Selecting behaviour change techniques to reduce sedentary behaviour in people with stroke using the Behaviour Change Wheel

Background Research has shown that sedentary behaviour increases the risk of stroke, cardiovascular disease and mortality. People with stroke are highly sedentary. Therefore, reducing sedentary behaviour might reduce the risk of secondary events and death. Personalized strategies using behavioural change techniques directed at reducing sedentary behaviour in people with stroke are currently lacking. Purpose To systematically determine the behaviour change techniques (BCTs) for a behavioural change intervention, directed at reducing sedentary behaviour in community-dwelling people with stroke, using the Behaviour Change Wheel (BCW). Method To complete the stages of the BCW, information on understanding the behaviour, identifying intervention functions, identifying BCTs and modes of delivery were needed. To acquire this information, per stage a literature search was conducted and nominal group technique (NGT) sessions were conducted to identify BCTs. The NGT sessions were conducted with professionals working with people with stroke and with international researchers working in the stroke or sedentary behaviour eld. Participants made their choice by rating the BCTs, starting from most important (eight points) down to zero points. Results In total, 75 eligible BCTs were identied. Five BCTs should always be included: ‘goal setting’, ‘action planning’, ‘social support’, ‘problem solving’ and ‘restructuring of the social environment’. For patients without cognitive impairments, ‘self-monitoring’, ‘feedback on behaviour’, ‘information about health consequences’ and ‘goal setting on outcome’ were advised to be included, while for patients with cognitive impairments, ‘prompts/cues’, ‘graded tasks’, ‘restructuring the physical environment’ and ‘social support practical’ should be considered. Conclusion Behaviour change techniques were identied for a behavioural change intervention aiming to reduce sedentary behaviour in community-dwelling people with rst-ever stroke. BCTs recommendations depend on the presence of physical and cognitive impairments, although ‘goal setting’, ‘action planning’, ‘social support’, ‘problem solving’ and ‘restructuring of the social environment’ are recommended in all people with rst-ever stroke. The identied BCTs serve as the basis for further development of a personalized blended care intervention to reduce sedentary behaviour in people with stroke.

Therefore, secondary prevention after a rst-ever stroke is important. Sedentary behaviour increases the risk of all-cause mortality and cardiovascular disease including stroke (5)(6)(7)(8)(9). Studies show that a reduction in the total amount of sedentary time reduces metabolic risk factors, like hypertension and impaired glucose tolerance, associated with an increased risk of cardiovascular diseases (7,10,11). Additionally, prolonged uninterrupted sedentary time, independent of total sedentary time, is associated with poor health and elevated cardiovascular risk factors (7,(11)(12)(13)(14)(15). In people with stroke a clinical relevant decrease of blood pressure was found by reducing and interrupting sedentary behaviour (16). Decreasing sedentary behaviour could already produce health bene ts in people with stroke (6,10,14,17).
Research has shown that people with stroke are even more sedentary compared to healthy peers and sedentary time is accumulated in longer uninterrupted sedentary bouts (18)(19)(20)(21). Since up to 40% of people with stroke experience a decline in activities of daily living after rehabilitation, it is important for patients to have self-management skills to preserve physical functioning (22). In an elderly population, even small reductions in sedentary behaviour increase physical functioning and decrease the prevalence of cardiovascular risk factors and mortality (23)(24)(25). Additional to possible health bene ts, decrease of sedentary behaviour could contribute to prevention of decline in physical functioning in people with stroke.
Only two intervention studies evaluated the effect of in uencing sedentary time in a stroke population. The results of these studies are promising (26,27). The rst study focused on increasing physical activity instead of reducing sedentary behaviour, in addition, sedentary behaviour was a secondary outcome measure (26).
When targeting the reduction of sedentary behaviour, the focus of an intervention should be primarily on reducing sedentary time and interrupting sedentary bouts (6,23). The second study was a feasibility study focussing on decreasing sedentary time and with a small sample (27). At this moment, a systematically developed intervention to reduce sedentary behaviour in stroke survivors is lacking.
Before developing a behaviour change intervention, well-de ned intervention techniques for people with stroke need to be identi ed. The Behaviour Change Wheel (BCW), is a step-by-step theory-based approach to develop behaviour change interventions. The BCW is based on all behaviour change frameworks and theories that currently exist (28,29) (see gure I). The wheel has four layers. The rst layer, the green part of the wheel, starts with Capability (physical and psychological), Opportunity (social and physical) and Motivation (automatic and re ective) in uencing behaviour model (COM-B). These three factors enhance the likelihood of performing a speci c behaviour. The second layer, the yellow part, is the Theoretical Domains  (29). The fourth and nal layer, the grey part, are the policy categories. These categories can be used to support the delivery of the intervention functions.
An intervention to reduce sedentary behaviour in people with stroke should be personalized to improve outcome (30). Additionally, personalization improves adherence and the uptake to the prescribed therapy (30). Therefore, this study aims to systematically determine the behaviour change techniques (BCTs) for a behavioural change intervention, directed at reducing sedentary behaviour in community-dwelling, using the stages of the BCW.

