Device-measured physical activity, sedentary behaviour and cardiometabolic health and fitness across occupational groups: a systematic review and meta-analysis

It is well established that regular physical activity (PA) prevents several chronic conditions (e.g. cardiovascular disease, diabetes, cancer, hypertension, obesity, depression, and osteoporosis) and premature all-cause mortality [1, 2]. The World Health Organization recommends that adults accumulate a minimum of 150 min of moderate-to-vigorous intensity PA (MVPA) per week for good health [3]. Reported adherence to these guidelines in high-income countries remains low (36.8%) [4] and the prevalence of inactivity is more than double that in low income countries [5]. Independent of MVPA, there is strong evidence that time spent in sedentary activities (i.e. sitting, reclining, lying down while awake) [6] increases one’s risk for cardiometabolic diseases, cancers, depression, and premature death [7–10]. Much of the evidence looking at the relationships between PA, sedentary behaviour and health outcomes has focused on behaviours that occur outside of work hours (e.g. recreational PA, TV or leisure-time screen use) [10, 11].

In the landmark London Transport Workers Study, Morris et al. found that men who worked more physically active jobs (i.e. conductors) had a lower incidence of heart disease compared to those with inactive jobs (i.e. drivers) [12]. Although this study was published more than 50 years ago, the relationship between movement behaviours and health outcomes across occupations remains poorly understood. The “physical activity paradox” suggests that PA undertaken at work differs in quality, quantity and health effects than PA performed during leisure-time [13]. For example, occupational PA carried out at lower intensities may not elicit substantial improvements in cardiorespiratory fitness, such as in the case of nurses who have been shown to spend a large proportion of their wakeful time in light intensity PA (LPA) [14]. Recent systematic review evidence also suggests that higher levels of occupational PA (even while controlling for leisure-time MVPA) are associated with an increased risk in premature mortality in men [15]. It is important to build on past research to understand the intensities of PA at which workers engage both within and outside of working hours and the associations with cardiometabolic health.

With approximately 8 hours of one’s waking day spent at work [16], occupational tasks and environments can heavily influence PA and sedentary time (ST). Occupational PA and ST are broadly defined as the amount of these behaviours accumulated while at work (typically within an 8-h timeframe), whereas leisure-time PA or ST occurs outside of work hours [17]. In the only known systematic review examining occupation and leisure-time PA, Kirk et al. found that white-collar workers accumulated higher amounts of leisure-time PA than blue-collar workers, and occupations with low occupational PA and long work hours were associated with lower leisure-time PA [18]. However, research has shown that individuals are more sedentary at work than during leisure-time, irrespective of occupation type [19]. Beyond white- vs. blue-collar classifications, little research has compared PA levels (i.e. ST, LPA, MVPA) and cardiometabolic and fitness outcomes between occupational groups. Identifying the specific occupations that enable higher rates of inactivity can inform workplace and lifestyle interventions aimed at improving workers’ PA levels and health status.

Most of the research examining occupational PA has been measured using self-report tools (e.g. questionnaires) [20], introducing considerable measurement errors (i.e. social desirability, recall bias, poor validity) [21, 22]. It is likely that individuals misperceive some of their occupational PA as MVPA when in fact, it is spent at lower intensities [23]. Activity monitors such as accelerometers, inclinometers, and pedometers have been validated for measuring ST (including posture in the case of inclinometers), steps, and PA of varying intensity, bouts, and duration [24, 25]. These devices allow movement to be continuously and more precisely monitored, thus effectively studied in many occupational environments.

No rigorous systematic investigations have examined device-measured PA across occupations. With higher income countries experiencing a transition towards more sedentary occupations compared to lower income countries, it is important to understand the PA levels experienced by these workers. Therefore, the primary objective of this systematic review was, to summarize and compare device-measured PA across occupations in adults from high-income countries. This includes step counts, ST, and, time spent in LPA, moderate intensity PA (MPA), vigorous intensity PA (VPA) and MVPA. The secondary objective was to examine whether cardiometabolic and fitness outcomes differ by occupation.

