Factors for adherence to a physical activity promotion program in the workplace: a systematic review

Abstract

Introduction

The health benefits of physical activity (PA) are now widely accepted and proven. Promoting PA in the workplace is therefore of major public health interest, but is limited by employees' adherence.

Method

A systematic review was therefore carried out to identify factors for adherence to PA promotion program in the workplace (primary outcome); health outcomes were to be regarded as secondary outcomes. Four databases, PubMed, Web of Science, Cochrane Central Register of Controlled Trials (Cochrane CRCTs) and PsycInfo were searched to find all pertinent articles published from 2000 until June 2024. Only randomized controlled trials (RCTs) and clinical trials were selected.

Results

More than 9000 publications were analyzed and 91 were retrieved. Two main types of study were identified: 46 non-supervised PA programs (NSPAPs) supported by socio-cognitive theories, and 45 supervised (tailored) programs (SPAPs). Concerning the primary outcome, the main factors identified for adherence were the levels of baseline PA, health and motivation of the individual; intervention individualization at the interventional level; and work environment quality at the organizational level. This review highlighted significant health benefits in both types of study, with effect sizes ranging from small to large.

Discussion

Assessing these factors for adherence emerges as an essential prerequisite before implementing a PA promotion program in the workplace. According to our results, implementing NSPAPs, supported by socio-cognitive theories, is rather complex, and such programs can be difficult to operationalize in their entirety; consequently, coach-supervised PAPs based on RCT programs tend to be more effective.

Conclusion

Our results prove the short and medium-term beneficial effect on health of PAP in the workplace based on rigorous methodology such as RCTs. Management's support through work organization and the follow-up of actions in the long term are an essential factor for adherence to these programs. Finally, we suggest what this literature review contributes for future research or entrepreneurial and/or political projects. In fine, new models of working time will have to be considered.

Introduction

Physical activity is defined as all body movements produced by the activation of skeletal muscles and resulting in an energy expenditure greater than the resting metabolism [1, 2]. It provides a large number of health benefits to individuals: reduction in chronic non-communicable diseases such as coronary heart disease, stroke, diabetes, certain cancers, depression, anxiety, or obesity [3,4,5]; primary prevention of more than 35 chronic diseases [6]; and first-line treatment for more than 24 chronic non-communicable diseases [7]. These benefits are now widely accepted, and are found in all individuals, regardless of their gender, age or comorbidities [8,9,10], even for modest physical activity levels [11]. Despite the WHO's guidelines on physical activity [12], it is clear that the physical activity level of most individuals is insufficient [13, 14]. In France, fewer than 50% of individuals aged 15 to 78 can be considered sufficiently active, with three quarters of the population failing to reach the 10,000 daily step target; indeed, while 36 million people stated that they had engaged in at least one physical or sports activity in the year, only 14 million of them practiced activity more than once a week [15]. Physical inactivity (not reaching the recommended physical activity levels [16]) results in a significant health decline in populations, through the increase in a number of chronic non-communicable diseases [17,18,19,20,21]. It may well even be the cause of one in six all-cause deaths worldwide [22]. There is also an increase in sedentary behaviors, defined as a state of wakefulness characterized by an energy expenditure < 1.5 times the resting energy expenditure (1.5 MET) [23], with deleterious effects on health different from those due to physical inactivity [24, 25], and present regardless of the physical activity level of the individual [26,27,28]. The harmful effects of sedentary behavior are very difficult to overcome through the practice of physical activity: indeed, > 60 min of moderate-to-high intensity physical activity per day are needed to overcome > 8 h of seated position per day [29]. In France, people in their forties spend an average of 12 h a day in a seated position during working days, and about nine hours during non-working days, with a total of more than 70% of the population spending more than eight hours a day seated [30].

The decrease in physical activity level and the concomitant increase in sedentary behaviors have been rising in developed countries for several decades. The forecasts of a 2012 study suggested that the average physical activity level in the United Kingdom would reach 153 MET hours per week by 2020, and 140 MET hours per week by 2030 [31]. The significant work changes observed since the industrial revolution appear to be one of the major causes of this phenomenon. An increase by > 20% in sedentary occupations was observed in the United States between 1960 and 2008, with a concomitant drop in physically active occupations [32]. The same is true in France, where working adults now spend more than ten hours a day seated during the working day and 7.58 h during non-working days with a concomitant increase in screen time [33]. Moreover, the association between sedentary behaviors and the type of work (office work, manual work, work in a care setting, etc.) is clearly demonstrated, with office workers spending 66% of their waking time in a seated position, compared to 59% for all workers, and barely 47% for manual workers [34]. The health of workers is thus impaired, with more cases of obesity, absenteeism, depression and anxiety among the most sedentary workers [24, 35,36,37,38].

The workplace has thus become the main target of health promotion policies [5], and even appears to be the ideal place to fight against sedentary behaviors and promote physical activity. Indeed, due to the time spent by employees in the workplace and the possibility of involving the employer and the social network of the company in the process, the chances of success of programs promoting physical activity and fighting sedentary behavior are enhanced [39]. Thus, many barriers are overcome, in particular the lack of time and proximity [40], with lack of time being the first limitation to the implementation of regular physical activity reported by employees [41]. The employer also becomes a major lever, whether at the material, organizational or financial level [42, 43]. Finally, the benefits for the company are well documented, with an increase in productivity [43, 44] and a drop in absenteeism [22] and health costs [45]. Finally, the benefits for the companies are often questioned, with a potential increase in productivity [43, 44], a drop in absenteeism [22], an impact on healthcare costs [45], provided that the programs are based on a high-quality methodology.

Physical activity promotion programs in the workplace have therefore been widely developed in recent years. Through its recommandations, the World Health Organisation encouraged two main categories of physical activity promotion programs [5, 12]: tailored physical activity programs, supervised by a coach (SPAPs) and/or non-supervised physical activity programs (NSPAPs), essentially based on motivational/socio-cognitive theories. According to Väänänen et al. [46], we distinguished three types of NSPAP: (a) informative programs, (b) behavioral programs; and (c) motivational programs supported by socio-cognitive theories (Fig. 1).

