Associations among workplace environment, self-regulation, and domain-specific physical activities among white-collar workers: a multilevel longitudinal study

Associations among workplace environment, self-regulation, and domain-specific physical... Background: Psychological and environmental determinants have been discussed for promoting physical activity among workers. However, few studies have investigated effects of both workplace environment and psychological determinants on physical activity. It is also unknown which domains of physical activities are promoted by these determinants. This study aimed to investigate main and interaction effects of workplace environment and individual self-regulation for physical activity on domain-specific physical activities among white-collar workers. Methods: A multi-site longitudinal study was conducted at baseline and about 5-month follow-up. A total of 49 worksites and employees within the worksites were recruited. Inclusion criteria for the worksites (a) were located in the Kanto area, Japan and (b) employed two or more employees. Employee inclusion criteria were (a) employed by the worksites, (b) aged 18 years or older, and (c) white-collar workers. For outcomes, three domain-specific physical activities (occupational, transport-related, and leisure-time) at baseline and follow-up were measured. For independent variables, self-regulation for physical activity, workplace environments (parking/bike, signs/bulletin boards/advertisements, stairs/elevators, physical activity/fitness facilities, work rules, written policies, and health promotion programs), and covariates at baseline were measured. Hierarchical Linear Modeling was conducted to investigate multilevel associations. Results: Of the recruited worksites, 23 worksites and 562 employees, and 22 worksites and 459 employees completed the baseline and the follow-up surveys. As results of Hierarchical Linear Modeling, stairs/elevator (γ=3.80 [SE=1.80], p<0.05), physical activity/fitness facilities (γ=4.98 [SE=1.09], p<0.01), and written policies (γ=2.10 [SE=1.02], p<0.05) were significantly and positively associated with occupational physical activity. Self-regulation for physical activity was associated significantly with leisure-time physical activity (γ=0.09 [SE=0.04], p<0.05) but insignificantly with occupational and transport-related physical activity (γ=0.11 [SE=0.16] and γ=−0.00 [SE=0.06]). Significant interaction effects of workplace environments (physical activity/fitness facilities, work rules, and written policies) and self-regulation were observed on transport-related and leisure-time physical activity. (Continued on next page) * Correspondence: kzwatanabe-tky@umin.ac.jp Department of Mental Health, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 2 of 14 (Continued from previous page) Conclusions: Workplace environments such as physical activity/fitness facilities, written policies, work rules, and signs for stair use at stairs and elevators; self-regulation for physical activity; and their interactions may be effective to promote three domain-specific physical activities. This study has practical implications for designing multi-component interventions that include both environmental and psychological approaches to increase effect sizes to promote overall physical activity. Keywords: Physical activity, Workplace environment, Multilevel study Background entire lifestyle [18–20]. Thus, workplace environment is The need for promotion of physical activity had been also an important determinant for physical activity among strongly recognized among workers since rapid changes workers. to the modern labor market have resulted in a large in- However, few studies have investigated effects of both crease of workers engaged in sedentary behavior and workplace environment and psychological determinants low-activity occupations [1]. Physical activity is one of on physical activity among workers using a multilevel the most important health behaviors in public health [2], study design, and no study has dealt with self-regulation and promoting this behavior is one of the most evident for physical activity. It is also unknown which domains workplace interventions for primary prevention of com- of physical activities are promoted by workplace envir- mon mental disorders [3]. Moreover, significant asso- onment and psychological determinants, since most ciations between physical activity and work-related studies measured only overall physical activity or one outcomes have repeatedly been reported, such as specific domain of activities. It is important to stratify well-being and presenteeism [4], absenteeism, job stress, the domains of physical activity because each activity employee turnover [5], and work ability [6]. These find- could arise in different contexts, could be caused by dif- ings indicate that promoting workers’ physical activity is ferent determinants [21], and could influence different indispensable for occupational health promotion and a outcomes [22]. Previous observational studies among sustainable workforce. three large worksites in Canada indicated that To promote physical activity among workers, multilevel self-efficacy for physical activity mediated the relation- factors across different levels have been suggested as de- ship between perceived workplace environment and terminants and targets for intervention [7, 8]. Of the psy- physical activity in the workday [23, 24]. However, in chological determinants, self-efficacy and self-regulation these studies, physical activity was measured only during for physical activity are recognized as important factors in the workday. Furthermore, they analysed the associa- theoretical models [9–11]. Especially, self-regulation (e.g., tions using simple regression as opposed to multilevel planning, scheduling, and self-organisational behaviors) analyses and did not address self-regulation. Another has recently been indicated to be strongest mediator be- previous study was a multilevel longitudinal study tween physical activity interventions and behavioral among 16 worksites and 129 employees in Japan [25], changes in physical activity in a non-clinical adult popula- which showed that employers providing fitness facilities tion [12]. Environmental determinants have also been dis- increased the positive association between self-efficacy cussed for promoting physical activity [13]. Workplace for physical activity and overall physical activity. How- environment can largely influence activities among ever, they did not measure specific domains of activities workers [14]. A workplace environment for promoting and did not address self-regulation either. For a clearer physical activity typically includes organising assessments/ understanding of how workplace environment, psycho- counseling/educations, informational support (e.g., post- social determinants, and physical activity interact, fur- ers/flyers/bulletin boards/maps), organisational policy, in- ther studies are needed. ternal physical environment (e.g., physical activity In this study, we aimed to investigate the associations equipment/stairs/lockers/showers/office connectivity), among workplace environment, self-regulation for phys- co-workers’ social support, and external environment (e.g., ical activity, and three domain-specific physical activ- walkability/parking/active commuting/physical activity fa- ities—occupational, transport-related, and leisure-time— cilities outside the workplace) [15–17]. These environ- in white-collar workers using a multi-worksite longitu- ments might be effective for promoting not only physical dinal design. White-collar workers are mainly engaged activities during work but also activities out of work (e.g., in sedentary work and are a primary target among the transport-related and leisure-time). Several observational working population [19]. Our hypotheses were as fol- studies suggested that the effects of workplace environ- lows: (H1) several aspects of supportive workplace envir- ment spilled over to leisure-time physical activity and onment will be positively associated with each domain Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 3 of 14 of physical activity among white-collar workers; (H2) Measures self-regulation for physical activity will be positively as- We measured worksite- and employee-level variables at sociated with higher scores in each domain of physical the baseline survey and measured physical activity both activity among white-collar workers; (H3) two-level at the baseline and the follow-up survey. Worksite-level interaction effects of workplace environment and variables were measured by a validated measurement, self-regulation for physical activity will be positively as- which consisted of observations by two independent sociated with higher scores in each domain of physical raters from the research team and a self-report question- activity among white-collar workers. This means that naire for worksite representatives (usually human re- the positive association between self-regulation of source personnel). Employee-level variables were white-collar workers and higher scores in each domain measured by a self-report questionnaire distributed to of physical activity will be stronger under well-facilitated workers. workplace environments. Workplace environment Methods Workplace environment to promote physical activity Study design and setting was measured by the Japanese version of the Environ- This was a multi-site longitudinal study. Multilevel mental Assessment Tool (EAT) [25, 27]. In this study, nested data were collected twice—at baseline (Oct–Dec workplace environment was operationally defined as 2015) and at follow-up (Feb–Apr 2016) —from work- EAT scores. Higher scores on the EAT indicate a more sites and workers employed by the worksites. We supportive environment for physical activity promotion approached worksites in the Kanto area, the economic and more invested environment by employers [28]. The center of Japan including metropolitan Tokyo, through concurrent validity with the rate of health abnormality some of the health insurance associations and chambers and test–retest reliability of the EAT were confirmed in of commerce in the area, using snowball sampling previous studies in both the US and Japan [25, 28]. In methods. Each worksite was explained the study in detail this study, subordinate components of the EAT that and asked to join the study. After the worksite represen- were used in the previous study [25] and suggested posi- tatives agreed to partake in the study, nested employees tive association with physical activity from the literature were recruited. Informed consent was obtained from all review [16, 17, 29–32] were used: parking/bike (4 points; representatives and workers via a consent form and the e.g., bike rack spaces), signs/bulletin boards/advertise- self-report questionnaire. The consent form and ques- ments (4 points; e.g., signs with physical activity mes- tionnaire informed participants that we assured protec- sages), stairs/elevator (4 points; e.g., signs encouraging tion of personal information and that the data would be stair use at building entrance or at elevators), physical anonymous and only used for academic purposes. The activity/fitness facilities (14 points; e.g., onsite fitness fa- study protocol received ethical approval by the research cilities), work rules (6 points; e.g., own lockers em- ethics committee of the Graduate School of Medicine ployees can have), written policies (6 points; e.g., written and Faculty of Medicine, The University of Tokyo, Japan policies supporting employees’ physical activity), and (No. 10919). Our study has been reported according to health promotion programs (20 points; e.g., education the Strengthening the Reporting of Observational studies classes for physical activity). Scores of health promotion in Epidemiology (STROBE) guidelines [26]. programs for physical activity, diet/nutrition, and weight management were combined into one variable to avoid Participants complexity in models. These scores were calculated by At baseline, a total of 49 representatives of worksites using the EAT scoring system based on a were provided with an explanation of the study and representative-reported questionnaire (section 1) and an asked to participate by the corresponding author. Work- observation form completed by researchers (section 2). site inclusion criteria (a) were located in the Kanto area, Scores of the observation (section 2) were rated by two the economic center of Japan including metropolitan of seven trained members from our research team for Tokyo, and (b) employed two or more employees. There each worksite and averaged. The research team members were no exclusion criteria for worksites. Within the were graduate students in psychology, nursing, public worksites which agreed to partake in the study, nested health, and health science. Inter-rater reliability of the employees were recruited. Employee inclusion criteria EAT scores ranged from to .46 to 1.00, considered suffi- were (a) employed by the worksites, (b) aged 18 years or cient values. older, and (c) white-collar workers (managerial, profes- sional, technical clerical, and other job types that re- Self-regulation for physical activity quired deskwork or sitting work). There were no specific Self-regulation for physical activity was measured by the exclusion criteria. Japanese version of the 12-item Physical Activity Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 4 of 14 Self-Regulation scale (PASR-12) [33, 34]. The PASR-12 Self-efficacy was measured using a scale developed by asked the workers how often they had utilized cognitive Marcus et al. [38]and Oka [39]. Because the original scale and behavioral methods for physical activity in the past was developed to assess self-efficacy for exercise, we re- four weeks (e.g., “I mentally kept track of my physical vised the word “exercise” to “physical activity.” The scale activity”). The 12-item measure consists of six factors: consists of four items (e.g., “I have the confidence to per- self-monitoring, goal-setting, eliciting social support, re- form physical activity even if I am a little tired”). All inforcements, time management, and relapse prevention. items are rated on a 5-point Likert-type scale (1 = Not All items are rated on a 5-point Likert scale (1 = Never, at All,5= Almost). Internal consistency and unidimen- 5= Very Often). Internal consistency, convergent valid- sional structural validity were confirmed in a previous ity, and structural validity of the Japanese version of the study [39]. In the present study, the total score of PASR-12 were confirmed in a previous study [33]. In the self-efficacy for physical activity was tripled to match present study, the total scale score was used in analyses its score range to that of self-regulation for physical ac- to avoid complexity. Cronbach’s alpha for the scale at tivity (12–60) to compare the effect sizes on physical the baseline survey was .93, considered an excellent activity in multilevel models. Cronbach’s alpha for the value. scale at baseline survey was .87, considered an excellent value. Physical activity Additionally, job strain as determinants for physical Three domain-specific physical activities were measured activity was assessed. Job strain was measured