduced here; see the original article.)

FIGURE 3 Flow diagram of study participants and clusters from baseline to follow-up. a, Three participants submitted their questionnaire booklets at the 6-month follow-up time point, despite withdrawing previously; b, two participants submitted their questionnaire booklets at the 6-month follow-up time point, despite withdrawing previously; c, one participant withdrew and died within the same period; they are counted as a withdrawal in this diagram; d, one participant submitted their questionnaire booklet at the 12-month follow-up time point, despite withdrawing previously; e, one participant submitted their questionnaire booklet at the 12-month follow-up time point, despite withdrawing previously. (Figure not reproduced here; see the original article.)

Intervention training

Intervention training took place over a 24-month period (February 2021–February 2023). Characteristics of staff trained are detailed in Appendix 5, Table 20. In total, 103 training sessions were delivered, primarily remotely (86.4%), with a median session length of 60 minutes. The median number of attendees was 4 (with a maximum of 23 attendees) per session (381 in total). It was provisionally planned that sites would have a supported 10-week ‘run-in’ period to adapt intervention delivery to their specific sites.

Screening and recruitment

Conversations with interested stroke services began in October 2019. Recruitment of stroke survivors and carers started in January 2021, after being postponed due to the start of the COVID-19 pandemic. A total of 15 clusters were randomised; 8 were randomised to deliver GSG plus UC and 7 were randomised to deliver UC. A total of 5984 patients were screened: 1029 (17.2% of screened) were approached, 969 (94.2% of approached, 16.2% of screened) were eligible and 334 (34.6% of those eligible; 5.6% of those screened) were registered into the trial (Figure 4); 181 participants were registered to the GSG arm and 153 to the UC arm.

FIGURE 4 Recruitment graph. (Figure not reproduced here; see the original article.)

A carer was recruited for 73 (21.9%) of the registered stroke survivors, 46 in the GSG arm and 27 in the UC arm.

Eligibility violators and registration errors

Three participants were determined to be registered in error; one participant was registered using a paper form but withdrew before the registration was logged using the online system and two participants were registered on the online system, but no corresponding paper form was collected, and no baseline assessments/further data recorded. These three participants registered in error were excluded from the analysis population.

There were six participant eligibility violations (1.8%): five in the GSG arm (2.8%) and one in the UC arm (0.7%). In the GSG arm, three participants were discharged outside of the trial area, one participant was discharged to a care home for respite care and became a resident (not discharged to the community), and one participant’s diagnosis was updated to transient ischaemic attack after recruitment. In the UC arm, one participant was diagnosed with a clinical stroke but later had an additional scan which showed no evidence of stroke.

There were no eligibility violations for registered carers.

Participant characteristics

Participants across the treatment arms were similar in terms of age, gender and living arrangements (Table 2). Stroke survivors had an average age of 69 years (range 23–98) and 60.4% were female. However, there were some differences between arms in ethnicity of participants; 94.4% were White British in the GSG arm compared to 85.4% in the UC arm. Ethnicity differences could have been due to regional variations between the recruiting sites.

There were differences in the type and presentation of stroke. The percentage of participants with primary intracerebral haemorrhage was 15.0% in the GSG arm and 11.3% in the UC arm. The percentage of participants with left hemiparesis was 51.7% in the GSG arm and 36.4% in the UC arm. However, participants in each arm had similar levels of Modified Rankin Scale,53 National Institutes of Health Stroke Scale54 scores, and cognitive impairment.

Withdrawals

Of the 331 participants, 96 (29.0%) participants withdrew from at least 1 element of the study; 56 (31.1%) in the GSG arm and 40 (26.5%) in the UC arm. Baseline characteristics were very similar between participants who withdrew and did not withdraw.

Withdrawal rates for individual elements were: 93 (28.1%) for the activity monitor, 82 (24.8%) for questionnaires, 74 (22.4%) for researcher-administered data and 46 (13.9%) for medical records.

Participant follow-up and questionnaire completion

Follow-up rates were similar across the two arms (Table 3). Baseline characteristics were broadly similar across those included in the primary analysis compared to those lost to follow-up; however, there were some small differences. Participants lost to follow-up were slightly more likely to live alone (33.1% vs. 27.0% of those followed up) or have speech and language impairments (13.1% vs. 7.9% of those followed up).

Researcher unblinding

There were five researchers unblinded during the study: four researchers at GSG sites, and one at a UC site. In this cluster trial, they were therefore unblinded for all subsequent recruitment.

Safety and deaths

Of the 331 stroke survivors, 18 died during the study (5.4%); 15 in the GSG arm (8.3%) and 3 in the UC arm (2.0%). The mean time from registration to death was 359.7 days, ranging from 30 to 709 days. No carers died during the study.