Fig. 1

Physical activity promotion programs in the workplace. Based on Global action plan on Physical Activity 2018–2030: more active people for a healthier world [Internet]. World Health Organization; 2018 [notified 3 sept 2021]. 101p. https://apps.who.int/iris/handle/10665/272722.and Väänänen et al. 2022. Workplace physical activity practices in real-life: A scoping review of grey literature for small- and medium-sized enterprises. Eur J Public Health 32:1, i22–i27

(a) Informative programs deliver messages to employees about the benefits of physical activity through working groups, seminars, posters and newsletters, and have been shown to be effective in boosting physical activity levels and reducing sedentary behaviors [47,48,49].

(b) Behavioral programs consist in the development of behavioral strategies by employees, allowing them to include more physical activity in their daily lives while decreasing their sedentary behaviors. This can be done by the progressive implementation of walking sessions in the work routine (at lunch break, while going to a meeting, during a phone call…) [50], through a change in transportation mode [51,52,53,54], or through the implementation of active breaks (short interruptions of the usual work tasks to practice simple physical exercises of moderate intensity, for example via an application sending messages at regular intervals encouraging the user to leave the seated position) [55,56,57]. The effectiveness of such programs is mainly seen in the increase in cumulative physical activity level during the day, with a lesser effect on sedentary time, even though it is known that for an equivalent sedentary time, the interruption of long sedentary periods is beneficial for health [58,59,60]. Effectiveness may be improved by individualized counseling and support provided during individual interviews, helping employees when choosing physical activity, setting progressively increasing goals, or monitoring their activity level [61,62,63].

(c) Motivational programs include a motivational component, for example by using a pedometer to monitor physical activity in real time [64, 65], or by setting up challenges among colleagues [66,67,68]. As lack of motivation is the second most important barrier to physical activity reported by employees [41], it is easy to understand the importance of this motivational component. In some programs, motivation has been targeted through symbolic, material or even financial rewards [69,70,71], with significant effectiveness.

The development of (b)/(c) components may be supported by theoretical models, such as the transtheoretical model of health behavior change [72], social cognitive theory [73], self-determination theory [74], goal-setting theory [75], theory of planned behavior [76], learning theories [77], self-reguation theory [78], behavior change wheel [79], or motivational interviewing [80, 81]. The reliability of these models has been discussed with divergent results, although the interest of such models has been repeatedly emphasized [82,83,84,85].

Supervised Physical Activity Programs (SPAP) may differ in many types of exercises: a) specific muscle strengthening exercises, probably effective on the prevalence of musculoskeletal disorders; b) aerobic exercises, such as endurance, that could also be beneficial at the musculoskeletal, cardiovascular and metabolic levels [85,86,87,88].

Yet the abundance of physical activity programs in the workplace should not obscure their limitations. For instance, the impact on sedentary behaviors seems limited; a Cochrane meta-analysis (Shrestha N., 2018) finds a reduction in seated time of less than 30–60 mn per day [89], whereas the mean sedentary time may exceed ten hours per day [33]. Another Cochrane meta-analysis (Parry SP., 2019) found no impact for reducing musculosketal disorders [90]. On the other hand, the above-mentioned physical activity programs rarely include more than one hour of additional cumulative physical activity per week, whereas a minimum of 60 min of moderate-intensity daily activity (brisk walking at 5–6 km/h, or cycling for pleasure at 16 km, for instance) is needed to significantly counteract the effects of sedentary behaviors [29].

Meta-analyses show contradictory results, with certain interventions demonstrating efficacy on shoulder pain, but no effect on lower back or upper limb pain [91]; certain interventions demonstrate no effect on metabolic risk factors, and only a small effect on the prevention of mental health disorders, but strong evidence was found for the prevention of musculoskeletal disorders through workplace interventions, especially resistance exercise training [92].

Secondly, numerous publications have shown that physical activity promotion programs in the workplace have failed to target the employees who needed them most, i.e., the most sedentary, least active, and with the worst health indicators; and that most participating employees were already physically active and healthy [93,94,95,96,97].

Thirdly, the adherence of employees appears to be the main limiting factor for such programs. The adherence rate is often ≤ 50% [98], both in terms of participation in physical activity sessions as part of physical activity programs [99,100,101,102] and the use of tools provided as part of NSPAPs [103, 104]. It should also be noted that this adherence rate decreases during the course of the interventions, thus highlighting the progressive disinterest of employees, despite a fairly satisfactory initial participation [99, 102].

Only a few research studies have attempted to analyze this low adherence, and even fewer have identified its causes. As stressed by Genin PM. et al., most studies, rather than trying to understand the determinants of the employees'low adherence, prefer to use different incentive methods to maintain an acceptable participation level (e.g., financial incentives), without real success [105]. It is therefore necessary, before considering the development of any new physical activity promotion program, to try to understand the factors promoting or preventing the adherence of employees.

Research hypothesis

• H1: Factors influencing an exercise program at work depend on:

  • 1. Individual factors

  • 2. Intervention factors related to the type of exercise program

  • 3. Organizational factors

• H2: Physical activity programs can have a beneficial effect on health.

Aim of the study

A systematic review was therefore carried out to:

1: identify factors for adherence to a physical activity promotion program in the workplace (primary outcome);

2: Identify the health effects based on the type of program, and specify, if possible, the effect size (secondary outcome).

So the study aims to provide an overview of the current state of knowledge, help entrepreneurial and/or political new projects, establish a framework for new research.

Method

Selection criteria

We aimed to include all relevant original research studies with a quantitative design.

Filters were applied. The first filter was the publication date, which had to be comprised among 2000 and june 2024. The second filter was the type of study conducted: clinical trial and/or randomized controlled trial. Studies had to be written in English, internationally published, and peer-reviewed. We did not include studies with purely qualitative design, studies only reporting descriptive statistics, dissertations, book chapters, or theoretical work such as editorials, short communications, or conference abstracts.

