Included RCTs

To address objective 1, and provide context for the process evaluations, online supplemental file 3 presents data from trials with included process evaluations, including: aims, inclusion/exclusion criteria, sample size, participant characteristics, study design, intervention and control descriptions, data collection and follow-up time points and outcome measures used.

RCT aims

Associated trials where sedentary behaviour was measured as an outcome were published between 2007 and 2020. Five trials focused specifically on reducing sedentary behaviour19–21 or sitting time.22 23 The remaining 12 trials aimed to increase physical activity or promote healthier lifestyles but measured sedentary behaviour as an outcome (online supplemental file 3).

Trial location and participant characteristics

Seven trials were conducted in the UK,20–22 24–27 the remainder in the USA,19 23 28 Netherlands,29 30 Brazil,31 Ireland,32 Canada,33 Hong Kong34 and Belgium.35 Participants recruited into the trials varied, including: mothers or parents of infants, pregnant women, adults, older adults, overweight adults, individuals with chronic illnesses and individuals with intellectual disabilities or serious mental illnesses. Most trials included males and females, however, three included females only.19 26 28 Participants’ ages ranged between 30 and 75; the majority of trials included participants aged between 40 and 50 years.19–21 24 25 29 30 32 33 Only nine trials reported ethnicity, the most ethnically diverse study was by Albright et al28 which reported the following ethnicities: Native Hawaiian, Pacific Islander, Asian, mixed race, white, black-Native American.

Included process evaluations

Online supplemental file 4 presents data specific to the process evaluations including: aims and whether process evaluations were prespecified, sample size and sampling methods, study design and data collection methods, and theoretical frameworks used. These data provide further context for the narrative synthesis.

Thirteen process evaluations were pre-specified in published protocols, or trial register records. Five studies19 26 30 32 35 were published prior to the MRC guidance,4 which was developed to provide a more systematic approach for planning and conducting process evaluations. The majority were published in the same year or after the guidance was published.20–25 27–29 31 33 34 Despite this, nine studies did not report using a framework or guidance19 21 23 24 28 32–35 only four authors cited the MRC guidance20 22 25 27 and only one reported using this to guide the process evaluation.25 As shown in online supplemental file 4, five studies cited other frameworks20 26 29–31 the most common alternative to the MRC framework being the RE-AIM (Reach, Effectiveness, Adoption, Implementation, and Maintenance) framework.36 Fourteen used the term ‘process evaluation’ within the publication. Three did not use this term.23 24 34

Process evaluation aims

There was considerable variation in process evaluation aims. Some studies had a broad focus on participants’ experiences, for example, Elramli24 aimed to explore participants’ views regarding the effectiveness of a walking intervention for rheumatoid arthritis (RA). Others focused more specifically on barriers to achieving activity goals,28 or barriers and facilitators to the sustainability of an intervention.29 Some focused on the feasibility and/or acceptability of interventions among different participant groups, including those at risk of chronic disease;33 older adults;23 individuals with intellectual disabilities;20 and individuals with serious mental illnesses.21 Only two process evaluations were conducted with a view to refine the intervention.26 27

Study design and data collection methods

As outlined in online supplemental file 4, sample sizes of participants recruited to the process evaluations varied, from 521 to 411.29 A total of 1553 participants were included from intervention groups across the 17 studies and 340 from control groups in four studies.26 29 34 35

Nine studies19–22 25–27 29 31 used mixed-methods, most commonly combining quantitative questionnaires with semistructured interviews (telephone and face to face). In five studies, questionnaires were used to ask participants about their satisfaction with the intervention, intervention fidelity, and about suggested improvements to interventions.19 27 29 32 35 In two studies, questionnaires focused on intervention providers’ experiences of delivering and participating in interventions.25 30

Semi-structured interviews explored intervention contexts, barriers and facilitators to intervention delivery, and experiences from the perspective of intervention providers, participants, and their family members or carers.21–27 29 31 33 34 Other methods used included: non-participant observations,19 focus groups,20 25 27 31 34 healthcare professionals’ registries and log books.29

Screening questions

The majority of studies had clear research questions or aims, and appropriate data were collected.

