Introduction
Impaired mental health presents a significant global health challenge, placing a substantial burden on affected individuals and resulting in notable costs through sickness absences and reduced productivity [
1]. There is an urgent need to mitigate the burden of impaired mental health, and workplaces can play an important role in mental health promotion since many working-age adults can be reached via the workplace [
1]. Employers in some Western countries are legally obliged to ensure working conditions that prevent not only physical but also mental health impairment [
2].
Physical activity (PA) and sedentary behavior (SED) are important determinants of mental health; numerous studies confirmed positive effects of PA and negative effects of extensive SED on various mental health outcomes, such as mental wellbeing, depression or anxiety symptoms, and stress [
3‐
7]. Considering this evidence, office workers may be at an increased risk of poor mental health since they spend up to 82% of work time sedentary [
8‐
10]. In addition, sedentariness at work is associated with sedentariness during non-work time [
8]. Thus, it is important to investigate whether office workers’ mental health can be improved by supporting them in reducing SED by interrupting long periods of sitting or by replacing sitting with standing or light-intensity physical activity (LIPA), or by supporting them in increasing time spent in moderate to vigorous physical activity (MVPA), for example by engaging in structured exercise or active commuting. However, only few studies have investigated whether workplace-delivered interventions focusing on SED or PA can also improve mental health outcomes [
11‐
14] and the World Health Organization (WHO) recommends them only conditionally as workplace mental health promotion strategies due to very low certainty of evidence [
1].
Workplace-delivered interventions among office workers can lead to meaningful reductions in SED of up to 100 min per 8 h workday [
15,
16]. The evidence regarding effectiveness of workplace interventions among office workers focusing on increasing MVPA is inconsistent [
17,
18]. The available evidence indicates that interventions which focus on such movement behaviors may reduce symptoms of depression or anxiety, relieve stress [
11], and improve mental wellbeing [
12] and mental health-related quality of life [
14]. However, systematic reviews of previous interventions point to a lack of high-quality randomized controlled trials (RCTs) with adequate sample sizes and control conditions, which might explain the inconsistency of results across studies [
11‐
13]. In addition, there are considerable challenges with previous workplace interventions among office workers that aimed at increasing MVPA or reducing SED. For example, time spent in these different movement behaviors has predominantly been measured with self-reports that are less precise and reliable than accelerometers due to, for example, a social desirability or recall bias [
19]. Furthermore, previous interventions usually do not take individual preferences of participants into account, offering one standardized program for all participants [
18,
20,
21]. Most interventions that focus on reducing SED make changes to the physical work environment, for example by providing sit-stand-desks, only sometimes in combination with behavioral support such as self-monitoring or goal setting [
16,
20,
22]. It might be that previous studies have not achieved desired effects because they do not take individual needs and preferences regarding the choice of activity into account and do not provide sufficient behavioral support for changing behavior. Engaging in activities that people enjoy and find personally meaningful increases the likelihood of establishing and sustaining new behaviors [
23], and might be especially important for improving mental health outcomes via changes in movement behavior [
24]. Previous literature suggests that factors such as enjoyment, mastery of skills or goals, choice, social interaction, and a sense of belonging likely influence the relationship between PA and mental health [
25].
Furthermore, previous literature hypothesized that multi-level interventions including several components that provide support not only on the individual but also via the physical and social work environment and organization might be more efficacious compared to those that only address one of these levels of influences [
22,
26]. This is in line with widely established ecological models of health behavior which propose that movement behavior is determined by factors at these different levels [
27]. Very few multi-level RCT interventions among office workers investigated whether interventions targeting SED or PA can lead to improved mental health outcomes [
28‐
30]. The results of these studies suggest that multi-level interventions can improve sitting time [
28‐
30] and LIPA [
29], daily anxiety, occupational fatigue, job performance, work engagement [
28], stress, wellbeing and vigor [
30].
Several biochemical mechanisms have been suggested for explaining the mental health effects of increased PA on certain mental health outcomes, such as changes in neuroplasticity, neuroendocrine responses, or reductions in inflammatory processes [
31]. In addition, psychosocial mechanisms have been suggested, such as increased self-efficacy, self-esteem or social support [
31]. However, it is yet unknown in how far these biochemical or psychosocial mechanisms are depending on PA intensity [
32]. Therefore, more research is needed to establish whether interventions among office workers should focus on reducing SED by increasing LIPA, or whether increases in MVPA are necessary for improving mental health outcomes.
