Implementation strategies, and barriers and facilitators for implementation of physical activity at work: a scoping review

This scoping review is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) [26].

A systematic search of the following databases was undertaken: Embase, MEDLINE, Scopus, and PsycINFO. All databases were searched from 2008 until March 2018. Reference and citation searching were also undertaken. The searches were performed by the first author, who was guided by an experienced information specialist from the library at the University of Southern Denmark.

The search strategy included subject indexing terms and free-text terms for title, abstract, and keyword searching. The research question directed the entire literature search, and based on this, the search terms were grouped into three concepts and arranged in accordance to relevance in a search matrix: 1) implementation, 2) physical activity, and 3) workplaces. Search terms under each of the three concepts were selected from keywords identified in a preliminary search in PubMed and the list of Medical Subject Headings (MeSH), and after discussion with the review team. The full version of the search terms used, including specifications on use of title, keywords, or abstract screening, is documented in Additional file 1. In all four databases, the search was performed according to the block search method using the Boolean operators “OR” and “AND”. To maintain an overview throughout the literature search, a search protocol was continually populated with all search terms, number of hits, and combination of searches (Additional file 1).

All identified citations from the searched databases were uploaded to EndNote ×8 software. An integrated duplication detection tool was used to identify duplicates. All suggested duplicate pairs were screened for correctness by one reviewer (AGD). Title and abstract screenings were performed for each article by two independent reviewers (AGD and SM) [27]. Disagreement between the two reviewers resulted in inclusion of the citation to full-text screening. Full-text screening was similarly performed by two independent reviewers (AGD and SM) assessing the eligibility of the citation. Any disagreements were resolved through discussion mediated by a third reviewer (MJS) [27].

The eligibility criteria for the original studies are presented in Table 1. Unlike systematic reviews and meta-analyses, the eligibility does not necessarily have to be established before the literature search in a scoping review. These can be developed as the knowledge of the identified literature grows [29]. In accordance with this approach, the eligibility criteria for this scoping review were adjusted after the screening of the titles. Interventions relating to return-to-work, sickness absence management, and occupational safety were added to the exclusion criteria, due to irrelevance in accordance with the review focus.

Similar to the study selection process, data extraction was performed independently by two reviewers (AGD and SM) using predefined data extraction spreadsheets. Discrepancies in data extracted were negotiated until consensus was reached. Data were systematically extracted on study characteristics (year, country, study design); study participants (occupation, number of participants); intervention (content, duration); implementation approaches and evaluation (details on methods of implementation, evaluation methods applied, main findings); and barriers and facilitators (methods of data extraction, factors or themes of either enhancing or hampering effect on implementation). In qualitative studies, barriers and facilitators were extracted in their original format, unless the authors had coded the factors to specific themes indicating otherwise. In quantitative studies, factors were extracted if 50% or more of the participants indicated the factor as a barrier or facilitator. If the authors of the original study did not state the factors as barriers or facilitators, the reviewers would assess the influence on the implementation, as either enhancing/positive (facilitator) or hampering/negative (barrier).

To further organize and make sense of the data, we used TDF to code the extracted barriers and facilitators into domains. TDF contains 14 domains, which offers a theoretical perspective on the cognitive, affective, social, and environmental influences on behaviour [24, 30].

A coding manual (Additional file 2) was developed to guide the coding process [23, 30, 31]. Two coders (AGD and SM) independently coded barriers and facilitators. Barriers and facilitators could be coded to more than one domain if deemed relevant. Any disputes about the meaning of domain definitions or coding of factors were resolved by discussion until agreement was reached.

The study population, implementation strategies, evaluative methods and key findings of the included studies were descriptively summarised and presented. The results of the TDF coding process were summarised for each domain, and the absolute number and proportion of codes was calculated for the barriers and facilitators respectively. Lastly, the main themes from the coding process were identified and examples from identified barriers and facilitators were presented to illustrate the coding process.