The PMG discussed the causes of deaths and the between-arm differences; the causes of death were not determined to be related to the intervention. None of the deaths in the GSG arm occurred in hospital.

No RUSAEs were reported during the study. Hospital re-admissions, non-fatal strokes and myocardial events are summarised in Appendix 6, Tables 21–23.

Intervention delivery

Of the 181 participants receiving the GSG intervention, 103 (57.8%) inpatient therapy records, 51 (28.3%) nursing records and 76 (42.8%) community records were returned by sites.

From the inpatient records, 74.8% (77/103) of participants were provided the GSG guide, and for 71.8% (74/103) of participants staff shared their prescription with the nursing team and used the checklist to discuss the guide with the participant.

Weekly nursing records were completed for a median of 1 week per participant, ranging up to 5 weeks. Daily tasks such as prompting standing and moving according to prescription and encouraging participant to use the guide to record standing and moving were completed on approximately 90% of days (ranging from 89.3% to 91.8% of days across tasks).

From the community records, 86.8% (66/76) of participants had a GSG guide, either from hospital or replaced by community staff; however, only 14.5% (11/76) participants had the optional prompt magnet provided. Weekly community records were completed for a median of 4 weeks per participant, ranging up to 11 weeks and including a total of 380 days across the 76 participants. Daily tasks of checking the traffic light status and prescription and prompting participant to stand and move and use the guide were completed for approximately 87% of days (ranging from 85.8% to 91.8% of days across tasks).

Primary outcome analyses (exploratory as this trial was not powered to detect a difference)

The mean NEADL35 score pre-stroke was 57.4, with slightly greater independence in the UC arm (58.9 compared to 56.1; Table 4). At 6 and 12 months after stroke, the mean NEADL35 score was approximately 45.3 with a median score of 50.0 and a wide interquartile range (IQR) from 31 to 61 points.

Independence increased in both arms for the subset of participants followed up at 24 months. At both 6 and 12 months, the UC arm had slightly greater independence.

The results of two-stage cluster-level analysis showed no statistically significant difference between the two arms. The mean difference in the cluster-level adjusted residuals for NEADL35 score between the two arms at 12 months was −1.76 (95% CI −7.63 to 4.25, p-value 0.530).

In the linear mixed-effects model (Table 5), the estimate of intervention effect was −1.5 (95% CI −9.79 to 6.71, p-value 0.656), and the intervention–time point interaction estimate was −0.4 (95% CI −3.45 to 2.65, p-value 0.796). Neither the intervention term nor the intervention–time point interaction term was statistically significant, although this trial was not powered to detect a difference.

Secondary outcome analyses

The mean difference in the cluster-level adjusted residuals for mean daily sedentary time in minutes between the two arms at 12 months was −29.6 (95% CI −33.71 to 92.81, p-value 0.322). The difference was not statistically significant for mean daily sedentary time (Table 6), but this trial had insufficient power to detect a difference.

See Appendix 7 (Tables 24−38) for results of other activity monitor outcomes, secondary questionnaire outcomes, carer questionnaire outcomes, moderator outcomes, SWAT analysis, and Appendix 6 for routine data analysis.

Process evaluation

The process evaluation aimed to explore and understand the implementation of GSG and how it was experienced and understood by providers and recipients.

Objectives

1. Describe and clarify causal assumptions about the intervention, and its mechanisms of impact.
2. Assess implementation fidelity.
3. Explore views, perceptions and acceptability of the intervention to staff, stroke survivors and their carers (family/friends).
4. Establish the contextual factors that may influence implementation, intervention mechanisms and outcomes.

Methods

The process evaluation protocol is available at: https://doi.org/10.1136/bmjopen-2023-075363.

In brief, an embedded mixed-methods process evaluation was conducted in six sites (four intervention, two control). Data collection methods included observations of intervention training, observations of stroke care (inpatient and community), interviews (with staff, patients and, where appropriate, their family, friends or carers) and documentary analysis. Due to the process evaluation being conducted during the pandemic, we secured an ethical amendment to collect data remotely where needed.

Methods were guided by the Medical Research Council (MRC) framework for designing and conducting process evaluations.55 The intervention was designed to target behaviours of staff, patients and carers and implemented within complex stroke services; therefore, the process evaluation focused on individual-level behaviour change and implementation. We used the Theoretical Domains Framework and BCTs in the logic model when outlining intended mechanisms of impact. An implementation plan was developed to expand the logic model and detail processes that staff would ideally engage in to implement the intervention. This implementation plan incorporated NPT,49 and learning from WS4. Additionally, we used the theoretical framework of acceptability to inform the data collection and interpretation and analysis of findings.

