Background
Several studies have found that the relationship between physical activity (PA) and health differs depending on whether the activity occurs at work or during leisure [
1,
2]. Moderate and high leisure time physical activity (LTPA) is generally associated with favorable health outcomes (e.g., reduced risk of cardiovascular disease and mortality), while no clear association or even an inverse relationship is observed for occupational physical activity (OPA) [
2‐
6]. This apparent paradox still remains unexplained.
One possible explanation for the observed different effects of OPA and LTPA on health could be that temporal activity patterns differ between the two settings. OPA may be constrained to a particular type, duration, and intensity of PA, e.g., with limited opportunities for the worker to take breaks at discretion. In contrast, the individual is left to organize the contents and temporal structure of LTPA according personal preferences. Temporal patterns (i.e., variations across time) of different activity types, such as walking, standing and sitting, are considered important determinants for cardiovascular [
7], metabolic [
8] and musculoskeletal health outcomes [
9], independent of the total exposure dose. Accordingly, at least 30 min per day of moderate PA accumulated in periods >10 min has been recommended for better health [
10,
11]. Moreover, recent studies suggest that time spent in prolonged sitting (e.g., uninterrupted sitting periods >30 min) is particularly detrimental to health [
12,
13], while sitting is not a health hazard to the same extent if accumulated from shorter periods. Thus, breaking up prolonged sitting by brief periods of walking or standing is associated with reduced resting blood pressure [
14], enhanced endothelial function [
15] and beneficial changes in biomarkers related to metabolism [
8].
Epidemiological studies on health effects of OPA and LTPA have mainly relied on self-reported PA, and the instruments used for measuring PA have differed between work and leisure. Self-reports are prone to bias compared to objective methods for assessing OPA and LTPA [
16,
17], and less reliable [
18,
19], and they cannot be used to assess temporal activity patterns at any particular detail [
20]. Thus, a trustworthy and detailed record of temporal patterns of PA needs to be based on an objective, valid and precise method for measuring PA across several days, such as accelerometry, accompanied by an appropriate analytical tool to retrieve the temporal structure of data.
Exposure Variation Analysis (EVA) [
21] has been widely used to quantify temporal variation in long-term recordings of biomechanical exposures at work; most notably postures and muscle activity (e.g., [
22‐
25]), but lately even daily PA, including sedentary behavior [
26‐
28]. In the latter application, EVA splits up a time line of categorical PA data (expressed by type or intensity) into periods spent without interruption in the same PA category. Hence, using EVA, the temporal pattern of PA can be expressed as (proportions of) time spent in uninterrupted periods of different durations (e.g., <1 min, 1–5 min, 5–10 min, 10–30 min, 30–60 min and > 60 min) at different PA types (e.g., sitting, standing, running, cycling, walking).
Blue-collar work is associated with a substantial prevalence of musculoskeletal disorders and cardiovascular diseases [
3‐
6,
29,
30]. Also, blue-collar workers, as opposed to, for instance, office workers, more obviously face the paradox mentioned above, i.e., that health effects of moderate-to-high levels of OPA and LTPA seem to be different. Although there have been efforts to objectively assess PA levels at work among blue-collar workers [
31,
32], little is known about their temporal activity patterns. To our knowledge, no studies are available that compare these patterns in detail between work and leisure. Due to the dynamic nature of much blue-collar work, OPA may be distributed in relatively short periods of separate activities. In contrast, population studies suggest that leisure contains only few prolonged periods of (non-sedentary) PA and more periods of prolonged sitting than work. This PA pattern may be particularly pronounced among workers in blue-collar occupations due to their physically demanding work tasks [
31]. Several studies suggest that PA differ depending on age and gender [
33,
34], including the temporal activity pattern [
26,
35]. Thus, age and gender may be modifiers of the potential differences in the patterns of PA between work and leisure.
Therefore, our aims were, 1) to document temporal patterns of objectively measured PA at work and during non-occupational time (henceforth referred to as leisure) in a cross-sectional sample of blue-collar workers, 2) to determine the extent to which these patterns differ between work and leisure, and 3) to assess the extent to which differences between work and leisure are modified by age and gender. We expected that the temporal distribution of time in any particular activity, including sitting, would differ between work and leisure, and that this would occur independently of total time in a particular PA type. Specifically, we hypothesized that standing and walking would to a larger extent occur as brief periods between other activities at work than during leisure, while prolonged sitting periods would occur more during leisure than during work.
