Health Reports
Daily physical activity and sedentary behaviour across occupational classifications in Canadian adults

by Stephanie A. Prince, Karen C. Roberts, Jennifer L. Reed, Aviroop Biswas, Rachel C. Colley, and Wendy Thompson

Release date: September 16, 2020


Regular physical activity (PA) is known to protect against several chronic conditions (e.g., diabetes, cardiovascular disease, cancer) and premature all-cause mortality.Note 1Note 2 There is also evidence that greater sedentary behaviour (SB) (waking behaviours while sitting, reclining and lying downNote 3) increases one’s risk for many of the same chronic diseases and for premature mortality.Note 4 Available evidence suggests that large volumes (∼60 to 75 minutes per day) of walking or leisure/recreational moderate-to-vigorous intensity physical activity (MVPA) can offset the risks associated with prolonged sitting.Note 5 A large proportion of Canadian adults are at increased risk for poor health outcomes as the majority (82.5%) do not meet PA guidelines (⋝150 minutes per week of MVPA in ⋝10-minute bouts) and are sedentary for much of the day (9.6 hours).Note 6 Since adults spend a large amount of their day at work (∼8 hours) and PA levels vary considerably between occupations,Note 7 PA and SB at work may have a significant influence on overall daily PA and SB levels. This is particularly troubling since higher-income countries such as Canada are experiencing a transition toward more sedentary occupations.Note 8Note 9

Previous work based on device measures has identified that total and leisure-time steps, MVPA, and sedentary time differ by occupation.Note 10 Office- and desk-based workers have been found to obtain the lowest number of steps, the lowest amount of light-intensity PA, the most sedentary time but also the most MVPA (largely attributed to leisure time).Note 10 In contrast, those in more physically demanding jobs (e.g., labourers including agriculture workers, cleaners, construction workers, dry cleaners, farmers, freight mechanics) are some of the least sedentary, accumulate more steps per day, and spend more time engaged in light- and moderate-intensity PA.Note 10 Although there has been an increase in the number of studies reporting on device-measured PA and SB by occupation, the majority have been conducted in smaller, non-representative samples, have focused on a single PA outcome and have neglected to describe gender differences.Note 10 In Canada, there are known differences between women and men regarding the types of occupations they undertake,Note 11 their roles as caregivers outside of work hoursNote 11 and their occupational and leisure-time PA.Note 12 To the researchers’ knowledge, no studies have examined both device-measured alongside self-reported, domain-specific measures of PA and SB for Canadian full-time workers.

Steeves et al.(2018) recently described accelerometer-measured PA and SB levels across occupational groups in the United States using data from the National Health and Nutrition Examination Survey 2005–2006.Note 13 This was an update to their earlier workNote 14 and is one of the only large and representative examinations of device-measured PA levels across occupational groups. They condensed 22 occupational groups into three occupational PA categories based on total activity movement counts, MVPA and sedentary time measured by accelerometers. They found a strong association between occupational category and daily PA and SB levels. Workers in the high-activity group engaged in more MVPA and took more steps than those in the intermediate- and low-activity groups. Steeves et al. (2018) also found that the high-activity occupational group contained more men, individuals with less than a high school education and individuals with lower incomes.Note 13

The present study aimed to apply and build upon the methods used by Steeves et al. (2018)Note 13 to characterize the activity levels of Canadian workers across 10 occupational groups that represented skill level and type, using combined data from the nationally representative Canadian Health Measures Survey (CHMS). This study examined the hypothesis that PA and SB would differ across these 10 occupational categories; that male and female workers in the same occupational category would engage in different amounts of PA and SB; and that sociodemographic characteristics, clinical characteristics and self-reported PA and SB would differ across occupational categories based on device-measured activity groups.