Key findings

Data collection started in August 2021 (delayed due to pandemic restrictions), staggered along trial service setup, and completed in July 2023. Stroke services were approached for process evaluation inclusion by introduction by the central research team, to local principal investigators (PIs) and ‘Champions’ in intervention sites. All approached agreed. Service selection considered various factors, and resulted in a spread of service configurations.

We conducted a total of 11 observations of training; 44 intervention observations; 17 control observations; 52 staff interviews (40 intervention, 12 control); and 27 stroke survivor interviews (17 intervention, 10 control) across 4 time points each separated by approximately 3 months (Table 7).

Observations in intervention sites: key learning

A large number of staff were trained across the sites, but it was more difficult to train nursing staff and it was evident that therapy staff were more aware than others of GSG. Parts of GSG were visible in the environment, for example, posters, traffic light symbols and guides. It was implemented in what staff described as peaks and troughs (used more so after baseline, then time 1 but dropped off by time 3). The champions were key drivers in the implementation, but engagement tended to reduce over time or needed reigniting at points by the champions. Throughout all the time points, they were positive about the core message around standing and moving. Use of the guide was not always evident in the observations, some patients found it useful, others disengaged. The observations gave insights into how different sites managed the storage of materials and the systems they used to facilitate the implementation process (more space materials accessible, less space stored off the ward). Systems such as noting down who had received the guide in an existing log facilitated more continued use of GSG over time. Teams across all sites faced system-level pressures; the day-to-day practice of care was unpredictable, and staff needed to respond to the demands of what was happening. Although movement was supported in therapy sessions, patients were commonly inactive aside from these times. Staff faced challenges with the amount of paperwork associated with the trial and there was some confusion about trial processes.

Staff interviews

Staff interviewees were from a mix of professions (Table 8).

Key findings from intervention sites are very briefly summarised here (expanded in Appendix 8):

• The concept of training – moving beyond formal training sessions

This theme focuses on the notion that ‘training’ staff to implement complex interventions such as GSG in complex settings requires learning from more than just the training sessions. We identified additional steps that helped staff make sense of the information provided in training, in an attempt to translate this learning into practice. Our findings also highlight the challenges staff faced following initial training and the subsequent impacts on implementation.

• The role of the champion – a double-edged sword

The champions brought value to pushing implementation forward but also there were potentially negative impacts if it led to less involvement from the wider team.

• Suitability of patients for the intervention and the role of staff in determining this

Staff had differing perceptions of whether they thought the intervention was suitable for different stroke survivors and the extent to which this had an impact on whether conversations were had to introduce the guide and what these conversations involved for different patients is explored.

• Core message was buried under paperwork

This theme addresses the notion that the core message around reducing SB throughout the day unintentionally was obscured amid the complexities of the accompanying intervention and trial paperwork.

• Conflicts between intervention suitability in inpatient and community settings

Based on interview feedback, there were mixed views as to whether the intervention was best suited to inpatient settings or community settings. Staff discussed this in relation to the practicalities of implementation in the two different contexts and the benefits of different timing of intervention delivery for patients.

Patient interviews

Of the 63 patient participants we attempted to approach for interview, we consented 43% (their characteristics are described in Table 9). For the remaining 36, we were unable to check alive status or find working contact details (n = 10); they were re-admitted or in a care home (n = 2); out of area or in temporary address (n = 2); did not answer a phone call (n = 14) or e-mail (n = 3); or declined or were too unwell (n = 5). Interviews took place in patients’ homes (n = 24) or by telephone. Of carers recruited, two were male, four were in the intervention arm.

Of the intervention arm interviewees (n = 17), the majority had been given the intervention (n = 12), two had not, and three were unsure. However, only a minority of these seemed familiar with GSG, as many were unable to comment on the intervention and some said their memory of it may be affected by stroke. As well as the intervention, interviewees discussed stroke recovery and daily activity.

Key themes raised by patients at intervention sites were:

• Acceptability of intervention message and content

The central concept of getting moving after stroke was valued by patients who received GSG. For some, this helped them to see their potential or came as a revelation:

What meant the most to me was don’t be [sedentary], be active, and that came as a revelation. I think there’s a temptation mentally that if you’ve been ill … whatever I do I mustn’t move, and this is contrary to that, it’s saying no, the thing you must do is move, exercise, use the muscles.

The level and tone of the GSG materials were suited to some and not others; individual requirements differed due to stroke severity and personal preferences.

• Wanting to move but feeling restricted in hospital

Many required assistance getting to the toilet and had difficulty getting help from staff, sometimes resorting to going independently. Patients commonly felt they were stopped from standing and moving freely in hospital, some likened this to being told off:

I thought, need another wee-wee, up I get … and the nurse is all like … what are you up for? I’m going to the toilet, you know, thanks for asking.

