Introduction

Inactivity is a major issue in developed countries. Research has consistently shown that physical activity has a protective effect against chronic medical conditions (e.g., hypertension, heart disease, Type 2 diabetes, colon cancer, breast cancer, and osteoporosis) and premature mortality (World Health Organization [WHO], 2018). Nonetheless, approximately 23% of adults aged 18 years and over fail to attain the recommended level of physical activity (WHO, 2018), with women being less physically active than men of the same age in most countries (Edwards & Sackett, 2016; Hallal et al., 2012). The low prevalence of physical activity among women is a health issue that needs to be addressed. In Western countries, physical activity among women has been shown to decrease with age. In East Asian (including Taiwan), women aged 20–50 years have the lowest average physical activity (Bauman et al., 2012; Chao et al., 2011). In Taiwan, the prevalence of normal physical activity in the 25- to 49-years age group is relatively low (Chao et al., 2022; Sports Administration, Ministry of Education, Taiwan, ROC, 2021). The low physical activity level among working-age women has attracted considerable interest among researchers (Health Promotion Administration, Ministry of Health and Welfare, Taiwan, ROC, 2018). This is also an issue that governments must address, largely because of the increased participation of women with dependent children in the labor force.

How physical activity is affected by factors at the individual, environmental, and policy levels has been investigated over the past three decades (Bauman et al., 2012). However, level of physical activity has consistently been found to be lower among women than men (Belcher et al., 2010). Men's health status is generally explained by socioeconomic status, whereas women's health is more affected by social roles (Lahelma et al., 2002), perhaps because women often fulfill multiple roles. A number of studies have investigated the effects on women of employment status and of living with a partner and/or children (Fernández Lorca & Lay, 2020; Kuehner, 2017). However, there is a need to expand the scope of the analysis beyond single roles (Bauman et al., 2012; Prince et al., 2014), as few researchers have considered the integrated effects of multiple roles.

Two main theories, namely, role strain theory and role attachment theory, have been used to describe the impact of multiple roles on women (Fekete et al., 2019; Meighan, 2017; Shrestha et al., 2019). Role strain theory focuses on the role conflict model, which describes competing demands and obligations of multiple roles that may result in role conflict, role overload, or role contagion (Berger & Bruch, 2021). This theory posits that juggling employment and family responsibilities consumes significant time resources, demands focus and energy, and adversely affects personal well-being (Shrestha et al., 2019). The constraints on personal time involve life transitions that increase one's responsibilities and obligations such as starting or leaving a job, acquiring a mortgage, getting married, and having children (Joseph et al., 2015). Role strain theory posits that having a full-time job with dependent children is likely to lead to role strain with potentially negative effects on an individual's health (Shrestha et al., 2019).

Attachment theory takes the opposite position of role strain theory, positing that a paid job and responsibilities toward children and/or a partner are sources of social support that increase self-esteem (Fitton, 2013) and that paid employment provides income and financial independence. Improved self-esteem and greater financial independence can promote health by buffering against adverse health effects through the provision of social support (Fekete et al., 2019; Lahelma et al., 2002). Many studies have explored gender-specific differences in the relationship between multiple roles and health (Brim et al., 2019; Ju et al., 2018; O'Connor et al., 2021). Overall, the findings suggest that the multiple roles assumed by women impact their overall health significantly (Kuehner, 2017; Lahelma et al., 2002). However, the relationship between multiple roles and level of physical activity in the context of these two contending hypotheses remains uncertain.

Leisure-time physical activity (LTPA) is used to broadly describe the physical activity that an individual engages in during their free time (i.e., time not dedicated to work or commuting; Chao et al., 2011). LTPA is a public health issue with implications beyond physical activity (Lee et al., 2020). Numerous researchers have examined the impact of LTPA on health and have found evidence in support of psychosocial and spiritual benefits (Lin et al., 2010; Prince et al., 2021). Physical activity during leisure time has also been shown to slow or prevent age-related disability (Chen, Chiang, et al., 2016). LTPA is an important issue in the context of public health policy and health programs for older adults (Chen, Chen, et al., 2016; Chen et al., 2018). Nonetheless, researchers have yet to fully elucidate the underlying reasons why women lag behind men in terms of LTPA.