The objectives of this study were to: describe accelerometer-measured daily PA and SB across 10 occupational categories; characterize occupations into three activity groups (i.e., low, intermediate, high) based on accelerometer-measured PA and SB; examine gender differences in PA and SB across the 10 occupational categories; and examine differences in sociodemographic and clinical characteristics (e.g., age, gender, education, marital status, immigration status, smoking status, chronic conditions, body mass index [BMI]) known to influence activity and occupation class and status between activity groups.


Data source

This study combined data from cycles 3, 4 and 5 (2012 to 2013, 2014 to 2015, 2016 to 2017) of the CHMS. The CHMS is an ongoing cross-sectional survey conducted by Statistics Canada that collects self-reported and directly measured health information from a representative sample of the Canadian household-dwelling population aged 3 to 79 years living in the provinces. This study includes data from adults (⋝18 years) that self-reported working full time (⋝30 hours per weekNote 15) at the time they were surveyed. Cycles 1 and 2 were not included because they used a non-comparable occupational classification system. Activity group classification was based on 4,080 respondents with valid accelerometer data from combined cycles 3 to 5. The analysis of self-reported PA, SB, and sociodemographic and clinical characteristics across the occupational activity groups (see below) was based on 3,698 respondents with complete self-reported PA and SB information from the CHMS household survey from cycles 4 and 5.

Occupational categories

Respondents were grouped into occupational categories based on the National Occupational Classification (NOC) 2011 codesNote 16 for cycles 3 and 4, and NOC 2016 version 1.1Note 17 codes for cycle 5. The NOC categories were developed by Statistics Canada and Employment and Social Development Canada and are the nationally accepted taxonomy and organizational framework of occupations in the Canadian labour market.Note 18 The NOC classifies the occupational information provided by respondents and groups occupations by the types of work usually performed.Note 18 “An occupation is defined as a collection of jobs that are sufficiently similar in the work performed.”Note 18 Ten broad occupational categories were used, based on skill level and skill type. Although the NOC 2016 has been updated from the NOC 2011, Statistics Canada developed a variant to ensure comparability between CHMS cycles. Table 1 lists the 10 broad occupational categories from the NOC 2011, and their major occupation groups.

Accelerometer-measured PA and sedentary time

In the CHMS, PA and sedentary time are measured using Actical accelerometers (Philips Respironics, Oregon, United States). Respondents who attended a clinic visit were asked to wear the accelerometer over their right hip on an elasticized belt during waking hours for seven consecutive days. For this study, respondents were required to have a minimum of four valid days, defined as days with ⋝10 hours of wear time. Previously validated movement intensity thresholds using the 2007 to 2009 CHMSNote 19Note 20 were applied to the data to derive sedentary time (<100 counts-per-minute [CPM]) and time spent in light-intensity PA (LPA; 100 to <1535 CPM) and MVPA (⋝535 CPM).Note 21 The accelerometer data also provide total raw movement counts (CPM) and step counts.

The three accelerometer-measured outcomes used to derive the composite score for creating the three occupational activity groups were average proportion of daily wear time spent sedentary (SED%); average weekly minutes of MVPA in ⋝10-minute bouts (MVPAbouted); and average daily steps. Additionally, accelerometer-measured averages are described for CPM, proportion of wear time in LPA (LPA%) and proportion of wear time in non-bouted MVPA (MVPA%).

Self-reported, domain-specific PA

The CHMS uses self-reporting to assess domain-specific PA. Respondents were asked to report their total minutes of PA per week spent in recreation or leisure, transport, or other (occupational or household) for a minimum of 10 continuous minutes. Because the PA module changed between cycles 3 and 4, and did not assess the same domains, the estimates derived from the two cycles are not comparable.Note 12 Therefore, for the purpose of this analysis, only data from the new PA module were used (cycles 4 and 5). Self-reported PA outcomes include minutes per day of active transportation PA, recreation PA, work and domestic PA, and total PA.