Methods
Study population and design
The present study was conducted on a cross-sectional sample of male (N = 108) and female (N = 83) blue-collar workers from the ‘New method for Objective Measurements of physical Activity in Daily living (NOMAD)’ study in Denmark. Data were collected from October 2011 to April 2012. Danish surveys and registers were used to select seven occupational groups with a high prevalence of musculoskeletal disorders and with varying exposures to OPA (i.e., workers in the health service sector, assembly workers, cleaners, construction workers, manufacturing workers, garbage collectors, and mobile plant operators). Workers were then recruited by convenience from different workplaces, primarily through contact with trade unions or safety representatives. Workplaces were eligible if workers were allowed to participate in the study during paid working hours.
Individuals were allowed into the study if they performed blue-collar work as their main occupation for at least 20 h per week, and if they were between 18 and 65 years of age. Workers were excluded if they declined to sign an informed consent to participate, reported to predominantly perform white-collar work, were pregnant, were absent from work due to sickness on the day of testing, or reported skin allergy to adhesives.
In total 358 blue-collar workers were offered participation, out of which 259 volunteered to participate and 223 filled out a questionnaire and used the accelerometers. Out of the 223, 10 workers were excluded as they reported their working hours to be less than four hours per day, and 22 workers were excluded because not even one working day with valid objective measurements was available. Thus, 191 workers were included in further examination of OPA and LTPA. The study was approved by the regional Ethics Committee in Copenhagen, Denmark (journal number H-2-2011-047) and conducted in accordance with the Helsinki declaration.
Procedure
Prior to data collection, all workers were invited to information meetings where the objective, procedure, and requirements of the study were explained. Workers declaring an interest in taking part in the study completed a screening questionnaire containing general information about demographic variables. Each participating worker were instructed to wear accelerometers for collecting PA for 24 h per day over four consecutive days, with research staff visiting the worker at the workplace on the first and last day. On the first day, the worker (a) underwent anthropometric measurements, (b) was equipped with accelerometers for objective measurement of OPA and LTPA, and a written diary, and (c) completed a computer-based questionnaire (results presented elsewhere [
36]).
The worker was instructed to perform a reference measurement in upright standing for 15 s each day, to report the times of those reference measurements as well as non-wear time in the diary, and even to note times when getting up in the morning, starting and ending work, and going to bed. The worker was allowed to remove the accelerometers if they caused itching or any kind of discomfort such as disturbed sleep. After completing the four measurement days, the worker returned the objective measurement devices, and the accelerometer data were downloaded to a computer by the research staff.
Assessment of age and gender
Age was determined from the workers’ Danish civil registration numbers, while gender was assessed using self-report.
Objective assessment of physical activity
Physical activity was measured continuously using two accelerometers (Actigraph GT3X, ActiGraph LLC, Florida, USA) placed on the thigh and trunk using double sided adhesive (3 M, Hair-Set) and medical tape (Fixomull, BSN medical), as previously described [
36,
37]. The Actigraph is a small, water resistant device (19x34x45mm, weight 19 g), which records, samples and stores tri-axial acceleration data at a frequency of 30 Hz with a dynamic range of ± 6G, and a 12 bit precision.
The Actigraph was initialized for recording and downloading of data using the manufacturer’s software (Actilife Software version 5.5, ActiGraph LLC, Pensacola, FL, USA). The accelerometer data were further processed and analyzed using a custom-made MATLAB based software, Acti4 (The National Research Centre for the Working Environment, Copenhagen, Denmark and BAuA, Berlin, Germany), which determines the type and duration of different activities and body postures with a high sensitivity and specificity [
38‐
40]. In this software, accelerometer data were low-pass filtered using a 5 Hz 4th order Butterworth filter and then split up into 2 s sequences with 50 % overlap. Afterwards, using the individual’s reference measurement, the occurrence of different PA types (i.e., sitting, standing, walking, running and cycling) was identified from the accelerometer outputs using algorithms presented previously [
37,
38]. Walking periods interrupted by brief (<30 s) sequences of standing, moving, running or cycling where merged if the total duration of the walking period exceeded 10 min. Non-wear was identified when, (a) the software detected a period longer than 90 min with zero acceleration counts, or (b) the participant reported non-wear time, or (c) artefacts or missing data were detected by visual inspection.