We designed a 3-arm multi-level cluster RCT among office workers that aimed to increase MVPA (iPA intervention) or reduce SED (iSED intervention) to improve mental health outcomes among office workers [
33].
PA and SED were considered as proximal outcomes because the primary aim of the interventions was to support changes in movement behavior. Thus, PA and SED were expected to be directly impacted by the interventions. Mental health outcomes were instead considered distal outcomes since we originally hypothesized to impact mental health via changes in PA and SED. However, previous analyses based on the same RCT showed that participants already had high levels of accelerometer-measured PA (approximately 1.5 h of MVPA per day) but also high levels of SED (approximately 13 h) at baseline, despite efforts to enroll less active office workers [
34]. The interventions had no average effects on accelerometer-measured MVPA, SED [
35], 24-h movement behavior [
34], movement behavior during work or non-work time [
34], cardiorespiratory fitness [
34] or cognitive functions [
36]. However, the interventions increased autonomous and controlled motivation among iPA participants [
37]. Self-efficacy in relation to regulating own movement behavior also improved among participants who attended at least three of the five counseling sessions [
37].
Thus, we hypothesize that the interventions could elicit changes in mental health outcomes via mechanisms that are not directly related to the intended changes in movement behavior, but via activation of psychosocial mechanisms such as increased self-efficacy or social support [
31]. This hypothesis is further supported by the fact that the interventions included counseling sessions based on cognitive behavioral therapy (CBT) and motivational interviewing (MI) to provide individual-level support for increasing PA or reducing SED [
38]. CBT is an evidence based method for changing health behaviors and improving mental health outcomes [
39,
40]. The counseling sessions aimed at helping individuals understand how thoughts and emotions impact their movement behavior and at increasing their self-efficacy for finding and sustaining new helpful behaviors that are in line with their personal preferences and values. MI is a client-centered, goal-oriented counseling approach that seeks to elicit and strengthen a person's intrinsic motivation for positive behavior change by exploring and resolving ambivalence [
41]. As the interventions might have impacted on mental health also directly and not only via changes in movement behavior,To our knowledge, only a few studies with small sample sizes have used CBT to support movement behavior change; existing studies among healthy older adults showed that CBT improves exercise behavior and increases adherence to predefined exercise programs [
42,
43]. To our knowledge, no previous interventions have included CBT and MI for the specific aim of supporting movement behavior change to improve mental health outcomes among office workers.
The aim of this study was to investigate the effect of multi-level workplace interventions focusing on increasing MVPA or reducing SED on mental wellbeing, depression or anxiety symptoms, and stress.
Discussion
The aim of this study was to evaluate the effects on mental health outcomes of two 6-month multi-level interventions among office workers that focused on increasing MVPA or reducing SED. Compared to the control group, the intervention group which focused on reducing SED improved mental wellbeing. Mental wellbeing also increased for the intervention group focusing on increasing MVPA in comparison to the control group but did not reach statistical significance. No intervention effects were found for depression or anxiety symptoms or stress.
Interpretation of intervention effects on mental wellbeing
Mental wellbeing increased by eight points for the iSED group and by six points for the iPA group. This is below the 10-point improvement that experts suggest as clinically relevant on the individual level [
49,
50]. However, the potential public health relevance of intervention effect sizes should be judged in light of the study population’s baseline mental health status, whether interventions are scalable and whether similar effect sizes can be anticipated in the wider target population from which the study population was drawn [
57]. Universal interventions, which are designed to benefit high- and low-risk individuals, tend to yield smaller effect sizes compared to interventions specifically targeting high-risk individuals [
57]. Smaller health improvements at the group level can still have a significant impact on population health since they apply to a large number of people [
57]. Our interventions were delivered by trained health coaches and the companies’ own managers and HR staff. Thus, intervention conditions resembled those that would exist if such interventions were implemented at scale at the workplaces, and similar effect sizes might be found under implementation conditions.
Improvements in mental wellbeing were observed in both intervention groups, but statistically significant intervention effects were found for the iSED group but not for the iPA group with no significant difference between the groups. This might indicate that the iSED intervention was more effective for improving mental wellbeing. In fact, according to the qualitative RCT analysis, iSED participants perceived fewer barriers to reducing their time spent in SED compared to iPA participants who described it as very challenging to integrate more exercise into their everyday lives [
45]. In addition, participants spent substantial amounts of time in SED at baseline, suggesting that there was room for improvement, opposite to the already large amounts of time spent in PA at baseline. However, no intervention effects on either movement behavior have been found [
34,
35]. Another potential explanation for the difference in intervention effects between intervention groups might be related to retention rates. Drop-out from the iSED group was larger compared to the iPA group, potentially because the iPA intervention was perceived as more attractive since it included free gym cards [
45]. Thus, it is possible that only highly motivated iSED participants completed the iSED intervention and provided complete data. This might explain why effects were found for iSED but not iPA participants.