Self-reported leisure SB

The CHMS also uses self-reporting to assess type-specific leisure SB. Respondents were asked to report total hours per typical week spent in specific SBs during leisure time in the past three months. SBs included computer use; playing inactive video games; watching television, DVDs or videos; and reading. Total leisure screen time was calculated as the sum of time spent on computers, video games, and television, DVDs and videos.

Sociodemographic and clinical characteristics

The study examined differences between occupational activity groups for the following self-reported characteristics since they have been known to influence both activity and occupation status: age, gender (male or female), education (some postsecondary or less vs. post-secondary graduate), marital status (married or living with partner vs. single or not living with partner), immigration status, smoker status, presence of any chronic condition (i.e., asthma, fibromyalgia, arthritis, back problems, osteoporosis, high blood pressure, high blood cholesterol, chronic bronchitis, emphysema, chronic obstructive pulmonary disease, diabetes, diabetes, heart disease, cancer, thyroid condition, mood disorder, eating disorder, kidney dysfunction or disease, liver disease or gallbladder problems, hepatitis, developmental disability or disorder, attention deficit disorder, learning disability, or any other long-term physical or mental health condition diagnosed by a health professional), and objectively measured BMI (kg/m2).

Statistical analysis

All analyses were conducted using SAS Enterprise Guide v.7.1 (SAS, Inc., Cary, NC). Descriptive statistics including means and 95% confidence intervals (CIs) are presented for the accelerometer-derived variables and the self-reported PA and SB variables for each of the 10 occupational groups.

Each of the accelerometer-derived PA variables were ranked in ascending order from 1 to 10 for the occupational groups—except for SED%, which was ranked in descending order. A composite score was generated for each of the occupational groups by summing the rankings of three of the variables known to have strong associations with health outcomes (steps, MVPAbouted and SED%) and to harmonize with the methodology by Steeves et al. (2018).Note 13 In the case of a tie, a higher MVPAbouted ranking was used. Occupational activity groups were based on the primary ranked summary score, whereby the top three ranked occupations were in the “high-activity” group and those in the bottom three were in the “low-activity” group. The remaining four occupations were in the “intermediate- activity” group.

Differences between activity groups in accelerometer-derived variables, sociodemographic and clinical characteristics, and self-reported domain-specific PA and type-specific leisure SB were assessed using analysis of variance, with multiple contrasts adjusted using a Bonferroni correction for continuous data and chi-square and contrasts for proportions. Results for men and women are presented separately.

All analyses were weighted using combined-cycle survey weights.Note 22 Cases that were missing data for the accelerometer or survey variables examined were omitted from the respective analyses. Analyses of accelerometer data used accelerometer subsample weights, and analyses of self-reported PA and SB and sociodemographic and clinical characteristics used full household weights. In the analysis of combined cycles 3, 4 and 5, degrees of freedom were set at 33. In the combined cycles 4 and 5, degrees of freedom were set at 22. To account for survey design effects, 95% CIs were estimated using the bootstrap technique with 500 bootstraps.


Accelerometer-measured PA characteristics of Canadian full-time working adults

Descriptive information on the accelerometer-derived PA variables is presented in tables 2 to 5. On average, Canadians employed in full-time work were sedentary for 68.9% of their day (95% CI: 68.3% to 69.6%), took 8,984 steps per day (95% CI: 8,719 to 9,249) and accumulated 79.5 minutes per week of MVPAbouted (95% CI: 71.1 to 87.9). Among these Canadians, only 18.5% (95% CI: 15.3% to 21.7%) met the Canadian Physical Activity Guidelines. Most of the significant differences between occupational groups were observed for SED%, LPA% and steps per day. Few to no differences were observed for MVPA%, MVPAbouted or CPM.