Method
The step by step approach of the BCW was used to selected appropriate BCTs. The BCW involved a series of stages. These three stages are: 1. Understanding the behaviour; 2. Identify intervention functions; and 3.
Identify BCTs and modes of delivery. Per stage different methods were used to collected the information. Literature was searched until September 2018 within PubMed and Cinahl. Search strategies were formulated for Pubmed and adapted for use in Cinahl. Both the stages and the used methods are presented in gure II. Each stage is described in more detail below.

Stage 1: Understanding the behaviour
In stage 1, rst the target behaviour was de ned, selected and speci ed using existing literature and by discussion in the research team. The research team consisted of six experts in the eld of stroke, rehabilitation, physiotherapy, movement behaviour and/or behavioural change. Second, a literature search was conducted to get insight into the behavioural diagnosis. The researchers WH and RW conducted a literature study to identify motivators, barriers and opportunities regarding sedentary behaviour in people with stroke (see table I

Literature research
Effective BCTs and modes of delivery were retrieved from the literature by WH and RW. Search terms used are presented in table I. An overview of BCTs that were found to be effective, not effective, con icting evidence or no evidence to reduce sedentary behaviour was made. WH and RW independently recoded the BCTs of the retrieved intervention studies to the BCW method if needed. In case of disagreement, a third researcher (MP) was consulted. Effective modes of delivery were listed.

Nominal Groups Techniques
After the literature study, Nominal Groups Technique sessions were performed and facilitated by WH and RW. The Nominal Group Technique sessions were undertaken because it was expected that the retrieved BCTs were mainly based on a healthy population. Instead, interventions should be tailored to people with stroke and therefore other BCTs could be more suitable to the stroke population. Additional, symptoms after stroke are diverse and personalization of interventions is needed to improve the uptake of an intervention (30,31). Therefore, four pro les of people with stroke were formulated by the research team based on literature (31) and best practice experience: pro le 1. no physical or cognitive impairments; pro le 2. mainly cognitive impairments; pro le 3. mainly physical impairments; and pro le 4. both physical and cognitive impairments.
Two groups were empanelled formulated to carry out the Nominal Group Technique sessions. Group one, professionals, consisted of physiotherapists working with people with stroke in a hospital, rehabilitation centre and in private practice. All professionals were working in the stroke service of Utrecht. Group two, researchers, were working in the eld of behavioural change, people with stroke and movement behaviour.
International researchers were asked by email to participate in this study. Since the researchers reside in different parts of the world, it was decided to use individual interviews within the NGT structure to receive their input on the content of the intervention. Both the groups session and interviews were audio recorded.
Both professionals and experts received an overview of the BCTs found in literature to be effective, not effective, generating con icting evidence or no evidence before the interview or NGT face-to-face session.
The professionals and researchers were asked to identify all BCTs that might be relevant for the intervention. Based on the answers, the possible relevant BCTs were provided to all participants. The participants and researchers were asked to individually choose the eight most important BCTs per pro le to reduce sedentary behaviour in people with stroke (32). Each individual made their choice by rating the BCTs; eight points were given to the BCT deemed most important, seven points to the second most important BCT and so on. The scores of the individuals were summed per stroke pro le, resulting in an overview of the most important BCTs to reduce sedentary behaviour per pro le.

Stage 1: Understanding the behaviour
Based on existing literature and discussion within the research group two target behaviour were selected. The rst target behaviour is to reduce total time spent sedentary (5)(6)(7)(8)(9)(10)(11)(12)14,33). The BUST-study found a statistically and clinically relevant decrease of the systolic blood pressure by interrupting sedentary behaviour every thirty minutes with a walk of three minutes (16). Therefore, the second target behaviour is to reduce time spent in sedentary behaviour accumulated in bouts over thirty minutes.
From the literature study searching motivations, barriers and opportunities to reduce sedentary behaviour in people with stroke, one study including people after stroke was found (34). The study found that there is limited awareness of health risks of sedentary behaviour among people with stroke. The main reasons for sedentary behaviour were relaxation, comfort, sedentary occupation or inability to get back to work. It was concluded that participants encountered barriers in their daily lives that affect engagement in activity. The main barriers are motor impairments, fatigue, cognitive problems, depression, lack of support from friends and family and lack of motivation to be physically active.Strategies involving wearable technologies for selfmonitoring, movement throughout the day and action planning to reduce sedentary behaviour were found as potential ways to reduce sedentary behaviour according to people with stroke. An additional search focussing an elderly population resulted in one study including elderly women (35). This resulted in fteen reasons to sit, fourteen motivators and six opportunities. All motivators, barriers and opportunities were connected to the COM-B model and Theoretical Domains Framework and can be found in table II.