Some patients were keen to get moving and perceived GSG as giving them ‘permission’ to move appropriately. The intervention message was not consistently enacted across disciplines, as evidenced by behaviour around avoiding falls. This highlights complexities in promoting movement in ward culture which is also implementing safety regimes.

• The extent to which GSG is sustainable over time in the ward and at home

A minority of patients recognised that the intervention gave them motivation initially, for example, to get home. However, using GSG was difficult for acutely unwell patients and at admission, and some did not engage in recording their activity because they were already active. Some patients used the guide once they returned home, but few participants used GSG for the full 12 weeks, partly as priorities and routines changed once they were home.

It wasn’t so very relevant, because he was moving around a lot, whenever he could, and he didn’t have that problem where you’d have to say come on, get up and go, he wanted to get up and go.

The intervention message about standing and moving throughout the day was not at the forefront of their minds, and many were doing a lot once home and felt their abilities were beyond the intervention.

Comparing control and intervention services

Two control sites were included in the process evaluation, selected based on the same criteria as intervention sites. We selected two sites of differing service configuration. All six services were found to have similar barriers to patients standing and moving, including staffing issues; risks such as falls; and increased patient movement often not suitable to all patients or settings. Commonly across the six sites, staff valued the premise of movement after stroke and saw it as appropriate for some of their patients. During this process evaluation, we were aware of some sites having initiatives, alongside our research, which could have increased standing and moving. These included a ward-based training programme to support standing and moving (Site E), GAP research (Site D) and EndPJparalysis (Site A). While these make it difficult to understand whether changes over time can be attributed to GSG or not, they should also be considered in the context of initiatives which may oppose patient movement [e.g. fall prevention initiatives (Site E)]. In interviews, patients talked in very similar ways about movement, such as activities they are aiming to get back to and how they would go about this, suggesting their approach and behaviour may be the same regardless of GSG.

Implementation and study objectives

This section is guided by the MRC guidance for process evaluations55 and synthesises findings across all implementation sites.

Programme theory and logic model (objective 1)

Staff could appreciate the value of what GSG was trying to achieve in terms of reducing SB; however, they did not always feel confident in providing this to patients who they thought may be at risk of falls if they took ownership of GSG (e.g. patients with cognitive and communication needs). For patients where they felt more confident to provide GSG, that is those they did not feel were at risk of falls, there was often limited time for talking through the guide. This meant the quality of the conversation around the guide was lacking, yet this was one of the key intended mechanisms for patients taking ownership of GSG.

A minority of patients read the guide in its entirety and were fully on board with recording how much standing and moving they had done each day. However, many patients either used the guide for a short period of time and then stopped once they felt they were doing more or did not engage with the guide at all. This suggests the behavioural regulation element which was designed to form habits did not work for everyone. There were some positive unintended consequences of monitoring for some people, for example, being able to manage their fatigue and practising writing.

Summary of intervention implementation (objective 2)

Across the four intervention sites within the process evaluation, the intervention was partially delivered as intended. There were key parts of the intervention that tended to be more commonly in place/delivered and other parts less used. Competency of staff varied across the sites – physiotherapists tended to be the most competent staff group, partly related to the extent of involvement and subsequent opportunities to ‘practice’. Across all sites, the intervention did not filter through to the whole team. One of the key aspects that was not delivered with fidelity was communication of the message to reduce SB throughout the day. A second was communicating the reason for the tick boxes in the guide, which were designed to encourage the stroke survivors to form habits.

Responses to the intervention (from the perspective of staff, patients and carers) (objective 3)

Staff valued the overall principles of GSG; however, it was often perceived as an extra thing that the teams needed to think about within their already busy working days. At times they were less motivated to implement GSG because they perceived they already promoted standing and moving. Some of the staff thought it was not too different from what they did already; rather it was just a case of documenting what they do.

Get Set Go was viewed positively where patients felt motivated because they wanted to avoid another stroke as they could see the benefits, but not so in cases where movement was not connected to anything meaningful. Our findings highlight the importance of understanding what matters to patients so that movement can be encouraged in the context of their lives and interests.

Context (objective 4)

A number of contextual factors were influential in whether it was possible to deliver in the different services. Examples included staffing pressures; staff rotations across different parts of the stroke service; ward pressures due to COVID-19; the safety culture on the ward – staff being fearful of falls; and environmental factors such as storage space for materials and access to therapy space.

Summary of process evaluation

We identified that the intervention did not always have its intended effect because it did not become consistently habitual for staff or patients for reasons indicated. Revisions to the intervention are needed to enable these mechanisms to have an effect. Our findings provide important learning for implementing interventions in complex settings and are applicable to researchers beyond the field of SB.