Activity groupings

Tables 2 to 5 describe the three occupational activity groupings, summary scores and the rank order of the accelerometer-derived PA variables in the total sample, and for men and women separately. Lower summary scores indicate higher PA and lower SED%, resulting in a higher overall rank. Among respondents, 26.3% were classified into the high-activity group, 27.5% into the intermediate-activity group and 46.2% into the low-activity group. Those in the high-activity group took significantly more steps per day (high: 9,904 vs. intermediate: 9,020 [p=0.04] vs. low: 8,369 [p<0.0001]) and had a lower SED% (high: 65.1% vs. intermediate: 69.6% [p<0.0001] vs. low: 71.0% [p<0.0001]). However, those in the high-activity group had a higher LPA% (high: 31.8% vs. intermediate: 27.4% [p<0.0001] vs. low: 25.6% [p<0.0001]) than the intermediate- and low-activity groups. No significant differences were observed between the groups for MVPAbouted, CPM or MVPA%.

The intermediate-activity group had a significantly greater LPA% compared with the low-activity group (p=0.013). No other differences were observed between the intermediate- and low-activity groups.

Other than the first-ranked occupational category (personal and customer information services), the other two categories in the high-activity group had some of the lowest quantities of MVPAbouted. In the low-activity group, the lowest-ranked occupational category (professional) was ranked the lowest in three of the six variables, with the fewest number of steps and the lowest SED% and LPA%. However, this category was also ranked the highest in three of the six variables: MVPAbouted, CPM and MVPA%.

Characteristics by activity groupings

Using data from cycles 4 and 5 only, no significant differences were found between activity group for age, immigration status, BMI or presence of a chronic condition (Table 6). Compared with the low-activity group, the high- and intermediate-activity groups had significantly greater proportions of respondents with lower levels of education (i.e., some postsecondary or less), who were single or not living with a partner, and who were smokers. Self-reported PA and SB differed between groups (Figure 1). The high- and intermediate-activity groups reported significantly more work and domestic PA, and more leisure computer use, video game play, television watching and total screen time. The low-activity group self-reported more active travel and recreational PA compared with the intermediate-activity group. However, the low-activity group self-reported significantly less total PA compared with the high-activity group.

Gender differences

Women in full-time work spent a greater proportion of their total day sedentary compared with men in full-time work (69.8% vs. 68.3%, p=0.009), but a lower proportion of their total day in MVPA (3.0% vs. 3.4%, p=0.02). Compared with men, women also had lower CPM (211.0 vs. 232.8, p=0.005) and took fewer steps per day (8,361 vs. 9,446, p<0.0001). However, women had similar LPA% compared with men (27.3% vs. 28.3%, p=0.06).

Significant differences between men and women were found within occupational categories (tables 3 and 4). For example, women working in industrial, construction and equipment operation occupations spent a significantly greater proportion of their total day sedentary, a lower proportion of their total day in LPA, and took fewer steps per day than men in the same occupational category (SED%: 73.4% vs. 64.5%, p=0.0005; LPA%: 24.3% vs. 32.5%, p<0.0001; steps: 5,079 vs. 10,333, p=0.01). Women working in personal and customer information services engaged in less MVPA than men in the same occupation category (2.4% vs. 4.2%, p=0.015).

Rankings differed between men and women. Although personal and customer information services was in the high-activity group for both men and women, the other occupations in the high-activity group differed. For men, occupations in industrial, construction and equipment operation were in the high-activity group, whereas, for women, these occupations were in the low-activity group. The opposite was true for technical and paraprofessional occupations, which were in the high-activity group for women, but in the low-activity group for men. There were significantly more men than women in the overall high-activity group compared with the intermediate- or low-activity groups (Table 6).


This study is the first to describe accelerometer-derived and self-reported daily PA and SB variables across occupational categories among Canadian full-time working adults. It builds upon previous work in the United StatesNote 13 and other research in high-income countries that has demonstrated that device-measured PA and SB levels differ by occupation.Note 10

This study found that respondents in the highest-activity group took significantly more steps and had a higher LPA% and a lower SED% than those in the intermediate- and low-activity groups. However, no significant differences were observed between activity groups for MVPAbouted or MVPA%; there was little to no variation in MVPA between occupational categories. As a result, ranking largely favoured respondents who took more steps and had lower SED%. As previously mentioned, large volumes of MVPA (∼60 to 75 minutes per day) are needed to offset the risks associated with prolonged sitting.Note 5 Working Canadian adults fall short of this requirement, regardless of their occupation. In fact, none of the occupational groups met the Canadian Physical Activity Guidelines for Adults,Note 23 and all workers spent a high proportion (65% to 73%) of their day sedentary.

