Health Reports
Daily accelerometer-measured physical activity patterns and associations with cardiometabolic health among Canadian working adults

Abstract

Background

Previous studies examining the cardiometabolic risks associated with physical activity (PA) in workers have predominantly used self-reported measures. Little is known about workers’ distinct daily PA patterns and whether these are linked with cardiometabolic risks. This study examined associations between patterns of workers’ accelerometer-measured daily PA and four markers of cardiometabolic health.

Data and methods

Working adults (N=8,229; 47% women; average age: 42 years; standard deviation = 0.3) were sampled from the Canadian Health Measures Survey (five cycles: 2007 to 2017). Accelerometer devices measured daily PA, and hierarchical cluster analysis identified distinct activity patterns. Multiple linear regression analyses examined associations between activity patterns and cardiometabolic risk markers (waist circumference, systolic and diastolic blood pressure, and non-high-density lipoprotein [HDL] cholesterol).

Results

Workers were classified into six distinct activity patterns. On average, compared with workers classified in the “lowest activity” pattern, workers with the “moderate consistent activity,” “fluctuating moderate activity,” “high daytime activity” and “highest activity” patterns were associated with lower waist circumferences; workers with the “fluctuating moderate activity” and “highest activity” patterns were associated with lower systolic blood pressure; the “moderate evening activity” pattern was associated lower diastolic blood pressure; and workers with the “fluctuating moderate activity,” “high daytime activity” and “highest activity” patterns were associated with lower non-HDL cholesterol. “High daytime activity” was associated with lower waist circumference in women, compared with men, and the “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower diastolic blood pressure in younger workers (40 years or younger).

Interpretation

Workers with high daily PA levels tended to have the most optimal cardiometabolic health. Some evidence suggested that there are benefits to moderate levels of PA, particularly for lowering waist circumference and non-HDL cholesterol. Findings may assist in identifying workers for PA initiatives to promote cardiometabolic health benefits.

Keywords

Workers, occupation, physical activity, cardiovascular disease, epidemiology, exercise, chronic disease

Authors

Aviroop Biswas, Peter M. Smith and Cameron A. Mustard are with the Institute for Work & Health, Toronto and the Dalla Lana School of Public Health, University of Toronto. Peter M. Smith is also with the Department of Epidemiology and Preventive Medicine, Monash University, Australia. Cynthia Chen is also with the Institute for Work & Health. Stephanie A. Prince is with the Centre for Surveillance and Applied Research, Public Health Agency of Canada and the School of Epidemiology and Public Health, University of Ottawa.

 

What is already known on this subject?

• Most North American adults are physically inactive and sedentary for much of the day.
• Workers can accumulate their daily physical activity in different ways. At work, men and women can accumulate different levels of physical activity because of the gendered distribution of the labour force and differences in their work tasks. Outside of employment, differences in work and family responsibilities may offer little time for recreational physical activity.
• Declining physical activity levels in middle age can be attributed to the onset of health and functional problems, and work and family responsibilities.

What does this study add?

• This study used population-based accelerometer data and hierarchical cluster analysis to identify six distinct activity patterns of workers and examined associations with cardiometabolic risk markers.
• Compared with the “lowest activity” pattern, higher physical activity patterns—except for “moderate evening activity”—were associated with lower waist circumference. The “highest activity” pattern was also associated with lower levels of diastolic blood pressure and non-high-density lipoprotein (HDL) cholesterol. “High daytime activity” was associated with lower waist circumference and non-HDL cholesterol.
• The “highest activity” pattern was associated with lower waist circumference for women, compared with men, while both the “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower diastolic blood pressure in those aged 40 or younger, compared with those older than 40.
• The physical activity patterns identified in this study can have important public health implications, as they potentially represent more attainable physical activity targets for optimal cardiometabolic health and may be more feasible for large segments of workers who are unable to meet the Canadian physical activity recommendations.

