Characteristics of the built-up environment at work and sitting at work and for transportation between Japanese desk workers



Result variables

Sedentary behaviors

A validated Japanese questionnaire14th, which tries to evaluate the sitting time for six specific behaviors in three areas (in relation to work, traffic and leisure) separately for working days and non-working days (see additional table 1 for the complete questionnaire). Participants were asked to indicate their average daily sitting time for each behavior in the previous week. As a result measure, we used three specific sedentary behaviors on workdays: sitting time at work; Sitting time in cars; and sitting times in public transport. These behaviors were assumed to occur in or near the participants’ work environment, as workers engage in them on weekdays. This questionnaire showed a moderate to high test-retest reliability (intra-class correlation coefficient[ICC]= 0.83) for the work area with a one-week recall period14th. The criterion validity of the cross-domain sitting time for working days (rho = 0.57, p 14th.

Exposure variables

Perceived environment in the neighborhood of the workplace

The Japanese version of the Abbreviated Neighborhood Environment Wandering Scale (ANEWS-J) was used to measure environmental perception in the work environment. The workplace environment was defined as being within 10 to 15 minutes’ walk of the workplace. A total of six subscales were assessed: diversity of the land use mix (16 items), access to the land use mix (6 items), road connections (3 items), availability and quality of hiking / cycling infrastructures (4 items), aesthetics (4 items.)) And security against crime (5 points). The Cronbach-α, an indicator of internal consistency, for the diversity of the land use mix, access to the land use mix, road connections, the availability and quality of pedestrian / cycling infrastructure, aesthetics and crime safety was 0.91, 0.65, 0 , 64, 0.72, 0.73 and 0.56 and the subscales of residential density and traffic safety that did not apply to the study were not taken into account due to poor internal consistency (α = 0.26).fifteen. The details of the modified ANEWS-J used in this study are provided in supplemental Table 2. All subscale items were rated on a four-point scale, with the exception of those assessing the diversity of land use mixes (six-point scale). The evaluation of the subscales was carried out according to the methods of ANEWS-J, which were published online ( Higher values ​​indicate better walking ability. ANEWS-J has an acceptable test-retest reliability (ICCs = 0.76–0.96) for residential areas16. We examined the test-retest reliability of ANEWS-J for the workplace environment in a subsample of participants (n = 200). Participants reported twice within two weeks about their perception of their surroundings at work. The test-retest reliability of ANEWS-J was moderate to high for all subscales (ICC = 0.57–0.87) (Supplementary Table 3).

Objectively measured accessibility of the neighborhood at the workplace

The degree of accessibility in work environments was estimated using the Walk Score®. It is a measure of access to local destinations using a distance decay function to destinations such as grocery stores, restaurants, banks, parks and schools, with adjustment by two street connectivity metrics: intersection density and block length17th. Walk Score® can be assigned to locations (e.g. postcodes or addresses) and is normalized between 0 and 100. A higher Walk Score® indicates that more destinations can be reached on foot. Walk Score® uses open source data such as Google, and Open Street Map as source data to identify relevant goals17th. Walk Score® has been confirmed as a valid measure for assessing walking ability in the neighborhood in Japan18th. Around 60% of the participants gave their seven-digit workplace postcodes (n = 1360), 777 were unable to fully state their workplace postcodes. Each workplace zip code was manually entered into the Walk Score® website ( to get the score in July – August 2020. Walk Score® was available for 1163 participants. The website did not generate a Walk Score® for 197 participants who provided a workplace zip code due to limited data for spatial details from Japan. Since Walk Score® was negatively biased (median value = 82, 25th percentile = 63, 75th percentile = 94), we used Walk Score® as a categorical measure. We divided the participants into three groups according to the Walk Score®: car-dependent (0–69); somewhat accessible (70–89); and very accessible (90–100).


The covariates at the individual level included gender, age group (20–29, 30–39, 40–49 or 50–59 years), marital status (not married or married), level of education (tertiary education or below tertiary education), individual annual income (19th to assess the level of physical activity in three areas (work, transport and leisure). The GPAQ data were checked for valid responses using standardized procedures provided by the World Health Organization20th. The total amount of physical activity for these domains was used as the covariate. Four other participants were excluded due to a lack of data on total physical activity. Information on driving license possession (yes / no) was also collected for traffic-related sedentary behavior. The working time was assessed based on the question “How many hours did you work in the last 7 days?”. The covariate at the workplace level was the workplace size, which was measured using the self-reported number of employees at the participant’s workplace (

Statistical analysis

Differences in characteristics between the subsample categories were examined using Pearson’s chi-square tests for categorical variables and independent t-tests for continuous variables. The Spearman correlation was used to examine the correlations between the perceived properties of the built environment in the workplace and the Walk Score®, as the latter was biased.

We used linear regression models to examine the associations of workplace environment attributes with sedentary behavior in the workplace. For the associations, the non-standardized regression coefficients (β) and 95% confidence intervals (CIs) were estimated, which correspond to an increment of the standard deviation (SD) of the perceived environmental attributes. We also calculated β and 95% CI for the Walk Score® category, using the middle category (more walkable) as a reference. Each workplace neighborhood attribute was examined individually in the models. All regression analyzes were performed with Stata 15 (Stata Corp, College Station, Texas, USA) and the significance level was set at p



Comments are closed.