1. Qualitative studies

   • Thirteen of 17 included studies had a qualitative component. Four21 26 27 34 were rated as not meeting some of the criteria in this category, because descriptions of the analysis process lacked detail, and it was unclear how authors arrived at their findings. In these studies, findings were commonly presented as a series of quotes, in tables or online supplemental files but interpretation was considered too limited to constitute an in-depth analysis.

2. Randomised controlled trials

   • Each of the included studies was associated with an RCT. This category of the MMAT was used to assess the quality of the trials. The ‘can’t’ tell’ option was most commonly used in this section because authors often provided insufficient information to provide an answer, particularly regarding the randomisation process and blinding. Scoring was more mixed within this category and no studies scored yes for all questions.

3. Non-randomised studies

   • The associated trials were all RCTs; therefore, this category was not applicable.

4. Quantitative descriptive studies

   • Thirteen studies had a quantitative component. Overall, they were rated positively across all questions.

5. Mixed-methods studies

   • We considered studies which used methods meeting the criteria for both categories 1 and 4 as mixed-methods studies. This category was only applicable for nine studies. When studies were rated negatively on either the qualitative or quantitative component, it was reflected in the judgement for this category.

Description of the interventions and their causal assumptions

According to Moore et al,4 a clear description of the intervention and its causal assumptions are an important part of understanding how other factors (eg, implementation, context and mechanisms of impact) influence outcomes.

Online supplemental file 5 describes the content and delivery methods for all interventions. Intervention delivery periods ranged between 6 weeks and 18 months. All interventions included multiple components, examples include group based educational sessions combined with email input and self-monitoring tools19 or one-to-one counselling combined with tailored email input.28 In terms of delivery, interventions commonly incorporated some group based input or support.19 21 22 24–26 29 31 34 Interventions were delivered by a range of providers including researchers,19 health educators,22 28 exercise professionals, including personal trainers,20 29 coaches,21 23 33 advisors and nurses.25 30

Online supplemental file 5 also includes information about the mechanisms by which the interventions are intended to have an effect, and any theoretical underpinnings. All interventions were underpinned by theory or incorporated behavioural change techniques, the most common theory being social cognitive theory.37

Implementation and delivery approaches

Moore et al4 recognise that interventions can have limited effects due to weaknesses in how they are designed, or because they are not properly implemented. This section outlines the extent to which interventions were reported to be delivered as intended, common approaches used in intervention delivery, and whether this reportedly translated into changes in outcomes.

As indicated in online supplemental file 5, in three studies21–23 interventions were reportedly delivered as intended. In seven studies,19 20 25 28–30 33 adaptations were made to the interventions during the course of the trial. In the remaining seven studies,24 26 27 31 32 34 35 it was difficult to determine whether there were any adaptations as authors only reported the actual delivery, not the intended delivery.

Approaches for achieving intervention delivery included: ensuring staff were appropriately trained and prepared to deliver the intervention with fidelity;19 31 tailoring aspects of the programme to individuals and their needs (eg, ensuring activity consultations are appropriate for those with intellectual disabilities20); and allowing for flexibility in delivery methods. For example, in Poston et al26 pregnant women were provided with the option of receiving the intervention via phone or email, rather than sessions delivered at the hospital, and in Berendsen et al29 coaching meetings as part of the intervention were planned with consideration of holidays and health issues.

Despite these adaptations for enhancing fidelity, interventions were not always effective in achieving the intended outcomes. For example, in Poston et al26 despite flexibility in the delivery mode, objectively measured physical activity and sedentary behaviour did not change in the intervention group. In this particular participant group (pregnant women), the potential to achieve the targeted health outcome, optimal blood glucose level, via dietary changes, was greater than changes in physical activity, including sedentary behaviour, as for some participants increasing their activity led to feelings of discomfort. Similarly, in Matthews et al20 although individual tailoring was used, the intervention did not have a significant effect on any of the primary or secondary outcomes including time spent in MVPA and time spent sedentary. It was suggested that this intervention may need to be longer than 12 weeks for individuals with intellectual disabilities. This highlights the importance of understanding more about how an intervention is intended to have an effect, as outlined in the following section.