Search strategy and study selection

The Institute of Medicine guidelines for completing systematic reviews were used and reporting of findings followed PRISMA guidelines [106, 107]. Electronic databases searched for studies published between 2000 through June 2024 included PubMed, Web of Science, PsycINFO, and Cochrane CRCTs. All aspects of the review process were completed by two researchers (DT, QG). Titles and abstracts of all articles were screened for applicability. If the article appeared to meet the review’s inclusion criteria, the full article was reviewed and assessed to ensure that it met the criteria for inclusion. All discrepancies were reviewed by the same two researchers and consensus was met regarding the eligibility of the study. For each article included in this review, all references were reviewed for inclusion criteria. The keywords included the two main concepts: physical activity and workplace.

The algorithm and selected MeSH Terms included in the database were: physical activity intervention [MeSH Terms]) OR physical activity program OR physical exercise [MeSH Terms] OR training [MeSH Terms] OR resistance training [MeSH Terms] OR endurance training [MeSH Terms] AND workplace [MeSH Terms] OR worksite [MeSH Terms] OR employees [MeSH Terms] OR workers [MeSH Terms] AND compliance [MeSH Terms].

This query allowed the retrieval of the maximum number of scientific publications dealing with both physical activity and the workplace. Filters were applied. The first filter was the publication date, which had to be comprised between 2000 and 2024. The second filter was the type of study conducted: clinical trial and/or randomized controlled trial. We found no meta-analysis concerning our primary outcome for this review.

In addition to the references obtained through this search, references of interest cited by or citing the retrieved publications, or presented as similar to them were also analyzed.

Inclusion and exclusion criteria

The inclusion and exclusion criteria for the publications followed the PICO method in accordance with Richardson WS [108], and were therefore based on the study population, the intervention type, the type of assessment performed by the authors, and the selected endpoint. They also took into account the analysis of adherence and factors for adherence performed by the authors. Further, we included populations consisting of employed workers (public and private sector). We excluded studies of self-employed and/or students.

Manuscript selection

DT and QG analyzed the selected papers independently in the screening and eligibility steps to compare the selected/not selected articles; in the case of disagreement over the quality of the article or the lack of precision in the inclusion criteria, the articles were excluded (Fig. 2). The publications retrieved were sorted by year. This allowed the grouping of publications on the same study, and the rapid identification of papers published by the same research team. Firstly, the titles were analyzed to exclude all publications that were clearly unrelated to physical activity or the workplace. The abstracts of the publications were all then read, to exclude further publications. When it was not possible to exclude a publication based on its title and abstract, the full text was read to search for exclusion criteria, which included a lack of information about the design of the study, the number of participants, the type of work environment, the study's duration, and/or the theoretical model used to develop the intervention. The publication was included in the final analysis only after the full text was read (Fig. 2).

Fig. 2

Article selection flowchart. From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. https://doiorg.publicaciones.saludcastillayleon.es/10.1136/bmj.n71

When reviewing the different publications, CG and DT grouped the references related to the same study, and analyzed them together. For each publication selected, detailed criteria of inclusion were necessary: a) study design, b) number of participants, b) type of work environment, c) study duration, d) the theoretical model used to develop the intervention, e) description of the intervention. These data were compiled on two tables according to the type of intervention: non-supervised physical activity program (NSPAP) (Table 1) or supervised physical activity program (SPAP) (Table 2). For NSPAPs, we specified whether the intervention was based on a sociocognitive theory and its model (such as TTM, SCT…).

Table 1 Non-Supervised Physical Activity Programs (NSPAP) in the workplace

Table 2 Supervised physical activity programs with coaching sessions (SPAP) in the workplace

We also identified the tools and means used: individualized adjustment of the intervention content, use of pedometers, individual counseling and financial incentives in particular. For SPAP programs, we specified whether the exercise sessions took place in the workplace, whether they were supervised by a sports educator, and whether their content was adapted to an individual level. Then, for each of the publications, we indicated the reported adherence, either in terms of participation in the different components of the intervention (e.g., website access, participation in motivational interviews, participation in physical activity sessions), or in terms of changes in physical activity (e.g., changes in the number of daily steps, changes in sedentary behaviors, increase in physical activity level).

We then identified the different factors for adherence reported by the authors, classifying them into three categories: individual (employee-specific), interventional (physical activity promotion program-specific) and organizational (workplace-specific) factors. This classification has already been used and validated elsewhere [109]. For each of these factors, the type of association with adherence was specified (positive, negative, neutral).

Risk of bias and quality of evidence

In order to estimate the Quality Index [QI]: [Min = 0, Max = 5], we assessed and reported the methodological risk of bias of the included studies in accordance with the Cochrane Handbook (Higgins, 2011) and the guidelines of the Cochrane Consumers and Communication Review Group (Ryan, 2013), which recommend the explicit reporting of the following individual elements for randomized controlled trials: random sequence generation; allocation sequence concealment; blinding (participants, personnel); blinding (outcome assessment); completeness of outcome data; selective outcome reporting. We assigned the following scores according to the number of limitations described: one major limitation: QI = 4; two limitations: QI = 3; three limitations: QI = 2, and so on. Within Supplementary Files 1 and 2, we added columns describing the results (adherence factors, outcomes and effect size), and a column describing a quality index [QI] taking into account the studies’ limitations.

Results

The total number of records retrieved was 9619. Finally, we included 91 publications that met the inclusion and exclusion criteria (Fig. 2). The systematic review allowed for the identification of two main types of studies. There were 46 publications on non-supervised physical activity programs (NSPAP) and 45 on supervised physical activity programs (SPAP). The features of each publication are presented in Tables 1 and 2.

Results for employees’ adherence, outcomes, potential limitations, study quality and effect sizes (when they are indicated on the manuscript) are reported in Supplementary Files 1–2. Studies with significant outcomes results are described in Tables 3, 4 and 5, Figs. 3, and 4.

