Participants

This cRCT will include multilevel participants: worksites and employees who are employed by the worksites. As the intervention programme will include environmental modifications at the workplace, the cluster level is each worksite. There will be no inclusion and exclusion criteria for worksites; any Japanese worksites will be able to participate in the study if they are interested in the promotion of physical activity. Employees will be considered for inclusion in this study if they are: workers employed by the included worksites; workers who are 18 years of age or older; white collar workers (managerial, professional, technical, clerical and other job types which require desk work or sitting work). There will be no exclusion criteria for participants enrolled in this study.

Procedure

Figure 1 shows a participant flow chart for this study. We will recruit more than a hundred worksites in the Kanto area through some of the health insurance associations and chambers of commerce in the area, using snowball sampling methods. The corresponding author (KW) will send invitations to these organisations, asking them to participate in the recruitment of the worksites. If they agree to assist, the corresponding author will also provide an explanation asking each worksite to participate in the study. In our previous study conducted using the same sampling methods in Kanto area, approximately half of the worksites agreed to participate in the study.44 Therefore, this study is expected to recruit 50 or more worksites. After the worksite representatives’ agree to partake in the study, nested employees will be recruited. An average cluster size will be approximately 20 employees. In this study, coordinators in each worksite will be appointed with whom we will discuss sampling methods for the workers. Some worksites will recruit workers randomly, some will recruit workers using flyers and the others will recruit workers in one of the departments. As the response rate of workers was 87.8% in our previous study, about 878 workers are expected to agree and participate in the study.44 After the worksites and employees complete the baseline survey, the worksites will be allocated randomly to the intervention or control group. The intervention programme with environmental changes will last 3 months. The postsurveys immediately after the completion of the intervention (3-month follow-up) and 6-month follow-up surveys will be conducted in both the intervention and control groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Participant flow chat.

Intervention programme

We developed an original intervention programme that consists of multicomponent environmental changes based on a validated scale (the Environmental Assessment Tool (EAT)),38 39 good practices to promote physical activity among Japanese worksites and a literature review. Table 1 shows contents of the intervention programme. The EAT, used to define the workplace environment to promote physical activity in this study, was tested both in the USA and Japan, where its reliability and validity were confirmed.38 45 Higher scores on the EAT indicate a more supportive environment for physical activity promotion and a more invested environment by employers.39 The EAT consists of three subordinate scales.38 Of these, variables for Physical Activity Assessment and Organisational Characteristics and Support can be determined and measured for promoting physical activity among employees. We referred eight items in the two subordinate scales to develop the intervention programme: parking/bike, signs/bulletin boards/advertisements, shower/changing facilities, stairs/elevators, physical activity/fitness facilities, work rules, written policies and health promotion programmes for physical activity and weight management. Additionally, we conducted qualitative interviews for Japanese worksites in the Kanto area to learn about good practices with environmental changes already being conducted to promote physical activity. From the results, the 13 elements based on 7 items of the EAT (table 1) were considered feasible to conduct at Japanese workplaces. Each operational definition of the environmental change was defined by the EAT scoring system. Finally, literature reviews were conducted to investigate rationales and functions of each item to promote physical activity among workers. Three possible functions for the items were ascertained from the literature24 27 46: building awareness and social norms around physical activity, enhancing accessibility for physical activity and enhancing individual cognitive-behavioural skills. As our resources are limited, only seven elements (Nos 1–7, table 1) will be offered free of charge to the participating worksites. The other six elements (Nos 8–13) will be optional, offered with cofunding or no funding from the research team. If the worksite is not feasible for conducting the specific items, only feasible contents will be conducted. For instance, prompts for stair use (No 4) cannot be enacted if the worksite is located in a single-storied building. Taking these conditions into account, we will discuss with the coordinators which and how many elements can be feasibly conducted at each worksite. Here, details of the seven elements that will be offered in free are described.

Intervention programme elements

Intervention worksites

Worksites in the intervention group will be offered the intervention programme described above for 3 months. At the starting stage of the intervention, policies will be implemented and informational messages will be provided to employees (Nos 1–4). The individual competition programme (No 6) and the optional elements for facilitations (Nos 8–13) will also be started. The exercise programme (No 5) and the psychological education (No 7) will also be conducted in each worksite. In addition to the intervention programme, three-time feedback (baseline, 3-month and 6-month follow-up surveys) of the assessment of the amount of physical activity and basic occupational health service in each worksite will be offered as the TAU.