The review was prospectively registered with PROSPERO (#CRD42017070448).

This systematic review examined and compared device-measured PA and ST, and cardiometabolic health and fitness indicators across occupation groups. Working adults spent about 60% of their working and waking time in sedentary behaviour; a very small proportion (~ 4%) of the day included MVPA. On average, workers fell within the 7000 to 11,000 steps/day range associated with good health [161]. Office and call center workers’ steps/day were among the lowest, while labourers and postal delivery workers had the most. We found several occupational group-related differences in levels of ST, LPA and MVPA at work and during wakeful time. Occupations that were desk-based (i.e. office workers) had the highest ST and the lowest number of steps, LPA (work and wakeful time) and MVPA (work). They did however, have more daily minutes of MVPA compared to all other workers. Laborers had the lowest ST and spent a significantly greater proportion of their work time in LPA and MVPA. Healthcare and protective services workers had higher levels of LPA at work (and wakeful time in healthcare) compared to other workers. Workers in driving-based occupations tended to have higher BMIs and BPs. While office workers had a significantly higher WHR than other workers, their percentage of body fat did not appear to differ.

As mentioned, similar investigations have been limited to blue vs. white-collar occupational classifications only. Such previous systematic review evidence (that included self-report and device-measured outcomes) identified that white-collar workers (e.g. desk-based professionals) were more likely to engage in higher amounts of occupational ST and lower levels of occupational PA [20]. White-collar workers have also been shown engage in higher leisure-time PA compared to blue-collar workers [18]. We also found that office workers had higher occupational ST and lower occupational PA compared to blue-collar workers (e.g. laborers). Similar to our findings, higher step counts have been observed in blue-collar workers compared to white-collar workers [163]. Our finding that office workers had greater minutes of MVPA during wakeful time suggests that they are likely spending a greater amount of time engaged in non-work (e.g. travel, leisure) MVPA. However, given that evidence suggests a substantial amount (60–75 min) of daily MVPA is required to overcome large amounts of ST [162], office workers remain at health risk due to their high ST.

Adults who are employed in occupations with long work hours, stressful conditions and who engage in low levels of occupational PA are at the greatest risk for lower leisure-time PA [18, 164]. Those participating in manual labor are more likely to engage in sedentary behaviour in their leisure-time, whereas office workers are less likely to engage in these behaviours in leisure-time [165]. Lunde et al. found that construction workers had a greater number of minutes/day of sitting in their leisure-time/outside of work than within (282 vs. 157 min/day) [94]. Mansoubi et al. found that office workers engaged in a greater proportion of MVPA and a lower proportion of ST outside of work hours than during [96].

Higher levels of occupational, leisure-time and total PA are shown to be associated with a reduced risk of all-cause mortality, especially among women [166]. Additionally, higher levels of occupational sitting have been associated with a greater risk for diabetes and mortality [9]. However, the “PA paradox” suggests that occupational PA may not confer the same benefits as leisure-time PA as it may: occur at too low of intensity; elevate BP as a result of heavy lifting and static postures; not provide adequate recovery time; and, be performed outside of a worker’s control (e.g. work tasks, protective clothing, stressors, surrounding environment) [13, 167].

Globally, occupation-related energy expenditure and PA has decreased substantially since the 1960s and is projected to continue to decline [168–171]. Given that the majority of working adults’ wakeful time is spent on the job, workplaces are opportune settings for PA. Previous research has shown that workplace interventions may increase MVPA, decrease ST and improve cardiometabolic health indicators among workers [172–176]. In addition, recent review evidence suggests that workplace interventions may improve mental health [177, 178], the ability to perform occupational tasks fully [179], and adverse musculoskeletal symptoms (e.g. neck and shoulder pain) [176, 180–182], though higher quality studies are still needed to identify the most promising interventions. Evidence has also shown that increasing PA levels may be advantageous to employers through reduced absenteeism [183, 184] and increased productivity [185]. The findings of this review therefore highlight the need for workplace and public health interventions to promote these behaviours both within and outside of work (i.e., active travel and leisure/recreation) and their cardiometabolic sequelae among working adults. It is also a call to action to recognize that more physically demanding occupations may not provide the same opportunity for health-enhancing PA. It is important that future research examines lower intensities of PA, the postural components of occupational work to understand the effects on health (e.g. prolonged standing, walking) and consider measuring sleep as part of the 24-h movement continuum.