A growing body of research has shown the importance of single roles in analyzing physical activity levels (Chao et al., 2011; Hsu et al., 2019; Su et al., 2013). Employment status is a major factor contributing to poor health, particularly in employees engaged in low-activity occupations (Cook & Gazmararian, 2018; Quinn et al., 2020). Long working hours allow less time for LTPA (Cook & Gazmararian, 2018). Nonetheless, research on the association between marital status and physical activity has yielded mixed findings (Chao et al., 2011; Hilz & Wagner, 2018; Hull et al., 2010; Schoeppe et al., 2018). Taking on a parenting role has been shown across all races to reduce the opportunity to engage in dedicated physical activities (Edwards & Sackett, 2016; Eyler et al., 2002). Nonetheless, most research on women's single roles and LTPA has focused on the individual rather than multiple roles (Fernández Lorca & Lay, 2020; Kuehner, 2017; Lahelma et al., 2002). Therefore, the purpose of our research was to explore this issue.

Inactive lifestyle is often associated with lack of free time because of life transitions that increase personal responsibilities and obligations (Cook & Gazmararian, 2018; Joseph et al., 2015). Multiple roles refer to the responsibilities of employment and those related to the family (Lahelma et al., 2002). Multiple role theory has proven effective in several studies as a predictor of variations in health (Kuehner, 2017; Lahelma et al., 2002). However, the directions of the associations discovered have been inconsistent. The findings of several studies have found caregiving to be negatively associated with life satisfaction (Fernández Lorca & Lay, 2020), whereas other studies have found the health of women in two-parent families with children to be superior to that of women in other types of families (Kuehner, 2017; Lahelma et al., 2002). Despite extensive research on the relationship between multiple roles and female health (Brim et al., 2019; Fernández Lorca & Lay, 2020), relatively little investigation work has been done on the relationship between multiple roles and level of physical activity. The objectives of this study were to elucidate the relationship between multiple roles and LTPA from the perspective of multiple role theory and to determine whether LTPA is significantly related to the roles that women assume.

Methods Data

This study used data from the 2013 Taiwan National Health Interview Survey (NHIS), which is a large-scale cross-sectional survey based on stratified multistage systematic sampling, with samples obtained in each county/municipality (stratum) proportional to population size. This nationally representative survey was conducted using personal interviews with individuals between the ages of 12 and 64 years in private households. In addition to health-behavior-related information (e.g., smoking, drinking, exercise, betel nut chewing, eating patterns), data were also collected on respondents' health status, medical service utilization, and medical care utilization. Trained interviewers collected the data for this survey using both computer-assisted in-person interviews and computer-assisted self-administered interviews. A total of 30,960 participants completed the survey, representing a response rate of 75.2%. The focus population for this study was women aged 20–50 years, with the exclusion criteria being having difficulties with performing activities of daily living (e.g., bathing, dressing, eating, getting in and out of chairs, walking, and using the toilet), being retired or a student, having fallen or had been hospitalized within the 1-month period before the survey, and not providing their own answers. In addition, eligible individuals had to have answered the core items in both parts of the questionnaire. Thus, 5,147 respondents were qualified and included in the analysis in this study. The ethics committee of the National Health Research Institutes approved this research, and prior approval was given by the institutional review board of St. Martin De Porres Hospital (Code 19C004).