Control worksites

Worksites in the control group will be offered the three-time feedback and basic occupational health service as the TAU. Worksites and employees enrolled in the control group will be put on a waiting list to receive the same intervention programme with the intervention worksites after completing the 6-month follow-up survey.

Outcomes

All outcomes, including the primary and secondary outcomes (self-regulation for physical activity, psychological distress and subjective health) will be measured at the baseline survey and at the 3-month and 6-month follow-up surveys (table 2).

Physical activity

Physical activity will be measured using the Japanese version of Global Physical Activity Questionnaire (GPAQ V.2).49 This scale is widely used and has demonstrated reliability and convergent validity among nine countries, including Japan.50 The GPAQ can assess three domain-specific physical activities in moderate-to-vigorous intensity per week in fewer items than previous questionnaires (the International Physical Activity Questionnaire, IPAQ)51 52: occupational; transportation and leisure time and sitting time in a day. Although the GPAQ were developed as a tool for population surveillance across the world at first, it has also been used for assessment of the outcomes of intervention studies.53 54 In this study, we adopted the GPAQ because of its easiness and low cost to answer, while its criterion validity with pedometers and accelerometers was poor–fair.50 There will be certain limitation for overestimation of physical activity when compared with pedometers and accelerometers.50 55 Metabolic equivalents (METs) will be used as a unit of physical activity intensity. We will calculate the total amount of physical activity per week (METs—hours/week), according to the GPAQ analysis guide.56 We will assume that physical activity is promoted for employees with higher levels of physical activity.

Self-regulation for physical activity

Self-regulation for physical activity will be measured using the Japanese version of the 12-item Physical Activity Self-Regulation scale (PASR-12).57 58 The internal consistency, convergent validity and structural validity of the Japanese version of the PASR-12 have been confirmed in a previous study.58 The PASR-12 asks the workers how frequently they used cognitive and behavioural methods for physical activity in the past 4 weeks (eg, ‘I mentally kept track of my physical activity’). The PASR-12 consists of 12 items and 6 factors: self-monitoring; goal setting; eliciting social support; reinforcements; time management and relapse prevention. All items are rated on a five-point Likert scale (1=never to 5=very often). We will calculate individual 6-factor scores and total PASR-12 scores.

Psychological distress

Questions from the Brief Job Stress Questionnaire59 will be used to measure psychological distress at work in terms of vigour (three items, eg, ‘I have been very active’), irritation (three items, eg, ‘I have felt angry’), fatigue (three items, eg, ‘I have felt extremely tired’), anxiety (three items, eg, ‘I have felt tense’) and depression (six items, eg, ‘I have felt depressed’). This scale has been widely used to assess responses to stress in Japan and has demonstrated reliability and validity.60 61 All items are rated on a four-point Likert scale (1=hardly to 4=almost). In this study, we will calculate each factor and total scores. As vigour measures a positive aspect of psychological distress, vigour scores will be reverse coded in calculating the total scores; higher scores indicate higher psychological distress.

Subjective health

Overall subjective health status will be measured using one question ‘Overall, how good is your health?’ rated on seven-point Likert scale (1=not very good to 7=very good). Higher scores indicate better subjective health status.

Sample size calculation

The required sample size was calculated according to the guidelines in the Consolidated Standards of Reporting Trials (CONSORT) for cluster RCTs,62 taking into account intraclass correlations (ICC) of the outcomes nested by the worksites (table 3). Sample sizes in cRCTs should be multiplied by design effect (1+[m-1]ρ), where m is the average cluster size and ρ is ICC.63 In our previous study,38 ICC for physical activity among Japanese employees was 0.009. In another study64 1.1% in the variance of leisure time physical activity was found in working groups among employees. Therefore, the estimated ICC for the primary outcome in this study was set to 0.01 and cluster size was set to 2065. An effect size of the intervention programme for individual physical activity was estimated based on a previous meta-analysis17 and an cRCT.34 The former meta-analysis concluded standardised mean difference (d) of workplace interventions on physical activity was 0.21. Cohen’s d of the cRCT was calculated using the reported descriptive statistics and the standardised Cohen’s d for vigorous and moderate physical activity was 0.24. The required sample size ranges between 436 and 569 employees in each arm; thus, from 22 to 29 worksites they should be recruited in the case of alpha error probability of 0.05 and a power (1-β) of 0.90, using G*Power V.3.1.9.2 (table 3).66 67