The primary goal of the intervention was to support participants in increasing PA or reducing SED to ultimately improve their mental health. Considering that no interventions effects on PA or SED were found in previous analyses [
34,
35], the observed improvements in mental wellbeing might have been caused via other mechanisms. It is possible, that the observed increases in autonomous and controlled motivation, as well as self-efficacy, in relation to regulating own movement behavior led to improved mental wellbeing [
37]. In addition, the social and emotional support and attention that participants received through the interventions might have increased feelings of hope or expectations of positive change. Thus, there is a possibility that the interventions might have impacted on mental health also directly and not only via changes in movement behavior. Mental wellbeing might also be more easily influenced by these mechanisms than the other outcomes, i.e., symptoms of depression and anxiety and perceived stress.
Furthermore, the interventions were not placebo-controlled. Thus, we cannot exclude a potential placebo effect, i.e., participants experiencing positive changes in mental wellbeing due to their belief that they are receiving a beneficial treatment, even if the treatment itself had no therapeutic effect [
58,
59].
Interpretation of lack of intervention effect on depression and anxiety symptoms and stress
The population in this study was a generally healthy, non-clinical sample of working adults. The prevalence of depression symptoms was 2% among participants (
N = 3) compared to pre-pandemic levels of 6% among the general adult Swedish population [
60]. Anxiety symptoms were slightly more prevalent among participants (17%) compared to adult Swedish populations (14%) [
60]. The proportion of participants experiencing stress at least several times per week or daily at baseline was 30%, which is higher compared to the general Swedish population (15% of women, 10% of men) [
61]. Participants had good mental wellbeing at baseline, comparable to populations from other European countries [
49,
62‐
64]. Nonetheless, we found intervention effects on mental wellbeing but not depression or anxiety symptoms or stress. Differences in effects might be explained by the fact that the WHO-5 is designed to assess positive aspects of mental health, such as emotional well-being, vitality, and satisfaction with life, whereas the HADS measures indicators of mental disorder [
47]. Changes in mental wellbeing might be easier to achieve than changes in depression and anxiety symptoms among healthy office workers.
In addition, study participants already had higher than average levels of MVPA (approximately 100 min daily on average) [
34] at baseline compared to the average Swedish population [
65], indicating a potential ceiling effect. Dose‒response studies showed that even small increases in PA can improve depression or anxiety among less active populations, but the effects plateau around one hour of MVPA per day [
66], suggesting a curvilinear relationship.
It is also plausible that workplace-delivered movement behavior interventions are efficacious for improving some mental health outcomes, while others require other types of interventions, e.g., addressing workload, organization, or leadership. More research is needed to investigate this.
Comparison with previous research
The results of this study are partially in line with results from other RCTs among office workers that tested multi-level workplace interventions that were based on ecological models of health behavior. In contrast to our study, most previous studies focused on reducing sitting time and had longer intervention periods [
30]. Other RCTs reported improvements in daily sitting time [
30], overall SED and leisure PA [
67], occupational sitting and standing time [
28] and small but significant improvements in mental wellbeing (two points, assessed with the WHO-5) [
30], job performance, work engagement, occupational fatigue, sickness presenteeism, daily anxiety, and quality of life [
28]. However, one multi-level study did not find effects on total sitting, standing or stepping time and health- or work-related outcomes [
68]. Some [
30] but not all interventions [
67] improved stress but not depression and anxiety symptoms (assessed with the HADS) [
30].