Introduction

Cardiometabolic diseases, such as cardiovascular disease and type 2 diabetes, are leading causes of morbidity and mortality in North American adults.^Note 1 Being physically inactive can contribute to unhealthy weight gain and obesity, high cholesterol, elevated blood pressure, and blood glucose levels, all of which heighten the risk of developing metabolic syndrome and cardiometabolic diseases.^Note 2,^Note 3 Accordingly, the promotion of regular physical activity (PA) across a variety of settings throughout the day is an important component of public health recommendations.^Note 4,^Note 5

Most North American adults are physically inactive and sedentary for much of the day.^Note 6,^Note 7 Studies from Canada and the United States have demonstrated that PA levels decline over the life course, whereby levels are stable or increase in younger adulthood and begin to decline around middle age (40 to 45 years) and up to retirement.^Note 8, ^Note 9, ^Note 10 Declining PA levels in middle age can be attributed to a lack of time for PA because of increasing work and family responsibilities, and the onset of health and functional problems.^Note 11 Subsequently, a sharp increase in the prevalence of cardiometabolic diseases in men and women of middle age and onward has also been reported.^Note 12 Men and women can also differ in their PA levels. On average, men accumulate more moderate-to-vigorous intensity PA than women, regardless of age—a difference that generally persists throughout the day.^Note 13,^Note 14 Conversely, women generally accumulate more light-to-moderate intensity PA than men.^Note 15 Men and women can also accumulate different levels of PA at work because of their representation in the labour force, with more men employed in manual labour jobs involving physically demanding tasks, and women more represented in occupations that involve many sedentary tasks,^Note 16, ^Note 17, ^Note 18 although some female-dominated occupations in health and child care are also physically demanding.^Note 19,^Note 20 Men and women also have different biological responses to PA. Light-to-moderate intensity PA, including brisk walking, provides protection from cardiovascular disease and diabetes to a greater extent among women, compared with men, and the minimum duration and intensity of PA may differ by sex.^Note 21,^Note 22

While PA guidelines do not differentiate recommendations based on the settings where PA occurs, systematic reviews have found, paradoxically, that high levels of physically demanding work can have detrimental effects on cardiovascular health and all-cause mortality, with no beneficial (protective) effects for both men and women.^Note 23,^Note 24 Occupational tasks also have effects on recreational PA, with workers who report long work hours and psychologically demanding jobs less likely to accumulate moderate-to-vigorous intensity recreational PA.^Note 25,^Note 26 The increasing use of accelerometers in research has made it possible to collect continuous and detailed information about the time spent in various movement intensities during a person’s free-living activity patterns, while overcoming the recall bias associated with survey-measured PA. However, there exists an evidence gap of available large-scale epidemiological studies examining workers’ activity patterns using device-measured PA data associated with objectively measured clinical outcomes. Accordingly, little is known about how workers typically accumulate PA and if this is associated with different cardiometabolic risks.

This study aimed to identify relationships between distinct PA patterns accumulated by workers in their waking hours and markers of cardiometabolic health. Furthermore, the study examined similarities and differences between younger compared to middle-aged and older workers, and between men and women, to understand whether these groups are associated with different cardiometabolic risks. Focusing on working adults provided the opportunity to examine total daily PA patterns and identify the patterns potentially associated with optimal cardiometabolic health outcomes that are relevant to large segments of the adult working population. Study findings can guide the primary prevention of cardiometabolic diseases by informing the promotion of PA in specific domains and settings. It was hypothesized that, compared with the least active workers, those characterized by the highest daily PA accumulation would have the most optimal cardiometabolic health profile.

Methods

Participants and procedures

The Canadian Health Measures Survey (CHMS) is an ongoing cross-sectional survey conducted by Statistics Canada that collects computer-assisted self-reported surveys, accelerometry data and measured health biomarker information from the Canadian household-dwelling population aged 6 to 79 years, living in the 10 provinces.^Note 27 Participants living in rural areas more than 100 km from a mobile clinic (or urban areas more than 50 km from an examination centre) are excluded. The included population covers approximately 96% of the Canadian population.^Note 27 Details on the survey design and data collection methods have been described previously.^Note 28,^Note 29 The present study conducted a secondary analysis of data from five consecutive CHMS cycles (2007 to 2009, 2009 to 2011, 2012 to 2013, 2014 to 2015, and 2016 to 2017). The combined response rate of the household survey and clinical components ranged from 49% to 56%.^Note 27 Information on the characteristics of non-respondents is not publicly available; however, pre-testing of the CHMS had found that non-respondents were more likely to be younger adults to be non-White and to report high stress. Work responsibilities and the length of clinic appointments were barriers to participation.^Note 30 The sample selection criteria excluded participants who were younger than 18 years, reported not working the week prior to responding to the CHMS, were pregnant or had a pre-existing cardiovascular disease diagnosis.

Participants with less than four days of 10 hours or more of valid accelerometer data were also excluded. This is a common protocol for the minimum daily accelerometer wear time used in population-based studies from Canada and the United States.^Note 31,^Note 32 The final sample size was 8,229 (53% men and 47% women). The study was approved by the University of Toronto’s Research Ethics Board (REB #00037753). All participants provided Statistics Canada with informed consent for the use of their data for research purposes.

Exposure variables: Daily activity patterns

Free-living PA information was collected from ambulatory CHMS participants who attended a mobile clinic to conduct the physical measures portion of the survey. Participants’ PA counts were measured by a waterproof Actical accelerometer (Philips Respironics, Oregon, United States) worn by participants on their right hip for a week except when sleeping—even when swimming and bathing.^Note 27 Accelerometer-measured activity counts were summed over one-minute intervals for seven consecutive days. Accelerometer measurement and data treatment procedures have been described in full elsewhere.^Note 31,^Note 33 In this study, the total activity undertaken by each participant was derived by calculating the average counts per minute for 10-minute periods over the course of the waking hours of a 24-hour day.^Note 34 There was a maximum of 1,008 PA data points for each respondent, based on six 10-minute recordings every hour for 24 hours, for seven consecutive days. The accelerometer data were highly dimensional, noisy, discontinuous and non-independent. The distance between all PA time points of every pair of respondents in the sample was calculated using the dynamic time warping method to account for the correlations between temporal series by considering the overall shape. Hierarchical cluster analysis was applied on the dynamic time warping distance matrix. Ward’s minimum variance method was used to identify the strongest clustering structures. A combination of the elbow, average silhouette and gap statistics methods was assessed to determine the optimal number of clusters. Clusters of various activity patterns were extracted from activity count data using hierarchical cluster analysis. Over 10-minute periods, four PA intensity thresholds that have been used elsewhere were arbitrarily chosen:35 low PA was operationally defined as accumulating fewer than 1,000 counts, moderate PA as 1,000 to 2,000 counts, and high PA as over 2,000 counts.

Six distinct accelerometer-measured activity patterns were identified, and their weekday and weekend activity counts are summarized in Figure 1 (previously published).^Note 33 Detailed profiles of each of the activity patterns are described in another study,^Note 33 and the characteristics of each pattern are provided in Appendix Table 1. Briefly, pattern 1, “moderate consistent activity,” consisted of workers who mostly engaged in moderate levels of PA on both weekdays and weekends. Moderate levels of PA started at about 8:00 a.m. on weekdays and noon on weekends, continuing to about 5:00 p.m., and steadily decreasing over the evening. Pattern 2, “lowest activity,” consisted of the least active workers, with mostly low levels of PA on both weekdays and weekends. Low levels of PA started at about 8:00 a.m. on weekdays and slightly later on the weekend, continuing to about 7:00 p.m. on weekdays and 5:00 p.m. on weekends, and steadily decreasing over the evening. Pattern 3, “fluctuations of moderate activity,” consisted of workers with fluctuating levels of moderate PA on both weekdays and weekends, starting at about 8:00 a.m. on weekdays and weekends, fluctuating to the highest levels of moderate activity at 9:00 a.m., noon and 5:00 p.m. on weekdays and at 11:00 a.m., 2:00 p.m. and 4:00 p.m. on weekends, and steadily decreasing for the rest of the evening. Pattern 4, “high daytime activity,” consisted of workers with high levels of activity from 8:00 or 9:00 a.m. to 5:00 or 6:00 p.m. and low activity during evening hours. Activity levels were higher on weekdays, compared with weekends. Pattern 5, “moderate evening activity,” consisted of workers with activity levels steadily increasing to moderate levels until midnight, then steadily decreasing in the early morning. PA occurred from about 8:00 a.m. on weekdays and 10:00 a.m. on weekends, with activity levels on weekdays steadily increasing to moderate levels until midnight, then steadily decreasing in the early morning until 5:00 a.m. Pattern 6, “highest activity,” consisted of workers with the highest activity levels, with fluctuations of high and moderate activity occurring from 8:00 to 11:00 a.m., 12:30 to 2:00 p.m., and 5:00 to 7:00 p.m. on weekdays, with activity levels steadily decreasing afterward. Weekends followed a similar activity pattern, but with a peak of the highest activity at about 10:00 a.m. and fluctuations of moderate-to-high activity from noon to 4:00 p.m.