Dependent Variable: Leisure-Time Physical Activity

WHO defines physical activity as any bodily movement produced by skeletal muscles that requires energy expenditure. Physical activity may be assessed in terms of frequency, duration, or intensity (WHO, 2018). The focus in this study was on LTPAs, as these activities represent voluntary behavior that is easily altered and indicative of individual health status. LTPA has been argued as being more important to human health than other forms of physical activity (Lin et al., 2010). The metabolic equivalent (MET) score, commonly used to measure physical activity intensity (Ainsworth et al., 2011), is calculated as the ratio of the working metabolic rate to the resting metabolic rate, with 1 MET defined as the energy cost of sitting quietly (i.e., the equivalent to a caloric consumption of 1 kcal/kg per hour). The energy costs of moderate and vigorous activity are, respectively, 3–6 METs and > 6 METs (Centers for Disease Control and Prevention, 2018). On the basis of current physical activity guidelines, healthy women between 20 and 50 years old should dedicate ≥ 150 minutes per week to moderate physical activity and ≥ 75 minutes per week to vigorous physical activity or to an appropriate combination of the two (Piercy et al., 2018).

In this study, LTPA was assessed using a self-report questionnaire covering 13 types of physical activity and ranging in score from low to vigorous intensity. The question items, including “In the past week, in what types of sports did you participate?”, “How often do you engage in physical activities?”, “How much time do you spend on each physical activity?”, and “Do you experience shortness of breath when engaging in physical activity?”, were each asked once for each type of physical activity. The MET was then calculated based on the type of physical activity and breathing patterns during physical activity (Wen et al., 2007). The results for the five activities in which the respondent most frequently engaged were then summed. The respondents were categorized according to LTPA level based on the WHO recommendations, as follows:

Regular: moderate aerobic exercise (3–6 METs) for at least 150 minutes or vigorous aerobic exercise (> 6 METs) for at least 75 minutes during the previous week Insufficient: moderate aerobic exercise for < 150 minutes or vigorous aerobic exercise for < 75 minutes during the previous week Inactive: no physical activity or a combined LTPA of < 10 minutes during the previous week Independent Variable: Multiple Roles

LTPA was assessed as a function of the roles assumed by women based on employment and family status. Employment status was categorized as full-time (≥ 40 hours or more per week), part-time (≤ 39 hours per week), unemployed, and stay-at-home mother. As previously noted, retired, disabled, and others were omitted from the multivariate analysis. The distribution of employment status and other variables is listed in Table 1. Family type included living with a partner (yes for married or cohabiting and no for divorced, separated, or widowed) and living with children (0 or ≥ 1). Multiple roles were defined as having a partner, having at least one child, and being engaged in full-time employment. This variable was scored between 0 and 3, with 0 = unemployed and living alone, 1 = employed full-time or living with a partner or living with children, 2 = children and employment, partner and employment, partner and children, and 3 = employed full-time and living with a partner and children.