Table 3 NSPAP RCT program results: high quality RCT with significant effect size results

Table 4 SPAP RCT Program Results: high-quality RCT with significant effect size results

Fig. 3

Forest Plot illustrating the distribution of Cohen’s d effect sizes for SPAP and NSPAP RCT

Fig. 4

Funnel plot illustrating the distribution of Cohen’ s d effect sizes and standard error for SPAP and NSPAP group

Physical activity program interventions description

Forty-six randomized controlled trials (RCTs) assessed an NSPAP implemented in the workplace

All these interventions were comparative studies, except for three that were longitudinal [110,111,112]. Most of these interventions included more than 100 participants, and five [113,114,115,116,117] included over 1,000 individuals. In most cases, the participants were office workers. Three included exclusively manual workers [67, 118, 119] and four interventions were conducted in a care setting [69, 120,121,122].

There were also studies conducted in any type of work environment setting, usually with a multicenter enrollment and a large number of participants [110, 113,114,115, 117, 118, 123,124,125,126,127,128,129,130,131,132,133]. Other studies included a small number of subjects, fifty or less [134,135,136,137].

Regarding intervention duration, only a few interventions lasted over one year. The studies by Aittasalo M. et al. [110], Plotnikoff RC et al. [125] Robroek SJW et al. [138] and Hallam JS et al. [139], lasted one year. The remaining interventions lasted from a few weeks to a few months, with a total of 75% of interventions lasting ≤ 6 months.

Twelve interventions included an informative program systematically combined with a behavioral or motivational component, except for one intervention [110]. In about half of the cases, the informative program was clearly separate from the behavioral/motivational programs, while other interventions mixed the different components, for example individual interviews [111, 124] or guides/booklets provided to employees [69, 123, 125, 130, 134, 140, 141].

Seventeen interventions included a behavioral program, which mainly consisted of PA objectives that increased progressively over the course of the intervention, combined with behavioral strategies to achieve them more easily. The goals were defined and their evolution was assessed on an individual basis during the intervention, in particular during individual interviews with “facilitators” [111, 115, 118, 119, 125, 140, 142].

Sixteen interventions included an incentive component, mostly in the form of self-monitoring of PA using a pedometer. Pedometer data could be used to adjust the behavioral component of the program [67, 71, 110,111,112,113,114, 118,119,120,121, 126,127,128, 132, 134, 143, 144], to set up challenges [112,113,114, 119, 120], or to obtain material and financial rewards [69, 71, 113, 114, 126,127,128, 143].

Thirty seven interventions (81%) were based on a theoretical model

The most frequently used theoretical model is the social cognitive theory (SCT), nine interventions [69, 129, 130, 134, 139, 144,145,146,147]; six interventions were based on the transtheoretical model of health behavior change (TTM), [119, 123,124,125, 141, 148]; two on SCT/TTM combined [116, 140]; six studies were based on the learning theory [71, 121, 126, 127, 143, 149], seven studies on the self-determination continuum theory [118, 120, 121, 132, 142, 150, 151] two studies on the goal-setting theory [113, 114], two studies on the theory of planned behavior [114, 132], and one study combined SCT, TTM and theory of planned behavior [116]. One intervention was based on the behavior change wheel [152], one intervention was based on the socio-interactionist theory [131], and one intervention was based on the motivational interviewing theory [153].

For three interventions, the theoretical model was not used to design the intervention, but only to assess some of the endpoints [121, 126, 134].

Forty-five RCTs assessed an SPAP (tailored program) implemented in the workplace

All these studies were comparative, except for two that were longitudinal [135, 154]. Thirty-five studies compared an intervention group (IG) to a control group (CG), receiving either no intervention or a minimal intervention (e.g., a single message encouraging participants to participate in sufficient PA), while six studies compared different PA programs with each other [106, 155,156,157,158,159]. Three studies were nested studies from RCTs [109, 160, 161]; one study was a nested study from a RCT with a qualitative approach [109].

Twenty-one studies included fewer than 100 employees, five included more than 500 employees [154,155,156, 162, 163], and the others included an intermediate population of a few hundred participants.

Twenty-six studies were conducted in office worker populations, one in office care and manual workers [154], nine in manual worker populations such as laboratory technicians [162,163,164], women cleaners [157, 165, 166], industrial sectors [164, 165, 167], seven in a healthcare setting[136, 161, 168,169,170,171,172], others on military and/or civilian staff [173,174,175], university employees [88, 135, 137], university home and work place training [176], casino employees [177], and sitting workers [178].

Eighty-three per cent of the interventions lasted ≤ 6 months, and five interventions lasted ≥ 1 year [154, 155, 160, 179, 180].

With two exceptions [109, 176], all the proposed interventions consisted of PA sessions in the workplace, during the working time (or counted as working time), and were supervised by a professional sports educator. They mainly differed in their organizational mode:

All were strictly organized (training volume, intensity, type of exercises, session timing and group composition were determined by the experimental protocol and assigned to the participants), one had a flexible organization [109] that left the employees freedom regarding most of the protocol's components (choice of exercises, session timing, training volume and intensity, etc.), whilst the others were characterized by an intermediate organization (freedom for certain components, constraints for others).

Two intervention programs [181, 182] were based on two socio-cognitive theories (TTM and SCT). One intervention program [157] was based on cognitive behavioral theory.

Two interventions combined an incentive component with PA sessions, such as material rewards [177] or the use of a pedometer [135].

Adherence analysis

Adherence was assessed differently across studies, with most NSPAPs quantifying adherence in terms of changes in PA, such as changes in the number of daily steps during the intervention [109, 116, 119, 128, 141, 169, 183, 184], whereas SPAP programs assessed adherence in terms of participation in sessions, volume of exercices performed, anthropometric/cardiorespiratory parameters, blood pressure, plasma biomarkers or ergonomic office layout [135, 163, 175, 181, 182, 185].