by the by the Japanese version of the Global Physical Activity Japanese version of the Job Content Questionnaire Questionnaire (GPAQ v2) [35]. This scale is widely used 22-item version (JCQ) [40, 41]. The JCQ includes five and has demonstrated reliability and convergent validity items for job demands (e.g., “My job requires working in nine countries, including Japan [36]. The test-retest very fast”) and nine items for job control (e.g., “I have a reliability in each domain was moderate to strong both lot of say about what happens on my job”). All items are in the validation study (Spearman’ rho 0.67 to 0.81) [36] rated on a 4-point Likert-type scale (1 = Strongly dis- and in this study (Intra-Class Correlation 0.66 to 0.84). agree,4= Strongly agree). The scale has been widely The GPAQ can assess occupational, transport-related, used to assess job strain, and its reliability and validity and leisure-time physical activity with regard to intensity were confirmed by a previous study [40]. Cronbach’s al- of the activity (moderate-to-vigorous) and time spent phas for job demands and job control at the baseline doing the activity (minutes, hours, days) based on a typ- survey were .65 and .75, respectively. According to the ical week. In the present study, the amounts of three user’s guide for the JCQ [41], we calculated continuous domain-specific physical activities per week (MET- scores of job strain by the ratio of job demands to job s-hours/week) were calculated according to the GPAQ control. analysis guide [37]. Participants who responded with in- consistent answers or implausible values were excluded Analyses from all analyses in the study (e.g., more than seven days Descriptive statistics, intra-class correlation coefficients in any column of days spent doing the activity per for employee-level variables, and multilevel correlation week). coefficients for two-level variables were calculated. For the main analysis, Hierarchical Linear Modeling (HLM) Covariates was conducted to investigate multilevel relationships Worksite-level confounders included worksite size and among workplace environment, self-regulation for phys- worksite location (urban, suburban, and local). Worksite ical activity, and physical activity in each domain. Three size was operationalized as an ordinal variable and clas- domain-specific physical activities at follow-up were en- sified into four categories (10–49, 50–99, 100–299, and tered into the models as the dependent variables. Work- ≥300 employees). Worksite location was measured by a place environment, self-regulation for physical activity at single ordinal item (urban, suburban, and rural) for baseline, and interaction effects of workplace environ- worksite representatives, “Where is the worksite ment and self-regulation for physical activity were en- located?” tered as independent variables. Worksite size, worksite Self-efficacy for physical activity as another psycho- location, self-efficacy for physical activity, gender, age, logical determinant, gender, age, employment status employment status, job type, working hours, and job (regular, part-time, dispatched, contract, and others), oc- strain at baseline were controlled as the covariates. In cupational status (managerial, technical and professional, addition, physical activities at baseline were also con- clerical, and others), and working hours per week (1–34, trolled for to analyze these associations longitudinally. 35–40, 41–50, 51–60, 61–65, 66–70, and ≥71 hours) The categorical covariates were transformed into were measured as employee-level confounders. dummy variables: worksite size (10–49 [reference Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 5 of 14 group]), worksite location (urban [reference group]), missing variables, the worksites and employees that par- gender (men [reference group]), employment status tially had missing values or that dropped out during (regular [reference group]), occupational status (not follow-up were included in the analysed model. We de- managerial [reference group]), and working hours per scribe the number of missing samples on all variables week (≤ 40 hours [reference group]). Of the continuous used in the analyses in descriptive statistics. Mplus ver- variables, worksite-level variables (workplace environ- sion 7.4 [43] (Muthén & Muthén, 1998–2015) was used ment) were grand-mean centered, and employee-level for each analysis. variables (self-regulation and self-efficacy for physical ac- We treated three domain-specific physical activities as tivity, age, job strain, and each domain of physical activ- the continuous values (METs-hours/week). Although ity at baseline) were group-mean centered for the distribution of these scores was right-skewed and might multilevel analysis to make the correlations between not be suitable assuming normal distribution, estimating two-level variables equal to zero [42]. cross-level interaction effects between individual and We estimated an unconditional model (model 1), a workplace was so complex in the multi-level model; im- crude random slope model (model 2), a crude inter- proper solutions were observed when we handled the action model (model 3), an employee-level adjusted outcomes as ordinal values. In addition, ordinal out- model (model 4), and a worksite-level adjusted model comes were not applied in our model because we ap- (model 5) in HLM and referred to the Akaike Informa- plied FIML. tion Criteria (AIC) as a fit index. The equation for the adjusted model (model 5) was explained as follows. Results Level 1 (employee-level) Figure 1 shows a participation flow chart of the work- sites and the employees in the study. Of 49 representa- YðÞ each domain of physical activtiy at follow−up ij tives of the worksites, 23 representatives agreed and ¼ β þ β ðÞ self −regulation þ β signed a consent form (response rate = 46.9%). Of the 0 j 1 j ij 2 j worksites and employees, 23 worksite representatives ðÞ self −efficacy þ β ðÞ gender þ β ij 3 j ij 4 j and 562 employees (265 men, 293 women, and 4 un- ðÞ age þ β ðÞ Not regular employment þ β ij 5 j ij 6 j known; mean age = 43.5, SD = 11.1) completed the base- line survey (response rate = 87.8 %). The average ðÞ Managerial job þ β ij 7 j number of workers who completed the baseline survey ðÞ ≧41 working hours per week þ β 8 j ij from each worksite was 24.4 (SD = 22.3). At the ðÞ job strain þ β ij 9 j follow-up, 22 worksite representatives and 459 em- ployees (227 men, 229 women, and 3 unknown; mean ðÞ each domain of physical activity at baseline þ e ij ij age = 44.7, SD = 11.1) completed the survey (response Level 2 (worksite-level) rate = 71.7 %). One worksite and 103 workers refused or dropped out during follow-up because of being too busy β ¼ γ þ γ ðÞ parking þ γ ðÞ signs þ γ 0 j 00 01 j 02 j 03 or unable to link. ðÞ stairs þ γ ðÞ fitness facilities þ γ j 04 j 05 Table 1 shows characteristics of the worksites and em- ðÞ work rules þ γ ðÞ policies þ γ j 06 j 07 ployees at baseline (23 worksites and 556 employees). Of ðÞ health promotion programs þ γ 562 employees, 6 were excluded due to inconsistent an- j 08 swers on the questionnaire to measure physical activity ðÞ worksite scale 50  99 þ γ j 09 [35, 37]. Among the worksites, 12 (52.2%) employed 10– ðÞ worksite scale 100  299 þ γ j 010 49 workers, classified as small-size worksites. The other ðÞ worksite scale≥300 þ γ j 011 11 worksites employed 50–99 (6), 100–299 (3), and ðÞ local or suburban area þ μ j 0 j more than 300 workers (2). Most of the worksites (60.9%) were located in an urban area (e.g., 23 special β ¼ γ þ γ ðÞ parking þ γ ðÞ signs þ γ 1 j 10 11 j 12 j 13 wards in Tokyo) in the Kanto region of Japan. Types of ðÞ stairs þ γ ðÞ fitness facilities þ γ j 14 j 15 industries were service (nine worksites), manufacturing ðÞ work rules þ γ ðÞ policies þ γ j 16 j 17 (four), medical and welfare (three), information and ðÞ health promotion programs þ μ j 1 j communication (three), education (two), wholesale and retail (one), and transportation (one). Of the employees, 0 τ τ oj 00 01 β ¼ γðÞ q ¼ 2…9  N most (69.8%) were employed full-time. About half qj q0 μ 0 τ τ 1 j 10 11 (52.4%) of the employees were engaged in clerical jobs, Because the Full Information Maximum Likelihood 112 (20.4%) mainly had managerial jobs, and 73 (13.1%) (FIML) method was used for parameter estimation to were engaged in technical and professional jobs. A total avoid a selection bias due to dropout and to impute of 340 (62.0%) employees worked over 40 hours per Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 6 of 14 Fig. 1 A participation flow chart of the worksites and the workers. Note. Nj = the number of worksites; Ni = the number of workers week, considered long-working workers [44]. Between with self-regulation and self-efficacy for physical activity employees who completed the follow-up survey and who (γ = 0.40, p <0.01 and γ =0.34, p <0.01),and dropped out, dropped-out employees were significantly transport-related physical activity was also positively associ- younger than the completers (p = 0.001). There was no ated with them (γ = 0.11, p <0.01 and γ =0.17, p < 0.01). other significant difference between two groups for However, occupational physical activity was not signifi- demographic, exposure, and outcome variables. cantly associated with them. For the worksite-level vari- After excluding nine participants due to missing data on ables, scores for each workplace environment were all independent variables, multilevel correlation coefficients inconsistently associated with each other. Of those vari- among workplace environment, self-regulation and ables, work rules were significantly and negatively associ- self-efficacy for physical activity, and each domain of phys- ated with transport-related physical activity (γ = −0.56, ical activity among 547 employees are shown in Table 2. p < 0.05), and physical activity/fitness facilities and Intra-class correlation coefficients for domain-specific phys- written policies were significantly and positively associated ical activities ranged from 0.04 to 0.15, indicating 4–15% of with leisure-time physical activity (γ = 0.73, p <0.05 the variances were explained by these worksite-level vari- and γ = 0.61, p <0.05). ables. For the employee-level psychological determinants, Tables 3, 4, and 5 show the main results of HLM on leisure-time physical activity was most strongly associated domain-specific physical activity at follow-up, indicating Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 7 of 14 Table 1 Characteristics of the worksites and employees at baseline (Nj = 23, Ni = 556) Worksite-level variables (Nj = 23) N (%) Min–Max Mean (SD) Missing (%) Worksite size 0 (0.0) 10–49 12 (52.2) –– 50–99 6 (26.1) –– 100–299 3 (13.0) –– ≥300 2 (8.7) –– Worksite location –– 0 (0.0) Urban 14 (60.9) –– Local/suburban 9 (39.1) –– Workplace environment 0 (0.0) Parking/Bike – 0.00–3.00 1.28 (1.0) Signs/Bulletin boards/ – 0.00–4.00 1.15 (1.2) Advertisements Stairs/Elevator – 0.00–2.00 0.74 (0.6) Physical activity/ – 0.00–7.00 0.74 (1.7) Fitness facilities Work rules 3.00–6.00 5.04 (0.8) Written policies – 0.00–6.00 0.26 (1.3) Health promotion – 0.00–6.00 1.43 (1.9) programs Employee-level Total (Ni = 556) Follow-up completer Drop-out p for difference variables (Ni = 454) (Ni = 102) N (%) Missing (%) N (%) Missing (%) N (%) Missing (%) Mean (SD) [Min–Max] Mean (SD) Mean (SD) [Min–Max] [Min–Max] Demographic variables Gender 4 (0.7) 2 (0.4) 2 (2.0) 0.091 Male 263 (47.6) 223 (49.3) 40 (40.0) Female 289 (52.4) 229 (50.7) 60 (60.0) Age M=43.56 (11.1) 12 (2.2) M=44.27 (11.1) 6 (1.3) M=40.21 (10.2) 6 (5.9) 0.001 [19–84] [19–84] [21–65] Employment Status 3 (0.5) 0 (0.0) 3 (2.9) 0.980 Regular 386 (69.8) 317 (69.8) 69 (69.7) Non-regular (Part-time, 167 (30.2) 137 (30.2) 30 (30.3) contract, dispatched) Occupational status 6 (1.1) 3 (0.7) 3 (2.9) 0.926 Clerical 288 (52.4) 236 (52.3) 52 (52.5) Managerial 112 (20.4) 94 (20.8) 18 (18.2) Technical and professional 73 (13.1) 59 (13.1) 14 (14.1) Others 77 (14.0) 62 (13.7) 15 (15.2) Working hours per week 8 (1.4) 3 (0.7) 5 (4.9) 0.850 ≤40 hours 208 (38.0) 172 (38.1) 36 (37.1) ≥41 hours 340 (62.0) 279 (61.9) 61 (62.9) Job stressors Job strain (job demands M=0.48 (0.1) 19 (3.4) M=0.48 (0.1) 15 (3.3) M=0.48 (0.1) 4 (3.9) 0.849 by job control) [0.20–1.00] [0.20–1.00] [0.20–1.00] Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 8 of 14 Table 1 Characteristics of the worksites and employees at baseline (Nj = 23, Ni = 556) (Continued) Psychological determinants Self-regulation for M=24.66 (10.1) 9 (1.6) M=24.56 (10.1) 4 (0.9) M=25.09 (9.9) 5 (4.9) 0.640 physical activity [12–57] [12–57] [12–50, 53, 54] Self-efficacy for M=11.45 (3.6) 4 (0.7) M=11.46 (3.6) 3 (0.7) M=11.44 (3.8) 1 (1.0) 0.954 physical activity [4–20] [4-20] [4–20] Physical activity (METs-hours/week) Occupational M=3.21 (18.7) 5 (0.9) M=3.42 (19.8) 4 (0.9) M=2.31 (12.7) 1 (1.0) 0.590 [0–304] [0–304] [0–124] Transport-related M=10.98 (16.5) 6 (1.1) M=10.72 (16.1) 5 (1.1) M=12.12 (18.4) 1 (1.0) 0.442 [0–144] [0–144] [0–140] Leisure-time M=7.89 (14.6) 5 (0.9) M=8.17 (14.3) 4 (0.9) M=6.64 (15.8) 1 (1.0) 0.344 [0–124] [0–124] [0–124] Nj the number of worksites, Ni the number of employees. different associations with each other. Because the positively associated in the worksite-level adjusted models. worksite-level adjusted model (Model 5) in all three A negative interaction effect between parking/bike and domain-specific physical activities indicated the best self-regulation for physical activity was also significant model fit indices (AIC = 18,629.98, 18,150.18, and (γ = −0.43 [SE = 0.20], p < 0.05). With transport-related 17,980.61, respectively) among the five models, we physical activity (Table 4), self-regulation for physical ac- adopted Model 5 as our conclusive results. With occupa- tivity was not significantly associated (γ = −0.00 tional physical activity (Table 3), self-regulation for phys- [SE = 0.06]). For workplace environments, health promo- ical activity was not significantly associated (γ =0.11 tion programs were significantly and negatively associated [SE = 0.16]). For workplace environments, stairs/elevator (γ = −0.60 [SE = 0.25], p < 0.05). Significant and positive (γ =3.80 [SE =1.80], p < 0.05), physical activity/fitness interaction effects of workplace environments and (γ =4.98 [SE = 1.09], p < 0.01), and written policies self-regulation were observed on physical activity/fitness (γ =2.10 [SE =1.02], p < 0.05) were significantly and facilities (γ =0.06 [SE = 0.03], p <0.05),work rules 06 14 Table 2 Multilevel correlations among worksite- and employee-level variables (Nj = 23, Ni = 547) Variables Mean (SD) ICC [95% CI] 1 2 3 4 5 6 7 8 9 10 11 12 Worksite-level 1. Parking/Bike 1.28 (1.0) – 1.00 −0.16 0.23 −0.22 −0.39 0.36* −0.09 −0.47 0.31 −0.57 −0.14 −0.02 2. Signs/Bulletin boards/ 1.15 (1.2) – 1.00 −0.20 0.01 −0.15 −0.22* 0.49** 0.18 −0.36 −0.19 −0.07 0.02 Advertisements 3. Stairs/Elevator 0.74 (0.6) – 1.00 −0.35* 0.30 0.28* −0.23 −0.12 −0.59 0.10 0.16 0.53 4. Physical activity/ 0.74 (1.7) – 1.00 −0.40** 0.28 0.40 −0.04 0.33 0.22 −0.31 0.73* Fitness facilities 5. Work rules 5.04 (0.8) – 1.00 −0.28 −0.24 −0.07 −0.43 −0.22 −0.56* −0.20 6. Written policies 0.26 (1.3) – 1.00 0.18 0.51 0.32 −0.19 −0.12 0.61** 7. Health promotion 1.43 (1.9) – 1.00 0.22 0.33 −0.02 −0.07 0.02 analys Employee-level 8. Self-regulation 24.66 (10.1) 0.10* [.01, .19] 1.00 0.73 −0.36 0.54** 0.95** for physical activity 9. Self-efficacy for 11.45 (3.6) 0.00 [−0.10, 0.10] 0.41** 1.00 0.27 0.73 0.97** physical activity 10. Occupational 3.38 (17.8) 0.10 [−0.03, 0.23] 0.08 0.05 1.00 −0.25 −0.23 physical activity 11. Transport-related 9.31 (11.9) 0.15* [0.03, 0.26] 0.11** 0.17** 0.14 1.00 0.52* physical activity 12. Leisure-time 7.07 (13.0) 0.04 [−0.03, 0.10] 0.40** 0.34** 0.18 0.09* 1.00 physical activity Upper triangular matrix indicates worksite-level correlations, and lower triangular matrix indicates employee-level matrix. Nj the number of worksites, Ni a b the number of workers, ICC intra-class correlation coefficient, CI confidence interval. At baseline. At follow-up. *p < 0.05. **p < 0.01. Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 9 of 14 Table 3 The associations among workplace environment, self-regulation for physical activity, and occupational physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: occupational Model 1 Model 2 Model 3 Model 4 Model 5 physical activity at follow-up Unconditional Crude random Crude interaction Employee-level Worksite-level a b model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee-level Self-regulation for physical activity (γ ) – 0.23 (0.18) 0.13 (0.14) 0.11 (0.16) 0.11 (0.16) Worksite-level Workplace environment Parking/Bike (γ ) – −0.84 (1.33) −0.83 (1.33) −0.81 (1.20) −1.02 (1.41) Signs/Bulletin boards/Advertisements (γ ) – −0.50 (1.32) −0.49 (1.31) −0.17 (1.21) 0.70 (0.51) Stairs/Elevator (γ ) – 0.16 (2.49) 0.12 (2.50) 0.28 (2.05) 3.80 (1.80)* Physical activity/Fitness facilities (γ ) – 1.68 (1.48) 1.68 (1.48) 1.46 (1.22) 4.98 (1.09)** Work rules (γ ) – 0.34 (2.32) 0.38 (2.39) 0.12 (2.24) −1.25 (1.50) Written policies (γ ) – −0.92 (0.87) −0.89 (0.85) −0.81 (0.83) 2.10 (1.02)* Health promotion programs (γ ) – −0.52 (0.31) −0.53 (0.31) −0.53 (0.29) −0.01 (0.16) Interaction effects Parking × self-regulation (γ ) –– −0.41 (0.21)* −0.42 (0.20)* −0.43 (0.20)* Signs × self-regulation (γ ) –– 0.19 (0.18) 0.24 (0.16) 0.25 (0.16) Stairs × self-regulation (γ ) –– 0.42 (0.37) 0.34 (0.32) 0.35 (0.33) Fitness × self-regulation (γ ) –– 0.07 (0.10) 0.05 (0.09) 0.05 (0.09) Rules × self-regulation (γ ) –– 0.04 (0.32) 0.01 (0.33) 0.00 (0.32) Policies × self-regulation (γ ) –– 0.08 (0.10) 0.12 (0.10) 0.13 (0.10) Programs × self-regulation (γ ) –– −0.10 (0.07) −0.11 (0.08) −0.12 (0.08) Intercept (γ ) 3.90 (1.54)* 3.26 (1.08)** 3.27 (1.08)** 2.87 (0.98)** 7.07 (1.71)** Random intercept (τ ) 31.88 (26.72) 21.48 (12.45) 21.60 (12.37) 16.55 (13.65) 0.10 (1.39) Random slope: self-regulation (τ ) – 0.40 (0.34) 0.21 (0.11) 0.24 (0.09)** 0.25 (0.09)** Residual variance (e ) 289.41 (145.74)* 261.85 (133.85) 261.64 (133.62) 180.65 (99.80) 178.26 (96.23) ij AIC 22,792.08 18,780.05 18,786.13 18,646.91 18,629.98 Nj the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and occupational physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. (γ =0.28 [SE = 0.12], p < 0.05), and written policies Some of the components of workplace environments (γ = 0.04 [SE = 0.01], p < 0.05). With leisure-time phys- could directly increase occupational physical activity. ical activity (Table 5), self-regulation for physical activity However, transport-related and leisure-time physical ac- was significantly and positively associated (γ =0.09 tivity might not be increased by environmental modifica- [SE = 0.04], p < 0.05). Workplace environments did not tion at the workplace alone without combining with have any significant main effect. Positive interaction ef- enhancing self-regulation for physical activity. This may fects of workplace environments and self-regulation mean the necessity of conducting the studies and inter- were observed on physical activity/fitness facilities vention by domains. Another implication of this study is (γ =0.06[SE =0.03], p < 0.05) and written policies that, to promote overall physical activity among (γ =0.06[SE = 0.02], p < 0.05). white-collar workers, designing multi-component inter- ventions that include both of environmental and psycho- Discussion logical approaches might be important. This was the first longitudinal study to report multilevel The hypothesis of the association between workplace effects of workplace environment and self-regulation for environment and physical activity (H1) was supported domain-specific physical activities among white-collar mainly for occupational physical activity, suggesting the workers. The results of each analysis suggested that their direct effect of workplace environment may influence effects may differ among domain-specific activities. physical activity at work. The effectiveness of stairs/ Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 10 of 14 Table 4 The associations among workplace environment, self-regulation for physical activity, and transport-related physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: transport- Model 1 Model 2 Model 3 Model 4 Model 5 related physical activity Unconditional Crude random Crude interaction Employee-level Worksite-level a b at follow-up model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee-level Self-regulation for physical activity (γ ) – 0.13 (0.07) 0.10 (0.02)** −0.00 (0.06) −0.00 (0.06) Worksite-level Workplace environment Parking/Bike (γ ) – −2.98 (1.14)** −3.00 (1.14)** −2.48 (1.18)* −0.61 (0.98) Signs/Bulletin boards/Advertisements (γ ) – −1.00 (0.98) −0.97 (0.98) −0.82 (1.01) 0.59 (0.80) Stairs/Elevator (γ ) – 3.95 (1.96)* 3.93 (1.95)* 3.32 (1.96) 2.80 (2.01) Physical activity/Fitness facilities (γ ) – −1.10 (0.64) −1.11 (0.64) −0.95 (0.70) −0.39 (0.88) Work rules (γ ) – −7.21 (1.97)** −7.21 (1.97)** −6.87 (1.94)** −1.50 (2.89) Written policies (γ ) – −1.44 (0.63)* −1.47 (0.63)* −1.34 (0.65)* −0.72 (1.13) Health promotion programs (γ ) – −0.03 (0.35) −0.02 (0.35) −0.01 (0.30) −0.60 (0.25)* Interaction effects Parking × self-regulation (γ ) –– −0.01 (0.04) 0.07 (0.06) 0.07 (0.06) Signs × self-regulation (γ ) –– 0.07 (0.04) 0.08 (0.05) 0.08 (0.05) Stairs × self-regulation (γ ) –– −0.00 (0.10) 0.01 (0.10) −0.00 (0.11) Fitness × self-regulation (γ ) –– 0.03 (0.02) 0.05 (0.03)* 0.06 (0.03)* Rules × self-regulation (γ ) –– 0.15 (0.07)* 0.26 (0.12)* 0.28 (0.12)* Policies × self-regulation (γ ) –– 0.06 (0.01)** 0.04 (0.01)* 0.04 (0.01)* Programs × self-regulation (γ ) –– −0.02 (0.01) −0.02 (0.01) −0.02 (0.01) Intercept (γ ) 8.48 (1.21)** 10.12 (0.97)** 10.12 (0.98)** 9.37 (1.43)** 11.32 (2.05)** Random intercept (τ ) 23.19 (6.85)** 11.19 (6.22) 11.20 (6.23) 12.53 (5.98)* 0.22 (1.28) Random slope: self-regulation (τ ) – 0.00 (0.02) 0.00 (0.00) 0.00 (0.01) 0.00 (0.01) Residual variance (e ) 103.26 (14.10)** 101.92 (13.89)** 101.29 (13.65)** 80.41 (12.20)** 81.84 (10.47)** ij AIC 22,230.67 18,227.64 18,238.37 18.160.32 18,150.18 Nj the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and transport-related physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. elevator, physical activity/fitness facilities, and written employers for prompts and facilities might amplify the im- policies at the workplace has been repeatedly indicated portance of physical activity for employees, being social, by previous studies [16, 17, 45–48]. Thus, this study may and ecological supports for doing physical activity. add further evidence of the association among these The second hypotheses for self-regulation was sup- workplace environments and occupational physical ac- ported for leisure-time physical activity but not for occu- tivity, even after adjusting for worksite-level covariates pational/transport-related physical activities. A possible (i.e., worksite size and worksite location). Possible reason for the result could be difficulty to utilize their functions of these components might be enhancing regulation for activities in occupational/transport-related opportunities and accessibility for doing physical ac- settings. In most work time, employees must pay atten- tivity [16, 45, 46, 48, 49]. Especially, because using tion to their own jobs, and most activities related to jobs stairs and physical activity/fitness facilities can be directly basically occur incidentally [51]. Therefore, most em- connected to increasing the amount of physical activity at ployees may have difficulty using their psychological work, their associations with occupational physical activity functions for occupational physical activity within work- may be stronger than “soft”’ workplace environment ing time. Transport-related physical activity is also changes, such as mounting signs or establishing work largely influenced by environmental factors at both rules. Another function might be establishing social norms workplaces and communities, and most employees may at the workplace [50]. Policies and investment by establish their commuting habits. Therefore, there may Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 11 of 14 Table 5 The associations among workplace environment, self-regulation for physical activity, and leisure-time physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: leisure-time Model 1 Model 2 Model 3 Model 4 Model 5 physical activity at follow-up Unconditional Crude random Crude interaction Employee-level Worksite-level a b model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee -level Self-regulation for physical activity (γ ) – 0.57 (0.09)** 0.40 (0.04)** 0.10 (0.04)* 0.09 (0.04)* Worksite-level Workplace environment Parking/Bike (γ ) – −1.53 (0.51)** −1.55 (0.52)** −1.29 (0.53)* −1.05 (0.69) Signs/Bulletin boards/Advertisements (γ ) – −0.01 (0.39) −0.00 (0.39) −0.11 (0.34) 0.55 (0.43) Stairs/Elevator (γ ) – 1.31 (1.40) 1.32 (1.41) 1.13 (1.47) −0.18 (1.21) Physical activity/Fitness facilities (γ ) – 0.76 (0.31)* 0.74 (0.31)* 0.79 (0.49) −0.03 (0.66) Work rules (γ ) −2.10 (1.11) −2.13 (1.12) −2.39 (1.39) 0.55 (1.28) Written policies (γ ) – 1.29 (0.25)** 1.20 (0.24)** 0.96 (0.37)* 0.99 (0.59) Health promotion programs (γ ) – 0.16 (0.11) 0.15 (0.11) 0.13 (0.16) −0.01 (0.13) Interaction effects Parking × self-regulation (γ ) –– −0.24 (0.08)** −0.08 (0.07) −0.07 (0.08) Signs × self-regulation (γ ) –– 0.02 (0.04) 0.04 (0.03) 0.04 (0.03) Stairs × self-regulation (γ ) –– 0.23 (0.13) 0.08 (0.13) 0.06 (0.13) Fitness × self-regulation (γ ) –– 0.05 (0.03)* 0.05 (0.03) 0.06 (0.03)* Rules × self-regulation (γ ) −0.21 (0.13) 0.03 (0.12) 0.04 (0.12) Policies × self-regulation (γ ) –– 0.10 (0.02)** 0.06 (0.02)** 0.06 (0.02)** Programs × self-regulation (γ ) –– −0.03 (0.02)* −0.02 (0.01) −0.02 (0.01) Intercept (γ ) 7.05 (0.85)** 7.09 (0.42)** 7.04 (0.44)** 8.53 (1.26)** 8.73 (1.44)** Random intercept (τ ) 6.95 (6.91) 0.16 (0.44) 0.12 (0.71) 1.35 (.2.61) 0.03 (0.20) Random slope: self-regulation (τ ) – 0.06 (0.03) 0.00 (0.02) 0.00 (0.02) 0.00 (0.01) Residual variance (e ) 162.54 (25.71)** 129.94 (21.98)** 127.78 (21.34)** 75.04 (14.59)** 73.79 (13.93)** ij AIC 22,291.35 18,202.44 18,198.14 17,986.55 17,980.61 Nj = the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and leisure-time physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. be few choices to commute and little need to actively are well facilitated at the workplace. For leisure-time phys- regulate their strategies for transport-related activities. ical activity, strategies of regulation are more effective On the other hand, workers could thoroughly utilize when physical activity/fitness (γ =0.06, p < 0.05) and their cognitive and behavioral regulation for physical ac- written policies (γ =0.06, p < 0.05) are well facilitated. tivity in leisure time. Considering that there was a sig- These results have a practical implication for future nificant positive relationship between self-efficacy for multi-component interventions that either environmental physical activity and physical activity in the workday modification or psychological approach alone is insufficient [23, 24], functions of self-efficacy and self-regulation to promote transport-related and leisure-time physical ac- might also differ for each domain of physical activities. tivity among white-collar workers. These workplace envi- It is interesting that the effects of workplace environ- ronments include not only internal physical environments ments may spill over into outside of the worksite through at the workplace but also external and social environments. boosting the associations between self-regulation and Therefore, they might be useful for employees to be more physical activity (H3). For transport-related physical activ- active when employees plan physical activity outside of the ity, strategies of self-regulation (γ = −0.00) are ineffective workplace. For instance, setting a goal to increase themselves (H2) but effective only when physical activity/ transport-related physical activity during lunch break can fitness (γ = 0.06, p <0.05),work rules (γ = 0.28, be more effective if employers set rules that enable em- 14 15 p < 0.05), and written policies (γ = 0.04, p <0.05) ployees to go out or if employers have a written policy that 16 Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 12 of 14 encourages employees to walk around the workplace. An- Conclusion other possible mechanism is that they enhance awareness In summary, the components of workplace environ- and build social norms for physical activity [52]. Internal- ments such as physical activity/fitness facilities, written ized awareness and norms could spill over into one’s policies, signs for stair use at stairs and elevators, and transport-related and leisure-time activities and entire life- work rules and self-regulation for physical activity may style among white-collar workers. Although the question of be effective to promote three domain-specific physical how these components moderate the association between activities directly or by augmenting the positive associa- self-regulation for physical activity and physical activity was tions between self-regulation for physical activity and not the focus of the present study, it should be investigated physical activity. Those effects may differ by activity do- in future studies. In addition, we could only investigate one main; occupational activities may be increased by work- possible model that workplace environment would exagger- place environments while transport-related and ate the effect of self-regulation for physical activity on phys- leisure-time activities may also be increased by the inter- ical activity. The other possible relationships among actions of workplace environments and self-regulation. two-level factors, such as mediation model and subgroup This study has practical implications for designing effects should be investigated in future studies. multi-component interventions that include both of en- Incidentally, some associations were negative and sig- vironmental and psychological approaches to increase nificant: the direct association between health promotion effect sizes to promote overall physical activity. programs and transport-related physical activity and the Abbreviations interaction effect of parking/bike and self-regulation for AIC: Akaike information criteria; EAT: Environmental assessment tool; physical activity on occupational physical activity. These FIML: Full information maximum likelihood; GPAQ: Global physical activity questionnaire; HLM: Hierarchical linear modeling; JCQ: Job content