We identified a total of 8,715 citations. From these, 2,455 citations were excluded as duplicates and 3,846 citations were excluded due to publication date or language. A total of 2,414 titles and abstracts were screened, which resulted in screening of 50 full-text papers. Nine references met the inclusion criteria [32‐40]. Additionally, seven references were identified by searching reference lists of included studies [17, 41‐46]. This resulted in a total of 16 included articles concerning eight different studies. The PRISMA flow diagram demonstrating the selection process is illustrated in Fig. 1 [47].

Table 2 presents an overview of the included studies. Three of the eight studies were undertaken in the United States [32‐34, 39], three in Denmark [35, 38, 40], and two in the United Kingdom [36, 37]. The studies were published between 2013 and 2018. Four studies were qualitative [32, 33, 36, 39, 40], one study was quantitative [38], and three studies were mixed-methods [34, 35, 37]. The reports included a total of 2,686 participants with a range of 41–1,260 participants per study.

Five out of eight studies included office workers and/or white-collar workers with mostly sedentary work [32‐35, 38, 40]. One study included university employees [36], and one study included participants from five different companies (bus company, hospital, university, city council, and government agency) [37]. One study focused on employees who did not meet the recommendations for physical activity [36], and one study included employees who were experiencing pain in the neck and shoulders [40]. Only one study focused on an occupation with high physical work demands (firefighters) [39].

In five out of eight included studies the physical activity intervention was comprised by cardio [32, 33, 35], strength [32, 33, 35, 38‐40] or functional exercises [35]. The remaining three interventions encompassed treadmill workstations [34], an indoor bicycle HIIT program [36] and a motivating toolkit and team challenges [37]. Three interventions were group-based [32, 33, 36, 37] and five focussed on the individual [34, 35, 38‐40]. The interventions either took place at the worksite [32, 33, 38, 39], near the worksite [36] or both [35, 37, 40]. The duration of the interventions spanned from 10 weeks to 2 years and incorporated different time schedules. E.g. Taylor et al. applied exercises 15 min a day, one study applied indoor bicycle HIIT for 18–25 min, three times a week [36] and another scheduled exercise for 1 hour a week [35]. Also, different methods of instruction were applied across the interventions, differing from trained participants being in charge of instruction [32, 33], to professional instructors overseeing some [35, 38, 40] or all training sessions [36, 39].

Five out of eight included studies applied more than one method of evaluating the implementation process [34‐38]. The most commonly applied methods were focus group interviews (n = 5) [34‐37, 39] and surveys (n = 5) [33‐35, 37, 38]. Individual interviews were used in three studies [32, 35, 40], fieldnotes in three studies [35‐37], and one study used training diaries to evaluate implementation [38]. Six out of eight studies evaluated the implementation from the employees’ perspectives only [32‐34, 36‐38, 40], and two studies evaluated from the managers’ and employees’ perspectives [35, 39].

The interventions were initiated by information meetings for the participants in three studies [17, 38, 42, 44], and by a kick-off event in three other studies [34, 41, 43]. Three studies used handpicked employees, who were given between one and three days of instruction, as “change agents” or “ambassadors” to facilitate implementation [17, 35, 37, 41, 43]. Other methods of implementations were group exercise to enhance social relationships [46], and partnership through signed declarations of support (Break Buddies) [41]. Two studies offered prizes and economic incentives for participation in the intervention [34, 41]. Only two studies described the visual design of the intervention with regards to information material, logos, and posters [43, 46]. Bredahl et al., 2015, described a focus on the physical surroundings, in terms of light and friendly training environment and colourful posters on the walls showing the training exercises [46].