The impact of the baseline PA level on adherence was assessed in fourteen publications (eight assessing a NSPAP, six assessing a SPAP), with a positive correlation in nine publications: five assessing an NSPAP [69, 112, 121, 129, 141] and four assessing an SPAP [106, 135, 167, 186]. PA level at baseline was not significantly correlated with adherence in three NSPAP studies [110, 111, 113] and two SPAP studies [154, 177].

Eleven studies analyzed the relationship between the baseline health level and adherence (four assessing an NSPAP and six assessing an SPAP).

Six publications found a positive correlation: four concerning NSPAPs [115, 124, 129, 130] and two concerning SPAPs [106, 165]. Five publications found no significant correlation: three assessing an NSPAP [110, 112, 126], and two assessing an SPAP [158, 177]. One study found a negative correlation due to musculoskeletal disorders at baseline [158].

The association between adherence and the change in employees'health status during the intervention was also analyzed in ten studies. One study assessed an NSPAP and showed a positive correlation [146]. Eight studies assessed SPAPs, and six found a positive correlation [105, 109, 137, 160, 161, 173] while two found no significant correlation [155, 158].

The impact of the baseline Self-efficacy on participants'adherence was assessed in nine studies (seven studies on NSPAPs and two on SPAPs). Self-efficacy at baseline was found to be a positive factor for adherence in three studies (NSPAP: [127, 129]; SPAP: [156]), while one study (SPAP: [163]) found no significant correlation.

Five studies (NSPAP: [114, 116, 118, 139, 142]) highlighted a positive correlation between increased self-efficacy and adherence to the NSPAP, while one study (NSPAP: [128]) found no significant correlation between increased self-efficacy and the employees'adherence.

One study showed that external motivation (“I feel under pressure from my family/friends to exercise”) and identified motivation (“I value the benefits of exercise”) increased PA level until week three [121].

Based on TTM or SCT theory, two studies showed that being at a stage of change, as close as possible to “maintenance stage”, correlated with good adherence [119, 140], while another study showed that “having an inactive/precontemplation status” at baseline increased PA level [141].

The effect on adherence of the intervention's behavioral components was assessed in five NSPAP and two SPAP programs. In three NSPAPs [118, 124, 144], the presence of a behavioral program was associated with increased adherence; the other studies found no significant correlation.

The impact on adherence of a motivational program was assessed in four studies (on NSPAPs). One study [134] assessed the association between self-monitoring of PA using pedometers and adherence and found a positive correlation. Two studies assessed the association between adherence and financial rewards in the long-term follow-up [128, 143] and found no significant correlation. One study [120] assessed the effect of challenges on adherence and found no significant correlation.

Four publications (NSPAP programs) assessed the effect of adapting the intervention according to the principles of theoretical models. One study (SCT model) found significantly higher adherence among employees who participated in a program that was adapted to closely follow social cognitive theory [144], while another study (TTM model) highlighted that the stage-matched intervention outperformed the active control condition for physical activity, nutrition and psychological variables (physical activity intention, nutrition intention, nutrition planning) [123]. Two studies, however, found no significant correlation between the adaptation of the program to the employees'stage of change (transtheoretical model) and adherence [125, 140].

Four SPAP programs [136, 137, 156, 159] assessed the impact of session supervision on the employees'adherence and found no significant correlation. Yet it is interesting to observe that “minimal exercise supervision” could have a significant effect on health during a 20-week intervention [159].

Two studies found a positive correlation between strict session organization and the employees'adherence [155, 168], whereas one study highlighted that unexpected events in the program (absence of the instructor, change in instructor) led to negative correlation [157].

Several studies found a positive correlation between the flexibility of session organization and the employees'adherence [109, 130, 150, 152, 157, 187].

Twenty-three studies included an assessment of the relationship between occupational organizational factors and the employees'adherence

Eight studies on NSPAPs [113, 117, 122, 127, 130, 131, 142, 146] and eight on SPAPs [109, 154, 155, 157, 168, 169, 177, 188], assessed the impact of the working environment, and found a positive correlation. Workload was found to be a negative factor for adherence (lesser adherence with greater workload) in two studies (SPAP: [109], NSPAP: [130]).

Comparative analysis of the systematic review results

The analyzed studies fall into two categories: supervised physical activity programs (SPAP) and non-supervised physical activity programs (NSPAP). SPAPs involve direct supervision by a coach or an organizational structure, whereas NSPAPs rely on autonomous interventions such as motivational messages or tracking tools. Supervised and unsupervised studies show results of high variability in terms of adherence and compliance (Tables 3, and 4). In the latter, adherence varies based on initial motivation and organizational support, while individualized adjustments can improve effectiveness despite implementation complexity (Hallam JS, 2004; Opdenacker J, 2008; Dishman RK, 2009; Andersen CH, 2012).

The factors influencing adherence to physical activity programs fall into three categories. Individual factors include the initial level of physical activity (Macniven R, 2015; Losina E, 2017; Genin PM, 2018; Corbett DB, 2018; Brunet J, 2020), motivation, and self-efficacy (Hallam JS, 2004; Kaewthummanukul T, 2006; Dishman RK, 2010), as well as the use of behavioral tracking tools such as pedometers (Finkelstein EA, 2016; Murrray JM, 2019 and 2020). Interventional factors show that programs with regular support and progressive goals promote better adherence (Dishman RK, 2010; Raedeke TD, 2017). Finally, organizational factors play a crucial role: management involvement, work schedule adjustments significantly increase participation (Gazmararian JA, 2013; Bale JM, 2015).

The measured effects of interventions include physical health, psychological and social well-being, and organizational impact. Physical activity increases by an average of 10–20% in MET-minutes per week, and sedentary behavior decreases (Plotnikoff RC, 2005; Opdenacker J, 2008). Cardiovascular benefits are observed, such as a 5% reduction in BMI and a 10 mmHg decrease in blood pressure (Proper KI, 2003; Mainsbridge CP, 2015). Psychologically, stress decreases by 20%, while professional engagement increases by 15% (McEachan RRC, 2011; Jakobsen MD, 2015). Finally, organizational effects include a 10% reduction in absenteeism and a 5–10% perceived increase in productivity (Nurminen E, 2002; Justesen JB, 2017; Welsh A, 2020; Stenner HT, 2020).