Strengths of this review include a comprehensive search strategy developed with a research librarian (SV), an a priori established protocol with inclusion and exclusion criteria including device-measured movement intensities, cardiometabolic and fitness outcomes, the assessment of risk of bias, and the use of meta-analyses to provide overall summary measures across occupational groups. This review builds on previous research by focusing on device-measured outcomes, including varying intensities of PA and ST thereby removing sources of potential biases (recall and response bias). While device measures overcome biases, there was variation in the monitors used and cut-points/algorithms applied to determine intensity; we used a random effects model to account for such heterogeneity.

A limitation of this review is the small number of studies available for the meta-analyses of MPA, VPA, and the majority of the cardiometabolic outcomes. Further, there were insufficient studies to compare movement intensities across all specific occupations or occupational groups. There is a need for further studies to characterize the movement patterns (including comparisons between leisure- and work-specific movement and describing postural changes and bout duration of movement intensities) and cardiometabolic health and fitness in occupations, especially those that are not office workers. Future work would also benefit from examining the data in a more compositional manner looking at the 24-h daily period and how different behaviours at work might influence those outside. It is also important to recognize the limitations of many of the devices used which have the potential to misclassify behaviours. The cut-point thresholds employed, especially for the accelerometers, were often developed in controlled laboratory settings performing standard activities (e.g. walking on a treadmill) rather than during occupational tasks. This review included self-reported (largely BMI) and objectively measured cardiometabolic outcomes from studies which reported on the primary outcome of interest. Therefore, there is the potential for response bias. The meta-analytic software was unable to produce funnel plots and an Egger’s test for the steps meta-analyses due to large standard deviations and small sample sizes (S. Tarlow [Comprehensive Meta-Analysis support], personal communication, August 20, 2018). We were unable to perform a sex-specific analysis as most studies did not report on male and female data separately. Our review did not examine any of the reasons for why PA and ST differed by occupation. However, research suggests that the differences are likely affected by an individual’s work and leisure-time hours. During work hours, PA is most influenced by the type of tasks performed as part of the job, whereas, during leisure-time, socioeconomic background is an important determinant [186]. This is likely why there is a higher amount of leisure-time PA in white-collar workers. There was evidence of publication bias for many of the behaviour outcomes, as well as for body mass and BP, indicating that smaller studies had larger effect sizes. As most studies were smaller (N < 100) and non-representative, there is a need for larger, more representative studies examining device-measured PA and ST in occupational groups. Lastly, wide confidence intervals for many of the outcomes are indicative of heterogeneity among several occupation groups within each analysis and thereby reduce our confidence in the estimates. Finally, it is likely that our categorizations have misclassified some studies into the wrong occupation group. For example, we included nurses, nurses’ aides, physicians, diagnosticians and physiotherapists under healthcare workers. These groups may involve different types of work tasks.

This review identified that work-based and wakeful PA and ST differed between occupations. Office workers spent the greatest amount of time at work and throughout wakeful time sedentary, but also spend the most minutes/day in MVPA. Healthcare workers engage in more LPA, and laborers and postal delivery workers in more work time-specific MVPA. Future studies are needed to assess whether these patterns differ between men and women within the same occupation, describe leisure-time specific PA and ST and contrast them to occupational-specific PA and ST, and whether they translate to different levels of risk for poor cardiometabolic health. Ultimately, awareness that these health behaviours differ between occupations calls for the need to include workplace interventions that promote health-enhancing PA both at work and during leisure-time.