Despite previous interventions also adopting multi-level designs, several dissimilarities might explain differences in intervention effects. First, the context, available resources and health culture at the companies differ across companies and are likely to impact the efficacy of such interventions [
69‐
71]. However, such contextual information is usually not provided, and their impact on intervention efficacy not assessed. For example, participants in our study already had sit-stand desks, as is common in Swedish workplaces, whereas the introduction of sit-stand desks was a component of many previous multi-level studies [
28,
30,
67,
68]. Available multi-level interventions among office workers also differ by the type and number of included intervention components, which levels of influence they address, the degree of involvement of managers or higher leadership, reach within the company, and intervention duration, all of which might explain differences in efficacy. In fact, manager support has been identified as very important for supporting more PA and less SED in the workplace but was not delivered as planned in our interventions since it was difficult to recruit managers as team leaders for the interventions [
45]. Instead, regular employees acted as team leaders. This might have reduced overall intervention efficacy. Some of the previous interventions were more intensive in terms of the number of components included per level of influence. It seems that more intensive interventions regarding the number of included components and frequency of contact with intervention components [
28,
30] achieved greater improvements in movement behavior, mental health and work-related outcomes compared to less intensive interventions [
68]. Often, these multi-level interventions aim at changing movement behavior as the short-term, proximal outcome. Proximal outcomes are expected to change faster than distal outcomes, which require longer to appear [
57]. Thus, a potential explanation for the differences in effects on mental health outcomes between our and other studies might be the differences in the timing of the follow-up measurements.
Another study based on this RCT found that iPA participants with high executive function and perceived high demands but also control in relation to work substantially increased LIPA, and iSED participants with these characteristics showed a tendency of reduced SED compared to the control group [
72]. Executive function refers to the higher-level cognitive skills used to control and coordinate other cognitive abilities as well as behavior. This emphasizes the importance of better understanding which individuals benefit from such interventions and how interventions can be tailored to benefit a broader population.
Implications
The findings from this study have implications primarily for occupational health researchers and employers. Most importantly, our study confirmed that holistic workplace-delivered interventions which provide support for movement behavior change on multiple levels can improve office workers’ mental wellbeing. The CBT and MI techniques were useful in providing individualized support for movement behavior change [
45] and their potential should be further evaluated in the context of workplace health promotion.
Since only a few well-designed RCTs have investigated the efficacy of movement behavior interventions on mental health outcomes, more research among diverse groups of office workers is needed to understand their potential for addressing the burden of impaired mental health. Researchers may want to consider the following aspects when designing future similar multi-level interventions. To improve the effectiveness of such interventions, there is a need to identify how (analysis of mediators), for whom (analysis of personal moderators), and under what circumstances (analysis of situational moderators) such interventions can lead to health improvements [
73]. Future research may also want to consider including diverse types of office work with varying possibilities for social interactions, degrees of remote work, responsibility, possibility for taking breaks, etc., all of which might impact the efficacy of such interventions. In addition, the office design, e.g., activity-based or large open plan offices vs. smaller unit offices, might affect efficacy.future studies should consider including placebo control groups to better understand the mechanisms by which movement behavior interventions cause improvements in mental health [
59]. It is important to consider that universal interventions may produce smaller effect sizes than interventions that focus on at-risk populations, but they might also be less likely to cause stigmatization since participation in such interventions is not determined by mental health status [
1].
Strengths and limitations
This study was based on a large cluster RCT evaluating the mental health effects of interventions that aim at improving movement behavior. The design of the interventions was based on ecological models of health behavior. The interventions were delivered by trained health coaches as well as HR staff and employees acting as team leaders. Therefore, the results might reflect what could be expected if the interventions were implemented at scale.
Several limitations must be considered. First, the study was underpowered for detecting statistically significant changes in mental health outcomes. Two hundred sixty-three office workers were enrolled, but 330 were needed to achieve a power of 80% to detect changes in stress, according to the sample size calculation [
33]. Only the stress measure was included in the sample size calculation. Hence, non-significant findings could be attributed to either true absence of intervention effects or the possibility that the effects, if present, were too small to be detected with the given sample size.
Conducting complete case analysis was considered more robust since randomization was related to missingness of data on mental health outcomes, contraindicating maximum likelihood estimation including participants with missing data. However, this further reduced the statistical power.
We found no differences between participants with complete data and those with missing data regarding key demographic characteristics. However, there might have been differences regarding motivation or other unmeasured factors, introducing a potential selection bias. Participants included in the analyses might therefore not have been representative of all included RCT participants.
A selection bias might also have occurred regarding the inclusion of workplace and employees. Participants with good mental health and favorable movement behavior might be more inclined to participate and might find it easier to engage with and enact the intervention components, overestimating the health benefits. However, health benefits for those participants might also be more difficult to achieve given their better mental health and movement behavior, leading to a potential underestimation of intervention effects. In addition, participants in this study were predominantly female, highly educated and 42 years old and potentially not representative of all office workers. Mental health outcomes were assessed via self-report questionnaires and may have been subject to reporting bias.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.