Socioeconomic position and work, travel, and recreation-related physical activity in Japanese adults: a cross-sectional study

Background The aim of this study was to examine the association between socioeconomic position and the domains of physical activity connected with work, travel, and recreation in Japanese adults. Methods A total of 3269 subjects, 1651 men (mean ± standard deviation; 44.2 ± 8.1 years) and 1618 women (44.1 ± 8.1 years), responded to an Internet-based cross-sectional survey. Data on socioeconomic (household income, educational level) and demographic variables (age, size of household, and household motor vehicles) were obtained. To examine the associations between socioeconomic position and physical activity, logistic regression analysis was used to calculate the odds ratio (OR) and confidence interval (CI) for “active” domains of physical activity. Results Men with a household income of ≥7 million yen had significantly lower work-related physical activity than the lowest income group (OR 0.51; 95 % CI, 0.35–0.75), but significantly greater travel-related (OR 1.37; 1.02–1.85), recreational (OR 2.00; 1.46–2.73) and total physical activity (OR 1.56; 1.17–2.08). Women with a household income of ≥7 million yen had significantly greater recreational physical activity (OR 1.43; 1.01–2.04) than the lowest income group. Their total physical activity was borderline significant, with slightly more activity in the high-income group (OR 1.36; 1.00–1.84), but no significant differences for work- and travel-related physical activity. Men with higher educational level (4-year college or higher degree) had significantly lower work-related (OR 0.62; 0.46–0.82), and greater travel-related physical activity (OR 1.33; 1.04–1.71) than the lowest educated group, but there were no significant differences in recreational and total physical activity. Women with a 4-year college or higher degree had significantly greater travel-related physical activity than the lowest educated group (OR 1.49; 1.12–1.97), but there were no significant differences in any other physical activity. There was no relation between working full time and physical activity in men, but women working full-time have significantly lower and not higher travel related physical activity. (OR 0.75; 0.59–0.96). Conclusions This study suggests that lower socioeconomic position is associated with more work-related physical activity, and less travel-related, recreational and total physical activity, and that this was more pronounced in men than in women.


Background
Physical activity reduces mortality and the incidence of non-communicable diseases [1,2]. However, a large proportion of the population in Japan and in many other countries remains physically inactive [3]. Promotion of physical activity is therefore a major public health concern globally. In Japan, the number of walking steps per day has decreased over time from around 1998-2000 [4], and a health promotion policy is needed to increase physical activity again. To increase participation in physical activity, it is necessary to understand the factors that influence this, and to design relevant policies and effective interventions based on that understanding.
Socioeconomic position (SEP), including for example, household income and educational level, is a determining factor of health status, and it is particularly important to promote a healthy lifestyle in low SEP populations [5]. In previous studies, lower SEP was associated with poorer health status and unhealthy behaviour among Japanese adults [6][7][8].
Studies that examined the association between SEP and total physical activity have, however, been inconsistent [9][10][11][12][13]. In a systematic review of European studies of the association between SEP and physical activity across the domains of work, travel and recreation, lower SEP was associated with greater occupational physical activity, but with less recreational physical activity [14]. It is also important to examine the association between SEP and domains of physical activity (DPA) such as work, travel, and recreation, to identify lifestyle factors that can be targeted to increase physical activity [15].
There is no comprehensive study examining the association between SEP and DPA among Japanese subjects. Japanese culture and lifestyles are different from those in Europe, so examining the relevant associations between socioeconomic position and physical activity in each domain of life in Japan is important for improving health promotion activity [16]. The association between SEP and health behaviours varies according to gender [7]; there are, however, few studies that have examined this [14]. The aim of this study was therefore to examine the association between SEP and DPA in Japanese men and women.

Participants and data collection
The data sample for this study consisted of 3269 respondents recruited to an Internet-based cross-sectional survey, which was conducted through a Japanese Internet research company, which had approximately 106,281 voluntarily-registered subjects across Japan in 2014. The company had access to detailed sociodemographic information and was able to target specific attributes. The set sample size and attributes for this study were: approximately 3000 adults, 30-59 years of age, stratified by age, gender, and household income distributions in line with the basic Japanese resident register for 2013, the 2012 Comprehensive Survey of Living Conditions in Japan [17,18]. A total of 8284 potential respondents were randomly and blindly selected and subsequently invited via e-mail to participate in the Internet-based survey. There were 3269 respondents, a 39.5 % response rate. Of these, 137 were excluded because of missing data, leaving 3132 participants (final response rate: 37.8 %). The invitation e-mail contained a link directing the potential respondents to a protected area of the website where the questionnaire was located. They could then log on using their own login identity and password. As an incentive for participation, the respondents were offered reward points valued at 100 yen (1 U.S. dollar was approximately equivalent to 102 yen and €1 to 140 yen at the time). All the respondents voluntarily completed a demographic data information form and clicked on the "agree" button at the end of an online informed consent form approved by the Ethics Committee, Waseda University. The study received prior approval from the Ethics Committee, Waseda University, Japan.

Physical activity
The Global Physical Activity Questionnaire version 2 (GPAQ v2) is a questionnaire that was developed as a modification of the International Physical Activity Questionnaire for use in multi-ethnic settings by the World Health Organization [19]. GPAQ v2 was used to estimate the total weekly quantity (in minutes) of moderateto-vigorous physical activity across three separate domains (work, travel, and recreation) lasting at least 10 min per session.
Previous study has shown that GPAQ has good testretest reliability and validity [20]. Translation of the original content of GPAQ into Japanese, followed by back translation into English ensured the reliability of the questionnaire. No changes were made to the original content and wording of the questionnaire.