A total of 109 factors were identified in the eight included studies and were divided between 57 barriers and 52 facilitators. When applying the TDF, the 109 factors were given 181 codes: 94 codes to the barriers and 87 codes to the facilitators. Thus, 53 factors were given codes from more than one TDF domain. Table 3 presents the overall results of the TDF coding. For each study, the number of identified barriers ranged from two to 12, and for facilitators from three to nine. The barriers were coded under 11 of the 14 TDF domains and most frequently, to the TDF domains Environmental Context and Resources (ECR, n = 34, 36.2%), Social influences (n = 13, 13.8%), and Social/Professional Role and Identity (SPRI, n = 8, 8.5%). We did not find barriers relating to the domains Optimism, Reinforcement and Memory, Attention and Decision Processes. Likewise, the facilitators were most frequently coded to the TDF domains Social influences (n = 17, 19.5%), Environmental Context and Resources (ECR, n = 16, 18.4%), and Social/Professional Role and Identity (SPRI, n = 9, 10.3%). The facilitators covered all 14 TDF domains. For an overview of the TDF coding, see Additional file 3.

Barriers and facilitators coded to the ECR domain covered organisational culture and resources and employees’ interactions with, and influences of, the surrounding environment. Under the ECR domain, all eight studies identified lack of time, conflicting work time schedules, or alternating workstations/work locations as barriers [32, 34‐40]. E.g., in Taylor et al., 48% of the data texts collected from interviews regarding barriers concerned issues with scheduling the exercise, time constraints, and interruptions of work flow [32]. Further, vacation or sickness absence interrupting the program was perceived as a barrier [34, 37]. Several studies reported suboptimal implementation because of a lack of awareness among employees about the intervention [36, 37] or workstations not suitably equipped for both work and exercise [34]. For example, Tudor-Lock et al. reported how fitting a rigid schedule of shared use of treadmill desks with the equally rigid work schedule was challenging [34]. In addition, the treadmill desks lacked necessary equipment and compromised the confidentiality during telephone calls [34]. Four studies reported that study participants lost motivation because the exercise programs lacked variation [33, 38‐40], and in one study, some participants were disinclined to engage in workplace-based exercise, as they felt this activity belonged to leisure time and private life rather than work life [35]. Other barriers coded under the ECR domain were disturbing noises from workstation treadmills [34] and increased room temperature as a result of exercising [36].

Facilitators under the ECR codes related to the positive evaluations of the content of the exercise programs with respect to intensity and frequency of the exercises [36, 38], the simplicity of the program [39], and the flexibility of the programs in terms of location, time spent on exercising, and timing during the workday [36, 40]. In two studies, the opportunity to meet and spend time with colleagues was mentioned as a facilitating factor [37, 40]. Further, structured breaks [32], change agents [35], the work team working in the same location [37], and clear distribution of roles and responsibilities between project implementers and management [35] were reported as facilitating.

Factors that impact uptake of the interventions and coded to the Social influences domain covered interpersonal processes, social support and norms, group conformity and identity, and intergroup conflict. Three studies reported lack of managerial support [32, 33, 35, 37], and four studies reported lack of social/collegial support as factors which impede uptake [32, 36, 39, 40]. For example, lack of camaraderie amongst colleagues was reported as a barrier by Taylor et al. [32], and in Mayer et al., participants reported that they would have been more motivated by group-based exercise than the implemented individual exercise [39]. Bredahl et al. reported that doing exercise and sweating in a public place was perceived as a barrier for some participants. In the same study, colleagues were found to hinder PA if they were pressuring others to keep working instead of doing exercise [40]. Other barriers under the Social influences domain were lack of motivation/commitment [32, 37] and instructors without appropriate competences and/or behaviour [36, 40].

Just as the lack of social/collegial support was found to impede PA, the presence of this support was reported to facilitate workplace PA [36‐40]. Building a sense of team spirit or camaraderie amongst colleagues and doing activities together with colleagues were reported as means to create this support [36, 39, 40]. Further, supervision of the exercise by instructors [36, 38, 39] and support, acknowledgement, and active participation from management facilitated uptake [32, 35, 37, 40].