Mechanisms of impact influencing intervention effectiveness

Moore et al4 emphasised the importance of exploring mechanisms through which interventions bring about change, to learn more about how the intervention effects may have occurred and how they may be replicated in similar future interventions. This section outlines the mechanisms reported across the studies and the extent to which they impacted on behaviour and outcomes.

Social Cognitive Theory was the most commonly used theory, and the following mechanisms of action were reported in several studies: enhancing self-efficacy by rating confidence in completing goals;19 using behavioural cues, for example, standing up every hour, and leaving the remote at the TV;19 using resources, for example, websites combined with counselling calls to encourage goal setting28 providing social support in educational sessions or workshops, and input and engagement from carers.19 20 22 24 28

However, across the studies, the extent to which these mechanisms had their intended impact on behavioural change varied. In Elramli24 the intervention aim was increasing daily step count, social support was found to be a key factor in participants who increased their physical activity. However, behavioural change techniques including social support, feedback and self-monitoring were to a lesser extent associated with reduced sedentary behaviour in those with RA. In Matthews et al20 where the intervention aimed to increase walking and reduce sedentary behaviour, the social support component was not effective for adults with intellectual disabilities. In Biddle et al22 where the intervention aimed to reduce sitting time, there was no difference in sedentary time at 12 months between intervention and control arms. Reasons for a lack of change in sedentary behaviour included: a preference for adopting physically active behaviours rather than sitting less, and motivational drift after 3 months. In Adams and Gill19 which focused on reducing sedentary behaviour and increasing light physical activity, self-efficacy was not shown to be a predictor of change in sedentary behaviour. Behavioural cues, for example, leaving the remote at the TV, did not always influence behaviours either, because some participants were already doing the cued behaviour, and some did not have a TV.19

Studies underpinned by the Transtheoretical Model, Theory of Planned Behaviour and Self-Determination theory placed emphasis on encouraging participants to be aware of and monitor their own behaviour.20 29 30 Motivational interviewing was used in two studies to prompt participants to find solutions, rather than telling them how to change their behaviour.29 30 Berendsen et al29 found the feasibility of changing physical activity behaviours and dietary habits was not as high as expected and was likely associated with poor adherence. Some participants were unrealistic about how much of their own effort would be required, which influenced attendance at meetings. Lakerveld et al30 reported that practice nurses were competent and confident in the delivery of motivational interviewing and participants’ satisfaction was high, but even so, almost no effects were seen in the determinants of behavioural change in this population of individuals who were at risk of cardiovascular disease and diabetes.

In summary, these findings provide some insights into how mechanisms may or may not have an effect on sedentary behaviour, highlighting that it is important to fully understand the complexities of interventions.

Factors including context that facilitate or hinder implementation or how participants respond or interact with the intervention

Moore et al4 regard understanding context as an important part of interpreting factors influencing whether interventions are effective. They defined context as anything external to the intervention that may act as a barrier to its implementation or effects. They also considered participants’ responses to and interactions with the intervention as important mechanisms that could influence outcomes. Drawing on the coding framework, this section is divided into include barriers and facilitators to delivery of interventions, barriers and facilitators to participation and engagement, and understanding of participants experiences from different perspectives.

Barriers to delivery of interventions

Across the studies, there were a range of barriers to delivering interventions, including administrative or scheduling issues and organisational difficulties or challenges. In two studies, planning educational sessions around other commitments including holidays and childcare responsibilities was difficult for staff.24 34 In Blunt et al,33 a central research team were involved in scheduling appointments, intending to reduce the workload for coaches. However, this resulted in increasing time spent scheduling and it was recommended that coaches were best placed to take responsibility for their own scheduling.33

Organisational difficulties were apparent across two studies.20 31 A community health worker from one of the six health centres in Benedetti et al31 described the long absence of a doctor as a turbulent time in the unit, which added difficulties in trying to deliver the intervention. In Matthews et al20 the intervention was implemented at a time of significant change within the local learning disability service. Provision of support was affected by the closure of many day centres, which led to a low morale and increasing work pressures among the staff. In Berendsen et al29 there were factors that influenced adherence; additionally suspended government financial and policy support meant the programme could not continue.