Demographic variables and SEP indicator
The demographic variables obtained from the research company included gender (men and women), age (categorized as 30-39, 40-49, and 50-59 years), size of household (0, 1, 2, 3, 4, and 5 or more people) and the number of household motor vehicles (none, 1 and 2 or more). Body Mass Index (BMI) was calculated from selfreported body height and body weight using the equation BMI = body weight in kg/body height in m 2 . The socioeconomic variables included household income (categorized as <3 million yen, 3-7 million yen, ≥7 million yen), educational level (junior high and high school graduation, 2-year college degree or equivalent, and 4year college or higher degree), and employment status (non full time worker or full time worker).

Data analysis
Data from 1579 men and 1553 women who provided complete information for the study variables were analysed. All the analyses were stratified by gender. Means, standard deviations and relative frequencies (%) were used to describe the subjects. To examine the associations between SEP and DPA, a logistic regression analysis was conducted to calculate the odds ratio (OR) and confidence interval (CI) for "active" physical activity. The dependent variable was physical activity. For the analysis, each DPA was categorised into "active" (physical activity performed at least 10 min duration during one or more days a week) or "inactive" (physical activity not performed at least 10 min duration during a week). And, total physical activity was categorised by WHO physical activity recommendation (active: ≥ 150 min/week, inactive: < 150 min/week) [21]. The independent variables included in the multiple binomial logistic regression analyses were household income, educational level and employment status with adjustment for age, size in household, BMI, and household motor vehicle ownership. Significance was set at a level of p <0.05. Analyses were conducted using SPSS 21.0 for Windows (SPSS Japan Inc., Tokyo, Japan, 2012).

Results
Basic characteristics of the respondents    Table 2 also shows the associations between physical activity and employment status in men and women, with adjustment for age, size of household, body mass index, household motor vehicle ownership, and household income. Full-time working in men was not significantly related to physical activity, but women who worked full time had significantly lower travel-related physical activity (OR 0.75; 95 % CI 0.59-0.96).

Discussion
This study is the first to examine the gender-linked association between SEP and physical activity in Japanese adults. In accordance with the previous systematic review of European studies [14], this study found various associations between SEP and different DPA in Japanese adults. There were also differences in these associations between the sexes. It is therefore important to take into  account SEP, DPA, and gender when formulating proposals for the promotion of physical activity. In work-related physical activity, the men with higher household income or educational status were less active than men with lower SEP. However, the relationship between SEP and work-related physical activity was not significant in women. Previous European studies had shown a consistent association between SEP and workrelated physical activity in both sexes [14]. This difference in this study may be related to a difference in the assessment of work-related physical activity. Assessment of work-related physical activity in this study included housework, while previous studies had excluded this [22,23]. Since 70.2 % of women in this study reported that they worked part-time (data not shown), most work-related physical activity for women in this study was likely to be housework. This could account for the lack of association between SEP and workrelated physical activity in women.
In travel-related physical activity, the men with higher household income or educational status were more active than men with lower SEP. Also, the women with higher educational status were more active than women with lower educational status. However, full time employed women were less travel-related physically active than non-full time employed women The European review found no consistent association between SEP and travel-related physical activity in either sex [14]. In a Japanese study, Ishii et al. [24] reported that higher SEP was associated with more active commuting modes (e.g., walking, bicycle, public transport). In Japan, previous studies reported a significant association between the neighbourhood environment and travel-related physical activity [16,24]. The significant direct association between SEP and travel-related physical activity found in this study might therefore be affected by environmental factors. However, the reasons for the differences between men and women in this study are unknown, and further study is therefore required.
In recreational-related physical activity, both men and women who showed high household income were more active than lower household income workers. There was also a direct association between SEP and recreational physical activity in both sexes in the European review [14], and earlier Japanese studies [7,25]. Leslie et al. reported that high SEP residents had easier access to parks than lower SEP residents and used them more [26]. Psychological factors such as self-efficacy and self-control, and social factors, such as social interaction, mediate the relationship between SEP and recreational physical activity [27][28][29]. This study did not indicate why there was no significant association between educational level and recreational physical activity, though there was a trend towards an association in both sexes.
In total physical activity, both men and women who showed high household income were more active than lower household income workers. Previous Japanese studies have reported inconsistent associations between SEP and total physical activity [9][10][11][12][13]. Although high household income was associated with less work-related physical activity, this appears to be more than compensated for by greater travel and recreational physical activity.
This study had two major limitations. First, physical activity was not measured using an objective instrument. Objective measurements (using, for example, a pedometer) are generally recommended, but cannot provide data about physical activity in each domain. A selfreport questionnaire was therefore used in this study to examine the association between SEP and the different domains of physical activity. It is therefore possible that the validity and reliability of GPAQ v2 is lower than an objective measure. However, GPAQ has been standardized and is used worldwide [19,30].
Secondly, this study may have a sampling bias. Participants voluntarily registered with the Internet research company [31], and many Internet survey registrants expect an incentive from the research company. It is therefore possible that the study subjects were not representative of the local general population. This study may also have shown selection bias, since the response rate was lower in women, people who were younger and those with a lower income.

Conclusions
Although this study has some limitations, our results suggest that both men and women with higher household income are more active in recreational-related and total physical activity than the both sexes with lower income. However, the relationships between SEP and work or travel-related physical activities are different in gender. To increase physical activity in Japanese adults with lower SEP, it will therefore be important to focus on increasing travel and recreational physical activity. In a follow-up study, the mechanism of the association between SEP and DPA will be examined.