Stage 2 Identify intervention functions
No evidence was found on intervention functions speci c to reduce sedentary behaviour in people with stroke. Three systematic reviews were found on reducing sedentary behaviour in general populations (36)(37)(38). The following intervention functions were found to be effective and connected to the TDF domains (see No evidence on BCTs and modes of delivery speci c to reduce sedentary behaviour was found for people with stroke. Three systematic reviews were found on reducing sedentary behaviour in general populations (36)(37)(38). The overall conclusion of the reviews was that lifestyle interventions targeting sedentary behaviour speci cally or targeting sedentary behaviour and physical activity at the same time are effective (36-38) for reducing sedentary time. One review already recoded the content of the included interventions to BCT (37).
For the other two reviews, the authors RW and WH recoded the content of the included interventions to BCTs (36,38). An overview of BCTs that were found to be effective is provided in additional le I, table I. The identi ed modes of delivery were face to face group, web-based personal, written materials and activity monitors.

Nominal Group Techniques sessions
In total, six professionals and ve researchers participated in the Nominal Group Techniques sessions. The average age of the professionals was 36 years (range 23 to 51). The average work experience was 13 years (range 2 to 30). All had a bachelor's degree in physiotherapy, and two had an additional master's degree in physiotherapy sciences. Two currently worked in an academic hospital, two worked in a rehabilitation centre and two worked in private practice. All of the professionals were working with people with stroke on a regular basis. The average age of the researchers was 44 (range 41 to 49). All but one had a background as a physiotherapist; the other one was a neuropsychologist. All researchers had a PhD and worked at least parttime as a researcher. All had movement behaviour and/or stroke as their area of expertise.
The participants identi ed, in total, 75 BCTs as possibly eligible to include in an intervention to reduce sedentary behaviour. A mean of 30 BCTs per pro le received points (range 29-33 BCTs). Overall 'goal-setting', 'action planning', 'social support', 'problem solving' and 'restructuring the social environment' were selected in all four pro les. 'Self-monitoring', 'feedback on behaviour', 'information about health consequences' and 'goal setting on outcome' were selected for both pro les without cognitive impairments, and 'prompts/cues', 'graded tasks', 'restructuring the physical environment' and 'social support practical' were selected for both pro les with cognitive impairments. An overview of the ten most eligible BCTs per pro le can be found in table III. An overview of the ranking and frequency of the BCTs for the four different pro les can be found in Additional le I, table II -V .

Discussion
The aim of this study was to determine BCTs for a behavioural change intervention to reduce sedentary behaviour in people with stroke using the BCW. BCTs were ranked by professionals and researchers after the literature was reviewed and the main elements were extracted. In summary, 'goal-setting', 'action planning', 'social support', 'problem solving' and 'restructuring the social environment' were found to be main elements to be included in an intervention to reduce sedentary behaviour in all people with stroke.

Target behaviour
Reducing sedentary behaviour needs to be the target behaviour and the focus within an intervention, rather than enhancing physical activity (37). Sedentary behaviour and reaching su cient levels of physical activity are two different behavioural constructs (39). Additionally, it is di cult for people with stroke to achieve adequate levels of moderate to vigorous physical activity (40). Focussing entirely on sedentary behaviour can already contribute to secondary prevention and could be more achievable for people with stroke, including those with ambulatory di culties. However, a part of the population could be able to reach su cient amounts of physical activity. In this subpopulation, sedentary interventions should be implemented alongside physical activity and exercise interventions to reach an optimal reduction of cardiovascular risk factors (39).
It remains unclear how much reduction is needed in total sedentary time and in breaking up prolonged bouts of sedentary behaviour to gain health bene ts. Already, small improvements seem to have health bene ts in other populations (9,13,14).

Motivators, Barriers and Opportunities
Only one study is conducted investigating the barriers and motivators to reduce sedentary behaviour in people with stroke. This study provided important information with regards to the capabilities, opportunities and motivators in people with stroke to remain sedentary (34). However, for further development of the intervention content it will be important to include people with stroke and their carers to be sure the content connects to the target population (41,42).