Description of Figure 1



Figure 1 Accelerometer-measured activity patterns and their weekday and weekend activity counts

The figure shows six distinct patterns of physical activity (PA). Pattern 1, “moderate consistent activity,” consisted of workers who mostly engaged in moderate levels of PA on both weekdays and weekends. Moderate levels of PA started at about 8:00 a.m. on weekdays and at noon on weekends, continuing to about 5:00 p.m., and decreasing steadily over the evening. Pattern 2, “lowest activity,” consisted of the least active workers, with mostly low levels of PA on both weekdays and weekends. Low levels of PA started at about 8:00 a.m. on weekdays and slightly later on the weekend, continuing to about 7:00 p.m. on weekdays and 5:00 p.m. on weekends, and steadily decreasing over the evening. Pattern 3, “fluctuations of moderate activity,” consisted of workers with fluctuating levels of moderate PA on both weekdays and weekends, starting at about 8:00 a.m. on weekdays and weekends, fluctuating to the highest levels of moderate activity at 9:00 a.m., noon and 5:00 p.m. on weekdays and at 11:00 a.m., 2:00 p.m. and 4:00 p.m. on weekends, and steadily decreasing for the rest of the evening. Pattern 4, “high daytime activity,” consisted of workers with high levels of activity from 8:00 or 9:00 a.m. to 5:00 or 6:00 p.m. and low activity during evening hours. Activity levels were higher on weekdays compared with weekends. Pattern 5, “moderate evening activity,” consisted of workers with activity levels steadily increasing to moderate levels until midnight, then steadily decreasing in the early morning. PA occurred from about 8:00 a.m. on weekdays and 10:00 a.m. on weekends, with activity levels on weekdays steadily increasing to moderate levels until midnight, then steadily decreasing in the early morning until 5:00 a.m. Pattern 6, “highest activity,” consisted of workers with the highest activity levels, with fluctuations of high and moderate activity occurring from 8:00 to 11:00 a.m., 12:30 to 2:00 p.m., and 5:00 to 7:00 p.m. on weekdays, with activity levels steadily decreasing afterward. Weekends followed a similar activity pattern, but with a peak of the highest activity at about 10:00 a.m. and fluctuations of moderate-to-high activity from noon to 4:00 p.m.

Outcomes: Cardiometabolic risk markers

The cardiometabolic risk markers examined in this study included waist circumference, systolic blood pressure, diastolic blood pressure and non-high-density lipoprotein (non-HDL) cholesterol. These markers are significant components of metabolic syndrome and are strongly associated with physical inactivity and cardiometabolic risk.^Note 36,^Note 37 Non-HDL cholesterol levels were derived from the total cholesterol minus HDL cholesterol. While body mass index (BMI) is used in population studies to examine obesity and cardiometabolic health markers, only waist circumference was used as a measure of obesity-related risk because it measures visceral fat, which poses a greater cardiovascular risk and is associated with insulin resistance and dyslipidemia, more accurately than BMI.^Note 38,^Note 39 Plasma glucose and triglyceride levels were not consistently measured across full samples (only in two CHMS cycles) and were excluded from the analysis to preserve the analytical sample size.