Potential intervention refinements are detailed in Appendix 8 (Table 39), alongside an implementation plan (Table 40).

Additional planned research

We were successfully awarded further funding as part of the stream B post-award programme grant scheme to tailor the intervention using an ideal types approach: www.fundingawards.nihr.ac.uk/award/NIHR206938

Exploratory economic evaluation

Objectives

The revised economic evaluation objectives were to address the following questions:

1. How likely is the GSG intervention to be cost-effective?
2. What is the difference in estimates for the cost of hospital services utilisation using data from self-report versus routinely collected Hospital Episode Statistics (HES)?
3. What is the impact on cost-effectiveness estimates if the economic evaluation is based on self-report versus HES data?
4. What is the agreement level between estimates of hospital services utilisation based on HES versus self-reported data?
5. What additional information does a longer-term (24 months vs. 12 months vs. 6 months) follow-up provide for the evaluation in terms of cost difference and completion rates?
6. Can HES data be used as a replacement for self-report service use data for study participants for whom data are missing?

To address these objectives, it was still necessary to undertake the originally planned within-trial economic evaluation to serve as a base case against which to consider some of the revised objectives. Undertaking the economic evaluation thus remained a significant portion of the planned work. Methods for this are outlined below with further details available in a separate output (https://doi.org/10.1101/2025.10.18.25338286). In this context, it should be noted that subsequent cost-effectiveness/cost–utility findings should be viewed as exploratory only.

Methods

Overall approach for the within-trial exploratory economic evaluation

The exploratory within-trial economic analyses were carried out on participant-level clinical trial data and according to the ITT principle.56 Cost-effectiveness and cost–utility analyses were based on measures of resource use and outcomes over a 12- and 24-month follow-up period from the point of randomisation. Details about the trial population and overall follow-up rates are described earlier and highlight that a very significant loss to follow-up affected running the analyses over a 24-month period.

The primary evaluation perspective was NHS and Personal Social Services (NHS/PSS). This included costs associated with the intervention, secondary care, and primary/community-based health and social care. An additional societal perspective further added costs associated with non-NHS community-based services (where costs were borne by the patient) and unpaid help to the total NHS/PSS costs.

Service utilisation was measured using an adapted version of a CSRI developed and used in WS4 adapted from previous stroke rehabilitation trials44,57 and administered in self-report format as part of the case report form, at baseline (covering the use of resources over the previous 3 months), at 6-month (for the period since baseline), at 12- and 24-month follow-ups (covering the previous 6 months); respondents were asked to report the frequency of use for each of the afore-listed resources within these assessment periods. Relevant unit costs were applied to estimate individual-level costs for each time point.

Costs associated with the intervention (including training) were estimated using a mix of top-down and bottom-up approaches for different components with specifically designed pro-forma completed at each site (further details: https://doi.org/10.1101/2025.10.18.25338286). In essence, fixed or semi-fixed costs were assigned to staff training time and management activities, and variable costs for activities related to staff interactions with participants.

All costs are reported at 2022–3 levels using the PSS Pay and Price Index and the NHS Cost Inflation Index58–60 to adjust unit costs as necessary. Discounting to adjust for the differential timing of costs and quality-adjusted life-years (QALYs) estimated between 12- and 24-month period was implemented using 3.5% discount rate.61

Details of resource use identification, measurement and unit cost estimation, including for the GSG intervention, are provided elsewhere (https://doi.org/10.1101/2025.10.18.25338286).

Differences in mean total costs with associated bias-corrected 95% CIs between the two arms from NHS/PSS and from societal perspectives over the period from baseline to 12 months were analysed using bootstrap multilevel mixed-effect linear models adjusting for patients’ baseline costs with randomisation stratification variables as covariates.

The mean difference in QALYs over the period from baseline to 12-month period was estimated using bootstrap multilevel mixed-effect model with associated bias-corrected 95% CIs. Baseline utility score with randomisation stratification variables were used as covariates. The mean difference in NEADL was estimated using the model specification mentioned for estimating the difference in QALYs.

Probabilities that the true (unobserved) population difference in mean cost, QALYs and NEADL between the two treatment arms are positive (i.e. cost, QALYs, NEADL score are higher in the intervention arm when compared to control) based on trial sample information were estimated.

To make exploratory inferences about the cost-effectiveness of the intervention, a joint distribution of mean difference in cost and QALYs/NEADL was produced by fitting cost and outcome models jointly to each bootstrap replicate. Results are presented as scatterplots of these distributions.

Hospital Episode Statistics data

Use of inpatient, day patient and outpatient services were available for the whole follow-up period, and are split between 1–12 months and 12–24 months. We applied the same unit costs as for the CSRI44,45 data, and subsequent cost estimates are used to address the feasibility-related questions.