Randomisation

Enrolled worksites which meet the inclusion criteria will be randomised to intervention or control groups. The randomisation will be stratified into three strata based on worksite size (≤49, 50–299 and ≥300 employees) because the intervention effect might be different based on this factor; it has been proven that it is easier for large worksites to facilitate health and welfare systems.38 Permuted-blocked randomisation (blocked size=2) will be adopted for equal randomisation. Because each employee who participate in the study will not be notified of the result of the randomisation, assessment of the amount of physical activity (self-reported) will be blinded. On the other hand, the coordinators in the participating worksites will be notified of the result of the randomisation. Data analysis conducted by the corresponding author (KW) will also be open to blinded. A stratified permuted block random table will be created by an independent biostatistician. This table will be managed by another research assistant and will be blinded to the researchers (KW and NK).

Statistical analysis

Multilevel latent growth modelling (LGM)34 68 using robust maximum likelihood estimation will be conducted as the main analysis to examine the effects of the intervention programme on the promotion of physical activity among white collar employees (figure 2). In this study, we will determine three levels of information: repeated measures for employees at level 1, physical activity within employees at level 2 and workplace environment within worksites at level 3. However, because estimation for the total amount of physical activity at baseline and changes of physical activity can be accounted for in latent variables in LGM, the number of hierarchical levels will be 1 less than the number of hierarchical levels in other multilevel modelling approach.68 We will investigate the significance of the coefficient from a dummy variable for the intervention (control=0, intervention=1) compared with the linear slope of physical activity as the effect of the intervention programme. We will reference some model fit indices, such as χ², comparative fit index (CFI), Tucker-Levis index (TLI) and root mean square error of approximation (RMSEA). We will consider that the model demonstrates good fit if CFI and TLI exceed 0.95 and RMSEA is less than 0.06.69 Intention-to-treat analysis using full information maximum likelihood estimation will be conducted, including all employees who complete the baseline survey. When results of LGM are mis-specification or improper solutions, we will consider conducting three-level mixed model analysis using the restricted maximum likelihood estimation. Mplus V.7.470 for LGM and the PASW statistics V.18 (IBM SPSS software) for mixed model analysis will be used. Analyses for secondary outcomes (self-regulation for physical activity, psychological distress and subjective health) will also be conducted using the same methods.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Latent growth modelling for the study.  T1, enrollment; T2, 3-month follow up; T3, 6-month follow-up.

Potential subgroup analyses will be conducted, stratified by worksite size, age, gender, job categories (eg, manufacture, services, construction, transportation) and initial levels of physical activity or self-regulation for physical activity.

Some mediation analyses will also be conducted using multilevel structural equation modelling. We will model (1) the intervention dummy variable for physical activity, mediated by self-regulation for physical activity and (2) the intervention dummy variable for psychological distress and subjective health, mediated by physical activity.

Data monitoring

A data monitoring committee (DMC) will be set up, consisting of the corresponding author (KW) and the coordinator in each worksite because human resources are limited. The DMC will be held every 3 months following the randomisation in each worksite. The purpose of the meetings will be to review the participation rates and reasons for study dropout. The DMC will be independent from any sponsor and competing interest.

Ethics and dissemination

Strengths and limitations

To the best of our knowledge, this study will be the first cRCT of a multicomponent workplace intervention programme with environmental changes among Japanese white collar employees. The study will include the majority of the worksites in Japan, including small-sized worksites. In addition, operational definitions and components of the intervention programme are developed based on a validated scale. The findings will be generalisable because future intervention programmes will be developed according to the same scale. The study will also be the first to investigate the effects of a multicomponent workplace intervention both on physical activity and psychological distress.

This study is subject to several limitations. All measurements are self-reported; therefore, there will likely be measurement errors and information bias. Especially, the assessment of the amount of physical activity can be overestimated. This bias can also be applicable for conducting the specific element among the intervention programme (No 6, individual competition). Additionally, since there is a shortage in our human and monetary resources, we will not be able to recruit worksites from large areas within Japan. Sampling method will not be at random for both the worksite and the employee level, possibly causing a selection bias. Furthermore, we will not be able to offer all of our interventions for free, potentially limiting the impact of our proposed interventions. Other possible limitations will be small sample sizes and high attrition rates.