Prior to participating in physical assessments at a mobile clinic, respondents were required to fast for 12 hours if they had been selected for a morning appointment, or to fast for 2 hours if selected for an afternoon appointment. Participants were asked to refrain from exercising on the day of their clinic visit. Waist circumference was directly measured at the highest lateral point of the iliac crest, according to the procedures outlined in the Canadian Physical Activity, Fitness and Lifestyle Approach manual.^Note 40 Resting seated blood pressure measurements were obtained using the BpTRUTM BPM-300 monitor (BpTRU Medical Devices Ltd., Coquitlam, British Columbia). Average systolic and diastolic blood pressure readings were calculated based on the last five of six blood pressure readings taken at one-minute intervals following a five-minute rest period, with a minimum of three valid readings required to determine an average. A certified phlebotomist collected and performed tests on participants’ venous blood samples to obtain cholesterol and plasma glucose profiles.

Covariates

Sociodemographic and health covariates included age, sex, educational attainment, household income, marital status, cohabiting with dependent children, regularity of alcohol consumption, and smoking status. Work-related variables included usual hours worked per week, work stress, and employed or self-employed status. Potential seasonality effects of PA across different domains and settings were estimated based on whether accelerometers were worn during colder (November to March) or warmer months (April to October). Accelerometer wear time was also included as a covariate.

Statistical analysis

Descriptive statistics were used to examine the distributions of the activity patterns (mean [standard deviation] for continuous variables, percentages for dichotomous and nominal variables). Chi-square and ANOVA tests were conducted to examine how variables differed for different activity patterns.

Ordinary least squares regression modelling was used to examine associations between activity patterns and the cardiometabolic risk markers of waist circumference, systolic blood pressure, diastolic blood pressure and non-HDL cholesterol. Model A examined crude associations. A second model further adjusted for accelerometer wear time, and subsequent partial models also included the separate adjustment for sociodemographic, health and work-related variables. A final model (Model B) was adjusted for all covariates. To examine similarities and differences in cardiometabolic risk markers between younger and middle-aged to older workers, and between men and women, models were stratified by age (younger than 40 [younger adults] and 40 and older [middle-aged and older adults]) and sex. Estimates were reported as beta (b) coefficients with 95% confidence intervals (CIs). To examine potential gender differences related to non-work-related responsibilities, models were also stratified based on whether CHMS participants reported cohabiting with dependent children. Differences between estimates from the sex- and age-stratified regression models were assessed post hoc using methods that considered the estimate and standard error around the estimate.^Note 41,^Note 42 Based on Wald tests and associated p-values using the post hoc approach, separate effect estimates were compared for each group (e.g., for men vs. women, or younger vs. middle-aged or older workers) and identified if one group had a more negative association with a cardiometabolic risk marker. All estimates were weighted using Statistics Canada’s sampling weights to generate results representative of the Canadian population. Variance estimates around each statistical estimate were adjusted using 500 bootstrap replicate weights to account for the clustered design of the CHMS.^Note 43 Tests were two-sided and p < 0.05 was considered statistically significant. The descriptive and regression analyses were conducted using SAS version 9.4.

Results

Descriptive characteristics of the sample

Weighted characteristics are shown in Table 1. Overall, 8,229 individuals were analyzed. The sociodemographic profiles of the analytical sample vs. the sample excluded because of less than four days of valid accelerometer data were similar, except for average age, marital status and educational attainment. The analytical sample was, on average, slightly older (42 years vs. 39 years), had a greater proportion of married individuals (66% vs. 58%), and more had a postsecondary education (69% vs. 60%). The distributions of average blood pressure, cholesterol ratio, waist circumference and BMI were similar for the included and excluded samples. Compared with men, women in the analytical sample had a higher proportion of part-time workers (25% vs. 12%), non-smokers (84% vs. 76%) and blood pressure within an acceptable range (89% vs. 83%). Men had a higher proportion of regular drinkers (77% vs. 68%), higher average waist circumference (96 cm vs. 88 cm) and higher average non-HDL cholesterol (3.7 mmol/L vs. 3.3 mmol/L).