Results

Exploratory economic evaluation

Resource use patterns at baseline were similar between groups for formal services, but the GSG group used more unpaid care; this greater use of unpaid care was also evident at follow-ups, coupled with higher use of hospital services (data available: https://doi.org/10.1101/2025.10.18.25338286). Consequently, total costs were higher in the GSG group across baseline and all follow-ups (data available: https://doi.org/10.1101/2025.10.18.25338286). The GSG group showed higher variability which means there was wider cost dispersion and implying a highly skewed cost distribution with a SD twice as the mean at 12 months follow-up. Utility estimates also suggested lower QALYs for the GSG group over the follow-up period (data available: https://doi.org/10.1101/2025.10.18.25338286). In combination, from both NHS/PSS and societal perspectives, costs are higher in the GSG group, while outcomes measured in QALYs are worse, suggesting that the GSG intervention is unlikely to be cost-effective (Table 10). This is further evidenced by the bootstrapped replications in Figures 5 and 6 which show that almost all replications lie in the northwest quadrant (represents GSG group is more costly and less effective than the alternative, meaning they are dominated and not cost-effective) of the cost-effectiveness plane.

FIGURE 5 Cost-effectiveness plane for costs from NHS and PSS perspective and QALY differences. (Figure not reproduced here; see the original article.)

FIGURE 6 Cost-effectiveness plane for costs from societal perspective and QALY differences. (Figure not reproduced here; see the original article.)

Feasibility questions

Hospital Episode Statistics data suggested that hospital service use is consistently higher across both randomised groups compared to the self-reported CSRI44,45 data, resulting in higher estimates for hospital costs (data available: https://doi.org/10.1101/2025.10.18.25338286). Of note, estimates derived from HES did not provide evidence of higher costs in the GSG group compared to the UC group. Therefore, the direction of difference in service use and costs, which for self-report were consistently higher in the GSG group compared to the UC group, was not mirrored in the equivalent estimates derived from HES data.

Limitations are summarised below:

• COVID-19 caused issues in delivery, which led to slower than predicted recruitment.
• The trial was not powered to detect statistical significance.
• The population was less severe than expected, which may have been due to recruitment bias.
• There was an imbalance between arms in some participant characteristics, such as ethnicity differences; however, this was potentially due to case mix and regional population variation between sites rather than selection bias.
• Follow-up rates were lower than anticipated.
• There was a high level of missing data on intervention delivery.
• Sites likely represented stroke services with a higher-than-average interest and willingness to changing standing and moving practice. There was no indication that there was a difference in this between intervention and control sites. The research paperwork may have hampered intervention implementation; such paperwork would not be a factor in wider dissemination if a definitive trial demonstrated the intervention was effective.
• Purposive sampling in the process evaluation was challenging due to the limited availability of participant characteristics to process evaluation researchers.

Discussion

We have successfully delivered a complex programme on an important topic, with considerable learning generated. It is unfortunate that our work was derailed by the COVID-19 pandemic.

The importance of SB has been increasingly emphasised since initiation of this programme grant. It is an important issue both at the level of the public and for survivors of stroke. There is increasing research activity in this area (SB and stroke) which includes (as examples): observational work reporting the association between SB and stroke recurrence after minor stroke;62 exploration of a link with cognitive impairment;63 and using statistical modelling to identify factors associated with SB.64 However, the Cochrane Review (2021)19 concluded that current evidence is insufficiently strong to guide practice and recommended more high-quality randomised trials, demonstrating the relevance of our research question.

Patient and public involvement

Patient and public involvement engagement underpins all our work, to ensure that it addresses the priorities and needs of those affected by stroke and intervention development is relevant and appropriate. These aims were achieved through regular meetings and consultation. Our Consumer Research Advisory Group (CRAG) has been established for 20 years. Members helped develop the research question, the application, refinement and implementation of this programme of work. In addition to ongoing engagement with the local CRAG, we also visited local stroke groups on a regular basis (see Report Supplementary Material 1). Our co-applicant Gill Carter is a stroke survivor with speech impairment, so her input was particularly useful in reviewing the intervention materials to ensure they were acceptable to all. The intervention was coproduced (WS3) with stroke survivors and carers in two groups (one Scotland, one Yorkshire). Their input ensured that the intervention was understandable and reflected the realities of life after stroke. Gill was a member of our PMG and Amanda North (stroke survivor) was a member of our PSC. In our ongoing work relating to this programme, we are disseminating outputs through cultural activities (art exhibition). Reflections and recommendations for PPI work are provided in Report Supplementary Material 1.