SPAP programs show high-effect-size (Hunter JR. (2018), Brandt T. (2024), Eather N. (2020)). NSPAP interventions are more variable—some programs show large effects (e.g., step count, VO2 max), but many are low to moderate in impact. η² values for self-regulation (η² = 0.64) and MAP reduction (η² = 0.67) in NSPAP studies suggest that behavioral-focused interventions can be effective without supervision (Table 5).

Table 5 Key differences between SPAP and NSPAP based on effect sizes (Cohen’s d & η²)

SPAP leads to better physiological Improvements, while NSPAP enhances behavioral outcomes. SPAP programs show greater improvements in strength, mobility, and cardiovascular fitness. NSPAP programs with high adherence (e.g., Hallam JS. 2004) show significant behavioral changes but may lack the physiological benefits of structured, supervised training. Large η² values in self-efficacy (η² = 0.06) and motivation (η² = 0.022) in NSPAP suggest stronger psychological engagement, but not necessarily lasting physical changes.

NSPAP (Non-Supervised Programs) show greater variability but includes some of the largest effect sizes observed (Mansi S. (2015): Step count (d = 1.94), self-reported PA (d = 2.57) Metcalfe RS. (2020): VO2 max (d = 1.4), Samuels TY. (2011): 10,000-step goal (d = 1.1). Hallam JS (2004) and Mainsbridge CP (2015) reported η² > 0.5, indicating very large effects on exercise adherence and health indicators. Hunter RF (2018), η² = 0.06, indicate a moderate effect on physical activity maintenance. Murray JM (2019), η² = 0.54, indicating a large effect related to engagement with intervention components (Table 5).

Studies like Hunter RF. (2018) reported R² values of 0.56—0.76, indicating a strong intervention model, even though Cohen’s d effect sizes were small. Brunet J. (2020) (η² = 0.022) and Krebs S. (2019) confirm that low-magnitude effects exist even in well-structured interventions (η² = 0.006). Dropout rates remain high in NSPAP, suggesting that long-term behavioral change without supervision is more challenging (Table 5).

Conclusion and limitations: supervised (SPAP) vs. non-supervised (NSPAP) RCTs

SPAP interventions produce more consistent, higher effect sizes (both d and η²), particularly for physiological health outcomes. The structured supervision ensures better adherence but does not always lead to the highest effect sizes (Figs. 3–4).

NSPAP interventions can be effective in behavioral change and self-regulation but are highly variable and often lack long-term adherence (Figs. 3–4).

Eta Squared (η²) data confirms that some NSPAP programs (e.g., Hallam JS. 2004, Mainsbridge CP. 2015) achieve strong results, particularly in self-efficacy and adherence.

R² and SRMR suggest some NSPAP models are well-structured but experience high dropout rates.

SPAP confirmed physical health improvements, while NSPAP interventions show higher variability but can produce very large effect sizes when engagement is strong (e.g., Hallam, Mainsbridge, Murray). However, they also risk lower long-term adherence due to the lack of supervision.

The main limitations of the analyzed studies include, the lack of long-term follow-up, methodological heterogeneity, wide variability on results, and the lack of exploration of cultural or sectoral differences (Taylor, 2016; Losina, 2017; Fernandez La Puente de Battre, 2020; Metcalfe, 2020; Eather, 2020; Perez-Bilbao, 2021). To improve future research, it would be recommended to integrate additional indicators of motivation and perceived barriers, deepen the analysis of organizational impact, and compare program effects across different professional contexts (Bredahl, 2015; Krebs, 2019; Murray, 2020; Stenner, 2020; Welsh, 2020; Higham, 2023).

Discussion

The objectives we set for this study have been achieved, namely that:

Firstly, our review confirmed two main types of interventions/programs/RCT: supervised/tailored PA programs and non-supervised PA programs. Supervised studies tend to show better results. This is probably explained by the greater complexity of the implementation of the programs based on different socio-cognitive theories. Moreover, we have excluded from these tables studies with very low participation rates [172, 189] or studies showing a non-statistically significant trend towards improved health or aerobic capacity [150].

Secondly, based on our literature review, the main factors for adherence emerge as baseline PA level, health level, individual self-efficacy, the use of a behavioral component consisting of individual counseling, the individual adaptability of the program at the intervention level, the workload, and the quality of the working environment at the organizational level: these key results are summarized in Table 6. 

Thirdly, our review highlighted two parameters used in a differentiated or undifferentiated way in RCTs: adherence and compliance. We therefore considered that adherence was the commitment of employees to participate in sports activities (with or without incentive methods) within or outside the company. Compliance refers to the number of subjects still present in the study at the end of the intervention program, compared to the number of subjects included in the study after randomization. We therefore chose to consider these two parameters in our result tables. 

Table 6 What this literature review contributes, and what this literature review has to offer for future research, or entrepreneurial and/or political projects

Individual factors

Baseline PA level was found to be a positive factor for adherence in most publications included in this review. This finding is in line with the literature: employees who participate the most in PA programs are those who already have a good PA level [93,94,95]. However, this finding can be balanced, especially with respect to PA programs. Indeed, the studies that found a positive correlation between baseline PA level and adherence assessed PA programs consisting of high-intensity exercises from the beginning of the intervention, with little or no individualized support. Conversely, a study that found a significant effect on physical health assessed programs consisting of moderate-intensity exercises, such as walking sessions, that were accessible to most people [154]. Thus, while it clearly appeared that PA promotion programs were more likely to be adopted by employees who were already physically active, it was possible to achieve the same adherence in inactive employees provided that the programs were well adapted; indeed, inactive employees were more inclined to participate in workplace.