Associations of workers’ activity patterns and cardiometabolic risk markers

The associations between activity patterns and cardiometabolic risk markers unadjusted (Model A) and adjusted for all potential confounders (Model B) are shown in Table 2. The b coefficients indicate associations between each activity pattern (compared with the “lowest activity” reference category) and a per unit increase or decrease in a cardiometabolic risk marker. In the adjusted models, compared with the “lowest activity” pattern, people with the “highest activity” (b = -7.65 [95% CI = -10.40, -4.90], p ≤ 0.0001), “high daytime activity” (b = -5.21 [95% CI = -7.25, -3.18], p ≤ 0.0001), “fluctuating moderate activity” (b = -5.19 [95% CI = -6.97, -3.40], p ≤ 0.0001) and “moderate consistent activity” (b = -2.99 [95% CI = -4.60, -1.39], p = 0.00) patterns had significantly lower waist circumference. The “highest activity” (b = -3.18 [95% CI = -5.29, -1.07], p = 0.00) and “fluctuating moderate activity” (b = -2.25 [95% CI = -3.91, -0.58], p = 0.01) patterns were associated with lower systolic blood pressure, while only the “highest activity” pattern was associated with lower diastolic blood pressure (b = -1.83 [95% CI = -3.33, -0.34], p = 0.02), while the associations with other activity patterns were not statistically significant. The “highest activity” (b = -0.35 [95% CI = -0.47, -0.22], p ≤ 0.0001), “high daytime activity” (b = -0.18 [95% CI = -0.33, -0.04], p = 0.02), and “fluctuating daytime activity” (b = -0.18 [95% CI = -0.28, -0.08], p = 0.00) patterns were associated with significantly lower non-HDL cholesterol, compared with the “lowest activity” pattern. Estimates from the unadjusted models were similar to the fully adjusted models, with notable differences between the “moderate evening activity” pattern and diastolic blood pressure, and between the “high daytime activity” pattern and non-HDL cholesterol, where non-statistically significant associations in unadjusted models were statistically significant in adjusted models.



Fully adjusted (Model B) associations between activity patterns and cardiometabolic risk markers stratified by age and sex are shown in Table 3. There were no statistically significant differences observed between men and women, except for the association between the “highest activity” pattern and waist circumference, with a lower associated waist circumference for women (b = -9.96 [95% CI = -13.26, -6.67], p ≤ 0.0001), compared with men (b = -4.83 [95% CI = -8.12, -1.54], p = 0.00). In terms of age-related differences, the “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower diastolic blood pressure in those aged 40 or younger but not for those older than 40. Models were also stratified based on cohabiting with dependent children (Appendix Table 2). The “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower systolic blood pressure among people with no dependent children, but not for those cohabiting with dependent children. The “highest activity” pattern was associated with lower systolic blood pressure in those cohabiting with children. “Fluctuating moderate activity” was associated with lower diastolic blood pressure in those with no dependent children, while the “highest activity” pattern was associated with lower diastolic blood pressure in those cohabiting with children, but not for those with no dependent children. The “fluctuating moderate activity” and “moderate evening activity” patterns were associated with lower non-HDL cholesterol in those with no dependent children, but not for those with cohabiting children.