Behaviour Change Techniques
The identi cation of BCTs was accomplished through the comprehensive use of the BCW. The BCW ensures that there is a clear de nition of the behaviour and the change needed; this is to make sure there is a thorough understanding of all the aspects of the behaviour.
At least seven BCTs should be included in an intervention. In a review on reducing sedentary behaviour in a general population, it was found that effective interventions included at least seven BCTs (37). Little is known about the amount of BCTs therefore, we presented the top ten BCTs per pro le. However, more research is needed to include the su cient amount of BCTs in an intervention.
Personalization of care is important especially in the stroke population were complaints after stroke are divers (30). Although self-monitoring seems to be one of the most important BCTs to reduce sedentary behaviour, this could be di cult to implement, interpret and translate into behaviour change in people with stroke with cognitive problems (43). A different approach for these patients could be more effective. The results of our study show that social support needs to be included in the intervention for people with stroke with cognitive impairments. The involvement and support of family and friends is therefore highly recommended. Additionally, 68% of people with stroke have at least one cognitive complaint (44), and the variety of physical limitations is wide (45). This underlines the importance of tailoring the intervention (30).
When the individual needs, limitations and motivators of people with stroke are taken into account adherence to the intervention will increase (30). The pro les used in our study can guide the selection of BCTs and the personalization of the intervention.
In this study, the most important BCTs to reduce sedentary behaviour in people with stroke were identi ed.
Further research should focus on the effectiveness of the BCTs for both target behaviours, i.e., sedentary behaviour, in people with stroke. In such research, it is essential to describe BCTs using the Behaviour Change Technique Taxonomy (29). Thorough intervention descriptions in protocol articles are needed and intervention protocols should be available to use in practice. Description of included BCTs, the frequency of use, the intensity and the way BCTs are delivered are important as is education to implement and execute BCTs in daily practice. For example goal-setting is one of the most important BCTs recommended in stroke rehabilitation (46). However, the determination of goal-setting seems to be di cult, and health care professionals nd it di cult to make goals that are patient-centred (47)(48)(49). Education to overcome these problems could be explored and implemented to improve the quality of goal-setting.

Modes of delivery
The identi ed modes of delivery were face to face contact, group delivery, web-based personal, written materials standard and activity monitors. The results of our study underline the importance of a blended care intervention. To optimize personalized secondary prevention, blending care seems to be promising. The use of computer, mobile and wearable device (eCoaching) can be effective to reduce sedentary behaviour (50). Persuasive eCoaching, the use of technology during coaching to motivate and stimulate people to change attitudes, behaviour and rituals (41), could be a useful in reducing sedentary behaviour in people with stroke but this needs further research. ECoaching on its own showed only short term effects (50).
Whereas, eCoaching and face-to-face contacts together showed more sustainable behavioural changes (51).
However, this is not yet investigated in people with stroke. Activity monitors are highly important to gain insight into the individual behaviour and give real time feedback on behaviour (37). Therefore, an intervention including activity trackers, persuasive eCoaching and face-to-face contact could be a promising approach (41). Although the most important modes of delivery and BCTs are identi ed, a detailed description of an intervention needs to be further explored.

Study limitations
Based on the amount of consistent literature found and the thoroughness of the search, the literature research seems complete and comprehensive, although this is not a systematic review. Some information was retrieved out of other populations and should be further investigated in a population with people with stroke. Another limitation is that even though the description of the BCTs is quite elaborate, there is still some room for interpretation. Care was taken to make comprehension of the BCTs as clear as possible.
To get the insights of the researchers, the original Nominal Group Techniques process could not be followed. To make sure the most renowned researchers were involved in the selection of the BCTs, it was decided to include not just Dutch experts but researchers from around the world. Therefore, the NGT method was converted into an interview-based method. Although some of the group dynamics were compromised, a step-based method was used to ensure that all participants were informed of the identi ed possible BCTs before the individual ranking.
Almost all participants stated that their choice of the use of a BCT in clinical practice is partially based on the person in front of them and their limitations caused by the stroke. This is in line with the distinction made in the ranking by using the four pro les; these pro les are an attempt, at this point in the development, to do as much justice as possible to the individual differences. However, personal factors have to been taken into account. Additional to stroke characteristics personal factors like coping style, neuroticism and optimism are associated with functioning after stroke (52). When personalize an intervention these factors should be taken in to account. This study provides important information to personalize an intervention by selecting the right BCTS and mode of delivery based on the individual. Within the development of an intervention all stakeholders should be included. Within the design team for behavioural change interventions in stroke patients all professionals involved in stroke care, people with stroke themselves, proxies, behavioural experts and as well as technology experts should be included from the start of the design process (42). Ethics approval and consent to participate Ethical approval was not applicable to this study since researchers and professionals were included. Informed consent was given by all participants and the studies were performed in accordance with the Declaration of Helsinki. Professionals gave written informed consent and researchers gave oral consent.

Conclusion
Oral consent was audio reordered.

Consent for publication
Not applicable Availability of data and material The NGT ratings are included as additional le.

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