Factors coded to the SPRI domain covered behaviours and personal qualities, professional, social and group identity, and organisational commitment and leadership. Barriers coded to SPRI related to inconsistent or lack of support from management [33, 35, 37]. Lawton et al. reported that despite being supportive at the beginning, management would not allow employees the time and resources needed for the intervention [37]. In Taylor et al., participants suggested that management should participate more in the exercise sessions and be more encouraging [33]. Further, lack of project management and unclear roles between implementers/project managers and workplace managers impeded uptake [35].

The SPRI facilitators included sense of acceptance and legitimacy of participation as a result of managerial awareness, attitude and support [33, 35, 37, 40]. Other facilitators included team spirit and improved social work atmosphere [32].

Together, the ECR, Social influences and SPRI codes make up 58.5% of the coded barriers and 48.2% of the coded facilitators. The remaining 41.5% of the barriers are divided between eight domains, with the domains Skills, Beliefs about consequences, Intentions and Beliefs about capabilities with the highest representation. For the barriers, these domains cover lack of knowledge regarding health promotion, the intervention being too difficult or too easy [34, 35, 39], a sense of interference with one’s private life, the timing of the intervention [35, 37, 39], prioritisation of tasks, lack of commitment [35, 37], difficulties in the execution of the intervention, or loss of motivation in the event of negative results [34, 36]. The remaining 51.8% of the facilitators are divided between the remaining 11 domains, with the domains Beliefs about capabilities, and Goals, Knowledge and Emotions being the most prevalent. Acting as facilitating factors were feelings of enhanced competence and self-confidence [36, 39], or motivation to behaviour change and exercise [33, 40], increased health [32, 34, 39], and decreased pain [39, 40], or the experience of positive feelings towards, and greater joy with, physical activity [32, 34, 36].

We have systematically searched and descriptively summarized the literature pertaining to implementation approaches of workplace-based PA interventions, and the barriers and facilitators affecting uptake of such interventions. We found a large degree of variety regarding intervention content, delivery, and implementation; thereby, making comparisons between interventions difficult. We applied TDF to organise and describe barriers and facilitators, and found factors related to all 14 TDF domains, which underpins the multilevel dimensions and complexity of implementation of workplace-based PA interventions.

The majority of factors affecting uptake of workplace-based PA were coded to three TDF domains, ECR, Social influences, and SPRI. In most cases, we consistently found that absence of these factors hindered uptake, whereas presence facilitated uptake of the intervention. This indicates that focus and action must be directed toward these factors, otherwise they will hinder the uptake when implementing PA in workplaces. We found both intrinsic factors in the individual participants, like unwillingness to participate in workplace PA as this was believed to infringe on private matters, or participants feeling exposed when exercising in front of colleagues, and extrinsic factors related to different levels in the organisation and the persons acting within the organisation. These included organisation of work task and work schedules, prioritisation and allocation of resources, managerial support and commitment, and the importance of social coherence and group dynamics. The content, flexibility, variation of the intervention and competency of those delivering the intervention also impact uptake.

We found that a variety of methods have been applied in the implementation of workplace-based PA interventions, but in general, the engagement strategies used were not described in detail. Information meetings, kick-off events, and the application of handpicked, educated employees as “change agents” or “ambassadors” were the most common methods of implementation. The concept of “change agents” is advocated by several theories and models of behaviour change and implementation, and is based on the theory that any process of change needs someone to take the lead [48, 49]. “Change agents” have been found useful in diverse settings and conditions, such as prevention and health promotion programs among children [50]; weight management in adults with intellectual disabilities [51]; application of ergonomics among industrial workers [52]; and resident-oriented care in nursing homes [53]. We found that individually tailored interventions, adaptability, and flexibility in intervention programs increase uptake. This finding is supported by reviews of both workplace-based health programs [16], and in implementation studies from a wide range of health conditions and situations [50]. Other authors have advocated for short rather than long exercise programs [15]. Further, in concordance with other workplace-based health programs, which have shown that various incentives (e.g., gifts and gift vouchers) improve participation and intervention adherence [16], we found that offering financial incentives increases uptake. Also in line with our results, several other studies stress the importance of leadership support, a collective sense of ownership, allowing employees to take part in the development, a supportive workplace culture, and focus on economic advantages, when implementing health promoting interventions [16, 20, 50, 54]. Identifying which engagement and recruitment strategies are most effective for different groups of workers would be beneficial. Conceptual frameworks and metrics have been developed in other areas of health research to help determine which approaches are most successful [55, 56]. It will be important that future studies describe engagement or recruitment strategies in greater detail to improve the fidelity and impact of these approaches using resources and guides like the template for intervention description and replication (TIDieR) checklist [56, 57].