Discussion

This study used accelerometer data and hierarchical cluster analysis to identify distinct activity patterns of workers and assessed whether these activity patterns were associated with differences in cardiometabolic risk markers. Compared with workers with the “lowest activity” pattern, workers with the “fluctuating moderate activity” and “highest activity” patterns were associated with lower levels of three of four cardiometabolic risk markers: waist circumference, non-HDL cholesterol, and systolic blood pressure (for the “fluctuating moderate activity” pattern) and diastolic blood pressure (for the “highest activity” pattern). This was followed by the “high daytime activity” pattern, which was associated with lower waist circumference and lower non-HDL cholesterol. “Moderate consistent activity” was also associated with lower waist circumference. No differences in cardiometabolic risk were observed between men and women, except that the “highest activity” pattern was associated with a lower waist circumference in women. In terms of age-related differences, the “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower diastolic blood pressure in workers aged 40 or younger, but not for those older than 40.

These findings agree with the hypothesis that, compared with the least active workers, those with the “highest activity” pattern would be associated with the largest reductions in cardiometabolic health risk markers among the identified activity patterns, reflecting compelling evidence that supports the health benefits of accumulating PA throughout the day.^Note 44, ^Note 45, ^Note 46 The findings also highlight other PA strategies for conferring cardiometabolic health, notably daily patterns of “fluctuating moderate activity.” Workers with a daily “fluctuating moderate activity” pattern alternate between light and moderate PA throughout the day. This group may include workers who mix low levels of PA with short periods of higher PA during their work hours, such as nurses and childcare workers,^Note 47,^Note 48 and participate in some forms of moderate PA outside work. This pattern can have important public health implications, as it can potentially represent a more attainable PA target for promoting cardiometabolic health and may be more feasible for large segments of workers who are unable to accumulate high levels of daily PA, but still accumulate PA in their daily lives.

Previous studies based on self-reported PA measures found that workers with high daily PA are at lower risk for metabolic syndrome, abdominal obesity, high blood pressure and reduced HDL cholesterol, compared with the most sedentary workers.^Note 49, ^Note 50, ^Note 51 These findings corroborate the physiological mechanisms of high PA linked to increased energy expenditure, reduced arterial stiffness and increased vascular relaxation, which contribute to blood pressure control, as well as the increased production of HDL-C, which promotes reverse cholesterol transport.^Note 52 By comparison, the present study findings were mixed, where adjusted regression model estimates showed mostly non-statistically significant associations, except in relation to waist circumference (strong beneficial association with most activity patterns except for “moderate evening activity”) and non-HDL cholesterol (small beneficial association with the “fluctuating moderate activity” and “highest activity” patterns). While activity patterns could not be classified according to work schedules because this information was not collected in all CHMS cycles, it is expected that most of the sample participated in daytime work—87% of working respondents in CHMS cycles 1 and 2 reported daytime work hours.^Note 53 Therefore, “typical” daytime hours were used to define occupationally based PA, and those outside these hours as recreationally based PA. It is possible that workers with the “fluctuating moderate activity” and the “highest activity” patterns were engaging in more health-promoting PA from spending more time doing recreational PA and lower occupational PA.^Note 33 The higher levels of occupational PA among workers with “moderate consistent activity,” “high daytime activity” and “moderate evening activity” patterns might have unhealthy effects on cardiometabolic health if physically demanding and without much opportunity for rest.^Note 54 However, other studies have also suggested that high occupational PA is positively associated with reduced BMI or waist circumference,^Note 55 or that there is no difference in effect.^Note 56 It is possible that the accumulated PA did not attenuate the negative effects of prolonged sedentary time outside work hours and obesity.^Note 57 Physically demanding work may also lead to higher caloric requirements and unhealthy eating choices, which are harder to regulate. Another explanation for these mixed associations could be the “healthy worker effect.”^Note 58 The associations with lower waist circumference might also reflect the energy expenditure benefits of being active throughout the day (i.e., less sedentary time and more movement), particularly because the magnitude of associations was stronger for activity patterns with higher daily PA accumulation. Work-related PA (e.g., in blue-collar and technical occupations) contributes to some or a large proportion of daily movement and energy expenditure, even if workers have been found to be sedentary outside work.^Note 59,^Note 60, ^Note 61 However, this study did not adjust for energy expenditure or intake because of unavailable data. Reverse causation might also partially explain the findings, whereby workers with lower body weight and lower central adiposity were more likely to participate in PA in their leisure time. Further evidence from prospective studies with repeated measures of exposure and outcome variables would help characterize the direction of such associations more definitively.