PA programs that did not require taking part in a specific class or attendance at a gym and which involved accessible activities like walking.

The baseline health level (before the intervention) was also identified as a positive factor for adherence. This is consistent with several previous studies that have found that employees participating in PA promotion programs were healthier than those who did not [93, 94, 96, 197, 198]. However, some interventions included in this review showed conflicting results, including programs specifically targeting distinct health problems such as musculoskeletal disorders [158, 162] and programs with individual counseling and support sessions [111, 177]. In both cases, it is likely that employees with the poorest health status were more aware of the benefits of participating in such programs, either because they felt directly concerned or because the intervention content could be adapted to them [199]. In the literature, we noted that the interventions that successfully targeted employees with the poorest health status [119, 157, 200] were characterized by adapted communication before the beginning of the intervention, highlighting the link between physical inactivity and work-related health problems as well as the expected benefits of such an intervention.

Improvement in health indicators during the intervention was also significantly associated with adherence. Indeed, it may be considered that good adherence increases the effectiveness of the intervention and thus enhances improvement in health indicators related to the practice of PA (body mass index, VO₂, resting heart rate, and so on). It may also be considered a phenomenon of positive reinforcement: a significant and early improvement in certain health indicators encourages the employee to continue participating in the intervention, for example by increasing comfort when practicing the exercises (improved exertion tolerance, reduced musculoskeletal disorders). An early improvement in indicators was associated with a higher adherence in the long term after the intervention [105, 186] or even remotely [201]. The improvement in health indicators also increased the benefits of PA perceived by employees, as confirmed by a literature review that observed that the perceived health benefits of PA directly correlated with employees'participation [191].

Self-efficacy appeared to be another leading factor for adherence. Since self-efficacy is the belief that an individual has the ability to perform a task [73], this seems consistent. The importance of self-efficacy in the practice of PA has been shown in several studies, and a literature review has even concluded that self-efficacy was the best predictor for employees'participation in PA in the workplace [191, 202,203,204,205]. There was also a phenomenon of positive reinforcement, with a correlation between increased self-efficacy and adherence. This finding is supported by the social cognitive theory, according to which the main component of self-efficacy is self-control or a control experience, i.e., success in past personal experiences [109]. Thus, successfully initiating PA from the start of the intervention could be a control experience, reinforcing the employees'self-efficacy and thus their motivation to continue the promotion program.

The level of the employees'stage of change, according to the transtheoretical model was not significant, either in terms of the impact of the employees'initial stage of change on their adherence, or in terms of the impact of intervention adaptation based on the stage of change at baseline [206]. The five stages of change correspond to different levels of determination to change a behavior, and are thus considered by several authors as important levers to increase the effectiveness of PA promotion programs in the workplace [207,208,209,210,211]. However, it was observed that the stage of change primarily determined the initiation of a behavior change, i.e., the initial commitment to a PA promotion program in the workplace, rather than maintaining this new behavior, i.e., the adherence to the program [138]. Taking into account the initial stage of change is therefore insufficient, and it is especially important to adapt to the evolution in the stage of change during the promotion program.

Interventional factors

Individualizing the intervention, whether through individual counseling sessions, flexibility in the organization of exercise sessions, or adjustment of session supervision, is associated with better adherence of the participants. This finding was reported by several authors [212, 213]. The other features of the intervention, such as session supervision, location of the intervention or the existence of material rewards, only had a significant impact on employees'adherence if they allowed the intervention to be better adjusted to the individual level.

The use of financial incentives was not identified as a significant factor for adherence. Studies assessing programs based on financial incentives showed a positive influence on the participants'motivation: according to the authors, financial incentives strengthened intrinsic motivation (i.e., motivation based on what individuals consider to be good for them, independently of external views), contrary to what is usually found in the literature [71, 143]. However, the authors also reported that financial incentives did not increase the long-term adherence of employees showing high sensitivity to immediate benefits [143]. The literature also showed that the increase in activity level directly correlated with the period during which the financial incentives were paid, with a sharp decrease as soon as their payment stopped [70].

The use of a pedometer or accelerometer, especially when combined with real-time self-monitoring of PA levels, was associated with higher employee adherence [134]. Van Hoye et al. showed that the feedback provided by the pedometer was most effective in increasing PA when combined with individualized counseling sessions [214]. This individual support helped participants to know what to do with this feedback, and to adapt their behaviors as best as possible.

Organizational factors

An original result of this review is that it highlights the work environment as a major determinant of adherence to physical activity programs. Some studies had already shown the predominant role of social-cognitive factors, support by supervisors, colleagues, friends or family [215, 216].

The quality of the working environment, particularly in terms of interpersonal relationships, work organization and workload, was a significant factor for adherence. Management, psychosocial factors (work autonomy, work pace, workload and interpersonal relationships at work) played a central role in employees'participation in a PA promotion program such as NSPAP [113, 117, 122, 127, 130, 131, 142, 146] or SPAP [109, 154, 155, 157, 168, 169, 177, 188]. Their quality explained the highly variable influence of co-workers and management, found in this review, bringing to the fore a negative influence in the case of impaired psychosocial factors, and a positive influence in the case of high-quality psychosocial factors. Strong social support, particularly in its functional component (when employees feel they are being supported by their professional environment), was thus a powerful factor for adherence. Recent investigations highlighted that a training adherence threshold of 70% has been recommended to attain clinically meaningful benefits and that it is important to implement initiatives before and during an intervention to maintain motivation [183, 189]. A favorable psychosocial environment and strong social support require the employees'involvement in the implementation of PA programs in the workplace, the open involvement of management in promoting PA, and the implementation of a corporate policy that openly supports PA.

Physical work load

According to the results of our analysis, physical workload was associated with poorer employee adherence. The accumulation of physical fatigue owing to work tasks together with participation in PA exercises was a fairly obvious explanation for this phenomenon, as suggested by Korsohj et al., who found that a PA program based on intensive aerobic exercises, implemented in maintenance workers with high workload, was accompanied by a rise in employees'blood pressure [165].