inverse associations were not consistent with our expect- questionnaire; PASR: Physical activity self-regulation; STROBE: Strengthening ation. Although the reasons for negative impact of the reporting of observational studies in epidemiology workplace environments for physical activity were un- known in the present study, some components of work- Acknowledgements We thank Yumi Asai, Risa Kotake, Yoshifumi Nin, Asuka Sakuraya, Utako place environment could be adverse for specific Sawada, and Peiying Tsay for their cooperation in worksite observation to domains. Not only possible benefits but also potential assess workplace environment. adverse effects of workplace environment should be in- vestigated in the future. Funding This work was supported by the Grant-in-Aid for Japan Society for the Promotion There are several limitations to this longitudinal study. of Science Fellows (15J04085). First, the response rate of worksite representatives that agreed to participate was not high (46.9%), which can Availability of data and materials The dataset used and analysed during the current study are available from cause selection bias resulting in an underestimation of the corresponding author on reasonable request. the associations. In addition, participants who were dropped were significantly younger. Although we ad- Authors’ contributions dressed attrition at follow-up by imputation of FIML, it KW had contributions to the initial conception and design. KW did the statistical analysis. KW wrote the first draft manuscript. NK, YO, and SI made may cause underestimation. Second, all employee-level revision of the manuscript. The all authors made critical comments on the variables and some worksite-level variables were mea- draft manuscript and approved its final version. sured by self-report questionnaires. Measured values Ethics approval and consent to participate contained information bias and measurement errors. It Informed consent was obtained from all representatives and workers via a has been repeatedly indicated that self-reported physical consent form and the self-report questionnaire. The consent form and activity was often distorted with actual physical activity questionnaire informed participants that we assured protection of personal information and that the data would be anonymous and only used for and often overestimated [53]. In addition, light physical academic purposes. The study protocol received ethical approval by the activity could not be measured due to the questionnaire. research ethics committee of the Graduate School of Medicine and Faculty of Furthermore, distribution of the three domains of activ- Medicine, The University of Tokyo, Japan (No. 10919). ities was skewed and might not much suitable for HLM. Competing interests Third, the results could be distorted by confounding fac- The authors declare that they have no competing interests. tors that could not be considered in this study, such as other types of job stressors [54] and environmental de- Publisher’sNote terminants outside the workplace. Finally, the samples Springer Nature remains neutral with regard to jurisdictional claims in were not extracted at random and were from a restricted published maps and institutional affiliations. area in Japan. Thus, there are limitations to the Author details generalizability of the results. These limitations need to Department of Mental Health, Graduate School of Medicine, The University be addressed by conducting higher-quality studies and 2 of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Faculty of randomized controlled trials. Human Sciences, University of Tsukuba, 3-29-1, Otsuka, Bunkyo-ku, Tokyo Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 13 of 14 112-0012, Japan. Department of Preventive Medicine and Public Health, 21. Smith L, McCourt O, Sawyer A, Ucci M, Marmot A, Wardle J, et al. A review Tokyo Medical University, 6-1-1 Shinjuku, Shinjuku-ku, Tokyo 160-8402, Japan. of occupational physical activity and sedentary behaviour correlates. Occup Med. 2016;66:185–92. Received: 28 December 2017 Accepted: 20 May 2018 22. Wang F, Orpana HM, Morrison H, de Groh M, Dai S, Luo W. Long-term association between leisure-time physical activity and changes in happiness: analysis of the prospective national population health survey. Am J Epidemiol. 2012;176:1095–100. 23. Prodaniuk TR, Plotnikoff RC, Spence JC, Wilson PM. The influence of self- References efficacy and outcome expectations on the relationship between perceived 1. Kirk AA, Rhodes RE. Occupation correlates of adults’ participation in leisure- environment and physical activity in the workplace. Int J Behav Nutr Phys time physical activity: a systematic review. Am J Prev Med. 2011;40:476–85. Act. 2004;1:7. 2. World Health Organization. Global health risks: Mortality and burden of 24. Plotnikoff RC, Pickering MA, Flaman LM, Spence JC. The role of self-efficacy disease attribute to selected major risks. 2009. http://www.who.int/ on the relationship between the workplace environment and physical healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf activity: a longitudinal mediation analysis. Health Educ Behav. 2010;37:170–85. 3. Joyce S, Modini M, Christensen H, Mykletun A, Bryant R, Mitchell PB, et al. 25. Watanabe K, Otsuka Y, Shimazu A, Kawakami N. The moderating effect of Workplace interventions for common mental disorders: a systematic meta- health-improving workplace environment on promoting physical activity in review. Psychol Med. 2016;46:683–97. white-collar employees: a multi-site longitudinal study using multi-level 4. Brown HE, Gilson ND, Burton NW, Brown WJ. Does physical activity impact structural equation modeling. J Occup Environ Med. 2016;58:178–84. on presenteesim and other indicators of workplace well-being? Sports Med. 26. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, 2011;41:249–62. et al. Strengthening the Reporting of Observational Studies in Epidemiology 5. Proper KI, Staal BJ, Hildebrandt VH, van der Beek AJ, van Mechelen W. (STROBE): explanation and elaboration. PLos Med. 2007;e297:4. Effectiveness of physical activity programs at worksites with respect to 27. DeJoy DM, Wilson MG, Goetzel RZ, Ozminkowskim RJ, Wang S, Barker KM, et work-related outcomes. Scand J Work Environ Health. 2002;28:75–84. al. Development of the Environmental Assessment Tool (EAT) to measure 6. van den Berg TI, Elders LAM, de Zwart BCH, Burdorf A. The effects of work- organizational physical and social support for worksite obesity prevention related and individual factors on the Work Ability Index: a systematic programs. J Occup Environ Med. 2008;50:126–37. review. Occup Environ Med. 2009;66:211–20. 28. Parker KB, DeJoy DM, Wilson MG, Bowen HM, Goetzel RZ. Application of the 7. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJ, Martin BW. The Lancet environmental assessment tool (EAT) as a process measure for a worksite Physical Activity Series Working Group. Correlates of physical activity: why weight management intervention. J Occup Environ Med. 2010;52:S42–51. are some people physically active and others not? Lancet. 2012;380:258–71. 29. Matson-Koffman DM, Brownstein JN, Neiner JA, Greaney ML. A site-specific 8. Sallis JF, Owen N, Fisher EB. Ecological models of health behavior. In: Glanz literature review of policy and environmental interventions that promote K, Rimer BK, Viswanath K, editors. Health behavior and health education: physical activity and nutrition for cardiovascular health: what works? Am J theory, research, and practice. 4th ed. New York: Jossey-Bass A Wiley Health Promot. 2005;19:167–93. Imprint; 2008. p. 465–85. 30. Mozaffarian D, Afshin A, Benowitz NL, Bittner V, Daniels SR, Franch HA, et al. 9. Bandura A. Self-efficacy: toward a unifying theory of self-change. Psychol Population approaches to improve diet, physical activity, and smoking Rev. 1977;84:191–215. habits: a scientific statement from the American Heart Association. 10. Bandura A. Social foundations of thought and action: a social cognitive Circulation. 2012;126:1514–63. theory. Englewood Cliffs NJ: Prentice-Hall; 1986. 31. Schröer S, Haupt J, Pieper C. Evidence-based lifestyle interventions in the 11. Prochaska JO, DiClemente CC. Stages and processes of self-change of workplace: an overview. Occup Med. 2014;64:8–12. smoking: toward an integrative model of change. J Consult Clin Psychol. 32. World Health Organization. Interventions on diet and physical activity: What 1983;51:390–5. works: Summary report. 2009. http://www.who.int/dietphysicalactivity/ 12. Rhodes RE, Pfaeffli LA. Mediators of physical activity behavior change summary-report-09.pdf among non-clinical populations: a review update. Int J Behav Nutr Phys Act. 33. Watanabe K, Kawakami N, Adachi H, Inoue S, Umstattd MR. Internal 2010;7:37. https://doi.org/10.1186/1479-5868-7-37. consistency, convergent validity, and structural validity of the Japanese 13. Sallis JF, Spoon C, Cavill N, Engelberg JK, Gebel K, Parker M, et al. Co- version of the Physical Activity Self-Regulation scale (PASR-12) among benefits of designing communities for active living: an exploration of Japanese workers: a validation study. J Occup Health. 2017;59:24–32. literature. Int J Behav Nutr Phys Act. 2015;12:30. https://doi.org/10.1186/ 34. Umstattd MR, Motl R, Wilcox S, Saunders R, Watford M. Measuring physical s12966-015-0188-2. activity self-regulation strategies in older adults. J Phys Act Health. 2009;6: 14. Hutchinson AD, Wilson C. Improving nutrition and physical activity in the S105–12. workplace: a meta-analysis of intervention studies. Health Promot Int. 2011; 35. World Health Organization. Global physical activity surveillance. http://www. 27:238–49. who.int/chp/steps/GPAQ/en/ 15. Hipp JA, Reeds DN, van Bakergem MA, Marx CM, Brownson RC, Pamulapati 36. Bull FC, Maslin TS, Armstrong T. Global physical activity questionnaire SC, et al. Review of measures of worksite environmental and policy supports (GPAQ): nine country reliability and validity study. J Phys Act Health. 2009;6: for physical activity and healthy eating. Prev Chron Dis. 2015;12:E65. https:// 790–804. doi.org/10.5888/pcd12.140410. 37. World Health Organization. Global Physical Activity Questionnaire (GPAQ): 16. Lin YP, McCullagh MC, Kao TS, Larson JL. An integrative review: work Analysis guide. http://www.who.int/chp/steps/resources/GPAQ_Analysis_ environment factors associated with physical activity among white-collar Guide.pdf workers. West J Nurs Res. 2014;36:262–83. 17. Soler RE, Leeks KD, Buchanan LR, Brownson RC, Health GW, Hopkins DH. 38. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and stages of exercise Task Force on Community Preventive Services. Point-of decision prompts to behavior change. Res Q Exerc Sport. 1992;63:60–6. increase stair use: a systematic review update. Am J Prev Med. 2010;38: 39. Oka K. Stages of change for exercise behavior and self-efficacy for exercise S292–300. among middle-aged adults. Jap J Public Health. 2003;50:208–15. Japanese 40. Kawakami N, Kobayashi F, Araki S, Haratani T, Furui T. Assessment of job 18. Choi B, Schnall PL, Yang H, Dobson M, Landsbergis P, Israel L, et al. stress dimensions based on the job demands-control model of employees Psychosocial working conditions and active leisure-time physical activity in of telecommunication and electric power companies in Japan: reliability middle-aged us workers. Int J Occup Environ Health. 2010;23:239–53. and validity of the Japanese version of the Job Content Questionnaire. Int J 19. Crespo NC, Sallis JF, Conway TL, Saelens BE, Frank LD. Worksite physical Behav Med. 1995;2:358–75. activity policies and environments in relation to employee physical activity. 41. Karasek R. Job Content Questionnaire and user’s guide. Lowell, MA: Am J Health Promot. 2011;25:264–71. University of Massachusetts at Lowell; 1985. 20. Fransson EI, Heikilä K, Nyberg ST, Zins M, Westerlund H, Westerholm P, et al. Job strain as a risk factor for leisure-time physical inactivity: an individual- 42. Yasuda M, Nakazawa K. Consideration of the color-form problem in a two- participant meta-analysis of up to 170,000 men and women: The IPD-Work dimensional oblique model. Jap J Personal. 2012;20:271–328. Japanese th Consortium. Am J Epidemiol. 2012;176:1078–89. https://doi.org/10.1093/aje/ 43. Muthén LK, Muthén BO. Mplus user’sguide (7 ed.). Los Angeles, CA: Muthén kws336. & Muthén; 1998–2015. https://www.statmodel.com/ugexcerpts.shtml Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 14 of 14 44. International Labour Office. Hours of work: From fixed to flexible? Geneva: International Labour Office; 2005. http://www.ilo.org/public/english/ standards/relm/ilc/ilc93/pdf/rep-iii-1b.pdf 45. Almeida FA, Wall SS, You W, Harden SM, Hill JL, Krippendorf BE, et al. The association between worksite physical environment and employee nutrition and physical activity behavior and weight status. J Occup Environ Med. 2014;56:779–84. 46. Dodson EA, Lovegreen SL, Elliott MB, Haire-Hoshu D, Brownson RC. Worksite policies and environments supporting physical activity in midwestern communities. Am J Health Promot. 2008;23:51–5. 47. Kolbe-Alexander T, Greyling M, da Silva R, Milner K, Patel D, Wyper L, et al. The relationship between workplace environment and employee health behaviors in a South African workforce. J Occup Environ Med. 2014;56: 1094–9. 48. Watts AW, Masse LCI. access to workplace amenities associated with leisure- time physical activity among Canadian adults? Can J Public Health. 2013; 104:e87–91. 49. Engbers LH, van Poppel MN, Chin A, Paw MJ, van Mechelen W. Worksite health promotion programs with environmental changes: a systematic review. Am J Prev Med. 2005;29:61–70. 50. Tabak RG, Hipp JA, Marx CM, Brownson RC. Workplace social and organizational environments and healthy weight behaviors. PLoS One. 2015;10:e0125424. https://doi.org/10.1371/journal.pone.0125424. 51. Engelen L, Dhillon HM, Chau JY, Hespe D, Bauman AE. Do active design buildings change health behaviour and workplace perceptions? Occup Med (Lond). 2016;66:408–11. 52. Beresford SA, Bishop SK, Brunner NL, Duncan GE, McGregor DF, McLerran DF, et al. Environmental assessment at worksites after a multilevel intervention to promote activity and changes in eating: The PACE project. J Occup Environ Med. 2010;52:S22–8. 53. Watson ED, Micklesfield LK, van Poppel MNM, Norris SA, Sattler MC, Dietz P. Validity and responsiveness of the Global Physical Activity Questionnaire (GPAQ) in assessing physical activity during pregnancy. PLoS One. 2017;12: e0177996. https://doi.org/10.1371/journal.pone.0177996. 54. Kouvonen A, Vahtera J, Oksanen T, Pentti J, Väänänen AKP, Heponiemi T, et al. Chronic workplace stress and insufficient physical activity: a cohort study. Occup Environ Med. 2013;70:3–8. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Behavioral Nutrition and Physical Activity Springer Journals

Associations among workplace environment, self-regulation, and domain-specific physical activities among white-collar workers: a multilevel longitudinal study

Free
14 pages

Loading next page...
 
/lp/springer_journal/associations-among-workplace-environment-self-regulation-and-domain-C7ssj7MUEX
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Medicine & Public Health; Clinical Nutrition; Behavioral Sciences; Health Promotion and Disease Prevention
eISSN
1479-5868
D.O.I.