Barriers to participation and engagement

Across the studies, there was a range of barriers to participation and engagement in the interventions. The most common barriers to engagement were: having a pre-existing illness or injury and associated problems, for example, pain,19 23–29 33 having other commitments, for example, work, caring responsibilities;23 24 26 28 and being too tired.22 26 33 Other, less common barriers to engagement included loss of accountability for behaviour over time,33 fluctuating mental health21 and lack of motivation.24

Some participants also experienced difficulties with pedometers and accelerometers used as an outcome measure for the trial, in terms of understanding how to use them, side effects of wearing them, for example, skin irritation19 23 and lost devices.19 22 In Biddle et al,22 half the participants experienced problems with the software for the ‘Gruve’ accelerometer, including: computer synchronisation issues, incompatible computers, website navigation problems, device malfunction, short battery life and charging issues.

Some barriers may be more applicable to specific groups. For example, in Benedetti et al31 a community health worker perceived some older people to be apprehensive about new things which may have been a barrier to participation. In another study, a participant thought that sitting was deserved in old age and he was looking forward to this aspect of retirement to indulge in some of his passions, for example, reading and studying, which made him resent the idea of standing more.23

Some barriers were specific to particular contexts. In Elramli,24 participants who had RA worried about using the gym because they lacked knowledge of suitable, safe exercises. Although workplace interventions were not included in this review, participants who had received educational based interventions reflected on how this applied to other parts of their lives, and therefore, provided some insight into how the work setting impacts on sedentariness. For example, participants felt that it was not appropriate to be standing in a work context which could cause embarrassment, for example, the expectation to be seated for meetings.19 22 23 Further barriers at work included having no access to stairs and no standing desks.22

The context of other parts of everyday life was also influential for some participants who had developed ingrained sedentary habits, as a result of their usual activities or hobbies, for example, reading, eating, socialising, TV viewing and knitting.23 Religious festivals had an impact on willingness to reduce sitting time at certain times of the year, for example, Christmas and Ramadan.25

Facilitators to the delivery of interventions

Some of the approaches for achieving implementation and delivery could be regarded as facilitators, including: allowing flexibility in delivery methods, tailoring aspects of the programme to individuals, initial preparation and planning. A range of other factors facilitated intervention delivery.

For example, in Blunt et al33 coaches valued the simplicity and structure of the programme. They also appreciated that the programme did not require extensive background knowledge or preparation over and above their existing working requirements. Coaches had the option of referring back to the Canadian Physical Activity Guidelines to ensure they were providing the right level of support to participants. In another study, not requiring too much additional trial focused expertise, and having access to useful trial related resources was valued by social workers.34 In this study, the research team prepared and organised most of the materials which facilitated delivery. As a contrast to low morale among staff,20 having a committed team was also important for facilitating delivery.34

Facilitators to participation or engagement in intervention

There were a range of facilitators to participation and engagement in the interventions. The most common facilitator was support and encouragement from providers and peers; participants valued personal interaction and having someone to keep them on track with the intervention.20 24 25 27 31 33

In some studies, group environments facilitated engagement and provided opportunities for sharing experiences and meeting other peers in a similar situation.21 24 27 In Matthews et al20 many participants liked one-to-one engagement with intervention providers. This was particularly beneficial to the group who had intellectual disabilities, partly because the conflicting needs of participants in group activities were occasionally disruptive. This group faced challenges to engagement with the intervention, compared with the general population. Matthews et al suggested the need for providing interventions to people with intellectual disabilities for longer than 12 weeks, so that consultations with providers can address more barriers.20

Being accountable to someone, for example, a health coach, also facilitated engagement in three studies because the participants felt being monitored provided motivation.20 23 25 33 While use of a step count monitor was a barrier for some, others found this was a good motivator.23 24 Adams and Gill19 recommended that in order for pedometers to be beneficial they need to be more accurate. It was also suggested that technology should be tailored to detect movement in older adults which may be different from younger adults.23

Participants valued textual resources that were considered attractive through using appropriate text and images.20 31 Adams and Gill19 made recommendations for making resources more accessible including embedding videos in emails rather than asking participants to use YouTube, and printing cue cards out rather than asking participants to do so themselves. Less common facilitators were: already being involved in health programmes,33 and becoming more aware of the extent of their own sedentary behaviour.23

Understanding experiences of interventions from different perspectives