As a conclusion, we suggested below, in Table 6, what this literature review contributes for future research or entrepreneurial and/or political projects.

Limitations of the study

The main limitation of our review is the evaluation of the potential limits of each study. We used the Cochrane guide for randomized controlled trials (RCTs) as a model, but interventional studies in the human and social sciences, although they may allow for randomization, do not always allow for blinding of the participants, or blinding the results throughout the study. For this reason, our quality index ranges from 1 to 4. For real RCTs, the score varies from 2 to 4. For non-RCT studies, the score ranges from 1 to 2, in an"arbitrary"way, because of the lower level of evidence. However, our results remain consistent with the existing literature, as discussed in the synthesis of results.

In addition, although broad categories of intervention were identified, the type of intervention could vary greatly from one study to another. Thus, for the same type of program, the frequency and duration of PA sessions could vary; the frequency of counseling sessions and the type of exercises or the educational material content were rarely the same.

Others have emphasized the need to improve the design of studies aimed at assessing the effectiveness of different PA promotion interventions in the workplace, in particular to be able to assess the relative effectiveness of various programs (PA vs. informative/educational strategy vs. counselling/support strategy according to their classification) [13]. Indeed, this was difficult to do in our review, because the analysis of interventional factors for adherence was challenging due to the broad heterogeneity in study methodology.

The choice of the endpoints, and especially their measurement method, was also highly heterogeneous. This was especially the case for PA, that was, depending on the studies, measured directly by an accelerometer or assessed using a self-questionnaire, measured on an ad hoc or continuous basis, and expressed in different units.

The methodology for data collection could also change from one study to another, compromising the reproducibility of results.

Finally, the aims of the studies were also highly heterogeneous. Some studies aimed at increasing PA level, some at improving the cardiorespiratory condition of the employees, and others at improving well-being at work. This obviously influenced the study protocol, methodology, choice of endpoints, etc. and affected the comparability of the results from one study to another. In addition, only a few interventions were designed to directly assess specific factors for adherence, in this case the baseline PA level. Other studies (NSPAP: [71, 114, 121, 126, 153]; SPAP: [109, 157, 160, 161, 163, 188]) investigated specific factors for adherence, but as part of nested/secondary analyses of interventions designed for a different objective. In all other studies, these assessments were marginal, and the authors paid little attention to the results. Another systematic review [98] has found the same results, highlighting that too few studies have assessed the influence of health, lifestyle and organizational factors on the participation of employees in health promotion programs in the workplace. The scientific value of the determinants of adherence reported by the authors was therefore, according to them, significantly reduced. The same is true in our review, where some of the publications included were not based on a robust and rigorous methodology, limiting the level of evidence of the results.

Perspectives

Currently, PA promotion programs in the workplace are complex, demanding onerous implementation (creation of sports structures, changes in the employees'work schedule, involvement of external stakeholders) requiring a great deal of resources (human, material and financial) for very limited results. The purpose of these programs is often unclear, as they target sedentary behaviors, physical and mental health, well-being in the workplace or even productivity, while a PA program has little impact on any of these factors. Thus, the determinants of sedentary behaviors are inherent to the nature of the work itself that will not be changed by a PA program. The same is true for musculoskeletal disorders, that are mainly caused by the physical constraints of the position (repetitive movements, carrying of loads, prolonged static position, etc.) that remain unchanged after a PA promotion program. Finally, well-being in the workplace is above all determined by the workload, the precariousness of the work, poor psychosocial relationships, and not, or only marginally, by a lack of PA. These enthusiastic but totally inappropriate goals for a PA promotion program in the workplace overlook the inability of these programs to sustainably increase the PA levels of all employees. As already pointed, few studies have attempted to understand the determinants of the employees'adherence to these programs. It is therefore necessary to focus PA promotion programs in the workplace on the employees'adherence, from the outset of the designing of such programs.

To this end, the factors for adherence identified in this review should be assessed prior to the development of any PA promotion program, in order to better tailor it to the different profiles of employees in the company. This could be achieved by administering a standardized questionnaire, which would therefore assess the levels of PA, health and motivation, and more specifically self-efficacy. The employees'expectations regarding the type of program offered should also be assessed. Based on the data from this review, it would be necessary to assess the benefits of exercise supervision, the organization (location, timing, group, etc.) of the sessions, the type of exercises proposed, and the need for individualized support (counseling sessions). Finally, the quality of the working environment, and more specifically the quality of psychosocial factors and social support within the company, should be assessed. Specifically, strengthening empowering leadership for managers, employees, interns, and students [217, 218].

"And to venture further": The establishment of such programs (e.g., SPAPs vs NSPAPs) might also prompt a collective, societal, entrepreneurial, and political reorganization of working hours and/or work schedules. With the rise of AI, should we remain committed to a traditional weekly work model (35 h, 39 h, 40 h, 50 h in certain countries…)? Might we not envision travel times, workdays, or schedules confined to the morning or the afternoon alone? Could we not also imagine, for instance, the possibility of"shifting the hours of students/apprentices/teachers (starting early in the morning) away from those of employees/managers (beginning later in the morning)?"These ideas, we believe, deserve thoughtful debate, as they hold the potential to breathe new life into our so-called industrialized societies, all in pursuit of enhancing the collective well-being.

Conclusion

This literature review showed that a number of factors determine the adherence of employees to PA promotion programs. Until now, these factors for adherence were poorly taken into account in the design of these programs, reducing their effectiveness.

It is now necessary to develop practical tools for occupational health stakeholders so that they can help companies to implement an effective and sustainable PA promotion policy. The primary goal of such a policy should be to increase the PA level of as many employees as possible, since the solution to other health issues in the modern workplace cannot be reduced solely to promoting physical activity. Hybrid models combining supervised and non-supervised elements could help balance adherence with flexibility, with long follow up. Future research should evaluate the sustainability of PA behavior changes beyond 12 months. In fine, new models of work time will have to be considered.