10.1186/s12966-018-0681-5
Publisher site
See Article on Publisher Site

Abstract

Background: Psychological and environmental determinants have been discussed for promoting physical activity among workers. However, few studies have investigated effects of both workplace environment and psychological determinants on physical activity. It is also unknown which domains of physical activities are promoted by these determinants. This study aimed to investigate main and interaction effects of workplace environment and individual self-regulation for physical activity on domain-specific physical activities among white-collar workers. Methods: A multi-site longitudinal study was conducted at baseline and about 5-month follow-up. A total of 49 worksites and employees within the worksites were recruited. Inclusion criteria for the worksites (a) were located in the Kanto area, Japan and (b) employed two or more employees. Employee inclusion criteria were (a) employed by the worksites, (b) aged 18 years or older, and (c) white-collar workers. For outcomes, three domain-specific physical activities (occupational, transport-related, and leisure-time) at baseline and follow-up were measured. For independent variables, self-regulation for physical activity, workplace environments (parking/bike, signs/bulletin boards/advertisements, stairs/elevators, physical activity/fitness facilities, work rules, written policies, and health promotion programs), and covariates at baseline were measured. Hierarchical Linear Modeling was conducted to investigate multilevel associations. Results: Of the recruited worksites, 23 worksites and 562 employees, and 22 worksites and 459 employees completed the baseline and the follow-up surveys. As results of Hierarchical Linear Modeling, stairs/elevator (γ=3.80 [SE=1.80], p<0.05), physical activity/fitness facilities (γ=4.98 [SE=1.09], p<0.01), and written policies (γ=2.10 [SE=1.02], p<0.05) were significantly and positively associated with occupational physical activity. Self-regulation for physical activity was associated significantly with leisure-time physical activity (γ=0.09 [SE=0.04], p<0.05) but insignificantly with occupational and transport-related physical activity (γ=0.11 [SE=0.16] and γ=−0.00 [SE=0.06]). Significant interaction effects of workplace environments (physical activity/fitness facilities, work rules, and written policies) and self-regulation were observed on transport-related and leisure-time physical activity. (Continued on next page) * Correspondence: kzwatanabe-tky@umin.ac.jp Department of Mental Health, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 2 of 14 (Continued from previous page) Conclusions: Workplace environments such as physical activity/fitness facilities, written policies, work rules, and signs for stair use at stairs and elevators; self-regulation for physical activity; and their interactions may be effective to promote three domain-specific physical activities. This study has practical implications for designing multi-component interventions that include both environmental and psychological approaches to increase effect sizes to promote overall physical activity. Keywords: Physical activity, Workplace environment, Multilevel study Background entire lifestyle [18–20]. Thus, workplace environment is The need for promotion of physical activity had been also an important determinant for physical activity among strongly recognized among workers since rapid changes workers. to the modern labor market have resulted in a large in- However, few studies have investigated effects of both crease of workers engaged in sedentary behavior and workplace environment and psychological determinants low-activity occupations [1]. Physical activity is one of on physical activity among workers using a multilevel the most important health behaviors in public health [2], study design, and no study has dealt with self-regulation and promoting this behavior is one of the most evident for physical activity. It is also unknown which domains workplace interventions for primary prevention of com- of physical activities are promoted by workplace envir- mon mental disorders [3]. Moreover, significant asso- onment and psychological determinants, since most ciations between physical activity and work-related studies measured only overall physical activity or one outcomes have repeatedly been reported, such as specific domain of activities. It is important to stratify well-being and presenteeism [4], absenteeism, job stress, the domains of physical activity because each activity employee turnover [5], and work ability [6]. These find- could arise in different contexts, could be caused by dif- ings indicate that promoting workers’ physical activity is ferent determinants [21], and could influence different indispensable for occupational health promotion and a outcomes [22]. Previous observational studies among sustainable workforce. three large worksites in Canada indicated that To promote physical activity among workers, multilevel self-efficacy for physical activity mediated the relation- factors across different levels have been suggested as de- ship between perceived workplace environment and terminants and targets for intervention [7, 8]. Of the psy- physical activity in the workday [23, 24]. However, in chological determinants, self-efficacy and self-regulation these studies, physical activity was measured only during for physical activity are recognized as important factors in the workday. Furthermore, they analysed the associa- theoretical models [9–11]. Especially, self-regulation (e.g., tions using simple regression as opposed to multilevel planning, scheduling, and self-organisational behaviors) analyses and did not address self-regulation. Another has recently been indicated to be strongest mediator be- previous study was a multilevel longitudinal study tween physical activity interventions and behavioral among 16 worksites and 129 employees in Japan [25], changes in physical activity in a non-clinical adult popula- which showed that employers providing fitness facilities tion [12]. Environmental determinants have also been dis- increased the positive association between self-efficacy cussed for promoting physical activity [13]. Workplace for physical activity and overall physical activity. How- environment can largely influence activities among ever, they did not measure specific domains of activities workers [14]. A workplace environment for promoting and did not address self-regulation either. For a clearer physical activity typically includes organising assessments/ understanding of how workplace environment, psycho- counseling/educations, informational support (e.g., post- social determinants, and physical activity interact, fur- ers/flyers/bulletin boards/maps), organisational policy, in- ther studies are needed. ternal physical environment (e.g., physical activity In this study, we aimed to investigate the associations equipment/stairs/lockers/showers/office connectivity), among workplace environment, self-regulation for phys- co-workers’ social support, and external environment (e.g., ical activity, and three domain-specific physical activ- walkability/parking/active commuting/physical activity fa- ities—occupational, transport-related, and leisure-time— cilities outside the workplace) [15–17]. These environ- in white-collar workers using a multi-worksite longitu- ments might be effective for promoting not only physical dinal design. White-collar workers are mainly engaged activities during work but also activities out of work (e.g., in sedentary work and are a primary target among the transport-related and leisure-time). Several observational working population [19]. Our hypotheses were as fol- studies suggested that the effects of workplace environ- lows: (H1) several aspects of supportive workplace envir- ment spilled over to leisure-time physical activity and onment will be positively associated with each domain Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 3 of 14 of physical activity among white-collar workers; (H2) Measures self-regulation for physical activity will be positively as- We measured worksite- and employee-level variables at sociated with higher scores in each domain of physical the baseline survey and measured physical activity both activity among white-collar workers; (H3) two-level at the baseline and the follow-up survey. Worksite-level interaction effects of workplace environment and variables were measured by a validated measurement, self-regulation for physical activity will be positively as- which consisted of observations by two independent sociated with higher scores in each domain of physical raters from the research team and a self-report question- activity among white-collar workers. This means that naire for worksite representatives (usually human re- the positive association between self-regulation of source personnel). Employee-level variables were white-collar workers and higher scores in each domain measured by a self-report questionnaire distributed to of physical activity will be stronger under well-facilitated workers. workplace environments. Workplace environment Methods Workplace environment to promote physical activity Study design and setting was measured by the Japanese version of the Environ- This was a multi-site longitudinal study. Multilevel mental Assessment Tool (EAT) [25, 27]. In this study, nested data were collected twice—at baseline (Oct–Dec workplace environment was operationally defined as 2015) and at follow-up (Feb–Apr 2016) —from work- EAT scores. Higher scores on the EAT indicate a more sites and workers employed by the worksites. We supportive environment for physical activity promotion approached worksites in the Kanto area, the economic and more invested environment by employers [28]. The center of Japan including metropolitan Tokyo, through concurrent validity with the rate of health abnormality some of the health insurance associations and chambers and test–retest reliability of the EAT were confirmed in of commerce in the area, using snowball sampling previous studies in both the US and Japan [25, 28]. In methods. Each worksite was explained the study in detail this study, subordinate components of the EAT that and asked to join the study. After the worksite represen- were used in the previous study [25] and suggested posi- tatives agreed to partake in the study, nested employees tive association with physical activity from the literature were recruited. Informed consent was obtained from all review [16, 17, 29–32] were used: parking/bike (4 points; representatives and workers via a consent form and the e.g., bike rack spaces), signs/bulletin boards/advertise- self-report questionnaire. The consent form and ques- ments (4 points; e.g., signs with physical activity mes- tionnaire informed participants that we assured protec- sages), stairs/elevator (4 points; e.g., signs encouraging tion of personal information and that the data would be stair use at building entrance or at elevators), physical anonymous and only used for academic purposes. The activity/fitness facilities (14 points; e.g., onsite fitness fa- study protocol received ethical approval by the research cilities), work rules (6 points; e.g., own lockers em- ethics committee of the Graduate School of Medicine ployees can have), written policies (6 points; e.g., written and Faculty of Medicine, The University of Tokyo, Japan policies supporting employees’ physical activity), and (No. 10919). Our study has been reported according to health promotion programs (20 points; e.g., education the Strengthening the Reporting of Observational studies classes for physical activity). Scores of health promotion in Epidemiology (STROBE) guidelines [26]. programs for physical activity, diet/nutrition, and weight management were combined into one variable to avoid Participants complexity in models. These scores were calculated by At baseline, a total of 49 representatives of worksites using the EAT scoring system based on a were provided with an explanation of the study and representative-reported questionnaire (section 1) and an asked to participate by the corresponding author. Work- observation form completed by researchers (section 2). site inclusion criteria (a) were located in the Kanto area, Scores of the observation (section 2) were rated by two the economic center of Japan including metropolitan of seven trained members from our research team for Tokyo, and (b) employed two or more employees. There each worksite and averaged. The research team members were no exclusion criteria for worksites. Within the were graduate students in psychology, nursing, public worksites which agreed to partake in the study, nested health, and health science. Inter-rater reliability of the employees were recruited. Employee inclusion criteria EAT scores ranged from to .46 to 1.00, considered suffi- were (a) employed by the worksites, (b) aged 18 years or cient values. older, and (c) white-collar workers (managerial, profes- sional, technical clerical, and other job types that re- Self-regulation for physical activity quired deskwork or sitting work). There were no specific Self-regulation for physical activity was measured by the exclusion criteria. Japanese version of the 12-item Physical Activity Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 4 of 14 Self-Regulation scale (PASR-12) [33, 34]. The PASR-12 Self-efficacy was measured using a scale developed by asked the workers how often they had utilized cognitive Marcus et al. [38]and Oka [39]. Because the original scale and behavioral methods for physical activity in the past was developed to assess self-efficacy for exercise, we re- four weeks (e.g., “I mentally kept track of my physical vised the word “exercise” to “physical activity.” The scale activity”). The 12-item measure consists of six factors: consists of four items (e.g., “I have the confidence to per- self-monitoring, goal-setting, eliciting social support, re- form physical activity even if I am a little tired”). All inforcements, time management, and relapse prevention. items are rated on a 5-point Likert-type scale (1 = Not All items are rated on a 5-point Likert scale (1 = Never, at All,5= Almost). Internal consistency and unidimen- 5= Very Often). Internal consistency, convergent valid- sional structural validity were confirmed in a previous ity, and structural validity of the Japanese version of the study [39]. In the present study, the total score of PASR-12 were confirmed in a previous study [33]. In the self-efficacy for physical activity was tripled to match present study, the total scale score was used in analyses its score range to that of self-regulation for physical ac- to avoid complexity. Cronbach’s alpha for the scale at tivity (12–60) to compare the effect sizes on physical the baseline survey was .93, considered an excellent activity in multilevel models. Cronbach’s alpha for the value. scale at baseline survey was .87, considered an excellent value. Physical activity Additionally, job strain as determinants for physical Three domain-specific physical activities were measured activity was assessed. Job strain was measured by the by the Japanese version of the Global Physical Activity Japanese version of the Job Content Questionnaire Questionnaire (GPAQ v2) [35]. This scale is widely used 22-item version (JCQ) [40, 41]. The JCQ includes five and has demonstrated reliability and convergent validity items for job demands (e.g., “My job requires working in nine countries, including Japan [36]. The test-retest very fast”) and nine items for job control (e.g., “I have a reliability in each domain was moderate to strong both lot of say about what happens on my job”). All items are in the validation study (Spearman’ rho 0.67 to 0.81) [36] rated on a 4-point Likert-type scale (1 = Strongly dis- and in this study (Intra-Class Correlation 0.66 to 0.84). agree,4= Strongly agree). The scale has been widely The GPAQ can assess occupational, transport-related, used to assess job strain, and its reliability and validity and leisure-time physical activity with regard to intensity were confirmed by a previous study [40]. Cronbach’s al- of the activity (moderate-to-vigorous) and time spent phas for job demands and job control at the baseline doing the activity (minutes, hours, days) based on a typ- survey were .65 and .75, respectively. According to the ical week. In the present study, the amounts of three user’s guide for the JCQ [41], we calculated continuous domain-specific physical activities per week (MET- scores of job strain by the ratio of job demands to job s-hours/week) were calculated according to the GPAQ control. analysis guide [37]. Participants who responded with in- consistent answers or implausible values were excluded Analyses from all analyses in the study (e.g., more than seven days Descriptive statistics, intra-class correlation coefficients in any column of days spent doing the activity per for employee-level variables, and multilevel correlation week). coefficients for two-level variables were calculated. For the main analysis, Hierarchical Linear Modeling (HLM) Covariates was conducted to investigate multilevel relationships Worksite-level confounders included worksite size and among workplace environment, self-regulation for phys- worksite location (urban, suburban, and local). Worksite ical activity, and physical activity in each domain. Three size was operationalized as an ordinal variable and clas- domain-specific physical activities at follow-up were en- sified into four categories (10–49, 50–99, 100–299, and tered into the models as the dependent variables. Work- ≥300 employees). Worksite location was measured by a place environment, self-regulation for physical activity at single ordinal item (urban, suburban, and rural) for baseline, and interaction effects of workplace environ- worksite representatives, “Where is the worksite ment and self-regulation for physical activity were en- located?” tered as independent variables. Worksite size, worksite Self-efficacy for physical activity as another psycho- location, self-efficacy for physical activity, gender, age, logical determinant, gender, age, employment status employment status, job type, working hours, and job (regular, part-time, dispatched, contract, and others), oc- strain at baseline were controlled as the covariates. In cupational status (managerial, technical and professional, addition, physical activities at baseline were also con- clerical, and others), and working hours per week (1–34, trolled for to analyze these associations longitudinally. 35–40, 41–50, 51–60, 61–65, 66–70, and ≥71 hours) The categorical covariates were transformed into were measured as employee-level confounders. dummy variables: worksite size (10–49 [reference Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 5 of 14 group]), worksite location (urban [reference group]), missing variables, the worksites and employees that par- gender (men [reference group]), employment status tially had missing values or that dropped out during (regular [reference group]), occupational status (not follow-up were included in the analysed model. We de- managerial [reference group]), and working hours per scribe the number of missing samples on all variables week (≤ 40 hours [reference group]). Of the continuous used in the analyses in descriptive statistics. Mplus ver- variables, worksite-level variables (workplace environ- sion 7.4 [43] (Muthén & Muthén, 1998–2015) was used ment) were grand-mean centered, and employee-level for each analysis. variables (self-regulation and self-efficacy for physical ac- We treated three domain-specific physical activities as tivity, age, job strain, and each domain of physical activ- the continuous values (METs-hours/week). Although ity at baseline) were group-mean centered for the distribution of these scores was right-skewed and might multilevel analysis to make the correlations between not be suitable assuming normal distribution, estimating two-level variables equal to zero [42]. cross-level interaction effects between individual and We estimated an unconditional model (model 1), a workplace was so complex in the multi-level model; im- crude random slope model (model 2), a crude inter- proper solutions were observed when we handled the action model (model 3), an employee-level adjusted outcomes as ordinal values. In addition, ordinal out- model (model 4), and a worksite-level adjusted model comes were not applied in our model because we ap- (model 5) in HLM and referred to the Akaike Informa- plied FIML. tion Criteria (AIC) as a fit index. The equation for the adjusted model (model 5) was explained as follows. Results Level 1 (employee-level) Figure 1 shows a participation flow chart of the work- sites and the employees in the study. Of 49 representa- YðÞ each domain of physical activtiy at follow−up ij tives of the worksites, 23 representatives agreed and ¼ β þ β ðÞ self −regulation þ β signed a consent form (response rate = 46.9%). Of the 0 j 1 j ij 2 j worksites and employees, 23 worksite representatives ðÞ self −efficacy þ β ðÞ gender þ β ij 3 j ij 4 j and 562 employees (265 men, 293 women, and 4 un- ðÞ age þ β ðÞ Not regular employment þ β ij 5 j ij 6 j known; mean age = 43.5, SD = 11.1) completed the base- line survey (response rate = 87.8 %). The average ðÞ Managerial job þ β ij 7 j number of workers who completed the baseline survey ðÞ ≧41 working hours per week þ β 8 j ij from each worksite was 24.4 (SD = 22.3). At the ðÞ job strain þ β ij 9 j follow-up, 22 worksite representatives and 459 em- ployees (227 men, 229 women, and 3 unknown; mean ðÞ each domain of physical activity at baseline þ e ij ij age = 44.7, SD = 11.1) completed the survey (response Level 2 (worksite-level) rate = 71.7 %). One worksite and 103 workers refused or dropped out during follow-up because of being too busy β ¼ γ þ γ ðÞ parking þ γ ðÞ signs þ γ 0 j 00 01 j 02 j 03 or unable to link. ðÞ stairs þ γ ðÞ fitness facilities þ γ j 04 j 05 Table 1 shows characteristics of the worksites and em- ðÞ work rules þ γ ðÞ policies þ γ j 06 j 07 ployees at baseline (23 worksites and 556 employees). Of ðÞ health promotion programs þ γ 562 employees, 6 were excluded due to inconsistent an- j 08 swers on the questionnaire to measure physical activity ðÞ worksite scale 50  99 þ γ j 09 [35, 37]. Among the worksites, 12 (52.2%) employed 10– ðÞ worksite scale 100  299 þ γ j 010 49 workers, classified as small-size worksites. The other ðÞ worksite scale≥300 þ γ j 011 11 worksites employed 50–99 (6), 100–299 (3), and ðÞ local or suburban area þ μ j 0 j more than 300 workers (2). Most of the worksites (60.9%) were located in an urban area (e.g., 23 special β ¼ γ þ γ ðÞ parking þ γ ðÞ signs þ γ 1 j 10 11 j 12 j 13 wards in Tokyo) in the Kanto region of Japan. Types of ðÞ stairs þ γ ðÞ fitness facilities þ γ j 14 j 15 industries were service (nine worksites), manufacturing ðÞ work rules þ γ ðÞ policies þ γ j 16 j 17 (four), medical and welfare (three), information and ðÞ health promotion programs þ μ j 1 j communication (three), education (two), wholesale and retail (one), and transportation (one). Of the employees, 0 τ τ oj 00 01 β ¼ γðÞ q ¼ 2…9  N most (69.8%) were employed full-time. About half qj q0 μ 0 τ τ 1 j 10 11 (52.4%) of the employees were engaged in clerical jobs, Because the Full Information Maximum Likelihood 112 (20.4%) mainly had managerial jobs, and 73 (13.1%) (FIML) method was used for parameter estimation to were engaged in technical and professional jobs. A total avoid a selection bias due to dropout and to impute of 340 (62.0%) employees worked over 40 hours per Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 6 of 14 Fig. 1 A participation flow chart of the worksites and the workers. Note. Nj = the number of worksites; Ni = the number of workers week, considered long-working workers [44]. Between with self-regulation and self-efficacy for physical activity employees who completed the follow-up survey and who (γ = 0.40, p <0.01 and γ =0.34, p <0.01),and dropped out, dropped-out employees were significantly transport-related physical activity was also positively associ- younger than the completers (p = 0.001). There was no ated with them (γ = 0.11, p <0.01 and γ =0.17, p < 0.01). other significant difference between two groups for However, occupational physical activity was not signifi- demographic, exposure, and outcome variables. cantly associated with them. For the worksite-level vari- After excluding nine participants due to missing data on ables, scores for each workplace environment were all independent variables, multilevel correlation coefficients inconsistently associated with each other. Of those vari- among workplace environment, self-regulation and ables, work rules were significantly and negatively associ- self-efficacy for physical activity, and each domain of phys- ated with transport-related physical activity (γ = −0.56, ical activity among 547 employees are shown in Table 2. p < 0.05), and physical activity/fitness facilities and Intra-class correlation coefficients for domain-specific phys- written policies were significantly and positively associated ical activities ranged from 0.04 to 0.15, indicating 4–15% of with leisure-time physical activity (γ = 0.73, p <0.05 the variances were explained by these worksite-level vari- and γ = 0.61, p <0.05). ables. For the employee-level psychological determinants, Tables 3, 4, and 5 show the main results of HLM on leisure-time physical activity was most strongly associated domain-specific physical activity at follow-up, indicating Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 7 of 14 Table 1 Characteristics of the worksites and employees at baseline (Nj = 23, Ni = 556) Worksite-level variables (Nj = 23) N (%) Min–Max Mean (SD) Missing (%) Worksite size 0 (0.0) 10–49 12 (52.2) –– 50–99 6 (26.1) –– 100–299 3 (13.0) –– ≥300 2 (8.7) –– Worksite location –– 0 (0.0) Urban 14 (60.9) –– Local/suburban 9 (39.1) –– Workplace environment 0 (0.0) Parking/Bike – 0.00–3.00 1.28 (1.0) Signs/Bulletin boards/ – 0.00–4.00 1.15 (1.2) Advertisements Stairs/Elevator – 0.00–2.00 0.74 (0.6) Physical activity/ – 0.00–7.00 0.74 (1.7) Fitness facilities Work rules 3.00–6.00 5.04 (0.8) Written policies – 0.00–6.00 0.26 (1.3) Health promotion – 0.00–6.00 1.43 (1.9) programs Employee-level Total (Ni = 556) Follow-up completer Drop-out p for difference variables (Ni = 454) (Ni = 102) N (%) Missing (%) N (%) Missing (%) N (%) Missing (%) Mean (SD) [Min–Max] Mean (SD) Mean (SD) [Min–Max] [Min–Max] Demographic variables Gender 4 (0.7) 2 (0.4) 2 (2.0) 0.091 Male 263 (47.6) 223 (49.3) 40 (40.0) Female 289 (52.4) 229 (50.7) 60 (60.0) Age M=43.56 (11.1) 12 (2.2) M=44.27 (11.1) 6 (1.3) M=40.21 (10.2) 6 (5.9) 0.001 [19–84] [19–84] [21–65] Employment Status 3 (0.5) 0 (0.0) 3 (2.9) 0.980 Regular 386 (69.8) 317 (69.8) 69 (69.7) Non-regular (Part-time, 167 (30.2) 137 (30.2) 30 (30.3) contract, dispatched) Occupational status 6 (1.1) 3 (0.7) 3 (2.9) 0.926 Clerical 288 (52.4) 236 (52.3) 52 (52.5) Managerial 112 (20.4) 94 (20.8) 18 (18.2) Technical and professional 73 (13.1) 59 (13.1) 14 (14.1) Others 77 (14.0) 62 (13.7) 15 (15.2) Working hours per week 8 (1.4) 3 (0.7) 5 (4.9) 0.850 ≤40 hours 208 (38.0) 172 (38.1) 36 (37.1) ≥41 hours 340 (62.0) 279 (61.9) 61 (62.9) Job stressors Job strain (job demands M=0.48 (0.1) 19 (3.4) M=0.48 (0.1) 15 (3.3) M=0.48 (0.1) 4 (3.9) 0.849 by job control) [0.20–1.00] [0.20–1.00] [0.20–1.00] Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 8 of 14 Table 1 Characteristics of the worksites and employees at baseline (Nj = 23, Ni = 556) (Continued) Psychological determinants Self-regulation for M=24.66 (10.1) 9 (1.6) M=24.56 (10.1) 4 (0.9) M=25.09 (9.9) 5 (4.9) 0.640 physical activity [12–57] [12–57] [12–50, 53, 54] Self-efficacy for M=11.45 (3.6) 4 (0.7) M=11.46 (3.6) 3 (0.7) M=11.44 (3.8) 1 (1.0) 0.954 physical activity [4–20] [4-20] [4–20] Physical activity (METs-hours/week) Occupational M=3.21 (18.7) 5 (0.9) M=3.42 (19.8) 4 (0.9) M=2.31 (12.7) 1 (1.0) 0.590 [0–304] [0–304] [0–124] Transport-related M=10.98 (16.5) 6 (1.1) M=10.72 (16.1) 5 (1.1) M=12.12 (18.4) 1 (1.0) 0.442 [0–144] [0–144] [0–140] Leisure-time M=7.89 (14.6) 5 (0.9) M=8.17 (14.3) 4 (0.9) M=6.64 (15.8) 1 (1.0) 0.344 [0–124] [0–124] [0–124] Nj the number of worksites, Ni the number of employees. different associations with each other. Because the positively associated in the worksite-level adjusted models. worksite-level adjusted model (Model 5) in all three A negative interaction effect between parking/bike and domain-specific physical activities indicated the best self-regulation for physical activity was also significant model fit indices (AIC = 18,629.98, 18,150.18, and (γ = −0.43 [SE = 0.20], p < 0.05). With transport-related 17,980.61, respectively) among the five models, we physical activity (Table 4), self-regulation for physical ac- adopted Model 5 as our conclusive results. With occupa- tivity was not significantly associated (γ = −0.00 tional physical activity (Table 3), self-regulation for phys- [SE = 0.06]). For workplace environments, health promo- ical activity was not significantly associated (γ =0.11 tion programs were significantly and negatively associated [SE = 0.16]). For workplace environments, stairs/elevator (γ = −0.60 [SE = 0.25], p < 0.05). Significant and positive (γ =3.80 [SE =1.80], p < 0.05), physical activity/fitness interaction effects of workplace environments and (γ =4.98 [SE = 1.09], p < 0.01), and written policies self-regulation were observed on physical activity/fitness (γ =2.10 [SE =1.02], p < 0.05) were significantly and facilities (γ =0.06 [SE = 0.03], p <0.05),work rules 06 14 Table 2 Multilevel correlations among worksite- and employee-level variables (Nj = 23, Ni = 547) Variables Mean (SD) ICC [95% CI] 1 2 3 4 5 6 7 8 9 10 11 12 Worksite-level 1. Parking/Bike 1.28 (1.0) – 1.00 −0.16 0.23 −0.22 −0.39 0.36* −0.09 −0.47 0.31 −0.57 −0.14 −0.02 2. Signs/Bulletin boards/ 1.15 (1.2) – 1.00 −0.20 0.01 −0.15 −0.22* 0.49** 0.18 −0.36 −0.19 −0.07 0.02 Advertisements 3. Stairs/Elevator 0.74 (0.6) – 1.00 −0.35* 0.30 0.28* −0.23 −0.12 −0.59 0.10 0.16 0.53 4. Physical activity/ 0.74 (1.7) – 1.00 −0.40** 0.28 0.40 −0.04 0.33 0.22 −0.31 0.73* Fitness facilities 5. Work rules 5.04 (0.8) – 1.00 −0.28 −0.24 −0.07 −0.43 −0.22 −0.56* −0.20 6. Written policies 0.26 (1.3) – 1.00 0.18 0.51 0.32 −0.19 −0.12 0.61** 7. Health promotion 1.43 (1.9) – 1.00 0.22 0.33 −0.02 −0.07 0.02 analys Employee-level 8. Self-regulation 24.66 (10.1) 0.10* [.01, .19] 1.00 0.73 −0.36 0.54** 0.95** for physical activity 9. Self-efficacy for 11.45 (3.6) 0.00 [−0.10, 0.10] 0.41** 1.00 0.27 0.73 0.97** physical activity 10. Occupational 3.38 (17.8) 0.10 [−0.03, 0.23] 0.08 0.05 1.00 −0.25 −0.23 physical activity 11. Transport-related 9.31 (11.9) 0.15* [0.03, 0.26] 0.11** 0.17** 0.14 1.00 0.52* physical activity 12. Leisure-time 7.07 (13.0) 0.04 [−0.03, 0.10] 0.40** 0.34** 0.18 0.09* 1.00 physical activity Upper triangular matrix indicates worksite-level correlations, and lower triangular matrix indicates employee-level matrix. Nj the number of worksites, Ni a b the number of workers, ICC intra-class correlation coefficient, CI confidence interval. At baseline. At follow-up. *p < 0.05. **p < 0.01. Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 9 of 14 Table 3 The associations among workplace environment, self-regulation for physical activity, and occupational physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: occupational Model 1 Model 2 Model 3 Model 4 Model 5 physical activity at follow-up Unconditional Crude random Crude interaction Employee-level Worksite-level a b model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee-level Self-regulation for physical activity (γ ) – 0.23 (0.18) 0.13 (0.14) 0.11 (0.16) 0.11 (0.16) Worksite-level Workplace environment Parking/Bike (γ ) – −0.84 (1.33) −0.83 (1.33) −0.81 (1.20) −1.02 (1.41) Signs/Bulletin boards/Advertisements (γ ) – −0.50 (1.32) −0.49 (1.31) −0.17 (1.21) 0.70 (0.51) Stairs/Elevator (γ ) – 0.16 (2.49) 0.12 (2.50) 0.28 (2.05) 3.80 (1.80)* Physical activity/Fitness facilities (γ ) – 1.68 (1.48) 1.68 (1.48) 1.46 (1.22) 4.98 (1.09)** Work rules (γ ) – 0.34 (2.32) 0.38 (2.39) 0.12 (2.24) −1.25 (1.50) Written policies (γ ) – −0.92 (0.87) −0.89 (0.85) −0.81 (0.83) 2.10 (1.02)* Health promotion programs (γ ) – −0.52 (0.31) −0.53 (0.31) −0.53 (0.29) −0.01 (0.16) Interaction effects Parking × self-regulation (γ ) –– −0.41 (0.21)* −0.42 (0.20)* −0.43 (0.20)* Signs × self-regulation (γ ) –– 0.19 (0.18) 0.24 (0.16) 0.25 (0.16) Stairs × self-regulation (γ ) –– 0.42 (0.37) 0.34 (0.32) 0.35 (0.33) Fitness × self-regulation (γ ) –– 0.07 (0.10) 0.05 (0.09) 0.05 (0.09) Rules × self-regulation (γ ) –– 0.04 (0.32) 0.01 (0.33) 0.00 (0.32) Policies × self-regulation (γ ) –– 0.08 (0.10) 0.12 (0.10) 0.13 (0.10) Programs × self-regulation (γ ) –– −0.10 (0.07) −0.11 (0.08) −0.12 (0.08) Intercept (γ ) 3.90 (1.54)* 3.26 (1.08)** 3.27 (1.08)** 2.87 (0.98)** 7.07 (1.71)** Random intercept (τ ) 31.88 (26.72) 21.48 (12.45) 21.60 (12.37) 16.55 (13.65) 0.10 (1.39) Random slope: self-regulation (τ ) – 0.40 (0.34) 0.21 (0.11) 0.24 (0.09)** 0.25 (0.09)** Residual variance (e ) 289.41 (145.74)* 261.85 (133.85) 261.64 (133.62) 180.65 (99.80) 178.26 (96.23) ij AIC 22,792.08 18,780.05 18,786.13 18,646.91 18,629.98 Nj the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and occupational physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. (γ =0.28 [SE = 0.12], p < 0.05), and written policies Some of the components of workplace environments (γ = 0.04 [SE = 0.01], p < 0.05). With leisure-time phys- could directly increase occupational physical activity. ical activity (Table 5), self-regulation for physical activity However, transport-related and leisure-time physical ac- was significantly and positively associated (γ =0.09 tivity might not be increased by environmental modifica- [SE = 0.04], p < 0.05). Workplace environments did not tion at the workplace alone without combining with have any significant main effect. Positive interaction ef- enhancing self-regulation for physical activity. This may fects of workplace environments and self-regulation mean the necessity of conducting the studies and inter- were observed on physical activity/fitness facilities vention by domains. Another implication of this study is (γ =0.06[SE =0.03], p < 0.05) and written policies that, to promote overall physical activity among (γ =0.06[SE = 0.02], p < 0.05). white-collar workers, designing multi-component inter- ventions that include both of environmental and psycho- Discussion logical approaches might be important. This was the first longitudinal study to report multilevel The hypothesis of the association between workplace effects of workplace environment and self-regulation for environment and physical activity (H1) was supported domain-specific physical activities among white-collar mainly for occupational physical activity, suggesting the workers. The results of each analysis suggested that their direct effect of workplace environment may influence effects may differ among domain-specific activities. physical activity at work. The effectiveness of stairs/ Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 10 of 14 Table 4 The associations among workplace environment, self-regulation for physical activity, and transport-related physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: transport- Model 1 Model 2 Model 3 Model 4 Model 5 related physical activity Unconditional Crude random Crude interaction Employee-level Worksite-level a b at follow-up model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee-level Self-regulation for physical activity (γ ) – 0.13 (0.07) 0.10 (0.02)** −0.00 (0.06) −0.00 (0.06) Worksite-level Workplace environment Parking/Bike (γ ) – −2.98 (1.14)** −3.00 (1.14)** −2.48 (1.18)* −0.61 (0.98) Signs/Bulletin boards/Advertisements (γ ) – −1.00 (0.98) −0.97 (0.98) −0.82 (1.01) 0.59 (0.80) Stairs/Elevator (γ ) – 3.95 (1.96)* 3.93 (1.95)* 3.32 (1.96) 2.80 (2.01) Physical activity/Fitness facilities (γ ) – −1.10 (0.64) −1.11 (0.64) −0.95 (0.70) −0.39 (0.88) Work rules (γ ) – −7.21 (1.97)** −7.21 (1.97)** −6.87 (1.94)** −1.50 (2.89) Written policies (γ ) – −1.44 (0.63)* −1.47 (0.63)* −1.34 (0.65)* −0.72 (1.13) Health promotion programs (γ ) – −0.03 (0.35) −0.02 (0.35) −0.01 (0.30) −0.60 (0.25)* Interaction effects Parking × self-regulation (γ ) –– −0.01 (0.04) 0.07 (0.06) 0.07 (0.06) Signs × self-regulation (γ ) –– 0.07 (0.04) 0.08 (0.05) 0.08 (0.05) Stairs × self-regulation (γ ) –– −0.00 (0.10) 0.01 (0.10) −0.00 (0.11) Fitness × self-regulation (γ ) –– 0.03 (0.02) 0.05 (0.03)* 0.06 (0.03)* Rules × self-regulation (γ ) –– 0.15 (0.07)* 0.26 (0.12)* 0.28 (0.12)* Policies × self-regulation (γ ) –– 0.06 (0.01)** 0.04 (0.01)* 0.04 (0.01)* Programs × self-regulation (γ ) –– −0.02 (0.01) −0.02 (0.01) −0.02 (0.01) Intercept (γ ) 8.48 (1.21)** 10.12 (0.97)** 10.12 (0.98)** 9.37 (1.43)** 11.32 (2.05)** Random intercept (τ ) 23.19 (6.85)** 11.19 (6.22) 11.20 (6.23) 12.53 (5.98)* 0.22 (1.28) Random slope: self-regulation (τ ) – 0.00 (0.02) 0.00 (0.00) 0.00 (0.01) 0.00 (0.01) Residual variance (e ) 103.26 (14.10)** 101.92 (13.89)** 101.29 (13.65)** 80.41 (12.20)** 81.84 (10.47)** ij AIC 22,230.67 18,227.64 18,238.37 18.160.32 18,150.18 Nj the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and transport-related physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. elevator, physical activity/fitness facilities, and written employers for prompts and facilities might amplify the im- policies at the workplace has been repeatedly indicated portance of physical activity for employees, being social, by previous studies [16, 17, 45–48]. Thus, this study may and ecological supports for doing physical activity. add further evidence of the association among these The second hypotheses for self-regulation was sup- workplace environments and occupational physical ac- ported for leisure-time physical activity but not for occu- tivity, even after adjusting for worksite-level covariates pational/transport-related physical activities. A possible (i.e., worksite size and worksite location). Possible reason for the result could be difficulty to utilize their functions of these components might be enhancing regulation for activities in occupational/transport-related opportunities and accessibility for doing physical ac- settings. In most work time, employees must pay atten- tivity [16, 45, 46, 48, 49]. Especially, because using tion to their own jobs, and most activities related to jobs stairs and physical activity/fitness facilities can be directly basically occur incidentally [51]. Therefore, most em- connected to increasing the amount of physical activity at ployees may have difficulty using their psychological work, their associations with occupational physical activity functions for occupational physical activity within work- may be stronger than “soft”’ workplace environment ing time. Transport-related physical activity is also changes, such as mounting signs or establishing work largely influenced by environmental factors at both rules. Another function might be establishing social norms workplaces and communities, and most employees may at the workplace [50]. Policies and investment by establish their commuting habits. Therefore, there may Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 11 of 14 Table 5 The associations among workplace environment, self-regulation for physical activity, and leisure-time physical activity among white-collar workers (Nj = 23, Ni = 547) Dependent: leisure-time Model 1 Model 2 Model 3 Model 4 Model 5 physical activity at follow-up Unconditional Crude random Crude interaction Employee-level Worksite-level a b model slope model model adjusted model adjusted model Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE) Employee -level Self-regulation for physical activity (γ ) – 0.57 (0.09)** 0.40 (0.04)** 0.10 (0.04)* 0.09 (0.04)* Worksite-level Workplace environment Parking/Bike (γ ) – −1.53 (0.51)** −1.55 (0.52)** −1.29 (0.53)* −1.05 (0.69) Signs/Bulletin boards/Advertisements (γ ) – −0.01 (0.39) −0.00 (0.39) −0.11 (0.34) 0.55 (0.43) Stairs/Elevator (γ ) – 1.31 (1.40) 1.32 (1.41) 1.13 (1.47) −0.18 (1.21) Physical activity/Fitness facilities (γ ) – 0.76 (0.31)* 0.74 (0.31)* 0.79 (0.49) −0.03 (0.66) Work rules (γ ) −2.10 (1.11) −2.13 (1.12) −2.39 (1.39) 0.55 (1.28) Written policies (γ ) – 1.29 (0.25)** 1.20 (0.24)** 0.96 (0.37)* 0.99 (0.59) Health promotion programs (γ ) – 0.16 (0.11) 0.15 (0.11) 0.13 (0.16) −0.01 (0.13) Interaction effects Parking × self-regulation (γ ) –– −0.24 (0.08)** −0.08 (0.07) −0.07 (0.08) Signs × self-regulation (γ ) –– 0.02 (0.04) 0.04 (0.03) 0.04 (0.03) Stairs × self-regulation (γ ) –– 0.23 (0.13) 0.08 (0.13) 0.06 (0.13) Fitness × self-regulation (γ ) –– 0.05 (0.03)* 0.05 (0.03) 0.06 (0.03)* Rules × self-regulation (γ ) −0.21 (0.13) 0.03 (0.12) 0.04 (0.12) Policies × self-regulation (γ ) –– 0.10 (0.02)** 0.06 (0.02)** 0.06 (0.02)** Programs × self-regulation (γ ) –– −0.03 (0.02)* −0.02 (0.01) −0.02 (0.01) Intercept (γ ) 7.05 (0.85)** 7.09 (0.42)** 7.04 (0.44)** 8.53 (1.26)** 8.73 (1.44)** Random intercept (τ ) 6.95 (6.91) 0.16 (0.44) 0.12 (0.71) 1.35 (.2.61) 0.03 (0.20) Random slope: self-regulation (τ ) – 0.06 (0.03) 0.00 (0.02) 0.00 (0.02) 0.00 (0.01) Residual variance (e ) 162.54 (25.71)** 129.94 (21.98)** 127.78 (21.34)** 75.04 (14.59)** 73.79 (13.93)** ij AIC 22,291.35 18,202.44 18,198.14 17,986.55 17,980.61 Nj = the number of worksites, Ni the number of employees. CI confidence interval, AIC Akaike information criteria. Full Information Maximum Likelihood estimation (FIML) was used using Mplus 7.4. Adjusted by self-efficacy for physical activity, gender, age, employment status, occupational status, working hours, job strain, and leisure-time physical activity at baseline. Adjusted by employee-level covariates, worksite size, and worksite location. *p < 0.05. **p < 0.01. be few choices to commute and little need to actively are well facilitated at the workplace. For leisure-time phys- regulate their strategies for transport-related activities. ical activity, strategies of regulation are more effective On the other hand, workers could thoroughly utilize when physical activity/fitness (γ =0.06, p < 0.05) and their cognitive and behavioral regulation for physical ac- written policies (γ =0.06, p < 0.05) are well facilitated. tivity in leisure time. Considering that there was a sig- These results have a practical implication for future nificant positive relationship between self-efficacy for multi-component interventions that either environmental physical activity and physical activity in the workday modification or psychological approach alone is insufficient [23, 24], functions of self-efficacy and self-regulation to promote transport-related and leisure-time physical ac- might also differ for each domain of physical activities. tivity among white-collar workers. These workplace envi- It is interesting that the effects of workplace environ- ronments include not only internal physical environments ments may spill over into outside of the worksite through at the workplace but also external and social environments. boosting the associations between self-regulation and Therefore, they might be useful for employees to be more physical activity (H3). For transport-related physical activ- active when employees plan physical activity outside of the ity, strategies of self-regulation (γ = −0.00) are ineffective workplace. For instance, setting a goal to increase themselves (H2) but effective only when physical activity/ transport-related physical activity during lunch break can fitness (γ = 0.06, p <0.05),work rules (γ = 0.28, be more effective if employers set rules that enable em- 14 15 p < 0.05), and written policies (γ = 0.04, p <0.05) ployees to go out or if employers have a written policy that 16 Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 12 of 14 encourages employees to walk around the workplace. An- Conclusion other possible mechanism is that they enhance awareness In summary, the components of workplace environ- and build social norms for physical activity [52]. Internal- ments such as physical activity/fitness facilities, written ized awareness and norms could spill over into one’s policies, signs for stair use at stairs and elevators, and transport-related and leisure-time activities and entire life- work rules and self-regulation for physical activity may style among white-collar workers. Although the question of be effective to promote three domain-specific physical how these components moderate the association between activities directly or by augmenting the positive associa- self-regulation for physical activity and physical activity was tions between self-regulation for physical activity and not the focus of the present study, it should be investigated physical activity. Those effects may differ by activity do- in future studies. In addition, we could only investigate one main; occupational activities may be increased by work- possible model that workplace environment would exagger- place environments while transport-related and ate the effect of self-regulation for physical activity on phys- leisure-time activities may also be increased by the inter- ical activity. The other possible relationships among actions of workplace environments and self-regulation. two-level factors, such as mediation model and subgroup This study has practical implications for designing effects should be investigated in future studies. multi-component interventions that include both of en- Incidentally, some associations were negative and sig- vironmental and psychological approaches to increase nificant: the direct association between health promotion effect sizes to promote overall physical activity. programs and transport-related physical activity and the