Equality, diversity and inclusion

We were mindful throughout that survivors of stroke may suffer cognitive and communication impairments which may impede participation in research. Our observations in stroke units (WS2) facilitated inclusion of all survivors of stroke, and the information obtained informed intervention development. Purposive sampling was used to identify participants for the coproduction groups, with the aim of including stroke survivors who were diverse in age, gender, communication and mobility status. The intervention materials were aphasia-friendly and developed with a local artist.

We undertook considerable work to increase the accessibility of the study; this included creating aphasia-friendly consent forms and facilitating consultee declarations.

In WS5, our ambition was to recruit a wide range of sites from areas with differing demographic profiles in terms of ethnicity and deprivation. We in part succeeded, but due to the differential effects of the COVID-19 pandemic, sites which entered the study earlier and had higher recruitment tended to be in coastal/rural areas with a primarily White British population. Therefore, over 90% of the participants described themselves as White British. However, there were some differences between arms in ethnicity of participants; 94.4% were White British in the GSG arm compared to 85.4% in the UC arm. These differences likely reflect regional variations between the recruiting sites.

Our experiences demonstrated the challenges of recruiting an inclusive cohort of participants, particularly from minority ethnic groups. Our team have therefore recruited two community engagement research fellows to develop methods and resources to enhance engagement in research. Some of the lessons learnt are shared in Report Supplementary Material 1. In addition, team members have obtained research funding to understand how stroke survivors from the local south Asian community can best be engaged in research (https://ageingstrokeresearch.org/research-projects/developing-approaches-to-collaborating-with-the-south-asian-community-to-develop-stroke-research-priorities/).

Impact and learning

Our evidence review work and detailed qualitative exploration of SB in the stroke setting provide valuable insights into the context, and views and understanding of stroke survivors and professionals, which informed our work and is available for others. Learning from this work (e.g. enhancing understanding of SB and importance of behaviour change) informed intervention development. Stroke survivors and their carers were key participants in the coproduction work, enabling an intervention to be developed that was valid and acceptable to survivors of stroke. While the methodology of coproduction is becoming well established, reporting of the detailed methods has contributed to the practical implementation of this approach (article accessed > 8000 times and 35 citations).65 The RECREATE programme has made an important contribution to improving knowledge and understanding of SB in people after stroke, and contributed to the evidence base around development and evaluation of complex interventions in stroke care.

Trial methodology

This trial was established based on the culminate knowledge of a number of very similar trials.44,57,66 Some of the challenges were COVID-19-related, for example the identification of sites, and the set-up procedures were extremely resource intensive and time-consuming. Cluster trials also have their limitations relating to maintaining blindness and ensuring uniformity of recruitment to obtain a balanced trial population.

Recruitment of participants was slower than anticipated: it has to be recognised that recruitment to non-CTIMP trials, in which the ambition is to recruit a cohort of post-stroke survivors (inclusive of participants with communication challenges by virtue of the stroke or through not having English as a first language), takes considerable skills and patience and may possibly be not prioritised over studies which are easier to recruit to.

Follow-up: despite considerable thought and a range of proactive measures, the follow-up rates for this study were suboptimal. This might be due to the recruitment period, as the country was emerging from COVID-19 and as life moved on participants became less interested in this particular trial. However, assessment burden remains a concern in trials of complex interventions where multidimensional outcomes are required. We were unable to see clear reasons for why average health-related quality-of-life outcomes were worse in the GSG group, but one potential explanation is that attrition from the study over time was uneven between the groups.

Outcomes: We have demonstrated that activity monitors are a feasible and acceptable way to gather information on the movement patterns of people after stroke. We undertook a great deal of careful work to refine the administration protocol to enhance ease of use for both staff and participants and innovative work was undertaken to refine the analysis. The result is that we have one of the largest available data sets of activity monitor data for survivors of stroke.

Setting

Considerable background work was undertaken relating to the timing of intervention delivery, but this remains an ongoing dilemma. The busyness, concern relating to falls risk, competing time pressures and priorities in the inpatient setting must be balanced against a wish to implant the behaviour change messages as soon as possible after stroke. There also remains a query around the optimum timing to introduce self-management approaches for post-stroke survivors.67

Intervention delivery

The cluster nature of our trial design was to engage all the multidisciplinary team (MDT) in intervention delivery. Unfortunately, COVID-19 precluded site visits to provide the training as planned, but, nevertheless, attendance at our online events was good (103 training sessions delivered to 381 staff). It is difficult to know whether we could have engaged more people if delivered in person. Ongoing support and engagement were provided to sites. However, greater checking of understanding might have helped, as on occasions staff felt patients were ‘too good’ for the intervention (although reducing SB is important for all). While there was some nurse engagement with the training, the intervention was commonly perceived as a therapy activity despite attempts by the site ‘champions’ to engage the wider team. We worked hard to create clear and simple messaging about the intervention, but the overlay of the trial paperwork (to capture resource use, fidelity and adherence to the intervention) led to some confusion and likely decreased engagement.