The accelerometer data represent total, average daily MVPA. While the evidence is mixed, some studies have suggested a compensatory effect may exist, whereby those who engage in higher work activity may be less active during leisure time, and those who are more sedentary at work may engage in more physical activity during leisure time.Note 24Note 25 This could explain why no differences were observed between occupational groups when the average total daily MVPA was examined. This finding also aligns with evidence from a meta-analysis of device measures that found no occupational differences in total daily MVPA.Note 10 It is also possible that differences would have appeared if more occupational groups had been examined.

Results showed that workers in occupations which may be considered more “blue-collar” (e.g., construction, labourers, agriculture) reported more total movement (i.e., lower SED%, higher LPA% and more steps per day), lower daily MVPA, and more domestic and occupational PA. In comparison, workers in occupations that would be considered “white-collar” (e.g., management and professional, office-based occupations) tended to be more sedentary, but also spent more time across the whole day engaged in MVPA and reported higher levels of recreational PA. Other studiesNote 10Note 26Note 27 have similarly suggested that white-collar and office-based occupations engage in the least occupation-related MVPA and most sedentary time, but also report higher leisure-time and total-day MVPA. Differences in the types of PA in which workers in certain occupational groups engage are important since research has suggested that leisure-time PA infers different and potentially greater benefits for cardiovascular health compared with occupational PA (referred to as the “physical activity paradox”).Note 28 The types of tasks that workers in white-collar and blue-collar occupations perform are likely to affect what they do outside of work (i.e., during leisure time). The physical or psychological fatigue present in certain jobs may reduce workers’ motivation to participate in leisure-time PA.Note 29Note 30 Workers’ socioeconomic status (e.g., income, education, job status) may also affect their means and opportunities to engage in leisure-time and travel-related PA. Research suggests that an individual’s socioeconomic status is one of the greatest influences on their PA, with a higher status associated with more leisure-time PA, but lower occupational PA.Note 31Note 32 These factors may explain why individuals who perform blue-collar jobs have been found to be less active during their leisure time.Note 33 Understanding the socioeconomic inequalities that exist with respect to occupational, leisure-time and total PA levels is important for developing interventions and policies to reduce these disparities.

All self-reported PA variables differed significantly between activity groups. Active travel and recreational PA levels were higher in the low-activity group, while work and domestic PA levels were higher in the high- and intermediate-activity groups. Evidence suggests that office-based workers likely spend more time engaged in non-work MVPA (e.g., active travel and leisure) compared with other workers.Note 10Note 34In comparison, workers employed in more manual occupations are more likely to be sedentary in their leisure time.Note 35Note 36 However, there is contradictory evidence that suggests that greater self-reported occupational PA is positively associated with leisure-time PA.Note 26 Self-reported PA is subject to respondent bias and an individual’s perception of the duration and intensity of activity. A comparison of the self-reported PA module with accelerometer-derived PA within the CHMS identified that work and domestic PA were likely more representative of LPA than MVPA.Note 12 Other research has shown that the variability in PA between occupational activity groups (similarly defined) is greater when using self-reports compared with accelerometers.Note 37 This is reflected in the higher LPA% and number of steps per day observed in the high-activity group occupations, and in the lack of variability in accelerometer-measured MVPA across activity groups.