Abbreviations interaction effect of parking/bike and self-regulation for AIC: Akaike information criteria; EAT: Environmental assessment tool; physical activity on occupational physical activity. These FIML: Full information maximum likelihood; GPAQ: Global physical activity questionnaire; HLM: Hierarchical linear modeling; JCQ: Job content inverse associations were not consistent with our expect- questionnaire; PASR: Physical activity self-regulation; STROBE: Strengthening ation. Although the reasons for negative impact of the reporting of observational studies in epidemiology workplace environments for physical activity were un- known in the present study, some components of work- Acknowledgements We thank Yumi Asai, Risa Kotake, Yoshifumi Nin, Asuka Sakuraya, Utako place environment could be adverse for specific Sawada, and Peiying Tsay for their cooperation in worksite observation to domains. Not only possible benefits but also potential assess workplace environment. adverse effects of workplace environment should be in- vestigated in the future. Funding This work was supported by the Grant-in-Aid for Japan Society for the Promotion There are several limitations to this longitudinal study. of Science Fellows (15J04085). First, the response rate of worksite representatives that agreed to participate was not high (46.9%), which can Availability of data and materials The dataset used and analysed during the current study are available from cause selection bias resulting in an underestimation of the corresponding author on reasonable request. the associations. In addition, participants who were dropped were significantly younger. Although we ad- Authors’ contributions dressed attrition at follow-up by imputation of FIML, it KW had contributions to the initial conception and design. KW did the statistical analysis. KW wrote the first draft manuscript. NK, YO, and SI made may cause underestimation. Second, all employee-level revision of the manuscript. The all authors made critical comments on the variables and some worksite-level variables were mea- draft manuscript and approved its final version. sured by self-report questionnaires. Measured values Ethics approval and consent to participate contained information bias and measurement errors. It Informed consent was obtained from all representatives and workers via a has been repeatedly indicated that self-reported physical consent form and the self-report questionnaire. The consent form and activity was often distorted with actual physical activity questionnaire informed participants that we assured protection of personal information and that the data would be anonymous and only used for and often overestimated [53]. In addition, light physical academic purposes. The study protocol received ethical approval by the activity could not be measured due to the questionnaire. research ethics committee of the Graduate School of Medicine and Faculty of Furthermore, distribution of the three domains of activ- Medicine, The University of Tokyo, Japan (No. 10919). ities was skewed and might not much suitable for HLM. Competing interests Third, the results could be distorted by confounding fac- The authors declare that they have no competing interests. tors that could not be considered in this study, such as other types of job stressors [54] and environmental de- Publisher’sNote terminants outside the workplace. Finally, the samples Springer Nature remains neutral with regard to jurisdictional claims in were not extracted at random and were from a restricted published maps and institutional affiliations. area in Japan. Thus, there are limitations to the Author details generalizability of the results. These limitations need to Department of Mental Health, Graduate School of Medicine, The University be addressed by conducting higher-quality studies and 2 of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Faculty of randomized controlled trials. Human Sciences, University of Tsukuba, 3-29-1, Otsuka, Bunkyo-ku, Tokyo Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 13 of 14 112-0012, Japan. Department of Preventive Medicine and Public Health, 21. Smith L, McCourt O, Sawyer A, Ucci M, Marmot A, Wardle J, et al. A review Tokyo Medical University, 6-1-1 Shinjuku, Shinjuku-ku, Tokyo 160-8402, Japan. of occupational physical activity and sedentary behaviour correlates. Occup Med. 2016;66:185–92. Received: 28 December 2017 Accepted: 20 May 2018 22. Wang F, Orpana HM, Morrison H, de Groh M, Dai S, Luo W. Long-term association between leisure-time physical activity and changes in happiness: analysis of the prospective national population health survey. Am J Epidemiol. 2012;176:1095–100. 23. Prodaniuk TR, Plotnikoff RC, Spence JC, Wilson PM. The influence of self- References efficacy and outcome expectations on the relationship between perceived 1. Kirk AA, Rhodes RE. Occupation correlates of adults’ participation in leisure- environment and physical activity in the workplace. Int J Behav Nutr Phys time physical activity: a systematic review. Am J Prev Med. 2011;40:476–85. Act. 2004;1:7. 2. World Health Organization. Global health risks: Mortality and burden of 24. Plotnikoff RC, Pickering MA, Flaman LM, Spence JC. The role of self-efficacy disease attribute to selected major risks. 2009. http://www.who.int/ on the relationship between the workplace environment and physical healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf activity: a longitudinal mediation analysis. Health Educ Behav. 2010;37:170–85. 3. Joyce S, Modini M, Christensen H, Mykletun A, Bryant R, Mitchell PB, et al. 25. Watanabe K, Otsuka Y, Shimazu A, Kawakami N. The moderating effect of Workplace interventions for common mental disorders: a systematic meta- health-improving workplace environment on promoting physical activity in review. Psychol Med. 2016;46:683–97. white-collar employees: a multi-site longitudinal study using multi-level 4. Brown HE, Gilson ND, Burton NW, Brown WJ. Does physical activity impact structural equation modeling. J Occup Environ Med. 2016;58:178–84. on presenteesim and other indicators of workplace well-being? Sports Med. 26. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, 2011;41:249–62. et al. Strengthening the Reporting of Observational Studies in Epidemiology 5. Proper KI, Staal BJ, Hildebrandt VH, van der Beek AJ, van Mechelen W. (STROBE): explanation and elaboration. PLos Med. 2007;e297:4. Effectiveness of physical activity programs at worksites with respect to 27. DeJoy DM, Wilson MG, Goetzel RZ, Ozminkowskim RJ, Wang S, Barker KM, et work-related outcomes. Scand J Work Environ Health. 2002;28:75–84. al. Development of the Environmental Assessment Tool (EAT) to measure 6. van den Berg TI, Elders LAM, de Zwart BCH, Burdorf A. The effects of work- organizational physical and social support for worksite obesity prevention related and individual factors on the Work Ability Index: a systematic programs. J Occup Environ Med. 2008;50:126–37. review. Occup Environ Med. 2009;66:211–20. 28. Parker KB, DeJoy DM, Wilson MG, Bowen HM, Goetzel RZ. Application of the 7. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJ, Martin BW. The Lancet environmental assessment tool (EAT) as a process measure for a worksite Physical Activity Series Working Group. Correlates of physical activity: why weight management intervention. J Occup Environ Med. 2010;52:S42–51. are some people physically active and others not? Lancet. 2012;380:258–71. 29. Matson-Koffman DM, Brownstein JN, Neiner JA, Greaney ML. A site-specific 8. Sallis JF, Owen N, Fisher EB. Ecological models of health behavior. In: Glanz literature review of policy and environmental interventions that promote K, Rimer BK, Viswanath K, editors. Health behavior and health education: physical activity and nutrition for cardiovascular health: what works? Am J theory, research, and practice. 4th ed. New York: Jossey-Bass A Wiley Health Promot. 2005;19:167–93. Imprint; 2008. p. 465–85. 30. Mozaffarian D, Afshin A, Benowitz NL, Bittner V, Daniels SR, Franch HA, et al. 9. Bandura A. Self-efficacy: toward a unifying theory of self-change. Psychol Population approaches to improve diet, physical activity, and smoking Rev. 1977;84:191–215. habits: a scientific statement from the American Heart Association. 10. Bandura A. Social foundations of thought and action: a social cognitive Circulation. 2012;126:1514–63. theory. Englewood Cliffs NJ: Prentice-Hall; 1986. 31. Schröer S, Haupt J, Pieper C. Evidence-based lifestyle interventions in the 11. Prochaska JO, DiClemente CC. Stages and processes of self-change of workplace: an overview. Occup Med. 2014;64:8–12. smoking: toward an integrative model of change. J Consult Clin Psychol. 32. World Health Organization. Interventions on diet and physical activity: What 1983;51:390–5. works: Summary report. 2009. http://www.who.int/dietphysicalactivity/ 12. Rhodes RE, Pfaeffli LA. Mediators of physical activity behavior change summary-report-09.pdf among non-clinical populations: a review update. Int J Behav Nutr Phys Act. 33. Watanabe K, Kawakami N, Adachi H, Inoue S, Umstattd MR. Internal 2010;7:37. https://doi.org/10.1186/1479-5868-7-37. consistency, convergent validity, and structural validity of the Japanese 13. Sallis JF, Spoon C, Cavill N, Engelberg JK, Gebel K, Parker M, et al. Co- version of the Physical Activity Self-Regulation scale (PASR-12) among benefits of designing communities for active living: an exploration of Japanese workers: a validation study. J Occup Health. 2017;59:24–32. literature. Int J Behav Nutr Phys Act. 2015;12:30. https://doi.org/10.1186/ 34. Umstattd MR, Motl R, Wilcox S, Saunders R, Watford M. Measuring physical s12966-015-0188-2. activity self-regulation strategies in older adults. J Phys Act Health. 2009;6: 14. Hutchinson AD, Wilson C. Improving nutrition and physical activity in the S105–12. workplace: a meta-analysis of intervention studies. Health Promot Int. 2011; 35. World Health Organization. Global physical activity surveillance. http://www. 27:238–49. who.int/chp/steps/GPAQ/en/ 15. Hipp JA, Reeds DN, van Bakergem MA, Marx CM, Brownson RC, Pamulapati 36. Bull FC, Maslin TS, Armstrong T. Global physical activity questionnaire SC, et al. Review of measures of worksite environmental and policy supports (GPAQ): nine country reliability and validity study. J Phys Act Health. 2009;6: for physical activity and healthy eating. Prev Chron Dis. 2015;12:E65. https:// 790–804. doi.org/10.5888/pcd12.140410. 37. World Health Organization. Global Physical Activity Questionnaire (GPAQ): 16. Lin YP, McCullagh MC, Kao TS, Larson JL. An integrative review: work Analysis guide. http://www.who.int/chp/steps/resources/GPAQ_Analysis_ environment factors associated with physical activity among white-collar Guide.pdf workers. West J Nurs Res. 2014;36:262–83. 17. Soler RE, Leeks KD, Buchanan LR, Brownson RC, Health GW, Hopkins DH. 38. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and stages of exercise Task Force on Community Preventive Services. Point-of decision prompts to behavior change. Res Q Exerc Sport. 1992;63:60–6. increase stair use: a systematic review update. Am J Prev Med. 2010;38: 39. Oka K. Stages of change for exercise behavior and self-efficacy for exercise S292–300. among middle-aged adults. Jap J Public Health. 2003;50:208–15. Japanese 40. Kawakami N, Kobayashi F, Araki S, Haratani T, Furui T. Assessment of job 18. Choi B, Schnall PL, Yang H, Dobson M, Landsbergis P, Israel L, et al. stress dimensions based on the job demands-control model of employees Psychosocial working conditions and active leisure-time physical activity in of telecommunication and electric power companies in Japan: reliability middle-aged us workers. Int J Occup Environ Health. 2010;23:239–53. and validity of the Japanese version of the Job Content Questionnaire. Int J 19. Crespo NC, Sallis JF, Conway TL, Saelens BE, Frank LD. Worksite physical Behav Med. 1995;2:358–75. activity policies and environments in relation to employee physical activity. 41. Karasek R. Job Content Questionnaire and user’s guide. Lowell, MA: Am J Health Promot. 2011;25:264–71. University of Massachusetts at Lowell; 1985. 20. Fransson EI, Heikilä K, Nyberg ST, Zins M, Westerlund H, Westerholm P, et al. Job strain as a risk factor for leisure-time physical inactivity: an individual- 42. Yasuda M, Nakazawa K. Consideration of the color-form problem in a two- participant meta-analysis of up to 170,000 men and women: The IPD-Work dimensional oblique model. Jap J Personal. 2012;20:271–328. Japanese th Consortium. Am J Epidemiol. 2012;176:1078–89. https://doi.org/10.1093/aje/ 43. Muthén LK, Muthén BO. Mplus user’sguide (7 ed.). Los Angeles, CA: Muthén kws336. & Muthén; 1998–2015. https://www.statmodel.com/ugexcerpts.shtml Watanabe et al. International Journal of Behavioral Nutrition and Physical Activity (2018) 15:47 Page 14 of 14 44. International Labour Office. Hours of work: From fixed to flexible? Geneva: International Labour Office; 2005. http://www.ilo.org/public/english/ standards/relm/ilc/ilc93/pdf/rep-iii-1b.pdf 45. Almeida FA, Wall SS, You W, Harden SM, Hill JL, Krippendorf BE, et al. The association between worksite physical environment and employee nutrition and physical activity behavior and weight status. J Occup Environ Med. 2014;56:779–84. 46. Dodson EA, Lovegreen SL, Elliott MB, Haire-Hoshu D, Brownson RC. Worksite policies and environments supporting physical activity in midwestern communities. Am J Health Promot. 2008;23:51–5. 47. Kolbe-Alexander T, Greyling M, da Silva R, Milner K, Patel D, Wyper L, et al. The relationship between workplace environment and employee health behaviors in a South African workforce. J Occup Environ Med. 2014;56: 1094–9. 48. Watts AW, Masse LCI. access to workplace amenities associated with leisure- time physical activity among Canadian adults? Can J Public Health. 2013; 104:e87–91. 49. Engbers LH, van Poppel MN, Chin A, Paw MJ, van Mechelen W. Worksite health promotion programs with environmental changes: a systematic review. Am J Prev Med. 2005;29:61–70. 50. Tabak RG, Hipp JA, Marx CM, Brownson RC. Workplace social and organizational environments and healthy weight behaviors. PLoS One. 2015;10:e0125424. https://doi.org/10.1371/journal.pone.0125424. 51. Engelen L, Dhillon HM, Chau JY, Hespe D, Bauman AE. Do active design buildings change health behaviour and workplace perceptions? Occup Med (Lond). 2016;66:408–11. 52. Beresford SA, Bishop SK, Brunner NL, Duncan GE, McGregor DF, McLerran DF, et al. Environmental assessment at worksites after a multilevel intervention to promote activity and changes in eating: The PACE project. J Occup Environ Med. 2010;52:S22–8. 53. Watson ED, Micklesfield LK, van Poppel MNM, Norris SA, Sattler MC, Dietz P. Validity and responsiveness of the Global Physical Activity Questionnaire (GPAQ) in assessing physical activity during pregnancy. PLoS One. 2017;12: e0177996. https://doi.org/10.1371/journal.pone.0177996. 54. Kouvonen A, Vahtera J, Oksanen T, Pentti J, Väänänen AKP, Heponiemi T, et al. Chronic workplace stress and insufficient physical activity: a cohort study. Occup Environ Med. 2013;70:3–8.

Journal

International Journal of Behavioral Nutrition and Physical ActivitySpringer Journals

Published: May 31, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off