Considerations about the intervention

The strength of our intervention was that it was coproduced by staff and survivors of stroke. The message to sit less and move more was generally well received. Although a lot of the staff were on board with the overall principle, particularly therapists, the key message often became lost or confused with promoting physical activity or exercise.

A recent review concluded that interventions such as ours were the most promising.68 Recent work has reported potential interventions like our own, but in one69 there was concern about the definitions used70 and the other while promising included only 14 patients.71

Reflections

Our work was considerably impacted by COVID-19, not only through the national measures but perhaps also due to a general feeling of ‘moving on’ after these life events. Some stroke services had experienced considerable disruption (e.g. stroke units were relocated), and this inevitably affected staff morale and their commitment to this work. We endeavoured to develop an intervention that would continue in the community post-hospital discharge. This was technically challenging, as in many cases the hospital and the community service were in different Trusts (although all part of one stroke service), thereby increasing the bureaucratic burden of study implementation. Although we are fortunate in stroke care to have national standards and audit, the differences in configuration of these services were unexpected.

Conclusions

A Cochrane Review (2021) and our own reviews have reported the lack of effective interventions to decrease SB for survivors of stroke. Our work is novel and important, particularly in the context of the increasing emphasis on the negative impacts of SB in society at large. Our detailed systematic review work including RCT evidence from the wider SB literature provided guidance regarding components of an intervention to reduce SB for survivors of stroke. The real-world observations and interviews (WS2) provided important context and understanding. These findings were presented to the coproduction groups through infographics and through carefully facilitated work. The groups successfully developed an intervention (GSG) which aimed to reduce SB for survivors of stroke using coproduction methods and involving a range of stakeholders. The intervention had content validity for stroke survivors and healthcare professionals and was compatible with routine practice. As such, it is an important output of this programme. Due to COVID-19, we were unable to undertake the planned definitive trial and the external pilot trial we conducted was not powered to detect a between-group difference. The smaller sample has implications for our quantitative results. The exploratory findings reported must be viewed with caution but suggest that the GSG intervention as currently implemented is unlikely to be cost-effective.

Survivors of stroke and healthcare professionals did not always fully understand the concepts of SB and the importance of breaking up sedentariness throughout the day, a finding which could be addressed locally in stroke services, regionally (e.g. Integrated Stroke Delivery Networks) and nationally (by the Stroke Association). Participants appreciated the relevance of the work and components of the intervention developed, but implementation in busy inpatient settings with shorter lengths of stay was difficult. Consideration should be given to increasing awareness of SB in the inpatient setting with materials such as GSG being provided in-person or online in the post-discharge period. However, finding the optimum time to introduce behaviour change interventions to the widest and inclusive cohort post stroke remains a challenging dilemma, which needs consideration by the whole MDT.

Conducting a trial with multiple patient-completed outcome measures is burdensome, especially in this patient group, who may have cognitive impairment, symptoms of fatigue and are likely to be encouraged to prioritise focus on active participation in rehabilitation. It could be helpful to evaluate interventions using hybrid data collection approaches that make use of routine data and the SSNAP. Work is ongoing with others to develop adaptive and resource-efficient trials including platform trials.

Implications for decision-makers

Survivors of stroke and healthcare professionals did not fully understand the concept of reducing SB and the importance of breaking up sedentariness throughout the day. This is an important public health message for everyone and requires clear dissemination.

In clinical settings, there is a challenging dilemma between managing falls risk and promoting the reduction of SB. Staff require clear guidance on these competing priorities.

There is a need to develop interventions that are appropriate for use in practice with clear messages and which require minimal time resource for already busy NHS stroke professionals. Knowledge mobilisation approaches can allow for this.

Recommendations for future research

1. While this external pilot trial was successfully concluded, take-up of the intervention was patchy and participant dropouts greater than we would wish. A fully powered trial would be needed to generate confirmatory evidence of the effectiveness of GSG. We have demonstrated that conducting such a trial would be feasible; however, further refinement of the intervention implementation (reducing content and considering timing as below) and considering more efficient trial designs is required prior to a definitive trial.
2. The GSG intervention remains a potentially useful and important tool. Further work is being conducted to produce recommendations for how GSG can be tailored to different types of survivors of stroke based on a secondary analysis of the patient interview data using ideal-type analysis methods (NIHR206938).
3. Future exploratory analysis could explore trajectories and clustering, that is whether GSG works for some patient groups and not for others. Additional quantitative data to supplement the RECREATE WS5 data set could be beneficial to increase the power of these analyses.
4. Further work will explore clinical staff’s views of assessing stroke survivors’ ability to self-manage aspects of their life after stroke.