Most of the associations between activity patterns and cardiometabolic risk markers showed no differences based on age and sex, suggesting that interventions promoting the accumulation of PA might be relevant for all workers. However, the differences in age and sex distribution among the activity patterns (described in Supplementary Table 1) suggest that male and female workers may accumulate their daily PA time differently. The patterns with the highest proportion of men were “high daytime activity” (68%) and “moderate evening activity” (70%); these may reflect the higher representation of men in physically demanding occupations in Canada16 and more time available for recreational PA in evening hours related to having fewer family demands than women.^Note 62 The “fluctuating moderate activity” (55%) pattern had the greatest proportion of women. This might reflect the challenges faced by women in engaging in higher-intensity PA at work and outside work, differences in activity preferences or gendered differences in job tasks.^Note 63,^Note 64 Workers with the “lowest activity” pattern were, on average, older than workers with the other activity patterns. This reflects the general trend of declining recreational PA in adults beginning in midlife.^Note 8 Possible reasons for this decline include a lack of discretionary time related to work and family responsibilities that commonly increase during this point in the life course.^Note 8 The theory of cumulative advantage also suggests that gender differences in PA can widen over time for middle-aged and older adults because of different gender-related barriers associated with age.^Note 65 Future studies should explore the interaction of age- and gender-related social factors in terms of daily PA accumulation and the subsequent cardiometabolic health outcomes.

The strengths of this study include the large and diverse population-based sample of workers; the identification of accelerometer-measured daily activity patterns that provide more precise and continuous measures of PA, compared with self-reported information; the inclusion of work-related covariates in regression models; and the use of clinically assessed markers of cardiometabolic health. This study is limited by its cross-sectional design, which does not provide information on the temporality of the associations. Residual confounding by unmeasured factors, such as diet quality, caloric intake and the quality of the PA infrastructure available to workers, may also influence the results. There was no detailed data to establish gender-related differences beyond cohabiting with dependent children. Furthermore, accelerometry measures can underestimate activities that involve limited stepping-based activity or carrying heavy loads.^Note 66 Additionally, accelerometers primarily measure movement when worn over the hip and, in doing so, may not capture upper-body movement. Further, they cannot distinguish whether a person is carrying any weight (e.g., walking while carrying a heavy bag expends more energy than walking without one). Accelerometers also do not distinguish between body positions, including sitting and stationary standing—both of which are potentially important components of workers’ daily movements and non-movements. Furthermore, the inability to confirm PA intensity using indirect calorimetry data to confirm cut points is a limitation. Only two cycles of the CHMS collected information about participants’ work schedules; self-reported domain-specific PA; and information on plasma glucose, HbA1c and triglycerides. These were therefore omitted from the analysis. A summary Z-score was not calculated because it would be incomplete without the inclusion of plasma glucose measures. It was also not possible to confirm from accelerometer time stamps whether PA was accrued at work or outside work hours.

Conclusion

Among Canadian working adults, compared with the “lowest activity” pattern, higher PA patterns—except for “moderate evening activity”—were associated with lower waist circumference. Additionally, the “fluctuating moderate activity” and “highest activity” patterns were also associated with lower levels of systolic and diastolic blood pressure, and non-HDL cholesterol. This was followed by the “high daytime activity” pattern, which was associated with lower waist circumference and non-HDL cholesterol. The “highest activity” pattern was associated with lower waist circumference for women, compared with men, while both the “moderate consistent activity” and “fluctuating moderate activity” patterns were associated with lower diastolic blood pressure in those aged 40 and younger, compared with those older than 40. Overall, findings lend further support to the notion that not all patterns of PA accumulated by workers are equally associated with cardiometabolic health benefits. This may assist in identifying public health priorities for PA-promotion initiatives aiming to confer cardiometabolic benefits to different populations.