The majority of identified barriers and facilitators were coded to the TDF domains ECR, Social influences, and SPRI. This mirrors, to some extent, the results of other studies, which likewise found ECR and Social influences to be of greatest influence when implementing PA. This has been seen in implementation of PA in school settings [31, 58], and in mental and physical rehabilitation settings [59‐61]. Some factors (e.g., interaction with colleagues) were perceived differently among participants, and thus reported as both a barrier and a facilitator. This was observed both within studies and across studies, which further emphasizes the complexity of implementing PA and the importance of attention to individuality and context when designing interventions. Nearly all the included studies exclusively examined company employees’ perspectives on barriers and facilitators for implementation of workplace-based PA. Since successful implementation processes of health promoting interventions seem to rely on leadership support and a collective contribution [16, 20, 54], research of factors influencing other stakeholders, e.g., management or intervention deliverers could add to the understanding of success or failure of implementation efforts.

This scoping review was designed and reported in line with the recommendations of the PRISMA-ScR statement [26]. We searched multiple databases, and a thorough search strategy was designed iteratively by the research team, and an information specialist, to account for the three different dimensions of the search (implementation, physical activity, and workplaces). All aspects of data collection, data extraction, and data analysis were carried out independently by two researchers, with a third party available for mediation in case of disagreements. We applied TDF to provide a structured and systematic foundation on which implementers may base the design of future interventions [62]. The application of TDF provides a theoretical foundation for working with implementation and factors affecting uptake, and gives a comprehensive investigation of potential factors relating to implementational difficulties [23]. The primary limitation of this scoping review is the sparse literature related to our objectives. Due to the limited number and type of study designs of the original studies, and the purpose of our review, we were not able to elucidate any relationship between factors that impact uptake of physical activity and the population or the content of the intervention.

Additionally, our search was limited to studies published from 2009 and onwards, and in English, Danish, Swedish, or Norwegian. The restrictions in language could be taken as a limitation, although there is some indication that this has only marginal impact on results [63], and grey literature was not included. We did not assess the quality of the included studies, as this is usually not part of a scoping review, due to an effort to keep a wide perspective and include studies with different methods and designs [29].

We have found many different factors affecting the success of implementation of workplace-based PA interventions, which highlight the need for well-planned implementation processes considering multiple levels and factors. In workplace-based PA studies, only few have undertaken process evaluations and even fewer base these on a theoretical framework [64]. This is a well-known challenge in implementation science, and studies addressing these issues, have been sought by researchers in the fields of implementation and health promotion [21, 22, 64]. Future research should focus on the application of theoretically informed process evaluations including all stakeholders, to provide standardised information on successful and unsuccessful implementation methods. We recommend implementers of workplace-based PA to apply a systematic approach, which ensures consideration of all influencing factors. Incorporation of a thorough assessment of needs and available resources and involvement of all participating parties would likely assist the process positively. Using implementation frameworks, like TDF, the linking of theories of behaviour change to behaviour change techniques and approaches becomes more tangible and relevant [30], thus enabling a better connection between intervention functions and behaviour change techniques to further guide intervention design [65].