Table 1. - Sociodemographic Characteristics of the Respondents Variable Total Subgroups p Inactive Insufficient Regular N % n % n % n % Total 5,147 100 2,668 51.8 1,580 30.7 899 17.5 Living with partner .0916  Yes 2,825 54.9 1,448 54.3 854 54.1 523 58.2  No 2,322 45.1 1,220 45.7 726 45.9 376 41.8 Employment status .0406  Full-time a 3,389 65.8 1,794 67.2 1,042 66.0 553 61.5  Part-time 744 14.5 371 13.9 237 15.0 136 15.1  Housewife 665 12.9 327 12.3 194 12.3 144 16.0  Unemployed 349 6.8 176 6.6 107 6.8 66 7.3 Living with children .0005  0 2,177 42.3 1,132 42.4 693 43.9 352 39.2  ≥ 1 2,970 57.7 1,536 57.6 887 56.1 547 60.9 Multiple roles .0467  Single, without children, unemployed 482 9.4 255 9.6 140 8.9 87 9.7  1 (living with partner) 190 3.7 87 3.3 80 5.1 23 2.6  1 (employed) a 1,316 25.6 769 28.8 367 23.2 180 20.0  1 (living with children) 144 2.8 85 3.2 49 3.1 10 1.1  2 (living with children and employed) 269 5.2 134 5.0 76 4.8 59 6.6  2 (living with partner, employed) 334 6.5 154 5.8 98 6.2 82 9.1  2 (living with partner and children) 887 17.2 377 14.1 294 18.6 216 24.0  3 (living with partner and children and employed) 1,525 29.6 807 30.2 476 30.1 242 26.9 Age (years) < .0001  20–30 1,532 29.8 872 32.7 452 28.6 208 23.1  31–40 1,839 35.7 982 36.8 572 36.2 285 31.7  41–50 1,776 34.5 814 30.5 556 35.2 406 45.2 Household income per month (NTD) < .0001  < 50,000 1,917 37.3 1,100 41.2 528 33.4 289 32.2  50,000–100,000 1,811 35.2 884 33.1 581 36.8 346 38.5  > 100,000 856 16.6 368 13.8 307 19.4 181 20.1  Missing 563 10.9 316 11.9 164 10.4 83 9.2 Educational level (years) < .0001  0–9 2,788 54.2 1,503 56.3 838 53.0 447 49.7  10–12 1,159 22.5 655 24.6 321 20.3 183 20.4  > 12 1,200 23.3 510 19.1 421 26.7 269 29.9 Chronic disease .0040  0 4,383 85.1 2,310 86.6 1,331 84.2 742 82.5  ≥ 1 734 14.3 339 12.7 240 15.2 1,559 17.3  Missing 30 0.6 19 0.7 9 0.6 2 0.2 Depressed < .0001  Yes 1,370 26.6 790 26.6 383 24.3 197 21.9  No 3,770 73.3 1,873 70.2 1,195 75.6 702 78.1  Missing 7 0.1 5 0.2 2 0.1 0 0.0 Body mass index (BMI) .0001  < 18.5 554 10.8 327 12.3 158 0.1 69 7.7  18.5 ≤ BMI < 24.0 3,289 63.9 1,655 62.0 1,034 65.4 600 66.7  24.0 ≤ BMI < 27.0 748 14.5 367 13.8 245 15.5 136 15.2  ≥ 27.0 517 10.0 292 10.9 134 8.5 91 10.1  Missing 39 0.8 27 1.0 9 0.5 3 0.3

Note. NT$ = New Taiwan dollar.

a Full-time: ≥ 40 hours per week.

Control Variables

Several covariates were considered in this analysis, including age (20–30, 31–40, and 41–50 years), education (basic: ≤ 9 years; medium: 10–12 years; high: > 12 years), household income per month (< 50,000, 50,000–100,000, and ≥ 100,000 New Taiwan dollars), and health status. Health-related covariates included chronic disease (hypertension, diabetes, stroke, heart disease, chronic pulmonary disease, and high blood lipid; 0 = none, 1 = at least one), depression (0 = not depressed, 1 = depressed), and body mass index (BMI; underweight: < 18.5; normal: 18.5–23.9; overweight: 24–26.9; and obese: ≥ 27).

Statistical Methods

Analysis was conducted using SPSS Version 27.0 (IBM, Inc., Armonk, NY, USA) and SAS V9.1 (SAS Institute, Inc., Cary, NC, USA) for Windows. Multinomial logistic regression was used to investigate the relationship between multiple roles and types of physical activity. The results are presented as odds ratios (ORs) using the regular group as a reference category. Two models were used to determine whether the number of roles was correlated statistically with insufficient exercise or inactivity (Figure 1). Whether each role was statistically correlated with LTPA levels was then tested (Table 2), and statistical significance was set at α = .05.

Figure 1.:

Odds Ratios of Engaging in Insufficient Leisure-Time Physical Activities and Inactive by the Number of Roles

Table 2. - Employment Status of Women by Marital Status and Parental Status Variable Employee (%) Housewife (%) Other Nonemployed (%) p Marital status < .001  Married/cohabiting 62.8 35.0 2.2  Divorced/widowed/separated 75.1 15.3 9.6  Single 88.7 1.9 9.5 Married/cohabiting < .001  No children 76.8 19.2 4.0  One 65.5 31.8 2.7  Two 64.3 33.3 2.4  Three or more