On average, the data collection period included 2.0 working days containing both work and leisure (i.e., non-occupational time). As only working days were addressed in this study, non-working days were excluded from the analyses, as were periods of sleep and non-wear, as well as periods not coded in the diary. The total non-wear time in the population was 1 % for the thigh and 3 % for the trunk accelerometer. A working day was considered valid for further analysis only if it contained objective measurements for at least four hours of work and >75 % of the average (across days) reported working time. Also, a day was accepted only if it comprised at least four hours of leisure, and >75 % of the average (across days) reported leisure time. Prior to further analysis on OPA and LTPA, each included working day was split into periods of “work”, defined as self-reported time spent working, and periods of “leisure”, defined as the waking hours not spent working, as described above.
Exposure Variation Analysis of physical activity (EVA)
For each measurement day, the time-line of the processed accelerometer signal was analyzed using EVA, identifying the occurrence of uninterrupted periods of different durations (i.e., <1 min, 1–5 min, 5–10 min, 10–30 min, 30–60 min and > 60 min) in each PA type (i.e., sitting, standing, running, cycling and walking). For each worker, average time spent in different EVA cells (e.g., sitting without interruption for 1–5 min) was expressed in minutes per day (i.e., total minutes in a particular EVA cell divided by the number of measured working days) and as percentages (minutes/day in a particular EVA cell divided by the average measured minutes/day). This was done separately for OPA and LTPA. Referring to PA recommendations from the American College of Sports Medicine and the American Heart Association [
10], and the 2008 Physical Activity Guidelines for Americans [
11] we then derived four selected derivatives from the EVA matrix according to Straker et al. [
27], i.e., “
prolonged sitting” (time spent in uninterrupted sitting periods >30 min), “
brief bursts (BB) standing” (time spent in <5 min standing periods) and “
BB walking” (time spent in <5 min walking periods), and “
walking >10 min” (time spent in >10 min walking periods). These metrics comply with recommendations based on biomedical evidence [
10,
11,
15,
27,
41‐
43] and operationalize the characteristics of OPA and LTPA addressed in the driving hypotheses of our study.
Statistical analyses
Descriptive data are presented as mean and standard deviation (SD) between workers, or frequencies. Time spent in different PA categories were averaged across days and expressed in percentages or minutes. All statistical analyses were carried out using the software SPSS, version 22 (IBM, US). The level of significance (α) was set at p < .05. Variables were visually inspected for normal distribution, and no deviations were identified that preclude the use of parametric procedures as described below.
The difference in temporal activity patterns between work and leisure as expressed by the selected EVA derivatives (i.e., prolonged sitting, BB standing, BB walking and walking >10 min) was tested using repeated measures ANOVA; time period (two levels: work, leisure) was treated as a within-subject factor. To determine whether differences between patterns of OPA and LTPA occurred independently of total PA time, repeated measures ANCOVA was used to adjust for the difference in total PA time between work and leisure (ΔtotalPA, i.e., LTPA subtracted from OPA; ANCOVA). Partial eta squared (η2) was used as a measure of effect size. ANOVAs were also constructed with time period (two levels: work, leisure) as a within subject factor and gender as a between subject factor to investigate the main effect of gender and the interaction (gender × time) on each EVA derivative.
Linear regression models were used to identify a possible association between each EVA derivative and the total time in each activity type, both during work and leisure. Associations between the EVA derivatives in OPA and LTPA were expressed using Pearson correlation coefficients. For each EVA derivative, a multiple regression model was developed to retrieve the possible association of age and gender with the difference between work and leisure in that EVA derivative (i.e., the LTPA result subtracted from the OPA result).
Competing interests
The authors declare that they have no competing interests.
The original NOMAD study was conducted with financial support from the Federal Institute for Occupational Safety and Health (BAuA), Berlin, Germany and the National Research Centre for the Working Environment (NRCWE), Copenhagen, Denmark. The present study was supported by grants from the Swedish Research Council for Health, Working Life and Welfare (Forte Dnr. 2009–1761). The sponsors did not influence the collection, analysis and interpretation of data, writing of the manuscript or the decision to submit the manuscript for publication.
Authors’ contributions
DH designed the study, carried out the statistical analyses, and drafted and revised the manuscript. SEM contributed in designing the study, and in drafting and critically revising the manuscript. NG and MK participated in planning the study and critically revising the manuscript. AH conceptualized the original NOMAD study and was together with MK responsible for coordinating the data collection; he participated in designing the present study, and in critically revising the manuscript. All authors read and approved the final manuscript.