Notably, respondents in the high- and intermediate-activity groups self-reported more leisure screen time compared with the low-activity group (high: 3.1 hours/day, intermediate: 3.2 hours/day, low: 2.4 hours/day). The opposite was found for leisure reading, with the low-activity group reporting a greater amount compared with the other two groups (low: 0.7 hours/day vs. high: 0.4 hours/day vs. intermediate: 0.5 hours/day). It is not clear if this is a compensatory effect whereby those in more physically demanding jobs spend more of their leisure time sedentary and using screens, or if it is the result of socioeconomic influences on leisure behaviours.Note 38 There is evidence that adults with lower household or respondent education levels engage in more leisure screen time and less reading time than those with higher education levels.Note 39

To date, there has been a lack of occupational studies, especially using nationally representative cohorts, to examine gender differences in device-measured PA or SB.Note 10Note 13 This study begins to fill this gap, by showing that men and women in the same occupations experience different levels of PA and SB. In general, this study found that working women were more sedentary and took fewer steps, but had similar levels of MVPA compared with working men. There were also fewer women in the high-activity group. Similar to the analyses performed by Steeves et al. (2018)Note 13, the analyses in this study found that different occupations were classified into each activity group for men and women. This indicates that, within occupations, men and women engaged in a different volume and intensity of total-day movement. This may reflect that men and women in the same occupation perform different tasks.


This study is not without limitations. Firstly, accelerometer data were used to define the activity groups. Accelerometers help to remove many of the biases associated with self-reported PA, but they are not able to capture all types of activities (e.g., arm movements, cycling and water-based activities). They also apply pre-established cut points that may misclassify some movement intensity. For these reasons, ambulatory activities (i.e., activities that occur while stepping) are more likely to be captured than some of the activities associated with physically demanding occupations (e.g., carrying heavy loads, performing arm-based tasks, standing in place for prolonged periods), and standing may be captured as SB. Accelerometers also reflect total daily activity rather than providing domain-specific information (e.g., occupational PA). It is possible that the composite score that was based on these data introduced random error into the categorization of the occupations.

Secondly, the four valid days required for wear time did not require a specific composition of work and non-workdays, and total physical activity was examined regardless of when it occurred (during work or outside of work). Thirdly, it was not possible to describe the PA levels of specific occupations (e.g., nurse, teacher, engineer, custodian). Because of sample size limitations, the study was limited to the 10 broad NOC categories of the CHMS. Occupations within the categories are identified as having similar work performed (determined by tasks, duties and responsibilities). It is still possible, however, that individuals within the occupational categories do not perform all of the same tasks. This study’s findings are cross-sectional. It cannot be inferred if occupations influence PA or if workers’ preferences for PA could influence the type of occupations they choose.

Lastly, several of the occupation groups (e.g., workers and labourers in transport and construction) contained few women. These occupations have historically employed more men than women.Note 40 In addition to women being underrepresented in these occupations, the types of tasks they perform compared with men in the same category may not be accurately captured.


The majority of Canadian full-time working adults are not getting adequate MVPA and are spending a large proportion of their day sedentary—regardless of their occupation. Results of this large cross-sectional study demonstrate that both accelerometer-measured and self-reported PA and SB differ by occupation in Canadian working adults. The results also highlight that gender differences in the PA and SB levels within occupations. Adults working in lower-activity occupations report more recreational and travel-related PA and leisure reading, while those in higher-activity occupations report more work and domestic PA and more leisure screen time.

As a result, public health strategies focused on reducing daily sedentary time in all occupations, getting people in lower-activity occupations to move more at work, and getting people in high-activity occupations to get more leisure-time PA could be most beneficial. Further exploration is needed to more thoroughly understand how domain-specific PA and SB affect the health outcomes of Canadian workers in different occupations. These data support the potential for workplace policies to improve the uptake of health-enhancing MVPA among all Canadian workers and to promote awareness for the need for different PA and SB messaging based on occupation.


Stephanie Prince was funded by a Canadian Institutes of Health Research–Public Health Agency of Canada Health System Impact Fellowship.

Date modified: